doi
stringlengths 3
59
| license
stringclasses 95
values | pmc
stringclasses 0
values | pubmed_id
stringclasses 0
values | subset
stringclasses 1
value | text
stringlengths 301
1.8M
|
|---|---|---|---|---|---|
10.18632/GENESANDCANCER.23
|
CCBY
|
4162140
|
25221643
|
s2orc_pubmed_articles
|
CD133-positive cancer stem cells from Colo205 human colon adenocarcinoma cell line show resistance to chemotherapy and display a specific metabolomic profile
During the past decade, cancer stem-like cells (CSCs) have drawn substantial interest in cancer research since they have been described as major targets to improve treatment of tumors and to prevent recurrence and metastasis.In this paper, we report on the search for CSCs within the Colo205 human adenocarcinoma cell line.We describe that CD133 (prominin) was the only reliable marker for the isolation and characterization of CSCs within a Colo205 cell population.CD133-positive cells displayed many CSC characteristics, such as tumorsphere formation ability, expression of early-stage development markers, high invasiveness, raised tumor initiation potential and resistance to cisplatin chemotherapy treatment.In vitro analyses also highlighted a specific metabolomic profile of CD133-positive cells and we concluded that the chemotherapy resistance of CSCs could be related to the quiescence of such cells associated with their reduced metabolism.Furthermore, in vivo metabolome analyses suggested that a high level of circulating glutathione molecules could also promote treatment resistance.From the perspective of metabolomics, we also discuss the controversial use of serum-free in vitro cultures for CSC enrichment prior to further phenotype characterization.
# Introduction
It is now well established that cancer stem-like cells (CSCs) can be identified within in vivo tumor bulks or in vitro cell cultures thanks to several markers [bib_ref] ALDH1-positive cancer stem cells predict engraftment of primary breast tumors and are..., Ginestier [/bib_ref] [bib_ref] CD133 positive hepatocellular carcinoma cells possess high capacity for tumorigenicity, Yin [/bib_ref] [bib_ref] A human colon cancer cell capable of initiating tumour growth in immunodeficient..., O'brien [/bib_ref].Many studies have already reported the presence of CSCs in different solid tumors, such as breast, brain, prostate, lung, ovary, colon, pancreas, liver, melanoma, head and neck [bib_ref] Alternative splicing of CD44 mRNA by ESRP1 enhances lung colonization of metastatic..., Yae [/bib_ref] [bib_ref] Identification and expansion of human colon-cancer-initiating cells, Ricci-Vitiani [/bib_ref] [bib_ref] Isolation and characterization of cancer stem like cells in human glioblastoma cell..., Qiang [/bib_ref] [bib_ref] Cancer stem cells from human breast tumors are involved in spontaneous metastases..., Liu [/bib_ref].The most common membrane markers used for the sorting or analysis of CSCs are CD133, CD44 and aldehyde dehydrogenase (ALDH1), while many other specific markers have also been described, such as CD24 and epithelial-specific antigen [bib_ref] A human colon cancer cell capable of initiating tumour growth in immunodeficient..., O'brien [/bib_ref] [bib_ref] Alternative splicing of CD44 mRNA by ESRP1 enhances lung colonization of metastatic..., Yae [/bib_ref] [bib_ref] Improved purification of hematopoietic stem cells based on their elevated aldehyde dehydrogenase..., Christ [/bib_ref] [bib_ref] CD24(-/low) stem-like breast cancer marker defines the radiation-resistant cells involved in memorization..., Bensimon [/bib_ref] [bib_ref] Aldehyde dehydrogenase 1 is a marker for normal and malignant human colonic..., Huang [/bib_ref].Nevertheless expression of these markers is highly heterogeneous, depending on cancer localization, cell type and the tumor microenvironment.
CD133 (prominin) is a membrane glycoprotein that was first described on hematopoietic and neural stem/ progenitor cells.CD133 protein was later reported as a marker of poor prognosis within cancers (e.g.colorectal/ breast cancer and myeloid leukemia) and was subsequently confirmed as being specifically expressed by the CSC population [bib_ref] A human colon cancer cell capable of initiating tumour growth in immunodeficient..., O'brien [/bib_ref] [bib_ref] Isolation and in vitro expansion of human colonic stem cells, Jung [/bib_ref] [bib_ref] CD133 expression in different stages of gastric adenocarcinoma, Boegl [/bib_ref] [bib_ref] Higher percentage of CD133+ cells is associated with poor prognosis in colon..., Li [/bib_ref] [bib_ref] A subpopulation of CD133(+) cancer stem-like cells characterized in human oral squamous..., Zhang [/bib_ref].Another molecule, CD44, is expressed by a large number of mammalian cell types.This protein was first discovered on human hematopoietic stem cells and then identified in several cancers [bib_ref] Alternative splicing of CD44 mRNA by ESRP1 enhances lung colonization of metastatic..., Yae [/bib_ref] [bib_ref] CD24(-/low) stem-like breast cancer marker defines the radiation-resistant cells involved in memorization..., Bensimon [/bib_ref].Some studies also revealed that ALDH1, another common marker used for CSC identification, was also intimately correlated with tumorigenesis [bib_ref] ALDH1-positive cancer stem cells predict engraftment of primary breast tumors and are..., Ginestier [/bib_ref] [bib_ref] Improved purification of hematopoietic stem cells based on their elevated aldehyde dehydrogenase..., Christ [/bib_ref] [bib_ref] Aldehyde dehydrogenase-expressing colon stem cells contribute to tumorigenesis in the transition from..., Carpentino [/bib_ref] [bib_ref] Phenotypic characterization of murine primitive hematopoietic progenitor cells isolated on basis of..., Armstrong [/bib_ref].
Several studies have already reported the presence of CSCs within colon cancers; they were described as a rare population characterized by self-renewal capacity, clonogenicity, multipotency and chemoresistance [bib_ref] A human colon cancer cell capable of initiating tumour growth in immunodeficient..., O'brien [/bib_ref] [bib_ref] Identification and expansion of human colon-cancer-initiating cells, Ricci-Vitiani [/bib_ref] [bib_ref] Aldehyde dehydrogenase 1 is a marker for normal and malignant human colonic..., Huang [/bib_ref] [bib_ref] CD133 expression is not restricted to stem cells, and both CD133+ and..., Shmelkov [/bib_ref].The scarcity of CSCs within cancer unfortunately impedes their detection and isolation.However, it has been well established that serum-free cultures can lead to in vitro stem cell enrichment through tumorsphere formation [bib_ref] Isolation and characterization of cancer stem like cells in human glioblastoma cell..., Qiang [/bib_ref] [bib_ref] A subpopulation of CD133(+) cancer stem-like cells characterized in human oral squamous..., Zhang [/bib_ref].
Our study focused on the analysis of metabolome using capillary electrophoresis time-of-flight mass spectrometry (CE-TOF-MS).We characterized and quantified over 100 intracellular metabolites involved in human metabolic pathways.Several metabolomic approaches in cancer research have been reported yet [bib_ref] Breast cancer stem cells: tools and models to rely on, Ginestier [/bib_ref] [bib_ref] A pilot study of gas chromatograph/mass spectrometry-based serum metabolic profiling of colorectal..., Ma [/bib_ref] [bib_ref] Targeting cellular metabolism to improve cancer therapeutics, Zhao [/bib_ref] [bib_ref] Stem cells: Metabolism regulates differentiation, Mcgraw [/bib_ref] and many proteomic applications for analyzing urine or serum of patients have also been conducted, confirming the high resolution and sensitivity of such techniques for clinical diagnoses [bib_ref] Single-cell metabolomics: analytical and biological perspectives, Zenobi [/bib_ref].
In this study, we highlighted that CD133 is the only reliable marker for CSC characterization within the Colo205 colon adenocarcinoma cell line.Besides, metabolome profiles further revealed that the serumfree expansion protocol commonly used for in vitro proliferation of progenitors may create too many artifacts in cell metabolism, reducing the efficacy of such a method prior to phenotype analyses or sorting.
# Results
## Colon adenocarcinoma cell lines can form tumorspheres in vitro.
We compared the in vitro culture of cells in a basal condition (10% FBS) and in a serum-free condition.The cultures revealed that the Colo205 cell line could give rise to tumorspheres in serum-free conditions only.In contrast, cultures under FBS conditions only led to a layer of adherent confluent cells (Figure [fig_ref] Figure 1: Figure 1 [/fig_ref].To rule out the possibility that cells may aggregate due to culture at a high concentration of cells, only 100 cells were seeded in each well.Tumorspheres could also be observed under these conditions.These results confirmed that tumorsphere-like colonies could be obtained from the Colo205 cell line and expanded in serum-free medium supplemented with EGF and bFGF, even under conditions with an extra-low cell concentration.
## In vitro characterization of colo205 cell line.
As in vitro serum-free conditions could lead to floating cell enrichment and colonies, we decided to analyze the phenotype further.mRNA expression levels in Colo205 tumorspheres were not significantly different from those under basal conditions (FBS 10%), even after five weeks of culture, with regard to the expression of early-development CD133, hTERT and ABCG-2 mRNA (Figure [fig_ref] Figure 1: Figure 1 [/fig_ref].Nevertheless, immunofluorescence and cytometry analyses showed an evolution of phenotype when cells were exposed to serum-free medium.The analyses confirmed the loss of early and late differentiation markers such as nestin and cytokeratin 20 (CK20), while the expression of embryonic and stem cell markers such as oct3/4 and CD133 was increased in non-serum cultures (by two and five times, respectively, compared with the control) (Figures [fig_ref] Figure 2: Figure 2 [/fig_ref].
## Cd133+ colo205 cells exhibit csc characteristics.
To assess the stem-cell profile of different cell fractions, we further performed cell selection on the Colo205 cell line.RT-PCR analyses revealed that Colo205 CD133+ purified cells exhibited significantly increased expression of early-development mRNAs such as CD133, ABCG-2, hTERT, oct4, nanog and nestin (p<0.05)compared with basal Colo205 cells (neg.) (Figure [fig_ref] Figure 1: Figure 1 [/fig_ref].We also investigated the colony formation of both CD133+ and CD133-sorted cells from the Colo205 cell line in soft agar.Our results showed that colony formation efficiency levels for CD133+ and CD133-cells were 42.2±2.3%and 11.3±3.1% respectively, indicating that the CD133+ population displayed high clonogenicity compared with the resulting CD133-fraction (Figure [fig_ref] Figure 3: Figure 3 [/fig_ref].
## Cd133+ population from colo205 is enriched in aldh1-positive cells.
We investigated the presence of the ALDH1 marker on Colo205 cells after CD133 selection.Quantification was then performed by cytometry analyses.The results revealed that 1.7±0.2% of CD133+ cells co-expressed the ALDH1 molecule, while 0.2±0.1% of CD133-cells did.The Colo205 CD133+ cell population was thus shown to be enriched in terms of ALDH1-expressing cells.
## Cd133+ subpopulation displays increased invasiveness.
In vitro invasion assays were performed on several Colo205 sorted subpopulations using a collagen-based cell invasion method.The results showed that the CD133+ purified population was significantly more invasive than the CD133-resulting fraction or the control fraction (absorbance units (AU)=0.99,0.15 and 0.27, respectively; P<0.01).No significant difference was observed for both CD44+ and CD44-fractions (AU=0.30and 0.26, respectively) (Figure [fig_ref] Figure 3: Figure 3 [/fig_ref].These results confirmed that Colo205 CD133+ cells exhibited greater tumorigenic capacity than the CD133-and CD44+/-cells.
## Cd133+ cells are resistant to cisplatin chemotherapy.
To assess drug sensitivity, chemotherapy survival assays were performed after CD133 and CD44 selection.Cells were exposed to 10 µM cisplatin and 5-FU, and then analyzed for survival.The CD133+ cell population exhibited significantly increased resistance to cisplatin anti-tumor treatment, while no significant difference was observed concerning 5-FU drug treatment (Figure [fig_ref] Figure 3: Figure 3 [/fig_ref].These findings suggested that the Colo205 CD133+ population has increased resistance to the common colon anti-cancer chemotherapeutic drug cisplatin, while other cell populations do not show any resistance to both of the investigated drugs.
## Colo205 cd133+ cells have specific metabolome expression.
To identify specific metabolome features, CE-TOF-MS analyses were performed after cell sorting.The results showed that only the CD133+ purified fraction had a significantly different metabolome profile compared with the other cell fractions.No significant difference was observed for CD44+, CD44-, CD133-and parental Colo205 cells (Figure [fig_ref] Figure 3: Figure 3 [/fig_ref].On the one hand, the CD133+ cell fraction exhibited high overall increases in glycogen, components of the citrate cycle, nucleotides and components of co-factor metabolism pathways (from two to ten times higher than in parental Colo205 cells).On the other hand, amino acid metabolites were barely detectable within CD133+ cells compared with those in other fractions.CE-TOF-MS analyses revealed that 28 metabolites were upregulated while 53 metabolites were downregulated in CD133-positive cells (Figure [fig_ref] Figure 1: Figure 1 [/fig_ref].These metabolome results reinforced the exclusiveness of the phenotype observed within the CD133+ cell population derived from the Colo205 cell line.
## Serum-free cultures lead to restructuring of the whole metabolome.
As phenotypic differences had been highlighted among cultured Colo205 cells, we performed CE-TOF-MS on these cells.The results indicated that serum-free culture (for three to five weeks) led to overall increases in the metabolome related to glycogenosis, glycolysis, the citrate cycle, amino acid synthesis and nucleotide metabolism pathways.Compared with the control, 78 metabolites were upregulated while only five metabolites were downregulated after five weeks in serum-free culture (Figure [fig_ref] Figure 2: Figure 2 [/fig_ref].
## Cd133+ cells have a slow development rate in vitro.
We investigated the growth of unsorted Colo205, CD133+ and CD133-cell fractions in basal 10% FBS in vitro culture.While unsorted parental Colo205 cells had a population doubling time of less than 24 h (22±1.4h), CD133+ cells had a far slower development rate (126±8.7 h) (Figure [fig_ref] Figure 4: Figure 4 [/fig_ref].Cell cycle analyses revealed that CD133+ cells had a decreased transition rate out of the G 0 /G 1 phase into the S phase.In fact, less than 10% of CD133+ cells were in the S phase (6.9±0.7%), while most of them remained in the G 0 /G 1 phase (82±2.1%).In contrast, almost half of CD133-cells remained in the G 0 /G 1 phase (51.2±2.3%) or had started the S phase (43.1±1.6%)(Figure [fig_ref] Figure 4: Figure 4 [/fig_ref].
## Cd133+ cells are more tumorigenic in vivo.
To assess the in vivo tumorigenicity of CD133+ cells, various Colo205 cell fractions were injected into mice at different concentrations.BALB/c Nude mice (n=6) were subcutaneously (s.c.) inoculated with parental cells, or purified CD133+ or CD133-Colo205 cells.Tumor development was more efficient for CD133+ purified cells than for the other cell fractions (Figures [fig_ref] Figure 5: Figure 5 [/fig_ref].While only 5×10 4 CD133+ cells were shown to be able to expand and give rise to a complete tumor bulk after seven days (2/6), it took more than six weeks to observe formation of a single tumor bulk from 2×10 6 CD133purified inoculated cells (1/6) (Table .
As tumor growth from purified CD133+ cells was greater than from CD133-cells under identical conditions, secondary tumors were also found in a few mice inoculated with purified CD133+ cells only (2/12), suggesting increased metastatic potential for this cell population.No metastasis was found in "CD133-negative" and "parental" inoculated mice after 10 weeks.We also observed that 30% of the mice inoculated with CD133+ purified cells suddenly died in less than five weeks (3/10), independently of tumor size.Meanwhile, in the CD133group, no mice died, even 12 weeks after s.c.inoculation (0/10; data not shown).These findings confirmed that Colo205 CD133+ cells have higher tumorigenicity than other cell populations.
## Tumor development of colo205 cd133+ cells led to reorganization of metabolome expression in mice.
Cytometry analyses were performed on each sample to confirm the selection efficiency prior to inoculation.
CD133+ purity was higher than 97% after sorting and no CD133+ cells were detectable in the remaining CD133cell fraction (n=12).On the basis of our preliminary in vivo tumor growth development results, we chose to inject 2×10 5 CD133+ cells, while 1×10 7 cells were injected in the CD133-/parental group, so we could compare metabolite variations under similar tumor growth conditions at each stage of our study.Cytometry analyses first revealed that CD133+ purified cells gave rise to a heterogeneous population in vivo, mainly composed of CD133-cells.In fact, six weeks after inoculation of purified CD133+ cells, only 2.6% of remaining CD133+ cells could be detected within tumor bulks (2.6±0.7%;n=6).On the other hand, while no CD133+ cells could be detected in the CD133-fraction prior to injection, a few CD133+ cells were then detected in tumors after six weeks (0.29±0.15%; n=6).Thus, even if the CD133+ cell population was drastically reduced during the tumor development process, prominin-1 expression in tumors was still eight times higher in the CD133+ group than in tumors derived from the CD133-fraction (2.6% vs. 0.29%, respectively; n=6) (Table .
## Metabolome analyses were performed once a week on each mouse following s.c. injection.
Three mouse groups were thus analyzed: those with "parental" cells containing unsorted Colo205 cells, and those with "CD133+" and "CD133-" sorted populations.Despite precise CE-TOF-MS analyses (n=90), only limited variation in metabolite concentrations was highlighted among the three different groups tested.Specifically, metabolome analyses revealed that only components of the glycolysis pathway, dATP, ATP, ADP, NAD+, NADH and glutathione (mono-GSH and divalent γ-GCS forms) molecules, were significantly overexpressed in the serum of mice that developed tumors from CD133+ purified cells compared with those in tumors derived from the CD133-population (Figure [fig_ref] Figure 5: Figure 5 [/fig_ref].These results confirm the specific features of the CD133+ CSC population within the Colo205 adenocarcinoma cell line.
# Discussion
Tumor tissues have been shown to contain highly heterogeneous cell populations [bib_ref] CD133 positive hepatocellular carcinoma cells possess high capacity for tumorigenicity, Yin [/bib_ref] [bib_ref] CD24(-/low) stem-like breast cancer marker defines the radiation-resistant cells involved in memorization..., Bensimon [/bib_ref] [bib_ref] CD133 expression is not restricted to stem cells, and both CD133+ and..., Shmelkov [/bib_ref].However, the role of each population and their interrelationships remain unclear and need to be determined to establish alternative effective diagnostic methods and therapeutics.Even though the existence of CSCs remains controversial, some authors have argued that new revolutionary therapeutic approaches could be achieved by focusing on stem-like cells subpopulation [bib_ref] Breast cancer stem cells: tools and models to rely on, Ginestier [/bib_ref] [bib_ref] Targeting cellular metabolism to improve cancer therapeutics, Zhao [/bib_ref].Indeed, CSCs were shown to have high clonogenicity and increased invasiveness, being able to give rise to a complete tumor bulk and to enhance metastasis, even from a few cells remaining after surgical resection [bib_ref] Tumor-initiating cells and tumor vascularization, Zhang [/bib_ref] [bib_ref] Treatment-induced damage to the tumor microenvironment promotes prostate cancer therapy resistance through..., Sun [/bib_ref].Moreover, these cells have been shown to be more resistant to common chemotherapy and radiotherapy treatments in many solid tumors [bib_ref] A subpopulation of CD133(+) cancer stem-like cells characterized in human oral squamous..., Zhang [/bib_ref] [bib_ref] CD133 expression predicts for nonresponse to chemotherapy in colorectal cancer, Ong [/bib_ref].Thus, we postulate that eradicating such resistant cells could lead to a loss of clonogenicity and drastically reduce the risk of recurrence or metastasis after resection and treatment.
Previous studies have already described many markers to characterize CSCs [bib_ref] ALDH1-positive cancer stem cells predict engraftment of primary breast tumors and are..., Ginestier [/bib_ref] [bib_ref] CD24(-/low) stem-like breast cancer marker defines the radiation-resistant cells involved in memorization..., Bensimon [/bib_ref] [bib_ref] CD133 expression is not restricted to stem cells, and both CD133+ and..., Shmelkov [/bib_ref].Nevertheless, the presence of a universal marker for colon CSC characterization remains controversial.We demonstrated here that CD133 was the only reliable marker for phenotypic characterization of CSCs within the Colo205 cell line.In fact, the small population of CD133+ cells (1%-2%) exhibited both in vitro and in vivo characteristics of CSCs, such as high clonogenicity, increased invasiveness and high expression of specific stem cell markers (i.e.MAP2, nanog, oct3/4).Moreover, we highlighted that Colo205 CD133+ cells were more resistant to the anti-tumor drug cisplatin.Our results also showed high tumorigenic and metastatic potential of the CD133+ population after inoculation in mice.Even though the CD44v8-10 variant form was previously shown to be related to CSCs in stomach cancer, no relationship could be clearly established between CD133 and CD44 variant populations within our Colo205 colon cancer model [bib_ref] CD24(-/low) stem-like breast cancer marker defines the radiation-resistant cells involved in memorization..., Bensimon [/bib_ref] [bib_ref] CD44 variant regulates redox status in cancer cells by stabilizing the xCT..., Ishimoto [/bib_ref].This reinforces the apparent absence of a common universal surface marker for CSCs.
Several metabolome analyses have recently been conducted in the field of cancer research to discover new diagnostic markers or potential therapeutic targets [bib_ref] Breast cancer stem cells: tools and models to rely on, Ginestier [/bib_ref] [bib_ref] A pilot study of gas chromatograph/mass spectrometry-based serum metabolic profiling of colorectal..., Ma [/bib_ref] [bib_ref] Single-cell metabolomics: analytical and biological perspectives, Zenobi [/bib_ref] [bib_ref] Quantitative metabolome profiling of Illicium anisatum by capillary electrophoresis time-of-flight mass spectrometry, Urakami [/bib_ref].The high resolution and sensitivity of this technique combined with molecular analyses revealed an additional powerful tool for advanced research.Our results highlighted that Colo205 CD133+ cells that harbor a stem-cell-like phenotype also have a specific metabolome profile.CD133+ cells exhibited increased expression of components related to the glycolysis pathway, the citrate cycle pathway and co-factors.Moreover, CD133+ cells exhibited a drastically increased level of nucleoside mono-and diphosphate molecules (e.g.ADP, cAMP, AMP), while other nucleotides like ATP were slightly decreased compared with those in CD133negative cells.Additionally, amino acid content almost completely disappeared in CD133+ cells, suggesting that protein synthesis was drastically inhibited in CSCs.Cellular differentiation has been shown to be associated with pronounced downregulation of glycolysis [bib_ref] A subpopulation of CD133(+) cancer stem-like cells characterized in human oral squamous..., Zhang [/bib_ref] [bib_ref] Stem cells: Metabolism regulates differentiation, Mcgraw [/bib_ref] [bib_ref] A restricted cell population propagates glioblastoma growth after chemotherapy, Chen [/bib_ref] [bib_ref] Evidence for an alternative glycolytic pathway in rapidly proliferating cells, Vander Heiden [/bib_ref].Thus, overexpression of this metabolic pathway in CD133+ cells can be considered to be suggestive of the "non-proliferative" state of this population.Moreover, acetyl-CoA, which is required for chromatin acetylation, cholesterol and glucose-dependent lipid synthesis, was completely inhibited in CD133+ cells, in contrast to that in the control.This specific metabolomic profile and the fact that most (>90%) CD133+ cells remained in the G 0 / G 1 phase confirm the hypothesis supported by previous studies, namely, that CSCs have a propensity to remain in a relatively quiescent state [bib_ref] CD44+ slow-cycling tumor cell expansion is triggered by cooperative actions of Wnt..., Ishimoto [/bib_ref] [bib_ref] Role of glutathione in the regulation of cisplatin resistance in cancer chemotherapy, Chen [/bib_ref].We postulate here that the accumulation of mono-and diphosphate nucleosides associated with strong down-regulation of amino acid content could be associated with an energy-saving process.Indeed, cAMP, AMP and ADP can be easily converted to ATP via aerobic respiration in the mitochondria [bib_ref] The ATP synthase: the understood, the uncertain and the unknown, Walker [/bib_ref].This "pre-ATP" stock may be a versatile energy source for amino acid synthesis, which is known as a strongly endergonic process [bib_ref] Warburg revisited: regulation of mitochondrial metabolism by voltage-dependent anion channels in cancer..., Maldonado [/bib_ref].Thus, the conditions required for entry into the G 2 /S phase seem to be possible if the proliferation of CD133+ cells is needed.
In our study, CD133+ cells were shown to be resistant to cisplatin.While several mechanisms of cisplatin resistance have already been described, including changes in cellular uptake and drug efflux, increased drug detoxification, inhibition of apoptosis and increased DNA repair [bib_ref] CD133 expression predicts for nonresponse to chemotherapy in colorectal cancer, Ong [/bib_ref] [bib_ref] Role of glutathione in the regulation of cisplatin resistance in cancer chemotherapy, Chen [/bib_ref] [bib_ref] The cellular interaction of 5-fluorouracil and cisplatin in a human colon carcinoma..., Johnston [/bib_ref] , we postulate here that the metabolic quiescence and slow proliferation of CD133+ cells could also provide another opportunity for CSCs to escape from antitumor chemotherapy treatments.As most CD133+ cells stay in the G 0 /G 1 phase, a strained activation of all CSCs to differentiate could be an interesting potential therapy to explore, promoting the subsequent targeting of cells by chemotherapy drugs that are specifically aimed to affect cell division [bib_ref] Self-renewal as a therapeutic target in human colorectal cancer, Kreso [/bib_ref].
In vivo analyses also confirmed the existence of a specific metabolic profile when mice were inoculated with purified CD133+ cells.We reported a significant increase of components of the glycolysis pathway, glutathione molecules, NAD+/NADH and ATP in mouse serum after tumor development.Besides overexpression of GSH and γ-GSH molecules, which has been shown to enhance cisplatin resistance, a high concentration of circulating ATP and co-factors could also promote tumor cell proliferation [bib_ref] Role of glutathione in the regulation of cisplatin resistance in cancer chemotherapy, Chen [/bib_ref] [bib_ref] The cellular interaction of 5-fluorouracil and cisplatin in a human colon carcinoma..., Johnston [/bib_ref].This specific metabolome observed in mice indicated that a high concentration of CD133+ cells in a tumor might lead to a strong Warburg effect in vivo [bib_ref] Tumor initiating cells and chemoresistance: which is the best strategy to target..., Paldino [/bib_ref] [bib_ref] Tyrosine phosphorylation inhibits PKM2 to promote the Warburg effect and tumor growth, Hitosugi [/bib_ref] [bib_ref] Tumor bioenergetics: an emerging avenue for cancer metabolism targeted therapy, Kee [/bib_ref].As an increased rate of metastasis was observed after CD133+ cell inoculation, rearrangements of the cellular micro-environment could also lead to CD133+ progenitors circulating and enhance the development of metastasis [bib_ref] Stem cells: Metabolism regulates differentiation, Mcgraw [/bib_ref] [bib_ref] Tumor-initiating cells and tumor vascularization, Zhang [/bib_ref] [bib_ref] Effects of inflammatory factors on mesenchymal stem cells and their role in..., Liu [/bib_ref] [bib_ref] Plasticity of supporting cells in a stem cell factory, Kincade [/bib_ref] [bib_ref] Quantitative metabolome profiling of colon and stomach cancer microenvironment by capillary electrophoresis..., Hirayama [/bib_ref].
Our in vitro studies also highlighted that the metabolome profile was unnaturally elevated in tumorspheres compared with that in control cells.Almost 100 molecules were regulated in opposite ways depending on the method used for culture (i.e.10% FBS vs. serum-free) and important variations were also observed in mRNA and protein levels between CD133+ purified cells and tumorsphere-derived ones.Even though tumorsphere culture is the best way to expand and enrich CD133+ cells in vitro, we accept that the serum-free expansion protocol conventionally used for stem/progenitor cell proliferation is not reliable for use prior to CSC molecular analyses [bib_ref] Changes of the metabolism of the colon cancer cell line SW-480 under..., Hartmann [/bib_ref].
In this work, we confirmed that Colo205 tumor bulks consisted of a heterogeneous population in which two clear cell subpopulations are maintained [bib_ref] Cancer stem cells: impact, heterogeneity, and uncertainty, Magee [/bib_ref] [bib_ref] Complexity of cancer stem cells, Sugihara [/bib_ref].On the one hand, there is the CD133+ cell population, which is a small population with clear CSC characteristics: resistance to chemotherapy, slow development rate, low metabolism, strong capacity for tumor formation, tissue invasion and metastasis formation [bib_ref] A human colon cancer cell capable of initiating tumour growth in immunodeficient..., O'brien [/bib_ref] [bib_ref] Higher percentage of CD133+ cells is associated with poor prognosis in colon..., Li [/bib_ref] [bib_ref] CD133 expression predicts for nonresponse to chemotherapy in colorectal cancer, Ong [/bib_ref] [bib_ref] Expressions and clinical significances of CD133 protein and CD133 mRNA in primary..., Yu [/bib_ref].On the other hand, the remaining CD133-cells, with a rapid proliferation cycle, could barely give rise to tumors in vivo.Despite that, accurate discrimination of the stem cell subpopulations remains a difficult process.Indeed, some CD133+ cells are barely detectable as the externalization of the CD133/prominin marker on the membrane has not yet been completed at the time of the sorting/analysis process [bib_ref] The AC133 epitope, but not the CD133 protein, is lost upon cancer..., Kemper [/bib_ref].Thus, a few cells expressing CD133/prominin but displaying a false-negative phenotype cannot be detected or isolated by conventional cytometry/sorting techniques, and might contaminate the pure CD133-cell fraction.This could explain why a few CD133+ cells could be detected after the development of a tumor derived from the purified CD133-population.Even though the need for collaboration between stem and non-stem cell subpopulations for efficient proliferation/ differentiation processes in different models has already been described [bib_ref] CD133 expression is not restricted to stem cells, and both CD133+ and..., Shmelkov [/bib_ref] [bib_ref] Cord blood-derived neurons are originated from CD133+/CD34 stem/progenitor cells in a cell-to-cell..., Zangiacomi [/bib_ref] [bib_ref] CD133 negative cancer stem cells in glioblastoma, Beier [/bib_ref] [bib_ref] Characteristics of glioma stem cells, Sampetrean [/bib_ref] , interactions among CD133+ cells, CD133-cells and the tumor environment remain unclear [bib_ref] CD133 expression is not restricted to stem cells, and both CD133+ and..., Shmelkov [/bib_ref].If the differentiation of CD133+ cells giving rise to CD133-cells is proven, we also support the hypothesis that, in the absence of CD133+ cells, some CD133-cells might dedifferentiate and give rise to CD133+ CSCs able to promote tumor development and metastasis.Unfortunately, given the probable rarity of this and its occurrence only under specific in vivo conditions, spontaneous dedifferentiation is assumed to be a complicated process to observe and a major challenge to identify in future studies.
# Methods
## Cell lines and in vitro cultures
The Colo205 human adenocarcinoma colon cell line was obtained from the American Type Culture Collection (ATCC).Cells were cultured in RPMI 1640 medium supplemented with 10% fetal bovine serum (FBS, Gibco).Tumorsphere formation was performed in serum-free medium supplemented with 20 ng/mL human recombinant EGF (PeproTech) and 10 ng/mL human recombinant bFGF (PeproTech).
## Cd133+ and cd44+ cell selection
Colo205 cells were collected using collagenase (Sigma-Aldrich) and were then fractionated using a CD133 or CD44 cell isolation kit (Miltenyi Biotec).Magnetic sorting was performed at least twice for each sample.The purity of sorted cells was evaluated by flow cytometry using FACSCalibur (BD Biosciences) after labeling with anti-human CD133/2 or CD44 antibody (Miltenyi Biotec).
## Immunofluorescent staining
For intracellular staining, cells were counted, washed twice in phosphate-buffered saline (PBS, Gibco) and fixed in PBS/3.7% formaldehyde (Sigma-Aldrich) for 20 minutes at 4°C.The cells were then permeabilized with PBS/0.1% Triton X-100 (Sigma-Aldrich) for 10 minutes at room temperature.They were subsequently washed and incubated for one hour at 4°C with mouse monoclonal primary antibodies: anti-nestin (MAB1259, 5 µg/mL; R&D Systems), anti-CK20 (M7019, 1/200; Dako) and anti-sox2 (MAB2018, 5 µg/mL; R&D Systems), or with rat monoclonal primary antibody anti-oct3/4 (MAB1759, 5 µg/mL; R&D Systems).After several washes, cells were incubated for 30 minutes with the appropriate secondary antibody: anti-mouse antibody conjugated to fluorescein isothiocyanate (FITC) (115-096-062, 1/200; Jackson Immunoresearch), PE-conjugated anti-mouse antibody (115-116-068, 1/200; Jackson Immunoresearch), rhodamine-conjugated anti-mouse antibody (115-026-003, 1/200; Jackson Immunoresearch) or FITC-conjugated anti-rat antibody (112-096-003, 1/200; Jackson Immunoresearch).For membrane staining, cells were incubated for 30 minutes at 4°C with anti-CD133-PE (130-090-853, 1/200, Miltenyi Biotec), anti-CD44-PE (130-095-180, 1/200, Miltenyi Biotec) or primary rat monoclonal antibody anti-human CD44v8-10 (ALG011, 1/1000, Link Genomics).For anti-CD44v8-10 detection, cells were further stained with secondary FITC-conjugated anti-rat antibody (112-096-003, 1/200; Jackson Immunoresearch).Nuclear DNA staining was also performed with 4',6'-diamidino-2-phenylindole (DAPI; D8417, 1/5000; Sigma-Aldrich).Cells were then washed and viewed under a fluorescent microscope.Negative controls with anti-human IgG antibodies were used to discard false-positive cells in the immunofluorescent staining (Jackson Immunoresearch).Aldefluor staining was also performed to quantify ALDH1-positive cells (Aldagen).
# Flow cytometry analysis
To assess the sorting process of the Colo205 cell line, selected cells were stained by both relevant monoclonal antibodies: PE-conjugated AC133/2-PE antibody (130-090-853, 1/200; Miltenyi Biotec) and PE-conjugated CD44-PE antibody (130-095-180, 1/200; Miltenyi Biotec), or with similarly conjugated isotype-matched antibodies, for 30 minutes at 4°C.Cytometry analyses revealed that 95.2±2.9%(n=19) of the selected cells were CD133+ after sorting, while 94.3±3.6% (n=15) of the selected cells were CD44+ after sorting.Furthermore, no CD133+ or CD44+ cells could be detected within the respective remaining negative fractions (i.e.CD133-or CD44-).
## Reverse-transcription polymerase chain reaction (rt-pcr)
Total RNA was isolated using Trizol Reagent (Invitrogen), according to the manufacturer's procedures.Primers were synthesized by Sigma-Aldrich.The specific oligonucleotide primers for the GAPDH gene were: TGA AGG TCG GAG TCA ACG GAT TTG G (sense) and CAT GTA GGC CAT GAG GTC CAC CAC (antisense), for the nestin gene: AGG ATG TGG AGG TAG TGA GA (sense) and TGG AGA TCT CAG TGG CTC TT (antisense), for the CD133 gene: TTA CGG CAC TCT TCA CCT (sense) and TAT TCC ACA AGC AGC AAA (antisense), for the oct4 gene: CGC ACC ACT GGC ATT GTC AT (sense) and TTC TCC TTG ATG TCA CGC AC (antisense), for the ABCG2 gene: CTG AGA TCC TGA GCC TTT GG (sense) and TGC CCA TCA CAA CAT CAT CT (antisense), for the Nanog gene: AAT ACC TCA GCC TCC AGC AGA TG (sense) and CTG CGT CAC ACC ATT GCT ATT CT (antisense), and for the hTERT gene: AGC CAG TCT CAC CTT CAA CCG C (sense) and GGA GTA GCA GAG GGA GGC CG (antisense).RNA was quantified with the Qubit RNA BR Assay Kit (Invitrogen).cDNA was synthesized using Prime Script 1 st strand cDNA Synthesis Kit (Takara Bio).The PCR reaction mixture contained 5 μL (20 μM) of the above specific primers, 5 μL of Taq DNA Polymerase, 16 μL of 4×dNTP, 20 μL of 10× buffer, 20 μL of cDNA and 133 μL of ddH 2 O.The conditions used were as follows: denaturation at 95°C for 5 min; 30 cycles of annealing at 65°C (for GAPDH and hTERT), 63°C (for oct4 and nanog), 60°C (for ABCG2), 56°C (for nestin) or 54°C (for CD133) for 15 s, and extension at 72°C for 1 min; and then heating at 72°C for 7 min.The PCR products were resolved on a 2% agarose gel containing fluorescent nucleic acid gel stain GelRed™ (Biotium).Acquisition of gel pictures and quantification were then performed with ImageQuant TL 7.0 (GE Healthcare).
## Colony formation assay
Colony formation assay was performed in soft agar.A base layer was prepared by mixing 1% soft agar (Invitrogen) in the medium.Then, cells were suspended in growth medium containing 0.3% soft agar and seeded upon the base layer at a density of 2500 cells per well.All experiments were conducted at least in triplicate.
Plates were maintained at 37°C in a humidified 5% CO 2 incubator and medium was added every three days.After three weeks, colonies (>10 cells) were counted under a microscope.
## Cell invasion assay
QCM Collagen-based Cell Invasion Assay Kit (Chemicon, Millipore) was used following the manufacturer's procedures.Cells were seeded into the upper insert at 1×10 5 cells per insert in serum-free medium.Outer wells were filled with RPMI medium containing 10% FBS as a chemoattractant.Cells were then incubated for 48 hours.Non-invading cells were removed by swabbing the top layer of collagen and cells able to migrate through the gel insert to the lower surface of the membrane were stained, solubilized and quantified by colorimetric measurements at 560 nm (Glomax Multidetector System, Promega).All experiments were conducted in triplicate.
## In vitro drug sensitivity assays
Drug sensitivity was evaluated using cisplatin and 5-fluorouracil (5-FU) (Sigma-Aldrich).In accordance with the manufacturer's instructions, cells were exposed to 10 µM cisplatin or 10 µM 5-FU for 72 h.Viability was then evaluated using the CellTiter 96 AQueous proliferation assay kit, by measuring absorbance at 490 nm (Promega).Experiments were conducted in triplicate for each sample.
## Doubling time calculation
Cells were plated at 10 5 cells/well in six-well plates and manually counted with a hemocytometer.Doubling time (Td) was calculated using the following equation: Td=(t2-t1)×log(2)/log(q1/q2), where q1 and q2 represent the numbers of cells at times t1 and t2, respectively (n=6).
# Cell cycle analysis
The cell cycle was evaluated by cytometry using propidium iodide (PI) solution (Sigma-Aldrich).After fixation in 70% ethanol solution, cells were washed twice in PBS and stained using 250 µl of RNase solution (2 mg/ ml, Sigma-Aldrich) added to 250 µl of PI solution (0.1 mg/ml in 0.6% Triton-X in PBS) for 45 minutes in the dark at room temperature.Cells were then transferred through capped tubes to avoid clumps during fluorescence detection.Samples were kept on ice and protected from light until cytometry analysis (Accuri C6, BD Biosciences).
## In vivo analyses
Male BALB/c mice (all six weeks of age) were obtained from the National Cancer Institute (Frederick, MD, USA).The mice were subcutaneously (s.c.) injected with 1×10 7 Colo205 cells.CD133+ and CD133-cells were also injected after sorting (5×10 4 to 2×10 6 cells).At specified times after tumor inoculation, the mice were euthanized in a CO 2 chamber and tumor cell suspensions were prepared from solid tumors by enzymatic digestion as follows.Minced tissues (<1 mm 3 ) from tumors were incubated at 37°C for 90 minutes in standard medium containing 2% FBS, 50 U/ml collagenase I, 100 U/ ml collagenase IV, 200 U/ml DNase-I and 2.5 U/ml protease XIV.Cells were then harvested for viability and immunocytometry characterization.
## Capillary electrophoresis-time of flight-mass spectrometry analyses (ce-tof-ms)
Metabolite standards, instrumentation and CE-TOF-MS conditions.Instrumentation and CE-TOF-MS conditions followed the Human Metabolome Technologies guidelines (HMT Inc.).CE-TOF-MS experiments were performed on an Agilent capillary electrophoresis system coupled to an Agilent 6224 CE/TOF-MS analyzer (Agilent Technologies).System control and data acquisition were processed using the Agilent Chemstation software.CE-TOF-MS was conducted in both positive and negative ion modes for each sample.Molecule separation was processed in fused-silica capillaries filled with 1 M formic acid as a background electrolyte.Samples were injected at 50 mbar, and a voltage of 27 kV (for cation mode) or 30 kV (for anion mode) was applied.Capillary temperature was maintained at 20°C while sample tray temperature was kept below 10°C.Sheath liquid was delivered at 10 μL/min.Capillary voltage was set at 4 kV for cation mode and 3.5 kV for anion mode; the flow rate of nitrogen gas (heater temperature 300°C) was set at 0.35 bar.Fragmentor, skimmer and Oct RFV voltages were set according to HMT setting recommendations (HMT Inc.).Exact mass data were acquired at a rate of 1.5 cycles per second over the range of 50-1000 m/z.
## Processing of ce-tof-ms data
Raw data were extracted with MassHunter software (Agilent Technologies).Data processing was then performed using MasterHands software developed by HMT (MasterHands v2.8.0.3,HMT Inc.).Data analysis included noise-filtering, baseline correction, peak detection and integration of the peak area from sliced electropherograms (width 0.02 m/z).Alignment was carried out and accurate m/z values were determined for each detected peak.All peak areas were then quantified by comparison to the values of internal standard molecules (relative area) to normalize signal intensities among multiple measurements.Undetected peaks with a threshold signal-to-noise ratio of 2 were given a peak area of 0 and then discarded.
## Metabolite identification
To discriminate metabolites of interest, peak identification was carried out based on matched m/z values and normalized migration times of standard compounds (standard mixtures H3301-10024, HMT Inc.).After alignment, all data were processed with MasterHands software to confirm and refine the surface of each peak according to the m/z values.
## Mapping process
Statistical analysis was carried out by calculating mean error and standard deviation for each group.Results were finally processed with Vanted software (V1.9) to create metabolic pathway mapping, including all metabolites of interest.
[fig] Figure 1: Figure 1: Serum-free cultures enrich Colo205 cells in CSCs. A. Colo205 cells cultured in 10% FBS or serum-free conditions.Scale bar = 50 µm.B. Relative expression of ABCG-2, nanog and hTERT mRNA of Colo205 cells grown under 10% FBS conditions (control), CD133+ sorted cells and serum-free growing cells (week 1 to week 5).C. Immunofluorescence analyses of nestin, CD133, CK20 and Oct3/4 proteins.Images show 10% FBS Colo205 growing cells (control) and serum-free growing cultures (week 1 to week 5).Scale bar = 5 µm. [/fig]
[fig] Figure 2: Figure 2: Serum-free cultures lead to the loss of early and late development markers and increase of stemlike markers.A-B.Cytometry analyses of Nestin, CK20, CD133, Oct3/4, CD44 and CD44v9 on 10% FBS Colo205 growing cells (control) and serum-free growing cultures (week 1 to week 5). [/fig]
[fig] Figure 3: Figure 3: Colo205 CD133+ cells exhibit the phenotypic profile of CSCs. A. Tumorsphere evaluation of CD133+ and CD133-sorted cells.B. Invasiveness of different cell fractions measured using a collagen-based invasion kit (relative DO measured at 560 nm).C. Survival assay after chemotherapy treatment with cisplatin and 5-FU.D. Metabolite quantification after CE-TOF-MS experiments.Only metabolites of interest for Colo205 cells cultured in 10% FBS (control; green bar), CD133+ sorted cells (red bar) and CD44+ sorted cells (yellow bar) are reported here. [/fig]
[fig] Figure 4: Figure 4: Growth evaluation of CD133+ cells derived from Colo205. A. Doubling time calculation of unsorted Colo205 cells vs. CD133+ sorted cells.B. Cell cycle analysis (PI staining) of Colo205 CD133-and CD133+ sorted cells. [/fig]
[fig] Figure 5: Figure 5: CD133+ cells are more tumorigenic in vivo than unsorted Colo205 cells.A. Evaluation of mouse weight after cell inoculation.B. Tumor size evaluation after inoculations of parental Colo205 cells (unsorted) and CD133+ cells in Balb/c mice.C. Metabolite quantification after CE-TOF-MS experiments.Metabolites were quantified from mouse serum after tumor development from Colo205 CD133+ sorted cells (purple bar) and parental unsorted Colo205 cells (green bar).Metabolites of interest reported here were quantified six weeks after s.c.inoculation. [/fig]
|
10.1186/s12919-016-0008-y
|
CCBY
|
5133484
|
27980614
|
s2orc_pubmed_articles
|
Omics-squared: human genomic, transcriptomic and phenotypic data for genetic analysis workshop 19
Background: The Genetic Analysis Workshops (GAW) are a forum for development, testing, and comparison of statistical genetic methods and software. Each contribution to the workshop includes an application to a specified data set. Here we describe the data distributed for GAW19, which focused on analysis of human genomic and transcriptomic data. Methods: GAW19 data were donated by the T2D-GENES Consortium and the San Antonio Family Heart Study and included whole genome and exome sequences for odd-numbered autosomes, measures of gene expression, systolic and diastolic blood pressures, and related covariates in two Mexican American samples. These two samples were a collection of 20 large families with whole genome sequence and transcriptomic data and a set of 1943 unrelated individuals with exome sequence. For each sample, simulated phenotypes were constructed based on the real sequence data. 'Functional' genes and variants for the simulations were chosen based on observed correlations between gene expression and blood pressure. The simulations focused primarily on additive genetic models but also included a genotype-by-medication interaction. A total of 245 genes were designated as 'functional' in the simulations with a few genes of large effect and most genes explaining < 1 % of the trait variation. An additional phenotype, Q1, was simulated to be correlated among related individuals, based on theoretical or empirical kinship matrices, but was not associated with any sequence variants. Two hundred replicates of the phenotypes were simulated. The GAW19 data are an expansion of the data used at GAW18, which included the family-based whole genome sequence, blood pressure, and simulated phenotypes, but not the gene expression data or the set of 1943 unrelated individuals with exome sequence.
# Background
Genetic Analysis Workshop 19 (GAW19) concentrated on approaches to identify and characterize loci and genetic variants influencing quantitative and complex phenotypes through analysis of genome sequence and gene expression levels. Data distributed for the workshop included whole genome sequence (WGS) and gene expression in 959 individuals in a set of 20 Mexican American families [bib_ref] Data for Genetic Analysis Workshop 18: human whole genome sequence, blood pressure,..., Blangero [/bib_ref] and whole exome sequence in a set of 1943 unrelated Mexican American individuals drawn from a larger multi-ethnic case/ control study [bib_ref] The genetic architecture of type 2 diabetes, Fuchsberger [/bib_ref]. The genotype calls provided for odd-numbered autosomes previously underwent quality control screening and were pre-cleaned. Both real and simulated phenotype data were provided. The real phenotypes concentrated on systolic and diastolic blood pressure and related covariates, including age, year of examination, use of antihypertensive medications, and tobacco smoking. Simulated phenotypes were designed to mimic various aspects of these real phenotypes, including the distribution of the quantitative traits, overall heritabilities, and correlations between traits. Both real and simulated phenotypic data from the family sample were longitudinal, with up to four time points in the real data and three time points in the simulated data. The unrelated data set was cross-sectional, with a single time point available in both the real and simulated data sets.
# Methods
## Whole genome sequence in families
The family data set distributed for GAW19 is an expanded version of that used for GAW18, which has been previously described in detail [bib_ref] Data for Genetic Analysis Workshop 18: human whole genome sequence, blood pressure,..., Blangero [/bib_ref]. We provide here only a brief summary of this data set. The core of the family data set was WGS data for 464 individuals in 20 large Mexican American families drawn from Type 2 Diabetes Genetic Exploration by Next-generation sequencing in multi-Ethnic Samples (T2D-GENES) Project 2, and measures of systolic and diastolic blood pressure. Called WGS variants were available for 464 key individuals. These individuals were selected to provide comprehensive data on all alleles present in a pedigree and their phase, with two parents and one child per sibship sequenced when possible or multiple children sequenced when one or more parents was unavailable. Genotypes were imputed for the remaining 495 family members based on a genome-wide association study (GWAS) framework of dense SNPs genotyped in all family members. Directly typed or imputed WGS data for odd numbered autosomes were provided for 959 individuals in these 20 families. Systolic and diastolic blood pressures were measured at up to four time points over a span of 20 years, and were available for 932 of the 959 family members. Three or more measures were available for 503 individuals (52.5 %) and 686 individuals (73 %) had at least two measurements. Hypertension was defined as systolic blood pressure (SBP) > 140, diastolic blood pressure (DBP) > 90, and/or use of antihypertensive medications at that examination. The prevalence of hypertension varied from 18 % at the first exam to 52 % at the fourth exam as the cohort aged. Accompanying covariate data included sex, age at examination, year of examination, current use of antihypertensive medication, and current tobacco use. Year of examination was provided to allow for examination of temporal trends as the examinations spanned a 20-year period.
## Gene expression in the family data set
For GAW19, the GAW18 family data set was expanded with the addition of genome-wide gene expression measures in a subset of the T2D-GENES WGS families drawn from the San Antonio Family Heart Study (SAFHS). Measures of gene expression were generated using version 1 of Illumina Sentrix Human Whole Genome (WG-6) microarrays containing 47,293 probes in total. The SAFHS transcript data set is described in G ring et al. [bib_ref] Discovery of expression QTLs using large-scale transcriptional profiling in human lymphocytes, Göring [/bib_ref] and details of laboratory procedures can be found there. However, the GAW19 data set was constructed using a somewhat different analytical processing pipeline than that described previously.
Briefly, gene expression data were generated from peripheral blood mononuclear cells (PBMCs) from 1,371 samples in total (including controls of various types and duplicate samples). Based on the per-sample number of probes with significantly detectable expression level (counts of reported "detection p-values" of less than or equal to 0.05 across all probes), the mean raw expression level (reported "average signals") across all probes, and the mean correlation of any sample against all other samples (in the reported "average signals" of all probes), 1244 unique samples (out of 1280 in total) were identified as yielding expression data of adequate quality and were kept for further processing. Of these 1244 high quality samples, 647 come from individuals in the 20 T2D-GENES WGS families and were included in the GAW19 family data set.
Among these samples, we tested separately for each probe whether there was significant detectable expression, by conducting a binomial test based on counts of samples with and without reported "detection p-values" of less than or equal to 0.05. Subsequently, we calculated the false discovery rate (FDR) across all probes. A total of 20634 transcripts were significant at a FDR of 0.05 and were kept for further processing. Subsequently, we shifted all "average signals" upwards so that the observed minimum value (in any probe in any sample) was 1.0, conducted a log2 transformation followed by a quantile normalization transformation. The resulting data were distributed for GAW19.
## Exome data in unrelated individuals
The second data set distributed for GAW19 was a large sample of unrelated Mexican American individuals drawn from T2D-GENES Project 1 [bib_ref] The genetic architecture of type 2 diabetes, Fuchsberger [/bib_ref]. Overall, this project included 10,000 whole exome sequences from five ancestry groups, with approximately 1000 cases with T2D and 1000 controls from each group. This project was designed to study the role of uncommon variation and to identify potential functional variants behind previously identified GWAS signals.
The GAW19 exome data set included 1,943 Hispanic samples whole-exome sequenced as part of T2D-GENES Project 1. These samples were drawn from five separate family-based studies, the San Antonio Family Heart Study [bib_ref] Genetic and environmental contributions to cardiovascular risk factors in Mexican Americans. The..., Mitchell [/bib_ref] , the San Antonio Family Diabetes/Gallbladder Study [bib_ref] Genome-wide linkage analyses of type 2 diabetes in Mexican Americans: the San..., Hunt [/bib_ref] , the Veterans Administration Genetic Epidemiology Study [bib_ref] Genome-wide linkage scan for genes influencing plasma triglyceride levels in the Veterans..., Coletta [/bib_ref] , the Family Investigation of Nephropathy and Diabetes study family component [bib_ref] The Family Investigation of Nephropathy and Diabetes (FIND): design and methods, Knowler [/bib_ref] , and a study from Starr County, Texas [bib_ref] Diabetes among Mexican Americans in Starr County, Hanis [/bib_ref] [bib_ref] Genome-wide association and meta-analysis in populations from Starr County, Texas, and Mexico..., Below [/bib_ref]. Approximately 75 % of the sample of unrelated individuals came from the Starr County study. The 1943 individuals include 1021 with T2D and 922 non-diabetic controls. Information on T2D diagnosis was not provided as part of the GAW19 data set.
Phenotypic data provided included year of examination, SBP, DBP, and use of anti-hypertension medication; however, some of these variables were only available from a subset of the studies in the data set. Year of examination was available for 409 individuals and ranged from 1991 to 2012. Use of anti-hypertensive medications was available for 407 individuals, of whom 147 were on such medications and 260 were not. A total of 1851 individuals had measurements of systolic and diastolic blood pressure. SBP ranged from 66-213, with a mean of 125. DBP ranged from 32-123, with a mean of 73.5. These values are similar to those in the GAW19 family data set, where mean SBP was 122-128 across the four examinations and mean DBP was 71-78. Data on nicotine use, provided for the GAW19 family sample, were not available for the exome sample.
Exomic regions were isolated using Agilent Truseq capture reagents, and individually-barcoded samples were sequenced on Illumina HiSeq2000 instruments. Across the coding sequence of 18,281 genes the average read depth was 81.7-fold. Sequence reads were processed and aligned to the reference genome (hg19) with Picard (http://broadinstitute.github.io/picard/). Polymorphic sites and genotypes were called with GATK [bib_ref] The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing..., Mckenna [/bib_ref].
Samples and variants were excluded on the basis of multiple quality control metrics: array genotype concordance (where available), mean heterozygosity and homozygosity, high singleton counts for samples, Variant Quality Score Recalibration (VSQR) for single nucleotide variants (SNVs), and hard filtering for small insertion-deletion variants (INDELs). Within each ethnicity in the overall T2D-GENES Project 1 data set, variants were excluded on the basis of call rate (<90 % in any study in ancestry group), deviation from Hardy-Weinberg equilibrium (exact p < 10 −6 in any study in ancestry group) or differential call rate between T2D cases and controls (p < 10 −4 in all studies combined across ancestry group). Autosomal variants that passed extended QC and with MAF > 1 % in all ancestry groups were used for trans-ethnic kinship analyses. Identity-by-state (IBS) sharing between each pair of samples was calculated on the basis of independent variants (trans-ethnic r2 < 0.05) and axes of genetic variation were constructed through principal components analysis implemented in EIGENSTRAT [bib_ref] Principal components analysis corrects for stratification in genome-wide association studies, Price [/bib_ref] to identify ethnic outliers. Only individuals in the Mexican American subset of T2D-GENES Project 1 were included in the GAW19 exome data set.
For GAW19, variant call format (VCF) files were provided for odd-numbered autosomes. These included genotype calls in the NALTT field, which contained only high quality (GQ > 20) genotypes scored as 0/1/2, genotypes subjected to only minimal quality control in the GT field, and likeihood-based estimates of allele dosages in the DOS field. These VCF files included 1,689,048 SNVs (some of which were multiallelic) and 76,397 INDEL variants. Because the Mexican American exome data set provided for GAW19 was drawn from T2D-GENES' larger multi-ethnic case/control sample, some markers included in the GAW19 exome data were monomorphic as they varied in the overall T2D-GENES sample but not in the Mexican American subset. Considering only SNVs with at least five observed copies of the minor allele, which might be individually analyzed, [fig_ref] Table 1: Variant annotation, by minor allele frequency category, for variants present in at... [/fig_ref] shows the distribution of these variants across annotation categories in the unrelated exome sample and in the family WGS sample. In general, although the family sample had fewer individuals sequenced, there were more variants present in five or more copies in each annotation category. Some general patterns are similar across the unrelated exome and family WGS data sets, with 51 % of coding variants in each case being non-synonymous and with the proportion of SNVs with minor alleles frequencies < = 1 % increasing when moving from coding variants to non-synonymous variants to variants rated as highly deleterious by PolyPhen-2 [bib_ref] A method and server for predicting damaging missense mutations, Adzhubei [/bib_ref].
Although the exome sample was intended to be unrelated individuals, analysis of the SNV-based kinship estimates among individuals in the GAW19 unrelated data set shows that there are a few relative pairs present in the sample. While most of these relationships are third degree or more distant, a few first-or second-degree relative pairs are present in the GAW19 exome sample.
## Simulated phenotypes
A set of simulated phenotypes was constructed using the same model in both the exome and family data sets to closely match the observed phenotypic data. Simulated SBP and DBP had the same mean, variance, heritability and correlations with each other as observed in the real data. The observed data were also used to model covariate effects, with blood pressures being higher in males than females and increasing with age. A total of 200 replicates of the simulated phenotypes were generated. For the WGS family data set, phenotypes were simulated longitudinally, at three time points at 5-year intervals. Genetic parameters remained the same across all three exams and random environmental components were given a correlational structure based on that seen in the real data. In the exome data set of unrelated individuals, simulated SBP and DBP were modeled for a single time point, with the same mean, variance, age, sex, and medication effects as in the family sample. The age and sex of each individual were drawn from the real data and did not vary across the 200 simulation replicates that were generated.
'Functional' genes for the simulation were selected based on correlations of gene expression with measures of SBP and DBP in the SAFHS. To meet inclusion criteria, a gene's expression levels in the SAFHS had to be both phenotypically and genetically correlated with observed SAFHS SBP or DBP. Within these selected genes, non-coding variants within 5 kb upstream and downstream of the gene that were associated with expression levels of that gene were declared 'functional' for the simulations as were coding variants predicted by PolyPhen-2 [bib_ref] A method and server for predicting damaging missense mutations, Adzhubei [/bib_ref] to be deleterious. Effect sizes in the simulation for each SNP were determined using the observed correlation between mRNA levels and blood pressure in the SAFHS for the non-coding variants and using a function of PolyPhen-2 score (PP2S) for the coding variants:
[formula] percentile of ranked PP2S ð Þ Â PP2S 2 À Á  ρ g  k  l [/formula]
where ρ g is the genetic correlation between mRNA levels and SBP or DBP, k is an overall constant, and l is a gene-specific constant. There were 245 genes selected to influence simulated SBP and/or DBP. In the family data set, these 245 genes contained 1458 functional variants whose effect sizes ranged from < 0.001 to 2.78 % of the total phenotypic variance. The gene with the largest effect, MAP4 on chromosome 3, accounted for 7.79 % of phenotypic variance in simulated SBP and 6.48 % in simulated DBP when effects of all 'functional' variants within and flanking the gene were combined. A list of the functional variants with the largest effect sizes in the family data set is available in the GAW18 data description [bib_ref] Data for Genetic Analysis Workshop 18: human whole genome sequence, blood pressure,..., Blangero [/bib_ref].
The variants designated as 'functional' for phenotype simulations in the GAW19 exome data set differ slightly from those in the family data set, due to non-coding variants present in the WGS but not covered by the exome sequencing and due to new coding variants present in the larger set of unrelated individuals in the exome data set that had not been represented in the smaller family data set. The 245 'functional' genes used in simulating phenotypes for the family data set were screened for new non-synonymous coding variants present in the unrelated exome data set and new variants were assigned effect sizes by the same formula, as a function of their PolyPhen-2 scores. This resulted in a total of 1730 functional variants in the exome data set. The 20 variants with the largest effect sizes in the exome sample are shown in [fig_ref] Table 2: Top 20 variants influencing simulated SBP and DBP in the GAW20 unrelated... [/fig_ref]. All are non-synonymous coding variants. Effect sizes were somewhat larger for the GAW19 exome simulation than for the simulations in the family data set. The MAP4 variant with the largest effect in the families, is only the fourth largest effect in the unrelated exome simulation.
In simulations in the family data set, frequency of simulated anithyptertensive medication usage was modeled on the observed data. Simulated SBP and DBP were each reduced by treatment, except in individuals carrying deleterious variants in the CYP3A43 gene, producing a genotype-by-medication interaction effect. In the exome data set, because only a subset of individuals had information available on medication usage, medication status was randomly assigned, varying across replicates, and was based on the proportion of participants on antihypertensive medications in exam 1 in the family sample.
In addition to the measured genetic effects generated from the sequence data, an aggregate, unspecified additive genetic residual was modeled based on pedigree-derived kinship estimates in the family data set and on empirical kinship estimates among all pairs of individuals estimated using the program LDAK [bib_ref] Improved heritability estimation from genome-wide SNPs, Speed [/bib_ref] in the exome data set. This residual additive genetic correlation was set to maintain the heritabilities of simulated SBP and DBP and the genetic correlations between them as observed in exam 1 of the family data set.
Pedigree-derived and empirical kinship estimates also were used to simulate a phenotype called Q1 that had a heritability of 68 % but was independent of the WGS or exome sequence variants. Q1 was simulated as a normally distributed quantitative trait and was designed primarily for testing of type I error. Only a single time point was simulated for 200 replicates of Q1 in both the family and exome data sets.
# Conclusions
The GAW19 data provide a broad range of analytical possibilities, including both genomic and transcriptomic data; both family and unrelated cohort data sets; both real and simulated phenotypes; and both longitudinal and cross sectional data sets. At the workshop, investigators used these data to address a wide variety of topics. Analytical issues addressed included methods for population- [bib_ref] Above and beyond state-of-the-art approaches to investigate sequence data: summary of methods..., Bermejo [/bib_ref] and family-based [bib_ref] Family-based approaches: design, imputation, analysis, and beyond, Wijsman [/bib_ref] association, machine learning and data mining approaches to gene localization [bib_ref] Machine learning and data mining in complex genomic data-a review on the..., König [/bib_ref] , and methods for joint analysis of mutiple phenotypes [bib_ref] Joint analysis of multiple phenotypes: summary of results and discussions from the..., Schillert [/bib_ref]. Some groups concentrated on approaches to dealing with multiple testing in these high dimensional sequence data by filtering sequence variants or placing informative priors for association analyses [bib_ref] Filtering genetic variants and placing informative priors based on putative biological function, Friedrichs [/bib_ref] , by pathway-based approaches for gene localization [bib_ref] Pathway-based analyses, Jr [/bib_ref] , or by other variant collapsing approaches [bib_ref] Progress in methods for rare variant association, Santorico [/bib_ref]. Other contributions focused on utilizing unique aspects of the GAW19 family data set, including genetic analyses of longitudinal data [bib_ref] Longitudinal analytical approaches to genetic data, Chiu [/bib_ref] , and analysis of gene expression data [bib_ref] Gene expression in large pedigrees: analytic approaches, Cantor [/bib_ref]. The variety of topics addressed in these GAW19 contributions illustrate the utility and versatility of the GAW19 data. As many genetic studies of complex human phenotypes are currently focusing on exome and whole genome sequence, and their integration with gene expression data, we anticipate that the GAW19 data will continue to provide a rich resource for statistical genetic methods development, comparison, and testing for years to come.
[table] Table 1: Variant annotation, by minor allele frequency category, for variants present in at least 5 copies in the unrelated exome sample and in the family WGS sample [/table]
[table] Table 2: Top 20 variants influencing simulated SBP and DBP in the GAW20 unrelated exome sample, in decreasing order of variance explained Chromosome Position (bp) Gene Frequency of non-reference allele Beta a DBP Beta a SBP DBP variance explained (%) SBP variance explained (%) Beta = change in mean phenotype value per non-reference allele carried [/table]
|
10.1089/trgh.2018.0062
|
CCBY
|
6657294
|
31346546
|
s2orc_pubmed_articles
|
Recruitment and Enrollment of a National Sample of Transgender Youth via Social Media: Experiences from Project Moxie
This report compares social media strategies for enrolling transgender youth (TY) into online HIV prevention research. Over 12 months, 202 TY enrolled in Project Moxie, a randomized trial of an at-home HIV testing intervention. Free Craigslist advertisements showed promising success in enrolling TY, especially those of color. Paid Facebook advertising was successful in reaching a large sample of TY, as was participant referral. This supports previous literature suggesting peer referral as an effective strategy for reaching TY. High levels of attempted fraud were detected and mitigated. Findings demonstrate that recruitment and enrollment of a diverse TY sample is possible online.
# Introduction
Transgender youth (TY; 15-24 years) face a multitude of health disparities that increase their vulnerability to poor health outcomes. [bib_ref] Transgender youth: invisible and vulnerable, Grossman [/bib_ref] Although sexual and gender minority (SGM) populations are often regarded as ''hidden,'' online recruitment has been lauded as a tool for successfully engaging hard-to-reach populations in sexual health research. [bib_ref] Recruitment and retention of an online sample for an HIV prevention intervention..., Bull [/bib_ref] [bib_ref] Getting wired: exploiting the internet for the collection of valid sexuality data, Mustanski [/bib_ref] [bib_ref] The intersection of youth, technology, and new media with sexual health: moving..., Allison [/bib_ref] [bib_ref] Reaching adolescent gay, bisexual, and queer men online: development and refinement of..., Prescott [/bib_ref] Online interventions have gained popularity in HIV prevention research, specifically with regard to SGM youth, yet their success depends on the ability to recruit and enroll diverse samples while detecting fraudulent activity. [bib_ref] The intersection of youth, technology, and new media with sexual health: moving..., Allison [/bib_ref] Recent reports show 95% of teens aged 13-17 years and 88% of adults aged 18-29 years use some form of social media, indicating a wide potential reach of online recruitment.Previous findings suggest that SGM youth spend more time online than do their non-SGM counterparts.Furthermore, researchers have observed a significant role of social media in identity formation and community building for SGM individuals, and specifically TY. [bib_ref] You can form a part of yourself online: the influence of new..., Craig [/bib_ref] [bib_ref] Transgender representation in offline and online media: LGBTQ youth perspectives, Mcinroy [/bib_ref] [bib_ref] Transgender youth of color and resilience: negotiating oppression and finding support, Singh [/bib_ref] [bib_ref] Queer identity online: informal learning and teaching experiences of LGBTQ individuals on..., Fox [/bib_ref] [bib_ref] Trans youth and social media: moving between counterpublics and the wider web, Jenzen [/bib_ref] Collectively, this suggests that online recruitment through social media may be a successful method to reach TY given its appeal and cultural relevance. Despite the widespread use of social media recruitment, little is known about the feasibility and success of recruiting TY across different platforms. This report compares enrollment rates and sample characteristics between online strategies for recruiting TY in an HIV prevention study (Project Moxie), with the aim of highlighting relative merits for several popular social media platforms.
# Methods
Project Moxie is a randomized trial that aims to test the feasibility of pairing a HIPAA secure, online video counseling intervention with at-home HIV testing for TY. A comprehensive study protocol was approved by the University of Michigan Institutional Review Board (IRB; HUM0012 3412) and is available elsewhere. [bib_ref] Providing home-based HIV testing and counseling for transgender youth (Project Moxie): protocol..., Stephenson [/bib_ref] After recruitment, participants were asked to take a baseline survey before randomization into two study arms: (1) the control arm, in which participants were sent an OraQuick at-home rapid HIV test and asked to report their results in an online study portal, or (2) the intervention arm, in which participants were sent an OraQuick at-home rapid HIV test and asked to participate in a remote motivational interviewing/counseling, testing, and referral (MI/CTR) session. MI/CTR sessions were completed using VSee, an online HIPAA secure video calling service. The study protocol was approved by the University of Michigan Institutional Review Board (IRB; HUM00123412) and is available elsewhere. [bib_ref] Providing home-based HIV testing and counseling for transgender youth (Project Moxie): protocol..., Stephenson [/bib_ref] Recruitment of TY took place from June 2017 to June 2018, using advertisements and postings placed on the following social media websites: Facebook, Instagram, Twitter, Tumblr, and Craigslist. Recruitment advertisements featured photos representing a spectrum of transgender and gender variant persons, and directing interested individuals to the Project Moxie website to earn up to $150 for participating.
The Project Moxie landing page provided basic study information, including a short description of activities, and an informed consent to screen for eligibility. Those who provided consent were directed to an eligibility screener. Eligible individuals then underwent a comprehensive online consent before participation. A waiver of need for parental consent to screen and enroll those under the age of 18 years was approved by the IRB.
Eligibility for the study included the following: (1) selfidentification as noncisgender, indicated by a current gender identity differing from sex assigned at birth; (2) aged 15-24 years; (3) negative or unknown HIV status; (4) current U.S. residency; (5) willingness to receive an at-home HIV test; and (6) access to a computer, smartphone, or tablet that supports VSee video-calling software to implement the MI/CTR intervention. Pronouns and recruitment route were also asked at this time.
Eligible participants were instructed to create an account. All identifying information was stored on a password-protected server accessible only to IRBapproved study staff. Duplicate accounts and accounts that could be linked across inconsistent IP address, name, physical address, email, or phone number were flagged for verification. The information from these accounts was verified by staff using Spokeo, an online information aggregator; if the search was inconclusive, participants were contacted and asked to confirm their account information. After verification, participants were given the opportunity to refer friends into the study.
If multiple accounts were detected for a verified individual, they were notified by staff that their longest standing account would be retained while any others would be deleted. In the case of fraudulent activitythe persistent creation of numerous accounts with inconsistent information-the associated IP address was restricted from accessing the study server.
The current analysis describes the process of recruitment, rates of enrollment, demographic differences of participants, and associated costs across the social media and participant referral platforms. Proportion tests and Fisher's exact tests were conducted using Stata/SE 15.1 to determine statistically significant (p < 0.05) differences in enrollment rates and demographic data across recruitment platforms.
# Results
Project Moxie social media advertisements generated 1,113,955 impressions-the total number of times the advertisements were displayed to any user-resulting in 33,182 (3.0%) clicks. Electronic consent to screen for eligibility was obtained from a sample of 2707 individuals, 1365 (50.4%) of whom started the eligibility screener. Of these individuals, 698 (51.1%) met the study eligibility criteria. Reasons for ineligibility included self-identification as cisgender (275, 20.1%), age (303, 22.2%), HIV status (12, 0.9%), unwillingness to receive an at-home rapid HIV test (61, 4.5%), and lack of access to a computer (7, 0.5%).
Of the 698 eligible individuals, 480 (68.8%) consented for participation and created an account. Information from 216 (45%) accounts was verified, whereas 264 (55%) accounts were determined duplicate or fraudulent and dropped from the study. Of the 216 verified individuals who created accounts, 202 (93.5%) took the baseline survey through the study website. For the purpose of this analysis, participants are categorized as enrolled once they have provided consent to screen for eligibility, met eligibility criteria, provided consent to participate, and taken the baseline survey. The largest volume of eligible (219, 31.4%) and enrolled (69, 34.2%) individuals came from Facebook, followed by participant referral (170, 24.3%; 57, 28.2%; [fig_ref] Table 1: Eligibility and Enrollment by PlatformBold indicates statistically significant values [/fig_ref].
The overall sample had an enrollment rate of 28.9%, with a significantly greater proportion of eligible individuals enrolling when recruited through Craigslist (49.3%; p = 0.000) and a significantly lesser proportion enrolling when recruited through Tumblr (7.3%; p = 0.000).
The Moxie participant sample was composed of mostly nonbinary individuals (83, 41.1%) and transgender men Fisher's exact tests revealed statistically significant variance of race ( p = 0.000) and region ( p = 0.036) by platform [fig_ref] Table 2: Demographics of Enrolled Participants [/fig_ref]. Although some trends between platforms may be observed among other variables, these did not achieve significance. Participants enrolled from Facebook (69, 34.2%) share a similar demographic makeup to that of the overall sample.
[formula] % (n) %( n) %( n) %( n) %( n) %( n) p [/formula]
Compared with the overall sample, a significantly higher percentage of those enrolled from Instagram (22, 10.9%) identified as white (non-Hispanic; 21, 95.5%; p = 0.002), and were from the northeast (8, 36.4%; p = 0.003). Conversely, a significantly higher percentage of those enrolled from Craigslist (35, 17.3%) identified as non-white (including Hispanic; 27, 77.1%; p = 0.000), comprising 40.3% of the nonwhite TY in the overall sample.
Few individuals were enrolled from other routes, including Twitter (5, 2.5%), Tumblr (9, 4.4%), or recruitment through local organizations (5, 2.5%). These small subsamples have highly varied demographic makeups, and skew heavily toward midwest residence (10, 52.6%; p = 0.011).
The total direct cost for all paid advertising was $2493.75 across the 12-month recruitment period, with $2482.73 spent on Facebook and $11.02 spent on Instagram. All other advertisements were placed using free, registered accounts across the various platforms. Cost can be broken down as follows: $0.08 per click, $0.92 per consent, $1.83 per eligibility screener, $3.57 per eligible participant, and $12.35 per baseline survey. The largest cost differential is observed between eligible participants and baseline surveys, which may be attributed to insufficient motivation to return to the study website for account setup, high levels of fraudulent account creation, or some combination of these factors.
# Discussion
Facebook generated the largest portion of the Project Moxie sample, but was also the most costly. Of the social media websites used, Facebook is the most popular among U.S. adults aged 18-24 years with 80% usage compared with 71% on Instagram and <50% on all others.For those aged 13-17 years in the United States, Facebook is less popular (51%) but second only to Instagram (72%).The advertising interface on Facebook and Instagram allowed TY to be targeted based on age and interests independent of subscription (e.g., friend request, follow), something unavailable through unpaid advertising routes. Although Facebook advertising took place throughout the entirety of recruitment, Instagram advertising only began approximately half way through this period. Although fewer individuals were recruited from Instagram than from Facebook, enrollment was similarly efficient.
Peer referral has been demonstrated as an effective recruitment strategy for hard-to-reach populations, including TY, the success of which relies on participating community members' connection to other interested individuals. [bib_ref] Interventions to improve cultural competency in healthcare: a systematic review of reviews, Truong [/bib_ref] [bib_ref] Differential HIV risk for racial/ ethnic minority trans*female youths and socioeconomic disparities..., Wilson [/bib_ref] In addition, previous data have shown that web-based referral to be successful in the recruitment of a diverse target population online. [bib_ref] Innovative recruitment using online networks: lessons learned from an online study of..., Bauermeister [/bib_ref] Data from Project Moxie support existing findings on the utility of participant referral, and demonstrate its viability in the efficient recruitment of TY online.
Placing free Craigslist advertisements was also a viable strategy for recruiting TY, especially those of color, with the highest enrollment rate in Project Moxie. Craigslist has been identified as an online space utilized by racially diverse SGM communities, with previous literature demonstrating the ability to recruit SGM samples of significantly lower Caucasian-identifying proportion through Craigslist. [bib_ref] Displacing the dominant ''down low'' discourse: deviance, same-sex desire, and Craigslist.org, Robinson [/bib_ref] [bib_ref] Recruiting rural and urban LGBT populations online: differences in participant characteristics between..., Warren [/bib_ref] Although Craigslist provided a valuable opportunity to reach TY of color, it has limitations for use in research recruitment. Routine removal of research advertisements in some areas forced study staff to spend considerable time (4-6 hours per week) reposting advertisements on Craigslist.
Few individuals were enrolled using free post-based advertisements on either Twitter or Tumblr. Although Tumblr contributed the third highest volume of eligible individuals, it had an enrollment rate >20% lower than any other platform. Whether this is due to high levels of disinterest in participating or fraudulence is unknown. The ability to build and maintain a following, and/or network of reliable community members to share (e.g., retweet, reblog) advertisement posts to their own audience, is central to generating substantial exposure on these platforms without utilizing platform-provided advertising services. Future research should explore the establishment of effective community engagement protocols or use of paid advertisements when recruiting TY on these platforms.
# Limitations
Fraud is clearly a risk when recruiting online, making mechanisms of detection and mitigation essential in ensuring the validity of the enrolled sample. Although more than half of all Moxie accounts created were deemed fraudulent or duplicate, the implementation of protocol described in this report's methods allowed for verification of legitimate accounts and removal of fraudulent accounts. We did not collect data on the recruitment route for fraudulent participants, and thus it is not possible to determine whether differential fraudulence occurred between social media platforms.
Implementation of fraud detection mechanisms such as browser cookies or captcha during the registration process, in addition to retroactive fraud detection and mitigation (as used in Project Moxie), has the potential to save valuable resources and protect the integrity of online data. [bib_ref] Detecting, preventing, and responding to ''fraudsters'' in internet research: ethics and tradeoffs, Teitcher [/bib_ref] [bib_ref] Lessons learned from an online study with dual-smoker couples, Choi [/bib_ref] Conclusion Evidence from Project Moxie serves to add TY to the list of hard-to-reach populations who can be successfully recruited online. [bib_ref] Recruitment and retention of an online sample for an HIV prevention intervention..., Bull [/bib_ref] [bib_ref] Getting wired: exploiting the internet for the collection of valid sexuality data, Mustanski [/bib_ref] [bib_ref] The intersection of youth, technology, and new media with sexual health: moving..., Allison [/bib_ref] [bib_ref] Reaching adolescent gay, bisexual, and queer men online: development and refinement of..., Prescott [/bib_ref] However, when recruiting online, consideration for both proactive and retroactive fraud detection and mitigation mechanisms is crucial to most effectively preserve the integrity of data.
[table] Table 1: Eligibility and Enrollment by PlatformBold indicates statistically significant values (p < 0.05).Percentage of those screened eligible who were successfully enrolled in the study.Local sources of recruitment include on-site by study staff, newsletters/listservs, and palm card. [/table]
[table] Table 2: Demographics of Enrolled Participants [/table]
|
10.3390/foods11233878
|
CCBY
|
9737139
|
36496686
|
s2orc_pubmed_articles
|
Lipid and Protein Oxidation of Brown Rice and Selenium-Rich Brown Rice during Storage
# Introduction
Selenium is an essential trace element for the human body and is closely related to human health. It has significant effects on antioxidation [bib_ref] Purification, identification, and in vitro antioxidant activities of selenium-containing proteins from selenium-enriched..., Liu [/bib_ref] , the prevention of cardiovascular and cerebrovascular diseases, the improvement of body immunity [bib_ref] Toward improved human health: Efficacy of dietary selenium on immunity at the..., Xia [/bib_ref] and antitumor and antiaging properties [bib_ref] Beneficial and paradoxical roles of selenium at nutritional levels of intake in..., Zhang [/bib_ref]. The human body mainly supplements selenium through diet. Two thirds of the world's soil is deficient in selenium, resulting in insufficient selenium intake [bib_ref] Selenium as a bioactive micronutrient in the human diet and its cancer..., Radomska [/bib_ref]. Selenium is mainly involved in the regulation of physiological functions in the form of proteins in the human body. Rotruck et al. [bib_ref] Biochemical role as a component of glutathione peroxidase, Rotruck [/bib_ref] proposed in 1973 that selenium participates in the formation of glutathione peroxidase (GPx). With further research, selenium has been proved to replace sulfur to form selenocysteine (SeCys), which exists at the GPx active center. Moreover, it is the cofactor of GPx, thioredoxin reductase and other biological enzymes [bib_ref] Current knowledge in species-related bioavailability of selenium in food, Thiry [/bib_ref].
Among the three major grains in the world, rice is the main grain supply in China. About 50% of people in the world regard rice as the main source of energy, and it is widely used in food industry production [bib_ref] Importance for global nutrition, Fukagawa [/bib_ref]. China's rice production ranks first in the world, accounting for about 30% of the world production.
The organic selenium in Se-rich rice is mainly present in the form of selenomethionine (SeMet), selenocysteine (SeCys 2 ) and methylselenocysteine (MetSeCys) [bib_ref] Purification, identification, and in vitro antioxidant activities of selenium-containing proteins from selenium-enriched..., Liu [/bib_ref] [bib_ref] Selenium accumulation in protein fractions during germination of Se-enriched brown rice and..., Liu [/bib_ref]. The characteristics of rice production and consumption indicate that rice needs to be stored for a certain period of time. Owing to the long-term storage process, unstable factors, such as temperature and humidity, may affect the quality of rice after storage. Lipids and proteins in rice undergo complex physical and chemical changes during storage. Studies have shown that selenium-enriched rice has stronger antioxidant activity than ordinary rice [bib_ref] Comparative proteomics analysis reveals the effect of germination and selenium enrichment on..., Li [/bib_ref] , which may be related to selenium-substituted proteins participating in different metabolic pathways of proteins.
The lipid content of brown rice accounts for 1-2%, but during storage, unsaturated fatty acids are easily oxidized and decomposed, producing peroxidized free radicals and other secondary oxidation products. Free radicals, as a potential initiating factor, initiate chain polymerization in the protein reaction system through hydrogen extraction [bib_ref] Light exposure accelerates oxidative protein polymerization in beef stored in high oxygen..., Zainudin [/bib_ref] , which promotes complex changes in proteins and aging and deterioration of rice while affecting the edible quality and nutritional value of rice [bib_ref] Effects of rice aging on its main nutrients and quality characters, Peng [/bib_ref]. The protein content of brown rice accounts for 8-12%. Similar to the process of lipid oxidation, protein/Se-protein oxidation is affected by environmental factors and reactive oxygen species in grains (such as O 2 − , OH, H 2 O 2 , alkoxy radicals and lipid free radicals) during storage, resulting in intramolecular or intermolecular cross-linking of proteins and affecting the structure, nutrition and function of proteins [bib_ref] Protein oxidation: Basic principles and implications for meat quality, Zhang [/bib_ref]. Methionine (Met), tyrosine, histidine, phenylalanine, tryptophan, lysine and arginine are sensitive to the oxidation of proteins, and they are important targets for lipid-induced oxidation [bib_ref] Effect of oxidation on the gel properties of porcine myofibrillar proteins and..., Shen [/bib_ref] [bib_ref] Proteomic responses to oxidative damage in meat from ducks exposed to heat..., He [/bib_ref]. Some studies have shown that the solubility, foam stability, gelation and thermal stability of proteins increase after being subjected to appropriate oxidative attack, which may be related to the proper development of protein structure, indicating that appropriate oxidation can improve the functional properties of proteins in some aspects, which has positive significance for improving food quality. However, excessive oxidation will seriously break the structure of protein, leading to the decline of the protein's functional properties, flavor deterioration and nutritional loss, and even produce toxic and harmful substances. The current research on the interaction between lipid oxidation products and proteins has involved the formation and transfer of lipid free radicals, the aggregation of protein molecules induced by lipid oxidation products and the change of functional properties [bib_ref] High-pressure homogenization combined with sulfhydryl blockage by hydrogen peroxide enhance the thermal..., Chen [/bib_ref] [bib_ref] Effects of oxidation by malondialdehyde on the structure and function of rice..., Zhou [/bib_ref].
Proteomics is a research technique that finds certain specific protein molecules between samples by studying protein characteristics on a large scale, including protein expression, translation modification and protein-protein interaction [bib_ref] Protein structural properties and proteomic analysis of rice during storage at different..., Zhao [/bib_ref]. Proteomics has been widely used in many fields to reveal the structure, abundance and interactions between proteins. Wu et al. [bib_ref] Ubiquitin-specific protease 3 promotes cell migration and invasion by interacting with and..., Wu [/bib_ref] analyzed the mechanism of gastric cancer metastasis promoted by deubiquitinating enzyme ubiquitin-specificprotease3 via the isobaric tags for relative and absolute quantitation (iTRAQ) technique. Yang et al. [bib_ref] Pattern of protein expression in developing wheat grains identified through proteomic analysis...., Yang [/bib_ref] identified differentially expressed proteins (DEPs) in the early grain development of large-grain and small-grain wheat cultivars by using proteomic methods and elucidated the molecular and cellular processes leading to these differences.
At present, there are few reports on the changes in the protein oxidation mechanism and the functional properties of Se-rich rice during storage. The purpose of this study is to analyze the changes in lipids and proteins in Se-rich brown rice and ordinary brown rice after storage under the same storage conditions for a certain period of time. The mechanism of quality deterioration of the two kinds of rice during storage is also clarified from the level of proteomics.
# Materials and methods
## Sample
Ordinary brown rice (BR) and Se-rich brown rice (Se-BR) were obtained from Yuanyang Basu Rice Industry Co., LTD (Xinxiang, China) in 2021. The rice variety is Huangjinqing 22438, which is commonly cultivated in China. The moisture content of BR and Se-BR was 12.99% and 13.20%, respectively.
## Storage conditions
Two kinds of samples of brown rice were packed in clean cloth bags and placed in a constant temperature and humidity incubator at 35 - C and 75% RH for 180 days. Samples were taken every 30 days for further analysis.
## Preparation of brown rice oil
The brown rice powder was extracted in hexane for 12-16 h at room temperature. The filtered supernatant was evaporated in a water bath to obtain brown rice oil, which was stored at 4 - C.
## Preparation of brown rice protein
Brown rice protein was extracted according to method reported by Liu et al.. Ordinary brown rice protein (BRP) and selenium-rich brown rice protein (Se-BRP) were extracted with deionized water at pH 9.0. The mixture was rotated at 45 - C for 3 h and then centrifuged at 4000× g for 20 min. The supernatant was adjusted to pH 4.5 and then centrifuged at 5000× g for 10 min. Protein precipitation was obtained after the supernatant was removed. After freeze-drying, the protein was stored at −20 - C.
## Determination of peroxide value (pv)
The PV was determined according to method reported by Mehta et al. [bib_ref] Comparison of five analytical methods for the determination of peroxide value in..., Mehta [/bib_ref]. Briefly, 0.2 g (accurate to 0.01) of sample was taken into a 15 mL centrifuge tube. Then, 9.8 mL of chloroform-methanol (7/3, v/v) solution was added, and the mixture was oscillated for 5 s. Next, 100 µL of xylenol orange solution (10 mmol/L) and 50 µL of FeCl 2 solution (36 mmol/L) were added, and the absorbance at 560 nm was measured after a 5-min reaction.
## Determination of fatty acid value (fav)
The FAV was determined according to method reported by Shu et al. [bib_ref] Selected quality attributes of paddy rice as affected by storage temperature history, Shu [/bib_ref]. Herein, 10 g of rice powder was extracted with benzene for 30 min. Subsequently, 0.04% phenolphthalein ethanol was added into 25 mL of filtrate and was titrated with 0.01 mol/L potassium hydroxide.
## Determination of carbonyl value (cv)
The CV was determined according to method reported by Hasan et al. [bib_ref] Principal component analysis of lipid and protein oxidation products and their impact..., Hasan [/bib_ref]. For this, 6 mL of 2,4-dinitrophenylhydrazine (DNPH, 10 mmol/L) solution was used to react with 2.1 mL of protein solution (3.5 g/mL) to form a brownish-red precipitate. Next, 2.7 mL of trichloroacetic acid (40%) was added to react for 20 min. After centrifugation, 9 mL of ethyl acetate and ethanol solution (1:1, v/v) were used to wash the precipitate 3 times. The precipitate was dissolved with 6 mL of guanidine hydrochloride (6 mol/L). The absorbance at 367 nm of the solution was determined and the extinction coefficient was 22,000 M −1 cm −1 .
## Fourier-transformed infrared (ftir) spectroscopy
Samples were mixed with potassium bromide in a ratio of 1:100 and scanned 32 times in the wavenumber range of 4000-400 cm −1 by the Nicolet6700 Fourier-Transform Infrared Spectrometer (Waltham, MA, USA). PeakFit (version 4.12, Palo Alto, CA, USA) software was used to analyze the infrared spectra of amide I bands (1600-1700 cm −1 ) of protein. The relative content of the secondary structure was calculated [bib_ref] Protein structural properties and proteomic analysis of rice during storage at different..., Zhao [/bib_ref].
2.9. Proteomic Analysis 2.9.1. Protein Extraction Brown rice protein was extracted by the method of Wang et al. [bib_ref] iTRAQ-based proteomic reveals cell cycle and translation regulation involving in peanut buds..., Wang [/bib_ref]. The treated protein clumps were stored at −80 - C for further use.
## Protein quantification
Lysate L3 was added to dissolve the protein, and ultrasonic treatment-assisted solubilization was performed after freezing centrifugation. The BSA was taken to make a standard curve. The protein solution (200 µg) was kept in a centrifuge tube, and 4 µL of TCEP reducing reagent (60 - C for 1 h) and 2 µL of MMTS cysteine (Cys)-blocking reagent (room-temperature reaction for 30 min) were added subsequently. The reduced alkylated protein solution was added to a ultrafiltration tube (10 kDa) and centrifuged (12,000× g at 4 - C for 20 min), and the bottom solution was discarded. Likewise, 100 µL of 8 mol/L urea (pH 8.5) was added and centrifuged, and the bottom solution was discarded; this process was repeated 2 times. Afterward, 100 µL of 0.25 mol/L TEAB (pH 8.5) was added and centrifuged, and the bottom solution was discarded; this process was repeated 3 times. The collection tube was then replaced. TEAB (50 µL of 0.5 mol/L) was added to an ultrafiltration tube, followed by trypsin (1:50 trypsin/protein, 37 - C reaction overnight). The next day, trypsin (1:100 trypsin/protein, 37 - C for 4 h), was added and centrifuged (12,000× g for 20 min). Subsequently, 50 µL of 0.5 mol/L TEAB was added to the ultrafiltration tube and centrifuged (12,000× g at 4 - C for 20 min), and this step was merged with the previous step. Finally, a total of 100 µL of the enzymatic digested sample at the bottom of the tube was collected.
## Labeling and analysis of itraq reagent
The samples were labeled with a 6-standard iTRAQ kit and preserved for use after vacuum freeze-centrifugal drying.
Liquid chromatography-mass spectrometry (LC-MS) was performed according to Zhou et al.. The mass spectrum parameters were as follows: (1) primary mass spectrometry: resolution = 70,000; AGC target = 3,000,000; maximum IT = 100 ms; scan range = 350 to 1800 m/z; (2) secondary mass spectrometry: resolution = 17,500; AGC target = 50,000; maximum IT = 120 ms; topN = 20; collision energy = 30.
# Statistical analysis
Data were calculated as mean ± standard deviation with 3 replicates. Significant differences were determined by one-way ANOVA followed by Duncan's test (p < 0.05 was considered as statistically significant). IBM SPSS Statistics (version 26.0, Chicago, IL, USA) and GraphPad Prism (version 8.0, San Diego, CA, USA) were used to calculate and plot.
# Results and discussion
## Effect of storage on peroxide value
The PV indicates the content of lipid hydroperoxide in brown rice oil under the condition of self-oxidation and photo-oxidation, which reflects the oxidation degree of the sample [bib_ref] Effect of heating on compositional characteristics and oxidative stability of crude and..., Ali [/bib_ref]. [fig_ref] Figure 1: Peroxide value of ordinary brown rice [/fig_ref] shows the change in PV of Se-BR and BR during storage. The PV of the two kinds of brown rice first increased and then decreased. Specially, the PV of BR increased to the highest point of 1.232 mEq/kg from 0 day to 60 days, and then began to decline, indicating that the degradation rate of hydroperoxide in BR was higher than the formation rate after 60 days. As an intermediate product of lipid oxidation, hydroperoxide can be further degraded into secondary metabolites such as aldehydes and ketones [bib_ref] Inhibition of lipid and aroma deterioration in rice bran by infrared heating, Yan [/bib_ref]. The PV of Se-BR slowly increased to 1.115 mEq/kg from 0 day to 90 days and then tended to be flat. The upward and downward trend of the PV of BR was steeper than that of the PV of Se-BR. The increase in the PV of Se-BR was significantly lagging behind, which, to some extent, indicates that Se-BR has higher oxidation resistance stability than BR. Hence, selenium can effectively reduce the lipid oxidation rate during the storage of Se-BR. The PV of Se-BR slowly increased to 1.115 mEq/kg from 0 day to 90 days and then tended to be flat. The upward and downward trend of the PV of BR was steeper than that of the PV of Se-BR. The increase in the PV of Se-BR was significantly lagging behind, which, to some extent, indicates that Se-BR has higher oxidation resistance stability than BR. Hence, selenium can effectively reduce the lipid oxidation rate during the storage of Se-BR.
## Effect of storage on fatty acid value
The FAV is an indicator of the content of free fatty acids in oil, which reflects the quality change of brown rice during storage. As shown in [fig_ref] Figure 2: Fatty acid value of BR and Se-BR [/fig_ref] , the changes in FAV of the two kinds of brown rice were positively correlated with the storage time, indicating that the production rate of fatty acids was greater than that of the decomposition rate. The FAV of BR decreased from 150 days to 180 days, which may be related to the activity of enzymes that produce fatty acids [bib_ref] Inhibition of lipid and aroma deterioration in rice bran by infrared heating, Yan [/bib_ref]. Similar results were found in the study of Li et al. [bib_ref] Effect of selenium enrichment on the quality of germinated brown rice during..., Li [/bib_ref] , demonstrating that selenium exerts a positive effect on the preservation of brown rice.
## Effect of storage on carbonyl value
Under the oxidative stress of free radicals, reactive oxygen species, nitrogen and other substances, the protein structure is easily changed, including amino acid side chain modification, polypeptide chain cross-linking and disulfide bond formation. Carbonyl derivatives are mainly produced through direct oxidation of amino acid side chains, cleavage of peptide bone chains and binding of other nonprotein carbonyl groups. Given the various mechanisms involved, carbonyl derivatives are widely considered one of the landmark products for describing protein oxidative damage [bib_ref] Protein oxidation and peroxidation, Davies [/bib_ref]. [fig_ref] Figure 3: Carbonyl value of BR and Se-BR [/fig_ref] shows the
## Effect of storage on fatty acid value
The FAV is an indicator of the content of free fatty acids in oil, which reflects the quality change of brown rice during storage. As shown in [fig_ref] Figure 2: Fatty acid value of BR and Se-BR [/fig_ref] , the changes in FAV of the two kinds of brown rice were positively correlated with the storage time, indicating that the production rate of fatty acids was greater than that of the decomposition rate. The FAV of BR decreased from 150 days to 180 days, which may be related to the activity of enzymes that produce fatty acids [bib_ref] Inhibition of lipid and aroma deterioration in rice bran by infrared heating, Yan [/bib_ref]. Similar results were found in the study of Li et al. [bib_ref] Effect of selenium enrichment on the quality of germinated brown rice during..., Li [/bib_ref] , demonstrating that selenium exerts a positive effect on the preservation of brown rice.
The PV of Se-BR slowly increased to 1.115 mEq/kg from 0 day to 90 days and then tended to be flat. The upward and downward trend of the PV of BR was steeper than that of the PV of Se-BR. The increase in the PV of Se-BR was significantly lagging behind, which, to some extent, indicates that Se-BR has higher oxidation resistance stability than BR. Hence, selenium can effectively reduce the lipid oxidation rate during the storage of Se-BR.
## Effect of storage on fatty acid value
The FAV is an indicator of the content of free fatty acids in oil, which reflects the quality change of brown rice during storage. As shown in [fig_ref] Figure 2: Fatty acid value of BR and Se-BR [/fig_ref] , the changes in FAV of the two kinds of brown rice were positively correlated with the storage time, indicating that the production rate of fatty acids was greater than that of the decomposition rate. The FAV of BR decreased from 150 days to 180 days, which may be related to the activity of enzymes that produce fatty acids [bib_ref] Inhibition of lipid and aroma deterioration in rice bran by infrared heating, Yan [/bib_ref]. Similar results were found in the study of Li et al. [bib_ref] Effect of selenium enrichment on the quality of germinated brown rice during..., Li [/bib_ref] , demonstrating that selenium exerts a positive effect on the preservation of brown rice.
## Effect of storage on carbonyl value
Under the oxidative stress of free radicals, reactive oxygen species, nitrogen and other substances, the protein structure is easily changed, including amino acid side chain modification, polypeptide chain cross-linking and disulfide bond formation. Carbonyl derivatives are mainly produced through direct oxidation of amino acid side chains, cleavage of peptide bone chains and binding of other nonprotein carbonyl groups. Given the various mechanisms involved, carbonyl derivatives are widely considered one of the landmark products for describing protein oxidative damage [bib_ref] Protein oxidation and peroxidation, Davies [/bib_ref]. [fig_ref] Figure 3: Carbonyl value of BR and Se-BR [/fig_ref] shows the
## Effect of storage on carbonyl value
Under the oxidative stress of free radicals, reactive oxygen species, nitrogen and other substances, the protein structure is easily changed, including amino acid side chain modification, polypeptide chain cross-linking and disulfide bond formation. Carbonyl derivatives are mainly produced through direct oxidation of amino acid side chains, cleavage of peptide bone chains and binding of other nonprotein carbonyl groups. Given the various mechanisms involved, carbonyl derivatives are widely considered one of the landmark products for describing protein oxidative damage [bib_ref] Protein oxidation and peroxidation, Davies [/bib_ref]. [fig_ref] Figure 3: Carbonyl value of BR and Se-BR [/fig_ref] shows the changes in CV of BR and Se-BR during storage. The CV of BR and Se-BR increased from 2.647 and 2.638 mEq/kg at 0 day to 8.571 and 8.0263 mEq/kg at 90 days, respectively, followed by a flattening, downward trend. The rate of increase in the CV of the two proteins is similar, which is the same as the results of Li et al. [bib_ref] Comparative proteomics analysis reveals the effect of germination and selenium enrichment on..., Li [/bib_ref]. This finding indicates that the proteins in the two brown rice varieties were oxidatively damaged during storage. changes in CV of BR and Se-BR during storage. The CV of BR and Se-BR increased from 2.647 and 2.638 mEq/kg at 0 day to 8.571 and 8.0263 mEq/kg at 90 days, respectively, followed by a flattening, downward trend. The rate of increase in the CV of the two proteins is similar, which is the same as the results of Li et al. [bib_ref] Comparative proteomics analysis reveals the effect of germination and selenium enrichment on..., Li [/bib_ref]. This finding indicates that the proteins in the two brown rice varieties were oxidatively damaged during storage.
## Analysis of ftir spectra
FTIR is a commonly used method in studying the structure of proteins. The secondary structure of the two species of brown rice protein is shown in [fig_ref] Table 1: Secondary structure composition of BR and Se-BR [/fig_ref] , from which the relative content of α-helix, β-sheet, β-turn and random coil changed significantly with the change in storage time. The β-sheet content of BRP first increased and then decreased, which might be due to the inversion structure of the β-turn of 180° to the extension structure of β-sheet, making the spatial conformation of the protein more stable. The increase in the random coil structure at 90 days of storage was insignificant, indicating that the αhelix also underwent a transition to a β-sheet structure, which may be related to the fact that, compared with other secondary structures, β-sheet is more sensitive to environmental changes [bib_ref] Comparative study of structural and functional characterization of bran protein concentrates from..., Singh [/bib_ref]. However, with the deepening of the deterioration of brown rice, the three relatively ordered structures were destroyed, and the content of random coil increased significantly, demonstrating that the disordered structure in the protein increased and the ordered structure decreased, which may lead to a decrease in edible quality [bib_ref] Effect of composition and secondary structure of soybean protein isolate on surface..., Qi [/bib_ref]. Different from BRP, the β-sheet structure of Se-BRP did not change significantly. The β-sheet gradually decreased during storage, whereas the content of random coil gradually increased, indicating the transformation of the secondary structure of the protein to a disordered one. The α-helix rose during 90-180 days of storage, implying that some kind of alteration in Se-BRP may promote an increase in ordered structure, thereby increasing the
## Analysis of ftir spectra
FTIR is a commonly used method in studying the structure of proteins. The secondary structure of the two species of brown rice protein is shown in [fig_ref] Table 1: Secondary structure composition of BR and Se-BR [/fig_ref] , from which the relative content of α-helix, β-sheet, β-turn and random coil changed significantly with the change in storage time. The β-sheet content of BRP first increased and then decreased, which might be due to the inversion structure of the β-turn of 180 - to the extension structure of β-sheet, making the spatial conformation of the protein more stable. The increase in the random coil structure at 90 days of storage was insignificant, indicating that the α-helix also underwent a transition to a β-sheet structure, which may be related to the fact that, compared with other secondary structures, β-sheet is more sensitive to environmental changes [bib_ref] Comparative study of structural and functional characterization of bran protein concentrates from..., Singh [/bib_ref]. However, with the deepening of the deterioration of brown rice, the three relatively ordered structures were destroyed, and the content of random coil increased significantly, demonstrating that the disordered structure in the protein increased and the ordered structure decreased, which may lead to a decrease in edible quality [bib_ref] Effect of composition and secondary structure of soybean protein isolate on surface..., Qi [/bib_ref]. Different from BRP, the β-sheet structure of Se-BRP did not change significantly. The β-sheet gradually decreased during storage, whereas the content of random coil gradually increased, indicating the transformation of the secondary structure of the protein to a disordered one. The α-helix rose during 90-180 days of storage, implying that some kind of alteration in Se-BRP may promote an increase in ordered structure, thereby increasing the thermal stability of the protein to prevent further damage to the protein caused by storage [bib_ref] Formation and stability of beta-hairpin structures in polypeptides, Blanco [/bib_ref]. This result suggests that, compared with BR, Se-BR has higher oxidative resistance stability of proteins.
# Proteomic analysis
To study the proteome changes in the two kinds of brown rice before and after storage, we identified a total of 2992 trusted proteins (confidence interval ≥ 95%). Credible DEPs were screened out (filter criteria: number of peptides ≥ 2; AVG ≥ 1.5 was selected as the upregulated protein and AVG ≤ 0.67 as the downregulated protein), and the statistical results of the DEPs were obtained, as shown in [fig_ref] Table 2: Statistics of different proteins [/fig_ref]. Groups Se-0:Se-6 and BR-0:BR-6 identified 235 and 237 total DEPs, respectively, groups BR-0:Se-0 and BR-6:Se-6 identified 113 and 213 total DEPs, respectively. This result showed that both types of brown rice had different degrees of quality change during storage as the storage time increased. Through comparing the DEPs between groups, we can study the DEPs and metabolic pathways that play an important role in the storage quality of the two brown rice varieties under different storage conditions. Se-0 refers to Se-BR that is stored for 0 day; Se-6 refers to Se-BR that is stored for 180 days; BR-0 refers to BR that is stored for 0 day; BR-6 refers to BR that is stored for 180 days.
## Gene ontology (go) analysis
GO analysis was performed of four groups of DEPs. [fig_ref] Figure 1: Peroxide value of ordinary brown rice [/fig_ref] shows a bubble chart based on the number and significance of DEPs annotated to secondary classification items.
For groups Se-0:Se-6 and BR-0:BR-6, in terms of biological process (BP), the DEPs were mainly enriched in the negative regulation of hydrolase activity, cellular glucan metabolic process, negative regulation of molecular function, negative regulation of catalytic activity, negative regulation of proteolysis and negative regulation of peptidase activity. In terms of cellular component (CC), the DEPs were mainly enriched in amyloplast, aleurone grain, organelle outer membrane, vacuole and mitochondrial outer membrane. In terms of molecular function (MF), the DEPs were mainly enriched in nutrient reservoir activity, serine-type endopeptidase inhibitor activity, immunoglobulin binding and peptidase inhibitor activity.
For groups BR-0:Se-0 and BR-6:Se-6, in terms of BP, the DEPs were mainly enriched in the negative regulation of molecular function, negative regulation of catalytic activity, negative regulation of proteolysis, regulation of proteolysis, negative regulation of cellular protein metabolism process, regulation of cellular protein metabolism process and regulation of endopeptidase activity. In terms of CC, the DEPs were mainly enriched in extracellular regions, amyloplast, aleurone grain, vacuole, preribosome, intrinsic component of the chloroplast outer membrane and integral component of the plastid membrane. In terms of MF, the DEPs were mainly enriched in immunoglobulin binding, serine-type endopeptidase inhibitor activity, peptidase inhibitor activity, nutrient reservoir activity, endopeptidase regulator activity, molecular function regulator and antioxidant activity.
## Kyoto encyclopedia of genes and genomes (kegg) pathway analysis
In the Se-0:Se-6 and BR-0:BR-6 groups, the DEPs were mainly enriched in carbon fixation in photosynthetic organisms, pyruvate metabolism, starch and sucrose metabolism, peroxisomes, glutathione (GSH) metabolism and phenylpropanoid biosynthesis. In Se-0:Se-6, the DEPs were enriched in Cys and Met metabolism and sulfur metabolism pathways, which indicates that selenium may play a role in isotope substitution of sulfur in metabolism. Moreover, enrichment of the base excision repair and the zeatin biosynthesis pathway showed that the protein expression of Se-BR was more significant in reducing gene expression errors, reducing the protein solubilization rate and delaying seed aging. In BR-0:BR-6, the DEPs were enriched in phenylalanine, tyrosine and tryptophan biosynthesis, isoquinoline alkaloid biosynthesis, tropane, piperidine and pyridine alkaloid biosynthesis and ubiquinone and other terpenoid-quinone biosynthesis. Accordingly, under adverse conditions, BR initiates a certain defense mechanism to regulate its own information expression and cell signal transduction.
In the BR-0:Se-0 and BR-6:Se-6 groups, the DEPs were mainly enriched in carbon fixation in photosynthetic organisms, phenylpropanoid biosynthesis, base excision repair, starch and sucrose metabolism and phenylalanine, tyrosine and tryptophan biosynthesis. This result showed that, in addition to the primary metabolism and secondary metabolism, there were also differences in the regulatory effects of Se-BR and BR in information expression and antistress conditions under the same external conditions.
In terms of fatty acid synthesis and degradation, Os01g0919900 is a protein involved in the negative regulation of rice defense reactions, responsible for regulating fatty acid desaturase activity [bib_ref] Suppression of the rice fatty-acid desaturase gene OsSSI2 enhances resistance to blast..., Jiang [/bib_ref]. As shown in [fig_ref] Table 3: Expression of DEPs by different metabolic pathways [/fig_ref] , the expression of this protein was upregulated in BR-0:BR-6 but showed no significant change in Se-0:Se-6, implying that BR had higher unsaturated fatty acid accumulation than Se-BR. 3-Ketoacyl-CoA thiolase-like protein plays an important role in fatty acid β-oxidative cleavage [bib_ref] Peroxisomal plant 3-ketoacyl-CoA thiolase structure and activity are regulated by a sensitive..., Pye [/bib_ref] , expressing upregulation in both BR-0:BR-6 and Se-0:Se-6. 2-Hydroxyacyl-CoA lyase (2-HPCL) is a peroxisomal enzyme that depends on TPP to degrade the C-C bond of 3-methyl-branched fatty acids or shorten 2-hydroxy long-chain fatty acids [bib_ref] Role of thiamine pyrophosphate in oligomerisation, functioning and import of peroxisomal 2-hydroxyacyl-CoA..., Fraccascia [/bib_ref]. The expression of 2-HPCL and ADH1 was upregulated in the BR-6: Se-6 group, indicating that Se-BR had higher expression of fatty acid degradation mechanism and lower lipid oxidation pressure on protein, which is consistent with the change trend of FAV measured previously. Catalase catalyzes the decomposition of hydrogen peroxide into oxygen and water and reduces free radical stress. The expression of catalase was downregulated in BR-0:Se-0 and BR-0:BR-6 but upregulated in Se-0:Se-6 and BR-6:Se-6. Superoxide dismutase (SOD)[Mn] is an important antioxidant enzyme capable of catalyzing the disproportionation of superoxide anions and hydrogen peroxide [bib_ref] Catalase and superoxide dismutase enzymes activities in flaxseed (linum usitatissimum) extract, Yucebilgic [/bib_ref]. The expression of SOD[Mn] was upregulated in Se-0:Se-6 and BR-6:Se-6, but no significant change was observed in BR-0:Se-0 and BR-0:BR-6. This result showed that Se-BR expressed more antioxidant enzyme activity than BR in a high-temperature storage environment, which is consistent with the previous experimental results; that is, Se-BR has stronger antioxidant stability. Glycolate oxidase 5 is a key enzyme in photorespiration metabolism and has been demonstrated as an alternative source for H 2 O 2 production during gene-for-gene and non-host resistance in plants [bib_ref] Association-dissociation of glycolate oxidase with aatalase in rice: A potential switch to..., Zhang [/bib_ref]. The expression of glycolate oxidase 5 was upregulated in BR-0:BR-6 but had no significant change in Se-0:Se-6, suggesting that regular brown rice has developed a source of H 2 O 2 production independent of NADPH (nicotinamide adenine dinucleotide phosphate) oxidases and may be subjected to more oxidative stress than Se-rich proteins.
Glutathione transferase (GST) is one of the most important enzymes in biotransformation, which participates in cell anti-injury, anticarcinogenesis and other important metabolic processes [bib_ref] Glutathione transferases: A structural perspective, Oakley [/bib_ref] [bib_ref] Glutathione transferases, regulators of cellular metabolism and physiology, Board [/bib_ref]. The proteomic analysis showed that both kinds of brown rice underwent varying degrees of change: Q93WM2 was upregulated in BR-0:BR-6 but downregulated in BR-6:Se-6; Q945W6 was downregulated in BR-0:Se-0 and BR-0:BR-6; Q93WY5 was downregulated in BR-0:BR-6 but upregulated in Se-0:Se-6 and BR-6:Se-6; Q0JJ25 was upregulated in Se-0:Se-6. Meanwhile, the expression of L-ascorbate peroxidase was downregulated in both BR-0:BR-6 and Se-0:Se-6, suggesting that the two kinds of brown rice had protein antioxidant regulation to reduce the oxidation caused by high-temperature storage stress [bib_ref] Role of L-ascorbate in alleviating abiotic stresses in crop plants, Venkatesh [/bib_ref].
Peroxiredoxin (Prdx) and glutathione reductase (GR) (P48642, Q8S5T1) were upregulated in Se-0:Se-6 expression. Prdx is a peroxidase that detoxifies peroxides, such as hydrogen peroxide, has broad substrate specificity and plays a key role in protecting cells from stress and maintaining homeostasis in many cellular processes. The active action of Prdx is based on conserved Cys residues and depends on intermolecular or intramolecular mercaptan dioxide as an electron donor regeneration cycle, e.g., GSH and thioredoxin [bib_ref] Role for plant peroxiredoxin in cadmium chelation, Smiri [/bib_ref]. Therefore, the upregulation expression in Se-BRP may be associated with the isotope substitution of selenium and sulfur. GSH has the functions of maintaining cell stability and preventing protein denaturation. With both oxidized and reduced forms, GSH is transformed into oxidized glutathione (GSSG) under oxidative stress and into GSH under the action of GR. Thus, GR is a key enzyme for maintaining the GSH/GSSG concentration ratio [bib_ref] A sensitive and simple impedance sensing strategy for glutathione and glutathione reductase..., Wu [/bib_ref].
GR was upregulated in Se-GBP, indicating that, after selenium enrichment, GR was more sensitive to oxidative stress and expressed a stronger antioxidant effect. GSH can be metabolized to produce Cys and converted into Met, providing sulfur-containing amino acids for protein synthesis. Phosphoserine aminotransferase (PSAT), 5-methyltetrahydropteroyltriglutamatehomocysteine methyltransferase 2 and cysteine synthase (CS) [bib_ref] Structural analysis of phosphoserine aminotransferase (Isoform 1) from Arabidopsis thalianathe enzyme involved..., Sekula [/bib_ref] were upregulated in Se-0:Se-6, indicating that the sulfur metabolism in Se-BRP was faster. This finding is consistent with the higher GSH metabolism in Se-BRP; that is, Se-BR had higher levels of expression against oxidative stress.
Proteins associated with sugar metabolism, such as starch synthase, isoamylase 2, α-1,4 glucan phosphorylase and malate dehydrogenase (MDH), were upregulated or downregulated in both brown rice proteins after 180 days of storage, suggesting that storage promoted glucose metabolism and accelerated the aging rate of brown rice [bib_ref] Protein structural properties and proteomic analysis of rice during storage at different..., Zhao [/bib_ref]. Some enzymes involved in glucose metabolism are also involved in other biological metabolic processes. MDH is a key enzyme in the tricarboxylic acid (TCA) cycle; acetyl-CoA carboxylase (ACC) is involved in regulating the synthesis rate of fatty acids; pyruvate phosphate dikinase (PPDK) is a rate-limiting enzyme of the C4 pathway; hydroxyacylglutathione hydrolase (HAGH) is involved in the production of reductive glutathione. The different degrees of expression of these enzymes in the four groups of samples indicate that the protein expression of the two brown rice varieties involves distinct metabolic pathways. [fig_ref] Figure 4: DEPs in the expression of Se-BR and BR in different metabolic pathways [/fig_ref] shows the differences in the expression of Se-BR and BR through TCA cycle, glycolysis/gluconeogenesis, GSH metabolism, starch metabolism and fatty acid metabolism after storage.
1,4 glucan phosphorylase and malate dehydrogenase (MDH), were upregulated or downregulated in both brown rice proteins after 180 days of storage, suggesting that storage promoted glucose metabolism and accelerated the aging rate of brown rice [bib_ref] Protein structural properties and proteomic analysis of rice during storage at different..., Zhao [/bib_ref]. Some enzymes involved in glucose metabolism are also involved in other biological metabolic processes. MDH is a key enzyme in the tricarboxylic acid (TCA) cycle; acetyl-CoA carboxylase (ACC) is involved in regulating the synthesis rate of fatty acids; pyruvate phosphate dikinase (PPDK) is a rate-limiting enzyme of the C4 pathway; hydroxyacylglutathione hydrolase (HAGH) is involved in the production of reductive glutathione. The different degrees of expression of these enzymes in the four groups of samples indicate that the protein expression of the two brown rice varieties involves distinct metabolic pathways. [fig_ref] Figure 4: DEPs in the expression of Se-BR and BR in different metabolic pathways [/fig_ref] shows the differences in the expression of Se-BR and BR through TCA cycle, glycolysis/gluconeogenesis, GSH metabolism, starch metabolism and fatty acid metabolism after storage.
# Conclusions
Lipid and protein oxidation occurred during the 180-day storage of Se-BR and BR. PV, FAV and CV showed that selenium-rich brown rice had higher lipid and protein oxidation resistance stability compared with ordinary brown rice; FTIR showed that storage led to the changes of protein secondary structure from ordered to disordered. By comparing the differences in protein expression before and after storage in BR and Se-BR, 235 DEPs were identified in group Se-0:Se-6, 237 DEPs were identified in group BR-0:BR-6, 113 DEPs were identified in group BR-0:Se-0 and 213 DEPs were identified in group BR-6:Se-6. The metabolic pathway of DEPs was analyzed by KEGG; it was found that storage promoted the glucose metabolism of the two kinds of brown rice and accelerated the aging rate of brown rice. The expression of catalase, SOD[Mn], GST, Prdx and GR was upregulated in Se-BRP, proving the oxidation resistance stability of Se-BR is higher than that of BR. The DEPs related to the synthesis and degradation of fatty acids show that Se-BR has higher efficiency in decomposing fatty acids and higher stability in lipid oxidation, which is consistent with the previous conclusions. This study provides insights for the storage and oxidation mechanism of ordinary brown rice and selenium-rich brown rice. On this basis, further research on the kinetics of protein oxidation induced by lipid oxidation products is helpful to control the quality and prolong the storage period of rice, which is of great significance for improving food quality and reducing resource waste.
[fig] Figure 1: Peroxide value of ordinary brown rice (BR) and selenium-rich brown rice (Se-BR). Each data point is expressed as the mean and standard error of three replicates. [/fig]
[fig] Figure 2: Fatty acid value of BR and Se-BR. Each data point is expressed as the mean and standard error of three replicates. [/fig]
[fig] Figure 3: Carbonyl value of BR and Se-BR. Each data point is expressed as the mean and standard error of three replicates. [/fig]
[fig] Figure 4: DEPs in the expression of Se-BR and BR in different metabolic pathways. Different metabolic pathways are distinguished by different colored boxes. Metabolic process substances are connected by arrows, and the direction of the arrow represents the direction of metabolism. Upregulated and downregulated proteins are marked in red and blue, respectively. [/fig]
[fig] Supplementary: Materials: The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/foods11233878/s1, Figure S1: Biological process, cellular component and molecular function of BR and Se-BR. Author Contributions: M.Z.: investigation, data curation, methodology and writing-original draft. K.L.: funding acquisition, methodology, supervision and writing-review and editing. All authors have read and agreed to the published version of the manuscript. [/fig]
[fig] Funding: This study was supported by the National Natural Science Foundation of China (32172259, U1904110), the Natural Science Foundation of Henan Province (212300410033), the Program for the Top Young Talents of Henan Associate for Science and Technology (2021) and the Innovative Funds Plan of Henan University of Technology (2021ZKCJ03). [/fig]
[table] Table 1: Secondary structure composition of BR and Se-BR. [/table]
[table] Table 2: Statistics of different proteins (DEPs) identified by different groups. [/table]
[table] Table 3: Expression of DEPs by different metabolic pathways. [/table]
|
10.1007/s13659-020-00248-y
|
CCBY
|
7367987
|
32656629
|
s2orc_pubmed_articles
|
Arthrinins E–G, Three Botryane Sesquiterpenoids from the Plant Endophytic Fungus Arthrinium sp. HS66
Arthrinins E-G (1-3), three new sesquiterpenoids possessing non-isoprenoid botryane skeleton, were isolated from the fermentation of an endophytic fungus named Arthrinium sp. HS66 which colonized in the stems of Isodon xerophilus. Their structures were determined by extensive spectroscopic methods. Furthermore, the structure of 1 was unambiguously confirmed by X-ray diffraction, while those of 2 and 3 were verified through quantum chemical calculation of NMR data and ECD spectra.Keywords Isodon xerophilus · Endophytic fungus · Arthrinium · Botryane sesquiterpenoid · Quantum chemical calculationResults and DiscussionCompound 1 was isolated as colorless oil, and it was assigned a molecular formula of C 15 H 26 O 3 according to its positive HRESIMS ion peak at m/z 277.1775 ([M + Na] + , calcd for 277.1774), which required three degrees of unsaturation. The IR spectrum showed the presence of hydroxyl (3374 cm -1 ) group.
# Introduction
Fungal metabolites have received considerable attention nowadays because their diverse chemical structures and bioactivities greatly facilitate the progress of drug discovery. Botryanes are a class of fungus-derived sesquiterpene metabolites featuring characteristic bicyclic non-isoprenoid system, and they were found to possess broad spectrum of biological activities, such as cytotoxicity, phytotoxicity, and antimicrobial property. Furthermore, fascinating by the botryane sesquiterpenoids, researchers have conducted plenty of in-depth investigations about their structure-activity relationships, synthesisand biosynthetic pathway.
Fungal endophytes have become an important source for discovering structurally novel and biologically active secondary metabolites. Over the past several years, our groups have made great efforts to study the secondary metabolites from endophytic fungi inhabiting the Isodon species. As a result, isopenicin A, a potent inhibitor of Wnt signaling, as well as several antineoplastic compounds like phomopchalasins A and Bhave been successfully obtained. In the present research, an endophytic fungus colonizing in the stems of Isodon xerophilus was discovered and identified as Arthrinium sp. HS66. Subsequent large fermentation and chemical investigation on this strain resulted in the isolation of three new botryane sesquiterpenoids named arthrinins E-G. Notably, compound 2 possesses uncommon 15-nor-botryane skeleton. Herein, details of the isolation, structure elucidation, and cytotoxicity of these compounds were reported [fig_ref] Figure 1: Chemical structures of compounds 1-3 [/fig_ref]. the existence of three singlet methyls (δ H 0.97, 1.18 and 1.38), one doublet methyl (δ H 1.10, d, J = 6.9 Hz), a pair of nonequivalent oxygenated methylene protons at δ H 4.04 (dd, J = 11.9, 2.1 Hz)/4.33 (d, J = 11.9 Hz), one hydroxylated methylene signal at δ H 3.30 (overlap), and one hydroxylated methine proton at δ H 3.63 (ddd, J = 11.9, 9.6, 3.6 Hz). The analysis of its 13 C NMR and DEPT spectra revealed 15 carbon resonances which were assigned as four methyls, four sp 3 methylenes, three sp 3 methines, two sp 3 quaternary carbons and two olefinic carbons. These data suggested that compound 1 might be a botryane sesquiterpenoid.
Specifically, a structural subunit of C-11/C-2/C-3/C-4/C-5 could be established as an isolated spin-system according to the 1 H-1 H COSY correlations of H 3 -11/H-2/ H 2 -3/H-4/H-5. Then, the subunit can be assigned to ring A through the HMBC correlations from H-5 to C-9 (δ C 145.9), from H 2 -10 to C-1 (δ C 136.0), C-2 (δ C 34.4) and C-9, from H 3 -11 to C-1. Meanwhile, the HMBC correlations from H 2 -7 to C-13 (δ C 30.9) and C-14 (δ C 26.1), from H 3 -12 to C-6 (δ C 40.0), from H 3 -13 to C-5 (δ C 61.4) and C-12 (δ C 23.9), from H 3 -14 to C-8 (δ C 47.5), from H 2 -15 to C-9 and C-14, in combination with the remaining one degree of hydrogen deficiency, successfully established the ring B as well as its connection with ring A . Hence, the planar structure of compound 1 was resolved.
The relative configuration of 1 was deduced by analyses of ROESY and 1 H NMR data. Randomly assigning H-2 as α-oriented, the cross peaks of H-4/H-2 and H-5/H 2 -15, together with the coupling constant (J = 9.6 Hz) between H-4 and H-5 indicated that both H-4 and CH 3 -14 adopted α-orientation, while both H-5 and CH 2 OH-15 adopted β-orientation . Fortunately, colorless square crystals of 1 were eventually obtained through slow evaporation of methanol. Then a single-crystal X-ray diffraction experiment with Cu Kα radiation of 1 [Flack parameter = 0.07 (3)] unambiguously verified the aforementioned deduction and assigned the absolute configuration of 1 as 2R,4S,5S, and 8S [fig_ref] Figure 3: X-ray crystallographic structure of 1 [/fig_ref]. Arthrinins A-D from the sponge derived fungus Arthrinium sp. has been reported, so we gave compound 1 the trivial name from arthrinin E. Compound 2 was obtained as colorless oil. Its molecular formula C 14 H 24 O 3 was ascertained by the positive HRESIMS ion peak (m/z 263.1618, [M + Na] + , calcd for 263.1618), indicating three degrees of unsaturation. Comparison of the 1 H NMR spectrum of 2 with that of 1 disclosed that 2 was structurally analogous to compound 1 [fig_ref] Table 1 1: H NMR data [/fig_ref]. Furthermore, the 13 C NMR and DEPT spectra, which demonstrated 14 carbons resonances involving three methyls, four sp 3 methylenes, four sp 3 methines, one quaternary carbon and two olefinic carbons, suggested that the structure of 2 was highly similar to that of boledulin C, a 15-nor-botryane sesquiterpenoid.
More credible evidence was obtained from 2D NMR spectra (Figs. S20-23) of 2. On one hand, the structure of ring A was established by the 1 H-1 H COSY correlations of H 2 -10/H-1/H-2/H 2 -3/H-4/H-5 and H-2/H 3 -11 and HMBC correlations from H-5 to C-9 (δ C 139.8), from H 2 -10 to C-9. On the other hand, the HMBC correlations from H-5 to C-8 (δ C 136.9) and C-13 (δ C 31.6), from H 2 -7 to C-5 (δ C 60.9), from H 3 -12 to C-6 (δ C 39.7) and C-7 (δ C 51.7), from H 3 -13 to C-12 (δ C 25.6), from H 2 -14 to C-7 and C-8 defined the ring B fragment and planar structure of 2 ultimately .
As for the configuration of 2, randomly assigning H-5 as β-oriented, the observed correlations of H 2 -10/H 3 -11/H-5 in the ROESY spectrum, together with the coupling constant (J = 4.4 Hz) between H-1 and H-2 implied that both CH 2 OH-10 and CH 3 -11 adopted β-orientation, while the analysis of coupling constant (J = 10.8 Hz) between H-4 and H-5 suggested that H-4 was α-oriented . Then, (1R*,2R*,4S*,5S*)-2 was subjected to quantum chemical calculation of NMR chemical shifts and spin-spin coupling constants at mPW1PW91-SCRF/6-31 + G(d,p)//B3LYP-D3BJ-SCRF/6-31G(d) and B972-SCRF/pcJ-1//B3LYP-D3BJ-SCRF/6-31G(d) level of theory (both with methanol as solvent and SMD solvent model), respectively, and the calculated results matched their experimental counterparts very well [fig_ref] Table 3: Analyses of the NMR computation results of [/fig_ref] , which verified the established planar structure and relative configuration of 2. Moreover, TDDFT ECD calculation of (1R,2R,4S,5S)-2 was run at CAM-B3LYP-SCRF/def2-SVP//B3LYP-D3BJ-SCRF/6-31G(d) level of theory in MeOH with SMD solvent model, and the obtained curve, which is entirely consistent with its experimental counterpart, supported the absolute configuration of 2 to be 1R,2R,4S, and 5S [fig_ref] Figure 4: Experimental ECD spectrum of 2 [/fig_ref]. Therefore, compound 2 was confirmed to possess 15-nor-botryane skeleton and named as arthrinin F.
Compound 3 was isolated as a colorless oil. The [M+Na] + peak at m/z 285.1100 (calcd for 285.1097) existing in the positive HRESIMS spectrum revealed the molecular formula of 3 to be C 15 H 18 O 4 , which implied seven indices of hydrogen deficiency. Its IR spectrum indicated the presence of hydroxyl (3419 cm -1 ), methyl (2932 and 2869 cm -1 ), phenyl (1607 and 1467 cm -1 ) and ester (1743 and 1040 cm -1 ) groups. The 1 H NMR spectrum gave two characteristic aromatic protons at δ H 7.43 (d, J = 7.9 Hz), 7.72 (d, J = 7.9 Hz) and two singlet methyl groups (δ H 1.37, 1.34) [fig_ref] Table 1 1: H NMR data [/fig_ref]. The 13 C NMR and DEPT spectra displayed 15 carbon signals, including one ester carbonyl carbon, six aromatic carbons, two sp 3 quaternary carbons, four sp 3 methylenes and two methyls. The above data suggested that the structure of 3 resembled to that of dehydrobotrylactone with a botryane scaffold. The 1 H-1 H COSY correlation between H-3 and H-4, along with the HMBC correlations from H-3 to C-1 (δ C 145.2), C-5 (δ C 157.8) and C-11 (δ C 173.9), from H-4 to C-2 (δ C 126.0) and C-9 (δ C 144.7), from H 2 -10 to C-1, C-2 and C-11 established the structure of the butyrolactone motif (ring C), and its connection to a tetra-substituted benzene ring (ring A). Moreover, the HMBC cross peaks from H-4 to C-6 (δ C 50.1), from H 2 -7 to C-13 (δ C 27.2) and C-15 (δ C 70.8), from H 2 -12 to C-5 and C-6, from H 3 -13 to C-12 (δ C 71.3), from H 3 -14 to C-8 (δ C 50.2) and C-9, from H 2 -15 to C-14 (δ C 24.9) determined the existence of a cyclopentane fragment (ring B) and its linkage with ring A. As a result, the planar structure of compound 3 was elucidated as shown in .
As for the stereochemistry of 3, randomly assigning H-7a as α-oriented, according to the ROESY correlations observed from H-7α to both H 3 -13 and H 3 -14, as well as from H-7β to both H 2 -12 and H 2 -15, it can be concluded that both CH 3 -13 and CH 3 -14 adopted α-orientation. Then, the NMR chemical shifts of (6R*,8S*)-3 were calculated at mPW1PW91-SCRF/6-31 + G(d,p)//B3LYP-D3BJ-SCRF/6-31G(d) level of theory with methanol as solvent and SMD solvent model, the predicted chemical shifts matched their experimental counterparts very well [fig_ref] Table 3: Analyses of the NMR computation results of [/fig_ref] , which supported the above deduction concerning the relative configuration of 3. Furthermore, the theoretical ECD spectrum of (6R,8S)-3 was obtained at CAM-B3LYP-SCRF/def2-SVP// B3LYP-D3BJ-SCRF/6-31G(d) level of theory in MeOH with SMD solvent model, and the calculated curve matched the experimental one very well and thus supported the absolute configuration of 3 to be 6R,8S. . Compound 3 was given the trivial name arthrinin G.
Additionally, compounds 1-3 were evaluated for their cytotoxicity against five human cancer cell lines (HL-60, A549, SMMC-7721, MCF-7, SW480), with cis-platin and paclitaxel as positive controls, however, no compounds showed activity against the tested cell lines .
## Experimental
## General experimental procedures
Optical rotations were measured with a JASCO P-1020 polarimeter. UV spectra were obtained using a Shimadzu UV-2401 PC spectrophotometer. A Tensor 27 spectrophotometer was used for scanning IR spectroscopy with KBr pellets. 1D and 2D NMR spectra were recorded on Bruker DRX-600 and 500 spectrometers with TMS as internal standard. Chemical shifts (δ) are expressed in parts per million (ppm) with reference to the solvent signals. HRESIMS was performed on an API QSTAR spectrometer. Semipreparative HPLC was performed on an Agilent 1200 liquid chromatograph with a Zorbax SB-C18 (9.4 mm × 250 mm) column. LC-MS/MS was performed on Agilent 6530 Accurate-Mass Q-TOF spectrometer coupled to an Agilent 1290 LC system with Zorbax SB-C18 (9.4 mm × 250 mm) column. Column chromatography was performed with silica gel (100-200 mesh, Qingdao Marine Chemical, Inc., Qingdao, People's Republic of China). Fractions were monitored by TLC, and spots were visualized by heating silica gel plates sprayed with 10% H 2 SO 4 in EtOH.
## Fungal material, identification and fermentation
The fungal strain of Arthrinium sp. HS66 was isolated from the fresh stems of Isodon xerophilus that was collected from Kunming Botanical Garden, Kunming City, Yunnan Province, People's Republic of China, in August 2018. The isolate was identified based on sequence (GenBank Accession No. MT355097) analysis of the ITS region of the rDNA. The fungal strain was cultured on slants of potato dextrose agar at 28 °C for 7 days. Agar plugs were cut into small pieces (about 0.5 × 0.5 × 0.5 cm 3 ) under aseptic conditions, and 15 pieces were used to inoculate three Erlenmeyer flasks (500 mL), each containing 200 mL of media (0.4% glucose, 1% malt extract, and 0.4% yeast extract); the final pH of the media was adjusted to 7.0, and the flasks were sterilized by autoclave. Three flasks of the inoculated media were incubated at 28 °C on a rotary shaker at 170 rpm for 7 days to prepare the seed culture. The detailed lager fermentation procedure was as following: Fermentation was carried out on solid rice medium in 125 Fernbach flasks (500 mL, 90 mL distilled water was added to 80 g rice and kept overnight before autoclaving). Each flask was inoculated with 5.0 mL of the spore inoculum and incubated for 30 days at 28 °C in a static incubator.
## Cytotoxicity assay
Five human cancer cell lines, human myeloid leukemia HL-60, lung cancer A-549 cells, hepatocellular carcinoma SMMC-7721, breast cancer MCF-7, and colon cancer SW480, were purchased from the Shanghai Institute of Biochemistry and Cell Biology, Chinese Academy of Sciences (Shanghai, China). Cells were cultured according to the manufacturer' recommendations. All mediums were supplemented with 10% fetal bovine serum (FBS), 100 units/ml penicillin G sodium and 100 μg/ml streptomycin (HyClone). All the cells were incubated at 37 ℃, 5% CO 2 in a humidified atmosphere. Cytotoxicity of compounds was determined by MTS method. Briefly, 5 × 10 3 cells were plated in 96-well plates 12 h before treatment and continuously exposed to test compounds for 48 h. Then MTS (Promega) was added to each well. The samples were incubated at 37 ℃ for 1-4 h and the optical density (OD) was measured at 490 nm using a microplate reader (Bio-Rad Laboratories). The IC 50 values were calculated by Reed and Muench's method.
## Extraction and isolation of compounds 1-3
The culture medium was overlaid and extracted with MeOH by maceration. With filtration and concentration, the resultant extract was partitioned with EtOAc. Then the solvent was evaporated in vacuo to afford a crude extract (130 g).
The extraction was subjected to column chromatography on silica gel with a CHCl 3 /Me 2 CO gradient system (1:0
## Physical constants and spectroscopic data of compounds 1-3
# Computational method
Conformational searching of (1R*,2R*,4S*,5S*)-2 and (6R*,8S*)-3 were undertaken with the CREST code (version 2.8) using the default iMTD-GC procedure. The first 20 conformers of (1R*,2R*,4S*,5S*)-2 and (6R*,8S*)-3 were subjected to DFT geometry optimization at B3LYP-D3BJ-SCRF/6-31G(d) level of theory (with MeOH as solvent and SMD solvent model). Frequency analyses of all optimized conformers were undertaken at the same level of theory to ensure that no imaginary frequency exists. Then, thermal correction to Gibbs free energies obtained by frequency analyses were added to the electronic energies obtained at B3LYP-D3BJ-SCRF/6-311 + G(d,p) level of theory (with MeOH as solvent and SMD solvent model) to get the Gibbs free energies of each conformer. Subsequently, Room-temperature (298.15 K) equilibrium populations were calculated according to Boltzmann distribution law:
where P i is the population of the ith conformer; n i the number of molecules in ith conformer; ΔG is the relative Gibbs free energy (kcal/mol); T is room temperature (298.15 K) here; R is the ideal gas constant (0.0019858995). Those conformers with a population of over 2% were subjected to subsequent NMR and ECD calculations.
NMR shielding constants were calculated with the GIAO method at mPW1PW91-SCRF/6-31 + G(d,p) level (with MeOH and SMD solvent model). The obtained shielding constants were converted into chemical shifts by referencing to TMS at 0 ppm (δ cal = σ TMSσ cal ), where the σ TMS was the shielding constant of TMS calculated at the same level of theory. The parameters a and b of the linear regression δ cal = aδ exp + b; the correlation coefficient, R 2 ; the mean absolute error (MAE) defined as Σ n |δ calδ exp |/n; the corrected mean absolute error (CMAE) defined as Σ n |δ corr δ exp |/n, where δ corr = (δ calb)/a were calculated. Calculation of coupling constants were run at B972/pcJ-1 level of theory (with MeOH as solvent and SMD solvent model).
TDDFT ECD calculations were run at CAM-B3LYP/ def2-SVP level of theory (with MeOH as solvent and SMD solvent model). For each conformer, 30 excited states were calculated. The calculated ECD curves were generated using the Multiwfn software (version 3.7).
[formula] p i = n i ∑ j n j = e −ΔG i ∕RT ∑ j e −ΔG j ∕RT [/formula]
The geometry optimization, single-point energy calculation, NMR shielding constant calculation, coupling constant calculation, and TDDFT ECD calculation were all completed in Gaussian 09 program.
[fig] Figure 1: Chemical structures of compounds 1-3 [/fig]
[fig] Figure 3: X-ray crystallographic structure of 1 [/fig]
[fig] Figure 4: Experimental ECD spectrum of 2 (black); Calculated ECD spectra of (1R,2R,4S,5S)-2 (shift = 12.5 nm, red) and (1S,2S,4R,5R)-2 (shift = 12.5 nm, red dash)Fig. 5 Experimental ECD spectrum of 3 (black); Calculated ECD spectra of (6R,8S)-3 (shift = 12 nm, red) and (6S,8R)-3 (shift = 12 nm, red dash) [/fig]
[table] Table 1 1: H NMR data (δ in ppm, J in Hz) of compounds 1-3 a Recorded at 600 MHz, Recorded in CD 3 OD b Recorded at 500 MHz, Recorded in CD 3 OD [/table]
[table] Table 3: Analyses of the NMR computation results of (1R*,2R*,4S*,5S*)-2 and (6R*,8S*)-3 [/table]
|
10.1038/s41598-022-25863-2
|
CCBY
|
9744848
|
36509873
|
s2orc_pubmed_articles
|
Place cells dynamically refine grid cell activities to reduce error accumulation during path integration in a continuous attractor model
Navigation is one of the most fundamental skills of animals. During spatial navigation, grid cells in the medial entorhinal cortex process speed and direction of the animal to map the environment. Hippocampal place cells, in turn, encode place using sensory signals and reduce the accumulated error of grid cells for path integration. Although both cell types are part of the path integration system, the dynamic relationship between place and grid cells and the error reduction mechanism is yet to be understood. We implemented a realistic model of grid cells based on a continuous attractor model. The grid cell model was coupled to a place cell model to address their dynamic relationship during a simulated animal's exploration of a square arena. The grid cell model processed the animal's velocity and place field information from place cells. Place cells incorporated salient visual features and proximity information with input from grid cells to define their place fields. Grid cells had similar spatial phases but a diversity of spacings and orientations. To determine the role of place cells in error reduction for path integration, the animal's position estimates were decoded from grid cell activities with and without the place field input. We found that the accumulated error was reduced as place fields emerged during the exploration. Place fields closer to the animal's current location contributed more to the error reduction than remote place fields. Place cells' fields encoding space could function as spatial anchoring signals for precise path integration by grid cells.
# Results
## Salient visual features enabled the place cell's fields to emerge. we have employed deepmind
Labto simulate the spatial exploration of a rodent-like organism in a square arena (animal in brief from now on) [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref]. We took visual scenes from the animal's perspective during random exploration [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] to identify salient visual features that elicited activity in place cells. The DeepMind Lab simulator allowed the animal to move and adapt to a configurable synthetic environment. Across all the simulations presented here, the animal's navigation was only controlled by the search-and-seek factory definition provided in the simulator, but not by the place or grid cell networks.
We registered the movements across the spatial exploration (from a zenithal standpoint) of the square arena without goals (i.e., food or water for a rat). Two example trajectories of random exploration with different strategies are shown in [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] : a trajectory displaying mostly unevenly distributed paths around the center of the arena (S1), and a trajectory primarily close to the walls (S2). In both cases, the animal covered the entire arena over time. There was no information carried over from one exploration to another because all the synaptic weights of both networks were erased between experiments. The rationality of testing our hypothesis based on S1 and S2 was that these trajectories are typically observed in rodents in an open field. Based on S1 and S2, we quantified place and grid cell properties and path integration.
A fully connected feed-forward network was implemented to determine activations of place cells and sensors as presented in [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref]. Visual scenes were used to train the place cell network post navigation, speeding up the learning process of the place cell network and statistical analyses. Based on Hodgkin-Huxley (H-H) model with metabolic requirements and energy consumption features (Suppl. [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] , the place cell network linked proximity sensors ( N s = 3 ) to place cells ( N pc = 2000) through synaptic-like weights (Suppl. . Sensors measured spatial distances (proximity) between the animal's current position and walls. A measured distance value was sent from each proximity sensor to every place cell through different synaptic weights. These synaptic weights coded the relationship between proximity and place cell activation which affected the firing of place cells. Every place cell responding to the current location with a firing power above a certain threshold modified their weights from sensors through a learning rule as reported in. Weights were the basis vectors in the place cell model that computed the firing powers of these neurons [bib_ref] Neural energy supply-consumption properties based on Hodgkin-Huxley model, Wang [/bib_ref] [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref]. Using the firing rate definition as in [bib_ref] Geometric determinants of the place fields of hippocampal neurons, O'keefe [/bib_ref] and [bib_ref] Modeling place fields in terms of the cortical inputs to the hippocampus, Hartley [/bib_ref] , we considered the energy consumption of a place cell during an action potential (or spike) [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref]. Place cells also detected whether preferred visual features appeared on scenes to maximize place field emergence when detecting these features from the sensed environment [bib_ref] A visually driven hippocampal place cell model, Fuhs [/bib_ref].
For a place field to emerge on the animal's current position, three conditions must be met simultaneously: (i) a place cell must fire as a product of the dynamics of the place cell network based on the H-H model (where each neuron receives proximity information) [bib_ref] Neural energy supply-consumption properties based on Hodgkin-Huxley model, Wang [/bib_ref] [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref] ; (ii) a place cell must receive enough input from grid modules (see. (C) Two typical example trajectories (S1 and S2) during random exploration of a 2D square arena. (D) Place field centers for S1 and S2. (E,F) Example place fields for S1 and S2 at different locations with different field sizes. (G) Example of absolute decoding error from the place cell population along the navigation. Plots indicated that there was a tendency for location error to decrease along the simulation for both trajectories. www.nature.com/scientificreports/ scene [bib_ref] A visually driven hippocampal place cell model, Fuhs [/bib_ref]. Only when these conditions are met simultaneously, will a place field on the animal's current position emerge (see Suppl. . More details are provided as follows. Place cells had individual differences from each other, such as sensory preferences for detecting salient features and metabolic-like features (Suppl. . These differences caused the firing power (Suppl. and the distribution of maximum firing power across neurons (Suppl. vary along the network during the simulation. The place fields were defined as the set of all the locations with firing power larger than a predefined value. Each place cell acquired a unique spatial selectivity to form its place field from the detected salient visual features, where most cells developed fewer place fields (Suppl. . This selectivity occurred according to the neurons' randomly defined activation preferences. Furthermore, some neurons showed sparse firing (i.e., firing rates with a highly peaked, non-Gaussian distribution) that defined their place fields, while other neurons presented firing profiles across the arena or close to corners for both trajectories (S1: [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref]. We observed a non-uniform distribution of place field centers across the arena for S1 trajectory [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] , and mostly along the walls for S2 trajectory [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref].
In addition, we evaluated whether place fields conveyed precise spatial information to estimate the animal's position under different trajectories. The estimated location error (Loc) of the animal from place cell activity was computed by the weighted average of place field centers according to the response set by where the place fields emerged, and by the activity power of neurons at each time step. The Loc was normalized using the joint activity power of all neurons [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref]. The error between the actual position and the estimated position from the place cell activities was profiled using 100 bins to characterize the absolute decoding error average (S1: [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] -left; S2: [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref]. In general, we observed that for both trajectories there was a tendency for the location error to decrease over time, but the average was more variable for the S2 trajectory. In this model, a decoding error of 1 indicates that most place neurons fired initially without selectivity to a certain location. As the place cells received different inputs regarding proximity information from sensory neurons, they started selective firing in the arena. Therefore, the normalized Loc over joint activity power of all neurons got closer to 0.
Place fields emergence remained stable after the animal's early exploration. Having shown that place cells can detect salient features from visual scenes as local or distal reference points, we analyzed the spatial distribution of place fields at time t for both trajectories, S1 and S2, using entropy H (i.e., Shannon's entropy), spatial entropy S and information density Z . H measured information on how the arena was encoded by the population of place cells. S measured how random the place fields were distributed in the arena. Since H was related to the number and distribution of the place fields, Z measured H per place in the area (see "Methods" for details). A change in H , S , or Z measured how much information was gained/lost about the arena. As the animal explored the arena more, place fields would get stabilized, and the change in information would be less. By looking at the dynamics of the change, we were able to infer the formation dynamics of the place fields, but more importantly, we were able to infer the stability of place fields.
As shown in , S and Z sharply increased in magnitude after a few simulated time steps (i.e., ~ t = 20), approaching their maximum value from ~ t = 2000 on. For ~ t < 2000 in in both trajectories, small reversals with entropy H and information density Z were observed (insets in , left and right). The reversal effect was largely due to the dynamics of the emerged place fields that increased in number non-uniformly overtime. For ~ t > 2000 in , left and right, there was a decay in H and S , but not for Z . This decay suggested that no further spatial information was acquired after a certain number of place fields emerged. However, these measurements were difficult to interpret because they did not show strong links to the density distributions of place fields given the animal's trajectories. This was also the case for [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] , sharp decreases were observed in the mean estimated location error for both trajectories. Still, for S1 the error remained relatively less variable than for the S2 trajectory across time.
Thus, we next examined changes in the place field distribution when the number of place fields ( n ) were increased during the sampling of the scenes. We plotted the H entropy and its decomposition into S spatial entropy and Z information density against the place field number n in . We did not see significant differences between S1 and S2 trajectories for these measures. This relationship revealed that despite some marginal changes, n is by far the largest determinant of the level of complexity (refer to [bib_ref] Entropy, complexity, and spatial information, Batty [/bib_ref] for an associated measure).
Because the distribution of place fields was largely due to changes in the number of fields, we decided to analyze further the changes in place field distribution through time. As shown in for both S1 and S2, the number of place fields across time increased till late exploration (~ t = 2000), and then remained stable. Reversals in value H were related to the reversals in the number of emerged place fields . With a small number of place fields (i.e., when the animal did not yet cover much space exploring the arena), we should assume then that the changes in entropy were proportionately larger by simply reflecting on changes in the maximum entropies (log(n) + 1 − log(n))/log(n) , which gets smaller as n increases. also showed that the growth in the number of place fields n dominated the change in information I for early exploration. This change was correlated with an early decay in I (i.e., ~ t < 20) because adding place fields impacted the place field spatial information across timeless. There was no further change in the information I after the number of place fields reached a plateau around t = 2000. This time point ( t = 2000) represented when the place fields were stabilized after repeated exploration of the environment. Our analyses suggested that till t = 2000, some place fields emerged fast, while others emerged late, which is similar to empirical findings. For example, when a rat runs a simple linear track, some place fields are formed only after several laps of run even if the rat had explored the linear track in the first run. In our simulation, although the animal had explored all parts of the arena, it took t = 2000-time steps for place fields to cover the arena and have a stable representation as a population. www.nature.com/scientificreports/ Comparing the H entropy for t < 2000 vs. t > 2000, we found that the entropy before and after reaching the stability represented different distributions for the S1 trajectory (p < 1e−200; Mann-Whitney U test; . However, when performed the same analysis for the S2 trajectory, we observed that the H entropy did not represent a clearly different distribution before and after reaching place field stability (p = 0.03; Mann-Whitney U test when assuming p < 0.01; .
Computing the complement of the ratio between the entropy and its maximum value, we also defined a complexity ratio R , which revealed the percentage of information needed at any period to move to maximum complexity of place field distribution and plotted it across the number of emergent place fields and magnified in . At the beginning of the simulation (~ t < 20), the number of place fields was relatively low and then increased (till ~ t = 2000) as the animal explored the arena. Examining the complexity ratio R , it increased quickly and reached a plateau in early exploration, but afterward continued increasing. The late increase after reaching plateau happened for both trajectories steadily but only significantly for the S1 trajectory (S1: p < 1e−200; S2: p = 0.44; Mann-Whitney U test; t in 100-to-2000-time steps vs. in 2001-to-end of the simulation . Importantly, R indicated how significant the effect of emerging place fields was on the complexity measures. This observation was also reflected in the value of H , which correlated with the complexity ratio as illustrated in (the mean complexity ratio is illustrated in .
A comparison of the place field peak location per time step was performed considering the distance of the peak between two consecutive time steps. Supp. [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] showed that the mean change in place field peak location (Δ PF peak location) was significantly higher before place fields reached stability (t = 2000; S1: p = 9.50e−135; S2: p = 4.74e−153; Two-sample t-test). These analyses complemented those reported in in that stability of place fields was reached after initial exploration.
Overall, these results showed a fast growth in the number of place fields from the start of the simulation. A sufficient pool of place cells after early exploration with their individual preferences for detecting salient features in visual scenes enabled space coding.
Grid patterns were resilient but contingent on the animal's trajectory. We next computed mean activity maps (spike densities across the arena; [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] ,B-upper two rows) of grid cells as a function of the animal's position, and the rate map autocorrelation [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] ,B-bottom row) for each dorsal to the ventral module in network organization (S1: [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] ; S2: [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref]. For trajectory S1, these maps showed a coherent and stable activity of the multiple grid subfields from dorsal to ventral modules with visually clear, regular triangular tessellating subfields [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref]. For trajectory S2, since the animal remained closer to walls and less often visited the middle of the arena, we expected a visually less rich grid firing pattern, meaning a less visible but still present triangular tessellation. This phenomenon was shown in the rate (autocorrelations) maps in [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref]. The spatial periodicity of grid cells firing was maintained in S2 [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] , bottom row), but grid formation was distorted because the animal did not uniformly cover the arena [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] , top row). Importantly, our observation of the S2 trajectory can be correlated to recording sessions where animals move in 1D circular tracks 67 , as well as an animal exploring the whole arena before observing the characteristic pattern of grid cells; c.f..
To compare the tessellations of grid cells between trajectories, we examined the distribution of the hexagonal grids using a mean (hexagonal) gridness score. This score measured the degree to which a hexagonal grid resembled a hexagonal pattern during the simulations. Considering all the grid cells, we noticed that the mean gridness score was higher for the S1 than the S2 trajectory (S1: 0.41 ± 0.003; S2: 0.38 ± 0.003; Supp. [fig_ref] Figure 4: Place [/fig_ref]. A comparison between the gridness distributions for S1 indicated that the mean was significantly different than S2 (p = 1.2e−12; Two-sample t-test), and both trajectories represented different distributions (p = 1.6e−32; Twosample Kolmogorov-Smirnov test). Based on 69 , we also measured a "squareness gridness" score; i.e., the squareness of the grids was measured as opposed to 'hexagonality' . The rationale for using the squareness gridness score was to evaluate if the trajectory affected the convergence to hexagons. Our motivation for using the squareness measure was that an initial visual inspection of grid rate maps [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] suggested there were deformations of the gridness across modules and trajectories. We found a higher square-gridness score for the S1 trajectory compared to the S2 trajectory (S1: 0.45 ± 0.008; S2: 0.39 ± 0.007; mean ± sem; Supp. [fig_ref] Figure 4: Place [/fig_ref]. Considering the squareness gridness score, we observed similar results indicating a difference in mean gridness for S1 vs. S2 trajectories (p = 51e−10; Two-sample t-test) and distribution comparison (p = 2.5e−14; Two-sample Kolmogorov-Smirnov test). In brief, these results suggested that gridness was present for both trajectories, even though S1 had higher gridness values than S2.
We also analyzed the effect of S1 and S2 trajectories on position estimates only from grid cells activity among the dorsal to ventral modules in each time step. We expected that the regular hexagonal tessellation as in S1 would present more consistent position estimates between the modules than the less clear tessellation as in S2.
We computed the variance of the animal's current position estimates across modules to detect differences between S1 and S2. The variance of position estimates among modules fluctuated less in S1 [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] -left) compared to the S2 trajectory [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] -middle) throughout the exploration. This difference in variance indicated dissimilar means (p = 1.6e−179; Two-sample t-test) and variance levels (p < 1e−200; Two-sample F-test for equal variances; [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] -right) between trajectories; similar results were observed excluding the most ventral module to test whether the variance is driven by that module (p = 9.91e−144; Two-sample F-test for equal variances). Furthermore, when computing the absolute (Euclidean) distance error about the animal's current position estimates across modules, we observed less fluctuation (variance) over time for the S1 compared to the S2 trajectory only for modules 1, 4, and 5 (p = 5.0e−14, p = 3.7e−18, p = 4.7e−48, respectively), but not for modules 2 and 3 (p = 0.32, p = 0.48, respectively; Two-sample F-test for equal variances; [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref]. For both S1 and S2, the most ventral module (module 5) differed in error compared to dorsal modules [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref]. The average variance across modules for S1 was higher than for the S2 (p = www.nature.com/scientificreports/ www.nature.com/scientificreports/ t-test; [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref]. These observations indicated that the scaling properties of grid cells influenced the estimates, where scaling along the dorsal-to-ventral modules did not seem strictly necessary for optimal error reduction. These results indicated that the grid tessellation depended on the animal's trajectory. Since in S2 the animal was mostly near the walls, the tessellation was distorted compared to S1 because the trajectory sampling did not cover the arena uniformly. This observation does not convey that when firing fields were less regular (as during the S2 trajectory), the information represented by grid cells is less accurate for estimating the animals' position as we will see in the next section. Our results just suggested that the grid's firing fields along the modeled dorsalto-ventral modules might represent a simple solution by equally contributing to the estimations of grid cells contingent on the animal's trajectory.
The grid network reduced the error accumulation using place fields. The collective activity distribution of grid cells as implemented here had one prominent feature: an "activity packet" (bump) per grid module (i.e., dorsal to ventral) that was driven by the animal's movement. The dynamical space where this bump moved was defined by a 2D matrix represented by 20 × 20 grid cells for each module [fig_ref] Figure 4: Place [/fig_ref]. When the animal's velocity changed, the activity packet moved in the grid neuronal dynamic space in a similar proportion. This activity packet and its dynamic move are commonly referred to as "continuous attractor" in the biological and modeling literature 2 . In this scenario, the activity packet created a grid cell's dynamics that enabled estimating the animal's position.
We analyzed here whether place field emergence had a role in reducing the error accumulation of grid cells. The presented analyses were based on four assumptions in our model: (I) both place and grid cells were simultaneously active during navigation and represented two different but concurrent neural populations; (II) when place fields emerged in each time step of the simulation, they received inputs from grid cells as explicitly defined for our model (see grid-place cell relationship in the Method section); (III) by computing the position change of the grid cells' activity bump between consecutive time steps, we intended to estimate animal's next movement; (IV) inputs from the grid to place cells were implemented based on a diversity of spacings represented by the grid modules [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] , where place cells feedbacked to grid cells to correct for drift and anchor grids to environmental cues [bib_ref] Experience-dependent rescaling of entorhinal grids, Barry [/bib_ref]. Regarding the assumption (II), it is worth noting that only the activities of grid cells projecting to place cells varied across the environment. These inputs considered the maximum grid activity per module in each time step and were multiplied by the symmetrical weight matrix to represent the diversity of information provided by the different spacings from the dorsal-to-ventral grid modules. Place fields would only emerge for the active place cells above a certain threshold, given the grid cell input. Note also that place fields were not necessarily defined in the places with above the threshold grid cell input but determined by the place network activities. These properties between grid and place cells remained fixed through all simulations. A simple schematic of the implemented model under these assumptions is shown in [fig_ref] Figure 4: Place [/fig_ref] (see also Supp. for a more detailed explanation).
The activity of the grid network for each module was separately used to compute the spatial position estimates. Emerging place fields affected only the position estimates, not our model's grid cell firing patterns. Figure 4B represents the activity of the grid network across modules that was used to compute the spatial position estimates. Considering the S1 spatial trajectory with and without place cell input to grid cells [fig_ref] Figure 4: Place [/fig_ref] -top and bottom, respectively), the animal's position was estimated accurately only when place field input was present. In the absence of that input, no grid module could estimate the animal's trajectory with precision right from the start position of the simulated animal [fig_ref] Figure 4: Place [/fig_ref] shows the whole estimated trajectory). Supplemental shows similar results but for an S2-like trajectory. Furthermore, [fig_ref] Figure 4: Place [/fig_ref] represents the estimated trajectories with and without place field input for the most ventral grid module. The defined place field centers for the animal's trajectory were plotted in [fig_ref] Figure 4: Place [/fig_ref]. Interestingly, place fields that were defined in the. (B) Example of the "activity packet" (bump) for the grid cell network considering the dorsal to ventral modules individually. The squared dynamical space is represented by the 20 × 20 grid cells. (C) Example of the estimated trajectories made by the grid cell modules during an S1 movement across the arena for 1000-time steps at the top row with place field's input. To evidence that the error accumulation deviates the estimated position from the actual trajectory, the bottom row represents the prediction of each module without place fields' input for a short trajectory only (100-time steps; red trace) from the starting position (see Supp. [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref] for the whole trajectory). Plots indicate a better prediction of the actual animal's trajectory when place field inputs are given to grid cells. (D) The same trajectories as previously shown for the ventral module were magnified for comparison purposes. The green dot represents the starting position, the magenta dot indicates the predicted trajectory ending, and the pink one is the actual ending position. Asterix represents the places where the activities of grid cells across modules could enable place fields. (E) An example of place field centers that emerged during the short trajectory is depicted in panned (D). (F,I) The Euclidean distance error measure compared the estimated and the actual trajectory for dorsal (module 1) to ventral (module 5) modules. Plots indicate that the distance was higher in (I) when place field information to the grid cell network was absent. (G,J) The measure of the variance across estimates showed a similar observation. (H,K) The estimated error was lower when place field input was provided (H) compared to the absence of input (K). The inset numbers represent mean ± sem for the associated data for all the panels. www.nature.com/scientificreports/ places where the input from grid cells was low (i.e., below 1.1 a.u.), were shown as uncertain places in [fig_ref] Figure 4: Place [/fig_ref] ,D. This observation indicated that the place and grid cells were dynamically defined and depended on the neural activations in each time step. Computing the Euclidean distance between the position estimates and the animal's current position, the place field input to grid cells [fig_ref] Figure 4: Place [/fig_ref] resulted in less fluctuation compared to the absence of the input [fig_ref] Figure 4: Place [/fig_ref] across modules (p = 1.0e-53, p = 9.0e−52, p = 1.1e−46, p = 2.6e−51, p = 3.9e−56, respectively; Two-sample F-test for equal variances). We observed that the most ventral module (module 5) had a different fluctuation than other modules in the presence of the place cell input [fig_ref] Figure 4: Place [/fig_ref] , but not in the absence [fig_ref] Figure 4: Place [/fig_ref]. A similar observation was obtained when plotting the variance across modules, indicating different means (p = 7.1e−19; Two-sample t-test) and variance levels (p = 8.5e−177; Two-sample F-test for equal variances) with or without the place field's input for S1 [fig_ref] Figure 4: Place [/fig_ref]. The estimate indicated a higher mean of error in the absence of place field inputs to the grid cell estimate than in the presence of the input (p = 3.7e−29; Two-sample t-test; [fig_ref] Figure 4: Place [/fig_ref] ,K; see "Methods"). Performing the same analyses for the S2 trajectory presented similar results as for S1 (Supp. . When place field input to grid cells was provided (Supp. , it resulted in less absolute distance error compared to the absence of the input (Supp. across modules (p = 3.9e−43, p = 1.7e−42, p = 4.1e−42, p = 9.2e−44, p = 6.7e−49, respectively; Two-sample F-test for equal variances). Plotting the variance across modules in these situations indicated different means (p = 1.5e−24; Two-sample t-test) and variance levels (p = 8.1e−187; Two-sample F-test for equal variances) for S1 and S2 trajectories (Supp. . The estimated error showed a higher mean of error in the absence of place field inputs to the grid cell estimate than in the input's presence (p = 9.2e−46; Two-sample t-test; Supp. . These results informed us that the grid cell network could reduce the error accumulation using the emerged place fields during navigation in our model.
Overall, the rate maps in [fig_ref] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the... [/fig_ref] with an isolated grid network showed stable grid firing patterns throughout the session. However, our results indicated an intrinsic accumulation of error when estimating animal's location even without an induced noise source to the grid network. The estimation error is rooted in the grid cells' connectivity matrix that determined the activity bump (see "Methods"-position estimation). The connectivity matrix affected the position of the activity bump because the matrix represented a discretization of the bump's movements in the neural space. When estimating the position of the animal based on the previous position and the magnitude and direction of the bump's movement, there was an error accumulated in the estimation that depended on where the activity peak was placed in the grid matrix. Consequently, if the activity bump moved to certain coordinates in the matrix at time t relative to time t-1, then there was a discretization of how many relative translational units within the matrix (i.e., how many positions in the matrix and the direction of that relative movement) we considered determining the magnitude of the animal's movement.
Closer place fields increased spatial precision for grid cells' location estimation drift. We next analyzed whether closer or distal place fields to the animal's current position differently reduced the error accumulation. To address this issue, the number and the distribution of place fields were manipulated. Their effect on place estimation during a short trajectory was evaluated [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref]. The place fields were randomly placed across the arena during the S1 trajectory under two scenarios: a fewer number of place fields (PF = 20; [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref] -top) and a larger number of place fields (PF = 200; [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref]. The rationale in proposing this configuration was that the distance between the place fields and the animal's current position followed a Gaussian distribution in both scenarios, but because of the difference in sampling (i.e., PF = 20 vs. PF = 200) the distances of place fields to the animal's current position would have been higher in the fewer place fields scenario.
Our model with place field inputs on grid cells showed a good estimate of position across modules only when the number of place fields was high [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref]. Specifically, we compared simulations in which there were relatively fewer place fields [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref] to those with a larger number of fields [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref]. Computing the Euclidean distance between the position estimates and the animal's current position resulted in less fluctuation with a larger number of place fields across modules (p = 3.9e−06; p = 0.05; p = 0.04; p = 0.01; p = 1.1e−04, respectively; Twosample F-test for equal variances). A similar observation was obtained when plotting the variance across modules, indicating different means (p = 1.0e−04; Two-sample t-test) and variance levels (p = 2.8e−17; Two-sample F-test for equal variances) for S1 and S2 trajectories [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref]. The estimated error for such comparison regarding place field positions indicated a higher error mean when a relatively small number of place fields was provided compared to higher numbers (p = 1.3e−05; Two-sample t-test; [fig_ref] Figure 5: A lower variance and estimated error across grid modules were observed with... [/fig_ref] ,G; see "Methods").
Overall, these results indicated that when place fields were statistically closer to the actual animal trajectory, there were less variance and error across grid modules. The high number and closer place fields to the animal's trajectory were the most critical factors in reducing the error. As the number of place cells increased, the possibility of having a field near the animal's trajectory also increased. The evidence was in line with the notion that place fields correct for drift and anchor grids to environmental cues 70 . Our model supported this hypothesis by showing that these defined place fields must be relatively close to the animal's trajectory.
# Discussions
Several modeling studies in the literature have proposed that place cell inputs can stabilize and reduce accumulated error in grid firing patterns [bib_ref] Modular realignment of entorhinal grid cell activity as a basis for hippocampal..., Monaco [/bib_ref] [bib_ref] A spin glass model of path integration in rat medial entorhinal cortex, Fuhs [/bib_ref] [bib_ref] A hybrid oscillatory interference/continuous attractor network model of grid cell firing, Bush [/bib_ref]. Our work goes beyond these observations by examining the influence of dynamic integration of emerging place fields in a new environment, rather than defining static preassigned place fields, reaching similar conclusions to those previous works.
The main contribution of this work is that we characterized how to incorporate place cells-like information into grid path integration for error reduction. We hypothesized that the dynamic coupling between place-and grid-like cells with position information in the form of place fields would have integrated velocity inputs and spatial information in the grid network for path integration. Spatial entropy and information density measures characterized the evolution of place fields' emergence during exploration. This study effectively demonstrated www.nature.com/scientificreports/ www.nature.com/scientificreports/ that place cells could dynamically anchor grid cell activities to reduce the error accumulation in a continuous attractor model. The results showed a simple but realistic representation of place cells allowing a simulated animal to encode spatiotemporal information relative to its visual and distal information for place field definition. The model identified salient features in visual scenes obtained from the 3D DeepMind Lab environment in a realistic fashion. Furthermore, we showed that idiothetic information enabled the grid network to obtain an updated current position estimate. The analyses finally characterized how increasing the number of emerging place cells through their place fields affected grid error reduction. Our position estimates considered each grid modules' activity separately. We could also have combined grid modules' activity for decoding, as commonly proposed in empirical studies. It is still uncertain, however, how the brain makes such a combination, given the different grid scales along the dorso-ventral axis. It is also uncertain which brain structure sustains global estimates for position representation. In fact, it has been recently proposed that the coordination of activity across individual modules is necessary to assess how grid cells encode the brain's internal representation of position [bib_ref] Grid-cell modules remain coordinated when neural activity is dissociated from external sensory..., Waaga [/bib_ref]. In this respect, Waaga et al. indicated that there are specific mechanisms between grid-cell modules, but it is unclear from that work how an overall estimate would emerge from grid modules. Supp. shows the estimated trajectories using a global trajectory estimate when averaging estimates from all the modules in each time step with and without place fields' input. It can be seen in Supp. that even after averaging across modules, the position estimate is accurate only with input from place cells.
Vision is sufficient to prompt place cell responses to encode local cues. It has been reported that animals recognize landmarks through visual identification, and the animal's current location relative to a landmark is encoded and remembered [bib_ref] A visually driven hippocampal place cell model, Fuhs [/bib_ref] [bib_ref] Visual stimulus features that elicit activity in object-vector cells, Andersson [/bib_ref] [bib_ref] Slow feature analysis: Unsupervised learning of invariances, Wiskott [/bib_ref]. Within this cognitive framework, coding places is possible even for a simple neural network that processes an image frame in each time step. Previous studies discussed consequently that a complicated cognitive mechanism seems unnecessary [bib_ref] A visually driven hippocampal place cell model, Fuhs [/bib_ref]. In this respect, the simplicity of a cognitive mechanism was satisfied in our model by two conditions to determine one's location relative to the landmarks 74 : (a) a function of the feature space that changed among landmarks but did not change with different viewpoints of the same landmark; (b) a function of the feature space that changed with one's relative position to the landmark. These functions were implicitly embedded in the model described in this work, and the model conveyed enough spatiotemporal information from visual processing to enable dynamic place information; c.f. [bib_ref] Vector-based navigation using grid-like representations in artificial agents, Banino [/bib_ref] [bib_ref] Emergent elasticity in the neural code for space, Ocko [/bib_ref]. These sensory neurons provided a complementary but complete representation of the experienced environmental layout, regardless of specific landmark features and boundaries in the simulated environment. Considering our previous works on place field emergence 78,79 , the described simulation showed that vision is sufficient to prompt place cell responses that resemble local cues and other spatial distal cue identification 9,57,80 . However, we cannot conclude whether these cell-like profiles differ in their underlying mechanisms. We can only say that the modeled place cells represented the animal's position relative to a visually recognized local and a distal landmark.
## Place cells can emerge as a by-product of the experience.
Our results agree with the idea of experience dependent place field formation. However, the experience-dependent formation of place fields contradicts evidence for an underlying hardwired circuit in the hippocampus [bib_ref] Place cells, grid cells, and memory, Moser [/bib_ref] [bib_ref] Cell assemblies, sequences and temporal coding in the hippocampus, Dragoi [/bib_ref] [bib_ref] Selection of preconfigured cell assemblies for representation of novel spatial experiences, Dragoi [/bib_ref] [bib_ref] Distinct preplay of multiple novel spatial experiences in the rat, Dragoi [/bib_ref] [bib_ref] Preplay of future place cell sequences by hippocampal cellular assemblies, Dragoi [/bib_ref]. In more detail, in 86 sought to address this issue by recording place fields as rats entered a novel environment. Ten of the 12 recorded cells appeared to have spatial firing fields immediately, indicating that the place-cell map was largely predetermined. Although this evidence suggests the possibility of preexisting maps, experience also has a critical role in shaping hippocampal maps of space. Place fields are expressed in some form when animals are put into an environment for the first time, although the map may evolve further with experience. In 88 this phenomenon was also shown on a novel arm of a radial maze, in which some place fields only became evident after 1-2 min experience. If place cell formation is emerging from experience instead of being hardcoded in the brain, that would indicate the existence of a specific group of place cells responding solely to spatiotemporal relative information. Verifying this will help us to get a deeper insight into how we perceive and interact with space and time in our daily experiences. Our study suggests that place cell formation can emerge from experience instead of being hardcoded in the brain. This phenomenon presupposed the existence of a specific group of place cells responding solely to spatiotemporal relative information. Our entropy analyses enabled us to characterize aspects that emerge from experiencing the environments. Those emerging aspects were relevant to spatial information based on the shape of place field distribution (i.e., how the population of place fields spread across the environment) and on the number of events (i.e., the number and the spiking activity of place cells). Our analyses also identified a tradeoff between the shape of the distribution and the number of place fields. When the number of place fields increased, the spatial entropy got larger. Nevertheless, the shape of the distribution made a difference. This work suggested that the animal's movement strategy was helpful for place field development because the network quickly reached high coding complexity and spatial information after early exploration of the animal-like model. The place cells processed plain visual features as anchoring points, allowing navigation to proceed in environments with a few free-standing landmarks. We saw a complex tradeoff between the density and the number of events that emerge through experience.
There are still open questions on the place-grid cell relationship. We showed that a combination of idiothetic and allocentric sensory inputs enabled error reduction in grid cells through place field information. Experimental observations contradict the assumption of place cells taking a role in incremental (and relatively slow) spatial coding. Furthermore, some theoretical models commonly define fixed place fields during the training of grid networks from the beginning of simulations, disregarding the dynamics of place cell-like emergence [bib_ref] Prediction of the position of an animal based on populations of grid..., Guanella [/bib_ref] www.nature.com/scientificreports/ between internally generated path integration signals and sensory processing [bib_ref] Grid cell symmetry is shaped by environmental geometry, Krupic [/bib_ref]. In this, further studies should evaluate the effect of place field size (see [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] on the grid cells' error reduction for path integration. We expect that smaller place fields would produce a better error reduction than wider fields. Our work proposed that allocentric place cells (e.g., based on vision) can be used to recalibrate the activity of the grid cells in case of path integration errors, and, in turn, the activity of grid cells is used to enable place cells to disambiguate between two visually similar places [bib_ref] A model of grid cells based on a path integration mechanism, Guanella [/bib_ref]. Our observations align with the experimental evidence indicating that a toroidal topology of grid cell populations tends to accumulate error over time [bib_ref] Toroidal topology of population activity in grid cells, Gardner [/bib_ref]. Our model implied that grid cells could exist without place field input, which has some parallelism with experimental data showing that grid cells maintain synchrony during hippocampal inactivation even when the grid pattern is lost. This experimental evidence showed that grid cells require an excitatory drive from the hippocampus as one prerequisite for forming and translocating grid patterns in the MEC [bib_ref] Grid cells require excitatory drive from the hippocampus, Bonnevie [/bib_ref]. Consequently, we can infer that when the grid pattern is lost, the grid cells cannot estimate the animal's position in biological networks. However, further experimental evidence is needed to prove our observation.
# Methods
All data analyses and scripting reported in this manuscript were made through custom code using Matlab R2016b. This work is licensed under a Creative Commonstaken from a simulated atop camera on a simulated, rodent-like animal (i.e., a mobile robotic agent). The images referred to partially observed and visually diverse landmarks. Each experiment reported in this work enabled the animal to explore and quickly interact with the environment. Images of 84 × 84 pixels with a bin depth of 24 were obtained from multiple viewpoints depending on the animal's movements. Because the image set had a certain redundancy of scenes (i.e., images taken with the same orientation in nearby places), the dataset processed in each experiment was a down sample of all the image sets taking every 40 images starting with the first.
Blob detection. Visual blobs were detected from each visual scene (image frame) using the algorithm reported in [bib_ref] An attention model for extracting components that merit identification, Jahangiri [/bib_ref] (image processing module from Blob Detection Toolbox-Imperial College London). We defined visual blobs as a set of pixels after three main steps: (i) toboggan edge-preserving smoothing; (ii) applying an interest operator which was designed to be scale and color invariant; (iii) employing morphological operators and connected component analysis to extract convex regions, which did not constitute straight lines and occupied a considerable amount of the whole scene. There was no restriction on the number of blobs detected in an image to ensure the representation of the entire scene. The images themselves did not comprise far fewer regions of approximately uniform intensity. Without merging blobs, multiple blobs occupying part of the area of a particular region with uniform intensity did compete. By intermittently saving the state of each blob at each time step of the simulation, the algorithm created a database of blobs at varying sensitivity levels to intensity changes. To use blob information for matching the place cell's preferences, the algorithm detected the area of each blob, the angle of orientation of the central axis, solidity, and centroid coordinates in the image. Few other simple features such as average intensity, elevation, azimuth (i.e., the vertical and horizontal positions of the blob center in the image, respectively), size, eccentricity (i.e., the ratio of the lengths of the major and minor axes) were not included but measured for future investigations.
## Experimental definition. environmental configuration. we set up a google deepmind lab environment
for each experiment to obtain visual scenes and recorded the positional statistics of the simulated animal in each time step. For the most straightforward configuration (i.e., search and avoid factory in a squared arena), a square arena with a side length of L = 100 (dimensions: 100 × 100 × + ∞ arbitrary units) was defined. At the beginning of the simulation, the animal was placed in the center of the arena with a random orientation, leaving the animal to explore the arena freely. Despite the animal moving in 2D space, the six borders in a 3D space were regarded as landmarks. The animal perceived environmental cues using its visual (simulated atop camera as previously indicated) or distance sensing, then processed the information through its place cell network. The place cell model. Proximity sensor model. Considering the 3D simulated environment without any references other than the delimiting walls of the arena, the simulated animal only got visual or proximity information from walls or distal landmarks placed further than the walls (landmarks). To study the locating function of place cells, we referred to that egocentric information of position as front (F), back (B), left (L), right (R), up (U), and down (D) in analogy to [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref]. The distance up and down was not considered in the model and was kept constant. In this way, we naturally approached the referencing place in animals (c.f., 2D reference in [bib_ref] Geometric determinants of the place fields of hippocampal neurons, O'keefe [/bib_ref]. The actual location of the animal at time t was described, however, based on just L-R, B-F, D-U main axis information. This part was made through the actual distances for the 2D space to walls in a vector representation X(t) = ( x i (t) , y i (t) , z i (t) ), were z i is kept constant. Because it was assumed here that the acquired perception should not be accurate due to sensory noise of its locations, then the input in each coordinate added an error of sensory perception as X ′ (t) = x i (t), y i (t), z i (t) .(1 + αη) , where α represented the error rate of the sensory (visual or proximity) perception. The random number, η , was taken from a uniform distribution within the interval [− 1, 1], that is η ∼ U [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] www.nature.com/scientificreports/ represent the distance to the animal 65 , which affected the firing of place cells [bib_ref] Odor supported place cell model and goal navigation in rodents, Kulvicius [/bib_ref]. A feed-forward network was implemented to determine the locating system constituted of place and sensors [bib_ref] Geometric determinants of the place fields of hippocampal neurons, O'keefe [/bib_ref] [bib_ref] Odor supported place cell model and goal navigation in rodents, Kulvicius [/bib_ref].
Energy consumption. To compute the firing profile of the place cell network, the implemented model was based on [bib_ref] Neural energy supply-consumption properties based on Hodgkin-Huxley model, Wang [/bib_ref] [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref] , calculating the energy consumption of a place cell through the Hodgkin-Huxley (H-H) definition. The neural energy was measured as:
These terms represented the neuron components as follows: V r = −65 mV as the resting membrane potential; m = 0.5 , h = 0.06 , n = 0.5 ; I = 10 ; the input current as 0.1 mA; base V = 65 ; three different ion currents: g Na = 120 maximum Na + conductance mS/cm 2 ; e Na = 150 mV indicating the Nernst potentials of Na +; g k = 36 indicated the maximum K + conductance mS/cm 2 ; E k = −12 mV represented the Nernst potentials of K +; g l = 0.3 was the leakage conductance mS/cm 2 ; E l = 10.6 mV as Nernst potentials while there is no leakage current; dt = 0.01 time step for forward Euler method. The additional 'gating' variables m , n , h model the probability that a channel is open at a given moment in time. The combined action of h controls the Na+ channels while the K + gates are controlled by n 97 . The specified values were used for the simulation parameters 1 of the H-H model. For simulation parameters 2, the number of place cells was as set to N pc = 2000 ; Number of proximity sensors was N s = 3 ; Sensory error rate as a = 0.1 ; Learning rate as µ = 0.001 ; Maximum firing rate as R m = 20 Hz ; Firing threshold as P thr = 0.3 ; Number of simulation steps sets to 10,000;
Step length as 1.
Place cell learning rule. A feed-forward network was defined as fully connected, linking proximity sensors ( N s = 3 ) and place cells ( N pc = 2000 ) through a synaptic-like weights matrix W(t) , with w ij (t) the connection weight from the i-th sensor to the j-th place cell at time t. Based on 96 , weights were initialized as w ij (t) = (1 + exp((γ − E(γ ))/2σ 2 )) −1 , with γ ∼ U(0, 1), and expectation E(γ ) = 0.5 . Using that definition of w ij (t) , we imposed that place fields could emerge both near and far from the boundary of the environment them; c.f., [bib_ref] Odor supported place cell model and goal navigation in rodents, Kulvicius [/bib_ref]. Weights were the basis vectors in the model, which were used to compute the firing powers of place cells. When the competitive learning rule was employed, place cells became tuned to a specific input, which led to the spatial selectivity [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref]. Using the firing rate model proposed in [bib_ref] Geometric determinants of the place fields of hippocampal neurons, O'keefe [/bib_ref] [bib_ref] Modeling place fields in terms of the cortical inputs to the hippocampus, Hartley [/bib_ref] , we defined the firing power P f j (t) of the j -th place cell at time t as: P f j (t) = CR m exp(−(1/n�X ′ (t)/L − w j (t)�) 2 /(2σ 2 j )) , where R m is the maximum firing rate of a single place cell, which is about 20 Hz. The norm represents the Euclidean distance, n is the number of sensory inputs, and w j (t) is the j-th row of W(t). C is the energy consumption by a place cell during an action potential which was ~ 188 nJ (Suppl. [fig_ref] Figure 1: Place field formation from visual scenes enabled estimating the animal's position [/fig_ref] -see "Methods"-H-H model simulation parameters 1) as the energy consumed to transmit a spike as previously described; C was normally distributed from N(188, 10) nJ, and σ j ~ N(0.03, 0.005) both reflecting the diversity of place cells' metabolic environment [bib_ref] An energy model of place cell network in three dimensional space, Wang [/bib_ref]. The learning rule was applied not only in a batch manner 63 but also on an energy level as follows:
[formula] dW j (t)/dt = µ(X ′ (t)/L − W j (t)) [/formula]
with the responding set J = {j|P f j (t) > P f thr } . The learning rate was defined as µ and P f thr was the minimum firing power of neurons that are activated. Consequently, every place cell responding to the current location with a firing power above the threshold (together with the feature matching criteria explained beneath) did modify the weights from sensors. Spatial location. As indicated in 62 , the spatial location X(t) can be seen as a function of the firing power of place cell j . Then the place field of neuron j in our model was defined as the set of all the locations X(t) with firing power larger than P f thr . The field centers were determined from the positions within the corresponding place fields. The center of the place field associated with the j-th place cell during this dynamical process is
[formula] C J = +∞ 0 P f j (t)X ′ (t)dt/ +∞ 0 P f j (t)dt . [/formula]
The location of the animal was estimated by the weighted average of place field centers according to the response set as the complement of Loc(t) = J P f j (t)C j / J P f j (t) with J = j|P f j (t) > P f thr , where the place field was defined as C j and P f j (t) was the activity power of the jth neuron at moment t. www.nature.com/scientificreports/ The feature matching model. Place cells matched features from the sensed environment as in [bib_ref] A visually driven hippocampal place cell model, Fuhs [/bib_ref]. Each place cell was tuned randomly to detect two visual blobs describing four features: area, orientation, solidity and centroid coordinates. Each place cell was therefore represented by two sets of feature values, M1 and M2. This pair of feature values modeled the neuron's preference that would maximize the place activation. The similarity of a place cell's ideal feature value, µ f , to the corresponding feature value of a blob x f was measured using a Gaussian function:
[formula] C m dV m dt = g l (E l − V m ) + g Na m 3 h(E na − V m ) + g k n 4 (E k − V m ) + I dn dt = α n (1 − n) − β n n dm dt = α m (1 − m) − β m m dh dt = α h (1 − h) − β h h α n = 0.01(10 + V m − V r ) exp[((10 + V m − V r )/10) − 1] β n = 0.125exp (V m − V r ) 80 α m = 0.1(25 + V m − V r ) exp[((25 + V m − V r )/10) − 1] β m = 4exp (V m − V r ) 18 α h = 0.07exp((V m − V r )/20)β h = 1 exp[((30 + V m − V r )/10) + 1] β mG(x f , µ f , σ f ) = e −((x f −µ f )/σ f ) 2 [/formula]
. When x f and µ f were close, G tended to 1; as their difference increased, the function approached zero. The magnitude of σ f determined the sensitivity of the function to the difference between x f and σ f , which was set to 0.5 in this work. The response of a place cell to a particular blob B i was determined by the product of feature similarities over the set of features F : A Bi = f ∈F G(x f , µ f , σ f ) . All features of B i did have to be like the place cell's maximal response value for a high activation. In that case, the place cell responded to conjunctions of features. Note that there was no predetermination of which blobs were to be associated with which place cell. Those visual blobs B i and B j that generated the strongest response to the two sets of place cell feature values M1 and M2 determined the activation of the place cell: A = max i∈B A Bi .max j∈B,j� =i A Bj . Note that from one simulation to another, the same modeled place cell would show different place fields, because the neuron's preference for blob detection and other parameters were randomized at the beginning of each simulation.
Place field center adaptation. The modeled place cells were initially tuned to random feature values, which were unlikely to correspond to any environmental blob. Therefore, place fields were initially quite broad and weak, conveying little spatial information. To increase the spatial information content, a competitive learning algorithm was used to adapt each place cell's feature value µ f to the corresponding blob feature value x f . For each iteration t in the competitive algorithm, the place cell whose activation was the highest at a particular position was determined to be the 'winner' of that position. For each position, the pair of visual blobs that maximized the activation of the winning place cell were used to train the feature values of the winning place cell. Each feature value of the winning place cell was adapted to the corresponding blob feature value according to the following equation: µ t+1
[formula] f = αx f + (1 + α)µ t f . [/formula]
The coefficient α controled the rate at which feature values were adapted. For our simulations, the α value of 0.5 was used.
## The grid cell model. continuous attractor network (can).
Theoretical works [bib_ref] Size matters: How scaling affects the interaction between grid and border cells, Santos-Pata [/bib_ref] [bib_ref] A model of grid cells based on a twisted torus topology, Guanella [/bib_ref] presented evidence that grid cells can be modeled through toroidal representations and as a symmetric locally connected neural network 71 which has been recently confirmed experimentally 98 . The model of grid cells that we implemented was based on a twisted torus topology formed by neurons and weighted connections between them. Excitatory and inhibitory connections were defined between nodes to obtain local cooperation and distal inhibition [bib_ref] A model of grid cells based on a path integration mechanism, Guanella [/bib_ref] [bib_ref] Size matters: How scaling affects the interaction between grid and border cells, Santos-Pata [/bib_ref] [bib_ref] Prediction of the position of an animal based on populations of grid..., Guanella [/bib_ref]. This caused neurons that were closer together to become mutually excited, increasing their activity, while the activity of distant neurons got inhibited. The model dynamics caused an "activity packet" (bump) that was modified by the input to the neurons. The shape of the bump was modified by the neurons' weights, while the localization of the activity in the network was guided by external inputs (i.e., the velocity information of the animal's movements). The model was implemented in Matlab R2016b based on N = 20 neurons organized in a rectangular matrix, such that N = N x xN y (with N = 400; N x = 20; N y = 20 ) represented the repetitive structure of grid subfields. The synapses connect neuron i with j , where i , j ∈ {1, 2, . . . , N} , were defined by Gaussian weighted function. The connection between neuron i and j , are w ij = I ai exp −�c i − c j � 2 /σ 2 − T , where I ai was defined as the activation intensity parameter, T was the shift parameter affecting the inhibitory-excitatory balance, σ regulated the size of the Gaussian, and c l defined the position of the neuron as represented by c l = (c l x − c l y ) , with c l x = (l x − 0.5)/N x and c l y = √ 3/2(l y − 0.5)/N x , where l x and l y were the column and row of neuron l . The dynamics of the grid cells utilized in this work were governed by A i (t + 1) = f (B j (t + 1) + τ (B j (t + 1)/B j (t + 1) − B j (t + 1))) , where A i represented the activity level of the neuron i , with i ∈ {1, 2, . . . , N} , τ = 0.95 was defined as the parameter that determined the stabilization strength, and f as a simple rectification, non-linear function such that f (x) = x for x > 0 and was 0 otherwise. The activity of the neurons depended on the transfer function B j (t
[formula] + 1) = N i=1 A i (t)w ij (t) , [/formula]
where w ij (t) represented the weight connecting neuron i and j . The network was initialized with a random activity using a uniform distribution between 0 and 1/ √ N . The position estimation depended on the imputed planar velocity that subsequently was integrated to obtain a position estimation. The network took as input a modulated version of the planar velocity defined as v(t) = [v x (t), v y (t)] T where v x (t) and v y (t) were the velocities in the x and y direction at time t . The input to the network was then modulated by a gain parameter and a rotation matrix such that v R (t) = αR β v(t) , where R β defined a rotation matrix that depended on the bias angle β = 0 and the gain α ∈ +realnumbers (with α = 1 ). Here the vector v did not carry any information about the current position of the animal. The grid remained stable when no velocity input was introduced (i.e., v = [0; 0] ). However, when the animal moved, this affected the activity of the network shifting the bumps. The effect of the velocity in the network was introduced in the synaptic weight equation as www.nature.com/scientificreports/ Position estimation. The activity of the grid cells (i.e., the bump activity) was used to obtain an updated estimation of the animal's position. Considering at t 0 the estimated position pos(t 0 ) = [0; 0] , it was updated as pos(t + 1) = pos(t) + γ �A in every time step, where A defined the change in activity bump (i.e., the magnitude and direction of the bump's movement). This update was individually made by the network of grid cells in each grid module; γ was a parameter that regulated the grid spacing. Varying the spacing and the orientation of the grids by changing the gain and bias parameters, we tested whether there was a difference in module place estimates (see the main text). To determine whether a correction of the estimated position was made in every time step of the simulation, an estimated error was computed as follows: E error = std(pos x ) 2 + std(pos y ) 2 , where pos x and pos y represented vectors of the estimated animal's position along grid modules in both spatial coordinates, respectively. If the estimated error E error was higher than a predefined threshold (i.e., 1.5 a.u.), the animal's current position represented an 'uncertain' place indicating that the grid modules did not reach enough precision to estimate the actual animal's position. This situation enabled a place cell's field to emerge on that position only if a place cell was active and the grid modules provided enough input to place cells as previously explained. The correction of the position estimate pos(t + 1) was made based on the nearest place field center from the animal's current position. It was taken 2.5 a.u. as the minimum distance to select a near a place field as a reference point to correct the position estimate. The rationale for using such a correction is that it was previously reported 'resetting' at grid network level. This understanding was discussed in grid cells' activity to the agent's correct location [bib_ref] Hebbian plasticity realigns grid cell activity with external sensory cues in continuous..., Mulas [/bib_ref]. It was also discussed that the dynamics between place and grid cells could have a role through feedback projections from the hippocampus to grid cells, anchoring grid cells' activity to specific spatial locations, thereby resetting the accumulated error to the ground truth [bib_ref] A model of grid cells based on a twisted torus topology, Guanella [/bib_ref]. There is experimental evidence supporting this view [bib_ref] Grid cells require excitatory drive from the hippocampus, Bonnevie [/bib_ref]. Init was observed that place cells' firing could shift toward goal locations, and while the grid cells maintain information overnight, the place cells can reset.
[formula] w ij = I ai exp −�c i − c j + v R (t)� 2 /σ 2 − T , [/formula]
Grid-place cell relationship. The connection from grid to place cells was implemented through the diversity of spacings from the dorsal-to-ventral grid modules 39 (Suppl. . This connectivity was implemented based on a symmetrical weight matrix of a Gaussian curve function for each module. Each Gaussian was distributed evenly in the [0;10] range. The maximum grid activity per module was computed in each time step and multiplied by the symmetrical weight matrix to represent the diversity of information provided by the spacings from the dorsal-to-ventral grid modules. After multiplication, the mean value across neurons of the resulting matrix was computed to determine the place cells to be evaluated. The grid input enabled the emergence of place fields for firing place cells on the animal's current position with a neural activity above 1.1 a.u. (Supp. . Place fields were not necessarily defined in the places where the input from grid cells was high (i.e., above 1.1 a.u.) but were determined by the place network activities. For active place cells, a plasticity rule was computed as
[formula] W G_P (t + 1) = W G_P (t) + µ * (X ′ (t) − W G_P (t)) [/formula]
, where µ = 0.001 represented the learning rate, and X ′ was the error of sensory perception as previously described. Regarding the inputs from place cells to grid cells, this information was possible in estimating the animal's current position (i.e., path integration error reduction as indicated in [fig_ref] Figure 4: Place [/fig_ref]. Because there is no evidence suggesting that the MEC (or even the hippocampal formation) provides movement-related signals to motor areas, we did not control the motor activities of our simulated animal. Instead, we only used the neural activities to estimate the animal's position. The estimate was made from the network's stable activity bump after computing the path integration from the network that was updated based on the animal's movements.
Gridness measure. To test the hexagonality of grid cells, we used a hexagonal gridness score of the spatial fields based on 69 and 11 . The score was calculated from a cropped ring of their autocorrelogram considering the six maxima points closest to the center. The ring was rotated 30 degrees per rotation starting from 30 to 150 degrees. For the six rotated angle, the Pearson correlation C was obtained considering the original un-rotated map. Combining as follows the correlations for these specific rotation angles, the final gridness score was: gridness = 1/2 * (C 60 + C 120 ) − 1/3 * (C 30 + C 90 + C 150 ) . Following descriptions in, we also computed a squareness gridness score to evaluate how square-like grid autocorrelograms were spatially defined. This measure was computed by rotating the autocorrelogram 45 degrees for every iteration to reach angles of 45, 90 ,135 degrees. This score was calculated as: square gridness = C 90 − 1/2 * (C 45 + C 135 ).
Definition of grid modules. Previous works [bib_ref] Microstructure of a spatial map in the entorhinal cortex, Hafting [/bib_ref] have shown that regular triangular grids can be observed from dorsal to ventral MEC, where the spacing of the grid increases isometrically along the dorsoventral axis. Firing fields of grid cells located at dorsal-most MEC represent small and highly tuned locations, while ventral-most cells display broader and less location-specific firing activity. Based on a previously defined grid cell model [bib_ref] A model of grid cells based on a path integration mechanism, Guanella [/bib_ref] , we used the orientation, the phase of the grid, the spacing (minimal inter-subfields distance), and the size of its subfields in our model to represent the topographical organization of MEC. These settings controlled the spacing and the orientation of the grids across modules in the model. We based the dynamics of a population of neighboring grid cells as observed in dorsal-to-ventral MEC, whose grids share the same orientation and spacing, but have different phases [bib_ref] A model of grid cells based on a path integration mechanism, Guanella [/bib_ref] www.nature.com/scientificreports/ spiking activity of a place cell), then the information gained (i.e., the coding of that place in the population of place cells) varied inversely with the size of the probability. When the probability of the event was minimal and the event occurred, then the information gained was high compared to a situation where the probability was very likely. In the extreme case, where the probability of the event occurring was 1, then no information was gained. In general, it is assumed that the raw information gained varied as 1/p i , but the best form of this function needs to be determined for other criteria. In addition, based on the expected value H , the overall information for n events was defined as: H = H(n) = − n i=1 p i log(p i ) , which is Shannon's information [bib_ref] A mathematical theory of communication, Shannon [/bib_ref] , and it is equivalent to the Boltzmann-Gibbs formula for entropy, where H ∈ [0; log(n)] . When H = 0 , then one event dominated (i.e., p k = 1, ∀p k = 0, i = k and H = log(n) , then p k = 1/n , ∀i ). Thus, the entropy was maximum when the probabilities were all equal and H = log(n) . If all the events happened at the same location x , the information gained was zero when a new event occurred. Once the population distribution of place fields was uniformly spread across many locations ( ∀x i ) , and an event occurred, this event gave complete information. There was a trade-off between the spread of the probability distribution across spatial and the number of events. In brief, the newer places were visited in the arena, the more complex and more extensive the internal map represented by place cells was characterized by more and more events. Furthermore, the increase of events did change the shape of the representation, which led to an increase or decrease in information. This trade-off between spatial density and the number of events can be represented by H as an index of complexity. In short, it was assumed that as the exploration of the environment extended, the complexity in terms of information increased. However, this criterion was purely based on the number of their place fields, and there may be strong ordering effects that discount this increase. Entropy calculations also presupposed the additive independence of events [bib_ref] Entropy, complexity, and spatial information, Batty [/bib_ref].
Entropy ratio. In the case of H , and to get a measure of relative information, a primary probability distribution acted as a prior since the information was relative to what occurred in which the event sequence is unknown. Thus, to characterize relative information given the prior information, it was defined the entropy ratio as r = H/H max = −� i p i log p i /log(n) . By normalizing H concerning the maximum value H max = log(n) it was possible to represent a uniform distribution.
Complexity ratio. R Represented the complexity ratio based on the complement of the entropy ratio R = 1 − r , which can be written as: (H max − H)/H max = I/H max , indicating the percentage of information that the system could achieve by adjusting itself to the most probable state (i.e., a uniform distribution). This measure represented the relative information as information difference I . Assuming q i = 1/n and i q i = 1 , information can be written following 66 as: I = H max − H = −� i p i log(p i /(1/n)) = −� i p i log(p i /q i ) , which clarify the role of the number of events or place fields. Spatial entropy. The complexity ratio tended to discount the effect as the system changed through more events because information formulas measured relative change [bib_ref] Entropy, complexity, and spatial information, Batty [/bib_ref]. Consequently, we can consider probability densities to discriminate the distribution density from the number of events. If the measured area increased (i.e., the animal explored more of the arena), it was expected that the probability over that area also increased, and then the density changed. Thus, by defining an approximation to the density over an area x i , there was computed the total area of the system X as i x i = X . The density can be written as p(x i ) = p i /�x i or p i = p(x i )�x i , and it was assumed that in the limit, the equation converged to the probability density p(x) as p(x) = p i /�x i . After few steps, the entropy formula H in probability density terms can be represented as the discrete form of entropy H = − i p i log p i /�x i − i p i log(�x i ) = S + Z , with H ∈ [0; log(n)] , where S is the approximation to the continuous entropy. The term Z is the approximation to the information associated with the sizes of the events. S is the formula that was called 'spatial entropy' . This H is composed of S and the 'information density' Z -the amount of information represented by the place cell space. In short, when we examined H , we did this for the numerical co-variation of its elements S and Z . The size of the place fields was kept fix across simulations in x = 20 a.u.
## Varying spatial information.
To examine changes in the number of events and their density, we did consider the distribution q i , which normalizes the distribution of the visited places to sum to unity [bib_ref] Entropy, complexity, and spatial information, Batty [/bib_ref] , and form the expected value of its logarithm H(p) . Compared to the entropy of the distribution itself H q , where these measures were defined as H(p) = − i p i log q i and H(q) = − i q i log q i H p , where H(p) represents the evenness of the density distribution, while H(q) provides the same but for the density. Note that this is a measure of difference indicating how the place field was represented. Based on 101 we can have: H(p/q) = H(q)-H p = i p i log q i − i q i log q i , which resembles the previously introduced equation for the generic information measure in its classical form. Consequently, we can expect that H(p/q) co-vary with the information difference I = i p i log p i − i p i log q i 66 . The difference in absolute terms became greater as the difference between the distributions of observations and the size of the environment increased. To this end, computing complexity on information from the animal's navigation in each time step, we analyzed the relationship between the number of place fields n and the density of population p i / x i .
## Data availability
Datasets supporting the findings of this paper are available upon request from the corresponding author.
## Code availability
Custom code used in this paper is available upon reasonable request from the corresponding author.
[fig] Figure 1: Place field formation from visual scenes enabled estimating the animal's position. (A) Schematic of the simulated animal. (B) Examples of the animal's perspective images showing local and distal cues. (A) and (B) were adapted from Google DeepMind Lab open source software (https:// www. deepm ind. com/ open-source/ deepm ind-lab) [/fig]
[fig] Figure 3: Grid patterns were maintained across the animal's trajectory but depended on the trajectory features. (A,B) Spikes across trajectories, spike density and rate map plots from dorsal (right side) to ventral (left side) grid network configurations. Note that for each column, the plots indicate the same neuron (5 different neurons in total separately for (A) and (B) groups of plots). Regular triangular tessellation can be observed for both trajectories. Yellow represents maximum activity in the spike density and rate map plots, and red indicates spike clusters of the same neurons as shown in the spike density plots along the trajectory. Over each rate map, the gridness score (G) and the squared gridness (GS) are shown. (C) Variance across modules over place estimates through time for both trajectories. (D) The Euclidean distance between estimates for each module from dorsal (module 1) to ventral (module 5) modules. The inset numbers represent mean ± sem for the associated data for all the panels. All panels were made using custom code in Matlab R2016b (https:// www. mathw orks. com/). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License (https:// creat iveco mmons. org/ licen ses/ by/4. 0/).Scientific Reports | (2022) 12:21443 | https://doi.org/10.1038/s41598-022-25863-2 [/fig]
[fig] Figure 4: Place . deepm ind. com/ open-source/ deepm ind-lab) [/fig]
[fig] Figure 5: A lower variance and estimated error across grid modules were observed with closer place fields to the animal trajectory. (A) A brief trajectory example and the estimated error from the grid cell network when a small (PF = 20) or high (PF = 200) number of place field centers were spread randomly across the arena. Blue lines represent actual trajectories, and red ones indicate trajectory estimates. (B,C) An example of Euclidean distance error measure shows that (B) when place field input was provided and place fields were set to PF = 20, the error was higher than (C) when the number of place fields was set to PF = 200. Module 1 refers to dorsal and module 5 to ventral. (D,E) The measure of variance in fewer place fields (D) vs a larger number of place fields (E), indicated a higher variance for the former case. (F,G) Example of the estimated error when (F) PF = 20 vs (G) when PF = 200 across modules. The inset equations represent mean ± sem for the associated data for all the panels. All panels were made using custom code in Matlab R2016b (https:// www. mathw orks. com/). This work is licensed under a Creative Commons Attribution 4.0 (CC BY 4.0) International License (https:// creat iveco mmons. org/ licen ses/ by/4. 0/). [/fig]
[fig] .: Statistical measures Entropy The proportion of place cells showing activity at a specific space and time was computed as p i = P i /P, with p i ∈ [0; 1] the occurrence of an event i that varied concerning to size x i , and n i=1 p i = 1 with n (the number of place cells) indicating the range of probable events This probability was based on population densities of place cells which represented the chance to observe spiking activity by a place cell at the animal's current position (ie, probability of locating a spike in one place cell) If an event occurred (ie, the Scientific Reports | (2022) 12:21443 | https://doiorg/101038/s41598-022-25863-2 [/fig]
[table] 63: That error X ′ (t) conveyed to place cells resembling how boundary vector neurons [/table]
|
10.3390/ijms21072313
|
CCBY
|
7178064
|
32230799
|
s2orc_pubmed_articles
|
miR-216a Acts as a Negative Regulator of Breast Cancer by Modulating Stemness Properties and Tumor Microenvironment
Breast cancer is the most frequent malignancy in females in terms of both incidence and mortality. Underlying the high mortality rate is the presence of cancer stem cells, which divide indefinitely and are resistant to conventional chemotherapies, so causing tumor relapse. In the present study, we identify miR-216a-5p as a downregulated microRNA in breast cancer stem cells vs. the differentiated counterpart. We demonstrate that overexpression of miR-216a-5p impairs stemness markers, mammosphere formation, ALDH activity, and the level of Toll-like receptor 4 (TLR4), which plays a significant role in breast cancer progression and metastasis by leading to the release of pro-inflammatory molecules, such as interleukin 6 (IL-6). Indeed, miR-216a regulates the crosstalk between cancer cells and the cells of the microenvironment, in particular cancer-associated fibroblasts (CAFs), through regulation of the TLR4/IL6 pathway. Thus, miR-216a has an important role in the regulation of stem phenotype, decreasing stem-like properties and affecting the cross-talk between cancer cells and the tumor microenvironment.the progression of cancer[2,3]. Thus, a main challenge is to develop therapies that selectively target the CSC population.MicroRNAs (miRNAs) are small (20-23 nucleotide), endogenous, single-stranded RNA molecules involved in the post-transcriptional regulation of gene expression. Several miRNAs regulate genes involved in mechanisms such as proliferation, apoptosis, and stemness maintenance. Mutation in the biogenesis of these miRNAs can contribute to cancer development. Several miRNAs have been found to be involved in the regulation of the stemness phenotype[4]. For instance, let-7 is one of the main regulators of self-renewal in breast cancer cells [5]; miR-127 and miR-128 have been shown to be important in the modulation of stemness in glioma, since their ablation enhances stem cell renewal and proliferation[6]; and, miR-24 was proven to be a powerful positive regulator of breast CSC features during hypoxic conditions[7]. Moreover, miR-216a-5p-encoded on chromosome 2 on the reverse strand and highly conserved across species-has been shown to act as a tumor suppressor in gastric cancer, breast cancer, hepatocellular carcinoma, small cell lung cancer and renal cancer[8][9][10][11]. Little is known about the miR-216a regulation in cancer, but a recent manuscript by Tao W and colleagues reported that a long noncoding RNA named DANCR is able to target miR-216a-5p in breast cancer, regulating the expression of stemness markers such as Nanog, SOX2 and OCT4[12].The tumor microenvironment is largely involved in the regulation of neoplastic processes; it includes cells belonging to innate immunity, such as macrophages, neutrophils, and mast cells, and to adaptive immunity, such as T and B lymphocytes as well as stromal cells (fibroblasts, endothelial cells and mesenchymal cells)[13][14][15]. All these components of the tumor microenvironment interact with each other and cancer cells by either direct contact or through the release of cytokines and chemokines, which can act as paracrine or autocrine effectors[16][17][18]. The level of activation of these different cell types and the relative expression of the various mediators are able to tilt the balance in favor or against cancer progression[19,20].Through modulation of gene expression related to the inflammation, miRNAs influence cancer-related inflammation, enhancing cancer tumorigenicity and aggressiveness. Cancer-associated fibroblasts (CAFs) are the main players in tumor stroma[21]. These cells release inflammatory cytokines, leading to the activation of pathways enhancing proliferation and stemness maintenance of cancer cells[22]. On the other hand, cancer cells release pro-inflammatory factors able to educate normal fibroblasts (NFs) into CAFs, generating positive feedback loops between cancer cells and CAFs[23]. A link between the microenvironment and cancer cells are the Toll-like receptors (TLRs), which have recently generated great interest in cancer research: these receptors are involved in the defense against microbial infection, but they can also support tumor cell growth in vitro and in vivo[24]. TLRs in general, and TRL4 in particular, play a significant role in breast cancer progression and metastasis by leading to the release of pro-inflammatory molecules[25]. Here we show that miR-216a regulates the stemness state by acting on TLR4. The up-regulation of this gene is implicated in stemness state regulation, controlling stemness pathways and interactions in the microenvironment.
# Introduction
Breast cancer is the leading cause of death among women: In fact, it represents 15% of all cancer deaths in females [bib_ref] Global cancer statistics, Torre [/bib_ref]. Normal adult stem cells are tissue-specific cells with particular features, such as the abilities to self-renew and to differentiate into all cell types of the tissue of origin. These cells undergo an unlimited number of cell divisions, and through symmetrical or asymmetrical division, they produce daughter cells that also have the ability to self-renew or differentiate, respectively. In a similar manner, cancer stem cells (CSCs)-which are resistant to conventional treatment and are implicated in tumor relapse and to the metastatic process-play a main role in the development and
# Results
## Mir-216a negatively regulates the stemness features and adlh activity of breast cancer stem cells (bcscs)
To evaluate the role of miRNAs in breast cancer stemness maintenance, we previously performed a microarray analysis comparing miRNAs expression levels in differentiated vs. stem cells obtained from three patients' specimens [bib_ref] MiR-221 promotes stemness of breast cancer cells by targeting DNMT3b, Roscigno [/bib_ref]. BCSCs were obtained by biopsy digestion and characterized by their ability to produce tumors when injected into immunocompromised mice and by real time PCR for the expression of the stem markers Nanog and Oct4 [bib_ref] MiR-221 promotes stemness of breast cancer cells by targeting DNMT3b, Roscigno [/bib_ref] [bib_ref] Erythropoietin Activates Cell Survival Pathways in Breast Cancer Stem-like Cells to Protect..., Todaro [/bib_ref]. The microarray results revealed there was a significant down-regulation of miR-216a [bib_ref] Erythropoietin Activates Cell Survival Pathways in Breast Cancer Stem-like Cells to Protect..., Todaro [/bib_ref]. We confirmed this finding by RT-PCR [fig_ref] Figure 1: MiR-216a-5p expression in breast cancer stem cells [/fig_ref]. In order to obtain stem cell populations, we adopted suspension cultures of T47D and MDA-MB-231 breast cancer cells. In this condition, stem cells are enriched and grow as mammospheres. Through Western blot analysis and real-time PCR, we verified that the expression of stem cell markers such is down-regulated in enriched stem cell cultures. Data are mean values ± SD of three independent experiments. Significance was calculated using Student's t-test. *, p < 0.05, **, p < 0.01, ***, p < 0.001, ****, p < 0.0001. It has been reported that both normal and cancer stem cells express high levels of ALDH and that ALDH is a powerful predictor of poor clinical outcome as well as being a marker of stem/progenitor cells. We compared ALDH activity levels in MDA-MB-231 and T47D stem and differentiated cells. We found that mammospheres had higher levels of ALDH activity [fig_ref] Figure 1: MiR-216a-5p expression in breast cancer stem cells [/fig_ref]. Interestingly, miR-216a was down-regulated in the stem cell cultures in a similar manner to primary breast cancer cells [fig_ref] Figure 1: MiR-216a-5p expression in breast cancer stem cells [/fig_ref].
In an effort to elucidate the role of miR-216a on stemness properties, we modulated miR-216 levels in stem cells and then performed a limiting dilution assay. The frequency of MDA-MB-231 and T47D stem cells in a mixed population decreased upon miR-216a overexpression [fig_ref] Figure 2: MiR-216a-5p regulates BCSC phenotype [/fig_ref]. Interestingly, miR-216a overexpression induced a decrease in ALDH activity in both MDA-MB-231 and T47D stem cells [fig_ref] Figure 1: MiR-216a-5p expression in breast cancer stem cells [/fig_ref]. Additionally, overexpression of miR-216a decreased the expression of stemness markers (Nanog and Slug) and Epithelial Mesenchymal Transition (EMT) markers (Vimentin) in stem cells, as analyzed by Western blotting [fig_ref] Figure 2: MiR-216a-5p regulates BCSC phenotype [/fig_ref]. Conversely, downregulation of miR-216a in differentiated cells enhanced the capability of differentiated breast cells to form mammospheres, while overexpression decreased the propensity to become stem cells, as assessed by a mammosphere assay [fig_ref] Figure 3: MiR-216a-5p increases stemness phenotype in differentiated breast cancer cells [/fig_ref] ,B, Supplementary [fig_ref] Figure 1: MiR-216a-5p expression in breast cancer stem cells [/fig_ref]. Moreover, down-regulation of miR-216a in differentiated T47D and MDA-MB-231 cells induced upregulation of ALDH activity [fig_ref] Figure 3: MiR-216a-5p increases stemness phenotype in differentiated breast cancer cells [/fig_ref] and stem and EMT markers [fig_ref] Figure 3: MiR-216a-5p increases stemness phenotype in differentiated breast cancer cells [/fig_ref]. Moreover, to further confirm the role of miR-216a on stemness properties, we took advantage of an organoid culture system as a model closer to an in vivo experiment. For this, we transfected anti-miR-216 in breast cancer patient-derived organoids. Interestingly, we observed an increased ALDH activity [fig_ref] Figure 3: MiR-216a-5p increases stemness phenotype in differentiated breast cancer cells [/fig_ref]. Therefore, miR-216a negatively regulates stemness properties.
## Mirna-216a targets tlr4
We then investigated hypothetical miR-216a targets involved in the stemness phenotype. We used Target Scan Human as a bioinformatics tool for miRNA target prediction, focusing on a putative target, TLR4, which plays a significant role in breast cancer progression and metastasis [bib_ref] Links between Toll-like receptor 4 and breast cancer, Ahmed [/bib_ref] [bib_ref] TLR4 Promotes Breast Cancer Metastasis via Akt/GSK3beta/beta-Catenin Pathway upon LPS Stimulation, Li [/bib_ref] and that has also been reported to be a direct target of miR-216a-5p in renal carcinoma [bib_ref] MiR-216a exerts tumor-suppressing functions in renal cell carcinoma by targeting TLR4, Wang [/bib_ref]. We found that TLR4 expression was higher in MDA-MB-231 and T47D stem cells compared to differentiated cells at both protein and mRNA levels [fig_ref] Figure 4: Toll-like receptor 4 [/fig_ref]. Interestingly, overexpression of miR-216a in stem cells decreased TLR4 expression [fig_ref] Figure 4: Toll-like receptor 4 [/fig_ref]. Conversely, expression of TLR4 was enhanced by downregulating miR-216a in differentiated MDA-MB-231 and T47D cells [fig_ref] Figure 4: Toll-like receptor 4 [/fig_ref]. transfection of miR-216, as assessed by Western blotting, in MDA-MB-231 and T47D stem cells. Data are mean values ± SD of three independent experiments. Significance was calculated using Student's t-test. *, p ≤ 0.05; **, p < 0.01, ***, p < 0.001, ****, p < 0.0001.
## Role of mir-216a in inflammation and tumor microenvironment crosstalk
Inflammation is a critical component of tumor progression [bib_ref] Inflammation and cancer, Coussens [/bib_ref]. Cancer cells release inflammatory cytokines and activate pathways that stimulate further cytokine production in CSCs. This generates a positive feedback loop driving CSC self-renewal and the activation of cells within the tumor microenvironment into tumor promoter cells. In particular, the interaction between cancer cells and fibroblasts results in the activation of the latter into CAFs [bib_ref] Stromal cells in tumor microenvironment and breast cancer, Mao [/bib_ref]. Recent studies have demonstrated that TLR4 expression and signaling induce an upregulation of IL-6 in different context such as colon cancer, Fragile X, and human bladder epithelial cells [bib_ref] A novel TLR4-mediated signaling pathway leading to IL-6 responses in human bladder..., Song [/bib_ref] [bib_ref] Regulation of TLR4-induced IL-6 response in bladder cancer cells by opposing actions..., Qian [/bib_ref] [bib_ref] Regulation of IL-6 Secretion by Astrocytes via TLR4 in the Fragile X..., Krasovska [/bib_ref]. For this reason, we investigated whether miR-216a is involved in the crosstalk between cancer cells and the tumor microenvironment through the regulation of inflammatory cytokines IL-6. Interestingly, we found that IL-6 protein and mRNA levels were higher in MDA-MB-231 and T47D stem cells compared to their differentiated counterparts, due to higher levels of TLR4 [fig_ref] Figure 4: Toll-like receptor 4 [/fig_ref] , and even more interestingly, IL-6 expression levels changed upon TLR4 modulation induced by miR-216a [fig_ref] Figure 4: Toll-like receptor 4 [/fig_ref] , confirming the crosstalk between TLR4, miR-216a, and IL-6.
## Role of mir-216a in the tumor microenvironment
We next evaluated if downregulation of miR-216a in BCSCs induced fibroblast activation into CAFs. We treated CAFs with conditioned media collected from MDA-MB-231 and T47D stem cells transfected with miR-216a. After 48 h, cells were harvested and the expression of CAF activation markers (FAP and α-SMA) evaluated by Western blotting. We found that miR-216a-conditioned media reduced CAF activation compared to the control . Conversely, the conditioned media from differentiated MDA-MB-231 and T47D cells upon downregulation of miR-216a induced education of NFs into CAFs . In addition, CAFs treated with conditioned media collected from MDA-MB-231 and T47D stem cells transfected with miR-216a decrease their migration , while conditioned media from differentiated MDA-MB-231 and T47D transfected with anti-miR-216a increased the migratory ability of normal fibroblasts . Thus, in BCSCs, miR-216a decreases release of soluble factors that induce CAF activation, regulating the stem cell-microenvironment network.
# Discussion
Cancer stem cells are supposed to be at the root of a tumor since they are able to self-renew indefinitely and generate mature cells [bib_ref] Cancer stem cells in solid tumours: Accumulating evidence and unresolved questions, Visvader [/bib_ref]. Therefore, even if conventional therapy has improved life expectation, many patients still go through relapses because of the presence of CSCs in the tumor site [bib_ref] Stem cells in mammary development and carcinogenesis: Implications for prevention and treatment, Dontu [/bib_ref]. Several studies have demonstrated that the expression of microRNAs is altered in the tumor setting [bib_ref] overexpession in human glioblastoma increases invasiveness by targeting the protein phosphate PTPmu, Quintavalle [/bib_ref] , including in cancer stem cells [bib_ref] Downregulation of microRNA-196a inhibits stem cell self-renewal ability and stemness in non-small-cell..., Liu [/bib_ref] , by modulating gene expression and, hence, stemness properties. Consequently, it appears evident that the study of microRNAs involved in the regulation of stemness features could provide valuable data for understanding the mechanisms involved in cancer progression and relapse. Here, we demonstrate that miR-216 acts as a stemness repressor in breast cancer stem cells, acting at different nodes. By microarray analysis performed on tumor patient specimens [bib_ref] MiR-221 promotes stemness of breast cancer cells by targeting DNMT3b, Roscigno [/bib_ref] , we found that miR-216a-5p was downregulated in BCSCs compared to differentiated cells. Then, to elucidate the role of miR-216a in the regulation of the stem properties, we modulated the expression of the microRNA, upregulating it in breast cancer stem cells and downregulating it in differentiated cells. We verified that miR-216a-5p negatively regulates the number of stem cells, EMT, stem markers and ALDH activity. We then looked for possible targets, focusing our attention on TLR4, which belongs to a family of transmembrane receptors involved in the innate immune system [bib_ref] Proinflammatory cytokines (IL-6, IL-8), cytokine inhibitors (IL-6sR, sTNFRII) and anti-inflammatory cytokines (IL-10,..., Sikora [/bib_ref]. TLRs protect the host against viral and bacterial infections and are able to trigger proinflammatory signaling cascades linking innate immunity to inflammation. TLR4 activation triggers the proinflammatory response, leading to the production and secretion of cytokines, including interleukin 6 (IL-6) [bib_ref] TLR4-induced NF-kappaB and MAPK signaling regulate the IL-6 mRNA stabilizing protein Arid5a, Nyati [/bib_ref] [bib_ref] Role of interleukin-6 in toll-like receptor 4 and 2 expressions induced by..., Inoue [/bib_ref] and IL-8 [bib_ref] LPS induces IL-8 expression through TLR4, MyD88, NF-kappaB and MAPK pathways in..., He [/bib_ref]. Several studies have been conducted to elucidate the links between TLR4 and breast cancer, and it has been shown that TLR4 is highly expressed in breast cancer cells and plays a crucial role in invasiveness, migration and angiogenic potential of cancer at primary or metastatic sites [bib_ref] Links between Toll-like receptor 4 and breast cancer, Ahmed [/bib_ref] [bib_ref] Proinflammatory cytokines (IL-6, IL-8), cytokine inhibitors (IL-6sR, sTNFRII) and anti-inflammatory cytokines (IL-10,..., Sikora [/bib_ref]. In addition, TLR4 knockdown reduces IL-6 and IL-8 secretion, cell survival and proliferation [bib_ref] Reduced expression of Toll-like receptor 4 inhibits human breast cancer cells proliferation..., Yang [/bib_ref]. Through Western blot analysis, we found that this protein is regulated by miR-216, inducing us to consider it as possible axis involved in stemness features and microenvironmental cross talk. Interestingly, it has been already shown that miR-216a targets TLR4 in renal cancer [bib_ref] MiR-216a exerts tumor-suppressing functions in renal cell carcinoma by targeting TLR4, Wang [/bib_ref] , corroborating our hypothesis. Since this protein is involved in inflammation pathways, we speculate that miR-216a has a role in the regulation of the inflammation network. Cancer cells interact with cells present within the tumor microenvironment, such as fibroblasts, immune and mesenchymal cells. It is known that breast CSCs modulate stromal cell activity (for example, transforming fibroblasts into cancer-associated fibroblasts) through cytokine secretion, leading them to release other cytokines that promote tumor malignancy. Moreover, in bladder cancer it has been reported that TLR4 signaling upregulates IL-6 in a dose-and time-dependent manner [bib_ref] Regulation of TLR4-induced IL-6 response in bladder cancer cells by opposing actions..., Qian [/bib_ref]. Conversely, TLR4 knockdown reduces IL-6 and IL-8 secretion, cell survival and proliferation [bib_ref] Reduced expression of Toll-like receptor 4 inhibits human breast cancer cells proliferation..., Yang [/bib_ref]. We demonstrate here that miR-216a-5p modulates TLR4 to regulate the release of proinflammatory cytokines, such as IL-6, and that CAFs treated with miR-216-conditioned media express lower levels of CAF-activation markers compared to controls.
In the present study, we explored the role of miR-216a in breast cancer, demonstrating that miR-216a acts as a negative regulator of stemness in BCSCs, influencing the expression of stem markers, their number and the interaction between BCSCs and cells within the tumor microenvironment. miR-216a regulates pathways involved in the control of stemness properties and cytokine production. Our finding shows for the first time that miR-216 modulates different aspects of cancer stem cells, regulating stemness pathways and microenvironment education. Our data demonstrate that miR-216 has pleiotropic effects and that it is downregulated in CSCs, exacerbating their malignant phenotype.
# Materials and methods
## Cells and mammosphere culture
Breast tumor differentiated cells from two patients (#3,#4) and BTSCs (Breast tumor stem cells) were obtained as previously described [bib_ref] MiR-221 promotes stemness of breast cancer cells by targeting DNMT3b, Roscigno [/bib_ref] [bib_ref] Isolation and in vitro propagation of tumorigenic breast cancer cells with stem/progenitor..., Ponti [/bib_ref] , and were used for the miR array. T47D (ER-positive) and MDA-MB-MB-231 (TNBC) differentiated cells were grown in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM L-glutamine and 100 U/mL penicillin/streptomycin [bib_ref] Cancer Cell Lines Are Useful Model Systems for Medical Research, Mirabelli [/bib_ref]. For mammosphere cultures, single cells were plated at a density of 1,000 cells/mL for T47D and 50,000/mL for MDA-MB-231. T47D and patients' cells were grown in serum-free DMEM-F12 (Sigma, Milan, Italy), supplemented with B27 (Life technologies, Milan, Italy), 10 ng/mL EGF (Sigma), 20 ng/mL FGF (BD Biosciences, Milan, Italy) and 1X antibiotic-antimycotics (Life technologies). MDA-MB-231 stem cells were grown in Mammary Epithelial Cell Growth Medium (Lonza, Milan, Italy) supplemented with BPE, hEGF, insulin, hydrocortisone, GA-1000 (Lonza), B27 (Life technologies), and 20 ng/mL FGF (BD Biosciences, Milan, Italy). After 4-7 days, mammospheres, appearing as spheres of floating viable cells, were collected by gentle centrifugation (800 rpm) and dissociated with 0.25% trypsin for 5 min.
## Cell transfection
For miRs transient transfection, cells at 50% confluency were transfected using Oligofectamine (Invitrogen, Life technologies) for differentiated cells, and Lipofectamine 3000 for stem cells (Invitrogen, Life technologies), with either 100 nM of pre-miR-216a-5p or scramble, or 200 nM of anti-miR-216a-5p and anti-scramble (Ambion, Life technologies).
## Breast primary cell culture
Breast carcinoma specimens were collected at the Surgical Unit of Clinica Mediterranea SPA (Naples, Italy). Prior consent was given by the donor before the collection, acquisition or use of human tissue. To obtain the cells, samples were mechanically and enzymatically disaggregated, and the lysates grown in DMEM-F12 medium supplemented with 10% FBS and 1% penicillin-streptomycin (Sigma-Aldrich).
## Organoids cultures
Tumor organoids were isolated using Hans Clevers protocol as reported before [bib_ref] Highly Homogeneous Biotinylated Carbon Nanodots: Red-Emitting Nanoheaters as Theranostic Agents toward Precision..., Scialabba [/bib_ref]. Organoids pellet was suspended in cold Cultex growth factor reduced BME type 2 (Trevigen, 3533-010-02) and 40 mL drops were allowed to solidify in a 24 multi-well suspension plates (Greiner, M9312) for 20 min at 37 - C. At the end, 400 µl of organoid medium were added for each well to cover the drop. Medium was replaced every 3-4 days. Organoids transfection was performed with Lipofectamine 3000 (ThermoFisher Scientific) according to manufacture instructions.
## Mammosphere formation assay
T47D and MDA-MB-231 cells were transfected with anti-miR-216a for 24 h. Mammospheres were generated according to the protocol published elsewhere [bib_ref] Stem cells in mammary development and carcinogenesis: Implications for prevention and treatment, Dontu [/bib_ref] [bib_ref] Comparison of mammosphere formation from breast cancer cell lines and primary breast..., Wang [/bib_ref] [bib_ref] Propofol Reduced Mammosphere Formation of Breast Cancer Stem Cells via PD-L1/Nanog In..., Zhang [/bib_ref]. Briefly, T47D and MDA-MB-231 differentiated cells were dissociated with 0.25% trypsin for 5 min. Then, cells were washed in serum-free DMEM-F12, centrifuged and the pellet was resuspended in stem media. 1000 or 5000 differentiated cells were counted and suspended in 2 mL of stem medium and finally plated into 6-well plates previously treated with polyHEMA to stop cell attachment. After 4-7 days, mammospheres, appeared as spheres of floating viable cells for T47D (ER-positive) cell line, while the TNBC cell line MDA-MB-231 appeared as bead-like structure. After 7/10 days for T47D and 10/15 days for MDA-MB-231, mammospheres were counted. Spheres of a diameter >50 µm for each well were counted under microscope. Representative images from mammospheres assay were used for each experiment [bib_ref] Comparison of mammosphere formation from breast cancer cell lines and primary breast..., Wang [/bib_ref].
## Limiting dilution assay
After a 24-h transfection, 1, 5, or 10 cells were seeded per well of 96-well plates in stem cell medium. For MDA-MB-231 stem cells, we used polyHEMA-pretreated plates. Two weeks after seeding, the number of wells containing spheroids for each cell was counted. The analysis was performed using software available at http://bioinf.wehi.edu.au/software/elda13. We calculated the reciprocal of 95% confidence intervals for 1/(stem cell frequency) generated by ELDA software and the results shown in graph form as reported previously [bib_ref] RYK promotes the stemness of glioblastoma cells via the WNT/beta-catenin pathway, Adamo [/bib_ref].
## Conditioned media
Media collected from stem cells cultures or differentiated cells were centrifuged at 1200 rpm for 5 min to remove cells and stored at −80 - C until used to treat fibroblasts for 48 h.
## Flow cytometry
The ALDEFLUOR assay was carried out according to the manufacturer's guidelines (Stem Cell Technologies). Briefly, dissociated single cells were suspended in Aldefluor assay buffer containing an ALDH substrate, bodipy-aminoacetaldehyde (BAAA) (3 µL per milliliter of sample), and incubated for 40 min at 37 - C. A fraction of cells was incubated under identical conditions in the presence of an ALDH inhibitor (diethylaminobenzaldehyde (DEAB); 4 µL per milliliter of sample). Cells were analyzed using a BD Accuri C6 (BD Biosciences) flow cytometer with BD Accuri C6 software.
## Rna extraction and qrt-pcr
Total RNA (miR and mRNA) was extracted using Trizol (Invitrogen, Milan, Italy) according to the manufacturer's protocol. Reverse transcription of total miR was performed starting from equal amounts of total RNA/sample (500 ng) using SuperScript III Reverse Transcriptase (Invitrogen). Quantitative analysis of Zeb, Nanog, TLR4, IL-6, and β-actin (as an internal reference) were performed by Real Time PCR using specific primers and iQ SYBR Green Supermix (Bio-Rad, Milan, Italy). The reaction for detection of mRNAs was performed as follows: 95 - C for 15 , 40 cycles of 94 - C for 15 , 58 - C for 30 , and 72 - C for 30 .
For miRNA expression analysis, 250 ng of the extracted RNA was reverse-transcribed using a TaqMan MicroRNA Reverse Transcription Kit. Quantitative analysis for microRNA-216a-5p and U6 (as an internal reference) were performed by qRT-PCR using specific Taqman probes.
All reactions were run in triplicate. The threshold cycle (CT) is defined as the fractional cycle number at which the fluorescence passes the fixed threshold. For relative quantization, the 2 −∆∆CT method was used. Experiments were carried out in triplicate for each data point, and data analysis was performed by using Bio-Rad software.
## Protein isolation and western blotting
Cells were washed twice in ice-cold PBS, and lysed with JS buffer (50 mM HEPES pH 7.5 containing 150 mMNaCl, 1% glycerol, 1% Triton X100, 1.5mM MgCl 2 , 5mM EGTA, 1 mM Na 3 VO 4 , and 1X protease inhibitor cocktail). Protein concentration was determined by the Bradford assay (BioRad) using bovine serum albumin as the standard, and equal amounts of proteins analyzed by SDS-PAGE (12.5% or 15% acrylamide). Gels were electroblotted onto nitrocellulose membranes (Millipore, Bedford, MA, USA), membranes then blocked for 1 h with 5% non-fat dry milk in Tris Buffered Saline (TBS) containing 0.1% Tween-20, and incubated at 4 - C overnight with the primary antibody. Detection was performed by peroxidase-conjugated secondary antibodies using the enhanced chemiluminescence system (Thermo, Euroclone, Milan, Italy). Primary antibodies used were: anti E-Cadherin, anti-Nanog, anti-Slug, anti-Vimentin, anti-TLR4, anti-IL6, anti-FAP, anti-αSMA (Santa Cruz Biotechnologies, MA, USA), anti-β-actin (Sigma), and anti-vinculin (Cell Signaling). Western blots are from representative experiments.
## Migration assay
8.0 µm polycarbonate membrane permeable 6.5 mm transwell inserts (Corning Incorporated, NY, USA) were used to carry out the migration assay. Normal fibroblasts were harvested with a trypsin-EDTA solution (Sigma Aldrich, USA). 5 × 10 4 cells were washed with PBS, then resuspended in 1% fetal bovine serum containing DMEM-F12 medium and seeded in the upper chamber. The lower chamber of the transwell was filled with 600 µL of culture medium containing 10% FBS. Cells were incubated at 37 - C for 24 h. The transwells were stained with 0.1% Crystal Violet in 25% methanol.
Non-migrated cells were scraped off the top of the transwell with a cotton swab. The percentage of migrated cells was evaluated by eluting Crystal Violet with 1% SDS and reading the absorbance at λ 570 nm.
# Statistical analysis
Data are presented as mean ± standard deviation (SD). For comparisons between two groups, the Student's t test was used to determine differences between mean values for normal distribution. All data were analyzed for significance using GraphPad Prism 6 software (San Diego, CA, USA). p values less than 0.05 were considered significant.
## Ethics
The study was conducted according to the criteria set by the declaration of Helsinki and each subject signed an informed consent before participating to the study. The study was approved by the Research Ethics Committee of the University of Naples Federico II n - 119/15ES1.
[fig] Figure 1: MiR-216a-5p expression in breast cancer stem cells (BCSCs). (A) MiR-216a expression levels were analyzed by qRT-PCR in primary breast cancer cells cultured in suspension (stem cells) or adherent (differentiated cells) conditions. (B) MDA-MB-231 and T47D stem cells express higher levels of stem markers compared to differentiated cells, as assessed by Western blotting. (C) MDA-MB-231 [/fig]
[fig] Figure 2: MiR-216a-5p regulates BCSC phenotype. (A,B) MiR-216a overexpression decreases the frequency of stem cells in MDA-MB-231 and T47D cultures, as assessed by limiting dilution assay. Significance was calculated using ELDA software. (C) ALDH activity decreases after transfection with miR-216a in MDA-MB-231 and T47D stem cells. (D) Stemness and EMT markers are decreased after [/fig]
[fig] Figure 3: MiR-216a-5p increases stemness phenotype in differentiated breast cancer cells. (A) MiR-216a anti-sense oligonucleotide increases the number of mammospheres when transfected into differentiated MDA-MB-231 cells. Images were taken at 10x magnification, scale bar 250 µm. (B) MiR-216a overexpression decreases the number of stem cells in MDA-MB-231 and T47D cultures, as assessed by mammosphere assay. Images were taken at 5x magnification, scale bar 500 µm. (C) miR-216a downregulation, mediated by anti-miR-216a transfection, increases ALDH in MDA-MB-231 and T47D differentiated cell cultures. (D) Downregulation of miR-216a leads to a higher expression of stemness and EMT markers in differentiated MDA-MB-231 and T47D cells, as assessed by Western blotting. (E) ALDH activity increases after transfection with miR-216a anti-sense in patient-derived organoid cultures. Data are mean values ± SD of three independent experiments. Significance was calculated using Student's t-test; **, p < 0.01, ***, p < 0.001. Int. J. Mol. Sci. 2019, 20, [/fig]
[fig] Figure 4: Toll-like receptor 4 (TLR4) is targeted by miR-216a-5p and regulates interleukin 6 (IL-6) levels. (A,B) TLR4 and IL6 protein and mRNA levels are higher in MDA-MB-231 and T47D stem cells compared to differentiated cells, as assessed by Western blotting and qRT-PCR, respectively. (C) MiR-216a overexpression induces a decrease in TLR4 and IL6 protein expression in MDA-MB-231 and [/fig]
[fig] of 18, Figure 5: Int. J. Mol. Sci. 2019, 20, x FOR PEER REVIEW 11 Cont. [/fig]
[fig] Figure 5, Figure 5: Effect of MiR-216a on cancer-associated fibroblasts (CAFs) and normal fibroblasts (NFs). (A) CAFs were treated with conditioned media collected from MDA-MB-231 and T47D stem cells overexpressing miR-216a. MiR-216a reduces CAF activation compared to the negative control. (B) Differentiated MDA-MB-231 and T47D cells transfected with anti-miR-216a upregulates CAF markers Effect of MiR-216a on cancer-associated fibroblasts (CAFs) and normal fibroblasts (NFs). (A) CAFs were treated with conditioned media collected from MDA-MB-231 and T47D stem cells overexpressing miR-216a. MiR-216a reduces CAF activation compared to the negative control. (B) Differentiated MDA-MB-231 and T47D cells transfected with anti-miR-216a upregulates CAF markers in NFs. (C) CAFs treated with conditioned media collected from MDA-MB-231 and T47D stem cells transfected with miR-216a. MiR-216a impaired fibroblast migration compared to the negative control.Images were taken at 5× magnification. (D) NFs treated with conditioned media collected from MDA-MB-231 and T47D differentiated cells transfected with anti-miR-216a. Anti-MiR-216a sequence induces migration of cells compared to the negative control. Images were taken at 5× magnification. Data are mean values ± SD of three independent experiments. Significance was calculated using Student's t-test. *, p < 0.05; **, p < 0.01. ***, p < 0.001. [/fig]
|
10.1186/2046-4053-1-58
|
CCBY
|
3563607
|
23176742
|
s2orc_pubmed_articles
|
Depression screening and mental health outcomes in children and adolescents: a systematic review protocol
Background: Depression is an important cause of disability among children and adolescents. Depression screening is one possible method for managing depression, and screening programs have been initiated in some school and medical settings. However, in 2005, the Canadian Task Force on Preventive Health Care and the United Kingdom National Institute of Clinical Excellence did not recommend depression screening among children and adolescents. By contrast, in 2009, the United States Preventive Services Task Force recommended that all adolescents, but not younger children, be screened for depression in medical settings with integrated depression management services, although no trials of screening were identified. The objectives of this systematic review are to evaluate in children and adolescents the accuracy of depression screening tools; depression treatment efficacy; whether depression screening improves depression outcomes; and potential harms related to depression interventions and screening. Methods/design: Data sources will include the bibliographic databases MEDLINE, Cochrane CENTRAL, PsycINFO, EMBASE, LILACS and Web of Science, supplemented by reference harvesting of eligible articles, relevant systematic reviews, relevant guidelines and recommendations, and selected journals, and by searches for unpublished studies. Eligible studies will report data for children and adolescents aged 6 to 18 years. Eligible diagnostic accuracy studies must compare a depression screening tool to a validated diagnostic interview for major depressive disorder and report diagnostic accuracy data. Eligible treatment studies must be randomized controlled trials of pharmacological, psychotherapeutic, or other depression treatments commonly available for children and adolescents in pediatric, primary-care, and family medicine settings. Eligible screening studies must be randomized controlled trials that compare depression outcomes between children or adolescents who underwent depression screening versus those who did not. Studies of harms will include randomized controlled trials and observational studies that evaluate harms from depression screening or treatment. Two investigators will independently review titles and abstracts, followed by full article review. Disagreements will be resolved by consensus. Two investigators will independently extract the data, with discrepancies resolved via consensus. Discussion: The proposed systematic review will determine whether there is sufficient evidence of benefits in excess of harms and costs to support screening for depression in childhood and adolescence.Key Question #1: What is the accuracy of depression screening instruments to detect cases of MDD among (a) children and (b) adolescents? Key Question #2: Is treatment of MDD during (a) childhood and (b) adolescence effective in improving symptoms of depression? Key Question #3: Is depression screening during (a) childhood and (b) adolescence more effective than usual care in (i) improving depressive symptoms or (ii) reducing the number of MDD diagnoses? Key Question #4: What are the potential harms associated with depression screening, including treatment, during (a) childhood and (b) adolescence?
# Background
Depression in children and adolescents is a disabling condition that is associated with long-term mental and physical health problems [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref]. Screening for depression is one possible solution to improve depression management. In 2005, however, the Canadian Task Force on Preventive Health Care (CTFPHC) determined that there was not sufficient evidence on health outcomes from depression screening to recommend the practice among children and adolescents in primary health care settings [bib_ref] Canadian Task Force on Preventive Health Care: Screening for depression in primary..., Macmillan [/bib_ref]. Consistent with this, a 2005 guideline from the United Kingdom's National Institute for Health and Clinical Excellence (NICE) concluded that universal screening was not advised based on available evidence. In 2009, by contrast, the United States Preventive Services Task Force (USPSTF) concluded that adolescents, but not younger children, should be routinely screened for depression in primary health care settings 'when systems are in place to ensure accurate diagnosis, psychotherapy (cognitive-behavioral or interpersonal), and follow-up' (page 1223). This recommendation was made based on a systematic review of the literature through May 2006 even though, as the USPSTF recognized, there was little data on the accuracy of screening tools and no data to compare health outcomes between screened and unscreened adolescents [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref]. In the more than 5 years since that review, a substantial amount of research relevant to depression screening in children and adolescence has been published.
Screening is potentially costly in health care resources and has the potential to cause harm to some children and adolescents who may be diagnosed or treated inappropriately. An important empirical question is whether screening, given these concerns, would be more effective than alternative care models that do not rely on universal screening. As described in a recent critique of calls for depression screening, there are a number of reasons why simply assuming that depression screening would result in more benefit than harm may not be reasonable, including high false positive rates if screening tools are not sufficiently precise, generally small depression treatment effects, and the inconsistent quality of routine care that is often provided [bib_ref] Rethinking recommendations for screening for depression in primary care, Thombs [/bib_ref]. Thus, the objective of the proposed systematic review is to evaluate the mental health effects of screening children and adolescents aged 6 to 18 years for major depressive disorder (MDD) in medical or community settings. To do this, we identified the following key questions based on wellestablished criteria delineated by the World Health Organizationand the United Kingdom National Screening Committee, as well as the methodological framework of the USPSTF for evaluating screening programs [bib_ref] Methods Work Group, Third US Preventive Services Task Force: Current methods of..., Harris [/bib_ref] :
## Major depressive disorder in children and adolescents
Depression is primarily characterized by the core symptoms of persistent sad mood or irritability and/or a loss of interest or pleasure in activities that are normally enjoyed. Among children and adolescents, irritability may be more prominent than sad mood, and tantrums and other disruptive behavior may also be important markers. In North America, MDD is diagnosed among children and adolescents based on criteria established by the American Psychiatric Association and delineated in the 4th Edition of the Diagnostic and Statistical Manual (DSM-IV). Criteria for a diagnosis of MDD include having at least five of nine depressive symptoms for 2 weeks or more, at least one of which must be depressed mood, including irritability, or loss of interest in activities. Symptoms must cause significant distress or impairment in daily function. MDD should be distinguished from bipolar disorder, which is characterized by abnormal and disruptive elevated moods, in addition to depression. The DSM-IV also includes dysthymic disorder, which is a longer lasting, but less severe, mood manifestation than MDD, as well as minor depressive disorder, which requires only two, rather than five, depressive symptoms. This review, consistent with earlier reviews by the CTFPHC [bib_ref] Canadian Task Force on Preventive Health Care: Screening for depression in primary..., Macmillan [/bib_ref] and the USPSTF [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] will assess evidence relevant to screening for MDD, but not dysthymic disorder or minor depression, for which treatment options and efficacy are much less well delineated. It will also include studies relevant to screening for depression based on the International Classification of Diseases (ICD) definition of depression, which is similar to the DSM-IV definition.
## Prevalence and burden of depression in children and adolescents
A 2006 meta-analysis that synthesized data from over 60,000 adolescents aged 13 to 18 years estimated the point prevalence of MDD to be 5.6% in the community with rates slightly higher among girls than boys [bib_ref] Is there an epidemic of child or adolescent depression?, Costello [/bib_ref]. In Canada, a study of almost 18,000 respondents to the National Population Health found that 4.8% of boys and 8.7% of girls had an episode of MDD in the past year. Among youth aged 12 to 14 years, the rates were 2.7% for boys and 2.6% for girls. For adolescents aged 15 to 19 years, 6.1% of boys and 12.5% of girls had experienced at least one episode [bib_ref] Gender differences in the prevalence of depression among Canadian adolescents, Cairney [/bib_ref]. Compared to community samples, in medical settings the rate may be up to twice as high [bib_ref] Correlates of ADHD among children in pediatric and psychiatric clinics, Busch [/bib_ref]. Some research has suggested that the lifetime prevalence of an episode of MDD among adolescents may be as high as 20% [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] [bib_ref] Major depressive disorder in older adolescents: prevalence, risk factors, and clinical implications, Lewinsohn [/bib_ref] [bib_ref] Uncommon troubles in young people: prevalence estimates of selected psychiatric disorders in..., Whitaker [/bib_ref]. In Canada, however, a study that used data from the Canadian Community Health Survey Cycle 1.2 reported a lifetime prevalence rate among almost 3,000 adolescents aged 15 to 19 years of 7.6% for MDD, including 4.3% for males and 11.1% for females [bib_ref] Canadian Community Health Survey: major depressive disorder and suicidality in adolescents, Cheung [/bib_ref]. Among youth under age 13, the point prevalence of MDD has been estimated to be approximately 3% [bib_ref] Is there an epidemic of child or adolescent depression?, Costello [/bib_ref].
MDD in childhood and adolescence is associated with many negative outcomes, including behavioral problems and poor school performance, early pregnancy, and impaired social, work, and family functioning in adolescence and into adulthood [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref]. Many studies have shown that patients with depression in adulthood had at least one episode of MDD in childhood or adolescence, and there is a high rate of recurrence among youth with MDD [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] [bib_ref] Uncommon troubles in young people: prevalence estimates of selected psychiatric disorders in..., Whitaker [/bib_ref] [bib_ref] Psychiatric diagnosis in child and adolescent suicide, Shaffer [/bib_ref] [bib_ref] Depressed adolescents grown up, Weissman [/bib_ref] [bib_ref] Mental health, educational, and social role outcomes of adolescents with depression, Fergusson [/bib_ref]. Despite this substantial burden, relatively few children and adolescents with depression receive treatment [bib_ref] Patterns and predictors of treatment contact after first onset of psychiatric disorders, Kessler [/bib_ref] [bib_ref] Epidemiology of DSM-III-R major depression and minor depression among adolescents and young..., Kessler [/bib_ref] [bib_ref] Mental health service use in the community and schools: results from the..., Leaf [/bib_ref].
What is depression screening and when should it be recommended?
Screening is a preventive strategy that is traditionally used to detect disease in patients who otherwise have no signs or symptoms. Screening programs are premised on the assumption that early detection of disease will enable earlier and more effective intervention. Unlike screening for most other medical diseases, however, screening for depression does not seek to achieve early identification of pre-symptomatic cases that will subsequently evolve into psychiatric disorder. Rather, screening for depression involves the use of depression symptom questionnaires or small sets of questions about depression to identify patients who may have current depression, but who have not sought treatment and whose depression has not already been recognized by health care providers. Patients identified as possible cases need to be further assessed and, if appropriate, offered treatment. Depression screening is potentially useful only to the extent that it improves patient outcomes beyond any treatment that is provided as part of standard care. Thus, to be successful, a screening program must identify a significant number of depressed patients who are not already diagnosed with depression, engage those patients in treatment, and obtain sufficiently positive treatment results to justify costs and potential harms from screening [bib_ref] Rethinking recommendations for screening for depression in primary care, Thombs [/bib_ref].
In 1968, the World Health Organization issued a report that delineated criteria to determine whether conditions are suitable for screening, and the main elements of those criteria continue to be used today. Generally, it is reasonable to consider screening when the condition in question is important and prevalent, can be effectively treated, and cannot be readily detected without screening. Further, screening methods should be accurate and carry only a tolerably small risk of false positive results, which could lead to unnecessary diagnostic testing, adverse effects, costs of inappropriate treatment, and to sequelae of being incorrectly labeled, such as stigma. False reassurance for false negatives may also need to be considered in some circumstances. The principal criterion is that there must be evidence that benefits from screening outweigh potential harms. Ideally, benefits in excess of potential harms should be demonstrated consistently in well-conducted randomized controlled trials (RCTs) with sufficiently long follow-up to cover the time horizon of important patient-oriented outcomes.
## Current practices in depression screening in childhood and adolescence
Studies from community [bib_ref] Improving adolescent preventive care in community health centers, Klein [/bib_ref] and health maintenance organization [bib_ref] Jr: Preventive services in a health maintenance organization: how well do pediatricians..., Halpern-Felsher [/bib_ref] settings in the United States published in 2000 and 2001 suggested that 40% to 60% of adolescent patients may be screened for depression based on physician report, but this number fell to only 3% when medical charts were examined for documentation of screening [bib_ref] Improving adolescent preventive care in community health centers, Klein [/bib_ref]. More recently, TeenScreen, an American organization based at Columbia University, has aggressively promoted widespread mental health screening for adolescents and has reported that they received more than 400,000 requests for screening questionnaires from schools, primary-care physicians and managed-care organizations in 2010 alone [bib_ref] Will students take a mental health test?, Landro [/bib_ref]. There are numerous reports of mental health screening supported by TeenScreen being conducted routinely in medical practices and across entire school districts in the United States [bib_ref] Enhancing access through TeenScreen, Flynn [/bib_ref] [bib_ref] Identifying adolescents at risk through voluntary school-based mental health screening, Husky [/bib_ref] [bib_ref] Mental health screening of adolescents in pediatric practice, Husky [/bib_ref]. In Canada, the governments of Alberta and Manitoba, for instance, have called for widespread depression screening in school settings as part of longterm plans to improve youth mental health.
## Existing guidelines and recommendations
In Canada, the last guideline statement on the topic was a CTFPHC recommendation from 2005 [bib_ref] Canadian Task Force on Preventive Health Care: Screening for depression in primary..., Macmillan [/bib_ref]. The CTFPHC recommendation focused on adult depression, but noted that, consistent with previous findings from the USPSTF, there was not enough evidence to support a recommendation for screening in child or adolescent medical care settings. Similarly, a 2005 NICE guideline on depression management for children and youth in the United Kingdom emphasized the importance of better and more consistent care, but concluded that the evidence needed to support a screening recommendation was not available. More recently, in 2009, the USPSTF issued a recommendation that adolescents in medical settings should be screened for depression in primary-care settings when integrated care systems are in place to ensure accurate diagnosis, competent psychotherapeutic and medical support, and follow-up. The recommendation document focused on the prevalence and burden of depression among adolescents, as well as the existence of screening tools and treatments, but not on evidence from any RCT that mental health outcomes improved among youth who were screened compared with youth who were not screened for depression. Calls for depression screening have gone beyond medical settings. The 2003 United States President's New Freedom Commission on Mental Health called for screening in primary medical care, school, and child welfare settings as the key to reducing the community burden of depression, and school-based screening programs have been implemented in Canada. Existing guidelines and recommendations, however, have not directly addressed whether school-based screening outside of the context of health care settings is recommended.
## Previous systematic reviews and meta-analyses
A number of reviews have assessed the accuracy of screening tools for detecting MDD, the efficacy of treatments, and harms that may be associated with depression treatment in children or adolescents (see Additional file 1). Only one review [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] , a United States Agency for Healthcare Research and Quality (AHRQ) report, which was done in conjunction with a USPSTF guideline, has assessed the various components of a depression screening program for children or adolescents (accurate screening tools, effects of treatment, effects of screening, potential harms) and whether there was evidence that mental health outcomes were better for screened versus non-screened children and adolescents. The review was published in 2009, but only included studies through a search that was done through May 2006. The review included nine studies on the accuracy of six different depression screening tools, but reported that all had serious methodological flaws, including non-random patient selection, excessive delays between administration of screening tools and diagnostic interviews, high levels of attrition, poor reporting of methods and results, and small samples. No studies were rated as being of good quality. The review identified 18 fair-to goodquality RCTs on treatment for MDD with psychotherapy, medication, or a combination of psychotherapy and medication, and concluded that treatments were generally effective, but that not all medications appeared to work well. There were no studies identified that examined mental health outcomes among children or adolescents who were screened compared to children and adolescents who were not screened.
# Methods/design
This systematic review has been funded by the Canadian Institutes for Health Research (Funding Reference Number KA1 -119795). The protocol has been registered in the PROSPERO prospective register of systematic reviews (CRD42012003194).
The review methodology was designed to be consistent with reporting guidelines described in the Meta-analysis of Observational Studies in Epidemiology statement [bib_ref] Meta-analysis of observational studies in epidemiology: a proposal for reporting. Meta-analysis of..., Stroup [/bib_ref] and the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement [bib_ref] The PRISMA statement for reporting systematic reviews and meta-analyses of studies that..., Liberati [/bib_ref] [bib_ref] Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement, Moher [/bib_ref]. Key questions for this systematic review are based on the USPSTF analytic framework [bib_ref] Methods Work Group, Third US Preventive Services Task Force: Current methods of..., Harris [/bib_ref]. The USPSTF logic model describes a process for determining whether screening benefits likely outweigh harms, including the specific populations, screening procedure, and health outcomes to be considered [bib_ref] Methods Work Group, Third US Preventive Services Task Force: Current methods of..., Harris [/bib_ref]. The model assumes that patients with undetected depression undergo screening, which classifies patients as cases or non-cases of MDD, and that the evaluation of the accuracy of screening tests is a key component. Treatment studies are evaluated to determine the degree to which they improve symptoms of depression. Trials of actual screening programs, in which health outcomes are evaluated for screened versus unscreened children and adolescents are, if they exist, the final standard by which the value of screening can be tested in terms of its ability to improve depression outcomes and produce benefits to children and adolescents that exceed potential harms from screening.
## Search strategy
Four distinct sets of searches will be conducted, one for Key Question #1 (accuracy of screening tools), one for Key Question #2 (effects of treatment), one for Key Question #3 (effects of screening), and one for Key Question #4 (harms). To identify studies relevant to Key Question #1 (accuracy of screening tools), the MEDLINE, EMBASE, PsycINFO, HAPI, and LILACS databases will be searched. For Key Questions #2 (effects of treatment), #3 (effects of screening), and #4 (harms of screening), the MEDLINE, EMBASE, PsycINFO, Cochrane CENTRAL, and LILACS databases will be searched. All search strategies have been peer-reviewed. There will be no language restrictions. See Additional file 2 for specific search terms that will be used.
Searches will be run from January 2006 to the present because an AHRQ systematic review on depression screening in children and adolescents [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] searched to May 2006. Eligible studies from that review will be included in the present review. For the effects of depression treatment, the previous AHRQ review [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] included only studies on the effects of selective serotonin reuptake inhibitors (SSRIs) and psychotherapy. However, serotonin-norepinephrine reuptake inhibitors (SNRIs), other non-SSRIs (bupropion, mirtazapine, trazodone), and exercise are also treatments for depression that may be commonly used with children or adolescents in general pediatric settings or primary-care and family medicine settings. Thus, for Key Question #2 (effects of treatment), an additional search with no date restriction will be run for studies on the effects of SNRIs, a select group of non-SSRIs (bupropion, mirtazapine, trazodone), and exercise.
In addition, manual searching will be done for the period of 3 months prior to the final database search on the references of eligible original articles, relevant systematic reviews and meta-analyses, and selected journals, including journals in psychiatry, psychology, and pediatrics. To identify unpublished or ongoing trials, we will use methods that include reviewing trial registries and results databases (for example, ClinicalTrials.gov), pharmaceutical industry trial registries and results databases, regulatory agency online databases, regulatory agency submissions, litigation documents, and conferences abstracts, as well as contacting trialists and sponsors for unpublished or ongoing trials [bib_ref] Out of sight but not out of mind: how to search for..., Chan [/bib_ref].
## Identification of eligible studies
Eligible studies will include children and adolescents aged 6 to 18 years. We will include studies conducted in general medicine clinics, schools, and community settings. Studies including medically ill children are eligible. Studies of college and university populations will be excluded, as college and university students have a different pathway to mental health treatment than adolescents under their parents' care.
Key Question #1 (accuracy of screening tools): Studies on the accuracy of screening tools will be included if they compared a screening instrument with a valid criterion standard, defined as a DSM diagnosis of MDD or an ICD diagnosis of depressive episode based on a validated diagnostic interview procedure, and if they reported data allowing determination of sensitivity and specificity, positive predictive value, and negative predictive value. Examples of validated psychiatric interviews that have been used in assessments of children or adolescents include, but are not limited to, the Composite International Diagnostic Interview, the Diagnostic Interview for Children and Adolescents, the Revised Diagnostic Interview Schedule for Children, and the Schedule for Affective Disorders and Schizophrenia for School-Aged Children. Studies that assess broader diagnostic categories, such as any depressive disorder or dysthymia, will be included only if they report data for MDD separately. Studies must administer the screening tool and the diagnostic assessment within 2 weeks of each other to be eligible. Studies in which only parent-or teacher-completed measures, but no child or adolescent self-report measures, are compared to a diagnosis of MDD will be excluded. Studies will be excluded if a cutoff score above a threshold on a depression screening tool is used as an eligibility criterion to receive assessment for MDD with a validated structured interview. Studies conducted in high-risk populations where many or most children and adolescents may have a psychiatric disorder, such as in psychiatric or youth-protection settings, will be excluded. Key Question #2 (effects of treatment): Studies on treatment will include RCTs with placebo or usual care controls that evaluated pharmacological, psychotherapeutic, or other interventions that would typically be available to children or adolescents in general pediatric settings or primary-care and family medicine settings as treatment for depression as diagnosed with a validated psychiatric interview and DSM criteria for MDD or ICD criteria for a depressive episode. Specifically, eligible interventions will include SSRIs (fluoxetine, sertraline, paroxetine, citalopram, escitalopram, and fluvoxamine), SNRIs (venlafaxine, duloxetine, desvenlafaxine), other non-SSRIs prescribed for children or adolescents with depression (bupropion, mirtazapine, trazodone), psychotherapy, educational interventions, and exercise. We will require, as an eligibility criterion, a DSM diagnosis of MDD or ICD diagnosis of depressive episode based on a valid diagnostic interview because unassisted clinician diagnoses have poor reliability and because a large proportion of patients scoring above cutoffs on selfreport questionnaires do not have MDD. Head-to-head trials of different interventions without a comparison to usual care or placebo are not eligible. Key Question #3 (effects of screening): Eligible studies will be RCTs that compared depression outcomes between children or adolescents who underwent depression screening and those who did not. Studies in which comprehensive depression care programs were provided to children or adolescents with depression as part of the screening program will be included if children and adolescents in the unscreened group could also access these treatment programs if identified as depressed by means other than screening. Otherwise, it will not be possible to disaggregate effects of screening from effects of providing comprehensive depression management. Changes in rates of depression recognition and treatment will be noted, but not included as depression outcomes. This is because increased treatment is not a benefit and, if improved depression outcomes are not obtained, exposes children and adolescents to costs and potential harms of treatment without benefit. Screening is defined per the United Kingdom National Screening Committee's definition. Thus, eligible screening trials had to include a case identification strategy based on an a priori defined cutoff score on a depression screening tool to make decisions regarding further assessment or treatment. Key Question #4 (harms): Studies that document harms from depression treatment or screening will include RCTs, as well as observational studies.
Two investigators will independently review titles and abstracts for eligibility with full-text review of articles identified as potentially eligible by one or both. Disagreements after full-text review will be resolved by consensus. Chance-corrected agreement will be assessed with Cohen's κ. See Additional file 3 for the coding manual.
## Data extraction and outcomes
Two reviewers will independently extract relevant data from each eligible article and enter the data directly into formatted Excel spreadsheets. Entries will be compared for accuracy and any discrepancies will be resolved by consensus. Authors of original studies will be contacted if necessary to clarify inconsistencies in reported results. See Additional file 4 for variables included in data extraction templates.
Key Question #1 (accuracy of screening tools): Outcomes reported will include sensitivity, specificity, positive predictive value, and negative predictive value, based on child or adolescent self-report measures. Key Questions #2 (effects of treatment) and #3 (effects of screening): The primary outcome will be standardized mean difference effect size based on a continuous measure of depressive symptoms. When multiple depression outcomes are reported, primary outcomes as identified in each study will be given highest priority. Then observer-rated scales will be prioritized over self-report measures. Outcomes will be reported as intent-to-treat analyses where possible and will be prioritized over those presented for completers only. Key Question #4 (harms): Outcome reporting will be based on harms identified from the review, but we expect the main potential harm to be suicidal ideation.
## Appraisal of risk of bias
For Key Question #1 (accuracy of screening tools), we will use the Quality Assessment of Diagnostic Accuracy Studies-2 tool [bib_ref] QUADAS-2 Group: QUADAS-2: a revised tool for the quality assessment of diagnostic..., Whiting [/bib_ref]. This tool incorporates assessments of risk of bias across four core domains: patient selection, the index test, the reference standard, and the flow and timing of assessments. For Key Questions #2 (effects of treatment) and #3 (effects of screening), we will use the Cochrane Risk of Bias tool, which includes assessments of six possible sources of bias related to randomization sequence generation; allocation concealment; blinding of participants, personnel, and outcome assessors; incomplete outcome data; selective outcome data; and other sources of bias. In addition to these domains, based on recent recommendations regarding potential bias due to financial conflicts of interest [bib_ref] Reporting of conflicts of interest in meta-analyses of trials of pharmacological treatments, Roseman [/bib_ref] [bib_ref] Reporting of conflicts of interest from drug trials in Cochrane reviews: cross..., Roseman [/bib_ref] , we will code two further domains, for pharmaceutical industry funding, and author-industry financial ties or employment by industry.
## Synthesis of included studies
The decision as to whether data can be synthesized using meta-analysis will be determined based on search results. Based on the studies included in the 2009 United States AHRQ systematic review [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] and a rough preliminary search of MEDLINE conducted in October 2011, we expect that there will be substantial heterogeneity between diagnostic accuracy studies with respect to assessment timing, criterion standards, screening instruments, and scoring thresholds. Most studies have used population-specific cutoff thresholds based on receiver operator characteristic curves, which yields overly optimistic estimates of screening accuracy that do not replicate consistently [bib_ref] Clinical versus actuarial judgment, Dawes [/bib_ref]. Similarly, for studies of MDD treatment, there appears to be substantial heterogeneity in the nature of interventions, outcome measures, and length of follow-up. Thus, due to clinical heterogeneity, it appears that data pooling will likely not be appropriate and that a systematic review, rather than a meta-analysis, will be conducted. If, following the full literature review, we find that meta-analytic data pooling is feasible, then we will use appropriate models that have been used previously in meta-analyses of diagnostic accuracy [bib_ref] Screening for depression in medical settings with the Patient Health Questionnaire (PHQ):..., Gilbody [/bib_ref] , screening effectiveness [bib_ref] Screening and case-finding instruments for depression: a meta-analysis, Gilbody [/bib_ref] , and treatment outcomes from depression care interventions [bib_ref] Educational and organizational interventions to improve the management of depression in primary..., Gilbody [/bib_ref] [bib_ref] Collaborative care for depression: a cumulative meta-analysis and review of longer-term outcomes, Gilbody [/bib_ref].
# Discussion
Depression is a chronic and disabling condition that is the leading global cause of life-years lived with disability and the fourth leading cause of disability-adjusted lifeyears, which takes into account premature mortality [bib_ref] Research on major depression: strategies and priorities, Insel [/bib_ref]. Depression during childhood and adolescence can have a devastating impact, both because of its effect on important childhood and adolescence outcomes and because it is a robust predictor of ongoing mental health problems into adulthood [bib_ref] Screening for child and adolescent depression in primary care settings: a systematic..., Williams [/bib_ref] [bib_ref] Uncommon troubles in young people: prevalence estimates of selected psychiatric disorders in..., Whitaker [/bib_ref] [bib_ref] Psychiatric diagnosis in child and adolescent suicide, Shaffer [/bib_ref] [bib_ref] Depressed adolescents grown up, Weissman [/bib_ref] [bib_ref] Mental health, educational, and social role outcomes of adolescents with depression, Fergusson [/bib_ref].
It is crucial that optimal depression prevention and care programs be developed and implemented to reduce the burden of suffering from childhood and adolescent depression. Universal depression screening has been recommended as a solution. Guidelines and recommendations, however, are sometimes made without full consideration of evidence or clinical practice realities [bib_ref] Why guideline-making requires reform, Sniderman [/bib_ref]. Current guidelines for the management of childhood and adolescent depression do not all agree about whether universal screening should be conducted [bib_ref] Canadian Task Force on Preventive Health Care: Screening for depression in primary..., Macmillan [/bib_ref] , and depression screening among children and adolescents has been implemented in some settings without evidence of benefit [bib_ref] Will students take a mental health test?, Landro [/bib_ref] [bib_ref] Enhancing access through TeenScreen, Flynn [/bib_ref] [bib_ref] Identifying adolescents at risk through voluntary school-based mental health screening, Husky [/bib_ref] [bib_ref] Mental health screening of adolescents in pediatric practice, Husky [/bib_ref].
Although screening is sometimes portrayed as the only alternative to simply ignoring depression [bib_ref] American Psychiatric Association: Depression and coronary heart disease: recommendations for screening, referral,..., Lichtman [/bib_ref] , researchers have accumulated evidence suggesting that patients may benefit more by investing resources into improving programs to better manage depression rather than seeking to identify otherwise unidentified patients, many of whom have lesser levels of symptomatology, and adding them to an already poorly functioning depression care system [bib_ref] Should we screen for depression?, Gilbody [/bib_ref] [bib_ref] A framework for describing the impact of antidepressant medications on population health..., Patten [/bib_ref]. Indeed, Gilbody et al. found that screening does not appear to be a necessary component of effective collaborative care programs for depression in primary care [bib_ref] Collaborative care for depression: a cumulative meta-analysis and review of longer-term outcomes, Gilbody [/bib_ref]. At this juncture, we know very little about potential benefits of depression screening in childhood and adolescence versus potential harms. A real danger is that implementation of routine screening without evidence of how this is best done or whether it will benefit patients could result in overtreatment of depression due to the prescription of antidepressants based on positive screens without follow-up diagnostic interviews on the one hand, and the continued inadequate treatment of children and adolescents with MDD on the other because resources are consumed by attempting to find new cases rather than providing adequate treatment to children and adolescents who are otherwise identified.
The proposed systematic review will determine whether there is sufficient evidence to support screening for depression in childhood and adolescence. The conclusions drawn from this systematic review, once disseminated to policy-makers, health care providers, and researchers, will allow decisions to be made about whether screening programs are likely to benefit children and adolescents.
## Additional files
Additional file 1: Relevant systematic reviews and meta-analyses.
[fig] Additional file 2: Search strategies. Additional file 3: Coding manual. Additional file 4: Variables included in data extraction form. Abbreviations AHRQ: Agency for Healthcare Research and Quality; CTFPHC: Canadian Task Force on Preventive Health Care; DSM: Diagnostic and Statistical Manual; ICD: International Classification of Diseases; MDD: Major depressive disorder; NICE: National Institute for Health and Clinical Excellence; RCT: Randomized controlled trial; SNRI: Serotonin-norepinephrine reuptake inhibitor; SSRI: Selective serotonin reuptake inhibitor; USPSTF: United States Preventive Services Task Force. [/fig]
|
10.1007/s11554-021-01070-6
|
CCBY
|
7818701
|
33500738
|
s2orc_pubmed_articles
|
Implementing a real-time, AI-based, people detection and social distancing measuring system for Covid-19
COVID-19 is a disease caused by a severe respiratory syndrome coronavirus. It was identified in December 2019 in Wuhan, China. It has resulted in an ongoing pandemic that caused infected cases including many deaths. Coronavirus is primarily spread between people during close contact. Motivating to this notion, this research proposes an artificial intelligence system for social distancing classification of persons using thermal images. By exploiting YOLOv2 (you look at once) approach, a novel deep learning detection technique is developed for detecting and tracking people in indoor and outdoor scenarios. An algorithm is also implemented for measuring and classifying the distance between persons and to automatically check if social distancing rules are respected or not. Hence, this work aims at minimizing the spread of the COVID-19 virus by evaluating if and how persons comply with social distancing rules. The proposed approach is applied to images acquired through thermal cameras, to establish a complete AI system for people tracking, social distancing classification, and body temperature monitoring. The training phase is done with two datasets captured from different thermal cameras. Ground Truth Labeler app is used for labeling the persons in the images. The proposed technique has been deployed in a low-cost embedded system (Jetson Nano) which is composed of a fixed camera. The proposed approach is implemented in a distributed surveillance video system to visualize people from several cameras in one centralized monitoring system. The achieved results show that the proposed method is suitable to set up a surveillance system in smart cities for people detection, social distancing classification, and body temperature analysis.
# Introduction
COVID-19 is a disease caused by a new coronavirus which appeared in China in December 2019. COVID-19 symptoms include mainly fever, cough, chills, and shortness of breath, body aches, loss of taste, and smell. COVID-19 can be severe, and in many cases, it has caused death. The coronavirus can spread from one person to another as diagnosed by researchers in laboratories. This pandemic has spread to over 188 countries around the world. On October 15, 2020, WHO (World Health Organization) declared that there have been 38,394,169 confirmed COVID-19 cases and 1089,047 deathsaround the world. The uncertainty, underpinning, and complexity of the coronavirus have made it difficult to predict the duration and spread of this pandemic. As of yet, there is no vaccine available. Prevention involves wearing masks and washing hands frequently. An infected person should stay at home when people are sick to prevent spreading this pandemic to the others. This situation forces the global community and governments to find the best mitigation plan to stop the spread of coronavirus. Nations stopped their business and closed the border and public places such as schools and workplaces to avoid people's interactions. It has been reported that all infected countries who applied the lock-down for their communities achieved a reduction of the number of COVID-19 cases and the number of deaths from this pandemic.
Fever or chills are common symptoms of coronavirus. Researchers in China found that 99% of people infected with the coronavirus presented with a high temperature. Thermal cameras and non-contact infrared thermometers, which 1 3 are non-contact instruments, can be used to measure body temperature. This approach can monitor a person's surface temperature to limit the spread of coronavirus infections.
Based on the information from the World Health Organization, social distancing is the best practice where individuals can minimize physical contact with possible COVID-19 carriers by maintaining a certain distance between one person and another. The main target is to provide a comprehensive tool and effective technologies that can be utilized to enforce social distancing. Technologies could play an important role to facilitate social distancing practice. In such a context, Artificial Intelligence (AI) and information and communication technology (ICT) can be used in addressing this challenge.
This research aims at mitigating the spread of this virus in communities and saving the lives of people. In this work, we propose a deep learning object detection model for people detection in combination with an implemented algorithm for social distancing classification on thermal images. Hereafter, the paper is organized as follows: after the introduction of COVID-19 in Sects. 1, 2 presents the research background and related work. Section 3 shows an overview of object detection. Section 4 presents the proposed methodology to define a measuring system for people detection and social distancing check. Section 5 shows the experimental results and Sect. 6 describes the implementation of the proposed approach on embedded hardware. Conclusions are drawn in Sect. 7.
## Research background and related work
Social distancing and temperature screening are effective tools for preventing the spread of disease. They have been suggested by many organizations, including the World Health Organization (WHO). Russel et al. [bib_ref] The effect of control strategies to reduce social mixing on outcomes of..., Prem [/bib_ref] studied the effects of social distancing techniques on the spread of coronavirus. This paper presented scientific location contact patterns to produce the trajectory of an outbreak by utilizing susceptible exposed infected removed (SEIR) methods. The authors also mentioned that the sudden lifting of social distancing could increase the infection and spread of the virus between people. Nabil Kahale [bib_ref] On the economic impact of social distancing measures, Kahale [/bib_ref] highlighted the impact of social distancing measures. The study aimed to derive an approximation that shows how early social distancing measures can reduce economic loss and the number of new infections significantly. At the time when coronavirus is begun spreading across the individuals and society, research and scientists are starting to find out the best solution to eliminate the spread of this pandemic. Jennifer Berglund, suggested tracking a person infected with COVID-19 using GPS and built-in applications in smartphones. However, this technology has limitations on tracking individuals who have no Wi-Fi or cell signals. On the other hand, some authorities utilize drones with mounted video cameras to track the gathering of individuals in the outdoor area [bib_ref] The use of drones during mass events, Robakowska [/bib_ref]. Such technology is suitable for monitoring COVID-19 which could amid the coronavirus outbreak.
Recently, the problem of classifying and detecting the objects in an image is solved, thanks to the improvements in computer vision and deep learning in general. Accordingly, computer vision development has focused on various interesting and challenging topics, such as neural style transfer, segmentation, and tracking, and of course object detection.
Deep learning is an artificial intelligence function (AI) that emulates the tasks of the human brain in data processing and object detection. It can be referred to as a neural network with a sophisticated algorithm. The history of the neural network dates to 1940 [bib_ref] How we know universals the perception of auditory and visual forms, Pitts [/bib_ref]. The original intention of the neural network is to solve learning problems ethically [bib_ref] Object detection with deep learning: a review, Zhao [/bib_ref]. A convolutional neural network (CNN) is widely used in deep learning models for object detection. CNN is a deep learning algorithm that takes an input image and assigns the learnable weights and biases for various classes in an image and differentiates them from one to another. The convolutional neural network has been made evolution which can be implemented on an embedded system with a low-resolution input and low complexity [bib_ref] Lapped convolutional neural networks for embedded systems, Wang [/bib_ref]. There are various deep learning models such as R-CNN, Single-shot detector (SSD), and YOLO which are applied in different applications for object detection. These models are efficient algorithms for movement estimation in video scenes. Ebrahim et al. [bib_ref] People detection and finding attractive areas by the use of movement detection..., Kajabad [/bib_ref] , proposed a technical approach for detecting people using video frames. The author utilized a background subtraction and Gaussian mixture with a deep learning detector for people detection. In method [bib_ref] Neurocomputing computer vision and deep learning techniques for pedestrian detection and tracking:..., Brunetti [/bib_ref] , the authors presented a deep learning (CNN) technique for human detection. They utilized a combination of deep learning and machine learning methods to achieve high accuracy and less computation for people detection. Unfortunately, this method had problems with low speed for real-time detection. In method [bib_ref] Detection of static groups and crowds gathered in open spaces by texture..., Manfredi [/bib_ref] , researchers suggested a method on static crowds for a group of people that stayed in the same location for a long time. They utilized the mean of class as support vector machine (SVM) to categorize patches as essential crowds and these patches are extracted by text features.
Recent developments showed that the identification of individuals through video surveillance cameras can be achieved by face [bib_ref] Shape similarity for 3d video sequences of people, Huang [/bib_ref] , and a person's manner of walking. However, the detection of a person under crowds' technique is difficult and hard to optimize.
In method [bib_ref] Robust real-time pedestrians' detection in urban environments with low-resolution cameras, Alahi [/bib_ref] , the authors presented a solution for detecting pedestrians with a low-resolution camera by utilizing background subtraction by extracting foreground silhouettes and classifying them in real-time.
## 3 3 overview of object detection
Object detection systems place a bounding box around the objects and associate the correct object's category with each bounding box. Deep learning is an effective method to perform object detection. In [bib_ref] Rich feature hierarchies for accurate object detection and semantic segmentation, Girshick [/bib_ref] , Ross Girshick explored a regional convolutional neural network detector (R-CNN). This model consists of four stages. It starts with introducing the images into the input layer, then it extracts the regional proposals, after that it computes the features by CNN, and finally, it classifies these features, see [fig_ref] Figure 1: The schematic diagram for R-CNN detector [/fig_ref]. R-CNN uses selective search algorithms to generate region proposals. It takes a huge amount of time as it would have to classify the regions per image. R-CNN cannot be implemented in realtime object detection as it takes 47 s for each image. R-CNN cannot be trained at one time. Rather, it needs to train every part independently.
Fast R-CNN is another version of the regional proposal algorithm, which was presented by the same author of R-CNN model [bib_ref] Fast R-CNN, Girshick [/bib_ref]. Fast R-CNN enhanced the drawbacks from R-CNN to build faster object detection algorithm. It is similar to the R-CNN algorithm. However, the input image is fed to the convolutional neural network to generate a convolutional feature map instead of feeding region proposals to CNN. The region proposal is warped into squares in this model. Using region of interest (ROI) pooling layer, these regions are reshaped into a fixed size which can be fed into a fully connected layer. Softmax layer is used in this architecture to predict the class of region and the offset values of the bounding box. [fig_ref] Figure 2: The schematic diagram for Fast R-CNN detector [/fig_ref] shows the schematic diagram for fast R-CNN detector.
Both algorithms (R-CNN and Fast R-CNN) use selective search to find the region proposals. This process a slow and time-consuming which is affecting the performance for neural network algorithm.
Recent improvements in object detection deep learning include other algorithms such as YOLO and YOLOv2. You look at one or (YOLO) is a state-of-the-art deep learning object detection. It was presented by Joseph Redmon et al. [bib_ref] You only look once: Unified, real-time object detection, Redmon [/bib_ref]. YOLO uses a single neural network to the whole image. It divides the image into regions and predicts the bounding boxes and the probabilities for each region. These bounding boxes are weighted by predicted probabilities.
YOLO detector looks full image at one time; therefore, its predictions are informed by the context in the image. It predicts with single network evaluation, unlike other object detectors such as (R-CNN) which requires thousands for a single image. YOLO algorithms take the input image and split into S ×S grids. It extracts the features from each grid. It predicts the bounding boxes with confidence scores for the predicted classes in the bounding boxes, see [fig_ref] Figure 3: Schematic diagram for YOLO [/fig_ref]. Each grid cell detects bounding boxes and confidence scores. The bounding box consists of five predictions which are represented with (x, y, w, h) and the confidence score. The (x, y) coordinates reflect the center of the bounding box of the grid cell. The (w, h) represents the width and the height of the full image. The confidence scores represent the measurement of how confident the detector is that the box contains the object to be predicted.
YOLO predicts several bounding boxes for each grid cell. In the training stage, it only requires one predictor of the bounding box to be responsible for each class. The predictor is assigned to predict an object which has the highest Intersection over Union value (IoU) for the ground truth. This process leads to specialization within the bounding boxes prediction. YOLO algorithms use sum-squared error between the ground truth and predictions of bounding boxes for loss. This sum squared error computes the classification, localization, and confidence losses for the model. Therefore, YOLO is optimized with the following loss function Schematic diagram for YOLO: input image which splits into S×S grids, each grid predicts the bounding boxes and the confidence scores and finally, the score encodes the probability with bounding box on the detected class to enhance its performance during the training process, see Eq. (1). YOLOv2 is the second version of YOLO. It is an object detection system targeted for real-time processing. It has several improvements to YOLO as explored in YOLO9000 paper. YOLOv2 resolved the issues which were encountered with YOLO, thus improving the processing accuracy and speed for the architecture. It enhanced the errors of localization for the classes to be predicted in the images. It uses batch normalization in all convolutional layers.
where: coord is a constant used to increase the weight for the first two terms of the loss function. B is the number of box predictions for each cell. s 2 is the number of cells. 1 obj i,j is equal to 1 if there is an object in cell i and confidence of the jth predictor of this cell is the highest among all the predictors of this cell. x i ,y i represent the location of centroid of the anchor box. w i is the width of the anchor box. h i is the height of the anchor box. C i is the confidence score whether there is an object or no. Batch normalization helps the regularization of the model. It eliminated the requirement for using the dropout layers to overcome the overfitting problems. It improves the normalization for its input by defining the variance values and means over the mini-batch and it calculates the activation as seen in Eq. (2) where x i is the normalize value. xi is the element of the input. B is the mini-batch mean.B is the batch variance. ∈ is the property of Epsilon and enhances the mini batch when the variance value is small.
[formula] (1) coord s 2 � i=0 B � j=0 1 obj i,j [(x i − ∧ x i ) 2 + � y i − ∧ y i ) 2 � + coord s 2 � i=0 B � j=0 1 obj i,j [( √ w i − � ∧ w i ) 2 + � √ h i − � ∧ h i ) 2 � + s 2 � i=0 B � j=0 1 obj i,j (C i − ∧ C i ) 2 + noobj s 2 � i=0 B � j=0 1 noobj i,j (C i − ∧ C i ) 2 + s 2 � i=0 1 obj i � c∈class (p i (c) − ∧ p i (c)) 2 ,(2)x i = xi − B √ 2 B + , [/formula]
We used anchor boxes to make bounding boxes on the detected objects in the images. These boxes are a set of predefined rectangular boxes with a specific width and height. Anchor boxes are defined to capture the scale and ratio of certain classes that are to be detected and typically selected based on the sizes of the objects in the training dataset. K-mean clustering was used to select a good set of labeled boxes in the training dataset in MATLAB. It is essential to have the correct sizes of these bounding boxes (height, width) for YOLOv2 to detect the targeted objects accurately. Intersection over Union (IoU) score of k-means was measured to determine the required number of these bounding boxes for the detector. The advantage of using anchor boxes is to prevent utilizing more boxes which could lead to overfitting and poor performance for YOLOv2 model.
# Proposed methodology
## Social distancing detector steps
This section discusses the essential steps which are attempted to establish a workflow for monitoring social distancing on thermal images as seen in 1. Prepare the thermal images or streaming a video from a thermal camera which contains people. 2. Applying the deep learning object detector to detect people in thermal images or video streams. 3. Check the number of persons that are in the images or video stream. 4. Compute the distance between the centroid of the bounding boxes which are enclosed to the detected people. 5. Finally, the algorithm will decide for safe or unsafe social distancing based on the number of persons and The steps involved for people detection and social distancing classification on thermal images the measured distance between the centroid of bounding boxes.
## Neural network design
A Deep Neural Network (DNN) application is used in MAT-LAB to construct YOLOv2 neural network layers. Then the designed DNN is ported in embedded platforms like NVIDIA Jetson Nano. We built a CNN with 29 layers, see [fig_ref] Figure 5: Architecture of YOLOv2 Neural Network [/fig_ref]. This is to establish a light-weight model to fit the real-time implementation of CNN inference also in low-cost embedded platforms, such as those of IoT nodes. The neural network layers include the input layer, middle layers, and subnetwork of YOLOv2 layers. The proposed approach starts with the input image layer, which introduces the input image with a size of (224 × 224 × 3) for our detector. A set of middle layers was used, which includes batch normalization, convolutional, max-pooling, and Relu (rectified linear unit) layers. Convolutional layers were used to map the features for the images. The size of the filter was set to . It defines the height and width of the regions in the input image. Batch normalization layers were used to regularize the model and eliminate the overfitting problem. ReLU activation functions were utilized to introduce the non-linearity to the neural network.
Maxpooling layers were used to downsample the images into pooling regions. We applied (2 × 2) for the size of pooling with a stride of (2 × 2) for all max-pooling layers in a neural network. 'ReLU_5' was used as the feature extraction layer. This is to extract the features from neural network layers and then given as input to YOLOv2 subnetwork layers. YOLOv2 layers were used in this detector which constructs YOLOv2 detection network. YOLOv2 Subnetwork consists of a batch of layers that include convolutional (yolov2cov), batch normalization (yolov2Batch), ReLU (yolov2ReLU), transform, and output layers. The transform layer was utilized in YOLOv2 detector to stabilize the network for object localization. This layer transforms the raw CNN output into a form required to produce object detections. YOLOv2 output layer was used which refines the location of bounding boxes to the detected objects. The model was examined with a neural network analyzer and reported zero errors.
## Training
The designed network was trained with two different datasets of thermal images. Dataset I consists of 775 thermal images of humans captured in various scenarios while walking, running, or sneaking and in different body positions, as well as different motion speeds, maximizing the simulated conditions for detecting people in the surveillance and monitoring areas. These images were collected from different sources on the internet. Dataset II consists of 800 images. These images are infrared images that were created by FLIR company for thermal cameras. We used ground truth labeler application in MATLAB for labeling the persons in the thermal images . We split the images into 70% for training, 20 for validation, and 10% for testing for each dataset. The model has been trained with stochastic gradient descent (sdgm). The learning rate parameter in the training option was used to control the model change in response to the error. We started the learning rate with 10 -2 . However, we noticed that the model was unstable during the training process. The learning rate was fine-tuned at 10 -3 , and the loss curve for mini-batch was steady with small fluctuation, see [fig_ref] Figure 6 a: Mini-Batch Loss Curve before fine-tuning, b Mini-Batch Loss Curve after fine-tuning [/fig_ref]. [fig_ref] Table 1 Training: where [/fig_ref] shows Training Hyper-Parameters for the proposed neural network.
## Algorithm for distancing classification
We also implemented code in MATLAB to work with bounding boxes of a detected person in the thermal images. This code classifies and decides if persons in the image are within safe distancing or not. We assigned a green color for safe social distancing and red color for unsafe social distancing for the bounding boxes. First, we find the number of persons in the images. If it is one person, a green color is assigned for a bounding box of detected persons. When we have two or more persons, then color is decided from the function which is called find Color. This function will determine if the bounding box is 2 or more and in addition to that, it will calculate the distance between the centers of bounding boxes for the detected person. The center points, C (x, y) of bounding boxes is measured using the equation as seen in Eq..
where: C is the center point of the bounding box. X min and X max are the minimum and maximum values for the corresponding width of the bounding box. Y min and Y max are the minimum and maximum values for the corresponding height of the bounding box.
To measure the distance C 1 (X max -X min ), and C 2 (Y max -Y min ), between the center of each bounding box, we used the Euclidean formula, see Eq. (4), where the distance between pixels is translated in a metric distance (knowing the range and field of view covered by the camera) and then compared to a threshold value. In case of finding-color function detects two bounding boxes and the distance is less than the threshold value, these boxes will have a red color. If this function detects two bounding boxes and the distance is more than the threshold value, the color will be green for these boxes. provides the measured distance (D) between the center of each bounding box for a detected person.
where: D is the distance between the centers of bounding boxes.
Closed-circuit television (CCTV) cameras are installed in such a way that they provide angle views on the ground plan. To calculate the distance between the people effectively, a top view of the ground plane is required. This can be performed by applying a homography transformation to the four points coordinates in the angled view. These four points can be transformed as shown in the Eq. (5). To estimate the distance between people in the real world, the distance is calculated between the individuals using Eq. 4 and four points coordinates with homography matrix value. This distance is then scaled by factor S to have the real-world distance between the individuals. The scaling factor S is obtained by measuring a number of pixels in an image that represents 1 m in the real-world.
[formula] (3) C(X, Y) = X min + X max 2 , Y min + Y max 2 ,(4)D(C 1 , C 2 ) = √ (X max − X min ) 2 + (Y max − Y min ) 2 , [/formula]
# Experimental results
The technique proposed in Sect. 4 was examined with two testing datasets to evaluate the capability of detection and localization of persons in the thermal images. These datasets have been made challenging, which encountered a realistic situation by capturing body temperature on people from real thermal cameras. Motivating to that, we selected
[formula] (5) ⎡ ⎢ ⎢ ⎢ ⎣ X corn.top Y corn.top 1 ⎤ ⎥ ⎥ ⎥ ⎦ = M * ⎡ ⎢ ⎢ ⎣ X corn.ang Y corn.top 1 ⎤ ⎥ ⎥ ⎦ , [/formula]
these datasets for our experiments. YOLOv2 and distance classification algorithms were applied to these thermal images. YOLOv2 model detects people and provides the bounding box information. After people detection, the Euclidean distance between each detected centroid pair is computed using the detected bounding box and its centroid information based on dimensions of (x, y) for each bounding box. As a further step, we designed and trained R-CNN and Fast R-CNN models for people detection with the same training datasets. We compared these R-CNN and Fast R-CNN architectures with the technique proposed in Sect. 4 using the same testing datasets of thermal images.
To measure the efficiency of the proposed approach, the parameters on which the three architectures are evaluated include accuracy, precision, and recall values using confusion matrix criteria, see Eq.. Based on the results from these experiments, the new proposed detector showed good performance for people detection, social distancing classification on thermal images in both datasets, see [fig_ref] Figure 8: Sample Images from a, b Dataset I, c, d Dataset II [/fig_ref]. It achieved significant results with two datasets and overcomes R-CNN and Fast R-CNN detectors see [fig_ref] Table 2: Performance of this work vs [/fig_ref].
YOLOv2 neural network looks the entire image at one time, unlike R-CNN and Fast R-CNN methods which see only the generated region proposals. Therefore, the proposed technique reduces the problem of background mistakes and improves the localization of detected persons in the image. In addition to that, the proposed approach shows better accuracy in comparison to other methodologies [bib_ref] Two-person interaction recognition via spatial multiple instances embedding, Sener [/bib_ref] [bib_ref] Real-time social distancing detector using social distancingnet-19 deep learning network, Rinkal [/bib_ref] , and [bib_ref] Deep learning based safe social distancing and face mask detection in public..., Yadav [/bib_ref] , see [fig_ref] Table 3: Performance of the proposed approach vs [/fig_ref]. According to these [bib_ref] Two-person interaction recognition via spatial multiple instances embedding, Sener [/bib_ref] 93.3 Rinkal et al. [bib_ref] Real-time social distancing detector using social distancingnet-19 deep learning network, Rinkal [/bib_ref] 92.8 Yadav et at [bib_ref] Deep learning based safe social distancing and face mask detection in public..., Yadav [/bib_ref] 91 results, the methodology proposed in Sect. 4 is a promising one for people detection and social distancing classification on thermal images.
where TP stands for the number of true positive; TN stands for the number of true negative; FP stands for the number of false positive; FN stands for the number of false negative. Experiments were carried on a computer with Intel® Core TM I3-6006U CPU @ 2 GHz. MATLAB2020a was adopted with its built-in applications such as Ground Truth Labeler, Neural Network Designer. Jetson nano was used as an embedded system test platform in Sect. 6.
## Real-time measurement of this work vs other object detectors
The main objective of this research is to detect and recognize individuals in real-time. We have to monitor and track people's movements by utilizing a video camera. The invention and evolution of deep learning have improved the traditional ways of object detection and recognition systems. This technology is applied in several applications to identify and locate the objects in images, and it showed encouraging results for real-time detection [bib_ref] Real-time object detection using deep learning: a survey, Shubham [/bib_ref]. To understand further, experiments were carried to compare the proposed approach and other deep learning detectors such as R-CNN and Fast R-CNN. MATLAB was used with our test bench of videos that were captured from a thermal camera. The three models run simultaneously while frames per second were calculating for each model. Based on results from this experiment, the neural network proposed in this work runs faster than the other two detectors (Fast R-CNN and R-CNN). Note that this work showed better results for real-time detection comparable to the method, which proposed YOLOv3 detector. It is observed that R-CNN and Fast R-CNN have low frames per second, which make them not suitable for real-time applications. shows the comparative realtime detection of this work versus other deep learning object detectors. MATLAB environments and third-party packages were utilized to generate the C code of the proposed approach in the NVIDIA device. GPU coder was used for converting MATLAB code into an optimized CUDA code. Compute unified device architecture or CUDA is an extension of C programming language which is designed for NVIDIA frameworks. Jetson nano was connected to the host computer using an ethernet cable. MATLAB coder was utilized to generate C code to Jetson nano. We used a parallel The comparison of this work vs other competing deep learning detectors (R-CNN, Fast R-CNN, and YOLOv3) for real-time detection computing toolbox to solve complex computational and data processing using a multicore processor and GPU. A deep learning toolbox was utilized to provide a framework to implement the neural network and algorithms in Jetson nano. GPU support package for NVIDIA is used to deploy the proposed algorithms in Jetson nano. This support package application enables the communication remotely to the targeted NVIDIA hardware. Embedded coder was used for code generation on Jetson nano. This tool improves the code generation on hardware effectively. A JetPack developer AI tool was installed in the NIVIDIA device. It is an environment variable application which is to be applicable for code generation of the proposed deep learning architecture in Jetson nano. Microsoft visual studio 2019 was installed as a compiler generate GPU code in Jetson nano. CUDA Deep Neural Network libraries were used to accelerate primitives for neural network architecture.
## Test the proposed algorithm on jetson nano
The proposed algorithm was deployed in Jetson nano and run as a standalone application to evaluate its performance. A Raspberry Pi camera model V2 was exposed to another personal computer that simulated a number of videos that were captured from a thermal camera. While the proposed algorithm was running in the NIVIDIA device, we recorded various parameters. We measured the average frames per second for the proposed approach on Jetson nano and we compared the achived results with other different methods, see [fig_ref] Table 4: The real-time measurement for the proposed approach vs other methods [/fig_ref]. According to the results from this experiment, our approach showed the best result for the real-time which reached up to 27 fps.
We measured also the power consumption for Jetson nano. We removed all Jetson nano accessories such as a mouse, monitor, and keyboard. We measured the power consumption at 1.24 W when the deployed algorithm is off. When the distance classification algorithm was executed, the power consumption was recorded at 4.4 W. [fig_ref] Table 5: Power consumption measurement in different scenarios [/fig_ref] shows the power measurement of the NVIDIA device in different scenarios.
We recorded the measurement of Graphics Processing Unit (GPU) and the Central Processing Unit (CPU) % resource utilization in Jetson device. The GPU is designed to process the graphic operations and the CPU runs the operating system and applications. These characteristics are essential to assess its computation processing. The table shows the measured values for the GPU and CPU processors while the proposed algorithm was executed in the targeted hardware. Moreover, we measured the temperature of Jetson nano while the proposed approach was in execution. The temperature was measured at 54.5 °C for GPU and 54.1 for CPU, see [fig_ref] Table 6: The % resource utilization and temperature measurement for the GPU and CPU... [/fig_ref]. Further to our experiments, we measured the memory size for the deployed [fig_ref] Figure 1: The schematic diagram for R-CNN detector [/fig_ref] Comparison of the proposed approach verses other pretrained models in terms of memory size algorithm in Jetson nano, which is 14 MB. This is the advantage of the proposed approach in comparison to the other pre-trained models such as VGG16, Alexnet, and Resnet50 in method [bib_ref] Real-time social distancing detector using social distancingnet-19 deep learning network, Rinkal [/bib_ref]. [fig_ref] Figure 1: The schematic diagram for R-CNN detector [/fig_ref] shows the comparison between the proposed approach with other pre-trained models in terms of memory size. These architectures use massive CNN layers which need a large disk size for the deployment on the targeted embedded system. This could affect real-time performance while the algorithm is running on low-cost embedded devices. This is the advantage of the proposed approach versus the pre-trained CNN models and it can be superior for real-time detection.
## Distributed surveillance video system for social distancing
Video surveillance cameras are an effective monitoring system for authorities to visualize how people are acting and their compliance with social distancing. In our research, we implemented a distributed surveillance camera system based on embedded devices. The proposed system is composed of multiple Jetson nanodevices with, each combined with a video camera. Each camera is connected to one Jetson nanodevice, which represents a smart node in the system architecture. Jetson nanodevices were upgraded with Wi-Fi for internet connection. All Jetson nanodevices were connected to the computer through a router using a static IP address from each node. The router directs the video streaming of each node and serves as a networking device to the centralized surveillance management system (personal computer), see [fig_ref] Figure 1: The schematic diagram for R-CNN detector [/fig_ref].
MobaXterm application in Windows 10 was used to establish communication between the centralized surveillance management system (personal computer) and Nvidia Jetson nano nodes. The communication was established using an OpenSSH application with respect to the defined IP address of each node. Secure Shell or OpenSSH is a remote information communication technology protocol that allows users to control and transfer data between computers. The system is built with a multi-access point of IP addresses through OpenSSH sessions in the MobaXterm software. Each OpenSSH session communicates with Jetson nano node through its defined IP address. The latency time was measured between the computer and Jetson boards at 0.3 ms. The proposed approach is suitable for a distributed surveillance system that can visualize people detection and social distancing classification on thermal images from several Jetson nanodevices in one centralized surveillance management system.
## Conclusion and future work
This research presented an intelligent surveillance system for people tracking and social distancing classification based on thermal images. The proposed technique achieved promising results for people detection in terms of evaluation the accuracy and precision of the detector comparable to the other deep learning models. A specific algorithm was implemented on bounding boxes to distinguish between safe and unsafe conditions, respectively, marking as green and red the bounding box for detected persons. The proposed technique showed better results for real-time performance vs other object detectors. The proposed approach can be implemented in a distributed video surveillance system; indeed, it is a suitable solution for the authorities to visualize the compliance of people with social distancing and at the same time screening their body temperature. In the future, we will utilize this methodology on mobile cameras, e.g., mounted on an autonomous drone system, and hence drones are simpler to operate and more effective to capture fast actions of the detected objects from different angles. We will extend our research to use and experiment people detection by also applying 3-D dimensions to have three parameters (x, y, z), in which we can perceive uniform distribution distance in the entire image and eliminating the perspective effect. In addition to that, the newly released YOLOv4 detectorwill be also considered.
Funding Open Access funding provided by Università di Pisa.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated [fig_ref] Figure 1: The schematic diagram for R-CNN detector [/fig_ref] Smart Surveillance distributed video system for people detection and social distancing classification otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.
[fig] Figure 1: The schematic diagram for R-CNN detector [/fig]
[fig] Figure 2: The schematic diagram for Fast R-CNN detector [/fig]
[fig] Figure 3: Schematic diagram for YOLO: input image which splits into S×S grids, each grid predicts the bounding boxes and the confidence scores and finally, the score encodes the probability with bounding box on the detected class [/fig]
[fig] ∧: C i is the box confidence score of the box j in cell i . noobj weights down the loss when detecting background. object appears in the cell i , otherwise 0. p i (c) is the classification loss. p i (c) is the conditional class probability for class c in cell i. [/fig]
[fig] Figure 5: Architecture of YOLOv2 Neural Network [/fig]
[fig] Figure 6 a: Mini-Batch Loss Curve before fine-tuning, b Mini-Batch Loss Curve after fine-tuning [/fig]
[fig] Figure 8: Sample Images from a, b Dataset I, c, d Dataset II [/fig]
[fig] 6: Implementation of the proposed approach on embedded hardware .1 Jetson nano (NVIDIA device) Jetson nano NVIDIA system is a low-cost embedded device but a powerful computer. It costs approximately $100[33]. Jetson nano can run various advanced neural networks including a full version of most popular deep learning (DL) and machine learning (ML) frameworks such as Pytorch, Caffe, Keras, and TensorFlow. This embedded device uses TensorRT accelerator libraries which include Jetpack packages. Jetson nano is suitable for real-time applications in different scenarios and is capable to process multiple highdefinition video streams.Jetson nano includes CPU QUAD-core ARM A57 at 1.43 GHz and GPU 128-core Maxwell. The memory of the device is 4 GB, 4-bit, LPDDR4 25. GB/s. Jetson nano has a USB 2.0 Micro-B, 4 × USB 3.0. A standard camera module with 8 M pixel resolution has been used in our experiments. The camera was connected to the camera serial interface (CSI) in Jetson nano. The trained neural network model and social distancing classification algorithm defined in Sect. 4 has been deployed in Jetson nano and it runs as a standalone application. [/fig]
[table] Table 1 Training: where: X corn.ang and Y corn.ang represent the pixel coordinates of one of the four points in the CCTV view image X corn.top , Y corn.top represent the same point after transformed to the top view. M is the homography matrix. [/table]
[table] Table 2: Performance of this work vs. other object detectors [/table]
[table] Table 3: Performance of the proposed approach vs. [/table]
[table] Table 4: The real-time measurement for the proposed approach vs other methods [/table]
[table] Table 5: Power consumption measurement in different scenarios [/table]
|
10.1038/srep04551
|
CCBY
|
3983607
|
24722502
|
s2orc_pubmed_articles
|
Enhanced Antibacterial Activity of Silver Nanoparticles/Halloysite Nanotubes/Graphene Nanocomposites with Sandwich-Like Structure
A sandwich-like antibacterial reagent (Ag/HNTs/rGO) was constructed through the direct growth of silver nanoparticles on the surface graphene-based HNTs nanosheets. Herein, various nanomaterials were combined by adhesion effect of DOPA after self-polymerization. Ag/HNTs/rGO posses enhanced antibacterial ability against E. coli and S. aureus compared with individual silver nanoparticles, rGO nanosheets or their nanocomposites.H alloysite nanotubes (HNTs), a kind of aluminosilicate clay, exhibit novel physical and chemical properties due to predominantly hollow tubular structure, thereby providing opportunities for advanced applications in many fields. A significant advantage of HNTs is their easiness to obtain and low price in the comparison with other nanomaterials, like carbon nanotubes 1-4 . Interestingly, HNTs discovered and used by our group are provided with higher purity and quality, which have been developed into a vast of applications, for instance, phase change material 5 , enzyme immobilization 6 , adsorbents 7 and membranes 8 .Antibacterial reagents, including antibiotics 9 , metal ions 10 , enzymes 11 , and quaternary ammonium compounds 12 have been extensively used to defend the public health in our daily life. However, the above materials have some drawbacks, such as, antibiotic resistance, environmental damages, relatively high cost 13 . Currently, a number of attempts have been made to develop novel, efficient and environment-friendly antibacterial materials. For instance, graphene and graphene related materials were extensively studied which exhibit strong antibacterial activity[14][15][16]. Physical damages on cell membranes are likely to occur in consequence of membrane stress induced by sharp edges of graphene nanosheets, thereby contributing to the loss of bacterial membrane integrity and the leakage of RNA 14 . In addition, silver-based nanocomposites synthetized by loading silver on the carrier were also often reported as antibacterial materials 17,18 . In our previous studies, copper19and silver 8,20 respectively were loaded on HNTs for reducing their release to fabricate antibacterial nanocomposites which showed good antibacterial performance against Gram-negative bacteria (Escherichia coli) and Gram-positive bacteria (Staphylococcus aureus). However, the antibacterial performance of individual graphene nanosheets is limited compared with silver which are easy to agglomerate contributing to the decrease of the antibacterial effects. In early approaches, graphene-AgNPs-based nanocomposites were fabricated by a number of workers which show enhanced antibacterial properties 15,21-24 . It is supposed that the direct growth of silver nanoparticles on graphene is quite capable of strengthening their antibacterial performance. Therefore, the increase of the surface area of nanocomposites may provide opportunities for the direct growth of silver nanoparticles and thus enhance the antibacterial effect deriving from the direct attachment to bacteria.Bioadhesive proteins of marine organisms, such as mussels(Figure 1a), have attracted the public attention in the past few decades in consequence of the formation of the adhesive force on variety kinds of substrates even in wet environments 25 . The adhesive ability is attributed to L-3, 4-dihydroxyphenylala-nine (DOPA), which was discovered at a high level in adhesive proteins 26 . Lee et al.27identified dopamine(Figure 1b)as a structural mimic of DOPA and demonstrated the self-polymerization of dopamine onto a wide range of inorganic or organic materials even ''nonsticking surfaces'' to form thin, surface-adherent polydopamine films.Herein, HNTs were introduced to graphene-AgNPs-based nanocomposites combining the adhesive ability of DOPA to fabricate a sandwich-like nanomaterial based on the interpenetrative nanocomposites. This new approach is designed to reinforce the synergistic effects between graphene and HNTs in the interest of the direct OPEN SUBJECT AREAS: NANOBIOTECHNOLOGY BIOMATERIALS
growth of silver nanoparticles. [fig_ref] Figure 1 |: Fabrication process for sandwich-Like, silver nanoparticles/halloysite nanotubes/graphene nanocomposites [/fig_ref] illustrates the typical procedure of the construction of the sandwich-like nanomaterial. The whole procedure was accomplished under a mild environment where there are no environmentally hazardous chemicals involved in the reaction and no temperature or pressure control instrument was needed as well. DOPA deposited on the surface of HNTs through self-polymerization. The sandwich-like nanomaterial was constructed by a one-step method in which silver nanoparticles were formed via the reduction of DOPA and located on the surface of HNTs or graphene oxide (GO) nanosheets. By this means, HNTs constantly locate between GO nanosheets during the concentration and drying process. Hence, the GO nanosheets were partially reduced to graphene (rGO) and the reunion of twodimensional graphene tends to fall off. For this reason, the specific surface area of resulted nanocomposites is inclined to increase in contrast to individual rGO or GO nanosheets in dry state. Consequently, this approach may provide an opportunity to prepare high-surface-area nanocomposites with antibacterial performance benefited from the synergistic effects from different nanomaterials. [fig_ref] Figure 2 |: TEM images of [/fig_ref] gives Transmission electron microscopy (TEM) images of the as-prepared sandwich-like nanomaterial (Ag/HNTs/rGO). As shown, HNTs we use [fig_ref] Figure 2 |: TEM images of [/fig_ref] have an admirable hollow tubular structure and wrinkles of GO nanosheets [fig_ref] Figure 2 |: TEM images of [/fig_ref] are clearly visible indicating the GO nanosheets are extremely thin. The resulted GO suspension is homogeneous, transparent and typically golden or brown colored. From [fig_ref] Figure 2 |: TEM images of [/fig_ref] and [fig_ref] Figure 2 |: TEM images of [/fig_ref] , it can be seen that onedimensional HNTs intricately distribute between different GO nanosheets and silver nanoparticles directly grown on the surface of both HNTs and GO nanosheets. In [fig_ref] Figure 2 |: TEM images of [/fig_ref] , it is found that the silver nanoparticle size ranges from 5-15 nm. A HRTEM image of single entity silver nanoparticles is shown in [fig_ref] Figure 2 |: TEM images of [/fig_ref]. The crystal lattice of silver nanoparticles are resolved in most regions and the fringe spacing is found to be 0.23 nm corresponding to the (111) crystal plane [bib_ref] A Facile and Novel Synthesis of Ag-Graphene-Based Nanocomposites, Pasricha [/bib_ref] [bib_ref] Growth of silver nanocrystals on graphene by simultaneous reduction of graphene oxide..., Tang [/bib_ref] [bib_ref] Preparation, characterization and antibacterial properties of silver-modified graphene oxide, Ma [/bib_ref]. In addition, energy dispersive spectrum (EDS) of Ag/HNTs/rGO is shown in [fig_ref] Figure 3 |: Energy dispersive spectrum of Ag/HNTs/rGO nanomaterial [/fig_ref] which revealed the existence of various elements in the nanocomposites. Interestingly, a paper-like antibacterial film was prepared from Ag/HNTs/rGO by incorporation with a small amount of polyethersulfone, which shows excellent flexibility, convenience to use and may have some potential specific applications [bib_ref] Graphene-based antibacterial paper, Hu [/bib_ref].
# Results and discussion
Fourier transform infrared spectroscopy (FTIR) of Ag/HNTs/ rGO was performed to justify whether GO nanosheets was reduced to rGO during the construction process of the sandwich-like nanomaterial. It can be seen from [fig_ref] Figure 4 |: FTIR spectra of [/fig_ref] , the curve of GO represented a considerable absorption peak around 1720 cm 21 , which ascribes to C5O stretching vibrations in the carboxyl group of GO after the oxidation of natural graphite. However, after the construction of Ag/ HNTs/rGO, the characteristic peak around 1720 cm 21 vanished and the whole curve of Ag/HNTs/rGO was almost identical to that of natural graphite. The two small peaks around 2930 cm 21 and 2840 cm 21 of Ag/HNTs/rGO are supposed to be in the consequence of the partial destruction of carbon skeleton of natural graphite during the oxidation process, which tends not to be restored thoroughly by reduction. These results indicated that GO nanosheets were reduced to rGO by DOPA during the fabrication process of Ag/ HNTs/rGO.
To investigate the surface area change of Ag/HNTs/rGO, the specific surface area (BET) was measured using the adsorption of N 2 at the temperature of liquid nitrogen. The specific surface area (SBET) for Ag/HNTs/rGO showed an obvious enhancement with a value 287.10 m 2 /g. Nevertheless, the SBET value for Ag/HNTs and Ag/ rGO were 59.6 m 2 /g and 128.92 m 2 /g, respectively. That is to say, a high-surface-area substrate was synthetized as our initial design.
The antibacterial properties of Ag/HNTs/rGO was studied against Gram negative bacterial strains, E. coli and Gram positive bacterial strains, S. aureus as bacterium models. Minimal inhibitory concentration (MIC) was observed by the tube double dilution method to evaluate the effectiveness of as-prepared nanomaterials. In the tube double dilution method, MIC is defined as the lowest concentration (in mg/ml) of the antimicrobial agent that prevents visible growth of a microorganism under defined conditions. The MIC results of the sandwich-like nanomaterial, Ag/HNTs/rGO and the controls including Ag, Ag/HNTs, rGO/Ag is shown in [fig_ref] Table 1 |: MIC of as-synthesized different nanomaterials to E [/fig_ref]. Note that all the mentioned Ag is at the nanoscale. The MIC of all the prepared nanomaterials followed the order: rGO . Ag . Ag/HNTs . Ag/ rGO . Ag/HNTs/rGO. Compared to all the control groups, Ag/ HNTs/rGO exhibited a lowest MIC of 2 mg/ml indicating excellent antibacterial effect. The phenomenon may attribute to the synergistic antibacterial effect from Ag and rGO. Furthermore, the enhanced surface area of Ag/HNTs/rGO is another important factor for the excellent antibacterial effect which may reduce the reunion of silver nanoparticles. Generally, the aggregation of antibacterial nanomaterial would greatly decrease their antibacterial activities [bib_ref] Facile synthesis of silver@graphene oxide nanocomposites and their enhanced antibacterial properties, Xu [/bib_ref]. In this work, silver nanoparticles were loaded onto the surface of the sandwich-like nanomaterial and in-situ reduced by DOPA. Thus, the aggregation of silver nanoparticles was effectively decreased.
Additionally, the sandwich-like nanomaterial showed excellent antibacterial performance through bacteriostasis rate test. It can be seen in [fig_ref] Figure 5 |: Photographs of bacterial colonies formed by [/fig_ref] that colonies of E. coli or S. aureus without treated by Ag/HNTs/rGO are observed as small white dots and almost fill the whole plate as shown in [fig_ref] Figure 5 |: Photographs of bacterial colonies formed by [/fig_ref] , c. In the comparison, in [fig_ref] Figure 5 |: Photographs of bacterial colonies formed by [/fig_ref] , d few colonies can be visibly observed on the plates treated by Ag/ HNTs/rGO. The superior bacteriostasis rate nearing 100% against both E. coli and S. aureus demonstrate the effectiveness and popularity of the antibacterial properties of Ag/HNTs/rGO.
The morphology changes of E. coli cells were further analyzed by using TEM. As shown in [fig_ref] Figure 6 |: TEM images of E [/fig_ref] , the original E. coli cells were evenly fuscous with well-defined membranes. However, after treated with Ag/HNTs/rGO for 16 h, the E. coli cells were ruptured, pale along with the release of cytoplasm and a large portion of the cells was decomposed as can be seen in [fig_ref] Figure 6 |: TEM images of E [/fig_ref]. Such irreversible cellular damage demonstrates the effectiveness of the antibacterial properties of Ag/HNTs/rGO. The optical density at 600 nm (OD 600 ) of E. coli suspension treated with different prepared reagents in test tubes for 5 h and 12 h were determined, which is a widely used method to determine the growth of bacteria for the assessment of antibacterial abilities [bib_ref] GFAJ-1 Is an Arsenate-Resistant, Phosphate-Dependent Organism, Erb [/bib_ref] [bib_ref] Possible Role for the Essential GTP-Binding Protein Obg in Regulating the Initiation..., Vidwans [/bib_ref]. Generally, the lower optical density at 600 nm of bacteria suspension after cultivation for a definite time means the better antibacterial ability of antibacterial reagent. As can be seen from [fig_ref] Figure 7 |: Optical density at 600 nm [/fig_ref] , for different prepared antibacterial reagent, the OD 600 values followed the order: control . rGO . Ag . Ag/HNTs . Ag/rGO . Ag/ HNTs/rGO, which is almost the same with the MIC of them. Silver nanoparticles which grow directly from Ag/HNTs, Ag/rGO, and Ag/ HNTs/rGO showed lower OD 600 after both 5 h and 12 h cultivation. This phenomenon can be on account of the decrease of conglomeration of silver nanoparticles. Additionally, each group of data presents a good stability with a very minor anova. It is important to emphasize that the sandwich-like nanomaterial, Ag/HNTs/rGO, showed the lowest OD 600 especially after 12 h cultivation in contrast to other antibacterial reagents. These results indicated that Ag/HNTs/rGO possesses a better ability in resisting bacteria growth agreeing strongly with the MIC results, which probably benefits from enhanced surface area of Ag/HNTs/rGO.
In summary, a sandwich-like antibacterial reagent was constructed through the direct growth of silver nanoparticles on the surface graphene-HNTs-based nanocomposites. Different nanomaterials were combined by adhesion effect of DOPA after self-polymerization. A series of experiments were performed and the results showed that Ag/ HNTs/rGO has enhanced antibacterial ability against E. coli and S. aureus compared with individual silver nanoparticles, rGO nanosheets or their nanocomposites. It is anticipated that the sandwich-like antibacterial nanomaterial may be used in some potential antibacterial applications, antibacterial ultrafiltration membranes used for water treatment [bib_ref] Preparation and characterization of HPEI-GO/PES ultrafiltration membrane with antifouling and antibacterial properties, Yu [/bib_ref].
# Methods
Materials. Graphite powders (spectral pure) were purchased from Sinopharm Chemical Reagent Co., Ltd., and were used as received. Halloysite clay from Henan Province (China) was milled and sieved to obtain halloysite nanotubes (HNTs). Tris (hydroxymethyl) aminomethane (AP) and dopamine (AP) were purchased from Sigma Aldrich. Polyethersulfone (PES, Ultrason E6020P with Mw 5 58 kDa) was obtained from BASF, Germany. All the other chemicals (analytical grade) were obtained from Tianjin Kermel Chemical Reagent Co., Ltd., China, and were used without further purification. The used water is deionized water.
Fabrication of sandwich-like nanomaterial (Ag/HNTs/rGO). GO nanosheets were prepared by oxidizing natural graphite based on an improved method [bib_ref] Improved Synthesis of Graphene Oxide, Marcano [/bib_ref] and was used in aqueous solution. Dopamine, 200 mg, was dissolved in a 100 mL Tris-HCl solution. HNTs, 0.3 g, were dispersed in the above solution by ultrasonic to react for 16 h under stirring at ambient condition. Subsequently, the treated particles were washed by deionized water thoroughly and dispersed in a 250 ml GO aqueous solution (2 mg/mL). Then, Tollens' reagent was added to the mixed solution to react overnight at room temperature under stirring. Finally, the product was collected by centrifugation and washed with deionized water and ethanol, respectively. The wet product was vacuum dried at room temperature for 24 h.
Fabrication of Ag/HNTs/rGO film. As prepared Ag/HNTs/rGO particles, 0.3 g, were added in 5 mL N, N-Dimethylacetamide (DMAc) and dispersed thoroughly by stirring. Then PES, 0.9 g, was added under stirring and the stirring process continued at least 24 h. After PES completely dissolved in the solution, the mixture was poured into a home-made glass groove with about 3 mm deepness. The solvent was evaporated thoroughly under an open environment at room temperature and the Ag/ HNTs/rGO film was peeled off from the glass carefully. The resulted Ag/HNTs/rGO has a 20 3 60 mm 2 area with about 1 mm thickness.
Characterization. FTIR spectra were performed at 2 cm 21 resolution with Thermo Nicolet IR 200 spectroscope (Thermo Nicolet Corporation, USA). Typically, 64 scans were signal-averaged to reduce spectral noise. The spectra were recorded in the 400-4000 cm 21 range using KBr pellets. A FEI model TECNAI G 2 transmission electron microscope (FEI, USA) was used to study the morphology of Ag/HNTs/rGO. The samples were dispersed in solvent with the aid of ultrasound. The suspended particles were transferred to a copper grid (400 meshes) coated with a strong carbon film and dried. Energy Dispersive System (EDS) was carried out in a JEOL JSM-7500F FE-SEM with samples sputtered with gold. Bacteriostasis rate. A 100 mL of the above bacterial suspension after being diluted 10 5 times was cultured on an agar plate. After incubated at 37uC for 16 h, the numbers of colonies were counted and the bacteriostasis rate was determined by dividing the number of colony-forming units (CFU) which are killed by antibacterial reagent by the number of CFU of the control group.
Transmission Electron Microscopic (TEM) Measurements: E. coli cells treated with Ag/HNTs/rGO solution for 30 min were fixed with 2.5% glutaraldehyde. The cells were washed with PBS and then postfixed with 1% aqueous OsO 4 (Fluka) for 1 h and washed again twice with PBS. The cells then were dehydrated through ethanol series (70% for 15 min, 90% for 15 min, and 100% for 15 min twice) and embedded in Epon/Araldite resin (polymerization at 65uC for 15 h). Thin sections (90 nm) containing the cells were placed on the grids and stained for 1 min each with 4% uranyl acetate (151 acetone/water) and 0.2% Raynolds lead citrate (water), air dried, and examined under a transmission electron microscope (FEI, USA).
[fig] Figure 1 |: Fabrication process for sandwich-Like, silver nanoparticles/halloysite nanotubes/graphene nanocomposites (Ag/HNTs/rGO). [/fig]
[fig] Figure 2 |: TEM images of (a) HNTs, (b) GO nanosheets and (c), (d), sandwich-like nanomaterials at different magnifications. (e) HRTEM image of single entity silver nanoparticles with fringe spacing. (f) Photograph of antibacterial film prepared from Ag/HNTs/rGO. www.nature.com/scientificreports SCIENTIFIC REPORTS | 4 : 4551 | DOI: 10.1038/srep04551 [/fig]
[fig] Figure 3 |: Energy dispersive spectrum of Ag/HNTs/rGO nanomaterial. [/fig]
[fig] Figure 4 |: FTIR spectra of (a) natural graphite, (b) GO and (c) Ag/HNTs/ rGO. [/fig]
[fig] Figure 5 |: Photographs of bacterial colonies formed by (a), (b) E. coli and (c), (d) S. aureus: (a), (c) control groups, (b), (d) treated with Ag/HNTs/ rGO. [/fig]
[fig] Figure 6 |: TEM images of E. coli cells before (a) and after (b) treated with Ag/HNTs/rGO. [/fig]
[fig] Figure 7 |: Optical density at 600 nm (OD 600 ) of bacterial suspension treated with different reagents for 5 and 12 h. [/fig]
[table] Table 1 |: MIC of as-synthesized different nanomaterials to E. coli (mg/ml) [/table]
|
10.1371/journal.pone.0230797
|
CCBY
|
7100954
|
32218603
|
s2orc_pubmed_articles
|
A multi-state model analysis of the time from ethical approval to publication of clinical research studies
BackgroundResults of medical research should be made publicly available in a timely manner to enable patients and health professionals to make informed decisions about health issues. We aimed to apply a multi-state model to analyze the overall time needed to publish study results, and to examine predictors of the timing of transitions within the research process from study initiation through completion/discontinuation to eventual publication.MethodsUsing a newly developed multi-state model approach, we analysed the effect of different study-related factors on each of the transitions from study approval to eventual publication, using a data set of clinical studies approved by a German research ethics committee between 2000 and 2002.ResultsOf 917 approved studies, 806 were included in our analyses. About half of the clinical studies which began were subsequently published as full articles, and the median time from study approval to publication was 10 years. Differences across model states were apparent; several factors were predictive of the transition from study approval to completion, while funding source and collaboration were predictive of the transition from completion to publication.ConclusionsThe proposed multi-state model approach permits a more comprehensive analysis of time to publication than a simple examination of the transition from approval to publication, and PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.
# Introduction
Results of medical research should be made publicly available in a timely manner to enable patients and health professionals to make informed decisions about health issues [bib_ref] Evidence based medicine: what it is and what it isn't, Sackett [/bib_ref]. However, there are several barriers inhibiting the translation of research knowledge into practice. Firstly, a large proportion of clinical studies are prematurely discontinued [bib_ref] Recruitment failure and futility were the most common reasons for discontinuation of..., Van Den Bogert [/bib_ref] [bib_ref] Time to publication of oncology trials and why some trials are never..., Chapman [/bib_ref] [bib_ref] Early termination of cardiovascular trials as a consequence of poor accrual: analysis..., Baldi [/bib_ref] [bib_ref] Discontinuation and non-publication of randomised clinical trials supported by the main public..., Amstutz [/bib_ref] [bib_ref] Premature Discontinuation of Prospective Clinical Studies Approved by a Research Ethics Committee-A..., Blumle [/bib_ref] [bib_ref] Prevalence, characteristics, and publication of discontinued randomized trials, Kasenda [/bib_ref] [bib_ref] Discontinuation and nonpublication of surgical randomised controlled trials: observational study, Chapman [/bib_ref]. Secondly, not all research findings are ultimately published in peer-reviewed journals [bib_ref] The "file drawer problem" and tolerance for null results, Rosenthal [/bib_ref]. A systematic review of the extent of non-publication of studies approved by research ethics committees (REC) or included in trial registries showed that only about half of all studies were eventually published in a peer-reviewed journal [bib_ref] Discontinuation and nonpublication of surgical randomised controlled trials: observational study, Chapman [/bib_ref] [bib_ref] Extent of Non-Publication in Cohorts of Studies Approved by Research Ethics Committees..., Schmucker [/bib_ref] [bib_ref] Increasing value and reducing waste: addressing inaccessible research, Chan [/bib_ref].
A Health Technology Assessment report found that the main reason given by investigators for not publishing their studies was 'lack of time or low priority', followed by 'results not important enough' and 'journal rejection' [bib_ref] Dissemination and publication of research findings: an updated review of related biases, Song [/bib_ref] [bib_ref] Authors report lack of time as main reason for unpublished research presented..., Scherer [/bib_ref]. Lack of financial and personnel resources are also frequently-cited reasons for non-publication [bib_ref] Barriers towards the publication of academic drug trials. Follow-up of trials approved..., Berendt [/bib_ref]. Specific study-related factors are also associated with whether or not a study is published, particularly the direction of the study results [bib_ref] Extent of Non-Publication in Cohorts of Studies Approved by Research Ethics Committees..., Schmucker [/bib_ref]. "Negative" study results, e.g. no significant difference between treatment arms, are less likely to be published than "positive" results, i.e. significant effects favouring the experimental treatment [bib_ref] Publication bias in clinical research, Easterbrook [/bib_ref] [bib_ref] Effect of the statistical significance of results on the time to completion..., Ioannidis [/bib_ref] [bib_ref] Publication bias: evidence of delayed publication in a cohort study of clinical..., Stern [/bib_ref] [bib_ref] Positive outcomes influence the rate and time to publication, but not the..., Sune [/bib_ref] [bib_ref] Reporting Bias G. Systematic review of the empirical evidence of study publication..., Dwan [/bib_ref]. This is also associated with time to publication, with positive results published on average about 1-3 years more quickly than negative results . Source of funding is also important; negative results are more likely to be published when studies are funded by non-profit institutions than when they are commercially funded [bib_ref] Non-publication and delayed publication of randomized trials on vaccines: survey, Manzoli [/bib_ref]. In a cohort of biomedical research studies approved by a German REC we showed that multicentre and international collaboration, large sample size, a declared study funding source and the sponsor's involvement in the study were also positively associated with subsequent publication [bib_ref] Fate of Clinical Research Studies after Ethical Approval-Follow-Up of Study Protocols until..., Blümle [/bib_ref] [bib_ref] Clinical research projects at a German medical faculty: follow-up from ethical approval..., Blümle [/bib_ref].
The 'selective publication' of studies results in a restricted and frequently over-optimistic body of evidence. Treatments are often erroneously accepted as effective, whereas null or adverse effects are underestimated. The problem is magnified when studies are combined in meta-analyses or systematic reviews, and can lead to inappropriate health care recommendations and ultimately non-optimal treatment that can prolong patients' suffering, cause harm and in the worst case, cost lives [bib_ref] Increasing value and reducing waste: addressing inaccessible research, Chan [/bib_ref] [bib_ref] Evidence-informed recommendations to reduce dissemination bias in clinical research: conclusions from the..., Meerpohl [/bib_ref] [bib_ref] The answer is 17 years, what is the question: understanding time lags..., Morris [/bib_ref].
The process of conducting and publishing a study can be divided into multiple event types or "multi-states" occurring over a period of time. All studies start in an initial state, e.g. REC approved, may pass through one or more intermediate states, and may finally result in an absorbing state, e.g. published. The process is typically visualized by a directed graph, often called a multi-state model, where the states are drawn as nodes and the possible state transitions are indicated by arrows [fig_ref] Fig 1: Illustration of the four-state model [/fig_ref]. For studies that have begun, two intermediate states are possible: a) the study is completed, i.e. the study has ended as planned and participants are no longer being followed up; or b) the study is discontinued (e.g. was stopped early due to adverse events). Either of these scenarios can then lead to eventual publication of the findings. This four-state model, originally developed in a Master's thesis written by one of the co-authors (TH) [bib_ref] Multi-state models for analyzing the publication process of clinical studies, Haag [/bib_ref] , allows for the systematic examination of variations in the total time from ethical approval to publication, distinguishing between the time required to conduct the study (i.e., study initiation to either study completion or discontinuation), and the time required to analyse, write up and publish the study findings (i.e., study completion/discontinuation to publication).
To demonstrate this novel analytic approach we conducted a secondary analysis of a data set collected for a previous investigation of study protocols submitted to a German REC [bib_ref] Fate of Clinical Research Studies after Ethical Approval-Follow-Up of Study Protocols until..., Blümle [/bib_ref]. In Germany (like in many countries), registration in a clinical trial registry is not mandatory, but all studies, regardless of their design, are required to be approved by an REC before they can legally begin. Using the aforementioned multi-state model, we analysed the effect of different factors on each of the illustrated transition intensities within the research process from study approval to eventual publication.
# Methods
## Study design and data
The data set we used to demonstrate our newly developed four-state model approach included all research protocols of human research studies submitted to and approved by the University of Freiburg REC between 2000 and 2002 [bib_ref] Fate of Clinical Research Studies after Ethical Approval-Follow-Up of Study Protocols until..., Blümle [/bib_ref]. The design of each study was determined according to predefined criteria (S1 . Study characteristics were extracted by a research assistant (FV or AH, see acknowledgments). In cases of uncertainty, the issue was discussed with the lead author to reach a consensus. The lead author cross-checked the other database entries. Subsequent publications were identified via electronic literature searches (AH, PO) and a survey sent to the applicants (AH, AB, actively supported by the REC). The study was piloted by investigating protocols approved by the REC in 2000. The search for publications was conducted in 2006, the survey to applicants was sent in 2007 (followed by a reminder three months later) asking them to confirm the publications that had been identified and to add further publications. It was also asked whether the project (a) had been completed as planned, (b) had been discontinued, (c) had not been started at all or at the local study site, or (d) was still ongoing, and if so its current status (i.e., recruitment closed, data collection completed, or preliminary results published) [bib_ref] Fate of Clinical Research Studies after Ethical Approval-Follow-Up of Study Protocols until..., Blümle [/bib_ref]. The results of the pilot project were published in For the present study, in order to comply with the survey information received, we defined the study initiation date as the year of REC approval, the study completion date as the year in which data collection was completed, the study discontinuation date as the year in which the study was discontinued, and the study publication date as the year of electronic publication of the first manuscript presenting findings of the study.
Data from REC files, incl. application, protocol, correspondence with the REC, and from the survey were recorded for several covariates potentially associated with the occurrence of state transitions. These were: (a) whether the study was an RCT (y/n); (b) sample size; (c) funding source (commercial, non-commercial or not reported); (d) industry involvement in the conduct and/or analysis of the study (y/n); (e) a composite variable of number of study centres and international collaboration (categorized as 'multi-centre, international collaboration', 'multi-centre, national collaboration' and 'single centre'); and (f) whether the study had a primary outcome (y/n). The only covariate for which there was a non-trivial amount of missing data was sample size; for these 37 (4.6%) studies the median number of participants (n = 120) was imputed.
For two covariates, funding source and industry involvement, information was also extracted from the final publication. There was occasional inconsistency between the REC protocol and the publication, likely because it was not an explicit requirement for the protocol to state the study funding source or (anticipated) involvement of the sponsor. In these cases we took the pragmatic approach of designating studies as (non-)commercially-funded if (non-) commercial funding was mentioned either in the protocol or the publication; the same applies for industry involvement.
# Statistical analysis
We aim to analyze the time needed to publish study results, and to estimate predictors of the timing of transitions within the research process from study initiation through completion/ discontinuation to eventual publication. To accomplish this, we employ time-to-event approaches, which make use of the information that (or whether) a certain event occurs and the time until which the transition to this event occurs. Thereby, the multi-state model [bib_ref] Multi-state models for event history analysis, Andersen [/bib_ref] is central. Statistical inference is made on the basis of the transition intensities (hazards), the instantaneous risk of moving from one state to another. Even though they were developed for situations where transitions are observed in continuous time (potentially subject to right-censoring), multi-state model approaches can be used to deal with more challenging observation patterns, e.g., interval censoring [bib_ref] Inference for multi-state models from interval-censored data, Commenges [/bib_ref] , based on the assumption that patterns of missing timeto-event information are non-informative for the event history process. In our case, the observation pattern is retrospective event collection in a prospective cohort, which we have outlined along with the corresponding censoring mechanisms in the S1 Appendix and S1 [fig_ref] Table: Estimated covariate effects with their 95% confidence intervals from a logistic regression... [/fig_ref] We first investigated a simplified two-state model with only one possible transition, Approved ! Published, corresponding to a standard survival analysis. We fitted the nonparametric Kaplan-Meier estimator to estimate the cumulative distribution function of total time to publication and the Cox model to estimate potential effects of the covariates on the hazard of publication using the R package survival. From the estimation procedure, we further obtained the median time to publication incl. corresponding confidence intervals based on the Brookmeyer-Crowley approach [bib_ref] The mean, median, and confidence intervals of the Kaplan-Meier survival estimate-computations and..., Barker [/bib_ref] [bib_ref] A confidence interval for the median survival time, Brookmeyer [/bib_ref].
Our main analysis was based on the four-state model, where we used numerical estimators based on the likelihood function to estimate covariate effects on each transition intensity from interval-censored data. We assumed piece-wise constant baseline intensities, allowing us to compute the full likelihood analytically (see section 2 in S1 Appendix). Maximization of the log-likelihood used the Method of Moving Asymptotes [bib_ref] The method of moving asymptotes-a new method for structural optimization, Svanberg [/bib_ref]. All covariates in the multivariable analyses were entered simultaneously, and no interactions were hypothesized.
The function to estimate the transition intensities as well as covariate effects was written in C by one of the authors (TH) and can be called from Rvia a shared library object. Mathematical details as well as documentation of the function can be found in the appendix of [bib_ref] Multi-state models for analyzing the publication process of clinical studies, Haag [/bib_ref].
# Results
## Baseline and outcome characteristics
Of 917 approved studies, 806 were eligible for inclusion in our analyses [fig_ref] Fig 2: Flowchart of study protocols approved between 2000 and 2002 by the research... [/fig_ref].
Study characteristics are displayed in [fig_ref] Table 1: Baseline characteristics of included studies [/fig_ref].
The overall survey response rate was 91%. By 2010 / 2012, 576 (71%) studies were completed and 128 (16%) had been discontinued. A further 41 were classified as 'ongoing' based on survey responses, while for 61 studies the status was unclear (i.e., no survey response, and no evidence of publication). Of the completed studies 363 (63%) had been published as a full article, as had 32 (25%) of the discontinued studies. Overall, 49% of studies had been published We included one additional study compared to our previous analyses [bib_ref] Fate of Clinical Research Studies after Ethical Approval-Follow-Up of Study Protocols until..., Blümle [/bib_ref] , because it later turned out that the survey results were incorrect (study started at local study site). For the intermediate states of study completion and discontinuation, time-to-event information was only available in 23% and 13% of cases respectively, thus the majority of observations were interval-censored. S1 the different observation cases within the four-state model with potential interval-or right-censoring and reports the corresponding counts and percentages in the data set.
## Two-state model
The median time overall was 10 years (i.e., after 10 years half of all studies had been published) with an approximate 95% confidence interval of [bib_ref] Discontinuation and nonpublication of surgical randomised controlled trials: observational study, Chapman [/bib_ref] [bib_ref] Evidence based medicine: what it is and what it isn't, Sackett [/bib_ref] , and 4 years (95% CI [bib_ref] Early termination of cardiovascular trials as a consequence of poor accrual: analysis..., Baldi [/bib_ref] [bib_ref] Discontinuation and non-publication of randomised clinical trials supported by the main public..., Amstutz [/bib_ref] for studies that were actually published. The latter, however, is an underestimate of the true time to publication as it is biased by 'conditioning on the future' [bib_ref] Interpretability and importance of functionals in competing risks and multistate models, Andersen [/bib_ref] ; that is, only including studies for which the outcome was successful.
The publication rate, i.e., the probability that a previously unpublished study is published, is highest four to seven years after REC approval. Studies that have not been published after nine years are very unlikely to be subsequently published (publication probability < 10%) [fig_ref] Fig 4: Complement of Kaplan-Meier estimate with its 95% confidence interval of the cumulative... [/fig_ref]. [fig_ref] Table 2: Estimated covariate effects with their 95% confidence intervals [/fig_ref] , displaying the findings of the two-state model predicting the transition from approval to publication, shows that non-commercially funded studies were published faster than both commercially-funded studies and those in which the funding source was not reported. Studies with industry involvement in the conduct or analysis were published faster than those without industry involvement, as were studies with a primary outcome. Finally, single-centre studies tend to be published faster than multi-centre studies with an international partner. We note here that the variance inflation factors from the covariance matrix of the parameter estimates in the two-state Cox model were overall below three, i.e., there are probably no strong issues due to multicollinearity.
## Four-state model
We now consider each transition in the four-state model in detail. As shown in [fig_ref] Table 3: Estimated covariate effects with their 95% confidence intervals [/fig_ref] , several factors predicted study completion. Funding source was strongly predictive of subsequent study completion; non-commercially-funded studies were completed faster than commercially-funded studies and those in which the funding was unstated. Studies in which the sponsor was involved in the conduct or analysis were completed considerably faster than others. Also, studies with a predefined primary outcome tended to be completed faster than others. The main factors predicting discontinuation included study design and sample size: the hazard for an RCT to be discontinued after approval was higher than that for a non-RCT type study, while studies with a smaller sample size were also more likely to be discontinued. Funding source also tended to be predictive (borderline significant), with commercial funding or unstated funding increasing the hazard for study discontinuation as compared to non-commercial funding. Once completed, funding source was strongly predictive of subsequent publication; again, non-commercially-funded studies were published faster than commerciallyfunded studies and those in which the funding was unstated. Single-centre studies were published faster after completion than multi-centre/international studies. Prediction of publication following discontinuation was limited by the low number of studies in this category (n = 32), which led to wide confidence intervals around some of the estimates. However, among discontinued studies those with a larger sample size tended to be published considerably faster than others, whereas the hazard for an RCT to be published after discontinuation was considerably lower than that for a non-RCT type study.
## Sensitivity analyses
Given the high proportion of missing information on the timing of model transitions, analyses were repeated using logistic regression. The outcomes were successful completion of each state
## Plos one
transition, without taking into account time to event. Overall the results of the logistic regression analysis predicting transition from approval to publication show no discrepancies to the results of the two-state model (S2 . The results (S3 were similar to those obtained in the multi-state model analysis, but with larger standard errors. Funding source emerged as strongly predictive of the transition from approved to completed, with both commercially-funded studies and those in which the funding was unstated less likely to be completed than non-commercial studies. Moreover, two additional covariates emerged as predictive of the transition from completed to published: a pre-specified primary outcome, and industry involvement. With these changes, predictors of the transition from approved to completed were largely identical to those of the transition from completed to published.
# Discussion
In this study, we applied a newly-developed four-state model to examine the research process from ethical approval to publication among clinical studies approved by a German REC between 2000 and 2002. We determined the overall time needed to publish the study results, distinguishing between the time required to conduct the study (i.e., study initiation to either study completion or discontinuation), and the time required to analyse, write up and publish the study findings (i.e., study completion/discontinuation to publication). We also examined predictors of the timing of transitions within the research process as described by the fourstate model, i.e. from study initiation through the completion/discontinuation of data collection to eventual publication. About half of initiated clinical studies were subsequently published as full articles by 2010 / 2012, with the median time from study approval to publication 10 years. Discontinued studies were less often (25%) published than completed studies (63%). There is little consensus in the literature regarding time to publication, with wide variation depending on the cohorts investigated [bib_ref] Time to publication of oncology trials and why some trials are never..., Chapman [/bib_ref] [bib_ref] Barriers towards the publication of academic drug trials. Follow-up of trials approved..., Berendt [/bib_ref] [bib_ref] Time to publication among completed diagnostic accuracy studies: associated with reported accuracy..., Korevaar [/bib_ref] [bib_ref] Time to publication for publicly funded clinical trials in Australia: an observational..., Strand [/bib_ref]. Previous evaluations, however, excluded studies with unknown completion date, which likely led to underestimation of time to publication [bib_ref] Time to publication among completed diagnostic accuracy studies: associated with reported accuracy..., Korevaar [/bib_ref].
The primary findings of this study were that a number of study-related factors which predicted transitions in the conduct of research studies from initiation to eventual publication could be identified, and that these factors differed across the transitions specified in the fourstate model. The latter finding demonstrates the value of differentiating the time required to collect data from the time required to analyse and write up the findings for publication. The use of a four-state model permits a more comprehensive analysis of the research process than is possible if one simply examines the period from approval to publication in a single analysis. It thus represents an advance on previous studies of time to publication [bib_ref] Extent of Non-Publication in Cohorts of Studies Approved by Research Ethics Committees..., Schmucker [/bib_ref].
The hazard from completion to publication was mainly predicted by funding source, with the findings consistent with previous research finding industry-funded studies to have a lower publication rate than studies funded by medical centres. This may suggest that commercial funders have lower incentive to publish findings in academic journals. Interestingly, Goldacre et al recently found that among clinical trials registered in the European Union clinical trials registry, those with a commercial sponsor were more likely to include result data in the registry [bib_ref] Compliance with requirement to report results on the EU Clinical Trials Register:..., Goldacre [/bib_ref]. For commercial funders, journal publication may not always be the preferred means of communicating study findings to the wider community. It may also point to characteristics of the findings, e.g. whether or not the main hypotheses were supported, influencing the decision whether or not to publish to a greater extent for commercially-funded studies, as has been found previously [bib_ref] Extent of Non-Publication in Cohorts of Studies Approved by Research Ethics Committees..., Schmucker [/bib_ref]. The influence of funding source is also reflected in the transition from approved to discontinued, where commercial studies were more likely to be discontinued. Again, this is suggestive of emerging study findings influencing the decision whether or not to continue a study to a greater extent if the study is commercially funded. Consistent with Ioannidis [bib_ref] Effect of the statistical significance of results on the time to completion..., Ioannidis [/bib_ref] , we did not find sample size affected overall time to publication in the two-state model analysis, and in the four-state model analysis sample size was related only to the time from approval to discontinuation, with smaller studies more likely to be discontinued than larger studies.
The major limitation of our study was that in over three-quarters of studies the year of completion or discontinuation was unavailable. As it was not a focus of investigation at that time, survey respondents were not asked to specify the year in which the study was completed or discontinued. Rather, this information was sourced from annual reports submitted to the REC, where it was inconsistently reported. The underlying data set was, therefore, insufficient to perform an adequate time-to-event analysis (with 2 or more states). However, the method we used to maximise the log-likelihood in the time-to-event analyses, the Method of Moving Asymptotes [bib_ref] The method of moving asymptotes-a new method for structural optimization, Svanberg [/bib_ref] , has been extensively studied with simulated data assuming different underlying baseline intensities and a similarly large amount of missing information on the timing of intermediate states still resulting in almost unbiased covariate effect estimates [bib_ref] Multi-state models for analyzing the publication process of clinical studies, Haag [/bib_ref]. It is also important to note that we were mainly missing only time-to-event data for these state transitions, rather than whether or not the study was completed or discontinued. It is reassuring that the sensitivity analyses we performed, which did not use time-to-event information, mostly resulted in similar findings to the primary analyses with regard to the predictors of state transitions. Moreover, while model estimates in the time-to-event analyses had wider confidence intervals than we would have had with full data, even with limited data the estimates were more precise than in the logistic regression analyses, which did not take into account time-to-event data. Despite the presence of missing time-to-event data, Cox regression is preferable to logistic regression because of the long follow-up period we considered and the large heterogeneity in follow-up times between our two cohorts. With extended follow-up, the odds ratio is a poor estimate of the hazard ratio, tending to overestimate the risk factor effect particularly when there are many observed events [bib_ref] A comparison of the logistic risk function and the proportional hazards model..., Green [/bib_ref].
A further limitation was that the exact timing of events was often not determinable. One problem is that it is not always easy to define the completion date of a study. For example, the study completion date may be the date on which the last patient completed the final follow-up, the date on which the database was closed, or the date of the study end report, e. g. final report to the funder. Therefore, we decided not to consider specific dates, but rather to restrict our analysis to the year in which events occurred. We note here that instead of fitting a continuous-time model, a model for time-discrete or grouped time-to-event data would also have been an adequate choice towards fitting this kind of data observed in time periods. Still, our approach can be seen as a suitable approximation. Since REC approval always occurs prior to data collection, this would have had the effect of increasing the estimate of the time required from study initiation to completion/discontinuation. In most cases, we expect the difference would be minimal or non-existent, however it is conceivable that studies of complex behavioural interventions may require a considerable period of intervention development work before participants can be recruited. Additionally, while the 2000 cohort of studies was followed over 11 years, the 2001/2002 cohort was only followed until 2010, see S2 Fig. This reduced the number of studies from these cohorts which we were able to identify, which may have led to an overestimate of the median time to publication. However, this would not lead to a systematic bias if the assumption of independency between the publication process and the censoring mechanism is not violated. There were also a number of studies for which the planned sample size was unknown. Imputing the median sample size likely led to underestimating the variance of this predictor, thus reducing our capacity to find an effect. Finally, only a few trials were published following discontinuation, and confidence intervals for the effect estimates for this transition were partly broad (indeed, no estimate could be computed for the comparison between non-commercially funded studies and funding unstated). These findings should therefore be interpreted with caution.
We believe that poor availability of such data is typical, at least from 2000-2002 in Germany. Things have since changed for the better, particularly due to the founding of the German Registry of Clinical Trials (DRKS) https://www.drks.de/drks_web/ in 2008. Study investigators who apply for an ethics approval are now requested to register there, such that a study is identifiable by a unique study ID. This is intended to ensure it is easier to track studies over time, allowing more reliable data on time to publication to be collected.
Finally, we are obliged to call attention to the importance of careful data collection. Many of the limitations of this study we have discussed are a direct consequence of our failure (in retrospect) to ask the right questions in the survey to ensure we could establish the existence and timing of state transitions. Readers who are considering undertaking a multi-state model analysis should ensure that a data analysis plan is drafted before data collection begins, to help prevent important factors in the analysis being overlooked.
# Conclusions
The use of a four-state model permitted a comprehensive examination of the research process, allowing for the identification of predictive relationships between states that were not apparent from a simple two-state analysis. While the presence of missing time-to-event information was an important limitation, it did not appear to have an undue influence on the findings. Since this study was initiated, the proportion of studies which are registered in a clinical trial registry prior to data collection has greatly increased [bib_ref] Registration of published randomized trials: a systematic review and meta-analysis, Trinquart [/bib_ref] , and therefore future investigations of time to publication should have access to more complete information regarding the timing of state transitions. We recommend using modern statistical methods of event history analysis (multistate models) to permit a detailed study of the publication process in medicine, epidemiology, or other areas of research.
## Supporting information
[fig] Fig 1: Illustration of the four-state model. The boxes indicate the states, the arrows the possible state transitions. https://doi.org/10.1371/journal.pone.0230797.g001 2008 [25]. This was then complemented by protocols approved in 2001 and 2002. For those studies, the search for publications took place between August 2009 and January 2010, and surveys were sent in February 2010. For the pilot year an updated search was carried out between July 2011 and January 2012 (S2 Fig). [/fig]
[fig] Fig 2: Flowchart of study protocols approved between 2000 and 2002 by the research ethics committee of the University of Freiburg/Germany. [/fig]
[fig] Fig 3: Four-state model. The numbers next to the arrows indicate the total numbers of studies for each transition. https://doi.org/10.1371/journal.pone.0230797.g003 [/fig]
[fig] Fig 4: Complement of Kaplan-Meier estimate with its 95% confidence interval of the cumulative distribution function of the total time to publication. Time point 0 is the REC's positive vote, i.e. study approval. https://doi.org/10.1371/journal.pone.0230797.g004 [/fig]
[fig] S1: Fig. Criteria for study design classification. (TIFF) S2 Fig. Timeline of study research activities. (TIFF) Appendix. Extended statistical information. (DOCX) Table. Four-state model observation cases with coding of potential censoring times and corresponding counts and percentages in the data set. (DOCX) S2 [/fig]
[table] Table 1: Baseline characteristics of included studies (n = 806).https://doi.org/10.1371/journal.pone.0230797.t001 [/table]
[table] Table 2: Estimated covariate effects with their 95% confidence intervals (two-state model). [/table]
[table] Table 3: Estimated covariate effects with their 95% confidence intervals (four-state model). [/table]
[table] Table: Estimated covariate effects with their 95% confidence intervals from a logistic regression model (see Sensitivity analysis in main manuscript). (DOCX) S3Table. Estimated covariate effects with their 95% confidence intervals from separate logistic regression models (see Sensitivity analysis in main manuscript). (DOCX) 9. Bernardez-Pereira S, Lopes RD, Carrion MJM, Santucci EV, Soares RM, de Oliveira Abreu M, et al. Prevalence, characteristics, and predictors of early termination of cardiovascular clinical trials due to low recruitment: Insights from the ClinicalTrials.gov registry. American Heart Journal. 2014; 168 (2):213-9.e1. https://doi.org/10.1016/j.ahj.2014.04.013 PMID: 25066561 [/table]
|
10.3390/cancers12102754
|
CCBY
|
7601447
|
32987868
|
s2orc_pubmed_articles
|
Three-Dimensional Culture System of Cancer Cells Combined with Biomaterials for Drug Screening
Simple Summary: For the research and development of drug discovery, it is of prime importance to construct the three-dimensional (3D) tissue models in vitro. To this end, the enhancement design of cell function and activity by making use of biomaterials is essential. In this review, 3D culture systems of cancer cells combined with several biomaterials for anticancer drug screening are introduced.Abstract: Anticancer drug screening is one of the most important research and development processes to develop new drugs for cancer treatment. However, there is a problem resulting in gaps between the in vitro drug screening and preclinical or clinical study. This is mainly because the condition of cancer cell culture is quite different from that in vivo. As a trial to mimic the in vivo cancer environment, there has been some research on a three-dimensional (3D) culture system by making use of biomaterials. The 3D culture technologies enable us to give cancer cells an in vitro environment close to the in vivo condition. Cancer cells modified to replicate the in vivo cancer environment will promote the biological research or drug discovery of cancers. This review introduces the in vitro research of 3D cell culture systems with biomaterials in addition to a brief summary of the cancer environment.
# Introduction
The basic concept of regenerative medicine is to achieve the regeneration and repairing of damaged or injured tissues by utilizing the natural healing potential of the body itself. Regenerative medicine consists of regenerative therapy and regenerative research. Regenerative therapy is to treat patients through the in vivo enhancement of cell activity. Regenerative research is positioned as the scientific support for the regeneration therapy of the next generation. Drug discovery is defined as regenerative research. The therapeutic efficacy, metabolism or toxicology of drugs are efficiently evaluated by taking advantage of activated cells. To enhance the cell activity, two methodologies have been recently noted. One is to utilize three-dimensional (3D) cell culture technologies. Cells are usually cultured in a two-dimensional (2D) system, with a plate or dish. However, the functions of cells cultured in the 2D system are lower than those of body cells because cells tend to interact with each other for the enhancement of their own activities in the body [bib_ref] Novel hepatocyte culture system developed using microfabrication and collagen/polyethylene glycol microcontact printing, Fukuda [/bib_ref] [bib_ref] Energy metabolism transition in multi-cellular human tumor spheroids, Rodriguez-Enriquez [/bib_ref] [bib_ref] Methods for inducing embryoid body formation: In vitro differentiation system of embryonic..., Kurosawa [/bib_ref] [bib_ref] Recent advances in three-dimensional multicellular spheroid culture for biomedical research, Lin [/bib_ref]. Due to the difference in the cell condition, the drug effect evaluated by the in vitro drug screening is not always the same as that in preclinical or clinical study, which leads to the failure of drug research and development [bib_ref] Three-dimensional cell culture: The missing link in drug discovery, Breslin [/bib_ref] [bib_ref] Anticancer drug development: The grand challenges, Hait [/bib_ref] [fig_ref] Figure 1: Research and development process of drug development [/fig_ref]. The comparison of cancer cell culture between 2D and 3D systems is shown in [fig_ref] Table 1: Comparison of cancer cells culture between 2D and 3D systems [/fig_ref]. There are merits or demerits between the two culture systems. Although the systems have been used depending on the purpose, the 3D culture is superior in terms of drug discovery which well reflects the in vivo cancer environment. The other methodology to enhance cell functions is the active utilization of biomaterials. Cell culture is often performed on the dish or plate which is mainly composed of polystyrene. This condition of an artificial environment is quite different from the in vivo body environment of cancer cells, and consequently, the drug effect or cytotoxicity evaluation is technologically limited. Biomaterials which consist of extracellular matrix (ECM) components are effective in enhancing the cell activity or functions. The interaction with biomaterials will enable cells to enhance their proliferation, differentiation, and biological functions, leading to the realization of cancer cell-environment interaction. Anticancer drug screening is often performed by using the 2D culture system of cancer cells. As mentioned above, to mimic the cancer environment in the body, the combination of 3D cell culture technology and biomaterials is important. In addition to the technological methods, the interaction of cancer cells with stromal cells should be considered [bib_ref] Engineering Tumors: A Tissue Engineering Perspective in Cancer Biology, Burdett [/bib_ref] , because the cancer environment is composed of several stromal cells, such as cancer-associated fibroblasts (CAF) [bib_ref] Cancer-Associated Fibroblasts: Their Characteristics and Their Roles in Tumor Growth, Shiga [/bib_ref] [bib_ref] The biology and function of fibroblasts in cancer, Kalluri [/bib_ref] , tumor-associated macrophages (TAM) [bib_ref] Tumor-Associated Macrophages and Neutrophils in Tumor Microenvironment, Kim [/bib_ref] [bib_ref] Tumor-associated macrophages: From basic research to clinical application, Yang [/bib_ref] , mesenchymal stem cells (MSC) [bib_ref] Interaction of MSC with tumor cells, Melzer [/bib_ref] [bib_ref] Tumor microenvironment: Bone marrow-mesenchymal stem cells as key players, Barcellos-De-Souza [/bib_ref] or endothelial cells [bib_ref] Cellular abnormalities of blood vessels as targets in cancer, Baluk [/bib_ref] [bib_ref] Isolated tumor endothelial cells maintain specific character during long-term culture, Matsuda [/bib_ref]. It has been demonstrated that cancer cells interact with stromal cells, leading to the promotion of cancer diseases [bib_ref] Cancer and the tumor microenvironment: A review of an essential relationship, Mbeunkui [/bib_ref] [fig_ref] Figure 2: Cancer cells interact with various stromal cells of cancer-associated fibroblasts [/fig_ref]. Moreover, several humoral factors secreted from cells are also important to construct the cancer environment [bib_ref] Matrix metalloproteinases: Regulators of the tumor microenvironment, Kessenbrock [/bib_ref] [bib_ref] The tumor microenvironment and its role in promoting tumor growth, Whiteside [/bib_ref] [bib_ref] Tumor-associated macrophages as major players in the tumor microenvironment, Chanmee [/bib_ref]. Therefore, to mimic the cancer environment or cancer diseases in vitro, a coculture system of cancer cells with stromal cells is essential.
Nowadays, to replicate the cancer environment and diseases in vitro, several studies have been reported on 3D cancer models combined with biomaterials. In this review, first, the important stromal cells and their characterization are briefly described. Second, we introduce 3D cancer models by making use of several biomaterials.
## Stromal cells in cancer environment
There are four types of stromal cells which are composed of the cancer environment. The biological functions of stromal cells and the humoral factors secreted are briefly explained. summarizes some key cytokines in the cancer environment. . Cytokines secreted in cancer environment and the biological function.
## Cytokines functions
Transforming growth factor-β (TGF-β) Support of cancer cells proliferation Promotion of endothelial-mesenchymal transition (EMT) and the consequent invasion or metastasis Recruitment of fibroblasts Differentiation of fibroblasts or MSC into CAF Promotion of tumorigenicity Promotion of angiogenesis Tumor necrosis factor-α (TNF-α) Disruption of epithelial barrier Promotion of inflammatory cell infiltration Stimulation of TGF-β-induced EMT Induction of vascular endothelial growth factors (VEGF) secretion
## Cancer-associated fibroblasts
Cancer-associated fibroblasts (CAF) are major stromal cells. CAF of a large-spindle shape are perpetually activated and never undergo apoptosis [bib_ref] Cancer-Associated Fibroblasts: Their Characteristics and Their Roles in Tumor Growth, Shiga [/bib_ref]. Although the origin of CAF is not completely clear, normal fibroblasts [bib_ref] MicroRNAs reprogram normal fibroblasts into cancer-associated fibroblasts in ovarian cancer, Mitra [/bib_ref] [bib_ref] Autocrine TGF-beta and stromal cell-derived factor-1 (SDF-1) signaling drives the evolution of..., Kojima [/bib_ref] [bib_ref] BM-MSCs promote prostate cancer progression via the conversion of normal fibroblasts to..., Wen [/bib_ref] , mesenchymal stem cells (MSC) [bib_ref] Bone marrow-derived myofibroblasts contribute to the mesenchymal stem cell niche and promote..., Quante [/bib_ref] [bib_ref] Bone marrow contribution to tumor-associated myofibroblasts and fibroblasts, Direkze [/bib_ref] , or endothelial cells [bib_ref] Endothelial-to-mesenchymal transition contributes to cardiac fibrosis, Zeisberg [/bib_ref] [bib_ref] The role of endothelial-mesenchymal transition in development and pathological process, Lin [/bib_ref] are potential sources of CAF. As CAF markers, alpha-smooth muscle actin (α-SMA), fibroblast activation protein (FAP), and fibroblast specific protein-1 are well known [bib_ref] Heterogeneity of stromal fibroblasts in tumors, Orimo [/bib_ref]. In particular, approximately 90% of cancer cell types show the expression of FAP [bib_ref] Cell surface glycoprotein of reactive stromal fibroblasts as a potential antibody target..., Garin-Chesa [/bib_ref]. The interaction between cancer cells and CAF plays a key role in cancer diseases. An experimental trial to indicate the importance of CAF has been reported by Weinberg et al. Human CAF and breast cancer cells are injected to nude mice. It is demonstrated that cancer cells with CAF effectively proliferate compared with CAF-free cancer cells or cancer cells cocultured with normal fibroblasts groups. This proliferation enhancement was induced by stromal cell-derived factor-1 (SDF-1) secreted [bib_ref] Stromal fibroblasts present in invasive human breast carcinomas promote tumor growth and..., Orimo [/bib_ref]. This study clearly indicates the importance of CAF existence for cancer cell activity. CAF not only promote cancer proliferation but also increase the invasion of cancer cells via the cancer-CAF interaction. The interaction also promotes the secretion of various matrix-degrading proteinases. Among them, matrix-metalloproteinase (MMP) has a key role in the cancer invasion or metastasis. MMP can degrade type IV collagen and laminin, which are major components of basement membrane [bib_ref] PAR1 is a matrix metalloprotease-1 receptor that promotes invasion and tumorigenesis of..., Boire [/bib_ref] [bib_ref] Fibroblasts and extracellular matrix differently modulate MMP activation by primary and metastatic..., Koontongkaew [/bib_ref] [bib_ref] In vivo glioma growth requires host-derived matrix metalloproteinase 2 for maintenance of..., Takahashi [/bib_ref]. In addition to SDF-1 and MMP, transforming growth factor-β1 (TGF-β1) [bib_ref] Cancer associated fibroblasts stimulated by transforming growth factor beta1 (TGF-beta 1) increase..., Casey [/bib_ref] [bib_ref] Cancer-associated fibroblasts induce epithelial-mesenchymal transition of breast cancer cells through paracrine TGF-beta..., Yu [/bib_ref] and interleukin (IL)-6 [bib_ref] IL-6 Secreted from Cancer-Associated Fibroblasts Mediates Chemoresistance in NSCLC by Increasing Epithelial-Mesenchymal..., Shintani [/bib_ref] are also important factors for the cancer-CAF interaction.
## Tumor-associated macrophages
Macrophages are usually polarized to M1 or M2 phenotypes responding to the environment. M1 macrophages (proinflammatory) have a capacity of inflammation induction, chronic inflammation, and pathogen defense. On the other hand, M2 macrophages (anti-inflammatory) are involved in noninflammatory response, wound healing, and tissue regeneration [bib_ref] Alternative activation of macrophages, Gordon [/bib_ref] [bib_ref] Expression of vascular permeability factor (vascular endothelial growth factor) by epidermal keratinocytes..., Brown [/bib_ref]. TAM are generally recognized as M2-type macrophages [bib_ref] Macrophage polarization: Tumor-associated macrophages as a paradigm for polarized M2 mononuclear phagocytes, Mantovani [/bib_ref] [bib_ref] M2 polarization of tumor-associated macrophages is dependent on integrin beta3 via peroxisome..., Shu [/bib_ref]. Due to the M2-type phenotype, CD163 and CD204 are well known as the TAM markers [bib_ref] CD163 as a marker of M2 macrophage, contribute to predicte aggressiveness and..., Hu [/bib_ref] [bib_ref] Tumor-associated CD204(+) M2 macrophages are unfavorable prognostic indicators in uterine cervical adenocarcinoma, Kawachi [/bib_ref]. The stimulation of macrophages by lipopolysaccharide (LPS) and adenosines can induce TAM in vitro [bib_ref] Polarizing Macrophages In Vitro, Huang [/bib_ref]. TAM play an important role in cancer progression. Grivennikov et al. indicate that IL-23 and IL-17 secreted from TAM promote the cancer proliferation [bib_ref] Adenoma-linked barrier defects and microbial products drive IL-23/IL-17-mediated tumour growth, Grivennikov [/bib_ref]. Tumor-necrosis factor-α (TNF-α), vascular endothelial growth factor (VEGF), and TGF-β1 secreted from TAM can promote the cancer metastasis [bib_ref] Imbalance of Clara cell-mediated homeostatic inflammation is involved in lung metastasis, Tomita [/bib_ref]. Taken together, TAM are recognized as important cells for cancer diseases. This promising TAM-targeted therapy has been investigated [bib_ref] Tumour-associated macrophages as a novel target of VEGI-251 in cancer therapy, Dong [/bib_ref] [bib_ref] Macrophages in the microenvironment of head and neck cancer: Potential targets for..., Evrard [/bib_ref].
## Cancer-associated fibroblasts and tumor-associated macrophages for different cancer types
CAF and TAM are major components of stromal cells in the cancer environment. However, their biological contribution and influence on cancer cells generally depend on the cancer regions. For example, in brain, liver, or kidney cancer, contribution of TAM is larger than that of CAF, while the effect of CAF on the lung or pancreatic cancer is high compared with that of TAM. This is mainly because of the existence ratio [bib_ref] CAFs and TAMs: Maestros of the tumour microenvironment, Komohara [/bib_ref]. Therefore, the CAF/TAM contribution ratio should be considered to understand the characteristics of various cancer cell types.
## Mesenchymal stem cells
Mesenchymal stem cells (MSC) have been noted in the field of tissue regeneration because MSC have a capacity of differentiation into bone, cartilage, or fat cells [bib_ref] Mesenchymal stem cells are recruited into wounded skin and contribute to wound..., Sasaki [/bib_ref] [bib_ref] Osteoblastic cells: Differentiation and trans-differentiation, Kassem [/bib_ref] [bib_ref] Transdifferentiation potential of human mesenchymal stem cells derived from bone marrow, Song [/bib_ref]. Therefore, MSC transplantation would be effective in regenerative medicine [bib_ref] Mesenchymal stem cells: Environmentally responsive therapeutics for regenerative medicine, Murphy [/bib_ref]. However, the differentiation capacity of MSC is unfavorable for cancer patients. For example, TGF-β1 secreted from several cells in the cancer environment can differentiate MSC into CAF [bib_ref] Mesenchymal Stem Cells are Recruited and Activated into Carcinoma-Associated Fibroblasts by Prostate..., Barcellos-De-Souza [/bib_ref]. Chowdhury et al. also report that exosomes secreted from cancer cells promote the differentiation MSC into CAF [bib_ref] Cancer exosomes trigger mesenchymal stem cell differentiation into pro-angiogenic and pro-invasive myofibroblasts, Chowdhury [/bib_ref]. In addition to the differentiation into CAF, MSC also allow TAM to migrate into the cancer environment via C-C chemokine receptor type 2 (CCR2) [bib_ref] MSCs, macrophages, and cancer: A dangerous menage-a-trois, Mantovani [/bib_ref]. Moreover, IL-6 and angiopoietin-1 secreted from primary human MSC can promote the angiogenesis [bib_ref] Concise Review: Cancer Cells, Cancer Stem Cells, and Mesenchymal Stem Cells: Influence..., Papaccio [/bib_ref]. Recently, it has been reported that MSC can polarize into a proinflammatory MSC-1 and an immunosuppressive MSC-2 phenotype. MSC-2 can enhance the cancer proliferation, spread, and promotion, while MSC-1 suppress the cancer proliferation [bib_ref] Tumor microenvironment: Bone marrow-mesenchymal stem cells as key players, Barcellos-De-Souza [/bib_ref] [bib_ref] A new mesenchymal stem cell (MSC) paradigm: Polarization into a pro-inflammatory MSC1..., Waterman [/bib_ref] [bib_ref] Mesenchymal stem cell 1 (MSC1)-based therapy attenuates tumor growth whereas MSC2-treatment promotes..., Waterman [/bib_ref]. The understanding of MSC roles at cancer sites would provide an important aspect for further cancer research and therapies.
## Endothelial cells
It is important for cancer cells to induce angiogenesis in terms of nutrient and oxygen supply, the elimination of waste products, invasion, and metastasis. However, since a vascularization suddenly advances at the cancer sites under a nonphysiological condition, it is well recognized that the blood vessels in the cancer environment are fragile and the wall is highly permeable. Enhanced permeation and retention effect (EPR effect) is a concept to symbolize this condition of cancer blood vessels [bib_ref] Heterogeneity of tumor endothelial cells, Hida [/bib_ref]. Based on the EPR effect concept, a positive targeting of micelles containing anticancer drug to cancer has been reported [bib_ref] Progress of drug-loaded polymeric micelles into clinical studies, Cabral [/bib_ref] [bib_ref] Preclinical and clinical studies of anticancer agent-incorporating polymer micelles, Matsumura [/bib_ref]. Thus, there are some structural and functional differences between the cancer and normal blood vessels. To study cancer characteristics or therapeutic efficacy, the blood vessel properties and the cancer-endothelial cell interaction are important to consider. Some research has been reported to demonstrate that tumor endothelial cells (TEC) differ from normal endothelial cells in properties, such as the cell proliferation, the gene expression, the response to growth factors, or migration [bib_ref] Cytogenetic abnormalities of tumor-associated endothelial cells in human malignant tumors, Akino [/bib_ref] [bib_ref] Tumor endothelial cells express epidermal growth factor receptor (EGFR) but not ErbB3..., Amin [/bib_ref]. High metastatic tumor-derived TEC (HM-TEC) and low metastatic tumor-derived TEC (LM-TEC) can be isolated from mice. It is demonstrated that the secretion levels of VEGF, MMP-2, MMP-9, and SDF-1 from HM-TEC are higher than from that of LM-TEC [bib_ref] Heterogeneity of tumor endothelial cells, Hida [/bib_ref] [bib_ref] HuR keeps an angiogenic switch on by stabilising mRNA of VEGF and..., Kurosu [/bib_ref]. It is reported that coculture with endothelial cells facilitates the in vitro culture of cancer cells [bib_ref] Recapitulating spatiotemporal tumor heterogeneity in vitro through engineered breast cancer microtissues, Mazio [/bib_ref].
## 3d culture system of cancer cells with biomaterials
Biomaterials classify into natural biomaterials derived from animals or plants and synthetic biomaterials artificially prepared. Natural biomaterials are composed of polysaccharide (amylose, cellulose, alginate, chitosan, or hyaluronic acid), peptide (collagen or gelatin), nucleic acid, or polyhydroxyalkanoates. Since the degradative enzyme and metabolic system have already existed in the body, most natural biomaterials can enzymatically be degraded. Because the components constitute the cancer environment as the ECM and contribute to cancer diseases, natural biomaterials are often used to design the 3D culture system of cancer cells. Although natural biomaterials are of high biocompatible, there are some limitations of immunogenicity or homogeneity to use. To avoid the issues, synthetic biomaterials are used. Synthetic biomaterials are mainly degraded nonenzymatically based on simple hydrolysis. There are some merits of synthetic biomaterials, such as the characteristics control, the high stiffness, and the clarity of properties.
In this chapter, several 3D culture systems of cancer cells combined with biomaterials are introduced. To date, two types of biomaterials have been applied to the 3D culture system of cancer cells. One is the culture system of cancer cells with the biomaterials of a spherical shape. When incubated with microspheric hydrogels of biomaterial, cancer cells naturally form a cell aggregate of a tissue-like 3D structure, which mimics the cancer environment. The disadvantages of this system are the difficulty of cells separation from the cell-hydrogel aggregates, and consequently, the result is often of low repeatability. The other is the culture system of cancer cells with the biomaterials of nonspherical type, such as sponge shapes or nonwoven fabrics. In this system, cells effectively proliferate and migrate on the scaffold. This is suitable for immunohistochemical analysis. [fig_ref] Table 3: 3D culture system of cancer cells combined with biomaterials [/fig_ref] summarizes the 3D culture systems of cancer cells combined with various types of biomaterials.
# Biomaterials characteristics
## Types of cancer cells cultured with biomaterial scaffolds of spherical or other shapes stromal cells cocultured with cancer cells spherical (a) other (sponges shapes or nonwoven fabrics) (b)
Chitosan Derived from crustacean shells Linear cationic polymer Formation of polyelectrolyte complexes with anionic polymers
Breast cancer [bib_ref] Characterization and evaluation of chitosan matrix for in vitro growth of MCF-7..., Dhiman [/bib_ref] Liver cancer [bib_ref] Chitosan-alginate scaffold culture system for hepatocellular carcinoma increases malignancy and drug resistance, Leung [/bib_ref] Glioblastoma [bib_ref] Porous chitosan-hyaluronic acid scaffolds as a mimic of glioblastoma microenvironment ECM, Florczyk [/bib_ref] [bib_ref] Culture on 3D Chitosan-Hyaluronic Acid Scaffolds Enhances Stem Cell Marker Expression and..., Wang [/bib_ref] [bib_ref] Chitosan-alginate 3D scaffolds as a mimic of the glioma tumor microenvironment, Kievit [/bib_ref] [bib_ref] Proliferation and enrichment of CD133(+) glioblastoma cancer stem cells on 3D chitosan-alginate..., Kievit [/bib_ref] Lung cancer [bib_ref] Chitosan-hyaluronan based 3D co-culture platform for studying the crosstalk of lung cancer..., Han [/bib_ref] [bib_ref] Acquisition of epithelial-mesenchymal transition and cancer stem-like phenotypes within chitosan-hyaluronan membrane-derived 3D..., Huang [/bib_ref] Prostate cancer [bib_ref] 3D Porous Chitosan-Alginate Scaffolds as an In Vitro Model for Evaluating Nanoparticle-Mediated..., Wang [/bib_ref] [bib_ref] 3D porous chitosan-alginate scaffold stiffness promotes differential responses in prostate cancer cell..., Xu [/bib_ref] [bib_ref] 3D porous chitosan-chondroitin sulfate scaffolds promote epithelial to mesenchymal transition in prostate..., Xu [/bib_ref] MSC [bib_ref] Chitosan-hyaluronan based 3D co-culture platform for studying the crosstalk of lung cancer..., Han [/bib_ref] Alginate Derived from seaweed Water-soluble Crosslinked by ions Easy cell encapsulation Nonadhesive nature to cells Easy stiffness control Thermally stable High water-holding capacity Breast cancer [bib_ref] Modelling the tumour microenvironment in long-term microencapsulated 3D co-cultures recapitulates phenotypic features..., Estrada [/bib_ref] [bib_ref] A new cell-laden 3D Alginate-Matrigel hydrogel resembles human breast cancer cell malignant..., Cavo [/bib_ref] Liver cancer [bib_ref] Role of three-dimensional matrix stiffness in regulating the chemoresistance of hepatocellular carcinoma..., Liu [/bib_ref] [bib_ref] Mesenchymal stem cells enhance the metastasis of 3D-cultured hepatocellular carcinoma cells, Liu [/bib_ref] Head and neck squamous cell carcinoma [bib_ref] Potential effect of matrix stiffness on the enrichment of tumor initiating cells..., Liu [/bib_ref] Leukemia [bib_ref] Characterization of leukemic cell behaviors in a soft marrow mimetic alginate hydrogel, Vu [/bib_ref] Liver cancer [bib_ref] Chitosan-alginate scaffold culture system for hepatocellular carcinoma increases malignancy and drug resistance, Leung [/bib_ref] Breast cancer [bib_ref] 3D culture broadly regulates tumor cell hypoxia response and angiogenesis via pro-inflammatory..., Delnero [/bib_ref] [bib_ref] Hybrid collagen alginate hydrogel as a platform for 3D tumor spheroid invasion, Liu [/bib_ref] Glioblastoma [bib_ref] Chitosan-alginate 3D scaffolds as a mimic of the glioma tumor microenvironment, Kievit [/bib_ref] [bib_ref] Proliferation and enrichment of CD133(+) glioblastoma cancer stem cells on 3D chitosan-alginate..., Kievit [/bib_ref] Prostate cancer [bib_ref] 3D Porous Chitosan-Alginate Scaffolds as an In Vitro Model for Evaluating Nanoparticle-Mediated..., Wang [/bib_ref] [bib_ref] 3D porous chitosan-alginate scaffold stiffness promotes differential responses in prostate cancer cell..., Xu [/bib_ref] Oral squamous cell carcinoma [bib_ref] 3D culture broadly regulates tumor cell hypoxia response and angiogenesis via pro-inflammatory..., Delnero [/bib_ref] Lung cancer [bib_ref] 3D culture broadly regulates tumor cell hypoxia response and angiogenesis via pro-inflammatory..., Delnero [/bib_ref] Gastric cancer [bib_ref] 3D culture broadly regulates tumor cell hypoxia response and angiogenesis via pro-inflammatory..., Delnero [/bib_ref] Fibroblasts [bib_ref] A new cell-laden 3D Alginate-Matrigel hydrogel resembles human breast cancer cell malignant..., Cavo [/bib_ref] [bib_ref] ECM microenvironment regulates collective migration and local dissemination in normal and malignant..., Nguyen-Ngoc [/bib_ref] Breast cancer [bib_ref] The morphologies of breast cancer cell lines in three-dimensional assays correlate with..., Kenny [/bib_ref] [bib_ref] Human mammary fibroblasts stimulate invasion of breast cancer cells in a three-dimensional..., Olsen [/bib_ref] [bib_ref] Cells assemble invadopodia-like structures and invade into matrigel in a matrix metalloprotease..., Yu [/bib_ref] [bib_ref] Enhanced invasion of metastatic cancer cells via extracellular matrix interface, Zhu [/bib_ref] [bib_ref] Establishment of a heterotypic 3D culture system to evaluate the interaction of..., Augustine [/bib_ref] [bib_ref] Human bone marrow stromal cells enhance breast cancer cell growth rates in..., Sasser [/bib_ref] [bib_ref] Malignant stroma increases luminal breast cancer cell proliferation and angiogenesis through platelet-derived..., Pinto [/bib_ref] Fibrosarcoma [bib_ref] Cells assemble invadopodia-like structures and invade into matrigel in a matrix metalloprotease..., Yu [/bib_ref] [bib_ref] Invasion of reconstituted basement membrane matrix by metastatic human tumor cells, Kramer [/bib_ref] Melanoma [bib_ref] Cells assemble invadopodia-like structures and invade into matrigel in a matrix metalloprotease..., Yu [/bib_ref] Fibroblasts [bib_ref] Human mammary fibroblasts stimulate invasion of breast cancer cells in a three-dimensional..., Olsen [/bib_ref] [bib_ref] Malignant stroma increases luminal breast cancer cell proliferation and angiogenesis through platelet-derived..., Pinto [/bib_ref] T REG lymphocyte [bib_ref] Establishment of a heterotypic 3D culture system to evaluate the interaction of..., Augustine [/bib_ref] NK cells [bib_ref] Establishment of a heterotypic 3D culture system to evaluate the interaction of..., Augustine [/bib_ref] MSC [bib_ref] Human bone marrow stromal cells enhance breast cancer cell growth rates in..., Sasser [/bib_ref] Endothelial cells [bib_ref] Malignant stroma increases luminal breast cancer cell proliferation and angiogenesis through platelet-derived..., Pinto [/bib_ref] Poly (lactic-coglycolic acid)
Porosity morphology Biodegradability Hydrophobic property
Ovarian cancer [bib_ref] The controllable preparation of porous PLGA microspheres by the oil/water emulsion method..., Zhang [/bib_ref] Breast cancer [bib_ref] Characterization of porous PLGA/PLA microparticles as a scaffold for three dimensional growth..., Sahoo [/bib_ref] Breast cancer [bib_ref] Incorporation of hydroxyapatite into nanofibrous PLGA scaffold towards improved breast cancer cell..., Luo [/bib_ref] [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Prostate cancer [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Melanoma [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Ovarian cancer [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Lung cancer [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Liver cancer [bib_ref] Characterization of porous poly(D,L-lacticco-glycolic acid) sponges fabricated by supercritical CO2 gas-foaming method..., Zhu [/bib_ref] Polyethylene glycol Chemical modification Water-holding capacity
Breast cancer [bib_ref] Three-dimensional-engineered matrix to study cancer stem cells and tumorsphere formation: Effect of..., Yang [/bib_ref] [bib_ref] PEG-fibrinogen hydrogels for three-dimensional breast cancer cell culture, Pradhan [/bib_ref] [bib_ref] A three-dimensional spheroidal cancer model based on PEG-fibrinogen hydrogel microspheres, Pradhan [/bib_ref] Lung cancer [bib_ref] 3D modeling of human cancer: A PEG-fibrin hydrogel system to study the..., Bufalo [/bib_ref] Prostate cancer [bib_ref] A three-dimensional spheroidal cancer model based on PEG-fibrinogen hydrogel microspheres, Pradhan [/bib_ref] Colon cancer [bib_ref] A three-dimensional spheroidal cancer model based on PEG-fibrinogen hydrogel microspheres, Pradhan [/bib_ref] Breast cancer [bib_ref] Mechanical confinement via a PEG/Collagen interpenetrating network inhibits behavior characteristic of malignant..., Reynolds [/bib_ref] [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] [bib_ref] Cancer cell migration within 3D layer-by-layer microfabricated photocrosslinked PEG scaffolds with tunable..., Soman [/bib_ref] Lung cancer [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Melanoma [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Ovarian cancer [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] Prostate cancer [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] [bib_ref] Phenotypic Characterization of Prostate Cancer LNCaP Cells Cultured within a Bioengineered Microenvironment, Sieh [/bib_ref] Fibrosarcoma [bib_ref] A peptide functionalized poly(ethylene glycol) (PEG) hydrogel for investigating the influence of..., Singh [/bib_ref] Glioblastoma [bib_ref] Bioengineered 3D brain tumor model to elucidate the effects of matrix stiffness..., Wang [/bib_ref] Fibroblasts [bib_ref] 3D modeling of human cancer: A PEG-fibrin hydrogel system to study the..., Bufalo [/bib_ref] Endothelial cells [bib_ref] 3D modeling of human cancer: A PEG-fibrin hydrogel system to study the..., Bufalo [/bib_ref] (a) 3D cell constructs are readily formed; (b) cells well proliferate and migrate on the scaffold.
## Chitosan
Chitosan of poly (1, 4 D-glucosamine), a partially deacetylated derivative of chitin, is a natural cationic linear polysaccharide [bib_ref] Spatial distribution of mammalian cells dictated by material surface chemistry, Healy [/bib_ref]. Chitin is known as primary structural polymers in arthropod exoskeletons. The antigenic response of chitosan is rather low among organonitrogen compounds, and the stiffness is also enough for the cell scaffold. Therefore, chitosan is used as a blood anticoagulant [bib_ref] Sulfated chitin and chitosan as novel biomaterials, Jayakumar [/bib_ref] , a wound healing accelerator [bib_ref] Topical formulations and wound healing applications of chitosan, Ueno [/bib_ref] , and a surgical suture [bib_ref] The use of chitin and chitosan as drug carriers, Miyazaki [/bib_ref] and also for cardiac [bib_ref] Functional 3-D cardiac co-culture model using bioactive chitosan nanofiber scaffolds, Hussain [/bib_ref] , neural [bib_ref] Polylysine-functionalised thermoresponsive chitosan hydrogel for neural tissue engineering, Crompton [/bib_ref] , bone [bib_ref] Osteoconduction Exerted by Methylpyrrolidinone Chitosan Used in Dental Surgery, Muzzarelli [/bib_ref] , or vein endothelial [bib_ref] 3-D culture of human umbilical vein endothelial cells with reversible thermosensitive hydroxybutyl..., Wei [/bib_ref] tissue engineering. Chitosan is also an effective biomaterial for 3D culture of cancer cells because glycosaminoglycan (GAG), closely to the structure of chitosan, is one major component of ECM in the cancer environment [bib_ref] Chitosan-based biomaterials for tissue engineering, Croisier [/bib_ref]. A chitosan scaffold is reported for the 3D culture system of cancer cells. When human breast MCF-7 cancer cells were cultured on the chitosan scaffold, the cell attachment and proliferation were superior to the regular culture of plastic dish [bib_ref] Characterization and evaluation of chitosan matrix for in vitro growth of MCF-7..., Dhiman [/bib_ref].
## Alginate
Alginate, purified from seaweed, is a naturally-occurring anionic polysaccharide composed of α-l-guluronic acid and β-d-mannuronic acid [bib_ref] Alginate composition of some New Zealand brown seaweeds, Miller [/bib_ref]. As a pharmaceutical application, sodium alginate has already been used for the treatment of peptic ulcer. One of the alginate merits is the quick gelation or cell encapsulation by ionic crosslinking using divalent metal ions of calcium or ferric ions [bib_ref] Alginate-Based Biomaterials for Regenerative Medicine Applications, Sun [/bib_ref] [bib_ref] Degradable alginate hydrogels crosslinked by the macromolecular crosslinker alginate dialdehyde, Jejurikar [/bib_ref]. Second, alginate is thermally stable [bib_ref] Alginate based polyurethanes: A review of recent advances and perspective, Zia [/bib_ref]. The molecular structure of alginate is similar to that of polysaccharide in vivo [bib_ref] Supramolecular organization of extracellular matrix glycosaminoglycans, in vitro and in the tissues, Scott [/bib_ref]. Therefore, for the 3D culture system of cancer cells, there are many studies on the encapsulation of cancer cells by using alginate gels. Liu et al. prepare alginate gels to encapsulate head and neck squamous carcinoma cells. In addition, three types of gels with different stiffness are prepared by changing the alginate concentration. It is found that the tumorigenicity, the metastatic ability, and the drug resistance increased at the moderate stiffness [bib_ref] Potential effect of matrix stiffness on the enrichment of tumor initiating cells..., Liu [/bib_ref]. The system is also applied to not only neck squamous cell carcinoma but also the hepatocellular carcinoma reaction [bib_ref] Role of three-dimensional matrix stiffness in regulating the chemoresistance of hepatocellular carcinoma..., Liu [/bib_ref]. In addition, it is reported that IL-8, inflammatory cytokines, secreted from cancer cells cultured within alginate gels under the hypoxia, was high compared with in 2D culture system [bib_ref] 3D culture broadly regulates tumor cell hypoxia response and angiogenesis via pro-inflammatory..., Delnero [/bib_ref]. Alginate is widely used as a material of cell encapsulation or scaffold for the 3D culture system of cancer cells.
## Collagen
Collagen is the main protein of most tissues and contributes to the physical support of tissues [bib_ref] Collagens-Structure, function, and biosynthesis, Gelse [/bib_ref]. Therefore, collagen is widely used as a material for nerve [bib_ref] Electrospun Nanofibrous Materials for Neural Tissue Engineering, Lee [/bib_ref] [bib_ref] Nanofibrous Collagen Nerve Conduits for Spinal Cord Repair, Liu [/bib_ref] [bib_ref] Pre-differentiation of mesenchymal stromal cells in combination with a microstructured nerve guide..., Boecker [/bib_ref] , bone [bib_ref] 3D printing of composite calcium phosphate and collagen scaffolds for bone regeneration, Inzana [/bib_ref] [bib_ref] 3D Scaffolds with Different Stiffness but the Same Microstructure for Bone Tissue..., Chen [/bib_ref] [bib_ref] Cell-free scaffolds with different stiffness but same microstructure promote bone regeneration in..., Chen [/bib_ref] , cartilage [bib_ref] Enhancement of chondrogenic differentiation of rabbit mesenchymal stem cells by oriented nanofiber..., Zheng [/bib_ref] [bib_ref] Articular cartilage repair with recombinant human type II collagen/polylactide scaffold in a..., Muhonen [/bib_ref] [bib_ref] Mechanical characterization of matrix-induced autologous chondrocyte implantation (MACI(R)) grafts in an equine..., Griffin [/bib_ref] [bib_ref] Matrix-induced autologous chondrocyte implantation (MACI) in the knee: Clinical outcomes and challenges, Basad [/bib_ref] , tendon [bib_ref] Purified collagen I oriented membrane for tendon repair: An ex vivo morphological..., Gigante [/bib_ref] , ligament [bib_ref] Electrospun fibre diameter, not alignment, affects mesenchymal stem cell differentiation into the..., Cardwell [/bib_ref] [bib_ref] A novel fabrication method to create a thick collagen bundle composed of..., Yunoki [/bib_ref] , or skin [bib_ref] Collagen/chitosan porous scaffolds with improved biostability for skin tissue engineering, Ma [/bib_ref] [bib_ref] Electrospinning of collagen nanofibers: Effects on the behavior of normal human keratinocytes..., Rho [/bib_ref] tissue engineering. Chen et al. report that the expression of proangiogenic growth factors and the transcript of MMP of human breast MCF-7 cancer cells cultured on collagen sponges increased [bib_ref] The enhancement of cancer stem cell properties of MCF-7 cells in 3D..., Chen [/bib_ref]. For the 3D cancer cell culture, collagen is often used to evaluate the invasion ability of breast cancer cells. This may be mainly because it has been reported that breast cancer cells prefer to migrate into collagen I [bib_ref] ECM microenvironment regulates collective migration and local dissemination in normal and malignant..., Nguyen-Ngoc [/bib_ref]. When high-invasive breast MDA-MB-231 cancer cells were cultured on a collagen scaffold, the migration ability increased via the epithelial-mesenchymal transition (EMT) [bib_ref] Development of three-dimensional collagen scaffolds with controlled architecture for cell migration studies..., Campbell [/bib_ref]. For the bone metastasis models, Bersini et al. prepared collagen hydrogels containing osteoblasts cells on a microfluidic device. Human breast MDA-MB-231 cancer cells were invaded into the collagen hydrogels embedding osteoblasts cells effectively via the CXCL5/CXCR2 system compared with the collagen hydrogel without cells [bib_ref] A microfluidic 3D in vitro model for specificity of breast cancer metastasis..., Bersini [/bib_ref]. It is demonstrated that the migration ability of breast cancer cells was induced by the degree of collagen fiber alignment or the fibril bending stiffness of the collagen matrix [bib_ref] Fibril bending stiffness of 3D collagen matrices instructs spreading and clustering of..., Sapudom [/bib_ref].
## Hyaluronic acid
Mucopolysaccharide, namely GAG, repeating units of amino acid and uronic acid, is a major ECM component in connective, epithelial, and neural tissues. Hyaluronic acid (HA) is a GAG family and is composed of D-glucuronic acid and D-N-acetylglucosamine [bib_ref] Hyaluronic acid (hyaluronan): A review, Necas [/bib_ref]. The advantageous characteristic of HA is recognized by the CD44 surface receptor [bib_ref] CD44-mediated uptake and degradation of hyaluronan, Knudson [/bib_ref]. The interaction between HA and cells via the CD44 receptor affects the cell functions [bib_ref] Chondrogenesis in a hyaluronic acid scaffold: Comparison between chondrocytes and MSC from..., Jakobsen [/bib_ref]. For cancer, the HA-CD44 interaction leads to the cancer invasion [bib_ref] Hyaluronan-CD44 interaction promotes c-Src-mediated twist signaling, microRNA-10b expression, and RhoA/RhoC up-regulation, leading..., Bourguignon [/bib_ref] , MMP-2 secretion [bib_ref] Hyaluronan-CD44s signaling regulates matrix metalloproteinase-2 secretion in a human lung carcinoma cell..., Zhang [/bib_ref] , RhoGTPase activation or c-Src phosphorylation [bib_ref] CD44 interaction with c-Src kinase promotes cortactin-mediated cytoskeleton function and hyaluronic acid-dependent..., Bourguignon [/bib_ref] , and the expression of TGF-β1 and basic fibroblast growth factor (b-FGF) [bib_ref] CD44 is the principal mediator of hyaluronic-acid-induced melanoma cell proliferation, Ahrens [/bib_ref]. Moreover, HA affects the stemness maintenance of cancer cells, leading to tumorigenesis, EMT, or drug resistance because CD44 is a major surface marker for stem cells [bib_ref] The role of the hyaluronan receptor CD44 in mesenchymal stem cell migration..., Zhu [/bib_ref] [bib_ref] MSC surface markers (CD44, CD73, and CD90) can identify human MSC-derived extracellular..., Sanchez-Abarca [/bib_ref]. It has been demonstrated that the higher expression of HA in the cancer environment increased the cancer progression, leading to the poor mortality rate [bib_ref] Hyaluronan in peritumoral stroma and malignant cells associates with breast cancer spreading..., Auvinen [/bib_ref]. In addition, the molecular weight of HA is also one of the most important factors for cell response. Rayahin et al. report that the molecular weight of HA affects the macrophage phenotypes. At a low molecular weight (5 kDa), the secretion of TNF-α and nitrite production increased. HA of high molecular weight (3 MDa) enhanced the alginase activity which is the characteristic of M2-type macrophages [bib_ref] High and low molecular weight hyaluronic acid differentially influence macrophage activation, Rayahin [/bib_ref]. Therefore, when HA is selected for a 3D cell culture system, the molecular weight of HA should be sufficiently considered because macrophage phenotypes affect the characterization of cancer cells. David et al. report a 3D culture system of cancer cells by use of HA hydrogels crosslinked with adipic dihydrazide to evaluate the invasion ability of several cancer cell lines [bib_ref] Reticulated hyaluronan hydrogels: A model for examining cancer cell invasion in 3D, David [/bib_ref]. It is found by the same groups that the drug resistance enhanced on the same culture systems compared with that in the 2D culture [bib_ref] Hyaluronan hydrogel: An appropriate three-dimensional model for evaluation of anticancer drug sensitivity, David [/bib_ref].
## Matrigel
Basement membrane (BM), a thin layer of ECM, is between the epithelial and stromal sites [bib_ref] Matrigel: From discovery and ECM mimicry to assays and models for cancer..., Benton [/bib_ref] [fig_ref] Figure 2: Cancer cells interact with various stromal cells of cancer-associated fibroblasts [/fig_ref]. BM has a major role in tissue integrity, specificity, and separation [bib_ref] Laminin isoforms in endothelial and perivascular basement membranes, Yousif [/bib_ref]. The components of BM are collagen type IV, laminin, heparan sulfate proteoglycan, various growth factors, cytokines, and chemokines [bib_ref] Matrigel: Basement membrane matrix with biological activity, Kleinman [/bib_ref]. Although BM is an essential material for biological research, human BM of physiological integrity cannot be obtained. As an alternative, matrigel, an extract of Engelbreth-Holm-Swarm tumor derived from wild mice, is used in vitro and in vivo [bib_ref] Matrigel: Basement membrane matrix with biological activity, Kleinman [/bib_ref]. The major component of matrigel is laminin-111, and gelation is formed at 37 - C [bib_ref] Laminin-111-derived peptides and cancer, Kikkawa [/bib_ref].
Kramer et al. report on the investigation method of human HT1080 fibrosarcoma cells by use of matrigel [bib_ref] Invasion of reconstituted basement membrane matrix by metastatic human tumor cells, Kramer [/bib_ref]. After that, matrigel is often used for cancer invasion assay [bib_ref] Cells assemble invadopodia-like structures and invade into matrigel in a matrix metalloprotease..., Yu [/bib_ref] [bib_ref] Enhanced invasion of metastatic cancer cells via extracellular matrix interface, Zhu [/bib_ref]. Matrigel enables the evaluation of not only the cancer invasion ability but also morphology. High-invasive MDA-MB-231 breast cancer cells cultured on matrigel grew, forming a star-like appearance (invasive characterization), while near-sphere cell aggregates were formed when low-invasive breast MCF-7 cancer cells were cultured [bib_ref] The morphologies of breast cancer cell lines in three-dimensional assays correlate with..., Kenny [/bib_ref]. Nowadays, the Boyden chamber has been developed to widely investigate cancer invasion as a reliable method [bib_ref] A rapid in vitro assay for quantitating the invasive potential of tumor..., Albini [/bib_ref] [bib_ref] The 'chemoinvasion' assay, 25 years and still going strong: The use of..., Albini [/bib_ref] [bib_ref] MicroRNA-218 inhibits glioma invasion, migration, proliferation, and cancer stem-like cell self-renewal by..., Tu [/bib_ref]. The two chambers are separated via matrigel-coated porous filter. Cancer cells are plated in the upper chamber, while the medium with or without invasion modulators are in the feeder chamber. When the high-invasion cancer cells are plated, the filter is degraded, leading to the migration of cancer cells and their localization on the feeder surface of filter. Cancer cells migrated are easily counted by the trypan blue stain or fluorescence intensity. The merit of this assay is not to take a long time (12-24 h) to evaluate [bib_ref] Matrigel: From discovery and ECM mimicry to assays and models for cancer..., Benton [/bib_ref]. The Boyden chamber is a powerful tool to evaluate the cancer invasion ability or perform a drug screening.
## Poly (lactic-co-glycolic acid)
Poly (lactic-co-glycolic acid) (PLGA) of biodegradable lactic acid (LA) and glycolic acid (GA) copolymers are widely used for biomedical applications [bib_ref] An overview of poly(lactic-co-glycolic) acid (PLGA)-based biomaterials for bone tissue engineering, Gentile [/bib_ref]. As an example, leuprolide-loaded PLGA microparticles are used for the treatment of breast or prostate cancer. The microparticles realize an extended release of leuprorelin, which enables once every few months [bib_ref] In situ forming PLGA implant for 90days controlled release of leuprolide acetate..., Enayati [/bib_ref]. The basic properties of PLGA are usually given by molecular weight and the LA/GA ratio. For example, PLGA7520 indicates a copolymer of 20,000 molecular weight, and 75 wt % PLA and 25 wt % PGA. Both the molecular weight and LA/GA ratio determine the crystallinity or glass transition temperature [bib_ref] Factors affecting the degradation rate of poly(lactide-co-glycolide) microspheres in vivo and in..., Tracy [/bib_ref] , which enables the control of the size, porosity, or stiffness of PLGA particles or scaffolds easily [bib_ref] An overview of poly(lactic-co-glycolic) acid (PLGA)-based biomaterials for bone tissue engineering, Gentile [/bib_ref] [bib_ref] Polymer and microsphere blending to alter the release of a peptide from..., Ravivarapu [/bib_ref] [bib_ref] Effects of the conformation of PLGA molecules in the organic solvent on..., Nii [/bib_ref] [bib_ref] Poly(lactide-co-glycolide) porous scaffolds for tissue engineering and regenerative medicine, Pan [/bib_ref] [bib_ref] Physicochemical Properties and Applications of Poly(lactic-co-glycolic acid) for Use in Bone Regeneration, Lanao [/bib_ref].
Due to the easiness of the functional control, PLGA particles or scaffolds are also used for the 3D culture system of cancer cells. Sahoo et al. prepare PLGA scaffolds for the human breast MCF-7 cancer cell line by a solvent evaporation method. Since the PLGA scaffolds are hydrophobic, the difficulty of wetting and swelling in the culture medium is often a problem. The incorporation of poly (vinyl alcohol) (PVA) into the scaffolds enhanced the hydrophilic nature, leading to improved cell adherence and proliferation [bib_ref] Characterization of porous PLGA/PLA microparticles as a scaffold for three dimensional growth..., Sahoo [/bib_ref]. Besides breast cancer cells, several PLGA sponges have been prepared for a cell line of human liver Hep3B cancer by changing the LA/GA ratio. The sponges were prepared by a supercritical CO 2 gas-foaming method. The growth, mitochondrial activity, DNA amounts, hepatic function, and invasion ability of Hep3B cells on the sponges became maximum at the ratio of 85/15 [bib_ref] Characterization of porous poly(D,L-lacticco-glycolic acid) sponges fabricated by supercritical CO2 gas-foaming method..., Zhu [/bib_ref]. In addition, PLGA porous microparticles have been prepared for ovarian HO-8910 cancer cell growth [bib_ref] The controllable preparation of porous PLGA microspheres by the oil/water emulsion method..., Zhang [/bib_ref].
## Polyethylene glycol
Polyethylene glycol (PEG) is widely used for chemical modification in the field of drug delivery system or biomaterials [bib_ref] PEG-A versatile conjugating ligand for drugs and drug delivery systems, Kolate [/bib_ref]. PEG-based hydrogels are studied for the 3D cell culture system to investigate the migration of human fibrosarcoma HT-1080 cell line [bib_ref] A peptide functionalized poly(ethylene glycol) (PEG) hydrogel for investigating the influence of..., Singh [/bib_ref] or to mimic the prostate cancer environment [bib_ref] Phenotypic Characterization of Prostate Cancer LNCaP Cells Cultured within a Bioengineered Microenvironment, Sieh [/bib_ref]. PEG scaffolds in a layer-by-layer fashion with tunable stiffness are reported to evaluate the cell mortality [bib_ref] Cancer cell migration within 3D layer-by-layer microfabricated photocrosslinked PEG scaffolds with tunable..., Soman [/bib_ref]. In addition, Yang et al. report that the mouse breast 4T1 cancer cells are encapsulated in inert PEG hydrogels. The PEG hydrogels enabled cancer cells to form tumorspheres and maintain the cancer stemness [bib_ref] Three-dimensional-engineered matrix to study cancer stem cells and tumorsphere formation: Effect of..., Yang [/bib_ref].
## 3d culture system of cancer cells with combination of several biomaterials
Considering unique properties and functions of each biomaterial, different biomaterials are often combined to use for 3D culture system of cancer cells. In this chapter, the 3D culture systems of cancer cells with combined biomaterials are introduced.
## Chitosan-alginate
Chitosan forms insoluble ionic complexes with alginate to improve the mechanical strength or replicate cancer environment [bib_ref] Study on Alginate-Chitosan Complex Formed with Different Polymers Ratio, Kulig [/bib_ref] [bib_ref] Chitosan and alginate types of bio-membrane in fuel cell application: An overview, Shaari [/bib_ref] [bib_ref] Polyelectrolyte Complexes of Chitosan: Formation, Properties, and Applications, Krayukhina [/bib_ref]. Chitosan and alginate (CA) hybrid materials are used to create a 3D material with an interconnected and porous structure. The CA materials have a mechanical strength and shape maintenance significantly improved as compared with chitosan only. This is due to the electrostatic interaction between the amine groups of chitosan and the carboxyl groups of alginate [bib_ref] Chitosan-alginate hybrid scaffolds for bone tissue engineering, Li [/bib_ref]. When human liver HepG2 cancer cells were cultured on the CA scaffolds, both the malignancy and drug resistance increased [bib_ref] Chitosan-alginate scaffold culture system for hepatocellular carcinoma increases malignancy and drug resistance, Leung [/bib_ref]. The CA scaffolds can be applied not only for hepatocellular carcinoma cells, but also for human glioblastoma U-87 MG and U-118 MG cell lines. The expression levels of genes involved in EMT or cancer stem cells were rapidly promoted [bib_ref] Chitosan-alginate 3D scaffolds as a mimic of the glioma tumor microenvironment, Kievit [/bib_ref] [bib_ref] Proliferation and enrichment of CD133(+) glioblastoma cancer stem cells on 3D chitosan-alginate..., Kievit [/bib_ref].
## Chitosan-hyaluronic acid
The mixed hydrogel of chitosan and hyaluronic acid (CH) is often used as a nonadhesive material for spheroids formation. The CH has an ability to maintain the stemness of MSC spheroids through the Rho/Rock activation. A short time of spheroid formation and the enlargement of spheroid size were achieved compared with the conventional culture system [bib_ref] Spheroid formation of mesenchymal stem cells on chitosan and chitosan-hyaluronan membranes, Huang [/bib_ref]. When the 3D spheroids of human nonsmall cell lung cancer cells were prepared on the CH membrane, the expression level of EMT marker, the stemness, or the drug resistance increased compared with those of cells in the 2D culture system [bib_ref] Acquisition of epithelial-mesenchymal transition and cancer stem-like phenotypes within chitosan-hyaluronan membrane-derived 3D..., Huang [/bib_ref]. In addition, upon culturing on the CH scaffolds, the expression of stem cell marker and drug resistance of 3D human glioblastoma cancer stem cells was enhanced [bib_ref] Culture on 3D Chitosan-Hyaluronic Acid Scaffolds Enhances Stem Cell Marker Expression and..., Wang [/bib_ref]. A porous CH scaffold promoted the formation of cancer spheroids and their stemness [bib_ref] Porous chitosan-hyaluronic acid scaffolds as a mimic of glioblastoma microenvironment ECM, Florczyk [/bib_ref].
## Matrigel-collagen or alginate
Nguyen-Ngoc et al. formulate matrigel hydrogels embedding human breast cancer cell aggregates. Cancer cells are individually dissociated from aggregates to promote their invasion nature because matrigel gives cancer cells a suitable environment. Moreover, the addition of collagen type I into the matrigel increased further cancer invasion [bib_ref] ECM microenvironment regulates collective migration and local dissemination in normal and malignant..., Nguyen-Ngoc [/bib_ref]. It is reported that the mixed alginate matrigel hydrogel (a mixing ratio of 50:50) enabled human breast cancer cells incorporated to replicate the cancer invasion [bib_ref] A new cell-laden 3D Alginate-Matrigel hydrogel resembles human breast cancer cell malignant..., Cavo [/bib_ref].
# Polyethylene glycol-other biomaterials
For the formation of cancer cell scaffolds, PEG is often conjugated with various biomaterials of collagen [bib_ref] Mechanical confinement via a PEG/Collagen interpenetrating network inhibits behavior characteristic of malignant..., Reynolds [/bib_ref] , HA [bib_ref] Bioengineered 3D brain tumor model to elucidate the effects of matrix stiffness..., Wang [/bib_ref] , PLGA [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref] , fibrin [bib_ref] 3D modeling of human cancer: A PEG-fibrin hydrogel system to study the..., Bufalo [/bib_ref] , and fibrinogen [bib_ref] PEG-fibrinogen hydrogels for three-dimensional breast cancer cell culture, Pradhan [/bib_ref] [bib_ref] A three-dimensional spheroidal cancer model based on PEG-fibrinogen hydrogel microspheres, Pradhan [/bib_ref] [bib_ref] PEG-A versatile conjugating ligand for drugs and drug delivery systems, Kolate [/bib_ref]. PEG/collagen hydrogels of interpenetrating network are prepared to investigate the functions of human breast cancer cells, such as their proliferation, viability, or migration [bib_ref] Mechanical confinement via a PEG/Collagen interpenetrating network inhibits behavior characteristic of malignant..., Reynolds [/bib_ref]. PEG/HA hydrogels with different stiffness are prepared by changing the PEG concentration to investigate the behavior of brain cancer cells embedded into the hydrogels [bib_ref] Bioengineered 3D brain tumor model to elucidate the effects of matrix stiffness..., Wang [/bib_ref]. Lipke groups have intensively studied the function of cancer cells cultured with PEG/fibrinogen materials [bib_ref] PEG-fibrinogen hydrogels for three-dimensional breast cancer cell culture, Pradhan [/bib_ref] [bib_ref] A three-dimensional spheroidal cancer model based on PEG-fibrinogen hydrogel microspheres, Pradhan [/bib_ref]. Fibrinogen is one of the ECM components and has an important role in the polymerization or deposition of collagen [bib_ref] Polymerization of type I and III collagens is dependent on fibronectin and..., Velling [/bib_ref]. Breast cancers [bib_ref] PEG-fibrinogen hydrogels for three-dimensional breast cancer cell culture, Pradhan [/bib_ref] and colon or prostate cancer cells [bib_ref] A three-dimensional spheroidal cancer model based on PEG-fibrinogen hydrogel microspheres, Pradhan [/bib_ref] are embedded in the 3D PEG/fibrinogen hydrogel to experimentally confirm the possibility of a long-time culture. Girard et al. culture several cancer cells on the 3D nanofibers of PLGA-PLA-PEG. Tight irregular aggregates were formed similarly to those of cancers in vivo, and the EMT was induced [bib_ref] A 3D fibrous scaffold inducing tumoroids: A platform for anticancer drug development, Girard [/bib_ref].
## 3d coculture system of cancer and stromal cells combined with biomaterials
## Alginate
Coculture of cancer cells and stromal cells with alginate has been investigated. Alginate hydrogels encapsulating human breast MCF-7 cancer cell aggregates were cocultured with human fibroblasts. The oestrogen receptor and the membrane E-cadherin expression increased, the polarity was lost, and the cell migration and angiogenesis increased, in contrast to the monoculture of MCF-7 cells [bib_ref] Modelling the tumour microenvironment in long-term microencapsulated 3D co-cultures recapitulates phenotypic features..., Estrada [/bib_ref]. These phenotypic alterations are important at the advanced stage of cancer. Liu et al. embed hepatocellular carcinoma in the algiante hydrogels, and then, the hydrogels are cocultured with MSC. In this culture system, efficient induction of EMT and the metastasis of cancer cells via TGF-β were observed [bib_ref] Mesenchymal stem cells enhance the metastasis of 3D-cultured hepatocellular carcinoma cells, Liu [/bib_ref].
## Collagen
Nikkhah groups prepare a 3D microengineered cancer model composed of breast cancer cells and CAF embedded into collagen hydrogels. This culture system enabled cancer cells and CAF to achieve their interaction in vitro, which leads to better evaluation of invasion level of cancer cells, MMP secretion, and drug resistance [bib_ref] The Role of Desmoplasia and Stromal Fibroblasts on Anti-cancer Drug Resistance in..., Saini [/bib_ref]. 3D lung or pancreatic cancer cell aggregates embedded in collagen hydrogels are cocultured with CAF. Cancer cells were attached to CAF and quickly migrated on the CAF protrusions, while CAF-free cancer cells hardly invaded into the matrix [bib_ref] Cancer cell migration on elongate protrusions of fibroblasts in collagen matrix, Miyazaki [/bib_ref].
## Gelatin
Collagen of one ECM components is often used in the research field of 3D cell culture. However, collagen is water-insoluble and has biological activities, such as blood coagulation and a specific affinity for humoral factors. Considered as a material to design the cell culture system, the inherent properties are sometimes not suitable. Gelatin, a denatured form of collagen, is a cell friendly (high cell adhesion and low inflammation induction) material and is water-soluble [bib_ref] Osteogenic differentiation of mesenchymal stem cells in biodegradable sponges composed of gelatin..., Takahashi [/bib_ref] [bib_ref] Effect of gelatin hydrogel incorporating fibroblast growth factor 2 on human meniscal..., Narita [/bib_ref]. In addition, it is technologically easy to prepare gelatin with various physicochemical properties by changing the preparation process from collagen [bib_ref] Protein release from gelatin matrices, Ikada [/bib_ref] [bib_ref] The effect of amines added to an alkali-pretreatment on the solubilisation of..., Fujii [/bib_ref]. Hydrogel formulations of water-insoluble gelatin can be freely prepared by the physical or chemical crosslinking methods, while the degradation profile can be modified as well [bib_ref] Protein release from gelatin matrices, Ikada [/bib_ref] [bib_ref] Tissue regeneration based on growth factor release, Tabata [/bib_ref]. The gelatin material is used for a coculture system of cancer cells and stromal cells. Netti groups have extensively investigated cancer microtissues by use of gelatin porous microbeads (GPM). Gelatin scaffolds with interconnected pores of about 20 µm diameter are designed for a 3D culture system, and the microtissues of cancer are formulated [bib_ref] The role of microscaffold properties in controlling the collagen assembly in 3D..., Imparato [/bib_ref]. 3D CAF microtissues with GPM showed the higher deposition of collagen, fibronectin, and hyaluronic acid than that of GPM-free 3D CAF. GPM are effective materials to replicate the 3D cancer-stroma condition in vitro [bib_ref] 3D is not enough: Building up a cell instructive microenvironment for tumoral..., Brancato [/bib_ref]. Moreover, human MCF-7 breast cancer and CAF microtissues with GPM are prepared to mimic the cancer microenvironment. The diffusion coefficient of anticancer drugs and the drug action for the 3D MCF-7-CAF microtissues with GPM were higher than those for the GPM-free 3D MCF-7-CAF. In addition, there was a good correlation of the expression of some cancer biomarkers related to cell junctions between the 3D MCF-7-CAF microtissues combined with GPM and in vivo cancer site [bib_ref] 3D breast cancer microtissue reveals the role of tumor microenvironment on the..., Brancato [/bib_ref]. The combination of endothelial cells with the culture system is reported [bib_ref] Recapitulating spatiotemporal tumor heterogeneity in vitro through engineered breast cancer microtissues, Mazio [/bib_ref].
## Hyaluronic acid
As a coculture system of cancer cells and stromal cells with HA, a multilayer system of high-invasive prostate C4-2B cancer cells, or endometrial Ishikawa cancer cells and stromal cells with HA hydrogels is reported. This culture system enables the evaluation of the cytotoxicity of compounds used clinically for both prostate and endometrial cancer cells in vitro. In addition, it is technically possible to anticipate and identify drugs that fail in clinical trials [bib_ref] Hyaluronic Acid-Based Hydrogel Formulations Suitable for Automated 3D High Throughput Drug Screening..., Engel [/bib_ref]. Han et al. prepare multicellular spheroids of human cell lung carcinoma cell line A549 and human MSC isolated from adipose tissue on CH coating plates. It is found that the gene expression levels of tumorigenicity markers in cancer cells associated with cancer stemness, EMT property, and cell mobility were up-regulated in the MSC-tumor multicellular spheroids [bib_ref] Chitosan-hyaluronan based 3D co-culture platform for studying the crosstalk of lung cancer..., Han [/bib_ref].
## Matrigel
There are several reports on matrigel-assisted coculture systems with stromal cells, such as fibroblasts [bib_ref] Human mammary fibroblasts stimulate invasion of breast cancer cells in a three-dimensional..., Olsen [/bib_ref] , regulatory T lymphocyte (T REG lymphocyte) or natural killer cells (NK cells) [bib_ref] Establishment of a heterotypic 3D culture system to evaluate the interaction of..., Augustine [/bib_ref] , and MSC [bib_ref] Human bone marrow stromal cells enhance breast cancer cell growth rates in..., Sasser [/bib_ref]. Augustine et al. culture both T REG lymphocytes and NK cells with luminal phenotype MCF-7 and basal phenotype MDA-MB-231 to study the immune reaction of breast cancer progression. Cancer morphology, the expression of biomarkers, and CC-chemokine 4 (CCL4) secretion were influenced by the phenotype of breast cancer cells and their immune stimulation [bib_ref] Establishment of a heterotypic 3D culture system to evaluate the interaction of..., Augustine [/bib_ref]. MSC are cocultured with estrogen receptor-positive breast cancer cells embedded in matrigel. Cancer cells rapidly proliferated compared with the MSC-free cells [bib_ref] Human bone marrow stromal cells enhance breast cancer cell growth rates in..., Sasser [/bib_ref].
## Collagen-alginate
Mixed hydrogels of collagen and alginate are investigated to form the multicellular spheroids of human breast cancer cells and fibroblasts. The hydrogel system developed in this study enables the control of the stiffness without altering the major gel components, since the concentration of alginate and collagen in the hydrogel remains constant. The change in the degree of calcium crosslinking does not affect the cell adhesion on the collagen network [bib_ref] Hybrid collagen alginate hydrogel as a platform for 3D tumor spheroid invasion, Liu [/bib_ref]. Alginate has been extensively used as a material whose stiffness can be readily regulated.
An increase in ECM stiffness is involved in the cancer progression [bib_ref] Mechanical stiffness of liver tissues in relation to integrin beta1 expression may..., Zhao [/bib_ref]. In addition, there have been reports on the relationship between the stiffness and drug resistance [bib_ref] Extracellular matrix proteins protect small cell lung cancer cells against apoptosis: A..., Sethi [/bib_ref] [bib_ref] The desmoplastic reaction surrounding hepatic colorectal adenocarcinoma metastases aids tumor growth and..., Conti [/bib_ref]. Based on these findings, it is important to design the 3D culture system of cancer cells by making use of biomaterials of which the stiffness can be changed. It has been recently reported that the stiffness of biomaterials affects the characteristics of cancer cells, such as drug resistance [bib_ref] Role of three-dimensional matrix stiffness in regulating the chemoresistance of hepatocellular carcinoma..., Liu [/bib_ref] [bib_ref] Substrate stiffness influences the outcome of antitumor drug screening in vitro, Feng [/bib_ref] [bib_ref] Cell responses to the mechanochemical microenvironment-implications for regenerative medicine and drug delivery, Rehfeldt [/bib_ref] [bib_ref] Hypoxia-mediated drug resistance: Novel insights on the functional interaction of HIFs and..., Rohwer [/bib_ref]. It is promising for the 3D coculture of cancer and stromal cells to use biomaterials of the right material for the right place.
## 3d coculture system of cancer and stromal cells combined with biomaterials of drug delivery system
The drug delivery system (DDS) is defined as a technology and methodology to enhance the biological activities of drugs or reduce the adverse effects by appropriately combining with biomaterials. To date, the DDS has been mainly used for in vivo cancer therapy through drug delivery [bib_ref] Preclinical and clinical studies of anticancer agent-incorporating polymer micelles, Matsumura [/bib_ref] [bib_ref] A Phase I clinical study of cisplatin-incorporated polymeric micelles (NC-6004) in patients..., Plummer [/bib_ref] [bib_ref] Novel cisplatin-incorporated polymeric micelles can eradicate solid tumors in mice, Nishiyama [/bib_ref]. However, the technology and methodology are also applicable for drug screening because cancer-environmental normal cell interaction is biologically supported by humoral factors secreted from the cells [bib_ref] Cancer-Associated Fibroblasts: Their Characteristics and Their Roles in Tumor Growth, Shiga [/bib_ref] [bib_ref] Interaction of MSC with tumor cells, Melzer [/bib_ref] [bib_ref] Tumor microenvironment: Bone marrow-mesenchymal stem cells as key players, Barcellos-De-Souza [/bib_ref] [bib_ref] Cancer and the tumor microenvironment: A review of an essential relationship, Mbeunkui [/bib_ref] [bib_ref] Tumor-associated macrophages as major players in the tumor microenvironment, Chanmee [/bib_ref] [bib_ref] Heterogeneity of stromal fibroblasts in tumors, Orimo [/bib_ref] [bib_ref] Cancer associated fibroblasts stimulated by transforming growth factor beta1 (TGF-beta 1) increase..., Casey [/bib_ref]. The combination of humoral factors in the DDS will enable the enhancement of the interaction between cancer and stromal cells which physiologically takes place in the body.
Gelatin hydrogel microspheres (GM) for regenerative medicine have been explored. GM can incorporate various growth factors, such as b-FGF [bib_ref] Vascularization effect of basic fibroblast growth factor released from gelatin hydrogels with..., Tabata [/bib_ref] [bib_ref] Biodegradation of hydrogel carrier incorporating fibroblast growth factor, Tabata [/bib_ref] [bib_ref] Neovascularization effect of biodegradable gelatin microspheres incorporating basic fibroblast growth factor, Tabata [/bib_ref] [bib_ref] In vitro sorption and desorption of basic fibroblast growth factor from biodegradable..., Tabata [/bib_ref] , TGF-β1 [bib_ref] In vitro proliferation and chondrogenic differentiation of rat bone marrow stem cells..., Ogawa [/bib_ref] [bib_ref] Controlled release of growth factors based on biodegradation of gelatin hydrogel, Yamamoto [/bib_ref] , insulin-like growth factor-1 [bib_ref] Novel therapy for hearing loss: Delivery of insulin-like growth factor 1 to..., Lee [/bib_ref] [bib_ref] Topical insulin-like growth factor 1 treatment using gelatin hydrogels for glucocorticoid-resistant sudden..., Nakagawa [/bib_ref] , or SDF-1 [bib_ref] Controlled release of stromal-cell-derived factor-1 from gelatin hydrogels enhances angiogenesis, Kimura [/bib_ref] for controlled release. Growth factors and gelatin molecules effectively interact by physicochemical interaction (e.g., ionic or hydrogen interaction) [bib_ref] Protein release from gelatin matrices, Ikada [/bib_ref]. Due to the interaction, the mechanism of gelatin matrix-degradation-driven drug release is achievable. This is different from the conventional release system where the drug is usually released from release matrices by the drug diffusion. In addition, GM are in vivo and in vitro enzymatically degraded with time, and finally disappear. The characteristic behavior of GM disappearance is essential as a material for drug release used for tissue regeneration. To repair the damaged tissues, cells should migrate, proliferate, and differentiate. If drug release materials remain for a long time period after drug release is completed, the material remaining will cause the physical impairment of tissue regeneration. The speed of tissue regeneration should be synchronized to that of material degradation. Taken together, the growth factor release as the result of GM degradation with time is effective in realizing tissue regeneration based on the cell activity enhancement for natural healing potential [bib_ref] A Cancer Invasion Model Combined with Cancer-Associated Fibroblasts Aggregates Incorporating Gelatin Hydrogel..., Nii [/bib_ref] [bib_ref] Effect of gelatin hydrogel incorporating fibroblast growth factor 2 on human meniscal..., Narita [/bib_ref] [bib_ref] Effect of culture substrates and fibroblast growth factor addition on the proliferation..., Hori [/bib_ref] [bib_ref] Initial bone regeneration around fenestrated implants in Beagle dogs using basic fibroblast..., Akagawa [/bib_ref] [bib_ref] In situ regeneration of adipose tissue in rat fat pad by combining..., Hiraoka [/bib_ref] [bib_ref] In vivo degradability of hydrogels prepared from different gelatins by various cross-linking..., Ozeki [/bib_ref] [bib_ref] Promotion of bone regeneration by CCN2 incorporated into gelatin hydrogel, Kikuchi [/bib_ref]. In addition, a water phase of GM matrices is a pathway to permeate oxygen or nutrients [bib_ref] Influence of shaking culture on the biological functions of cell aggregates incorporating..., Nii [/bib_ref]. This permeability is very important considering the 3D cell culture because cells in cell aggregates easily die because of the lack of oxygen or nutrients [bib_ref] Determination of oxygen gradients in engineered tissue using a fluorescent sensor, Kellner [/bib_ref] [bib_ref] The role of hypoxia and platelets in air travel-related venous thromboembolism, Bradford [/bib_ref] [bib_ref] Preparation and functional evaluation of cell aggregates incorporating gelatin microspheres with different..., Tajima [/bib_ref]. As a trial to break through the issue and culture 3D cell aggregates for a long time period, GM incorporation into the aggregates has been attempted [bib_ref] Preparation of stem cell aggregates with gelatin microspheres to enhance biological functions, Hayashi [/bib_ref] [bib_ref] Preparation of cell aggregates incorporating gelatin hydrogel microspheres of sugar-responsive water solubilization, Inoo [/bib_ref] [bib_ref] Preparation of cell aggregates incorporating gelatin hydrogel microspheres containing bone morphogenic protein-2..., Tajima [/bib_ref]. Moreover, to enhance the cell activity, drugs to activate the cell function can be impregnated into GM for sustained release. Incorporation of GM containing drugs in cell aggregates is useful to give cells cultured in the 3D system a better condition. It is reported that CAF aggregates incorporating GM containing TGF-β1 (3D CAF-GM-TGF-β1) showed an activated function of CAF. When the activated CAF aggregates and cancer cells were cocultured via a model basement membrane, the invasion rate of cancer cells through the membrane was significantly higher than that of 2D cultured CAF [fig_ref] Figure 3: Illustration of cancer invasion based on a combination of 3D cell culture... [/fig_ref]. The findings indicate that the combination of 3D cell culture and DDS technology is promising to enhance the activity of cancer cells in the 3D culture system. TAM aggregates incorporating GM containing adenosines (3D TAM-GM-adenosines) were formulated to activate and maintain TAM functions. It is found that a 3D cancer cell coculture system of combined 3D CAF-GM-TGF-β1 and 3D TAM-GM-adenosines enabled the effective evaluation of the in vitro invasion of various cancer cells [bib_ref] A co-culture system of three-dimensional tumor-associated macrophages and three-dimensional cancer-associated fibroblasts combined..., Nii [/bib_ref]. The body tissue fundamentally consists of cells and the surrounding environment. The environment generally is made of ECM and nutrients for cells. In the case that the two factors of cell environment were not biologically sufficient, the functions of cells would rapidly decrease. The gelatin hydrogel microspheres (GM) function not only as the cell scaffold, but also as the release carrier of TGF-β1 and adenosines of nutrients for CAF and TAM.
## Future prospective and conclusion
Biomaterials can assist the 3D culture system of cancer cells through the biological induction of ECM components. Several studies have reported on 3D culture systems by taking advantage of biomaterials. For further development of the 3D culture system of cancer cells, several biomaterials should be combined considering their unique properties and functions. In addition, substantial and close interaction between tissue engineering and the biological research of cancer cells or cancer environment would bring about further development of the 3D cell culture system for anticancer drug screening. In future, patient-derived cancer cells or stromal cells should be combined with biomaterials selected to allow the culture system to approach a more realistic cancer environment. The 3D culture system with biomaterials is a promising tool for cancer research and anticancer drug screening.
## Conflicts of interest:
The authors declare no conflict of interest.
[fig] Figure 1: Research and development process of drug development. The difference in the environment condition between in vitro and in vivo leads to that in drug effects, which often causes a failure in drug development. [/fig]
[fig] Figure 2: Cancer cells interact with various stromal cells of cancer-associated fibroblasts (CAF), tumor-associated macrophages (TAM), mesenchymal stem cells (MSC), and endothelial cells (EC), leading to the pathological maintenance and promotion of cancer characteristics. [/fig]
[fig] Figure 3: Illustration of cancer invasion based on a combination of 3D cell culture and drug delivery system technology. [/fig]
[fig] Author: Contributions: Conceptualization, T.N., K.M., and Y.T.; writing-original draft preparation, T.N. and Y.T.; writing-review and editing, T.N. and Y.T. All authors have read and agreed to the published version of the manuscript. [/fig]
[fig] Funding: This research received no external funding. [/fig]
[table] Table 1: Comparison of cancer cells culture between 2D and 3D systems. [/table]
[table] Table 3: 3D culture system of cancer cells combined with biomaterials. [/table]
|
10.1016/S2589-7500(21)00064-9
|
CCBY
| null |
34020933
|
s2orc_pubmed_articles
|
The hopes and hazards of using personal health technologies in the diagnosis and prognosis of infections.
Work from numerous groups has shown the potential of using data from wearable devices to characterise each individual's unique baseline, identify deviations from that baseline suggestive of a viral infection, and to aggregate that data to better inform population surveillance trends. With a growing global population of connected wearable users, this could potentially help improve the earlier diagnosis and management of infectious individuals and improve timeliness and precision of tracking infectious disease outbreaks. However, despite these possibilities, there are important considerations when interpreting wearable data, including generalisability of user populations, sensor accuracy and comparability, and overfitting of models. Additionally, before deploying such tools as a global health solution, the proactive integration of community engagement through a user-centred design is, at a minimum, required to begin mitigating the risk of digital health illiteracy, structural inequality, and social marginality. Regional frameworks establishing transparent standards for participant data protection as data privacy, data use, and data rights are required to truly protect participant's rights, empower participants, and minimise risk of harm. As these sensors continue to evolve, standardised data and reporting, as well as collaboration and data sharing across study and technology groups will also be necessary.Personal health technologies-technologies (consumer or medical grade) used by an individual that generate data from that individual-offer an unprecedented opportunity to completely alter how we currently detect and manage infectious diseases at an individual as well as population level. Currently, management decisions in the setting of possible infection are made based on symptoms and objective physiological measurements, whether the person primarily making those decisions is the patient or a health-care provider. Because symptoms are often non-specific, the recognition of abnormalities in physiological parameters have an important role in diagnosing and guiding therapeutic interventions. However, abnormality has historically been defined by what is normal for a healthy population, rather than what is normal for an individual. For example, an elevated temperature, which has been recognised as a hallmark of infection since the beginning of recorded history, is commonly (but not consistently) classified as a temperature of 38°C or higher. 1 Similarly, a respiratory rate of more than 20 breaths per min or a heart rate of more than 100 beats per min in an adult would also be considered abnormal and might indicate a person who requires expedited, or a higher level of care. However, what is actually a person's "normal" varies substantially between individuals. For oral temperature, that range falls between 35·7°C and 37·4°C, 2 respiratory rate between 12 and 20 breaths per min, 3 and resting pulse rate between 40 and 109 beats per min. 4 Therefore, these broad population ranges are imprecise when applied to a specific individual, especially for identifying an early change in their physiological status. Individuals experience daily, weekly, and seasonal fluctuations unique to them in a range of physiological parameters and activities.4,5Only by knowing what is normal for an individual when they are well is it possible to identify the earliest possible deviations from that normal. This is what personal health technologies make possible, in the real world outside of a health-care setting, and in a nearly passive manner.Because of the emergence of COVID-19, numerous studies examining personal sensor data have found digital signals supporting the potential benefit of using these technologies to identify and track viral illness. Continuously evolving personal health technologies can be harnessed to collect unique baseline data on individuals and populations, which allows for earlier and more precise detection of viral illness; but for these tools to be successful, biases related to health inequity, loss to follow-up, an absence of data harmonisation, and others need to be addressed.Usability of personal health technologiesIt is relatively recent that personal health technologies have become available that enable investigators to even consider addressing the potential value of identifying individual changes in physiological parameters in large populations. Wearable sensors available to individuals can now track not only activity, but also body position, heart rate and rhythm, skin temperature, oxygen saturation, an electrocardiogram, and electrodermal activity and sound (figure 1). From these metrics respiratory rate, heart rate variability, heart arrhythmias, sleep, and sleep stages are derived. Other medical-grade wearables, not yet designed for a large consumer market, can also continuously track blood pressure as well as derive systemic vascular resistance and cardiac output.6Wearable sensor technologies are available in a wide range of form factors beyond the most commonly used wrist-worn sensors. There are options for rings, arm bands, earbuds, adhesive patches, and clothing that all are capable of tracking multiple physiological parameters. Each offers certain advantages and disadvantages, but the most effective will be the one that provides the most reliable information that the intended user will be willing
## The hopes and hazards of using personal health technologies in the diagnosis and prognosis of infections
Jennifer M Radin, Giorgio Quer, Marwa Jalili, Dina Hamideh, Steven R Steinhubl Work from numerous groups has shown the potential of using data from wearable devices to characterise each individual's unique baseline, identify deviations from that baseline suggestive of a viral infection, and to aggregate that data to better inform population surveillance trends. With a growing global population of connected wearable users, this could potentially help improve the earlier diagnosis and management of infectious individuals and improve timeliness and precision of tracking infectious disease outbreaks. However, despite these possibilities, there are important considerations when interpreting wearable data, including generalisability of user populations, sensor accuracy and comparability, and overfitting of models. Additionally, before deploying such tools as a global health solution, the proactive integration of community engagement through a user-centred design is, at a minimum, required to begin mitigating the risk of digital health illiteracy, structural inequality, and social marginality. Regional frameworks establishing transparent standards for participant data protection as data privacy, data use, and data rights are required to truly protect participant's rights, empower participants, and minimise risk of harm. As these sensors continue to evolve, standardised data and reporting, as well as collaboration and data sharing across study and technology groups will also be necessary.
Personal health technologies-technologies (consumer or medical grade) used by an individual that generate data from that individual-offer an unprecedented opportunity to completely alter how we currently detect and manage infectious diseases at an individual as well as population level. Currently, management decisions in the setting of possible infection are made based on symptoms and objective physiological measurements, whether the person primarily making those decisions is the patient or a health-care provider. Because symptoms are often non-specific, the recognition of abnormalities in physiological parameters have an important role in diagnosing and guiding therapeutic interventions. However, abnormality has historically been defined by what is normal for a healthy population, rather than what is normal for an individual. For example, an elevated temperature, which has been recognised as a hallmark of infection since the beginning of recorded history, is commonly (but not consistently) classified as a temperature of 38°C or higher. [bib_ref] Fever: its history, cause, and function, Atkins [/bib_ref] Similarly, a respiratory rate of more than 20 breaths per min or a heart rate of more than 100 beats per min in an adult would also be considered abnormal and might indicate a person who requires expedited, or a higher level of care. However, what is actually a person's "normal" varies substantially between individuals. For oral temperature, that range falls between 35·7°C and 37·4°C, 2 respiratory rate between 12 and 20 breaths per min,and resting pulse rate between 40 and 109 beats per min. [bib_ref] Inter-and intraindividual variability in daily resting heart rate and its associations with..., Quer [/bib_ref] Therefore, these broad population ranges are imprecise when applied to a specific individual, especially for identifying an early change in their physiological status. Individuals experience daily, weekly, and seasonal fluctuations unique to them in a range of physiological parameters and activities. [bib_ref] Inter-and intraindividual variability in daily resting heart rate and its associations with..., Quer [/bib_ref] [bib_ref] Association of sleep duration and variability with body mass index: sleep measurements..., Jaiswal [/bib_ref] Only by knowing what is normal for an individual when they are well is it possible to identify the earliest possible deviations from that normal. This is what personal health technologies make possible, in the real world outside of a health-care setting, and in a nearly passive manner.
Because of the emergence of COVID-19, numerous studies examining personal sensor data have found digital signals supporting the potential benefit of using these technologies to identify and track viral illness. Continuously evolving personal health technologies can be harnessed to collect unique baseline data on individuals and populations, which allows for earlier and more precise detection of viral illness; but for these tools to be successful, biases related to health inequity, loss to follow-up, an absence of data harmonisation, and others need to be addressed.
## Usability of personal health technologies
It is relatively recent that personal health technologies have become available that enable investigators to even consider addressing the potential value of identifying individual changes in physiological parameters in large populations. Wearable sensors available to individuals can now track not only activity, but also body position, heart rate and rhythm, skin temperature, oxygen saturation, an electrocardiogram, and electrodermal activity and sound (figure 1). From these metrics respiratory rate, heart rate variability, heart arrhythmias, sleep, and sleep stages are derived. Other medical-grade wearables, not yet designed for a large consumer market, can also continuously track blood pressure as well as derive systemic vascular resistance and cardiac output. [bib_ref] Comparing blood pressure measurements between a photoplethysmography-based and a standard cuff-based manometry..., Nachman [/bib_ref] Wearable sensor technologies are available in a wide range of form factors beyond the most commonly used wrist-worn sensors. There are options for rings, arm bands, earbuds, adhesive patches, and clothing that all are capable of tracking multiple physiological parameters. Each offers certain advantages and disadvantages, but the most effective will be the one that provides the most reliable information that the intended user will be willing e456 www.thelancet.com/digital-health Vol 3 July 2021 Viewpoint and able to use as frequently and for as long as needed. There are little data available on long-term use of any of these wearables, with one survey 7 in Canada finding that of people who purchased their own activity tracker, just 55% were still using it, but those that were still wearing it wore it an average of 23 days in the previous month. In another study 8 that gave participants a wrist wearable, approximately 25% wore the sensor for the majority of the 4-month requested monitoring period. [bib_ref] Usability of a wrist-worn smartwatch in a direct-to-participant randomized pragmatic clinical trial, Galarnyk [/bib_ref] A study of nearly 4·5 million insurance plan members who were offered financial incentives for achieving activity goals found that only 1·2% activated a device, but of those that did, 80% were still using the device at 6 months. [bib_ref] Using wearable devices and smartphones to track physical activity: initial activation, sustained..., Patel [/bib_ref] A meta-analysis 10 found that median participant retention across eight studies was only 5·5 days, and that most studies did not include a population representative of the ethnicities and diversity of the USA.
For many, a non-wearable, passive multiparametric sensor can provide the best option for longitudinal data. For example, under-mattress pads that can monitor heart rate, respiratory rate, and various sleep parameters have been found in preliminary work to identify early physiological decompensation. [bib_ref] Home monitoring of heart failure patients at risk for hospital readmission using..., Bennett [/bib_ref] A wide range of contactless sensors for in-home use that use computervision, infrared thermography, radar, and audio can track individual vital signs, as well as cough frequency and quality, in some cases, even when in different rooms. [bib_ref] Smart homes that monitor breathing and heart rate, Adib [/bib_ref] [bib_ref] Contactless vital signs measurement system using RGB-thermal image sensors and its clinical..., Negishi [/bib_ref] [bib_ref] A broader look: camera-based vital sign estimation across the spectrum, Antink [/bib_ref] [bib_ref] The present and future of cough counting tools, Hall [/bib_ref] As remarkable as progress has been over the last several years in the availability of consumer technologies that monitor and report physiological variables, it is important to recognise that the accuracy of the information provided is variable and should not be considered clinically dependable unless appropriate validation evidence exists, and ideally regulatory agency approval. [bib_ref] An evaluation of biometric monitoring technologies for vital signs in the era..., Manta [/bib_ref] For example, there have been recent concerns raised of the accuracy of oxygen saturation determined not only from the wrist with consumer sensors, 17 but also from hospital-based finger-tip devices due to skin pigmentation differences. [bib_ref] Should you trust Apple's new blood oxygen sensor? View from the valley, Perry [/bib_ref] [bib_ref] Racial bias in pulse oximetry measurement, Sjoding [/bib_ref] To aggregate findings from multiple sensors it will also be important to understand how sensor-specific and usually proprietary algorithms for calculating metrics such as daily resting heart rate and respiration rate can differ across devices.
## Studies of personal health technologies in infectious diseases
There is typically a delay from the time someone gets sick to when they develop symptoms, seek care, get tested, and finally receive a test result for COVID-19 and other viral infections. It then takes an additional 1-3 weeks before test results are collected and aggre gated into a central surveillance system, which often relies on outdated reporting methods such as fax machines.When wearable data is aggregated for a population, it is possible to so called nowcast or track viral activity in real time. Previous studies have shown that identifying resting heart rate and sleep data outside of an individual's normal levels, can be used to improve real-time predic tions for influenza-like illness at the state level. [bib_ref] Wearable sensor data and self-reported symptoms for COVID-19 detection, Quer [/bib_ref] Similarly, models to predict COVID-19 anomalies in China and south-central Europe using wearable data from 1·3 mil lion users who wore Huami smartwatches (Hefei, China) also showed promise [fig_ref] Table 1: Observational studies of wearables in the prediction of viral illnesses [/fig_ref].The Robert Koch institute, Berlin, Germany, has launched a similar fever trend tracker based on resting heart rate and activity data from more than 500 000 participants.Kinsa smart thermometers (San Francisco, CA, USA) have also shown utility in predicting influenza-like illness activity [bib_ref] A smartphone-driven thermometer application for real-time population-and individuallevel influenza surveillance, Miller [/bib_ref] [bib_ref] Assessing the utility of a smart thermometer and mobile application as a..., Ackley [/bib_ref] [bib_ref] Improving state-level influenza surveillance by incorporating real-time smartphone-connected thermometer readings across different..., Miller [/bib_ref] and potentially COVID-19.Novel data streams from sensors can offer key insight into trends, timing of outbreaks, and identifying specific geo graphical hotspots of infec tion. They might prove especially useful when integrated with both traditional clinical and laboratory surveillance and other novel surveillance data, including those from wastewater, internet search terms, social media, and mobility data (figure 2).
Individual-level diagnosis using continuously collected data from wearables has many benefits, including improved COVID-19 screening, especially when routine diagnostic tests are not readily available at scale. During infections, continuously collected heart rate data has shown that an individual's heart rate tends to increase about 8-10 beats per min for every 1°C rise in fever, [bib_ref] The relationship between body temperature, heart rate and respiratory rate in children, Davies [/bib_ref] [bib_ref] Fever and cardiac rhythm, Karjalainen [/bib_ref] and data from animal models suggested that changes in diurnal heart rate can be detected 4 days before fever.Wearables have also shown early promise to improve models to differentiate symptomatic individuals who have COVID-19 versus those who have other infections. App-based collection of self-reported symptom data has identified key symptoms predictive of COVID-19
## Figure 1: wearable sensors and return of health information
Multiple personal sensors can be used to characterise an individual's unique baseline and trends, which can help identify subtle, early changes that might indicate the onset of a viral illness.
## Covid-19 risk score
Sept 15 16 17 18 19 20 Viewpoint compared with other infections, with loss of smell and taste as the main predictor. [bib_ref] Real-time tracking of self-reported symptoms to predict potential COVID-19, Menni [/bib_ref] The addition of wearable data into these models has shown potential to further improve the discriminative ability of these models, with an area under the curve increasing from 0·71 (95% CI 0·63-0·79; considering symptoms only) to 0·80 (0·73-0·86; using also wearable data; table 1). [bib_ref] Wearable sensor data and self-reported symptoms for COVID-19 detection, Quer [/bib_ref] Individual sensor data also has shown promise to identify pre-symptomatic infection, [bib_ref] Digital health: tracking physiomes and activity using wearable biosensors reveals useful health-related..., Li [/bib_ref] including for individuals who tested positive for COVID-19, [bib_ref] Pre-symptomatic detection of COVID-19 from smartwatch data, Mishra [/bib_ref] which if confirmed would be especially valuable. Researchers have also looked at identifying the need for hos pitalisation looking at symp toms alone, [bib_ref] Assessment of physiological signs associated with COVID-19 measured using wearable devices, Natarajan [/bib_ref] how elevations in peripheral temperature correlate with self-reported fever, [bib_ref] Feasibility of continuous fever monitoring using wearable devices, Smarr [/bib_ref] and how symptoms and physiological changes are more severe for COVID-19 positive individuals than for influenza positive individuals. [bib_ref] Characterizing COVID-19 and influenza illnesses in the real world via person-generated health..., Shapiro [/bib_ref] As new metrics are added to sensors, substantially greater research is needed to better understand wearable changes for different infections, asymptomatic infections, non-infectious insults, and tracking long-term consequences, such as with post-acute sequelae of SARS-CoV-2 infection. Identifying early signs of decompensation can be especially useful for early initiation of antivirals, monoclonal antibodies, supportive care, and closer individual monitoring. Additionally, as vaccines are rolled out, sensor-based monitoring might prove useful for identifying immune response to vaccination, and potential side-effects and improve our ability to track infections and quantify vaccine effectiveness.
## Prediction models
The true value of the growing availability of personal, multiparametric, continuous sensor technologies is dependent on the development of meaningful detection or prediction models, which should follow existing standards before being implemented in the clinic [bib_ref] DECIDE-AI: new reporting guidelines to bridge the development-to-implementation gap in clinical artificial..., Vasey [/bib_ref] to ensure their reproducibility. [bib_ref] Challenges to the reproducibility of machine learning models in health care, Beam [/bib_ref] AUC=area under the curve. Some data not available because specific numbers of COVID-19 positive individuals were not reported. Studies needed to have a peer-reviewed, preprint article or post supporting data on their website. *As of Jan 2, 2020. [bib_ref] Prediction models for diagnosis and prognosis of covid-19 infection: systematic review and..., Wynants [/bib_ref] There are often issues in reporting important parameters such as the length of the follow-up or the concurrent prevalence of COVID-19 and other viral infections such as influenza in the considered population, although different models report performance with different statistical measures, making it more difficult to compare them for the same prediction task. Among the studies overviewed in the work of Wynants and colleagues, [bib_ref] Prediction models for diagnosis and prognosis of covid-19 infection: systematic review and..., Wynants [/bib_ref] all reported moderate to excellent prediction performance, but because of the high risk of bias due to both poor reporting and methodological issues, the results could be overly optimistic and not represent reality. This bias is a serious issue which was underscored in analyses after the publication of a mortality prediction model for COVID-19 patients that was based on several blood sample biomarkers and found a very high predictive accuracy of 90%. [bib_ref] An interpretable mortality prediction model for COVID-19 patients, Yan [/bib_ref] Subsequent studies [bib_ref] Limited applicability of a COVID-19 specific mortality prediction rule to the intensive..., Dupuis [/bib_ref] [bib_ref] Replication of a mortality prediction model in Dutch patients with COVID-19, Quanjel [/bib_ref] [bib_ref] Consider laboratory aspects in developing patient prediction models, Reeve [/bib_ref] [bib_ref] Clinical interpretation of an interpretable prognostic model for patients with COVID-19, Giacobbe [/bib_ref] and perspectives highlighted a number of limitations, including the limited clinical use and applicability of this model, the difficulty in reproducing the results in a different dataset, the potential bias due to different sources of the predictor biomarkers, and the inter pretability of the model. Although wearable sensors might offer a convenient means of individual data collection and model development that can potentially enable the detection for COVID-19 and other viral infections, there are best practices that must be followed in the analysis of these data [bib_ref] Best practices for analyzing largescale health data from wearables and smartphone apps, Hicks [/bib_ref] and multiple limitations to this approach. Obtaining sufficiently large data to provide the needed specificity for the identification of a given disease could be challenging and is dependent on individuals obtaining and then selfreporting if they test positive. Furthermore, individuals owning a smartwatch and participating in these studies are probably not representative of the broader population; therefore, the results might not be generalisable to the whole population. In the future, these models will need to address these limitations, taking into account other more technical issues related to the prediction of rare events, the model complexity that could drive to overfitting, and the fact that data from wearables are particularly affected by artifacts. 51
## Health inequities
The rapid proliferation of digital health innovations to address the COVID-19 pandemic could be especially valuable as a means to improve health delivery for vulnerable or typically hard to reach populations. However, many of the social, environmental, economic, and cultural
## Figure 2: multi-layered approach to improve infectious disease surveillance by incorporating novel and traditional data streams
Optimal disease detection should take advantage of all layers of available data to get a complete picture of transmission risk, pre-symptomatic and symptomatic infection, health-care seeking trends, and recovery. Viewpoint determinants of health that already contribute to these populations being underserved in existing systems of care (eg, race and ethnicities bias, economic and educational disadvantages, health illiteracy, inadequate healthcare access and quality, and complexity of individual behaviour) are compounded when digital technologies are expected to be incorporated organically. [bib_ref] COVID-19 and racial and ethnic disparities, Hooper [/bib_ref] In fact, additional disparities in digital technology access, digital engage ment, and digital health literacy, can layer on additional inequality. [bib_ref] COVID-19 and digital inequalities: reciprocal impacts and mitigation strategies, Beaunoyer [/bib_ref] Even worse, the race of the person using the technology can influence its accuracy. [bib_ref] Limiting racial disparities and bias for wearable devices in health science research, Colvonen [/bib_ref] The impact this can have on health outcomes was highlighted in a study [bib_ref] Racial bias in pulse oximetry measurement, Sjoding [/bib_ref] involving nearly 10 000 hospitalised patients in which Black patients had nearly three times the frequency of occult hypoxaemia not detected by pulse oximetry. The historical scarcity of culturally contexted interventions has contributed to poor uptake and scalability of digital health technologies, which risks increasing disparities. Designing solutions with (rather than for) the intended end users, alongside providing the infrastructure needed to assure universal access, and the digital health education frameworks necessary are what is needed for digital technologies to serve as vital public health tools that help minimise disparities rather than contribute to them.
Meaningful community engagement such as with religious leaders, local authorities, and other community leaders at early stages of study inception is fundamental to understand the communities' needs to address the inequity and action gaps experienced at a community level. [bib_ref] Back to the future: achieving health equity through health informatics and digital..., Brewer [/bib_ref] Empowering communities through participation and advocacy of digital health implementation and research establishes community trust, addresses and resolves hesitancies, helps effectively communicate the importance of research findings, ensures successful digital health implementation, and ultimately increases representation of historically marginalised communities. Additionally, providing educational opportunities for community members on the use of the digital tools in general, and for specific health conditions, through easy-access, free, educational services is necessary. These opportunities will require designing remote, culturally tailored educational tools for all digital and health literacy population levels.
The large and growing proportion of the world's population with a cell phone and internet access can provide a good start to the digital infrastructure needed, but recognising and addressing that the internet is still not accessible to approximately 40% of people living in emerging economies and that just under 1 billion people in the world are still without access to electricity are important to creating truly equitable solutions.International and domestic government sectors need to build alliances among stakeholders to discuss standards to build tools and platforms that can be shared equally across populations. This includes the need to increase internet access alongside digital networks and digital tools to enable access to health services while establishing transparent policies that can protect individuals' privacy and security and enable self-control of the use of their data for government, commercial, and research purposes. Participant autonomy alongside privacy and data rights are paramount if digital health technologies are to be used for combating major health issues. Full transparency of data use and the presentation of the risks to benefit Viewpoint ratio to participants is crucial to ensure consumer trust and adoption. [bib_ref] Ethical considerations when creating evidence from real world digital health data, Nebeker [/bib_ref] The COVID pandemic has highlighted unacceptable major gaps in care that exist due to long-standing disparities. However, some responses to addressing COVID- [bib_ref] Racial bias in pulse oximetry measurement, Sjoding [/bib_ref] have provided examples of how digital health tools can help to minimise disparities. For example, WHO partnered with instant messaging platforms WhatsApp (with over 2 billion active users in 180 countries) and Viber (1·2 billion users in 190 countries) to create a dedicated coronavirus informational service, with plans to be available in over 20 languages.Further, existing digital platforms that had been specifically designed for underserved communities were able to be rapidly modified to address COVID-19 care needs. For example, an app created in 2018 for the Syrian refugee population in Turkey was adapted for COVID-19 to assess symptoms, disseminate health information, and support prevention efforts.Before the pandemic, digital health tools have provided the opportunity to extend access to health research and management well beyond the geographical confines of bricks and mortar health systems to rural communities. 60
## Future directions
The value of remote monitoring of individuals to help maintain health is in its earliest stages of exploration. The COVID-19 pandemic has rapidly accelerated the work needed to move that field of research forward. The desire to address an unprecedented health need motivated investigators, funders, and many individuals worldwide to support a large number of studies made possible through the use of personal health technologies. The findings from these studies will inform future work with everimproving sensor technologies, enhanced user experiences, and more sophisticated analytics to provide unique solutions to support individual and public health needs. However, the pandemic has also exposed long enduring gaps and inequities in existing systems of care that can only be addressed through their thoughtful, collaborative, and user-centric implementation.
There remains much to learn. Ongoing studies will provide valuable lessons that will guide future use of personal health technologies to improve the diagnosis and treatment of infectious diseases in a more individualised, safer, and equitable manner. Such efforts could provide the opportunity to establish coherent policies and frameworks that keep up with the advancement of technologies and will require multi-stakeholders from government, science, technology, innovation, and civil society to develop sustainable solutions.
## Contributors
All authors were equally involved in original drafting, writing, reviewing, and editing of the manuscript. JMR and SRS contributed to the conceptualisation of the manuscript.
# Declaration of interests
SRS is a part-time employee of physIQ, Chicago, Illinois, USA. All other authors declare no competing interests.
[table] Table 1: Observational studies of wearables in the prediction of viral illnesses [/table]
[table] Table 2: Bias associated with personal sensor data modeling and future solutions [/table]
|
10.3390/nu13124441
|
CCBY
|
8706614
|
34959993
|
s2orc_pubmed_articles
|
Protective Effect of Lactiplantibacillus plantarum 1201 Combined with Galactooligosaccharide on Carbon Tetrachloride-Induced Acute Liver Injury in Mice
Citation: Ren, Z.; Huo, Y.; Zhang, Q.; Chen, S.; Lv, H.; Peng, L.; Wei, H.; Wan, C. Protective Effect of Lactiplantibacillus plantarum 1201 Combined with Galactooligosaccharide on Carbon Tetrachloride-Induced Acute Liver Injury in Mice. Nutrients 2021, 13, 4441. https://doi.
# Introduction
The liver plays many important roles in the body, such as the detoxification of chemicals, including drugs and environmental contaminants, which, in turn, damage the liver [bib_ref] Liver regeneration: From myth to mechanism, Taub [/bib_ref]. The liver is also able to maintain metabolic homeostasis [bib_ref] Role of grape seed extract on methotrexate induced oxidative stress in rat..., Cetin [/bib_ref]. Acute liver injury (ALI) is a life-threatening clinical syndrome characterized by rapid loss and abnormal liver cell function in patients with or without previous liver disease [bib_ref] Mechanism investigation of dioscin against CCl4-induced acute liver damage in mice, Lu [/bib_ref]. The molecular process of ALI is thought to involve complex interactions between oxidative stress, apoptosis, autophagy, and necrosis [bib_ref] Augmenter of liver regeneration protects against carbon tetrachloride-induced liver injury by promoting..., Shi [/bib_ref] [bib_ref] Oxidant stress, mitochondria, and cell death mechanisms in drug-induced liver injury: Lessons..., Jaeschke [/bib_ref]. In the early stages of ALI, the initial purpose of the liver is to induce the generation of reactive oxygen species (ROS) in hepatocytes and endoplasmic reticulum stress, thereby triggering hepatocyte apoptosis and necrosis. If the inactivated liver cells are not cleared quickly by the phagocytic cells, such as the phagocytic Kupffer cells, the membranes of these liver cells will break down, leading to secondary necrosis [bib_ref] Comparative hepatic transcriptome analyses revealed possible pathogenic mechanisms of fasiglifam (TAK-875)-induced acute..., Urano [/bib_ref] [bib_ref] Anthocyanin-rich Riceberry bran extract attenuates gentamicin-induced hepatotoxicity by reducing oxidative stress, inflammation..., Arjinajarn [/bib_ref] [bib_ref] Protective effect of lipoic acid against methotrexate-induced oxidative stress in liver mitochondria, Tabassum [/bib_ref]. Therefore, a focus on reducing the oxidative damage of the liver is of great significance for the prevention of ALI.
In recent years, with the introduction of the concept of the gut-liver axis, some researchers have boldly proposed that ALI is also related to intestinal flora. When the normal L. plantarum 1201 was cultured under anaerobic conditions at 37 - C in sterile deMan, Rogosa, Sharpe broth (Beijing Solarbio Science & Technology, Co. Ltd., Beijing, China). Subsequently, cells were harvested, centrifuged at 6000 rpm for 5 min at 25 - C, and washed in sterile 1 × PBS. L. plantarum 1201 isolated from Fuzhou Preserved Pickles is deposited in the China Center for Type Culture Collection with the collection number CCTCC M 2021050.
## . preparation culture supernatant and the intracellular extract
We set up four groups to measure the antioxidant capacity of L. plantarum 1201, namely a cell-free supernatants group (CFS), a cell suspension group (CS), a cell free extracts group (CFE), and a positive control group (PC). Lacticaseibacillus rhamnosus GG has been proven by previous studies to have a good antioxidant capacity, and some researchers have used LGG as a control to evaluate the antioxidant capacity of probiotics [bib_ref] Antioxidative and Probiotic Activities of Lactic Acid Bacteria Isolated from Traditional Artisanal..., Shi [/bib_ref]. So, in this study,
LGG was used as a control strain to evaluate the antioxidant capacity of L. plantarum 1201.
A total of 10 mL MRS was used to incubate L. plantarum 1201 at 37 - C for 24 h, then we transferred aliquots of the culture to 2 mL polypropylene tubes, and centrifuged at 12,000 rpm for 15 min at 4 - C. Next, 1 M NaOH was used to neutralize (pH 7.0) the supernatant and the mixture was heated at 95 - C for 5 min. According to the previous research [bib_ref] Probiotic potential, antimicrobial and antioxidant activities of Enterococcus durans strain LAB18s, Pieniz [/bib_ref] , ultrapure water was used to wash the cell pellet and suspend it, then the cell pellet was sonicated (Unique USC 700) at 40 kHz for 15 min, with 5 intervals of 1 min in an ice bath [bib_ref] Probiotic potential, antimicrobial and antioxidant activities of Enterococcus durans strain LAB18s, Pieniz [/bib_ref]. Cellular debris was removed by centrifugation at 12,000 rpm for 15 min at 4 - C. Finally, we obtained the intracellular extracts.
## Dpph radical scavenging assay
A DPPH radical solution (0.2 mmol/L) in 95% ethanol was prepared. A total of 1 mL of DPPH in ethanol was added to 1 mL of the sample (culture supernatant or intracellular extract), well vortexed, and incubated for 30 min in a dark room at room temperature. Subsequently, the mixture was centrifuged at 7200 rpm for 10 min. The absorbance of the supernatant was measured with a spectrophotometer at 517 nm. An equal volume of distilled water, was used for the control group and ethanol was used as a blank. The antioxidant activity was expressed as a percentage of DPPH activity calculated as: DPPH activity (%) = Absorbance of blank − absorbance of sample Absorbance of blank × 100
## Reducing power assay
The 0.5 mL of the 2 mM phosphate buffer (pH6.6) and 1% of Potassium ferricyanide (0.5 mL) were added to 0.5 mL of the sample and kept at 50 - C for 20 min. A total of 10% trichloroacetic acid (0.5 mL) was added to the reaction mixture and spun at 3000 rpm for 10 min. Distilled water (1 mL) and 1% ferric chloride (1 mL) was added to incubate for 10 min, then the absorbance of the mixture was measured at 700 nm. We used an equal volume of the phosphate buffer in the blank group to replace the sample solution [bib_ref] Characterization, antioxidant and immunomodulatory potential on exopolysaccharide produced by wild type and..., Adebayo-Tayo [/bib_ref].
Reducing power (%) = Absorbance of blank − absorbance of sample Absorbance of blank × 100
## Hydroxyl free radical scavenging activity assay
The reaction mixture containing 1.0 mL of PBS (20 mM, pH 7.4), 0.5 mL of 1,10phenanthroline (2.5 mM, Sigma-Aldrich, Shanghai, China), 0.5 mL of FeSO 4 (2.5 mM), 0.5 mL of H 2 O 2 (2.5 mM), and 0.5 mL of the samples was incubated at 37 - C for 60 min. The absorbance of the mixture was measured at 536 nm after centrifugation (3000 rpm, 10 min).
LGG was used to evaluate the ability of L. plantarum 1201.
Hydroxyl radical scavenging activity (%) = Absorbance of sample − absorbance of control Absorbance of blank − absorbance of control × 100
## Growth experiments
The medium was treated with 20, 30, and 40 g/L GOS, and the MRS medium was set as the control group. A total of 100 µL of L. plantarum 1201 was inoculated with 10 mL of the medium. The optical density at 630 nm was measured every 3 h for 24 h in an anaerobic cabinet to monitor the growth of L. plantarum 1201 (36 - C ± 1 - C; 80% N 2 , 10% CO 2 , 10% H 2 ). Three independent experiments, each performed in triplicate, were run.
## Animals
Six-week-old male specific pathogen-free C57BL/6N mice (Beijing Vital River Laboratory Animal Technology, Co. Ltd., Beijing, China) were housed in Nanchang Royo Biotech, Co. Ltd., Nanchang, China. under standard conditions with a light/dark cycle of 12 h. All mice were provided with ad libitum access to food and water. All experimental procedures were in accordance with the guidelines of the Institutional Animal Care and Use Committee of Nanchang Royo Biotech, Co. Ltd.
## Experimental groups
Mice were randomly assigned to 5 groups (ND, MD, LD, PD, LP) with 10 mice in each group. (1) the ND group was given PBS for 2 weeks, and 1 h after the last gavage, 2 µL/g peanut oil solution was injected intraperitoneally; (2) mice in the MD group were given PBS for 2 weeks, and 1 h after the last gavage, 2 µL/g 50% CCl 4 peanut oil solution was injected intraperitoneally; (3) mice in the LD group were administered intragastrically with 3 × 10 8 cfu/mL L. plantarum 1201 for 2 weeks, and 1 h after the last gavage, 2 µL/g 50% CCl 4 peanut oil solution was injected intraperitoneally; (4) mice in the PD group were given PBS for 1 week first, then were administered 100 µL GOS intragastrically at a concentration of 0.5 g/kg for 1 week. Moreover, 1 h after the last gavage, 2 µL/g 50% CCl 4 peanut oil solution was injected intraperitoneally; (5) mice in the LP group were given 3 × 10 8 cfu/mL L. plantarum 1201 for 2 weeks, and 100 µL GOS at a concentration of 0.5 g/kg for 1 week. Moreover, 1 h after the last gavage, 2 µL/g 50% CCl 4 peanut oil solution was injected intraperitoneally.
A total of 16 h after intraperitoneal injection, the mice were euthanized with ether. The mouse serum, liver, colon, colon contents, ileum, and cecum contents were separately collected in a sterile centrifuge tube and stored at −80 - C for subsequent experiments. The plasma was placed in a refrigerator at 4 - C for 1 night and centrifuged at 3000 rpm for 10 min, then the upper layer of serum was taken for storage.
## Organ index determination
The mice were weighed with an electronic scale (Kaifeng Group, Co. Ltd., Yongkang, China). Then the neck was severed, and the liver tissue was dissected and separated. Connective tissues such as fat and fascia were removed. Then they were rinsed with normal saline, and after absorbent paper was used to absorb the surface water, the organs were weighed and recorded. The liver index was calculated as the liver weight divided by the body weight.
## Aminotransferase measurement
Serum alanine aminotransferase (ALT) and apartate transaminase (AST) were measured with commercial Assay Kits (Jiancheng Bioengineering, Nanjing, China).
## Examination of oxidation markers
Liver samples (100 mg) were homogenized in 900 µL of saline at 4 - C and then centrifuged at 4000 rpm at 4 - C for 10 min. According to the manufacturer's instructions, superoxide dismutase (SOD), glutathione (GSH), and malondialdehyde (MDA) in the liver tissue were measured using commercial test kits (Jiancheng Bioengineering, Nanjing, China).
## Hematoxylin and eosin staining
Fresh colon and liver tissues were collected and soaked in 10% formalin tissue fixative solution. After the alcohol was dehydrated, the tissue blocks were sequentially transparent, waxed, encased, sliced, and baked in turn. The conventional dewaxing process was to sequentially put paraffin sections into xylene I (10 min)-xylene II (10 min)-ethanol I (5 min)-ethanol II (5 min)-95% alcohol (3 min)-90% alcohol (3 min)-80% alcohol (2 min)-70% alcohol (2 min). Next, the slices were soaked in distilled water and washed for 2 min and stained with hematoxylin and eosin (H&E), then the slices were put into anhydrous ethanol I for 5 min, anhydrous ethanol II for 5 min, xylene I for 5 min, and xylene II for 5 min, and dried to a neutral gum sealing piece before microscopy.
## Dna extraction from the mouse cecal contents and 16s rrna gene sequencing
The genomic DNA of mouse cecal contents was extracted, and the purity and concentration of DNA were detected. High-throughput sequencing was conducted by Biomarker Tech (Beijing, China). Clean Data were obtained by stitching, filtering, and removing the chimera of the original Data. Then the clustering and species classification of the Operational Taxonomic Units (OTUs) were analyzed based on the available data. Based on the results of the OTU cluster analysis, a multiple diversity index analysis and sequencing depth detection were conducted for OTUs. Based on the taxonomic information, the community structure was statistically analyzed at each taxonomic level.
The representative sequences of OTUs were annotated with the Greengenes database. A Venn diagram was drawn based on the OTU distribution. Based on the analysis of the species composition, the community structure component map, the community Heatmap, and the UPGMA clustering tree based on Unweighted Unifrac and weighted Unifrac distance were obtained. The difference analysis, PCA analysis, PCOA analysis, and NMDS analysis from the Beta diversity were performed using a t-test, a Wilcox Rank Sum test, and a Tukey test. The LEfSe analysis was based on taxonomic composition, and the linear discriminant analysis (LDA) was performed on the samples according to the different grouping conditions to find out the communities or species with significant differences in sample classification.
## Gene expression
High quality RNA was isolated from the frozen colon and liver tissue using the TaKaRa RNA extraction kit (Takara, Otsu, Japan), and then used for cDNA synthesis with a transcriptor cDNA kit (Takara, Otsu, Japan).
The mRNA levels of the pro-inflammatory factors (IFN-γ, IL-1β, TNF-α, IL-6), antiinflammatory factors (IL-10, IL-22, TGF-β), fibronectin 1 (FN1), Chemokines (CCL4, CCL5), Phospho-extracellular regulated protein kinases (p-ERK), c-Jun N-terminal kinase (JNK), and B-cell lymphoma-2 (Bcl-2) in the liver tissue were measured using a real-time PCR. The mRNA levels of the pro-inflammatory factors (IFN-γ, IL-1β, TNF-α, IL-6), antiinflammatory factors (IL-10, IL-22, TGF-β), Chemokines (CCL4, CCL5), intestinal permeability protein-related gene, (clan3, ocln, ZO-1), p-ERK, JNK, and Bcl-2 in the colon tissue were measured using a real-time PCR. The PCR reaction procedure had three steps: 5 µL SYBR Green Mix, 0.8 µL primer (10 µM), 1 µL cDNA (1000 ng/µL) plus a ddH 2 O supplement in the 10 µL system. The reaction conditions involved preheating at 95 - C for 30 s; a cycling stage: denaturation at 95 - C for 5 s; 59 - C return for 1 min; 72 - C extension for 30 s, 40 cycles; and a melting stage: 65 - C 5 s, 95 - C, 5 s. The primer sequence information is shown in the Supplementary Materials .
## Enzyme-linked immunosorbent assay
According to the kit instructions (Shanghai C-reagent Biotechnology, Co. Ltd., Shanghai, China), the double antibody sandwich method was used to determine the protein expression levels of mouse Caspase-3, BAX, Bcl-2, Beclin1, Atg5, and microtubule-associated protein 1 light chain 3 (LC3-II) in serum.
# Statistical analysis
All values are expressed as mean ± standard error of the mean (SEM). The Graph Pad Prism 6 (GraphPad, Inc. La Jolla, CA, USA) was employed for statistical analysis and graph preparation. A one-way ANOVA was used for multiple comparisons. Statistical significance was considered at p < 0.05. [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref] , the growth curves of L. plantarum 1201 in 20, 30, and 40 g/L of the GOS-added media were all higher than that of the control, with 20 g/L of the GOS medium having the best effect, which is consistent with previous reports [bib_ref] Streptococcus mutansInfluence of 2 -fucosyllactose and galactooligosaccharides on the growth and adhesion..., Salli [/bib_ref] [bib_ref] Prebiotics: The Concept Revisited, Roberfroid [/bib_ref]. expression levels of mouse Caspase-3, BAX, Bcl-2, Beclin1, Atg5, and microtubule-associated protein 1 light chain 3 (LC3-II) in serum.
# Statistical analysis
All values are expressed as mean ± standard error of the mean (SEM). The Graph Pad Prism 6 (GraphPad, Inc. La Jolla, CA, USA) was employed for statistical analysis and graph preparation. A one-way ANOVA was used for multiple comparisons. Statistical significance was considered at p < 0.05.
# Results
## Gos promoted the growth of l. plantarum 1201 with high-quality antioxidant power
L. plantarum 1201 growth accelerated after 6 h. After 21 h, the cell number reached its highest level. At this time, the OD 630 nm value of all the bacterial suspensions reached levels above 1.0, and the proliferation and death of the strain gradually balanced. The growth of L. plantarum 1201 was affected by the addition of GOS. As shown in [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref] , the growth curves of L. plantarum 1201 in 20, 30, and 40 g/L of the GOS-added media were all higher than that of the control, with 20 g/L of the GOS medium having the best effect, which is consistent with previous reports [bib_ref] Streptococcus mutansInfluence of 2 -fucosyllactose and galactooligosaccharides on the growth and adhesion..., Salli [/bib_ref] [bib_ref] Prebiotics: The Concept Revisited, Roberfroid [/bib_ref]. CFS, cell-free supernatants group; CS, cell supernatants group; CFE, cell-free extracts group; PC, Positive control group. * p < 0.05; ** p < 0.01; *** p < 0.001; * is compared with C in [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref] ; paired twotailed t-test.
The different components of L. plantarum 1201 showed significant differences in relation to the scavenging rate of DPPH free radicals [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref]. LGG has been proven to have a strong antioxidant capacity [bib_ref] Thirty Years of Lactobacillus rhamnosus GG: A Review, Capurso [/bib_ref]. Compared with LGG, L. plantarum 1201 had a similar effect on DPPH free radical scavenging. Their scavenging rates were 83.0% and 82.1% in the CFS group, while the scavenging abilities of the CS and CFE groups were CFS, cell-free supernatants group; CS, cell supernatants group; CFE, cell-free extracts group; PC, Positive control group. * p < 0.05; ** p < 0.01; *** p < 0.001; * is compared with C in [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref] ; paired two-tailed t-test.
The different components of L. plantarum 1201 showed significant differences in relation to the scavenging rate of DPPH free radicals [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref]. LGG has been proven to have a strong antioxidant capacity [bib_ref] Thirty Years of Lactobacillus rhamnosus GG: A Review, Capurso [/bib_ref]. Compared with LGG, L. plantarum 1201 had a similar effect on DPPH free radical scavenging. Their scavenging rates were 83.0% and 82.1% in the CFS group, while the scavenging abilities of the CS and CFE groups were both below 20%. This shows that L. plantarum 1201 has the active substance related to the DPPH free radical scavenging ability mainly in extracellular metabolites. The hydrogen peroxide activity of L. plantarum 1201 is shown in [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref]. In the CS group, the scavenging rates of L. plantarum 1201 and LGG on hydroxyl free radicals were 82.3% and 91.5%, respectively, indicating that the living bacteria had strong antioxidant activity. The scavenging rate of the cell extract (CFE) of L. plantarum 1201 on hydroxyl free radicals was only 7.7%, nearly half of LGG, indicating that its intracellular products had a weak scavenging ability in relation to hydroxyl free radicals.
As shown in [fig_ref] Figure 1: GOS promoted the growth of L [/fig_ref] , the absorbance of the L. plantarum 1201 CFS group was 0.69, while the absorbances of the CS group and the CFE group were 0.27 and 0.23, respectively. The results of LGG were consistent with 1201, the absorbance of the CFS group was 0.81, and the absorbance of the CS group and the CFE group were, respectively, 0.25 and 0.23. This shows that the active substance related to the reducing ability of L. plantarum 1201 and LGG may be the secretion of extracellular metabolism products.
## Synbiotics alleviated the dysfunction and pathological damage of the liver in ccl 4 -induced ali mice
As shown in [fig_ref] Figure 2: Synbiotics alleviated the dysfunction and pathological damage of the liver in CCl4-induced... [/fig_ref] , CCl 4 treatment caused an increase in the organ index (p < 0.001). However, compared with MD, the levels of the organ index were significantly reduced by L. plantarum 1201, GOS, and synbiotic pretreatment (p < 0.01).
To detect whether L. plantarum 1201 and GOS could reduce the liver dysfunction induced by CCl 4 , serum enzymes were analyzed. As shown in [fig_ref] Figure 2: Synbiotics alleviated the dysfunction and pathological damage of the liver in CCl4-induced... [/fig_ref] ,D, CCl 4 treatment increased the activities of ALT and AST (p < 0.001) in the serum. However, L. plantarum 1201, GOS, and synbiotic pretreatment reduced the activities of ALT and AST. Among them, GOS and the synbiotics had significant effects (p < 0.01).
We also used H&E staining of the liver tissues to analyze the histological changes in these different groups. Through the observation method of the H&E stained sections in the liver, we found severe hepatocellular necrosis and inflammatory cell infiltration, especially around the central veins in the model group [bib_ref] The protective effect of Zizyphus jujube fruit on carbon tetrachloride-induced hepatic injury..., Shen [/bib_ref]. However, the liver lesions of the LD, PD, and LP groups were drastically attenuated, with synbiotic pretreatment showing the most significant relief [fig_ref] Figure 2: Synbiotics alleviated the dysfunction and pathological damage of the liver in CCl4-induced... [/fig_ref].
## Synbiotics reduced oxidative stress and relieved liver inflammation in ccl 4 -induced ali mice
Oxidative stress is a key part of the pathogenesis of ALI. We examined the SOD activities and the contents of MDA and GSH in liver homogenates, the key markers of oxidation. We found that the CCl 4 challenge decreased the SOD activity by 39% and reduced the content of GSH by 25.98%, while it increased the content of MDA in the liver by 44.89%. However, L. plantarum 1201, GOS, and synbiotic treatment had the opposite effect, as they significantly increased the SOD activity and the content of GSH, while reducing the content of MDA (p < 0.01, [fig_ref] Figure 3: Synbiotics reduced oxidative stress and relieved liver inflammation in CCl4-induced ALI mice [/fig_ref].
To investigate whether L. plantarum 1201, GOS, and synbiotics could attenuate liver inflammation, we detected the expression levels of some pro-inflammatory cytokines (IL-6, IL-1β, TNF-α, IFN-γ), anti-inflammatory factors (IL-10, IL-22, TGF-β), and chemokines (CCL4, CCL5) in liver tissues. The CCl 4 challenge decreased the expression level of IL-10 in the liver by 53.27%, while it drastically increased the mRNA level of IL-22, IL-1β, TNF-α, IL-6, IFN-γ, TGF-β, CCL4, and CCL5 by 200% or more. However, with L. plantarum 1201, GOS, and synbiotic pretreatment, the LD, PD, and LP groups yielded similar results to the control group [fig_ref] Figure 3: Synbiotics reduced oxidative stress and relieved liver inflammation in CCl4-induced ALI mice [/fig_ref]. Furthermore, the mRNA level of Bcl-2 was also noticeably decreased in the MD group compared with the ND and diet intervention groups, whereas the mRNA expression level of FN-1 was significantly elevated with diet intervention [fig_ref] Figure 3: Synbiotics reduced oxidative stress and relieved liver inflammation in CCl4-induced ALI mice [/fig_ref].
## Synbiotics reduced oxidative stress and relieved liver inflammation in ccl4-induced ali mice
Oxidative stress is a key part of the pathogenesis of ALI. We examined the SOD activities and the contents of MDA and GSH in liver homogenates, the key markers of oxidation. We found that the CCl4 challenge decreased the SOD activity by 39% and reduced α, IL-6, IFN-γ, TGF-β, CCL4, and CCL5 by 200% or more. However, with L. plantarum 1201, GOS, and synbiotic pretreatment, the LD, PD, and LP groups yielded similar results to the control group [fig_ref] Figure 3: Synbiotics reduced oxidative stress and relieved liver inflammation in CCl4-induced ALI mice [/fig_ref]. Furthermore, the mRNA level of Bcl-2 was also noticeably decreased in the MD group compared with the ND and diet intervention groups, whereas the mRNA expression level of FN-1 was significantly elevated with diet intervention [fig_ref] Figure 3: Synbiotics reduced oxidative stress and relieved liver inflammation in CCl4-induced ALI mice [/fig_ref]. (F,G) the mRNA level of Bcl-2 (F) and FN-1 (G) (n = 3). * p < 0.05; ** p < 0.01; *** p < 0.001; * is compared with MD; paired two-tailed t-test.
## Synbiotics alleviated intestinal flora disturbance in ccl4-induced ali mice
Due to the presence of the liver-gut axis, we speculated that the gut microbiota would change with liver damage. We used both α and β diversity analyses to assess the differences between the groups in the diversity of intestinal flora. A principal coordinates (PCoA) analysis was used to analyze the β-diversity among groups. The distribution of the intestinal flora in the MD group was quite different from that in the ND group. However, with L. plantarum 1201, GOS, and synbiotic treatment, the microbial composition was similar to the ND group . Moreover, the species rarefaction curves tended to
## Synbiotics alleviated intestinal flora disturbance in ccl 4 -induced ali mice
Due to the presence of the liver-gut axis, we speculated that the gut microbiota would change with liver damage. We used both α and β diversity analyses to assess the differences between the groups in the diversity of intestinal flora. A principal coordinates (PCoA) analysis was used to analyze the β-diversity among groups. The distribution of the intestinal flora in the MD group was quite different from that in the ND group. However, with L. plantarum 1201, GOS, and synbiotic treatment, the microbial composition was similar to the ND group . Moreover, the species rarefaction curves tended to be relatively flat, illustrating that the sequencing quantity and depth were qualified, and we could carry out subsequent analysis . The diversity (Shannon) and community abundance index (Chao) were used to reflect the α diversity of the intestinal microbial population [bib_ref] Arsenic exposure induces intestinal barrier damage and consequent activation of gut-liver axis..., Zhong [/bib_ref]. Compared with those of the ND group, the Shannon and Chao indexes were significantly decreased in the MD group, and the dietary intervention groups yielded similar indexes to the control group , which indicated that CCl 4 significantly altered the richness and diversity of the intestinal microbiota in mice, and L. plantarum 1201, GOS, and synbiotic showed preventive effects.
## (c) rarefaction curve (n = 5). (d) shannon index (n = 5). (e) chao index (n = 5). (f) analysis of variance between groups at the phylum level. (g) analysis of variance between groups at the family level. (h,i)
Relative abundance of the gut bacterial composition at the level of the phylum and species (mean relative abundance > 0.1%). (J) Family-level network diagram of each species, "the circle represents the species, the size of the circle represents the average abundance of the species; the line represents the correlation between the two species, the thickness of the line represents the strength of the correlation, and the color of the line: orange represents a positive correlation, and green represents a negative correlation". * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001; * is compared with MD; paired two-tailed t-test.
We further analyzed the changes in the microbiota at the phylum and family levels. At the phylum level , a Metastats analysis showed that the CCl 4 challenge significantly decreased the abundance of Firmicutes while increasing the abundance of Bacteroidetes, Proteobacteria, and Actinobacteria. Additionally, L. plantarum 1201, GOS, and synbiotic treatment prevented these changes . At the family level, the relative abundances of Aeromonadaceae, Akkermansiaceae, Bacteroidaceae, and Lachnospiraceae were more abundant in the CCl 4 -treated group than in the control group, while the relative abundances of Lactobacillaceae and Christensenellaceae were more abundant in the control and dietary intervention groups .
Through the family-level flora correlation network, we found that Lactobacillaceae and Christensenellaceae were positively correlated . This is consistent with the changes in the flora abundance at the family level and is also consistent with the previous literature.
## Synbiotics relieved intestinal inflammation in ccl4-induced ali mice
Flora disorders are usually accompanied by inflammation. Therefore, we tested the expression of inflammatory factors in the intestinal tissue. Similar to the inflammation of the liver, the CCl4 challenge decreased the expression level of IL-10 in the colon, while it drastically increased the mRNA expression level of IFN-γ, IL-1β, TNF-α, IL-6, IL-22, TGF -β, CCL4, and CCL5. The mRNA expression level in the diet intervention groups was close to normal level [fig_ref] Figure 5: Synbiotics relieved intestinal inflammation in CCl4-induced ALI mice [/fig_ref]. Additionally, a significant difference in the mRNA expression level of Bcl-2 occurred in CCl 4 -treated group, while it was similar between the control and dietary intervention groups [fig_ref] Figure 5: Synbiotics relieved intestinal inflammation in CCl4-induced ALI mice [/fig_ref].
Intestinal tight junctions (TJs) have been shown to be associated with intestinal barrier integrity [bib_ref] Layered defense: How mucus and tight junctions seal the intestinal barrier, Capaldo [/bib_ref]. To test whether there was intestinal barrier destruction in the CCl 4 -induced ALI mice, and whether synbiotics have a recovery effect on intestinal permeability, we detected the mRNA expression of TJ proteins in the colon. As apparent from [fig_ref] Figure 5: Synbiotics relieved intestinal inflammation in CCl4-induced ALI mice [/fig_ref] , expressions of claudin3, occludin, and zonula occludin-1 were significantly reduced in ALI mice, while they increased in diet intervention groups. The results demonstrate that synbiotic treatment promoted the expression of the TJ protein.
Generally speaking, changes in intestinal permeability are accompanied by the destruction of intestinal tissues. To detect the pathological damage of the colon, we used H&E to stain the colon tissues. As shown in [fig_ref] Figure 5: Synbiotics relieved intestinal inflammation in CCl4-induced ALI mice [/fig_ref] , the mucosa in the ND group was intact without inflammatory cell infiltration; the colonic mucosa in the MD group was significantly shed, crypts were destroyed, and a large number of inflammatory cells were infiltrated; the LD group and the PD group had lighter mucosal defects and fewer inflammatory cells; and the LP group was almost the same as the ND group. L. plantarum 1201, GOS, and synbiotic treatment can alleviate pathological damage of the colon in ALI mice, and the effect of the synbiotic was the most significant.
## Synbiotics inhibited the apoptotic signaling pathway and the autophagic signaling pathway through inhibiting the mapk/nf-κb signaling pathway in ccl4-induced ali mice
In both the liver and colon, the mRNA level of p-ERK was elevated by the CCl4 challenge while that of JNK was reduced, but with the L. plantarum 1201, GOS, or synbiotic pretreatment, the mRNA levels of p-ERK and JNK were both similar to the control group [fig_ref] Figure 6: Synbiotics inhibited the MAPK/NF-κB signaling pathway, then inhibited the apoptotic signaling pathway... [/fig_ref]. Moreover, the concentrations of IKK-β and NF-κB were increased in MD, while the concentration of of IκB-α was reduced. However, in the LD, PD, and LP groups, the concentrations of these proteins returned to normal [fig_ref] Figure 6: Synbiotics inhibited the MAPK/NF-κB signaling pathway, then inhibited the apoptotic signaling pathway... [/fig_ref]. We found that the (E) H&E pathological section of the mouse colon tissue (n = 4). * p < 0.05; ** p < 0.01; *** p < 0.001; * is compared with MD; paired two-tailed t-test.
## Synbiotics inhibited the apoptotic signaling pathway and the autophagic signaling pathway through inhibiting the mapk/nf-κb signaling pathway in ccl 4 -induced ali mice
In both the liver and colon, the mRNA level of p-ERK was elevated by the CCl 4 challenge while that of JNK was reduced, but with the L. plantarum 1201, GOS, or synbiotic pretreatment, the mRNA levels of p-ERK and JNK were both similar to the control group [fig_ref] Figure 6: Synbiotics inhibited the MAPK/NF-κB signaling pathway, then inhibited the apoptotic signaling pathway... [/fig_ref]. Moreover, the concentrations of IKK-β and NF-κB were increased in MD, while the concentration of of IκB-α was reduced. However, in the LD, PD, and LP groups, the concentrations of these proteins returned to normal [fig_ref] Figure 6: Synbiotics inhibited the MAPK/NF-κB signaling pathway, then inhibited the apoptotic signaling pathway... [/fig_ref]. We found that the CCl 4 challenge activated the MAPK/NF-κB signaling pathway, while L. plantarum 1201, GOS, or synbiotic pretreatment inhibited it. In addition, the synbiotic had the most significant effect. Compared with the ND group, the MD group exhibited higher protein expression levels of Caspase3, BAX, and Bcl-2. The CCl4 challenge activated the apoptotic signaling pathway, but L. plantarum 1201, GOS, or synbiotic pretreatment inhibited this activation Compared with the ND group, the MD group exhibited higher protein expression levels of Caspase3, BAX, and Bcl-2. The CCl 4 challenge activated the apoptotic signaling pathway, but L. plantarum 1201, GOS, or synbiotic pretreatment inhibited this activation [fig_ref] Figure 6: Synbiotics inhibited the MAPK/NF-κB signaling pathway, then inhibited the apoptotic signaling pathway... [/fig_ref]. The change in the Bcl-2 mRNA expression level is consistent with the change in the protein expression level [fig_ref] Figure 3: Synbiotics reduced oxidative stress and relieved liver inflammation in CCl4-induced ALI mice [/fig_ref].
The expression levels of Beclin1, Atg5, and LC3-II in the MD group were higher than those in the ND group. On the contrary, the expression levels of autophagic proteins in the LD, PD, and LP groups were lower than those in the MD group, and similar to those in the ND group [fig_ref] Figure 6: Synbiotics inhibited the MAPK/NF-κB signaling pathway, then inhibited the apoptotic signaling pathway... [/fig_ref]. These results suggest that L. plantarum 1201, GOS, or synbiotics may normalize the autophagy induced by CCl 4 . This beneficial effect was most significant in the LP group.
# Discussion
The liver is an important organ that ensures the health of the body [bib_ref] Antioxidant and Antiapoptotic Polyphenols from Green Tea Extract Ameliorate CCl4-Induced Acute Liver..., Diao [/bib_ref]. It is well known that ALI induced by CCl 4 is mainly related to oxidative stress and secondary inflammation caused by oxidation [bib_ref] Hepatoprotective effects of baicalein against CCl 4 -induced acute liver injury in..., Huang [/bib_ref] , while L. plantarum has been proven to have a good antioxidant capacity [bib_ref] In vitro probiotic characteristics of Lactobacillus plantarum ZDY 2013 and its modulatory..., Huang [/bib_ref]. Since oxidative stress is vital in the process of acute intoxication with CCl 4 [bib_ref] Hepatoprotective effects of baicalein against CCl 4 -induced acute liver injury in..., Huang [/bib_ref] , we studied the mechanism by which synbiotics protect hepatocytes from these two aspects. SOD is the primary substance for scavenging free radicals in organisms [bib_ref] Effect of quercetin-5 -sulfonic acid sodium salt on SOD activity and ADMA/DDAH..., Trocha [/bib_ref] [bib_ref] Comparison of the effects of three different Baccaurea angulata whole fruit juice..., Ibrahim [/bib_ref] , and MDA is the degradation product of plasma lipid peroxide. In the case of liver cell damage, the content of GSH in the cell decreases, and various oxidative free radicals increase [bib_ref] L-cysteine supplementation upregulates glutathione (GSH) and vitamin D binding protein (VDBP) in..., Jain [/bib_ref]. When the concentration of GSH in the body is lower than the critical value, all kinds of GSH-dependent enzymes will be inactivated, and the protection of oxidative free radicals is weakened [bib_ref] Manganese doped silica nano GSH-cleaner for treatment of Liver Cancer by destroying..., Tang [/bib_ref] [bib_ref] Decoding Aging: Understanding the Complex Relationship among Aging, Free Radicals, and GSH, López-Navarro [/bib_ref]. Base changes in the activity of thiol enzymes in liver cells lead to the degeneration and necrosis of liver cells [bib_ref] Gut microbiota mediates diurnal variation of acetaminophen induced acute liver injury in..., Gong [/bib_ref]. Therefore, the decrease in GSH content is a potential early activation signal of apoptosis [bib_ref] Role of glutathione depletion and reactive oxygen species generation in apoptotic signaling..., Knox [/bib_ref]. SOD, MDA, and GSH can all reflect the body's oxidation/anti-oxidation status. In our research, L. plantarum 1201, GOS, and synbiotics drastically attenuated the AST and ALT activities in the serum, ameliorated the pathological damage of the central vein, and showed an obvious antioxidant potential.
According to previous studies, if injured, the liver can change the secretion of bile through the gut-liver axis and reduce the intestinal blood supply and peristalsis, among other factors, leading to the destruction of the intestinal mucosa and an imbalance in the intestinal flora [bib_ref] Gut-Liver Axis, Microbiome and Environmental Factors. A Never-Ending Bidirectional Cross-Talk, Di [/bib_ref]. To confirm whether the intestinal environment was dysregulated in ALI mice, we measured the intestinal flora and intestinal permeability. Our study showed a significant microbiome disruption, indicating that ALI can alter the abundance and diversity of the gut bacterial community, which may directly affect intestinal function. The relative abundances of Lactobacillaceae and Christensenellaceae were observably reduced, which was also seen in the liver-injury model [bib_ref] Semen hoveniae extract ameliorates alcohol-induced chronic liver damage in rats via modulation..., Qiu [/bib_ref] [bib_ref] Ursolic Acid Improves Intestinal Damage and Bacterial Dysbiosis in Liver Fibrosis Mice, Wan [/bib_ref]. In previous studies, the mixture of Lactobacillus regulated the gut microbiota, increased the number of short chain fatty acids (SCFAs), inhibited liver lipid accumulation and oxidative stress, improved the intestinal epithelial permeability, and decreased the LPS entering the portal vein, thereby inhibiting liver inflammation [bib_ref] Lactobacillus plantarum KLDS1.0344 and KLDS1.0901 Mixture Prevents Chronic Alcoholic Liver Injury in..., Li [/bib_ref]. Similarly, in our research, we found that L. plantarum 1201, GOS, and synbiotic pretreatment regulated the gut microbiota, which had deteriorated as shown by the Firmicutes/Bacteroidetes ratio [bib_ref] Arsenic exposure induces intestinal barrier damage and consequent activation of gut-liver axis..., Zhong [/bib_ref] and increased the abundance of Lactobacillaceae. Then, ALI mice also showed an impaired intestinal barrier, and the expression of TJs was significantly down-regulated. This was relieved after synbiotic pretreatment.
With disorders of the intestinal flora, microbiota products provide a certain concentration of pathogen-associated molecular patterns (PAMPs), such as peptidoglycan and bacterial lipopolysaccharide, which bind to Toll-like receptors and nod-like receptors on the surfaces of intestinal dendritic cells, activating NF-κB in the cytoplasm, leading to the production and secretion of inflammatory cytokines and chemokines. On the one hand, this can destroy the tight junction of the intestinal epithelium and down-regulate the intestinal barrier function. On the other hand, the broken barrier can promote the entry of PAMPs into the hepatic portal circulatory system to exert biological effects and promote liver inflammation [bib_ref] Comparative hepatic transcriptome analyses revealed possible pathogenic mechanisms of fasiglifam (TAK-875)-induced acute..., Urano [/bib_ref] [bib_ref] Role of toll-like receptors in immune activation and tolerance in the liver, Nakamoto [/bib_ref]. CCl 4 metabolites can also stimulate Kupffer cells in the liver, release proinflammatory cytokines, and further aggravate liver injury [bib_ref] Pathological bacterial translocation in liver cirrhosis, Wiest [/bib_ref] [bib_ref] FXR modulates the gut-vascular barrier by regulating the entry sites for bacterial..., Sorribas [/bib_ref]. Therefore, we hypothesize that the change in inflammatory cytokines may be related to the MAPK/NF-κB signaling pathway. Previous studies have demonstrated that novel selenium-glutathione-enriched probiotics (SGP) are effective in attenuating liver fibrosis by attenuating hepatic oxidative stress, inflammation, and MAPK signaling [bib_ref] Protective effects of selenium-glutathioneenriched probiotics on CCl-induced liver fibrosis, Liu [/bib_ref]. Likewise, L. casei Zhang reduced hepatic inflammation through modulating the TLR-MAPK-PPAR-γ signaling pathways and the intestinal microbiota in ALI mice [bib_ref] Probiotic Lactobacillus casei Zhang reduces pro-inflammatory cytokine production and hepatic inflammation in..., Wang [/bib_ref]. Furthermore, Fructosetooligosaccharides (FOS) could reduce the release of nonalcoholic fatty liver disease (NAFLD) by inhibiting the p38 MAPK signaling pathway and reducing the progression of cirrhosis [bib_ref] Lactobacillus paracaseiEffect of N1115 and fructooligosaccharides in nonalcoholic fatty liver disease, Yao [/bib_ref]. Our previous studies have shown that L. plantarum 1201 suppressed NF-κB activation in mice (unpublished). Remarkably, we found that the expression of p-ERK, JNK, IKK-β, IκB-α, and NF-κB was reduced with synbiotic pretreatment in CCl 4 -treated mice. Then, we detected inflammatory cytokines and chemokines in the liver and colon. After activation of the MAPK/NF-κB signaling pathway, the expression of most inflammatory cytokines and chemokines was up-regulated, while the expression of IL-10 was downregulated, and the expression in LD, PD, and LP groups was similar to that in the ND group. Collectively, pretreatment with a synbiotic may regulate the balance of intestinal flora and inhibit MAPK/NF-κB signaling, thus inhibiting the secretion of inflammatory cytokines and alleviating liver damage.
Another major finding was that synbiotics mitigated apoptosis and, consequently, attenuated CCl 4 -induced ALI. CCl 4 can aggravate apoptosis, while Polydeoxyribonucleotide inhibited hepatocyte apoptosis through the regulation of the NF-κB/MAPK signaling pathway in mice [bib_ref] Polydeoxyribonucleotide Exerts Protective Effect Against CCl-Induced Acute Liver Injury Through Inactivation of..., Ko [/bib_ref]. Indeed, administration of CCl 4 contributed to an increase in the protein expression of Caspase-3, the proapoptotic factor BAX, and the antiapoptotic protein Bcl-2. BAX directly opens the mitochondrial permeability transition pore, thereby facilitating the release of cytochrome c into the cytosol and activating the apoptotic signaling pathways [bib_ref] A tale of two mitochondrial channels, MAC and PTP, in apoptosis, Kinnally [/bib_ref]. L. plantarum C88 decreased the levels of BAX and Caspase-3 and elevated the level of Bcl-2 in the liver to ameliorate AFB-induced excessive apoptosis by regulating the mitochondrial pathway and cell death receptor pathways [bib_ref] Lactobacillus plantarum C88 protects against aflatoxin B-induced liver injury in mice via..., Huang [/bib_ref]. Thus, the synbiotic may alleviate autophagy by inhibiting the MAPK/NF-κB signaling pathway.
It is known that there are some similar pathways of cell necrosis, apoptosis, and autophagy and there is significant crosstalk between these different forms of cell death [bib_ref] Chloroquine ameliorates carbon tetrachloride-induced acute liver injury in mice via the concomitant..., Dai [/bib_ref] [bib_ref] Trolline Ameliorates Liver Fibrosis by Inhibiting the NF-kappaB Pathway, Promoting HSC Apoptosis..., Bai [/bib_ref]. Autophagy is critical in the regulation of cell survival/death, and more and more reports show that autophagy can worsen liver injury [bib_ref] Autophagy mediates epithelial cytoprotection in eosinophilic oesophagitis, Whelan [/bib_ref]. We speculate that L. plantarum 1201, GOS, and synbiotics can alleviate CCl 4 -induced liver injury through autophagy. To test this hypothesis, we tested the expression of Beclin1, Atg5, and LC3-II in mice. LC3 has been widely used as a marker of autophagy [bib_ref] Methods in mammalian autophagy research, Mizushima [/bib_ref]. The content of LC3II was increased after acute CCl 4 treatment for 16 h, which undoubtedly indicated that autophagy was involved in addition to the known mechanism of CCl 4 hepatotoxicity, because the increased expression of Beclin1 and Atg5 is evidence of the autophagy pathway. Therefore, we found that the synbiotic could inhibit autophagy by inhibiting the MAPK/NF-κB signaling pathway.
# Conclusions
In summary, synbiotics protect the liver and intestines through the gut-liver axis. Synbiotics protect the liver from injury by enhancing the host's antioxidant capacity, inhibiting the MAPK/NF-κB signal pathway and the downstream Beclin1-mediated autophagy cascade, thereby reducing CCl 4 -induced autophagy [fig_ref] Figure 7: Schematic representation of the role of intestinal flora in the improvement of... [/fig_ref]. It is expected that the synbiotic of L. plantarum 1201 and GOS can be used as a new target to protect human health by regulating the balance of intestinal flora through diet. Considering the low sample size, the experimental results may have certain deviations and further studies could be optimized by expanding the sample size. Informed Consent Statement: Not applicable.
# Data availability statement:
The data presented in this study are available in the article and its Supplementary Materials. for making the research data of this project more reliable. Thanks to the National Natural Science foundation of China (81760102 and 31770133) and the College Student Innovation and Entrepreneurship Training Program of Nanchang University (2021CX201). We sincerely thank the Editor and reviewers for their contributions and suggestions.
## Conflicts of interest:
The authors declare that they have no known competing financial interests or personal relationships that could have influenced the work reported in this paper. Informed Consent Statement: Not applicable.
# Data availability statement:
The data presented in this study are available in the article and its Supplementary Materials.
[fig] Figure 1: GOS promoted the growth of L. plantarum 1201 with high-quality antioxidant power. (A) Growth curve of L. plantarum 1201 in GOS treatment mediums and the MRS medium. (B) Picrylhydrazyl free radical (DPPH) radical scavenging assay (%) of L. plantarum 1201. (C) Hydrogen peroxide capacity (%) on L. plantarum 1201. (D) Reducing power concentration on L. plantarum 1201. [/fig]
[fig] Figure 2: Synbiotics alleviated the dysfunction and pathological damage of the liver in CCl4-induced ALI mice. (A) Experimental set-up: C57BL/6N mice with an intraperitoneal injection of 2 μL/g 50% CCl4 peanut oil solution treated with probiotics, prebiotics, or synbiotics by daily gavage. (B) Liver index (n = 9 mice/group; each data point represents one mouse). (C,D) ALT and AST enzyme activity in mouse serum (n = 3). (E) H&E pathological section of mouse liver tissue (n = 4). * p < 0.05; ** p < 0.01; *** p < 0.001; * is compared with MD; paired two-tailed t-test. [/fig]
[fig] Figure 3: Synbiotics reduced oxidative stress and relieved liver inflammation in CCl4-induced ALI mice. (A-C) Homogenizing the liver tissues to examine the activities of SOD (A), the MDA (B), and GSH content (C) (n = 3). (D) the expression levels of IFN-γ, IL-1β, TNF-α, IL-6 and Chemokines (CCL4, CCL5) (n = 3). (E) the expression levels of anti-inflammatory factors (IL-10, IL-22, TGF-β). [/fig]
[fig] 1, Figure 4: Synbiotics alleviated intestinal flora disturbance in CCl 4 -induced ALI mice. (A) PCoA analysis chart, PC1 vs. PC2, "the dots represent each sample; the horizontal and vertical coordinates are the two characteristic values that cause the largest differences between the samples" (n = 5). (B) Three-dimensional PCoA analysis chart, PC1 vs. PC2 vs. PC3. [/fig]
[fig] Figure 5: Synbiotics relieved intestinal inflammation in CCl4-induced ALI mice. (A) the expression levels of IFN-γ, IL-1β, TNF-α, IL-6, CCL4, and CCL5 in the colon tissues (n = 3). (B) the expression levels of IL-10, IL-22, and TGF-β in the colon tissues (n = 3). (C) the mRNA level of Bcl-2 (n = 3). (D) the mRNA levels of claudin3, occludin, zonula occludin-1 (n =3). (E) H&E pathological section of the mouse colon tissue (n = 4). * p < 0.05; ** p < 0.01; *** p < 0.001; * is compared with MD; paired twotailed t-test. [/fig]
[fig] Figure 6: Synbiotics inhibited the MAPK/NF-κB signaling pathway, then inhibited the apoptotic signaling pathway and the autophagic signaling pathway in CCl4-induced ALI mice. (A,B) mRNA levels of the MAPK signaling pathway-related genes in the liver (A) and colon (B) (n = 3). (C-E) Quantitative analysis of protein expressions of NF-κB signaling (C), apoptotic signaling (D), and autophagic signaling (E) pathway-related proteins (n = 3-4). * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001; * is compared with MD; paired two-tailed t-test. [/fig]
[fig] Figure 7: Schematic representation of the role of intestinal flora in the improvement of ALI through synbiotics. After intraperitoneal injection, CCl4 damages the liver, causing oxidative damage and inflammation of the liver; then it affects the balance of intestinal flora through the liver-gut axis, causing intestinal inflammation and damage, and intestinal damage in turn aggravates liver damage, causing autophagy and apoptosis. However, synbiotics act directly on the intestinal flora, relieving intestinal inflammation, inhibiting the autophagy and apoptosis pathways, and protect ing the liver from inflammation and oxidative damage.Supplementary Materials:The following are available online at www.mdpi.com/xxx/s1,Table S1: Primer sequences for quantitative Real-Time polymerase chain reactions.Author Contributions: Z.R., Conceptualization, Methodology, Investigation, Software, Writing-Original Draft; Y.H., Writing-Original Draft; Q.Z., Investigation, Validation; S.C., Investigation, Date Curation; H.L., Software, Investigation; L.P., Formal analysis; H.W., Writing-Review and Editing; C.W., Conceptualization, Methodology, Resources, Funding acquisition, Writing-Review and Editing. All authors have read and agreed to the published version of the manuscript. Funding: This work was supported by the National Natural Science foundation of China (81760102 and 31770133). Institutional Review Board Statement: The study was approved by the Ethics Committee of the Institutional Animal Care and Use Committee of Nanchang Royo Biotech Co. Ltd. (RYE2020061101) and conducted according to the Guide for the Care and Use of Laboratory Animals (China). [/fig]
|
10.1186/s13000-015-0378-x
|
CCBY
|
4528780
|
26249370
|
s2orc_pubmed_articles
|
Association between interleukin 8–251 T/A and +781 C/T polymorphisms and glioma risk
Background: Gliomas are aggressive tumors of the central nervous system that rely on production of growth factors for tumor progression. Interleukin 8 (IL-8) is up-regulated in gliomas to promote angiogenesis and proliferation. The aim of this study was to evaluate the association of the IL-8 -251 T/A and +781 C/T polymorphisms and glioma risk. Methods: We enrolled 300 glioma patients and 300 age-and gender-matched healthy controls. A prospective hospital-based case-control design and logistic regression analysis were utilized. The IL-8 gene polymorphisms were genotyped by polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). Results: Glioma patients had a significantly higher frequency of IL-8 -251 AA genotype [odds ratio (OR) =1.91, 95 % confidence interval (CI) = 1.22, 3.00; P = 0.005] and IL-8 -251 A allele (OR =1.36, 95 % CI = 1.08, 1.70; P = 0.009) than controls. When stratified by the grade of glioma, patients with WHO IV glioma had a significantly higher frequency of IL-8 -251 AA genotype (OR =1.56, 95 % CI = 1.01, 2.39; P = 0.04).Conclusions: To the best of our knowledge, this is the first report in the literature that the IL-8 -251 AA genotype and A allele were at a higher risk for glioma.
# Background
Gliomas make up about 30 % of all brain and central nervous system tumors and 80 % of all malignant brain tumors [bib_ref] Genetics of adult glioma, Goodenberger [/bib_ref]. Despite the growing number of preclinical and clinical trials focused on the treatment of malignant gliomas, the prognosis for this disease remains grim [bib_ref] Epidemiology and etiology of gliomas, Ohgaki [/bib_ref] [bib_ref] Comparison of a strategy favoring early surgical resection vs a strategy favoring..., Jakola [/bib_ref] [bib_ref] CD8+ T cell-independent tumor regression induced by Fc-OX40L and therapeutic vaccination in..., Murphy [/bib_ref]. Only rare familial syndromes and exposure to high therapeutic doses of ionizing radiation are known causes of glioma [bib_ref] Epidemiology and molecular pathology of glioma, Schwartzbaum [/bib_ref]. Some other factors are also found to affect glioma risk, such as: high levels of processed meat consumption and obesity during adolescence [bib_ref] Association between processed meat and red meat consumption and risk for glioma:..., Wei [/bib_ref] [bib_ref] Height, body mass index, and physical activity in relation to glioma risk, Moore [/bib_ref]. Recent research has focused on identifying germline polymorphisms associated with risk of glioma, and using molecular markers to classify glial tumors into more homogenous groups [bib_ref] Epidemiology and molecular pathology of glioma, Schwartzbaum [/bib_ref] [bib_ref] Variants in the CDKN2B and RTEL1 regions are associated with high-grade glioma..., Wrensch [/bib_ref].
Interleukin-8 (IL-8), which is best known for its leukocyte chemotactic properties and associated role in inflammatory and infectious diseases [bib_ref] Essential involvement of interleukin-8 (IL-8) in acute inflammation, Harada [/bib_ref] , is now known to possess tumorigenic and proangiogenic properties as well [bib_ref] Interleukin-8 and human cancer biology, Xie [/bib_ref] [bib_ref] The role of interleukin-8 and its receptors in gliomagenesis and tumoral angiogenesis, Brat [/bib_ref]. IL-8 is encoded by the IL-8 gene located on chromosome 4q13-21, consisting of four exons, three introns, and the proximal promoter region [bib_ref] Genomic structure of the human monocytederived neutrophil chemotactic factor IL-8, Mukaida [/bib_ref]. Several SNPs have been reported in the IL-8 gene and some of them, such as −251 T/A (rs4073) and +781 C/T (rs2227306), can regulate the IL-8 production [bib_ref] An interleukin-8 (IL-8) cDNA clone identifies a frequent HindIII polymorphism, Fey [/bib_ref] [bib_ref] Increased in vivo transcription of an IL-8 haplotype associated with respiratory syncytial..., Hacking [/bib_ref] [bib_ref] The polymorphism interleukin 8-251 A/T influences the susceptibility of Helicobacter pylori related..., Ohyauchi [/bib_ref]. In human gliomas, IL-8 is expressed and secreted at high levels both in vitro and in vivo, and recent experiments suggest it is critical to glial tumor neovascularity and progression [bib_ref] The role of interleukin-8 and its receptors in gliomagenesis and tumoral angiogenesis, Brat [/bib_ref].
We hypothesized that common genetic variants in IL-8 gene influenced the risk of glioma. To test this hypothesis, we performed a prospective hospital-based case-control study to evaluate the association of the IL-8 -251 T/A and +781 C/T polymorphisms and glioma risk.
# Methods
## Study population
In this prospective hospital-based case-control study, we enrolled 300 glioma patients and 300 age-and gender-matched healthy controls between May 2012 and May 2014 in the Department of Neurosurgery, The First Affiliated Hospital of Xi'an Jiao Tong University, China. Tumor type and stage were determined according to the WHO criteria [bib_ref] The 2007 WHO classification of tumours of the central nervous system, Louis [/bib_ref]. In addition, similar to the cases the controls were all required to be born in China to native Chinese Han parents. To confirm the diagnosis, two physicians reviewed the hospital records and validated each case. Collected clinical data included sex, age, mean education, smoking, alcohol drinking and family history of cancer. All data points were collected through interviews with the patient or their families/ surrogates. All parts of the study were approved by the Institutional Ethical Committee of the First Affiliated Hospital of Xi'an Jiao Tong University, and informed consent according to the Declaration of Helsinki was obtained from all participants or their families/ surrogates.
## Dna extraction and genotyping
Genomic DNA was isolated from white blood cells by the commercially available Qiagen kit (QIAGEN Inc., Valencia, CA, USA). IL-8 -251 A/T and +781 C/T polymorphisms were determined by polymerase chain reactionrestriction fragment length polymorphism (PCR-RFLP) as previous described [bib_ref] Polymorphisms in the interleukin-8 gene in patients with chronic obstructive pulmonary disease, Arinir [/bib_ref] [bib_ref] The IL-8 gene polymorphisms and the risk of the hepatitis B virus/infected..., Qin [/bib_ref] [bib_ref] Findings related to IL-8 and IL-10 gene polymorphisms in a Mexican patient..., Olivo-Diaz [/bib_ref]. Restriction enzyme and primer sequences of IL-8 promoter SNPs were listed in . The PCR cycling conditions were 5 min at 94°C followed by 35 cycles of 1 min at 94°C, 1 min at 61°C, and 2 min at 72°C, with a final step at 72°C for 20 min. The PCR product was subjected to digestion for 4 h with restriction enzyme at 37°C. Electrophoresis in a 2.5 % agarose gel followed by ethidium bromide staining and ultraviolet illumination allowed detection of the alleles. For quality control, two independent observers, read all genotypes without knowing about the case or control status. When replicate quality control samples were evaluated, genotypes showed 100 % concordance.
# Statistical analysis
Data are presented as means ± standard deviation (SD) or as percentages for categorical variables. Differences between continuous variables were assessed by Student's t test, while those between categorical variables were evaluated using Pearson x 2 test. The existence of differences in genotypic frequencies between groups were assessed by means of Pearson x 2 test and calculating the odds ratio (OR) with the 95 % confidence intervals (CI). Statistical significance was taken at nominal P-value < 0.05 for all comparisons. SAS version 9.1 (SAS Institute, Cary, NC) was used for all statistical tests.
# Results
## Characteristics of participants
Characteristics of glioma cases and healthy controls were presented in [fig_ref] Table 2: Characteristics of glioma cases and healthy controls [/fig_ref]. No significant differences were found between the glioma cases and healthy controls in sex, age, mean education, smoking, alcohol drinking and family history of cancer [fig_ref] Table 2: Characteristics of glioma cases and healthy controls [/fig_ref]. Among 300 glioma cases, 223 cases had astrocytomas, 43 cases had ependymomas, 19 cases had oligodendrogliomas, and 15 cases had mixed gliomas [fig_ref] Table 2: Characteristics of glioma cases and healthy controls [/fig_ref]. In these cases, 17 were WHO grade I gliomas, 106 were WHO grade II gliomas, 76 were WHO grade III gliomas, and 101 were WHO grade IV gliomas [fig_ref] Table 2: Characteristics of glioma cases and healthy controls [/fig_ref].
## Il-8 -251 t/a polymorphisms and glioma risk
Glioma patients had a significantly higher frequency of IL-8 -251 AA genotype (OR =1.91, 95 % CI = 1.22, 3.00; P = 0.005) and IL-8 -251 A allele (OR =1.36, 95 % CI = 1.08, 1.70; P = 0.009) than controls [fig_ref] Table 3: Genotype and allele frequencies of IL-8 gene polymorphisms among glioma cases and... [/fig_ref]. When stratified by the grade of glioma, patients with WHO IV glioma had a significantly higher frequency of IL-8 -251 AA genotype (OR =1.56, 95 % CI = 1.01, 2.39; P = 0.04) [fig_ref] Table 4: Stratification analysis of IL-8 -251 T/A polymorphism in glioma [/fig_ref]. When stratified by the histology of glioma, there was no significant difference in the distribution of each genotype [fig_ref] Table 4: Stratification analysis of IL-8 -251 T/A polymorphism in glioma [/fig_ref].
## Il-8 + 781 c/t polymorphisms and glioma risk
No association was found between IL-8 + 781 C/T polymorphisms and glioma risk [fig_ref] Table 3: Genotype and allele frequencies of IL-8 gene polymorphisms among glioma cases and... [/fig_ref].
# Discussion
Recent research has focused on identifying germline polymorphisms associated with risk of glioma. A study by Zong et al. suggested that the Cyclin D1 gene G870A polymorphism was a risk factor for the development of glioma [bib_ref] Association between the G870A polymorphism of Cyclin D1 gene and glioma risk, Zong [/bib_ref]. There was a potential decreased susceptibility to glioma in association with the XRCC1 Arg399Gln polymorphism, especially in Asians [bib_ref] Assessment of the association between XRCC1 Arg399Gln polymorphism and glioma susceptibility, Zhu [/bib_ref]. Regulator of telomere elongation helicase 1 (RTEL1) rs6010620 polymorphism was likely to be associated with increased glioma risk [bib_ref] Regulator of telomere elongation helicase 1 (RTEL1) rs6010620 polymorphism contribute to increased..., Zhao [/bib_ref]. A study by Yuan et al. suggested that common variants in ERCC1 may contribute to susceptibility to glioma, especially in Asians [bib_ref] DNA repair gene ERCC1 polymorphisms may contribute to the risk of glioma, Yuan [/bib_ref]. The AA genotype of ERCC1 C8092A might be associated with a higher risk of adult glioma than the CA and CC genotypes and that the risk allele of ERCC2 K751Q confered a significant susceptibility to adult glioma, especially in Asian populations [bib_ref] Association between ERCC1 C8092A and ERCC2 K751Q polymorphisms and risk of adult..., Xu [/bib_ref]. Strong evidence Restriction enzyme and primer sequences of IL-8 SNPs for the association between XRCC3 C18607T polymorphism and glioma risk was found in a metaanalysis [bib_ref] Quantitative assessment of the association between XRCC3 C18607T polymorphism and glioma risk, Wang [/bib_ref]. There was a significant association between EGF +61A > G polymorphism and glioma risk among Asians [bib_ref] Quantitative assessment of the association between +61A > G polymorphism of epidermal..., Tao [/bib_ref]. The polymorphism of interleukin-4 receptor alpha (IL-4Ralpha) rs1801275 played a protective role in the glioma pathogenesis, particularly among Asians [bib_ref] Association of the interleukin-4Ralpha rs1801275 and rs1805015 polymorphisms with glioma risk, Guo [/bib_ref]. The IL-8 gene polymorphisms were also associated with many other cancers. A recent prospective casecontrol study found that IL-8 polymorphism was associated with ovarian cancer [bib_ref] Polymorphism of the IL-8 gene and the risk of ovarian cancer, Koensgen [/bib_ref]. The IL-8 -251 AA genotype was at a higher risk for oral cancer in the Caucasian population [bib_ref] Association of IL-8-251A > T polymorphisms with oral cancer risk: evidences from..., Yang [/bib_ref] [bib_ref] Association between -251A > T polymorphism in the interleukin-8 gene and oral..., Wang [/bib_ref]. The IL-8-251 TA or AA genotype conferred risk of cardia gastric cancer in a population in Southwestern China [bib_ref] Interleukin-4 and −8 gene polymorphisms and risk of gastric cancer in a..., Pan [/bib_ref]. The IL-8 -251 AA genotype was associated with the overall risk of developing gastric cancer and may seem to be more susceptible to overall gastric cancer in Asian populations [bib_ref] A meta-analysis of interleukin-8 -251 promoter polymorphism associated with gastric cancer risk, Xue [/bib_ref]. The IL-8 -251 T/A polymorphism was associated with lung cancer susceptibility in Asians and the −251 A allele may increase risk of lung cancer in Asians [bib_ref] Association between interleukin-8 -251A/T polymorphism and risk of lung cancer: a meta-analysis, Gao [/bib_ref]. A HuGE review and meta-analysis based on 42 case-control studies found that -251A allele was susceptible in the development of low-penetrance cancers [bib_ref] −251 T/A polymorphism of the interleukin-8 gene and cancer risk: a HuGE..., Wang [/bib_ref]. Two case-control studies found that IL-8 -251 T/A polymorphism was significantly associated with colorectal cancer susceptibility risk [bib_ref] Risk modification of colorectal cancer susceptibility by interleukin-8 -251 T > A..., Mustapha [/bib_ref] [bib_ref] The lL-8 and IL-13 gene polymorphisms in inflammatory bowel disease and colorectal..., Walczak [/bib_ref]. The IL-8-251 T/A polymorphism was associated with breast cancer risk [bib_ref] IL-8-251A > T polymorphism is associated with breast cancer risk: a meta-analysis, Huang [/bib_ref]. A study found that IL-8 -251 T/A polymorphism was associated with bladder cancer susceptibility and outcome after bacillus Calmette-Guerin immunotherapy in a northern Indian cohort [bib_ref] IL-8 -251 T > A polymorphism is associated with bladder cancer susceptibility..., Ahirwar [/bib_ref].
The mechanisms of the IL-8 -251 AA genotype and A allele as risk factors of glioma are still unclear. In human gliomas, IL-8 is expressed and secreted at high levels both in vitro and in vivo, and recent experiments suggest it is critical to glial tumor neovascularity and progression [bib_ref] The role of interleukin-8 and its receptors in gliomagenesis and tumoral angiogenesis, Brat [/bib_ref]. There is strong evidence that IL-8 secretion is associated with glioma formation and malignant progression [bib_ref] The role of interleukin-8 and its receptors in gliomagenesis and tumoral angiogenesis, Brat [/bib_ref]. The IL-8 -251A allele in a homozygous state has been associated with increased expression of the IL-8 gene [bib_ref] Interleukin-8 promoter polymorphism increases the risk of atrophic gastritis and gastric cancer..., Taguchi [/bib_ref] [bib_ref] Submit your next manuscript to BioMed Central and take full advantage of:..., Hull [/bib_ref]. It is entirely plausible that the the IL-8 -251 AA genotype and A allele will affect the risk of gliomas.
There is limitation that needs to be acknowledged and addressed regarding the present study. First of all, these results should be interpreted with caution because the population was only from China, which reduces the possibility of confounding from ethnicity, so it does not permit extrapolation of the results to other ethnic groups. Second, this study is limited by its size and lack of replication. Additional large scale studies are needed to confirm this finding. Third, since controls were recruited from those who came to hospitals for routine health
# Conclusion
To the best of our knowledge, this is the first report in the literature that the IL-8 -251 AA genotype and A allele were at a higher risk for glioma. Additional large scale studies are needed to confirm this finding. Genetic risk profiling has the potential to identify individuals who have not yet had glioma develop but are at high risk. Analysis of polymorphic variants will potentially provide a tool to assess the risk of glioma decades before diagnosis. This process provides ample opportunity to implement behavioral and environmental changes.
# Ethical approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
## Informed consent
Informed consent was obtained from all individual participants included in the study.
## Competing interests
All authors declare that they have no competing interests.
Authors' contributions HL and PM carried out the molecular genetic studies and drafted the manuscript. CX and WX carried out the genotyping. MW and HJ participated in the design of the study and performed the statistical analysis. HL, PM, CX, WX, MW and HJ conceived of the study, and participated in its design and coordination and helped to draft the manuscript. All authors read and approved the final manuscript.
[table] Table 2: Characteristics of glioma cases and healthy controls [/table]
[table] Table 3: Genotype and allele frequencies of IL-8 gene polymorphisms among glioma cases and healthy controls [/table]
[table] Table 4: Stratification analysis of IL-8 -251 T/A polymorphism in glioma [/table]
|
10.1002/ijc.28944
|
CCBY
| null |
24789676
|
s2orc_pubmed_articles
|
A preclinical evaluation of a novel multikinase inhibitor, SKLB‐329, as a therapeutic agent against hepatocellular carcinoma
Hepatocellular carcinoma (HCC) is a serious life-threatening malignant disease of liver. Molecular targeted therapies are considered a promising strategy for the treatment of HCC. Sorafenib is the first, and so far the only targeted drug approved by the US Food and Drug Administration (FDA) for clinical therapy of HCC. Despite being effective in some HCC patients, some demerits of sorafenib in the treatment of HCC, such as modest survival benefits, and drug resistance, have also been reported, which highlights the unmet medical need among patients with HCC. Here, we report a novel multikinase inhibitor discovered by us, SKLB-329, which potently inhibits angiogenesis-related kinases including VEGFR1/2/3, and FGFR2, and the Src kinase. SKLB-329 significantly inhibited endothelial cell growth, migration, invasion and tube formation. It showed potent antiangiogenic activity in a transgenic zebrafish model. Moreover, SKLB-329 could efficiently restrain the proliferation of HCC cells through down-regulation of Src-mediated FAK and Stat3 activity. In vivo, oral administration of SKLB-329 considerably suppressed the tumor growth in HCC xenograft models (HepG2 and SMMC7721) in a dose-dependent manner. In all of the in vitro and in vivo assays of this investigation, sorafenib was used as a positive control, and in most assays SKLB-329 exhibited a higher potency compared with the positive control. In addition, SKLB-329 also bears favorable pharmacokinetic properties. Collectively, the results of preclinical studies presented here demonstrate that SKLB-329 is a promising drug candidate for HCC treatment.
Hepatocellular carcinoma (HCC) is one of the most common malignances worldwide with 630 thousands new cases reported per year. [bib_ref] Advanced hepatocellular carcinoma. Review of targeted molecular drugs, Alves [/bib_ref] Unfortunately, most HCC patients present with tumor in advanced and/or incurable stages, making them ineligible for curative therapies such as surgical resection or liver transplantation. For these HCC patients, systemic treatments are the main option. Nevertheless, chemotherapy with conventional cytotoxic agents is often very toxic and fairly ineffective. [bib_ref] Advanced hepatocellular carcinoma. Review of targeted molecular drugs, Alves [/bib_ref] [bib_ref] Management of hepatocellular carcinoma: an update, Bruix [/bib_ref] The lack of effective systemic therapy for HCC patients has changed with the approval of the first molecular targeted agent sorafenib for clinical use in advanced HCC by the US FDA. [bib_ref] Development of molecularly targeted therapies in hepatocellular carcinoma: where do we go..., Finn [/bib_ref] Sorafenib is a multikinase inhibitor and mainly targets angiogenesis and RAF/MEK/ERK signaling cascade. Despite being effective in some HCC patients, some demerits for sorafenib, such as modest survival benefits, and drug resistance, have also been reported, [bib_ref] Mechanisms of resistance to sorafenib and the corresponding strategies in hepatocellular carcinoma, Zhai [/bib_ref] [bib_ref] Activation of phosphatidylinositol 3-kinase/Akt signaling pathway mediates acquired resistance to sorafenib in..., Chen [/bib_ref] [bib_ref] EGFR activation is a potential determinant of primary resistance of hepatocellular carcinoma..., Ezzoukhry [/bib_ref] which highlights the unmet medical need among patients with HCC.
In development of molecular targeted agents, the selection of targets is critical. Currently, a number of targets have been demonstrated to be associated with HCC. However, very specific targets for HCC have not been identified, which might be due to the fact that HCC is a biologically heterogeneous malignant disease. In this case, an optimal choice is multitarget drugs, which have been demonstrated to be superior to single-target ones in other cancer types. [bib_ref] Multi-target drugs: the trend of drug research and development, Lu [/bib_ref] [bib_ref] Inhibition of autophagy potentiates the antitumor effect of the multikinase inhibitor sorafenib..., Shimizu [/bib_ref] [bib_ref] Multi-target therapeutics: when the whole is greater than the sum of the..., Zimmermann [/bib_ref] Among targets associated with HCC, receptor tyrosine kinases related to angiogenesis, such as VEGFR, and FGFR, could be one of the most important targets because HCC is a highly vascular tumor. [bib_ref] Sorafenib blocks the RAF/MEK/ERK pathway, inhibits tumor angiogenesis, and induces tumor cell..., Liu [/bib_ref] [bib_ref] A novel monoclonal antibody to fibroblast growth factor 2 effectively inhibits growth..., Wang [/bib_ref] Another important target is the nonreceptor tyrosine kinase Src. Src plays a broad and crucial role in activating PI3K/Akt, Ras/Raf/MAPK, JAK/Stats and SFK/FAK/ p130CAS signaling pathways; these pathways are associated with cellular proliferation, survival, adhesion, migration and invasion, as well as angiogenesis. [bib_ref] Src signaling in cancer invasion, Guarino [/bib_ref] [bib_ref] Src kinases catalytic activity regulates proliferation, migration and invasiveness of MDA-MB-231 breast..., Sanchez-Bailon [/bib_ref] [bib_ref] Regulation of SRC family kinases in human cancers, Sen [/bib_ref] [bib_ref] Latent bone metastasis in breast cancer tied to Srcdependent survival signals, Zhang [/bib_ref] [bib_ref] Current status of SRC inhibitors in solid tumor malignancies, Puls [/bib_ref] Recent studies have demonstrated that activation of Src is highly related to the early stage phenotype of HCC, and Src could be an independent prognostic marker of HCC patient. [bib_ref] Crosstalk between activated and inactivated c-Src in hepatocellular carcinoma, Chen [/bib_ref] [bib_ref] Activation of c-Src gene product in hepatocellular carcinoma is highly correlated with..., Ito [/bib_ref] [bib_ref] Expression of Src and FAK in hepatocellular carcinoma and the effect of..., Lau [/bib_ref] These results highlight Src as a potential target for HCC therapy. We therefore hypothesized that agents that simultaneously target angiogenesis and Src might exhibit improved anti-HCC efficacy. Thus, we performed a rational drug design study and lead optimization (related investigations will be reported elsewhere), and obtained a novel compound termed SKLB-329 [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref] , which potently inhibits VEGFR1/2/3, FGFR2 and Src. SKLB-329 displayed a more potent anti-tumor activity than sorafenib in HCC xenograft models. In this account, we report the preclinical assessment results of SKLB-329.
# Material and methods
## Compounds
The 1-(4-((1H-pyrazolo pyrimidin-4-yl)oxy)-2-fluorophenyl)-3-(4-chloro-3-(trifluoromethyl)phenyl)urea (SKLB-329, [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref] was synthesized at the State Key Laboratory of Biotherapy, Sichuan University (details for the synthesis of SKLB-329 see Supporting Information). Sorafenib was obtained from commercial source. A stock solution of SKLB-329 for all assays in vitro was prepared in DMSO and then diluted in optimal medium. The final concentration of DMSO in the incubation mixture did not exceed 0.1% (v/v) in each experiment.
## Kinase inhibition assays
Kinase inhibition assays were performed by KinaseProfiler service (Millipore, USA). SKLB-329 (0.001-10 lM) or vehicle was incubated with protein kinases in reaction solution containing 8 mM 3-(N-morpholino) propanesulfonic acid (MOPS, pH 7.0), 0.2 mM EDTA, 0.33 mg mL 21 myelin basic protein, 10 mM Mg acetate, 10 mM [g-33 P-ATP], and corresponding peptide as substrate. After incubation for 40 min at room temperature, the reaction was stopped by 3% phosphoric acid solution, and 10 lL of the reaction solution was spotted onto a P30 filtermat. The samples were then washed 3 times with 75 mM phosphoric acid and once in methanol before detecting the kinase activity by scintillation counting.
## Cells and cell culture conditions
Human cancer cell lines and normal liver cell line (HL-7702) used in this investigation were obtained from American Type Culture Collection (ATCC, Rockville, MD) and National Platform of Experimental Cell Resources for Sci-Tech (China), respectively, and were grown in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% FBS (Gibco, Eggenstein, Germany), 100 U mL 21 penicillin (Sigma-Aldrich, USA), and 100 lg mL 21 streptomycin (Sigma-Aldrich, USA). All cell lines were maintained at 37 C, surrounded by a humidified atmosphere of 5% CO 2 , and passaged for <6 months after receipt or resuscitation. Human umbilical vein endothelial cell (HUVEC) was isolated from human umbilical cord veins according to a standard procedure and grown in EGM-2 medium (EBM2 medium containing EndoGRO TM VEGF supplement kit, Millipore, USA). HUVECs at passage 2-6 were used for all related experiments.
## Cell viability and proliferation assays
Cell viability was measured using MTT assay. The cell lines were seeded in 96-well plates at 2500-4000 cells/well (depending on cell type) in DMEM supplemented with 10% FBS and incubated overnight. The cells were treated with drugs in different concentrations for 72 hr in serumcontaining media, and then 20 lL of MTT solution (5 mg mL 21 , Sigma, USA) was added to each well. After incubation for 2-4 hr at 37 C, the formazan crystals were dissolved with 50 lL of acidified SDS (20%, w/v). Absorbance was determined at 570 nm on Multiskan MK3 (Thermo Scientific, USA) the next day. Each assay was performed in three replicates and all experiments were repeated at least twice. For the HUVEC growth inhibition assay, the cells seeded in 96well plates (1 3 10 4 per well) were starved overnight in What's new? Targeted therapy is a potentially effective means of treating hepatocellular carcinoma (HCC), but only a single targeted agent, sorafenib, is FDA-approved, and its clinical benefits are limited. In the interest of meeting the need for the identification of new drug candidates, the present report describes the preclinical investigation of a novel multikinase inhibitor, SKLB-329. In HCC xenografts, SKLB-329 was found to possess anti-angiogenic activity and to exert anti-tumor effects that were more potent than those of sorafenib. SKLB-329 anti-tumor activity was associated with the inhibition of HCC cell proliferation via blockade of Src/FAK and Src/Stat3 signalings.
EBM2 medium and incubated with SKLB-329 for 45 min. Then 50 ng mL 21 growth factor (VEGF, bFGF or EGF) or 5% FBS was introduced into the assays and the cells were continued to incubate for 72 hr. The subsequent procedures were performed as described above.
For the HCC cell proliferation assay, cells were seeded in 96-well plates and treated with 20 lM SKLB-329 the next day. After exposure to the drug-containing medium for 24 hr, EdU incorporation assay was conducted on the cells following the manufacturer's instruction (RIBOBIO, China), and the pictures were captured using ArrayScan VTI HCS reader (Thermo Scientific, USA).
## Colony formation assay
Cells were seeded in six-well plates at a density of 5,000 per well and treated with vehicle or SKLB-329 the next day. The medium containing vehicle or SKLB-329 was replaced every 3 days. Cells were fixed with methanol and stained with crystal violet after treatment for 10 days.
## Flow cytometry for cell cycle analysis
The cell cycle was measured by FCM assay. Approximately 1 3 10 5 cells were collected after treatment with vehicle or SKLB-329 in various concentrations for 48 hr and fixed with 70% ethanol overnight. The cells were then washed with cold PBS and stained with 50 lg mL 21 propidium iodide containing 100 lg mL 21 RNase, and 0.1% Triton X-100. The cell cycle profiles were determined on a FACS Calibur flow cytometer (Becton Dicknson, USA) and analyzed using the ModFit LT 3.2 software (Verity Software House, USA).
## Wound healing assay
HUVECs were cultured to confluence in 24-well plates and wounded using a sterilized yellow pipette tip to make a straight scratch. Cells were rinsed with sterile PBS gently, and then PBS was replaced with EGM2 medium containing vehicle, sorafenib or SKLB-329. Pictures were taken by an OLYMPUS digital camera attached to a light microscope after 18 hr.
## Transwell invasion assay
Transwell invasion assay was done as described previously. [bib_ref] SKLB1206, a novel orally available multikinase inhibitor targeting EGFR activating and T790M..., Pan [/bib_ref] In brief, 50 lL per well of diluted Matrigel (BD Biosciences, USA) was added to the Transwell compartments (Millipore, USA), which have been inserted into a 24-well plate. HUVECs were suspended in EBM-2 medium (without growth factors) 1 hr later and seeded in the upper chamber (3 3 10 4 /100 lL). Afterward, another 100 lL medium containing vehicle, sorafenib or SKLB-329 was added to each upper chamber. The lower compartments were filled with 500 lL EGM-2 medium (EBM-2 medium supplemented with various growth factors). After incubation for 24 hr at 37 C, the migrated cells were fixed with methanol and stained with 0.05% crystal violet for 15 min, followed by rinsing twice with PBS. The cells were photographed under a light microscope (Leica, Germany).
## Tube formation assay
The tube formation assay was conducted as described previously. [bib_ref] SKLB1002, a novel potent inhibitor of VEGF receptor 2 signaling, inhibits angiogenesis..., Zhang [/bib_ref] Briefly, 50 mL per well of Matrigel was added to the 96-well plate and incubated for 30 min at 37 C to allow the gel to solidify. HUVECs were harvested and seeded on the matrigel for 10 min, followed by treatment with vehicle, sorafenib or serial dilutions of SKLB-329. Cells were photographed with an OLYMPUS digital camera 6 hr later.
## Angiogenesis in live fluorescent zebrafish assay
Anti-angiogenesis activity of SKLB-329 and sorafenib was assessed in transgenic zebrash (FLK-1: EGFP) according to the protocol reported previously. [bib_ref] SKLB610: a novel potential inhibitor of vascular endothelial growth factor receptor tyrosine..., Cao [/bib_ref] Zebrafish embryos at the 13-somite stage (30 embryos per group) were incubated overnight with vehicle, sorafenib or SKLB-329. Then zebrafish were anesthetized and imaged using a fluorescence microscope (Carl Zeiss Microimaging, Germany).
# Western blot analysis
For the HUVEC immunoblot studies, subconfluent cells were serum starved overnight in EBM2 medium, and then incubated with vehicle, sorafenib, or SKLB-329 for 2 hr, followed by treatment with 50 ng mL 21 recombinant human VEGF or bFGF (Peprotech, USA) for 10 min. The cells were lysed in RIPA buffer (Beyotime, China) containing Roche protease inhibitor cocktail, and the protein concentrations were determined by the Bradford method. Proteins were separated by gel electrophoresis on 5-10% SDS-PAGE gels and probed with specific antibodies (Cell Signaling Technology, USA) including anti-VEGFR2, anti-pVEGFR2 Tyr1175 , anti-FAK, anti-pFAK Tyr925 , anti-Src, anti-pSrc Tyr416 , anti-AKT, anti-pAKT Ser473 , anti-ERK, anti-pERK Thr202/Tyr204 and anti-b-actin. All of the antibodies were used at a 1:1,000 dilution, and the horseradish peroxidase-coupled secondary antibodies (Zhong Shan Golden Bridge Bio-technology,China) were used at 1:5,000.
For the HepG2 and PLC/PRF/5 Western blot assays, cells were incubated for 24 hr in medium containing vehicle, SKLB-329, or sorafenib, and then lysed in RIPA buffer. Western blots were done on whole-cell extracts as described above.
## In vivo antitumor effects
All animal studies were conducted according to the guidelines of the Animal Care and Use Committee of Sichuan University (Chengdu, Sichuan, China). Female Balb/c athymic mice were purchased from Chinese Academy of Medical Science (Beijing, China). Nearly 5 3 10 6 tumor cells were implanted subcutaneously into the hind flank region of mice, and tumors were allowed to reach a size of 200 mm 3 before initiation of treatment. The solvent containing 10% Cremophor EL (Sigma, USA), 10% ethanol, and 80% water was used as vehicle. Tumor sizes were monitored twice-weekly with vernier caliper and calculated as length 3 width 2 /2. Studies were typically terminated when tumors in vehicletreated animals reached an average size of 1,500 mm 3 . Inhibition rate of tumor growth was calculated as 1003{12[(tumor volume final 2tumor volume initial ) for the compound-treated group]/[(tumor volume final 2tumor volume initial ) for the vehicle-treated group]}.
## Immunohistochemistry
Tumor-bearing mice were given vehicle or SKLB-329 (15 mg kg 21 ) once daily by oral gavage, and parafin-embedded tumor sections were made after harvesting the tumors which had been treated for 2 weeks. For immunohistochemistry studies, antigen retrieval was performed on tumor sections and the following antibodies were used: phospho-Src (Cell Signaling Technology, 1:100), phospho-FAK (Cell Signaling Technology, 1:100), phospho-Stat3 (Cell Signaling Technology, 1:200). A DAKO polymer secondary antibody system (Dako Envision 1K4007) was used for secondary detection, and the sections were counterstained with Carazzi's hematoxylin. Additionally, Rat anti-mouse CD31 antibody (BD Biosciences, USA) and anti-Ki67 antibody (Thermo Fisher Scientific, USA) were used to determine vessel density and cell proliferation following the manufacturer's protocol, respectively. Images were captured using an Olympus digital camera.
## Pharmacokinetic assessments
The pharmacokinetic properties of SKLB-329 were investigated in male Sprague-Dawley rats (Chinese Academy of Medical Science, Beijing, China). A catheter was surgically placed into the jugular vein of the rats, and the animals were Statistical analysis SPSS 13.0 was used to evaluate the statistical significance of differences between the means. Data were analyzed by the Student t test and ANOVA. Differences were considered significant if p < 0.05.
# Results
## Kinase inhibition potency of sklb-329
The kinase inhibition potency of SKLB-329 against various human protein kinases were measured with gold-standard 33 P radiolabeled technology. The results are shown in [fig_ref] Table 1: In vitro kinase inhibition profile of SKLB-329 [/fig_ref]. SKLB-329 potently inhibited VEGFR1/2/3 with IC 50 values of 5, 18, and 5 nM, respectively, which are more potent than that of sorafenib (Corresponding IC 50 values for sorafenib are 26, 90, and 20 nM, respectively). [bib_ref] Myelosuppression and kinase selectivity of multikinase angiogenesis inhibitors, Kumar [/bib_ref] [bib_ref] Discovery and development of sorafenib: a multikinase inhibitor for treating cancer, Wilhelm [/bib_ref] [bib_ref] BAY 43-9006 exhibits broad spectrum oral antitumor activity and targets the RAF/MEK/ERK..., Wilhelm [/bib_ref] It showed a considerable potency against FGFR2 and Src with IC 50 values of 26 and 18 nM, respectively. Though sorafenib has some activity against FGFR2 and Src, its potencies are very weak (IC 50 s: 825 nM for FGFR2, 390 nM for Src). [bib_ref] Myelosuppression and kinase selectivity of multikinase angiogenesis inhibitors, Kumar [/bib_ref]
## Sklb-329 inhibits huvec growth, migration, invasion and tube formation
To assess the anti-angiogenesis effects of SKLB-329 in vitro, the inhibitory activity of SKLB-329 against various growth factors-induced growth of HUVECs was measured by MTT assay. SKLB-329 showed a strong inhibitory effect on VEGF or bFGF-stimulated cells, and the VEGF-stimulated cells seemed more sensitive to SKLB-329 compared with bFGFstimulated cells [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref]. SKLB-329 displayed a very weak anti-viability effect on EGF or FBS stimulated HUVECs. These results indicated that SKLB-329 inhibited endothelial cells growth mainly through the suppression of VEGF and FGF signaling.
Angiogenesis arising from endothelial cells in the preexisting vessels is a complex process, in which cell migration and invasion are both pivotal steps. The inhibitory effect of SKLB-329 on HUVEC migration was evaluated using wound healing assay. After exposure to SKLB-329 in various concentrations for 18 hr, the number of migrating cells was significantly diminished by SKLB-329 as compared with vehicle [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref] ; the IC 50 was about 2.5 lM. In transwell invasion assay, treatment of cells with 2.5-10 lM SKLB-329 strongly inhibited invasion by 40.2-84.4% [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref]. Tube formation of endothelial cells is also indispensable for angiogenesis. We used Matrigel-based tube formation assay to assess the effect of SKLB-329 and found that it dose-dependently inhibited the ability of HUVEC to assemble into branched capillarylike structures, with inhibition rates of 31.6, 68.4, and 85.4% at concentrations of 2.5, 5, and 10 lM, respectively [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref]. As a positive control, sorafenib also showed comparable suppression effects in all functional assays. Taken together, these data demonstrated that SKLB-329 could effectively inhibit angiogenesis in vitro.
## Sklb-329 inhibits embryonic angiogenesis in zebrafish
Transgenic zebrafish assays were adopted to further assess the anti-angiogenesis effect of SKLB-329. As shown in [fig_ref] Figure 2: Anti-angiogenesis effect of SKLB-329 in transgenic zebrafish embryos and its anti-angiogenesis mechanism... [/fig_ref] , SKLB-329 could dose-dependently inhibit the intersegmental blood vessel growth. A concentration of 1.25 lM could lead to entire blockade of the intersegmental blood vessel growth, and 0.625 lM SKLB-329 resulted in the interseg-mental blood vessel growth suppressed by 65.4%. By comparison, a complete inhibition of the intersegmental blood vessel growth needed a concentration of 2.5 lM for sorafenib, and 1.25 lM sorafenib led to a suppression of 52.7% [fig_ref] Figure 2: Anti-angiogenesis effect of SKLB-329 in transgenic zebrafish embryos and its anti-angiogenesis mechanism... [/fig_ref]. These results showed that SKLB-329 could effectively restrain angiogenesis in vivo and had higher potency of anti-angiogenesis than sorafenib.
## Inhibition of vegfr and fgfr signalings in huvecs
The ability of SKLB-329 to inhibit the VEGFR and FGFR signalings in intact cells was evaluated in HUVECs using Western blot analysis. As shown in [fig_ref] Figure 2: Anti-angiogenesis effect of SKLB-329 in transgenic zebrafish embryos and its anti-angiogenesis mechanism... [/fig_ref] , the VEGFR signaling was activated upon stimulation by VEGF. This activation could be effectively inhibited by SKLB-329: VEGFR phosphorylation was completely blocked at concentrations higher than 10 nM, and downstream kinases including AKT, extracellular signal-regulated kinase (ERK), FAK and Src were also significantly inhibited with IC 50 s being about 10 nM. Very similar results were observed in FGF-stimulated HUVECs. As indicated in [fig_ref] Figure 2: Anti-angiogenesis effect of SKLB-329 in transgenic zebrafish embryos and its anti-angiogenesis mechanism... [/fig_ref] , FGF stimuli also activated VEGFR and the downstream effector proteins of VEGFR and FGFR, and this activation was significantly suppressed when exposed to SKLB-329. The IC 50 value for VEGFR phosphorylation was between 1 and 10 nM, and those for downstream kinases were between 10 and 100 nM.
Here one may notice that the downstream kinase Src could not be fully inactivated even at high concentrations (>100 nM) of SKLB-329. One of the possible reasons could be that Src is also regulated by other signaling pathways in addition to VEGFR/FGFR signaling. Collectively, these data suggested that SKLB-329 inhibited angiogenesis mainly through blockade of VEGFR and FGFR signaling pathways.
## Inhibition of tumor cell viability and colony formation in vitro
The anti-viability activity of SKLB-329 against HCC cells was measured by MTT assay. SKLB-329 could efficiently inhibit the viability of HCC cells including HepG2, SMMC7721, and PLC/ PRF/5, with IC 50 values of 15.6, 16.3, and 14.8 lM, respectively [fig_ref] Figure 3: SKLB-329 inhibited HCC cell viability and colony formation in vitro [/fig_ref] , which were comparable to those for sorafenib (the corresponding IC 50 values were 11.0, 12.2, and 11.1 lM, respectively). In addition, both SKLB-329 and sorafenib showed weaker anti-viability activity on a normal human liver cell line, HL-7702; the IC 50 values were 22.8 and 16.9 lM for SKLB-329 and sorafenib, respectively. Colony formation assays were then performed to visually assess the cytoreductive activity. SKLB-329 at a concentration of 10 lM notably decreased the formation of colonies in all the three HCC cell lines, but showed no apparent influence on HL-7702 colonies [fig_ref] Figure 3: SKLB-329 inhibited HCC cell viability and colony formation in vitro [/fig_ref]. Collectively, these results indicated that SKLB-329 could efficiently inhibit cell viability and colony formation of HCC cells, but had lower toxicity to normal liver cells and no apparent impact on normal liver cell colonies.
## Sklb-329 inhibits hcc cell proliferation via blocking src/fak and src/stat3 signalings
The anti-proliferation effect of SKLB-329 was examined by Edu cell proliferation assay. As shown in , treatment with SKLB-329 markedly reduced the number of proliferating cells (red nuclei) compared with the control. Cell cycle studies showed that SKLB-329 dose-dependently induced cellcycle arrest in the G0/G1 phase in all the three HCC cell lines .
To further study the anti-proliferative mechanism of SKLB-329, changes in the phosphorylation levels of pivotal proteins modulated by Src were detected by Western blot analysis in HepG2 and PLC/PRF/5 HCC cell lines. SKLB-329 effectively inhibited Src phosphorylation, and down-regulated the phosphorylation levels of FAK and Stat3 at concentrations between 3 and 30 lM in HepG2 tumor cells, and between 10 and 30 lM in PLC/PRF/5 tumor cells . Nevertheless, there were no changes in the phosphorylation levels of AKT and ERK. These results imply a mechanism of action different from that of sorafenib, which exerts its anti-proliferative activity mainly through inhibition of MAPK signaling. [bib_ref] Sorafenib blocks the RAF/MEK/ERK pathway, inhibits tumor angiogenesis, and induces tumor cell..., Liu [/bib_ref] [bib_ref] Vitamin K enhancement of sorafenib-mediated HCC cell growth inhibition in vitro and..., Wei [/bib_ref] To further verify that sorafenib uses different mechanisms of action in suppressing HCC cell proliferation, we examined the influence of the two agents on phosphorylation level of MEK in the PLC/PRF/5 cell line.
As shown in , there was a significant change in phosphorylated MEK when the cells were treated with sorafenib, but not in the SKLB-329 treatment group. All of these results revealed that SKLB-329 inhibited the growth of HCC cells mainly through the suppression of Src/FAK and Src/ Stat3 signalings, but not the MAPK cascade, which is the main target of sorafenib.
## Antitumor efficacy of sklb-329 in human hcc xenograft models
The in vivo anti-tumor efficacy of SKLB-329 was assessed using HepG2 and SMMC7721 tumor xenograft models. Oral dosing of SKLB-329 at 7.5 and 15 mg kg 21 day 21 slowed down the tumor growth dose-dependently, with tumor growth inhibition rate of 62.0 and 80.2%, respectively, for the HepG2 model, and 57.4 and 77.0%, respectively, for the SMMC7721 model . As a positive control, 30 mg kg 21 sorafenib groups showed inhibition rates of 69.8 and 63.2% for the HepG2 and SMMC7721 models, respectively, indicating a slightly weaker antitumor activity compared with those of SKLB-329. No weight loss and pathological changes of major organs (Supporting Information [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref] were observed in all treatment groups during the whole experiments.
## Anti-tumor mechanisms of action of sklb-329
To determine the in vivo anti-tumor mechanisms of action of SKLB-329, immunohistochemical staining assays were carried out on the tumor tissues. show the immunohistochemical analysis results for the tumor tissues in the HepG2 and SMMC7721 models, respectively. Tumor tissues from the SKLB-329 treatment groups all showed a decrease in the phosphorylation levels of Src, FAK and Stat3, as well as in the tumor mitotic index (Ki67) compared with the corresponding control groups , suggesting that SKLB-329 was also able to inhibit the proliferation of tumor cells in vivo. In addition, a significantly reduced microvessel density was also observed in the SKLB-329 treatment groups compared with the control groups (Figs. 5c and 5d), indicating inhibition of angiogenesis. Taken together, SKLB-329 exerts its antitumor effects in vivo through inhibition of both tumor angiogenesis and tumor cell proliferation.
## Pharmacokinetic characteristics of sklb-329
The pharmacokinetic characteristics of SKLB-329 were assessed on rats following peros administration. The plasma concentration versus time profile is presented in .
The key pharmacokinetic parameters calculated are summarized in Supporting Information [fig_ref] Table 1: In vitro kinase inhibition profile of SKLB-329 [/fig_ref]. The area under the concentration-time curve (AUC 0-72h ) is 386.7 lg mL 21 h 21 , which is slightly larger than that of sorafenib (336.5 lg mL 21 h 21 ). SKLB-329 was absorbed very well, achieving a maximum plasma concentration (C max ) of 14.1 lg mL 21 within about 9.2 hr, and displayed long half-life (t 1/2 ) of 11.7 hr, slow clearance rate (CL) of 0.05 L h 21 kg 21 , and small apparent distribution volume (V ss ) of 0.8 L kg 21 , indicating that SKLB-329 was mainly distributed in vasculature and eliminated slowly from the body.
# Discussion
The anti-angiogenesis therapy has been established as an efficacious strategy for the treatment of solid tumors. [bib_ref] SKLB1002, a novel potent inhibitor of VEGF receptor 2 signaling, inhibits angiogenesis..., Zhang [/bib_ref] [bib_ref] Vitexicarpin acts as a novel angiogenesis inhibitor and its target network, Zhang [/bib_ref] [bib_ref] Identification of phosphorylase kinase as a novel therapeutic target through high-throughput screening..., Camus [/bib_ref] The success of sorafenib in treating HCC could be one of the most convincing proofs. SKLB-329 reported here is also a small molecule anti-angiogenesis agent. In terms of the potency of anti-angiogenesis, SKLB-329 is more potent than sorafenib. For example, in biochemical assays, SKLB-329 displayed higher activities in inhibiting receptor tyrosine kinases associated with angiogenesis, including VEGFR1/2/3, FGFR2; the IC 50 values for SKLB-329 against VEGFR1/2/3 and FGFR2 are 5, 18, 5 and 26 nM, respectively, and those for sorafenib are 26, 90, 20 and 825 nM, respectively. In functional assays, such as HUVEC assays and zebrafish embryonic assay, SKLB-329 also exhibited higher or comparable potency compared with sorafenib. Here, it is necessary mentioning that though SKLB-329 has low nanomolar enzymatic activities (IC 50 s) against VEGFR1/2/3 and FGFR2, and almost ablated phosphorylation of VEGFR in HUVEC at nanomolar range (10-1,000 nM), inhibition of HUVEC migration, invasion, and tube formation still required much higher concentrations (2.5-10 lM, [fig_ref] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays [/fig_ref]. The most plausible explanation is the significant binding of SKLB-329 with proteins in the medium which shifts potency into micromolar range because extracts for western blots were prepared from serum starved HUVECs whereas all phenotypic assays were performed in a presence of serum. In addition, various growth factors contained in the medium could also play a role since these growth factors might activate angiogenesis-related signaling pathways, which may partially compensate the inhibition action of SKLB-329. Another attractive point for SKLB-329 is its ability to inhibit Src. Src plays critical roles in cell morphology, adhesion, differentiation, proliferation, survival, migration and invasion. Activation of Src can stimulate a variety of downstream signaling pathways, such as PI3K/Akt, Ras/ Raf/MAPK, JAK/Stats and SFK/FAK/p130CAS cascades; these signaling pathways play crucial roles in regulating the development of cancer. Over-expression or abnormal activation of Src protein kinase was often detected in a wide variety of human cancers, including tumors of the breast, prostate, pancreas, lung, ovarian, colon and brain. [bib_ref] c-Src protein expression is increased in human breast cancer. An immunohistochemical and..., Verbeek [/bib_ref] [bib_ref] SRC family kinases as novel therapeutic targets to treat breast cancer brain..., Zhang [/bib_ref] [bib_ref] Role of Src expression and activation in human cancer, Irby [/bib_ref] [bib_ref] The Src signaling pathway: a potential target in melanoma and other malignancies, Homsi [/bib_ref] [bib_ref] pp60c-src activation in lung adenocarcinoma, Masaki [/bib_ref] [bib_ref] Activated SRC protein tyrosine kinase is overexpressed in late-stage human ovarian cancers, Wiener [/bib_ref] [bib_ref] Src family kinases in tumor progression and metastasis, Summy [/bib_ref] [bib_ref] Overexpression and activation of the tyrosine kinase Src in human ancreatic carcinoma, Lutz [/bib_ref] [bib_ref] Action of the Src family kinase inhibitor, dasatinib (BMS-354825), on human prostate..., Nam [/bib_ref] Though Src is also expressed in the normal tissues of these organs, the expression level is much lower than that in tumors. Src was thus recognized as a good target for these cancers. Src inhibitors or their combination with other agents have exerted efficacy in the treatment of some of these cancers. [bib_ref] Src kinases as therapeutic targets for cancer, Kim [/bib_ref] [bib_ref] Dasatinib: a potent SRC inhibitor in clinical development for the treatment of..., Araujo [/bib_ref] [bib_ref] SRC: a century of science brought to the clinic, Aleshin [/bib_ref] [bib_ref] Src-family tyrosine kinases as therapeutic targets in advanced cancer, Gelman [/bib_ref] Recently, a pathological analysis conducted by Lau and colleagues [bib_ref] Expression of Src and FAK in hepatocellular carcinoma and the effect of..., Lau [/bib_ref] indicated a high level of Src in most HCC cases, but a low level in normal liver and in chronic persistent hepatitis caused by chronic hepatitis C. Further studies have shown that the prognosis of HCC patients with a high level of Src is significantly worse than that of patients with a low level of Src. [bib_ref] Crosstalk between activated and inactivated c-Src in hepatocellular carcinoma, Chen [/bib_ref] Src has thus also been considered an attractive target for HCC. Some small molecule Src inhibitors indeed showed strong anti-HCC effect in vitro. [bib_ref] Expression of Src and FAK in hepatocellular carcinoma and the effect of..., Lau [/bib_ref] [bib_ref] Molecular mechanisms of action and potential biomarkers of growth inhibition of dasatinib..., Chang [/bib_ref] [bib_ref] Molecular subtype and response to dasatinib, an Src/Abl small molecule kinase inhibitor,..., Finn [/bib_ref] SKLB-329 is a good Src inhibitor with an IC 50 value of 18 nM. In cellular assays, SKLB-329 could efficiently inhibit HCC cell proliferation and arrest the cell cycle in the G0-G1 phase. Further studies showed that the anti-proliferation effect of SKLB-329 was mainly due to the blockade of Src/FAK and Src/Stat3 signaling cascades, but not PI3K/PTEN/Akt or Ras/Raf/ MAPK signaling pathway. Again, it is worth mentioning that though SKLB-329 has low nanomolar IC 50 against Src, effective concentrations in inhibition of HCC cell viability and proliferation were still in micromolar range (10-20 lM, [fig_ref] Figure 3: SKLB-329 inhibited HCC cell viability and colony formation in vitro [/fig_ref]. A possible reason could be that the growth of HCC cells is regulated by multiple signaling pathways, and Src kinase is not the only pivotal regulatory molecule driving HCC cell survival and proliferation.
In HCC xenograft models, SKLB-329 exhibited more potent anti-tumor activity than sorafenib. Studies of mechanisms of action showed that SKLB-329 inhibited both tumor angiogenesis and HCC cell proliferation in tumor tissues. Though it is difficult to exactly differentiate how much of the observed in vivo effects of SKLB-329 are due to antiangiogenesis effects and how much due to its tumor cell suppression, roughly speaking, anti-angiogenesis effects should play more important roles on the anti-tumor efficacy; this speculation was based on biochemical and functional potencies of SKLB-329.
In conclusion, SKLB-329 is a novel multikinase inhibitor that potently inhibits VEGFR1/2/3, FGFR2, and Src.
It showed significant activity in inhibiting angiogenesis and considerable potency in suppressing tumor cell proliferation both in vitro and in vivo. SKLB-329 has the convenience of oral administration, favorable pharmacokinetic properties. Taken together, the results of preclinical evaluation of SKLB-329 support the use of SKLB-329 as a promising candidate for clinical studies in patients with HCC.
[fig] Figure 1: The chemical structure of SKLB-329 and its anti-angiogenesis effects in HUVECs assays. (a) The chemical structure of SKLB-329. (b) SKLB-329 inhibited various growth factors induced growth of HUVECs. (c) SKLB-329 inhibited HUVECs migration in wound healing assay. The number of migrated cells was used for statistics and quantified by manual counting. (d) SKLB-329 inhibited HUVECs invasion in transwell invasion assay. The number of invaded cells was used for statistics and quantified by manual counting. (e) SKLB-329 inhibited tube formation of HUVECs. The number of vessel branch points was used for statistics and quantified by manual counting. Nearly 5 lM sorafenib was used as positive control, and the percentage of inhibition was expressed using vehicle treated cells at 100%. Scale bars represent 100 lm. Column, mean; bars, SD (n 5 3; **, p < 0.01 vs. the control; ANOVA). [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] [/fig]
[fig] Figure 2: Anti-angiogenesis effect of SKLB-329 in transgenic zebrafish embryos and its anti-angiogenesis mechanism of action. (a) and (b), SKLB-329 inhibited zebrafish embryonic angiogenesis. Brightfield and fluorescent images were taken after treatment with vehicle, sorafenib or SKLB-329, and the length of intersegmental vessels (ISVs) was used for statistics. Scale bars represent 100 lm (left) and 50 lm (right). Column, mean; bars, SD (n 5 10; **, p < 0.01 vs. the vehicle; ANOVA). (c) SKLB-329 inhibited VEGFR phosphorylation and activity of its downstream signaling proteins in HUVECs. (d) SKLB-329 inhibited the downstream signaling of FGFR in FGF-stimulated HUVECs. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] [/fig]
[fig] Figure 3: SKLB-329 inhibited HCC cell viability and colony formation in vitro. (a) HCC cells and normal human liver cells were treated with SKLB-329, sorafenib, or vehicle for 72 hr and cell viability was determined by MTT assay. Points, mean values; bars, SD. (b) HCC cells and normal human liver cells were seeded in 6-well plates and treated with vehicle or SKLB-329 for 10 days, colonies were stained with crystal violet and photographed. [/fig]
[fig] Figure 4, Figure 5: Anti-proliferation effect of SKLB-329 on HCC cells. (a) Edu cell proliferation assay. The red (EDU-positive) and blue (Hoechst 33358) represent proliferating cells and cell nucleus, respectively. (b) Influence of SKLB-329 on cell cycle progression in HCC cells. (c) Inhibitory effect of SKLB-329 against Src kinase and its downstream signaling molecules in HepG2 and PLC/PRF/5 cell lines. (d) Inhibition of MEK phosphorylation in PLC/PRF/5 cells. The PLC/PRF/5 cells were incubated for 24 hr in medium containing vehicle, SKLB-329, or sorafenib, and then lysed for western blot assay. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.] In vivo pharmacodynamic and pharmacokinetic studies of SKLB-329. (a) and (b), In vivo anti-HCC studies of SKLB-329 against HepG2 and SMMC7721 tumor xenograft models. Tumor volume and body weight were monitored twice weekly. Points, mean tumor volume (mm 3 ) or mean body weight (g); bars, SD. Immunohistochemical staining analysis was used to determine the anti-angiogenesis and antiproliferative effects of SKLB-329 on HepG2 (c), and SMMC7721 models (d). Scale bars represent 100 lm. (e) The plasma concentrationtime curves of SKLB-329 and sorafenib in Sprague-Dawley rats after a single oral dose of 20 mg kg 21 . Blood was collected at indicated time and the plasma concentration was determined by HPLC. Points, mean; bars, SD; n 5 6. [/fig]
[table] Table 1: In vitro kinase inhibition profile of SKLB-329 [/table]
|
10.1038/s41598-019-44095-5
|
CCBY
|
6529442
|
31114004
|
s2orc_pubmed_articles
|
Essential role of prostaglandin E2 and the EP3 receptor in lymphatic vessel development during zebrafish embryogenesis
Lymphatic endothelial cells arise from the venous endothelial cells in embryonic lymphaticdevelopment. However, the molecular mechanisms remain to be elucidated. We here report that prostaglandin (pG) e 2 plays essential roles in the embryonic lymphatic development through the EP3 receptor, one of the PGE 2 receptors. Knockdown of the EP3 receptor or inhibition of cyclooxygenases (COX; rate-limiting enzymes for PG synthesis) impaired lymphatic development by perturbing lymphatic specification during zebrafish development. These impairments by COX inhibition were recovered by treatment with sulprostone (EP1/3 agonist). Knockdown of the EP3 receptor further demonstrated its requirement in the expression of sex determining region Y-box 18 (sox18) and nuclear receptor subfamily 2, group F, member 2 (nr2f2), essential factors of the lymphatic specification. The EP3 receptor was expressed in the posterior cardinal vein (region of embryonic lymphatic development) and the adjacent intermediate cell mass (ICM) during the lymphatic specification. COX1 was expressed in the region more upstream of the posterior cardinal vein relative to the EP3 receptor, and the COX1selective inhibitor impaired the lymphatic specification. On the other hand, two COX2 subtypes did not show distinct sites of expression around the region of expression of the EP3 receptor. Finally, we generated EP3-deficient zebrafish, which also showed defect in lymphatic specification and development. Thus, we demonstrated that COX1-derived PGE 2 -EP3 pathway is required for embryonic lymphatic development by upregulating the expression of key factors for the lymphatic specification.The lymphatic system is a major component of the vertebrate vasculature and plays pivotal roles in the collection of interstitial fluid, absorbance of dietary lipids, and trafficking of immune cells 1 . Development of the lymphatic system begins in the early developmental stages, and lymphatic endothelial cells develop from endothelial cells in the posterior cardinal vein 2,3 . Vascular endothelial growth factor c (vegfc) and its receptor, fms-related tyrosine kinase 4 (flt4; also known as vascular endothelial growth factor receptor 3, vegfr3) are key factors for lymphatic specification 4-8 . Vegfc released by the dorsal aorta induces the lymphatic specification through binding to flt4 expressed in venous endothelial cells. Sex determining region Y-box 18 (sox18), a member of sox transcription factor family, is also essential for the lymphatic specification through the induction of transcriptional factors, such as nuclear receptor subfamily 2, group F, member 2 (nr2f2; also known as COUP transcription factor 2, COUP-TFII) 8-10 . Nr2f2 controls the expression of lymphatic genes, such as lymphatic vessel endothelial hyaluronic receptor 1b (lyve1b; a lymphatic marker) 11-13 . Additionally, apelin (apln), collagen and calcium binding EGF domains 1 (ccbe1), and wingless-type MMTV integration site family, member 5b (wnt5b) (secreted grown in modified Eagle's medium (Sigma-Aldrich) supplemented with 10% fetal bovine serum at 37 °C in a fully humidified CO 2 atmosphere. Wild type or mutant EP3 receptor constructs were transfected into HeLa cells using FuGENE HD (Promega, Madison, WI) according to the manufacturer's instructions. After 24 h, transfected cells were labeled with calcium 5 loading buffer for 1 h, and stimulated with sulprostone. Fluorescence (excitation, 485 nm; emission, 515 nm) was monitored for 90 seconds on FlexStation III (Molecular Devices) and the area under curve (AUC) was evaluated as induced intracellular Ca 2+ mobilization.Statistical analysis.Data are shown as the mean ± SEM. Comparison of two groups was analyzed by the Student's t-test. For comparison of more than two groups with comparable variances, one-way ANOVA was performed, and the Tukey's test was subsequently used to evaluate the pairwise group difference. P-values less than 0.05 were considered to indicate significant differences.
proteins) were reported to promote lymphatic development [bib_ref] Ccbe1 regulates Vegfc-mediated induction of Vegfr3 signaling during embryonic lymphangiogenesis, Le Guen [/bib_ref] [bib_ref] Ccbe1 is required for embryonic lymphangiogenesis and venous sprouting, Hogan [/bib_ref] [bib_ref] Essential role of Apelin signaling during lymphatic development in zebrafish, Kim [/bib_ref] [bib_ref] Lymphatic vessels arise from specialized angioblasts within a venous niche, Nicenboim [/bib_ref]. On the other hand, bone morphogenetic protein 2b (bmp2b) was shown to negatively modulate the development of lymphatic endothelial cells [bib_ref] Bone morphogenetic protein 2 signaling negatively modulates lymphatic development in vertebrate embryos, Dunworth [/bib_ref].
Prostaglandin (PG) E 2 is an arachidonate metabolite that is synthesized via a pathway with cyclooxygenase (COX) as the rate-limiting enzyme. PGE 2 has been shown to exert a variety of actions by binding to four specific G-protein-coupled receptors (EP1, EP2, EP3, and EP4) on the plasma membrane of neighboring cells in humans and mice [bib_ref] Prostanoids and their receptors, Coleman [/bib_ref]. The EP1 receptor is coupled to Gq to increase intracellular Ca 2+ concentrations. The EP2 and EP4 receptors are coupled to Gs and increase intracellular cAMP concentrations via the activation of adenylyl cyclase. The EP3 receptor is coupled to Gi, and mediates the inhibition of adenylyl cyclase and the upregulation of intracellular Ca 2+ concentrations [bib_ref] International Union of Pharmacology classification of prostanoid receptors: properties, distribution, and structure..., Coleman [/bib_ref] [bib_ref] International Union of Basic and Clinical Pharmacology. LXXXIII: classification of prostanoid receptors,..., Woodward [/bib_ref]. The formation of lymphatic vessels is known to occur during not only embryogenesis but also the progression of various cancers [bib_ref] Tumor lymphangiogenesis and new drug development, Dieterich [/bib_ref] , and COX2-derived PGE 2 was found to facilitate lymphangiogenesis during tumor development through the EP3 receptor in mice [bib_ref] Host prostaglandin EP3 receptor signaling relevant to tumor-associated lymphangiogenesis, Kubo [/bib_ref] [bib_ref] Prostanoid induces premetastatic niche in regional lymph nodes, Ogawa [/bib_ref]. In addition, PGE 2 was shown to accelerate the formation of lymphatic vessels through the EP3 receptor in granulation tissues [bib_ref] Roles of prostaglandin E2-EP3/EP4 receptor signaling in the enhancement of lymphangiogenesis during..., Hosono [/bib_ref] [bib_ref] Signaling of prostaglandin E receptors, EP3 and EP4 facilitates wound healing and..., Hosono [/bib_ref]. These data suggest the possibility that the PGE 2 -EP3 pathway plays important roles in the formation of lymphatic vessels. However, it remains unknown whether the PGE 2 -EP3 pathway is involved in the development of lymphatic vessels during embryogenesis.
To identify the role of the PGE 2 -EP3 pathway in lymphatic vessel development during embryogenesis, we used zebrafish as a model organism, because zebrafish embryos are optically clear and undergo rapid early development outside the maternal body [bib_ref] Live imaging of lymphatic development in the zebrafish, Yaniv [/bib_ref]. In addition, zebrafish share many similarities in their molecular mechanisms of lymphatic vessel formation with other vertebrates, and express three COX subtypes (COX1, COX2a, and COX2b) and eight PGE 2 receptor subtypes (EP1a, EP1b, EP2a, EP2b, EP3, EP4a, EP4b, and EP4c) [bib_ref] Developmental expression of functional cyclooxygenases in zebrafish, Grosser [/bib_ref] [bib_ref] The zebrafish genome contains two inducible, functional cyclooxygenase-2 genes, Ishikawa [/bib_ref] [bib_ref] Molecular and pharmacological characterization of zebrafish 'contractile' and 'inhibitory' prostanoid receptors, Iwasaki [/bib_ref] [bib_ref] Molecular and pharmacological characterization of zebrafish 'relaxant' prostanoid receptors, Tsuge [/bib_ref]. Here, we report novel functions of the PGE 2 -EP3 pathway in the formation of lymphatic vessels during early development.
# Results
## The pge 2 -ep3 pathway is involved in lymphatic vessel formation during early development.
To investigate whether the PGE 2 -EP3 pathway regulates lymphatic vessel formation in embryogenesis, we analyzed the effects of EP3 receptor knockdown on zebrafish lymphatic development using Tg(fli1a:egfp) embryos, in which both blood and lymphatic vessels are labeled [bib_ref] Live imaging of lymphatic development in the zebrafish, Yaniv [/bib_ref] [bib_ref] In vivo imaging of embryonic vascular development using transgenic zebrafish, Lawson [/bib_ref]. We used two splice-blocking morpholino antisense oligos (MOs), EP3 MO1 and MO2 [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref] , . Quantitative analysis showed that injection of EP3 MO1 or MO2 both markedly decreased the mature mRNA expression level of the EP3 receptor at 24 hours post fertilization (hpf) [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. To evaluate lymphatic vessel formation, we analyzed the parachordal lymphangioblast (PL) at the horizontal myoseptum, which is commonly used to study lymphatic development in zebrafish [bib_ref] Sox18 genetically interacts with VegfC to regulate lymphangiogenesis in zebrafish, Cermenati [/bib_ref] [bib_ref] Live imaging of lymphatic development in the zebrafish, Yaniv [/bib_ref]. In control (Cont) MO-injected embryos, the PL was fully formed in most of the segments at 52 hpf [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. On the other hand, PL formation was severely impaired in morphants injected with EP3 MO1 or MO2 [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref] , and the ratio of PL-positive segments was also significantly reduced in these morphants [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. We then investigated whether indomethacin, a non-selective COX inhibitor 28 , induces lymphatic vessel defects. In vehicle-treated embryos, PL was normally formed in the horizontal myoseptum at 52 hpf. On the other hand, treatment with indomethacin inhibited PL formation and significantly reduced the ratio of PL-positive segments [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. This effect of indomethacin was substantially recovered by cotreatment with sulprostone, an EP1/3 agonist 30 [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. Furthermore, we investigated the formation of thoracic duct (TD), which is located immediately ventral to the dorsal aorta in the trunk, at 5 days post fertilization (dpf) [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. In Cont MO-injected embryos, the TD was fully formed in most of the segments. On the other hand, TD formation was severely impaired in morphants injected with EP3 MO1 and MO2 [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref] , and the ratio of TD-positive segments was significantly reduced in the EP3 MO1-injected morphants [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. These results indicated that the PGE 2 -EP3 pathway plays an important role in lymphatic vessel formation during early development. the pGe 2 -EP3 pathway is required for lymphatic specification from venous to lymphatic endothelial cells. Lymphatic endothelial cells are generated from pre-existing endothelial cells of the posterior cardinal vein from approximately 24 to 36 hpf, and subsequently sprout from the posterior cardinal vein after 36 hpf and migrate to colonize embryonic tissues. To identify the lymphatic development process in which the PGE 2 -EP3 pathway is involved, we performed time-dependent inhibition of PG synthesis by treatment with indomethacin for a limited time [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. There was no significant difference in the ratios of PL-positive segments between vehicle treatment and indomethacin treatment from 12 to 24 hpf, when the posterior cardinal vein develops. On the other hand, indomethacin treatment from 24 to 36 hpf significantly reduced the ratio of PL-positive segments. The degree of reduction by the treatment from 24 to 36 hpf was equivalent to that by the treatment from 24 to 60 hpf. These data suggest the importance of the PGE 2 -EP3 pathway in the lymphatic specification process. Therefore, to determine whether the PGE 2 -EP3 pathway acts during the first steps of lymphatic development, we analyzed the expression levels of lyve1b, which is a lymphatic marker [bib_ref] Visualization of embryonic lymphangiogenesis advances the use of the zebrafish model for..., Flores [/bib_ref]. Compared with control morphants, EP3 receptor morphants had significantly lower expression levels of lyve1b at both 24 and 36 hpf [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. Whole-mount in situ hybridization (WISH) analysis demonstrated that lyve1b was expressed around the posterior cardinal vein in Cont MO-injected embryos at both 24 and 36 hpf [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. By contrast, embryos injected with EP3 MO1 showed substantial decreases in lyve1b-derived signals around the posterior cardinal vein at both time points [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. Embryos treated with indomethacin also showed decreased expression levels of lyve1b at 24 hpf [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. This effect of indomethacin was significantly recovered by cotreatment with sulprostone but not ONO-AE1-259 (an EP2 agonist) [bib_ref] Molecular and pharmacological characterization of zebrafish 'relaxant' prostanoid receptors, Tsuge [/bib_ref] or ONO-AE1-329 (an EP4 agonist) [bib_ref] Molecular and pharmacological characterization of zebrafish 'relaxant' prostanoid receptors, Tsuge [/bib_ref] [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. WISH analysis of embryos at 24 hpf demonstrated that indomethacin markedly reduced lyve1b-derived signals around the posterior cardinal vein, where lyve1b-derived signals were observed in vehicle-treated embryos [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. Lyve1b-derived signals that were decreased by indomethacin were recovered to the levels similar to that of vehicle-treated embryos by the cotreatment of sulprostone [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. These results indicated that the PGE 2 -EP3 pathway contributes to the formation of lymphatic vessels by regulating the lymphatic specification, which is the first step of lymphatic development during embryogenesis.
www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ Expression levels of genes involved in the lymphatic specification. To analyze the function of the PGE 2 -EP3 pathway in the lymphatic specification, we investigated the mRNA expression level of various genes (vegfc, flt4, sox18, nr2f2, apln, ccbe1, wnt5b, and bmp2b) crucially involved in the lymphatic specification 4-10,14-17 . By reverse transcription-quantitative PCR (RT-qPCR), we analyzed the expression levels of these genes in EP3 receptor morphants at 24 and 36 hpf [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref] , which are early and late phases of lymphatic specification, respectively. At both 24 and 36 hpf, expression levels of sox18 were significantly decreased in EP3 receptor morphants compared with control morphants [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. Expression levels of nr2f2 at 36 hpf were also significantly decreased in EP3 receptor morphants, although expression levels of nr2f2 at 24 hpf did not change [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. There was no significant difference in the expression levels of vegfc, flt4 (also known as a vein marker at 24 hpf), apln, ccbe1, wnt5b, and bmp2b [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref] ,B,E-H). Because nr2f2 is expressed in not only the posterior cardinal vein but also the cranial and spinal cord 10 , we then examined the expression of nr2f2 around the posterior cardinal vein at 24 and 36 hpf by WISH analysis [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. Signals of nr2f2 were detected in the posterior cardinal vein and spinal cord in the trunk of Cont MO-injected embryos. In the trunk of EP3 MO1-injected embryos, nr2f2-derived signals were decreased only in the posterior cardinal vein but not the spinal cord, specifically in 36 hpf but not 24 hpf. These data indicated that the EP3 receptor plays important roles in the expression of sox18 and nr2f2 in the lymphatic specification. www.nature.com/scientificreports www.nature.com/scientificreports/ COX1-derived PGE 2 regulates the lymphatic specification through the EP3 receptor. We performed WISH analysis to identify the sites of EP3 receptor expression during the lymphatic specification. As previously reported 33 , EP3 receptor-derived signals were detected around blood vessels of the trunk at 24 hpf [fig_ref] Figure 4: EP3 receptor mRNA is expressed in the posterior cardinal vein and the... [/fig_ref]. To analyze these expression sites in more detail, we compared the signals of the EP3 receptor with those of several tissue markers at 24 hpf, namely, hes-related family bHLH transcription factor with YRPW motif 2 (hey2; an aorta marker) [bib_ref] Gridlock, an HLH gene required for assembly of the aorta in zebrafish, Zhong [/bib_ref] , flt4 (a vein marker) [bib_ref] Notch signaling is required for arterial-venous differentiation during embryonic vascular development, Lawson [/bib_ref] and GATA binding protein 1a (gata1a; an erythrocyte progenitor marker) [bib_ref] Characterization of expanded intermediate cell mass in zebrafish chordin morphant embryos, Leung [/bib_ref] , which is reported to be exclusively expressed in the intermediate cell mass (ICM) between the dorsal aorta and the posterior cardinal vein at this developmental stage [bib_ref] Characterization of expanded intermediate cell mass in zebrafish chordin morphant embryos, Leung [/bib_ref]. The signals of the EP3 receptor were located ventral to those of hey2 and around those of flt4 and gata1a [fig_ref] Figure 4: EP3 receptor mRNA is expressed in the posterior cardinal vein and the... [/fig_ref]. Furthermore, we made transverse sections of these embryos that were subjected to WISH analysis [fig_ref] Figure 4: EP3 receptor mRNA is expressed in the posterior cardinal vein and the... [/fig_ref]. As previously reported 34 , hey2-derived signals were detected under the notochord in transverse sections. Signals of flt4 were located in a more ventral region than signals of hey2, and signals of gata1a were observed in a region between those of hey2 and flt4. EP3 receptor-derived signals appeared to be located in the region with both flt4-derived and gata1a-derived signals. These results indicated that the EP3 receptor is expressed in the ICM and the posterior cardinal vein, where lymphatic endothelial cells are generated.
COX are rate-limiting enzymes for the biosynthesis of PGE 2 . As zebrafish have three subtypes of COX (COX1, COX2a, and COX2b), we tried to identify the specific COX subtypes involved in the lymphatic specification. We first performed WISH analysis at 24 hpf to identify the region of expression of each COX subtype during the lymphatic specification. COX1-derived signals were detected in the region more upstream of the posterior cardinal vein relative to EP3 receptor-derived signals [fig_ref] Figure 5: COX1-derived PGE 2 is involved in the lymphatic specification [/fig_ref]. On the other hand, COX2a-and COX2b-derived distinctive signals were not detected in the trunk of zebrafish embryos [fig_ref] Figure 5: COX1-derived PGE 2 is involved in the lymphatic specification [/fig_ref]. To clarify whether COX1 is involved in the lymphatic specification, we treated zebrafish embryos with a COX1-selective inhibitor, SC-560 [bib_ref] Prostaglandin E2 regulates vertebrate haematopoietic stem cell homeostasis, North [/bib_ref] and investigated the expression levels of lyve1b at 24 hpf using WISH analysis. SC-560 diminished lyve1b-derived signals around the posterior cardinal vein [fig_ref] Figure 5: COX1-derived PGE 2 is involved in the lymphatic specification [/fig_ref]. The decrease in lyve1b-derived signals was recovered by cotreatment with sulprostone [fig_ref] Figure 5: COX1-derived PGE 2 is involved in the lymphatic specification [/fig_ref]. These results suggested that COX1-derived PGE 2 plays important roles in the lymphatic specification from venous to lymphatic endothelial cells during early development.
## Impairment of lymphatic specification and development by ep3 receptor deficiency.
To confirm the role of the EP3 receptor in the lymphatic specification, we finally generated EP3 receptor-deficient (EP3 −/− ) zebrafish using transcription activator-like effector nucleases (TALEN) [bib_ref] Targeting DNA double-strand breaks with TAL effector nucleases, Christian [/bib_ref]. Sequencing analysis showed that a stop codon was introduced into the transmembrane II region of the wild-type EP3 receptor by the deletion of seven base-pairs in transmembrane I [fig_ref] Figure 6: Effect of EP3 receptor deficiency on lymphatic specification and development [/fig_ref]. EP3 +/− and EP3 −/− zebrafish were born at approximately expected Mendelian ratio and were viable [fig_ref] Figure 6: Effect of EP3 receptor deficiency on lymphatic specification and development [/fig_ref]. We then performed the Ca 2+ mobilization assay to confirm whether this deletion actually results in a loss of the function of the EP3 receptor. Although HeLa cells transfected with a construct encoding the wild-type EP3 receptor dose-dependently induced Ca 2+ mobilization by sulprostone, HeLa cells transfected with a construct encoding the mutant EP3 receptor did not respond to sulprostone [fig_ref] Figure 6: Effect of EP3 receptor deficiency on lymphatic specification and development [/fig_ref]. Subsequently, we investigated the expression levels of lyve1b in EP3 +/− and EP3 −/− zebrafish to evaluate the effects of EP3 receptor deficiency on the lymphatic specification. Compared with EP3 +/+ zebrafish, EP3 −/− zebrafish but not EP3 +/− zebrafish had significantly lower expression levels of lyve1b at 24 hpf [fig_ref] Figure 6: Effect of EP3 receptor deficiency on lymphatic specification and development [/fig_ref]. Additionally, WISH analysis showed that the expression levels of lyve1b around the posterior cardinal vein were remarkably decreased in EP3 −/− zebrafish but not EP3 +/− zebrafish at 36 hpf [fig_ref] Figure 6: Effect of EP3 receptor deficiency on lymphatic specification and development [/fig_ref]. Finally, we investigated the expression of lyve1b at 52 hpf using WISH analysis to analyze PL formation in EP3 −/− zebrafish [fig_ref] Figure 6: Effect of EP3 receptor deficiency on lymphatic specification and development [/fig_ref]. In EP3 +/+ zebrafish, lyve1b signals were detected in the horizontal myoseptum, indicating adequate PL formation. On the other hand, signals of lyve1b were not detected in EP3 −/− zebrafish, indicating failure of PL formation. These results indicated the importance of EP3 receptor expression in lymphatic specification and development. Thus, the defects of lymphatic specification and development observed upon EP3 receptor knockdown and inhibition of PG synthesis were also observed in EP3 receptor gene-deficient mutants. www.nature.com/scientificreports www.nature.com/scientificreports/
# Discussion
In this study, we showed for the first time that the PGE 2 -EP3 pathway plays an essential role in embryonic lymphatic development, by regulating the lymphatic specification, which is the first step of embryonic lymphatic development [fig_ref] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation [/fig_ref]. We found that the EP3 receptor is important for the expression of sox18 and nr2f2 but not for vegfc, flt4, apln, ccbe1, wnt5b, and bmp2b [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. The EP3 receptor was expressed in the posterior cardinal vein and the neighboring ICM [fig_ref] Figure 4: EP3 receptor mRNA is expressed in the posterior cardinal vein and the... [/fig_ref]. On the other hand, COX1, but not two COX2 subtypes, was expressed in the region more upstream of the posterior cardinal vein relative to the region of expression of the EP3 receptor, and contributed to lymphatic specification [fig_ref] Figure 5: COX1-derived PGE 2 is involved in the lymphatic specification [/fig_ref]. Lymphatic vessels develop from endothelial cells in the posterior cardinal vein. Therefore, normal development of the posterior cardinal vein is important for lymphatic vessel development 2,3 . Our data using RT-qPCR and WISH analyses indicated that the expression levels of flt4, used as a maturation marker of venous endothelial cells, were not altered in EP3 receptor morphants at 24 hpf, when the posterior cardinal vein is fully developed [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. Our observation in Tg(fli1a:egfp) zebrafish also showed that the posterior cardinal vein of EP3 receptor morphants was similar to that of control morphants at both 24 and 36 hpf (data not shown). Additionally, lymphatic vessel development was not impaired by the inhibition of PG synthesis from 12 to 24 hpf, when the posterior cardinal vein develops [fig_ref] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from... [/fig_ref]. These data indicated that the EP3 receptor does not affect the development of PCV, at least on evaluation of lymphatic development. However, further detailed studies are needed to precisely determine whether the PGE 2 -EP3 pathway is involved in the differentiation and/or maturation of the posterior cardinal vein, at a level at which there is no effect on lymphatic development.
We found that the EP3 receptor plays important roles in the expression levels of sox18 and nr2f2 [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. These genes are transcriptional factors that facilitate the lymphatic specification [bib_ref] Transcription factor COUP-TFII is indispensable for venous and lymphatic development in zebrafish..., Aranguren [/bib_ref] [bib_ref] Sox18 genetically interacts with VegfC to regulate lymphangiogenesis in zebrafish, Cermenati [/bib_ref] [bib_ref] SoxF factors and Notch regulate nr2f2 gene expression during venous differentiation in..., Swift [/bib_ref]. It was reported that sox18 genetically interacts with vegfc in the early phase of lymphatic development 9 , and that sox18 is required for the expression of nr2f2 in zebrafish at 24 hpf 11 . Therefore, most of the effects of the EP3 receptor on the lymphatic specification might be exerted by the regulation of sox18 expression. These two transcriptional factors are expressed in the posterior cardinal vein [bib_ref] SoxF factors and Notch regulate nr2f2 gene expression during venous differentiation in..., Swift [/bib_ref] [bib_ref] Zebrafish Sox7 and Sox18 function together to control arterial-venous identity, Pendeville [/bib_ref] , where embryonic lymphatic development begins. Interestingly, EP3 receptors were found to be expressed in the posterior cardinal vein and the neighboring ICM [fig_ref] Figure 4: EP3 receptor mRNA is expressed in the posterior cardinal vein and the... [/fig_ref]. These results suggested that the functions of the EP3 receptor in lymphatic specification were exerted directly (in the posterior cardinal vein) or indirectly (by certain secreted or plasma membrane-associated factors supplied from the adjacent ICM through a paracrine or juxtacrine route). Then, we investigated in vitro whether the EP3 receptor expressed in venous endothelial cells accelerates differentiation toward lymphatic endothelial cells as a direct consequence of endothelial cell-autonomous activation of the EP3 receptor. However, stimulation by the selective human EP3 receptor agonist ONO-AE-248 did not upregulate the expression levels of the lymphatic marker LYVE1 in human umbilical vein endothelial cells (HUVECs), even when the human EP3 receptor was overexpressed in HUVECs (data not shown). On the other hand, expression levels of secreted regulatory factors such as apln, ccbe1, wnt5b, and bmp2b were not affected by knockdown of the EP3 receptor at both 24 and 36 hpf [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. Additionally, there have been no reports to our knowledge regarding plasma membrane-localized molecules that are involved in the lymphatic specification. Further studies are required to fully understand the molecular mechanism of the lymphatic specification promoted by the PGE 2 -EP3 pathway.
In this study, we found that COX1-derived PGE 2 accelerated the lymphatic specification during embryonic lymphatic development [fig_ref] Figure 5: COX1-derived PGE 2 is involved in the lymphatic specification [/fig_ref] , and the EP3 receptor had no effect on the expression levels of vegfc and flt4 [fig_ref] Figure 3: Expression of genes involved in lymphatic specification [/fig_ref]. In contrast to our study, COX2-derived PGE 2 was reported to accelerate lymphangiogenesis through the EP3 receptor in tumor implantation and granulation formation models [bib_ref] Host prostaglandin EP3 receptor signaling relevant to tumor-associated lymphangiogenesis, Kubo [/bib_ref] [bib_ref] Roles of prostaglandin E2-EP3/EP4 receptor signaling in the enhancement of lymphangiogenesis during..., Hosono [/bib_ref] [bib_ref] Signaling of prostaglandin E receptors, EP3 and EP4 facilitates wound healing and..., Hosono [/bib_ref]. Mice and human cells were used in these models and PGE 2 upregulated the expression of Vegfc and Flt4 through the EP3 receptor. Although the www.nature.com/scientificreports www.nature.com/scientificreports/ www.nature.com/scientificreports www.nature.com/scientificreports/ reason for these differences is unclear, a likely explanation is that these discrepancies may be owing to differences of venous endothelial cell types among species and tissues, and differences between embryos and adults. Additionally, the different results might be explained by the absence or presence of inflammation. In tumor implantation and granulation formation models, inflammation was induced and immune cells invaded the tissues. It has been reported that migrated macrophages produce Vegfc through the PGE 2 -EP3 pathway in granulation formation models [bib_ref] Roles of prostaglandin E2-EP3/EP4 receptor signaling in the enhancement of lymphangiogenesis during..., Hosono [/bib_ref] [bib_ref] Signaling of prostaglandin E receptors, EP3 and EP4 facilitates wound healing and..., Hosono [/bib_ref]. On the other hand, there have been no reports stating that immune cells are involved in embryonic lymphatic development, particularly in lymphatic specification. Sox18 was reported to be critical for tumor-induced lymphangiogenesis [bib_ref] Genetic ablation of SOX18 function suppresses tumor lymphangiogenesis and metastasis of melanoma..., Duong [/bib_ref]. It is therefore possible that the PGE 2 -EP3 pathway might promote lymphangiogenesis in tumors and granulation tissues by the upregulation of not only Vegfc and Flt4 expression but also Sox18 expression.
In summary, we found that the PGE 2 -EP3 pathway plays crucial roles in the lymphatic specification from venous to lymphatic endothelial cells through the upregulation of sox18 and nr2f2. These data strongly suggest a novel function of COX1-PGE 2 -EP3 pathway in the formation of the lymphatic system during early development.
# Materials and methods
Materials. The following materials were obtained from the sources indicated; PGE 2 , sulprostone, and SC-560 Zebrafish line and maintenance. A wild-type zebrafish strain was obtained from National BioResource Project Zebrafish (RIKEN, Japan). The transgenic zebrafish line Tg(fli1a:egfp) has been described previously [bib_ref] Live imaging of lymphatic development in the zebrafish, Yaniv [/bib_ref] [bib_ref] In vivo imaging of embryonic vascular development using transgenic zebrafish, Lawson [/bib_ref] , and was used to monitor lymphatic development. Zebrafish were maintained at 28.5 °C under a 14 h-light/10 h-dark cycle. Embryos were maintained in 1/3 ringer solution buffer at 28.5 °C. All experimental protocols were approved by Kumamoto University . All experiments with zebrafish were performed in accordance with the guidelines of Kumamoto University.
Morpholino antisense oligos. EP3 MO1 (targeting the splice site between exon 1 and intron 1), EP3 MO2 (targeting the splice site between intron 1 and exon 2), and Cont MO were purchased from Gene Tools, LLC (Philomath, OR). Each MO (10 ng) was injected into the yolk of 1-2 cell stage embryos. The sequence of each MO is shown in RNA extraction and RT-qPCR. Total RNA was extracted from zebrafish embryos at the indicated stages using Sepasol RNA I Super G (Nacalai Tesque, Kyoto, Japan), and was subjected to RT with PrimeScript RT Master Mix (Takara Bio, Shiga, Japan). Synthesized cDNA was subjected to qPCR using a LightCycler (Roche Applied Science, Penzberg, Germany) and Fast Start DNA Master SYBR Green I according to the manufacturer's instructions. Crossing point values were acquired by the second derivative maximum method. The expression level of each gene was quantified using external standardized dilutions. Relative expression levels among samples were normalized by the value of gapdh. Sequences of the used primers are shown in . The specificity of qPCR was confirmed by the lengths and melting temperatures of the amplified products.
Whole-mount in situ hybridization. Total RNA was isolated from zebrafish embryos, and cDNA was synthesized using SuperScript III (Invitrogen, San Diego, CA) and oligo (dT) primers. The coding sequence of each gene (lyve1b, nr2f2, hey2, and flt4,) were amplified from the cDNA by PCR and cloned into the pTA2 vector (Toyobo, Osaka, Japan). Primer sequences used in the PCR are shown in . Cloning of the coding sequences of gata1a, the EP3 receptor, COX1, COX2a, and COX2b was performed as previously described [bib_ref] International Union of Pharmacology classification of prostanoid receptors: properties, distribution, and structure..., Coleman [/bib_ref] [bib_ref] Characterization of the heme synthesis enzyme coproporphyrinogen oxidase (CPO) in zebrafish erythrogenesis, Hanaoka [/bib_ref]. These plasmids were linearized by restriction enzymes. Digoxigenin (DIG)-labeled anti-sense RNA was transcribed from each linearized vector by in vitro transcription using DIG mix and transcription buffer (Roche Diagnostics). These RNA was purified by ethanol precipitation, dissolved in hybridization buffer (50% formamide, 5× SSC, 5 mM ethylenediaminetetraacetic acid, 0.1% Tween-20, and 1 mg/mL torula RNA), and used as hybridization probes. WISH was performed as previously described [bib_ref] Characterization of the heme synthesis enzyme coproporphyrinogen oxidase (CPO) in zebrafish erythrogenesis, Hanaoka [/bib_ref]. After fixation by 4% paraformaldehyde, zebrafish embryos at 52 hpf were incubated in 3% hydrogen peroxide solution to remove dark pigments. Embryos stained with WISH were embedded in OCT Compound (Sakura, Tokyo, Japan) and transverse sections (10-20 μm) were prepared using a cryostat (Leica Microsystems, Wetzlar, Germany). Images were taken with a fluorescence microscope (BZ-X700; KEYENCE, Osaka, Japan) and were processed using the attached software or Adobe Photoshop (Adobe, San Jose, CA).
Generation of EP3 receptor-mutant zebrafish. TALEN plasmids were constructed using a two-step assembly system, as described previously [bib_ref] Efficient design and assembly of custom TALEN and other TAL effector-based constructs..., Cermak [/bib_ref]. Briefly, six or fewer TAL effector repeat domains were ligated into the pFUS vector [bib_ref] Efficient TALEN construction and evaluation methods for human cell and animal applications, Sakuma [/bib_ref]. Subsequently, each pFUS vector and last TAL effector repeat were ligated into the pCS2 vector as a TALEN plasmid [bib_ref] Simple methods for generating and detecting locus-specific mutations induced with TALENs in..., Dahlem [/bib_ref]. The TALEN plasmids were linearized by Not I digestion, and TALEN mRNA was transcribed using the mMESSAGE mMACHINE SP6 kit (Life Technologies, Gaithersburg, MD) and purified using the RNeasy Mini Kit (QIAGEN, Hilden, Germany). Forward and reverse TALEN mRNA (400 pg each) was simultaneously injected into zebrafish blastomeres at the one-cell stage. To detect genome modification in zebrafish, we utilized the heteroduplex mobility assay as reported previously [bib_ref] Efficient identification of TALEN-mediated genome modifications using heteroduplex mobility assays, Ota [/bib_ref]. The sequences of the primers used in the heteroduplex mobility assay are shown in . To check the functional ability of the TALEN-induced EP3 receptor mutant, the Ca 2+ mobilization assay was performed using the FLIPR Calcium 5 Assay Kit (Molecular Devices, Sunnyvale, CA) according to the manufacturer's instructions. HeLa cells were
[fig] Figure 1: Role of the PGE 2 -EP3 pathway in lymphatic vessel formation. (A) Tg(fli1a:egfp) embryos were injected with Cont MO, EP3 MO1, or MO2. Images of the trunks were taken at 52 hpf. (B) The ratio of PLpositive segments in nine consecutive segments of (A) was quantitated. Each value represents the mean ± SEM (N = 3-4). (C) Tg(fli1a:egfp) embryos were treated with vehicle (Veh) or indomethacin (Indo; 100 μM) in the absence or presence of sulprostone (Sulp; 1 μM) from 0 to 52 hpf. Images were taken at 52 hpf. (D) The ratio of PL-positive segments in nine consecutive segments of (C) was quantitated. Each value represents the mean ± SEM (N = 3). (E) Tg(fli1a:egfp) embryos were injected with Cont MO, EP3 MO1, or EP3 MO2. Images of the trunks were taken at 5 dpf. (F) The ratio of TD-positive segments in eight consecutive segments of Cont MO-or EP3 MO1-injected morphants was quantitated. Each value represents the mean ± SEM (N = 5-9). **P < 0.01 vs Cont MO. DA: dorsal aorta; ISV: intersomitic vessel; PL: parachordal lymphangioblast; TD: thoracic duct. The PL is indicated by arrowheads, and the TD is indicated by arrows. The number at the bottom right of each panel indicates the number of embryos demonstrating the phenotype shown in the panel over the total number of embryos analyzed in a representative experiment. [/fig]
[fig] Figure 2: Role of the PGE 2 -EP3 pathway in the lymphatic specification from venous to lymphatic endothelial cells. (A) Tg(fli1a:egfp) embryos were treated with Veh or Indo (100 μM) for the indicated times. The ratio of PL-positive segments at 60 hpf was quantified. Each value represents the mean ± SEM (N = 5). **P < 0.01 vs Veh. (B) Relative expression levels of lyve1b were quantified by RT-qPCR in morphants at 24 and 36 hpf. Values are shown relative to the value obtained with Cont MO at 24 hpf. Each value represents the mean ± SEM (N = 3-4) *P < 0.05, **P < 0.01 vs Cont MO at each corresponding time. (C,D) Expression of lyve1b was analyzed by WISH in morphants at 24 hpf (C) and 36 hpf (D). (E,F) Zebrafish embryos were treated with Veh or Indo (100 μM) in the absence or presence of EP agonists (10 μM) from 0 to 24 hpf. The expression level of lyve1b was quantified by RT-qPCR (E). The values are shown relative to the value obtained with Veh. Each value represents the mean ± SEM (N = 3-4). Expression of lyve1b was analyzed by WISH at 24 hpf (F). The number at the bottom right of each panel indicates the number of embryos demonstrating the phenotype shown in the panel over the total number of embryos analyzed in a representative experiment. [/fig]
[fig] Figure 3: Expression of genes involved in lymphatic specification. (A-H) Relative expression levels (at 24 and 36 hpf) of genes involved in lymphatic specification were quantified by RT-qPCR in morphants injected with Cont MO or EP3 MO1. Values are shown relative to the value obtained with Cont MO at 24 hpf. Each value represents the mean ± SEM (N = 3). *P < 0.05, **P < 0.01 vs Cont MO at each corresponding time. (I,J) Expression of nr2f2 was analyzed by WISH in morphants at 24 and 36 hpf. The number at the bottom right of each panel indicates the number of embryos demonstrating the phenotype shown in the panel over the total number of embryos analyzed in a representative experiment. [/fig]
[fig] Figure 4: EP3 receptor mRNA is expressed in the posterior cardinal vein and the neighboring ICM. (A,B) Regions of expression of marker genes and the EP3 receptor were analyzed by WISH at 24 hpf (A) and subsequently by microscopic examination of their cross-sections (B). Arrowheads indicate the regions where the signals derived from each gene were observed. [/fig]
[fig] Figure 5: COX1-derived PGE 2 is involved in the lymphatic specification. (A) Expression of COX1, COX2a, and COX2b was analyzed by WISH at 24 hpf. (B) Zebrafish embryos were treated with Veh or SC-560 (25 μM) in the absence or presence of Sulp (10 μM) from 0 to 24 hpf, and the expression of lyve1b was analyzed by WISH at 24 hpf. The number at the bottom right of each panel indicates the number of embryos demonstrating the phenotype shown in the panel over the total number of embryos analyzed in a representative experiment. [/fig]
[fig] Figure 6: Effect of EP3 receptor deficiency on lymphatic specification and development. (A) Sequence alignment of the wild-type and mutant EP3 receptor. The seven base-pair deletion in mutant EP3 receptor is indicated by a dotted line. Transmembrane (TM) regions of the wild-type EP3 receptor are indicated by gray boxes. (B) Number of live births of EP3 +/+ , EP3 +/− , and EP3 −/− zebrafish. (C) HeLa cells were transfected with an expression vector encoding either the wild-type or mutant EP3 receptor. After 24 h, cells were incubated with loading buffer for 1 h and then treated with Veh or Sulp. Induced intracellular Ca 2+ mobilization was evaluated by AUC analysis. *P < 0.05, **P < 0.01 vs Veh. Each value represents the mean ± SEM (N = 3). (D) Relative expression levels of lyve1b were quantified by RT-qPCR in EP3 +/+ , EP3 +/− , and EP3 −/− zebrafishs at 24 hpf. Values are shown relative to the value obtained with EP3 +/+ . Each value represents the mean ± SEM (N = 8-12) *P < 0.05 vs EP3 +/+ . (E) Expression of lyve1b was analyzed by WISH in EP3 +/+ , EP3 +/− , and EP3 −/− zebrafish at 36 hpf. (F) Expression of lyve1b was analyzed by WISH in EP3 +/+ and EP3 −/− zebrafish at 52 hpf. The PL is indicated by an arrowhead. The number at the bottom right of each panel indicates the number of embryos demonstrating the phenotype shown in the panel over the total number of embryos analyzed in a representative experiment. [/fig]
|
10.1038/s41598-018-25192-3
|
CCBY
|
5934370
|
29725125
|
s2orc_pubmed_articles
|
Development of a generic β-lactamase screening system for improved signal peptides for periplasmic targeting of recombinant proteins in Escherichia coli
Targeting of recombinant proteins to the Escherichia coli periplasm is a desirable industrial processing tool to allow formation of disulphide bonds, aid folding and simplify recovery. Proteins are targeted across the inner membrane to the periplasm by an N-terminal signal peptide. The sequence of the signal peptide determines its functionality, but there is no method to predict signal peptide function for specific recombinant proteins, so multiple signal peptides must be screened for their ability to translocate each recombinant protein, limiting throughput. We present a screening system for optimising signal peptides for translocation of a single chain variable (scFv) antibody fragment employing TEM1 β-lactamase (Bla) as a C-terminal reporter of periplasmic localisation. The Pectobacterium carotovorum PelB signal peptide was selected as the starting point for a mutagenic screen. β-lactamase was fused to the C-terminal of scFv and β-lactamase activity was correlated against scFv translocation. Signal peptide libraries were generated and screened for β-lactamase activity, which correlated well to scFv::Bla production, although only some high activity clones had improved periplasmic translocation of scFv::Bla. Selected signal peptides were investigated in fed-batch fermentations for production and translocation of scFv::Bla and scFv without the Bla fusion. Improved signal peptides increased periplasmic scFv activity by ~40%.The Gram negative bacterium Escherichia coli is a mainstay of the biopharmaceutical industry, and is the most common non-mammalian cell production system for recombinant protein biopharmaceuticals 1 . Advantages of E. coli include a long history of safe use, high biomass and protein yields, and ease of genetic engineering 2 . E. coli is used for the production of relatively simple recombinant protein biopharmaceuticals such as insulin, Growth Hormone and Granulocyte-Colony Stimulating Factor 1 . Whereas E. coli lacks the ability to make many post-translational modifications (such as glycosylation) that are required for some recombinant protein biopharmaceuticals, which necessitate the use of eukaryotic hosts such as CHO cells, it is able to generate disulphide bonds between cysteine residues. Disulphide bonding in E. coli naturally occurs in the periplasm, catalysed by the Dsb proteins (reviewed by 3 ). Therefore, recombinant proteins must be transported into the periplasm in order for disulphide bonding to occur. This is of particular relevance to antibody fragments which often require disulphide bonding for correct folding and function 4 . An example is the human biopharmaceutical Certolizumab pegol (Cimzia ® ), which is a PEGylated anti-Tumour Necrosis Factor α antigen-binding (Fab) antibody fragment
produced in E. coli. In addition to disulphide bonding, periplasmic targeting can decrease proteolysis of recombinant proteins by cytoplasmic proteases, and allows simplified product release and capture without the need for whole cell lysis [bib_ref] Advances in product release strategies and impact on bioprocess design, Balasundaram [/bib_ref].
E. coli exploits multiple mechanisms for transport of proteins into the periplasm that include the SecB, SRP and twin-arginine (Tat) pathways (reviewed by [bib_ref] The twin-arginine translocation (Tat) protein export pathway, Palmer [/bib_ref] [bib_ref] Protein export through the bacterial Sec pathway, Tsirigotaki [/bib_ref]. The SecB and SRP pathways both employ a common transport mechanism. The SecYEG complex comprises a pore in the inner membrane, which transports unfolded polypeptide chains from the cytoplasm to the periplasm. The SecB pathway is post-translational, whereby polypeptide chains are translocated after complete translation, whereas the SRP pathway is co-translational, as translocation occurs while the polypeptide chain is still being translated by the ribosome. The third mechanism, Tat, consists of a larger pore made up of the TatABC proteins, which is able to transport fully folded proteins into the periplasm. Although the Tat system has recently been successfully developed for recombinant protein production (RPP) applications [bib_ref] High-level secretion of a recombinant protein to the culture medium with a..., Albiniak [/bib_ref] , the majority of recombinant proteins translocated to the periplasm have been directed via the SecB and SRP pathways.
Targeting of polypeptide chains to the periplasm via SecB, SRP or Tat requires an N-terminal signal peptide that specifically interacts with components of the three pathways. This signal peptide is cleaved from the polypeptide chain by a protease during translocation, resulting in a mature protein in the periplasm. The destination (cytoplasmic or periplasmic) and route (SecB, SRP or Tat) of the polypeptide chain is therefore specified by the sequence of the signal peptide. Multiple factors affect the functionality of the signal peptide. It must interact, via electrostatic and hydrophobic interactions, with the inner membrane and the translocation apparatus to facilitate polypeptide transport [bib_ref] SecA-mediated targeting and translocation of secretory proteins, Chatzi [/bib_ref]. The increased incidence of rare codons in the signal peptide has been revealed to play a role in control of translation speed and protein folding (reviewed by [bib_ref] Biased codon usage in signal peptides: a role in protein export, Zalucki [/bib_ref]. The structure of the mRNA encoding the signal peptide has also been shown to have an influence on translocation in Lactococcus lactis [bib_ref] Engineering signal peptides for enhanced protein secretion from Lactococcus lactis, Ng [/bib_ref] , and in E. coli via translational pausing. Therefore, the signal peptide affects protein translation and translocation via a variety of mechanisms.
Selection and optimisation of signal peptides for effective translocation of recombinant proteins to the periplasm has been the subject of much research (reviewed by [bib_ref] Optimisation of signal peptide for recombinant protein secretion in bacterial hosts, Low [/bib_ref]. There are several important factors to consider when selecting a signal peptide and matching it to a RPP process. First, the rate of recombinant protein translation must match the rate of transport to the periplasm to ensure that unfolded polypeptides do not accumulate in the cytoplasm [bib_ref] Optimizing heterologous protein production in the periplasm of E. coli by regulating..., Schlegel [/bib_ref]. This would increase the risk of protein misfolding and inclusion body formation, thus inducing the cytoplasmic heat shock response [bib_ref] Consequences of membrane protein overexpression in Escherichia coli, Wagner [/bib_ref]. Second, recombinant protein translocation must not prevent translocation of native proteins, which would negatively affect bacterial physiology and viability. These requirements for the design of periplasmic RPP are additional to the requirements for any RPP process, such as balancing the metabolic requirements of RPP and growth to prevent metabolic burden or the stringent response, and prevention of excessive protein misfolding that would trigger the heat shock response (reviewed by 2 ). Therefore signal peptide choice is one element of process design, along with optimisation of fermentation conditions [bib_ref] Periplasmic expression in and release of Fab fragments from Escherichia coli using..., Hsu [/bib_ref].
This study stemmed from a simple question: "Which signal peptide should be used for targeting a recombinant protein to the periplasm?" The answer to this question is far from straightforward and cannot at this time be answered generically. There is no way at present to select or design a signal peptide in silico based on the nucleotide or peptide sequence of the recombinant protein (RP) of interest. Both academic and industrial groups have approached the problem by using multiple signal peptides and analysing RPP and periplasmic export; a "trial and error" approach (reviewed by [bib_ref] Optimisation of signal peptide for recombinant protein secretion in bacterial hosts, Low [/bib_ref]. Bioinformatic analysis has been used to construct the consensus signal peptide for E. coli [bib_ref] Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage..., Nielsen [/bib_ref] ; however, this was found to be suboptimal for periplasmic production of a single chain variable (scFv) antibody fragment, reinforcing the need for experimental selection of optimal signal peptides for each recombinant protein [bib_ref] Broad-host-range plasmid pJB658 can be used for industrial-level production of a secreted..., Sletta [/bib_ref]. Mutagenic screening approaches have sought to identify optimal signal peptides in E. coli [bib_ref] Combinatorial mutagenesis and selection of improved signal sequences and their application for..., Heggeset [/bib_ref] , Bacillus subtilis [bib_ref] Improvement of Sec-dependent secretion of a heterologous model protein in Bacillus subtilis..., Caspers [/bib_ref] and Lactococcus lactis [bib_ref] Engineering signal peptides for enhanced protein secretion from Lactococcus lactis, Ng [/bib_ref]. However, for E. coli, selection of optimal signal peptides for one recombinant protein does not translate to a generic signal peptide that is suitable for other recombinant proteins, reinforcing the need for rapid screening procedures [bib_ref] Combinatorial mutagenesis and selection of improved signal sequences and their application for..., Heggeset [/bib_ref]. Unfortunately, analysis of periplasmic targeting of RPs is time-consuming and low throughput without extensive robotics, requiring separation of cytoplasmic and periplasmic protein fractions followed by SDS-PAGE.
The primary aim of the current work was to develop a method for screening randomly produced libraries for signal peptides that improve the periplasmic targeting of any specific protein of interest. Our approach combined several well-established rapid techniques to generate and detect colonies that merit more detailed analysis. They include the use of error-prone PCR for library generation, C-terminal β-lactamase fusion to report secretion of the target into the periplasm [bib_ref] A rapid protein folding assay for the bacterial periplasm, Mansell [/bib_ref] and nitrocefin to detect β-lactamase activity of colonies after overnight growth [bib_ref] Outer membrane permeability in Pseudomonas aeruginosa: comparison of a wild-type with an..., Angus [/bib_ref]. We started with a model scFv antibody fragment, 13R4 24 , targeted to the periplasm using three commonly-used signal peptides. Detection of periplasmic scFv was then simplified and accelerated using a C-terminal TEM-1 β-lactamase (Bla) fusion. Signal peptide libraries were generated using error-prone PCR and chemical oligonucleotide synthesis routes and independently cloned upstream of the scFv::bla fusion to generate libraries, which were screened using β-lactamase assays. This assay system allowed higher-throughput screening using microplates, which would be difficult to achieve for the subcellular fractionation of E. coli and analysis of proteins. Promising clones were then grown in small scale and bioreactor cultures and scFv production and location assessed. Finally, the system was validated using selected signal peptide-scFv constructs minus their bla fusion. encoding scFv 13R4 on plasmid pLBAD2; expression was driven by the arabinose-inducible pBAD promoter [bib_ref] Tight regulation, modulation, and high-level expression by vectors containing the arabinose pBAD..., Guzman [/bib_ref]. Transformed E. coli BL21-A cultures were grown in terrific broth (TB) at 25 °C and RPP was induced at an OD 600 of 0.5 with 0.02% (w/v) arabinose to activate the pBAD promoter and 0.25% (w/v) glucose to regulate RPP by catabolite repression, conditions previously optimised for production of this scFv. Growth data revealed that induction of RPP inhibited growth of STII sp -scFv and DsbA sp -scFv cultures ; CFU analysis corroborated this by way of decreased culturability, indicating stress , whereas PelB sp -scFv cultures displayed less growth inhibition and a lower decrease in culturability. SDS-PAGE analysis of bacteria harvested after 24 hours growth revealed higher accumulation of scFv in STII sp -scFv and DsbA sp -scFv than PelB sp -scFv cultures, although relatively little transport to the periplasm and virtually none in the case of DsbA sp . In summary: STII sp gave rise to relatively high scFv production (20% of total cell protein, TCP), moderate periplasmic accumulation (only 15% of scFv was in the periplasmic fraction) but poor growth; DsbA sp high scFv production (22.8% of TCP), high periplasmic accumulation (22% of TCP) and poor growth; and PelB sp moderate scFv production (8.2% of TCP), moderate periplasmic accumulation (15%) and better growth. These data confirm that choice of signal sequence has a large impact on not only translocation but also overall production of scFv 13R4, and significant effects on bacterial growth.
## Figure 1.
Production of scFv 13R4 in the periplasm using STII, DsbA and PelB signal peptides. E. coli BL21-A carrying the vector encoding scFv 13R4 fused to STII sp , DsbA sp or PelB sp was grown at 25 °C; RPP was induced with 0.02% arabinose at an OD 600 ≈ 0.5 (blue arrow). Upon induction, 0.25% glucose was added to the culture medium. (a) The OD 600 and (b) CFU of the culture were measured over time. Data are shown as mean values from replica flasks, error bars are ±1 standard deviation. (c) SDS-PAGE and (d) Western blot analysis (anti-His) of total (T), periplasm (P) and spheroplast (Sp) fractions after 24 hours growth. A sample of the culture carrying STII sp fused to scFv 13R4 grown under non-induced conditions was used as control. Percentages of total cell protein that was scFv and the distribution of scFv between periplasmic and spheroplast fractions shown below (c). Uncropped gel images are available in Supplemental Development of a more high-throughput method for assessing periplasmic targeting. Since growth followed by harvest and separation of bacteria into periplasmic and spheroplast fractions and analysis by SDS-PAGE is a time-consuming process, and unsuited to use as a high throughput screening method without the involvement of robotic platforms, we utilised the TEM-1 β-lactamase (Bla) as a reporter of periplasmic location [bib_ref] A rapid protein folding assay for the bacterial periplasm, Mansell [/bib_ref]. Bla confers resistance to β-lactam antibiotics such as penicillin by cleaving the β-lactam ring. The Bla protein is targeted to the periplasm, the site of action of β-lactam antibiotics. Mature, periplasmic Bla also contains a single disulphide bond, although disulphide bonding is not required for activity [bib_ref] The precursor of beta-lactamase: purification, properties and folding kinetics, Laminet [/bib_ref]. Bla has previously been used as a C-terminal fusion reporter of periplasmic recombinant protein folding [bib_ref] A rapid protein folding assay for the bacterial periplasm, Mansell [/bib_ref] and as a screening tool for selection of improved E. coli signal peptides [bib_ref] Combinatorial mutagenesis and selection of improved signal sequences and their application for..., Heggeset [/bib_ref] , although in the latter study not as a fusion to recombinant proteins.
The gene encoding scFv 13R4 was fused at its C-terminus to the bla gene from pUC18; four different signal peptide sequences were used (STII sp , DsbA sp , PelB sp and the native Bla sp ; . In addition, controls comprising scFv-bla without a signal sequence and bla with and without its native signal sequence were constructed. Resultant constructs were transformed into E. coli BL21-A and transformants were grown at 25 °C with induction (0.02% arabinose) at OD 600 = 0.5. Growth analysis revealed repression of growth and/or culturability in all cultures expressing periplasmic scFv (in order of decreasing severity) DsbA sp -scFv::bla, >Bla sp -scFv::bla >STII sp -scFv::bla >PelB sp -scFv::bla. These data confirm growth repression observed in cultures expressing scFv without the bla fusion .
SDS-PAGE and Western blot analysis of cultures expressing bla with and without the Bla signal peptide confirmed expression and the requirement of Bla sp to direct Bla to the periplasm (Supplemental . Bla was also observed in the medium of cultures expressing periplasmic Bla, indicating some leakage. SDS-PAGE and Western blot analysis of cultures expressing scFv::bla revealed relatively high accumulation of scFv::Bla in STII sp and Bla sp cultures (both having 24% of TCP as scFv::Bla at 24 h), lower accumulation with DsbA sp (15%), and lowest accumulation with PelB sp (9%; . The proportion of scFv::Bla in the periplasm was highest with STII sp (22%), followed by PelB sp (21%), then DsbA sp and Bla sp (both 16%; .
To simplify quantification of periplasmic transport of scFv::Bla fusions, the activity of β-lactamase was quantified using two methods. First, the minimal inhibitory concentration (MIC) of the β-lactam antibiotic ampicillin was determined for cultures expressing scFv::bla fusions. Transformants were grown in liquid culture without induction of RPP, then serially diluted and plated onto M-H agar plates supplemented with kanamycin, 0.2% or 0.02% arabinose (to induce RPP) and concentrations of ampicillin from 3 to 1600 µg·mL −1 . Plates containing 0.2% arabinose were incubated at 37 °C and plates containing 0.02% arabinose at 25 °C; colonies were enumerated ; CFU data presented in . A signal peptide was required to confer resistance to ampicillin. For cultures grown at 37 °C and induced with 0.2% arabinose, the DsbA sp and Bla sp did not confer ampicillin resistance. As these two signal peptides were deleterious to growth and culturability even during culture at 25 °C (low-stress conditions; , it was concluded that this was most likely due to a lack of viability at 37 °C in the presence of arabinose (high-stress conditions). Growth was detected on the control plates (containing kanamycin but not induced with arabinose) for DsbA sp and Bla sp , confirming this hypothesis. Growth at 25 °C and induction with 0.02% arabinose restored ampicillin resistance for all periplasmically-targeted scFv::bla fusions. Ampicillin agar plates could thereby not only select for clones generating high quantities of scFv::Bla, but also counterselect against clones with growth defects due to RPP.
Second, β-lactamase activities were determined using the chromogenic β-lactam nitrocefin as a substrate [bib_ref] Novel Method for Detection of β-Lactamases by Using a Chromogenic Cephalosporin Substrate, O'callaghan [/bib_ref]. Cells were grown in liquid culture at 25 °C and RPP was induced with 0.02% arabinose, after which β-lactamase activity of whole cells was measured at intervals . As reported previously, cytoplasmically targeted Bla was active due to the lack of a requirement of disulphide bonding for enzymatic activity [bib_ref] The precursor of beta-lactamase: purification, properties and folding kinetics, Laminet [/bib_ref] [bib_ref] The role of the beta-lactamase signal sequence in the secretion of proteins..., Kadonaga [/bib_ref]. However, the β-lactamase activity of cytoplasmic scFv::Bla was very low, presumably due to relatively low accumulation of scFv::Bla in the cytoplasm, possibly due to protease degradation , lane scFv(C)). All four signal peptides permitted some level of activity from scFv::Bla. DsbA sp and Bla sp displayed relatively high activity after 4 hours, reflecting the rapid accumulation of scFv::Bla observed in SDS-PAGE . PelB sp had both the lowest scFv::Bla accumulation as judged by SDS-PAGE and β-lactamase activity. Overall, we have demonstrated that β-lactamase fusions are a good method for evaluation of periplasmic targeting of scFv 13R4.
## Development of a screen for selecting improved signal peptides.
Based on the successful testing of scFv::Bla fusions, a screen was developed to select optimal signal peptides. The design of the screen is summarised in [fig_ref] Figure 3: Workflow for β-lactamase screening for periplasmic protein production [/fig_ref]. Two random libraries of signal peptides were generated, one using error-prone PCR (epPCR) and one chemically synthesised (CS). The libraries were cloned upstream of scFv::bla fusions in the plasmid pSCREEN (construction detailed in Supplementary materials and methods) and the resulting colonies screened first for growth on agar plates containing ampicillin up to a concentration of of 200 μg·mL −1 (in effect an enrichment step), followed by β-lactamase assay using nitrocefin. Signal peptides were then selected on the basis of β-lactamase activity and analysed for growth and scFv::Bla production, first in shake flasks and then in fed-batch fermentation using an Ambr ® 250 modular fermentation system. Finally, fed-batch fermentation was repeated for the selected signal peptides directing translocation of the scFv without the Bla fusion. The PelB signal peptide was used as a starting point, as it allowed translocation of scFv and scFv::Bla to the periplasm but did not affect growth to a great extent in the above experiments.
Error-prone PCR library. For the epPCR library, four successive epPCR reactions were used to introduce mutations into the PelB signal peptide. This approach was taken as the chosen enzyme (Genemorph II, Agilent) can introduce 9-16 mutations per kilobase. As the signal peptide sequence is only 66 nucleotides in length, four successive rounds of epPCR were used to introduce additional mutations, aiming for up to 8 nucleotide mutations per signal peptide. The resultant signal peptide pools were named epPCR1, epPCR2, epPCR3 and epPCR4, with a further pool (epPCR5) comprising a mixture of pools 1-4. The linearised screen plasmid (pSCREEN;
SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3
## Figure 2.
Construction and characteristics of signal peptide-scFv-β-lactamase constructs. (a) Plasmid constructs generated for the β-lactamase screening; signal peptide (blue), scFv 13R4 (green) and the bla reporter (red). Cytoplasmic versions of β-lactamase and scFv::Bla without a signal peptide are included as controls, marked as (C). E. coli BL21-A carrying the vectors in (a) were grown at 25 °C, induced with 0.02% of arabinose at an OD 600 ≈ 0.5 (blue arrow), and the OD 600 (b) and CFU (c) measured. Data are shown as mean values from replica flasks, error bars are ±1 standard deviation. (d) Cells were harvested and analysed by SDS-PAGE and Western blotting (anti myc). The quantity of scFv::Bla is expressed as a percentage of whole cell protein at the bottom of the gel. Controls are scFv::Bla (C) and empty vector (−ve). Percentages of total cell protein that was scFv are shown below the gel. (e) The periplasmic (P) and spheroplast (Sp) fractions of samples taken after 6 hours growth were separated and analysed by SDS-PAGE and Western blotting using anti-myc. Negative control is the empty vector (pLBAD2), total protein. Percentage distributions of scFv between periplasmic and spheroplast fractions are shown below the gel. (f) β-lactamase activity was determined using nitrocefin. Transformants were grown at 25 °C and induced with 0.02% arabinose after 2 hours growth; samples were taken at intervals and β-lactamase activity of whole cells was measured. Error bars represent the 5% uncertainty in the calculation of the slopes corresponding to β-lactamase activity (n = 3), which is expressed in terms of change in OD 495 per minute per OD 600 . Uncropped gel images are available in Supplemental [fig_ref] Figure 3: Workflow for β-lactamase screening for periplasmic protein production [/fig_ref].
Supplemental [fig_ref] Figure 3: Workflow for β-lactamase screening for periplasmic protein production [/fig_ref] was ligated with each signal peptide pool, electroporated into E. coli ElectroSHOX, recovered in LB medium without selection for one hour, then transferred to a flask and grown for 12-18 hours with selection, after which plasmid DNA was isolated and purified. The resultant plasmid library was then transformed into the expression strain BL21-A, serially diluted and plated onto agar plates containing: 0.2% arabinose to induce scFv::Bla production, kanamycin to select for transformants, and ampicillin at concentrations between 3 and 1600 µg·mL −1 to select for Bla + bacteria. Plates were incubated at 37 °C. These growth conditions represent high induction so will counterselect for clones with low viability or those whose growth and physiology are negatively affected by RPP (Supplemental .
Growth was observed on agar plates at up to 200 µg·mL −1 ampicillin. Two hundred clones were selected at random (40 from each epPCR pool 1-5), and split into 5 libraries (A-E), that were grown in duplicate in 96-well plates in TB at 25 °C (conditions chosen as in to permit effective production of scFv), and induced with 0.02% arabinose after 2 hours growth. Relative β-lactamase activities were determined for each clone (Supplemental [fig_ref] Figure 4: Production of scFv [/fig_ref] and compared to PelB sp -scFv::bla, scFv::bla without a signal peptide and the empty vector (pLBAD2) as controls. The majority of clones had a similar β-lactamase activity to PelB sp -scFv::bla. Some clones had a lower activity (eg epA22, denoting error-prone PCR library A clone 22, and epB3); some clones had higher activity than PelB sp -scFv::bla after 8 hours growth but no increase in activity after 24 hours (eg epA7, epA19). However, the higher activity of some clones after 24 hours growth showed not only that the current objective of high-level production of functional scFv::Bla after 8 hours growth had been achieved, but also suggests clone stability.
The robustness of the β-lactamase assay as a screening method was assessed by measurement of the β-lactamase activity of 40 different colonies of E. coli BL21-A transformed with the vector encoding PelB sp -scFv::Bla (Supplemental [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. Colony to colony variability was low after 8 hours growth (CV = 0.13), although slightly higher after 24 hours growth (CV = 0.17), likely reflecting heterologous bacterial responses to stress imposed by periplasmic scFv::Bla production. Nonetheless, colony to colony variability was far lower than clone to clone variability observed in the epPCR screen (Supplemental [fig_ref] Figure 4: Production of scFv [/fig_ref] ; CV = 0.58 at 8 h and 0.58 at 24 h), confirming the validity of this approach.
Chemically synthesised signal peptide library screening. A similar workflow to the epPCR signal peptide library was followed for the chemically synthesised (CS) signal peptide library. The library was based on PelB sp and comprised 10 000 signal peptides, designed to contain 2-4 mutations each. A further 200 clones were assayed for β-lactamase activity (Supplemental [fig_ref] Figure 6: Sequence alignment of the signal peptides selected from epPCR and CS libraries [/fig_ref]. Compared to the epPCR signal peptide library, there was a larger number of high activity clones in the CS library. This was not unexpected as the CS library was designed to contain a comparable number of mutations per signal peptide as the epPCR4 pool. As with the epPCR library, individual clones had higher, comparable or lower β-lactamase activity when compared to PelB sp . Some clones in the CS library had higher β-lactamase activity than the highest activity clones in the epPCR library (eg csA3, 0.142 units; csD4, 0.165 units; compared to epPCR library maximum epC28, 0.118 units).
Evaluation of signal peptide clones from the epPCR library. Ten clones were selected for further analysis to investigate the variability generated by the epPCR library, epC25-epC34, representing three phenotypic groups: high activity clones epC26, epC31 and epC33; clones with activity close to the wild type PelB sp (epC27, epC28, epC29 and epC30); and clones with low activity (epC25, epC32 and epC34). Clones were grown in shake flasks in terrific broth (TB) at 25 °C and induced with 0.02% arabinose at OD 600 = 0.5. Growth and CFU data comparing the 10 selected clones with PelB sp (Supplemental [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] were inconclusive. Analysis of scFv accumulation by SDS-PAGE revealed that all high activity clones had higher scFv::Bla accumulation than the wild type PelB sp (20-24% versus 12% of TCP was scFv::Bla at 24 h; [fig_ref] Figure 4: Production of scFv [/fig_ref]. Conversely, all low activity clones had lower scFv::Bla accumulation (around 2% of TCP); most 'medium activity' clones had comparable scFv::-Bla accumulation to PelB sp . It should be noted that clone epC29 accumulates a large quantity of scFv (24% of TCP at 24 h), as revealed by SDS-PAGE, although has a 'medium' β-lactamase activity (~0.04 units at 24 h). This highlights that, although β-lactamase activity assays are a useful rapid screening technique to select for potential high-accumulating clones, SDS-PAGE must be used to measure protein concentration. Specific productivity values [fig_ref] Figure 4: Production of scFv [/fig_ref] confirm these observations. However, fractionation of periplasmic and spheroplast fractions revealed that no high activity clones had an increased proportion of scFv::Bla in the periplasm compared to PelB sp [fig_ref] Figure 4: Production of scFv [/fig_ref]. Therefore, it can be concluded that the increase in activity in the high activity epPCR clones investigated here was primarily driven by an increase in overall scFv::Bla accumulation rather than an increased proportion of scFv::Bla being transported to the periplasm.
Evaluation of signal peptide clones from the CS library. Five clones were selected from the CS library and analysed as above [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. The β-lactamase activity of all five was higher than with PelB sp , and four had higher activity than the highest activity epPCR signal peptides. Growth data revealed that growth of clones csD4 and csE18 was inhibited following induction of RPP [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. SDS-PAGE analysis [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref] revealed that none of the CS library clones analysed accumulated as much scFv::Bla as the high producers analysed from the epPCR screen (11-19% of TCP was scFv::Bla compared to 20-24% for the epPCR clones; [fig_ref] Figure 4: Production of scFv [/fig_ref] ; in fact, two clones (csA19 and csE18) had comparable accumulation to the wild type PelB sp . Specific productivity data for total scFv confirmed this [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. However, analysis of periplasm and spheroplast fractions revealed increased proportions of scFv::Bla in the periplasm of four clones (25-40% of scFv being in the periplasmic fraction compared to 10% for PelB sp ). In only one clone (csD4) was partitioning similar to PelB sp [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. Specific productivity of periplasmic scFv was higher than PelB sp for all four analysed clones, and much higher for the three best-growing clones (csA11, csA19 and csB2; 1.23 to 1.52 mg scFv·g −1 dcw·h −1 versus 0.31 for PelB sp ).
Taken together, the data reveal two mechanisms by which elevated β-lactamase activity was achieved by clones from the mutant signal peptide libraries. First, overall accumulation of scFv::Bla was elevated (named here high expression signal peptides) as seen in all high activity epPCR library clones analysed and clone csD4 from the CS library. These signal peptides are likely to allow increased translation of the scFv::Bla but not enhance translocation to the periplasm. This phenomenon has been observed previously [bib_ref] The presence of N-terminal secretion signal sequences leads to strong stimulation of..., Sletta [/bib_ref]. Second, increased partitioning of scFv::-Bla into the periplasm was observed in CS library clones csA11, csA19, csB2 and csE18, suggesting enhanced periplasmic translocation (named here high transporting signal peptides).
## Sequence analysis of signal peptides.
To better understand the effect that the signal peptides evaluated above had on production and transport of scFv::Bla, they were sequenced [fig_ref] Figure 6: Sequence alignment of the signal peptides selected from epPCR and CS libraries [/fig_ref]. Sequences are split into four categories: signal peptides giving rise to low (L) and medium (M) β-lactamase activity; those with high β-lactamase activity due to increased accumulation of scFv::Bla (high expression, Hex); and those with high β-lactamase activity due to enhanced partitioning of scFv::Bla to the periplasm (high transporting, HT). Signal peptides targeting proteins to the SecB pathway comprise three regions: the N-terminal n-region, which has a positive charge thought to allow the signal peptide to associate with the negatively-charged inner membrane; a central h-region, which is hydrophobic and α helical in conformation and associates with SecA; and a C-terminal c-region, which has a β conformation [bib_ref] Identification of prokaryotic and eukaryotic signal peptides and prediction of their cleavage..., Nielsen [/bib_ref] [bib_ref] Structure and mechanism of Escherichia coli type I signal peptidase, Paetzel [/bib_ref]. The C-terminus of the c-region contains the highly conserved "AXA" motif (where X is any amino acid) that is recognised by signal peptidase I and is the site of signal peptide cleavage [bib_ref] Structure and mechanism of Escherichia coli type I signal peptidase, Paetzel [/bib_ref] [bib_ref] Minimum substrate sequence for signal peptidase I of Escherichia coli, Dev [/bib_ref].
Considering the low activity signal peptides, epC25 has an insertion and epC34 a deletion giving rise to a frameshift, resulting in deletion of the signal peptide. Signal peptide epC32 has a large number of nucleotide mutations giving rise to two additional proline residues (amino acids 9 and 10) in the h-region, likely to break the α-helical conformation of this region thereby inhibiting function. The four medium activity signal peptides analysed either have no amino acid changes in signal peptide sequence, or relatively small changes.
All four HT signal peptides have a mutation at amino acid 6 in the n-region, which is a proline in PelB sp . In signal peptides csA11, csA19 and csB2, this residue is mutated to a positively-charged amino acid; these three signal peptides gave rise to the highest partitioning of scFv::Bla into the periplasm [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. The mutations in csA19 and csB2 resulted in an increased overall positive charge of the n-region; this has been shown in previous studies to increase translocation [bib_ref] Optimisation of signal peptide for recombinant protein secretion in bacterial hosts, Low [/bib_ref]. The only sequenced signal peptides in this study with increased positive charge in the n-region are these two HT signal peptides. Signal peptide csA11 has an arginine residue at position 6, but position 2 is mutated from lysine to leucine, maintaining overall charge. Nonetheless, the position of charged residues as well as overall charge is known to be important in the n-region [bib_ref] Optimisation of signal peptide for recombinant protein secretion in bacterial hosts, Low [/bib_ref]. The fourth HT signal peptide (csE18) also has a mutation at proline 6, but to a glutamine. Three Hex signal peptides (epC26, epC31 and epD4) also have mutations at this position, although none to a charged amino acid. Taken together, this suggests that mutation of proline 6 significantly modulates signal peptide functionality. Future work will focus on the effect of specific mutations at this position in the signal peptide. Within the h-region, three HT signal peptides (csA11, csA19 and csB2) have increased hydrophobicity when compared to PelB sp (hydrophobicity values calculated according to [bib_ref] Analysis of membrane and surface protein sequences with the hydrophobic moment plot, Eisenberg [/bib_ref] ; the hydrophobicity of the h-region of these signal peptides correlates to the proportion of scFv::Bla in the periplasm [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. Numerous studies have found increased hydrophobicity in the h-region to increase protein translocation, although this is not always the case, and is dependent upon the protein being targeted to the periplasm and by the need to balance transport of recombinant proteins with native periplasmic proteins (reviewed by [bib_ref] Signal peptides: exquisitely designed transport promoters, Izard [/bib_ref] [bib_ref] Enhancing full-length antibody production by signal peptide engineering, Zhou [/bib_ref]. Again, none of the other sequenced signal peptides had increased hydrophobicity in the h-region, and two of the Hex signal peptides (epC26 and epC33) have dramatically lower hydrophobicity than PelB sp . Two HT signal peptides (csB2 and csE18) have mutations at glycine 11; glycine can act as a helix breaker, so these mutations could increase the helical nature of the h-region [bib_ref] Signal peptides: exquisitely designed transport promoters, Izard [/bib_ref]. Again, one Hex signal peptide (epC26) also has a mutation at glycine 11, though combined with a dramatically decreased h-region hydrophobicity.
The relationship between the amino acid sequence of the c-region and function is not well understood, except for the consensus AXA motif which is recognised by signal peptidase I 34 . The HT signal peptides have few mutations in this region: csA19 has a proline in position −5, proline being a common amino acid in this region, thought to promote a β-turn 38 . Signal peptide csE18 has an isoleucine at position −3, breaking the AXA rule. Karamyshev et al. [bib_ref] Processing of Escherichia coli alkaline phosphatase: role of the primary structure of..., Karamyshev [/bib_ref] replaced alanine with leucine at position −3, resulting in a functional signal peptide, but did not attempt to insert isoleucine at this position; nonetheless, position −3 is less conserved than position −1 [bib_ref] Structure and mechanism of Escherichia coli type I signal peptidase, Paetzel [/bib_ref] [bib_ref] Processing of Escherichia coli alkaline phosphatase: role of the primary structure of..., Karamyshev [/bib_ref].
Three Hex signal peptides (epC26, epC33 and scD4) have mutations at positions −4 and −5 in relation to the peptidase cleavage site. It has been reported that medium-sized amino acids are preferred at positions −4 and −5 [bib_ref] Optimisation of signal peptide for recombinant protein secretion in bacterial hosts, Low [/bib_ref] [bib_ref] Processing of Escherichia coli alkaline phosphatase: role of the primary structure of..., Karamyshev [/bib_ref]. The mutations at position −5 in epC26 and csD4 (lysine and tryptophan) are rare at this position in Gram negative bacterial signal peptides [bib_ref] Processing of Escherichia coli alkaline phosphatase: role of the primary structure of..., Karamyshev [/bib_ref]. There is little information about the role of position −4, although the mutations in epC33 and epD4 at this position (glutamine and valine) are underrepresented in known signal peptides [bib_ref] Processing of Escherichia coli alkaline phosphatase: role of the primary structure of..., Karamyshev [/bib_ref].
In summary, signal peptide performance is thought to be determined by a combination of amino acid sequence attributes: the overall charge and position of charged residues in the n-region; the hydrophobicity and conformation of the h-region; and attributes of the c-region. These changes alter the way in which the signal peptide interacts with the membrane, Sec translocase and signal peptidase. The enhanced performance of the HT signal peptides csA11, csA19 and csB2 is likely due to a combination of increased and/or redistributed positive charge in the n-region and increased hydrophobicity of the h-region. However, the reasons for the enhanced performance of csE18 (although not as enhanced as csA11, csA19 and csB2) are unknown.
In addition to changes in amino acid sequence, codon usage can also affect signal peptide performance, changing translation rates and thereby folding and translocation [bib_ref] Combinatorial mutagenesis and selection of improved signal sequences and their application for..., Heggeset [/bib_ref] [bib_ref] Whole genome analysis reveals a high incidence of nonoptimal codons in secretory..., Power [/bib_ref]. Rare codons corresponding to <10% usage in E. coli B 40 are present in three HT signal peptides: csA11 has arginine 6 encoded by AGG and isoleucine 9 by ATA; csA19 has proline 18 encoded by CCC; and csE18 has a silent mutation changing the codon encoding leucine 13 from CTG to CTA. However, some rare codons were also added to signal peptides in the low, medium and Hex groups, so codon bias is only one factor in determining signal peptide functionality.
The data presented here reflect the complexities of correlating amino acid sequence to signal peptide function, and confirm the need for screening to select improved signal peptides. The data also confirm that signal peptide amino acid sequence is not only important for interaction with the Sec apparatus and translocation, but also regulates translation of the recombinant protein.
Performance of signal peptides in fed-batch fermentations for production of scFv::Bla. Five high activity signal peptides were evaluated in fed-batch high cell density fermentation cultures. Two high expression (epC26 and epC33) and three high transporting (csA11, csA19 and csB2) signal peptides were compared with PelB sp for production and periplasmic transport of scFv::Bla using an Ambr ® 250 modular fermentation system. These five signal peptides were chosen on the basis of high β-lactamase activity, scFv::Bla accumulation and growth. A semi-defined medium and a glycerol-based feed using an exponential feeding profile to maintain µ = 0.1 were used [bib_ref] Studies related to antibody fragment (Fab) production in Escherichia coli W3110 fed-batch..., Want [/bib_ref]. Initial cultures were induced with arabinose at a low cell density (OD 600 = 0.5); only epC26 and epC33, the two high expression clones, were able to achieve high cell density, the remaining cultures (including the PelB sp ) displaying poor growth and culturability (Supplemental [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref]. Growth conditions were therefore optimised to permit growth of all cultures to high cell density; cultures were induced at a high cell density (OD 600 between 70 and 80), with growth at 30 °C before induction and 25 °C afterwards. Growth and culturability were comparable for all cultures, with no growth defects observed following induction [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] although culturability decreased following induction, reflecting increased stress during RPP [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref]. Specific growth rates are compared in . SDS-PAGE analysis of whole cell lysates [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] showed that for high expression signal peptides epC26 and epC33, scFv::Bla accumulated at early stages of induction but by 28 hours (8 hours post-induction) had decreased, indicating loss of productivity, probably caused by plasmid loss or proteolysis. The accumulation of scFv::Bla (in terms of percentage of total cell protein) in these fed-batch cultures (maxima of 10% and 9% for epC26 and epC33) was far lower than in shake flasks (22% & 24% respectively; [fig_ref] Figure 4: Production of scFv [/fig_ref]. Fractionation [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] revealed that epC26 and epC33 did not direct scFv::Bla into the periplasm, corroborating shake flask experiments [fig_ref] Figure 4: Production of scFv [/fig_ref]. Specific productivity data [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] reflect the above findings. Specific productivity was also lower for fed-batch fermenters than the corresponding shake flask experiments [fig_ref] Figure 4: Production of scFv [/fig_ref].
The high transporting signal peptides all showed equivalent or higher accumulation of scFv::Bla in whole cell lysates (between 7.2% and 9% of TCP) when compared to PelB sp (7%; [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] , although (as with the high expression signal peptides) lower accumulation than in shake flask cultures [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref]. Taken together with the data from the high expression clones, this suggests a general decrease in RP expression levels in the fed-batch fermentation system, probably due to increased metabolic requirements of high cell density culture. Given that fed-batch cultures induced with arabinose at an OD 600 of 0.5 displayed growth defects (Supplemental [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref] , this also suggests that fed-batch growth needs to be carefully optimised in terms of titration of inducer concentration and induction point to balance not only metabolic requirements of growth and RPP but also potentially balance The feed was started after 10 hours (blue arrow) and cultures were induced with 0.02% arabinose at an OD 600 ≈ 70-80 (20 h, dotted line) whereupon the temperature was changed to 25 °C. The OD 600 (a) and CFU (b) were measured. (c) SDS-PAGE analysis of whole cell protein before (20 h) and after induction. Samples obtained from cultures carrying the scFv::Bla without the signal peptide (c) were used as a control. M is a molecular size marker. (d) Periplasm (P) and spheroplast (Sp) fractions from cultures after 28 hours of growth. The quantity of the scFv::Bla is expressed as percentage of whole cell protein (c) and the percentage of protein accumulated in each fraction (d). (e) Specific productivity of total scFv and periplasmic scFv at 28 hours growth. Values calculated from densitometry data from panels (c,d) as described in supplemental materials and methods. Uncropped gel images, and gel images for other time points, are available in [fig_ref] Figure 6: Sequence alignment of the signal peptides selected from epPCR and CS libraries [/fig_ref]. throughput of RP and native proteins through the Sec translocase. Fractionation revealed that a comparable proportion of scFv::Bla was present in the periplasm for the high transporting clones (36-44%) and PelB sp (42%; [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref]. However, comparison of shake flask fractionation data [fig_ref] Figure 5: Analysis of selected clones from chemically synthesised [/fig_ref] with fed-batch data [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] reveals that PelB sp was able to direct far more scFv::Bla into the periplasm at high compared to low cell density (42% versus 10%). Again, this probably reflects the balance between pBAD promoter strength and RP versus native protein transport through Sec.
Fed-batch production of scFv without a β-lactamase fusion. The bla gene was removed from each plasmid construct assessed in [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] and fed-batch fermentation was repeated. Although growth and culturability were good (Supplemental , scFv did not accumulate to high levels; Western blotting was needed to detect scFv on SDS-PAGE gels (Supplemental . High expression signal peptide epC26 accumulated scFv mainly in a precursor form in the spheroplast fraction; epC33 permitted slightly better transport to the periplasm and signal peptide cleavage. All CS library signal peptides allowed good partitioning to the periplasm and signal peptide cleavage. The scFv was also detected in samples obtained from the culture medium.
As high levels of scFv did not accumulate under these growth conditions, the fermentation was repeated with induction using 0.2% arabinose. In comparison to the fed-batch experiment with induction using the lower arabinose concentration, growth was similar (growth curves and CFU data in Supplemental ; specific growth rates in but accumulation of scFv was far higher [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref] , visualisation of scFv being possible by SDS-PAGE. As with the scFv::Bla fusion, the Hex signal peptides accumulated more scFv than the HT signal peptides (6-6.4% of TCP at 30 h versus 3.5-5.1%), but very little was transported to the periplasm [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref]. However, scFv accumulation under these conditions was still lower than accumulation of scFv::Bla during fed-batch growth and 0.02% arabinose induction (up to 10%; [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref]. This decrease in scFv accumulation when Bla is not present could indicate that Bla acts to increase the efficiency of translation or transport, or enhance the stability of scFv during synthesis and transport to the periplasm. This is a similar principle to the use of periplasmic fusion tags such as maltose binding protein (MBP) or DsbA, although these fusions tend to be at the N-terminal of the recombinant protein of interest [bib_ref] Increased production of human proinsulin in the periplasmic space of Escherichia coli..., Winter [/bib_ref] [bib_ref] Protein fusion tags for efficient expression and purification of recombinant proteins in..., Malik [/bib_ref].
Culture medium samples revealed that scFv leaked from cells with the HT signal peptides and PelB sp . Analysis of fractions confirmed transport of scFv to the periplasm by these signal peptides [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref]. As Hex signal peptides epC26 and epC33 poorly translocated scFv to the periplasm, there was no leakage.
Finally, scFv activity was assayed using ELISA [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref] and compared with densitometry analysis of SDS-PAGE gels [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref]. As expected, epC26 and epC33 samples had relatively low scFv activity as determined by ELISA; epC33, having better translocation of scFv than epC26, had a higher periplasmic activity. The relatively high activity of the spheroplast fraction in epC26 probably reflects scFv activity in the absence of disulphide bonding. Although a disulphide bond is required for correct folding of Fv domains 4 , scFv 13R4 is an engineered scFv selected for enhanced folding in the cytoplasm of E. coli where disulphide bonding is not achieved [bib_ref] Expression of an antibody fragment at high levels in the bacterial cytoplasm, Martineau [/bib_ref]. Both high expression signal peptides (epC26 and epC33) gave rise to precursor scFv in the spheroplast fraction, [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref] indicative of cytoplasmic accumulation.
Comparing the HT signal peptides with PelB sp , csA11 and csB2 both had higher periplasmic and culture medium scFv activity and higher [scFv] as determined by densitometry than the wild type PelB sp ; periplasmic scFv activities from csA11 and csB2 cultures were around 40% higher than PelB sp . HT signal peptide csA19 had lower performance than PelB sp in terms of active scFv as determined by ELISA, and comparable [scFv] as determined by SDS-PAGE. Specific productivity data, based both on the scFv concentration measured using . Specific growth rates for fed-batch fermentation cultures. SDS-PAGE and ELISA, [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref] confirm these observations, with csA11 and csB2 having higher specific productivity than PelB sp . The decrease in specific productivity between scFv::Bla [fig_ref] Figure 7: Fed-batch fermentation for the production of scFv [/fig_ref] and scFv [fig_ref] Figure 8: Production of scFv by fed-batch fermentation induced with 0 [/fig_ref] can be explained in part by the smaller molecular mass of scFv compared to the scFv::Bla fusion. Growth of cultures containing csA11 and csB2 was comparable to PelB sp (Supplemental , indicating suitability for scale-up. We can therefore confirm that the screening system has successfully generated signal peptides with better performance than PelB sp for the periplasmic production of this scFv. Peak productivity was determined as 0.65 g scFv·L −1 in the periplasmic fraction. Despite the drop in scFv accumulation when compared to scFv::Bla, the screening system nonetheless generated signal peptides capable of better production of scFv 13R4 than PelB sp . Future work should focus on further development of the system, including fermentation development to optimise process conditions and minimise the quantity of scFv in the culture medium for selected signal peptides 17 .
# Conclusions
Selection of signal peptides for translocation of recombinant proteins to the E. coli periplasm remains an important tool for bioprocessing, especially with the development of novel antibody fragment therapeutics. If E. coli is going to compete with CHO cells as an expression system for antibody fragments, platform technology (similar to the CHO platforms for monoclonal antibody production 44 ) needs to be developed. Selection of signal peptides is an important part of this platform development: we have shown that signal peptide sequence not only affects translocation but also production of scFv 13R4. This study has demonstrated for the first time the utility of C-terminal β-lactamase fusions for screening signal peptides for the translocation of recombinant proteins to the periplasm. Screening remains an essential step in signal peptide development due to the non-generic nature of signal peptides [bib_ref] Optimisation of signal peptide for recombinant protein secretion in bacterial hosts, Low [/bib_ref] ; the methods developed here will assist in this process. We have also demonstrated that, although some signal peptides increased translation but not translocation of scFv::Bla, and overall expression levels of scFv 13R4 were lower when the β-lactamase fusion was removed, two signal peptides that generated elevated β-lactamase activities in the scFv::Bla screen (csA11 and csB2) were able to improve production of periplasmic scFv 13R4 by ~40% (comparing active scFv as determined by ELISA) compared to PelB sp . Specific productivities of csA11 and csB2 were also greater than that of PelB sp . We propose that a workflow comprising a β-lactamase activity screen followed by assessment of high activity clones using growth and analysis of periplasmic protein content is a good approach to develop improved production strains for antibody fragments or other periplasmically-targeted recombinant proteins.
# Materials and methods
Strains and plasmids. E. coli strains used were the expression strain BL21-A, derived from E. coli B (F − ompT gal dcm lon hsdSB (rB − mB − ) [malB + ]K-12(λS) Δ(ara)) sourced from Cobra Biologics, and the electrocompetent ElectroSHOX sourced from Bioline, UK (F -mcrA Δ(mrr-hsdRMS-mcrBC) Φ80lacZ ΔM15 ΔlacX74 recA1 endA1 ara Δ139 Δ(ara, leu)7697 galU galK λ − rpsL (Str R ) nupG), which was used for cloning. Plasmids are detailed in and PCR and other oligonucleotide primers in . Purification of scFv 13R4 for ELISA standards. Reference scFv 13R4 was generated using a fed-batch fermentation of E. coli BL21-A carrying pLBAD2-PelB-scFv13R4 using growth conditions as described in Supplemental . Culture medium samples were taken (this fraction only contained mature scFv 13R4), filtered through a 0.2 µm filter and pretreated with 40 mM imidazole and 500 mM NaCl. A 1 mL nickel Sepharose HisTrap FF column (GE Healthcare) was washed with 5 column volumes of deionised water then 5 column volumes of binding buffer (PBS pH 7.2, 40 mM imidazole, 500 mM NaCl). The column was loaded with 10 mL of culture sample, washed with 15 column volumes of binding buffer, and eluted with 5 column volumes of elution buffer A (PBS pH 7.2, 100 mM imidazole, 500 mM NaCl) then 5 column volumes of elution buffer B (PBS pH 7.2, 500 mM imidazole, 500 mM NaCl). Eluted fractions were analysed by SDS-PAGE (Supplemental .
## Subcellular fractionation.
A modified cold osmotic shock fractionation procedure was used 9 except for data shown in . Cell pellets (equivalent to 1 OD 600 ·mL) were suspended in 50 µL of ice-cold spheroplast buffer (100 mM Tris-HCl pH 8.2, 500 mM sucrose, 5 mM EDTA) then lysozyme was added to 0.8 mg·mL −1 followed by 50 µL of ice-cold deionised water, and incubated on ice for 5 min. MgSO 4 was added to a concentration of 20 mM then the samples were centrifuged (12 000 g, 2 min). The supernatant was the periplasmic fraction, the pellet (resuspended in 100 µL of 100 mM Tris-HCl pH 8) was the spheroplast fraction. For the data in , a cold osmotic shock fractionation procedure was used. Cell pellets (equivalent to 1 OD 600 ·mL) were resupended in 150 μL of ice-cold spheroplast buffer (100 mM Tris-HCl pH 8, 500 mM sucrose, 0.5 mM EDTA) and stored on ice for 5 min. Fifty microliters were transferred to a new centrifuge tube, constituting the whole lysate protein sample. The cell suspension was centrifuged at 12 000 g for 1 minute. The supernatant was discarded and the pellet was resuspended in 100 μL of ice-cold deionised water. Ice-cold MgCl 2 was added to a final concentration of 20 mM, the cell suspension was centrifuged at 12 000 g for 2 minutes. The supernatant SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 was removed, comprising the periplasmic protein fraction, and the pellet was resuspended in 100 μL of 10 mM Tris-HCl pH 8, comprising the spheroplast protein fraction.
Cloning & screening. Details of cloning and methods can be found in the Supplementary methods section.
## Mic determination.
A single colony was used to inoculate 10 mL of LB plus 50 µg·mL −1 kanamycin, which was grown at 37 °C to an OD 600 of 1-2, then serially diluted in sterile PBS to a concentration of 10 5 -10 8 CFU·mL −1 . Dilutions were plated onto Mueller-Hinton agar supplemented with 50 µg·mL −1 kanamycin, 0.2% or 0.02% arabinose and 3-1600 µg·mL −1 ampicillin. Plates were incubated at 37 °C or 25 °C.
Beta lactamase assay. β-lactamase activities of whole cells were measured according to Angus et al. [bib_ref] Outer membrane permeability in Pseudomonas aeruginosa: comparison of a wild-type with an..., Angus [/bib_ref] and O'Callaghan [bib_ref] Novel Method for Detection of β-Lactamases by Using a Chromogenic Cephalosporin Substrate, O'callaghan [/bib_ref]. Cells were grown in 96-well plates: cells from a colony were added to 200 µL of TB supplemented with 50 µg·mL −1 kanamycin in each well and grown at 25 °C overnight. Ten µL of each culture was added to each well of a fresh 96-well plate containing 200 µL TB supplemented with 50 μg · mL −1 of kanamycin, which was grown at 25 °C for 2 h whereupon arabinose was added to 0.02% to induce RPP. At time intervals, 96-well plates were centrifuged (1 218 g, 15 min) and pellets resuspended in 10 mM Na-HEPES pH 7.0 and 100 µL of nitrocefin working solution (1 mg·mL −1 nitrocefin (Carbosynth, Reading, UK) in 200 µL DMSO and 9.8 mL Na-HEPES pH 7.0) was added to each well. The OD 495 was measured continuously for 30 minutes; reaction rates were normalised by division by OD 600 (as a measure of biomass concentration) and are in units of change in OD 495 per minute per OD 600 . The initial reaction velocity was calculated by linear regression and adapted to obtain an R ≥ 0.9 (where possible) using GraphPad Prism 7 software. The 5% uncertainty of the prediction of the slopes was also calculated and plotted as error bars.
DNA sequencing and protein sequencing. Vector modifications were confirmed by sequencing. DNA sequencing was carried out by Genewiz, formerly Beckman Coulter Genomics, with the exception of the sequencing of signal peptide libraries, which was carried out by Eurofins Genomics.
Fermentations using Ambr ® 250 modular fermentation system. Starter cultures were grown in 10 mL of LB with 50 µg·mL −1 kanamycin at 25 °C to an OD 600 of 2, where they were added to 50 mL of LB with 50 µg·mL −1 kanamycin in a 250 mL baffled shake flask which was incubated at 25 °C at 200 rpm until an OD 600 of between 4 and 6.
Fed-batch fermentations used the Ambr ® 250 modular fermentation system (TAP Biosystems, Sartorius Stedim) which comprises 250 mL single-use bioreactors. Fermentations started with 150 mL of batch medium and 100 mL of feed. The batch medium was from Want et al. [bib_ref] Studies related to antibody fragment (Fab) production in Escherichia coli W3110 fed-batch..., Want [/bib_ref] The pH was maintained at 7.0 using 10% NH 4 OH and 1 M HCl. Polypropylene glycol (PPG 2 000) was used as antifoam. The dissolved oxygen was maintained at above 20% using cascade control (increasing the stirrer speed followed by addition of O 2 ). Bioreactors were inoculated to an OD 600 of 0.1. Exponential feeding was used according to equation 1.
[formula] μ = × + × × μ F S Y m X e 1 (1) XS t 0 [/formula]
where F is the feed rate in L·h −1 , S is the substrate concentration in the feed (714 g·L −1 ), μ is the required specific growth rate (0.1 h −1 ), Y XS is the yield coefficient (0.4 g biomass per g glycerol), m is the maintenance coefficient (0.04), X 0 is the biomass in g at the start of the feed and t is time. The feed was started when the DO increased, indicating nutrient limitation. Data availability. All data generated or analysed during this study are included in this published article and its supplementary information files.
[fig] Figure S12: SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 [/fig]
[fig] Figure 3: Workflow for β-lactamase screening for periplasmic protein production. SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 [/fig]
[fig] SCiEntifiC: RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 [/fig]
[fig] Figure 4: Production of scFv::Bla by clones selected from the epPCR library. E. coli BL21-A carrying vectors coding for different signal peptides from the epPCR library (epC25-epC34) or PelB sp fused to scFv::Bla were grown at 25 °C and induced with 0.02% arabinose at an OD 600 ≈ 0.5. (a) SDS-PAGE gels showing accumulation of scFv::Bla (black arrow) from whole cell lysates. Samples were obtained before (3 h) and after induction (9 and 24 hours post-inoculation). M is the molecular size marker. Cultures carrying PelB sp fused to scFv::Bla, the cytoplasmic scFv::Bla (C) or the empty vector (-ve) after 24 hours were used as controls. (b) Specific productivity of total scFv and periplasmic scFv at 24 hours growth. Values calculated from densitometry data from panel (a,c) as described in supplemental materials and methods. (c) The periplasm (P) and spheroplast (Sp) fractions of samples after 9 hours of growth were analysed by SDS-PAGE. The quantity of the scFv::Bla is expressed as percentage of whole cell protein (a) and the percentage of protein accumulated in each fraction (c). (d) The β-lactamase activity of each clone evaluated here. Error bars represent the 5% uncertainty in the calculation of the slopes corresponding to β-lactamase activity (n = 2), which is expressed in terms of change in OD 495 per minute per OD 600 . ****Indicates values statistically significant (P < 0.0001) compared to the control (PelB sp -scFv::Bla) after selecting adjusted p < 0.001 as the level of significance during statistical analysis of data.Uncropped gel images are available in Supplemental Fig. S14. SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 [/fig]
[fig] Figure 5: Analysis of selected clones from chemically synthesised (CS) signal peptide library. E. coli BL21-A carrying the vector encoding scFv::Bla with five high-activity signal peptides from the CS library were grown at 25 °C and indu ced with 0.02% arabinose at an OD 600 ≈ 0.5 (blue arrow). The OD 600 (a) and CFU (b) were measured. Data shown are mean values of two replica flasks and error bars are ±1 standard deviation (n = 2). (c) SDS-PAGE analysis of whole cell lysates obtained before (3 h) and after induction (9 and 24 hours post-inoculation). M is the molecular size marker. (d) Specific productivity of total scFv and periplasmic scFv at 24 hours growth. Values calculated from densitometry data from panels (c,e) as described in supplemental materials and methods. (e) Periplasm (P) and spheroplast (Sp) fractions from cultures after 9 hours of growth. Negative control is the empty vector (−ve) after 24 hours growth. The quantity of the scFv::Bla is expressed as percentage of whole cell protein (c) and the percentage of protein accumulated in each fraction (e). (f) The β-lactamase activity of each clone evaluated here. Error bars represent the 5% uncertainty in the calculation of the slopes corresponding to β-lactamase activity (n = 2), which is expressed in terms of change in OD 495 per minute per OD 600 . Asterices indicate values statistically significant (****p < 0.0001; ***0.001 > p > 0.0001) compared to the control (PelB sp -scFv::Bla) after selecting adjusted p < 0.001 as the level of significance during statistical analysis of data. Uncropped gel images are available inSupplemental Fig. S15.SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 [/fig]
[fig] Figure 6: Sequence alignment of the signal peptides selected from epPCR and CS libraries. (a) Nucleotide sequence alignment; mutated bases are shown in red. L, M, HT and Hex refer to the β-lactamase activity conferred by the signal peptides: low, medium, high transporting and high expression respectively, see text for details. (b) Peptide sequence alignment. Signal peptides regions are highlighted. Underlined resides are those mutated. 'h-region change' states the change in hydrophobicity of the h-region, calculated using values from 34 . NC: no change. SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 [/fig]
[fig] Figure 7: Fed-batch fermentation for the production of scFv::β-lactamase. E. coli BL21-A transformed with vectors encoding scFv::Bla with five high-activity signal peptides (Hex, high expression; HT, high transporting) from the epPCR and CS libraries or PelB sp was grown in an Ambr ® 250 modular fermentation system at 30 °C. [/fig]
[fig] Figure 8: Production of scFv by fed-batch fermentation induced with 0.2% arabinose. E. coli BL21-A transformed with vectors encoding scFv with five high-activity signal peptides or PelB sp was grown in an Ambr ® 250 modular fermentation system at 30 °C. Cultures were induced with 0.2% arabinose at an OD 600≈ 70-80 whereupon the temperature was changed to 25 °C. (a) Whole cell lysates and culture broth were analysed by SDS-PAGE before (18 h) and after induction (26 h and 30 h). Controls are 24 hour cultures of empty vector (−ve) and cytoplasmic scFv (C). M is a molecular size marker. Periplasmic, spheroplast and culture broth fractions from 30 hours growth were analysed by SDS-PAGE (b) and Western blotting using anti-myc (c). Arrows show precursor (p) and mature (m) scFv, as determined by comparison of molecular weight. The quantity of scFv is expressed as percentage of whole cell or culture medium protein (a) or percentage in each fraction (b) at the bottom of the gel. The amount of scFv was determined using densitometry analysis of the SDS-PAGE gels (d) and active scFv was determined using ELISA (e). Uncropped gel images are available in Supplemental Fig. S17. (f) Specific productivity data based on (d) and (e). SCiEntifiC RepORTS | (2018) 8:6986 | DOI:10.1038/s41598-018-25192-3 [/fig]
|
10.1371/journal.pone.0139389
|
CCBY
|
4591971
|
26431458
|
s2orc_pubmed_articles
|
Change in Quality of Life for Patients with Irritable Bowel Syndrome following Referral to a Gastroenterologist: A Cohort Study
BackgroundIrritable bowel syndrome (IBS), a chronic functional condition, considerably reduces quality of life (QoL) and referral to gastroenterology is common. Until now, however, the impact of seeing a gastroenterologist for IBS on patients' QoL and utility has not been assessed.MethodsPatients referred with "probable IBS" to the Nottingham Treatment Centre between October 2012 and March 2014 were invited to complete a QoL questionnaire (EuroQol-5 Dimension) before their first appointment. Patients with confirmed IBS who completed this baseline assessment were sent follow-up questionnaires three and twelve months later. Global QoL and utility were measured at each time point and change from baseline calculated. Paired t-tests analysed the significance of any change.ResultsOf 205 invited patients, 69 were eligible and recruited. Response at three and twelve months was 45% and 17% respectively. Median global QoL at baseline was 67.5 (Interquartile range [IQR] 50.0 to 80.0), with a mean increase of 3.25 (95% confidence interval [CI] -5.38 to 11.88) three months later and a mean decrease of -1.82 (95% CI -16.01 to 12.38) after one year. The median utility at baseline was 0.76 (IQR 0.69 to 0.80), with a mean increase of 0.06 (95%CI -0.01 to 0.14) at three months and no change, 0.00 (-0.16 to 0.16), after one year.ConclusionPatients experienced a small but not statistically significant increase in QoL and utility three months after seeing a gastroenterologist for IBS, which was not maintained. Gastroenterology referral does not appear to appreciably improve Qol for most people with IBS.
# Background
Irritable Bowel Syndrome (IBS) is a chronic condition characterised by abdominal pain with associated diarrhoea, constipation or both but with no structural abnormality of the bowel. IBS has no attributable mortality [bib_ref] The epidemiology of irritable bowel syndrome, Canavan [/bib_ref] but it is important due to the effect it has on quality of life (QoL) [bib_ref] Impairment in work productivity and health-related quality of life in patients with..., Dean [/bib_ref] [bib_ref] Quality of life and illness costs in irritable bowel syndrome, Simrén [/bib_ref] and the large number of people it effects, with a global prevalence of 11%. [bib_ref] Global prevalence of and risk factors for irritable bowel syndrome: a meta-analysis, Lovell [/bib_ref] IBS causes considerable reductions in all dimensions of QoL. [bib_ref] Impairment in work productivity and health-related quality of life in patients with..., Dean [/bib_ref] The effects on work, social life and ability to travel are the greatest [bib_ref] Performance of the EQ-5D in patients with irritable bowel syndrome, Bushnell [/bib_ref] [bib_ref] The burden of illness of irritable bowel syndrome: current challenges and hope..., Hulisz [/bib_ref] and most IBS patients report at least moderate pain and moderate anxiety or depression. [bib_ref] Performance of the EQ-5D in patients with irritable bowel syndrome, Bushnell [/bib_ref] [bib_ref] The burden of illness of irritable bowel syndrome: current challenges and hope..., Hulisz [/bib_ref] [bib_ref] Health-related quality of life, work productivity, and health care resource utilization of..., Paré [/bib_ref] [bib_ref] A comparison of irritable bowel syndrome patients managed in primary and secondary..., Smith [/bib_ref] Symptoms fluctuate over time, but QoL does not change over three months without any intervention. [bib_ref] Health-related quality of life and cost impact of irritable bowel syndrome in..., Akehurst [/bib_ref] [bib_ref] Developing valid and reliable health utilities in irritable bowel syndrome: results from..., Spiegel [/bib_ref] This morbidity leads to high levels of health care utilization, [bib_ref] Health-related quality of life and cost impact of irritable bowel syndrome in..., Akehurst [/bib_ref] [bib_ref] Irritable bowel syndrome in Germany. A cost of illness study, Müller-Lissner [/bib_ref] [bib_ref] Costs of care for irritable bowel syndrome patients in a health maintenance..., Levy [/bib_ref] [bib_ref] Determinants of healthcareseeking behaviour among subjects with irritable bowel syndrome, Williams [/bib_ref] [bib_ref] Irritable bowel syndrome in a community: symptom subgroups, risk factors, and health..., Talley [/bib_ref] [bib_ref] Economic impact of irritable bowel syndrome: what does the future hold?, Cash [/bib_ref] [bib_ref] Irritable bowel syndrome: a 10-yr natural history of symptoms and factors that..., Ford [/bib_ref] with over half of primary care consultations for IBS being because the patient is not satisfied with their previous treatment. [bib_ref] Determinants of healthcareseeking behaviour among subjects with irritable bowel syndrome, Williams [/bib_ref] This dissatisfaction is the principle reason primary care physicians refer patients with IBS to see a gastroenterologist. [bib_ref] Irritable bowel syndrome: the view from general practice, Thompson [/bib_ref] Uncertainty that IBS is the correct diagnosis is the second most frequent reason. [bib_ref] Irritable bowel syndrome: the view from general practice, Thompson [/bib_ref] This uncertainty seems most common when patients have diarrhoea predominant symptoms. [bib_ref] Quality of life in patients with irritable bowel syndrome seen in referral..., Simrén [/bib_ref] [bib_ref] Irritable bowel syndrome in general practice: prevalence, characteristics, and referral, Thompson [/bib_ref] Consequently, despite guidelines recommending a positive clinical diagnosis and management in primary care,around 20% of IBS patients see a gastroenterologist. [bib_ref] Costs of care for irritable bowel syndrome patients in a health maintenance..., Levy [/bib_ref] [bib_ref] Irritable bowel syndrome in general practice: prevalence, characteristics, and referral, Thompson [/bib_ref] Some studies have assessed how successful seeing a gastroenterologist is for confirming the IBS diagnosis, [bib_ref] The yield of colonoscopy in patients with non-constipated irritable bowel syndrome: results..., Chey [/bib_ref] [bib_ref] A positive diagnostic strategy is noninferior to a strategy of exclusion for..., Begtrup [/bib_ref] but none have addressed how it affects QoL, the most frequent reason for the referral. [bib_ref] Irritable bowel syndrome: the view from general practice, Thompson [/bib_ref] Consequently, we conducted this study to measure how consulting a gastroenterologist for IBS affects QoL. This information is essential to enable clinicians to make appropriate referral decisions and for healthcare commissioners and decision makers to optimise resource allocation.
# Methods
We screened referral letters from general practitioners to the gastroenterology outpatient clinic at the Nottingham Treatment Centre between October 2012 and March 2014 to identify patients likely to have IBS. We excluded patients with previous secondary care attendance for their symptoms mentioned in their referral letters. Potentially eligible patients with referral letters describing symptoms in keeping with a diagnosis of IBS, or who had IBS diagnosed already by the general practitioner were sent a QoL questionnaire before seeing a gastroenterologist. The questionnaire consisted of the EuroQol Questionnaire of 5-Dimensions (EQ-5D) and some supplementary questions that asked about demographics, symptoms, time off work and asked patients to confirm this was their first attendance at a gastroenterology clinic. The QoL data from this instrument is converted easily to a utility score on a scale from 0 (death) to 1 (perfect health) which is necessary for use in economic evaluation. EQ-5D has also been shown to be valid in IBS, it is sensitive to change in disease captured on disease specific instruments [bib_ref] Performance of the EQ-5D in patients with irritable bowel syndrome, Bushnell [/bib_ref] and has good longitudinal validity. [bib_ref] Measuring irritable bowel syndrome patient-reported outcomes with an abdominal pain numeric rating..., Spiegel [/bib_ref] If patients wished to participate, they completed the questionnaire before attending for their appointment and brought it with them. Final diagnosis of IBS was confirmed by checking participants' medical records eight to ten weeks following their clinic appointment. Those found not to have IBS were then withdrawn from the study. Participants with confirmed IBS were sent a second questionnaire three months following their first appointment. A third questionnaire was sent to all the eligible participants one year after their initial clinic appointment, regardless of whether they returned the second questionnaire.
## Exclusion criteria
We did not invite patients aged under 16 or currently inmates in a prison to take part in the study. Patients who declined to participate, were unable to consent, had not completed the first questionnaire adequately to assess EQ-5D (including for reasons of illiteracy in written English) or did not attend the appointment were not recruited. Recruits were excluded from the study if they were diagnosed with a condition other than IBS following their clinic appointment or the clinic appointment was not their first referral for their IBS symptoms. Participants who returned invalid second or third questionnaires had that questionnaire excluded and were sent a replacement.
## Statistics
The EQ-5D health state utility value was calculated using the UK specific valuation algorithm.Baseline characteristics were analysed using simple descriptive statistics. The mean difference in utility, VAS and time off work for each patient before their clinic appointment and at three and twelve months afterwards were analysed using paired t-tests with 95% confidence intervals. We stratified VAS and utility results by sex, age, whether the person had a diagnosis of IBS before they were referred and by the underlying symptoms causing the referral (pain, diarrhoea, constipation or alternating bowel habit). Mean difference from before seeing a gastroenterologist to three and twelve months afterwards was calculated with 95% confidence intervals and significance assessed using paired t-tests.
Ethics approval was granted by the East Midlands Regional Ethics Committee (code: 11/ EM/0298).
# Results
We invited 205 patients to participate after screening referral letters. Of these, sixty declined, 31 did not have an initial questionnaire collected before their appointment as required in our protocol and 22 did not meet the eligibility criteria. Ninety-two patients completed the first questionnaire. Review of their notes after the appointment found that 5 had been seen previously by gastroenterologist and 18 were not diagnosed with IBS. This left 69 patients with IBS referred to see a gastroenterologist for the first time who were recruited to our study. At three months, 29 (42%) returned a follow-up questionnaire and at one year 12 (17%) returned a further questionnaire [fig_ref] Fig 1: Flowchart showing the recruitment process [/fig_ref]. [fig_ref] Table 1: Demographics of all patients invited to participate in this study, those who... [/fig_ref] shows the demographic details of all invited patients, the initial recruits and the responders at each time point. From those invited, 41% of referrals already had a diagnosis of IBS and over half were referred with diarrhoea. Over two thirds of the patients were female, with 73% of those invited and 69% of responders being female. The mean age of all invited patients was 41.4 years (95% CI 39.3 to 43.6) and the mean age of those participating was 40.2 years (95% CI 37.2 to 43.3). The proportions of patients in each age group differed between those invited and the participant sample, with the young somewhat underrepresented amongst participants. Twenty-percent of the responders were diagnosed with a condition other than IBS following their clinic appointment and investigations [fig_ref] Table 2: Conditions diagnosed by a gastroenterologist during outpatient assessment following the first referral... [/fig_ref]. Of these patients, 35% (8 patients) had a previous diagnosis of IBS. Referral symptoms in those diagnosed with something other than IBS were proportionate to the total sample invited [fig_ref] Table 1: Demographics of all patients invited to participate in this study, those who... [/fig_ref].
## Qol domains
Participants reported that pain or discomfort and depression or anxiety were the two domains that contributed to the greatest loss of QoL at each time point [fig_ref] Table 3: EQ-5D responses by domain and median and VAS scores overall utility [/fig_ref]. Only 17% reported no problems with pain before they saw a gastroenterologist, this increased to 38% at three months, but fell again to 17% at one year. Across the three questionnaires, around 8% reported extreme pain or discomfort and around 10% reported extreme depression or anxiety. No participants reported extreme problems with mobility, self-care or activities of daily living at any time point [fig_ref] Table 3: EQ-5D responses by domain and median and VAS scores overall utility [/fig_ref].
## Overall utility
The median overall utility score for the cohort before seeing a gastroenterologist was 0.76 (Interquartile range [IQR] 0.69 to 0.80). Three months after the gastroenterology appointment, 44% had improved utility, 42% experienced no change and 14% had worse utility [fig_ref] Fig 2: Proportion of respondents reporting a change in their utility from baseline [/fig_ref]. The mean utility increased by 0.04 to 0.80 (IQR 0.62 to 1.00). One year after the appointment, a third of responders had improved utility from the baseline, a third experienced no change and a third had worse utility [fig_ref] Fig 2: Proportion of respondents reporting a change in their utility from baseline [/fig_ref]. The mean utility fell by 0.07 from the 3 month peak to 0.73 (IQR 0.65 to 0.76), a 0.04 decrease from the pooled baseline. None of these changes were statistically significant, however. When responses at three and twelve months were compared to the baseline response of the same patients, as opposed to the whole cohort initially recruited, the mean utility three months after seeing a gastroenterologist had increased by 0.06 (95% CI -0.01 to 0.14), but this change reduced to 0.00 (95% CI -0.16 to 0.16) after a year . When the results were stratified, the two groups who showed sustained increased utility were those aged over 50 years and those referred with diarrhoea. Men and women had similar utility at baseline, at three months men has greater mean improvement but this was not maintained at one year. At baseline, utility was higher in patients aged under 30. At three months those aged over 30 years reported greater mean improvement in utility which was sustained at one year, whilst those aged under 30 reported a mean decrease in utility. None of these findings were statistically significant. No changes in utility at three or twelve months were statistically significantly different from before seeing the gastroenterologist . Mean values have been reported in [fig_ref] Table 3: EQ-5D responses by domain and median and VAS scores overall utility [/fig_ref] to make the results amenable to future cohort modelling.
## Global qol (vas)
Before seeing a gastroenterologist, the median VAS score for the cohort was 67.5 (IQR 50.0 to 80.0). Three months after the gastroenterology appointment, this increased to 75.0 (58.0 to 89.0), but had fallen to 60.0 (IQR 50.0 to 80.0) at one year. The mean change in VAS for individuals from before seeing a gastroenterologist to three months afterwards was not statistically significant nor was it one year later [fig_ref] Table 5: Median overall VAS score at each time point stratified by demographic variables,... [/fig_ref]. [fig_ref] Table 5: Median overall VAS score at each time point stratified by demographic variables,... [/fig_ref] shows that men experienced greater mean improvement in their global QoL three and twelve months after their gastroenterology appointment. Patients referred with diarrhoea also reported greater improvement in global QoL at both time points, whilst those referred with constipation or alternating bowel habit reported decreased global QoL after three and twelve months. None of the changes within the strata were statistically significant, however.
## Time off work
In the month preceding the first appointment with a gastroenterologist, the median number of days patients with IBS took off work (or away from their usual daily activities if they were not employed) for symptoms related to IBS was zero (IQR 0 to 4). Fifty-five percent of patients took no time off work due to their IBS and for those who did take time off the median time off work in a month was 4 days (IQR 2 to 14). Five patients reported that their IBS meant they were off work for the whole month. There was no statistically significant difference in the number of days patients took off work before seeing a gastroenterologist and afterwards, either at three or twelve months.
# Discussion
This is the first study to examine the effect of gastroenterology referral on QoL of patients with IBS. We found that in this group there was on average low initial quality of life, which improved somewhat 3 months after consultation, but declined to baseline after a year. We also found that in this group, who were identified as being likely to have IBS by their GPs, or from screening letters by a gastroenterology Registrar (CC), almost 20% had another organic gastrointestinal diagnosis made. Although our study is not directly comparable to any other, and so arguably gives the best available measure of the benefit of gastroenterology referral in IBS patients, its quality needs to be examined if our results are to be correctly interpreted. We have based our work in a large teaching hospital where both sub-specialist neuro-gastroenterologists and more general luminal gastroenterologists cared for the recruited patients. To ensure that our results will be generalisable to UK secondary care referrals we therefore excluded anyone who had previously been seen in secondary care for their problem. To generate utility values alongside QoL measures and ensure that our results can be compared to others in IBS as well as other conditions, we used the generic EQ-5D questionnaire. This has been validated against the disease specific instruments for IBS [bib_ref] Performance of the EQ-5D in patients with irritable bowel syndrome, Bushnell [/bib_ref] and the utility scores are widely used and accepted for health economic evaluations.
[27] Our study has an obvious weakness in its response rates, however, particularly at one year. This level of loss to follow up inevitably raises concerns about bias in the non-response. Though we cannot exclude the possibility of bias, when examining the known characteristics of respondents and non-respondents at each stage the groups appear similar, so such bias is at least not obvious. Despite this, it is still possible that those who are lost to follow-up have all been sufficiently reassured or had their management optimised following the gastroenterology appointment resulting in considerable improvement in QoL. If this is the case, then our study underestimates the positive impact of a referral to gastroenterology for patients with IBS. If the mean utility change for all respondents at each time point is assessed, then utility increased by 0.04 over the first three months and then decreased by 0.07 over the next nine months, so that it was 0.03 lower than baseline at one year. The magnitude of the increase in utility at three months is the same when the baseline is restricted to only those responding at three months. The magnitude of the decrease at 12 months is reduced, however. This means that the baseline utility for those patients responding at 12 months had a lower mean than those who were lost to follow-up. It is possible, therefore, that there is a response bias due to the respondents lower baseline utility representing patients who are less likely to benefit from a gastoenterology appointment. This would need further assessment in future studies. A further problem consequent upon the loss to follow up is that we have very limited power. We are therefore unable to be certain that the non-significant increase in QoL we found at 3 months is not important. In fact, the minimum clinically important difference in utility score using the EQ-5D in IBS is only 0.03, [bib_ref] Performance of the EQ-5D in patients with irritable bowel syndrome, Bushnell [/bib_ref] so the 0.06 point mean increase which is our central estimate would clearly be important if true. Yet the results at one year suggest that even if there is a real effect on QoL three months after referral it is likely to be transient. Finally, the low power limits our ability to assess differences in subgroups of IBS patients. Notwithstanding this, the non-significant benefit we found was more pronounced in older patients with diarrhoea.
Even though there are no directly comparable studies yet published, our results should be considered in conjunction with what is already known. In keeping with other studies that report greater referral of patients with diarrhoea predominant symptoms rather than constipation or alternating bowel function [bib_ref] Irritable bowel syndrome: the view from general practice, Thompson [/bib_ref] [bib_ref] Irritable bowel syndrome in general practice: prevalence, characteristics, and referral, Thompson [/bib_ref] [bib_ref] Prevalence of functional gastrointestinal disorders in consecutive new patient referrals to a..., Shivaji [/bib_ref] , a higher proportion of patients in our study had diarrhoea predominant IBS than is seen in primary care. [bib_ref] The epidemiology of irritable bowel syndrome, Canavan [/bib_ref] Likewise, almost 20% of the 92 patients who completed a first questionnaire were diagnosed with a condition other than IBS, which is similar to a previous report in the UK. [bib_ref] Prevalence of organic disease at colonoscopy in patients with symptoms compatible with..., Patel [/bib_ref] In that study, between 15% and 28% of similar patients undergoing colonoscopy had a diagnosis other than IBS apparently responsible for their symptoms. [bib_ref] Prevalence of organic disease at colonoscopy in patients with symptoms compatible with..., Patel [/bib_ref] These results are at variance however with those from America where a structural lesion has been reported in over 40% of patients with IBS who received a colonoscopy. [bib_ref] The yield of colonoscopy in patients with non-constipated irritable bowel syndrome: results..., Chey [/bib_ref] The proportion with such lesions, however, was not significantly different to that found in screening colonoscopy in the general population and only changed the diagnosis of IBS in 2% of patients. [bib_ref] The yield of colonoscopy in patients with non-constipated irritable bowel syndrome: results..., Chey [/bib_ref] In contrast to the higher American prevalence of apparent organic disease, in Denmark organic disease was identified in only 10% of patients diagnosed with IBS . Median overall utility score at each time point stratified by demographic variables, with mean difference from baseline and the p-value from a paired t-test. Clinically significant mean changes are in bold. No results were statistically significant at the 5% level. who subsequently received extensive gastroenterological investigation. [bib_ref] A positive diagnostic strategy is noninferior to a strategy of exclusion for..., Begtrup [/bib_ref] In the Danish study, however, patients were randomly assigned gastroenterological investigation as opposed to having been referred for further investigation as in our own. Across Europe and North America utility values between 0.62 and 0.75 have been reported by patients with IBS [bib_ref] Performance of the EQ-5D in patients with irritable bowel syndrome, Bushnell [/bib_ref] [bib_ref] Health-related quality of life, work productivity, and health care resource utilization of..., Paré [/bib_ref] [bib_ref] Developing valid and reliable health utilities in irritable bowel syndrome: results from..., Spiegel [/bib_ref] (equivalent to an average patient being willing to sacrifice between 10 and 15 years of their remaining life expectancy for an immediate cure [bib_ref] Developing valid and reliable health utilities in irritable bowel syndrome: results from..., Spiegel [/bib_ref]. Our baseline findings are comparable, with a median EQ-5D utility value of 0.76 (0.69 to 0.80). This is 0.10 points lower than the UK population average,consistent with the reduction previously reported in IBS. [bib_ref] Performance of the EQ-5D in patients with irritable bowel syndrome, Bushnell [/bib_ref] In our study, only 74% of patients had at least some pain at referral. This might seem low given that the presence of pain is required to meet the Rome III criteria. It is however slightly higher than a previous UK community based study in which only two-thirds reported some pain, [bib_ref] A comparison of irritable bowel syndrome patients managed in primary and secondary..., Smith [/bib_ref] perhaps suggesting that complaints of pain affecting QoL increase the likelihood of referral. Pain was less common in a recent study of patients referred to a gastroenterologist with functional gastrointestinal conditions, [bib_ref] Prevalence of functional gastrointestinal disorders in consecutive new patient referrals to a..., Shivaji [/bib_ref] though the inclusion of conditions not requiring pain for diagnosis could clearly affect this result. Our finding that almost half experienced at least moderate depression and anxiety was the same as found in the UK community study. [bib_ref] A comparison of irritable bowel syndrome patients managed in primary and secondary..., Smith [/bib_ref] Despite the UK National Institute for Health and Clinical Excellence (NICE) recommending the measurement of utilities using the EQ-5D[27] for health care decision making, there are very few studies that have reported health utilities in IBS using the EQ-5D. Current guidelines are therefore based on models calculated from assumed utility changes.Since these guidelines were published there have however been some studies reported of utility change following interventions in patients with IBS. [bib_ref] Economic evaluation of tegaserod vs. placebo in the treatment of patients with..., Bracco [/bib_ref] [bib_ref] Cost-effectiveness of acupuncture for irritable bowel syndrome: findings from an economic evaluation..., Stamuli [/bib_ref] [bib_ref] Randomised clinical trials: linaclotide phase 3 studies in IBS-C-a prespecified further analysis..., Quigley [/bib_ref] [bib_ref] Eluxadoline benefits patients with irritable bowel syndrome with diarrhea in a phase..., Dove [/bib_ref] Only one of these found a change in utility which was both clinically meaningful and statistically significant, and that was from the use of a serotonin-receptor partial agonist in patients with constipation predominant IBS. [bib_ref] Economic evaluation of tegaserod vs. placebo in the treatment of patients with..., Bracco [/bib_ref] All the other studies like our own have been unable to demonstrate such changes, [bib_ref] Cost-effectiveness of acupuncture for irritable bowel syndrome: findings from an economic evaluation..., Stamuli [/bib_ref] [bib_ref] Randomised clinical trials: linaclotide phase 3 studies in IBS-C-a prespecified further analysis..., Quigley [/bib_ref] [bib_ref] Eluxadoline benefits patients with irritable bowel syndrome with diarrhea in a phase..., Dove [/bib_ref] perhaps suggesting that few if any available interventions improve global QoL or overall utility in patients with IBS. [bib_ref] Systematic review: the efficacy of treatments for irritable bowel syndrome--a European perspective, Tack [/bib_ref] Summary and conclusions
In an era of health austerity, it will be increasingly necessary to demonstrate the efficacy of our care. We have shown a small, non-significant, but potentially clinically meaningful mean increase in QoL for IBS patients three months after seeing a gastroenterologist. This improvement was not maintained at one year, however. Although larger studies, in particular randomised control trials, of this complex intervention may in future provide better evidence, at present the most important benefit from referral that we have been able to demonstrate may be the diagnosis of other organic pathology. Until better data are available however, the figures we provide now permit some assessment of the cost/utility of referring patients with IBS to gastroenterology services.
Supporting Information S1 Checklist. STROBE checklist.
(DOCX)
[fig] Fig 1: Flowchart showing the recruitment process. [/fig]
[fig] Fig 2: Proportion of respondents reporting a change in their utility from baseline (remaining responders reported no change) at three and twelve months with the mean utility score change and 95% confidence intervals.doi:10.1371/journal.pone.0139389.g002 [/fig]
[table] Table 1: Demographics of all patients invited to participate in this study, those who agreed to participate, those diagnosed with something other than IBS and the responders in each round who contributed the questionnaires to the analysis. doi:10.1371/journal.pone.0139389.t001 [/table]
[table] Table 2: Conditions diagnosed by a gastroenterologist during outpatient assessment following the first referral of patients potentially with IBS from primary care. [/table]
[table] Table 3: EQ-5D responses by domain and median and VAS scores overall utility. [/table]
[table] Table 5: Median overall VAS score at each time point stratified by demographic variables, with mean difference from baseline and the p-value from a paired t-test for significance of the mean change.doi:10.1371/journal.pone.0139389.t005 [/table]
|
10.1007/s10653-022-01240-w
|
CCBY
|
9675645
|
35262826
|
s2orc_pubmed_articles
|
Potential fluoride exposure from selected food crops grown in high fluoride soils in the Makueni County, south-eastern Kenya
determination of 20 samples. Mean F concentration in the food crops was in the order; 700, 288, 71.2, 36.6, and 29 mg/kg in kale, cowpeas leaves, green grams, cowpeas (legume portion), and maize, respectively. The F concentration in farm soils ranged from 0 to 3.47 mg/kg (mean of 0.87 mg/kg) and showed a significant strong positive correlation (p = 0.03, r = 0.89) with F values in the crops. Apatite, muscovite, and biotite were identified as the F-rich minerals present. While considering two hypothetical F absorption fractions (75 and 100%), the estimated average daily dose (EADD) of F from consuming the crops ranged between 0.004 and 65.17 mg/kg/day where the highest values were from the vegetables. Most of these values were higher than the F reference dose (RfD) of 0.06 mg/kg. The estimated EADD values of several hypothetical meals prepared from the analyzed crops revealed that steamed kale and maize porridge pose the highest health risk of F associated diseases to the local population, whereas boiled cowpeas pose no health risk. Children, due to their higher daily energy requirement and low body weight, were the most vulnerable group at risk of high daily F intake relative to the RfD. These results suggest that consumption of the analyzed food crops in Makueni County may significantly contribute to F related diseases in the local population. This creates a food security issue for the area because of the potential health risks associated with these crops which are highly relied upon in the semi-arid area with a limited selection of food crops available and viable to grow.Abstract Makueni County, located in south-east
# Introduction
Food security and good health are among the United Nations sustainable development goals . Attaining these goals means ensuring the provision of sufficient nutritious food to the population that is free from contaminants [bib_ref] Attaining sustainable development goals in sub-saharan Africa the need to address environmental..., Omisore [/bib_ref]. In Kenyan rural areas, about 70% of the consumed food is produced locally [bib_ref] Food and nutrition scenario of Kenya, Mohajan [/bib_ref] [bib_ref] Determinants of household food security in Lugari and Makueni Sub-counties, Mungai [/bib_ref] hence, the local environment directly influences the populations health status. Globally, there is evidence and awareness of the adverse health issues from chemically contaminated food crops. Among these elements is fluoride (F), a widely known natural groundwater contaminant, whose health risks in food crops are not well documented [bib_ref] Occurrence of fluorosis in a population living in a high-fluoride groundwater area:..., Gevera [/bib_ref] [bib_ref] Fluoride contamination of selected food crops, domestic water, and milk consumed by..., Memba [/bib_ref] [bib_ref] Impacts of soil and water fluoride contamination on the safety and productivity..., Rizzu [/bib_ref].
F in crops grown in contaminated soils and/or irrigated with high F water could exacerbate its health effects on the consumers [bib_ref] Distribution of aluminum and fluoride in tea plant and soil of tea..., Xie [/bib_ref]. Fluoride accumulation in plants is controlled by factors such as plant species and the availability of soluble F in farm soils [bib_ref] Assessment of potential health risk of fluoride consumption through rice, pulses, and..., Bhattacharya [/bib_ref]. Weathering of F-rich rocks releases F in soils where it is absorbed by the food crops through root transfer. Its mobility is governed by factors including soil particle size, pH, the presence of organic matter, and clays [bib_ref] Assessment of fluoride content in tropical surface soils used for crop cultivation, Abugri [/bib_ref] [bib_ref] Analysis of particle size distribution of landfill contaminated soils and their mineralogical..., Nyika [/bib_ref]. Understanding these factors is crucial in controlling F dose received by local populations to improve the health status in high F regions.
Makueni County a semi-arid region in south-eastern Kenya, has low and unreliable rainfall patterns with limited surface water resulting in high dependence on groundwater sources for domestic and agricultural use [bib_ref] Naturally occurring potentially harmful elements in groundwater in Makueni County, South-Eastern Kenya:..., Gevera [/bib_ref] [bib_ref] The influence of precambrian metamorphic rocks on groundwater in the Chyulu area..., Mailu [/bib_ref] [bib_ref] Groundwater quality assessment and water quality indexing: Case study of Makueni County,..., Ng'ang'a [/bib_ref]. F levels up to 7.17 mg/L in these water sources as well as cases of dental fluorosis have been reported in the area (e.g., [bib_ref] Naturally occurring potentially harmful elements in groundwater in Makueni County, South-Eastern Kenya:..., Gevera [/bib_ref].
Despite the concerns about the high F content in groundwater used for irrigation in the study area, to the best of our knowledge, this is the first study investigating the F concentrations in these food crops, in the farm soils where the crops are grown, as well as the potential health risks from consuming these foods. The purpose of this paper is to address this significant knowledge gap by linking F concentrations in farm soil to the mineralogy of the soil and some selected commonly grown and consumed food crops in the area. The exposure dose and health risk associated with F in the food samples were also calculated using standard methodologies.
## Study area
Location, population, and climate Makueni County is located in the south-eastern region of Kenya and borders three counties, Machakos, Kajiado, Taita-Taveta, and Kitui [fig_ref] Figure 1: The geological map of Makueni County showing the Makindu-Kibwezi area, bounded in... [/fig_ref]. The county's population is approximately 884,500 where most of the people live in a predominately rural setting (Kenya National Bureau of . This paper reports a study conducted in the county's southern region covering the Makindu-Kibwezi area as demarcated by the black outline in [fig_ref] Figure 1: The geological map of Makueni County showing the Makindu-Kibwezi area, bounded in... [/fig_ref]. The area was recently assessed for its drinking water quality and the health implications of F and salinity on the local population [bib_ref] Naturally occurring potentially harmful elements in groundwater in Makueni County, South-Eastern Kenya:..., Gevera [/bib_ref].
Unreliable rainy seasons produce an average annual rainfall range between 200 and 700 mm, [bib_ref] Determinants of access and utilisation of seasonal climate information services among smallholder..., Muema [/bib_ref] , where planting usually occurs during the two rainy seasons in March-April and November-December. Irrigation is practiced in farms close to rivers, springs, and in few farms that use borehole water. Common food crops grown and consumed in the area include leaf and legume cowpeas (Vigna unguiculata), beans (Phaseolus vulgaris), maize (Zea mays), green grams (Vigna radiata), kale (Brassica oleracea var. viridis), cabbage (Brassica oleracea var. capitata), mangoes (Mangifera indica), and watermelons (Citrullus lanatus).
## Geology
Makueni County falls under the metamorphic Mozambique Mobile Belt (MMB) in Kenya [bib_ref] A reappraaisal of the geology geochemistry structures and tectonics of the Mozambique..., Nyamai [/bib_ref]. The geology of the area is dominantly metasediments composed of biotite, muscovite, and hornblende schists and gneisses as well as granitoid gneisses and granites, which are overlain by Pleistocene basalts flows and volcanic cones in the southern region [bib_ref] A reappraaisal of the geology geochemistry structures and tectonics of the Mozambique..., Nyamai [/bib_ref] as shown in [fig_ref] Figure 1: The geological map of Makueni County showing the Makindu-Kibwezi area, bounded in... [/fig_ref]. These rocks are rich in biotite, muscovite, and apatite which are potential F-hosting minerals. Soils in the area range texturally from clayey, sandyclay loam to sandy-clay and have a general porous massive structure [bib_ref] Small scale digital soil mapping in Southeastern Kenya, Mora-Vallejo [/bib_ref].
# Materials and methods
Sampling, preparation, and analysis All collected samples (soil and food crops) were transported to South Africa where preparation and analysis were conducted at the University of Johannesburg, Geology Department, and Agricultural Research Council (ARC) laboratories in Pretoria.
## Sample collection and preparation
Food samples Five commonly grown and consumed food crop samples, including three grain-based (maize, green grams, and cowpeas) and two vegetables (kale and cowpeas leaves), were collected in five farms in the southern region of Makueni County [fig_ref] Figure 1: The geological map of Makueni County showing the Makindu-Kibwezi area, bounded in... [/fig_ref] in January 2020. The samples include the edible parts of the crops (grains and leaves) as shown in . Grain and leaves crop samples were collected to determine how F concentrates in these two food crop types commonly consumed in the area. The samples were collected close to previously sampled [bib_ref] Naturally occurring potentially harmful elements in groundwater in Makueni County, South-Eastern Kenya:..., Gevera [/bib_ref] drinking water sources which showed elevated F levels up to 7.17 mg/l [bib_ref] Naturally occurring potentially harmful elements in groundwater in Makueni County, South-Eastern Kenya:..., Gevera [/bib_ref]. For each food crop, samples were collected from four spots in the specific farm and mixed to attain a representative sample size of 100 g. The food samples were then stored in plastic bags kept refrigerated until they were ready to submit to the ARC laboratories for analysis.
At the laboratory, the samples were crushed and sieved through a 1.4 mm mesh size to attain uniformity and oven-dried for two hours.
Soil samples 30 soil samples were collected in 30 farms, including those where food crops were sampled from, in December 2018 (20 samples) and January 2020 (10 samples). Farms sampled did not Samples of maize a green grams b cowpeas c grains and leaves of cowpeas leaves d and kale e from Makueni County that were collected for F analysis use fertilizers. Samples were collected from four locations in each farm and thoroughly mixed to get a homogenous representative sample size of approximately 500 mg. Each sample was collected by digging a 30 cm hole after the top layer of about 3 cm was removed to prevent the collection of plant material and other debris. The sampling depth chosen was within the root zone of the sampled food crops. Sampling was done distal to homesteads, buildings, and roads to avoid any anthropogenic contamination. The samples were kept in tightly closed plastic bags and later air-dried overnight.
A portion of the soil samples collected during the first field visit (n = 20) was prepared for XRD analysis at the Spectrum facility of the Faculty of Science at the University of Johannesburg to determine the presence of F-rich minerals in the farm soil which would warrant further F analyses in the soil samples. The samples were air-dried overnight and then crushed using an auger milling machine for three minutes to attain a 100 μm sieve size before analyzed. Subsequently, after ten more samples were collected during the second field visit, all the soil samples (n = 30) were submitted to the ARC laboratories for the analysis for water-soluble F concentration.
# F analysis
At the ARC laboratories, the five food crops (one sample for each crop) and 30 soil samples were analyzed for F concentration. For both soil and food crop samples, the partial leaching method was used to dissolve the water-soluble F from the soil. The concentrations of F were determined using the EPA Method 300.0.
The 1:10 (sample: water) ratio was used for partial dissolution. Five grams of each sample was weighed into a glass beaker and mixed with 50 ml of deionized water. The mixture was shaken in a mechanical shaker for 30 min to liberate the water-soluble F fraction in the samples. The mixture was then filtered into a volumetric flask, and the concentration of F was determined, in triplicate, using a Dionex ICS 1600 ion chromatograph. The standard solution was prepared by dissolving 2.21 g of sodium fluoride (NaF) in deionized water and diluted to 1 L.
# Mineralogical analysis
The semi-quantitative abundance of minerals in 20 samples of farm soil material was determined using the Panalytical X'Pert powder diffractometer. The milled soil samples were pressed into sample holders and loaded into the XRD instrument. Operational conditions were: Cu-K α radiation with the generator settings of 40 mA and 40 kV, and scanning angle range of 4.01 2 -89.98 2θ. The data were collected by the X'Pert-Pro data collection software and interpreted using the X'Pert HighScore Plus software package.
# Data analysis
F transfer factor (TF) The concentration of F in the selected food crops is correlated to the water-soluble F values in the farm soil where the crops were grown to determine the transfer factor. The determination of the transfer of potentially harmful elements such as F from the soil to food crops is essential to advise what crops are suitable for cultivation in these specific soils [bib_ref] Fluoride accumulation in crops and vegetables and dietary intake in a fluorideendemic..., Gupta [/bib_ref]. The TF values were calculated using the following equation,. TF = Concentration of element in crop-vegetable (mg/kg)/ concentration of element in soil (mg/kg).
F exposure dose F exposure dose associated with consumption of the analyzed food crops is essential to establish the health risk to the local population. The F estimated average daily dose (EADD) linked to the consumption of the five analyzed food crops was calculated from the following equation, [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref] :
[formula] EADD F = C x DCI x FEC x AF/E. Where; C is the concentration (mg/kg) of F in the different food crops analyzed. [/formula]
DCI is the daily calorie intake per person: the values of 90 kcal per day/kg of body weight which is the recommended energy consumption for children below one year and 24 kcal per day/kg of body weight recommended for adolescents and adults [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref] were used.
FEC is the food energy contribution of each analyzed food crop. The values of 47% for cereals (maize), 2% for grain legumes, and 3% for vegetables were used, based on their energy contribution to the diet of a study population in Kitui County, north of Makueni [bib_ref] Dietary patterns, food and macronutrient intakes among adults in three ethnic groups..., Hansen [/bib_ref].
AF is the absorption fraction of F in in the gastrointestinal tract, where the hypothetical rates of 75 and 100% [bib_ref] Assessment of potential health risk of fluoride consumption through rice, pulses, and..., Bhattacharya [/bib_ref] [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref] is adopted. These are the two limits at which F ingested in food is absorbed in the body [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref].
E is the energy contribution of each specific analyzed food crop (Kcal/Kg). The energy contribution (E) of most locally consumed foods in Kenya is reported by FAO and the Government of Kenya. The E values used (Kcal/kg) for the analyzed food crops (raw and dry) were 3450, 3110, 3130, 30, and 29 for maize, green grams, cowpeas legumes, cowpeas leaves, and kale, respectively.
Calculation of the F EADD values of cooked meals Given that the analyzed food crops are prepared and consumed in several forms by the population in the Makueni County, we determined the F dose associated with different meals commonly consumed in the area. This is because, meals prepared from the same crop or ingredient have different energy values (E), depending on the cooking method, which will in turn impact the F EADD value. A study by FAO and the Government of Kenya (FAO & Government of Kenya, 2018) reports the metabolizable energy values (E) of commonly consumed meals in Kenya.
In the study, the nutrition values of 100 g dry weight of several foods cooked in different ways were determined (FAO & Government of Kenya, 2018). The E values from cooking methods including boiling, steaming, and stir frying (for vegetables) we used to determine the EADD of the analyzed food crops in this study. The methods used in preparing the meals include boiling and draining the water after the meals were cooked, then the nutritional value of the dry cooked meals were determined (FAO & Government of Kenya, 2018).
The energy (E) values in used for the EADD calculations in the current study, as obtained from the FAO and the Government of Kenya (FAO & Government of Kenya, 2018) study are; 1460 for boiled maize, 1410 and 520 for whole maize floor meal (ugali) and porridge, 1160 and 1090 for boiled and stewed green grams, 1170 and 1320 for dry boiled and fresh boiled cowpeas, 280 for cowpeas leaves, and 1150, 1050, and 540 for boiled, steamed, and stir-fried kale as obtained from the Kenyan food nutritional data.
F health risk assessment The hazard quotient (HQ) associated with the five analyzed food crops was calculated to determine the individual health risk of food dietary F intake using the following formula [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref] , HQ = EADD/RfD. Where; RfD is the reference dose for humans associated with the 'no observable adverse effect level' (NOAEL) for F, which is set as 0.06 mg/kg by USEPA. Then, the cumulative hazard index (HI) was evaluated by adding the HQ of all the five food crops analyzed.
H I = H Q m a i z e + H Q c ow p e a s + H Q g r e e n grams + HQkale + HQcowpeas leaves.
It should be noted that, the calculated HI is an estimate of only the five analyzed food crops and addition of other F sources into the diet, such as drinking water and beverages, would increase the risk.
Statistical analyses Several statistical analyses were employed to determine the relationships between the different variables in the dataset, these include:
Bar graphs used to visually present F concentrations in different food crops, Spearman's correlation analysis was used to determine the strength (r value) and significance (p value) of correlation between variable such as F concentrations in farm soils and food crops, Linear correlation graphs were used to correlate the F EADD and the daily energy intake per Kg of body weight of the analyzed food crops to determine the possible health effects of F due to the food consumption. Microsoft Excel and IBM SPSS Statistics 25 were used to determine the statistical characteristics of the data.
# Results
## F concentration in food crops
The concentrations of F in the five food crop samples, as dry weight, are presented in . The order of concentrations increased from grains to vegetables. For the grains, maize had the lowest F concentration of 29 mg/kg, followed by cowpeas (36.6 mg/kg), while green grams had the highest value (71.2 mg/ kg). F concentrations in the two vegetables ranged between 388 mg/kg in cowpeas leaves and 700 mg/ kg in kale.
## F concentration in the farm soil
The water-soluble F concentration of 30 soil samples from farms in the area ranged from 0 to 3.47 mg/ kg with a mean value of 0.87 mg/kg and a standard deviation of 0.98. This F concentration represent a fraction readily available for plant uptake [bib_ref] Fluoride in the Context of the Environment, García [/bib_ref]. The mineralogical composition of the soil was analyzed to indicate the possible geogenic source of the water-soluble F observed in the farm soil in the area. The total semi-quantitative mineral compositions of the analyzed soil samples, their range, average, and standard error are shown in [fig_ref] Table 1 The: Fluoride concentration in the five food crops, the five farm soil where... [/fig_ref]. According to , apatite, biotite, and muscovite are the likely F-hosting minerals present in the soil. These authors reported that F occurs in mineral lattice in these minerals and will only be available for plant absorption upon weathering of these minerals. In the studied soil samples, these minerals (apatite, biotite, and muscovite) make up to 23% of the total mineral composition (Table 1), indicating that they are possible geogenic sources of F.
## F transfer factor (tf)
The concentrations of water-soluble F in the soils from the five farms where the analyzed food crops were collected are shown in . The transfer factor of F in all the analyzed food crops ranged from 26 to 257 . There was a significant strong positive (p = 0.03, r = 0.89) correlation between the concentration of water-soluble F in farm soil and F concentration in food crops grown in those soils. A TF value greater than 1 indicates a high accumulation factor of F in the crops [bib_ref] Fluoride accumulation in crops and vegetables and dietary intake in a fluorideendemic..., Gupta [/bib_ref].
F exposure dose and health risk assessment i) Exposure dose in the raw food crops
The estimated F average daily dose (EADD) of the analyzed food crop samples at the two hypothetical absorption limits of 75 and 100% against the daily calorie intake (DCI) range of 20 kcal for adults and 90 kcal for children are presented in . The doses were divided into grain-based foods and vegetables . Among the grains, both cowpeas and green grams had lower EADD values than the recommended F RfD values of 0.06 mg/Kg at both hypothetical absorption rates. Cowpeas values ranged from 0.004-0.016 between 20 and 90 kcal Fluoride concentration in selected five food crops including grains (maize, green grams, and cowpeas) in brown and vegetables (cow peas leaves and kale) in blue from Makueni County DCI, whereas green grams values ranged from 0.007-0.03 between the two DCI values. In the 75% absorption rate, maize had a lower EADD value (0.05) compared to the F RfD in the 20-kcal energy intake, but it had a higher value of 0.27 in the 90 kcal energy intake. In the 100% absorption rate, maize had higher EADD values (0.08-0.36) than the RfD in both the 20 and 90 kcal energy intake range.
Considering the two vegetables, both kale and cowpeas leaves had higher EADD values than the RfD of 0.06 mg/kg in both the 20 and 90 kcal energy intakes We determined the hypothetical F dose in different meals prepared from the sampled and analyzed food crops as highlighted in the data analyzed section. The resultant EADD F values calculated using the E values (FAO & Government of Kenya, 2018) of the various meals against the daily energy intake range of 20 kcal and 90 kcal are presented in .
The common maize-based foods whose E values were reported by the Government of Kenya (FAO & Government of Kenya, 2018) include stewed dry maize, maize floor meal (ugali), and maize porridge.. The F intake was considered at the two hypothetical absorption rates of 75% and 100% [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref] in the top and bottom frames, respectively
## Fig. 5
The EADD due to the consumption of different meals prepared from the three analyzed grain-based food crops at different levels of daily energy intake per Kg of body weight, at the two hypothetical absorption rates of 75% and 100% from Makueni County All the maize-based foods had higher EADD values than the RfD of 0.06 mg/kg within both the 20 kcal and 90 kcal energy intakes at both the 75 and 100% absorption rates . Maize porridge had the highest EADD values at both 75 and 100% absorption rates (0.39-1.77 and 0.52-2.36, respectively), followed by stewed dry maize (0.18-0.83 and 0.25-1.11, respectively), whereas maize floor meal (ugali) had the lowest values ranging (0.15-0.65 and 0.19-0.87, respectively). The energy content (E) of dry boiled cowpeas was the only cowpeas meal reported. Within both the 75% and 100% absorption rates, the EADD values of boiled cowpeas were below the F RfD value of 0.06 mg/kg . The values were between 0.01-0.04 and 0.01-0.06 in the two absorption rates, respectively. The green grams meal used was boiled. There was a similar trend in EADD values in these two meals . At the 75% absorption rate, the EADD values ranged from 0.02 to 0.09, where values below the RfD value were between the 20 kcal and 60 kcal energy intakes. In the 100% absorption rate, the EADD values ranged between 0.02 and 0.12, where values below the RfD were observed between the 20 kcal and 50 kcal energy intakes .
Among the vegetables, the E values of boiled, steamed, and stir-fried kale and boiled cowpeas leaves were reported . All the kale meals had EADD values higher than the RfD of 0.06 mg/ kg in both the 75 and 100% absorption rates . Steamed kale had the highest EADD values (1.26-5.67 and 1.68-7.56) in the two absorption rates followed by boiled (1.13-5.06 and 1.5-6.75) and stir-fried (0.58-2.63 and 0.78-3.5) meals. Similarly, boiled cowpeas leaves had higher EADD values than the RfD in both absorption rates (0.62-2.81 and 0.83-3.74) .
The Hazard Quotient (HQ) and cumulative hazard index (HI) of all the analyzed food samples (uncooked) are presented in [fig_ref] Table 3: The hazard quotient and cumulative hazard index of the five analyzed food... [/fig_ref]. All the cumulative HI values between the two absorption rates (75 and 100%) were higher than the 'no adverse effect' level of 1. The cumulative HI values ranged between 279 and 1256 in the 75% F absorption rate and 372 and 1675 in the 100% F absorption rate. Kale and cowpeas leaves had the highest HQ values and thus were the major contributors to the high values of cumulative HI observed. The EADD due to the consumption of different meals prepared from the two analyzed vegetables at different levels of daily energy intake per Kg of body weight, at the two hypothetical absorption rates of 75% and 100% from Makueni County
# Discussion
## F concentrations in food crops
Vegetables (cowpeas leaves and kale) showed higher F concentrations (388 to 700 mg/kg) than the grainbased crops (29 mg/kg in maize to 71.2 mg/kg in green grams). These concentrations are also higher than those (7-215 mg/kg) reported in kale, cabbage, amaranth, pumpkin, and cowpeas leaves from other parts of Kenya [bib_ref] Fluoride content of common vegetables from different parts of Kenya, Owuor [/bib_ref] , as well as those reported in cabbage (296 mg/kg) from India [bib_ref] Environmental fluoride level in Anekal Taluk of Bangalore district, Begum [/bib_ref] [bib_ref] Fluoride accumulation in crops and vegetables and dietary intake in a fluorideendemic..., Gupta [/bib_ref] [bib_ref] Fluoride consumption in endemic villages of India and its remedial measures, Saxena [/bib_ref] and in curly kale, endive, and lettuce (40-146 mg/kg) from the Netherlands.
## F concentration and sources in farm soil and transfer factor (tf)
The concentration of soil water-soluble F (up to 3.47 mg/kg) was higher than those reported in farm soils in Zaria, northern Nigeria (up to 0.2 mg/kg) [bib_ref] Fluoride content of soil and vegetables from irrigation farms on the bank..., Okibe [/bib_ref] and in Pennsylvania, USA (up to 1.5 mg/kg) [bib_ref] Fluorine in agricultural soils of Southeastern Pennsylvania, Gilpin [/bib_ref]. However, higher values (up to 133.1 mg/Kg) were reported in farm soils in northern Tanzania [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref].
The probable source of water-soluble F in farm soils in the study area is geogenic. This origin is supported by the presence of F-bearing minerals such as apatite, muscovite, and biotite, which make up to 23% of the soil mineralogical composition. This links to studies which consider biotite, muscovite, and apatite as the main sources of F in granitic regions based on the rate of weathering of these minerals [bib_ref] Controls on elevated fluoride and arsenic concentrations in groundwater from the Yuncheng..., Currell [/bib_ref] [bib_ref] Fluoride in natural waters, Edmunds [/bib_ref] [bib_ref] Fluoride in the Context of the Environment, García [/bib_ref]. Soil pH ranging between 3.8 and 7.2 was reported in farm soils in Makueni County , suggesting a relatively high F solubility potential in soils in the area [bib_ref] Fluoride and environmental health: A review, Ozsvath [/bib_ref]. The high F concentration in farm soil in the area is reflected in the TF values observed, which are higher in the vegetable samples (TF = 255-257) than in the grains (TF = 24-73). The significant positive correlation between the F concentration in food crops and farm soils suggests that farm soils in the study area contribute to the F absorbed by food crops. This relationship was also reported by several other studies [bib_ref] Accumulation and interaction of fluoride and cadmium in the soil-wheat plant system..., Li [/bib_ref] [bib_ref] Accumulation of fluoride and aluminium related to different varieties of tea plant, Ruan [/bib_ref] [bib_ref] The transfer of fluorine in the soil-wheat system and the principal source..., Wang [/bib_ref]. In addition to uptake from soil, the high F concentration in the analyzed crops could also result from irrigation with high F water. A recent study by [bib_ref] Naturally occurring potentially harmful elements in groundwater in Makueni County, South-Eastern Kenya:..., Gevera [/bib_ref] showed that different groundwater sources, used for domestic and agricultural purposes, in the study area were enriched in F (of up to 7.17 mg/l). Additionally, the farms where kale and cowpeas samples were collected, use borehole water with high F concentrations (between 5.19 and 7.17 mg/l) for irrigation. F from irrigation water can be absorbed into crops directly through leaves stomata and root uptake [bib_ref] Impacts of soil and water fluoride contamination on the safety and productivity..., Rizzu [/bib_ref].
## F exposure and health risk assessment
The calculated EADD and HI values suggest that the analyzed food crops considerably contribute to the daily F intake of the local population of the study area which may negatively impact their health. The EADD values of the grain-based crops (in both hypothetical 75% and 100% F absorption rates) were higher than the F reference dose for infants and children compared to adults. Vegetables had very high EADD values, well above the F reference dose for both adults and children in both hypothetical F absorption rates. This indicates that, children in the study area have a high risk of F related diseases from consuming both grain-based and vegetables, whereas adults are more vulnerable if they follow a largely vegetable diet.
The vulnerability of children to higher dietary F intake than adults was also reported in northern Tanzania [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref] , eastern India [bib_ref] Assessment of potential health risk of fluoride consumption through rice, pulses, and..., Bhattacharya [/bib_ref] , and southern Pakistan [bib_ref] Bioaccumulation of arsenic and fluoride in vegetables from growing media: Health risk..., Kazi [/bib_ref]. This was associated with a higher intestinal absorption rate and higher energy requirements relative to lower body weight in children compared to adults [bib_ref] Assessment of potential health risk of fluoride consumption through rice, pulses, and..., Bhattacharya [/bib_ref] [bib_ref] Bioaccumulation of arsenic and fluoride in vegetables from growing media: Health risk..., Kazi [/bib_ref] [bib_ref] Fluoride uptake and translocation in food crops grown in fluoriderich soils, Rizzu [/bib_ref]. Additionally, factors such as genetic determinants, sex, health status, dietary habits, substance abuse, physical fitness status, and concomitant exposure to other chemicals can influence the health implications of high F in diet in the area.
When considering F dose from different meals prepared, all vegetable meals had high EADD values compared to the F reference dose for both adults and children while maize meals showed the highest EADD contribution among the grains. It should be noted that, the energy (E) value is what differs between these meals (and therefore impacts the F EADD value) and the preparation methods do not add extra F in the meals. However, F can be present in the cooking material and dust.
The high EADD values in maize porridge and maize flour meals, which are among the most consumed meals in Makueni County and across Kenya, indicate that they contribute significantly to the daily F intake. Boiled cowpeas had no F health risks indicating a potentially desirable target grain-based crop in the region. Kale is a common vegetable component of many Kenyan meals and showed the highest EADD values in the steamed meal while stir-fried kale had the lowest value. Clearly cooking method makes a significant difference on EADD. The HI values of all the analyzed food crops were greater than 1 indicating that all the food crops analyzed pose noncarcinogenic health effects to the local population. The high prevalence dental fluorosis affecting the local population [bib_ref] Naturally occurring potentially harmful elements in groundwater in Makueni County, South-Eastern Kenya:..., Gevera [/bib_ref] , attests to the link between F availability and health impacts. Similarly, when a health survey was conducted to determine the health effects of high F and salinity in drinking water and public awareness in the area, 91% of the participants reported knowing at least one person with dental fluorosis in their village .
## Concluding remarks and recommendations
The findings of this study show that vegetables (cowpeas leaves and kale) have higher F concentrations (388 and 700 mg/kg) than the grain-based crops (maize, cowpeas, and green grams) (29 to 71.2 mg/ kg). The water-soluble F concentrations in the 30 samples of farm soil ranged between 0 to 3.47 mg/kg. One of the probable sources of F in the soil was geogenic, hosted by F-bearing minerals such as apatite, muscovite, and biotite present in the farm soils. The EADD analysis in the food samples indicated a potential F health risk mostly from consumption of the vegetables and maize meals. Children were at a higher risk of chronic F exposure in all the food samples due to their high daily energy requirement (high metabolic rate) per body weight. In addition, the use of high F irrigation water can contribute to the elevated F observed in the food crops. This causes a food security problem in the area that already has a constrained agricultural output due to semi-aridity. Therefore, besides the health risks associated with drinking of high F water in Makueni County, some food crops grown and consumed in the area could also contribute to the total daily F intake in a substantial way. In addition, different meals prepared from these food crops can also influence the level of F intake.
Based on these results, consideration should be given to the type of crops grown and consumed in the area. For example, grain-based crops, such as cowpeas legumes, with no F health risk should be encouraged for farming instead of green grams. Similarly, the types of meals prepared for the different age groups should be considered as evidenced by the high F exposure risk in maize porridge, which is a common meal for infants and children who are the most common vulnerable group in the area. Additionally, the use of high F water for irrigation should be discouraged in the area due to its potential contribution to F uptake by crops. Finally, further research is required to determine the amount of F in food consumed in the area as well as the dietary habits of the local population to determine the total F intake, which will help in establishing F risk factor in the area.
[fig] Figure 1: The geological map of Makueni County showing the Makindu-Kibwezi area, bounded in black, where soil and food crop samples were collected close to sampled water sources (black dots). Adapted from [/fig]
[table] Table 1 The: Fluoride concentration in the five food crops, the five farm soil where they were grown, and their Transfer Factor (TF) from Makueni CountyFig. 4The EADD due to the consumption of the five analyzed food crops (raw) at different levels of daily energy intake per Kg of body weight from Makueni County. The reference dose (RfD), in red, indicates the no observable adverse effects level (NOAEL), the level below which F has on humans health effects [/table]
[table] Table 3: The hazard quotient and cumulative hazard index of the five analyzed food crops (raw) at different levels of daily energy intake per Kg of body weight, at the two hypo- [/table]
|
10.1111/jcmm.14492
|
CCBY
|
6653234
|
31225720
|
s2orc_pubmed_articles
|
Dehydrocostus lactone attenuates osteoclastogenesis and osteoclast‐induced bone loss by modulating NF‐κB signalling pathway
Bone remodelling is a delicate homeostasis involving the regulation of both osteoblast and osteoclast activities. Maintenance of this balance helps prevent disorders such as rheumatoid arthritis, osteoporosis, and multiple myeloma-related bone loss. 1 In clinical practice, inflammatory bone loss remains difficult to treat, especially in complicated orthopedic trauma and arthroplasty cases. Inflammation of the skeletal system activates immune reactions and produces cytokines involved in osteoclast recruitment and differentiation, leading to excessive bone resorption. 2 However, an effective regimen for the treatment of inflammatory bone loss is lacking. Osteoclasts are multinucleate, tartrate-resistant acid phosphatase (TRAP)-positive cells originating from the monocyte-macrophage lineage. They play a crucial role in both physiological bone remodelling and pathological bone loss for the unique process of bone resorption. Osteoclastogenesis is induced by receptor activator of nuclear factor-κB ligand (RANKL) in the presence of Abstract Osteolysis is characterized by overactivated osteoclast formation and potent bone resorption. It is enhanced in many osteoclast-related diseases including osteoporosis and periprosthetic osteolysis. The shortage of effective treatments for these pathological processes emphasizes the importance of screening and identifying potential regimens that could attenuate the formation and function of osteoclasts. Dehydrocostus lactone (DHE) is a natural sesquiterpene lactone containing anti-inflammatory properties. Here, we showed that DHE suppressed receptor activator of nuclear factor-κB ligand (RANKL)-induced osteoclast formation and osteoclast marker gene expression. It also inhibited F-actin ring formation and bone resorption in a dose-dependent manner in vitro. Moreover, DHE inhibited the RANKL-induced phosphorylation of NF-κB, mitigated bone erosion in vivo in lipopolysaccharide-induced inflammatory bone loss model and particle-induced calvarial osteolysis model. Together, these results suggest that DHE reduces osteoclast-related bone loss via the modulation of NF-κB activation during osteoclastogenesis indicating that it might be a useful treatment for osteoclast-related skeletal disorders. K E Y W O R D S dehydrocostus lactone, NFATc1, osteoclast, RANKL How to cite this article: Hu B, Wu F, Shi Z, et al. Dehydrocostus lactone attenuates osteoclastogenesis and osteoclast-induced bone loss by modulating NF-κB signalling pathway.
macrophage colony-stimulating factor (M-CSF).Receptor activator of nuclear factor-κB ligand belongs to the tumour necrosis factor (TNF) superfamily; binding with its receptor, RANK, on the surface of osteoclast precursors recruits adaptor TNF receptor-associated factors (TRAFs). This process is important for osteoclast formation because it activates several downstream signalling pathways and the transcription factor activator protein 1 (AP-1), leading to full activation of T-cell cytoplasmic 1 (NFATc1).The expression of NFATc1 modulates osteoclast differentiation and functions via the induction of osteoclast-specific genes, including those encoding dendritic cell-specific transmembrane protein (DC-STAMP), TRAP, cathepsin K, calcitonin receptor (CTR), and matrix metalloproteinase 9 (MMP-9).Natural products have been an increasingly important source for drug development.We recently identified dehydrocostus lactone (DHE) as an effective inhibitor of osteoclastogenesis and osteoclast-related bone loss. Dehydrocostus lactone is a natural sesquiterpene lactone derived from the roots of Saussurea lappa, and it has been reported to have anti-inflammatory, anti-ulcer, anti-tumour, and immunomodulatory properties.However, the effect of DHE on osteoclast formation has not been reported. It has been shown that proinflammatory cytokines promote osteoclastogenesis,and overactivation of osteoclast activity plays an essential part in the origin of bone homeostatic imbalances.Thus, osteoclasts could be a key target in the treatment of inflammatory bone loss. Lipopolysaccharide (LPS) and titanium particles are both potent inducers of the immune system, and they play critical roles in the development of osteolytic bone loss.The purpose of this study was therefore to investigate the effects of DHE on osteoclastogenesis and on murine models of LPS-induced bone loss and particle-induced calvarial osteolysis, and to characterize the underlying mechanism of these processes.
## | material s and me thods
## | reagents and antibodies
Foetal bovine serum and Minimum Essential Medium Eagle Alpha Modification (α-MEM) were obtained from Gibco (Sydney, Australia).
## | cell culture and osteoclast differentiation
Primary murine bone monocyte/macrophage precursors were isolated from the unfractionated long bones of 6-week-old C57BL/6 mice and differentiated into bone marrow-derived macrophages (BMMs) in α-MEM containing 30 ng/mL of M-CSF for 3-4 days.
The cells were cultured in at 37°C in 5% CO 2 incubator, and the medium was changed every 2 days. The effect of DHE on BMM viability was determined using a CCK-8 assay. Cells were plated into 96-well culture plates at 8 × 10 3 /well for 1 day, and then incubated with different concentrations of DHE for 4 days. The cells were then incubated for 4 hours after the addition of 10 μL of CCK-8 buffer to each well, and the optical density (OD) at 450 nm was measured on an ELX808 absorbance microplate reader
## | immunofluorescence staining
The effect of DHE on F-actin ring formation was visualized using rhodamine-phalloidin staining. Bone marrow-derived macrophages were plated into 96-well plates and treated with the indicated concentrations of DHE in the presence of 30 ng/mL of M-CSF and 100 ng/mL of RANKL for 5 days. The cells were then fixed in 4% paraformaldehyde at room temperature; and permeabilized with 0.1% Triton X-100 in phosphate-buffered saline. F-actin was stained with rhodamine-phalloidin in 2% bovine serum albumin for 1 hour. The nuclei were stained with 4′,6-diamidino-2-phenylindole for 5 minutes. Fluorescence images were acquired using a Leica DMI4000B fluorescence microscope (Wetzlar, Germany). Three fields were randomly selected, and the numbers and sizes of the F-actin rings were calculated using Imagej (NIH).
## | bone resorption pit assay
The effects of DHE on osteolytic bone resorption in vitro were assessed on bone discs. An equal number of BMM-derived pre-osteoclasts were seeded onto bone discs and incubated overnight. The cells were then treated with or without 4 μmol/L DHE in osteoclastogenic medium for 2-3 days. The discs were fixed with 4% paraformaldehyde, and adhered cells were removed by gentle brushing. Three random views in each disc were visualized using a Hitachi S-3700N Scanning Electron Microscopy (Tokyo, Japan), and the resorbed areas were quantified by Imagej (NIH).
## | quantitative polymerase chain reaction
## | western blot analysis
To investigate the effects of DHE on RANKL-induced signalling pathways, RAW264.7 cells were seeded into six-well plates at 8 × 10 5 / well in complete α-MEM and incubated overnight. Cells were then pretreated with or without 4 μmol/L DHE for 4 hours before being stimulated by 50 ng/mL of RANKL for 0, 5, 15, 30, or 60 minutes.
To determine the effects of DHE on osteoclast-related protein expression, BMMs were seeded into six-well plates at 24 × 10The membranes were blocked with 5% (w/v) nonfat milk in 0.1% Tris-buffered saline and Tween 20 for 1 hour at room temperature, and then incubated with primary antibodies at 4°C overnight with gentle shaking. After washing and incubation with specific secondary antibodies for 1 hour at room temperature, the bands were detected using Enhanced Chemiluminescent Detection Reagent (Fude Biological Technology) with the Bio-Rad XRS Chemiluminescence Detection System (Hercules, CA, USA).
## | murine model of lps-induced bone loss and particle-induced calvarial osteolysis
This study was carried out in strict accordance with the protocols
# | statistical analysis
Data are expressed as the mean ± SD. All experiments were independently carried out at least three times. Unpaired Student's t tests were used to compare two groups, and one-way ANOVA tests were used for ≥three groups, with SPSS for Windows, version 19.0 (IBM Crop., Armonk, NY). A value of P < 0.05 indicated a significant difference between groups.
## | re sults
## | dhe inhibited the differentiation of bmms into osteoclasts without cytotoxicity
The cytotoxicity of DHE toward osteoclast precursor cells was determined by the CCK-8 assay with a range of doses of DHE, and the results indicate that the IC 50 of DHE for BMMs was 17.5 μmol/L .
To investigate the effect of DHE on osteoclast formation, BMMs were cul- . Together, these results showed that DHE inhibited osteoclast differentiation without cytotoxicity.
## | dhe suppressed rankl-induced osteoclast marker gene expression
## | dhe interfered with f-actin ring formation and bone resorption in vitro
The formation of F-actin rings is a prerequisite to the adhesion of osteoclasts to bone. To determine whether DHE impaired the formation of F-actin rings, BMMs were seeded into 24-well plates and
## | dhe attenuated rankl-induced nf-κb signalling activation
The downstream pathways of RANK are important for osteoclast differentiation, so we determined whether DHE modulated the phosphorylation of MAPK and NF-κB after RANKL stimulation. The phosphorylation peaked within 60 minutes after RANKL stimulation . The phosphorylation of MAPK pathway was unaffected. RANKL-induced NF-κB activation is also crucial for osteoclast formation. As shown in activation of the upstream IKK complex, p65, and IκBα was suppressed by DHE.
Taken together, these results suggest that DHE inhibited osteoclast formation by modulating the RANKL-induced activation of NF-κB signalling.
## | dhe impaired nfatc1 induction
NFATc1 is an expression inducer of osteoclast marker genes, and a master transcription factor during osteoclast formation.To determine whether DHE attenuated the expression of NFATc1, we investigated protein expression in the presence of DHE during osteoclastogenesis of BMMs. shows that the expression of NFATc1 reached a peak at day 3, while its expression was significantly down-regulated in the presence of DHE. In addition, it attenuated the expression of cathepsin K and c-Src in BMMs during osteoclastogenesis , which is critical for osteoclast formation and function.
## | the lps-induced bone volume loss and particle-induced calvarial osteolysis were partially rescued by dhe
To characterize the antiresorptive property of DHE in vivo, models of LPS-induced bone loss and particle-induced calvarial osteolysis were adopted. Mice that received intraperitoneal LPS showed a significant bone volume loss at the secondary spongiosa of the tibia, while mice that received LPS + DHE showed a dose-dependent increase in bone volume at this region . Three-dimensional micro-CT confirmed massive trabecular bone loss in LPS-treated mice compared with control mice . Consistent with the histological data, a morphometric study by micro-CT indicated significant reductions in BV/TV, Tb.Th, and Tb.N, and a remarkable increase in Tb.Sp after LPS administration. Treatment with DHE partially prevented inflammatory bone volume loss in a dose-dependent manner. What's more, protective effects of DHE were also observed in a model of particle-induced calvarial osteolysis. The resorption area of calvaria was significantly reduced in DHE treated groups . Together, these results confirmed the protective effects of DHE on osteoclast-related bone loss in vivo.
## | d iscuss i on
DHE is a naturally occurring sesquiterpene lactone reported to have many biological effects, including anticancer, bactericidal and antioxidative stress properties.However, there has been no report examining the effects of DHE on osteoclasts. In this study, we investigated its effects on RANKL-induced osteoclast formation and function, and determined the underlying mechanism in vitro. We also examined its effects on LPS-induced bone loss and particle-induced calvarial osteolysis in vivo.
NFATc1 is a central regulator of the expression of marker genes, such as DC-STAMP, cathepsin K, CTR, TRAP, MMP-9 and V-ATPase a3 in the differentiation of precursors into functional osteoclasts.
In this study, we showed that treatment with DHE suppressed the mRNA expression of osteoclast marker genes, including those encoding NFATc1, DC-STAMP, cathepsin K, CTR, TRAP, MMP-9, V-ATPase a3, and OSCAR. An in vitro study also confirmed the inhibitory effects of DHE on F-actin ring formation, which is essential reported in many studies.TRAF6 is recruited after the binding of RANKL and RANK, and it eventually forms a complex with TAK1 and TAK1-binding protein 2.The complex activates TAK1, leads to the stimulation of NF-κB and AP-1 via of IKK and p38 phosphorylation, respectively.The current results indicate that DHE inhibited the phosphorylation of NF-κB signalling, and attenuated the expression of c-Src, cathepsin K and NFATc1. However, DHE had little effect on the RANKL-induced phosphorylation of MAPK pathway.
The key role of osteoclasts in inflammatory bone loss makes it attractive as a potential therapeutic agent. Lipopolysaccharide is a membrane component of Gram-negative bacteria and has been used as a potent inducer of osteolytic bone loss.Studies have suggested that microbial-associated molecular patterns can adhere to generate wear debris in arthroplasty cases and lead to substantial inflammation 23 ; among them, LPS is the best known. It is capable of promoting the secretion of proinflammatory cytokines
## | con clus ion
Our findings demonstrate that DHE attenuated RANKL-induced osteoclast formation and inhibited F-actin ring formation and bone resorption in vitro. Moreover, DHE partially restored LPS-induced bone loss in cancellous bone and particle-induced calvarial osteolysis in vivo. Its mechanism involved suppression of the expression of NFATc1 and other marker genes during osteoclast formation by targeting the activation of NF-κB signalling. These data suggest that DHE could be used for the treatment of osteoclast-related skeletal disorders.
## Ack n owled g em ents
This work was supported by grants from the National Science Foundation of China (81772360 to Shigui Yan, 81401785 to Xiang Zhao) and from the Natural Science Foundation of Zhejiang Province (LY17H060004, Y17H060027 to Haobo Wu).
## Co n fli c t o f i nte r e s t
The authors confirm that there are no conflicts of interest.
## Auth o r s' co ntr i b uti o n
BHu, FFW and ZLS performed the research, BHu and XZ analysed the data, BHe performed part of the in vivo study, BHu, HBW and SGY designed the research study and wrote the paper.
## O rci d
## Shigui yan
https://orcid.org/0000-0001-9096-2868
|
10.3389/fncel.2020.00135
|
CCBY
|
7248338
|
32508598
|
s2orc_pubmed_articles
|
Modeling a Nociceptive Neuro-Immune Synapse Activated by ATP and 5-HT in Meninges: Novel Clues on Transduction of Chemical Signals Into Persistent or Rhythmic Neuronal Firing
Extracellular ATP and serotonin (5-HT) are powerful triggers of nociceptive firing in the meninges, a process supporting headache and whose cellular mechanisms are incompletely understood. The current study aimed to develop, with the neurosimulator NEURON, a novel approach to explore in silico the molecular determinants of the long-lasting, pulsatile nature of migraine attacks. The present model included ATP and 5-HT release, ATP diffusion and hydrolysis, 5-HT uptake, differential activation of ATP P2X or 5-HT3 receptors, and receptor subtype-specific desensitization. The model also tested the role of branched meningeal fibers with multiple release sites. Spike generation and propagation were simulated using variable contribution by potassium and sodium channels in a multi-compartment fiber environment. Multiple factors appeared important to ensure prolonged nociceptive firing potentially relevant to long-lasting pain. Crucial roles were observed in: (i) co-expression of ATP P2X2 and P2X3 receptor subunits; (ii) intrinsic activation/inactivation properties of sodium Nav1.8 channels; and (iii) temporal and spatial distribution of ATP/5-HT release sites along the branches of trigeminal nerve fibers. Based on these factors we could obtain either persistent activation of nociceptive firing or its periodic bursting mimicking the pulsating nature of pain. In summary, our model proposes a novel tool for the exploration of peripheral nociception to test the contribution of clinically relevant factors to headache including migraine pain.
# Introduction
The mechanisms responsible for migraine pain, a common and devastating condition, remain poorly understood. Nevertheless, the prevailing opinion suggests that one major component of migraine pain originates from the trigeminal nerve fibers located in meningeal tissues to send nociceptive signals to brainstem nuclei. The meninges harbor a large population of mast cells, which can release various chemicals that activate nearby nerve fibers. According to the original purinergic hypothesis of migraine, extracellular ATP (eATP) is one such player for the onset of migraine pathology. In fact, in addition to the effect of eATP on blood vessels, this endogenous purine can interact with pro-nociceptive P2X3 receptors located almost exclusively on sensory neurons, indicating P2X3 receptors as important contributors to pain generation.
Previous studies have described the unique properties of P2X3 receptors including fast activation and desensitization and slow rate of resensitization. To describe these characteristics, we developed a cyclic model for the function of P2X3 receptors that closely replicates P2X3 mediated responses. However, the original data used for modeling were obtained from cultured sensory neurons and recombinant receptor systems, leaving open the question of their in vivo applicability. One paradoxthat remains unsolved is how the strong desensitization of P2X3 receptors commonly observed with a patch-clamp recording from cultured neurons is compatible with the well-known role of this ATP-driven receptor in sustained pain signaling.
Our recent work has further supported the purinergic hypothesis of migraine by showing the ability of ATP and its chemical analogs to trigger persistent spiking in trigeminal nerve fibers present in the whole-mount rat meninges. Furthermore, using mast cell-deficient mice, we have shown that eATP could activate trigeminal nerves both directly as well as via release of 5-hydroxytryptamine (5-HT) originating from degranulation by immune cells. Interestingly, 5-HT is not only a powerful trigger for prolonged nociceptive firing in meningeal afferentsbut also a well-known sensitizing agent.
The complex interplay among ATP, 5-HT, and their mast cell release process remains, however, to be elucidated. To address this complex phenomenon, the present study applied a modeling approach to explore the impact of ATP and 5-HT release from immune cells (meningeal mast cells), ATP hydrolysis and diffusion, 5-HT uptake, ATP-activated P2X3and P2X2 receptors, and of 5-HT-activated 5-HT3 receptors. In addition to the standard role of sodium and potassium channels in membrane excitability, former modeling studies have highlighted the importance of certain subtypes of the sodium channel in coding sensory information by nociceptive sensory neurons. Thus, one computational model has described their role in sensory signaling by dorsal DRG neurons innervating the urinary bladder.have shown that the density of sodium channels determines the fidelity and precision of neuronal sensory coding. Likewise, the model of C-fibers byhas shown the characteristics of axonal spike propagation in human C-nociceptors. Whereas several subtypes of sodium channel are expressed by nociceptive neurons, the subtypes Nav1.7 and Nav1.8 play are particularly important ones for generation and propagation of action potentials. For instance,have shown the key role of sodium Nav1.7 channels in several pathological pain syndromes.
The current report provides novel information concerning not only fundamental molecular properties but also strategic topography of neuro-immune crosstalk underlying purinergic and serotonergic signaling and their impact on voltage-gated channels that may contribute to the peripheral mechanisms of migraine pain.
# Materials and methods
## Model of meningeal nociception
To simulate rat meningeal trigeminal fiber activity, we used the NEURON environment version 7.5. The fiber was assumed to be 3 cm longwith a diameter from 0.25 to 2 µm corresponding to C-and Adeltafibers, respectively. All A-fibers in the dura belong to the Adelta subtype and are known to be present in the meninges. shows the basic features of the model with the fiber (green) surrounded by a mast cell containing the secretory vesicles (as sources of eATP or 5-HT;and forming the neuro-immune synapse. Each segment of the fiber is referred to as a compartment and is indicated as a green rectangle (0.25 µm wide and 250 µm long). In our preliminary tests, we found just one P2X3 receptor in the compartment to be sufficient to depolarize the membrane potential up to the spike threshold . The properties of the following sodium and potassium channels examined in the present report are listed in : Nav1.7, Nav1.8, Nav1.3, potassium delayed rectifier (K-DR;and A-type currents.
It has been previously shown that meningeal nerves often branch to make a sort of ''neuronal dendritic tree''. Thus, we simulated a dendritic tree of the trigeminal nerve fiber with two branches. The tree was 3 cm long and comprised 170 compartments: main dendrite (70 compartments), and each branch with 50 compartments (example in. Two branches of the trigeminal nerve were joined at 1.75 cm from the trigeminal ganglion (TG). This is a site that can block the propagation of neuronal signals in the refractory state. This property was FIGURE 1 | Main components of the model for ATP-induced activation of meningeal afferent (nerve fibers). (A) The experimental approach of meningeal spike recording by glass electrode from the local nerve (Aa) and the image of meninges with labeled mast cells (Ab). MMA-main meningeal artery. (B) Schematic presentation of the model neuroimmune synapse with a mast cell ("presynaptic cell") as the source of the transmitter of ATP (or 5-HT) and meningeal nerve fiber ("postsynaptic cell") consisting of several compartments with ATP-gated P2X3 receptors. (C) Changes in the membrane potential of the nerve fiber (spike generation) triggered by activation of the P2X3 receptor. (D) Ionic currents through several subtypes of sodium, potassium, and P2X3 receptor channels. (E) The graph showing the number of P2X3 receptors required to trigger a spike as a function of nerve fiber diameter. (F) The kinetics of P2X3 receptor-induced currents and the recovery time course from desensitization. further explored in simulation experiments. To implement the ATP induced activation of nerve fibers, we integrated the cyclic scheme of the P2X3 receptor operationinto the model. To simulate ATP release from mast cells, we used the 3D diffusion model described bywhich is based on the following equation:
[formula] C(x, y, z, t) = 2 M (4πDt) 3/2 exp −kt − x 2 + y 2 + z 2 4Dt [/formula]
where C is the concentration of ATP, D is the diffusion coefficient (0.8 µm2/ms,, k is the degradation coefficient, M is the initial amount of ATP released from mast cells, x, y, z are the space coordinates and t stands for time. The ATP profile in the model was simulated as instant rise and slow decay determined by diffusion.
## Experimental part
The experimental part was performed on 10-12-week-old male WT C57BL/6J mice and adult Wistar rats provided by the Animal Facilities of the University of Eastern Finland (UEF). All procedures were approved by the Committee for the Welfare of Laboratory Animals of the University of Eastern Finland and the Provincial Government of Kuopio. Experiments were conducted according to the European Community Council guidelines (Directives 86/609/EEC). All efforts were made to minimize the number of animals used and their suffering.
## Recording of spikes from meningeal afferents
For the aim of validation, model data were compared with experimental results obtained with application of ATP or 5-HT onto meningeal afferents and published by us earlier. For those experiments we used isolated wholemount mouse hemiskulls as previously described. Hemiskulls were isolated, keeping the dura mater with meningeal nerves and vessels intact. The meningeal branch of the trigeminal nerve was cleaned from surrounding tissue, then cut and placed inside a saline-filled glass electrode. All recordings of electrical activity from trigeminal nerves were performed from hemiskull preparations continuously perfused with ACSF oxygenated with 95% O 2 /5% CO 2 . Trigeminal nerve spiking activity was recorded using DAM80 amplifier (World Precision Instruments, Sarasota, FL, USA). Electrical signals were digitized using a NI PCI6221 board (National Instruments, USA) and stored on a PC for off-line analysis. Signals were visualized by WinEDR v.3.2.7 software (University of Strathclyde, Glasgow, UK) and analyzed with Matlab-based software.
## Mast cells labeling
To demonstrate abundance and localization of mast cells in the meninges we used rat hemiskull preparations as described previously. After decapitation the head was cut mid-sagittally, brain was carefully removed to leave dura mater untouched. Then preparation was fixed in 4% PFA for 4 h, meninges were mounted on microscopy slides and stained in 0.1% Toluidine Blue.
## Immunolabeling
For nerve fibers staining (tubilin-beta III positive filaments), dissected meninges (from five ad hoc prepared mice) were incubated in a blocking solution containing 2% normal goat serum, 1% BSA, 0, 05% Tween20 and 0.1% Triton X-100 (Sigma Aldrich) for 1 h at room temperature. Meninges were incubated overnight at 4 - C, with rabbit anti-tubilin-beta III (1:1,000; Sigma T2200) primary antibody. Meninges were then washed three times in PBS and incubated for 2 h in the dark at room temperature with goat anti-rabbit AF488 (Invitrogen A11008) secondary antibody. After washing three times in PBS, meninges were mounted with Fluoromount-G (Thermo Fisher Scientific 00-4958-02), and images acquired with a Zeiss LSM 710 confocal microscope. shows the two principal components of the meningeal trigeminal system, namely the middle meningeal artery (MMA) and nearby trigeminal nerves with the glass electrode for spike recordings. In note the abundance of mast cells located in close proximity to meningeal blood vessels, which is a region densely innervated with trigeminal afferents. For modeling the meningeal neuro-immune synapse, we first focused on the pro-nociceptive action of extracellular ATP (eATP;and well-known kinetic properties of eATP-gated P2X2 and P2X3 receptors whose kinetics were presented in the earlier models.
# Results
## P2x3 receptor probability of spike generation
We assumed that meningeal mast cells known to be in close contact with nerve terminals serve as the main source of the local ATP release (''presynaptic cell''). However, we cannot exclude other sources of local eATP as it is well-known that various cell types can release ATP upon mechanical stress. Among them, there is release of ATP from endothelial cells during shear stress. As a first approximation, we did not take into account cellular stress as source of ATP release and modeled just eATP diffusion over time. Segments of the nerve fiber (''postsynaptic cell'' in the frame of the hypothesis of neuro-immune synapse) in our first models, express ATP-gated P2X3 receptors . shows a simulated nociceptive spike, while shows sodium, potassium and P2X3 receptor currents underlying this spike. depicts a plot obtained from data modeling to determine the number of receptors necessary for spike generation by a single 250 µm long nerve fiber. We assumed 0.25-2 µm diameter range and observed quadratic dependency of the number of P2X3 receptors for spike generation in one compartment of the nerve fiber. Thus, for the 0.25 × 250 µm compartment, an active P2X3 receptor already effectively generated nociceptive firing. As mentioned earlier, P2X3 receptors possess strong desensitization. shows the model of slow recovery from P2X3 receptor desensitization as the amplitude of the first current response (green deflection) is decreased following closely spaced ATP application and recovers as time lapses. This property was used in subsequent versions of our model (Models 1-14). As low concentration of ATP can produce the inhibitory action on P2X3 receptors via high-affinity desensitization (HAD;, we also simulated the prolonged action of 1 nM ATP on the stimulatory effect of this purinergic agonist (Supplementary. First, we found that, indeed, HAD reduced the probability of spikes generation by low concentrations of ATP. Thus, after HAD, the minimal number of active P2X3 receptors required for spike generation was doubled comparing with model without HAD (Supplementary. Interestingly, in this test, also the dependance between the number of receptors and the nerve fiber diameter became much steeper. Further, HAD reduced, by half, the amplitude of the P2X3 receptor mediated current induced by 1 µM ATP (Supplementary. Finally, the probability of repetitive firing via P2X3 receptors with HAD was also reduced as now the release of ATP triggered only one spike (Supplementary, whereas without HAD the same release produced two spikes (Supplementary.
Because of the large number of tested conditions, Supplementary shows the hierarchy of models used (see also Supplementary Table S3), starting from the simplest model 1 with ATP degradation coefficient = 0, conductance of Nav1.7 = 0.1 S/cm 2and P2X3 receptor function for the receptor potential necessary for spike generation. In model 2, we set the degradation coefficient of ATP to 0.01. Model 3 originates from model 1 with the degradation coefficient of ATP equal to 0.8. Model 5 stemmed from model 2 with the presence of P2X3 and P2X2 receptors in a 50/50 ratio, whereas in model 4 we used only P2X2 receptors (instead of P2X3 receptors) to obtain activation of the nerve fiber. Model 6 was developed from model 2 with a higher Nav1.7 conductivity (0.2 S/cm 2 ) used also for model 7 with Nav1.7 0.1 S/cm 2 and Nav1.8 0.1 S/cm 2 conductivity, while model 8 with Nav1.8 0.2 S/cm 2 conductivity. The subtype Nav1.3was present in all models with 0.2 S/cm 2 conductivity. Model 11 is based on model 8 with the fiber topology expressed as a tree, and model 12 was based on the same topology with two ATP release sites.
## The role of diffusion and atp hydrolysis
It is generally accepted that three ATP molecules must occupy three binding sites to activate successfully a single P2X3 receptor. However, in the tissues related to generation of initial migraine pain, such as the meninges, ATP is rapidly degraded by extracellular NTPDases to the less active ADP, and then to AMP and adenosine. Hence, in our model we introduced the role of ATP diffusion and hydrolysis in the control of trigeminal afferent firing. First, we explored the action of 1 µM ATP, which is close to its EC 50 on the P2X3 receptor; . In this simulation we assumed that P2X3 receptors were homogeneously distributed along the nerve fiber. The ATP concentration profile reaching compartments 1, 2, 3 and 4 of the nerve fiber without hydrolysis (lack of NTPDases) or with partial hydrolysis (active NTPDases) is indicated in(compartments 1, 2, 3 were close to the point of ATP release). Compartment 1 was located opposite the release point, whereas the distant compartment 4 was 55 compartments away from the compartment 3 (distance equal to 13.7 mm). The compartment 4, unavailable for ATP, served here only to indicate the propagation of spikes along the fiber. In model 1, in the compartment 1 without hydrolysis (degradation coefficient = 0;, ATP was as high as 0.9 µM ATP at the nerve fiber. To simulate ATP hydrolysis, we changed the rate of ATP degradation from 0.01 s −1 to 0.8 s −1 . Thus, model 2indicates the profile of ATP concentrations in compartments Frontiers in Cellular Neuroscience | www.frontiersin.org 1, 2, 3 and 4 with partial hydrolysis. In this case, the maximal ATP concentration in the compartment 1 (0.06 µM) was over 10 times less than in model 1, and was enough to activate 10% of P2X3 receptors. Strong rate of hydrolysis decreased the concentration of ATP in the compartments 2, 3 and 4 to almost undetectable values because of the exponential character of ATP diffusion (Equation 1). Without ATP hydrolysis, P2X3 receptor activation triggered spiking (two spikes) in compartment 1, which propagated through the compartments 2, 3 and 4 to reach the TG (Model 1;. In the case of strong hydrolysis (degradation coefficient = 0.01 from maximal 1), there was only one spike produced by 1 µM ATP (Model 2;. Neuronal activity decreased dramatically when the degradation coefficient was 0.8 (close to the maximal value of 1). In this case, even in compartment 1, the local receptor potential did not reach threshold for spike generation (Model 3;.
## Comparing the role of p2x2 vs. p2x3 receptors and effect of p2x2/3 heteromers
Although P2X3 receptors are expressed in approximately 80% of trigeminal neurons , sensory neurons also express slowly desensitizing P2X2 receptors which are implicated in pain signaling. Thus, we next explored the differential ability of P2X2 and P2X3 receptors to support spiking activity in trigeminal fibers. First, given that cells can release more than 1 µM ATP, we tested the role of higher concentrations of ATP limited by hydrolysisin the firing activity triggered by a homogeneous population of P2X3 receptors. Because of slow desensitization, we found that ATP signaling via P2X3 receptor provided just one (10 µM ATP) or two (100 µM ATP) spikes propagating to the TG.
Next, we simulated the role of P2X2 receptors based on a kinetic model with minimal desensitization state, which is a typical feature of this receptor. Unlike P2X3 receptors, this approach (Model 4) yielded much more prolonged receptor activationand multiple firing in the TG especially in the case of 100 µM ATP.
Because the potential co-expression of P2X2 and P2X3 receptors with distinct desensitization rates in trigeminal neurons and because P2X2 receptor subunits can co-assemble with P2X3 ones to form P2X2/P2X3 heteromers, we also explored the effect of such co-assembly (comprising 50% P2X3 and 50% P2X2) on meningeal neuronal firing (Model 5).indicated that in such a case, despite hydrolysis, 10 µM ATP produced repeated spikes, although less intensively than in the case of homogeneous P2X2 receptors. Simulating the release of 100 µM ATP led to strong, prolonged TG spiking (Model 5;. Thus, co-expression of P2X2 and P2X3 receptors appeared to be a powerful process to generate sustained nociceptive activity.
Different Role of Nav1.7 and Nav1.8 Sodium Channels
Among the subtypes of sodium channel expressed by nociceptive neurons, the subtypes Nav1.7 and Nav1.8 play a special role in the onset and propagation of spiking activity. Thus, we explored the impact of Nav1.7 and Nav1.8 channels on ATP-induced firing of trigeminal fibers. First, we tested the role of Nav1.7 subunit density on fiber activity induced by a single ATP release event (ATP 1 µM) acting on P2X3 receptors in compartment 1 (Model 2;. When the Nav1.7 conductance was 0.1 S/cm 2 , and the P2X3 receptor current depolarized membrane potential to −20 mV, this condition was sufficient to activate sodium channels and trigger limited spiking. When the Nav1.7 conductance was doubled to 0.2 S/cm 2 , the propagation rate of spikes was slightly increased, yet the number of spikes was unchanged.
Similar tests were performed with Nav1.8 channels (coexpressed with Nav1.7) starting from 0.1 S/cm 2 conductance like with Nav1.7 channels. In this case, after a single ATP release event (ATP 1 µM), multiple spiking emerged. In contrast to Nav1.7, firing largely increased when the conductance of Nav1.8 channels was doubled. During these simulations the activity of potassium channels was not changed, implying that changes in certain sodium channels (especially Nav1.8 subtype as indicated in models 7 and 8) were already sufficient to dramatically affect spiking activity of the nerve fiber.
## Repetitive atp release and distant release sites along the nerve fiber
Because meninges contain an abundance of mast cells along the vessels and the divergent branches of the sensory fibers , we next explored the role of repeated ATP release on the probability of repetitive spike generation in this histological arrangement. First, we simulated two release events delivering ATP to same site of the fiber (Model 9). This model consisted of 120 compartments along the nerve fiber, each compartment equipped with Nav1.7, Nav1.3, potassium DR, and potassium A-type conductances. In order to identify factors that can overcome lingering P2X3 desensitization, we again used signaling only via P2X3 receptors. This approach provided an initial condition with one ATP-driven spike to facilitate detection of other factors leading to multiple firing.
Using model 9, we set the different timing (∆t) of the two ATP release events in a range of 2-8 min and investigated generation and propagation of spikes from the release point to the TG. As expected, the generation of a second spike in the basal model with a simple one-release point simulation was mostly determined by the slow recovery of P2X3 receptors from desensitization. Thus, we did not observe a second spike with two ATP applications separated by less than 4 min ∆t interval. Hence, spikes were generated only when the second release occurred 6 min after the first one. The probability of the second spike firing was nonlinearly dependent on ∆t. Thus, at ∆t = 5 min the generation of a second spike had 0.5 probability, whereas at ∆t ≥ 6 min the function reached saturation with probability 1 for the second spike generation.
We next focused on the effect of the distance (∆l) between two distinct ATP release points (spatial factor) located along the nerve fiber to evoke a second spike and overcome the long-lasting desensitization of P2X3 receptors. The minimal starting distance between two release sites was set at 1 mm to minimize the effect of ATP diffusion from the first release site. Then, we varied the distance (l) and time (t) between two ATP applications (Model 10;.indicates that at ∆t = 20 ms and ∆l = 5 mm, only one spike emerged despite a dual ATP release, whereas at ∆t = 40 ms and ∆l = 7 mm two spikes appeared in the same fiber upon the second ATP release.shows the relation between ∆t and ∆l for generation of the second spike after two ATP release events. Thus, for triggering more than one spike, the minimal ∆t value was 15 ms with 1 mm ∆l distance between two ATP release events.
In summary, in the present simplified model with the linear morphology of the nerve fiber and only two mast cells as source of ATP, the probability of the second spike generation was nonlinearly dependent on the timing of the second ATP application (∆t) and was linearly dependent on the distance between ATP release events (∆l).
## Complex architecture of trigeminal fibers
We further developed our model to include a simple bifurcation of the single nerve fiber since it has been shown significant branching of trigeminal fibers in meninges. The morphology of mouse meningeal innervation with the branched nerve fibers is shown in. The schematic presentation of the branched tree is presented in. Model 11 shown inconsists of 170 compartments in two branches, each one with a P2X3 receptor and Nav1.8, Nav1.7, Nav1.3, potassium DR and A-type conductances. The bifurcation point was 70 compartments (1.75 cm) away from the TG, and there were two ATP release points 1 cm apart.
First, we found that a single (first) ATP release, indicated by arrow at simulation time = 50 ms) triggered (after some latency) multiple spiking of the TG branch (eight spikes in the period from 152 up to 485 ms;. Second, we tested the impact of a second ATP release (40 ms later than the first one) that also triggered repetitive spiking in the other The probability of successful 2 nd spike generation with a variable time (∆t) between two ATP applications. The distance (∆l) between two ATP applications is 0 (Model 9). (C) The dependency between the distance (∆l) and time (∆t) between two ATP applications (Model 10). The purple area identifies the conditions for the productive second spike generation (see inset) after two ATP applications whereas white area above-conditions when even a double release of ATP generates only one spike (see inset).
branch of the nerve (eight spikes, activity lasting from 193 ms to 538 ms;. Then, we explored the firing activity of a whole nerve sequentially activated by ATP released from two (first and second) distinct sites. This resulted in stronger spiking (12 spikes, activity lasting from 152 ms to 538 ms;which reached the TG.
Thus, the complex architecture of trigeminal fibers can also contribute to multiple firing.
## 5-ht induced activation of nerve fibers
Mast cells containing the principle mediator 5-HT are largely expressed in the meninges . We have previously shown the strong pro-nociceptive action of this monoamine on meningeal afferents via ligand gated 5-HT3 receptors in ratsand mice. In the present study, we applied the kinetic model of 5-HT3 receptorsto our model of C-fibers. The model implied mast cells releasing 2 µM 5-HT acting on 5-HT3 receptors and subjected to 5-HT uptake. For the sake of simplicity, we indicated only two compartments of the trigeminal nerve.shows the plot of 5-HT concentration with or without 5-HT uptake. Unlike the short time profile of ATP (rapidly degraded by ectoenzymes, gray trace), the concentration of 5-HT after single release from mast cells decayed much slower with or without uptake processes. In addition, 5-HT3 receptor recovery from desensitization was also faster; green) than the one for the ATP-gated P2X3 receptor, gray). Comparison of repeated spiking activity triggered by 5-HT3 receptor activation with or without monoamine reuptake and contribution of Nav1.8 channels is shown in. Spiking activity was, however, more intense with ATP application and Nav1.8 channel activity.
## Comparison of model and electrophysiological data
To validate our model, we compared simulated firing with experimentally observed action potentials triggered by ATP and 5-HT on mice meningeal afferents. In this study, we measured the firing of the mouse trigeminal nerve activated by application of exogenous ATP.shows the effect of 100 µM ATP for the 1st and 5th min of its action in meninges.
Next, we compared these experimental results with the simulation of neuronal activity induced by 100 µM ATP.(inset, left) shows a cross section of the model nerve with several active nerve fibers: either with P2X3 receptors or with a mixture of P2X2 and P2X3 receptors (Models 5 and 8, respectively). The inactive fibers (shown in gray) represent the absence of P2X receptors. Like the experimental approach, our model revealed an intense and prolonged ATP-induced spiking activity in the nerve fiber.
Comparable results were obtained also with experimental and simulated applications of 5-HT.shows the structure of the model nerve in which fibers express 5-HT3 receptors or lacking such receptors. In this case, as experimental application of 2 µM 5-HT induced a long-lasting activity in meningeal afferents, a similar phenomenon was also observed with simulated application of 2 µM 5-HT.
Then, we compared the inter-spike intervals (ISI) of experimental and modeling data. In experimental ATP application, ISI were 1,213 ± 499 ms, whereas in the model, the median ISI were 502 ms with 50%/50% mixture of P2X2 and P2X3 receptors, 1,704 ms with 75%/25% mixture and 56 ms with 25%/75% mixture. Thus, the 75%/25% ratio of P2X3/P2X2 receptors in the model was the closest combination of P2X2 and P2X3 receptors to the experimental data.
With experimental 5-HT application, the ISI of spiking activity was 4,039 ± 1,134 ms, and the closest model data were obtained with 10% activated fibers which had 5-HT3 receptors with conductances ranging from 0.25 to 0.5 S/cm 2 and 0.1-0.2 S/cm 2 (ISI: 3,041 ms and 6,556 ms, respectively;.
Together, these data indicated a high similarity of our model results with experimental ATP-and 5-HT-induced firing.
## Modeling prolonged and rhythmic nerve activity
Migraine pain has a still unexplained pulsating character perhaps due to mechanical fluctuations of meningeal vesselsthat might interact with nearby nerve fibers and mast cells. In order to investigate the potential role of pulsating blood flow as a trigger of meningeal nociception and its relation to ATP-and 5-HT-induced firing, we hypothesized that vasodilation (during migraine attacks) of pulsating meningeal vessels can stimulate release of these two mediators from local mast cells. In addition, we considered that ATP can be released directly from endothelial cells of meningeal vessels.
Based on the co-assembly of meningeal nerves, mast cells and vessels shown in and on our experimental data obtained from mice lacking mast cells, we hypothesized , shown by arrows) that the pulsatile blood flow can release ATP and 5-HT from mast cells adding more ATP released from the endothelium of meningeal vessels. The extracellular ATP (eATP) can act either directly on nerve fibers or indirectly via degranulation of mast cells. These interactions are also shown in the (Supplementary Movie).
The structure of modeled nerves for ATP and 5-HT actions in was the same as shown in. First, we modeled the neuronal activity in the rat induced by periodic pulsations (assuming an average heart beating rate = 400/. Thus, we examined the spiking activity assuming that these pulsations released 1 µM eATP (note arrows inwhich, in turn, activated nerve fibers. Interestingly, after the initial continuous spiking induced by ATP, the neuronal discharges (indicated by short vertical lines in Modeling also allows simulating pulse-induced spiking in human afferents. for ATP and for 5-HT, show that the assuming human pulsating blood flow (presumed to be 70 beats/min) could produce initial persistent firing, which was quickly transformed into periodic bursts of activity with expected lower rate. In this case, we took into consideration not only the lower human-specific heart rate but also the faster (circa two-fold) rate of recovery from desensitization of human P2X3 receptors.
# Discussion
The main result of the current study is a novel model of meningeal nociception based on the concept of a neuro-immune synapse composed by trigeminal nerve fibers and local mast cells, which may release two mediators, ATP and 5-HT. These two endogenous substances were taken here as chemical triggers of nociception based on their powerful and prolonged stimulatory effect on meningeal afferents experimentally demonstrated with FIGURE 9 | Firing of meningeal afferents induced by pulsatile ATP and 5-HT applications. (A) The schematic presentation of pulsating vessels inducing transmitter release from mast cells or endothelium. ATP acts either directly on nerve terminals (via P2X3 receptors) or by 5-HT released from mast cells and 5HT activates its own neuronal 5-HT3 receptors. (B) The cross-section of model nerve with P2X3 receptors and a mixture of 50% P2X2 and 50% P2X3 receptors. (Ba) Spiking activity in nerve fibers with ATP release events following at the frequency of the human heart rate (70 beats per minute). (Bb) The spiking activity of fibers with ATP action at the frequency of the rat heart rate (400 beats per minute). (C) The cross-section of the model nerve with 5-HT3 receptors. (Ca,Cb) Spiking activity of nerve fibers with ATP action at the frequency of the human heart rate (70 beats per minute) and rat heart rate (400 beats per minute), respectively. direct recordings of spikes from trigeminal nerve fibers in rats and mice.
## Role of atp receptor subtypes
To model purinergic (eATP-driven) signaling within the neuroimmune synapse, we first simulated the kinetic properties of eATP-gated P2X2 and P2X3 receptors, which are the major ATP sensitive receptor subtypes in sensory neurons generating pain signals.
Because of the fast onset of desensitization and slow recovery properties, P2X3 receptors activated by a single pulse of eATP (even without any ATP hydrolysis) evoked only limited activity in meningeal trigeminal nerve fibers (Model 1;. Assuming moderate ATP hydrolysis to mimic short ATP lifetime in meninges, repetitive firing of trigeminal fibers was transformed just into a single ATP-driven spike. Robust ATP hydrolysis left only a small receptor potential unable to trigger nociceptive spikes.
One crucial factor for the intensity and persistence of firing was the ATP receptor subtype. Unlike P2X3 receptors widely (up to 80%) expressed in the rodent trigeminal neurons , P2X2 receptors are less frequently expressed by sensory neurons, yet they show little desensitization despite relatively fast activation properties. Thus, the current simulations assuming a single release point indicate that the slow desensitization kinetics of P2X2 receptors could ensure high probability of repetitive firing.
The relative prevalence of P2X2 and P2X3 receptor subtypes is species and sensory neuron dependent. In the human DRG, there is a predominance of the P2X3 subtype, whereas in rodents, the P2X2 subtype is significantly present in trigeminal neurons . It should be noted that P2X2 and P2X3 subunits can heteromerize to form P2X2/3 heterotrimers. It has been proposed that P2X3 homomers are responsible for acute pain, whereas P2X2/3 heterotrimers are involved in chronic pain. Consistent with this view, our simulations indicated that the co-expression of P2X2/P2X3 receptors supported prolonged firing of fibers activated by eATP (Model 5;. Interestingly, unlike homomeric P2X3 or P2X2 receptors, in the case of heteromers, we observed strong increase in neuronal firing after raising the concentration of ATP from 10 µM to 100 µM. Thus, co-expression of P2X2 and P2X3 subunits enabled robust firing sensitive to ATP concentration.
The issue of relative contribution of parallel co-expression in the same neurons of homomeric P2X3 and homomeric P2X2 or the co-assembly of P2X2 and P2X3 subunits in the P2X2/3 heteromers is not fully solved in experimental setting. However, the nociceptive action of the P2X3 agonist α,β-meATP in rat meninges was comparable with the action of ATP, suggesting that the essential fraction of ionotropic ATP receptors in nerve terminals is presented by P2X2/3 heteromers. This issue might have a significant impact for suitability of P2X3 and P2X2/3 inhibitors for treatment of migraine pain. For instance, potent P2X3 and P2X2/3 inhibitors are already under the advanced stage of clinical trials for the treatment of chronic coup. On the other hand, highly potent P2X3 inhibitors such as various peptide toxins have been reported. Even emerging migraine treatments such as cannabinoidsdirectly inhibited P2X2 and P2X2/3 receptors in sensory neuronswhich can provide a dual anti-nociceptive effect. As ATP in living tissues is quickly degrades to ADP, it is also interesting to consider potential action of ADP via P2Y receptors, which might be the additional factor shaping the purinergic nociception in meninges. ADP sensitive excitatory receptors are expressed both in trigeminal neurons and in glial cells. In contrast, the inhibitory action of ADP on P2X3 receptors has been shown in isolated DRG neurons. However, our experimental studydid not show significant changes in the nociceptive firing of meningeal afferents after application of ADP suggesting that this type of modulation preferential takes places at the level of neuronal somata in the trigeminal ganglia or DRG. We cannot exclude also the region-specific differences in the relative contribution of P2Y vs. P2X receptors in analogy to the condition detected in the colon where the role of ADP sensitive P2Y receptors is dominating.
According to the purinergic hypothesis of migraine, the powerful algogen eATP is one of key mediators of this disease. In accordance with this notion, we have recently found that, when applied to mouse meninges, 100 µM ATP causes very strong firing (∼25-fold increase) of primary afferents. Our modeling of purinergic mechanisms at the meningeal neuro-immune synapse was facilitated by the known kinetics of P2X3 and P2X2 receptors. The availability of these input model parameters allowed exploring various factors determining the pattern of nociceptive fiber activity when mast cells were assumed as the source of ATP release. Nonetheless, one should bear in mind that, apart from mast cells, meninges are enriched with other immune cells. Thus, in the natural environment of these tissues, there could be multiple ATP releasing cell types possibly contributing in concert to neuro-immune interactions.
## Nociceptive signaling in meningeal afferents via 5-ht
Along with eATP, we simulated the action of 5-HT, which is a classical mediator released from mast cells. Recently we showed that 5-HT has a powerful pro-nociceptive action on rat and mouse meningeal afferents mainly via ligand-gated 5-HT3 receptors. The selection of 5-HT among other candidate transmitters was further supported by our experimental finding that histamine, also the known as the mast cell mediator, has only little if any excitatory action on nerve terminals. Unlike ATP, the lifespan of extracellular 5-HT in the meninges is expected to be much longer due to relatively slow reuptake. Furthermore, 5-HT3 receptors recover from desensitization much more rapidly than P2X3 receptors, rendering 5-HT potentially effective to trigger nociception in meninges. Nevertheless, ATP P2X3 receptors possess higher affinity (EC50 1 µM;and widespread expression in the majority of trigeminal neurons . Interestingly, at other non-traditional ''synapses,'' such as taste buds, release of ATP and 5-HT also activates nearby nerve terminals via P2X2/3 and 5-HT3 receptors, indicating that these two transmitters and their receptors may operate in a broader context than the proposed meningeal neuro-immune synapse. It is suggested that the functional outcome of dual activation via P2X and 5-HT3 receptors of afferents can multiply the total firing in meningeal nociception as a putative signal of migraine pain. The leading role in such scenario, most likely, belongs to ATP, which, apart from the direct excitation of terminals via P2X receptors, can degranulate meningeal mast cells to release serotonin, acting via excitatory 5-HT3 receptors.
This novel dual purinergic/serotonergic signaling reveals a new translational aspect of the present study, suggesting pharmacological interventions based on a combination of P2X3 and 5-HT3 antagonism.
## Role of sodium channels
For effective traffic of nociceptive signals to higher pain centers, sensory neurons must be equipped with a palette of sodium channels, which determine the fidelity and precision of neuronal sensory coding. Several subtypes of sodium channel, including Nav1.7, Nav1.8 and Nav1.9 isoforms, are expressed in the peripheral nervous system. Both the Nav1.7 and Nav1.8 subtypes found in nociceptive neurons can generate depolarizing slow currents with the characteristics to interact with the receptor potential evoked by ATP and 5-HT. In particular, human Nav1.8 channels display slower inactivation kinetics and produce large persistent currents than those observed in other species. In the present model we used the same set of sodium channel data reported by, and we observed that slow inactivation of Nav1.8 channels is one important determinant of the long-lasting pattern of spiking. Thus, unlike Nav1.7 channels, simulation with high density of Nav1.8 expression largely increased firing activated by ATP via P2X3 receptors. Thus, Nav1.8 channels could supply an additional mechanism to amplify the initial activation of nerve fibers by P2X3 receptors.
## Modeling multiple activation sites
Our modeling with a simple linear structure of the nerve fiber indicated that non-desensitizing P2X2 receptors and Nav1.8 channels could efficiently promote persistent firing after P2X3 receptor activation. We further explored other factors to transform the brief firing evoked by P2X3 receptors into a repetitive discharge. Emphasis on P2X3 receptors as initial condition was based on their wide expression by trigeminal sensory neurons , lack of inflammatory pain in mice genetically ablated for this receptorand by poor expression of P2X2 subunits by human nociceptive neurons.
To this end, we sought to improve our model by approximating its structure to the branching of the nerve fibers recently documented by functional and morphological data in dural afferents. One unexpected result was that P2X3 receptors alone were sufficient for repetitive firing when ATP was supposed to act on distinct branches of the trigeminal nerve (Model 12). In Model 12, we also quantified the spatial and temporal requirements, which determined the appearance of the second spike. Thus, our study showed that fiber branching could play a major role to generate repeated firing despite the intrinsic desensitization properties of the P2X3 receptor. Simulations indicated that this phenomenon could occur when ATP was applied to distinct branches with at least 40 ms interval. It seems likely that, in the naturally more complex structure of meningeal nerves and multiple release sites, repeated firing of trigeminal fibers equipped solely with P2X3 receptors might be even stronger as desensitization of P2X3 receptor represents a local phenomenon maintaining distant sites of the axonal tree sensitive to ATP.
Focusing on the factors promoting repetitive firing we also should consider factors which potentially can prevent it. Thus, the use-dependent inhibition of P2X3 receptor by low nanomolar concentrations of ATP known as HADcan reduce the operations of ATP via P2X3 (but not via P2X2) receptors. In our model we confirmed this possibility showing that HAD can reduce but not abolish the activity of homomeric P2X3 receptors. However, HAD was described in isolated neurons and was not previously tested in ex vivo preparations like our model of meningeal nociception and it is unknown whether it takes place in vivo. Thus, the (patho)physiological significance of HAD remains unclear and its efficiency in vivo could be reduced or neutralized by activity of ATP degrading NTPDases expressed in meningeal afferents.
Thus, the present study suggests how, despite their strong desensitization, P2X3 receptors can contribute to nociceptive pain signaling especially when supported by co-expression of P2X2 receptors, Nav1.8 sodium channels, co-release of 5-HT and branching of nerve fibers.
These data are schematically summarized in , where the efficiency of repetitive firing via only P2X3 vs. co-assembly of P2X3 with P2X2 , effect of Nav1.8 for P2X2/3 and for 5-HT3 receptors are indicated. showed the amplifying firing effect of neuronal branching along with other factors.
These novel results provide the clue to solve the paradox how strong P2X3 desensitization is consistent with in vivo evidence on the role of P2X3 in pain behavior.
## Potential pathophysiological implications in migraine
Our modeling approach together with previously published data provides a functional scenario to account how, during migraine attacks, ATP and 5-HT can persistently excite meningeal afferents. We assumed here the ATP and 5-HT co-release by meningeal immune cells (exemplified as mast cells). Mast cells react to mechanical stimuli, which can serve as a signal to release the content of their secretory granules including ATPto activate neighboring nerve fibers. Interestingly, cortical spreading depression (which underlies migraine aura,, also degranulates mast cells and opens pannexin channels as the pathway for ATP release. Further studies are necessary to identify all the endogenous triggers for 5-HT and ATP release during a migraine attack, also taking into account the blood flow-dependent endothelial release of ATPdue to shear stress or blood vessel pulsations.
As migraine pain is not only long-lasting but also pulsating (throbbing), one aim of this project was to elucidate the factors FIGURE 10 | Schematic presentation of factors determining repetitive activation of meningeal afferents. (A) P2X3 receptors activation produces one spike whereas activation of P2X2/3 heteromers triggers prolonged activity. (B) The co-expression of P2X2/3 or (C) 5HT-3 receptors with Nav1.8 channels enhances prolonged spiking activity. (D) Nerve branching further increases the duration of spiking activity which signals pain to higher brain centers.
which provide such migraine-typical activation of trigeminal fibers. Hence, to simulate the pulsating character of migraine pain, which is one of the main features of migraine [The International Classification of Headache Disorders 3rd edition (ICHD-3)], we compared the effect of experimental application of ATP or 5-HT to meningeal nerve fibers with similar modeling tests. Then, based on Models 5 and 8, which provided the best match with the experimental approach, we reconstructed pulsatile release of the two mediators ATP and 5-HT to monitor the spiking activity of nerve fibers.
Furthermore, we simulated the effect of rhythmic pulseinduced release of ATP and of 5-HT not only in rodents but in human nerves. In all cases after initial continuous activity, there were short very regular bursts of nociceptive activity, specific for each species. Notably, such a type of activity with high frequency of spikes within the burst should increase the probability of temporal summation and the windup phenomenon to transmit peripheral signals to the second order nociceptive neurons located in the brainstem. Although release of chemical mediators by pulsating blood vessels cannot per se be the primary process for migraine pain, it is feasible to assume that after the initial stimulation of ATP (and 5-HT) release by certain triggers (whose identity needs further study), pulsatile release might support long-lasting firing. Within this framework, the present basic model of meningeal nociception does not consider the early process of neuronal sensitization. Thus, it has been shown that the main migraine mediator, the neuropeptide CGRP largely increases the activity of P2X3 receptors and reduces their desensitization. Notably, mechanical activation of peptidergic nerve fibers containing CGRP can perhaps be achieved through mechanosensitive piezo channels; and see Supplementary Movie). This effect plus other sensitization mechanisms should enhance the role of ATP in nociceptive firing of meningeal afferents and supply the trigger for resilient firing. In fact, mast cells, apart from 5-HT, release other active substances such as NGF and histamine. Although histamine may preferentially act on dural vessels, NGF can directly target trigeminal neurons to enhance their responses to ATP via P2X3 receptors. Furthermore, for implementation of multiple migraine-related signaling pathways, the model should include G-protein coupled receptors (GPCR), including those for migraine-specific drugs such as triptans and neurotrophins.
In summary, the idea of the meningeal neuro-immune synapseis supported by the present model, which improves our understanding of the processes underlying peripheral mechanisms generating migraine pain within the concept of trigeminovascular dysfunction. The current model can, therefore, serve as a novel tool for further testing the mechanisms of meningeal trigeminal nociception in silico.
# Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher. Requests to access the datasets should be directed to [email protected].
# Ethics statement
The animal study was reviewed and approved by The Committee for the Welfare of Laboratory Animals of the University of Eastern Finland and the Provincial Government of Kuopio.
# Author contributions
|
10.1186/1471-2164-10-599
|
CCBY
|
2805695
|
20003355
|
s2orc_pubmed_articles
|
The transcriptional programme of Salmonella enterica serovar Typhimurium reveals a key role for tryptophan metabolism in biofilms
Background: Biofilm formation enhances the capacity of pathogenic Salmonella bacteria to survive stresses that are commonly encountered within food processing and during host infection. The persistence of Salmonella within the food chain has become a major health concern, as biofilms can serve as a reservoir for the contamination of food products. While the molecular mechanisms required for the survival of bacteria on surfaces are not fully understood, transcriptional studies of other bacteria have demonstrated that biofilm growth triggers the expression of specific sets of genes, compared with planktonic cells. Until now, most gene expression studies of Salmonella have focused on the effect of infection-relevant stressors on virulence or the comparison of mutant and wild-type bacteria. However little is known about the physiological responses taking place inside a Salmonella biofilm.Results:We have determined the transcriptomic and proteomic profiles of biofilms of Salmonella enterica serovar Typhimurium. We discovered that 124 detectable proteins were differentially expressed in the biofilm compared with planktonic cells, and that 10% of the S. Typhimurium genome (433 genes) showed a 2-fold or more change in the biofilm compared with planktonic cells. The genes that were significantly up-regulated implicated certain cellular processes in biofilm development including amino acid metabolism, cell motility, global regulation and tolerance to stress. We found that the most highly down-regulated genes in the biofilm were located on Salmonella Pathogenicity Island 2 (SPI2), and that a functional SPI2 secretion system regulator (ssrA) was required for S. Typhimurium biofilm formation. We identified STM0341 as a gene of unknown function that was needed for biofilm growth. Genes involved in tryptophan (trp) biosynthesis and transport were up-regulated in the biofilm. Deletion of trpE led to decreased bacterial attachment and this biofilm defect was restored by exogenous tryptophan or indole.Conclusions: Biofilm growth of S. Typhimurium causes distinct changes in gene and protein expression. Our results show that aromatic amino acids make an important contribution to biofilm formation and reveal a link between SPI2 expression and surface-associated growth in S. Typhimurium.
# Background
The ability to survive and overcome stress allows nontyphoidal Salmonella pathogens to be isolated from a diverse range of environments. Specific serovars of S. enterica, including Salmonella enterica serovar Typhimurium, are of particular concern to medicine and industry because they cause a significant proportion of foodborne disease worldwide. There has been a controversial suggestion that infection by Salmonella may subsequently cause an increase in mortality for up to one year [bib_ref] Short and long term mortality associated with foodborne bacterial gastrointestinal infections: registry..., Helms [/bib_ref] , and it is clear that we need to improve our understanding of the behaviour of this pathogen.
Experiments that tracked the spread of Salmonella from the farm and within food processing facilities have provided a direct link between bacterial biofilms and the contamination of the resulting food products [bib_ref] Sources of Salmonella on broiler carcasses during transportation and processing: modes of..., Corry [/bib_ref] [bib_ref] Crosscontamination with Salmonella on a broiler slaughterhouse line demonstrated by use of..., Olsen [/bib_ref] [bib_ref] Impact of the slaughter line contamination on the presence of Salmonella on..., Rasschaert [/bib_ref]. Surface-associated growth, termed biofilm growth, has been shown to promote the survival of Salmonella when exposed to limited nutrient availability, heat, acidic pH, low temperatures and antimicrobials [bib_ref] Survival and growth characteristics of Listeria monocytogenes and Salmonella Typhimurium on stainless..., Helke [/bib_ref] [bib_ref] Salmonella enteritidis phage type 4 isolates more tolerant of heat, acid, or..., Humphrey [/bib_ref] [bib_ref] Survival of Legionella pneumophila and Salmonella typhimurium in biofilm systems, Armon [/bib_ref] [bib_ref] Biofilm formation by Salmonella spp, Joseph [/bib_ref] [bib_ref] Adaptive resistance and differential protein expression of Salmonella enterica serovar Enteritidis biofilms..., Mangalappalli-Illathu [/bib_ref]. Salmonella cells attach and grow on a variety of abiotic and biotic surfaces, and remain viable for many weeks. Indeed, the number of outbreaks of salmonellosis caused by microbial growth on the surfaces of raw fruits and vegetables has increased dramatically in recent years [bib_ref] General outbreaks of infectious intestinal disease linked with salad vegetables and fruit, Long [/bib_ref] [bib_ref] Fresh produce: A growing cause of outbreaks of foodborne illness in the..., Sivapalasingam [/bib_ref] [bib_ref] Use of episcopic differential interference contrast microscopy to identify bacterial biofilms on..., Warner [/bib_ref]. This type of persistence of Salmonella in the food chain has become a major health concern, because the detachment of viable cells from a biofilm can cause subsequent contamination of foods during processing.
The profound consequences of biofilm formation in both nature and disease have led to increased efforts to define the characteristics that make biofilm cells physiologically distinct from planktonic (freely suspended) cells, and to identify the properties of microorganisms during surface-associated growth. Cells within a biofilm are heterogeneous, can grow at different rates, and resist antimicrobial treatments [bib_ref] Role of nutrient limitation and stationary-phase existence in Klebsiella pneumoniae biofilm resistance..., Anderl [/bib_ref] [bib_ref] Pseudomonas biofilm formation and antibiotic resistance are linked to phenotypic variation, Drenkard [/bib_ref] [bib_ref] Use of ribosomal RNA fluorescence in situ hybridization for measuring the activity..., Poulsen [/bib_ref]. Additionally, clusters of biofilm cells are typically encased in a bacterialderived extracellular matrix that is thought to provide adhesion and strength, as well as act as a physical barrier against the diffusion of antimicrobials [bib_ref] Diffusion in biofilms, Stewart [/bib_ref] [bib_ref] Biofilm exopolysaccharides: a strong and sticky framework, Sutherland [/bib_ref]. S. Typhimurium biofilms can form on abiotic surfaces (e.g. glass, polystyrene, stainless steel) and biotic surfaces (e.g. human epithelial cells or gallstones); in many strains, the bacterial cells are associated with an extracellular matrix composed of curli (thin aggregative fimbriae) and cellulose [bib_ref] Surface finishes on stainless steel reduce bacterial attachment and early biofilm formation:..., Jw [/bib_ref] [bib_ref] Differential binding to and biofilm formation on HEp-2 cells by Salmonella enterica..., Boddicker [/bib_ref] [bib_ref] Roles of curli, cellulose and BapA in Salmonella biofilm morphology studies by..., Jonas [/bib_ref] [bib_ref] Comparative analysis of Salmonella enterica Serovar Typhimurium biofilm formation on gallstones and..., Prouty [/bib_ref] [bib_ref] The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as..., Zogaj [/bib_ref]. Recent work has shown that strains exhibiting the rdar (red, dry and rough) morphotype, expressing curli and cellulose, are involved in colonisation but do not contribute to the persistence of Salmonella on food processing surfaces. S. Typhimurium also displays a swarming phenotype, a specialised motility that enables hyperflagellated bacteria to efficiently colonise surfaces and requires a combination of the chemotaxis, LPS synthesis, type III secretion and iron metabolism systems [bib_ref] Genetics of swarming motility in Salmonella enterica serovar typhimurium: critical role for..., Toguchi [/bib_ref] [bib_ref] Gene expression patterns during swarming in Salmonella typhimurium: genes specific to surface..., Wang [/bib_ref].
Biofilms have previously been studied with transcriptomic, proteomic and in vivo expression technologybased approaches for bacterial species, such as E. coli and Pseudomonas [bib_ref] Finding gene-expression patterns in bacterial biofilms, Beloin [/bib_ref] [bib_ref] Lessons from DNA microarray analysis: the gene expression profile of biofilms, Lazazzera [/bib_ref] [bib_ref] The genomics and proteomics of biofilm formation, Sauer [/bib_ref]. A large number of genes or proteins that are differentially regulated during biofilm development have been identified. Few studies have focused on the global response of Salmonella under environmental conditions relevant to food processing, where bacteria may encounter hydrodynamic stress and nutrient limitation [bib_ref] Transcriptome analysis of organisms with food safety relevance, Puttamreddy [/bib_ref]. Two studies have used a proteomic approach to identify S. Enteritidis proteins that are differentially regulated during biofilm growth in response to disinfectant and to different fluid flow rates [bib_ref] Adaptive resistance and differential protein expression of Salmonella enterica serovar Enteritidis biofilms..., Mangalappalli-Illathu [/bib_ref] [bib_ref] Architectural adaptation and protein expression patterns of Salmonella enterica serovar Enteritidis biofilms..., Mangalappalli-Illathu [/bib_ref]. However, the physiological and regulatory processes involved in the growth and persistence of Salmonella biofilms remain unclear. We have used a combination of physiological, transcriptomic and proteomic approaches to address this problem in S. Typhimurium.
# Results
## Growth of salmonella biofilms
The biofilm phenotype of Salmonella isolates is known to vary significantly depending on the strain, nutrient source, temperature and other factors [bib_ref] Oxygen tension and nutrient starvation are major signals that regulate agfD promoter..., Gerstel [/bib_ref] [bib_ref] Effect of temperature, pH, and water activity on biofilm formation by Salmonella..., Giaouris [/bib_ref] [bib_ref] Effect of heat, acidification, and chlorination on Salmonella enterica serovar Typhimurium cells..., Scher [/bib_ref]. Because Salmonella can encounter hydrodynamic environments at several stages in food processing [bib_ref] Architectural adaptation and protein expression patterns of Salmonella enterica serovar Enteritidis biofilms..., Mangalappalli-Illathu [/bib_ref] , we assessed the biofilm forming capacity of S. Typhimurium SL1344 on glass in a flowing system. We found that SL1344 produced substantially thicker biofilms in Colonising Factor Antigen (CFA) medium at 25°C, when compared to growth at 37°C, or growth in rich nutrient media (Luria Broth, LB), Brain Heart Infusion or M9 minimal media supplemented with glucose, sucrose or glycerol (data not shown). CFA medium has been previously shown to promote biofilm growth of S. Typhimurium [bib_ref] Contribution of the SirA regulon to biofilm formation in Salmonella enterica serovar..., Teplitski [/bib_ref].
SL1344 biofilms were then grown in a modified batch system for 72 h (Additional file 1) using silicone rubber tubing as a substratum for growth, as this surface permitted the isolation of sufficient quantities of mature biofilm and planktonic cells for proteomic and transcriptomic analyses (see Methods). The tubing was positioned vertically to avoid the isolation of bacterial aggregates following sedimentation. The pH of the medium effluent showed no detectable change (pH 7.0 ± 0.2) throughout the experiment. Influent samples were collected aseptically throughout the experiment showing that the number of cells increased from 1 × 10 6 CFU ml -1 (T = 0) to 5 × 10 8 CFA ml -1 (T = 72 h). No pellicle formation was observed in the influent flask during the 72 h experiment. To determine if the cells within a mature 72 h biofilm were metabolically active, an analogous biofilm grown in a glass flow cell was stained with Live/Dead BacLight. Images captured at ten random fields of view along the glass surface confirmed that more than 95% of the bacterial cells were alive after 72 h of growth (data not shown). Taken together, these experiments established that biofilms of SL1344 continued to accumulate biomass on glass for at least 72 h, and that the majority of these cells were viable.
## The global gene expression profile of salmonella biofilm
The transcriptome of mature S. Typhimurium biofilm (72 h) was compared to its planktonic counterpart (72 h) in three independent biological experiments. Capillary gel electrophoresis was used to confirm that the RNA obtained from the 72 h biofilm and planktonic samples was of good quality prior to labelling (data not shown). The transcriptomic data from the three biological replicates were statistically filtered and data are only presented for genes that showed significant changes in every replicate.
The transcriptomic data showed that a total of 560 genes were differentially expressed (P < 0.01) in biofilms compared with 72 h planktonic cultures, and 433 of these genes (10% of the S. Typhimurium genome) displayed more than a 2-fold change in expression. These included 229 genes that were up-regulated in biofilm (Additional file 2) and 204 down-regulated genes (Additional file 3). We noted that almost half of these genes were of hypothetical or unknown function. All differentially expressed genes were catalogued according to functional categories and were predominantly involved in cell motility, amino acid metabolism, stress response, outer membrane function and virulence [fig_ref] Figure 1: Whole genome expression profiling of S [/fig_ref]. [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] shows a subset of up-regulated genes that correspond to cellular processes previously implicated in biofilm growth (e.g. cell surface structures, motility, global regulation and oxygen availability) and [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] shows a subset of genes specific to Type III secretion. The complete transcriptomic data sets are presented in Additional files 2 and 3, and the key findings are discussed below.
In E. coli, cell surface structures such as fimbriae have been shown to be required for the initial colonisation of abiotic and biotic surfaces and the establishment of wellestablished biofilms [bib_ref] Global impact of mature biofilm lifestyle on Escherichia coli K-12 gene expression, Beloin [/bib_ref] [bib_ref] Global gene expression in Escherichia coli biofilms, Schembri [/bib_ref] [bib_ref] Temporal geneexpression in Escherichia coli K-12 biofilms, Domka [/bib_ref]. In mature biofilms of S. Typhimurium, several genes required for bacterial attachment and motility were up-regulated, including csgB and csgA that encode the curlin fimbrial subunits [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref]. The extracellular matrix of S. Typhimurium biofilms is composed of curli, along with cellulose, colanic acid and other polymers, depending on nutrient availability [bib_ref] The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as..., Zogaj [/bib_ref]. Under the conditions used in this study, biofilm cells showed increased expression of one gene, bcsE, required for cellulose biosynthesis. Genes involved in flagellar biosynthesis, assembly and regulation were up-regulated by up to 7-fold [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref]. Genes required for motility and chemotaxis were also up-regulated in S. Typhimurium biofilms, including motAB, cheA, cheY, cheR [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] and tsr. Other cell-surface associated genes that were highly expressed in the biofilm included those encoding the major outer membrane protein OmpX [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] , a regulator of LPS O-chain length WzzB, the OM lipoprotein Blc, the membrane bound lipid phosphotase PgpB and 13 genes of hypothetical cell envelope function.
Gradients in oxygen concentration have been experimentally demonstrated within biofilm clusters [bib_ref] Effects of biofilm structures on oxygen distribution and mass transfer, Debeer [/bib_ref]. Consistent with this, several genes that sense or respond to oxygen availability were up-regulated in flowing biofilm, including aer [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] and fnr [bib_ref] Two genetically distinct pathways for transcriptional regulation of anaerobic gene-expression in Salmonella..., Jamieson [/bib_ref]. cyo genes encode the cytochrome o ubiquinol oxidase subunits of the aerobic respiratory chain, and are regulated by the master flagellar regulator [bib_ref] FlhD/FlhC is a regulator of anaerobic respiration and the Entner-Doudoroff Pathway through..., Pruss [/bib_ref]. The cyo genes were more highly expressed in the biofilm suggesting that the environment of the biofilm was aerobic. Furthermore, several genes known to be repressed by aerobiosis were down-regulated in the biofilm, including genes of the fumarate (fum, frd), hydrogenase 2 (hyb), cytochrome o (cyd) and dicarboxylate (dcu) operons (Additional file 3). While biofilms are reported to be spatially heterogeneous in terms of oxygen levels, our results are consistent with some penetration of oxygen into cell clusters in a flowing environment, as suggested by Xu et al., [bib_ref] Spatial physiological heterogeneity in Pseudomonas aeruginosa biofilm is determined by oxygen availability, Xu [/bib_ref]. This may reflect the high oxygen permeability of the silicone surface used for growing our biofilms [bib_ref] Evaluation of silicone tubing toxicity using tobacco BY2 culture, Park [/bib_ref].
The precise growth phase of bacteria within the interior of a biofilm cell cluster has not been established, although several groups have reported slow growth rates, presumably due to nutrient limitation [bib_ref] Use of ribosomal RNA fluorescence in situ hybridization for measuring the activity..., Poulsen [/bib_ref] [bib_ref] Impact of rpoS deletion on Escherichia coli biofilms, Adams [/bib_ref]. Specific operons involved in translation (i.e. rps, rpl, rpm)
were down-regulated in biofilms of S. Typhimurium when compared to planktonic cells, suggesting that slower growth occurred in the attached population (Additional file 3). Transcriptomic comparisons with planktonic cells that were isolated at earlier time points Whole genome expression profiling of S. Typhimurium SL1344 flowing biofilms compared to planktonic cells when grown in CFA at 25°C for 72 h. Expression changes of genes belonging to functional groups and pathogenicity islands (numbers in parenthesis refer to the genes assigned to each functional group from the genome of S. Typhimurium LT2). The bars show the percentage of genes belonging to each group that were altered for expression > 2-fold between planktonic and flowing biofilm cells. The blue bars indicate the proportion of genes that are down-regulated and the red bars represent the proportion of up-regulated genes for each group during biofilm growth (P < 0.01).
of the experiment (i.e. 6 and 24 h), indicated that mature biofilm cells were most similar to a stationary phase planktonic population (data not shown). This observation is consistent with findings from studies in E. coli where biofilms cells have similarities to bacteria in stationary phase [bib_ref] Global impact of mature biofilm lifestyle on Escherichia coli K-12 gene expression, Beloin [/bib_ref] [bib_ref] Temporal geneexpression in Escherichia coli K-12 biofilms, Domka [/bib_ref].
Global gene regulators respond to environmental conditions, including nutrient limitation, oxygen availability and osmotic stress, and control a wide range of adaptive physiological and regulatory circuits. Several genes that encode global gene regulators were upregulated in flowing biofilms of S. Typhimurium, including csrA [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] and ihfA, confirming previous studies in well-established biofilms of Salmonella or E. coli [bib_ref] The csgD promoter, a control unit for biofilm formation in Salmonella typhimurium, Gerstel [/bib_ref] [bib_ref] Biofilm formation and dispersal under the influence of the global regulator CsrA..., Jackson [/bib_ref]. Other regulatory genes that respond to starvation conditions, including rpoH, cbpA and phoH, were up-regulated during biofilm growth (Additional file 2). RpoS (σ 38 ), the sigma factor that activates genes under growth arresting conditions, was highly expressed in both biofilm and planktonic populations. We observed that more than 25% of the S. Typhimurium RpoS regulon [bib_ref] Identification of RpoS (sigma(S))-regulated genes in Salmonella enterica serovar Typhimurium, Ibanez-Ruiz [/bib_ref] was up-regulated in biofilm cells (Additional file 2).
Following three days of growth, biofilms of S. Typhimurium showed up-regulation of genes that respond to oxidative stress (lexA, msrA, soda, sodC, gloA), heat shock (clpA, clpX), DNA replication and repair (recA, mug, phrB), cell envelope stress (pspB) and a putative stressrelated gene (yicC) when compared to planktonic cells. A possible role for MsrA, RecA, PspB and the cytoplasmic Clp protease in biofilm development of Gram-negative species has been previously reported [bib_ref] Global impact of mature biofilm lifestyle on Escherichia coli K-12 gene expression, Beloin [/bib_ref] [bib_ref] Initiation of biofilm formation in Pseudomonas fluorescens WCS365 proceeds via multiple, convergent..., O'toole [/bib_ref]. At least 19 genes that were up-regulated during biofilm growth of S. Typhimurium have a role in the osmotic stress and acid tolerance responses of E. coli [bib_ref] A microarray-based antibiotic screen identifies a regulatory role for supercoiling in the..., Cheung [/bib_ref] , including treF, talA, poxB, osmCY, himA, dps and aceB (Additional file 2).
Amino acid synthesis is energetically expensive for the cell but essential for protein production, nitrogen transfer and osmotic protection. We identified genes involved in alanine (dad) and glutamine/glutamate (gln, gltL, astE, nadE, STM1795) metabolism and transport that were more highly expressed in the biofilm (Additional file 2). We were intrigued to discover that the biosynthetic genes of the trp operon were overexpressed in well-established biofilms of S. Typhimurium [fig_ref] Table 1: Expression levels of S [/fig_ref]. These genes are required from the initial steps of tryptophan synthesis (i.e. from chorismate to indole) to the transfer of indole to tryptophan [bib_ref] Advancing our knowledge in biochemistry, genetics, and microbiology through studies on tryptophan..., Yanofsky [/bib_ref]. Additionally, there was a strong biofilm-dependent induction of mtr, a tryptophan-specific transporter [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref]. Recent studies reported the induction of tryptophan biosynthesis genes during early biofilm formation in E. coli, followed by repression of this operon at later time points [bib_ref] Temporal geneexpression in Escherichia coli K-12 biofilms, Domka [/bib_ref].
Expression profiles indicated that virulence genes located within Salmonella pathogenicity island 1 (SPI1) and the serotype-specific S. Typhimurium plasmid (pSLT) were not differentially expressed during biofilm growth. However, more than 30 genes belonging to the ssr, ssa and sse operons of SPI2, were down-regulated by up to 100-fold in the biofilm [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] ; Additional file 3); SPI2 encodes a type III secretion system that is required for intracellular survival in host phagocytes [bib_ref] Genes encoding putative effector proteins of the type III secretion system of..., Hensel [/bib_ref]. This prompted us to monitor the level of SPI2 expression observed in our CFA media-based experiments at 25°C compared with growth in LB at a higher temperature (data not shown). Surprisingly, our transcriptomic data showed that SPI2 was expressed at a significantly higher scale (e.g. value of 10 on the Y-axis corresponds to 10-fold up-regulated). All genes were significant at P < 0.01 (* denotes that the csgA gene was significant at P < 0.05).
level in planktonic CFA cultures grown at 25°C than at mid-log phase in LB at 37°C. This suggests that the CFA media contains a factor that induces SPI2 expression, even at this low temperature (data not shown). The only SPI2 gene observed to be up-regulated in biofilm was orf319, which encodes a putative transmembrane protein of unknown function and does not contribute to S. Typhimurium virulence [fig_ref] Figure 2: Biofilm-regulated S [/fig_ref] [bib_ref] Molecular and functional analysis indicates a mosaic structure of Salmonella pathogenicity island..., Hensel [/bib_ref]. In addition, sopB, a SPI5-encoded phosphatase required for invasion of epithelial cells [bib_ref] SopB, a protein required for virulence of Salmonella dublin, is an inositol..., Norris [/bib_ref] , was up-regulated in Salmonella biofilm.
Surface-associated growth of S. Typhimurium leads to changes in protein expression To expand our analysis from the transcriptomic to the proteomic level, and to identify biofilm-regulated proteins, total protein extracts from biofilm and planktonic populations of S. Typhimurium were compared using 2-D PAGE. Samples from the same flowing biofilm system were used for the proteomic and the transcriptomic experiments. A representative example of the biofilm and planktonic proteome with over 600 detected proteins spots per gel is shown in [fig_ref] Figure 3: Biofilm-regulated protein expression [/fig_ref]. Direct comparison of protein spots showed that the levels of at least 250 proteins remained similar in both populations. However, the expression of 124 proteins was altered (> 2-fold), with the expression of 59 proteins increasing and 65 proteins decreasing during surface-associated growth compared to planktonic culture. At least 175 proteins were only detected in the biofilm samples, and not in planktonic culture.
Identification of differentially regulated and unique proteins Fifty protein spots that showed differential expression between biofilm and planktonic gels, as well as spots detected as unique to either mode of growth [fig_ref] Figure 3: Biofilm-regulated protein expression [/fig_ref] , were selected and analysed by MALDI-TOF mass spectrometry (MS). These spots were specifically chosen as they were abundant and clearly separated from other spots, to facilitate unambiguously identification by MS. Forty-four proteins were successfully identified, and we discovered that the majority of proteins that were unique or highly expressed in the biofilm corresponded to the same functional groups identified by transcriptomic analyses. These included proteins involved in cell motility, amino acid and carbohydrate metabolism, as well as proteins of unassigned function (Additional file 4). In fact, 45% of the 24 up-regulated proteins corresponded to genes identified as differentially expressed at the transcriptional level [fig_ref] Table 2: Biofilm-regulated proteins that show similar trends in proteomic and transcriptomic analysis [/fig_ref]. A common theme emerged relating to cell motility, with the up-regulation of three external flagellar proteins FlgK [fig_ref] Figure 3: Biofilm-regulated protein expression [/fig_ref] , FlgL and FliD (also known as HAP1, HAP3, and HAP2) that are involved in the later stages of flagellar assembly. FlgK and FlgL were up-regulated in the biofilm compared to planktonic growth and FliD was only detected in the biofilm samples. Our transcriptomic analysis showed an increase in the expression of FliC in the biofilm, but this protein was not identified in the 2-D analysis. To confirm the biofilm-dependent regulation of proteins involved in late flagellar assembly, we performed a Western blot. As shown in [fig_ref] Figure 3: Biofilm-regulated protein expression [/fig_ref] , FliC protein was 3.4-fold more abundant in biofilm cells when compared with a planktonic culture.
Two periplasmic transport proteins were up-regulated in biofilm, namely DppA (10-fold; [fig_ref] Figure 3: Biofilm-regulated protein expression [/fig_ref] , a dipeptidebinding protein and ArgT (5-fold), which transports arginine, lysine and orthinine. Proteomic analysis of S. Typhimurium biofilm cells also showed differential expression of AnsB, TreA and GalE which are involved in asparagine metabolism, trehalose degradation and the conversion of UDP-galactose and UDP-glucose (Additional file 4). The remaining proteins that were unique or more highly expressed during biofilm growth were of unknown function (Additional file 4). The putative periplasmic protein YggE, which is highly conserved in Enterobacteriaceae and has been shown to possess immunogenic properties in Edwardsiella ictaluri [bib_ref] Cloning and characterization of Edwardsiella ictaluri proteins expressed and recognized by the..., Moore [/bib_ref] , was also more abundant (3.7-fold) during biofilm growth [fig_ref] Figure 3: Biofilm-regulated protein expression [/fig_ref].
## Targeted deletion of biofilm-regulated genes
To investigate the function of genes shown to be differentially expressed in S. Typhimurium biofilm, eight targeted gene deletions were constructed (Additional file 5AB). Genes that showed a biofilm-specific pattern of expression were selected for chromosomal mutation, including mtr, yhfG, ybaY and genes of the trp operon (Additional file 6). Disruption of trpE, which encodes anthranilate synthase component I, was chosen because four genes of the tryptophan operon were highly up-regulated during biofilm growth and the product of trpE catalyses the first reaction of the tryptophan pathway with TrpD [fig_ref] Table 1: Expression levels of S [/fig_ref]. Moreover, TrpE is the most important enzyme of this pathway from a regulatory point of view, as it is subject to feedback inhibition by tryptophan [bib_ref] Advancing our knowledge in biochemistry, genetics, and microbiology through studies on tryptophan..., Yanofsky [/bib_ref]. The other four genes were selected on the basis that they were they most highly expressed (ygaT) or repressed (STM0341, STM2779) in the biofilm, or because they were previously implicated in biofilms of E. coli (pspABCDEF) [bib_ref] Global impact of mature biofilm lifestyle on Escherichia coli K-12 gene expression, Beloin [/bib_ref]. To test whether the chromosomal deletions altered the fitness of the strains, the growth patterns of each mutant were analysed over 24 h in CFA and LB medium. No significant differences were observed in the planktonic growth rate of each mutant compared to the SL1344 parental strain (Additional file 7). ArgT ABC superfamily; lysine/arginine/orthinine tranport protein + 4.9
a. Proteins identified by MALDI-TOF mass spectrometry and peptide mass fingerprinting (Mascot). Proteins up-regulated in the biofilm that show the same pattern of expression at the transcriptional level (Additional files 2 and 3).
BMC Genomics 2009, 10:599 http://www.biomedcentral.com/1471-2164/10/599
To determine whether the deletion of these genes altered biofilm formation, each strain was tested for its ability to attach to polystyrene in a static biofilm assay. As shown in [fig_ref] Figure 4: A [/fig_ref] , the ΔtrpE mutant formed significantly less biofilm (P = 0.0001) than the parental strain after 24 h, showing that anthranilate synthase is required for biofilm formation. Similarly, biofilm growth of ΔSTM0341 was significantly reduced (P = 0.001). None of the other mutations affected growth of biofilms.
To determine whether the ΔtrpE mutant also had a biofilm defect on a hydrophilic glass surface, biofilm formation was compared to WT SL1344 in a oncethrough glass flow cell system. The attachment of ΔtrpE was monitored by light microscopy for 24 h and these results confirmed that the mutant formed reduced levels of biofilm (P < 0.01; [fig_ref] Figure 4: A [/fig_ref] ; n = 3). Although ΔtrpE cells did transiently attach and form small microcolonies on glass, they failed to grow and to form thick cell clusters on the flow cell surface, resulting in 3-fold less of the surface being covered with biofilm compared with the WT strain [fig_ref] Figure 4: A [/fig_ref]. The biofilm forming capacity of the ΔSTM0341 mutant was also reduced on glass when compared to WT SL1344 in one independent experiment (data not shown).
Inactivation of ssrA and rpoS alters biofilm formation in S. Typhimurium Our transcriptomic analysis revealed that genes of the SPI2 type 3 secretion system were differentially expressed (i.e. down-regulated up to 100-fold) during biofilm growth. SPI2 repression is activated by the two-component regulatory systems SsrAB and PhoP/Q and in response to high levels of phosphate and Mg 2+ , however this repression during biofilm growth has not been previously reported [bib_ref] Environmental regulation of Salmonella pathogenicity island 2 gene expression, Deiwick [/bib_ref] [bib_ref] The PhoP/PhoQ system controls the intramacrophage type three secretion system of Salmonella..., Bijlsma [/bib_ref] [bib_ref] Regulation of Salmonella pathogenicity island 2 genes by independent environmental signals, Lober [/bib_ref] [bib_ref] Identification of a virulence locus encoding a second type III secretion system..., Shea [/bib_ref]. To determine the effect of SPI2 gene expression on Salmonella biofilm formation, the ability of a ΔssrA deletion mutant to attach to polystyrene was compared to WT SL1344. The same ΔssrA strain has been used to identify the SsrA regulon of S. Typhimurium [bib_ref] SseL, a Salmonella deubiquitinase required for macrophage killing and virulence, Rytkönen [/bib_ref]. A Δorf319 deletion mutant was also tested as this was the only SPI2-associated gene that was up-regulated in the biofilm. No significant differences were observed in the planktonic growth rates of ΔssrA or Δorf319 compared to the SL1344 parental strain (Additional file 7). Deletion of ssrA caused more than a 40% decrease in attachment after 24 h and 48 h of growth, and this was restored to WT levels by complementation with a low copy plasmid encoding ssrAB [fig_ref] Figure 5: Static biofilm formations [/fig_ref]. The Δorf319 mutant did not show a significant defect in biofilm formation. To further examine the effect of SPI2 expression and its impact on biofilm formation, we over-expressed ssrAB from an arabinose-inducible plasmid [fig_ref] Figure 5: Static biofilm formations [/fig_ref]. These results showed that increased SPI2 expression significantly reduced biofilm formation in S. Typhimurium.
Transcriptomic comparison of biofilm and planktonic cells showed that rpoS was highly expressed in both populations, and that several RpoS-activated genes showed biofilm-specific patterns of expression. To determine the impact of RpoS expression on biofilm growth, we tested the fitness of a ΔrpoS mutant compared to WT SL1344 and its ability to colonise polystyrene (Additional file 7, [fig_ref] Figure 5: Static biofilm formations [/fig_ref]. We found that inactivation of rpoS significantly reduced the ability of S. Typhimurium to form a biofilm (P < 0.0001). Phenotypic characteristics of S. Typhimurium deletion mutants The biofilm-forming capacity of S. Typhimurium strains has been linked to the expression of curli and cellulose production [bib_ref] Roles of curli, cellulose and BapA in Salmonella biofilm morphology studies by..., Jonas [/bib_ref]. Likewise, the ability of E. coli to colonise and grow on surfaces has been shown to require flagella and RpoS [bib_ref] Impact of rpoS deletion on Escherichia coli biofilms, Adams [/bib_ref] [bib_ref] Genetic analysis of Escherichia coli biofilm formation: roles of flagella, motility, chemotaxis..., Pratt [/bib_ref]. Because of the importance of these cellular functions for biofilm formation, we examined whether the mutant strains showed alterations in swimming motility, RpoS activity or the ability to bind calcofluor and congo red (CR) dye, compared with the WT strain [fig_ref] Table 3: Phenotypic characteristics of deletion mutants and WT strain tested for motility, RpoS... [/fig_ref]. All deletion mutants showed the same pattern of flagellar-mediated motility as WT SL1344, indicating that the reduced biofilm formation observed in ΔtrpE, ΔSTM0341 or ΔssrA did not reflect a motility defect. When RpoS activity was examined indirectly by assessment of KatE-encoded catalase production, all of the mutants showed the same RpoS-positive phenotype as the WT SL1344 strain.
Calcofluor has previously been used to detect celluloseproducing strains of S. Typhimurium as this dye binds polysaccharides with 1,4β-glucopyranosyl units and fluoresces under UV light [bib_ref] The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as..., Zogaj [/bib_ref] To assess cellulose production, we grew each strain on CFA Calcofluor agar at 25 and 37°C and compared them to WT SL1344 and the cellulose-deficient rpoS mutant [bib_ref] Crl activates transcription initiation of RpoS-regulated genes involved in the multicellular behavior..., Robbe-Saule [/bib_ref]. Six of the strains that were positive for biofilm formation on CFA (i.e. ΔSTM2779, ΔygaT, ΔybaY, ΔyhfG, Δmtr and ΔpspABC-DEF) fluoresced brightly at 25°C. In contrast, three of the mutants (ΔSTM0341, ΔtrpE, and ΔssrA) that exhibited impaired levels of biofilm formation showed lower levels of calcofluor binding at 25°C, suggesting reduced levels of cellulose production [fig_ref] Table 3: Phenotypic characteristics of deletion mutants and WT strain tested for motility, RpoS... [/fig_ref]. The Δorf319 mutant also showed lower levels of binding to calcofluor but was not impaired in biofilm formation. Experiments to monitor cellulose production at 37°C showed that ΔssrA, ΔssrA/pssrAB and Δorf319 fluoresced more brightly than any other strain, suggesting a temperature-dependent production of cellulose in these SPI2 deletion strains.
Co-expression of genes specific to EPS, curli and cellulose production results in the rdar colony morphotype in S. Typhimurium [bib_ref] Crl activates transcription initiation of RpoS-regulated genes involved in the multicellular behavior..., Robbe-Saule [/bib_ref]. This morphotype is characterised by the binding of CR dye and the formation of red, dry and rugose spreading colonies. In strains not expressing curli and cellulose, a conventional smooth white colony is observed. We analysed the appearance of each strain on CFA-CR agar at 25°and 37°C compared to strains positive (S. Typhimurium LT2) and negative (ΔrpoS) for the rdar morphotype [bib_ref] Crl activates transcription initiation of RpoS-regulated genes involved in the multicellular behavior..., Robbe-Saule [/bib_ref] [bib_ref] Alteration of the rugose phenotype in waaG and ddhC mutants of Salmonella..., Anriany [/bib_ref]. Incubation at 25°C resulted in SL1344 and most of the mutants producing the rdar morphotype [fig_ref] Table 3: Phenotypic characteristics of deletion mutants and WT strain tested for motility, RpoS... [/fig_ref]. Three strains (ΔtrpE, ΔSTM2779, ΔssrA/pssrAB) showed altered morphology (i.e. smooth dark red colonies) at 25°C. However, the smooth phenotype disappeared when ΔtrpE and ΔSTM2779 were grown on CR at 37°C. These results indicate that chromosomal deletion of trpE, STM2779 (and rpoS) conferred temperature-dependent changes in EPS production, which were only apparent at 25°C. We noted that the ΔtrpE and ΔrpoS mutants showed a reduced ability to colonise polystyrene at 25°C, the same temperature at which a smooth or white colony phenotype was produced.
Previous work showed that the tnaA gene, encoding for tryptophanase, was required for E. coli biofilm formation on abiotic surfaces and human pneumocytes, and speculated that TnaA may modulate pH changes during attachment [bib_ref] Isolation of an Escherichia coli strain mutant unable to form biofilm on..., Martino [/bib_ref]. Since the tnaA gene is not present in S. Typhimurium, we investigated whether the ΔtrpE and ΔSTM0341 mutants, which showed reduced colonisation of abiotic surfaces and cellulose production, also showed altered levels of adherence or invasion of epithelial cells. We found that the ΔSTM0341 mutant was 9-times less adherent and 4-times less invasive than its parental strain (P < 0.01; data not shown), however Salmonella derivatives lacking the trpE gene did not show a significant difference in adherence and invasion compared to WT SL1344.
A role for aromatic amino acids in Salmonella biofilm growth Our data showed that inactivation of the tryptophan biosynthetic pathway altered the biofilm-forming capacity and EPS production by S. Typhimurium. To investigate the impact of amino acid availability on surface attachment, SL1344 was grown in CFA broth supplemented with different concentrations of aromatic and non-aromatic amino acids. Biofilm formation was determined after 12, 24 and 48 h of growth and compared to non-supplemented medium. Casamino acids, the main constituent of CFA medium, have been used in nutritional studies to determine bacterial growth requirements for peptides and amino acids [bib_ref] Metabolic pathways for cytotoxic end product formation from glutamate-and aspartate-containing peptides by..., Takahashi [/bib_ref]. Supplementation with non-aromatic amino acids did not alter the attachment of S. Typhimurium to polystyrene (data not shown), whilst a significant increase in biofilm formation occurred in the presence of aromatic amino acids [fig_ref] Figure 6: The effect of aromatic amino acids on static biofilm formation of wild-type... [/fig_ref]. This positive effect was noted even after 12 h of incubation, where the addition of all aromatic compounds, apart from tryptophan, significantly increased the number of adherent cells from 10 to 15-fold when compared with non-supplemented medium (P < 0.005). Similar positive effects on biofilm formation were observed after 24 h, with the greatest increase noted in wells containing indole (P < 0.001). Interestingly, supplementation with tryptophan had significant effects on surface associated growth at the later stages of biofilm development (i.e. 48 h). The addition of all aromatic amino acids significantly enhanced biofilm formation after 48 h of incubation, compared with non-supplemented CFA (P < 0.005).
Tryptophan and indole rescue the biofilm defect of the ΔtrpE mutant To determine whether the phenotypes associated with the ΔtrpE mutation were the direct result of cells being unable to synthesise or take up sufficient tryptophan or indole, the effect of these aromatic compounds on biofilm formation by ΔtrpE was investigated [fig_ref] Figure 7: Tryptophan biosynthesis is required for biofilm formation [/fig_ref]. Again, biofilm formation by the ΔtrpE strain was significantly reduced, when compared to SL1344 (P = 0.0001). The addition of tryptophan to the ΔtrpE mutant significantly increased biofilm formation, compared to growth in CFA alone (P = 0.01). In fact, the addition of 0.1 mM tryptophan completely restored the ability of the ΔtrpE strain to form a WT level of biofilm. Biofilm development by the ΔtrpE strain was significantly higher after supplementation with indole, compared to growth in CFA alone (P = 0.008). The highest concentration of indole (i.e. 0.1 mM) completely restored the ability of the ΔtrpE strain to form a biofilm on polystyrene. These findings are strikingly different to the situation for E. coli where exogenous indole reduces biofilm formation [bib_ref] Differential effects of epinephrine, norepinephrine, and indole on Escherichia coli O157:H7 chemotaxis,..., Bansal [/bib_ref].
[formula] + + + L L DR R DR R JH3185 ΔtrpE - + + L L D R S R R / S JH3187 ΔSTM0341 - + + L L D R R D R R JH3180 ΔssrA - + + L H D R R D R R JH3181 ΔssrA/pssrAB + + + L H D R S D R S JH3179 ΔORF319 + + + L H DR R DR R JH3182 ΔSTM2779 + + + H L DR S DR R JH3183 ΔygaT + + + H L D R R D R R JH3184 ΔybaY + + + H L D R R D R R JH3186 ΔyhfG + + + H L D R R D R R JH3188 Δmtr + + + H L D R R D R R JH3189 ΔpspABCDEF + + + H L D R R D R R JH3142 ΔrpoS - + - - - W R W S a [/formula]
# Discussion
It is clear that the ability of S. Typhimurium to grow as a biofilm on foods and processing surfaces represents an important survival strategy [bib_ref] Sources of Salmonella on broiler carcasses during transportation and processing: modes of..., Corry [/bib_ref] [bib_ref] Crosscontamination with Salmonella on a broiler slaughterhouse line demonstrated by use of..., Olsen [/bib_ref] [bib_ref] Use of episcopic differential interference contrast microscopy to identify bacterial biofilms on..., Warner [/bib_ref] [bib_ref] Effect of type of defeathering system on Salmonella cross-contamination during commercial processing, Clouser [/bib_ref] [bib_ref] Survival of foodborne pathogens on stainless steel surfaces and cross-contamination to foods, Kusumaningrum [/bib_ref]. We found that both nutrients and temperature had a profound effect on the attachment of S. Typhimurium SL1344. In this study biofilms were grown at 25°C, resulting in extensive colonisation of abiotic surfaces. Our finding that 95% of cells remain viable within established sessile communities of S. Typhimurium confirms that biofilms represent a potential reservoir for infection.
We used transcriptomic and proteomic approaches to analyze the changes that occur in S. Typhimurium in response to biofilm growth in the type of low nutrient, hydrodynamic environment that can be found during the processing of chickens [bib_ref] Sources of Salmonella on broiler carcasses during transportation and processing: modes of..., Corry [/bib_ref] [bib_ref] Crosscontamination with Salmonella on a broiler slaughterhouse line demonstrated by use of..., Olsen [/bib_ref] [bib_ref] Effect of type of defeathering system on Salmonella cross-contamination during commercial processing, Clouser [/bib_ref]. Transcriptomic analysis revealed the differential expression (i.e. more than a 2-fold change) of about 10% of S. Typhimurium genes during biofilm growth. The 229 genes that were significantly up-regulated during biofilm growth included several genes previously implicated in biofilm development of S. Typhimurium, including curli, cellulose and genes required for motility [bib_ref] The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as..., Zogaj [/bib_ref] [bib_ref] Biofilm formation and interaction with the surfaces of gallstones by Salmonella spp, Prouty [/bib_ref] , confirming the link between gene expression and gene function in our experiment. Although the value of whole genome studies for the comparison of heterogeneous biofilms has been a subject of debate, transcriptomic approaches have provided great insight into E. coli biofilm biology [bib_ref] Insights on Escherichia coli biofilm formation and inhibition from whole-transcriptome profiling, Wood [/bib_ref]. Prior to undertaking this study, the technical and biological reproducibility of a microarray-based approach for comparison of biofilm and planktonic S. Typhimurium cells was shown to be robust [bib_ref] Microarray: a useful tool for Salmonella biofilm research, Hamilton [/bib_ref].
Previous transcriptomic studies in other bacterial species showed that biofilm growth results in differential regulation of between 1% to 14.5% of the bacterial genome. While genetic variation between the organisms studied thus far may account for a proportion of these differences, we believe that the wide disparities between the proportions of biofilm-regulated genes reflect different experimental approaches. For example, cells were harvested at different times from different growth media and dissimilar model systems were used to cultivate the biofilm cells. Such technical differences greatly impact upon gene expression, and therefore the identification of biofilm-regulated genes, complicating the comparison of transcriptomic data between studies [bib_ref] The promise and peril of transcriptional profiling in biofilm communities, Parsek [/bib_ref]. A critique of the experimental approaches used to study biofilms has recently been presented [bib_ref] The developmental model of microbial biofilms: ten years of a paradigm up..., Monds [/bib_ref].
Proteomic analysis identified 24 proteins that were upregulated in the biofilm, and 45% of these corresponded to genes that were differentially expressed at the transcriptional level. Proteins required for late flagellar assembly were up-regulated in biofilms, when compared to planktonic cells. Proteomic profiling of P. putida identified the same pattern of up-regulation of flagellar genes within mature biofilms [bib_ref] Characterization of phenotypic changes in Pseudomonas putida in response to surfaceassociated growth, Sauer [/bib_ref]. In S. Enteritidis, thirty-two proteins were identified that showed differential expression in biofilms when compared to planktonic cultures [bib_ref] Architectural adaptation and protein expression patterns of Salmonella enterica serovar Enteritidis biofilms..., Mangalappalli-Illathu [/bib_ref] , including down-regulation of flagellar proteins and different patterns of expression of ArgT and Crr which are at variance with our study. The differing planktonic populations used for comparison may explain these conflicting reports [bib_ref] Lessons from DNA microarray analysis: the gene expression profile of biofilms, Lazazzera [/bib_ref]. It remains unclear whether flagella expression is limited to initial surface attachment or if it allows bacteria to move in and around the biofilm cell clusters and colonise new areas during the later stages of biofilm development [bib_ref] Development and dynamics of Pseudomonas sp. biofilms, Tolker-Nielsen [/bib_ref].
Unlike the proteomic approaches, transcriptomic studies have given several consistent messages; four studies comparing transcriptomic profiles of biofilm and planktonic-grown E. coli reported that biofilm growth leads to the up-regulation of genes involved in cell surface structures, amino acid metabolism, stress responses and anaerobic respiration [bib_ref] Global impact of mature biofilm lifestyle on Escherichia coli K-12 gene expression, Beloin [/bib_ref] [bib_ref] Global gene expression in Escherichia coli biofilms, Schembri [/bib_ref] [bib_ref] Temporal geneexpression in Escherichia coli K-12 biofilms, Domka [/bib_ref]. Our data suggest that mature biofilms of S. Typhimurium and E. coli involve similar physiological modifications, supporting the hypothesis that some adaptations required for biofilm growth and survival are conserved between bacterial species.
Our analysis showed that many of the RpoS (σ 38 )activated genes were up-regulated, suggesting that this regulon is important for the survival of Salmonella cells within the complex biofilm environment. Similar observations were made in E. coli biofilms [bib_ref] Global gene expression in Escherichia coli biofilms, Schembri [/bib_ref] [bib_ref] Impact of rpoS deletion on Escherichia coli biofilms, Adams [/bib_ref] , and RpoS is known to regulate the genes involved in curli and cellulose production [bib_ref] Crl activates transcription initiation of RpoS-regulated genes involved in the multicellular behavior..., Robbe-Saule [/bib_ref]. We used an RpoSdeficient mutant to confirm that RpoS plays a crucial role in mature biofilms of S. Typhimurium.
Previous transcriptomic analysis of E. coli cells showed that many biofilm-regulated genes were of unknown function [bib_ref] Finding gene-expression patterns in bacterial biofilms, Beloin [/bib_ref]. We found the same to be true in mature S. Typhimurium biofilms, with more than half of the differentially expressed genes having only putative or unknown function. Interestingly, seven of these genes, including ydcI, yebE and yceP, were also up-regulated in E. coli biofilms and the deletion of yceP has been shown to impair biofilm formation [bib_ref] Global impact of mature biofilm lifestyle on Escherichia coli K-12 gene expression, Beloin [/bib_ref] [bib_ref] YliH (BssR) and YceP (BssS) regulate Escherichia coli K-12 biofilm formation by..., Domka [/bib_ref].
To identify genes that were directly involved in biofilm growth, we mutated the five uncharacterised genes that were most highly up-and down-regulated. Our results suggest that STM0341, a putative inner membrane protein, is required for biofilm growth in S. epithelial cells than its parental strain, raising the possibility that expression of STM0341 could be directly or indirectly involved in efficient penetration of the gastrointestinal epithelial cell lining. [fig_ref] Table 3: Phenotypic characteristics of deletion mutants and WT strain tested for motility, RpoS... [/fig_ref] summarises the data from biofilm growth and further phenotypic analyses of the other uncharacterised mutant strains (i.e. STM2779, ygaT, ybaY, yhfG). Several of the mutations did not confer a detectable change in biofilm growth, or flagella, curli and cellulose production, although the WT genes were highly differentially regulated in the biofilm. Our observation that several individual genes were not required for biofilm formation by S. Typhimurium is consistent with biofilm formation involving multiple pathways [bib_ref] Initiation of biofilm formation in Pseudomonas fluorescens WCS365 proceeds via multiple, convergent..., O'toole [/bib_ref] or overlapping functions, as has been observed for the oxidative stress response of S. Typhimurium [bib_ref] Redundant hydrogen peroxide scavengers contribute to Salmonella virulence and oxidative stress resistance, Hébrard [/bib_ref] [bib_ref] Multiple redundant stress resistance mechanisms are induced in Salmonella enterica serovar Typhimurium..., Wright [/bib_ref].
The transcriptomic analysis revealed that genes involved in amino acid metabolism were up-regulated during biofilm growth, particularly tryptophan biosynthesis genes. The requirement for tryptophan was confirmed at the phenotypic level by showing that a trpE mutant formed significantly lower levels of biofilm. Domka and colleagues [bib_ref] Temporal geneexpression in Escherichia coli K-12 biofilms, Domka [/bib_ref] reported that tryptophan biosynthesis only occurred during early E. coli biofilm formation (<7 h) in LB medium and that repression of the trp operon was required during the later stages, suggesting that low intracellular levels of indole are required for biofilm development. This was confirmed by repression of the indole uptake gene (mtr) and up-regulation of indole efflux genes (acrEF) in E. coli [bib_ref] Temporal geneexpression in Escherichia coli K-12 biofilms, Domka [/bib_ref]. Such observations were consistent with earlier work showing that the absence of the regulatory YceP protein increased biofilm formation in E. coli by repressing genes that control indole transport into the cell and that a trpE mutation showed increased biofilm formation [bib_ref] YliH (BssR) and YceP (BssS) regulate Escherichia coli K-12 biofilm formation by..., Domka [/bib_ref]. We have reported several similarities between our data and biofilm-specific regulation in E. coli, but our findings suggest major differences in tryptophan metabolism between Salmonella and E. coli biofilms. We have shown that tryptophan biosynthesis plays a role at the late stages of biofilm development in S. Typhimurium, and that mtr and yceP are both up-regulated in the mature biofilm. The precise function of tryptophan and indole during biofilm formation of E. coli remains to be completely elucidated. It is clear that the tryptophanase (tna) operon that converts tryptophan to indole is required for biofilm formation in E. coli and other indole-producing bacteria [bib_ref] Isolation of an Escherichia coli strain mutant unable to form biofilm on..., Martino [/bib_ref] [bib_ref] Indole can act as an extracellular signal to regulate biofilm formation of..., Martino [/bib_ref].
Our data showed that exogenous tryptophan or indole restore the ability of the S. Typhimurium trpE mutant to form a biofilm. Moreover, both tryptophan and indole increased the biofilm-forming capacity of WT SL1344; in fact, low levels of indole increased attachment at all of the time points tested. We speculate that the highest concentration of indole (1 mM) used in this study is not biologically relevant, and imposes stress on the cell. While there are no other published reports on the effect of indole concentration upon biofilm formation in S. Typhimurium, lower levels of exogenous indole (312 to 625 μM) induced biofilm formation in E. coli, while higher levels of exogenous indole (1250 μM) caused indole toxicity and decreased bacteria growth [bib_ref] Indole can act as an extracellular signal to regulate biofilm formation of..., Martino [/bib_ref]. Taken together, these results suggest that in the absence of tryptophan, indole activates genes or pathways that contribute to biofilm formation.
Several reports have shown that indole acts as a signalling molecule in E. coli to: 1) prepare cells for a nutrient-poor environment and increase catabolism of amino acids, 2) up-regulate genes that encode drug exporters (e.g. acrDE, mdtAE, cusB) to increase bacterial tolerance to toxic compounds, 3) increase bacterial adherence to surfaces, and 4) delay cell division [bib_ref] Indole can act as an extracellular signal to regulate biofilm formation of..., Martino [/bib_ref] [bib_ref] Indole induces the expression of multidrug exporter genes in Escherichia coli, Hirakawa [/bib_ref] [bib_ref] Indole can act as an extracellular signal in Escherichia coli, Wang [/bib_ref]. All of these functions would be beneficial to cells within a well-established biofilm. Indole has recently been shown to act as an interspecies signal that controls biofilm formation by acting on oxygenases of bacteria that do not synthesise this molecule at temperatures below 30°C [bib_ref] Indole is an inter-species biofilm signal mediated by SdiA, Lee [/bib_ref] [bib_ref] Indole cell signalling occurs primarily at low temperatures in Escherichia coli, Lee [/bib_ref]. It is possible that we have observed a similar cell signalling phenomena in Salmonella. Several Gram-negative bacterial biofilms have been shown to secrete the amino acid valine that may play a role in signalling or function as a protective osmolyte [bib_ref] The amino acid valine is secreted in continuous-flow bacterial biofilms, Valle [/bib_ref]. Further functional analysis of biofilms grown in the presence of exogenous tryptophan (and indole) should provide insights into the role of these molecules in S. Typhimurium.
Microbial biofilms are inherently more resistant to host defences and antimicrobials than planktonic cells, but it is not known what proportion of this phenotype is due to bacterial factors such as the expression of virulence proteins, or external environmental influences such as the diffusion of nutrients or oxygen, or slow growth rate.
In this study, we report that mature biofilm growth in S. Typhimurium leads to significant down-regulation (up to 100-fold) of certain virulence genes located in SPI2. We confirmed the link between SPI2 expression and biofilm growth by showing that the SPI2-deficient ΔssrA strain formed significantly less biofilm than the parental strain, and that complementation with ssrAB effectively restored attachment. We discovered that overexpression of SPI2 led to a significant reduction in bacterial attachment by the WT strain. These data show that both increased and decreased expression of SPI2 interferes with biofilm formation. It is not clear how a TTSS impacts upon biofilm formation, but we note that it has been reported that an extracellular molecule (LPS) can interfere with the TTSS-mediated attachment of Shigella flexneri to mammalian cells [bib_ref] Optimization of virulence functions through glucosylation of Shigella LPS, West [/bib_ref]. By analogy, we speculate that aberrant expression of the SPI2 TTSS apparatus compromises the ability of Salmonella to form biofilm, perhaps by affecting the presentation of cell surface factors such as curli.
# Conclusions
We have shown that biofilm growth of S. Typhimurium involves processes that include amino acid metabolism, motility, and virulence. Three proteins, TrpE, STM0341 and SsrA, play a role in biofilm formation by S. Typhimurium. We have discovered that tryptophan metabolism is required for effective biofilm formation in our experimental system. The unexpected link between SPI2 expression and biofilm promises to be a fertile area for Salmonella research in the future. Further characterization of the mutations that led to a reduction in biofilm growth is ongoing and we are examining temporal expression within the biofilm using a GFP reporter system. We are now comparing S. Typhimurium biofilms grown in dissimilar model systems to identify the core genes that are required for survival of this pathogen on different surfaces.
# Methods
Bacterial strains, plasmids and growth conditions Bacterial strains and plasmids used in this study are described in [fig_ref] Table 4: Bacterial strains and plasmids [/fig_ref]. Liquid growth media were Colonising Factor Antigen (CFA) medium comprised of (w/v) casamino acids (Difco) 1.0%, yeast extract (Difco) 0.15%, MgSO 4 0.005% and MnCl 2 0.0005% [bib_ref] Regulatory circuitry of the CsrA/CsrB and BarA/UvrY systems of Escherichia coli, Suzuki [/bib_ref] , Brain-Heart Infusion (BHI; Oxoid), Luria Broth (LB) or M9 minimal medium supplemented with glucose, sucrose or glycerol at concentrations of 0.2, 2.0 and 20% (v/v), according to Sambrook. Bacto agar (Difco; 1.5% w/v) was used to make plates from these media, as required. All cultures were incubated at 25°C unless otherwise stated. Transductions were carried out using bacteriophage P22 followed by selection of non-lysogens on Green agar [bib_ref] Transduction by bacteriophage P22 in nonsmooth mutants of Salmonella typhimurium, Gemski [/bib_ref]. For complementation assays SL1344 was transformed with the low copy, F1-replicon-based, plasmid pssrAB, kindly provided by Michael Hensel [bib_ref] Simultaneous identification of bacterial virulence genes by negative selection, Hensel [/bib_ref]. Plasmids p1437-1 and p1437-6, the P15A-replicon-based, arabinose-inducible ssrAB+ and ssrAB-(reverse orientation) derivatives were used to over-express SsrA. Antibiotics were added at the following concentrations: kanamycin (Km), 50 μg ml -1 ; carbenicillin (Cb), 100 μg ml -1 ; chloramphenicol (Cm), 25 μg ml -1 . Media supplements were sterilised by filtration through 0.22 μm filters (Sartorius) and added to culture media. All non-aromatic amino acids (i.e alanine, cysteine, glutamate, glutamine, histidine, and serine) were added at 0.1, 1 and 10 mM. All aromatic amino acids were tested at the concentrations shown in [fig_ref] Figure 6: The effect of aromatic amino acids on static biofilm formation of wild-type... [/fig_ref].
Biofilm formation on glass and polystyrene The batch biofilm system used for in situ analysis of bacterial attachment to glass was designed at the Environmental Microbiology Research Group at Exeter. This system provides a way of monitoring biofilm formation on a hydrophilic glass surface under flowing conditions. Overnight cultures of S. Typhimurium SL1344 were grown in CFA and standardized to achieve an initial concentration of 10 6 CFU ml -1 and injected into a stirring influent flask containing 5 L of prewarmed (25°C) sterile CFA medium. The inoculated influent was then pumped through silicon tubing (5 mm internal diameter, Samco Silicon Products Ltd.) at 60 ml h -1 using a peristaltic pump (Minipulse 3, Gilson). In this manner, the bacteria were allowed to flow through the closed system and either attach to a borosilicate glass flow cell (3 × 3 mm, Camlab), held in a heated microscope stage, or flow out into a waste reservoir. A laminar flow rate was chosen to produce minimal fluctuations in velocity, requiring the bacteria to undergo active transport in order to interact with the surface [bib_ref] Biofilm formation in laminar flow using Pseudomonas fluorescens EX101, Brading [/bib_ref].
The entire system was kept at a constant temperature of 25°C and biofilm formation within the glass flow cell was imaged in situ without interrupting the flow for up to 72 h. Any possible backflow of media or bacteria within the system was eliminated using media breaks. Attached cells were then examined using light microscopy (see below).
Biofilm formation on polystyrene is based on the ability of bacteria to attach to the wells of microtitre plates, using a modification of a previously described technique [bib_ref] Genetic analysis of Escherichia coli biofilm formation: roles of flagella, motility, chemotaxis..., Pratt [/bib_ref]. This system provides a way to monitor biofilm formation on a hydrophobic polystyrene surface under static conditions. Overnight cultures of SL1344 were standardized (see above) and inoculated in quadruplicate into polystyrene 96-well plates (Nunc) containing sterile media. The plates were incubated statically for up to 48 h and then rinsed with dH 2 0 to remove any nonadherent bacteria, dried for 1 h at 60°C and stained with crystal violet (CV, BDH). The wells were destained with 20% acetone in ethanol (99.8%) solution and the amount of CV, indicative of the binding ability of each strain, was determined at A 590 using a Spectramax Plus (Molecular Devices) spectrophotometer. All experiments were repeated on three separate occasions. The results are reported as a mean value of absorbance and analyzed for significance (P < 0.05 or < 0.01) by the Student's twotailed t-test.
## Microscopy, image analysis and cell staining
Microscopy and image analysis of biofilms grown in glass flow cells (see above) were performed using a DMLB video light microscope (Leica). Images of bacterial attachment on the glass surface were visualized using a COHU 4612-5000 CCD camera (COHU) connected to a Macintosh G3 computer and Scion VG-5 PCI framestore board (Scion Corporation). Area measurements were calculated using Scion Image (Scion Corporation). Biofilm accumulation was measured as percentage surface cover on the top and bottom surfaces of the glass flow cell. The coordinates of 6 specific areas on the glass flow cell were marked in order to return to the same site at each time point. Images were captured for up to 72 h of flow with each time point representing an average of 30 frames. A threshold was applied to each image and the number of black pixels in each frame were measured and the percentage of biofilm surface cover was calculated as the proportion of white pixels to the total frame and analyzed for significance (P < 0.05 or <0.01) by the Student's two-tailed t-test. To determine cell viability of the biofilm based on membrane integrity, glass flow cells were rinsed with PBS to remove any unattached cells and stained with Live/Dead BacLight (Molecular Probes) for 1 h at room temperature. After 1 h, the staining solution was drained, the flow cell filled with fresh PBS and cells were viewed using a 40× objective of a CH-2 light microscope and illuminated by a BH2-RFCA fluorescent light source using a BP405 (nm) filter block (Olympus).
Isolation of biofilm and planktonic cells for total RNA and protein extractions Biofilm and planktonic cells used for total RNA and protein extractions were grown in a modified batch system (Additional file 1). This system is identical to the glass flow cell system described above apart from replacing the glass flow cell with silicone rubber tubing that permits the harvesting of large amounts of biofilm biomass [bib_ref] Architectural adaptation and protein expression patterns of Salmonella enterica serovar Enteritidis biofilms..., Mangalappalli-Illathu [/bib_ref] [bib_ref] Characterization of phenotypic changes in Pseudomonas putida in response to surfaceassociated growth, Sauer [/bib_ref] [bib_ref] Pseudomonas aeruginosa displays multiple phenotypes during development as a biofilm, Sauer [/bib_ref]. Our rationale for choosing a 'batch' biofilm system as opposed to a 'seeded' system was to mimic a food processing environment where bacteria attach and are then subjected to a hydrodynamic environment that lacks fresh nutrients. Overnight cultures of SL1344 grown in CFA broth were standardized to achieve an initial concentration of 10 6 CFU ml -1 and injected into a stirring influent flask containing 5 L of pre-warmed (25°C) sterile CFA medium. The inoculated influent was then pumped through silicon tubing (5 mm internal diameter, Samco Silicon Products Ltd.) at 60 ml h -1 using a peristaltic pump (Minipulse 3, Gilson). The bacteria were allowed to flow through the closed system and either attach to a vertical piece of silicon tubing (1 meter length, 16 mm internal diameter, Samco Silicon Products Ltd.) or flow out into a waste reservoir. The tubing was positioned vertically to collect biofilm cells adherent to the tubing and minimise collecting bacteria that had simply sedimented. Possible backflow of media or bacteria into the influent from the silicon tubing was eliminated using media breaks. Samples taken to determine the pH of the medium or bacterial cell counts were removed via a media port located near the effluent and influent vessel, respectively. Planktonic cells were removed after 6, 24 and 72 h of growth from the influent vessel to avoid any contamination with biofilm cells. The entire system was kept at a constant temperature of 25°C and biofilm cells were isolated after 72 h of growth. All planktonic and biofilm samples were isolated from the same experimental system and used for both RNA and protein isolation.
## Protein extraction and 2-d analysis
The silicon tubing used for isolating biofilm cells was rapidly rinsed three times in cold (4°C) PBS to remove any non-adherent cells. A sterile loop was used to detach the adherent cells from the silicon tubing into 20 ml of cold (4°C) PBS. Total protein was extracted using a previously described technique [bib_ref] Proteome of Salmonella Typhimurium SL1344: identification of novel abundant cell, Qi [/bib_ref]. Briefly, cells were [bib_ref] Campylobacter jejuni gene expression in response to iron limitation and the role..., Holmes [/bib_ref].
# Western blot analysis
Equal amounts of protein from S. Typhimurium biofilm and planktonic cell extracts were separated by SDS-PAGE (5-12% gradient) using a Mini Protean 3 electrophoresis system (Biorad) according to the methods of Sambrook et al.. For immunoblotting, samples were transferred to nitrocellulose Biodyne A membranes (Pall Corporation) using a Mini Trans-Blot Electrophoretic cell (Biorad) according to the manufactures instructions. The proteins were fixed with methanol and the membranes were blocked with 5% skim milk and 1% BSA. Immobilized protein was detected using monoclonal FliC antibody (primary antibody). Anti-mouse (titre 1:2000) or anti-rabbit (1:5,000) IgG conjugated to alkaline phosphatase (AP) were used as secondary antibodies and purchased from Sigma. 5-bromo-4chloro-3-indolyl-phosphate and nitro blue tetrazolium (BCIP/NBT) colour development substrate was used for antibody detection (Promega).
## Rna and dna extraction and quantification
The silicon tubing used for isolating biofilm cells was rapidly rinsed three times in cold (4°C) PBS to remove any non-adherent cells and immediately transferred to
# Microarray data analysis
Slides were scanned using a GenePix 4000A scanner (Axon Instruments, Inc.). Fluorescent spots and local background intensities were quantified using Genepix Pro 3.0 Software (Axon Instruments, Inc.). The data were filtered so that spots with a reference signal lower than the background plus two standard deviations of the background or obvious blemishes were not included in the analysis. Signal intensities were corrected by subtracting the background and the red/green (Cy5/Cy3) ratios were calculated. To compensate for unequal dye incorporation, data centring was performed by bringing the natural logarithm of the median of spots printed by the same pin to zero. Data from each microarray that passed the quality control procedures were then analyzed using Gene Spring 7.3 (Agilent). Differentially expressed genes were identified by performing a
## Genetic manipulations
Deletion of selected genes on the SL1344 chromosome was achieved by one-step inactivation using PCR products, as previously described [bib_ref] One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products, Datsenko [/bib_ref]. Red recombinase template plasmid pKD4 was amplified using primers (Sigma-Genosys) designed with homology extensions to regions 100 bp either side of the predicted ORF of selected genes (based on the LT2 genome sequence [bib_ref] Complete genome sequence of Salmonella enterica serovar Typhimurium LT2, Mcclelland [/bib_ref]. PCR was carried out using BIO-X-ACT DNA polymerase (Bioline) with the manufacturer's recommended 1 kb amplification parameters. Linear DNA products were introduced into SL1344/pKD46 by electroporation (Gene Pulser, Biorad). Gene disruption was verified by PCR and restriction enzyme analysis (the full list of primers used for construction and verification can be found in Additional file 5B). Following verification, the mutants were P22-transduced into a clean SL1344 background and non-lysogenic bacteria were selected on Green Agar plates.
## Phenotypic analysis of mutants
Growth curves of parent strains and mutants were performed using a Microbiology Reader Bioscreen C (ThermoLabsystems). Strains were tested for growth in CFA and LB broth over a period of 24 h. Each overnight culture was standardised to achieve an initial concentration of 10 6 CFU ml -1 and inoculated into four separate wells of a 100 well sterile honeycomb plate (Thermo-Labsystems) containing fresh medium and antibiotics as appropriate. The plates were incubated at 25°C with covers and measurements based on optical density (OD 600 ) were recorded every 15 min. EPS production was monitored by growth at 25°C on CFA without salt, supplemented with 0.01% Congo Red (CR) as based on a previously described method [bib_ref] Expression of two csg operons is required for production of fibronectin-and congo..., Hammar [/bib_ref]. Swimming motility was assessed by stabbing colonies into plates of CFA and LB containing 0.3% wt/vol agar and incubating at 25°C and 37°C for up to 48 h [bib_ref] Inorganic polyphosphate is needed for swimming, swarming, and twitching motilities of Pseudomonas..., Rashid [/bib_ref]. Calcofluor binding was assessed by growth on CFA agar supplemented with fluorescent brightener 28 (Sigma) as described by Zogaj et al. [bib_ref] The multicellular morphotypes of Salmonella typhimurium and Escherichia coli produce cellulose as..., Zogaj [/bib_ref]. RpoS activity was examined indirectly by assessing hydrogen peroxidase II (HPII) levels as described by Zambrano et al. [bib_ref] Microbial competition: Escherichia coli mutants that take over stationary phase cultures, Zambrano [/bib_ref].
Bacterial adherence and invasion of HeLa cells HeLa epithelial cells (European Collection of Cell Cultures (ECACC, 93021013) were grown in HEPESbuffered DMEM (Sigma) supplemented with 10% Fetal Bovine Serum at 37°C in presence of 10% CO 2 . The infection procedure was derived from the method of Steele-Mortimer et al., [bib_ref] Biogenesis of Salmonella typhimurium-containing vacuoles in epithelial cells involves interactions with the..., Steele-Mortimer [/bib_ref] as modified by Hautefort et. al., [bib_ref] During infection of epithelial cells Salmonella enterica serovar Typhimurium undergoes a time-dependent..., Hautefort [/bib_ref]. To assess the invasion ability of the Salmonella strains, bacteria that remained outside of the HeLa cells were subsequently killed by addition of HEPES-buffered DMEM containing 10% FBS and 30 ug ml -1 Gentamicin for 30 min at 37°C under 10% CO 2 atmosphere. Infected HeLa cells were washed twice in PBS (pH 7.4) and lysed for 10 min at room temperature in PBS containing 0.1% SDS. Intracellular bacterial viable counts were estimated by plating 10-fold dilutions on LB agar plates. To assess the adherence of the Salmonella WT and mutant strains, infection of HeLa cell monolayers was performed as described above. After 30 min invasion, HeLa cells were washed in PBS and immediately lysed in PBS containing 0.1% SDS. The total bacterial population was determined by plating dilutions of the lysate on LB agar plates. The adhering population was estimated by subtracting the intracellular viable counts obtained from the total counts. All assays were done with six biological replicates.
## List of abbreviations
[fig] Figure 1: Whole genome expression profiling of S. Typhimurium SL1344 flowing biofilms compared to planktonic cells when grown in CFA at 25°C for 72 h. Expression changes of genes belonging to functional groups and pathogenicity islands (numbers in parenthesis refer to the genes assigned to each functional group from the genome of S. Typhimurium LT2). The bars show the percentage of genes belonging to each group that were altered for expression > 2-fold between planktonic and flowing biofilm cells. The blue bars indicate the proportion of genes that are down-regulated and the red bars represent the proportion of up-regulated genes for each group during biofilm growth (P < 0.01). [/fig]
[fig] Figure 2: Biofilm-regulated S. Typhimurium gene expression. (A) A subset of genes up-regulated in the biofilm that encode cell surface structures, motility, global regulation, and oxygen diffusion. Representative examples of the functional categories (Figure 1) are shown. (B) Type III secretion genes. Transcriptomic data from the flowing biofilm system (72 h) was normalised to the planktonic samples and values are shown as fold change on a logarithmic [/fig]
[fig] Figure 3: Biofilm-regulated protein expression. SYPRO® Ruby stained 2-D gels of total protein extracts from flowing biofilm (A) and planktonic (B) cells of S. Typhimurium grown in CFA medium at 25°C for 72 h. Spots circled in red were excised from the gels and identified by mass spectrometry and peptide mass fingerprinting. (C) Magnification of 2-D gels comparing expression of FlgK (I), DppA (II), and YggE (III), which were all more highly expressed in the biofilm cells than in planktonic cells. (D) Western immunoblot of total protein extracts of mature S. Typhimurium biofilm (B) and planktonic (P) cells both grown in CFA medium for 72 h. Protein extracts were separated on a denaturing SDS-PAGE gel and probed with anti-FliC monoclonal antibody. Densitometric analysis showed that FliC was 3.4 fold induced in the biofilm compared with planktonic cells. [/fig]
[fig] Figure 4: A) Static biofilm formation of eight targeted gene deletion mutants, compared to attachment of WT SL1344. Following incubation at 25°C for 24 h in CFA medium, the level of biofilm formation is expressed as a percentage of WT SL1344 which had an A 590 nm of 0.56 ± 0.04 in this experiment. The ΔtrpE (JH3185) (* P = 0.0001) and ΔSTM0341 (JH3187) (* P = 0.001) mutants showed significantly less attachment to polystyrene than WT SL1344. The mean absorbance values from four wells are shown as a percentage of WT SL1344 and the error bars represent the SD between four technical replicates (n = 3). (B) The attachment of the ΔtrpE (black diamonds) mutant (JH3185) to the bottom surface of a glass flow cell compared with S. Typhimurium SL1344 (back circles). Bacteria were cultured at 25°C in CFA medium in a glass flow cell. Error bars represent the standard deviation between 6 images captured along the length of the flow cell over 24 h (n = 3). [/fig]
[fig] Figure 5: Static biofilm formations. (A) Static biofilm formation of SPI2 and rpoS deletion mutants. Following incubation at 25°C for 48 h in CFA medium, the level of biofilm formation is expressed as a percentage of WT SL1344 which had an A 590 nm of 0.15 ± 0.02 in this experiment. The ΔssrA (JH3180) and ΔrpoS (JH3142) mutants showed significantly less attachment to polystrene than WT SL1344 after 24 h (data not shown) and 48 h of growth (* P = 0.001-0.02). Complementation of ΔssrA with a low copy plasmid encoding ssrAB (JH3181) restored the ability of this mutant to form WT biofilm after 48 h (* P = 0.002). The mean absorbance values from four wells are shown as a percentage of WT SL1344 and the error bars represent the SD between 4 technical replicates (n = 2). (B) Static biofilm formation of SL1344 over-expressing SsrAB (p1437-1) in CFA medium (+ 0.1% L-arabinose where indicated) for 24 h at 25°C. The level of biofilm formation is expressed as a percentage of WT SL1344, which had an A 590 nm of 0.14 ± 0.01 in this experiment. The induction of SsrAB expression by an arabinoseinducible promoter significantly inhibited attachment (* P = 0.000007) when compared to WT SL1344. No significant difference in attachment was observed in the control strain over-expressing SsrAB in the reverse orientation (p1437-6). The mean absorbance values from six wells are shown and the error bars represent the SD between 4 technical replicates. [/fig]
[fig] Figure 6: The effect of aromatic amino acids on static biofilm formation of wild-type S. Typhimurium. CFA medium was supplemented with increasing concentrations of aromatic amino acids. The level of biofilm formation after 12, 24 and 48 h of growth at 25°C is expressed as a percentage of the biofilm formed after unsupplemented growth in CFA at 48 h, which had an A 590 nm of 0.15 ± 0.02 in this experiment. The mean absorbance values from four wells are shown and the error bars represent the SD between 4 technical replicates. [/fig]
[fig] Figure 7: Tryptophan biosynthesis is required for biofilm formation. The effect of the trpE mutation and addition of tryptophan (0.01 mM, 0.1 mM) and indole (0.01 mM, 0.1 mM) on static biofilm growth of ΔtrpE (JH3185) was determined after 24 h of growth in CFA broth at 25°C. The addition of tryptophan and indole significantly (*P = 0.02-0.00003) increased biofilm formation when compared to ΔtrpE grown in CFA alone. The mean absorbance values from four wells are shown and the error bars represent the SD between 4 technical replicates. [/fig]
[table] Table 1: Expression levels of S. Typhimurium tryptophan biosynthetic genes during biofilm growth under flowing conditions [/table]
[table] Table 2: Biofilm-regulated proteins that show similar trends in proteomic and transcriptomic analysis [/table]
[table] Table 3: Phenotypic characteristics of deletion mutants and WT strain tested for motility, RpoS activity, calcofluor binding and changes in extracellular matrix [/table]
[table] Table 4: Bacterial strains and plasmids [/table]
|
10.3389/fmed.2020.573522
|
CCBY
|
7575786
|
33117834
|
s2orc_pubmed_articles
|
Deep Learning Improves Osteonecrosis Prediction of Femoral Head After Internal Fixation Using Hybrid Patient and Radiograph Variables
Femoral neck fractures (FNFs) are a great public health problem that leads to a high incidence of death and dysfunction. Osteonecrosis of the femoral head (ONFH) after internal fixation of FNF is a frequently reported complication and a major cause for reoperation. Early intervention can prevent osteonecrosis aggravation at the preliminary stage. However, at present, failure to diagnose asymptomatic ONFH after FNF fixation hinders effective intervention at early stages. The primary objective of this study was to develop a predictive model for postoperative ONFH using deep learning (DL) methods developed using plain X-ray radiographs and hybrid patient variables. A two-center retrospective study of patients who underwent closed reduction and cannulated screw fixation was performed. We trained a convolutional neural network (CNN) model using postoperative pelvic radiographs and the output regressive radiograph variables. A less experienced orthopedic doctor, and an experienced orthopedic doctor also evaluated and diagnosed the patients using postoperative pelvic radiographs. Hybrid nomograms were developed based on patient and radiograph variables to determine predictive performance. A total of 238 patients, including 95 ONFH patients and 143 non-ONFH patients, were included. A CNN model was trained using postoperative radiographs and output radiograph variables. The accuracy of the validation set was 0.873 for the CNN model, and the algorithm achieved an area under the curve (AUC) value of 0.912 for the prediction. The diagnostic and predictive ability of the algorithm was superior to that of the two doctors, based on the postoperative X-rays. The addition of DL-based radiograph variables to the clinical nomogram improved predictive performance, resulting in an AUC of 0.948 (95% CI, 0.920-0.976) and better calibration. The decision curve analysis showed that adding the DL increased the clinical usefulness of the nomogram compared with a clinical approach alone. In conclusion, we constructed a DL facilitated nomogram that incorporated a hybrid of radiograph and patient variables, which can be used to improve the prediction of preoperative osteonecrosis of the femoral head after internal fixation.
Femoral neck fractures (FNFs) are a great public health problem that leads to a high incidence of death and dysfunction. Osteonecrosis of the femoral head (ONFH) after internal fixation of FNF is a frequently reported complication and a major cause for reoperation. Early intervention can prevent osteonecrosis aggravation at the preliminary stage. However, at present, failure to diagnose asymptomatic ONFH after FNF fixation hinders effective intervention at early stages. The primary objective of this study was to develop a predictive model for postoperative ONFH using deep learning (DL) methods developed using plain X-ray radiographs and hybrid patient variables. A two-center retrospective study of patients who underwent closed reduction and cannulated screw fixation was performed. We trained a convolutional neural network (CNN) model using postoperative pelvic radiographs and the output regressive radiograph variables. A less experienced orthopedic doctor, and an experienced orthopedic doctor also evaluated and diagnosed the patients using postoperative pelvic radiographs. Hybrid nomograms were developed based on patient and radiograph variables to determine predictive performance. A total of 238 patients, including 95 ONFH patients and 143 non-ONFH patients, were included. A CNN model was trained using postoperative radiographs and output radiograph variables. The accuracy of the validation set was 0.873 for the CNN model, and the algorithm achieved an area under the curve (AUC) value of 0.912 for the prediction. The diagnostic and predictive ability of the algorithm was superior to that of the two doctors, based on the postoperative X-rays. The addition of DL-based radiograph variables to the clinical nomogram improved predictive performance, resulting in an AUC of 0.948 (95% CI, 0.920-0.976) and better calibration. The decision curve analysis showed that adding the DL increased the clinical usefulness of the nomogram compared with a clinical approach alone. In conclusion, we constructed a DL facilitated nomogram that incorporated a hybrid of radiograph and patient variables, which can be used to improve the prediction of preoperative osteonecrosis of the femoral head after internal fixation.
# Introduction
Hip fracture is a significant public health concern that affects 4.5 million people worldwide each year and this number is expected to increase to 21 million in the next 40 years (1, 2). Femoral neck fracture (FNF) is one of the most common types of hip fracture, accounting for 49-80% of all hip fractures. Despite the availability of multiple effective internal fixation procedures, ∼10-48.8% femoral neck fractures require reoperation. Osteonecrosis of the femoral head (ONFH) is a major cause of reoperation for FNF. Joint disfunction, pain, disability, and mental anguish caused by ONFH result in great suffering for patients. End-stage ONFH often inevitably requires artificial joint replacement surgery, an invasive and economically costly technique. Early diagnosis can facilitate the application of interventions that can avoid or delay arthroplasty to a certain extent. However, misdiagnoses and delayed diagnoses are common due to the lack of preliminary symptoms, typical features, and internal fixation interference on radiographs. Different diagnostic criteria or simple visual estimates are used by radiologists for practical imaging diagnosis, resulting in unsatisfactory levels of diagnostic consistency and accuracy. Therefore, early accurate and consistent prediction of ONFH in patients after FNF internal fixation may hold the key for improving patient outcomes.
Deep learning (DL) using radiographs has a proven ability of classifying bone structures and features in specific sites with expert-level accuracy. Convolutional Neural Networks (CNNs) are the most suitable models for image recognition of DL, and have been widely used for the orthopedic diagnosis of wrists and ankles. Gale et al. developed a hip fracture detector using DL and achieved an AUC of 0.994. Cheng et al. reported on a deep convolutional neural network (DCNN) for the detection and localization of hip fractures using pelvic radiographs, which achieved an AUC of 0.98 for the identification of hip fractures. Recently, Chee et al. made a breakthrough discovery for the diagnosis of early ONFH using radiography through deep learning. This model achieved an AUC of 0.93 and sensitivity and specificity that were not inferior to the diagnosis made by both the less experienced and experienced radiologists. Their study indicated the potential of DL for the diagnosis and prediction of ONFH, especially for X-ray imaging. However, the implementation of DL for the diagnosis of postoperative ONFH using digital radiography remains unexplored. Postoperative X-rays are highly affected by interference, such as that of internal fixation devices, which cause difference between the images on radiographs and the original appearance of the femoral neck and femoral head. Since postoperative X-rays are the most common method used for early examination, a consistent diagnosis based on postoperative Xrays made using DL may improve the prediction of postoperative ONFH for better prognosis. In this study, we designed and assessed the diagnostic performance of a DL algorithm based on the CNN network model using postoperative X-rays. We also compared the accuracy of the diagnosis of postoperative ONFH between this DL model and assessments made by two orthopedic doctors of different levels of experience.
In previous studies, a large number of research studies have indicated that patient and interventional variables, including demography, fracture classification, laboratory examination, reduction quality, and initial postoperative rehabilitation, are significantly associated with postoperative ONFH. However, intraoperative, and postoperative factors, especially radiographic variables, including intraoperative reduction and fracture healing, have yet to be incorporated into routine clinical postoperative ONFH prediction. In this study, a DL facilitated predictive model using a hybrid of patient and artificial intelligence (AI) radiographic variables, was also developed. Comparisons were made with a single clinical prediction model was performed to estimate whether DL could improve the prediction of postoperative ONFH.
# Materials and methods
## Study population
Data were obtained from two urban tertiary hospitals, The First Affiliated Hospital of University of Science and Technology of China (FAH) and the Southern Branch of the First Affiliated Hospital of University of Science and Technology of China (SBH). One hundred thirty-nine FAH patients and 99 SBH patients who had received closed reduction and cannulated screw fixation from June 2013 to January 2015 were enrolled in this study. The patient inclusion criteria were as follows: (i) Patients over 18 years of age with fresh FNFs; (ii) Postoperative pelvic radiographs obtained 6 months after surgery; (iii) Continuous follow-up for a minimum of 5 years with the clinical characteristics available. The exclusion criteria were as follows: (i) Pathological fractures and bilateral fractures; (ii) Long-term hormone use. The treatment standard and strategy used for femoral neck fracture was the cannulated compression screws fixation technique, based on American Academy of Orthopedic Surgeons guidelines. Postoperative ONFH was diagnosed using pelvic MRIs or co-diagnosis by three experienced orthopedic surgeons based on the pelvic radiograph obtained at the last follow-up. This study was approved by the Ethics Committees of both hospitals. Exemption of the informed consent, the information disclosure, and a negative opportunity are guaranteed in the Ethical approval (20-P-049).
Demographics, comorbidities, smoking status, alcohol use, blood tests, preoperative Garden classification, Pauwels angle, preoperative interval from injury, operation associated data, postoperative Garden index, preoperative interval to weight bearing and other baseline patient and clinical data were derived from medical and follow-up records. The data were de-identified after patient variables were collected.
## Imaging studies
Image acquisition and retrieval procedures were conducted using Picture Archiving and Communication Systems (PACS) on FAH and SBH patients. Digital radiographs of the hip were obtained using Digital Diagnostics (Philips Healthcare) on FAH patients and Discovery XR656 (GE Healthcare) on SBH patients. The size of the stored images varied from 2,128 × 2,248 pixels to 2,688 × 2,688 pixels, with 8-bit grayscale color. Each radiograph was labeled based on the final diagnosis of postoperative ONFH. Geometric, smooth, concave, bandlike low-signal intensity lesions at the femoral head on the T1weighted images were regarded as pathognomonic MRI findings of ONFH. For MRI data not obtained at the last follow-up (45/238, 18.9%), diagnosis was based on pelvic plain radiographs obtained at the last follow-up and was set as a reference for labeling. The Association Research Circulation Osseous (ARCO) classification system was used as the diagnostic standard for ONFH.
Radiographic image files were loaded for processing using a MATLAB library (version 2017b, MathWorks, USA). The 7 × 7 cm images centered on the bilateral femoral heads were cropped. The center coordinates were manually recorded in advance. Radiographs were standardized to a common size and pixel intensity distribution. The images were down-sampled and padded to a final size of 120 × 120 pixels. Mean pixel intensity and standard deviation of each image was normalized.
## Algorithm development and extraction of image variables
For the development of a deep learning algorithm, we used MATLAB (version 2017b, MathWorks, USA) to implement a CNN model to compute abstract image features from input image pixel arrays. The design of the CNN model is shown in. The CNN model consisted of three convolutional blocks, a dropout and full connection layers. Each convolutional block comprised of convolutional operation, batch normalization, relu, and average pooling. The input used was Pixel values were set at 120 * 120 using a digital image. Cubic convolution and pooling were performed on each layer to adjust the weights of the neural network, using the difference between the output and true labels.
The patients in the dataset were assigned to different groups as follows: 149 (63%) for training, 17 (7%) for validation and 72 (30%) for testing. The output results underwent regression analysis. The network output was a probability distribution for the continuous variables of the regression coefficient from 0 to 1.25, which was divided at 0.25 intervals into classified labels, Higher label values were more likely to be considered to more strongly predict postoperative ONFH. In this study, this output label was referred to as the AI index classification.
## Algorithm evaluation
Seventy-two independent datasets were used to test the trained predictive model to evaluate its accuracy for postoperative ONFH prediction. The probability of the diagnosis being postoperative ONFH generated by the model was evaluated using the receiver operating characteristic (ROC) curve and the area under the curve (AUC). The sensitivity, accuracy, recall and specificity of the radiographs for the prediction of ONFH were measured using a cutoff level probability of 0.5. A training curve was used to determine root mean squared error (RMSE) and loss, while a precision-recall curve was used to determine precision and recall.
## Image predictive variable evaluation
We compared the AI index with the predictive measurement scores assigned by the two orthopedic surgeons of different levels of experience with the results of the DL algorithm based on the same X-rays to evaluate the performance of the algorithm. Radiographs obtained 6-months after anteroposterior hip operations were randomly divided into two IPAC sequences by the study coordinator. A less experienced orthopedic doctor (Doctor A, 3rd year of residency in orthopedics) and an experienced orthopedic doctor (Doctor B, 18 years in orthopedics) participated in the reading session. Both doctors were not involved in surgery, data collection or reference labeling. A score based on the subjective prediction of the doctors using the postoperative X-ray to determine the most likely outcome at final follow-up was assigned using a 1-5 grading system. One indicated that the development of ONFH was considered to be impossible, while 5 indicated that the development of ONFH was considered to be certain. Each doctor independently graded the predictive variables for ONFH. Comparison between the performance of the AI index and the evaluation made by the two doctors was conducted through calibration and ROC analysis.
## Development of prediction models
A multivariable logistic regression analysis was used to develop the clinical predication model based on patient and clinical variables. AI index classification was applied as a candidate predictor for univariate and multivariable logistic regression analyses for the construction of a DL-based postoperative ONFH prediction model using hybrid variables. A clinical prediction nomogram and a DL-based nomogram were then constructed based on multivariate logistic regression models. The work flowchart of this study is presented in.
## Assessment of nomogram performance
AI-based nomogram and clinical nomogram calibration were assessed using a calibration curve. The discrimination performance of both the AI-based nomogram and clinical nomogram were quantified using the AUC.
## Clinical use
Decision curve analysis (DCA) was performed by calculating the net benefits for a range of threshold probabilities to estimate the clinical utility of the nomogram.
# Statistical analysis
Median and mean standard deviation (SD) were used to describe continuous variables. Categorical variables were presented as frequencies and percentages. Statistical comparisons between groups were performed using the Mann-Whitney U-test and Chi-square test. R software version 3.0.1 was used to construct the nomogram. The "pROC" package was used to plot ROC curves. Nomogram construction and calibration plot creation were performed using the "rms" package. DCA was performed using the "dca.R" package. Model selection was based on the forward-backward step-wise method using the likelihood ratio test with Akaike's information criterion as the stopping rule. The model with the smallest Akaike Information Criterion was selected as the final model. The statistical significance levels reported are all two-sided, with statistical significance set at a P-value of 0.05.
# Results
## Patient and radiograph characteristics
Postoperative radiographs of a total of 238 patients, including 95 ONFH patients and 143 normal patients were used for the development of the DL model and construction of the predictive nomogram. Imaging feature variables were extracted from each radiograph and were referred to as the AI index of all patients.shows the baseline characteristics of the patients. Significant differences were found in BMI, Charlson comorbidity index, Injury Severity Score (ISS), d-dimer, timing of reduction, Garden classification and AI index between patients with ONFH and those without ONFH.
## Performance of the cnn model
A CNN model was established for the extraction of radiograph variables. The precision-recall curve of the test set is shown in, while the threshold value at the break-even point was 0.425. This point was set as the highest sum of sensitivity and specificity. Training accuracy values at this threshold for the training set was 0.903 and 0.873 for the test set. The change in RMSE and loss during the training process are shown in. Deviation of the RMSE in the training set and test set gradually decreased and the two curves leveled off (upper diagram) along with the increase of iterations. Similarly, as the number of iterations increased the deviation in loss between the training set and test set gradually decreased.
## Performance of the predictive radiograph ai variables
The calibration curve of the AI index for the prediction of postoperative ONFH demonstrated good agreement between prediction and actual observations, compared with that of Doctor A and Doctor B. The sensitivity value was 0.910 (95% CI, 0.871-0.949) for the AI index, 0.657 (95% CI, 0.591-0.724) for the less experienced Doctor A and 0.827 (95% CI, 0.776-0.879) for experienced Doctor B. The DCA curves shown inindicate that when the threshold probability for a doctor or a patient was within the range of 0.09-0.96, the AI index added more net benefits for the prediction, than that of Doctor A or Doctor B. It showed that if the threshold probability is between 0.09 and 0.96, then using the AI index adds more benefit than testing either all or no patients.
## Development of a hybrid prediction model
In the univariate logistic regression analysis, BMI, Injury Severity Score (ISS), timing of reduction, Garden classification and AI index were found to be significant factors associated with ONFH in the training cohort (all P < 0.05;. We then created a prediction nomogram that incorporated the above independent predictors and presented it as a hybrid nomogram. A clinical nomogram was also constructed based on independent predictors excluded from the AI index.
## Performance of the hybrid nomogram
The calibration curve of the hybrid nomogram for the prediction of postoperative ONFH demonstrated good agreement between prediction and actual observations, compared with that of the clinical nomogram. The AUC of the AIbased nomogram was 0.948 (95% CI, 0.920-0.976), while the AUC for the clinical nomogram was 0.696 (95% CI, 0.629-0.763). The difference was statistically significant, which indicated that the hybrid nomogram showed better discrimination and prediction ability for the diagnosis of ONFH.
## Clinical use
The DCA for the hybrid nomogram and for the clinical nomogram are presented in. The DCA indicated that when the threshold probability for a doctor or a patient was within the range of 0-0.98, the hybrid nomogram added more net benefits than "treat all" or "treat none" strategies. The range for the clinical nomogram was from 0.2 to 0.7, revealing that use of the hybrid nomogram to predict postoperative ONFH was more beneficial.
# Discussion
Early detection and identification of ONFH after femoral neck fracture fixation has been a long-term concern in clinical practice.
In this study, we developed and trained a DL model that could use postoperative pelvic radiographs to predict ONFH. The output values of the CNN model successfully stratified patients based on their risk of developing postoperative ONFH, which was referred to as AI index classification for prediction. The predictive The threshold probability is where the expected benefit of treatment is equal to the expected benefit of avoiding treatment. It showed that if the threshold probability is between 0 and 0.98, then using the AI-based nomogram adds more benefit in predicting ONFH than testing either all or no patients.
performance of the AI index was significantly superior to the predictive performance of a less experienced orthopedic doctor and non-inferior to that of an experienced orthopedic doctor. A combination of patient and radiograph variables were used to construct an AI-based nomogram for postoperative ONFH prediction. The hybrid nomogram showed better performance for the postoperative prediction of ONFH than a single clinical nomogram, indicating its potential in predicting and targeting ONFH during clinical follow-up to provide a decision base for orthopedic doctors. Hip pain is the most common postoperative symptom after FNF surgery. It may be associated with fractures, surgery, implant irritation, and early ONFH that should be identified during follow-up. Postoperative X-rays are the most common and readily available imaging examination used for routine clinical follow-up after internal fixation. The detection of sclerotic abnormalities and trabecular interruptions of the femoral head for the diagnosis of postoperative ONFH are subjective and depend on the level of experience and diagnostic criteria used by each doctor. Only radiologists who are rich in experience, may be able to accurately predict ONFH using postoperative Xrays. Even then, objectivity and consistency may be difficult to be achieved. The increased workload of radiologists worldwide has already had a significant impact on the diagnostic performance of radiologists. Therefore, DL can be used as a potential auxiliary diagnostic tool for orthopedic diagnoses to obtain stable and accurate diagnoses. In this study, we trained a DL model to read postoperative X-rays to predict ONFH. The accuracy and consistency of the DL model was significantly better than that of an orthopedic doctor with less experience. The DL model was similar in accuracy but better in consistency, compared with the experienced orthopedic doctor. This indicated the potential of the use of the DL model for the diagnosis and prediction of postoperative ONFH. Previous studies have indicated that an important feature of the DL model is its ability to detect key features of images through cyclic learning undergone by neural networks, which may be different from the existing understanding and research on image features in black box models. This makes it possible for the diagnostic path of the DL model to differ from existing known diagnostic and prediction criteria, resulting in a positive difference in the diagnostic accuracy of the DL model, compared with that of orthopedic doctors. The DL model created in Chee's study showed a high level of sensitivity and accuracy for the diagnosis of pre-collapse ONFH. When we applied the CNN network obtained from this non-traumatic ONFH prediction model to our postoperative ONFH prediction, internal fixation of the postoperative X-ray was found to be one of the major differences between the two models. Recent studies have suggested that different fixation constructs, such as cannulated screws or dynamic hip screws, produce different fracture fixation outcomes. The location differences under the implemented operations standard for the same fixation construct do not significantly affect outcomes. During training, we found that the output of the DL model could still reflect prediction efficiency and showed good calibration, even though the positions of the metal internal fixations were not exactly the same and occupied the recognition area in the finite image pixel.
Existing studies using clinical risk factors, such as demographic data, fracture classification, and preoperative interval, to make preoperative predictions for surgical decisions. Due to the lack of the incorporation of all perioperative variables, especially the intraoperative and postoperative radiograph variables, the preoperative prediction models in these studies have shown difficulties in achieving an ideal predictive ability. For example, the clinical nomogram constructed in our study achieved an AUC of 0.696 (95% CI, 0.629-0.763), which is similar to the AUC of 0.746 obtain by the Naive Bayes Classifier constructed by Cui et al.. The predictive ability of a preoperative model is limited for patients who have received certain internal fixation, for example dynamic hip screws and cannulated compression screws. The hybrid nomogram showed better prediction performance after the incorporation of patient and radiograph variables, compared with conventional clinical nomograms and the simple radiographic-based DL model for postoperative ONFH prediction. In this study, the hybrid classifier achieved an AUC of 0.948 (95% CI, 0.920-0.976). The variables we included after multivariate regression analysis of all risk factors were similar to that of conventional preoperative clinical prediction models. High-risk factors generally include fracture patterns, preoperative interval, and BMI. Inclusion of the DL model-based imaging prediction significantly improved the ONFH predictive ability of the traditional prediction models, indicating the value of using a combination of variables. The predictive model using hybrid variables more closely mimicked the diagnostic and predictive processes of orthopedic doctors, who are better at interpreting images based on the clinical status of patients. The addition of a combination of patient and hospital process variables associated with routine clinical care improved the ability of a DL model trained by Badgeley et al. to predict hip fractures. One explanation for this improvement was the presence of non-biological signals on radiographs that are predictive of diseases. Although multiple regression analyses were performed for risk factors, including intraoperative reduction, and postoperative weight bearing, the variables included in the single clinical nomogram were all preoperative variables. Among them, Garden classification showed the most assigned value, which was similar to the results of previous studies that found that fracture patterns are crucial for the prediction of postoperative ONFH. When the postoperative AI index was included, the attribution of Garden classification decreased significantly, which may be because the AI index already included certain manually incorporated graded variables from the images. The information was considered as a non-biological signal and contributed to the classification. The DL-based prediction model that incorporated a combination of patient and radiograph variables showed a significantly higher ability of prediction postoperative ONFH, and can be used to provide second opinions and a base for doctors to make decisions during clinical follow-up.
In the DCA curves analysis, prediction and diagnosis based on the DL model were found to be non-inferior to that of the two orthopedic doctors, while that of the AI-based nomogram using hybrid variables was superior to imaging prediction alone, allowing for more accurate diagnosis and prediction during clinical follow-up. There is no doubt that the gold standard imaging modality for the preliminary stages of ONFH is MRI. However, MRI is not the most common test used to evaluate treatment options and ONFH during postoperative FNF follow-up. MRIs are affected by metal implants, which may cause potential internal fixation losses and thermal effect. MRI tests are more expensive, take longer, and require the radiologist to have a higher level of diagnostic experience. Nomograms based on the DL model and clinical variables can improve the ability of positive diagnostic screening and provide doctors the opportunity of obtaining a second opinion.
The AI-based nomogram using hybrid variables may potentially assist in decision making during clinical followup as patients with early-stage ONFH may benefit from timely interventions. Although the definitive method of treatment for traumatic ONFH remains controversial, certain early interventions have been widely used during post-operative clinical follow-up. For patients with a high probability of developing ONFH, interventions for hip preservation or delayed joint replacement, including platelet-rich plasma (PRP)incorporated autologous granular and free vascularized fibular, have been proven to be safe and effective procedures for postoperative ONFH. Extracorporeal shock wave therapy and alendronate administration can also be potentially performed on patients with a moderate probability of a risk of developing ONFH. We assessed whether the AIbased nomogram assisted decisions that would improve patient outcomes to justify its clinical usefulness. Our study showed that if the threshold probability was between 0.06 and 0.96, as shown by the constructed decision curves, the AI-based nomogram could predict postoperative ONFH compared with treating either all or no patients. This indicated that early postoperative prediction using this hybrid of patient and radiograph variables can be useful for the application of early interventions that may even allow for a reasonable delay of the onset of arthroplasty. Substantial positive rehabilitation can be applied after accurate predictions are obtained after the operation for patients with a lower prediction probability, which will also relieve patient anxiety.
This study has some limitations. First, it was conducted on a retrospective cohort study, and is therefore likely to have been affected by selection bias. Second, due to the rarity of the disease, our study included only 238 images in the CNN model. The performance of the CNN model can be improved by using a larger multicenter sample size. Third, our diagnostic criteria for postoperative ONFH was based on follow-up MRIs and typical pelvic radiographs without the use of histopathological confirmation. Therefore, false-negative and false-positive values would not have been avoided due to the subjectivity of the imaging diagnosis method. At the same time, transverse comparison was not conducted with gold standard MRI when postoperative X-rays were included 6 months after surgery. The reason was that, as a retrospective study, MRIs had been performed on only 197 patients, probably due to their high cost. In the future, prospective clinical studies using larger cohorts should be preplanned to investigate strategies that can be used for ONFH prediction of patients after internal fixation.
# Conclusion
In conclusion, this study presents a DL facilitated nomogram that incorporates hybrid radiograph and patient variables, shows favorable predictive accuracy for preoperative osteonecrosis of femoral head in patients with femoral neck fractures after internal fixation.
# Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
# Ethics statement
The studies involving human participants were reviewed and approved by the First Affiliated Hospital of USTC. The patients/participants provided their written informed consent to participate in this study.
|
10.1186/1532-429X-18-S1-P296
|
CCBY
|
5032546
| null |
s2orc_pubmed_articles
|
Left and Right Ventricular Volumes and Global Systolic Function in Isolated Left Bundle Branch Block: A Cardiac Magnetic Resonance Imaging Study
# Methods
Subjects with isolated LBBB (n=18, 10F) were identified from the hospital echocardiographic database and invited to undergo CMR. Age-, sex-, and body-surface area (BSA)-matched controls (n=18) were identified from a large community-dwelling cohort who had undergone CMR, as part of a population study, on the same CMR scanner using the same SSFP sequence. Ten-mm thick contiguous slices in the LV short-axis orientation encompassing both ventricles were obtained. Right and left ventricular endocardial borders were manually segmented by a trained observer and volumes were determined in the standard fashion using a modified summation of disks method. Ejection fraction was computed as stroke volume divided by end-diastolic volume in each ventricle. Differences between the LBBB group and matched controls were assessed using the Wilcoxon paired signed rank test; a p < 0.05 was considered significant.
# Results
LBBB subjects ranged in age from 37 to 82 years. Left ventricular end-diastolic volume (EDV) and end-systolic volume (ESV) were larger in the LBBB group (p < 0.01), but LV stroke volume (p=0.13) and cardiac output (p=0.18) did not differ between LBBB subjects and controls. Subjects with isolated LBBB had significantly lower LVEF (56 ± 7%) than controls (68 ± 6%), p = 0.0002. Right ventricular volumes and ejection fraction did not differ between isolated LBBB and control groups.
# Conclusions
Using the reference standard of volumetric CMR, our study demonstrates that adults with an isolated LBBB have greater LVEDV and LVESV than age, sex and BSA-Cardiology, Beth Israel Deaconess Medical Center, Boston, MA
[table] Table 1: Baseline and biventricular characteristics of leftbundle branch block (LBBB) subjects and matched controls. Akhtari et al.: Left and Right Ventricular Volumes and Global Systolic Function in Isolated Left Bundle Branch Block: A Cardiac Magnetic Resonance Imaging Study. Journal of Cardiovascular Magnetic Resonance 2016 18(Suppl 1):P296. your next manuscript to BioMed Central and take full advantage of: • Convenient online submission • Thorough peer review • No space constraints or color figure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Akhtari et al. Journal of Cardiovascular Magnetic Resonance 2016, 18(Suppl 1):P296 http://www.jcmr-online.com/content/18/S1/P296 [/table]
|
10.1038/s41467-023-39520-3
|
CCBY
|
10319800
|
37402724
|
s2orc_pubmed_articles
|
Behavioral decomposition reveals rich encoding structure employed across neocortex in rats
Behavioral decomposition reveals rich encoding structure employed across neocortex in rats REVIEWER COMMENTS</B> Reviewer #1 (Remarks to the Author):The authors chronically recorded simultaneously from either motor and somatosensory (left cortex; N=4) or auditory and visual cortical areas (right cortex; N=3) with neuropixels probes in freely foraging rats while monitoring their movements with an array of imaged 3D markers and accelerometers. Recording locations were confirmed by decoding analysis of passive stimulation epochs, and a postmortem combined MRI and immunohistochemical / histological approach. All data were resampled and aligned at 120 Hz, then detrended and decomposed by Morlet wavelet. T-distributed stochastic neighbor embedding (t-SNE) was then used to identify 44 recurring action classes / behavioral motifs, and generalized linear models (GLM) used to determine which components of neural and inferred synaptic activity best predicted the action classes at each time point (binned at 8.33 ms or 1/120Hz). Just over half of neurons in all recorded areas predicted each action above chance. However, components of movement, posture, and self-motion were differentially represented across the four main cortical areas recorded from. Visual cortex represented horizontal scanning head movements and head position, auditory cortex represented gravity relative head position, motor cortex represented overall body and back position, and somatosensory cortex represented back position posture but also had a large proportion of "unclassified" neurons. Putative synaptic connections were identified by statistically enriched cross-correlogram peaks or troughs (i.e. excitatory or inhibitory). Synaptic connections within visual cortex tended to connect neurons with homogenous coding of behavioral subcategories (i.e. movement, posture, self-motion), although there was also an enriched probability of excitatory connections from posture encoding neurons onto movement encoding neurons, and of inhibitory connections from movement neurons onto posture neurons (feedback inhibition motif?). Movement neuron inhibition of posture neurons was the dominant motif in auditory cortex, and pairs of neurons in general that received common input (i.e. CCG peak at ~0 ms) tended to encode the same class of behavioral components.Overall this study seems to be taking the right approach to better understand the cortical neural encoding of spontaneous, ethological behaviors. However, the main findings come across as somewhat preliminary and disjoint, lacking an overall cohesive message. Furthermore, confidence in the main findings in terms of differential action encoding across cortical areas would be enhanced if recordings were simultaneous across all four main areas, and if comparable data existed for these areas in terms of the cortical layer/depth from pia distribution of recorded neurons. Inclusion of data regarding fine facial movements, which are known to be directly indicated arousal state and strongly predict cortical neural activity, would likely strengthen the validity of behavioral action classification. Lastly, use of longer recording sessions would likely enhance the validity of behavioral action transition matrices and allow for observation of the full span of the hierarchy of arousal states / behavioral actions. Were some or all -Separation of regular -and fast-spiking units (Extended dataFigure 2) based on spike duration, peak-totrough ratio, and end-slope is not very convincing.-Was head position really only updated when the rat was running faster than 10 cm / s (Methods, page 12)? This seems like too high of a threshold and likely results in ignoring many real changes in head position in the overall model.
of these concerns to be addressed by the authors, this reviewer may recommend this study for publication. Following are some additional major and minor points of critique.
Major Points:
1.-The identified behavioral actions are too granular and brief, perhaps missing organization from one level higher in the presumed hierarchy of behavioral organization. This can be seen as the lack of apparent sensible organization of the behavioral action transition diagram (Extended [fig_ref] Figure 3: Figure 3: [/fig_ref].
2.-tSNE organization of actions seems to have inherent organization unmentioned by authors -three vertically oriented strips (left, middle, right) seem to encode arousal level / locomotion speed, rightward orienting postures and movements, and leftward postures and movements, respectively 3.-Explanation for why auditory cortical areas would primarily encode gravity relative head movements is not convincing.
4.-The main finding that visual areas strongly encode head movements and head posture seems to not be novel, and may merely represent subtraction of changes in the visual scene due to self-movement, rather than representation of ethological behaviors in these areas, per se.
5.-Examination of recurring synaptic motifs across cortical areas is extremely interesting, but the n used here are perhaps insufficient to make strong statistical conclusions concerning their relative prevalence. In addition, showing that manipulation or modulation of these synapses during particular behavioral actions and/or transitions would greatly contribute to the cohesiveness of the main findings of the study.
6.-Extracted behavioral primitives do a relatively poor job of explaining neural activity in somatosensory cortex , and there is no good explanation for this. Is this due to a different layer recording position in somatosensory cortex, or to the lack of inclusion of fine whisker and facial movements in the overall model, for example? 7. Could the differences between areas result from different neurons being recorded? What if the same neurons (e.g. regular spiking cells in layer 5) were compared between areas -does that affect the results? What are the differences between layers within the same area? Are these differences similar to differences between areas? 8. How much of the widespread nature of activity related to behavior simply the result of the animal experiencing widespread changes in multiple inputs during behavior? E.g. when a animal moves, even in the dark, there is large scale changes in motor, somatosensory, auditory, vestibular, eye movement, etc. signals.
Minor Points: -It's not clear what the partial foraging paradigm adds to the study, other than to ensure a minimum amount of locomotion during the recordings. Also, it seems that the probability of naturalistic behaviors not directly related to foraging may have been decreased by this aspect of the experimental design.
-The vagueness of the final sentence of the abstract speaks to the overall suggestive nature of the results and the lack of a cohesive overall finding.
-More in-depth treatment of behavioral action transition times may allow for analysis of activity motifs and synaptic activity that predict these transitions.
-It's not clear if the authors looked for inter-areal synaptic connections (i.e. visual to auditory), or if they focused only on intra-areal connections (i.e. A1 to AuD).
-Use of alternative manifold embedding techniques, such as UMAP (see " , may allow for better inter-subject identification of high-level behavioral motifs.
-Is it possible to achieve a clearer alignment between the identified elementary poses and species typical actions with assigned ethological meaning? -A clearer explanation of the rationale for use of relative log likelihood ratio as a principal metric for assessing the degree to which each elemental behavioral is represented in each cortical region would be helpful to the general readership.
-Use of a 15 g head weight to control for the possible encoding of muscle contractions as proxies for head position seems unnecessary and perhaps a bit cruel.
-It would have been nice to provide passive somatosensory stimulation so that a hierarchical examination of behavioral vs. sensory encoding (see [fig_ref] Figure 3: Figure 3: [/fig_ref] could have been performed in somatosensory cortex as well as auditory and visual cortices, particularly given the preponderance of unclassified neurons in S1.
-It is not specifically mentioned that the activity of visual cortex neurons could not be used to decode the presence or absence of the white noise stimulus, although it is mentioned that the activity of auditory cortex neurons could not be used to detect the luminance condition.
-Has it already been shown in other studies that back movements tend to be represented in caudal motor and somatosensory cortex, while head movements tend to be represented in corresponding rostral areas? If this is a novel finding, perhaps it warrants more discussion (page 3).
-Does 12% seem like a low percentage of neurons in auditory cortex to be modulated "exclusively" by sound? What is the corresponding percentage for all other recorded areas? -The increased prevalence of "movement inhibited posture units" in auditory cortical areas seems to be of particular interest, and to merit further discussion, particularly in terms of the ability of the model to explain cortical activity during periods of low movement / low arousal.
-Is it possible that the observed increased probability of synaptic connection between neurons with heterogeneous behavioral encoding in "V2L" may actually be due to the mis-assignment of neurons that are actually in posterior parietal cortex (PPC)?
-Is it trivial that pairs of neurons receiving common synaptic inputs appear to be homogenously tuned for behavior (see final point in results)? Or, is this merely an artifact of the analysis? Please provide brief additional discussion on this point.
-Do your results suggest that studies not using video of the back and hindlimbs of rodents will have dramatically insufficient ability to account for variance in neural activity of primary somatosensory cortex? (See Discussion, paragraph 2). Please provide brief additional discussion on this point.
The goal of the study is to examine the neural representation of natural behaviors -both segmented 'momentary behaviors , and more "elementary" behavioral features . The reported findings are that (1) momentary behaviors were represented across all areas recorded, (2) behavioral features showed organization across areas, and (3) tuning properties of synaptically coupled cells differed across areas. The authors collected high-quality video recordings of freely behaving rats while recording single unit activity across multiple brain areas using Neuropixels probes. The methodology to quantify behavior appears sound and consistent with literature, and the statistical methodology behind fitting the array of decoders and GLMs in this paper appears sound.
The main weaknesses in the paper are that the primary evidence used to support claim (b) is based on models that appear to have very low performance in capturing neural responses and further that the differences in recording depth across different areas could be confounding the conclusions. Second, the data and analysis supporting claim (c), as presented in [fig_ref] Figure 4: InFig [/fig_ref] , are somewhat weak. Finally, too much of the substance of the paper is in the Extended Data, making it very challenging to grasp the significance of the results as presented. This is a novel and extensive dataset, and the results of these analyses are of interest, but there are serious shortcomings in its current form.
## Strengths of the paper:
It is clear that a tremendous amount of work went into acquiring, processing, and analyzing the data shown here, and the thoroughness of the documentation of technical details through the extended data figures is excellent. [fig_ref] Figure 3: Figure 3: [/fig_ref] clearly convey results of interest.
To highlight some of the strengths in these parts of the paper: in , the authors outline the methodology for segmenting natural behavior of freely moving rats, which produces a map of identifiable momentary behaviors (center of panel (c)). These momentary behaviors can be decoded from neural activity in all the brain areas recorded, with varying degrees of accuracy. Most momentary behaviors are decoded above a chance-level set by the prior distribution of momentary behaviors, some are decodable with very high accuracy. This is interesting, in that most previous decoding analysis have not focused on representation of a wide range of specific actions, and the fact that specific actions are represented is a novel finding. [fig_ref] Figure 3: Figure 3: [/fig_ref] shows how visual inputs and auditory inputs modulate activity, and that neurons across both areas have similarly broadly distributed representation of behavior variables and sensory variables. This broad representation of behavioral alongside sensory information is important for understanding what cortical computation is -at the most basic level, the inputs and the outputs of these computations are more complex than typically appreciated.
Detailed comments on major concerns:
1. Regarding evidence supporting claim (b)
The analysis supporting claim (b) in primarily in ( "Distinct cortices show differential tuning to posture, movement, and self-motion"). The authors fit individual cells with GLMs through a model selection procedure to identify which behavioral features contributed the most to the model performance. This figure represents how the distribution of those features differs across different brain areas.
Before discussing further, is missing any information on how well the GLMs explained spiking activity, which appears in Extended Data . The measure of model performance is the pseudo-R^2. Pseudo-R^2 would be 0 if the model explained no variance, and 1 if it exactly predicted the spiking of cell. Ext. Data shows that a small minority of cells have a pseudo-R^2 greater than 0.1. The medians (reported in the main text) across different areas range from 0.01 to 0.03. Thus, to interpret these values, the authors need to explain what a "good" value of pseudo-R^2 is for single units during freely moving behavior. This is challenging, as there is no repeated trial structure to use to assess variance across repeated conditions to provide a ceiling to the possible model variance explained. One possibility is to follow a similar approach to the Stringer (Harris, Carandini) 2019 paper, in which they compared feature-derived predictions of neural activity to predictions generated by simultaneously recorded cells. In other words, if the GLM instead used neighboring cell activity as a predictive feature, how well would it perform? This could at least provide a scale by which to assess model performance. Further, I want to know, if you restrict analysis to cells for which the GLM explained a meaningful amount of variance, how does the feature analysis in change?
Turning now to the presentation as-is in , there are four groupings of plots, each consisting of a pie chart ('proportion of n = ## cells'), a bar plot, and a polar chart ('contribution to model performance'). The proportion of cells refers to the proportion of cells for which each of these behavioral features was the top covariate; between ~20% (Motor) and ~50% (Somatosensory) of cells are "unclassified," and presumably not included in the wedge plot at right or the bar plots below. The bar plots show how many covariates were included in the selected model for each cell. This is where one possible difference across areas becomes apparent, where some 40% of somatosensory cells only have one covariate and < 2% have 6 or more covariates; in motor areas especially, cells tend to have more covariates. This is potentially interesting, except that the cells sampled in the somatosensory areas are reported to be primarily Layer 2/3, while the other areas are all sampling deeper layers, either instead of or in addition to L2/3 (manuscript, lines 82 -85). Related, extended data shows V1 vs V2 and A1 vs A2D differences in the number of covariates, but also tractography showing the probe crossing into different layers. If the authors have enough cells sampled from L5/6 in multiple areas, could they repeat this analysis restricted to that subset? Similarly, with L2/3? Do these trends persist?
Finally, the polar chart in each subpanel is represents 'contribution to model performance' of behavioral features. I am somewhat confused on how it was constructed. The angular extent of each wedge is 'the fraction to which each covariate was represented among all those selected.' Does this mean that you take every model for every single unit in the selected brain area, enumerate all the behavioral features that were ever used, and then report the fraction of that pool by feature? Is the contribution weighted by how much additional predictive power each feature contributed? Which cells are included in this analysis --for instance, a substantial fraction of cells have "unclassified" top features, are these excluded? and are cells with pseudo-R^2 of 0.01 included on the same footing as cells with pseudo-R^2 of 0.1 or 0.2?
As a somewhat minor point, I would argue that it is perceptually confusing to set the 'length' of each wedge (rather than the area) to denote the relative log-likelihood ratio. Is 0.6 really four times higher than 0.3 for rLLRs? Again, this is averaged across all the models the covariate was included in, but how does it change if restricted to models with high pseudo-R^2?
It may be worth considering if polar charts are the best way to convey this information in this figure. As is, given the concerns described above, I do not know how to interpret the data shown.
## Regarding evidence supporting claim (d)
In the abstract, the authors state, "tuning properties of synaptically coupled cells also exhibited connection patterns suggestive of different uses of behavioral information within and between regions." This is primarily addressed in [fig_ref] Figure 4: InFig [/fig_ref].
In [fig_ref] Figure 4: InFig [/fig_ref] , the authors use cross-correlograms between pairs of units to identify putative synaptic connections. [fig_ref] Figure 4: InFig [/fig_ref] shows an anatomical projection of auditory (green) and visual (pink) units, with grayscale lines showing putative connections. (The color bar (labeled log10 p) uses the same darkness to represent a p-value of 1 and of 10^-8 and should be reconsidered; at any rate, I cannot see where this colorscale is used in the figure.) The connectivity drawn in anatomical space is difficult to interpret; are cells truly unlikely to connect to their nearest neighbors, but likely to be connected across long A-P distance (roughly, left to right on this plot)? [fig_ref] Figure 4: InFig [/fig_ref] shows a scatter of synaptic strength against distance, with no significant trends. [fig_ref] Figure 4: InFig [/fig_ref] is where the main point is made: that tuning-specific connectivity suggests different uses of behavioral information within and between regions. First, with respect to "between regions," I do not see in this figure any analysis of synapses between regions, so perhaps something else is meant. The question is then, are there different tuning-specific patterns of connection within regions? The evidence for this is thin. [fig_ref] Figure 4: InFig [/fig_ref] the number of synapses found between functionally identified pairs of cells, where "function" is Mo (movement-responsive), Po (posture-responsive), LM (light-modulated) and SM (sound-modulated). (The authors claim that these subsets are largely non-overlapping, but this is not shown.)
In [fig_ref] Figure 4: InFig [/fig_ref] , a shuffle test within the synapses found in each area was performed. Note that there were different numbers of cells in visual vs auditory areas, and about 2x as many synapses found in visual areas. The overall number of detected synapses is also small: from 5 to 15 in any auditory cortex group, and from about 10 to 40 in visual areas. If this is out of the ~10k possible synapses in the 100s of simultaneously recorded cells, this is a very small fraction, and I do not know how much weight I should give these results.
The shuffle test represents the null hypothesis that synapses occur with equal rate independent of functional cell class. In visual areas, there appear to be more Mo to Mo synapses (both excitatory and inhibitory), more excitatory Po to Mo synapses, and more inhibitory Po to Mo synapses. Comparing that result to the auditory result is challenging. Due to the smaller number of cells recorded in auditory areas, it is harder to detect functional patterns of synaptic connectivity over the shuffle baseline. Slightly more inhibitory Mo to Po synapses and Mo to SM synapses were found. The cells and significant connections, for both areas are shown in a functional space, which cannot be understood without referring to the details in Ext. Data [fig_ref] Figure 3: Figure 3: [/fig_ref]. I would suggest that the authors consider whether [fig_ref] Figure 4: InFig [/fig_ref] truly belongs in this paper, and if it does, whether Ext. Data [fig_ref] Figure 3: Figure 3: [/fig_ref] should be included as well. How does it support the main point about 'encoding structure employed across neocortex'?
Minor concerns:
A general minor concern: burying the lede There are several interesting statements that are almost throw-aways in the paper as written. For instance, recordings were performed in dark and light conditions, with the decoding analysis restricted to the dark condition [lines 53-54 ] In [fig_ref] Figure 3: Figure 3: [/fig_ref] , the authors show a dramatic shift (in functional UMAP space, at least) of neural activity in visual areas in lights on vs lights off conditions. The analyses in also utilized the dark condition (lines 70-71), reporting that visual areas maintain information about head position in world coordinates. If this is really in the absence of visual inputs, how is a worldcoordinate representation maintained? Potentially through some sort of angular path integration or triangulation with auditory inputs (which, like visual inputs, have sources that could provide stable world-reference coordinates). At the least, are the GLM features and GLM performances in visual areas different when trained on dark vs. lights-on recordings?
Another result that seemed deserving of more attention was the perturbation experiment in which a 15 g weight was added to the head mount, with results reported in Extended Data and summarized in a few lines (106-113) in the results.
Lines 44 -46: "The animals combined ethogram consisted of 44 independent modular actions ... whose sequence order was best described by a set of transition probabilities"
- Is it shown that transition probabilities are the best description? My understanding of these ethograms is that the behavioral state dynamics tend to be highly non-Markovian, suggesting that transition probabilities alone are insufficient. See, e.g., Alba, Berman, Bialek and Shaevitz, arxiv, 2020.
## Figure 1:
The top panel is relatively clear, as is the behavior map and example postures in the bottom panel. I do not know how to interpret actions 1 to 44 without a map, and I don't understand what new information the map provides that bar heights do not. Could you pick one or the other?
Further, it is a non-intuitive choice to designate dark bars for accuracy of decoding with the prior and light bars for accuracy of decoding using neural activity, while in the maps of decoding, it appears to be light for low accuracy and dark for high accuracy.
A confusion matrix would be good to see in the extended data to understand how distinguishable similar behaviors are.
Minor grammar: "of each covariate across the set of models it was included in" -> "of each covariate across the set of models in which it was included" I do not understand why the average lines in the A1 plots in (B) are blue and yellow; do these correspond to blue and yellow clusters in the UMAP plots? I don't think so, as those seem more likely to correspond to the blue and yellow in panel (C), where light-modulation is demonstrated. Related, why is there no corresponding UMAP plot for sound modulation?
The gradients in (D) are intriguing, but also seem to move in and out of different layers in different areas (as portrayed in Extended Data . This makes it difficult to determine whether it is a proximal-distal gradient, or a laminar gradient.
Figure 3(E) is nice: it appears there is a distributed representation of sensory inputs and behavior across auditory and visual areas, and that this is more or less similar between FS and RS units. It would be helpful to have unit counts at each branch. The caption says 'We note that the outlined pair groups are largely separated." for [fig_ref] Figure 4: InFig [/fig_ref] This separation is not clear to me in the UMAP plot at right. More details on the functional subspace would be helpful for interpreting this plot. Extended Data [fig_ref] Figure 3: Figure 3: [/fig_ref] is useful in this regard, but [fig_ref] Figure 4: InFig [/fig_ref] needs to stand on its own. Extended Data [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] All x-axes should be the same The x-label says cross-validated log-likelihood, but the caption says cross-validated rLLRs.
Reviewer #3 (Remarks to the Author):
Summary:
Mimica*, Tombaz* et al present an impressive and unique study that examines the functional coding properties of neurons in somatosensory, motor, visual, and auditory cortices during natural behavior. By first parsing the natural behavior into actions (e.g. rearing), the authors find that most actions can be decoded in each region. By then parsing the behavior into continuous movements, such as head orientation in egocentric and allocentric coordinates, the authors uncover coding properties specific to each region. E.g. visual cortex encodes allocentric head movement, while motor cortex encodes egocentric movement. The authors then show that neurons in visual and auditory cortices encode visual and auditory information alongside behavioral information, and that there is structure to the putative synaptic connections in visual and auditory regions.
Overall, the data collected are quite impressive, as are the wide range of presented analyses. Understanding neural activity during natural behavior is of high interest to many but an incredibly difficult task. Here the authors have very bravely waded into those waters and come out with some interesting results. However, I felt there were some issues with the manuscript in its current form. First, while in general I can get behind the sharing of results even if their full impact or interpretation isn't entirely clear yet (as every paper is a small part of a bigger story), I felt that some of the results presented here really didn't have quite enough contextualization, interpretation, or motivation. There are a lot of facts in this paper, and in some places it was hard to know what to take away, or what exactly was learned that wasn't clear before. More details on this point are listed below, but overall, I felt this applied most to results in [fig_ref] Figure 3: Figure 3: [/fig_ref]. This issue is also exacerbated by the current short format of the article (which I'm not sure is necessary, although I admit I don't know the character limits here). Further, some of the figures were hard to parse, and some analyses or methodological details were either missing or hard to find. These issues and others are detailed below.
1. As mentioned above, I felt that in a couple places the figures were difficult to parse. While some figure panels are mentioned in other comments, issues pertaining to the rest have been collected into the list below.
a. -Artistically speaking, the circle looks nice. But it is very hard to visualize the decoding accuracy or compare the size of the bars in different regions in this format. While less exciting, it might be easier on the reader (and better convey the points you make in the text) to show rows of bar plots. In addition, the comparison to the control decoder, in which decoding is based on the prior, is vital, but difficult to see here.
b. [fig_ref] Figure 3: Figure 3: [/fig_ref] i. I find the lines in [fig_ref] Figure 3: Figure 3: [/fig_ref] ,c to be visually confusing -I see it is a double axis, but it makes it quite hard to interpret.
ii. The proportion breakdown in the bottom of [fig_ref] Figure 3: Figure 3: [/fig_ref] ii. Extended Data 10 -this figure has a ton of information in it, but with the number of panels and subpanels, it's difficult to pull out the relevant facts. In some cases, I was not sure what was being plotted -e.g., what the 'information rate' in panel e is being computed from (the encoding model?). I would suggest substantially simplifying this figure.
2. How were spikes sorted across multiple 'schedules' (the group of 4 sessions)? It's my understanding that the Neuropixels shank was implanted and in the same place across these schedules, but were the spikes sorted together? Or were cells sorted separately in each schedule? Unless I have missed something, it seems this latter method would lead to double-counting of the same cells.
3. In Line 51, the authors claim that each action is encoded by some population of neurons in each region. However, in Extended Data [fig_ref] Figure 3: Figure 3: [/fig_ref] , it looks like there are quite a few actions that are not encoded by neurons in any region. (To me, this figure actually highlights that only a subset of actions seem to be encoded for in each region.) Perhaps this can be directly quantified somewhere, either in the main text or the figure legend, to make it clearer that there are neurons encoding these actions. -the number of neurons will affect decoding accuracy (e.g. as in [fig_ref] Figure 3: Figure 3: [/fig_ref]. However, I could not find the average number of simultaneously recorded cells (or how much this number varied across sessions), or whether this was controlled for when presenting decoding accuracy across regions. Also, given that motor regions encode egocentric movement the strongest, did the authors also observe higher decoding accuracy for actions here as well?
## Related to the decoding result in
5. Regarding the model fits -20 minute sessions are quite short for fitting spike trains to natural behavior data, and the resulting pseudo-R2 values are quite low with medians ~ 0.02. While this is to be expected to some degree for natural behavior no matter how long the recordings are, the authors may be able to provide additional support for the encoding properties they present if they are indeed tracking cells across schedules (or even across light1 and light2 sessions). In this case, are the encoding properties (variables encoded and their tuning) generally similar? If not, the data presented in Extended Data Figure 5 is relevant and convincing; however, there are only 4 example cells shown. How often are the tuning curves similar across session halves?
6. Regarding the result regarding the addition of a weight -in the main text, it seems that the addition of the weight did not change the encoding properties of motor cortex cells. However, in extended data fig 10, it is difficult to see the extent to which this is the case. In particular, while the proportion of cells encoding different features is similar across the population in panel G, it is unclear whether other aspects of the encoding, such as the weights or log-likelihoods, are similar. Perhaps a more straightforward comparison would be to train on a light1 session, test the model on the weighted and light2 session, and then compare the resulting model fits. Further, one could directly compare the glm tuning for models fit with the same variables in each session. For what it is worth, I think this potentially a very interesting result, since this relates to ongoing arguments of encoding kinematics versus dynamics in motor cortex (which could also be referenced more directly in the text). [fig_ref] Figure 3: Figure 3: [/fig_ref] -currently, much more attention is paid to the overlap in sensory and behavioral variables in visual and auditory cortices than motor and somatosensory cortex. While I somewhat understand the logic for looking at sensory variables in sensory cortex, this has the unfortunate effect that the main results seem to be that visual cortex responds to visual stimuli, and auditory cortex responds to auditory stimuli. It's unclear what we've learned from this result, given that this would be expected. However, given that authors recorded from somatosensory and motor cortex during these trials as well, it might be interesting to include all four regions in these analyses and discuss the differences and similarities. For example, does the behavioral encoding in motor or somatosensory neurons depend on light/dark conditions? Are neurons in these regions modulated by luminance/sound and/or behavior? 8. Extended Data does not include p-values for the changes in encoding properties of neurons in the light and dark conditions. Do you think that the results reflect an actual change in the encoding of movement features, or that the actual features that the neurons are responding to are visual in nature and are only correlated with movement in the well-lit condition?
## Regarding the results in
9. There is a thread regarding FS/RS cells that I am not following. I think that part of the confusion is that FS/RS breakdowns in [fig_ref] Figure 3: Figure 3: [/fig_ref] are not mentioned in the text, and then FS/RS properties relating to [fig_ref] Figure 4: InFig [/fig_ref] are mentioned in the discussion, but not mentioned in [fig_ref] Figure 4: InFig [/fig_ref] or the related main text in the results (unless I missed something). Further, it's not clear to me what the main results are regarding FS/RS cells. For example, in [fig_ref] Figure 3: Figure 3: [/fig_ref] , both types of cells seem to have very similar properties. What is the result that the authors are trying to highlight here? 10. In [fig_ref] Figure 4: InFig [/fig_ref] , the take home message to me is not clear. I think perhaps part of the issue is that it is unclear what should be expected here, or how the results presented specifically update our knowledge of functional connectivity in visual and auditory cortex. I think additional text is needed to motivate and contextualize these results. There has been prior work on functional connectivity in at least visual cortex (Ko*, Hofer* et al 2011), which can be discussed here. Further, it is unclear what the functional space embeddings are supposed to convey.
a. In addition, it is not immediately clear to me why the focus is on auditory and visual cortex. The results in Extended Data 13 F-G could be compared to those in [fig_ref] Figure 4: InFig [/fig_ref] C-D. For example, does the increased precedence of Po->Mo excitatory connections in visual and somatosensory cortices as opposed to auditory and motor cortices align with a model similar to those described in the caption of 2. Line 51 -it is not completely clear what the numbers in the parentheses refer to. Is this the number of neurons that encode at least behavior in each region? I think the wording should be changed, or text added, so this is more transparent.
3. To help compare the number of simultaneously encoded variables across regions, the axis of the bar plot in should be the same as the other panels.
4. Lines 94-98, Lines 118-122 -the authors report a gradient in encoded features, but it is unclear to me whether the observed gradient is surprising or fits with the current literature. I think a sentence or two contextualizing these results would help. [fig_ref] Figure 3: Figure 3: [/fig_ref] is not correct -the legend says that the FS spikes are in green, but I do not see any no green spikes. 10. Lines 418-422 -it is unclear how the two halves of the split-half analyses were obtained.
## Either the legend or the panel in
## Response letter, mimica et al. 2022.
We thank the reviewers for their constructive input and suggestions for how to improve the prior version of our manuscript, 'Behavioral decomposition reveals rich encoding structure employed across neocortex'. Accordingly, the manuscript has undergone substantial modifications, with the inclusion of new sets of expanded and supporting analyses, revision and inclusion of new main and supplementary figures, and the addition of text (changes written in bold) throughout the manuscript to provide a clearer narrative and context for the results. New analyses include examination of tuning properties across cortical layers (including superficial and deep layers in visual and motor cortices, and granular and deep layers in auditory cortex) in relation to naturalistic actions and low-level pose and movement variables [fig_ref] Figure 3: Figure 3: [/fig_ref]. This extended our original results by showing that regionally distinct tuning to lower-level features was conserved across layers, but was more abundant in deeper layers in visual and auditory cortices. We also included examples of stable tuning curves for posture and movement features in superficial and granular and deep layers. In addition, we analyzed fast-and regular-spiking cell types in relation to higher-and lower-level behavioral features , which again showed largely similar tuning preferences, but with fast-spiking cells preferentially encoding movement features in somatosensory and visual cortices. The revision also includes confusion matrices to convey the specificity of the decoding results , and additional GLM analyses comparing tuning properties in the head-weight condition and across light and dark recordings . To make to make key supporting data more easily available, we moved figures showing tuning curves and the topographical analysis to the main text (Figs. 2 and 4, respectively). The revised manuscript, including new analyses, consists of 6 main figures and 21 supplementary figures.
The rebuttal also provides several critical tests of our analytical methods requested by the Reviewers, which validated the results of our GLM framework as well as our methodology for extracting individual actions. We have provided point-by-point responses to each of the Reviewers' concerns below.
## Reviewer #1 (remarks to the author):
The authors chronically recorded simultaneously from either motor and somatosensory (left cortex; N=4) or auditory and visual cortical areas (right cortex; N=3) with neuropixels probes in freely foraging rats while monitoring their movements with an array of imaged 3D markers and accelerometers. Recording locations were confirmed by decoding analysis of passive stimulation epochs, and a post-mortem combined MRI and immunohistochemical / histological approach. All data were resampled and aligned at 120 Hz, then detrended and decomposed by Morlet wavelet. T-distributed stochastic neighbor embedding (t-SNE) was then used to identify 44 recurring action classes / behavioral motifs, and generalized linear models (GLM) used to determine which components of neural and inferred synaptic activity best predicted the action classes at each time point (binned at 8.33 ms or 1/120Hz). Just over half of neurons in all recorded areas predicted each action above chance. However, components of movement, posture, and self-motion were differentially represented across the four main cortical areas recorded from. Visual cortex represented horizontal scanning head movements and head position, auditory cortex represented gravity relative head position, motor cortex represented overall body and back position, and somatosensory cortex represented back position posture but also had a large proportion of "unclassified" neurons. Putative synaptic connections were identified by statistically enriched cross-correlogram peaks or troughs (i.e. excitatory or inhibitory). Synaptic connections within visual cortex tended to connect neurons with homogenous coding of behavioral subcategories (i.e. movement, posture, selfmotion), although there was also an enriched probability of excitatory connections from posture encoding neurons onto movement encoding neurons, and of inhibitory connections from movement neurons onto posture neurons (feedback inhibition motif?). Movement neuron inhibition of posture neurons was the dominant motif in auditory cortex, and pairs of neurons in general that received common input (i.e. CCG peak at ~0 ms) tended to encode the same class of behavioral components.
Overall this study seems to be taking the right approach to better understand the cortical neural encoding of spontaneous, ethological behaviors. However, the main findings come across as somewhat preliminary and disjoint, lacking an overall cohesive message.
We appreciate the Reviewer's supportive view of our approach in the study, and have modified and added several portions of text throughout the manuscript to impart a more cohesive message than before, including the endings of the Abstract and Introduction, at the beginning of each of the Results section, and in the Discussion. The cohesive message we hope reads more clearly now is that, by quantifying behavior carefully and at multiple levels, we can generate sensible insights into how such signals are employed by different cortical systems during natural behavior, particularly in sensory areas.
Furthermore, confidence in the main findings in terms of differential action encoding across cortical areas would be enhanced if recordings were simultaneous across all four main areas, and if comparable data existed for these areas in terms of the cortical layer/depth from pia distribution of recorded neurons. Inclusion of data regarding fine facial movements, which are known to be directly indicated arousal state and strongly predict cortical neural activity, would likely strengthen the validity of behavioral action classification. Lastly, use of longer recording sessions would likely enhance the validity of behavioral action transition matrices and allow for observation of the full span of the hierarchy of arousal states / behavioral actions.
We thank the Reviewer for suggesting ways in which the action classification and transition analyses could be strengthened. This prompted us to include several new analyses and figures (listed below) which we feel have strengthened the study, though we are not able to oblige all suggestions. We provide responses below for each point raised.
Regarding simultaneous recordings across cortical areas: we wish to clarify that we do not claim to show differential coding of actions across cortical areas. On the contrary, we find action encoding is similar across regions, which is supported with an equivalence test [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref]. We contend that simultaneous recordings of all 4 regions would not make the claim of equivalence stronger because one could attribute regional similarities to behavioral confounds with recordings coming from the same animal.
Regarding recordings across cortical layers and cell types (RS vs. FS): this is an excellent suggestion, and we have included breakdowns of the data accordingly in a new , related to action encoding, and in new Supplementary Figs. 13 and 14, related to GLM results. Additional details on the results are included in the point-by-point responses below.
Both facial tracking and longer recordings were suggested as ways of validating our action classification and transition matrices using 20-minute recordings. We investigated this issue more closely on the assumption that, if our methods and recording durations were insufficient to extract meaningful actions, then the classification results should be arbitrary from one recording session to the next. However, we found that both action classification and the associated transitions were highly consistent across sessions (discussed below for point M1), which supports that our behavioral analysis approach is valid.
We otherwise agree that facial tracking would give added insight into internal states, and that longer recordings would add insight into the hierarchical organization of behavior. These are related and fundamental questions facing the field currently, but addressing them would require major technical and analytical developments which we feel fall beyond the scope of the present work. One is that the field has not yet solved how long behavioral recordings in rodents need to be before higher levels of the behavioral hierarchy emerge. The one such study of which we are aware that quantified rat behavior over long time scales recorded from animals in a home environment on a 24/7 schedule for >1 week. We feel that repeating our recordings following such a schedule, and in a necessarily new chronic recording setup, exceeds the scope of the present study. Likewise, inclusion of facial tracking would require starting over with new sets of recordings with facial cameras. This would necessitate the new design and incorporation of implants that house facial cameras alongside the electrophysiology hardware, as well as synchronization of the data streams, and expansion of our rate-coding, GLM and action classification pipelines. We respectfully argue that this would constitute grounds for a new standalone study.
We contend that our study, by focusing on time scales of seconds-to-minutes, still makes critical steps toward understanding how primary sensory and motor cortices represent unrestrained behavior at multiple levels of complexity. Such work is needed as a precursor before investigating more complex levels of hierarchical organization, and prior to considering additional variables such as facial features. We now make the distinction in the Results (lines 46-49) and at the close of the Discussion (lines 328-333) that our study focuses on action coding at shorter, rather than longer, time scales.
Major Points:
M1.-The identified behavioral actions are too granular and brief, perhaps missing organization from one level higher in the presumed hierarchy of behavioral organization. This can be seen as the lack of apparent sensible organization of the behavioral action transition diagram (Extended [fig_ref] Figure 3: Figure 3: [/fig_ref].
We regret that the structure of the behavioral transitions was obscured by the way it was presented originally, giving the impression that the data lacked a sensible organization with too-brief behaviors. To address this, we (i) arranged the action transition matrices to match the tSNE embedding in , so the actions and their transitions appear less arbitrary (shown in a new Supplementary [fig_ref] Figure 4: InFig [/fig_ref] ; see below). We (ii) checked the stability of the transition matrices in light vs. dark sessions and found they were very highly conserved across the recordings (Pearson's R of 0.968, p = 0.00058), which speaks very much against them being arbitrary. The average duration of actions across conditions was also similar across conditions (0.47±0.33 s (± SD) in light sessions vs. 0.49±0.36 s in dark sessions). We note that these M2.-tSNE organization of actions seems to have inherent organization unmentioned by authorsthree vertically oriented strips (left, middle, right) seem to encode arousal level / locomotion speed, rightward orienting postures and movements, and leftward postures and movements, respectively.
We more closely inspected the organization of behavioral features in tSNE space and indeed found that aspects of rudimentary pose and movement followed prevailing directions across the t-SNE embedding. Though not precise, there was an overall increase in running speed from the top to the bottom of the embedding, as well as changes along a left-right diagonal for the pitch of the head and back. We point out these patterns now in the legend of Supplementary [fig_ref] Figure 4: InFig [/fig_ref] , where the map of actions in the tSNE embedding is introduced:
## Head lowered in the upper left portions in t-sne space and raised at lower and rightward regions)."
M3.-Explanation for why auditory cortical areas would primarily encode gravity relative head movements is not convincing.
We have expanded our interpretation of the potential use of gravity relative postural signals (i.e. for head roll and pitch) in auditory cortex. We explain that rodents rely on interaural loudness differences (ILDs) for localizing sound sources in the environment, and that an animals' head posture will strongly influence how these inputs are sampled. Strong head roll and pitch modulation (i.e. in the vertical plane) will facilitate detection of differences in loudness in vertical space, allowing for 3D sampling of sound in the environment. Conversely, if the head and ears are perfectly level, then any ILD corresponds to a change in horizontal localization, and the individual is "blind" to changes in vertical localisation. Once the head rolls, that changes, therefore knowing the pitch and roll of the head is probably critical to interpreting a localization cue. The revised text in the Discussion (lines 313-319) reads:
"The roll and pitch of the head might be heavily represented because these features would strongly affect how interaural loudness differences (ILDs), used by rodents to locate sound sources, are sampled in the environment. If an animal's head is perfectly level, any ILD corresponds to a change in horizontal location, and the individual is "blind" to changes in vertical localization. Once the head rolls, this changes, and strong roll-and pitch-modulation would facilitate detection of ILDs in vertical space, allowing for 3D sampling. . We note, however, that the precise influence of posture on sound localization in rodents remains largely untested."
M4.-The main finding that visual areas strongly encode head movements and head posture seems to not be novel, and may merely represent subtraction of changes in the visual scene due to self-movement, rather than representation of ethological behaviors in these areas, per se.
We appreciate the concern regarding confounds due to visual scene changes during movement; however, we used recordings in darkness specifically to avoid the issue of visual scene changes. Since this is a critical methodological aspect, we note it again in the Discussion (below), in addition to the Results. Regarding the novelty of the visual cortex results, we added text to better contextualize our findings relative to prior work. In our view the novelty is that we measured several behavioral features at the same time and evaluate their relative importance, in addition to considering different frames of reference which the earlier studies did not do. The revised text from the Discussion (lines 275-280) is below:
"The activity of visual ensembles was best described by allocentric horizontal motion of the head and movement of the body over the surface of the arena, which confirms and extends earlier observations that features like angular head velocity , head direction , and running each modulate spiking activity in V1. The combination of these signals likely supports a variety of computations for stabilizing and predicting motion of the visual field, as well as optimizing visual processing during active movement ."
We also agree that movement representation in visual cortex would signal when visual scene changes are due to self-movement in lit conditions, which we note in the Discussion on functionally connected neurons (lines 290-295):
"More importantly, such fast spiking movement-responsive cells also inhibited postureencoding units (e.g., a left turn cell inhibiting a cell encoding rightward posture), and posturemodulated neurons were shown to excite luminance-modulated and movement-responsive cells (e.g., a rightward pose cell exciting a leftward-rotation neuron). This might assist visual networks in distinguishing whether optic flow is likely generated by self-movement : if postural cells convey static head position, their direct inhibition would indicate that visual scene changes are likely self-generated."
M5.-Examination of recurring synaptic motifs across cortical areas is extremely interesting, but the n used here are perhaps insufficient to make strong statistical conclusions concerning their relative prevalence. In addition, showing that manipulation or modulation of these synapses during particular behavioral actions and/or transitions would greatly contribute to the cohesiveness of the main findings of the study.
The total number of putative synaptic connections reported in the study is 650: 247 for visual, 107 for auditory, 35 for somatosensory and 261 for motor cortices. We agree in proceeding with caution regarding sampling that is limited, but we note that the inclusion criteria for counting cells as synaptically connected were strict (cross correlation p-values < 0.0001), as were criteria for significance in the rate of putative synaptic connections (exceeding 99% confidence intervals beyond shuffled data). We have acknowledged the limited sample size by modifying wording in the revised manuscript (lines 201-205): , b)."
## "we identified a limited, but informative number of putative excitatory synaptic connections (driven by rs units) and inhibitory connections (driven by fs units) of varying strength within auditory (n = 107 total connections) and visual cortices (n = 247) (fig. 6a, b; methods) as well as somatosensory (n = 35) and motor cortices (n = 181) (supplementary data
We nevertheless believe it is important to report these results for the following reasons: (1) The field of single-unit electrophysiology as a whole still very much relies on drawing statistical inferences on limited samples, e.g., in this study we make inferences based on samples of ~1000 neurons per region, though anatomically distinct areas are composed of different layers with tens of millions of neurons each. Given this, we believe it is still relevant to disclose our findings fully, given the particular rigor applied in constructing statistical arguments.
(2) We would also argue that recordings as expansive as the ones we are reporting, in freely-moving and freely-sensing animals, can still be considered a valuable rarity. They offer new insights which point to fundamental network properties of cortical areas which lack recurrent connections characteristic of other networks (e.g. in the head direction or hippocampal systems), which significantly lowers the probability of detecting putative synaptic relationships, but are still valuable as a reference point for future studies.
(3) Although the absolute number of reported putative connections is limited, we would argue that the analyses employed on such data still enable us to uncover meaningful effects. The shuffling method used in [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] , (formerly [fig_ref] Figure 4: InFig [/fig_ref] compared observed data to that if the single units we recorded were randomly connected. Because each shuffle iteration constructs the same number of random pairs as those that have been observed, having more or fewer pairs does not make the shuffle test easier or harder to pass. Notably, previously reported findings of movement-inhibited sound-responsive units in A1 were replicated in our experiments, arguably lending credence to other synaptic motifs we are reporting.
As for manipulations-now that we have generated observations regarding the nature of the functional connections (e.g. posture → luminance modulated cells), it would be possible to generate testable hypotheses about the function of the connections. Testing this, however, poses a major technological challenge, requiring selective perturbations of neurons whose functional characteristics are determined in vivo. The only approach that could work for this, to our knowledge, would be all-optical, using holographic silencing (or stimulation) of cells that were functionally identified using 2-photon microscopy in freely moving subjects. These combined approaches are still just on the horizon for experiments in unrestrained animals. We feel that developing and applying such novel methods would be a major technological advance, but would constitute a new stand-alone study outside the scope of the present work. We include this as a direction for future work in the final paragraph of the Discussion (lines 324-328):
"Pinpointing the mechanisms by which behaviorally generated signals shape sensory circuit computations will challenging in freely moving animals, requiring sufficiently resolved techniques, such as miniature 2-photon imaging and holographic stimulation , to identify then manipulate behaviorally-tuned neurons in vivo."
M6.-Extracted behavioral primitives do a relatively poor job of explaining neural activity in somatosensory cortex , and there is no good explanation for this. Is this due to a different layer recording position in somatosensory cortex, or to the lack of inclusion of fine whisker and facial movements in the overall model, for example? This is a shrewd observation, and we now give more attention to the lower rates of observed tuning in S1 in the revised manuscript. To check if the underlying reason was due to recording in superficial layers, we compared tuning in layer 2/3 across somatosensory and motor cortices but still found a comparatively lower proportion of assigned covariates in S1:
We did, however, uncover differences when we split S1 recordings into FS and RS neurons, with larger fractions of FS neurons encoding movement of the back and self-motion:
We note in Discussion that the propensity of FS neurons to represent movement features in S1 is quite consistent with earlier work from the Svoboda lab, reporting movement-driven increases in the spiking of parvalbumin neurons during active whisking in S1 barrel cortex (Yu et al., Neuron, 2019). Though we cannot prove the same circuit is at play in our recordings, our results are consistent, and we point this out in the Discussion (quoted below).
As for the overall larger number of unclassified neurons in S1, since it is unlikely attributable to the layers recorded, we suspect the unclassified cells respond to features which we did not track, such as the whiskers, face, limbs or tail. Given that the electrodes traversed the trunk and hindlimb regions of S1, we suggest that the limbs in particular could contribute, and have revised the Discussion (lines 260-270) accordingly:
"Somatosensory neurons, on the other hand, were better characterized by sparser models related to the trunk, and tuning was more common among FS than RS cells. Fast spiking neurons in particular encoded dynamic features such as trunk movement and self-motion, which fits broadly with known circuitry for movement-driven disinhibition of parvalbumin interneurons in S1, described in barrel cortex during active whisking . Overall, however, model performance in S1 was the lowest of all regions, which could be traced to different factors. It was not likely due to the recordings being confined to layer 2/3, since only 22% of layer 2/3 neurons in adjacent motor regions were unclassified. In this case, unclassified neurons could have been driven by features we did not track, such as the whiskers, face, limbs or tail. The hindlimbs may have been more influential in our recordings since the probes were placed in the hindlimb and trunk regions of S1, medial to barrel cortex, and the fact that trunk kinematics in quadrupeds are steadily affected by gross dynamics of the hindlimbs and hips ."
M7. Could the differences between areas result from different neurons being recorded? What if the same neurons (e.g. regular spiking cells in layer 5) were compared between areasdoes that affect the results? What are the differences between layers within the same area? Are these differences similar to differences between areas?
These are excellent questions, and in response we have included multiple new analyses with the recordings separated by RS and FS cell types (as noted above) and by cortical layers, where recorded.
As for action coding, this revealed that the distribution of encoded actions and the proportions of tuned cells were similar for RS and FS neurons in all regions except visual cortex, where actions were encoded by a slightly higher fraction of FS neurons (Supplementary shown below). In visual and motor areas, the distribution of encoded actions appeared similar across deep and superficial layers. The same was true between layer 4 and deep layers in auditory cortex (Supplementary . Similar percentages of cells encoded actions in superficial and deep layers of motor cortex, but there were notably fewer coding cells in superficial than deep layers in visual cortex (39% tuned cells in Layer 2/3, 41% in Layer 4, and 53% in layers 5&6) and auditory cortex (40% in layer 4, 58% in layers 5 & 6):
## Supplementary figure 7
For the lower-level features, GLM analyses showed that regional variations persisted across deep and superficial layers in visual and auditory cortices, which still primarily encoded head posture, head movement and self-motion (new Supplementary [fig_ref] Figure 3: Figure 3: [/fig_ref] , shown below). Similar to the action-encoding results, tuning was more widespread in deep layers of visual and auditory cortices which, at least for visual cortex, is consistent with earlier work showing that deep layer V1 neurons respond to head rotations in darkness . Both those and our findings point to the deeper cortical layers as a possible site of integration for motor input from midline motor cortex and vestibular signals relayed by retrosplenial cortex , noted in the Discussion (lines 280-284).
In motor cortex, the laminar analysis uncovered differences across deep and superficial layers, with the trunk being encoded more strongly in superficial layers and the head represented more in deep layers (see Supp. [fig_ref] Figure 3: Figure 3: [/fig_ref] , below). However, due to the tilt of the Neuropixels, layer ⅔ was sampled more posteriorly, and layer 5 was more anterior-which follows an established topography of head and trunk representation over the dorsal cortical surface in rats . Further, our own topographical analysis (now main [fig_ref] Figure 4: InFig [/fig_ref] showed a continuous evolution of back-to-head representation progressing from posterior to anterior recording sites. So, while we cannot rule out differences in cell layers, the result can be accounted for by known anatomical gradients. We note the earlier anatomical mapping studies and topographical organization in our data in the Discussion (lines 270-272). In separating recordings from FS and RS neurons, we found a larger proportion of FS neurons encoded dynamic features of the trunk in S1 (as discussed above), and a slightly larger fraction of FS neurons encoded allocentric head movement and selfmotion in visual cortex [fig_ref] Figure 4: InFig [/fig_ref]. The potential functions of such fast-spiking activity in visual cortex are considered in the context of functional connectivity in the 5 th paragraph of the Discussion (lines 273-284). FS and RS neurons in auditory and motor cortices encoded similar proportions of features.
## Supplementary figure 13
## Supplementary figure 14
M8. How much of the widespread nature of activity related to behavior simply the result of the animal experiencing widespread changes in multiple inputs during behavior? E.g. when a animal moves, even in the dark, there is large scale changes in motor, somatosensory, auditory, vestibular, eye movement, etc. signals.
We appreciate this as a fundamental question, but one that is difficult to address concretely. For decades animal subjects have been treated as a single point in space with a location, bearing and speed, as with studies of spatial coding in the hippocampal system. Our goal was to take a step beyond traditional approaches, to track more, and to start by using the simplest environment we could provide: an open arena, in darkness, without sound or other overt stimuli. We point out that we succeeded in finding stable tuning for the subsets of behavioral features we could track, then systematically manipulated inputs where feasible (e.g. light/dark sessions, sound stimuli, head weight).
It is also important to note that the regional differences in low-level features could not be explained by gross changes in inputs, since these would presumably have been the same across areas. To further test whether tuning across cortical regions changed following gross changes in environmental input, we performed new GLM analyses on recordings from all regions in dark and light conditions, and assessed whether the proportions of selected features changed with a two sample Z-test for proportions. This showed that only visual cortex was sensitive to the light-dark manipulation (Supplementary , and the lack of change in other regions is included in the Results (lines 188-195).
It remains a major question in the field as to how the many additional features which we could not (yet) track influence spiking activity, or how neural tuning in different systems would change in more complex environments or during different motivational states. We suggest these as directions for future work in the last paragraph of the Discussion (lines 328-330).
Minor Points: m1. It's not clear what the partial foraging paradigm adds to the study, other than to ensure a minimum amount of locomotion during the recordings. Also, it seems that the probability of naturalistic behaviors not directly related to foraging may have been decreased by this aspect of the experimental design.
The foraging paradigm was used to promote movement and exploration of the open recording arena, though it indeed probably influenced the proportion of foragingrelevant behaviors, such as bouts with lower speed, with the head lowered to retrieve food off the floor. This aspect of the task design is now acknowledged in the last sentence of the 2nd paragraph in the Results:
"The behavioral dynamics observed were also likely influenced by the foraging task, and would differ from purely exploratory patterns behavior in novel environments ."
m2. -The vagueness of the final sentence of the abstract speaks to the overall suggestive nature of the results and the lack of a cohesive overall finding.
We have changed the first and final sentences of the abstract to emphasize the main findings that (i) ongoing behavior is represented at multiple levels potentially throughout the dorsal cortex, and (ii) that low-level pose and movement representations vary by region and may support locally-relevant computations, particularly in sensory regions. The last sentence of the Abstract now reads:
"Together, our results indicate that ongoing behavior is encoded at multiple levels throughout the dorsal cortex, and that low-level features are differentially utilized by different regions to serve locally relevant computations."
We underscore these themes again in revised text in the last paragraph of the Introduction, and in motivating statements provided at the beginning of each section in the Results.
m3. -More in-depth treatment of behavioral action transition times may allow for analysis of activity motifs and synaptic activity that predict these transitions. This is a great suggestion and prompted us to more closely investigate the neural encoding of transitions between specific actions. We found potential evidence of differential encoding of actions (action "B") conditioned on the preceding action (action "A"), but ultimately could not conclude if apparent tuning to transitions were in fact tuning to only to action "B" due to limited sampling. The mean duration of behaviors conditioned on the preceding action was 0.09s (min=0.01s, max=1.9s, std=0.1s). This was for all combinations of behaviors for which there were at least 3 occurrences of the ordered pair (shown in the matrix directly below):
m4. -It's not clear if the authors looked for inter-areal synaptic connections (i.e. visual to auditory), or if they focused only on intra-areal connections (i.e. A1 to AuD).
Inter-areal (e.g. visual to auditory) synaptic connections were indeed too rare to perform quantitative analyses; we have modified the text to no longer mention interareal synaptic connections.
m5 -Use of alternative manifold embedding techniques, such as UMAP (see "BSOiD", Hsu et al, 2021), may allow for better inter-subject identification of high-level behavioral motifs. This is an interesting point, and prompted us to test whether identification of behavioral motifs was improved by other approaches such as UMAP. We took the watershed colorcoded scatter plot of training points used for the tSNE embedding (below, left), and performed 2D embedding with UMAP (below, right).
By comparing them, we observed that the UMAP and tSNE embeddings performed similarly in preserving both local structure (i.e. tSNE classified point are clustered together in the UMAP space) and the global structure (the gradient of color in the two embeddings is the same, with the exception of "rearing" and "rearing up"). We tested if the distributions were uncorrelated using the Jaccard's similarity for the k-nearest neighbors. At both micro (k=2) and meso (k=100) scales, the p-value associated with the median of the Jaccard's similarity was 0.001 (or lower) under a randomization test, indicating that UMAP and tSNE embeddings were very unlikely to be uncorrelated. Since we did not find evidence that the embeddings differed, we respectfully prefer to maintain our original usage of tSNE and associated results.
The "BSOiD" study by was published while we were preparing our manuscript last fall, but we now cite it with the other unsupervised approaches for behavioral classification in the Introduction. m6 -Is it possible to achieve a clearer alignment between the identified elementary poses and species typical actions with assigned ethological meaning? This is again an interesting suggestion. We also wished to know whether specific postural features aligned with the expression of specific actions, but did not find quantitative evidence to support it. Specifically, we tested whether postural variables occurred more or less often during specific behaviors vs. all other behaviors. A twosided Mann Whitney U test showed that none of the medians of the posture variables conditioned on any of the actions was different from the median of the same postural variable across all the other actions. P values < 5*10ˆ-5 = 0.05/44/23 (Bonferroni correction).
We did, however, re-align the lists of actions in Supplementary Figs. 5a and b (formerly 4a and b) to highlight commonalities in postural features among similar behaviors (e.g. when animals were walking with the head rolled, or while turning in place). The features by which actions are grouped are written in bold on the right of the groupings in the updated . m7 -A clearer explanation of the rationale for use of relative log likelihood ratio as a principal metric for assessing the degree to which each elemental behavioral is represented in each cortical region would be helpful to the general readership.
We used the relative log likelihood ratio (rLLR) because it provides a metric for how much the predictive power of the statistical model suffered when a behavioral covariate is removed. By setting the mean rLLR as the height of the wedges in the polar charts in [fig_ref] Figure 3: Figure 3: [/fig_ref] , the "importance" of each feature to the model is conveyed visually. We include this explanation in the Results (lines 98-99) and in the legend of [fig_ref] Figure 3: Figure 3: [/fig_ref].
m8 -Use of a 15 g head weight to control for the possible encoding of muscle contractions as proxies for head position seems unnecessary and perhaps a bit cruel.
We can appreciate this point of concern for the animals' welfare. However, the added weight to the head did not restrict the free movement of the animals during the experiments, and we limited the use of the weight to the time needed for the recordings. The only measurable behavioral difference was the tendency for the animals' heads to roll slightly leftward during the weight sessions. We have added text in the Discussion (lines 252-259) to better motivate the experiments in light of longstanding debates in the motor cortex literature:
"The robust encoding of head features allowed us to test whether M1 neurons primarily encoded spatial kinematics or muscle exertion controlling the head, a question classically debated in the context of hand and arm movements in primates . The addition of the head weight had only minor effects on tuning curve stability and statistical modeling of headrelated covariates, indicating that spiking activity was linked more closely with spatial aspects of movement or specific postures (see , rather the generation of mechanical force. Since the animals were freely moving, signals related to sensory feedback, planning or dynamic pattern generation may have also contributed and cannot be ruled out ." m9-It would have been nice to provide passive somatosensory stimulation so that a hierarchical examination of behavioral vs. sensory encoding (see [fig_ref] Figure 3: Figure 3: [/fig_ref] could have been performed in somatosensory cortex as well as auditory and visual cortices, particularly given the preponderance of unclassified neurons in S1.
We agree that such experiments could have been informative, but there would have been serious practical complications in delivering sensory stimuli to S1 without impeding the animals' natural movement in the arena. Unlike auditory or visual input, which are uniformly available throughout the recording arena and do not interfere with mobility, delivering somatosensory stimulation (e.g. stroking or applying vibration) to the body would have interfered with the animals' movement and biased behavioral sampling. We therefore restricted our tests to the more readily controlled visual and auditory sense modalities, and note this in the Results (lines 171-173):
"We next sought to characterize the degree to which sensory and behavioral and signals overlapped in sensory cortices, focusing on visual and auditory modalities since these could be manipulated without disrupting the animals' movement." m10-It is not specifically mentioned that the activity of visual cortex neurons could not be used to decode the presence or absence of the white noise stimulus, although it is mentioned that the activity of auditory cortex neurons could not be used to detect the luminance condition.
We re-checked the text and found that the result was included already. It can be found in the revision on lines 178-180:
"Decoding analyses confirmed that auditory, but not visual, units predicted sound stimulus presentation ." m11--Has it already been shown in other studies that back movements tend to be represented in caudal motor and somatosensory cortex, while head movements tend to be represented in corresponding rostral areas? If this is a novel finding, perhaps it warrants more discussion (page 3).
This is a relevant question to which we now give more attention in the revised Discussion (and noted above). Cortical microstimulation experiments in anesthetized rats have shown an overall topographical organization of the back and head spanning the posterior-to-anterior extent of somatomotor cortex. This topography, to our awareness, had not been considered in relation to 3D pose or movement in freely moving animals prior to our study. Our findings and the early microstimulation experiments are mentioned in the Discussion, as noted above. m12-Does 12% seem like a low percentage of neurons in auditory cortex to be modulated "exclusively" by sound? What is the corresponding percentage for all other recorded areas?
While 12% would seem low for the fraction of neurons exclusively modulated by a stimulus in a sensory cortical region, the total fraction of sound-modulated cells in our data was 35.3% (including neurons co-modulated by behavior). This number falls roughly in the middle of the range of sound modulated neurons reported prior studies in A1 of awake rats: Hromádka et al. 2008 reported 5-10% excited by white noise or pure tones; Abolafia et al. 2011 reported 70% responding neurons to white noise. In awake mice, reported 31% of layer ⅔ auditory cortical neurons responded to tone pips. By comparison, the total fraction of sound modulated neurons in visual areas was 7.4% (82 of 1107 tested neurons), which consisted of a subset of V2L neurons near the A1 border (see [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref]. We did not assess sound modulation in motor or somatosensory cortices. m13-The increased prevalence of "movement inhibited posture units" in auditory cortical areas seems to be of particular interest, and to merit further discussion, particularly in terms of the ability of the model to explain cortical activity during periods of low movement / low arousal.
We have added additional text on this point in the Discussion (lines 309-313):
"Movement-encoding fast spiking cells were also shown to inhibit allocentric postureencoding cells, which mostly occurred in A2D (e.g., a cell encoding leftward rolling of the head inhibiting a cell for right head roll posture). Circuits with this pattern of connectivity could report minute changes in posture whenever the animal moved and, conversely, maintain a strong readout of the current postural state when the rat was still." m14-Is it possible that the observed increased probability of synaptic connection between neurons with heterogeneous behavioral encoding in "V2L" may actually be due to the mis-assignment of neurons that are actually in posterior parietal cortex (PPC)?
We investigated this possibility by re-checking our anatomical delineations, and by conferring with anatomists with expertise in rat PPC. This re-affirmed that the recording locations in the right hemisphere (ranging from -5.5 to -6.5mm AP) were posterior to the caudal boundary of rat PPC, following regional boundaries of (Paxinos & Watson, 2013). m15-Is it trivial that pairs of neurons receiving common synaptic inputs appear to be homogenously tuned for behavior (see final point in results)? Or, is this merely an artifact of the analysis? Please provide brief additional discussion on this point.
The analysis used to establish the nature of connections between single units was agnostic to their functional properties, and relied solely on the spike train patterns. In that sense, neurons receiving common input did not have to be homogeneously tuned to behavior a priori. We wished to draw attention to the discrepancy between the synaptic and common input motifs that we observed, which in our view lends credence to these findings not being an artifact of the analysis.
The fact that the cells receiving common input were homogeneously tuned is relatable to findings where such properties have been hypothesized to be endowed by their upstream sources (e.g. . What is arguably not trivial is the difference between synaptic and common input pairs, where synaptic pairs form more functionally heterogeneous relationships, hopefully lending further credence to reporting the results in general (in relation to Point M5, above). In order to clarify the focus on this discrepancy, we added the following sentence in the Discussion (lines 306-309):
"The finding that homogeneously-and heterogeneously-tuned pairs showed different connectivity motifs, and that the analysis was agnostic to the functional properties of the cells, suggests that functional groupings were not an artifact of the analysis." m16-Do your results suggest that studies not using video of the back and hindlimbs of rodents will have dramatically insufficient ability to account for variance in neural activity of primary somatosensory cortex? (See Discussion, paragraph 2). Please provide brief additional discussion on this point.
We did not intend to suggest that behavioral tracking which omits the back and hindlimbs is dramatically insufficient to account for spiking variability in S1. We believe, however, that model performance will vary depending on where in S1 the recording takes place and which features are tracked. Tracking the hindlimbs, for example, may be of little service in explaining spiking activity in a barrel column. In our case the probes were placed in the trunk and hindlimb regions of S1, so tracking the trunk likely carries some explanation for our recordings. We include additional clarification of this point in the Discussion, lines 266-270 (quoted above in relation to Point M6). m17-Separation of regular -and fast-spiking units based on spike duration, peak-to-trough ratio, and end-slope is not very convincing.
The method employed to separate regular and fast spiking units was developed by to study visual cortical connections in mice, and has since been used by a number of labs to study behavior-related spiking in the rodent primary visual cortex. These include studies on head movement-driven spiking activity in mice and rats . The latter study, like ours, showed what FS/RS separation looked like in different datasets in Guitchounts at al.), which strongly resembled the separation we report. Because of the similar nature of the experiments, we reason that these studies and their methods provide useful points of comparison for our own work. Moreover, the unsupervised kmeans clustering approach used to uncover the separating boundary positioned it at 0.5 ms of spike duration in our data, which closely matches previously described separating points between these subtypes of cells (e.g. .
m18--Was head position really only updated when the rat was running faster than 10 cm / s (Methods, page 12)? This seems like too high of a threshold and likely results in ignoring many real changes in head position in the overall model.
Head position tracking was continuously monitored at all running speeds. The 10 cm / s threshold was used only to center the coordinate systems for head direction and the animals' running direction, since the head and body are likely to be better aligned when the animal is walking forward or running. This threshold was not used in any other analyses. We have clarified the text accordingly in the Methods (lines 475-6):
"Time points where the speed exceeded 10 cm / s were used to center the coordinate systems for the animals' head direction and running direction." m19-Use of the acronym "IMU" for "inertial measurement unit" is unnecessarily confusing. These instruments are typically referred to as "accelerometers".
We have changed our usage of "IMU" to accelerometer in the main text, but kept the term as used per manufacturer product information in the Methods. m20-Please provide a definition of "XYZ Euler angle method".
We have added the following explanation of the XYZ Euler angle method in the Methods, lines 482-484:
"The XYZ Euler angle method indicates the three elemental rotations are intrinsic about the axes of the rotating coordinate system XYZ, solidly with the moving body, which changes its orientation after each elemental rotation." m21--Behavioral feature vectors seem a bit overconstrained with 133 dimensions (Methods, page 13). Comparison of results after removal of redundant dimensions may be informative.
A 133-dimension feature vector is what was produced from the signal analysis, and we reduced redundant dimensions using PCA, keeping 97.2% of the variance from the first 22 PCs. We hope this clarification addresses the concern. We have updated the text (Methods, lines 520-522) so that the steps are clearer:
"Redundant dimensions were removed using Principal Component Analysis (PCA), which indicated that the first 22 principal components explained 97.2% of the variance".
m22--Downsampling of feature vectors to 1 Hz (Methods, page 13) may be excessive and prevent proper high-resolution alignment of neural activity with behavioral action transitions. A comparison with results after downsampling to 10-30 Hz, a range that better matches the frequency components of natural rodent movement, would be informative.
We fully agree that downsampling to 1Hz would have prevented proper alignment of neural activity and behavioral transitions, but 1Hz downsampling was only performed in our data during embedding, which was necessary to avoid artifacts driven by temporal autocorrelations. Our analysis of behavior and its correlation with neural activity (encoding, decoding) operated at 120Hz. This point has been clarified in the Methods (lines 518-520):
"Feature vectors were downsampled to 1 Hz, and data were pooled across animals and conditions. Downsampling was performed only for the sake of embedding, and did not affect behavioral analyses or correlations with neural activity (e.g. encoding, decoding), which used 120Hz tracking data." m23-The authors often state that movement drives or influences activityhowever, this is only a correlation. I recommend the authors use more circumspect language such as "movement is associated with….".
We have adopted more circumspect language as suggested by the Reviewer and, where applicable, employed more exact terms referring to neural "encoding" of actions, or that movement was associated with neural spiking activity (e.g. line 7 of the Introduction, or line 237 of the Discussion). m24-The results from the fast spiking category of cells is commented upon in the discussion but I couldn't find it presented in the main body of the results section.
We believe this referred to the functional connectivity analyses in the original [fig_ref] Figure 4: InFig [/fig_ref]. We have revised the text in that section of the Results so it is explicit that excitatory connections were driven by regular spiking units and that inhibitory connections were driven by fast-spiking units (lines 201-205, quoted above in relation to point M5).
## Reviewer #2 (remarks to the author):
General comments:
The goal of the study is to examine the neural representation of natural behaviorsboth segmented 'momentary behaviors' (e.g., rearing, turning), and more "elementary" behavioral features (posture, movement, position, etc.). The reported findings are that (1) momentary behaviors were represented across all areas recorded, (2) behavioral features showed organization across areas, and (3) tuning properties of synaptically coupled cells differed across areas. The authors collected high-quality video recordings of freely behaving rats while recording single unit activity across multiple brain areas using Neuropixels probes. The methodology to quantify behavior appears sound and consistent with literature, and the statistical methodology behind fitting the array of decoders and GLMs in this paper appears sound.
The main weaknesses in the paper are that the primary evidence used to support claim (b) is based on models that appear to have very low performance in capturing neural responses and further that the differences in recording depth across different areas could be confounding the conclusions. Second, the data and analysis supporting claim (c), as presented in [fig_ref] Figure 4: InFig [/fig_ref] , are somewhat weak. Finally, too much of the substance of the paper is in the Extended Data, making it very challenging to grasp the significance of the results as presented. This is a novel and extensive dataset, and the results of these analyses are of interest, but there are serious shortcomings in its current form.
## Strengths of the paper:
It is clear that a tremendous amount of work went into acquiring, processing, and analyzing the data shown here, and the thoroughness of the documentation of technical details through the extended data figures is excellent. [fig_ref] Figure 3: Figure 3: [/fig_ref] clearly convey results of interest. To highlight some of the strengths in these parts of the paper: in , the authors outline the methodology for segmenting natural behavior of freely moving rats, which produces a map of identifiable momentary behaviors (center of panel (c)). These momentary behaviors can be decoded from neural activity in all the brain areas recorded, with varying degrees of accuracy. Most momentary behaviors are decoded above a chance-level set by the prior distribution of momentary behaviors, some are decodable with very high accuracy. This is interesting, in that most previous decoding analysis have not focused on representation of a wide range of specific actions, and the fact that specific actions are represented is a novel finding. [fig_ref] Figure 3: Figure 3: [/fig_ref] shows how visual inputs and auditory inputs modulate activity, and that neurons across both areas have similarly broadly distributed representation of behavior variables and sensory variables. This broad representation of behavioral alongside sensory information is important for understanding what cortical computation isat the most basic level, the inputs and the outputs of these computations are more complex than typically appreciated.
We thank Reviewer 2 for their thorough consideration of the work, and we are encouraged to read that they feel our methods and analytical framework are sound, and that our dataset is novel and extensive. We agree with the Reviewer that it is interesting to see that a wide range of specific actions was broadly represented in cortex, and we were happy to read that they view the result showing side-by-side representation of sensory information and behavior as important to understanding cortical computation. We have worked to address each of the points of concern raised by the Reviewer, and respond to each of them below.
Detailed comments on major concerns: M1 (multi-part; continues below). Regarding evidence supporting claim (2) The analysis supporting claim (b) in primarily in ( "Distinct cortices show differential tuning to posture, movement, and self-motion"). The authors fit individual cells with GLMs through a model selection procedure to identify which behavioral features contributed the most to the model performance. This figure represents how the distribution of those features differs across different brain areas.
Before discussing further, is missing any information on how well the GLMs explained spiking activity, which appears in Extended Data . The measure of model performance is the pseudo-R^2. Pseudo-R^2 would be 0 if the model explained no variance, and 1 if it exactly predicted the spiking of cell. Ext. Data shows that a small minority of cells have a pseudo-R^2 greater than 0.1. The medians (reported in the main text) across different areas range from 0.01 to 0.03. Thus, to interpret these values, the authors need to explain what a "good" value of pseudo-R^2 is for single units during freely moving behavior. This is challenging, as there is no repeated trial structure to use to assess variance across repeated conditions to provide a ceiling to the possible model variance explained. One possibility is to follow a similar approach to the Stringer (Harris, Carandini) 2019 paper, in which they compared feature-derived predictions of neural activity to predictions generated by simultaneously recorded cells. In other words, if the GLM instead used neighboring cell activity as a predictive feature, how well would it perform? This could at least provide a scale by which to assess model performance. Further, I want to know, if you restrict analysis to cells for which the GLM explained a meaningful amount of variance, how does the feature analysis in change?
We wish to first clarify that the relative log-likelihood ratios (rLLR) values were indicated on the polar charts in original (ranging from 0.1 to 0.6), intended to convey mean importance of each covariate in the model. Due to space limitations, pseudo-R^2 values were shown fully in the original (now .
To address the concern regarding low pseudo-R^2 values from the original GLM (referred to here as the "Covariates Model"), we performed additional peer prediction analyses using spiking activity from simultaneously recorded cells (separated by >5 recording sites on the probe) as a predicted feature. We refer to this model as the "Peer Model", built according to the Stringer et al. (2019) paper suggested by the Reviewer.
We then compared the proportions of cells tuned to at least one behavioral feature from the Covariates Model vs. the subset of cells with larger pseudo-R^2 values than the corresponding Peer Models (i.e. cells with "good" pseudo-R^2 values). We found that the defining regional differences in encoding properties were upheld when only considering the cells with "good" pseudo-R^2 values. That is, head movement, head posture and self-motion still dominated in visual cortices; allo-and egocentric head posture still dominate in auditory cortex; motor regions still encoded the trunk and head, and somatosensory was still dominated by trunk representation (next page):
Since these additional analyses support our original findings and conclusions as they were, we respectfully suggest leaving the Figure and GLM portion of the paper as they were originally. , there are four groupings of plots, each consisting of a pie chart ('proportion of n = ## cells'), a bar plot, and a polar chart ('contribution to model performance'). The proportion of cells refers to the proportion of cells for which each of these behavioral features was the top covariate; between ~20% (Motor) and ~50% (Somatosensory) of cells are "unclassified," and presumably not included in the wedge plot at right or the bar plots below. The bar plots show how many covariates were included in the selected model for each cell. This is where one possible difference across areas becomes apparent, where some 40% of somatosensory cells only have one covariate and < 2% have 6 or more covariates; in motor areas especially, cells tend to have more covariates. This is potentially interesting, except that the cells sampled in the somatosensory areas are reported to be primarily Layer 2/3, while the other areas are all sampling deeper layers, either instead of or in addition to L2/3 (manuscript, lines 82 -85). Related, extended data shows V1 vs V2 and A1 vs A2D differences in the number of covariates, but also tractography showing the probe crossing into different layers. If the authors have enough cells sampled from L5/6 in multiple areas, could they repeat this analysis restricted to that subset? Similarly, with L2/3? Do these trends persist?
## M1.turning now to the presentation as-is in
This is an excellent point, and was raised by other Reviewers as well. We have now included analyses with the data separated by superficial and deep layers in visual and motor cortices, as well as for layer 4 vs. deep layers in auditory cortex:
For the lower-level features, GLM analyses showed that regional variations persisted across deep and superficial layers in visual and auditory cortices, which mainly encoded head posture, head movement and self-motion (new Supplementary [fig_ref] Figure 3: Figure 3: [/fig_ref] In motor cortex, the laminar analysis uncovered differences across deep and superficial layers, with the trunk being encoded more strongly in superficial layers and the head represented more in deep layers (below). However, due to the tilt of the neuropixels, layer ⅔ was sampled more posteriorly, and layer 5 was more anterior-which follows an established topography of head and trunk representation over the dorsal cortical surface in rats . Further, our own topographical analysis (now main [fig_ref] Figure 4: InFig [/fig_ref] showed a continuous evolution of back-to-head representation progressing from posterior to anterior recording sites. So, while we cannot rule out differences in cell layers, the result can be accounted for by known anatomy. We note these earlier anatomical mapping studies and the topographical organization in our recordings in the Discussion, lines 270-272.
To further substantiate that neurons in different layers encode low-level behavioral features, we included additional examples of stable tuning curves from single cells in main (superficial or granular layers) and Supplementary M1.Finally, the polar chart in each subpanel is represents 'contribution to model performance' of behavioral features. I am somewhat confused on how it was constructed. The angular extent of each wedge is 'the fraction to which each covariate was represented among all those selected.' Does this mean that you take every model for every single unit in the selected brain area, enumerate all the behavioral features that were ever used, and then report the fraction of that pool by feature? Is the contribution weighted by how much additional predictive power each feature contributed? Which cells are included in this analysis --for instance, a substantial fraction of cells have "unclassified" top features, are these excluded? and are cells with pseudo-R^2 of 0.01 included on the same footing as cells with pseudo-R^2 of 0.1 or 0.2?
We apologize for not explaining sufficiently the construction of the polar charts in the original manuscript. For clarification:
(i) The pie charts in the upper left of each panel show the proportion of all recorded cells which took a particular feature as the first covariate in the GLM and were assigned a color-coded wedge based on that covariate.
(ii) The polar charts were made using only those cells from the colored pie wedges ("Unclassified" cells were excluded), and only considered features that were included in the full statistical model after model selection. For example, a neuron best explained by "allocentric head pitch + azimuth" would fall in a dark purple wedge in the pie chart, but the contribution of each feature of (head pitch and azimuth) are shown in separate wedges in the polar chart.
Cells with low and high pseudo-R^2 values are not separated, since separating them did not change the trends in tuning observed in each region.
The legend for [fig_ref] Figure 3: Figure 3: [/fig_ref] has also been revised:
## Fig. 3 | distinct cortices show differential tuning to posture, movement and selfmotion. a, (top left) pie charts showing the fraction of single units in visual cortices (v1, v2l and v2m) that incorporated specific behavioral features as the first covariate (largest mean cross-validated log-likelihood among single covariates models) in model selection (for feature identification, refer to the color-coded legend at bottom). (lower left) bar graph indicating the percentages of single units statistically linked to one or any larger number of behavioral covariates. (right) polar charts denoting the relative importance of individual covariates included in the full model, after model selection, with "unclassified" cells excluded. wedge length denotes the mean cross-validated rllr (methods) of each covariate across the set of models it was included in. the width of each wedge reflects the fraction of times that feature was selected among the other covariates. wedge length and width are independent of each other (i.e. the area does not reflect effect size).
M1.As a somewhat minor point, I would argue that it is perceptually confusing to set the 'length' of each wedge (rather than the area) to denote the relative log-likelihood ratio. Is 0.6 really four times higher than 0.3 for rLLRs? Again, this is averaged across all the models the covariate was included in, but how does it change if restricted to models with high pseudo-R^2? It may be worth considering if polar charts are the best way to convey this information in this figure. As is, given the concerns described above, I do not know how to interpret the data shown.
Our goal with the polar charts in the current [fig_ref] Figure 3: Figure 3: [/fig_ref] is to convey the importance of each behavioral feature in each cortical area in a visually intuitive way (which we now clarify in the Results, lines 98-99). We previously tried other conventions, such as bubble plots, but found polar charts to be the most efficient way to fit the results in a single figure.
The width of the wedges in the polar charts indicates the number of times that a particular feature was selected. The length of the wedge indicates how much the predictive power of the model suffered when that feature was removed. The length and width of the wedges should be viewed as 2 independent properties; we do not merge the values, and the area is not meant to convey the size of the effect. We clarify this point in the revised legend of [fig_ref] Figure 3: Figure 3: [/fig_ref] (quoted above)
To further ease concerns about the way in which the data were presented, we include here comparisons of (i) polar charts showing mean rLLR values generated using full models as in the paper, and (ii) polar charts restricted to "good" cells (pseudo-R^2 values higher than corresponding peer-prediction models). This again confirms that the results presented in the paper hold up when only considering models with "good" performance (next page; note the difference in scale for Visual cortex polar charts):
M2. Regarding evidence supporting claim (d) In the abstract, the authors state, "tuning properties of synaptically coupled cells also exhibited connection patterns suggestive of different uses of behavioral information within and between regions." This is primarily addressed in [fig_ref] Figure 4: InFig [/fig_ref].
In [fig_ref] Figure 4: InFig [/fig_ref] , the authors use cross-correlograms between pairs of units to identify putative synaptic connections. [fig_ref] Figure 4: InFig [/fig_ref] shows an anatomical projection of auditory (green) and visual (pink) units, with grayscale lines showing putative connections. (The color bar (labeled log10 p) uses the same darkness to represent a p-value of 1 and of 10^-8 and should be reconsidered; at any rate, I cannot see where this colorscale is used in the figure.) The connectivity drawn in anatomical space is difficult to interpret; are cells truly unlikely to connect to their nearest neighbors, but likely to be connected across long A-P distance (roughly, left to right on this plot)? [fig_ref] Figure 4: InFig [/fig_ref] shows a scatter of synaptic strength against distance, with no significant trends.
To respond to the points in order: the color bar in the previous [fig_ref] Figure 4: InFig [/fig_ref] (now [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] indicated the significance of the cross correlogram values in each bin. The colors of the scale bar and data have been highlighted so that the color-code for significance is more visible than before. The middle of the scale bar reflects the significance threshold of p = 0.0001. We have grouped the scale bar with the cross-correlogram results so it is clearer that they go together.
We have adjusted the rendering of connectivity in anatomical space [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] , having moved the colored dots (neurons) to the back so that the lines depicting anatomical connections are more visible. Cells were connected to nearby neighbors as well as to cells over longer (~50 micron) distances, which were limited by the total width of the shank. The plot is only intended to provide a sense of how the cells are positioned relative to each other, and to show that they do not cluster only in one specific point along the probe. The scatter plots of synaptic strength and anatomical distance in panel 6b show that there were no significant trends between synaptic strength and distance. The lack of a trend is in our view also informativestrong connections, in any area we looked, were not limited to an immediate proximity, which motivated further functional analyses. [fig_ref] Figure 4: InFig [/fig_ref] is where the main point is made: that tuning-specific connectivity suggests different uses of behavioral information within and between regions. First, with respect to "between regions," I do not see in this figure any analysis of synapses between regions, so perhaps something else is meant.
The Reviewer is indeed correct that the putative synaptic connections occurred between cells within the same region, and we have modified the text in the Abstract and Results accordingly.
The question is then, are there different tuning-specific patterns of connection within regions? The evidence for this is thin. [fig_ref] Figure 4: InFig [/fig_ref] the number of synapses found between functionally identified pairs of cells, where "function" is Mo (movementresponsive), Po (posture-responsive), LM (light-modulated) and SM (soundmodulated). (The authors claim that these subsets are largely non-overlapping, but this is not shown.)
The direct evidence of non-overlapping cells pairs proved difficult to convey visually in the original UMAP plots in 4c, so they were removed from the main figure. The functional results are now shown together in , along with supporting statistics showing a lack of preferential coupling between functionally similar neurons . We felt this was an important result to include since, as opposed to orientation tuning in visual cortex, in which similarly tuned neurons form functional sub-groups (e.g. , functionally disparate neurons in our data were just as likely to have strong connections as functionally similar neurons. This raises the prospect that communication between functionally disparate cells is important, possibly even more so when the content (behavioral vs. sensory signals) is different. We make this point now in the last sentence of the Results section in the revision.
In [fig_ref] Figure 4: InFig [/fig_ref] , a shuffle test within the synapses found in each area was performed. Note that there were different numbers of cells in visual vs auditory areas, and about 2x as many synapses found in visual areas. The overall number of detected synapses is also small: from 5 to 15 in any auditory cortex group, and from about 10 to 40 in visual areas. If this is out of the ~10k possible synapses in the 100s of simultaneously recorded cells, this is a very small fraction, and I do not know how much weight I should give these results.
Regarding the number of putative synapses in visual and auditory areas, there were 3x more neurons recorded in visual areas, so this would be expected to be reflected in the number of putative synaptic connections identified as well. The total number of putative synaptic connections reported in the study is 650: 247 for visual, 107 for auditory, 35 for somatosensory and 261 for motor cortices. We agree in proceeding with caution regarding sampling that is limited, but we note that the inclusion criteria for counting cells as synaptically connected were strict (cross correlation p-values < 0.0001), as were criteria for significance in the rate of putative synaptic connections (exceeding 99% confidence intervals beyond shuffled data). We have acknowledged the limited sample size by modifying wording in the revised manuscript: [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] ; Methods) as well as somatosensory (n = 35) and motor cortices (n = 181) ."
## "we found a limited, but informative number of putative excitatory synaptic connections (driven by rs units) and inhibitory connections (driven by fs units) of varying strength within auditory (n = 107 total connections) and visual cortices (n = 247)
We nevertheless believe it is important to report these results for the following reasons: (1) The field of single-unit electrophysiology as a whole still very much relies on drawing statistical inferences on limited samples, e.g., in this study we make inferences based on samples of ~1000 neurons per region, though anatomically distinct areas are composed of different layers with tens of millions of neurons each. Given this, we believe it is still relevant to disclose our findings fully, given the particular rigor applied in constructing statistical arguments.
(2) We would also argue that recordings as expansive as the ones we are reporting, in freely-moving and freely-sensing animals, can still be considered a valuable rarity. They offer new insights which point to fundamental network properties of cortical areas which lack recurrent connections characteristic of other networks (e.g. in the head direction or hippocampal systems), which significantly lowers the probability of detecting putative synaptic relationships, but are still valuable as a reference point for future studies.
(3) Although the absolute number of reported putative connections is limited, we argue that the analyses employed on such data still enable us to uncover meaningful effects. Results reported in [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] , (formerly [fig_ref] Figure 4: InFig [/fig_ref] compare observed data to that if single units we recorded were randomly connected. Notably, previously reported findings of movementinhibited sound-responsive units in A1 were replicated in our experiments, arguably lending credence to other synaptic motifs we are reporting.
The shuffle test represents the null hypothesis that synapses occur with equal rate independent of functional cell class. In visual areas, there appear to be more Mo to Mo synapses (both excitatory and inhibitory), more excitatory Po to Mo synapses, and more inhibitory Po to Mo synapses. Comparing that result to the auditory result is challenging. Due to the smaller number of cells recorded in auditory areas, it is harder to detect functional patterns of synaptic connectivity over the shuffle baseline. Slightly more inhibitory Mo to Po synapses and Mo to SM synapses were found. The cells and significant connections, for both areas are shown in a functional space, which cannot be understood without referring to the details in Ext. Data [fig_ref] Figure 3: Figure 3: [/fig_ref].
The shuffle test does not assume that synapses occur with an equal rate, independent of functional cell class. On the contrary, the most represented functional classes (where "not classified" is included also) would, on average, tend to form more random synapses, as they would be more likely to get chosen by chance. This is precisely what we were after: we wanted to know whether the relative differences in putative connection subtypes were an artifact of some functional subtypes occurring more often. As the Reviewer rightly observed, certain synapses appear more often that others (movement parameters are, for example, more prevalent in visual and postural ones in auditory areas, as established by the GLM [fig_ref] Figure 3: Figure 3: [/fig_ref] , and our results indicate that the patterns we observe do not match those that would be established by forming completely random connections. Moreover, we respectfully point out that the comment that it is harder to detect functional patterns of synaptic connectivity in auditory pairs because of lower numbers is not correct. The method we employed can be utilized regardless of the total number of pairs, because each shuffle iteration constructs the same number of random pairs as those that have been observedin that sense, having more or fewer pairs does not make the shuffle test easier or harder to pass.
Regarding functional space plots, they are now only shown in , where the analysis is explained. I would suggest that the authors consider whether [fig_ref] Figure 4: InFig [/fig_ref] truly belongs in this paper, and if it does, whether Ext. Data [fig_ref] Figure 3: Figure 3: [/fig_ref] should be included as well. How does it support the main point about 'encoding structure employed across neocortex'?
The message of the paper, which we hope is clearer in the revised manuscript, is about differences in tuning in different cortical areas, and the functional analyses in [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] do not fall short in supporting that conclusion. We believe the added motivating text, along with additional explanation of the results, make it clearer that these analyses contribute to the narrative and should therefore remain in the paper.
Minor concerns: m1.A general minor concern: burying the lede There are several interesting statements that are almost throw-aways in the paper as written. For instance, recordings were performed in dark and light conditions, with the decoding analysis restricted to the dark condition [lines 53-54 ]. In [fig_ref] Figure 3: Figure 3: [/fig_ref] , the authors show a dramatic shift (in functional UMAP space, at least) of neural activity in visual areas in lights on vs lights off conditions. The analyses in also utilized the dark condition (lines 70-71), reporting that visual areas maintain information about head position in world coordinates. If this is really in the absence of visual inputs, how is a world-coordinate representation maintained? Potentially through some sort of angular path integration or triangulation with auditory inputs (which, like visual inputs, have sources that could provide stable world-reference coordinates). At the least, are the GLM features and GLM performances in visual areas different when trained on dark vs. lights-on recordings?
This is an interesting question and we are glad to provide more clarification. The world-based reference frame for pitch and roll here is defined by measuring rotations of the head relative to the arena floor (i.e. relative to the direction of gravity), so vestibular signaling, which is constant across lights-on and -off conditions, and which drives head-motion tuning in deep visual cortical layers , is very likely a key contributor to visual cortical tuning we found in darkness. Vestibular signals contributing to movement in azimuth will also be intact, regardless of whether it is measured relative to the room (allocentric) or the trunk (egocentric). Proprioceptive and efference copy signals also remain in darkness, the latter of which has also been shown to contribute to head movement representation in visual cortex . Together, these signals would still give strong information about the position and movement of the head. As requested, we performed new GLM analyses on visual cortex recordings (as well as all other regions) and compared stability across light-light and light-dark sessions. Visual cortex was the least stable of the 4 areas (shown in , and we note these findings in lines 188-195 of the Results:
"We next compared the stability of behavioral representation across light and dark conditions, and found that GLM-selected features in visual regions were more stable across light sessions than between light and dark sessions . Planar body motion, allocentric head movement, and allo-and egocentric head posture were the least stable covariates, but were still maintained in darkness, presumably by vestibular, proprioceptive or efference copy signals . Tuning in auditory cortex, on the other hand, was stable across light and dark conditions , as was the case in somatosensory and motor cortices (p > 0.05, two sample Z-test for proportions; not shown)." m2. Another result that seemed deserving of more attention was the perturbation experiment in which a 15 g weight was added to the head mount, with results reported in Extended Data and summarized in a few lines (106-113) in the results.
We have included additional context and discussion regarding the 15 g head weight experiments, both in the Results (lines 142-144) and Discussion, lines 252-259:
"The robust encoding of head features allowed us to test whether M1 neurons primarily encoded spatial kinematics or muscle exertion controlling the head, a question classically debated in the context of hand and arm movements in primates . The addition of the head weight had only minor effects on tuning curves and statistical models of head-related covariates, indicating that spiking activity was linked more closely with spatial kinematics or specific postures (see , rather than the generation of force. Since the animals were freely moving, signals related to sensory feedback, planning or dynamic pattern generation may have also contributed and cannot be ruled out ." m3. Lines 44 -46: "The animals combined ethogram consisted of 44 independent modular actions ... whose sequence order was best described by a set of transition probabilities" - Is it shown that transition probabilities are the best description? My understanding of these ethograms is that the behavioral state dynamics tend to be highly non-Markovian, suggesting that transition probabilities alone are insufficient. See, e.g., Alba, Berman, Bialek and Shaevitz, arxiv, 2020.
We did not test whether transition probabilities were in fact the best description for the behavioral data, and indeed behavioral transitions in rats have been shown to be highly non-Markovian at time scales of seconds to minutes (e.g. Marshall et. al 2021). We revised the description of the action transitions to state that the estimated transition probabilities were highly stable across recording sessions in different lighting conditions (Pearson's R of 0.968, p = 0.0006) (Results, lines 51-55):
"The animals' combined ethogram consisted of 44 independent modular actions and [fig_ref] Figure 4: InFig [/fig_ref] comprised of unique composites of rudimentary pose and movement features , which followed characteristic transition probabilities that were conserved across light and dark recording conditions [fig_ref] Figure 4: InFig [/fig_ref]." m4. : The top panel is relatively clear, as is the behavior map and example postures in the bottom panel. I do not know how to interpret actions 1 to 44 without a map, and I don't understand what new information the map provides that bar heights do not. Could you pick one or the other? Further, it is a non-intuitive choice to designate dark bars for accuracy of decoding with the prior and light bars for accuracy of decoding using neural activity, while in the maps of decoding, it appears to be light for low accuracy and dark for high accuracy.
We have revised substantially. We removed the circular bar plot and tSNE plots showing decoder accuracy, previously in . The decoder results are now shown in a stand-alone sub-panel in , with chance levels in the bar charts shown in dark grey. The tSNE plots with color-coded decoder accuracy were replaced by encoding results to complement the decoding results, and the distributions of encoded actions, which did not differ between areas, are now shown in . m5. A confusion matrix would be good to see in the extended data to understand how distinguishable similar behaviors are.
We have added confusion matrices for each brain region in a new and refer to them in the Results (lines 76-77). m6.Minor grammar: "of each covariate across the set of models it was included in" -> "of each covariate across the set of models in which it was included"
We have corrected the text in the Figure legend (now [fig_ref] Figure 3: Figure 3: [/fig_ref]. m7. [fig_ref] Figure 3: Figure 3: [/fig_ref] : I do not understand why the average lines in the A1 plots in (B) are blue and yellow; do these correspond to blue and yellow clusters in the UMAP plots? I don't think so, as those seem more likely to correspond to the blue and yellow in panel (C), where lightmodulation is demonstrated. Related, why is there no corresponding UMAP plot for sound modulation?
We have revised and simplified [fig_ref] Figure 3: Figure 3: [/fig_ref] (now , including changing the yellow and blue color schemes from before to red (luminance modulated) and blue (sound modulated) in and b. The color schemes match the luminance and sound modulation data shown on the Neuropixels schematic in 5c.
The UMAP plots, now in , are color-coded according to the same color scheme in the rest of the paper (auditory units in cyan, visual in pink/red). UMAP plots were generated to demonstrate light modulation effects because those data were not collected on a trial-to-trial basis-the room lights were on or off for the whole session. Sound modulation effects, on the other hand, could be shown clearly on a trial-by-trial basis in raw spike rasters each time the sound was presented, so UMAP plots were not necessary.
m8.The gradients in (D) are intriguing, but also seem to move in and out of different layers in different areas (as portrayed in Extended Data . This makes it difficult to determine whether it is a proximal-distal gradient, or a laminar gradient. This is an interesting possibility but is difficult to parse purely by layer, since layers were not sampled evenly across visual and auditory regions, and ~3x more neurons were recorded in visual than auditory cortices. We did, however, calculate the proportion of sound-or luminance-modulated neurons by layer within each region. This showed that both sound and luminance modulation were highest in layer 4 of auditory and visual cortices, respectively, with slightly lower values in superficial and deep layers, where recorded. We have added these results to the legend of :
"Soundand light-modulation within auditory and visual cortices, respectively, were highest in layer 4 (L4). In A1, 48% of L4 neurons had significant sound modulation indices (SMIs), followed by layers 5 and 6, with 26% and 25%, respectively. In V1, 30% of layer 4 neurons had significant luminance modulation indices (LMIs); layer 5 had 27%, layer 6 had 24%, and layer 2/3 had 23%." m9. [fig_ref] Figure 3: Figure 3: [/fig_ref] is nice: it appears there is a distributed representation of sensory inputs and behavior across auditory and visual areas, and that this is more or less similar between FS and RS units. It would be helpful to have unit counts at each branch.
We have added the proportions of FS and RS units at each level in the dendrograms in the . m10. [fig_ref] Figure 4: InFig [/fig_ref] : In [fig_ref] Figure 4: InFig [/fig_ref] , use the same x limits on all subplots. It would also be useful to know the number of possible connections.
We respectfully oppose the suggestion to set the same x-limits in this case since doing so potentially helps obscure true effects that we try to present, and in datasets where the expected number of synaptic connections is different. Given that different areas and connection subtypes have different numbers of total pairs, we argue that the number of pairs itself is not as relevant as the presentation of whether that number exceeds what is expected by chancefor different areas/connection subtypesthis will be a different value.
As for the number of possible connections, we have included in the Methods that a total of 557,620 possible connections were tested in the right hemisphere (visual and auditory cortices), and 606,301 possible connections were tested in the left hemisphere (somatosensory and motor areas) (Methods, lines 697-699). We again point out that our inclusion criteria were strict (p<0.0001 for CCG values in two consecutive bins within the ±1.6-4 ms window of the spike, and no threshold passing in the ±1.6-0 ms range). The proportion of detected connections relative to all possible connections says more about recurrent properties or propensity for strong connections in a network, and little about whether detected connection profiles could be observed by chance.
In [fig_ref] Figure 4: InFig [/fig_ref] D, do not use the same color for FS/RS outline that you use in [fig_ref] Figure 4: InFig [/fig_ref] indicate auditory regions. A1 and A2D are yellow and brown -why not shades of green to link to panel B and C?
We thank the Reviewer for catching this. The color scheme in the figure has been updated so that A1 and A2 are shades of green; FS neurons are outlined in magenta and RS neurons in purple (per the revised color scheme in .
The caption says 'We note that the outlined pair groups are largely separated." for [fig_ref] Figure 4: InFig [/fig_ref] This separation is not clear to me in the UMAP plot at right. More details on the functional subspace would be helpful for interpreting this plot. Extended Data [fig_ref] Figure 3: Figure 3: [/fig_ref] is useful in this regard, but [fig_ref] Figure 4: InFig [/fig_ref] needs to stand on its own.
Due to the difficulty in visually conveying that neurons forming different pair groups were largely distinct neurons (i.e., there were not many neurons that fell in multiple categories), we removed the UMAP embedding from [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref]. All functional subspace analyses are in , where they are explained. We have added text to motivate the functional embedding analyses, and contextualizing text for the results (noted above in relation to point M2). m11. Extended Data [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] All x-axes should be the same The x-label says cross-validated log-likelihood, but the caption says cross-validated rLLRs. The x-axes have been adjusted and the label has been updated, we thank the Reviewer for catching the error.
## Reviewer #3 (remarks to the author): summary:
Mimica*, Tombaz* et al present an impressive and unique study that examines the functional coding properties of neurons in somatosensory, motor, visual, and auditory cortices during natural behavior. By first parsing the natural behavior into actions (e.g. rearing), the authors find that most actions can be decoded in each region. By then parsing the behavior into continuous movements, such as head orientation in egocentric and allocentric coordinates, the authors uncover coding properties specific to each region. E.g. visual cortex encodes allocentric head movement, while motor cortex encodes egocentric movement. The authors then show that neurons in visual and auditory cortices encode visual and auditory information alongside behavioral information, and that there is structure to the putative synaptic connections in visual and auditory regions.
Overall, the data collected are quite impressive, as are the wide range of presented analyses. Understanding neural activity during natural behavior is of high interest to many but an incredibly difficult task. Here the authors have very bravely waded into those waters and come out with some interesting results. However, I felt there were some issues with the manuscript in its current form. First, while in general I can get behind the sharing of results even if their full impact or interpretation isn't entirely clear yet (as every paper is a small part of a bigger story), I felt that some of the results presented here really didn't have quite enough contextualization, interpretation, or motivation. There are a lot of facts in this paper, and in some places it was hard to know what to take away, or what exactly was learned that wasn't clear before. More details on this point are listed below, but overall, I felt this applied most to results in [fig_ref] Figure 3: Figure 3: [/fig_ref]. This issue is also exacerbated by the current short format of the article (which I'm not sure is necessary, although I admit I don't know the character limits here). Further, some of the figures were hard to parse, and some analyses or methodological details were either missing or hard to find. These issues and others are detailed below.
We thank the Reviewer for their thorough and largely positive evaluation of the study. We have taken their concerns into account, in particular by giving additional background and motivating text to link the experiments and findings in the paper. We have also moved data from the Supplementary material into main figures in areas of the paper, in particular to substantiate our finding that all cortical regions encoded posture and movement features in different cortical layers, where sampled. Responses to the concerns and suggestions are listed point-by-point below.
Major comments: As mentioned above, I felt that in a couple places the figures were difficult to parse. While some figure panels are mentioned in other comments, issues pertaining to the rest have been collected into the list below.
M1. -Artistically speaking, the circle looks nice. But it is very hard to visualize the decoding accuracy or compare the size of the bars in different regions in this format. While less exciting, it might be easier on the reader (and better convey the points you make in the text) to show rows of bar plots. In addition, the comparison to the control decoder, in which decoding is based on the prior, is vital, but difficult to see here.
These are helpful suggestions; we agree that, although the original version of was visually appealing, it was difficult to parse. We removed the circular bar plot showing decoder accuracy in and show the decoding results now as a stand-alone sub-panel in . Chance levels in the bar charts are shown in dark gray. The tSNE plots with color-coded decoder accuracy have also been removed, and in their place we show encoding results to complement the decoding results. Further we show the distributions of encoded actions in each overarching area in .
M1b. [fig_ref] Figure 3: Figure 3: [/fig_ref] i. I find the lines in [fig_ref] Figure 3: Figure 3: [/fig_ref] ,c to be visually confusing -I see it is a double axis, but it makes it quite hard to interpret.
We have simplified [fig_ref] Figure 3: Figure 3: [/fig_ref] (now in several ways. In this case, we included additional text in the figure legend explaining the left axis (unit number) and right (colorimetric scale for unit activity relative to max). To make sub-panels 3b and c easier to parse, we moved them to their own Supplementary , enlarged them, and changed the color scheme such that the colors for RS and FS cells (purple and magenta) are more visually distinct from colors indicating sound modulation (blue) or light modulation (red). [fig_ref] Figure 4: InFig [/fig_ref] b are difficult to interpret. Can they be summarized or restructured to highlight the most meaningful relationships between behaviors?
## "fig. 5 | auditory and visual cortices show similar prevalence and overlap of sensory and behavioral signals. a, (top) white noise stimulation paradigm schematic. (bottom)trial averaged activity for single cells (arranged by unit number, left axis) and ensemble averaged activity (overlaid trace) of all significantly suppressed (left) and sound-excited (right) single units in auditory cortices
## M1c. extended data figuresi. extended
Great suggestion. We re-grouped the actions in those figures (now Supplementary Figs. 5a and 5b) based on features which were most conspicuous when labeling the actions from video (e.g. if the animals were sitting still, walking, head up or turned, etc.). The features by which the actions were grouped are written in bold on the right of the groupings in the figure.
M1dii. Extended Data 10 -this figure has a ton of information in it, but with the number of panels and subpanels, it's difficult to pull out the relevant facts. In some cases, I was not sure what was being plotted -e.g., what the 'information rate' in panel e is being computed from (the encoding model?). I would suggest substantially simplifying this figure.
We have simplified Extended Data figure 10 (now Supplementary [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] to emphasize key points. It now consists of schematics and sample data in the top panel (a); the middle panel (b) shows stable tuning curves overlaid from each condition; (c) compares the properties of the tuning curves and spiking properties of the cells encoding head features; and in (d) the statistical results of compared features. We have added text in the figure legend to more clearly state that the comparisons were based on the tuning curves of the cells themselves and added that the information rate was calculated in bits / spike for each .
The calculation of baseline stability was removed, but is still described in the Methods, and the GLM results were placed in their own .
M2. How were spikes sorted across multiple 'schedules' (the group of 4 sessions)? It's my understanding that the Neuropixels shank was implanted and in the same place across these schedules, but were the spikes sorted together? Or were cells sorted separately in each schedule? Unless I have missed something, it seems this latter method would lead to double-counting of the same cells.
To clarify: each animal had a total of 8 recording sessions. In the first 4 sessions we recorded from bank0 (distal part of the probe), then the animals received a break, and we continued with the next 4 sessions from bank 1 (more proximal on the probe). The 4 recording sessions that were run consecutively were combined and processed together in kilosort, which meant that cell IDs were maintained across the 4 recordings and prevented double-counting. This also allowed us to track whether the spiking profiles (e.g. amplitude) changed over the 4-session recording period.
M3. In Line 51, the authors claim that each action is encoded by some population of neurons in each region. However, in Extended Data [fig_ref] Figure 3: Figure 3: [/fig_ref] , it looks like there are quite a few actions that are not encoded by neurons in any region. (To me, this figure actually highlights that only a subset of actions seem to be encoded for in each region.) Perhaps this can be directly quantified somewhere, either in the main text or the figure legend, to make it clearer that there are neurons encoding these actions.
The Reviewer is indeed correct that several actions were not encoded in the examples in Extended Data [fig_ref] Figure 3: Figure 3: [/fig_ref] (now [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] , which were from individual animals and single sessions. To the Reviewer's point specifically: now includes a complete summary of the fraction of cells encoding each action in each cortical region, and the most commonly encoded actions in each area are noted in the figure legend. It also highlights that the distributions of encoded actions were largely similar across cortical regions, and the Equivalence test in Suppl. [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] confirms this statistically. To address this point more generally, the revised manuscript gives more attention to action encoding results, including showing the overall prevalence of action encoding and inclusion of examples of action encoding by neurons in each region in Suppl. [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref]. We hope the added data make it clearer that nearly all actions were encoded, and to differing degrees, by subsets of neurons in each region. -the number of neurons will affect decoding accuracy (e.g. as in [fig_ref] Figure 3: Figure 3: [/fig_ref]. However, I could not find the average number of simultaneously recorded cells (or how much this number varied across sessions), or whether this was controlled for when presenting decoding accuracy across regions. Also, given that motor regions encode egocentric movement the strongest, did the authors also observe higher decoding accuracy for actions here as well?
## M4. related to the decoding result in
This is a good point, and we now include in the legend for that decoding analyses were restricted to 60 cells per session to match sampling across regions. We also added confusion matrices for each area in a new Supplementary , and a new showing how decoding accuracy in each region increased with the number of simultaneously recorded cells. The best decoder performance was achieved when the highest numbers of cells were recorded simultaneously in motor cortex. Visual cortex had the lowest rates of decoding accuracy for a given number of cells, but all regions were well above chance levels. These points are noted in the Results, lines 77-79.
M5. Regarding the model fits -20 minute sessions are quite short for fitting spike trains to natural behavior data, and the resulting pseudo-R2 values are quite low with medians ~ 0.02. While this is to be expected to some degree for natural behavior no matter how long the recordings are, the authors may be able to provide additional support for the encoding properties they present if they are indeed tracking cells across schedules (or even across light1 and light2 sessions). In this case, are the encoding properties (variables encoded and their tuning) generally similar? If not, the data presented in Extended Data is relevant and convincing; however, there are only 4 example cells shown. How often are the tuning curves similar across session halves? This is a fair point of concern, and was raised by other reviewers as well. There is no definition of what a "high" or "low" pseudo-R^2 is, so we took a suggestion from Reviewer to create a "Peer Model" GLM, in which spiking activity from simultaneously recorded cells (separated by >5 recording sites on the probe) was used as a predicted feature (as done by Stringer et al. (2019)). We then compared the proportions of cells tuned to at least one behavioral covariate in our original GLM (referred to here as the "Covariates Model") against the subset of cells with "good" pseudo-R^2 values, i.e. larger than their corresponding Peer Models. We found that the defining regional differences in encoding properties were upheld when only considering the cells with "good" pseudo-R^2 values. That is, head movement, head posture and self-motion still dominated in visual cortices; allo-and egocentric head posture still dominate in auditory cortex; motor regions still encoded the trunk and head, and somatosensory cortex was still dominated by trunk representation:
We also found that the mean rLLRs (importance) of each feature in the polar plots were also maintained when comparing the full dataset against models with "good" pseudo-R^2 values:
To address the question of tuning curve stability, we calculated Pearson's R-values across even and odd minutes using tuning curves for whichever feature had the highest mutual information for that cell's spiking activity. This analysis included all cells, regardless of whether they appeared to show genuine tuning or not. We compared even-odd minute R-values against the distributions of values produced with shuffled data, and counted cells as "stable" if they exceed the 95th percentile of the shuffled distribution (this explanation is included in the Methods, lines 502-506). Tuning curves from 30% of cells in visual cortices exceeded this criterion; 37% in auditory cortex; 43% in somatosensory, and 66% in motor cortex. We include the fraction of cells with stable tuning curves in each cortical area in the legend of .
We wish to point out that the criteria for a cell to have a "stable" tuning curve were more strict than for model selection by the GLM. This was due to practical differences in the way the analyses were constructed, including (i) tuning curves had 36 bins, whereas 1D features in the GLM consisted of 15 bins, (ii) we only considered one 1D covariate to test if a cell was stable, whereas the GLM considered all covariates, (iii) the GLM used 10-fold cross validation (train on 90%, test on 10%), whereas tuning curve stability was across session halves (testing 50% against 50% of sampled data).
M6. Regarding the result regarding the addition of a weight -in the main text, it seems that the addition of the weight did not change the encoding properties of motor cortex cells. However, in extended data fig 10, it is difficult to see the extent to which this is the case. In particular, while the proportion of cells encoding different features is similar across the population in panel G, it is unclear whether other aspects of the encoding, such as the weights or log-likelihoods, are similar. Perhaps a more straightforward comparison would be to train on a light1 session, test the model on the weighted and light2 session, and then compare the resulting model fits. Further, one could directly compare the glm tuning for models fit with the same variables in each session. For what it is worth, I think this potentially a very interesting result, since this relates to ongoing arguments of encoding kinematics versus dynamics in motor cortex (which could also be referenced more directly in the text).
We are glad the Reviewer found the result of the weight vs. no weight experiments interesting. As noted above, we included examples of tuning curves from cells encoding postural features of the head to better convey the minor effects the added weight had on the cells' tuning [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref]. Further, we performed new GLM analyses in which models were trained on data from the "light 1" session and tested on covariates in the "weight" and "light 2" sessions. The proportions of cells encoding ego-and allocentric head features were similar across weight and weightfree conditions, differing from 0.1% (for egocentric head posture) to 3.6% (for allocentric head posture)--shown in , and noted in lines 150-153 of the Results. We also added background and discussion on these experiments, and the debates which motivated them in the Discussion, lines 252-259. [fig_ref] Figure 3: Figure 3: [/fig_ref] -currently, much more attention is paid to the overlap in sensory and behavioral variables in visual and auditory cortices than motor and somatosensory cortex. While I somewhat understand the logic for looking at sensory variables in sensory cortex, this has the unfortunate effect that the main results seem to be that visual cortex responds to visual stimuli, and auditory cortex responds to auditory stimuli. It's unclear what we've learned from this result, given that this would be expected. However, given that authors recorded from somatosensory and motor cortex during these trials as well, it might be interesting to include all four regions in these analyses and discuss the differences and similarities. For example, does the behavioral encoding in motor or somatosensory neurons depend on light/dark conditions? Are neurons in these regions modulated by luminance/sound and/or behavior? [fig_ref] Figure 3: Figure 3: [/fig_ref] in the original manuscript in the revision) has been re-organized to emphasize the main result, that sensory and behavioral modulation overlap to similar extents in visual and auditory cortices, and in RS and FS neurons. The analyses demonstrating the specificity of visual tuning in visual cortex, and auditory sensitivity in auditory cortex, have been moved to a new . Characterization of the overlap of sensory and behavioral tuning in is followed by analyses investigating how behavioral signals are integrated in each area in [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref].
## M7. regarding the results in
To address the question of sensory modulation (at least for luminance) in all cortical regions, we performed new GLM analyses to quantify behavioral tuning across subsequent light-light and light-dark recordings in each cortical region. Neither motor, somatosensory nor auditory cortices differed in stability across the recording conditions, but tuning in visual cortex was significantly less stable across light-dark than light-light conditions (p-values calculated with a two sample Z-test for proportions). These findings are reported on lines 188-195 of the Results; the GLM results for auditory and visual cortices are shown in . We did not have the recordings to test whether other cortical regions were sound modulated. does not include p-values for the changes in encoding properties of neurons in the light and dark conditions. Do you think that the results reflect an actual change in the encoding of movement features, or that the actual features that the neurons are responding to are visual in nature and are only correlated with movement in the well-lit condition?
## M8. extended data
This is an excellent question, and (as stated above) we generated p-values for more rigorous comparisons of light-dark tuning changes in visual cortex. As a methodological note, we sought to simplify the question of whether apparentbehavioral tuning reflected the movement of the animal vs. movement of the visual scene by using recordings in darkness. Previous work by and reported that luminance conditions dramatically modify the activity of visual cortical neurons during movement, and an even more recent paper by showed multiplicative encoding of head and eye position with visual features in the environment. Since quantifying interactions of visual scene movement and head movement would require eye tracking and gaze reconstruction, which we did not perform, we cannot speculate about this potential (and intriguing) contribution to instability across light and dark conditions in our experiments.
M9. There is a thread regarding FS/RS cells that I am not following. I think that part of the confusion is that FS/RS breakdowns in [fig_ref] Figure 3: Figure 3: [/fig_ref] are not mentioned in the text, and then FS/RS properties relating to [fig_ref] Figure 4: InFig [/fig_ref] are mentioned in the discussion, but not mentioned in [fig_ref] Figure 4: InFig [/fig_ref] or the related main text in the results (unless I missed something). Further, it's not clear to me what the main results are regarding FS/RS cells. For example, in [fig_ref] Figure 3: Figure 3: [/fig_ref] , both types of cells seem to have very similar properties. What is the result that the authors are trying to highlight here?
We have revised the text and former [fig_ref] Figure 3: Figure 3: [/fig_ref] (now to more clearly convey the main results: that the amount of overlap between sensory and behavioral signals was uniform in auditory and visual cortices, and occurred to similar relative extents in FS and RS neurons in each area. To this end, we also include the fractions of FS and RS neurons encoding behavior or sensory features in the dendrogram in .
Whereas the point of was to highlight the striking uniformity of sensory and behavioral overlap in visual and auditory regions, the point of [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] was to uncover how behavioral signals might be utilized in these regions by considering functional connectivity (elaborated below). We also clarify in the text of the Results that inhibitory connections in the functional analyses (pertaining to [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] were driven by FS neurons, and that excitatory connections were driven by RS neurons.
M10. In [fig_ref] Figure 4: InFig [/fig_ref] , the take home message to me is not clear. I think perhaps part of the issue is that it is unclear what should be expected here, or how the results presented specifically update our knowledge of functional connectivity in visual and auditory cortex. I think additional text is needed to motivate and contextualize these results. There has been prior work on functional connectivity in at least visual cortex (Ko*, Hofer* et al 2011), which can be discussed here. Further, it is unclear what the functional space embeddings are supposed to convey.
Having explored the type and extent of behavioral modulations in three separate sensory cortices in the first two figures of the original manuscript, the final two figures (now [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] are an attempt to connect those findings to the sensory functions of these areas, to motivate future investigations into how behavioral modulation assists sensory processing. As opposed to , where the focus is on determining tuning overlap, in [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] we want to focus on communication between cells, as tuning overlap per se does not characterize whether, and if so, how information from behaviorally relevant signals flow within each network. Prior work on functional connectivity in the primary visual cortex (as in Ko*, Hofer* et al 2011) indeed established a relationship between connection probability and tuning similarity, where units with preferences for orthogonal stimuli had lower connection probabilities. In much the same vein, we wanted to learn whether functionally similar units exhibit greater likelihoods of forming synaptic connections. In our case, however, the functional subtypes could not be categorized based on firing rate differences along a variable that was essentially one-dimensional. Since we had numerous functional categories, each exhibiting directionality, it was not possible to visualize this plainly. Forming the functional space embeddings (via dimensionality reduction of filters emanating from our modeling analyses) was an attempt to visualize a potential relationship between connection probability and tuning similarity, a relationship which we ultimately show does not exist statistically, in the lower part of (no correlation between synaptic strength and functional distance).
The revised text motivates the analysis with a stronger introduction and reference to the earlier work on functional connectivity in visual cortex , and refers to the lack of relation between functional similarity and synaptic strength at the end of the section. We felt this was important to report since it suggests that communication between functionally disparate cells is important, possibly even more so when the content (behavioral vs. sensory signals) is different.
a. In addition, it is not immediately clear to me why the focus is on auditory and visual cortex. The results in Extended Data 13 F-G could be compared to those in [fig_ref] Figure 4: InFig [/fig_ref] C-D. For example, does the increased precedence of Po->Mo excitatory connections in visual and somatosensory cortices as opposed to auditory and motor cortices align with a model similar to those described in the caption of [fig_ref] Figure 4: InFig [/fig_ref] ?
Our focus on primarily reporting results of auditory and visual areas in the main figure, versus those of other cortices, stemmed from our ability to quantify some aspects of sensory processing in these regions, which we believe would potentially be interesting to a larger audience, but also the desire to not over-complicate what is already a result-heavy figure. As for the similarities in excitatory Po->Mo connections in visual and somatosensory corticesthough the functions of such connections have not been tested empirically-we speculate that they might serve to prime sensory networks for impending movement. In visual cortex, as we noted, a cell encoding maximally rightward head posture would excite a cell encoding leftward head movement, which may prime circuits to anticipate self-generated changes in the visual scene. In somatosensory cortex, such connections could anticipate whether a given change in posture or stance was generated by the individual or, if not, signal postural instability, for example when an animal encounters an unexpected obstacle or change in footing . We note this possibility for somatosensory cortex in the Discussion, lines 298-300.
Minor comments: m1. There were a few tiny typos: a. Line 442 -should be 'represent' b. Line 450 -'features' is misspelled c. Line 454 --there is an extra parentheses d. Line 65 -should be 'nearly all'
We thank the Reviewer for catching these typos; all have been corrected. m2. Line 51 -it is not completely clear what the numbers in the parentheses refer to. Is this the number of neurons that encode at least behavior in each region? I think the wording should be changed, or text added, so this is more transparent.
We have modified the text to more clearly indicate which numbers indicate encoding rates in which cortical region (lines 64-66):
"We found stable encoding of nearly every considered action by individual neurons in each cortical region (51% of neurons in visual cortex, 55% in auditory, 58% in motor and 56% somatosensory regions; , with most cells responding to multiple actions, and fewer to single actions [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] ; Supplementary Videos 1-6; Methods)" m3. To help compare the number of simultaneously encoded variables across regions, the axis of the bar plot in should be the same as the other panels.
The axis of the bar plot in revised [fig_ref] Figure 3: Figure 3: [/fig_ref] has been adjusted to match those in 3a-c.
m4. Lines 94-98, Lines 118-122 -the authors report a gradient in encoded features, but it is unclear to me whether the observed gradient is surprising or fits with the current literature. I think a sentence or two contextualizing these results would help.
We have added contextualizing sentences in the Results and Discussion for the topography results. In lines 122-124 we point out that previous studies characterizing head-motion tuning in visual cortex focused only on V1, so it had not been tested whether gradients exist for behavioral features across visual and neighboring regions. For somatosensory and motor cortices, classical microstimulation experiments in anesthetized rats showed an anterior-toposterior topography of the head and back from motor to somatosensory cortex. Posture and movement tuning in our study followed a consistent pattern. We also acknowledge this in the Discussion, lines 270-272. m5. Either the legend or the panel in [fig_ref] Figure 3: Figure 3: [/fig_ref] not correctthe legend says that the FS spikes are in green, but I do not see any no green spikes.
We have adjusted the wording (and color coding, for clarity) in the figure legend for what was shown in [fig_ref] Figure 3: Figure 3: [/fig_ref] (now in :
## "supplementary fig. 19: modality specificity and decoding using auditory and visual cortical units. a, (top) spike rasters (grey) and peri-event time histograms (peths) of sound-suppressed (left) and sound-excited (right) auditory cortical single units (purple line and shading depict trial averaged firing rate ±3 sem for regular spiking (rs) units; and the same for fast spiking (fs) units is shown in magenta;
sound stimulation in grey shading). (Bottom) Non-linear embedding of A1 population vector activity from light and dark recording sessions for an example rat (#26525); (right) same but for visual units."
2. At several points in the manuscript there is speculation regarding the interpretation of results and their implications. While I suspect that these statements were likely intended to contextualize the results in response to the first round of comments, some extend beyond the scope of the study and should be reworked. a. Line 28-29: "... in ways that could facilitate modality-specific functions, such as motion processing in visual cortices or sound localization in auditory cortex." Notably, as differences in sensory input cannot be ruled out to mediate the correlation of movement and postural variables with neural activity in visual and auditory cortex, it seems premature to conclude that the integration of these variables facilitates the proposed functions, particularly in the case of visual cortex. This point is supported by the authors' observation that the tuning of neurons in visual cortex changes between light and dark conditions. Walking back some of the related statements, or being more upfront with the caveats in the experiments, will help address this issue.
As the Reviewer points out, those interpretations of the results intended to offer context in response to the previous round of review. We have stepped back in the current revision on naming modality-specific perceptual processes which we did not test (e.g. visual motion processing, sound localization) in the Abstract and Introduction. We still feel it is important to explore those possibilities in the Discussion, however, where there is space to weigh both interpretations and the limitations of the experiments.
The 2nd to last sentence of the Abstract was changed to:
"The tuning properties of synaptically coupled cells also exhibited connection patterns suggestive of area-specific uses of pose and movement signals, particularly in visual and auditory regions."
Also, as noted above, the last paragraph of the Introduction was changed and no longer speculates about which sensory processes the observed tuning could underlie. We also added the following statement in last paragraph of the Discussion:
The widespread expression of such features speaks to the computational demand and importance of keeping sensory systems, and the individual, oriented while moving through complex environments, and may therefore reflect a general property of sensory coding in cortex.
## Although we did not test how behavioral and sensory information are combined here, our observations could guide future work investigating how pose and movement signals inform perceptual functions like visual self-motion subtraction or sound localization.
Pinpointing exactly how this integration happens at the circuit level may prove challenging in freely-behaving subjects, requiring sufficiently resolved techniques, such as miniature 2-photon imaging and holographic stimulation , to identify, then manipulate behaviorally-classified neurons in vivo.
b. Lines 313-319: It is hard to tell whether the authors are suggesting that neurons in auditory cortex encoding posture and movement variables influence the detection and processing of the ILD (a function typically attributed to the midbrain), or simply that one must incorporate posture and movement to localize in 3D space using ILD. Regardless, while I agree that the literature on the influence of pose on auditory localization in rats is sparse, work implicating neural activity in auditory cortex to sound localization in other species could be cited to support the authors' argument. Our interpretation was more in line with the latter alternative, that roll and pitch modulation may facilitate 3D localization of sound sources. We added support for this interpretation from observations made in barn owls and Tenglmalms owls; specifically, the ear openings in the heads of these species are offset vertically relative to each other, which has been shown to facilitate sound source localization using vertical ILDs . We noted that it is possible that the mammalian brain also uses roll information in particular to facilitate localization of sound sources in elevation. We further added that auditory cortical neurons in several species of mammal signal source locations over a broad range of spatial locations , so having added modulation by head position or movement within auditory cortex could facilitate locating sound sources relative to the head of the individual. We note that further work is required to test this idea empirically.
The revised passage in the Discussion now reads:
The roll and pitch of the head might be heavily represented because these features would strongly influence the detection of interaural loudness differences (ILDs) in a 3D environment. If a rat's head is perfectly level, for example, any ILD corresponds to a change in horizontal location, but the animal is blind to changes in vertical localization. Once the head rolls, this changes, and strong roll-and pitch-modulation would facilitate detection of ILDs in vertical space, allowing for 3D sampling . Further work will be needed to bear out if this is the case in rodents, but this interpretation is supported by observations in certain species of owl that have vertically offset ear openings in the head, which has been shown to facilitate sound localization using vertical ILDs
## . although additional recordings are also required to establish how postural signals contribute to sound localization at the level of cortex, auditory cortical neurons in mammals encode sound source locations uniformly and over broad ranges of spatial locations (middlebrooks et al., 1994; middlebrooks), 2015, so it is intuitive that added modulation by the position and motion of the head would help discriminate the location and direction of a sound relative to the individual (wallach, 1940).
c. Line 255-257: While this statement might have been included in response to a previous comment of mine, I believe this statement should actually be adjusted to focus on the finding that kinematic and postural encoding was not affected by the inclusion of a weight rather than suggesting that neural activity more closely resembles kinematic variables. There is a brief literature explaining why regression analyses like those performed in this manuscript are typically insufficient to distinguish between cortical control of force or end-point kinematics. We thank the Reviewer for clarifying this point, and for sharing these useful papers on the limitations of regression analyses in modeling highly interrelated features of motor behavior. We have adjusted the text away from a "kinematics vs. muscle force" distinction, and focus on the observation that kinematic & postural tuning were largely unaffected by the added head weight.
In the Results: Our recordings also uncovered dense representation of head kinematics, particularly in the deep layers of motor cortex [fig_ref] Figure 3: Figure 3: [/fig_ref]
## , which presented the opportunity to determine if neural encoding of spatial kinematics [46] changed with the added load of a 15 g weight on the head [47].
In the Discussion:
The robust encoding of head features allowed us to test whether the addition of weight on the head affected how neurons encoded posture or kinematics of the head, a question approached previously in the context of hand and arm movements in primates . We found that the added weight had only minor effects on the tuning properties of M1 neurons [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] , but acknowledge that signals related to sensory feedback, planning or dynamic pattern generation could also have contributed since the animals were moving freely .
3. I had some difficulty understanding the take-home message in [fig_ref] Figure 4: InFig [/fig_ref] and how it relates to the messages relayed in the main text. Overall, this figure seems to have been directly ported over from the Supplemental Figures, but with not quite enough rearrangement of the text to accommodate it. Some additional text that describes and contextualizes this figure would be extremely appreciated -the specific issues that I ran into are detailed below. a. First, [fig_ref] Figure 4: InFig [/fig_ref] is barely mentioned in the text, which makes its inclusion as a main figure confusing. [fig_ref] Figure 4: InFig [/fig_ref] introduced separately, which added to this confusion. Overall, adding more context and pointers to this figure will help. b. [fig_ref] Figure 4: InFig [/fig_ref] introduced in regards to head motion in visual cortex, but half of the panel is about planar body motion. If this distinction is relevant, it should be discussed in the text. c. In [fig_ref] Figure 4: InFig [/fig_ref] , it is introduced with the general finding that egocentric head position and movement is encoded in motor cortex while back-related variables are encoded in somatosensory cortex. While this is clear from the middle-left panel of 4b, it's not clear what the other panels of 4b are meant to indicate. d. More generally, the pie charts denoting the proportion of classified units in 4a or 4b are not referenced as far as I can tell. Do the authors mean to make a point about the gradient of classified units using this data? It seems like there could also be a gradient in the percentage of classified units in the motor cortex. Do the authors want to make a point about this? e. Is there a main message we should draw from the far-right subplots that is distinct from the middle-left subplots? We have rearranged the text to make a subsection for [fig_ref] Figure 4: InFig [/fig_ref] like the other figures in the paper, and broke the figure into 2 further sub-panels (a-d). We hope this and the added text in the new section address the questions raised above:
## Topographical mapping of behavioral features across cortical regions
Up to this point in the study, we found that discrete, naturalistic behaviors were represented throughout the cortical areas recorded, and were composed of simpler posture and movement primitives whose expression appeared to vary regionally. We therefore next sought to establish in the next series of experiments: (i) how the neural coding of pose and movement was organized within and between cortical regions, (ii) the extent to which pose and movement signals were integrated with sensory inputs in sensory cortices, and (iii) how pose and movement signaling might be utilized in the different areas.
In regard to the first question, we specifically asked whether region-specific differences revealed by the GLM analysis emerged abruptly between areas, or followed continuous topographies that spanned cortical boundaries. The first covariates we considered were allocentric head posture and movement in visual and auditory areas, since they were represented most prominently, and had previously only been studied within V1 . This revealed a graded increase in allocentric head posture coding that progressed laterally from V1 to V2L and peaked in A2D, (Xi 2 (7) , as well as a peak in allocentric head movement tuning nearby in V2L (Xi 2 (7)=13.09, p=.04; [fig_ref] Figure 4: InFig [/fig_ref] and Methods; data from individual animals shown in 4b). The representation of planar body motion features (e.g. self-motion and turning direction) also increased laterally across V1 and reached a maximum in deeper cortical layers at the border of V1 and V2L (Xi 2 (7)=18.2, p=.006) [fig_ref] Figure 3: Figure 3: [/fig_ref]. Together, these covariates largely accounted for the apex of coding cells around the visual-auditory cortical border [fig_ref] Figure 4: InFig [/fig_ref] , since the lesser represented features (egocentric head pose, movement, back tuning) were not organized topographically (p > .05 Xi 2 t; not shown). Across S1 and M1, on the other hand, we found continual gradients for egocentric head features and the back [fig_ref] Figure 4: InFig [/fig_ref]. Specifically, neurons encoding egocentric head posture and movement were more frequent in anterior than posterior M1 or in S1 (posture: Xi 2 (7)=37.7, p=1.3e-6; movement: Xi 2 (7)=106.9, p=8.8e-21), whereas back representations dominated in posterior motor areas and S1HL (posture: Xi 2 (7)=18.8, p=.004; movement: Xi 2 (7)=41.02, p=2.9e-7) [fig_ref] Figure 4: InFig [/fig_ref]. Features related to allocentric coding of the head and planar body motion were not organized topographically (p > .05 Xi 2 test; not shown), and the total fraction of classified cells was higher in anterior locations [fig_ref] Figure 4: InFig [/fig_ref] , peaking in M1.
The legend for [fig_ref] Figure 4: InFig [/fig_ref] has also been updated accordingly: but for cells classified along the length of S1-M1. d, Same as b, but using the probe view along S1 and M1 for each animal.
# Discussion
End of 3rd paragraph: This interpretation also fits the somatotopic organization of the trunk and limbs in rats , as well as the anatomical gradients we observed for trunk and head features, and the larger overall fraction of classified cells at anterior locations in motor cortex [fig_ref] Figure 4: InFig [/fig_ref].
4. Unfortunately, I still had difficulty with [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref]. While I agree that, in general, examining whether functionally-connected cells have similar function properties is interesting, the results here seem to be a bit unclear or speculative. Perhaps the most feasible way to address this issue is to add more text describing what these results mean and what hypotheses they support or falsify, and to clarify some aspects of this analysis. a. First, there is little explanation for why inhibitory (and not excitatory) Mo->Po and Mo->Sm connections may be more likely in auditory cortex. It is similarly unclear why excitatory Po->Mo and inhibitory Mo->Po connections may be more likely in visual cortex. Why might the system be organized specifically like this as opposed to any of the multitude of other ways one could combine motor and sensory information in these regions? Are there specific theories or previously-proposed mechanisms that could be supported or falsified? Perhaps if the authors provided some example potential circuit motifs, or further interpreted the ones found in light of what is known about circuit motifs in these regions, that would help. Inhibitory connections in auditory cortices (both Mo->Po and Mo->Sm) appeared more prevalent than would be expected by chance, sharply contrasting those observed for excitatory connections. The latter of these two sets (Mo->Sm) was likely the same phenomenon observed in prior studies in head-fixed animals , positing a role for inhibition in controlling the effects of self-generated sounds on audition (an interpretation previously noted in the Discussion, lines 307-310). Feedforward inhibition plays a known role in gain control (e.g., , and might have an upper hand over feedforward excitation in the auditory system, which is responsible for solving challenging problems like sound detection, attribution and localization. We also provided a hypothesis for why movement-tuned neurons might be inhibiting posture-tuned neurons (Mo->Po) and relate this to sound localization in the Discussion (lines 314-318), as head posture informs the auditory system about the position of its sensors relative to the environment.
As for visual cortices, the emergent picture is more complex, given that both excitatory and inhibitory connections are more prevalent than would be expected by chance. We attribute the discrepancy in connectivity patterns between auditory and visual to differences in problem sets these systems are tasked with addressing. The visual system needs to continuously resolve visual flow from head movements , and this problem might require both feedforward excitation and inhibition , with cross-talk between movement and posturemodulated cells. Future experiments with eye tracking could offer an additional layer of behavioral resolution necessary to tease apart this issue in more detail. It is an open question how much of what we are observing in terms of either posture or movement is related to eye kinematics.
Since the dashed boxes in [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] highlight connections discussed in the Results and shown in the schematic in [fig_ref] 6: Extended Data Fig 3 -it seems that panel c has actions in... [/fig_ref] , we instead tried giving additional direction to readers by numbering both the boxes in 6c and the connections to which they refer in 6d. We note the numbering of the examples in the figure legend:
[fig] Figure 3: Figure 3: [/fig]
[fig] Figure 4: InFig. 4C, use the same x limits on all subplots. It would also be useful to know the number of possible connections.InFig. 4 D,do not use the same color for FS/RS outline that you use inFig 4B toindicate auditory regions. A1 and A2D are yellow and brown -why not shades of green to link to panel B and C? [/fig]
[fig] 6: Extended Data Fig 3 -it seems that panel c has actions in a different configuration than the rest of the panels? This makes it difficult to visually compare with the others. 7. Extended Data Fig 6 -the x-axis should be labeled as 'relative log-likelihood' -just listing 'cross validated log-likelihood' had me confused for a little bit. 8. Extended Data Fig 7 -it would help if all were plotted on the same axis. 9. Extended Data Figure 13 -I believe the colors describing panels A and B in the caption are incorrect. [/fig]
[fig] ": The actions embedded in t-SNE space followed a coarse inherent organization of rudimentary features. These included an increase in running speed progressing from the top to the bottom of the tSNE map, as well as a tendency for the back to be low or hunched at the upper-left of the map, and raised vertically at the lower right of the map (i.e. at the rearing actions, 42 and 44). Head pitch followed a similar coarse diagonal ( [/fig]
|
10.3390/ijms21228603
|
CCBY
|
7696443
|
33203068
|
s2orc_pubmed_articles
|
Underestimated Peripheral Effects Following Pharmacological and Conditional Genetic Microglial Depletion
* * p < 0.01 and * * * p < 0.001.
Supplementary Figure S1: Splenic flow cytometry gating strategy. Supplementary Figure S2: Gating strategy of monocytes in the circulation. Supplementary Figure S3: Lower dose of PLX3397 treatment fails to effectively deplete microglia. (A) Representative flow cytometry plots of CD11b + CD45 low Ly6C -Ly6Gmicroglia of the hemi-brains in C57BL/6NTac mice following consecutive PLX3397 diet (7 and 21 days) with two distinct doses of 75mg/kg and 290mg/kg, respectively. Control mice were treated with normal diet. (B) Total CD11b + CD45 low Ly6C -Ly6Gmicroglial counts (±SEM) of the hemi-brains 7 days following PLX3397 treatment (control, black bars; PLX3397 treatment with the dose of 75mg/kg, blue bars; PLX3397 treatment with the dose of 290mg/kg, red bars). (C) TotalCD11b + CD45 low Ly6C -Ly6Gmicroglial counts (±SEM) of the hemi-brains 21 days following PLX3397 diet with a dose of 75mg/kg (control, black bars; PLX3397 treatment with the dose of 75mg/kg, blue bars; PLX3397 treatment with the dose of 290mg/kg, red bars). Statistical significance is indicated as * p < 0.05,
|
10.1155/2021/5585629
|
CCBY
|
8116163
|
33997021
|
s2orc_pubmed_articles
|
Importance in the Occurrence Distribution of Minimum Oropharyngeal Cross-Sectional Area in the Different Skeletal Patterns Using Cone-Beam Computed Tomography
Purpose. Obstructive sleep apnea is a condition involving repetitive partial or complete collapse of the pharyngeal airway, especially in patient with mandibular hypoplasia. The present study investigated the differences between the volume of the oropharyngeal airway and the minimum axial area in three skeletal patterns through the use of cone-beam computed tomography (CBCT). Materials and Methods. CBCT scans of 147 patients were collected to measure the upper oropharyngeal airway volume (UOV), lower oropharyngeal airway volume (LOV), upper oropharyngeal airway area (UOA), minimum upper oropharyngeal airway area (MUOA), lower oropharyngeal airway area (LOA), minimum lower oropharyngeal airway area (MLOA), anatomical structures (orbitale, Or; porion, Po; pogonion, Pog; hyoid, H; second cervical vertebra, C2; fourth cervical vertebra, C4), and relevant angles. Statistical analysis was performed using analysis of variance and Pearson's test. Results. Compared with patients in Class II, those in Class III and Class I exhibited a significantly anterior position of H and Pog. The vertical positions of H and Pog revealed no significant difference between the three skeletal patterns. Patients in skeletal Class III exhibited significantly larger oropharyngeal area (UOA, MUOA, LOA, MLOA) and oropharyngeal airway (UOV and LOV) than those in skeletal Class II did. The horizontal position of Pog had a moderately significant correlation with UOA (r = 0:471) and MUOA (r = 0:455). Conclusion. Patients in skeletal Class II had significantly smaller oropharyngeal airway areas and volumes than those in Class III did. The minimum oropharyngeal cross-sectional area had a 67% probability of occurrence in the upper oropharyngeal airway among patients in Class I and Class II and a 50% probability of occurrence among patients in Class III.
# Introduction
The pharynx is a conical channel linking the oral and nasal cavity to the esophagus and trachea. It is located at the intersection of the digestive and respiratory tracts and serves as the passage for food and air. Thus, the pharynx plays a crucial role in swallowing and breathing functions [bib_ref] The mouth, palate and pharynx, Berkovitz [/bib_ref]. The pharyngeal airway is divided into three parts, namely, the nasopharynx, oropharynx, and laryngopharynx. The nasopharynx and the oropharynx are demarcated by the soft palate to the rear of the palate, and the oropharynx and laryngopharynx are demarcated by the apex of the epiglottis. The structure of the pharynx affects the volume of the respiratory tract, facial growth pattern, masticatory pattern, and the risk of obstructive sleep apnea. The anatomical structure of the pharyngeal airway space varies according to the diverse growth patterns of the maxilla and mandible.
Face-driven orthodontics and mandibular setback surgery can cause the backward movement of teeth, leading to changes in the pharyngeal airway space [bib_ref] Effects of bodily retraction of mandibular incisors versus mandibular setback surgery on..., Keum [/bib_ref]. Thus, evaluating and measuring the pharyngeal airway space of patients are important before orthodontic treatment and orthognathic surgery. Such precautions can avoid the backward movement of teeth and prevent the mandible from pushing the tongue further backward, which ultimately oppresses and reduces the pharyngeal airway space, causing obstructive sleep apnea in more serious cases. Compared with the nasopharyngeal and laryngopharyngeal airways, the oropharyngeal airway is more likely to be influenced by the surrounding organs. The dimensions of the oropharyngeal airway are mainly affected by the anteriority or posteriority of the mandibular position and tongue size. The front and rim of the tongue are attached to the mandible, and the base of the tongue is linked to the hyoid bone; connections also exist between the tongue and the soft palate as well as the palatoglossus muscles.
In the present study, cone-beam computed tomography (CBCT) was used to explore differences in the volume and minimum cross-sectional area of the individual parts of the oropharyngeal airway in terms of skeletal patterns. In addition, this study involved evaluation of the relationships between the maxilla and mandible; the relationships between sex, age, and the cervical spine; other anatomical structures (including the mandible and hyoid bone positions); the related distances or angles (head and cervical spine positions) that might affect oropharyngeal airway dimensions; and relationships between oropharyngeal airway volume and the minimum cross-sectional area.
# Materials and methods
The CBCT scans (New Tom VGi evo, Imola, Italy) of 147 patients were collected from the dental department of Kaohsiung Medical University Chong-Ho Memorial Hospital. Patients with craniofacial disorders or malformation, those with pharyngeal or laryngeal pathology, and those with craniofacial injuries were excluded from the study. The patient characteristics included age, ANB angle, and body mass index (BMI). For analysis, the patients were divided into three groups according to their skeletal pattern: 30 patients (19 female and 11 male) in Class I (0°≤ ANB ≤ 4°; average age: 25.3 years), 40 patients (28 female and 12 male) in Class II (ANB > 4°; average age: 25.8 years), and 77 patients (44 female and 33 male) in Class III (ANB < 0°; average age: 23.8 years).
CBCT images were imported using the Digital Imaging and Communications in Medicine into Dolphin® 11.0 software (Dolphin Imaging and Management Solutions, Chatsworth, CA, USA). The reference points [fig_ref] Figure 1: Landmarks [/fig_ref] included sella (S), nasion (N), A point (A), B point (B), pogonion (Pog), the most superior and anterior point of the hyoid bone (H), tip at the end of the uvula (U), upper tip at the end of the epiglottis (E), inferoanterior point on the second cervical vertebra (C2), and inferoanterior point on the fourth cervical vertebra (C4). The coordinate system consisted of the X -axis (constructed by drawing a line through nasion 7°up from the SN line) and Y-axis (constructed by drawing a line through the S point perpendicular to the X-axis) [bib_ref] Cephalometrics for orthognathic surgery, Burstone [/bib_ref]. The horizontal and vertical positions of H and Pog were investigated. The related angles were measured and included the head positions (Or-Po-Pog angle, Or-Po-H angle, and Or-Po-C2 angle) and cervical spine positions [fig_ref] Table 2: The measured angles in the skeletal patterns using one-way ANOVA with Tukey's... [/fig_ref].
As shown in [fig_ref] Figure 2: Landmarks [/fig_ref] , the Frankfort horizontal (FH) plane was defined as the plane connecting the right orbitale (Or) and porion (Po) on both sides. The oropharyngeal airway space was divided into the upper oropharyngeal airway (velopharyngeal airway) and lower oropharyngeal airway (glossopharyngeal airway). In the upper oropharyngeal airway, the upper bound of the pharyngeal airway passes through the posterior nasal spine (PNS) and is parallel to the standard horizontal plane, and the lower bound passes through the tip at the end of the uvula and is parallel to the standard horizontal plane. In the lower oropharyngeal airway, the upper bound of the pharyngeal airway passes through the tip at the end of the uvula and is parallel to the standard horizontal plane, and the lower bound passes through the upper tip at the end of the epiglottis and is parallel to the standard horizontal plane.
In, three-dimensional (3D) model of airway space was obtained by Dolphin® 3D software. Airway semiautomatic segmentation (including borders and landmarks) was defined as aforementioned. The airway volume and airway area were automatically calculated by the Dolphin® 3D software. The upper oropharyngeal airway volume (UOV), lower oropharyngeal airway volume (LOV), and total oropharyngeal airway volume (TOV = UOV + LOV) were measured. The oropharyngeal airway areas (axial view) were measured as follows: upper oropharyngeal airway area (UOA: passes through the tip at the end of the uvula), minimum upper oropharyngeal airway area (MUOA: the minimum cross-sectional area of UOV), lower oropharyngeal airway area (LOA: passes through the upper tip at the end of the epiglottis), minimum lower oropharyngeal airway area (MLOA: the minimum cross-sectional area of LOV), and minimum total oropharyngeal airway area (MTOA: the minimum cross-sectional area of TOV).
The present study investigated the differences between the various skeletal patterns in terms of the volume and area of the oropharyngeal airway. Statistical analysis was performed using SPSS (version 20; IBM, Armonk, NY, USA), and p < 0:05 was the criterion for statistical significance.
The mean values among the groups were compared using one-way analysis of variance with post hoc Tukey HSD test. Pearson's correlation coefficient was used to compare the correlations among the variables of the groups. Regarding the absolute value of the correlation coefficient (r), 0-0.19 indicated a very weak correlation, 0.2-0.39 indicated a weak correlation, 0.40-0.59 indicated a moderate correlation, 0.6-0.79 indicated a strong correlation, and 0.8-1 indicated a very strong correlation. This study was approved by the Institutional Review Board of Kaohsiung Medical University Hospital (KMUHIRB-E(II)-20160066).
# Results
No significant difference was observed between the patients in the three groups in terms of age or BMI [fig_ref] Table 3: Oropharyngeal airway spaces in the skeletal patterns using one-way ANOVA with Tukey's... [/fig_ref] presents a comparison of oropharyngeal airway space in the three skeletal patterns. The UOA of patients in Classes III (468.5 mm 2 ) and I (443.9 mm 2 ) was significantly greater than that of patients in Class II (377. On the basis of patient characteristics [fig_ref] Table 4: Pearson test of oropharyngeal airway in the patient's characteristics [/fig_ref] , Pearson's test was performed to evaluate the correlations of pharyngeal airway space. Both the area and volume of each airway space were significantly positively correlated with sex: male patients had larger airways, indicating a positive correlation. Skeletal pattern had a significant positive correlation with MUOA and MTOA: the MUOA and MTOA of patients in Class III were larger, indicating a positive correlation. Age exhibited a significant negative correlation with MUOA, LOA, and MTOA; higher age was associated with As shown in [fig_ref] Table 5: Pearson test of measured angles and oropharyngeal airway [/fig_ref] , the Or-Po-Pog angle was significantly negatively correlated (weak or very weak) with UOA,
# Discussion
The volume of the pharyngeal airway can be affected by anatomical anomalies in both the soft tissue and craniofacial skeleton. According to functional matrix theory, proposed by Moss, the growth and development of the craniofacial area can be controlled by the functional activity of the soft tissue around the craniofacial skeleton. Thus, a direct interaction exists between the pharyngeal airway space and craniofacial morphology, and any anomaly in these spaces could affect the position of the surrounding bones. Related literature [bib_ref] A longitudinal study of the growth of the nasopharynx and its contents..., Jeans [/bib_ref] [bib_ref] Pharyngeal airway dimensions: a cephalometric, growth-study-based analysis of physiological variations in children..., Mislik [/bib_ref] has reported rapid and ongoing growth of the pharyngeal structure before the age of 13 years that ceases between 14 and 18 years of age. On the basis of relevant research results, this study focused on the pharyngeal airways of patients aged over 16 years, which constitutes the most mature and stable period. BMI is generally used to represent a patient's physical characteristics. In this study, no significant difference was observed between the age and BMI of the patients in the three groups, signifying similar demographic characteristics of all patients. Therefore, the results of this study were unaffected by differences in the physical characteristics of the patients, thereby revealing their actual oropharyngeal airway statuses with objective measurements. It is as expected that BMI exhibited no significant correlation with the area or volume of the oropharyngeal airway. Claudino et al. [bib_ref] Pharyngeal airway characterization in adolescents related to facial skeletal pattern: a preliminary..., Claudino [/bib_ref] and Tseng et al. [bib_ref] Evaluation of pharyngeal airway volume for different dentofacial skeletal patterns using cone-beam..., Tseng [/bib_ref] indicated that airway volume was significantly correlated with ANB angle, whereas Kula et al. [bib_ref] Three dimensional evaluation of upper airway volume in children with different dental..., Kula [/bib_ref] and Alves et al. [bib_ref] Evaluation of pharyngeal airway space amongst different skeletal patterns, Alves [/bib_ref] found no significant correlation between these elements. The current study confirmed the findings of Claudino et al. [bib_ref] Pharyngeal airway characterization in adolescents related to facial skeletal pattern: a preliminary..., Claudino [/bib_ref] and Tseng et al. [bib_ref] Evaluation of pharyngeal airway volume for different dentofacial skeletal patterns using cone-beam..., Tseng [/bib_ref] that the ANB angle is a crucial factor affecting airway dimensions. Alves et al. [bib_ref] Evaluation of pharyngeal airway space amongst different skeletal patterns, Alves [/bib_ref] reported significant differences between the airway volumes of male and female participants. However, a study by Solow et al. 5 BioMed Research International [bib_ref] Airway adequacy, head posture, and craniofacial morphology, Solow [/bib_ref] revealed that sex did not significantly affect airway dimensions. The current study also noted a significant correlation between the oropharyngeal airway space and sex; the oropharyngeal airway space of male patients was larger than that of female patients.
El and Palomo [bib_ref] Airway volume for different dentofacial skeletal patterns, El [/bib_ref] indicated that the relation between the position of the mandible and skull base also affects the oropharyngeal space. Kim et al. [bib_ref] Threedimensional analysis of pharyngeal airway in preadolescent children with different anteroposterior skeletal..., Kim [/bib_ref] indicated that compared with patients with normal skeletal anterior-posterior relationships, patients with a mandible positioned more to the rear had a smaller airway volume. Research reports [bib_ref] Relationship of the hyoid bone and posterior surface of the tongue in..., Yamaoka [/bib_ref] have highlighted the crucial role of the hyoid bone and the muscle tissue attached to it in maintaining a normal airway space, and different positions of the mandible are often accompanied by diverse hyoid bone positions. Yamaoka et al. [bib_ref] Relationship of the hyoid bone and posterior surface of the tongue in..., Yamaoka [/bib_ref] revealed that the tongue base of patients with skeletal Class II malocclusion was positioned farther back compared with that of patients with skeletal Class III malocclusion. In general, mandibles that are shorter and/or located farther back might push the tongue and soft palate back into the pharyngeal space, thus reducing the oropharyngeal volume. Patients in Class III had a more protruded mandible; thus, the hyoid bone had a more anterior position, accounting for the larger distance between the back of the tongue and the posterior pharyngeal wall. Therefore, patients in Class III had the largest airway volumes. Consistent with the aforementioned reports [bib_ref] Threedimensional analysis of pharyngeal airway in preadolescent children with different anteroposterior skeletal..., Kim [/bib_ref] [bib_ref] Relationship of the hyoid bone and posterior surface of the tongue in..., Yamaoka [/bib_ref] , the current study also found that the horizontal distance of the hyoid bone and Pog among patients in Class II was significantly smaller than that among patients in Class I and Class III.
When the related structural positions of Pog, H, and C2 on the FH plane were evaluated in terms of Or-Po-Pog and Or-Po-C2 angles, the angles of patients in Class II were significantly larger than those of patients in Class I and Class III. No significant difference was observed in the Or-Po-H angle between the three groups; however, the horizontal position of H in patients in Class II was significantly farther back than that in patients in Class I and Class III, indicating that patients in Class II would raise their heads to elevate the FH planes more to compensate for smaller airways, which explains the absence of a significant difference between the Or-Po-H angles of the patients in the three groups. When the airway was examined through the cervical spine and related structural positions through C2, no significant difference was observed in terms of the Po-C2-Pog and Po-C2-H angles of patients in Class I and Class III, reflecting that the pharyngeal airway spaces of those in Class I and III also showed no significant difference. By contrast, the angles of patients in Class II were the smallest, and their airways were also the smallest, probably representing a compensation mechanism for maintaining an airway patency and function when the glossopharyngeal airway volume decreases.
The minimum cross-sectional area is an important factor in the evaluation of the obstruction potency of the pharyngeal airway. Pharyngeal airway obstruction in patients with sleep apnea manifests through not only reduced airway volume but, more crucially, also compressed area (the minimum cross-sectional area). Trudo et al. [bib_ref] State-related changes in upper airway caliber and surrounding soft-tissue structures in normal..., Trudo [/bib_ref] had shown by statedependent imaging that the mean minimal cross-sectional airway area was reduced by 228% (p = 0:004) in the retropalatal region (UOA) and by 22% (p = 0:02) in the retrolingual region (LOA) during sleep in normal subjects. Therefore, both of UOA and LOA collapse partially and cause the changes of airflow dynamic during sleep, especially in UOA. Alves et al. [bib_ref] Evaluation of pharyngeal airway space amongst different skeletal patterns, Alves [/bib_ref] observed significant differences between the minimum cross-sectional areas of the airways of patients in Class I and Class II. Claudino et al. [bib_ref] Pharyngeal airway characterization in adolescents related to facial skeletal pattern: a preliminary..., Claudino [/bib_ref] reported that the minimum cross-sectional areas of the lower pharynx, velopharynx, and oropharynx as well as the mean crosssectional area of patients in Class II were all smaller than those of patients in Class III. The current research revealed that in terms of the minimum cross-sectional area (MUOA, MLOA, and MTOA) of the oropharyngeal airway, no significant difference was observed between patients in Class I and II. However, the minimum cross-sectional area (MUOA, MLOA, and MTOA) of patients in Class III was significantly greater than that of patients in Class II. More importantly, the present study revealed the area with the highest frequency of MTOA during pharyngeal airway obstruction. Two-thirds of the patients in Class I and Class II had an MTOA in the UOV, and approximately 50% of patients in Class III had an MTOA in both the UOV and LOV. This indicates that the position of Pog in patients in Class III could enlarge the MUOA and UOA more than the MLOA and LOA. Therefore, different obstruction areas of the pharyngeal airways were observed in the three skeletal patterns.
Grauer et al. [bib_ref] Working with DICOM craniofacial images, Grauer [/bib_ref] reported that the glossopharyngeal airway volumes of patients in Class II were smaller than those of patients in Class I. This reduction in pharyngeal airway volume was mainly due to the mandible position being farther [bib_ref] Cone-beam evaluation of pharyngeal airway space in class I, II, and III..., Castro-Silva [/bib_ref] reported that the mean volume of the pharyngeal airway space among patients in Class III was significantly greater than that among patients in Class I and Class II. The current results are consistent with those reported by Castro-Silva et al., [bib_ref] Threedimensional analysis of pharyngeal airway in preadolescent children with different anteroposterior skeletal..., Kim [/bib_ref] in which the oropharyngeal airway volumes of patients in Class III were significantly larger than those of patients in Class II. Contrary to the findings of Grauer et al. [bib_ref] Working with DICOM craniofacial images, Grauer [/bib_ref] , those of the present study revealed that no significant difference existed in oropharyngeal airway volume between patients in Class I and Class II. Analysis using Pearson's correlation coefficient revealed significant correlations between sex and airway space in terms of both oropharyngeal area and oropharyngeal volume; the values for airways were significantly higher in male patients than in female patients, and ANB was significantly negatively correlated with all of the airway spaces. In addition, when the correlation of airway space alone was considered with respect to the three skeletal pattern types, a significant positive correlation was observed in MUOA and MTOA, whereas a significant negative correlation was observed between the ANB and the oropharyngeal airway variables. The positions of the hyoid bone and Pog were nearly significantly correlated with the area or volume of the oropharyngeal airway space, and the correlation strength of PogX (horizontal position) was greater than that of HX (horizontal position). Moreover, the correlation strength of HY (vertical position) was greater than that of PogY (vertical position). From the Or-Po-Pog angles of the head position (FH plane), significant negative correlations were observed with the measurements of all of the oropharyngeal airways; that is, the airway space was smaller when the Pog was farther back. Moreover, when observed from the cervical spine, several negative correlations with Po-C2-C4 were observed, indicating that the angle of the cervical spine affected the volume and area of the oropharyngeal airway space.
# Conclusion
The oropharyngeal airway areas and volumes of patients in Class II were significantly smaller than those of patients in Class III. The positions of the mandible, head, and cervical spine were important factors affecting the oropharyngeal airway area and volume. The minimum oropharyngeal crosssectional area had a 66%-67% probability of occurrence in the upper oropharyngeal airway among patients in Class I and Class II and a 50% probability of occurrence among patients in Class III.
## Data availability
The data used to support the findings of this study are included within the article. The data used to support the findings of this study are available from the corresponding author upon request.
[fig] 2: mm 2 ). The MUOA (118.3 mm 2 ), LOA (289.7 mm 2 ), MLOA (113.4 mm 2 ), UOV (13801.9 mm 3 ), LOV (7773.5 mm 3 ), MTOA (96 mm 2 ), and TOV (21575.4 mm 3 ) of patients in Class III were significantly greater than the corresponding values of patients in Class II (78.8 mm 2 , 225.4mm 2 , 86.0 mm 2 , 10658.7 mm 3 , 6051.5 mm 3 , 69.6 mm 2 , and 16710.1 mm 3 , respectively). Evaluation of the distribution of MTOA revealed that 20 patients in Class I had MUOA and 10 had MLOA, 27 patients in Class II had MUOA and 13 had MLOA, and 39 patients in Class III had MUOA and 38 had MLOA. The MUOA represented the MTOA of the oropharyngeal airway in a two-thirds of patients in Class I (66.7%), a two-thirds of patients in Class II (67.5%), and one-half (50.6%) of patients in Class III. [/fig]
[fig] Figure 1: Landmarks: sella (S), nasion (N), A point (A), B point (B), pogonion (Pog), hyoid bone (H), second cervical vertebra (C2), and fourth cervical vertebra (C4). X-axis (white line): constructed by drawing a line through nasion 7°up from SN line. Y-axis (white line): a line through sella (S) perpendicular to the X-axis. The measured angles: brown color, ANB angle; green color, (1) Or-Po-Pog angle, (2) Or-Po-H angle, (3) Or-Po-C2 angle; and yellow color, (4) Po-C2-Pog angle, (5) Po-C2-H angle, (6) Po-C2-C4 angle. [/fig]
[fig] Figure 3: (a) In the 3D image, the minimum cross-sectional area (green color) of upper oropharyngeal airway in the oropharyngeal airway (pink color). (b) The minimum cross-sectional area (pink color) of upper oropharyngeal airway in the axial view. [/fig]
[table] Table 1: In terms of the horizontal distance of H and Pog, that of patients in Class II (27.1 mm and 63.7 mm, respectively) was [/table]
[table] Table 3: Oropharyngeal airway spaces in the skeletal patterns using one-way ANOVA with Tukey's HSD post hoc test. [/table]
[table] Table 4: Pearson test of oropharyngeal airway in the patient's characteristics. [/table]
[table] Table 5: Pearson test of measured angles and oropharyngeal airway. UOA: upper oropharyngeal area; MUOA: minimum upper oropharyngeal area; LOA: lower oropharyngeal area; MLOA: minimum lower oropharyngeal area; UOV: upper oropharyngeal volume; LOV: lower oropharyngeal volume; MTOA: minimum total oropharyngeal area; TOV: total oropharyngeal volume; Or: orbitale; Po: porion; Pog: pogonion; H: hyoid bone; C2: second cervical vertebra; C4: fourth cervical vertebra. * Statistically significant, p < 0:05.back. Moreover, Castro-Silva et al. [/table]
|
10.1039/d1cp02999a
|
CCBY
| null |
34473158
|
s2orc_pubmed_articles
|
Lifting the discrepancy between experimental results and the theoretical predictions for the catalytic activity of RuO2(110) towards oxygen evolution reaction.
Developing new efficient catalyst materials for the oxygen evolution reaction (OER) is essential for widespread proton exchange membrane water electrolyzer use. Both RuO 2 (110) andIrO 2 (110) have been shown to be highly active OER catalysts, however DFT predictions have been unable to explain the high activity of RuO 2 . We propose that this discrepancy is due to RuO 2 utilizing a different reaction pathway, as compared to the conventional IrO 2 pathway. This hypothesis is supported by comparisons between experimental data, DFT data and the proposed reaction model. Furthermore, our findings indicate that the reaction pathway utilized by RuO 2 (110) might be pH dependent, following the conventional pathway at high pH. view of proton exchange membrane water electrolysis on degradation mechanisms and mitigation strategies. Journal of Power Sources, 366:33-55, 2017. 7 G. Matute, J.M. Yusta, and L.C. Correas. Techno-economic modelling of water electrolysers in the range of several MW to provide grid services while generating hydrogen for different applications: A case study in Spain applied to mobility with the stability of oxygen evolution reaction catalysts: The importance of monitoring mass losses. Degradation of iridium oxides via oxygen evolution from the lattice: correlating atomic scale structure with reaction mechanisms. Energy Environ. Sci., 12:3548-3555, 2019.
# Introduction
As part of a transition to a future sustainable economy, there is a need for sustainable fuel and energy storage. Hydrogen gas is an ideal candidate for such a fuel and storage compound, as it can be readily produced by the electrolysis of water 1 . Furthermore, the hydrogen is completely sustainable if the source of electricity is renewable 2 . Currently, the major challenge facing widespread electrolyzer use is the sluggish kinetics of the Oxygen Evolution Reaction (OER) at the anode 3 , fundamentally limited by the universal scaling relations [bib_ref] Oxygen evolution reaction: a perspective on a decade of atomic scale simulations, Divanis [/bib_ref]. Further development of the water electrolyzer thus requires finding efficient and practical catalysts to facilitate the OER. There are three types of water electrolyzers: alkaline water electrolyzers, proton exchange membrane (PEM) water electrolyzers and solid oxide water electrolyzers [bib_ref] Importance of Surface IrOx in Stabilizing RuO2 for Oxygen Evolution, Feng [/bib_ref]. Of these three the alkaline water electrolyzer is the most mature and commercialized. Yet PEM technology has many advantages compared to the alkaline electrolyzer. Some examples include a much higher current density, purer gas, a smaller size for the same power and even the ability to operate at high pressure [bib_ref] Importance of Surface IrOx in Stabilizing RuO2 for Oxygen Evolution, Feng [/bib_ref]. Currently the best candidates for PEM electrolyzer anode material are IrO 2 and RuO 2 , as these are both stable and active 8-13 . Both iridium and ruthenium are however scarce materials and thus expensive [bib_ref] Availability of metals and materials, Hageluken [/bib_ref]. It is therefore unrealistic to expect that these catalysts can be used on an industrial scale that would have an impact on society [bib_ref] Powering the planet with solar fuel, Gray [/bib_ref] [bib_ref] Addressing the terawatt challenge: scalability in the supply of chemical elements for..., Peter [/bib_ref] [bib_ref] Electrolysis of water on oxide surfaces, Rossmeisl [/bib_ref] [bib_ref] Synthesis and Activities of Rutile IrO2 and RuO2 Nanoparticles for Oxygen Evolution..., Lee [/bib_ref]. This suggests that either the current reaction model is wrong in the case of RuO 2 or the under-evaluation is due to a computational artefact. In this work we propose that the discrepancy is due to RuO 2 utilizing an alternate reaction pathway for oxygen evolution as compared to IrO 2 . We therefore argue that it is not due to a computational artefact.
# Results-discussion
The conventional pathway describing the interaction between water molecules and the surface of an electrocatalyst was suggested in 2004 21 . During this reaction pathway three intermediates are produced via four electron-proton pair exchanges between the anode and the electrolyte, presented in the following reactions:
[formula] H 2 O + * → HO * + H + + e −(1)HO * → O * + H + + e −(2)H 2 O + O * → HOO * + H + + e −(3)HOO * → * + O 2 + H + + e −(4) [/formula]
where * indicates an active site of the surface and HO * , O * , HOO * the adsorbed intermediates on that particular site. The above reaction path describes the Oxygen Evolution Reaction taking place in an acidic environment but it can also be used for the thermodynamic description of the procedure happening in alkaline environment [bib_ref] Jónsson. Origin of the overpotential for oxygen reduction at a fuel-cell cathode, Nørskov [/bib_ref] [bib_ref] Electrolysis of water on (oxidized) metal surfaces, Rossmeisl [/bib_ref]. A schematic representation of the conventional OER reaction path is depicted in [fig_ref] Figure 1: Illustration of two oxygen evolution reaction pathways [/fig_ref]. analysis, is that the blue trend line is followed by experimental data produced in the work of Suntivic et.al.. In their experiments, RuO 2 (110) surfaces were synthesized and their electrochemical response in different pH is recorded. Furthermore, they assign the first and second pre-oxidation peaks observed at the cyclic voltammetries, as the HO * and O * intermediates respectively. The experimental HO * energies serve the role of the descriptor for the experimental data at the activity volcano in the above diagram [fig_ref] Figure 2: The OER activity volcano [/fig_ref]. The red triangles corresponding to IrO 2 , reproduced from another work of Suntivic et. al. [bib_ref] Influence of Surface Adsorption on the Oxygen Evolution Reaction on IrO2(110), Kuo [/bib_ref] , tend as an ensemble to be placed towards weaker HO * binding energies, closer to the strong binding side of the conventional activity volcano. This is an indication that IrO 2 follows the conventional reaction pathway. The RuO 2 experimental data points are however spread. The points corresponding to highly acidic electrolytes are placed right on top of the blue trend-line together with the theoretical prediction for pathway 1b. In contrast, those corresponding to neutral and alkaline electrolytes are placed closer to the conventional activity volcano, suggesting that RuO 2 at those conditions might follow the conventional pathway.
In [fig_ref] Figure 3: Scaling relation of HO * binding energies against O * on the... [/fig_ref] It is the relative strong binding of the oxygen intermediate, which makes the calculated activity of RuO 2 smaller than IrO 2 . Whereas the binding of HO * and HOO * on the RuO 2 cus site is similar, the O * binding is much stronger than that on IrO 2 . This could be an artefact of the DFT calculations, however, it is seen to hold across DFT implementations. The binding energies vary between the different methods, but the difference between HO* and O* binding is close to constant.
Previous experimental studies also show that RuO 2 binds oxygen stronger than IrO 2 for the same HO* binding 27 , even if the difference is smaller than that found in the DFT data. In contrast to the DFT data, differences in experimental data is due to varying electrolyte pH. The experiments measure the potentials for the first and second oxidation peaks, those are normally assumed to be
# Conclusion
In this work we are studying the discrepancy between DFT and experimental results, regarding the oxygen evolving reactivity of RuO 2 . We propose that the reaction pathway for electrochemical water oxidation on RuO 2 (110) surfaces, at least in acidic conditions, is slightly different from the reaction path on IrO 2 . In particular, the differences are located at the first and third intermediates,
where the protons of HO * and HOO * are migrating towards the bridge oxygen surface. The energy inter-dependency of HO * and HOO * is 2.7eV for the RuO 2 pathway, and is much closer to the ideal difference of 2.46eV . As a consequence the DFT activity is much higher than the one produced by the conventional mechanism and thus the structure is placed closer to the apex of the activity volcano. Furthermore the new placement of RuO 2 (110) on the activity volcano, is at the same region of the RuO 2 experimental results for high acidic electrolytes. By using the conventional pathway we have a very weak interaction of HO * with the surface's cus site. On the other hand, using the RuO 2 pathway widens the energy difference between HO * and O * , placing this DFT calculation closer to experimental trend-lines. This theoretical-experimental agreement, indicates that the RuO 2 mechanism, at least for highly acidic environments, is followed.
[fig] Figure 1: Illustration of two oxygen evolution reaction pathways. Each step is accompanied with the exchange of a proton-electron pair between the electrolyte and the catalyst surface. a: The conventional OER path. b: The proposed reaction mechanism for OER on RuO 2 surfaces. by a yellow dotted line). This hypothesis is supported by the experimental works of R. Rao et al, who identified a −OO species at high potentials 17,26 . This −OO species is the experimental equivalent of the third intermediate of the RuO 2 pathway depicted in Fig. 1b. [/fig]
[fig] Figure 2: The OER activity volcano. The red data points correspond to IrO 2 and the blue ones to RuO 2 . The triangles represent the experimental data while the circle and the the pentagon are theoretical data for the conventional and the RuO 2 pathway respectively. The blue trend line corresponds to the strong binding side of the volcano if the RuO 2 pathway is followed. The blue shaded area around the blue trend line, is the DFT error of ±0.2eV . The left y-axis(cyan) presents the theoretical overpotential and the right y-axis(magenta) presents the experimental overpotential.The theoretical and experimental overpotentials are calibrated by overlapping the theoretical and experimental value for IrO 2 (110) on the y-axis. Two different descriptors with the same scale are used on the x-axis. G O − G HO and G HO for the theoretical and the experimental data respectively.The different way that the first and third intermediates are adsorbed on the surface, has an effect on their binding energy and thereby on the overall activity. The energy inter-dependency of the HO * and HOO * intermediates thus changes from ≈ 3.2eV , as dictated by the universal scaling relations 4,5 , to ≈ 2.7eV , a value that is closer to the ideal value of 2.46eV . This relation is depicted in the activity volcano ofFig. 2by the blue trend line. The activity volcano supports our hypothesis, as the DFT data point corresponding to RuO 2 following pathway 1b (blue pentagon), holds a lower overpotential compared to the data point corresponding to the conventional pathway (blue circle). This places it right on top of the blue trend line. An observation that is strengthening our [/fig]
[fig] Figure 3: Scaling relation of HO * binding energies against O * on the cus site for IrO 2 and RuO 2 (110) surfaces. The circles correspond to DFT data following different DFT implementations as reproduced from Federico Calle-Vallejo et al19 . In contrast, the triangles correspond to experimental data in varying electrolyte conditions 27 . Each of these data sets have their own corresponding trend-line. The blue pentagon represents RuO 2 DFT data following pathway 1b, and the blue arrow represents the difference in ∆G HO * as a result of following this pathway. This indicated difference is the same size as the difference between the experimental and DFT trend-lines for RuO 2 . [/fig]
|
10.4014/jmb.2111.11026
|
CCBY
|
9705928
|
34949742
|
s2orc_pubmed_articles
|
A Promising Vaccination Strategy against COVID-19 on the Horizon: Heterologous Immunization
# Introduction
## Immune responses to variants of concern
Four variants of SARS-CoV-2 have been posing added threats to the ongoing pandemic: B.1.1.7 (Alpha), B.1.351 (Beta), P.1 (Gamma), and B.1.617.2 (Delta). A VOC is defined as a virus with mutations in multiple clusters in their genome with either detrimental changes in COVID-19 epidemiology, virulence, decreased effectiveness of public health measures, available diagnostics, vaccines, and/or therapeutics. Compared to the Wuhan-1 reference strain (wild-type), VOC predominantly have mutations in the spike gene, altering their interactions with the host receptor ACE2, which results in higher infection rates. For example B.1.1.7 (~43-90% more transmissible compared with previous circulating strains) has H69/V70 and Y144 deletions; N501Y, A570D, D614G, and P681H substitution [bib_ref] Changes in symptomatology, reinfection, and transmissibility associated with the SARS-CoV-2 variant B.1.1.7:..., Graham [/bib_ref] [bib_ref] Estimated transmissibility and impact of SARS-CoV-2 lineage B. 1.1. 7 in England, Davies [/bib_ref] [bib_ref] SARS-CoV-2 variants, spike mutations and immune escape, Harvey [/bib_ref]. B.1.351 (25% more transmissible) has K417N, E484K, N501Y, and D614G key substitutions [bib_ref] SARS-CoV-2 variants, spike mutations and immune escape, Harvey [/bib_ref] [bib_ref] Increased transmissibility and global spread of SARS-CoV-2 variants of concern as at, Campbell [/bib_ref]. P.1 (1.4-2.2 times more transmissible) has E484K, K417N/T, N501Y and D614G key substitutions [bib_ref] SARS-CoV-2 variants, spike mutations and immune escape, Harvey [/bib_ref] [bib_ref] Genomics and epidemiology of the P.1 SARS-CoV-2 lineage in Manaus, Faria [/bib_ref]. B.1.617.2 (97% more transmissible) has L452R, T478K, D614G, P681R key substitutions [bib_ref] Increased transmissibility and global spread of SARS-CoV-2 variants of concern as at, Campbell [/bib_ref] [bib_ref] SARS-CoV-2 spike mutations, L452R, T478K, E484Q and P681R, in the second wave..., Cherian [/bib_ref]. VOC can also alter the potency of neutralizing antibodies, resulting in compromised vaccine efficacy and effectiveness. A properly timed and effective immune response is important for outsmarting the SARS-CoV-2. Firstly, a proper neutralizing antibody response would substantially decrease the number of virions that could successfully infect angiotensin converting enzyme-2 (ACE-2) receptor-expressing cells [bib_ref] Antibody responses in COVID-19: A, Chvatal-Medina [/bib_ref]. Neutralizing antibodies are able to bind to the virus and directly block its ability to infect cells, usually through inhibition of the interaction between the viral spike protein and the cellular ACE2 receptor [bib_ref] Prospects for durable immune control of SARS-CoV-2 and prevention of reinfection, Cromer [/bib_ref]. Secondly, T cells might be playing
## Fig.1. proposed mechanism of action of covid-19 vaccines. (a)
In mRNA-vaccines, the spike mRNA is modified in which uridine is replaced by pseudouridine in order to escape immune responses. Moreover, the mRNA is stabilized in its prefusion conformation by two consecutive proline substitutions at amino acid positions 986 and 987, at the top of the central helix in the S2 subunit. mRNA is loaded in the lipid nanoparticles, which interact with the cell membrane and release modified mRNA in the cytoplasm of the muscle cell or antigen-presenting cell (1) (2) (3). mRNA is translated into spike protein in the cytoplasm, later presented by MHC class I to CD8 + T cells. The spike antigens are also released in the extracellular environment where they migrate to the draining lymph nodes and are endocytosed by APCs within the germinal centers. APCs at the site of injection may also be involved. These endocytosed antigens are processed by the presented by MHC class II to CD4 + T cells. Activated CD4 + T cells help to activate CD8 + T cells and B cells. CD8 + T cells kill the infected cells (not investigated in the context of SARS-CoV-2). With the help of CD4 + T cells, B cells mature to plasma secreting cells and synthesize antibodies to combat SARS-CoV-2. (B) In viral vectored vaccines, the full-length spike gene (DNA) is inserted in a harmless adenovirus vector (rAdenovirus) [bib_ref] Current status of epidemiology, diagnosis, therapeutics, and vaccines for novel coronavirus disease..., Ahn [/bib_ref]. rAdenovirus latches to the host cell and releases DNA in the cytoplasm (2) (3), which later migrates to the nucleus of the cell and is transcribed to mRNA (4). (C) Inactivated pathogen vaccines are chemically inactivated by βpropiolactone. The active component, along with the alum, generates immune responses. APCs process the antigens by MHC I machinery and present antigens to CD4 + T cells. (D) Full-length stabilized spike gene is engineered into baculovirus (rbaculovirus). rbaculovirus delivers spike gene into the Sf9 (1) (2), migrates to the nucleus of the cell, and is transcribed to mRNA. mRNA further migrates to the cytoplasm and is translated to spike protein [bib_ref] mRNA-1273 COVID-19 vaccine effectiveness against the B.1.1.7 and B.1.351 variants and severe..., Chemaitelly [/bib_ref]. Translated and glycosylated spike protein is eventually purified and mixed with adjuvant [bib_ref] Adverse rare events to vaccines for COVID-19: from hypersensitivity reactions to thrombosis..., Novak [/bib_ref]. The dotted line represents that the exact role of CD8 + T cells is not known. MHC = major histocompatibility complex. mRNA = messenger RNA. r = recombinant. Sf9 cell = insect cell line, a clonal isolate derived from the parental Spodoptera frugiperda cell line IPLB-Sf-21-AE. a role in combating SARS-CoV-2 [bib_ref] Prospects for durable immune control of SARS-CoV-2 and prevention of reinfection, Cromer [/bib_ref]. SARS-CoV-2-specific CD4 + T cells commonly differentiate into Th1 and Tfh T cells [bib_ref] Adaptive immunity to SARS-CoV-2 and COVID-19, Sette [/bib_ref] [bib_ref] Phenotype and kinetics of SARS-CoV-2-specific T cells in COVID-19 patients with acute..., Weiskopf [/bib_ref] [bib_ref] SARS-CoV-2-specific T cells exhibit phenotypic features of helper function, lack of terminal..., Neidleman [/bib_ref]. Th1 cells have antiviral properties, and Tfh cells are specialized in providing help to B cells and are critical for the development of neutralizing antibodies, memory B cells, and long-term humoral memory. Mostly Th1 skewed responses with little to none Th2 cytokines were detected in mRNA [bib_ref] COVID-19 vaccine BNT162b1 elicits human antibody and TH 1 T cell responses, Sahin [/bib_ref] and adenovirus vector-based vaccineswhile Tfh cells have also been detected in vaccinated individuals [bib_ref] COVID-19 vaccine BNT162b1 elicits human antibody and TH 1 T cell responses, Sahin [/bib_ref] [bib_ref] Rapid induction of antigen-specific CD4+ T cells is associated with coordinated humoral..., Painter [/bib_ref] [bib_ref] 2021. T follicular helper cells in the humoral immune response to SARS-CoV-2..., Koutsakos [/bib_ref]. In addition, CD8 + T cells can directly kill infected cells, which are also induced after SARS-CoV-2 infection or vaccination [bib_ref] Adaptive immunity to SARS-CoV-2 and COVID-19, Sette [/bib_ref] [bib_ref] Rapid induction of antigen-specific CD4+ T cells is associated with coordinated humoral..., Painter [/bib_ref]. The presence of virus-specific CD8 + T cells has been associated with better COVID-19 outcomes. SARS-COV-2-specific CD8 + T cells possess effector molecules, including IFN-γ, granzyme B, perforin, and CD107a [bib_ref] Robust T cell immunity in convalescent individuals with asymptomatic or mild COVID-19, Sekine [/bib_ref] [bib_ref] Antigen-specific adaptive immunity to SARS-CoV-2 in acute COVID-19 and associations with age..., Moderbacher [/bib_ref] [bib_ref] Characterization of pre-existing and induced SARS-CoV-2-specific CD8(+) T cells, Schulien [/bib_ref]. In the case of inactivated COVID-19 vaccine, Th1 and Th2 T cell subsets were not defined although cellular immune responses in vaccinated individuals exhibited antigen-specific CD4 + and CD8 + T cells [bib_ref] Dynamic SARS-CoV-2 specific B cell and T cell responses following immunization of..., Chen [/bib_ref]. However, the role of T cells, especially CD8 + T cells, against SARS-CoV-2 remains to be elucidated. Although natural infection with SARS-CoV-2 and different vaccines induce more or less protective immunity, the ability of such immune responses to recognize and provide protection against variants of SARS-CoV-2 is a matter of concern.
## Antibody responses
Planas et al. [bib_ref] Reduced sensitivity of SARS-CoV-2 variant Delta to antibody neutralization, Planas [/bib_ref] , evaluated the neutralizing potential of serum from BNT162b2, or ChAdOx1 nCoV-19 vaccinated individuals against D614G, B.1.1.7, B.1.351, and B.1.617.2 strains. After a single dose (post-3-weeks) of BNT162b2 vaccine, the levels of neutralizing antibodies were low against D614G and almost undetectable against the Alpha, Beta, and Delta variants. When evaluated 5-weeks after the booster, antibody titers significantly increased. However, in contrast to Alpha, 3-and 16-fold reductions in the neutralization titers against the Delta and the Beta variants, respectively, were observed. A similar pattern was observed with the ChAdOx1 nCoV-19. A single dose induced low levels of antibodies neutralizing the Delta and Beta variants (post-10-weeks) compared to the D614G and Alpha. Four weeks after the second dose, neutralizing titers were strongly increased. However, relative to the Alpha, 5-and 9-fold reductions in neutralization titers against the Delta and the Beta variants were observed. Further studies reported that BNT162b2 vaccinated individuals displayed 3.3-, 7.6-, 2.6-, and 2.5-fold reductions in the neutralization titer against Alpha, Beta, Gamma, and Delta variants, respectively, in contrast to Victoria strain [bib_ref] Reduced neutralization of SARS-CoV-2 B.1.1.7 variant by convalescent and vaccine sera, Supasa [/bib_ref] [bib_ref] Evidence of escape of SARS-CoV-2 variant B. 1.351 from natural and vaccine-induced..., Zhou [/bib_ref] [bib_ref] Antibody evasion by the P.1 strain of SARS-CoV-2, Dejnirattisai [/bib_ref] [bib_ref] Reduced neutralization of SARS-CoV-2 B.1.617 by vaccine and convalescent serum, Liu [/bib_ref]. Moreover, ChAdOx1 nCoV-19 vaccinated individuals showed a 2.33-, 9-, 2.9-, and 4.29fold loss in neutralization titer against Alpha, Beta, Gamma, and Delta variants, respectively, compared with Victoria [bib_ref] Reduced neutralization of SARS-CoV-2 B.1.1.7 variant by convalescent and vaccine sera, Supasa [/bib_ref] [bib_ref] Evidence of escape of SARS-CoV-2 variant B. 1.351 from natural and vaccine-induced..., Zhou [/bib_ref] [bib_ref] Antibody evasion by the P.1 strain of SARS-CoV-2, Dejnirattisai [/bib_ref] [bib_ref] Reduced neutralization of SARS-CoV-2 B.1.617 by vaccine and convalescent serum, Liu [/bib_ref]. In the following sections vaccination refers to full vaccination as recommended by respective manufacturers unless otherwise mentioned.
B.1.1.7 VOC. The sera from mRNA or viral vectored vaccinated individuals showed a small yet significant reduction (1.7 to 2.5-fold) in neutralizing activity against B.1.1.7 compared to the reference strains [bib_ref] Reduced neutralization of SARS-CoV-2 B.1.1.7 variant by convalescent and vaccine sera, Supasa [/bib_ref] [bib_ref] SARS-CoV-2 variants B.1.351 and P.1 escape from neutralizing antibodies, Hoffmann [/bib_ref] [bib_ref] Serum neutralizing activity elicited by mRNA-1273 vaccine, Wu [/bib_ref] [bib_ref] Neutralising antibody activity against SARS-CoV-2 VOCs B.1.617.2 and B.1.351 by BNT162b2 vaccination, Wall [/bib_ref] [bib_ref] Serum neutralizing activity of mRNA-1273 against SARS-CoV-2 variants, Choi [/bib_ref] [bib_ref] Resistance of SARS-CoV-2 variants to neutralization by monoclonal and serum-derived polyclonal antibodies, Chen [/bib_ref] [bib_ref] Efficacy of ChAdOx1 nCoV-19 (AZD1222) vaccine against SARS-CoV-2 variant of concern 202012/01..., Emary [/bib_ref]. However, a 9-fold loss in neutralization potential of serum from ChAdOx1 nCoV-19 vaccinated individuals was also reported. Serum from inactivated pathogen vaccine, CoronaVac, vaccinated individuals showed a 17.35-fold loss in geometric mean titer of neutralizing antibodies against the authentic virus, compared with wild-type SARS-CoV-2 [bib_ref] CoronaVac induces lower neutralising activity against variants of concern than natural infection, Vacharathit [/bib_ref]. Other studies suggest that neutralizing activity of mRNA or viral vectored vaccinated individuals sera likely maintains protective efficacy against B.1.1.7 [bib_ref] Antibody evasion by the P.1 strain of SARS-CoV-2, Dejnirattisai [/bib_ref] [bib_ref] COVID-19 mRNA vaccine induced antibody responses against three SARS-CoV-2 variants, Jalkanen [/bib_ref] [bib_ref] Antibody resistance of SARS-CoV-2 variants B.1.351 and B.1.1.7, Wang [/bib_ref]. All these in-vitro studies have certain limitations and variations in methodology, sample size, sampling time, and considering only the humoral arm of the immune response [fig_ref] Table 2: Antibody escape by SARS-CoV-2 variants [/fig_ref]. For example, volunteers in phase 2/3 vaccine efficacy study of ChAdOx1 nCoV-19 showed a 9-fold reduced serum neutralization activity against B.1.1.7 in comparison to a canonical non-B.1.1.7 Victoria strain [bib_ref] Efficacy of ChAdOx1 nCoV-19 (AZD1222) vaccine against SARS-CoV-2 variant of concern 202012/01..., Emary [/bib_ref]. In contrast, the vaccine was 70.4% effective against nucleic acid amplification test of nasal swabs for B.1.1.7 (81.5% effective against non-B.1.17 lineages). P.1 VOC. Serum neutralization assay using a pseudovirus system deciphered, 1.2 to 5.12-fold reductions in neutralization against P.1 for mRNA vaccinees serum [bib_ref] SARS-CoV-2 variants B.1.351 and P.1 escape from neutralizing antibodies, Hoffmann [/bib_ref] [bib_ref] Serum neutralizing activity elicited by mRNA-1273 vaccine, Wu [/bib_ref] [bib_ref] Neutralizing activity of BNT162b2-elicited serum, Liu [/bib_ref]. Serum from Ad26.CoV.S vaccinees showed a 3.3-fold reduction in neutralization potential compared with WA1/2020 [bib_ref] Immunogenicity of Ad26.COV2.S vaccine against SARS-CoV-2 variants in humans, Alter [/bib_ref]. Geometric mean serum neutralization titers against P.1 were reduced by 2.6-fold for the BNT162b2 and 2.9-fold for the ChAdOx1 nCoV-19 vaccinee's serum against authentic virus, relative to the Victoria strain [bib_ref] Antibody evasion by the P.1 strain of SARS-CoV-2, Dejnirattisai [/bib_ref].
B.1.351 VOC. In contrast to Alpha, a 9 to 16-fold reduction in the neutralization titers of serum from BNT162b2 vaccinees against the Beta variant has been observed. The neutralization potential of sera from BNT162b2 (1 to 5-weeks post-second dose) was 7.6 to 16-fold resistant to B.1.351 as compared with reference strains [bib_ref] Reduced sensitivity of SARS-CoV-2 variant Delta to antibody neutralization, Planas [/bib_ref] [bib_ref] Evidence of escape of SARS-CoV-2 variant B. 1.351 from natural and vaccine-induced..., Zhou [/bib_ref] [bib_ref] Antibody evasion by the P.1 strain of SARS-CoV-2, Dejnirattisai [/bib_ref] [bib_ref] Neutralising antibody activity against SARS-CoV-2 VOCs B.1.617.2 and B.1.351 by BNT162b2 vaccination, Wall [/bib_ref] [bib_ref] COVID-19 mRNA vaccine induced antibody responses against three SARS-CoV-2 variants, Jalkanen [/bib_ref] [bib_ref] Antibody resistance of SARS-CoV-2 variants B.1.351 and B.1.1.7, Wang [/bib_ref] [bib_ref] Sensitivity of infectious SARS-CoV-2 B.1.1.7 and B.1.351 variants to neutralizing antibodies, Planas [/bib_ref] [bib_ref] SARS-CoV-2 variant B.1.617 is resistant to bamlanivimab and evades antibodies induced by..., Hoffmann [/bib_ref]. Likewise, the neutralization potential of sera from mRNA-1273 vaccinees was 5 to 12.4-fold lesser, compared with reference strains [bib_ref] Serum neutralizing activity elicited by mRNA-1273 vaccine, Wu [/bib_ref] [bib_ref] Serum neutralizing activity of mRNA-1273 against SARS-CoV-2 variants, Choi [/bib_ref] [bib_ref] Antibody resistance of SARS-CoV-2 variants B.1.351 and B.1.1.7, Wang [/bib_ref]. Serum from ChAdOx1 nCoV-19 (4-weeks postsecond shot) or Ad26.CoV.S (71 days post-vaccination) vaccinees showed 9-and 10.6-fold reductions in neutralization potential compared with B.1.1.7 and WA1/2020, respectively [bib_ref] Reduced sensitivity of SARS-CoV-2 variant Delta to antibody neutralization, Planas [/bib_ref] [bib_ref] Immunogenicity of Ad26.COV2.S vaccine against SARS-CoV-2 variants in humans, Alter [/bib_ref]. Another study evaluated serum from CoronaVac vaccinees, reported a 22.11-fold loss in geometric mean titer of neutralizing antibodies against B.1.351, compared with wild-type SARS-CoV-2 [bib_ref] CoronaVac induces lower neutralising activity against variants of concern than natural infection, Vacharathit [/bib_ref].
In a phase III clinical trial, a single shot of Ad26.COV2.S has shown 64.0% and 80% efficacy for moderate to severe-critical COVID-19 against the B.1.351 and B.1.1.7 variants (96% efficacy against the original strain). Despite reduced neutralizing antibody titer (>10.6-fold lower against B.1.351 compared with WA1/2020), the protective efficacy of Ad26.COV2.S might be due to CD8 + T cells and functional non-neutralizing antibodies [bib_ref] Immunogenicity of Ad26.COV2.S vaccine against SARS-CoV-2 variants in humans, Alter [/bib_ref].
B.1.617.2 VOC. Post BNT162b2 vaccination, the sera from individuals showed a 1.14 to 5.8-fold reduction in neutralization titer against B.1617.2 compared with reference strains [bib_ref] Neutralising antibody activity against SARS-CoV-2 VOCs B.1.617.2 and B.1.351 by BNT162b2 vaccination, Wall [/bib_ref] [bib_ref] BNT162b2-elicited neutralization of B.1.617 and other SARS-CoV-2 variants, Liu [/bib_ref]. The serum from mRNA-1273 vaccinated individuals showed a 2.1-fold reduction compared with D614G [bib_ref] Serum neutralizing activity of mRNA-1273 against SARS-CoV-2 variants, Choi [/bib_ref]. Edra et al. [bib_ref] Infection and vaccine-induced neutralizingantibody responses to the SARS-CoV-2 B.1.617 variants. New Engl, Edara [/bib_ref] reported a 3.3-fold and 3-fold decrease in the neutralizing antibody titer in the serum of BNT162b2 and mRNA-1273 vaccinated individuals compared with the WA1/2020 SARS-CoV-2. Liu et al. [bib_ref] Reduced neutralization of SARS-CoV-2 B.1.617 by vaccine and convalescent serum, Liu [/bib_ref] , showed 2.5-fold and 4.29-fold reductions in the neutralizing antibody titers against B.1.617.2 in the serum of BNT162b2 or ChAdOx1 nCoV-19 vaccinated individuals, respectively, compared with Victoria strain. Compared with wild-type SARS-CoV-2, a 31.64-fold loss in geometric mean titer of neutralizing antibodies against authentic B.1.617.2 has been reported in CoronaVac vaccinated individuals [bib_ref] CoronaVac induces lower neutralising activity against variants of concern than natural infection, Vacharathit [/bib_ref].
Although the VOC more or less escape neutralization by antibodies and there are reports of infection by variants in the vaccinated population, the vaccines effectively reduce the severity of the disease [fig_ref] Table 1: Breakthrough cases [/fig_ref]. Currently, the prevention of severe disease and deaths is of utmost importance. However, the resilience of immune responses elicited by COVID-19 vaccines, especially against VOC, remains to be elucidated. Although, an 8month study in which 20 participants received the Ad26.COV2.S vaccine in 1 or 2 doses (either 5×10 10 viral [fig_ref] Table 2: Antibody escape by SARS-CoV-2 variants [/fig_ref].
## T cell responses
Currently, most of the vaccines contain spike [bib_ref] COVID-19 vaccine BNT162b1 elicits human antibody and TH 1 T cell responses, Sahin [/bib_ref] [bib_ref] Predicted cellular immunity population coverage gaps for SARS-CoV-2 subunit vaccines and their..., Liu [/bib_ref] [bib_ref] SARS-CoV-2 mRNA vaccine design enabled by prototype pathogen preparedness, Corbett [/bib_ref] , and mutations have been widely reported in the spike. Antibodies induced by spike of the prototypic strain of SARS-CoV-2 have less binding and neutralization abilities for newly emerging variants resulting in escape from the antibody responses. SARS-CoV-2 antibody responses have received a lot of attention. However, the arsenals of humoral and T cell responses may play diverse roles in different viral infections. In addition, T cells induced by vaccines are supposed to recognize SARS-CoV-2 variants [bib_ref] Impact of SARS-CoV-2 variants on the total CD4+ and CD8+ T cell..., Tarke [/bib_ref] [bib_ref] SARS-CoV-2 variants of concern partially escape humoral but not T-cell responses in..., Geers [/bib_ref]. For example, a study evaluated WT and variants of SARS-CoV-2 specific CD4 + and CD8 + T cell responses in BNT162b2 (n = 8) or mRNA-1273 (n = 11) vaccinees (samples were collected 2-4 weeks after the second dose of vaccination) [bib_ref] Impact of SARS-CoV-2 variants on the total CD4+ and CD8+ T cell..., Tarke [/bib_ref]. Peptide mega pools (MPs) spanning the entire SARS-CoV-2 proteins or only spike were used to stimulate PBMCs, and the response was evaluated based on activation-induced markers (AIM) in CD4 + (OX40 + CD137 + ) and CD8 + (CD69 + CD137 + ) T cells. The CD4 + /CD8 + T cell reactivity in the vaccinees was not substantially affected by mutations in B.1.1.7 and P.1. However, decreases of 14% and 22% were observed with the B.1.351 spike-pools for CD4 + and CD8 + T cells, respectively. AIM T cell responses in COVID-19 vaccinees displayed a memory phenotype irrespective of the variant analyzed, with preferential enrichment for central memory (T cm ) and effector memory (T em ) for CD4 + and T em and terminally differentiated effector memory (T emra ) for CD8 + T cells. These provide evidence that donors primed by the ancestral strain spike protein mount a memory T cell response that can cross-recognize the SARS-CoV-2 VOC. A limitation of this study is that overlapping peptide pools rather than individual peptides were used to evaluate the responses by which alterations in terms of antigen processing for either class I or class II MHC would be undetected. In another study, peptide pool (15-mers with 11 amino acids overlap) spanning mutated spike regions of B.1.1.7 and B.1.351 were used to detect the cross-reactivity of SARS-CoV-2 specific T cells with variants [bib_ref] SARS-CoV-2 variants of concern partially escape humoral but not T-cell responses in..., Geers [/bib_ref]. Blood samples were collected from COVID-19 naïve and recovered donors before and after the BNT162b2 vaccination. No differences in CD4 + T cell activation (based on AIM) were seen in response to variant antigens. However, in this study, the number of donors was limited to 20, and CD8 + T cells responses to VOC were not evaluated. In contrast, a study evaluated 747 SARS-CoV-2 virus isolates by deep sequencing and reported that MHC-I restricted mutant epitopes showed reduced (assessed based on melting temperature stabilizing capacity of wildtype or mutant peptides towards MHC-I) or even abrogated (HLA tetramers, loaded with WT or mutant peptide, were presented to expanded CD8 + T cells of HLA-matched COVID-19 patients) binding to MHC-I [bib_ref] SARS-CoV-2 mutations in MHC-Irestricted epitopes evade CD8(+) T cell responses, Agerer [/bib_ref]. Moreover, CD8 + T cells stimulated with respective epitopes showed decreased proliferation and cytotoxicity. The tetramer-sorted CD8 + T cells revealed qualitative differences at the transcriptional level to mutant peptides. However, this approach should be extended to evaluate the T cell response after vaccination/immunization.
## Breakthrough cases
Breakthrough cases are people who get an infection even after complete immunization, meaning the pathogen breaks the protective barrier developed by vaccination. As already stated, none of the COVID-19 vaccines is 100% effective. Moreover, SARS-CoV-2 variant strains have emerged continuously. Reduced antibody responses in susceptible populations might render them prone to breakthrough infections [bib_ref] BNT162b2 COVID-19 vaccine and correlates of humoral immune responses and dynamics: a..., Lustig [/bib_ref]. Moreover, VOC may escape immune responses, so breakthrough cases are expected [fig_ref] Table 1: Breakthrough cases [/fig_ref]. For example, a study including 1497 fully vaccinated health care workers reported 39 SARS-CoV-2 breakthrough infections. For 22 of the 39 workers with breakthrough infections, the results for peri-infection neutralizing antibodies were available. During the periinfection period, the neutralizing antibody titers in breakthrough cases were lower than those in matched uninfected vaccinated controls (n = 104) [bib_ref] Covid-19 breakthrough infections in vaccinated health care workers, Bergwerk [/bib_ref]. Although higher peri-infection neutralizing antibody titers were associated with lower infectivity, the levels of neutralizing antibodies in breakthrough cases were not significantly lower than matched uninfected vaccinated controls. Moreover, this analysis does not provide a specific level of antibodies that might be associated with protection [bib_ref] A blood marker predicts who gets 'breakthrough' COVID, Mallapaty [/bib_ref]. Another study reported lower levels of antibodies (S-RBD IgG, 3.469 arbitrary units /ml, AU/ml) in a 41-year-old woman 34-days post complete vaccination [bib_ref] Breakthrough COVID-19 case after full-dose administration of CoronaVac vaccine, Ulhaq [/bib_ref] compared to a previous study [bib_ref] Interpretive discrepancies caused by target values inter-batch variations in chemiluminescence immunoassay for..., Selingerova [/bib_ref]. This patient developed COVID-19 symptoms 40-days post-vaccination. Subsequently, 20-days post-symptom onset, the titer of the spike protein receptor-binding domain (S-RBD) IgG antibodies increased to 130 AU/ml. These results show that the vaccine failed to develop an effective immune response in the patient. In contrast, Hacisuleyman et al. [bib_ref] Vaccine breakthrough infections with SARS-CoV-2 variants, Hacisuleyman [/bib_ref] reported 2 breakthrough cases among 417 mRNA vaccinated individuals (19 and 36 days post-complete vaccination) [bib_ref] Vaccine breakthrough infections with SARS-CoV-2 variants, Hacisuleyman [/bib_ref]. One patient had extremely high titers of neutralizing antibodies. Moreover, the antibodies recognized the variants but were nonetheless insufficient to prevent a breakthrough infection. However, it can't be ruled out that the infection may have occurred before the booster shot took full effect.
## Heterologous vaccination
# Background
In March 2021, vaccinations with ChAdOx1 nCoV-19 were abruptly halted due to VITT [bib_ref] Covid-19: European countries suspend use of Oxford-AstraZeneca vaccine after reports of blood..., Wise [/bib_ref] [bib_ref] Thrombotic thrombocytopenia after ChAdOx1 nCov-19 vaccination, Greinacher [/bib_ref]. The activation of platelet factor 4 (PF4) by antibodies might be amplified by booster vaccination with an adenoviral vector, which might induce and/or aggravate its adverse reactions. In addition, immune responses to the viral vector itself might compromise vaccine efficacy. Thus, boosting with an mRNA-based vaccine have instead been recommended [bib_ref] Immunogenicity and reactogenicity of BNT162b2 booster in ChAdOx1-S-primed participants (CombiVacS): a multicentre,..., Borobia [/bib_ref]. Moreover, uneven availability issues for the approved vaccines around the world also compelled the switch to heterologous vaccination schedules [bib_ref] Heterologous ChAdOx1-nCoV19?NT162b2 vaccination provides superior immunogenicity against COVID-19, Richardson [/bib_ref]. A heterologous prime-boost vaccination (HtPBV) strategy could be an opportunity to make vaccination programs more flexible and reliable in response to fluctuations in supply or demand [bib_ref] Immunogenicity and reactogenicity of BNT162b2 booster in ChAdOx1-S-primed participants (CombiVacS): a multicentre,..., Borobia [/bib_ref]. However, HtPBV has also been evaluated before COVID-19, and in many scenarios, heterologous vaccination has been more immunogenic than homologous prime-boost vaccination (HmPBV) [bib_ref] Safety and Efficacy of NVX-CoV2373 Covid-19 vaccine, Heath [/bib_ref]. In the context of COVID-19, some initial reports demonstrate that HtPBV is better or at least as immunogenic as HmPBV [fig_ref] Table 3: Immune response to heterologous vaccination [/fig_ref].
## Safety and efficacy
Com-COV is a participant-blinded, randomized, phase 2, UK multicenter, non-inferiority study investigating the safety, reactogenicity, and immunogenicity of HtPBV COVID-19 vaccine schedules (interval between first and second shot = 28 days). As per the initial reactogenicity data, both heterologous vaccine schedules (ChAdOx1 nCoV-19-BNT162b2 prime-boost and BNT162b2-ChAdOx1 nCoV-19 prime-boost) induced greater systemic reactogenicity following the boost shot than their homologous counterparts (ChAdOx1 nCoV-19 prime-boost and BNT162b2-prime-boost [bib_ref] Heterologous prime-boost COVID-19 vaccination: initial reactogenicity data, Shaw [/bib_ref]. In this study, up to 80% of individuals receiving a HtPBV reported fatigue and other systemic reactions, an up to 40 times increase compared with the HmPBV. Feverishness was reported by 37 (34%) of 110 recipients of ChAdOx1 nCoV-19 for prime and BNT for boost compared with 11 (10%) of 112 recipients of ChAdOx1 nCoV-19 for both prime and boost. In addition, feverishness was also reported by 47 (41%) of 114 recipients of BNT for prime and ChAdOx1 nCoV-19 for boost, compared with 24 (21%) of 112 recipients of BNT for both prime and boost (difference 21%, 95% CI 8-33%). Similar increases were observed for chills, fatigue, headache, joint pain, malaise, and muscle ache. Most of this increase in reactogenicity was observed in 48 h after immunization. However, there were no hospitalizations due to solicited symptoms. In contrast, a prospective observational cohort study demonstrated no major differences in reactogenicity between the primeboost regimens [bib_ref] Safety, reactogenicity, and immunogenicity of homologous and heterologous prime-boost immunisation with ChAdOx1..., Hillus [/bib_ref]. Between frequent after prime immunization with ChAdOx1 nCoV-19. Reactogenicity of HmPBV (BNT162b2-BNT162b2), HmPBV (ChAdOx1 nCoV19-ChAdOx1 nCoV19), and HtPBV (ChAdOx1 nCoV19-BNT162b2) were similar, with slightly decreased systemic reactions after HtPBV (ChAdOx1 nCov-19-BNT162b2) and HmPBV (ChAdOx1 nCov-19-ChAdOx1 nCoV19). The difference in study design, population demographics, and prime-boost vaccination interval might be responsible for the discrepancy in these two studies. HtPBV induces effective humoral and cellular immune responses in vaccinees [fig_ref] Table 3: Immune response to heterologous vaccination [/fig_ref]. One of the initial studies compared ChAdOx1 nCoV-19, BNT162b2 HtPBV with ChAdOx1 nCoV-19 prime and no boost vaccinated groups [bib_ref] Immunogenicity and reactogenicity of BNT162b2 booster in ChAdOx1-S-primed participants (CombiVacS): a multicentre,..., Borobia [/bib_ref]. RBD antibody-titer, trimeric spike protein antibody titers, and neutralizing antibodies were significantly higher in HtPBV than ChAdOx1 nCoV-19 primed group. Moreover, overnight stimulated whole blood with pools of SARS-CoV-2 spike peptides displayed significantly higher INF-γ levels in HtPBV than ChAdOx1 nCoV-19 primed group. A prospective cohort study evaluated BNT162b2, ChAdOx1 nCoV-19 HmPBV and BNT162b2, ChAdOx1 nCoV-19 HtPBV and deciphered an increased spike S1-reactive T cell responses in HtPBV [bib_ref] Safety, reactogenicity, and immunogenicity of homologous and heterologous prime-boost immunisation with ChAdOx1..., Hillus [/bib_ref]. The geometric means of 50% inhibitory dose against Alpha and Beta variants were highest in recipients of ChAdOx1 nCov-19 BNT162b2 HtPBV compared with the recipients of ChAdOx1 nCov-19 or BNT162b2 HmPBV. Next study evaluated ChAdOx1 nCoV-19-ChAdOx1 nCoV-19 HmPBV with ChAdOx1 nCoV-19-BNT162b2 HtPBV [bib_ref] Immune responses against SARS-CoV-2 variants after heterologous and homologous ChAdOx1 nCoV-19/BNT162b2 vaccination, Barros-Martins [/bib_ref]. In contrast to HmPBV, HtPBV significantly induced higher frequencies of spike specific CD4 + and CD8 + T cells and in particular, induced high titers of neutralizing antibodies against VOC (B.1.1.7 and B.1.351, and P.1). Schmidt et al. [bib_ref] Immunogenicity and reactogenicity of heterologous ChAdOx1 nCoV-19/mRNA vaccination, Schmidt [/bib_ref] reported a significantly higher frequency of activated CD69 + IFN-γ + CD8 + T cells in HtPBV than HmPBV. A single case report deciphered that ChAdOx1 nCoV-19-BNT162b2 HtPBV elicited a robust humoral immune response [bib_ref] Heterologous immunization with Covishield and Pfizer vaccines against SARS-CoV-2 elicits a robust..., Ostadgavahi [/bib_ref] , exceeding the levels reported by Mulligan et al. [bib_ref] Phase I/II study of COVID-19 RNA vaccine BNT162b1 in adults, Mulligan [/bib_ref] in BNT162b2-BNT162b2 HmPBV. Other studies reported a comparable immune responses in BNT162b2-BNT162b2 HmPBV and ChAdOx1 nCoV-19-BNT162b2 HtPBV [bib_ref] Safety and immunogenicity of heterologous versus homologous prime-boost schedules with an adenoviral..., Liu [/bib_ref]. Overall, it seems that HtPBV elicits at least comparable or even better immune responses. In the context of heterologous vaccination, Com-COV study will recruit more individuals to evaluate mRNA-1273 and NVX-CoV2373 mixing [bib_ref] Safety and Efficacy of NVX-CoV2373 Covid-19 vaccine, Heath [/bib_ref]. HtPBV will at least help to counterbalance the shortage of one or more vaccines. However, the durability of such a regime to maintain protection over longer periods should be evaluated. Moreover, the efficacy and effectiveness of HtPBV against variants should be given more attention.
## Hybrid vigor immunity
Immunological memory induced by vaccines is a source of protection against infection. However, the vaccine effectiveness is more or less reduced against VOC [bib_ref] BNT162b2 mRNA Covid-19 vaccine in a nationwide mass vaccination setting, Dagan [/bib_ref] [bib_ref] mRNA-1273 COVID-19 vaccine effectiveness against the B.1.1.7 and B.1.351 variants and severe..., Chemaitelly [/bib_ref] [bib_ref] Effectiveness of the BNT162b2 Covid-19 vaccine against the B.1.1.7 and B.1.351 variants, Abu-Raddad [/bib_ref] [bib_ref] Effectiveness of an Inactivated SARS-CoV-2 Vaccine in Chile, Jara [/bib_ref]. On the other hand, natural infection by SARS-CoV-2 also induces memory immune responses. However, reinfections, especially with variants, including B.1.315 have been reported. What happens when previously infected individuals are vaccinated? The reports from several studies suggest that an impressive synergy results from a combination of natural immunity and vaccine-generated immunity called "hybrid vigor immunity" [bib_ref] Hybrid immunity, Crotty [/bib_ref]. Natural immunity to SARS-CoV-1 or SARS-CoV-2, combined with vaccine-generated immunity, generates broad immune responses. For example, Tan et al. [bib_ref] Pan-sarbecovirus neutralizing antibodies in BNT162b2-immunized SARS-CoV-1 survivors, Tan [/bib_ref] Stamatatos et al. [bib_ref] 2021. mRNA vaccination boosts cross-variant neutralizing antibodies elicited by SARS-CoV-2 infection, Stamatatos [/bib_ref] evaluated sera from 15 individuals who had previously been infected with SARS-CoV-2 and 13 individuals who had not been infected. The sera were collected before and after immunization with one of the mRNA vaccines (BNT162b2 or mRNA-1273). Prior to vaccination, sera from 12 of the 15 previously infected donors neutralized the Wuhan-Hu-1. However, the sera from these individuals showed weak and only sporadic neutralizing activity against the B.1.351. Interestingly, a single shot of vaccine in previously infected individuals with pre-existing virus-specific antibodies induced higher levels of virus-specific IgG and IgA than two vaccine doses in naive individuals. Compared to two vaccine doses in naïve individuals, a single dose of vaccine in previously infected individuals displayed 10-and 20-fold higher levels of neutralizing antibodies to the Wuhan-Hu-1 and B.1.351, respectively. Nevertheless, the serum of previously infected vaccinated individuals was 3 to 10fold less efficient in neutralizing the B.1.351 compared with Wuhan-Hu-1. Moreover, a second dose of the vaccine in the previously infected individuals within 3-4 weeks did not further boost neutralizing antibodies levels. Goel et al. [bib_ref] Distinct antibody and memory B cell responses in SARS-CoV-2 naïve and recovered..., Goel [/bib_ref] evaluated antibody and antigen-specific memory B cells in 33 SARS-CoV-2 naïve and 11 SARS-CoV-2 recovered subjects. Both groups received SARS-CoV-2 mRNA vaccines (BNT162b2 or mRNA-1273). SARS-CoV-2 naïve individuals required both vaccine doses for optimal increases in antibodies. Memory B cells specific for full-length spike protein and the RBD were also efficiently primed by mRNA vaccination and detectable in all SARS-CoV-2 naive subjects after the second vaccine dose. In SARS-CoV-2 recovered individuals, antibody and memory B cell responses were significantly boosted after the first vaccine dose. However, there was no increase in circulating antibodies, neutralizing titers, or antigen-specific memory B cells after the second dose. This robust boosting after the first vaccine dose strongly correlated with levels of pre-existing memory B cells in recovered individuals, identifying a key role for memory B cells in mounting recall responses to SARS-CoV-2 antigens.
In summary, hybrid vigor immunity is a potential field to explore the active components of COVID-19
vaccines. It is interesting to note that currently available vaccines mostly employ the spike protein as immunogen.
Including other viral genome components alongwith the spike in COVID-19 vaccines may mimic the natural virus more closely. And more importantly, development of replication-defective vaccines using the reverse genetics might pave way to better vaccines by inducing and mimicking the hybrid immunity described above. Idenfifying and deletion of viral factors [bib_ref] Chikungunya virus nsP2 impairs MDA5/RIG-I-mediated induction of NF-kappaB promoter activation: A potential..., Bae [/bib_ref] [bib_ref] Zika virus proteins NS2A and NS4A are major antagonists that reduce IFN-beta..., Ngueyen [/bib_ref] [bib_ref] Dose-dependent inhibition of melanoma differentiation-associated gene 5-mediated activation of type I interferon..., Myoung [/bib_ref] [bib_ref] Beyond viral interferon regulatory factors: Immune evasion strategies, Myoung [/bib_ref] [bib_ref] Methyltransferase of a cell culture-adapted hepatitis E inhibits the MDA5 receptor signaling..., Myoung [/bib_ref] [bib_ref] Middle east respiratory syndrome coronavirus-encoded ORF8b inhibits RIG-I-like receptors in a differential..., Lee [/bib_ref] [bib_ref] Middle East respiratory syndrome coronavirus-encoded ORF8b strongly antagonizes IFN-beta promoter activation: its..., Lee [/bib_ref] [bib_ref] Middle East respiratory syndrome coronavirus-encoded accessory proteins impair MDA5-and TBK1-mediated activation of..., Lee [/bib_ref] [bib_ref] Generation of full-length infectious cDNA clones of middle East respiratory syndrome coronavirus, Lee [/bib_ref] , which modulate the host interferfon reponses, need to be considered for the development of next-generation COVID-19 vaccines.
## Future strategies: fractional dosing of vaccines and route of vaccine administration
In the context of COVID-19, various public health and social measures have been implemented to control the transmission of SARS-CoV-2. However, being emergency measures, they are difficult to sustain for longer periods [bib_ref] Fractionation of COVID-19 vaccine doses could extend limited supplies and reduce mortality, Cowling [/bib_ref]. Besides, a shortage in the supply of vaccines is a matter of concern, especially in low-income countries. However, if dose-sparing is effective in preventing symptomatic and severe disease, it would extend the limited supply of vaccines and will play a significant role in bringing the pandemic to an end. More importantly, vaccinating more people with lesser doses may reduce the transmission of the virus, which might reduce the incidence and occurrence of the disease [bib_ref] Fractionation of COVID-19 vaccine doses could extend limited supplies and reduce mortality, Cowling [/bib_ref]. Dose sparing in case of COVID-19 vaccines shall be evaluated to answer a number of questions: Will dose sparing result in an abundant immune response to prevent symptomatic or severe disease and transmission of the virus; how effective will it be against VOC; how safe will it be to administer, including adverse reactions and emergence of new variants; will it be effective in different populations, including immunocompromised individuals? A primising example of successful vaccine dose fractionation is against yellow fever in Angola, the Democratic Republic of Congo. In 2015, in response to the yellow fever epidemic, emergency vaccination was required. However, due to the limited supply of vaccines, WHO's Strategic Advisory Group of Experts on Immunization reviewed the evidence on the immunogenicity and safety of fractional dosing of vaccines against yellow fever and recommended dose fractionation down to one-fifth of the standard dose [bib_ref] Fractional dosing of yellow fever vaccine to extend supply: a modelling study, Wu [/bib_ref]. Fractional dosing was predicted to substantially reduce population infection attack rates and save lives [bib_ref] Fractional dosing of yellow fever vaccine to extend supply: a modelling study, Wu [/bib_ref]. In the context of COVID-19 vaccines, a preliminary study comprising 600 individuals of different age groups evaluated 50 and 100 μg 2-dose regime (mRNA-1273) for safety and immunogenicity [bib_ref] A preliminary report of a randomized controlled phase 2 trial of the..., Chu [/bib_ref]. Anti-SARS-CoV-2 spike binding antibody levels increased substantially by day 14 after the second dose to geometric mean peak levels of 189 (173-207) and 239 (221-259) μg/ml at 50 and 100 μg dose respectively in younger participants (≥18 to <55-years age), and 153 (135-175) and 162 (142-185) μg/ml in older participants (≥55 years age). In addition, neutralizing antibody levels were increased to maximum geometric mean titers of 1733 (1611-1865) μg/ml at 50 μg dose and 1909 (1849-1971) μg/ml at 100 μg dose in younger adults, and 1827 (1722-1938) μg/ml at 50 μg and 1686 (1521-1869) μg/ml at 100 μg in older adults. Although no statistical evaluation was done for antibody levels in participants who received 50 or 100 μg doses, numerical antibody levels seem to be comparable which favors the feasibility of fractional dosing [bib_ref] A preliminary report of a randomized controlled phase 2 trial of the..., Chu [/bib_ref]. In an interim analysis of 4 randomized controlled trials, a subgroup of participants was primed with a half dose of ChAdOx1 nCoV-19 vaccine instead of a full dose, followed by a full-dose boost after a median of 12 weeks [bib_ref] Safety and efficacy of the ChAdOx1 nCoV-19 vaccine (AZD1222) against SARS-CoV-2: an..., Voysey [/bib_ref]. A vaccine efficacy of 90% (67-97%) was reported in this subgroup. Although only a small number of participants were included, the lower bound of 67% for the efficacy estimate is very reassuring [bib_ref] Fractionation of COVID-19 vaccine doses could extend limited supplies and reduce mortality, Cowling [/bib_ref]. However, fractional dosing of COVID-19 vaccines needs to be evaluated in larger populations especially because immune correlates of protection have not been established.
In the UK, a decision was made in December 2020 to delay the second vaccine dose to 12-weeks post-first dose, which aimed to vaccinate more people to develope at least some protection against SARS-CoV-2. A third wave of COVID-19 caused by a highly transmissible Delta variant has led to considerations of the potential need and optimal timing for a second booster shot for vaccinated populations [bib_ref] COVID vaccine boosters: the most important questions, Callaway [/bib_ref]. However, vaccinating more people appears more tempting. Two doses of COVID-19 vaccines are efficient in controlling severe disease, even those caused by VOC [bib_ref] BNT162b2 mRNA Covid-19 vaccine in a nationwide mass vaccination setting, Dagan [/bib_ref] [bib_ref] mRNA-1273 COVID-19 vaccine effectiveness against the B.1.1.7 and B.1.351 variants and severe..., Chemaitelly [/bib_ref] [bib_ref] Effectiveness of the BNT162b2 Covid-19 vaccine against the B.1.1.7 and B.1.351 variants, Abu-Raddad [/bib_ref]. Although there are concerns about waning antibody responses, however, the declining antibody responses do not necessarily mean reduced vaccine efficacy because the effect against disease is not only mediated by antibodies that might be relatively short-lived for some vaccines but also by long-living memory and cellular immune responses [bib_ref] SARS-CoV-2 mRNA vaccines induce persistent human germinal centre responses, Turner [/bib_ref]. For influenza, each annual vaccine is based on the most current data about circulating strains, increasing the likelihood that the vaccine will remain effective even if there is further strain evolution [bib_ref] Considerations in boosting COVID-19 vaccine immune responses, Krause [/bib_ref]. In the sense of COVID-19, there is an opportunity now to study variant-based boosters before there is a widespread need for them [bib_ref] Progress of the COVID-19 vaccine effort: viruses, vaccines and variants versus efficacy,..., Tregoning [/bib_ref]. In this context, Moderna has started clinical trials (NCT04785144) for mRNA-1273.351, targeting novel B.1.351 VOC. The study is divided into 2 cohorts. Cohort 1 who received two vaccinations of mRNA-1273 at dosages of 50 μg, 100 μg, or 250 μg in the Phase 1 clinical trial (DMID 20-0003) will be given a single intramuscular (IM) booster of mRNA-1273.351. Cohort 2, who have never received a COVID-19 vaccine, will be given 2 or 3 IM doses of mRNA-1273.351. Moreover, a multivalent booster candidate mRNA-1273.211 (Combines mRNA-1273 and mRNA-B.1.351) to adult participants who previously received 2 doses of mRNA-1273 (NCT04470427) is currently in Phase 2 and 3 (NCT04927065).
SARS-CoV-2 specific T cells have been detected even in asymptomatic individuals [bib_ref] Robust T cell immunity in convalescent individuals with asymptomatic or mild COVID-19, Sekine [/bib_ref] and those who don't seroconvert [bib_ref] Intrafamilial exposure to SARS-CoV-2 associated with cellular immune response without seroconversion, Gallais [/bib_ref]. T cells can be especially important in convalescents who don't seroconvert or immunocompromised individuals who are less likely to develop an effective antibody response. Sterilizing immunity completely stops viral replication in the host, which can be achieved by antibodies. Among T cells, CD8 + T RM could come closest to sterilizing immunity by eliminating the pathogens at the portal of entry [bib_ref] Characteristics of T-cell responses in COVID-19 patients with prolonged SARS-CoV-2 positivity -a..., Yang [/bib_ref]. The route of COVID-19 vaccine administration shall be given more attention as both route and vaccine formulation are key determinants for T RM formation [bib_ref] Immunological considerations for COVID-19 vaccine strategies, Jeyanathan [/bib_ref] [bib_ref] Location, location, location: Tissue resident memory T cells in mice and humans, Szabo [/bib_ref]. For example, the parenteral route of administration is unable to efficiently induce IgA and T RM in the lungs [bib_ref] New Tuberculosis vaccine strategies: taking aim at un-natural immunity, Jeyanathan [/bib_ref] [bib_ref] A single-dose intranasal ChAd vaccine protects upper and lower respiratory tracts against..., Hassan [/bib_ref] in comparison to intranasal (IN) vaccination. A single IN dose of Chimpanzee adenoviral vaccine encoding stabilized S in mice almost entirely prevented SARS-CoV-2 infection in both the upper and lower respiratory tracts by inducing a mucosal immune response, including high levels of SARS-CoV-2 S specific IgA in serum and lung. Of note, CD103 + CD69 + CD8 + T cells, likely of a resident memory phenotype, were induced by IN route and not by the IM route [bib_ref] A single-dose intranasal ChAd vaccine protects upper and lower respiratory tracts against..., Hassan [/bib_ref]. These results depict that intramuscular vaccination does not confer sterilizing immunity. Eventually, Hassan et al. [bib_ref] A single-dose intranasal ChAd vaccine protects upper and lower respiratory tracts against..., Hassan [/bib_ref] extended their strategy to non-human primates and found that a single dose of IN adenoviral vectored vaccine protects rhesus macaques against SARS-CoV-2. However, in this study, IM and IN routes were not compared. Currently, 7 vaccines are in clinical phase trials which will be administrated by IN route. However, how effective IN vaccination will be, primarily in the long run, need to be evaluated in a more controlled and strict manner. A typical exemplary to understand the immune kinetics of IN immunization is vaccination against Influenza A virus (IAV) [reviewed by [bib_ref] Memory T cell dynamics in the lung during influenza virus infection, Pizzolla [/bib_ref] ]. IAV specific lung T RM provides potent protection against heterosubtypic influenza challenge. However, this protection is transient because of increased apoptosis of T RM in the lung and airways, unlike populations in the skin, nasal tissue, and intestinal mucosae. In this regard, COVID-19 vaccines effectively inducing and stabilizing T RM in the lungs will be an exciting field to explore.
In conclusion, VOC, especially B.1.315 and B.1.617.2, escape the antibody responses. The failure to generate sufficient immune responses might lead to breakthrough cases. However, recommended doses of vaccines are effective against severe diseases and deaths that are of utmost importance in the present scenario.
The uneven availability of COVID-19 vaccines can be tackled by heterologous vaccination, which generates better or at least comparable immune responses. The reports about the adverse reactions of heterologous vaccination are rare and shall be evaluated in larger populations. An emerging concept of hybrid vigor immunity shall be given prime attention. In this context, the inclusion of different SARS-CoV-2 proteins along with spike may provide broader protection against SARS-CoV-2 variants.
[table] Table 1: Breakthrough cases.Ad26.COV2.S vaccinated and 1 mRNA-1273 vaccinated. c An elderly patient with multiple comorbidities who already was on in-home oxygen previous to post-vaccination COVID-19 infection and had a lengthy stay at the ICU. d diagnosis of breakthrough infection in patient 1 and patient 2, 106 and 122 days, respectively, following administration of the 2nd vaccine dose. e One patient had history of diabetes mellitus type 2, high blood pressure, and obesity degree I. [/table]
[table] Table 2: Antibody escape by SARS-CoV-2 variants.particles or 10 11 viral particles), reported durable humoral and cellular immune responses with expanding neutralizing antibody breadth against variants[47]. Individuals receiving a single-shot regimen had a median pseudovirus-neutralizing antibody titer of 272 and 184 against the parental WA1/2020 strain, 167 and 158 against the D614G, 60 and 147 against the B.1.1.7, 39, 107 against the B.1.617.2, 28 and 129 against the P.1, <20 and 62 against the B.1.351 on days 29 and 239, respectively. However, this study has its limitations of including low sample size, use of pseudovirus assay instead of authentic live virus, lack of comparison between different dose regimes, and lack of evaluation of memory B and T cells.Summary of different methodologies used to evaluate the antibody responses in vaccinated indididuals has been provided in [/table]
[table] Table 3: Immune response to heterologous vaccination. [/table]
|
10.1038/s41586-021-04364-8
|
CCBY
| null |
35296842
|
s2orc_pubmed_articles
|
Steady Floquet–Andreev states in graphene Josephson junctions
Engineering quantum states through light-matter interaction has created a new paradigm in condensed matter physics. A representative example is the Floquet-Blochstate, which is generated by time-periodically driving the Bloch wavefunctions in crystals. Previous attempts to realise such states in condensed matter systems have been limited by the transient nature of the Floquet states produced by optical pulses, which masks the universal properties of non-equilibrium physics. Here, we report the generation of steady Floquet Andreev (F-A) states in graphene Josephson junctions by continuous microwave application and direct measurement of their spectra by superconducting tunnelling spectroscopy. We present quantitative analysis of the spectral characteristics of the F-A states while varying the phase difference of superconductors, temperature, microwave frequency and power. The oscillations of the F-A state spectrum with phase difference agreed with our theoretical calculations. Moreover, we confirmed the steady nature of the F-A states by establishing a sum rule of tunnelling conductance, and analysed the spectral density of Floquet states depending on Floquet interaction strength. This study provides a basis for understanding and engineering non-equilibrium quantum states in nanodevices.Light is a powerful tool for tailoring new quantum states by driving them out of equilibrium 1, 2 . One prominent example is Floquet engineering, which involves dynamic control of the properties of quantum matter through a time-periodic electromagnetic drive. Similar to the space-periodic potential that copies Bloch energy bands along the crystal momentum direction, the time-periodic potential copies Floquet states along the energy direction. The Floquet state promises rapid and comprehensive control of the excitation spectrum and topological nature of
physical systems by simply tuning the intensity, frequency and polarisation of the light.
Therefore, there has been a great deal of research effort expended on the realisation and manipulation of a wide variety of long-sought-after quantum states, such as chiral topological orders with no equilibrium counterpart 3 , including Floquet Majorana fermions [bib_ref] Majorana Fermions in equilibrium and in driven cold-atom quantum wires, Jiang [/bib_ref] and a new braiding protocol in the energy dimension [bib_ref] Topologically protected braiding in a single wire using Floquet Majorana modes, Bauer [/bib_ref]. For example, the realisation and control of Floquet dynamics have been reported in both photonic and ultra-cold atomic systems [bib_ref] Interacting Floquet polaritons, Clark [/bib_ref] [bib_ref] Realization of an anomalous Floquet topological system with ultracold atoms, Wintersperger [/bib_ref]. Another important class of platforms are solid-state systems, in which transient Floquet states have been mainly investigated. As Floquet interaction strength is proportional to the ratio of the electric field to the square of the frequency of light [bib_ref] Theoretical description of timeresolved photoemission spectroscopy: application to pump-probe experiments, Freericks [/bib_ref]
## , previous condensed matter experiments realising
Floquet states at the optical frequency of 30-50 THz have applied pulsed lasers to achieve a large electric field of 2.4-4.0 × 10 7 Vm −1 (Refs. [bib_ref] Observation of Floquet-Bloch States on the surface of a topological insulator, Wang [/bib_ref] [bib_ref] Selective scattering between Floquet-Bloch and Volkov states in a topological insulator, Mahmood [/bib_ref] [bib_ref] Light-induced anomalous Hall effect in graphene, Mciver [/bib_ref]. However, Floquet states persist no longer than an order of picoseconds due to the transient nature of pulsed laser and their inherently short lifetime. This short timescale makes it difficult to fully investigate and understand these novel light-driven states and exploit them for practical applications. Another critical issue affecting most experimental realisations of Floquet states is heating by the energy absorption from time-periodic driving. The driving increases the entropy density of the system and eventually brings it to a featureless state with no local correlations [bib_ref] Long-time behavior of isolated periodically driven interacting lattice systems, D'alessio [/bib_ref]. Although a number of ways to reduce or avoid heating effects have been suggested [bib_ref] Effective Hamiltonians, prethermalization, and slow energy absorption in periodically driven many-body systems, Abanin [/bib_ref] [bib_ref] Floquet prethermalization in a Bose-Hubbard system, Rubio-Abadal [/bib_ref] [bib_ref] Thermalization and prethermalization in isolated quantum systems: a theoretical overview, Mori [/bib_ref] [bib_ref] Equilibrium states of generic quantum systems subject to periodic driving, Lazarides [/bib_ref] [bib_ref] Periodically driven ergodic and many-body localized quantum systems, Ponte [/bib_ref] , the heating problem still remains in photonic and ultra-cold gas systems because they are well isolated from the outside environment. On the other hand, condensed matter systems have well-defined electron cooling paths, such as electron-phonon coupling and Wiedemann-Franz cooling of conducting electrons. However, a large electric field in an optical domain, which has been used in previous condensed matter experiments, requires pulsed measurements to avoid heating problems. One 4 strategy to avoid thermal problems is to use lower frequency driving that requires a smaller electric field. There have been experimental studies in the microwave domain, but they relied on indirect methods for probing Floquet states either by tracing their time evolution [bib_ref] Observation of Floquet states in a strongly driven artificial atom, Deng [/bib_ref] [bib_ref] Gigahertz dynamics of a strongly driven single quantum spin, Fuchs [/bib_ref] , qubitresonator resonance conditions 21 , magnetic resonance conditions [bib_ref] Floquet spin states in OLEDs, Jamali [/bib_ref] or AC Josephson effect [bib_ref] Engineering the Floquet spectrum of superconducting multiterminal quantum dots, Melin [/bib_ref] [bib_ref] Simple Floquet-Wannier-Stark-Andreev viewpoint and emergence of low-energy scales in a voltage-biased threeterminal..., Melin [/bib_ref]. process when bias voltage V < 0 is applied between the tunnelling probe and graphene, such that the occupied DOS peak of the tunnel probe matches the empty DOS peak ( + ) of ABS.
e, Colour-coded plot of differential conductance dI/dV as a function of V and B for device 1.
The top horizontal axis shows the superconducting phase difference corresponding to B.
Circles and solid lines represent the dI/dV peaks and corresponding theoretical fittings in a short junction limit, respectively. The line cut at = 0 plotted in the right panel shows dI/dV peaks coming from upper ( + ) and lower ( − ) bands of ABS.
In contrast to previous studies, we report the experimental realisation of truly steady Floquet-Andreev (F-A) states based on Andreev bound states formed in a graphene Josephson junction (GJJ) and their spectra based on direct tunnelling spectroscopy. Here, we generated steady F-A states by continuously applying a monochromatic microwave drive and probed them by superconducting tunnelling spectroscopy with high energy resolution. We investigated the behaviour of the F-A states at various microwave frequencies and powers, and showed that their spectral features agreed well with our theoretical calculations. For example, we corroborated the steady nature of the F-A states by establishing a sum rule of the measured tunnelling conductance. This study clearly demonstrated steady Floquet states and will have a substantial impact on different areas of physics, including topological condensed matter, cold atoms in optical traps and nonequilibrium quantum statistical physics. Our technique, when 6 combined with rapidly developing microwave technologies, can be extended to Floquet-based novel quantum device applications.
A GJJ consists of a non-superconducting graphene layer sandwiched between two superconductors, as shown in [fig_ref] Figure 1 |: Schematics of Andreev bound state and device geometry [/fig_ref]. The GJJ allows the Josephson supercurrent by forming Andreev bound states (ABSs) of electron-and hole-like quasi-particles in the graphene, which are correlated by Andreev reflections at the graphene/superconductor interfaces. In the short junction limit where GJJ channel length L is much shorter than the superconducting coherence length = ℏ / ABS , a single pair of ABSs forms within the superconducting gap of ohmiccontacted Al electrodes ABS and oscillates with the macroscopic quantum phase difference between two superconductors as ± ( ) = ABS √(1 − sin 2 ( /2)) [bib_ref] Relating Andreev bound states and supercurrents in hybrid Josephson junctions, Nichele [/bib_ref] [bib_ref] Tunnelling spectroscopy of Andreev states in graphene, Bretheau [/bib_ref] [bib_ref] Andreev bound states in supercurrent-carrying carbon nanotubes revealed, Pillet [/bib_ref] [bib_ref] Superconducting quantum interference proximity transistor, Giazotto [/bib_ref]. Here, ℏ is a reduced Planck's constant, =10 6 ms −1 is the Fermi velocity of graphene, and D is the contact transparency. The schematic in [fig_ref] Figure 1 |: Schematics of Andreev bound state and device geometry [/fig_ref] shows the oscillation of ABS in a short junction limit and the corresponding density of state (DOS) at a fixed .
In this study, we performed tunnelling spectroscopy on ABS formed in graphene using Al superconducting tunnel contact with the graphene edge, as shown schematically in [fig_ref] Figure 1 |: Schematics of Andreev bound state and device geometry [/fig_ref] (see [fig_ref] Figure 1 |: Schematics of Andreev bound state and device geometry [/fig_ref]. Direct deposition of Al onto the graphene edge forms a sufficiently high potential barrier for the tunnelling probe due to the large inter-atomic distance between graphene and Al atoms [bib_ref] Ultimately short ballistic vertical graphene Josephson junctions, Lee [/bib_ref]. In this structure, = 2π / 0 was controlled by the external magnetic flux threading the superconducting ring in which GJJ was embedded. Here, 0 = ℎ/2 is the superconducting flux quantum with Planck's constant h and electron charge e < 0.
We measured the voltage difference (V) between the Al tunnel probe and graphene while biasing the current (I). The output impedance of the current source (1 GΩ) was much larger than the largest tunnelling resistance (16 MΩ) measured in this experiment. We obtained the tunnelling differential conductance (dI/dV) as a function of V, which represents the convolution of DOS of the ABS in GJJ and that of a tunnel probe [bib_ref] Andreev bound states in supercurrent-carrying carbon nanotubes revealed, Pillet [/bib_ref].
## 9
To generate steady F-A states of GJJ, the devices were continuously irradiated with microwaves and the tunnelling spectrum of graphene was measured using a superconducting tunnel probe.
As shown in , additional peaks in dI/dV emerged above and below the original ABS peaks as the microwave power P increased. The average voltage spacing between the two adjacent dI/dV peaks ΔV = 51.7 ± 11.0 µV at P = -5. Nevertheless, the overall agreement with the sum rule supports the steadiness of F-A states.
We proceeded to calculate the differential conductance dI/dV theoretically, and compared the results with experimental data. Two major factors that determine tunnelling conductance are the spectral weight of F-A states and background differential conductance (dI/dV)BG from the The experimental data gradually deviates from the theoretical fitting at P > −7 dBm ( ≈ 1.7); this can be attributed to, for instance, the significant heating of electrons that leads to a non-Fermi-Dirac distribution of electrons. We now discuss the superconducting phase dependence of F-A states. [fig_ref] Figure 3 |: Phase and frequency dependence of Floquet-Bloch states [/fig_ref] shows synchronised oscillations of the original ABS and Floquet replica states with phase difference , which further confirms that the F-A states are replications of the original ABS. For better visualisation of the oscillations, we subtracted the background obtained by averaging dI/dV over several peaks in V (see . With increasing P, the peak values of the background subtracted differential conductance (d /d ) BS of Floquet replica (n ≠ 0) states became larger, while that of the original ABS (n = 0) became smaller . In addition, the oscillation amplitude, which corresponds to ABS , becomes smaller with increasing P and vanished at P = 0 dBm. This can be explained by the significant electron heating due to strong microwave irradiation, such that electron temperature reaches the critical temperature of the ohmic-contacted Al superconductor (see . Solutions of the timedependent Schrodinger equation of the GJJ under the unpolarised electromagnetic field allowed us to compute the quasi-energy spectrum of the F-A states. We found that our theoretical calculations well reproduced the observed oscillation of the F-A spectrum, as shown in [fig_ref] Figure 3 |: Phase and frequency dependence of Floquet-Bloch states [/fig_ref].
Here, dI/dV for V < 0 represents only the replicas of + ( ), as electrons in the superconducting tunnel probe can hop only to replicas of + ( ) that are empty (see .
Finally, we discuss the microwave frequency dependence of F-A states. The power dependence of (dI/dV)BS at various microwave frequencies is shown in [fig_ref] Figure 3 |: Phase and frequency dependence of Floquet-Bloch states [/fig_ref]. For all frequencies, the decreasing intensity of the peak from the original ABS (n = 0, red arrow) with increasing P is accompanied by increasing intensity of the Floquet replica states (n ≠ 0, white arrows), as expected from the sum rule and Bessel function behaviour. The overall peak position shift to lower voltages with increasing P is attributed to the heating of electrons. We found that the voltage spacing ΔV between adjacent (dI/dV)BS peaks was linearly proportional to f [fig_ref] Figure 3 |: Phase and frequency dependence of Floquet-Bloch states [/fig_ref] with a slope of 4.34 ± 0.46 µV/GHz. This value is close to the slope expected for F-A states,
h/e = 4.14 µV/GHz, confirming the nature of F-A states.
In sum, we have realised the steady F-A states in a Josephson junction device by continuous microwave irradiation without significant heating, and directly measured their energy spectra by superconducting tunnelling spectroscopy. This technique is readily applicable to other lowdimensional topological materials, such as topological insulators and topological semi-metals, for studying and engineering topological Floquet physics [bib_ref] Topologically protected braiding in a single wire using Floquet Majorana modes, Bauer [/bib_ref] Theoretical calculation of differential conductance. We theoretically calculated the timeaveraged differential conductance of the ABS. We first modelled the GJJ without any external driving and obtained the ABS. We then included microwave driving within the graphene region and obtained the Floquet states for a given polarisation of the light. Next, we computed the spectral function and differential conductance of the Floquet states using the Floquet-Kubo formula.Finally, we averaged the differential conductance over the polarisation for comparison with our experiments where the microwaves were unpolarised. For further details, please see and Supplementary Section 9.
We first calculated the ABS spectra and corresponding wavefunctions using standard methods. [bib_ref] Andreev reflection and Klein tunneling in graphene, Beenakker [/bib_ref] [bib_ref] Josephson effect in ballistic graphene, Titov [/bib_ref] Essentially, the model is equivalent to "a (Dirac) particle in a box" under the superconducting boundary conditions. Formally, the Hamiltonian of graphene with superconducting electrodes is given as the Dirac-Bogoliubov-de Gennes (DBdG) equation. [bib_ref] Specular Andreev reflection in graphene, Beenakker [/bib_ref] To reflect the finite width of the GJJ, we allow several conduction modes to appear in the spectrum of ABS. Each mode is identified with a given y-directional momentum, i.e. the momentum along the width of the GJJ. We found that the 12 lowest modes are sufficient to explain the experiment, which gives small y-directional momentums compared to x-directional momentums. We numerically obtained the wavefunctions | , ⟩ and energy spectrum of the Thus, the time-averaged differential conductance is given as
[formula] ⟨ ⟩ = ℏ ∫ ∞ −∞ ∑ [ ( ) − (− )] ∂ T ( ) ( − ℏ ). [/formula]
Derivation of the sum rule of differential conductance. Here, we describe our derivation of a sum rule for differential conductance, by following methods in the literature 3, 28, 31 that discuss closely related Floquet sum rules. The detailed derivation of the sum rule is given in the section 9 of the Supplementary Information.
The sum of the time-averaged differential conductance can be computed as This can be computed using the tunnelling current obtained above, as
[formula] = 2 ℏ | probe | 2 ∫ ∞ −∞ ∑ ( ). [/formula]
Note that ∫ ∞ −∞ ∑ ( ) is the sum of the density of states of the Floquet states; it is known to respect a sum rule [bib_ref] Positivity of the spectral densities of retarded Floquet Green functions, Uhrig [/bib_ref] and should therefore satisfy a sum rule (see Supplementary Section 9-3 for details). Finally, invoking the particle-hole symmetry of the overall spectrum, we conclude that ∫ ∞ 0 ⟨ ⟩ = 1 2 also satisfies the sum rule, which was confirmed in our experimental data.
## Data availability
The data supporting the findings of this study are available from the corresponding author upon reasonable request.
[fig] Figure 1 |: Schematics of Andreev bound state and device geometry. a, Microscopic illustration of an Andreev bound state (ABS) formed in graphene in contact with two superconductors. Within the superconducting gap Δ, phase coherent electron-like (e) and hole-like (h) quasiparticles undergo Andreev reflection and subsequently form an ABS. b, In a short junction limit, a pair of upper ( + ) and lower ( − ) bands of ABS oscillates as a function of the phase difference of two superconductors, φ, as depicted in the left panel. The density of state (DOS) of ABS at a given is shown in the right panel. c, Schematic of the device showing that the graphene (green) is in ohmic contact with aluminium (Al) superconducting electrodes (yellow) 5 that form a Josephson junction, and in tunnel contact with the other Al electrode (blue). The magnetic field (B) threading a superconducting loop controls φ. The graphene Josephson junction has length L = 0.51 µm and width W = 3.0 µm. d, Schematic showing the tunnelling [/fig]
[fig] 8, Figure 2 |: 1d depicts the condition of the maximum dI/dV, where the DOS peak of a tunnel probe matches ABS at the bias voltage = ( Al + + )/ , and where Al is the superconducting gap of the Al tunnel probe (see Supplementary Fig. 2). The sharp peak in DOS of the tunnel probe near Al allows a high energy resolution in tunnelling spectroscopy. Here, we estimated an upper bound of energy resolution (11.0 µeV) with the half-width at half-maximum of background subtracted dI/dV (see Supplementary Fig. 3), which is three orders of magnitude better than that of time-resolved photoemission method 9, 10 .Figure 1e shows dI/dV measured at 20 mK as a function of V and B, the latter of which controls ( ) = 2 ( − 0 ) / 0 with a magnetic field offset B0 of a superconducting solenoid magnet and an effective area of the superconducting ring A. A clear oscillation of the dI/dV peak as a function of is shown, which fits well with our theoretical calculations (see Supplementary Fig. 4). In addition, the particle-hole symmetry of the superconductor manifests itself as a symmetric dI/dV with respect to zero-bias voltage V = 0. Microwave power dependence of Floquet-Andreev states. a, Colour-coded plot of differential conductance dI/dV as a function of bias voltage V and microwave power P, measured in device 2 with microwave frequency f = 12 GHz. Floquet replicas of an Andreev bound state (ABS) indicated by arrows emerge as P increases. b, Schematics of the evolution of the spectral density of Floquet-Andreev (F-A) states at different P. c, Line cuts of the plot in a at various P, from −12 to 0 dBm with 3-dBm intervals. Inset shows the integrated area of dI/dV line cuts at a given P. d Line cuts of the plot in a at dimensionless Floquet interaction strength parameter = 0.64, 0.92, 1.24 and 1.64. Red circles are the experimental data and black lines are the theoretical fittings. Note that each line has an offset of 0.75 . e, dI/dV values corresponding to the F-A states with the index = , ± , ± as a function of P. The colour of each line corresponds to the arrows in a. [/fig]
[fig] 2: dBm corresponds to the microwave energy quanta hf = 49.6 µeV with microwave frequency f = 12 GHz. The error in ΔV is estimated by the half-width at half-maximum of background subtracted dI/dV. This behaviour can be understood by the emergence of Floquet replica states of the ABS, with the microwave irradiation acting as a time-periodic perturbation (Fig. 2b). As P increased, the differential conductance of both the occupied lower band and unoccupied upper band of ABS were replicated to ± + ℎ with integer n = 0, ±1, ±2… In sharp contrast to previous experiments showing transient Floquet-Bloch states 9, 10 , our system supports the steady F-A states generated by continuous Floquet driving, which are well resolved by tunnelling spectroscopy in DC measurements. The steadiness of F-A states is further evidenced by the sum rule of the tunnelling conductance: for a steady Floquet system, several non-trivial sum rules of physical observables have been proposed based on theoretical studies. We showed analytically that experimentally measured dI/dV should satisfy a sum rule, i.e., F-A states are steady (see Methods). The sum rule is based on the fact that the integral of the spectral weights of the F-A states over the energies is a constant 31 . Indeed, S of our experimental data remained almost constant over a broad range of P (see Fig. 2c) and various microwave frequencies (see Supplementary Fig. 5). The slight decrease of S with increasing P shown in the inset of Fig. 2c can be attributed to the DOS of high-energy F-A states leaking out from the finite integration window used for calculating S [−0.5 mV, 0 mV]. [/fig]
[fig] Figure 3 |: Phase and frequency dependence of Floquet-Bloch states. a, Background subtracted differential conductance (d /d ) BS measured in device 2 as a function of bias voltage V and phase difference , with various microwave power P. The microwave frequency is f = 12 GHz. For P ≥ −8 dBm, V is slightly shifted for easier comparison of the data. The theoretical calculation is overlaid as red solid lines. The colour scale ranges of −2 and 0 dBm are changed to [−30 nS, 100 nS] and [−10 nS, 40 nS], respectively. b, (d /d ) BS measured in device 1 as a function of V and P at various f. Original Andreev bound state (ABS) peak and its Floquet replicas are denoted by red and black arrows, respectively. c, The average voltage difference between adjacent peaks (ΔV) in b is plotted as a function f. The errors are estimated by the halfwidth at half-maximum of background subtracted dI/dV. The red dotted line is a linear fit 13 crossing the origin. [/fig]
|
10.2196/jmir.7467
|
CCBY
|
5750416
|
29258975
|
s2orc_pubmed_articles
|
Public Perceptions Regarding Use of Virtual Reality in Health Care: A Social Media Content Analysis Using Facebook
Background: Virtual reality (VR) technology provides an immersive environment that enables users to have modified experiences of reality. VR is increasingly used to manage patients with pain, disability, obesity, neurologic dysfunction, anxiety, and depression. However, public opinion regarding the use of VR in health care has not been explored. Understanding public opinion of VR is critical to ensuring effective implementation of this emerging technology.Objective: This study aimed to examine public opinion about health care VR using social listening, a method that allows for the exploration of unfiltered views of topics discussed on social media and online forums.Methods:In March 2016, NBC News produced a video depicting the use of VR for patient care. The video was repackaged by NowThis, a social media news website, and distributed on Facebook by Upworthy, a news aggregator, yielding 4.3 million views and 2401 comments. We used Microsoft Excel Power Query and ATLAS.ti software (version 7.5, Scientific Software Development) to analyze the comments using content analysis and categorized the comments around first-, second-, and third-order concepts. We determined self-identified gender from the user's Facebook page and performed sentiment analysis of the language to analyze whether the perception of VR differed by gender using a Pearson's chi-square test.Results: Out of the 1614 analyzable comments, 1021 (63.26%) were attributed to female Facebook users, 572 (35.44%) to male users, and 21 (1.30%) to users of unknown gender. There were 1197 comments coded as expressing a positive perception about VR (74.16%), 251 coded as expressing a negative perception and/or concern (15.56%), and 560 coded as neutral (34.70%). Informants identified 20 use cases for VR in health care, including the use of VR for pain and stress reduction; bed-bound individuals; women during labor; and patients undergoing chemotherapy, dialysis, radiation, or imaging procedures. Negative comments expressed concerns about radiation, infection risk, motion sickness, and the ubiquity of and overall dependence on technology. There was a statistically significant association between the language valence of the Facebook post and the gender of the Facebook user; men were more likely to post negative perceptions about the use of VR for health care, whereas women were more likely to post positive perceptions (P<.001).Conclusions: Most informants expressed positive perceptions about the use of VR in a wide range of health care settings. However, many expressed concerns that should be acknowledged and addressed as health care VR continues to evolve. Our results provide guidance in determining where further research on the use of VR in patient care is needed, and offer a formal opportunity for public opinion to shape the VR research agenda.
# Introduction
## Virtual reality in health care settings
Virtual reality (VR) technology provides an immersive environment that enables users to have modified experiences of reality [bib_ref] The effectiveness of virtual reality distraction for pain reduction: a systematic review, Malloy [/bib_ref] [bib_ref] Virtual reality and pain management: current trends and future directions, Li [/bib_ref]. To date, VR has been used in various health care settings to help treat anxiety disorders, reduce fall risks in older patients, control pain, manage obesity, support physical rehabilitation, and distract patients during wound care [bib_ref] The effectiveness of virtual reality distraction for pain reduction: a systematic review, Malloy [/bib_ref] [bib_ref] Virtual reality and pain management: current trends and future directions, Li [/bib_ref] [bib_ref] The effect of virtual reality on pain and range of motion in..., Carrougher [/bib_ref] [bib_ref] Virtual reality distraction for pain control during periodontal scaling and root planing..., Furman [/bib_ref] [bib_ref] A rapid evidence assessment of immersive virtual reality as an adjunct therapy..., Garrett [/bib_ref] [bib_ref] Effectiveness of virtual reality for pediatric pain distraction during i.v. placement, Gold [/bib_ref] [bib_ref] Virtual reality as an adjunctive pain control during burn wound care in..., Hoffman [/bib_ref] [bib_ref] The effectiveness of virtual reality for dental pain control: a case study, Hoffman [/bib_ref] [bib_ref] Feasibility and potential effect of a low-cost virtual reality system on reducing..., Morris [/bib_ref]. By stimulating the visual and proprioceptive senses, VR acts as a distraction to limit the processing of nociceptive stimuli while refocusing the brain on cognitively stimulating, positive, and potentially therapeutic experiences [bib_ref] Virtual reality and pain management: current trends and future directions, Li [/bib_ref]. In a meta-analysis of randomized controlled trials, we found that VR is generally effective and well tolerated by patients across a range of clinical settings, although the existing literature is hampered by small studies of varying quality [bib_ref] Virtual reality and medical inpatients: a systematic review of randomized, controlled trials, Dascal [/bib_ref].
Despite increasing awareness about VR and its potential benefits, it remains unclear whether and how best to scale this technology in clinical practice. There are also questions about whether some patients are willing to accept VR in the clinical settings. We previously assessed the acceptability of VR in hospitalized patients [bib_ref] Feasibility of an immersive virtual reality intervention for hospitalized patients: an observational..., Mosadeghi [/bib_ref] and found that most patients find VR to be a positive and pleasant experience that eases anxiety and provides an escape from the confines of a distressful illness experience. Most patients report a willingness to use VR again if given the opportunity. However, we also found that younger patients are more willing to use VR than older patients; that some patients find the technology uncomfortable, intrusive, or confusing to use; and that patients occasionally report that the headsets are difficult to operate, can induce vertigo, or are of unclear benefit. In short, this study suggests that introducing VR into clinical practice requires careful thought, consideration of patient preferences, and an understanding of the risks and benefits of this emerging technology.
The potential health care uses of VR are far-reaching, but acceptability of the technology is an important consideration toward enabling its successful implementation. Addressing concerns about the safety, effectiveness, usability, and accessibility of VR is critical to ensuring that VR interventions and programs are effective. Moreover, it is important to survey public perception not only about concerns surrounding VR but also about use cases in which VR may be most impactful. Patients should have a voice in determining whether, when, and where to implement VR in their own care.
## Using social listening to examine opinions of virtual reality
Few studies have examined user perceptions of VR, and of these studies, most focused on narrow applications of VR in clinical settings [bib_ref] Using mixed methods to evaluate efficacy and user expectations of a virtual..., Schuster-Amft [/bib_ref] [bib_ref] Training safer surgeons: How do patients view the role of simulation in..., Akhtar [/bib_ref] [bib_ref] Virtual reality games for rehabilitation of people with stroke: perspectives from the..., Lewis [/bib_ref] [bib_ref] Benefits of activity and virtual reality based balance exercise programmes for adults..., Thornton [/bib_ref]. In this study, we used social listening techniques to examine unsolicited comments posted on Facebook in response to an online video about the use of VR in health care. We selected to study the Facebook posts in response to the video given that people use the immensely popular social network to share opinions about various topics with their online communities. Facebook is the largest social network with respect to logged-in users; in June 2017, the site had 2 billion monthly active users. In addition, Facebook was the ideal platform to study public opinions about VR, as the video was posted on the site by an online news aggregator (Upworthy) that specializes in posting viral videos. This presented a natural opportunity to study the Facebook posts about VR without explicitly soliciting opinions from users.
The objective of this study was to use social listening methods to examine how individuals perceive health care VR, including understanding general sentiments about the technology, concerns about the use of health care VR, and which settings may be useful future areas for VR research. The results can elucidate facilitators and barriers to the dissemination and implementation of VR in health care. We use quantitative content analysis to characterize online opinions about the use of VR for patient care.
# Methods
## Study overview
In March 2016, NBC News produced a video depicting the use of VR for patient care based on our research at Cedars-Sinai Medical Center and other sites employing VR in clinical practice. The video was edited and repackaged by "NowThis," a digital news company that distributes content to social media sites, and posted on Facebook by Upworthy, yielding 4.3 million views, 36,000 shares, 67,000 "likes," and 2401 spontaneous comments as of December 2016. The video depicts hospitalized patients using VR headsets, features children using VR to ease the experience of cancer treatment, includes brief interview quotes about how VR can be used in health care, and references the emerging role of VR for managing depression, anxiety, and various types of pain.
In this study, we expand on our previous research [bib_ref] Feasibility of an immersive virtual reality intervention for hospitalized patients: an observational..., Mosadeghi [/bib_ref] by examining the Facebook posts submitted in response to the video. We employ social listening techniques to capture online public opinion. Social listening offers unique advantages compared with traditional survey methods. One of these advantages is the ability to capture unsolicited discussions among participants without a researcher present, thus overcoming the Hawthorne effect, where individuals may change their behavior when they know they are being studied [bib_ref] Systematic review of the Hawthorne effect: new concepts are needed to study..., Mccambridge [/bib_ref]. Social listening can catalogue opinions from large informant groups and from those who might not otherwise participate in a research study. Furthermore, social listening allows researchers to capture data from a wide demographic and geographic spectrum.
Social listening is an established and efficient way to study online communities [bib_ref] Netnography: a method specifically designed to study cultures and communities online, Bowler [/bib_ref]. Previous studies have used Facebook to examine the social media activity of clinicians and pharmacists [bib_ref] How social are we? a cross-sectional study of the website presence and..., Mcevenue [/bib_ref] [bib_ref] Professional use of social media by pharmacists: a qualitative study, Benetoli [/bib_ref] ; to evaluate Facebook groups focused on diabetes and explore the information that patients request, the unsolicited information that is provided, and the nature of the virtual communities that congregate on Facebook [bib_ref] Health information sharing on facebook: an exploratory study on diabetes mellitus, Alqarni [/bib_ref] ; and to examine risk perceptions of obesity among social media users [bib_ref] Obesity is the new major cause of cancer": connections between obesity and..., Kent [/bib_ref]. We build on this methodology by using responses to the VR video posted on Facebook as a virtual focus group to characterize public knowledge, attitudes, beliefs, and preferences about the use of VR in health care. We used qualitative methods to develop a list of themes in the data and used this list to apply the codes to the rest of the Facebook posts. We subsequently quantified the results for analysis.
## Gender and social media use
We examined the gender of the informant, given that previous studies have found that men and women express themselves in different ways on social media. Several studies examining interactions and posts on Facebook found that there are notable variations in the use of language on social media by gender [bib_ref] Personality, gender, and age in the language of social media: the open-vocabulary..., Schwartz [/bib_ref] [bib_ref] Women are warmer but no less assertive than men: gender and language..., Park [/bib_ref]. Men are more likely to use swear words, whereas women are more likely to use positive, emotion-related words such as "excited" [bib_ref] Personality, gender, and age in the language of social media: the open-vocabulary..., Schwartz [/bib_ref] [bib_ref] Women are warmer but no less assertive than men: gender and language..., Park [/bib_ref]. Other studies have found that women are more likely to discuss social relationships (eg, friendships and family), whereas men are more likely to discuss topics such as online gaming, sports, and political topics [bib_ref] Women are warmer but no less assertive than men: gender and language..., Park [/bib_ref]. Additionally, given that various demographic characteristics, including gender, have been found to be important factors in the adoption and diffusion of technology [bib_ref] A longitudinal field investigation of gender differences in individual technology adoption decision-making..., Venkatesh [/bib_ref] [bib_ref] Why don't men ever stop to ask for directions? Gender, social influence,..., Venkatesh [/bib_ref] , we wanted to explore whether the same holds true for the use of VR. Previous studies have found, for instance, that men value the relative advantage and overall usefulness of technology more highly, whereas women tend to value ease of use [bib_ref] Why don't men ever stop to ask for directions? Gender, social influence,..., Venkatesh [/bib_ref].
## Data collection and analysis
We used Microsoft Excel Power Query (2016, Microsoft Corporation) to extract 2401 Facebook comments written in response to the video posted as of 4:20 pm on March 7, 2016. Out of the 2401 comments, we analyzed 67.22% (n=1614) of posts that expressed a measurable sentiment and excluded posts in which users simply tagged friends or included uninterpretable symbols with no other information.
We used the qualitative analysis software ATLAS.ti (Scientific Software Development, Berlin, Germany) to code the Facebook posts. We used ATLAS.ti, given the functionality of the software to support multiple coders and to perform quantitative analyses based on the codes and categories. We used multiple coders (HJP, MEC, and JEF) to enhance the coding process; using multiple coders allowed for the inclusion of multiple perspectives during the code and category development and the ability to discuss coding disagreements among the group. The first round of inductive coding was used to generate a codebook of themes grounded in the data. We used a consensus process to agree on the final list of codes. The coders then iteratively coded the data several times to categorize each of the Facebook posts into the sentiment categories (positive, negative, and neutral) and major (eg, mental health uses and pain relief) and minor (eg, VR as helpful for anxiety, depression, or stress) themes.
Facebook posts could contain more than one theme if they expressed multiple messages or sentiments within the same post. The unit of analysis was the entire Facebook post. We subsequently used ATLAS.ti to generate code count histograms within major and minor themes. We used the sentiment values and major and minor themes to create a map of attitudes, beliefs, and preferences about the use of VR in health care. We compared perceptions and beliefs by informant gender-a demographic variable accessible through each informant's Facebook page. If self-identified gender was missing, then we coded that individual's gender as unknown.
We used the ATLAS.ti code cooccurrence tool to explore patterns among code frequencies regarding gender differences in the categories of Facebook posts, as have been reported in previous research [bib_ref] Data mining emotion in social network communication: gender differences in MySpace, Thelwall [/bib_ref]. We used Pearson chi-square tests to examine the association between gender and sentiment valence of Facebook post.
The study was reviewed and approved by the Cedars-Sinai Medical Center's Institutional Review Board (Pro00044905). No individual subjects were contacted. The only study data used were public posts on Facebook, accessed in full accordance with the Facebook privacy policy.
# Results
## Facebook comments by gender, sentiment, and theme
Out of the 1614 Facebook comments analyzed, 1021 (63.25%) were attributed to female Facebook users, 572 (35.43%) to male users, and 21 (1.30%) to users of unknown gender [fig_ref] Table 1: Sentiment type of Facebook comments by gender [/fig_ref]. Overall, 1197 (74.16%) comments were coded as expressing a positive perception about VR, 251 (15.55%) coded as a negative perception or concern, and 560 (34.70%) coded as neutral. Comments often expressed overlapping themes; thus, the percentage total does not equal 100%. Thematic analysis of the of the comments yielded 50 unique codes, including 27 positive perception codes, 18 negative perception codes, and 5 neutral codes. [fig_ref] Table 2: Distribution of positive beliefs, attitudes, and perceptions toward virtual reality in Facebook... [/fig_ref] lists the 27 codes for positive beliefs, attitudes, and preferences toward VR. Positive perceptions were categorized into the following 2 major groups of first-order concepts: (1) specific health care uses and (2) general uses. The category specific health care uses included the following 5 second-order concepts organized around the use of VR in various settings and conditions: (1) lack of mobility, (2) pain, (3) mental health, (4) treatment and rehabilitation, and (5) drugs. Secondary concepts under general uses included interest in VR technology and desire for personal use and to share with others.
## Positive perception network
The second-order concepts of both primary groups encompass additional third-level concepts specifying positive uses and reactions to health care VR. For example, informants identified both acute and chronic pain conditions that may benefit from VR.
Under specific health care uses, the most common grouping of Facebook posts pertained to using VR to combat a lack of mobility (n=153), followed by managing pain (n=42), and the use of VR to positively influence mental health or for use in mental health treatment (n=31). Informants expressed opinions about how VR could benefit a variety of populations with a lack of mobility, including patients with long-term hospital stays (n=30), elderly patients who cannot move or travel (n=25), children in the hospital (n=15), cancer patients undergoing chemotherapy or radiation treatment (n=9), patients receiving scans or undergoing other types of imaging (n=8), and wheelchair-bound individuals (n=3). In addition, informants believed that VR could be beneficial for populations facing a lack of mobility, distracting individuals from boredom, and encouraging mental stimulation (n=60), as well as encouraging movement (n=3). Specifically, for mental health, informants noted how VR could be used to manage stress (n=14), anxiety (n=10), and depression (n=7).
General positive responses include the following second-order groups: (1) interest in VR technology and (2) desire for personal use and to share with others. Interest in VR technology had third-order concepts of strong general interest (n=665), positive use of technology (n=80), and remarks expressing that VR was better than television or movies (n=9). The most common third-order concept was general interest in VR (n=665), principally comprising nonspecific positive reactions to VR in health care (eg, "cool," "awesome," and "I love this!"). Secondary concepts under desire for personal use and to share with others include individuals expressing interest in VR for their personal use (n=61) or interest in trying VR (N=49), and wishing that VR had been available in their previous hospital stay (n=39) or that the technology was available or had been available for friends and family members in the hospital (n=31).
Acute pain conditions (quaternary concepts) under the second-level concept pain and under the third-level concept acute pain identified by informants include VR benefits for patients in labor and delivery (n=15), dentistry (n=13), burns (n=2), and dialysis (n=1) settings. [fig_ref] Table 3: Distribution of negative beliefs, attitudes, and perceptions toward virtual reality in Facebook... [/fig_ref] lists the 18 codes for negative beliefs, attitudes, and preferences toward VR. Negative perceptions were categorized into the following 2 major groups of first-order concepts: (1) concerns of VR effects in health care and (2) general hesitation/negative reaction. Concerns of VR effects in health care include the following 3 second-order concepts: (1) paranoia or barriers, (2) threats to patient health, and (3) negative effects on specific groups. General hesitation/negative reaction includes 2 second-order concepts, including disinterest and more research necessary to support claims.
## Negative perception network
The second-order concepts mapped to several third-order concepts specifying the negative reactions and concerns elicited by informants. For example, third-order groups under paranoia or barriers revealed user concerns over insurance coverage of VR and costs of use (n=77), VR increasing the societal dependence on technology (n=26), concerns about the information that is transmitted to the user (n=7), VR as a barrier to discharge (n=3), VR creating difficulties in caring for patients (n=2), and concerns about what happens in the room while the patient uses VR (n=2).
The most common codes under concerns of VR effects in health care included paranoia or barriers concerning the use of VR (n=153), followed by VR as a threat to patient health (n=31) and negative effects of VR on specific groups (n=16). Informants identified 5 potential threats to patient health with the use of VR in health care, including motion sickness (n=11), vision complications (n=10), infections due to bacteria and other microorganisms because of sharing equipment (n=4), the potential for radiation from the mobile phone to contribute to cancer (n=3), and concerns about patients falling off the bed when using VR (n=3). Informants expressed concerns about pregnant patients and patients in labor (n=16), psychiatric or traumatic brain injury patients (n=3), and those suffering from mental illness (n=3).
Other users wrote about various general hesitations and reactions, including VR as being generally unnecessary (n=42), the ability for drugs and alcohol to achieve the same goal (n=25), the belief that the idea can be improved or advanced (n=17), and the need for more research to support the claims (n=3).
## Neutral perception network
There were 5 codes for neutral beliefs, attitudes, and preferences toward VR. Neutral responses were grouped into either curious/inquiring responses or prior exposure/knowledge responses. Curious/inquiring responses were grouped into the following second-order responses: (1) wants to share information, (2) long-term outcomes/future uses of VR, and (3) availability outside of the United States. Responses in the first-order group prior exposure/knowledge comprised users who have already heard about this technology and users who compare VR to an existing project or media. The most common second-order concept was wants to share information (n=349). This was followed by the code compares VR to an existing project or media (n=140).
## Type of comments by gender
Pearson's chi-square test was used to determine if there was an association between language valence (positive, negative, or neutral) and gender. Men were significantly more likely to exhibit negative perceptions about the use of VR in health care than women (P<.001; [fig_ref] Figure 1: Distribution of Facebook statements regarding virtual reality technology in health care by... [/fig_ref].
# Discussion
## Principal findings
This case study provides rich data to better understand how the public perceives the use of VR in a health care setting. The results provide innovative, crowdsourced ideas for how to shape VR implementation in health care. For example, the study informants noted that VR could be used in areas such as labor and delivery, dentistry, chemotherapy administration, and in patients undergoing imaging tests. There is preliminary research in some of these areas [bib_ref] The effect of virtual reality during dental treatment on child anxiety and..., Sullivan [/bib_ref] [bib_ref] A pilot and feasibility study of virtual reality as a distraction for..., Gershon [/bib_ref] [bib_ref] Symptoms, functional status, and neuromuscular impairment following carpal tunnel release, Katz [/bib_ref] , and our findings illustrate that there is public interest in using health care VR for these types of applications. Health care organizations and the VR research community may use the resulting thematic network (Multimedia Appendix 1) to help determine where further research on the use of VR in patient care is needed.
A 2012 review examining users' perceptions of VR game-based interventions in the rehabilitation setting noted that users were primarily concerned with the technological limitations of VR, the ability of the user to use the VR system independently, the desire to engage in novel physical and cognitive challenges using VR, the ability to connect with other users using VR during rehabilitation, the ability to receive feedback on progress during rehabilitation sessions, and a desire for rich and varied virtual environments [bib_ref] Virtual reality games for movement rehabilitation in neurological conditions: how do we..., Lewis [/bib_ref]. Interestingly, this study found that although the study informants in our study noted that VR could be used for rehabilitation and to encourage movement, their concerns were quite different from individuals who had used VR in the rehabilitation setting. Although study informants in our study focused on cost, potential technology dependence, and safety issues (eg, falling off the bed), users who had used VR in the health care setting noted different types of concerns, including the desire to use VR in a more social manner, the failure of the technology to meet expectations, the ability to use the technology without assistance, and the desire for more realistic virtual worlds. Actual users of VR were less concerned with overdependence on technology, even those who had used game-based therapies, illustrating that clinicians hoping to use VR may need to address these unfounded fears with new users. This demonstrates that although our findings may be useful in facilitating the dissemination of VR by reducing barriers to use for novice users, more qualitative research is needed to understand how to improve the technology so that it can be used in specific health care settings.
Our findings from this natural experiment indicate that online perceptions of health care VR is generally positive, and most informants believe the technology can benefit various populations in diverse clinical settings. Informants expressed interest in using VR for patients with diminished mobility (eg, those experiencing long-term hospital stays; frail individuals; those receiving chemotherapy, dialysis, radiation, or imaging; and wheelchair-bound individuals) as a drug-free treatment for acute or chronic pain, and for those struggling with mental health conditions such as anxiety and depression. Our findings are in line with those of other studies, which have found that individuals perceive VR as a form of therapy with benefits for the mind and body [bib_ref] Stroke survivors' perceptions of a leisure-based virtual reality program, Farrow [/bib_ref] [bib_ref] Clinician perceptions of virtual reality to assess and treat returning veterans, Kramer [/bib_ref]. Previous studies have found VR to be effective in reducing chronic pain [bib_ref] Feasibility and safety of a virtual reality dodgeball intervention for chronic low..., Thomas [/bib_ref] , aiding in stroke rehabilitation [bib_ref] Virtual reality games for rehabilitation of people with stroke: perspectives from the..., Lewis [/bib_ref] [bib_ref] Stroke survivors' perceptions of a leisure-based virtual reality program, Farrow [/bib_ref] , improving movement in Parkinson's patients [bib_ref] Virtual reality-based training to improve obstacle-crossing performance and dynamic balance in patients..., Liao [/bib_ref] [bib_ref] Using virtual reality to explore the role of conflict resolution and environmental..., Matar [/bib_ref] [bib_ref] Effects of movement imitation training in Parkinson's disease: a virtual reality pilot..., Robles-García [/bib_ref] , treating nicotine addiction [bib_ref] A feasibility study of virtual reality-based coping skills training for nicotine dependence, Bordnick [/bib_ref] , reducing anxiety and post-traumatic stress disorder [bib_ref] Efficacy of virtual reality exposure therapy in the treatment of PTSD: a..., Gonçalves [/bib_ref] , and aiding in the rehabilitation of traumatic brain injury [bib_ref] Using virtual reality and videogames for traumatic brain injury rehabilitation: a structured..., Pietrzak [/bib_ref].
## Concerns about virtual reality and implications for use
However, it is important to acknowledge that many informants in this study expressed concerns about using VR for patient care. There are concerns about costs, whether insurance will cover VR as a therapeutic option, and an overarching concern about increasing dependence on technology. The rising cost of health care is a critical issue overall, and previous research has found that individuals are concerned about the gap between those who can afford experimental treatment and those who struggle to get basic medical care [bib_ref] Using virtual reality and videogames for traumatic brain injury rehabilitation: a structured..., Pietrzak [/bib_ref]. The results from this study illustrate that apprehension about access to and affordability of new technologies is of foremost importance in the public eye. The adoption and dissemination of VR in the health care setting, therefore, will be dependent on how health care systems price this service and whether insurance companies choose to cover the technology and associated VR services. Future studies that examine the cost-effectiveness of adding VR to standard inpatient care will be useful in understanding the added value of VR.
Interestingly, another important barrier to use expressed by informants was the potential of VR to add to the overdependence of technology in society. This finding reveals that individuals feel some trepidation about the current use-and overuse-of technology in society, and that they may not be willing to continue to accept more forms of technology in settings such as health care, despite the potential clinical benefits. Although some individuals may see the use of VR as a way to escape a hospital room, others may be hesitant to use and engage with additional consumer-facing technologies that could prove to be addictive. Researchers should examine whether game-based VR applications, such as those used in rehabilitation settings, have the potential to be addictive. If users are assured that these applications can be used without fears of overuse, they may be more likely to be open to using the technology.
## Gender differences in virtual reality use
We found that female users are more likely to comment positively, whereas male users express more concerns about VR for patient care. Our findings are in line with previous research showing that women express positive sentiments more frequently than men in a social media context [bib_ref] Women are warmer but no less assertive than men: gender and language..., Park [/bib_ref]. We also found that women are more likely to post any opinion about VR, independent of sentiment, which is also consistent with previous research about gender use and social media. Men have been found to be less likely to engage in social media and less likely to use Facebook. Women have also been found to express themselves in warmer, more compassionate, and more polite tones than men on social media sites [bib_ref] Personality, gender, and age in the language of social media: the open-vocabulary..., Schwartz [/bib_ref].
This study also aligns with other research that finds gender differences in attitudes and adoption of technology [bib_ref] Why don't men ever stop to ask for directions? Gender, social influence,..., Venkatesh [/bib_ref] , including the acceptance and use of technologies such as fitness trackers, mobile chat services, video games [bib_ref] Explaining intention to use mobile chat services: moderating effects of gender, Nysveen [/bib_ref] , and Internet use overall [bib_ref] Sex differences in video game play: a communication-based explanation, Lucas [/bib_ref]. Thus, VR, much like other consumer-facing technologies, may be adopted and used by diverse groups in different manners. Researchers studying the diffusion of VR technology should incorporate these gender differences into models of diffusion and should also explore more in depth how different demographic characteristics influence the use and acceptance of VR.
This study has important limitations. First, the video published by Upworthy portrayed VR in a generally positive light, which could bias Facebook users toward commenting more positively. However, our results indicate that the video still prompted a wide range of far-reaching negative comments, indicating a diversity of sentiment. Second, our sample was limited to people who have a Facebook account and who viewed the video on their feed; similar to other surveys, our sample may not express sentiments that represent the larger population. Compared with the general population, informants in this study may be more interested in new technologies or have social networks that view new technologies in a more favorable light. Third, social listening can only capture comments from users who choose to post. Thus, individuals who prefer to view content but do not post are not represented in our analysis. Another limitation is the relatively narrow scope of the study: we focused on responses to one video posted on Facebook. Future research may expand on this study by exploring public opinion about VR based on various sources. Finally, we were limited in the type of demographic data that we could capture through Facebook.
In conclusion, this social listening analysis of a large social media sample indicates wide acceptance of and interest in using VR for patient care but also reveals concerns that should be acknowledged and addressed as health care VR continues to evolve.
## Conflicts of interest
None declared.
# Multimedia appendix 1
Thematic network of sentiment in Facebook comments posted in response to virtual reality (VR) video.
[JPG File, 95KB-Multimedia Appendix 1]
[fig] Figure 1: Distribution of Facebook statements regarding virtual reality technology in health care by gender. [/fig]
[table] Table 1: Sentiment type of Facebook comments by gender. Number of codes by language valence Gender of Facebook user Each comment may include multiple statements; therefore, the number of codes is greater than the number of overall comments. Multimedia Appendix 1 depicts the thematic network encompassing positive, negative, and neutral sentiments. The network indicates that opinion about the role of VR in health care is varied; informants reported diverse attitudes, beliefs, preferences, and concerns about the use of the technology. Multimedia Appendix 2 shows specific examples of Facebook comments left in response to the video. [/table]
[table] Table 2: Distribution of positive beliefs, attitudes, and perceptions toward virtual reality in Facebook comments posted in response to the video. Value (N=1197 b ), n (%) Positive theme of Facebook comment identified through qualitative analysis Desire for personal use and to share with others (N=180) Total does not add up to 100% due to multiple codes per Facebook comment and rounding. J Med Internet Res 2017 | vol. 19 | iss. 12 | e419 | p. 6 http://www.jmir.org/2017/12/e419/ (page number not for citation purposes) [/table]
[table] Table 3: Distribution of negative beliefs, attitudes, and perceptions toward virtual reality in Facebook comments posted in response to the video. Negative theme of Facebook comment identified through qualitative analysisConcerns of VR b effects in health care (N=164)More research needed to support claims a Total does not add up to 100% due to multiple codes per Facebook comment and rounding. [/table]
|
10.1186/s13054-017-1623-3
|
CCBY
|
5356358
|
28302180
|
s2orc_pubmed_articles
|
Assessing preoperative plasma growth-differentiation factor-15 for prediction of acute kidney injury in patients undergoing cardiac surgery
Assessing preoperative plasma growth- differentiation factor-15 for prediction of acute kidney injury in patients undergoing cardiac surgeryAssessing preoperative plasma growth- differentiation factor-15 for prediction of acute kidney injury in patients undergoing cardiac surgery
L E T T E R Open Access See related research by Heringlake et al.
With great interest, we read the recent article published by Heringlake et al. [bib_ref] Preoperative plasma growth-differentiation factor-15 for prediction of acute kidney injury in patients..., Heringlake [/bib_ref] assessing whether preoperative plasma GDF-15 independently predicts postoperative AKI in patients undergoing elective cardiac surgery. Many factors in this study were done well. Other than strict inclusion and exclusion criteria of patients, the authors had also tried to control most of the known risk factors that can affect AKI. We thank the authors for their endeavor to validate that preoperative plasma GDF-15 independently predicts postoperative AKI in patients undergoing elective cardiac surgery. Nevertheless, other than the limitations described in the discussion, we note that several important issues of this study were not addressed.
First, preoperative aspirin therapy was not included in the data analysis. Actually, preoperative aspirin therapy is common among patients undergoing cardiac surgery for primary or secondary prevention of myocardial infarction, stroke, and death. Preoperative aspirin therapy in cardiac surgery has been associated with decreased overall morbidity, mortality, and cardiac surgery-associated acute kidney injury [bib_ref] Preoperative aspirin use and outcomes in cardiac surgery patients, Cao [/bib_ref].
Second, patients receiving valvular heart surgery or coronary artery bypass graft surgery were included in the study. However, some patients may use contrast agents before surgeries. The preoperative contrast angiography or ventriculography is independently associated with increased risks of postoperative adverse renal events [bib_ref] Pathophysiology of contrast medium induced nephropathy, Persson [/bib_ref]. Therefore, contrast angiography or ventriculography should be included in the data analysis.
Third, it was unclear whether serum creatinine levels applied in the diagnosis of postoperative AKI had been adjusted based on the perioperative fluid balance. Moore et al. [bib_ref] The impact of fluid balance on the detection, classification and outcome of..., Moore [/bib_ref] showed that using Acute Kidney Injury Network criteria for diagnosis of AKI, if not adjusting serum creatinine levels for fluid balance, can underestimate the severity and incidence of AKI and can confuse the association of AKI with postoperative adverse outcomes.
Finally, when assessing the association of preoperative plasma GDF-15 with postoperative AKI in patients undergoing cardiac surgery, the available literature shows that postoperative complications including anemia, low cardiac output syndrome, hypoalbuminemia, and sepsis are independent risk factors of AKI after cardiac surgery [bib_ref] Cardiac surgery-associated acute kidney injury, Vivesa [/bib_ref]. We are concerned that not taking postoperative covariates into account would have biased the true effect of preoperative plasma GDF-15 on the occurrence of postoperative AKI in this study. We thank Dr Zhang et al. for their interest in our work [bib_ref] Preoperative plasma growth-differentiation factor-15 for prediction of acute kidney injury in patients..., Heringlake [/bib_ref]. The authors raised concerns that our findings might be confounded by several variables not addressed specifically in our publication.
We are of course aware that every tool used for preoperative risk stratification must have relevant limitations, especially if aiming for the prediction of a multifactorial complication like cardiac surgery-associated acute kidney injury (CSA-AKI). However, we took the greatest care to take into account as many preoperative factors as possible.
Zhang et al. criticized that preoperative treatment with aspirin was not included in our analyses. However, the incidence of AKI in patients preoperatively treated with aspirin was not different (p = 0.720) from the AKI incidence in patients not being treated with aspirin. Consequently this variable was not entered into the multivariate analyses.
We are also aware of the association between immediate preoperative radio contrast medium exposure and postoperative AKI [bib_ref] Acute kidney injury in patients undergoing cardiac surgery and coronary angiography on..., Ranucci [/bib_ref]. To omit this potential confounder we focused only on elective surgery, and excluded urgent patients from the analysis; at least in our institution, elective patients are rarely subjected to an immediate preoperative contrast agent exposure.
We admit that not taking into account the postoperative fluid balance may have lead to underestimation of AKI incidence and severity. Several studies, including that by Moore et al. [bib_ref] The impact of fluid balance on the detection, classification and outcome of..., Moore [/bib_ref] cited by Zhang et al., have shown that adjusting creatinine levels to fluid balance and body weight will increase the probability to detect clinically significant AKI. However, precise data on fluid balance were not available for all patients and thus we could not adjust postoperative creatinine levels accordingly. Nonetheless, the current definitions of AKIdo not urge making such adjustments; and thus our findings are clearly valid according to current AKI criteria (based on creatinine).
We of course agree with Zhang et al.'s comment "that not taking postoperative covariates into account would have biased the true effect of preoperative plasma GDF-15 on the occurrence of postoperative AKI". However, as stated in our manuscript [bib_ref] Preoperative plasma growth-differentiation factor-15 for prediction of acute kidney injury in patients..., Heringlake [/bib_ref] , a tool for preoperative risk stratification will hardly be capable to predict the unpredictable: bleeding anemia (due to poor surgery!), unexpected low cardiac output syndrome (e.g., due to inadequate cardioplegia), postoperative hypoalbuminemia (due to excessive need for crystalloid fluids), and sepsis.
Taken together the authors' comments do not alter our statement that preoperative plasma GDF-15 is an independent predictor of postoperative AKI in patients undergoing elective cardiac surgery and improves the predictive ability of the Cleveland Clinic Acute Renal Failure score.
Abbreviation AKI: Acute kidney injury
|
10.1186/s12938-019-0683-9
|
CCBY
|
6528364
|
31109326
|
s2orc_pubmed_articles
|
Increased cardiorespiratory synchronization evoked by a breath controller based on heartbeat detection
Background: The cardiovascular and respiratory systems are functionally related to each other, but the underlying physiologic mechanism of cardiorespiratory coupling (CRC) is unclear. Cardiopulmonary phase synchronization is a form of cardiorespiratory coupling. However, it is difficult to study in experimental data which are very often inherently nonstationary and thus contain only quasiperiodic oscillations. So how to enhance cardiopulmonary synchronization and quantify cardiopulmonary synchronization, the changes in cardiac function under the conditions of cardiopulmonary synchronization, and the physiological mechanisms behind them are the main issues to be discussed in this paper.Results:The results showed that the cardiorespiratory synchronization significantly increased when breathing was controlled by heartbeat detection (p < 0.001). And the respiratory sinus arrhythmia (RSA) obviously decreased (p < 0.01) in the 2/2 mode and increased (p < 0.001) in the 4/4 mode. During the 2/2 breathing pattern compared with spontaneous breathing, systolic blood pressure (SBP) decreased (p < 0.05), and diastolic blood pressure (DBP), mean arterial blood pressure (MBP), and SV decreased significantly (p < 0.01). During the 4/4 breathing pattern compared to 2/2 breathing patterns, DBP, MBP, and cardiac output (CO) increased (p < 0.05), and stroke volume (SV) increased significantly (p < 0.01). When analyzing the relationships among these parameters, the RSA was found to be associated with the respiration rate in all respiratory patterns.Conclusions:We demonstrated that voluntary cardiorespiratory synchronization (VCRS) can effectively enhance cardiopulmonary phase synchronization, but cardiopulmonary phase synchronization and RSA represent different aspects of the cardiorespiratory interaction. It is found that cardiac function parameters such as the blood pressure and output per stroke could be affected by the number of heartbeats contained in the exhalation and inspiratory phase regulated through VCRS. So we can study cardiopulmonary phase synchronization by VCRS. It can be used to study in experimental data for the physiological mechanism of cardiopulmonary coupling. [bib_ref] 18:61 • fast, convenient online submission • thorough peer review by experienced..., Ren [/bib_ref] Background It is well known that the heart and lung control systems are coupled with each other. Cardiac and respiratory rhythms in humans are synchronized. However, the underlying physiologic mechanism of cardiopulmonary synchronization is unclear. Research on cardiopulmonary synchronization has a certain guiding significance for the diagnosis and treatment of some diseases and for healthcare. Studies have shown that the cardiopulmonary synchronization of athletes and swimmers is superior to that of ordinary people [bib_ref] Heartbeat synchronized with ventilation, Schäfer [/bib_ref] [bib_ref] Synchronization in the human cardiorespiratory system, Schäfer [/bib_ref] , and cardiopulmonary synchronization during the practice meditation is enhanced compare to that during natural breathing [bib_ref] Cardiorespiratory phase synchronization during normal rest and inward-attention meditation, Wu [/bib_ref] [bib_ref] Cardiorespiratory synchronization during Zen meditation, Büssing [/bib_ref]. These results seem to suggest that cardiopulmonary enhancement represents a better physiological state. So how to enhance cardiopulmonary synchronization and quantify cardiopulmonary synchronization, the changes in cardiac function under the conditions of cardiopulmonary synchronization, and the physiological mechanisms behind them are the main issues to be discussed in this paper.
The interaction between the cardiac and the respiratory systems is traditionally identified through the respiratory sinus arrhythmia (RSA), which accounts for the periodic variation of the heart rate within a breathing cycle. With the development of nonlinear dynamics, phase-synchronization analysis technology was used to analyze the relationships in cardiorespiratory coupling [bib_ref] Heartbeat synchronized with ventilation, Schäfer [/bib_ref] [bib_ref] Effects of acute hypoxia on heart rate variability, sample entropy and cardiorespiratory..., Zhang [/bib_ref] [bib_ref] Cardiovascular and cardiorespiratory phase synchronization in normovolemic and hypovolemic humans, Zhang [/bib_ref] [bib_ref] Characterization of cardiorespiratory phase synchronization and directionality in late premature and full..., Lucchini [/bib_ref]. More recently, phase synchronization between heartbeat and breathing has been studied using the synchrogram method [bib_ref] Quantifying cardio-respiratory phase synchronization-a comparison of five methods using ECGs of post-infarction..., Kuhnhold [/bib_ref] , which first applied to check the different synchronous states and conversion rates of cardiorespiratory coupling by Schafer C [bib_ref] Synchronization in the human cardiorespiratory system, Schäfer [/bib_ref]. Hoyer D improved the detection method of phase synchronization [bib_ref] A new approach to uncover dynamic phase coordination and synchronization, Hoyer [/bib_ref]. At present, the cardiorespiratory synchrogram method has been widely used in cardiopulmonary coupling research [bib_ref] Experimental evidence for phase synchronization transitions in the human cardiorespiratory system, Bartsch [/bib_ref] [bib_ref] Cardiorespiratory Phase Synchronization in OSA subjects during wake and sleep states, Sola-Soler [/bib_ref] [bib_ref] Cardiorespiratory Phase Synchronization increases during certain mental stimuli in healthy subjects, Sola-Soler [/bib_ref].
In this paper, we investigated the physiological mechanism of cardiopulmonary coupling based on enhancing cardiopulmonary synchronicity using the heartbeat to control breath. The only method of controlling respiratory-induced variation before 1964 was for subjects to hold their breath. In 1964, Schmitt was first put forward controlling breathing using synchronization with an electrocardiogram/vector cardiogram (ECG/VCG), which was called voluntary cardiorespiratory synchronization (VCRS) [bib_ref] Voluntary cardio-respiratory synchronization, Patterson [/bib_ref]. VCRS has been applied to study the changes in heart rate variability with human age and body position [bib_ref] Heart rate change as a function of age, tidal volume and body..., Patterson [/bib_ref] , to investigate respiratory effects on stroke volume [bib_ref] Respiratory effects on cardiac related impedance indices measured under voluntary cardio-respiratory synchronisation..., Wang [/bib_ref] , and to study the influence of respiration on changes in blood pressure and heart rate [bib_ref] Determining the relationship of heart rate and blood pressure using voluntary cardio-respiratory..., Mason [/bib_ref]. However, the enhancement of cardiopulmonary synchrony by VCRS has not been quantified and verified, and it is not discussed whether cardiopulmonary synchronization time is the influencing factor of cardiac function. These are important to further reveal the cardiopulmonary coupling mechanism.
Our aim is to use VCRS to study respiratory effects on blood pressure, heart rate, stroke volume, and cardiac output, using the phase-synchronization technique to quantify cardiopulmonary synchronization time. On the basis of synchronicity enhancement, we analyzed the effects of cardiopulmonary synchronicity on heart function and discussed the underlying physiologic mechanism of cardiorespiratory coupling.
# Methods
## Subjects
Thirty healthy male subjects (19-27 years old) voluntarily participated in the study. Each subject was given a medical survey questionnaire to ensure that they had no respiratory and cardiovascular diseases. The investigation was performed with the approval of the Xi'an Jiaotong University Ethics Committee, and all subjects signed an approved informed consent after the study procedures had been explained.
## Experimental protocol
The experiment was performed in a quiet laboratory between 8 p.m. and 10 p.m. and the temperature was controlled to 22-24 °C. The subjects were asked to avoid alcohol, tea, coffee, and strenuous exercises for 12 h and to get enough sleep the night before the experiment. Participants were seated comfortably. Before measurements were taken, participants rested and remained calm for 10 min. Then, the subjects received instructions for breathing in the sitting position. Three sets of data were tested with a rest period of 10 min between each set of measurements. The first measurement was made with subjects breathing spontaneously for 6 min. The subjects then exercised for 2 to 4 min to breath pace the VCRS, and the researchers made sure there were no problems during the procedure. The researcher was asked to follow the paced breathing VCRS mode for 6 min using a sound pattern from the ECG. This breath controller was developed in our laboratory. It outputs a sound signal by detecting the heartbeat signal and counting the heartbeats. The subject was instructed by the sound to breathe, synchronized with his/her heartbeat. The device signaled the subject when to inhale and exhale based on a fixed number of heart beats for each phase of the respiratory cycle: for example, inspire for the first two heartbeats and expire for the next two heartbeats in a 2/2 pattern. The 2/2 pattern was selected as the second measurement mode for 6 min. [fig_ref] Figure 1: The relationship between ECG and respiration signals of a subject using the... [/fig_ref] shows the ECG and respiration signals of a subject using this 2/2 breathing pattern. Another measurement mode was the 4/4 paced breathing VCRS pattern for 6 min. [fig_ref] Figure 1: The relationship between ECG and respiration signals of a subject using the... [/fig_ref] shows an example of this 4/4 breathing pattern. The 2/2 and 4/4 breathing patterns were chosen to increase and decrease the respiration rate, respectively, but still be subjectively comfortable for the participants.
## Physiological measurements
Respiration, ECG, and thoracic electric bioimpedance (TEB) signals were acquired online using a multichannel physiology recorder (MP150, Biopac, USA) and software (AcqKnowledge 4.2, Biopac systems) on a PC (Gateway) and collected at 1000 samples per second. Beat-to-beat arterial blood pressure was logged continuously via a noninvasive finger photoplethysmography (FMS, Finapres Measurement Systems, Arnhem, Netherlands). The synchronization method between the two instruments involves attaching the output interface (BNC) of the FMS to an analog input channel of the MP150.
The ECG was measured using disposable self-adhesive Ag/AgCl ECG electrodes, which were placed on the subject's right arm, left leg, and right leg. The connection method was a standard bipolar ECG lead II that the left leg was connected to during the in-phase input of the amplifier, with the right arm connected to reverse-phase input, and the right leg linked to ground wire.
The TEB was recorded using NICO100C (Biopac system, Inc.). The NICO100C noninvasive cardiac output amplifier records the parameters associated with cardiac output measurements. It incorporates a precision high-frequency current source, which injects a very small (400 μA) measurement current through the thoracic volume, which is defined by the placement of a set of current source electrodes.
Respiration was recorded with a respiratory effort transducer strain assembly belt TSD201 (Biopac Inc.) that measures thoracic expansion and contraction. Voluntary cardiorespiratory synchronized breathing (VCRS) was controlled with the device that generated a sound to instruct the subject when to inhale and exhale, which was introduced above.
# Data analysis
## Phase synchronization and synchrogram
Phase synchronization is a kind of cooperative performance between two weakly interacted oscillators. It is classically understood as phase locking and defined as [bib_ref] Synchronization and modulation in the human cardiorespiratory system, Lotrič [/bib_ref] :
where n and m are integers, Φ 1 and Φ 2 are phases of the two oscillators. The n : m phase locking manifests as a variation of ϕ n,m around a horizontal plateau. The synchrogram is a visualization tool which enables the detection of synchronization epochs in bivariate data. The synchrogram is constructed by plotting the corresponding normalized respiratory phase ψ m (t k ) for every heart beat within m respiratory cycles [bib_ref] Effects of mental tasks on the cardiorespiratory synchronization, Zhang [/bib_ref] :
where t k is the time of the k th R peak and Φ r is the corresponding respiratory phase.
The respiratory phase, Φ r (t k ) , is calculated by the method based on the marker events as the following formula:
where t k is the time of the onset of k th expiration. The local maximum of respiratory signal is taken for the marker events of respiratory oscillator. In a perfect m : n phase locking, ψ m (t k ) exactly attains the same n different values within m adjacent respiratory cycles, and the synchrogram consists of n horizontal strips.
The method of phase recurrences based on a heuristic approach was used to quantify the cardiorespiratory synchrogram. Detection with phase recurrences offered the best temporal resolution and the highest number of synchronized sequence [bib_ref] A quantitative comparison of different methods to detect cardiorespiratory coordination during night-time..., Cysarz [/bib_ref]. Generally, a n : m synchronization will be identified if the difference between the normalized relative respiratory phase corresponding to the (i + n) th R peak and the one corresponding to the i th R peak is within a defined tolerance ε. This condition has to be fulfilled for at least k successive R peaks:
where N r is the total number of R peaks. To be compatible with the description of parallel horizontal lines during synchronization, k ≥ m needs to be fulfilled. This procedure allows a detection of a structure of parallel horizontal lines with a length of 2 m successive normalized relative phases.
It is most valuable in cases where one of the signals resembles a point process. The synchrogram is a stroboscopic view of the phase of the respiration signal at the times of R waves. R peaks were detected using the combination of wavelet transforms and thresholding methods [bib_ref] R peak detection in electrocardiogram signal based on an optimal combination of..., Rabbani [/bib_ref]. The peaks and troughs at breathing signal were extracted using the algorithm of threshold value. The times of R peaks at ECG and inspiratory onsets at respiratory signal were obtained and served as event markers for synchrogram, as shown in [fig_ref] Figure 3: Typical cardiorespiratory synchrogram of one subject [/fig_ref]. The duration of the synchronization epochs is calculated, which is used as the index of synchronization strength in the article.
[formula] (1) ϕ n,m = |nΦ 1 − mΦ 2 − δ| < const, (2) ψ m (t k ) = 1 2π (Φ r (t k ) mod 2mπ), (3) Φ r (t k ) = 2π t − t k t k+1 − t k + 2π k, t k ≤ t < t k+1 ,(4)∃k > 1, |ψ m (t i+n ) − ψ m (t n )| < ε, i ∈ l, . . . , l + k − 1, 0 ≤ l ≤ N r − k + 1 , [/formula]
## Respiratory sinus arrhythmia (rsa)
Respiratory Sinus Arrhythmia is used to explore the connection between respiration and changes to heart rate. This RSA index is computed using the peak-valley method [bib_ref] A comparison of three quantification methods for estimation of respiratory sinus arrhythmia, Grossman [/bib_ref]. This method uses both a recorded ECG Lead II signal and a respiration signal. The respiration signal is used to locate periods of inhalation and exhalation. Inhalation begins at valleys in the signal while expiration at peaks. RSA index is the average difference between the highest and lowest heart rate (HR) during each respiratory cycle (HR Max − HR Min), expressed in milliseconds.
## Arterial blood pressure
The peaks and troughs of the blood pressure signal were extracted for systolic and diastolic blood pressures by combining them with the ECG signal. The mean arterial blood pressure (MAP) (= (2 × diastolic)/3 + systolic/3) was calculated from systolic and diastolic blood pressures. The method of extracting feature points in the blood pressure signal incorporated into the ECG signal is an advantage of synchronous collection of blood pressure and ECG signals. The maximum and minimum of the blood pressure signal matched the cardiac cycle between two adjacent R wave peaks. Each cardiac cycle includes one systolic blood pressure and one diastolic blood pressure. The algorithm has a high detection rate, as shown in [fig_ref] Figure 2: The peaks and troughs in the blood pressure signals of a subject [/fig_ref].
## Stroke volume and cardiac output
The stroke volume (SV) and cardiac output (CO) can be derived from the popular noninvasive method of impedance cardiography. CO is defined as the volume of blood pumped each minute ( = SV × HR ). SV can be calculated from the Kubicek formula [bib_ref] Respiratory effects on cardiac related impedance indices measured under voluntary cardio-respiratory synchronisation..., Wang [/bib_ref] : where SV is stroke volume, i.e., volume of blood pumped by left ventricle in a single beat (mL beat), ρ is blood resistivity at 100 kHz (typical value for a normal hematocrit = 150 Ω cm), L is mean distance between inner pair of electrodes (cm), Z 0 is the average basal thoracic impedance (Ω), LVET is left ventricular ejection time(s), (dZ/dt) max is the maximum value of ( dZ/dt ) signal (Ω s −1 ), and ( dZ/dt ) is the derivative of cardiac systolic impedance.
[formula] (5) SV = ρ L 2 Z 2 0 dZ dt max × LVET, [/formula]
# Statistical analysis
Statistical analyses were performed with SigmaPlot software (Systat Software, Inc. USA).
If the data were normally distributed, paired t tests were used. Otherwise, rank sum tests were used. A p value < 0.05 was statistically considered significant, and data were represented as the mean ± SEM. Pearson's correlation coefficients were used for the correlation analyses.
# Results
## Changes in the cardiorespiratory synchronization time (syn), rsa, and breath rate (br)
Compared with spontaneous breathing, Syn was significantly increased (p < 0.001 for the 2/2 breathing pattern, p < 0.01 for the 4/4 breathing pattern). RSA obviously decreased (p < 0.01) in the 2/2 mode and increased (p < 0.001) in the 4/4 mode. In contrast, the BR was significantly increased (p < 0.001) during the 2/2 breathing pattern, and it was significantly decreased (p < 0.001) during the 4/4 breathing pattern [fig_ref] Table 1: Syn and BR during spontaneous breathing and 2/2 and 4/4 breathing patternsData... [/fig_ref]. The results showed that cardiorespiratory synchronization was significantly increased by breath control based on heartbeat detection. A typical cardiorespiratory synchrogram of one subject is shown in [fig_ref] Figure 3: Typical cardiorespiratory synchrogram of one subject [/fig_ref]. During spontaneous breathing, synchronization was found in 26 subjects, and 4 subjects did not exhibit any synchronization. The average synchronization epoch was [bib_ref] Effects of mental tasks on the cardiorespiratory synchronization, Zhang [/bib_ref].79 s in 360 s, which was 5.50% of the total recordings. During the 2/2 breathing pattern, synchronization was found in 29 subjects, and 1 subject did not exhibit any synchronization. The mean synchronization epoch was 84.43 s in 360 s, which indicated that heartbeat and breath rate were synchronized in 23.45% of the total recording; this result was far longer than that of spontaneous breathing. During the 4/4 breathing pattern, the average synchronization epoch in all 24 subjects was 60.85 s in 360 s, which was 16.90% of the total recording; this result was far longer than that of spontaneous breathing. Furthermore, the breath rate was in accordance with the breathing controller based on heartbeat detection. These results suggest that the levels of cardiorespiratory coupling were significantly higher in the 2/2 and 4/4 breathing modes than during spontaneous breathing.
## Effect on blood pressure, hr, sv, and co
To determine the effect of breath triggered by ECG enhancing cardiopulmonary synchronicity on the cardiovascular system, we analyzed cardiovascular function parameters during the 2/2 and 4/4 breathing patterns relative to those during spontaneous breathing pattern and between the 2/2 and 4/4 breathing patterns. These cardiovascular function parameters include blood pressure (SBP, DBP, and MBP), heart rate (HR), stroke volume (SV), and cardiac output (CO). During the 2/2 breathing pattern compared with spontaneous breathing, SBP decreased (p < 0.05), and DBP, MBP, and SV decreased significantly (p < 0.01). During the 4/4 breathing pattern compared with spontaneous breathing, all of the parameters showed no obvious change. During the 4/4 breathing pattern compared to the 2/2 breathing pattern, SBP, DBP, MBP, and CO increased (p < 0.05), and SV increased significantly (p < 0.01); no obvious difference was found in SBP and HR.
## Correlation among syn, rsa, bp, and breath rate
To study the mechanism of cardiopulmonary coupling, we analyzed the correlation coefficients among the cardiopulmonary synchronization time and blood pressure, cardiac output, stroke output, and respiration rate. No significant correlation was found. When analyzing the correlation between RSA and these parameters, RSA was found to be associated with the respiration rate and BP in the 2/2 mode [fig_ref] Figure 4: Correlation analysis of the power spectral energy of HRV and breath rate [/fig_ref]. RSA was significantly negatively correlated with the respiration rate in all the respiratory mode (Spt, r = − 0.535, p = 0.002; 2/2 mode, r = − 0.741, p = 0.000003; r = − 0.757, p = 0.000001). RSA was obviously negatively correlated with DBP and MBP in the 2/2 respiratory mode (r = − 0.551, p = 0.002; r = − 0.468, p = 0.009). No obvious correlation was found in spontaneous breathing and 4/4 breath pattern.
# Discussion
Phase synchronization can be used to analyze cardiopulmonary coupling [bib_ref] Heartbeat synchronized with ventilation, Schäfer [/bib_ref] [bib_ref] Synchronization in the human cardiorespiratory system, Schäfer [/bib_ref] [bib_ref] Experimental evidence for phase synchronization transitions in the human cardiorespiratory system, Bartsch [/bib_ref]. We analyzed the synchronization time of the heart and respiration in 2/2, 4/4, and spontaneous breathing patterns by using phase synchronous graphs and found that controlling respiration by heartbeat can significantly enhance cardiopulmonary coupling and control cardiopulmonary synchronization. Compared with spontaneous breathing, the synchronous time of the heart rate and respiration in 2/2 and 4/4 modes was significantly enhanced, the respiratory rate increased significantly during the 2/2 pattern, and the respiratory rate decreased significantly during the 4/4 pattern [fig_ref] Table 1: Syn and BR during spontaneous breathing and 2/2 and 4/4 breathing patternsData... [/fig_ref]. In addition, compared to the 2/2 breathing mode, the breathing rate was significantly reduced in the 4/4 breathing mode, but the synchronization time did not change significantly. These results show that the synchronization time is inconsistent with the change in respiration rate, which is agreement with Ref. [bib_ref] Phase transitions in physiologic coupling, Bartsch [/bib_ref]. So the enhancement of synchronization time is not caused by the change of respiratory rate. At the same time, the method of adjusting phase synchronization by VCRS is to control the initial phase and the ratio of heartbeat to respiratory cycle through heartbeat, which is important for enhancing phase-synchronization time [bib_ref] Experimental evidence for phase synchronization transitions in the human cardiorespiratory system, Bartsch [/bib_ref]. This showed that VCRS can effectively enhance cardiopulmonary synchronization and control the number of heartbeats contained in the exhalation and inspiratory phase of the respiratory cycle. It provides experimental conditions for studying the physiological mechanism of cardiopulmonary coupling.
In this paper, we used VCRS to study the circulatory system and found that blood pressure and stroke volume were reduced in the 2/2 mode by VCRS. Through analysis, we think that this is related to the number of heartbeats contained in the exhalation and inspiratory phase regulated by VCRS. Compared with spontaneous breathing, the 2/2 respiratory pattern resulted in decreased blood pressure and stroke volume, and there were no significant changes in mean HR because vagal tone did not change. During the 2/2 respiratory pattern, the ratio of inhalation time to exhalation time of the respiratory cycle was approximately 1:1, and both the inhalation and exhalation phases contained two ECG cycles. In spontaneous breathing mode, the inhalation phase contained two ECG cycles, and the exhalation phase contained three ECG cycles [bib_ref] Basic technology of voluntary cardiorespiratory synchronization in electrocardiology, Almasi [/bib_ref] , so the ratio of inhalation time to exhalation time was approximately 1:1.5. Compared with spontaneous breathing, the duration of inhalation increased in the whole respiratory cycle in the 2/2 breath mode. When inhaling, the decrease in pleural pressure resulted in increased right ventricular filling, which decreased left ventricular filling and stroke volume and further reduced systolic blood pressure. Exhalation has the opposite effect and was accompanied by a delayed effect of the increased right ventricular stroke volume due to inhalation. The maximum left ventricular stroke volume occurred during the posterior half of the exhalation phase. When the breathing amplitude increased, namely, the amount of the moisture increased, the magnitude of change in pleural pressure increased, and the effect became more pronounced [bib_ref] Pressure-flow studies in man: effect of respiration on left ventricular stroke volume, Ruskin [/bib_ref] [bib_ref] A model of mechanical interactions between heart and lungs, Fontecave [/bib_ref]. Therefore, the blood pressure and stroke output decreased in the 2/2 mode relative to those during spontaneous respiration. The maximum left ventricular stroke volume occurred during the latter half of the exhalation phase, although the time ratio of the inhalation phase to the exhalation phase was 1:1 in the 4/4 mode, the exhalation time had 4 ECG cycles. The exhalation time was sufficient compared to that of the 2/2 mode and showed little difference compared to that of spontaneous breath. Therefore, in the 4/4 mode, blood pressure, stroke, and cardiac output increased compared to the 2/2 mode, and there was no significant change compared to spontaneous breath. It indicates that blood pressure and output per stroke could affected by changes in chest pressure. In our study, it showed that the decrease in blood pressure and output per stroke is related to the mode by VCRS.
Our results demonstrate that cardiopulmonary phase synchronization and the traditionally studied respiratory sinus arrhythmia represent different aspects of the cardiorespiratory interaction. This is consent to previous studies [bib_ref] Cardiorespiratory Phase Synchronization increases during certain mental stimuli in healthy subjects, Sola-Soler [/bib_ref] [bib_ref] Phase transitions in physiologic coupling, Bartsch [/bib_ref] [bib_ref] Three independent forms of cardio-respiratory coupling: transitions across sleep stages, Bartsch [/bib_ref]. In this study, the strength of phase synchronization is represented by phase-synchronization time, and the strength of RSA is represented by HR Max − HR Min. We investigated the relationship between phase synchronization and RSA. Our analyses did not reveal a statistical relation between the degree of cardiopulmonary phase synchronization and the strength of RSA. RSA was affected by respiration rate [fig_ref] Figure 4: Correlation analysis of the power spectral energy of HRV and breath rate [/fig_ref]. However, cardiopulmonary phase-synchronization time was not correlated with respiration rate. And the changes of cardiopulmonary phase-synchronization time are different with RSA in all breath patterns [fig_ref] Table 1: Syn and BR during spontaneous breathing and 2/2 and 4/4 breathing patternsData... [/fig_ref]. These differences are due to the fact that, whereas RSA is a measure of the amplitude of variation of the heartbeat intervals within the breathing cycles, phase synchronization is characterized by the clustering of heartbeats at specific phases in the breathing cycle. This clustering is independent of the amplitude of heart rate modulation [bib_ref] Phase transitions in physiologic coupling, Bartsch [/bib_ref]. Cardiorespiratory phase synchronization is a type of cardiorespiratory coupling that manifests through a predilection for heart beats to occur at specific points relative to the phase of the respiratory cycle [bib_ref] On the difference of cardiorespiratory synchronisation and coordination, Krause [/bib_ref] , which can be controlled by VCRS. However, the RSA is quite obviously another manifestation of the cardiorespiratory interaction [bib_ref] Cardiorespiratory coupling in young healthy subjects, Sobiech [/bib_ref]. In our study, RSA was significantly changed in all respiration modes, which was correlated with breath frequency and BP [fig_ref] Figure 4: Correlation analysis of the power spectral energy of HRV and breath rate [/fig_ref]. However, it is related to blood pressure only in 2/2 breathing mode. It is probable that RSA is mainly affected by breath frequency. Cardiopulmonary phase synchronization and RSA are different forms of the
[fig] Figure 1: The relationship between ECG and respiration signals of a subject using the 2/2 and 4/4 breathing patterns. Above is the ECG signal, and below is breathing signal. a ECG and respiration signals of a subject in the voluntary cardiorespiratory synchronization 2/2 mode, during which the subject inspired for two heartbeats and expired for two heartbeats in one respiration cycle. b ECG and respiration signals of a subject in the voluntary cardiorespiratory synchronization 4/4 mode, during which the subject inspired for four heartbeats and expired for four heartbeats. The top trace shows the ECG; the bottom trace shows respiration measured by a respiration belt Ren and Zhang BioMed Eng OnLine (2019)18:61 [/fig]
[fig] Figure 2: The peaks and troughs in the blood pressure signals of a subject. Systolic blood pressure (SBP) calculated by the peak is marked with cross symbols. Diastolic blood pressure (DBP) calculated by the trough is marked with circle Ren and Zhang BioMed Eng OnLine (2019)18:61 [/fig]
[fig] Figure 3: Typical cardiorespiratory synchrogram of one subject. Typical cardiorespiratory synchrograms of one subject for spontaneous breathing (a), the 2/2 breathing pattern (b), and the 4/4 breathing pattern (c). Synchronization, marked with cross symbols, is characterized by the arrangement of the wrapped phase in horizontal lines. In this subject, 4:1 phase synchronization was detected for both 2/2 breathing pattern and spontaneous breathing, but the synchronization epoch in 2/2 breathing pattern obviously longer. 8:1 phase synchronization was detected for the 4/4 breathing pattern Ren and Zhang BioMed Eng OnLine (2019)18:61 [/fig]
[fig] Figure 4: Correlation analysis of the power spectral energy of HRV and breath rate. Scatterplots for RSA and respiratory rate during spontaneous breathing (Spt), the 2/2 breathing pattern (2/2), and the 4/4 breathing pattern (4/4) (a). Scatterplots for RSA and DBP during spontaneous breathing (Spt), the 2/2 breathing pattern (2/2), and the 4/4 breathing pattern (4/4) (b). Scatterplots for RSA and MBP during spontaneous breathing and the 2/2 and 4/4 breathing modes (c) Ren and Zhang BioMed Eng OnLine (2019) 18:61 [/fig]
[table] Table 1: Syn and BR during spontaneous breathing and 2/2 and 4/4 breathing patternsData are presented as the mean ± SEM Spt, spontaneous breathing; 2/2, 2/2 breathing pattern; 4/4, 4/4 breathing pattern; Syn, synchronization time; RSA, HR Max minus HR Min; BR, breath rate ** and *** p < 0.01 and p < 0.001, respectively, vs the spontaneous breathing pattern (paired t test). ### p < 0.001, respectively, for a comparison of the = 4/4 breathing pattern and the 2/2 breathing pattern [/table]
|
10.3389/fnmol.2020.00031
|
CCBY
|
7094161
|
32256312
|
s2orc_pubmed_articles
|
An Alternative Pin1 Binding and Isomerization Site in the N-Terminus Domain of PSD-95
Phosphorylation-dependent peptidyl-prolyl cis-trans isomerization plays key roles in cell cycle progression, the pathogenesis of cancer, and age-related neurodegeneration. Most of our knowledge about the role of phosphorylation-dependent peptidyl-prolyl cis-trans isomerization and the enzyme catalyzing this reaction, the peptidyl-prolyl isomerase (Pin1), is largely limited to proteins not present in neurons. Only a handful of examples have shown that phosphorylation-dependent peptidyl-prolyl cis-trans isomerization, Pin1 binding, or Pin1-mediated peptidyl-prolyl cis-trans isomerization regulate proteins present at excitatory synapses. In this work, I confirm previous findings showing that Pin1 binds postsynaptic density protein-95 (PSD-95) and identify an alternative binding site in the phosphorylated N-terminus of the PSD-95. Pin1 associates via its WW domain with phosphorylated threonine (T19) and serine (S25) in the N-terminus domain of PSD-95 and this association alters the local conformation of PSD-95. Most importantly, I show that proline-directed phosphorylation of the N-terminus domain of PSD-95 alters the local conformation of this region. Therefore, proline-directed phosphorylation of the N-terminus of PSD-95, Pin1 association, and peptidyl-prolyl cis-trans isomerization may all play a role in excitatory synaptic function and synapse development.
# Introduction
The postsynaptic density (PSD) of excitatory synapses is a highly crowded space composed of transsynaptic proteins, extracellular matrix constituents, surface receptors, ion channels, and scaffolding proteins. The scaffolding proteins at the PSD are essential elements required for the enrichment of ionotropic α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)-type and N-methyl-D-aspartate receptor (NMDAR)-type glutamate receptors at the PSD. At a given PSD, there are over 300 copies of the postsynaptic density protein-95 .
Although PSD-95 is a rather stable element of the PSD, capable of sustaining harsh biochemical extractions, PSD-95 synaptic stability is regulated following the induction of synaptic plasticity. In particular, the induction of synaptic plasticity, namely NMDAR-dependent long-term depression (LTD), increases the phosphorylation state of the N-terminus domain of PSD-95 and this event regulates its stability, Delgado Pin1-Mediated Isomerization of the N-Terminus Domain of clustering, proteolytic cleavage, and PSD-95 palmitoylation. Specifically, the induction of NMDAR-dependent LTD increases threonine 19 (T19) and serine (S25) phosphorylation of PSD-95. Despite significant advances on this topic, there is a lack information about potential binding partners interacting with phosphorylated T19 and S25 in PSD-95 that may regulate the stability following its increase in phosphorylation.
A potential protein that could interact with phosphorylated T19 and S25 is the phosphorylation-specific peptidyl-prolyl cis-trans isomerase (Pin1) as Pin1 has been shown to bind phosphorylate T287, S290, and T295. Pin1 is a small cytosolic enzyme that exclusively binds phosphorylated S/T-Proline sites. Pin1 has two major domains: an N-terminal WW domain and a C-terminal peptidyl-prolyl isomerase (PPIase) domain. The WW domain of Pin1 recruits the protein to the phosphorylated serine/threonine-proline (S/T-P) residues of its target protein and the catalytically active peptidyl-prolyl isomerase (PPIase) triggers the cis-trans peptidylprolyl isomerization.
Pin1 serves various functions in neuronal and excitatory synaptic physiology. In the hippocampus, Pin1 is involved in late-phase LTP via the regulation of dendritic protein synthesis. At excitatory synapses of striatal MSN and pyramidal neurons of the hippocampus, Pin1 regulates NMDAR currents by interacting with the phosphorylated T287, S290, and T295 in PSD-95; however, there is reason to believe that Pin1 binds additional sites in PSD-95. A closer inspection of the data presented inshows that binding is still observed in the deletion mutant of PSD-95 (∆287-295,,, which is the presumed binding region in PSD-95. Furthermore, the phosphomutant of PSD-95 where T287, S290, and T295 are replaced to alanine show increased cellular proteolysis. The proteolytic fragments of PSD-95 do not contain the N-terminus phosphorylation sites T19 and S25 which can serve as alternative Pin1 binding sites. Therefore, the observed loss in Pin1 binding to the PSD-95 alanine mutants could be due to the loss of an alternative binding site at the N-terminus domain of PSD-95. Therefore, the question of whether Pin1 associate with the phosphorylated serine-threonine residues in the N-terminus domain of PSD-95 has not been thoroughly examined yet.
In this work, I demonstrate that Pin1 binds phosphorylated T19 and S25 in the N-terminus domain of PSD-95. Phosphorylation of these residues triggers a change in PSD-95 conformation and Pin1 association with the phosphorylated residues restores the conformation of this domain. These findings position Pin1 as a key protein with the potential to regulate forms of synaptic plasticity dependent on phosphorylation of the N-terminus domain of PSD-95.
# Materials and methods
## Cloning and cdna plasmids
The plasmid encoding PSD-95::EGFP was a gift from S. Okabe (Tokyo University, Japan). The triple T19A, S25A and S35A (N3A-PSD-95) and the double T287A and S295A (C2A-PSD-95) PSD-95::EGFP mutants were generated using site-directed mutagenesis following the manufacturer's recommendations (Agilent Technologies, Santa Clara, CA, USA) and sequence verified. For the N3A-PSD-95 mutant, we first introduced the T19A and S25A double mutation using the following primer set . We then introduced the S35A mutation. For the C2A mutant, we introduced the T287A and S295A double mutation. The GST-Pin1 K63A and GST-Pin1 WW (GST-Pin1) were obtained from Addgene, plasmid ID# 19027 as described inmutant was generated using the following primer set . The T19, T19E, S25, and S35 PSD-95 peptide sequences were cloned into the EKAR construct using the following primer set .
## Mutant
Sense Antisense
## Expression and purification of the gst fusion proteins
Cell amplification, protein harvesting, and isolation proceeded as described inwith the exception that purified proteins were dialyzed, flash-frozen and stored at −80 - C in 20 mM Tris-HCl buffer (pH 7.5) containing 0.1 M NaCl, 5 mM DTT and 20% Glycerol.
## Gst-pulldown
COS-7 or COS-1 cells were transfected with PSD-95::EGFP using X-TremeGENE (Roche, Basel, Switzerland). Two days post-transfection cells were scraped off and lysed in buffer (in mM), 50 Tris-HCl, 200 NaCl, 100 NaF, 10% Glycerol, 1% Triton X-100, pH 8) containing protease inhibitor cocktail III (Calbiochem) and homogenized using three brief (1-3 s long) pulses on an ultrasonic homogenizer. Cells were spun at 5,000 rpm for 5 min and the supernatant was collected. Protein concentration was measured using the BCA method (Pierce). Between two hundred micrograms of total protein (depending on cell confluence) were incubated with 200 µg of GST fusion proteins and incubated for a minimum of 4 h followed by 2 h incubation with 20 µl of glutathione coated magnetic beads pre-blocked with 1% BSA (Pierce). Protein complexes were thoroughly washed with 200 µl and eluted in 50 µl of homogenization buffer containing 10 mM reduced glutathione pH ∼8.0.
## Pin1 isomerase spectrophotometer assay
The peptide substrate N-Succinyl-Ala-Glu-Pro-Phe-p-Nitroanilide (Peptides International, Inc., Louisville, KY, USA) was dissolved to 16 mM in DMSO. One microliter of the peptide stock solution was diluted in 100 µl of 480 mM LiCl in trifluoroethanol for 10 min before the start of the reaction. The reaction was carried out in cold buffer (0.1 M NaCl, 50 mM Hepes, 2 mM DTT and 0.04 mg/ml BSA, pH 7.0) containing 1-2 µg of GST-Pin1, 6 µg of α-chymotrypsin (Sigma-Aldrich, St. Louis, MO, USA) in 1 mM HCl and 0.5 µg of the peptide substrate. The absorbance change was immediately measured at 390 nm using a UV-Vis spectrophotometer (UV-1800, Shimadzu) at room temperature in cold buffer.
## Isomerization of full-length psd-95
This assay was adapted. Transfected COS-7 cells were homogenized as described previously without serine protease inhibitors. 30 µg of total protein were aliquoted into individual Eppendorf tubes (4 - C) labeled: input, BSA or Pin1. Input samples were quickly denatured by the addition of 6× Laemmli sample buffer (LB) and boiled at 85 - C for 10 min. BSA or Pin1 samples received 1 µl of α-chymotrypsin at 200 ng/µl in 0.08 M Tris HCl buffer, pH 7.8 containing 0.1 M calcium and 1 µg of BSA or Pin1. The reactions were quickly mixed and left to proceed undisturbed for 30 s at 25 - C, stopped with the addition of 6× LB, and boiled for 10 min.
Immunoprecipitation COS-7 cells were transfected with PSD-95::EGFP and Pin1. Whole-cell lysates were incubated with 4 µg of the anti-Pin1 antibody for 4 h followed by a 2 h incubation with 30 µl of Sepharose A coated beads pre-blocked for 1 h. Immunoprecipitated protein complexes were thoroughly washed 3X with 200 µl of homogenization buffer. Protein complexes were eluted in 50 µl of lysis buffer with 2× LB. For immunoprecipitation of PSD-95 from cultured neurons, a 20 mm dish at a confluency of 200,000 cells was scraped in 100 µl of HB containing 50 Tris-HCl, 200 NaCl, 100 NaF, 10% Glycerol, 0.5% deoxycholate, pH 8 containing protease inhibitor cocktail III, briefly sonicated and diluted in equal volumes of 50 Tris-HCl, 200 NaCl, 100 NaF, 10% Glycerol, 1% Triton X-100, pH 8 containing protease inhibitor cocktail III.
## Western blotting
Proteins were subjected to SDS-PAGE electrophoresis, transferred onto PVDF membranes and immunoblotted using the anti-PSD-95 antibody and anti-GST. The Pierce West Femto ECL substrate was used to reveal the immune complex. Images were taken using a Syngene apparatusor on X-Ray filmand analyzed using ImageJ. We only analyzed images obtained on the Syngene apparatus. Brightness and contrast are adjusted.
## Egfp fluorescence lifetime imaging
Live-cell imaging forand fixed/mounted cell imaging for. Cells were imaged in extracellular solution, or mounted in Vectashield, containing (in mM, 125 NaCl, 3 KCl, 10 HEPES, 10 Glucose, 2 CaCl and 1 MgCl pH 7.34) using a 40× 1.25 NA HCX PL Apo oil immersion objective in a modulated Light-Emitted-Diode 451 nm (3W). Lifetime acquisition and measurements were performed on an inverted Leica DMI6000B (Leica Microsystem, Wetzlar, Germany) microscope equipped with a LIFA frequency-domain lifetime attachment (Lambert Instruments, Roden, The Netherlands) and the manufacturer's LI-FLIM software. Lifetimes were referenced with a 1 mM solution of fluorescein in phosphate-buffered saline (PBS; pH 10), 4.00 ns lifetime. Measurements were obtained offline from an area encompassing most of the cell excluding the nucleus as in.
## Hippocampal cultured neurons
Preparation of cultured neurons was performed by plating neurons at a density of 100-200K/well of a 6-well plate. In brief, hippocampal neurons from E18 embryos of either sex were cultured on glass coverslips coated with PEI. Neurons were plated in Neurobasal supplemented with B27, 5% FBS and glutamine. Two days post-plating neurons were treated with 1 µM Ara-C to stop glia and microglial proliferation. Feedings were done every 4 days using low cysteine-containing media. At day in vitro 8-10 neurons were transfected using Lipofectamine 2000 following the manufacturer's recommendation. Experiments were performed on neurons were between 11 and 20 DIV.
## Immunostaining
Three to five days post-transfection cells were fixed in 4% Paraformaldehyde at room temperature for 20 min. Cells were then rinsed three times with 1× PBS, then 5 min in 50 mM NH4Cl, and three more quick rinses in 1× PBS. Cells were permeabilized in 0.1% Tx100 PBS for 5 min followed by three quick PBS rinses and incubated in 2 mL of 1%BSA in PBS for 45 min followed by incubation in 100 µl of anti-Pin1 (Santa Cruz Biotechnology, Dallas, TX, USA) and the anti-phospho T19 (1:500) for 1 h. Cells were rinsed 3× in PBS and the Alexa fluor 647 anti-mouse (1:500), and Alexa fluor 488 anti-rabbit (1:500) for 1 h in 1% BSA in PBS. Cells were rinsed 5× in PBS, postfix in 4% PFA and mounted in slow fade mounting media (Life Technologies, Carlsbad, CA, USA).
## Spinning disk confocal
Cells were imaged using 3-I Marianas live-cell dual-camera Yokogawa CSU-X spinning disk confocal. AxioObserver platform with DualCam and two Evolve EM-CCD cameras, CFP/YFP and R/G cubes using 100× /1.45 oil objective. We used the solid-state 488, 561, and 640 lasers with fiber switcher to excite the corresponding fluorophores as needed. The objective was mounted onto a piezo MadCityLabs piezo Z insert which was used to collect Z-stacks.
## Peptide synthesis
Peptides were synthesized using 9-fluorenyl methoxycarbonyl (Fmoc) solid-phase peptide synthesis with rink amide 4-methylbenzhydrylamine resin (EMD Millipore). The synthesis was performed on a CEM Liberty automated microwave peptide synthesizer. After removal of Fmoc groups with 30% 4-methylpiperidine and 0.1 M hydroxybenzotriazole (HOBt) in N,N-dimethylformamide (DMF) at 75 - C for 3-4 min, each amino acid or biotin (4 equiv.) was coupled at 75 - C for 5-10 min using 4 equiv. of O-benzotriazole-N,N,N ,Ntetramethyluronium hexafluorophosphate (HBTU), and 8 equiv. of N, N-diisopropylethylamine (DIEA). After synthesis, peptides were cleaved from the resin with a 95:2.5:2.5 trifluoroacetic acid (TFA)/triisopropylsilane (TIPS)/water mixture for 3-4 h. Rotary evaporation and precipitation in cold diethyl ether yielded the crude peptide mixture. Crude peptides were purified by HPLC on a C18 Phenomenex Jupiter or Gemini column in a water-acetonitrile gradient containing 0.1% v/v TFA or NH4OH respectively. Pure fractions were collected and identified using ESI-MS. The combined fractions were subjected to rotary evaporation to remove volatile solvents, frozen, and lyophilized to dryness. The purity of the final lyophilized solid was verified by LCMS.
## Surface plasmon resonance
SPR was used to measure the yes or no interaction between GST::Pin1 (and its mutants) and the different PSD-95 peptides using a BIAcore 3000 biosensor. Various biotinylated PSD-95 peptides were bound to the streptavidin matrix of a sensor chip. The immobilization process was carried out at a flow rate of 10 µl/min. Running buffer was used to prepare the control surface. The running buffer consisted of (in mM): 10 HEPES, 150 NaCl, 0.050 EDTA pH 7.4, 0.005% Tween 20. GST:Pin1 and its mutants (analyte) at various concentrations (see ''Results'' section) were injected with a flow of 10 µl/min over the immobilized peptides. Care was taken to use a low amount of protein to keep the signal below 400 Refractive Units (RU) units. The binding was assessed by monitoring the change in the refractive index (given in arbitrary response units, RU). The association/dissociation phases were monitored for 300 s. After each binding experiment, the sensor chip was regenerated by sequential washing with 0.85% phosphoric acid buffer (Bio-Rad). Several rounds of injections and regeneration were performed.
## Experimental design and statistical analysis
At least two coverslips were used on each data set. Data collection was interleaved, control for time and order effects. Samples from all groups were acquired on a weekly basis to reduce variability, else no data were included in the final analysis. We tested for outliers on a weekly basis and they were eliminated after testing all groups using Prism online calculator at a significance level of p < 0.05. Normality testing was performed on every group using D'Agostino and Pearson omnibus normality test. Betweengroup statistical significance was calculated accordingly for each distribution and experiment design. Data were normalized on a weekly basis to compensate for week to week variability. Numerical averages are presented as mean ± SEM or as box plots Statistical analyses were created using GraphPad Prism 5.0. Exact p-values are reported when provided.;. To evaluate the hypothesis that Pin1 could bind phospho T19, S25 and S35 PSD-95, I performed a sequence alignment between these sites in PSD-95 and other Pin1 binders. The sequence alignment showed that T19, S25, and S35 contain many of the amino acids found in most Pin1 binding partners. The T19/S25 phosphorylated form of PSD-95 enriches, biochemically, to the PSD fraction II, but lower amounts can be detected in the PSD fraction number III. To test if this interaction occurs in vivo, I immunostained cultured neurons with antibodies against Pin1 and phospho-T19. Pin1 showed ubiquitous expression and co-localized with phospho-T19 PSD-95 on dendritic spines, suggesting that Pin1 is in close proximity to phosphorylated T19-PSD-95 and could interact in vivo. The interaction between PSD-95::EGFP and Pin1 were re-examined via co-immunoprecipitation and GST pulldown experiments. In both COS-1 and neuronal lysates, Pin1 and PSD-95 co-immunoprecipitatedand no binding was detected towards the lower molecular weight bands of PSD-95. As stated earlier, the lower molecular fragments of PSD-95 do not contain the N-terminus phosphorylation sites T19 and S25; suggesting that Pin1 interaction with PSD-95 requires the integrity of this domain.
To determine which Pin1 domain mediates the interaction with PSD-95, I performed a series of GST pull-down assays. GST-Pin1 but not with GST pulled-down PSD-95. This interaction was not affected by two isomerase dead mutants [the K63A and R68A, R69A (RR/AA)], and it was also present in beads coated with the WW domain of Pin1. Once again, Pin1 interacted only with full-length PSD-95.
The phosphorylation-dependent binding of Pin1 to PSD-95 was confirmed using the GST pull-down assay with lysates treated with calf intestinal alkaline phosphatase (a broadspectrum phosphatase). CIAP reduced Pin1 binding to PSD-95::EGFP.
To evaluate if the phospho-sites in PSD-95 and Pin1 interact in live cells, I performed fluorescence resonance energy transfer/fluorescent lifetime imaging (FRET/FLIM) experiments using the EKAR construct. The original EKAR construct contains the Pin1-WW domain fused in frame with a flexible serine/glycine linker and the phosphopeptide of CDC25C. The CDC25 phosphorylatable sequence served as a positive control. In addition, the EKAR construct contains mRFP and EGFP, which, respectively, serve as the acceptor and donor fluorescent proteins. In the experimental conditions, the sequence of CDC25C was replaced with a 10-mer phosphorylatable peptide of T19 and S25. The association between the WW domain of Pin1 and the PSD-95 peptides was measured via FLIM as a reduction in EGFP lifetime, scheme right hand). Kinase activity was triggered by the presence of serum in the growing medium. Cells expressing the T19 peptide of PSD-95 showed similar values of EGFP lifetime as cells expressing the CDC25C phosphopeptide, suggesting that the Pin1 WW domain binds equally well to T19 and T48 in CDC25C. To test if the WW domain prefers T19, S25 or S35 in PSD-95, I repeated the FLIM experiments and noticed similar binding for all peptides, although binding was a little weaker for S35, imaging performed in fixed mounted cells). The phosphomimetic forms of the T19 peptide showed increased binding (imaging performed in living cells,right, one-way ANOVA with Bonferroni posttest, F (3,448) = 381.3, * * * p < 0.001), indicating that Pin1 prefers acidic residues.
Next, surface plasmon resonance (SPR) experiments were performed to examine if the association occurs in a reduced system containing only purified Pin1 and the N-terminus peptides of PSD-95. The synthetic peptides were immobilized in the streptavidin-coated sensor chip via a biotin moiety. Only T19 and S25 peptides were used as baits since they are the only known sites implicated in activity-dependent forms of synaptic plasticity. The immobilized peptides were challenged with purified Pin1 proteins, and the association was scored as an increase in resonance units as a function of time (RU values). Wildtype Pin1, the WW domain of Pin1, and the RR/AA mutant all bound well to the phospho-T19 peptide (only data for wt Pin1 is shown). Only wildtype Pin1 bound both phospho-T19 and phospho-S25. No detectable binding was observed to non-phosphorylated peptides or GST/BSA, used as negative controls (data not shown). These results suggest that in cells, the WW domain of Pin1 can bind the phosphorylated residues in the N-terminus domain of PSD-95.
## The α-chymotrypsin-coupled cis-trans isomerization assay reveals sites of binding and isomerization
Phosphorylation of S/T-proline bonds triggers conformational changes in protein structure by transiently changing the cis-trans equilibrium of the peptidyl-prolyl amide bonds . The peptidyl-prolyl amide bonds can exist in four possible conformations: (1) cis non-phosphorylated, (2) transnon-phosphorylated, (3) cis phosphorylated, and (4) trans phosphorylated. Because Pin1 is the only phosphorylationdependent peptidyl-prolyl isomerase in most eukaryotic cells, I used the α-chymotrypsin cis/trans isomerization assay which is used to follow in real-time the cis/trans isomerization of a phosphorylated peptidyl-prolyl bond. This assay takes advantage of the inability of α-chymotrypsin to hydrolyzed peptide bonds when the amino acid preceding the proline is in the cis configuration. Therefore, this assay was modified to work with full proteins in lysates and confirm the site of Pin1 binding and isomerization on full-length PSD-95. In addition, I also tested if phosphorylation of T19 and S25 alters the local conformation of the N-terminus domain.
The purified Pin1 protein was catalytically active as evidence of the accelerated loss of the Suc-AEPY-pNA peptide, measured at 390 nm;. On full-length PSD-95, the degradation by α-chymotrypsin was measured via Western immunoblotting and the intensity of the band corresponding to full-length PSD-95 was quantified. Homogenates incubated with wt Pin1 showed a time-dependent loss of full-length PSD-95 intensity when compared to the isomerization mutant (RR/AA;. To further explore the sites of Pin1 association/isomerization, T19, S25, and S35 were mutated to alanine (referred to as N3A, short for three N-terminus residues mutated into alanine;. Mutating these sites to alanine residues should occlude Pin1-mediated isomerization because (1) the alanine-proline prolyl bond tends to be in the trans configuration 99% of the time, (2) Pin1 does not bind to non-phosphorylated residues, and (3) the Pin1 isomerase domain cannot isomerize these bonds. In contrast to the results obtained with wt PSD-95, reactions containing the N3A-PSD-95 mutant were insensitive to Pin1, while homogenates expressing wt PSD-95 show an accelerated loss of PSD-95left bar graph). Additional controls on mutants of the hinge domain, T287A and S295A phospho-mutant (C2A), were sensitive to Pin1, left bar graph, n = 8 **p < 0.01 unpaired t-test). Thus, indicating that these sites are not sensitive to the configuration of this assay or that they do not undergo much cis-trans isomerization. Although singly phosphorylated S290 could be isomerized by Pin1, it is highly unlikely due to bivalency requirements for Pin1 binding/isomerization. Furthermore, a mutant containing both sets of mutations (N3A/C2A) behaved just like the N3A-PSD-95 mutant. In all experiments, equal amounts of PSD-95 protein are present across all paired reactions (BSA vs. Pin1, right bar graphs). The lack of action by Pin1 in lysates expressing the N3A-PSD-95 mutant supports the idea that the phosphorylated N-terminus domain is a site of Pin1 binding and isomerization in PSD-95.
## N-terminal psd-95 phosphorylation alters its conformation
The data obtained using the α-chymotrypsin assay supports the idea that phosphorylation of the N-terminus domain in PSD-95 alters the conformation of PSD-95 and Pin1 restores its conformation. To further support this hypothesis, I compared the rates of α-chymotrypsin degradation from homogenates expressing either wt PSD-95 or the N3A-PSD-95 mutant without Pin1. In agreement with the idea that Pin1 isomerizes the N-terminus domain, reactions containing the N3A-PSD-95 mutant showed faster degradation than reactions containing wt PSD-95, * * * p < 0.001 unpaired t-tests). These results support the following ideas: (1) Pin1 interacts with T19 and S25 in PSD-95, (2) N-terminal phosphorylation of PSD-95 alters this section of PSD-95, (3) the alanine mutants of T19 and S25 adopt the correct conformation, and (4) Pin1 can restore its structure.
# Discussion
Previous reports have shown that Pin1 interacts with the hinge domain of PSD-95, and in this report, I show that Pin1 binds to full-length PSD-95 via the phosphorylated T19, S25, and S35. It is common for Pin1 to bind multiple regions within the same target. But how can Pin1 bind these distant sites in PSD-95? In the extended conformation, Pin1 is over 7.3 nm in length, which is potentially long enough to position one of its domains close to the hinge and the other domain close to the N-terminus domain of PSD-95. The interaction between Pin1 and PSD-95 could be further facilitated by conformational changes in PSD-95. Given that palmitoylation of PSD-95 strongly regulates its conformation, the interplay between PSD-95 conformational states and PSD-95 palmitoylation may be an integral mechanism regulating the association. Therefore, further studies will be required to better our understanding of the relationships between PSD-95/Pin1 association and how N-terminus domain phosphorylation/palmitoylation regulates this interaction. In an accompanying article, on this issue, we specifically evaluate the functional significance of this interaction.
Although the biochemical studies show that Pin1 accelerates the loss of full-length PSD-95 in the α-chymotrypsin assayobserved that Pin1 protected PSD-95 from subtilisin degradation. The reason for this discrepancy is unknown, but the data presented here differs from theirs in many ways. First, I provide data showing the activity of purified GST-Pin1. Second, I show the averages for the reactions. Third, I show evidence of equal loading among the pair of reactions. Fourth, I show the full gels showing the products of proteolytic degradation as shown by others. Last, a mutagenesis strategy was employed to confirm the region of Pin1-mediated binding and isomerization, which confirmed the idea that Pin1 mediate the cis-trans isomerization of the N-terminus domain of PSD-95.
The α-chymotrypsin data raises several interesting questions about the multiple conformations adopted by the N-terminus domain of PSD-95 and the biological activity of these conformations in neurons. For example, it would be interesting to quantitatively estimate the fractions of these conformations in cells or solution as has been done for the mGluR5 receptor. Because each S/T-P bond can exist in 4 different mutually exclusive conformations, the phosphorylated N-terminus domain of PSD-95 could adopt up to 64 possible conformations. If the other three hinge domain sites are included then up to 1,296 isomers of PSD-95 could be present within a given PSD. Therefore, it is important to identify which isomers of PSD-95 are important for normal excitatory synaptic function.
# Data availability statement
The datasets generated for this study are available on request to the corresponding author.
# Ethics statement
All procedures were in accordance with guidelines of and approved by the Institutional Animal Care and Use Committee at the University of Chicago, Brandeis University, and University of Bordeaux II.
# Author contributions
JD conceived the initial idea, performed experiments, generated graphs, analyzed data and wrote the manuscript.
# Funding
Funding sources: AMPAzeta IIF Marie Curie fellowship to JD, ERC Grant Nano-Dyn-Syn, the ERA-NET project MODIFSYN to Daniel Choquet, the T32 NS0072419-28 to Eve Marder and the NS0636853 to Gina G. Turrigiano.
|
10.1098/rsos.170694
|
CCBY
|
5717634
|
29291060
|
s2orc_pubmed_articles
|
Environmental influences and ontogenetic differences in vertical habitat use of black marlin (Istiompax indica) in the southwestern Pacific
# Background
Billfishes of the family Istiophoridae (marlins, sailfish and spearfishes) include some of the largest and most highly migratory species on earth. However, despite their ability to transverse ocean basins, billfish movements are restricted by their physiological tolerance to certain environmental conditions [bib_ref] Vertical habitat use of Atlantic blue marlin Makaira nigricans: interaction with pelagic..., Goodyear [/bib_ref]. Satellite tagging has provided important insights into the influence of environmental factors on the vertical habitat use of billfish. A recent review of their diving behaviour identified four physical drivers of ecology: temperature, light, oxygen and complex water mixing (e.g. fronts and eddies). These physical drivers have been shown to effect billfish physiology, whereby optimal conditions are thought to encourage the expansion of vertical habitat use [bib_ref] Habitat characterization for striped marlin in the Pacific Ocean, Lam [/bib_ref]. Therefore, to understand how changes in global climate patterns may influence the distribution or abundance of these large pelagic predators, it is important that vertical habitat use and environmental drivers of movement are better understood.
The effect of dissolved oxygen (DO) on diving behaviour has been extensively studied in some billfish species. Comparative investigations of sailfish (Istiophorus platypterus) and blue marlin (Makaira nigricans) vertical habitat use between the oxygen-rich waters of the Atlantic and oxygen-depleted waters of the tropical Eastern Pacific show that the extent of water column use is limited at depths where DO levels are below 3.5 ml O 2 l −1 [bib_ref] Vertical and horizontal movements of striped marlin (Tetrapturus audax) near the Hawaiian..., Brill [/bib_ref] [bib_ref] Ocean scale hypoxia-based habitat compression of Atlantic istiophorid billfishes, Prince [/bib_ref]. The availability of light and conditions influencing water mixing have also been associated with billfish vertical distribution. As billfish are visual predators, the degree of light penetration through the water column influences their vertical distribution, and they are able to forage in deeper waters during the day [bib_ref] Retinal specializations in the blue marlin: eyes designed for sensitivity to low..., Fritsches [/bib_ref]. Diel-diving behaviour has been noted for many istiophorid billfishes including striped marlin (Kajikia audax) in the southwestern Pacific, where tagged fish spent daylight hours foraging in deeper waters, before returning to surface waters at night [bib_ref] Habitat characterization for striped marlin in the Pacific Ocean, Lam [/bib_ref] [bib_ref] Investigating behaviour and population dynamics of striped marlin (Kajikia audax) from the..., Sippel [/bib_ref]. Complex water mixing such as eddies or fronts have been shown to play an important role in the foraging behaviour of pelagic predators. The formation of eddies as a driver in the aggregation of prey has been identified as a factor influencing vertical habitat use in opah (Lampris guttatus) [bib_ref] Vertical movement and habitat of opah (Lampris guttatus) in the central North..., Polovina [/bib_ref]. Complex water mixing has also been associated with shifts in species distribution and catch per unit effort (CPUE) increases for billfishes [bib_ref] Dynamic habitat suitability modelling reveals rapid poleward distribution shift in a mobile..., Hill [/bib_ref] [bib_ref] Comparative influence of ocean conditions on yellowfin and Atlantic bluefin tuna catch..., Teo [/bib_ref] ; however, the influence of water mixing on depth use has yet to be investigated.
The distribution of temperature at depth is another physical factor commonly associated with billfish vertical habitat use because temperature has been shown to affect the cardiac function of marine teleosts [bib_ref] Physiological and behavioral reactions of fishes to temperature change, Crawshaw [/bib_ref]. Decreases in ambient water temperature reduce metabolic rate, contractibility of cardiac muscle, cardiac output and core body temperature in marine teleosts [bib_ref] Effects of temperature on cardiac function in teleost fish, Kalinin [/bib_ref]. Billfish are often restricted in the amount of time they can spend in cold deep water presumably owing to the physiological costs of operating in cooler waters, often outside a species' optimal thermal window [bib_ref] Vertical habitat use of Atlantic blue marlin Makaira nigricans: interaction with pelagic..., Goodyear [/bib_ref] [bib_ref] Vertical and horizontal movements of striped marlin (Tetrapturus audax) near the Hawaiian..., Brill [/bib_ref]. Comparison of vertical habitat use among various species of billfish has led to the suggestion that larger-bodied individuals make greater use of their increased thermal inertia than smaller-bodied ones, by diving deeper [bib_ref] Vertical habitat use of sailfish (Istiophorus platypterus) in the Atlantic and eastern..., Hoolihan [/bib_ref]. This theory derives from the observation that the larger-bodied M. nigricans and K. audax [bib_ref] Vertical habitat use of Atlantic blue marlin Makaira nigricans: interaction with pelagic..., Goodyear [/bib_ref] [bib_ref] Investigating behaviour and population dynamics of striped marlin (Kajikia audax) from the..., Sippel [/bib_ref] have been found to dive deeper than the smaller-bodied I. platypterus and white marlin (Kajikia albida) [bib_ref] Vertical habitat use of sailfish (Istiophorus platypterus) in the Atlantic and eastern..., Hoolihan [/bib_ref] [bib_ref] Habitat utilization and vertical movements of white marlin (Tetrapturus albidus) released from..., Horodysky [/bib_ref]. Despite the suggestion that a species' body mass limits the amount of time spent outside of its optimal thermal range and subsequent vertical habitat use [bib_ref] Vertical habitat use of Atlantic blue marlin Makaira nigricans: interaction with pelagic..., Goodyear [/bib_ref] [bib_ref] Vertical habitat use of sailfish (Istiophorus platypterus) in the Atlantic and eastern..., Hoolihan [/bib_ref] , there have been no quantitative analysis of such a relationship in any billfish species. In addition, the suggestion of body mass influencing depth use also does not consider the large ecological differences between billfish species or actual mass of individuals tagged in studies [bib_ref] Vertical habitat use of sailfish (Istiophorus platypterus) in the Atlantic and eastern..., Hoolihan [/bib_ref]. Despite the black marlin (Istiompax indica) being one of the largest billfishes, tagging of I. indica has shown the deepest dives to be less than 250 m (minimum ambient temperate of 14.5°C) [bib_ref] Seasonal movements and diving behaviour of black marlin (Istiompax indica) in the..., Chiang [/bib_ref] [bib_ref] Post-release behaviour of black marlin, Makaira indica, caught off the Great Barrier..., Pepperell [/bib_ref] , which are equivalent to the maximum depths reported for I. platypterus (8.7°C) and K. albida (12°C), which maintain a much smaller average mass [bib_ref] Vertical habitat use of sailfish (Istiophorus platypterus) in the Atlantic and eastern..., Hoolihan [/bib_ref] [bib_ref] Habitat utilization and vertical movements of white marlin (Tetrapturus albidus) released from..., Horodysky [/bib_ref]. In addition, I. indica have been observed to primarily occupy the upper 50-100 m of the water column and no indications of diel-diving behaviour have been reported [bib_ref] Seasonal movements and diving behaviour of black marlin (Istiompax indica) in the..., Chiang [/bib_ref] [bib_ref] Post-release behaviour of black marlin, Makaira indica, caught off the Great Barrier..., Pepperell [/bib_ref]. To effectively evaluate the relationship between body mass and vertical habitat use, an intraspecific comparison is required to reduce the potentially large effect of variation owing to differing species' ecologies.
Istiompax indica is a highly migratory istiophorid billfish that occupies waters throughout the tropical and subtropical Indo-Pacific [bib_ref] Makaira indica (Cuvier), 1831, Nakamura [/bib_ref]. [bib_ref] Global overview of the major constituent-based billfish tagging programs and their results..., Ortiz [/bib_ref]. Tagging data have revealed that horizontal movement patterns vary across life stages, with juvenile and sub-adult movements presumably motivated by prey availability, while adult movements are also influenced by reproductive philopatry [bib_ref] Movements and variations in early year-class strength of black marlin (Makaira indica)..., Pepperell [/bib_ref]. Horizontal movements have been assessed in conjunction with physical factors to determine sea-surface habitat preferences in I. indica. Deployment of pop-up satellite archival tags (PSATs) on I. indica off the east coast of Australia revealed an affinity for waters with a sea-surface temperature (SST) between 26°C and 27°C. Using spatial positions derived from conventional tagging data in a species distribution model, the occurrence of juvenile I. indica has been strongly correlated with sea-surface height anomaly (SSHa) and chlorophyll-a concentrations (Chl-a) [bib_ref] Dynamic habitat suitability modelling reveals rapid poleward distribution shift in a mobile..., Hill [/bib_ref]. While considerable efforts have been made towards understanding the horizontal movements of I. indica in surface waters, the paucity of information regarding vertical habitat use represents a knowledge gap.
Istiompax indica is one of the most extensively tagged billfish species, yet there is little known about how they use the water column and the influence of physical factors on their diving behaviour. In this study, we investigate: (i) the variation of vertical habitat use among differing size classes of I. indica, and (ii) explore physical factors to better understand how they influence the depth use of I. indica.
# Material and methods
## Satellite tagging
A total of 102 PSATs were deployed on I. indica off the east coast of Australia (Queensland and New South Wales) in consecutive years from 2002 to 2014 as part of two separate tagging programmes. All tagged fish were captured using trolled baits in a rod-and-reel fishery with the assistance of recreational anglers, with 41 of the tags part of the Great Marlin Race (GMR) (https://igmr.igfa.org/Conserve/IGMR.aspx) and the remaining 61 tags being deployed by Domeier & Speare. Tag attachment methods have been previously described by Domeier & Speareand GMR tagging protocols by Carlisle et al. [bib_ref] Influence of temperature and oxygen on the distribution of blue marlin (Makaira..., Carlisle [/bib_ref]. The PSATs deployed were manufactured either by Wildlife Computers (WC; Redmond, WA, USA), models PAT0, PAT1, PAT2, PAT3, PAT5, PAT6, MK-10 and MiniPATs, or by recovered Microwave Telemetry (MT, Columbia, MD, USA) model PTT-100 tags. Tags were programmed to record pressure (depth), temperature and light at 20 or 60 s intervals. Tags were programmed to release from fish after 120 days (n = 14), 180 days (n = 86) or 270 days (n = 2) in order to capture both broad-scale and high-resolution data. Archival data containing depth, temperature and geolocation (based on light levels) were received through the Argos satellite system (Service Argos, Toulouse, France). High-resolution time-series data (recorded every 7 min) were collected from WC MiniPATs via satellite, as well as from 15 physically recovered tags (data recorded every 20 or 60 s), which were downloaded directly from each tag's data archive (electronic supplementary material, table S2).
## Data processing
All tag data were collated using R v. 3.3.2. For WC tags, percentage time at depth (PTD) and maximum depth (MD) data were aggregated daily based on summarized daily histogram and maximum depth files. To examine diel patterns of depth and temperature, only high-resolution time-series data were used. Tagged animals were separated into five size classes based on mass estimates made by anglers and charter captains at the time of capture. As females and males mature at 100 and 40 kg, respectively, it was assumed that all tagged animals that had an estimated mass over 100 kg were mature. However, owing to the inability to determine the age or sex of tagged animals for all size classes, additional groupings were based on five arbitrary mass ranges that had sufficient sample sizes for downstream analysis. These were: small (20-50 kg); intermediate (51-100 kg); medium (101-250 kg); large (251-400 kg) and very large (greater than 400 kg).
## Geolocation estimation
Daily position estimates were calculated using a state-space model accessed through the WC proprietary software, Global Position Estimator 3 (GPE3) (WC). The state-space model uses light-level, in situ temperature observations and their corresponding remotely sensed reference data (twilight, SST) into a diffusion-based space-state movement model to generate time-discrete gridded probability surfaces. Grids are produced at a resolution of 0.25°. To reduce the number of unrealistic positions, an estimate of maximum animal speed was input to represent the standard deviation of diffusion rate for the animal. A starting maximum speed of 200 km d −1 was input, based on the knowledge that I. indica have been recaptured to 500 km from their tagging location after 3 days (J. Pepperell 2008, unpublished data). Animal speed was then varied to ensure the geolocation estimation was within the model bounds and stabilized across runs, with optimal animal speed for all tagged I. indica being estimated at 250 km d −1 .
# Data analysis
Environmental data were derived from broad-scale climatological datasets to investigate associations with I. indica vertical habitat use. Monthly 1°-grid climatology data of DO at 100, 200 and 300 m depths were sourced from World Ocean Atlas 2009 (WOA09), and a suite of satellite remote-sensed data (electronic supplementary material, table S1) from National Oceanic and Atmospheric Administration (NOAA) Coastwatch were extracted along daily track positions using xtractomatic routines in Ras described by Lam et al. [bib_ref] Habitat characterization for striped marlin in the Pacific Ocean, Lam [/bib_ref]. To account for error in the position estimates and ensure consistent spatial coverage of remote-sensed data, an average value for each given position was calculated with a 0.25°e
rror. An estimate of T was calculated as the difference between remote-sensed SST and archival tag-sampled temperatures at depth. Positive T values reflect cooler measured temperature than SST [bib_ref] Vertical and horizontal movements of striped marlin (Tetrapturus audax) near the Hawaiian..., Brill [/bib_ref].
To account for any spatial variability when investigating differences in vertical habitat use, data were sub-sampled from an area of spatial overlap based on the daily position estimates from the most probable track of tagged animals across all five size classes. The area of spatial overlap was defined by observations within latitudes of 10°S-20°S and longitudes of 145°E-160°E. The sub-sampled data were plotted to inspect the data for normality and to check that the distribution of the data was consistent among size classes. To assess for differences in the MD and PTD data among size classes, a Friedman repeated-measures non-parametric analysis of variance was undertaken in R. To account for repeated observations by individual fish, a unique tag identification number was included as a random variable.
To determine whether the distribution of temperatures used at depth, relative to the SST ( T) varied among size classes, Kolmogorov-Smirnov tests were performed on the time-series data in R. To correct for multiple comparisons, a Bonferroni correction was applied to the α value.
The effect of environmental factors on the vertical habitat use of tagged I. indica was investigated using generalized additive mixed models (GAMMs). Two separate GAMMs were fitted to examine the relationship between environmental predictor variables and the response variables (MD and PTD <150 m). Environmental factors from the climatological databases, including Chl-a, SST, DO at 100 m depth intervals (DO 100-300 m ), mixed layer depth (MLD), sea-surface height deviation (SSHd), surface wind speed (Wind), location (longitude-latitude pair-derived from the most probable track) and month were modelled as fixed explanatory variables. GAMMs were constructed based on the same approach undertaken by Lam et al. [bib_ref] Habitat characterization for striped marlin in the Pacific Ocean, Lam [/bib_ref] , with analysis run using the package 'mgcv' in R and after visual assessment of the error distribution, a Gaussian family with an identity link function was applied. All explanatory variables were modelled as continuous variables and smoothed, with the exception of month as it was categorical. To account for multiple observations from the same tagged fish, each fish was modelled as a random variable. Correlations between variables were investigated, and the presence of colinearity between covariates was assessed against the variance inflation factor (VIF) using the corvif function by Zuur et al. [bib_ref] Additive and generalised additive modelling, Zuur [/bib_ref]. Variables were determined to be colinear if they contained VIF scores greater than 5 [bib_ref] Additive and generalised additive modelling, Zuur [/bib_ref]. If colinearity was identified, the variables with the highest VIF values were sequentially removed until all VIF values were less than the threshold [bib_ref] Additive and generalised additive modelling, Zuur [/bib_ref]. The initial, full factorial models were:
[formula] R i = s(Location) + s(MLD) + s(SST) + s(Wind) + s(Chl -a) + s(SSHd) + s(DO 100 m ) + s(DO 200 m ) + s(DO 300 m ) + Month + Fish i + ε, [/formula]
where R i is the response variable (MD or PDT) for Fish i , and i = 1, . . . , number of individual fish n; environmental predictor variables are defined earlier, is the Gaussian error term and smoothing functions were chosen automatically and evaluated manually using the 'gam.check' function. The amount of smoothing (k) applied to the splines was limited to ensure that the model did not overfit the data, yet sufficient to describe the nonlinearity between the response and explanatory variable [bib_ref] Additive and generalised additive modelling, Zuur [/bib_ref].
The selection of an optimal model was undertaken using a backward selection strategy, whereby non-significant explanatory variables were dropped in sequential order until a final model containing only significant predictors was reached, as described by Lewis. Statistical significance was set at α = 0.05 during the selection process. The final model was visually inspected to confirm the fit of all data and residuals in diagnostic plots. Explanatory variables were considered as strong predictors of vertical habitat use if they were significant in both PTD and MD models.
# Results
Of the 102 PSATs analysed, 14 tags were physically recovered after pop-off. The number of I. indica tagged from each size class comprised: 14 small (20-50 kg), 16 intermediate (51-100 kg), 43 medium (101-250 kg), 14 large (251-400 kg) and 15 very large fish (greater than 400 kg). Tags stayed on animals for a median duration of 56 days (range 1-201) (electronic supplementary material, table S2). The displacement distance travelled by I. indica after tagging ranged from 10 to 10 623 km, including transoceanic and trans-equatorial movements. After filtering of erroneous tracks, a total of 6788 geolocation days were included in the final analysis.
## Diving behaviour
Evaluation of time-series data resulted in the identification of two distinct diving behaviours: episodic deep-diving behaviour and bounce-diving behaviour (electronic supplementary material, figure S1). Episodic deep-diving behaviour was characterized by infrequent excursions below 400 m. These episodic deep dives were characterized by short dive durations (no longer than 20-30 min) and a 'V' shape dive profile, with animals spending less than 60 s at the maximum depth before ascending. Episodic deep-diving to depths greater than 400 m was observed in 28 individuals across all size classes. It was identified that episodic deep dives were more common in fish from adult size classes (greater than 50% of individuals from medium, large and very large classes) than in individuals from intermediate and small classes (less than 20%).
Bounce-diving was characterized by fish remaining in the upper water column (approx. 20 m deep) during the night and undertaking a number of short repetitive dives between 80 and 400 m throughout the day (figure 1). The number of repetitive dives was often greater than 10 d −1 , and bounce-diving behaviour appeared to start shortly after sunrise. Bounce-diving behaviour was found to be consistent among fish greater than 50 kg (intermediate, medium, large and very large size classes) with routine dives commencing at sunrise and occurring throughout the day until sunset. Although bounce-diving was detected in small fish, it appeared to be less frequent and occurred during both day and night (figure 1).
## Vertical habitat use
Tagged I. indica recorded a daily average MD of 206 m (±123 m), with 600 m being the deepest dive of any individual. Significant differences in the MD within the sub-sampled area of spatial overlap were detected among all five size classes (Friedman's test: χ 2 = 60.05, p < 0.05), with the exception of between medium and large I. indica (figure 2). Overall, the mean MD increased with increasing size. A similar pattern of increasing depth of vertical habitat use was also observed in the PTD, with significant differences observed among all size classes, except for medium and large size classes (Friedman's test: χ 2 = 49.08, p < 0.05). Large and very large fish spent greater than 10% of time below 150 m, contrasting with small and intermediate marlin which spent as little as 1.5% and 4.3% of time below 150 m, respectively.
## Temperature use
Istiompax indica used a wide range of water temperatures, from 7.4°C to 31.3°C, although there was a higher occupancy of waters around 25-27°C. The frequency of temperature at depth distribution experienced by I. indica relative to the SST ( T) was significantly different among size classes (Kolmogorov-Smirnov test, p < 0.01). The range of T values experienced by the various size classes varied from approximately 6.5°C in small to approximately 21.5°C in medium, large and very large I. indica. Despite differences in the magnitude of T that were exhibited by different size classes, the majority of temperature records for all size classes were within 5°C of SST, amounting to approximately 87% for very large, 91% for large and intermediate, 97% for medium and approximately 99% for small size classes (figure 3).
## Diel patterns
Pronounced diel patterns in vertical habitat use were detected for all individuals; however, they were not consistent across all size classes [fig_ref] Figure 4: Diel behaviour of I [/fig_ref]. Occupancy of deeper depths during daylight hours was found to increase with size class. All size classes greater than 50 kg (intermediate-very large) occupied greater depths during the day than during the night. This increase in depth use during daylight hours was driven by the onset of bounce-diving, which resulted in movement into deeper waters (figure 1). During the night, bounce-diving ceased and animals larger than 50 kg were found to shift their behaviour to occupy surface waters (figure 4). By contrast, tagged I. indica from the smallest size class occupied shallower depth ranges during daylight hours than they did at night. The shift in vertical habitat use between day and night was also less pronounced than that observed in the larger size classes.
## Factors influencing vertical habitat use
The effects of spatio-temporal and environmental factors on the daily vertical habitat use of I. indica were examined using two GAMMs. DO at 100 m, MLD, SSHd and location were all identified to significantly affect the PTD less than 150 m and MD [fig_ref] Table 1: Summary of GAMM results on vertical habitat use of I [/fig_ref]. The best-fit model for PTD also included month and DO at 200 and 300 m as significant predictors. SST was found to be the only other variable significantly related to MD, whereas Chl-a and wind were not found to be related to either of the response variables. The best-fit models for both MD and PTD explained (adjusted R 2 ) approximately 24% of the overall variability. Colinearity was not detected among any of the significant predictors in the MD or PTD models. The magnitude of the response of both MD and PTD changed when the MLDs were greater than 50 m. The MD increased when the SST was between 26°C and 28°C (figure 5). The spatial depth use appeared to vary on both a latitudinal and longitudinal gradient. The MD was found to increase notably as tagged animals occupied tropical waters to the north of the Tropic of Capricorn (approx. 23°S) and east of the Coral Sea basin (approx. 155°E). A similar spatial trend was noted in the percentage time at depth, where A linear trend was observed for the MD and PTD in response to changes in the SSHd. The maximum daily depths recorded for I. indica became deeper with increases in the SSHd. As the PTD showed a negative response to increasing SSHd, it suggests that less time spent in the upper 150 m water column occurred with rising SSHd (figure 6). Decreases in the time spent in the upper water column by I. indica were also identified to occur when the availability of DO at 100 m improved. The response of DO 100 m to changing MD was found to be a nonlinear trend that was weak in magnitude [fig_ref] Figure 5 10: rsos [/fig_ref]. This nonlinear response of MD to DO 100 m illustrated that greater maximum depths were observed when DO concentrations were less than 4.2 ml O 2 l −1 or greater than 4.8 ml O 2 l −1 . A nonlinear trend was also noted in the response of DO 300 m to PTD, with the highest percentage of time spent in waters deeper than 150 m occurring when DO 300 m concentrations were between 3.5 and 4.2 ml O 2 l −1 .
# Discussion
Accessing the world's largest PSAT tagging datasets on I. indica enabled us to undertake the most comprehensive analyses of the species' vertical habitat use and to examine differences in diving behaviour by size. In this study, we found that vertical habitat use expanded with increases in the mass of tagged animals. We also demonstrated the influence of environmental factors (including the MLD, DO and SSHa) and the spatial distribution on the maximum diving depth and time at depth. By characterizing the vertical habitat use of I. indica across the full-size range of the species, our investigation provides a foundation for exploring ontogenetic differences in other large pelagic predators.
## Vertical habitat characterization
Vertical habitat use differed from that of previous satellite tagging investigations of I. indica [bib_ref] Seasonal movements and diving behaviour of black marlin (Istiompax indica) in the..., Chiang [/bib_ref] [bib_ref] Post-release behaviour of black marlin, Makaira indica, caught off the Great Barrier..., Pepperell [/bib_ref]. Tagged animals in our study were observed to regularly access mesopelagic waters (greater than 200 m deep), with pronounced diel patterns in the diving behaviour, which had not previously been observed in I. indica. Our findings contrast with other studies in the South China Sea [bib_ref] Seasonal movements and diving behaviour of black marlin (Istiompax indica) in the..., Chiang [/bib_ref] and the southwestern Pacific Ocean, which concluded that I. indica was restricted to approximately 250 m of the upper water column, with individuals rarely diving below the mixed layer [bib_ref] Seasonal movements and diving behaviour of black marlin (Istiompax indica) in the..., Chiang [/bib_ref] [bib_ref] Post-release behaviour of black marlin, Makaira indica, caught off the Great Barrier..., Pepperell [/bib_ref]. These observed differences in vertical habitat use are likely in part owing to regional differences, with fish tagged in the South China Sea exposed to differing environmental conditions and the collection of depth data at 15 min intervals in that study which may have precluded the detection of fine-scale diving behaviour (i.e. bounce-diving) [bib_ref] Seasonal movements and diving behaviour of black marlin (Istiompax indica) in the..., Chiang [/bib_ref]. By comparison, the analysis of I. indica diving behaviour by Gunn et al.was limited to only two dive profiles (mean duration = 51 days), resulting in modest temporal and spatial coverage, which may have restricted the ability to capture the true extent of diving behaviour over longer time and area scales. However, the results reported here are similar to the diving patterns reported for many other large pelagic fish species including blue marlin (M. nigricans) [bib_ref] Vertical habitat use of Atlantic blue marlin Makaira nigricans: interaction with pelagic..., Goodyear [/bib_ref] , yellowfin tuna (Thunnus albacares) [bib_ref] Movements, behavior, and habitat utilization of yellowfin tuna (Thunnus albacares) in the..., Schaefer [/bib_ref] and tiger sharks (Galeocerdo cuvier) [bib_ref] Tiger shark (Galeocerdo cuvier) movement patterns and habitat use determined by satellite..., Holmes [/bib_ref] , indicating that the vertical habitat use observed by I. indica is not uncommon among large pelagic predators.
The dive profiles of I. indica were characterized by two types of diving behaviour, which were similar to those observed for T. albacares, skipjack (Katsuwonus pelamis) and bigeye tuna (Thunnus obesus) [bib_ref] Vertical movements and habitat utilization of skipjack (Katsuwonus pelamis), yellowfin (Thunnus albacares),..., Schaefer [/bib_ref] , opportunistic deep-diving and bounce-diving. Opportunistic deep dives by I. indica were characterized by infrequent and unpredictable vertical movements to below 250 m, the deepest of which was recorded to 600 m in the Coral Sea. In striped marlin (K. audax), opportunistic deep-diving has been suggested as an act of predator avoidance owing to its irregular nature [bib_ref] Habitat characterization for striped marlin in the Pacific Ocean, Lam [/bib_ref] , and it is possible that opportunistic deepdiving of I. indica may also reflect an act of predator avoidance. Predation of five tagged fish was observed in this study, suggesting the relevance of predator avoidance as a potential motive for opportunistic deep-diving. Although opportunistic deep-diving was useful in defining the vertical limits of I. indica, this behaviour was infrequent and is not considered representative of normal vertical habitat use.
## Ontogenetic variability
The depth exhibited by I. indica during bounce-diving behaviour varied considerably among size classes, but its commonality in occurrence indicated that it probably reflects an important behaviour. Bounce-diving has been extensively studied in several species of tuna and is suggested as a tactic to optimize time at depth when foraging [bib_ref] Vertical movements and habitat utilization of skipjack (Katsuwonus pelamis), yellowfin (Thunnus albacares),..., Schaefer [/bib_ref]. The use of bounce-dives enables an individual to regularly access warm surface waters to raise body temperature between dives, which then allows for an increase in the time available to forage below the thermocline [bib_ref] Tiger shark (Galeocerdo cuvier) movement patterns and habitat use determined by satellite..., Holmes [/bib_ref]. This bounce-dive behaviour in many large pelagic fishes is generally restricted to daylight hours when sunlight penetrates deep into the water column, facilitating the hunting ability of these large visual predators [bib_ref] Habitat characterization for striped marlin in the Pacific Ocean, Lam [/bib_ref]. Recent evidence from dietary investigations provides support for foraging as a motive for I. indica to undertake bounce-diving to mesopelagic waters. Stomach content analysis has shown that I. indica fed almost exclusively on the purple-back flying squid (Sthenoteuthis oualaniensis), a deep-water species, during the day in the eastern Arabian Sea [bib_ref] Diet composition, feeding niche partitioning and trophic organisation of large pelagic predatory..., Varghese [/bib_ref]. In the southwestern Pacific, the vertical distribution of S. oualaniensis closely correlates with the reported depth use of I. indica when bounce-diving [bib_ref] A review of the systematics, distribution, and biology of arrow squids genera..., Dunning [/bib_ref] , suggesting a causal link between diving behaviour and prey distribution.
Despite the importance of bounce-diving as a foraging behaviour, considerable differences in depth use were noted among size classes, which appeared to be reflected in the periodicity, frequency and depth of bounce-diving behaviour. Small I. indica were the only size class observed to undertake bounce-diving during the night. This was further highlighted by the small size class being the only one that exhibited a greater surface affinity during the day than at night. The repetitive-diving behaviour of small I. indica was often only to depths of 50-70 m, suggesting that if prey distribution is driving bounce-diving depth, then these smaller sized animals are likely to be targeting different prey sources than larger marlin. Gut content analysis of juvenile I. indica caught in the southwestern Pacific supports this theory, with small pelagic baitfishes (Amblygaster sirm and Sardinella gibbosa) identified as the dominant prey species of juveniles 10-40 kg. Both A. sirm and S. gibbosa occupy neritic waters (less than 40 m deep), suggesting that differences in diurnal patterns and depth use may well reflect an ontogenetic diet shift [bib_ref] Life histories of clupeids in north-eastern Australia: preliminary data, Williams [/bib_ref].
Differences in vertical habitat use by animals of different size classes have also been reported for other pelagic predators. For example, ontogenetic niche expansions observed in the depth distribution of small and medium sized tiger sharks (G. cuvier) were attributed to both a dietary shift and increased thermal inertia in larger sharks [bib_ref] Vertical movement patterns and ontogenetic niche expansion in the tiger shark, Galeocerdo..., Afonso [/bib_ref]. It was also observed that young of the year G. cuvier used neritic habitats to enhance growth before presumably shifting to oceanic habitats and food sources [bib_ref] Vertical movement patterns and ontogenetic niche expansion in the tiger shark, Galeocerdo..., Afonso [/bib_ref]. This may be similar for small I. indica, whose use of the water column may also be in part owing to their occupancy of neritic waters [bib_ref] Movements and variations in early year-class strength of black marlin (Makaira indica)..., Pepperell [/bib_ref]. As a result, depth use by I. indica and their prey species may be physically restricted by the shallow benthos of coastal waters. In the light of this, in this study, it was observed that when small fish occupied waters off the continental shelf (greater than 2000 m), they were seldom observed to dive to below 100 m [fig_ref] Figure 1 6: rsos [/fig_ref]. This indicates that while prey distribution is likely to provide a motive for ontogenetic niche expansion, in deeper waters it is likely that environmental factors play a role in limiting the extent of vertical habitat use.
## Physical drivers of depth use
Temperature influences the distribution and development of marine finfish, with preferred temperature ranges often coinciding with optimal growth rates [bib_ref] Effects of temperature on cardiac function in teleost fish, Kalinin [/bib_ref] [bib_ref] Temperature preference versus acclimation in fishes: selection for changing metabolic optima, Kelsch [/bib_ref]. For any fish species, optimal performance (e.g. oxygen uptake, heart rate and stroke volume, skeletal muscle contractility and cellular processes) is dependent on body temperature. For ectothermic species, as water temperature falls below their particular optimum operating temperature, they become physiologically compromised, which affects swimming performance [bib_ref] Cardiac function in an endothermic fish: cellular mechanisms for overcoming acute thermal..., Shiels [/bib_ref]. Therefore, it is not surprising that temperature has been identified as one of the primary physical factors associated with depth use of billfishes [bib_ref] Vertical habitat use of sailfish (Istiophorus platypterus) in the Atlantic and eastern..., Hoolihan [/bib_ref]. The physiological effects of temperate also appears to influence vertical habitat use of I. indica, as observed in this study, where increases in depth use were associated with SST and MLD, suggesting an expansion of the thermal niche for larger individuals. The effect of temperature on I. indica was similar to that observed in T. albacares, where greater time spent at depths by larger fish was accredited to their enhanced physiological tolerance to low temperatures and DO concentrations [bib_ref] 2014 Movements, behavior, and habitat utilization of yellowfin tuna (Thunnus albacares) in..., Schaefer [/bib_ref]. In the current study, the maximum diving depth and time-at-depth increased as I. indica moved into warmer equatorial waters, consistent with the effects of temperature on performance [bib_ref] Fisheries conservation on the high seas: linking conservation physiology and fisheries ecology..., Horodysky [/bib_ref].
It has been suggested that the temperature of the heart may play an important role in limiting the vertical movement of tunas and billfish [bib_ref] How water temperature really limits the vertical movements of tunas and billfishes-it's..., Brill [/bib_ref]. In Pacific bluefin tuna (Thunnus orientalis), the effect of prolonged exposure of the heart to cold temperatures has been shown to drive collapse of cardiac output [bib_ref] Cardiac function in an endothermic fish: cellular mechanisms for overcoming acute thermal..., Shiels [/bib_ref]. This may also be the case for istiophorid billfishes, as like tunas, the heart in marlin consists of a high-pressure, high stroke rate pump [bib_ref] Physiological investigations of marlin, Daxboeck [/bib_ref]. The heart in M. nigricans comprises about 0.15% of the body mass [bib_ref] Physiological investigations of marlin, Daxboeck [/bib_ref] and is relatively large compared with most other fishes. Marlin have well-developed coronary circulation which is unusual among teleost fishes, but common in other teleosts capable of sustained, fast swimming. The coronary arteries arise from efferent branchial arteries (cranial source) and the dorsal aorta (caudal source), and deliver oxygenated blood to the heart. This blood, having recently passed through the gills, which are efficient heat exchangers, will be close to ambient water temperature [bib_ref] Heat and oxygen exchange in the rete mirabile of the bluefin tuna,..., Carey [/bib_ref] [bib_ref] Warm-bodied fish, Carey [/bib_ref]. Therefore, as marlin descend into cooler waters, there will be a rapid cooling of the heart that will affect cardiac performance, which is likely to be size-independent. Therefore, if the temperature of a marlin's heart was the limiting factor, it would be expected that the diving of all I. indica size classes would be restricted to a similar depth. Our finding is in contrast to this, as vertical habitat use (temperature range and depth range) expanded with body mass of tagged animals. As a result, if I. indica vertical habitat use is restricted by temperature, then it is likely that other aspects of their physiology, such as skeletal muscle contractile performance or lactic acid build-up in the tissues, represent the limiting factor [bib_ref] Fisheries conservation on the high seas: linking conservation physiology and fisheries ecology..., Horodysky [/bib_ref]. Alternatively, it could be that marlin-diving behaviour is dictated by ecological factors such as prey availability and that it is not necessary for small marlin to access the colder mesopelagic waters as sufficient prey is available in the upper photic zone. However, in the absence of controlled data on the effect of temperature on istiophorid billfish cardiac (and skeletal) muscle, the physiological response of billfish to environmental change is restricted to that which may be surmised from tagging studies.
As with temperature, DO in surrounding water is a physical factor critical for maintaining physiological function in teleost fishes [bib_ref] Physiological and behavioral reactions of fishes to temperature change, Crawshaw [/bib_ref] [bib_ref] Influence of temperature and oxygen on the distribution of blue marlin (Makaira..., Carlisle [/bib_ref]. Despite billfishes having specialized gills for optimizing gas exchange, oxygen availability is still likely to affect their diving physiology owing to the high metabolic demands of being a large very active pelagic predator [bib_ref] Gill specializations in high-performance pelagic teleosts, with reference to striped marlin (Tetrapturus..., Wegner [/bib_ref]. Tagging studies on istiophorids off the tropical eastern Pacific have shown that hypoxic waters are associated with a compression of the vertical habitat used by I. platypterus and M. nigricans [bib_ref] Ocean scale hypoxia-based habitat compression of Atlantic istiophorid billfishes, Prince [/bib_ref]. Our findings also identified DO as a strong predictor of I. indica vertical habitat use in the southwestern Pacific. The presence of oxygen-rich waters in the southwestern Pacific and interaction between physical factors on the physiological response probably contributed to the nonlinear relationship between MD and DO 100 m . In the central Pacific where the water column is also characterized by high oxygen availability, changes in DO have been found to influence the movement of tagged M. nigricans when fish were already physiologically stressed by operating outside their optimal thermal window [bib_ref] Influence of temperature and oxygen on the distribution of blue marlin (Makaira..., Carlisle [/bib_ref]. Thus, as oxygen availability and temperature both decrease with increasing depth, this combination is likely to result in physiological stresses that ultimately limit available vertical habitat [bib_ref] Influence of temperature and oxygen on the distribution of blue marlin (Makaira..., Carlisle [/bib_ref] [bib_ref] Impacts of hypoxia on the structure and processes in pelagic communities (zooplankton,..., Ekau [/bib_ref].
Understanding the effects of physical factors on the depth use of pelagic fishes is imperative given recent declines in the DO content of oceans because of global warming [bib_ref] Decline in global oceanic oxygen content during the past five decades, Schmidtko [/bib_ref]. The equatorial Pacific Ocean has been associated with the largest declines in oxygen content of any oceanic region [bib_ref] Decline in global oceanic oxygen content during the past five decades, Schmidtko [/bib_ref]. This decline, combined with associated changes in water temperatures, could drive changes in both the vertical and horizontal distribution of I. indica stocks regularly accessing these waters [bib_ref] Decline in global oceanic oxygen content during the past five decades, Schmidtko [/bib_ref]. Along the east coast of Australia, the effects of ocean warming have reportedly shifted the preferred surface habitat of juvenile I. indica southward by approximately 88 km decade −1 and as much as approximately 300 km during El Nino years [bib_ref] Dynamic habitat suitability modelling reveals rapid poleward distribution shift in a mobile..., Hill [/bib_ref]. While considerable uncertainty remains as to how billfishes will respond to environmental change, changes to the migration phenology, spawning, vertical distribution and survival rate of larvae have all been suggested [bib_ref] Effects of ocean warming on Pelagic tunas, a review, Gilman [/bib_ref]. If pelagic species are unable to adapt to environmental change, it could also have broad-scale implications for fisheries, including a shift in global fishing effort, increased susceptibility to fishing gear (i.e. longline and gillnets) and even collapse of some fisheries [bib_ref] Expansion of oxygen minimum zones may reduce available habitat for tropical pelagic..., Stramma [/bib_ref].
In addition to the effects of temperature and oxygen availability, the other physical factor found to be a strong predictor of I. indica vertical habitat use was SSHd. Changes in SSH indicate the presence of mesoscale ocean features, resulting in SSHd increases often being associated with fronts or eddies [bib_ref] Using multi-sensor satellite remote sensing and catch data to detect ocean hot..., Zainuddin [/bib_ref]. Consequently, SSH is commonly investigated as a predictive variable for horizontal movement in billfish and has been linked with shifting species distributions and noted increases in CPUE [bib_ref] Dynamic habitat suitability modelling reveals rapid poleward distribution shift in a mobile..., Hill [/bib_ref] [bib_ref] Comparative influence of ocean conditions on yellowfin and Atlantic bluefin tuna catch..., Teo [/bib_ref]. The recent identification of SSH as a significant variable in depth use models for L. guttatus and K. audax suggests that cyclonic eddy systems may also play an important role in determining how large pelagic predators use the water column [bib_ref] Habitat characterization for striped marlin in the Pacific Ocean, Lam [/bib_ref] [bib_ref] Vertical movement and habitat of opah (Lampris guttatus) in the central North..., Polovina [/bib_ref]. However, this is not surprising as increased levels of primary productivity during the formation of eddies drive the aggregation of popular prey items [bib_ref] Using multi-sensor satellite remote sensing and catch data to detect ocean hot..., Zainuddin [/bib_ref]. The association of SSH anomalies with changes in vertical habitat use may therefore be reflective of feeding behaviour during periods of high prey availability.
# Conclusion and recommendations
Advancements in the field of wildlife telemetry have led to a growing number of investigations on the vertical habitat use of pelagic fishes. Our results add new insights to this knowledge base by revealing fundamental differences in the vertical habitat use of different sized I. indica. Future investigations of intraspecific size differences in other species of billfish will be important to identify whether this relationship exists across the broader species group or varies among species. Furthermore, the differences in diving behaviour of I. indica observed between regions highlight the need for tagging
[fig] 2: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/fig]
[fig] 4: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/fig]
[fig] 5: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/fig]
[fig] Figure 1 6: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Typical week long dive profiles of varying sized I. indica: (a) dive profile of a 150 kg medium; (b) 90 kg intermediate; and (c) 30 kg small. Light shading identifies night-time hours (19.00-05.00 h) as defined by light-level data. All dive profiles were in waters off the continental shelf with a bathymetric depth greater than 1000 m. [/fig]
[fig] 7: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/fig]
[fig] 8: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/fig]
[fig] Figure 3: Temperature at depth use of I. indica. [/fig]
[fig] 9: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/fig]
[fig] Figure 6 11: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (a) Percentage time-at-depth less than 150 m by I. indica. Tag observations are binned to generate average values in a 1°× 1°g rid and plotted in false colour (map). (b-e) Estimated individual effects (solid line) of environmental covariates on the percentage time at depth less than 150 m. Shaded areas show 95% confidence limits. Ticks on x-axis denote values for which there are data. To aid visualization, a horizontal line is added at 0 on the y-axis. Positive values on y-axis mean higher percentage. [/fig]
[fig] 13: rsos.royalsocietypublishing.org R. Soc. open sci. 4: 170694 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/fig]
[table] Table 1: Summary of GAMM results on vertical habitat use of I. indica. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . [/table]
|
10.1074/JBC.M102736200
|
CCBY
| null |
11406625
|
s2orc_pubmed_articles
|
Saposin D Solubilizes Anionic Phospholipid-containing Membranes*
Saposin (Sap) D is a late endosomal/lysosomal small protein, generated together with three other similar proteins, Sap A, B, and C, from the common precursor, prosaposin. Although the functions of saposins such as Sap B and C are well known (Sap B promotes the hydrolysis of sulfatides and Sap C that of glucosylceramide), neither the physiological function nor the mechanism of action of Sap D are yet fully understood. We previously found that a dramatic increase of Sap D superficial hydrophobicity, occurring at the low pH values characteristic of the late endosomal/lysosomal environment, triggers the interaction of the saposin with anionic phospholipid-containing vesicles. We have presently found that, upon lipid binding, Sap D solubilizes the membranes, as shown by the clearance of the vesicles turbidity. The results of gel filtration, density gradient centrifugation, and negative staining electron microscopy demonstrate that this effect is due to the transformation of large vesicles to smaller particles. The solubilizing effect of Sap D is highly dependent on pH, the lipid/saposin ratio, and the presence of anionic phospholipids; small variations in each of these conditions markedly influences the activity of Sap D. The present study documents the interaction of Sap D with membranes as a complex process. Anionic phospholipids attract Sap D from the medium; when the concentration of the saposin on the lipid surface reaches a critical value, the membrane breaks down into recombinant small particles enriched in anionic phospholipids. Our results suggest that the role played by Sap D is more general than promoting sphingolipid degradation, e.g. the saposin might also be a key mediator of the solubilization of intralysosomal/late endosomal anionic phospholipidcontaining membranes.
Sap D is a member of a family of four similar glycoproteins, Sap A, B, C, and D, generated from a common precursor, prosaposin, in late endosomes/lysosomes. The four saposins have been the subject of numerous structural and functional studies. They are small homologous proteins (each contains about 80 amino acids), have six cysteine residues at similar positions, and share the same disulfide structure, similar to that of a group of saposin-like proteins such as surfactant protein B, Entamoeba histolytica pore-forming peptides, and NK-lysin. The maintenance of the disulfide struc-ture is essential for the functional properties of saposins and saposin-like proteins.
The involvement of saposins in the lysosomal degradation of sphingolipids has been unequivocally demonstrated. In fact the deficit of prosaposin and, thus, of the four saposins causes the lysosomal accumulation of several sphingolipids such as ceramide, glucosylceramide, lactosylceramide, and ganglioside GM3. Moreover, the description of a variant form of Gaucher's disease due to a deficit of Sap C and of a variant form of metachromatic leukodystrophy due to a deficit of Sap B indicates that the two saposins are specifically involved in the degradation of glucosylceramides and sulfatides, respectively .
It has been reported that besides promoting sphingolipid degradation, saposins can have other functions. In fact, the four saposins as well as prosaposin have been described as serving as ganglioside binding and transport proteins. Moreover, prosaposin is a neurotrophic factor with neurotrophic activity localized within its Sap C domain.
For a long time the current hypothesis on the mechanism of action of saposins involved the concept that the activation of sphingolipid degradation was due to the interaction of saposins either with sphingolipids (e.g. Sap B with sulfatides) or with sphingolipid hydrolases (e.g. Sap C with glucosylceramidase). With a series of papers we have demonstrated that this was not the case at least for Sap C, whose main target is neither glucosylceramidase nor glucosylceramide but, instead, phospholipid membranes. Sap C, on interaction with anionic phospholipid-containing membranes, causes a dramatic destabilization of the lipid surface, that in turn favors the association of glucosylceramidase and the enzymatic degradation of glucosylceramide. Beside modulating the glucosylceramidase interaction with membranes, Sap C has also the capacity to aggregate and fuse anionic phospholipid vesicles at low pH values.
Recently, we demonstrated that Sap D, like Sap C, strongly binds to membranes and that anionic phospholipids promote and modulate its interaction with lipid surfaces. These similarities contrast with the different physiological role played by the two saposins, namely, Sap C is involved in glucosylceramide degradation, whereas Sap D has been proposed as an activator of ceramide or sphingomyelin degradation . A direct interaction of Sap D with these sphingolipids seems unlikely since neither the presence of ceramide nor that of sphingomyelin increases the affinity of Sap D for lipid surfaces. On the other hand, it is reasonable to suppose that the high affinity of Sap D for anionic phospholipids, as reflected in its behavior in vitro, will in large measure dictate its behavior in vivo.
The present study was originally conceived to search for differences between the membrane binding properties of Sap C and D, which might account for their different physiological roles. By comparing the interaction of the two saposins with vesicles, we met an unexpected property of Sap D not yet reported for any saposin, namely, the ability to disrupt anionic phospholipid-containing membranes with formation of small particles. This finding prompted us to characterize the phospholipid recombinant particles and to define the conditions that favor their formation. Our results indicate that Sap D has the potential to be a phospholipid membrane-solubilizing protein under physiological conditions.
## Experimental procedures
Materials-Phosphatidylcholine (PC) 1 from egg yolk and phosphatidylserine (PS) from bovine brain were from Avanti Polar Lipids, Inc. (Alabaster, AL). Phosphatidylinositol (PI) from bovine liver and cholesterol (Chol) were from Sigma. L-␣-Dipalmitoyl [2-palmitoyl-9,10-3 H(N)]PC (82 Ci/mmol) was from PerkinElmer Life Sciences, and 1, 2-dioleoyl-L-3-phosphatidyl-L-[3-14 C]serine (56 mCi/mmol) was from Amersham Pharmacia Biotech. All other chemicals were of the purest available grade.
Sap C and Sap D Preparation-Sap C and Sap D were purified from the spleens of patients with Type 1 Gaucher's disease after a previously reported procedure; it consisted essentially of heat treatment of a water homogenate followed by ion exchange chromatography on DEAE-Sephacel, gel filtration on Sephadex G-75, and reverse-phase high pressure liquid chromatography on a protein C4 column (Vydac). The purity of the final preparations of Sap C and Sap D was verified by N-terminal sequence analysis, SDS-polyacrylamide gel electrophoresis, and electrospray mass spectrometry.
Vesicle Preparation-Multilamellar vesicles (MLVs) were prepared by mixing appropriate amounts of lipids dissolved in chloroform and evaporating the solvent under N 2 . PC was supplemented with [ 3 H]PC and, in some experiments, PS with [ 14 C]PS. The specific activities of both phospholipids were in the range of 2 ϫ 10 6 dpm/mg. The dry lipids were dispersed by Vortex mixing in 2 mM L-histidine, 2 mM TES, 150 mM NaCl, 1 mM EDTA, pH 7.4. The suspension was submitted to 10 cycles of freezing and thawing and then blended by a Vortex mixer for 3-5 min. The vesicle concentration was determined by radioactivity measurements.
Light-scattering Experiments-The changes of vesicle turbidity were monitored by 90°light scattering. A FluoroMax-2 spectrofluorimeter equipped with a constant temperature cell holder and stirrer (Spex Industries Inc., Edison, NJ) was used with excitation and emission wavelengths set at 600 nm (slit widths of 3 nm). The temperature of the cuvette holder was maintained at 37°C. All solutions were preincubated at this temperature. Lipid vesicles were added to 1 ml of buffer A (10 mM acetate, 150 mM NaCl, 1 mM EDTA), pH 4.5, and monitored for 5 min. Saposin solutions were then added, and the change in light scattering was monitored as function of time.
Gel Permeation-Different mixtures of Sap D and labeled MLVs (see above) in buffer A, pH 4.5, were incubated for 30 min at 37°C and then applied to a Sepharose CL-4B column (1 ϫ 22 cm) pre-equilibrated and eluted at room temperature with the same buffer. In specified experiments, buffer A was adjusted to different pH values. The flow rate was 0.25 ml/min. Fractions of 0.4 ml were collected. The lipid distribution was determined by measuring the radioactivity of the fractions. When two labeled phospholipids were present, double isotope counting conditions were adopted.
Density Gradient Centrifugation-Samples for density gradient centrifugation were prepared as for the gel permeation experiments. A discontinuous KBr density gradient was formed in 13 ϫ 51-mm Ultra-Clear centrifuge tubes. The KBr solutions were prepared in buffer A, pH 4.5. Each gradient contained, from the bottom, 0.3 ml of KBr solution (density 1.39 g/ml), 0.3 ml of sample solution adjusted to a density of 1.30 g/ml with KBr, 1.8 ml of KBr solution (density 1.21 g/ml), 1.8 ml of KBr solution (density 1.06 g/ml), and 0.6 ml of KBr solution (density 1.02 g/ml). Samples were centrifuged in a Beckman SW 55 Ti rotor at 46,000 rpm (250,000 ϫ g), 25°C, for 4 h. After centrifugation, each sample was separated into 16 fractions (0.3 ml each). The lipid distribution was determined by measuring the fraction radioactivity.
Binding of Sap D to Vesicles-To compare the binding of Sap D to large and small vesicles, the fractions from the Sepharose CL-4B col-umns (see above) were concentrated by transferring them to Microcon-YM-3 centrifugal filter devices (molecular mass cut-off 3 kDa) and centrifuging at 13,000 ϫ g until all the liquid except 20 l was passed through the filter. The presence of Sap D in each retentate was tested by SDS-polyacrylamide gel electrophoresis (see the next paragraph).
SDS-Polyacrylamide Gel Electrophoresis-SDS-polyacrylamide gel electrophoresis was performed with 15% acrylamide-separating gels and 4.5% stacking gels. After electrophoresis, Sap D was visualized with the silver-staining kit Protein (Amersham Pharmacia Biotech).
Electron Microscopy-Different mixtures of Sap D and 20% PS containing MLVs in buffer A, pH 4.5, were incubated for 30 min at 37°C and then adjusted to pH 7.0 with 0.1 M imidazole. The samples were applied to carbon-coated copper grids (400 mesh). After 30 s on the grid, the samples were dried, negatively stained with 2% sodium phosphotungstate, and observed by transmission electron microscopy. Micrographs were taken using a Philips 208S electron microscope operating at 80 kV.
# Results
## Sap d induces clearance of anionic phospholipid-containing
Vesicles-We have previously shown that, within the four saposins, Sap C and Sap D have the strongest capacity to destabilize anionic phospholipid-containing vesicles at acidic pH values. Moreover, Sap C, under appropriate conditions, is able to aggregate and fuse PS vesicles. To address the question whether Sap C and D exert different actions on membranes, the two saposins were separately added to phospholipid MLVs, and the turbidity of the vesicles was followed by light-scattering measurements. Representative curves of the changes in scattered light intensity are shown in [fig_ref] FIG. 1: MLV light scattering as a function of time after the addition of... [/fig_ref]. The addition of Sap C to MLVs containing 20% PS caused a timedependent increase in light-scattering intensity, confirming the Sap C capacity to promote aggregation of anionic phospholipid-containing membranes. In contrast, the addition of Sap D to the same MLV preparation induced clearance of the suspension. The rate of turbidity clarification decreased on lowering the membrane content of PS and was dramatically reduced when MLVs were devoid of anionic phospholipids. These results strongly suggest that Sap D is able to disrupt membranes and that anionic phospholipids play a critical role in this process. [fig_ref] FIG. 2: Effect of the lipid composition on the elution profiles of MLV-Sap D... [/fig_ref]. Also, in the presence of another anionic phospholipid such as PI, the addition of Sap D shifted about 40% of vesicles to a second peak (maximum at fraction 34) [fig_ref] FIG. 2: Effect of the lipid composition on the elution profiles of MLV-Sap D... [/fig_ref]. The comparison of [fig_ref] FIG. 2: Effect of the lipid composition on the elution profiles of MLV-Sap D... [/fig_ref] , B and C, suggests that the recombinant particles formed in the presence of PI are smaller than those formed in the presence of PS. To confirm that Sap C was unable to dissolve phospholipid membranes, a mixture of 10% PS-containing MLVs and Sap C was analyzed by gel permeation at pH 4.5. The elution pattern of MLVs was not affected by the presence of Sap C (data not shown).
When the PS percentage in the vesicles was increased from 10 to 20%, the peak of the recombinant particles formed after the addition of Sap D sharpened and shifted toward lower molecular masses (compare [fig_ref] FIG. 2: Effect of the lipid composition on the elution profiles of MLV-Sap D... [/fig_ref]. This result is consistent with the light-scattering experiments showing that the addition of Sap D to 20%PS-containing MLVs induces a more rapid turbidity clarification than the addition to 10%PS-containing MLVs (see [fig_ref] FIG. 1: MLV light scattering as a function of time after the addition of... [/fig_ref]. Thus, the gel permeation experiments confirm that Sap D, but not Sap C, induces breakdown of anionic phospholipid-containing membranes, giving rise to smaller particles whose size and extent of formation depend on the type and percentage of the anionic phospholipid.
Effect of pH on the Formation of Recombinant Particles-Since we previously found that the interaction of Sap D with membranes is favored by an acidic environment, the pH dependence of the formation of recombinant particles was evaluated [fig_ref] FIG. 3: Effect of pH on the elution profiles of MLV-Sap D mixtures on... [/fig_ref]. When the gel filtration analysis of the SapDvesicles mixtures was performed at pH 6.0, essentially all the 20% PS-containing vesicles eluted in one peak, with a maximum at fraction 18, corresponding to the MLVs. At lower pH values, progressively larger proportions of the vesicles appeared as a second peak (maximum at fraction 32-33), indicating that low pH values are required for the formation of the small particles.
To check whether the recombinant particles had the same phospholipid composition as the original MLVs, the vesicles were doubly labeled in PC and PS. [fig_ref] FIG. 3: Effect of pH on the elution profiles of MLV-Sap D mixtures on... [/fig_ref] shows that the small particles have a significantly lower PC/PS ratio (r ϭ 2.4) than the parent vesicles (r ϭ 2.7). The remaining MLVs eluting in the void volume, on the other hand, were enriched in PC (r ϭ 3.0). Apparently Sap D selects PS over PC during the reaction with MLVs vesicles, promoting the pH-dependent formation of recombinant small particles richer in PS than the original membranes.
Effect of the Lipid/Sap D Molar Ratio on the Formation of Recombinant Particles-To ascertain at which lipid/saposin molar ratio the Sap D-induced breakdown of membranes becomes evident, a gel permeation analysis of Sap D mixed with different amounts of PS-containing MLVs was performed [fig_ref] FIG. 4: Effect of the lipid/saposin molar ratios on the elution profiles of MLV-Sap... [/fig_ref]. At a lipid/saposin molar ratio of 100:1, the presence of the saposin poorly affected the elution profile of the MLVs [fig_ref] FIG. 4: Effect of the lipid/saposin molar ratios on the elution profiles of MLV-Sap... [/fig_ref]. At a molar ratio of 50:1, about 35% PC and 40% PS eluted as a broad shoulder after the MLVs peak [fig_ref] FIG. 4: Effect of the lipid/saposin molar ratios on the elution profiles of MLV-Sap... [/fig_ref]. When the molar ratio was further decreased to 12.5:1, most of the MLVs was transformed into particles eluting in the second peak (a maximum at fraction 33) [fig_ref] FIG. 4: Effect of the lipid/saposin molar ratios on the elution profiles of MLV-Sap... [/fig_ref]. Judging from the elution profiles, the small particles were generated when the lipid/ saposin molar ratio was lower than 100:1.
Resolution of Recombinant Particles by Density Gradient Ultracentrifugation-To further characterize the recombinant particles and to determine their density, the Sap D-induced transformation of MLVs was analyzed by density gradient centrifugation. When the 20% PS-containing vesicles were subjected to centrifugation in a KBr gradient at pH 4.5, the lipids floated at the top, whereas after incubation with Sap D at a lipid/saposin molar ratio of 25:1, a major portion of the vesicles was shifted to a peak centered at fraction 13, corresponding to a density of about 1.23 . At a lipid/Sap D molar ratio of 12.5:1, an almost complete transformation of the original MLVs was observed . Vice versa, the Sap C addition to the PS-containing MLVs did not cause a significant transformation of the vesicles . The density gradient banding profiles of the recombinant particles were consistent with the elution patterns previously observed for the corresponding gel filtration samples, the peak at higher density (d ϭ 1.23) corresponding to the peak eluting at lower molecular masses (a maximum at fractions 32-33) (compare [fig_ref] FIG. 3: Effect of pH on the elution profiles of MLV-Sap D mixtures on... [/fig_ref] , upper left panel, and [fig_ref] FIG. 4: Effect of the lipid/saposin molar ratios on the elution profiles of MLV-Sap... [/fig_ref].
Analysis of the Sap D Binding Affinity to MLVs and Recombinant Particles-To check whether Sap D forms stable complexes with the vesicles and/or with the recombinant particles, the peaks obtained after gel permeation analysis of 20% PScontaining MLV-Sap D mixtures were analyzed by electrophoresis for the presence of the saposin. As shown in [fig_ref] FIG. 6: Electrophoretic analysis of the Sap D binding to vesicles [/fig_ref] , upper panel, when the lipid/Sap D molar ratio of the mixture was 100:1, Sap D was found under the single MLV peak (maximum at fraction 18) eluting from the gel permeation column under this condition (see [fig_ref] FIG. 4: Effect of the lipid/saposin molar ratios on the elution profiles of MLV-Sap... [/fig_ref]. At a lipid/saposin molar ratio of 25:1, when a first MLV peak and a second peak of small recombinant particles elute from the column (see [fig_ref] FIG. 3: Effect of pH on the elution profiles of MLV-Sap D mixtures on... [/fig_ref] , upper left panel), most of the Sap D was found in the second rather than in the first peak [fig_ref] FIG. 6: Electrophoretic analysis of the Sap D binding to vesicles [/fig_ref]. The same situation was observed when the lipid/saposin molar ratio was further decreased to 12.5:1, and most of the MLVs was transformed into small particles [fig_ref] FIG. 6: Electrophoretic analysis of the Sap D binding to vesicles [/fig_ref]. These results indicate that Sap D remains preferentially associated with the small particles obtained after the conversion of the large vesicles.
Electron Microscopic Structure of the Recombinant Particles-The ultrastructure of the recombinant particles formed on incubation of PS-containing MLVs with Sap D was examined by negative staining electron microscopy . Control MLVs are spherical in shape, with a variable diameter ranging from 400 to 1000 nm . The incubation of MLVs with Sap D at a lipid/saposin molar ratio of 25:1 gave mixtures of small spherical vesicular structures of varying sizes. The majority of the vesicles had diameters ranging from 20 to 50 nm , but vesicles with diameter larger than 200 nm were also observed (data not shown). At a lower lipid/Sap D molar ratio (12.5:1), the diameter of the vesicles further decreased , and discoidal structures forming rouleaux were clearly discerned . It is thus evident that the size and morphology of the recombinant particles depend on the Sap D concentration on the membrane. Saposins are involved in the lysosomal degradation of several sphingolipids in late endosomes/lysosomes. Most likely a crucial step in the cascade of events during the hydrolysis of these lipids is the interaction of saposins with the membranes where sphingolipids are located. We have previously demonstrated that at low pH values similar to those present in the endosomal/lysosomal environment, at least two saposins (Sap C and D) have a high affinity for phospholipid membranes, particularly in the presence of anionic phospholipids. In the present study several independent experimental approaches have provided the first demonstration that Sap D has the capacity to break down anionic phospholipid-containing membranes, giving rise to small recombinant particles. The criteria for this conclusion include (a) the ability of Sap D to transform MLVs into structures that scatter less light, (b) the isolation by gel filtration or by density gradient centrifugation of stable complexes of saposin and lipid particles having a different size and density than the starting membranes, and (c) the detection by electron microscopy of small spherical and discoidal structures after incubation of MLVs with Sap D.
The membrane-disrupting activity of Sap D is expressed only when the pH of the medium is below 5.5. This observation is consistent with our previous studies showing that Sap D un-dergoes a hydrophobic transition at low pH values, leading to interaction with membranes. The change of hydrophobicity of Sap D most likely has a physiological significance because the low pH values found in late endosomes/lysosomes correspond to those required by the saposin for binding, perturbing, and solubilizing anionic phospholipid-containing membranes. Even a small pH change in the acidic organelles might markedly affect the properties of Sap D.
The formation of recombinant lipid particles induced by Sap D is dependent on the presence of negatively charged phospholipids. We observed in the past that variation of the percentage of anionic phospholipids modulates the strength of the saposinmembrane interaction. The association of Sap D with membranes and the depth of its insertion are most likely governed by electrostatic interactions between the head groups of Another critical factor for the solubilization of the membranes is the concentration of Sap D on the lipid surface. In fact, our results show that large anionic phospholipid-containing vesicles bind the saposin up to a lipid/saposin molar ratio of about 100:1 before they break down into small recombinant complexes. It can be envisaged that electrostatic interactions provided by the anionic phospholipids attract Sap D from the medium and concentrate it on the lipid surface. When the local concentration of Sap D reaches a critical value the membrane breaks down, giving rise to small particles enriched in anionic phospholipids as compared with the original membrane. This last observation suggests that Sap D, upon association with vesicles, favors the formation of domains enriched in anionic phospholipids.
Our past investigation on the structural and membrane binding properties of saposins showed that Sap D and Sap C share several important features as follows. They have the ability to exist in both water-soluble and membrane-bound forms (16); they undergo a dramatic increase of hydrophobicity at pH 5.0 -5.5, as indicated by phase partitioning experiments in Triton X-114 and fluorescence measurements; they bind and perturb anionic phospholipid-containing large vesicles in a pH-dependent manner; and they have the same disulfide structure. On the other hand, the two saposins also exhibit profound differences. In fact, Sap C, but not Sap D, is able to promote the association of glucosylceramidase with large anionic phospholipid-containing vesicles reconstituting the enzyme activity, a finding consistent with the notion that only Sap C is the physiological activator of glucosylceramidase. Conversely, the present work demonstrates that Sap D can break down anionic phospholipid-containing membranes, an activity not exhibited by Sap C. Thus, our findings indicate that, although both saposins have a high affinity for phospholipid vesicles, the mode of Sap D interaction with membranes is remarkably different from that of Sap C.
Sap D has been proposed to have a role in ceramide degradation, since its addition to the culture medium of fibroblasts from a patient suffering from prosaposin deficiency resulted in the reduction of ceramide storage. Recently it has been reported that not only Sap D, but also Sap A and C, reduce the ceramide storage. Thus, a specific physiological function of Sap D in ceramide hydrolysis is still uncertain. We have previously shown that ceramide does not increase the affinity of Sap D for the membranes (6), the real target of the saposin being anionic phospholipids. The key role played by these lipids in governing the behavior of Sap D suggests that the saposin interacts with anionic phospholipids to elicit its biological effect(s).
In the late endosomes/lysosomes, where saposins are located, the presence of internal vesicles and structures has often been observed . Recently it has been found that the network of internal vesicles in the lumen of late endosomes contains large amounts of anionic phospholipids, especially lysobisphosphatidic acid and PI, forming specialized domains where specific proteins such as the multifunctional receptor for mannose 6-phosphate-bearing ligands preferentially segregate. In view of our past and present results on the role of anionic phospholipids in the interaction between Sap D and membranes, it is tempting to envisage that also Sap D segregates and concentrates onto the intraendosomal/lysosomal vesicles, where it might participate to their degradation by converting them into small recombinant particles. In cells of patients affected with prosaposin deficiency, where all saposins are missing, the accumulation of intraendosomal/lysosomal multivesicular structures has been observed. A putative role of Sap D in the degradation of anionic phospholipid-containing membranes inside acidic organelles would be consistent with this observation.
Interestingly, some of the Sap D properties characterized in the present paper are similar to those shown by exchangeable apolipoproteins, such as apoA-I, except that the Sap D properties are exhibited at low pH and the apolipoproteins properties are exhibited at neutral pH. In fact, apoA-I, under appropriate conditions, also has a marked ability to solubilize membranes, causing liposome structure to be disrupted so as to loose turbidity, especially in the presence of anionic phospholipids. Moreover, the Sap D-induced formation of small lipid particles resembles the recruitment of phospholipids by apoliprotein A-I to form apolipoprotein lipid spherical and discoidal recombinant particles that can be separated from the original vesicles by gel permeation or density gradient centrifugation . Exchangeable apolipoproteins have a functional role in the remodeling of lipoproteins. The apolipoprotein-like properties exhibited by Sap D at low pH values suggest that this saposin might have a role in the remodeling of phospholipid membranes inside acidic organelles.
In conclusion, the present work has highlighted for the first time the ability of Sap D to convert membranes of appropriate composition into small recombinant particles. This finding might give critical clues to the function of this saposin, whose precise physiological and pathogenic role is still relatively unknown. Our results suggest that Sap D might have more general functions than promoting the sphingolipid degradation, e.g. Sap D might be also involved in the processes of phospholipid membrane transformation in late endosomes/lysosomes.
[fig] FIG. 1: MLV light scattering as a function of time after the addition of Sap C (a) or Sap D (b-d). The light scattering of MLV solutions in buffer A, pH 4.5, was monitored as described under "Experimental Procedures" for 5 min. Then Sap C (a) or Sap D (b-d) were added as indicated by the arrow, and the changes in light scattering were followed with time. The MLV compositions were PC:Chol:PS, 55:25:20 (a and d), PC:Chol, 75:25 (b), or PC:Chol:PS, 65:25:10 (c). The final concentrations of lipids and saposins were 75 and 3 M respectively. The scattering intensity is expressed as I/I o , where I and I o are the scattering intensities in the presence and absence of saposins, expressed as percentage. The experiments were repeated at least three times with similar results. Resolution of Small Recombinant Particles by Gel Filtration-To look for further proof of a Sap D-induced breakdown of membranes, mixtures of vesicles and Sap D were analyzed by gel filtration on a Sepharose 4-B column. To characterize the role of anionic phospholipids, the experiments were performed with MLVs prepared with or without PS or PI. As shown in Fig. 2A, Sap D poorly affected the elution of MLVs composed of PC and Chol. Most of the vesicles eluted in a peak (maximum at fraction 18) in the void volume of the column, either in the presence or in the absence of the saposin. On contrast, Sap D induced the formation of a second broad peak centered at fractions 29 -30, when MLVs contained 10% PS [/fig]
[fig] FIG. 2: Effect of the lipid composition on the elution profiles of MLV-Sap D mixtures on Sepharose CL-4B columns. MLVs of different composition incubated in the absence (----) or in the presence ( ) of Sap D (30 g) to yield a lipid/saposin molar ratio of 25:1 were analyzed by gel filtration on Sepharose CL-4B columns. The samples were incubated and eluted in buffer A, pH 4.5, as described under "Experimental Procedures." Trace amounts of [ 3 H]PC have been included in the MLV preparations, and the elution of PC is given in terms of 3 H dpm. The compositions of MLVs are the following: A, PC:Chol, 75:25; B, PC:Chol:PS, 65:25:10; C, PC:Chol:PI, 65:25:10. The experiments repeated at least three times gave similar elution profiles. [/fig]
[fig] FIG. 3: Effect of pH on the elution profiles of MLV-Sap D mixtures on Sepharose CL-4B columns. Sap D (30 g) was mixed with MLVs (PC:Chol:PS, 55:25:20) to yield a lipid/saposin molar ratio of 25:1. Trace amounts of [ 3 H]PC and [ 14 C]PS have been included in the MLVs. The samples were incubated and eluted as in Fig. 2, except that buffer A was adjusted to different pH values as indicated. The elution of PC ( )and PS (----) is given in terms of 3 H and 14 C dpm, respectively. The molar ratios of PC/PS (q) are shown for the fractions across the two elution peaks in the upper left panel. The elution positions for MLVs and Sap D eluted separately are also indicated (arrows). The experiments repeated at least three times gave similar elution profiles. [/fig]
[fig] FIG. 4: Effect of the lipid/saposin molar ratios on the elution profiles of MLV-Sap D mixtures on Sepharose CL-4B columns. Sap D (30 g) was mixed with MLVs (PC:Chol:PS, 55:25:20) to yield lipid/saposin molar ratios of 100:1 (A), 50:1 (B), or 12.5:1 (C). Trace amounts of [ 3 H]PC and [ 14 C]PS have been included in the MLVs. The samples were incubated and eluted in buffer A, pH 4.5, as in Fig. 2. The elution of PC ( ) and PS (----) is given in terms of 3 H and 14 C dpm, respectively. The experiments repeated at least three times gave similar elution profiles. FIG. 5. Density gradient ultracentrifugation of MLVs in the absence (----) or presence ( ) of Sap D and Sap C. 30 g of Sap D (A and B) or Sap C (C) were mixed and incubated in buffer A, pH 4.5, with MLVs (PC:Chol:PS, 55:25:20) to yield a lipid/saposin molar ratio of 25:1 (A and C) or 12.5:1 (B). The density gradient ultracentrifugation was carried out as described under "Experimental Procedures." Trace amounts of [ 3 H]PC have been included in the MLV preparations, and the distribution of PC in the gradient is given in terms of 3 H dpm. Data are representative of experiments performed at least three times. anionic phospholipids and the cationic sites of the saposin. Both the anionic phospholipids used in the present work, namely PS and PI, favored the SapD-induced disruption of the membrane. Apparently, the action of Sap D on the lipid surface occurs in the presence of anionic phospholipids independently of their structure. [/fig]
[fig] FIG. 6: Electrophoretic analysis of the Sap D binding to vesicles. The presence of Sap D in fractions from Sepharose CL-4B columns was detected by electrophoretic analysis. The fractions corresponding to the vesicles peaks were concentrated and then subjected to SDS-polyacrylamide gel electrophoresis as described under "Experimental Procedures." Upper panel, fractions from the Sepharose CL-4B column depicted in Fig. 4A (lipid/Sap D molar ratio of 100:1). Middle panel, fractions from the column depicted in Fig. 3, upper left panel (lipid/Sap D molar ratio 25:1); lower panel, fractions from the column depicted in Fig. 4C (lipid/Sap D molar ratio 12.5:1). The electrophoretic patterns are representative of experiments performed at least three times. FIG. 7. Electron micrographs of negatively stained MLVs incubated in the absence (A) or in the presence (B and C) of Sap D. MLVs (PC:Chol:PS, 55:25:20) were incubated in buffer A, pH 4.5, in the absence (A) or in the presence of Sap D at a lipid/saposin molar ratio of 25:1 (B) or 12.5:1 (C). The samples were prepared for electron microscopy as described under "Experimental Procedures." The bar marker in A represents 300 nm, whereas in B and C it represents 100 nm. The inset in C is a ϫ3 magnification of the discoidal structure indicated by the arrow. [/fig]
|
10.3389/dyst.2022.10494
|
CCBY
|
10032351
|
36960404
|
s2orc_pubmed_articles
|
Quantification of Behavioral Deficits in Developing Mice With Dystonic Behaviors
Converging evidence from structural imaging studies in patients, the function of dystonia-causing genes, and the comorbidity of neuronal and behavioral defects all suggest that pediatric-onset dystonia is a neurodevelopmental disorder. However, to fully appreciate the contribution of altered development to dystonia, a mechanistic understanding of how networks become dysfunctional is required for early-onset dystonia. One current hurdle is that many dystonia animal models are ideally suited for studying adult phenotypes, as the neurodevelopmental features can be subtle or are complicated by broad developmental deficits. Furthermore, most assays that are used to measure dystonia are not suited for developing postnatal mice. Here, we characterize the earlyonset dystonia in Ptf1a Cre ;Vglut2 fl/fl mice, which is caused by the absence of neurotransmission from inferior olive neurons onto cerebellar Purkinje cells. We investigate motor control with two paradigms that examine how altered neural function impacts key neurodevelopmental milestonesThis is an open-seen in postnatal pups (postnatal day 7-11). We find that Ptf1a Cre ;Vglut2 fl/fl mice have poor performance on the negative geotaxis assay and the surface righting reflex. Interestingly, we also find that Ptf1a Cre ;Vglut2 fl/fl mice make fewer ultrasonic calls when socially isolated from their nests. Ultrasonic calls are often impaired in rodent models of autism spectrum disorders, a condition that can be comorbid with dystonia. Together, we show that these assays can serve as useful quantitative tools for investigating how neural dysfunction during development influences neonatal behaviors in a dystonia mouse model. Our data implicate a shared cerebellar circuit mechanism underlying dystonia-related motor signs and social impairments in mice.
# Introduction
Dystonia is a complex neurological movement disorder characterized by involuntary muscle contractions that can cause rigid limbs and/or twisting postures. These behaviors typically arise because agonist and antagonist muscles either co-contract or contract persistently, causing repetitive and sometimes sustained motions. The affected muscles can be found in a single body part as in focal dystonia or in multiple muscle groups as in generalized dystonia [bib_ref] Dystonia Updates: Definition, Nomenclature, Clinical Classification, and Etiology, Grütz [/bib_ref]. The motor behavioral features of dystonia are often the predominant signature of the disease and are thought to reflect the underlying neural dysfunctions that cause primary dystonia. However, dystonia can also occur as a secondary symptom in neurodevelopmental conditions, neurodegenerative diseases, or acquired neurologic dysfunction. Mechanistically, this diversity is an important consideration as dystonia onset can occur in patients of all ages. Current evidence indicates that the dystoniaassociated motor impairments arise from circuit deficits throughout the brain including the basal ganglia, cerebellum, thalamus, and motor cortex [bib_ref] The Anatomical Basis for Dystonia: the Motor Network Model, Jinnah [/bib_ref] [bib_ref] Current Opinions and Areas of Consensus on the Role of the Cerebellum..., Shakkottai [/bib_ref] [bib_ref] The Functional Neuroanatomy of Dystonia, Neychev [/bib_ref]. Despite the rapidly growing knowledge of the genetic and circuit bases of dystonia pathophysiology, the heterogeneity and complexity of the disease has hindered a full understanding of the etiology and neural deficits that cause the debilitating dystonia-associated symptoms. Specifically, how the altered behaviors arise during development remains unclear.
There is emerging evidence pointing towards impaired neurodevelopment as a key factor in some forms of pediatric dystonia [bib_ref] Hereditary Dystonia as a Neurodevelopmental Circuit Disorder: Evidence from Neuroimaging, Niethammer [/bib_ref]. Although certain hereditary childhood-onset dystonias have incomplete penetrance (DYT1 and DYT6), both manifesting and non-manifesting patients have abnormal neural circuit connectivity between multiple nodes within the motor circuit that are associated with dystonia. The main affected brain areas are thought to include the basal ganglia, thalamus, and cerebellum [bib_ref] Understanding the Anatomy of Dystonia: Determinants of Penetrance and Phenotype, Lerner [/bib_ref]. Thus, while the appearance of dystonia may rely on additional molecular and circuit modifying factors [bib_ref] Expression Profiling in Peripheral Blood Reveals Signature for Penetrance in DYT1 Dystonia, Walter [/bib_ref] , aberrant circuit development appears central to the behavioral expression when genetic mutations are associated with the disease. Furthermore, affected patients often display their first motor signs during childhood, whereas carriers who remain asymptomatic through childhood rarely develop symptoms later in life [bib_ref] TheDYT1 Phenotype and Guidelines for Diagnostic Testing, Bressman [/bib_ref]. This difference in behavioral onset could indicate a neurodevelopmental period during which network dysfunction leads to dystonia, coinciding with a critical developmental window that also appears relevant to other neurodevelopmental disorders including autism spectrum disorders [bib_ref] CNS Critical Periods: Implications for Dystonia and Other Neurodevelopmental Disorders, Li [/bib_ref]. In accordance with this hypothesis, a number of the recently identified dystonia-associated mutations involve genes that are also known for their role in neurodevelopment, including the autismassociated gene, CDH8 [bib_ref] Monogenic Variants in Dystonia: an Exome-wide Sequencing Study, Zech [/bib_ref] [bib_ref] Frequency and Phenotypic Spectrum of KMT2B Dystonia in Childhood: A Single-center Cohort..., Carecchio [/bib_ref] [bib_ref] A Novel Missense Mutation in GRIN2A Causes a Nonepileptic Neurodevelopmental Disorder, Fernández-Marmiesse [/bib_ref] [bib_ref] Childhood-onset Progressive Dystonia Associated with Pathogenic Truncating Variants in CHD8, Doummar [/bib_ref] [bib_ref] Loss-of-function Variants in HOPS Complex Genes VPS16 and VPS41 Cause Early-Onset Dystonia..., Steel [/bib_ref] [bib_ref] Recent Genetic Advances in Early-Onset Dystonia, Steel [/bib_ref]. In addition, dystonia is also a frequent comorbidity in infants with other genetic, neurodevelopmental disorders including Rett syndrome (MECP2 mutations) [bib_ref] Rett Syndrome as a Movement and Motor Disorder -A Narrative Review, Brunetti [/bib_ref] [bib_ref] Movement Disorders and Syndromic Autism: A Systematic Review, Bell [/bib_ref] , Partington syndrome (ARX mutations) [bib_ref] Infantile Spasms, Dystonia, and Other X-Linked Phenotypes Caused by Mutations in Aristaless..., Strømme [/bib_ref] , and as mentioned before, in an array of patients with autism spectrum disorders [bib_ref] A Neurological Model for Childhood Autism, Damasio [/bib_ref]. Together, these studies suggest that some iterations of pediatric dystonia may emerge from aberrant neurodevelopmental processes. To test this hypothesis in vivo, it is crucial to investigate how behavior is shaped during development in the context of dystonia.
Multiple reports now indicate that mouse models harboring loss-of-function mutations in genes that cause the most common hereditary pediatric-onset dystonias (DYT1 and DYT6) do not display the involuntary muscle contractions or abnormal posturing observed in infants with the disorder [bib_ref] Abnormalities of Motor Function, Transcription and Cerebellar Structure in Mouse Models of..., Ruiz [/bib_ref] [bib_ref] Loss of the Dystonia-Associated Protein torsinA Selectively Disrupts the Neuronal Nuclear Envelope, Goodchild [/bib_ref] [bib_ref] Mouse Models of torsinA Dysfunction, Dauer [/bib_ref]. However, mice that have brain-restricted loss of Tor1a or knock-in mutations that mimic the human condition show several limb and body motor abnormalities that are observed in human DYT1 [bib_ref] TorsinA Hypofunction Causes Abnormal Twisting Movements and Sensorimotor Circuit Neurodegeneration, Liang [/bib_ref]. Mouse models with overt dystoniaassociated impairments can also be induced by infusing drugs into the brain [bib_ref] Abnormal Cerebellar Signaling Induces Dystonia in Mice, Pizoli [/bib_ref] [bib_ref] The Neural Substrates of Rapid-Onset Dystonia-Parkinsonism, Calderon [/bib_ref] [bib_ref] Abnormal High-Frequency Burst Firing of Cerebellar Neurons in Rapid-Onset Dystonia-Parkinsonism, Fremont [/bib_ref] or by downregulating the expression of dystonia-associated genes [bib_ref] Aberrant Purkinje Cell Activity Is the Cause of Dystonia in a shRNA-Based..., Fremont [/bib_ref] [bib_ref] Acute Cerebellar Knockdown of Sgce Reproduces Salient Features of Myoclonus-Dystonia (DYT11) in..., Washburn [/bib_ref] in adult mice. Electrophysiological recordings in the brains of rodent models with overt dystonic postures showed irregular burst-like firing patterns in cerebellar neurons, suggesting abnormal cerebellar function as a critical and likely shared feature of dystonia pathogenesis [bib_ref] Abnormal Cerebellar Signaling Induces Dystonia in Mice, Pizoli [/bib_ref] [bib_ref] Abnormal High-Frequency Burst Firing of Cerebellar Neurons in Rapid-Onset Dystonia-Parkinsonism, Fremont [/bib_ref] [bib_ref] Aberrant Purkinje Cell Activity Is the Cause of Dystonia in a shRNA-Based..., Fremont [/bib_ref] [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref] [bib_ref] Single-unit Activity of Cerebellar Nuclear Cells in the Awake Genetically Dystonic Rat, Ledoux [/bib_ref] [bib_ref] Abnormal Spontaneous and Harmaline-Stimulated Purkinje Cell Activity in the Awake Genetically Dystonic..., Ledoux [/bib_ref] [bib_ref] Cerebellar Dysfunction as a Source of Dystonic Phenotypes in Mice, Brown [/bib_ref]. In line with these findings, abnormal cerebellar neuron function has been confirmed in non-manifesting genetic models for DYT1 and DYT6 [bib_ref] Abnormal Cerebellar Function and Tremor in a Mouse Model for Non-manifesting Partially..., Van Der Heijden [/bib_ref] [bib_ref] The Abnormal Firing of Purkinje Cells in the Knockin Mouse Model of..., Liu [/bib_ref]. Furthermore, cerebellar dysfunction is also compatible with childhood-onset dystonia of other etiologies; the protracted timeline of cerebellar development [bib_ref] Interactions between Purkinje Cells and Granule Cells Coordinate the Development of Functional..., Van Der Heijden [/bib_ref] [bib_ref] Functional Outcomes of Cerebellar Malformations, Gill [/bib_ref] facilitates the postnatal refinement of motor control (44,45) during a period in which even healthy infants exhibit dystoniaassociated features [bib_ref] Recognizing the Common Origins of Dystonia and the Development of Human Movement:..., Lin [/bib_ref] [bib_ref] Physiological Movement Disorder-like Features during Typical Motor Development, Kuiper [/bib_ref] [bib_ref] The Burke-Fahn-Marsden Dystonia Rating Scale Is Age-dependent in Healthy Children, Kuiper [/bib_ref]. Indeed, we have previously found that manipulations during cerebellar development result in early-onset dystonia in mice [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref] [bib_ref] Maturation of Purkinje Cell Firing Properties Relies on Neurogenesis of Excitatory Neurons, Van Der Heijden [/bib_ref]. In one of these models, impairing excitatory synaptic transmission from brainstem inferior olive neurons onto cerebellar Purkinje neurons causes dystonia without persistent gross cerebellar malformations (Ptf1a Cre ;Vglut2 fl/fl mice [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref] [bib_ref] Cerebellar Dysfunction as a Source of Dystonic Phenotypes in Mice, Brown [/bib_ref] , similar to what is found in infants with the disease. As a result, this engineered mouse model offers an excellent platform to investigate how aberrant neural circuit function alters neurodevelopment and subsequently impacts behavior.
However, studying dystonia in the context of neurodevelopment is often challenging. Motor control is refined during postnatal development such that even normally developing infants display high scores on the same dystonia rating scales that are commonly used to examine adults [bib_ref] Physiological Movement Disorder-like Features during Typical Motor Development, Kuiper [/bib_ref] [bib_ref] The Burke-Fahn-Marsden Dystonia Rating Scale Is Age-dependent in Healthy Children, Kuiper [/bib_ref]. Similarly, dystonia rating scales used to quantify dystonic movements in mature mice provide higher scores in developing control mice since the latter's motor control is not yet refined, resulting in the reduced sensitivity of these rating scales. Another approach to quantify dystonia in animal models includes EMG recordings [bib_ref] Electromyographic Evidence in Support of a Knock-In Mouse Model of DYT1 Dystonia, Deandrade [/bib_ref] [bib_ref] Kinematic and Electromyographic Tools for Characterizing Movement Disorders in Mice, Scholle [/bib_ref]. Unfortunately, these invasive EMG recordings require surgeries and implants that are hard to perform in growing animals with small limbs. Additionally, albeit not specific to dystonia- associated impairments, global measures that are often used to assess motor control in adult mice, including the accelerating rotarod or the open field assay [bib_ref] The Neural Substrates of Rapid-Onset Dystonia-Parkinsonism, Calderon [/bib_ref] [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref] , cannot be easily performed with reliability and reproducibility in young postnatal developing pups because rodents at this age demonstrate less ambulatory activity and rudimentary motor skills overall.
Importantly, several behavioral assays specifically designed to assess neural function in young rodents do exist [bib_ref] Reflex-ontogeny and Behavioural Development of the Mouse, Fox [/bib_ref] [bib_ref] Behavioral and Functional Analysis of Mouse Phenotype: SHIRPA, a Proposed Protocol for..., Rogers [/bib_ref] : the negative geotaxis reflex, the righting reflex, and the detection of ultrasonic vocalizations (USVs) after pups are separated from the mother [bib_ref] A Battery of Motor Tests in a Neonatal Mouse Model of Cerebral..., Feather-Schussler [/bib_ref] [bib_ref] Ultrasonic Vocalizations (USV) in the Three Standard Laboratory Mouse Strains: Developmental Analysis, Wiaderkiewicz [/bib_ref]. Here, using these tests, we quantify the performance of a mouse model with earlyonset dystonia, the Ptf1a Cre ;Vglut2 fl/fl mutant [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref] [bib_ref] Cerebellar Dysfunction as a Source of Dystonic Phenotypes in Mice, Brown [/bib_ref]. We have chosen to focus on a key period of neurodevelopment when these motor reflexes emerge in normally developing mice, from postnatal day (P) 7-11. We found impaired acquisition of these motor behaviors in pups with dystonia, demonstrating that these paradigms may be good quantitative assays for measuring aberrant neurodevelopment in mouse models of pediatric dystonia. Furthermore, we found that the dystonic motor behaviors are accompanied by alterations in USVs. Crucially, the co-expression of motor and non-motor defects mirrors the multi-domain deficits seen in some pediatric neurodevelopmental syndromes.
# Methods
## Animals
Mice used in this study were housed in a Level 3, AALAS-certified vivarium. Experiments and studies were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC) of Baylor College of Medicine (BCM). Mice were ear-tagged on the first day of behavioral investigation (P7) and their genotypes determined using allelespecific PCR amplification after the conclusion of experiments so that experimenters were blinded to the genotypes of the mice. The day a copulation plug was detected was considered embryonic day (E) 0.5. We defined P0 as the date of birth. For the experiments in this study, we crossed male mice that were heterozygous for Ptf1a Cre (JAX #023329)and homozygous for the LoxP-flanked glutamatergic vesicular transporter 2 gene, Vglut2 fl (JAX #012898) [bib_ref] Synaptic Glutamate Release by Ventromedial Hypothalamic Neurons Is Part of the Neurocircuitry..., Tong [/bib_ref] , to female mice that were homozygous for Vglut2 fl . The resulting Ptf1a Cre ;Vglut2 fl/fl offspring had a conditional deletion of the Vglut2 allele in the Ptf1a lineage, which prevents the loading of glutamate into presynaptic vesicles during fast neurotransmission and therefore eliminates neurotransmission at chemical synapses of glutamatergic, Ptf1a lineage neurons (58). Most Ptf1a lineage neurons are inhibitory neurons and therefore are not affected by the deletion of Vglut2, although inferior olive neurons that send climbing fiber projections to cerebellar Purkinje cells in the molecular layer of the cerebellar cortex are excitatory and do express Vglut2. Preventing communication between climbing fibers and Purkinje cells results in severe, early-onset dystonia-associated impairments [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref] [bib_ref] Cerebellar Dysfunction as a Source of Dystonic Phenotypes in Mice, Brown [/bib_ref]. In our study, we used all the pups from 4 litters, which provided us with 21 Vglut fl/fl control mice (7 female, 14 male) and 10 Ptf1a Cre ;Vglut2 fl/fl dystonic mice (5 female, 5 male). When comparing mice from each sex and genotype with each other, we did not observe any sex differences. For this assessment, we performed a two-way ANOVA (genotype, sex) and did not find interactions between genotype and sex (p > 0.05 for all tests) at any of the developmental time-points for any of the behavioral tests. Therefore, we combined the data collected from male and female mice when performing the statistical analyses reported in this study.
## Negative geotaxis reflex
Mice were tested in the negative geotaxis assay at P7, P9 and P11 [bib_ref] A Battery of Motor Tests in a Neonatal Mouse Model of Cerebral..., Feather-Schussler [/bib_ref] [bib_ref] Negative Geotaxis: An Early Age Behavioral Hallmark to VPA Rat Model of..., Ruhela [/bib_ref]. A cage top wrapped with a sterile Poly-Lined drape was used to create a ramp with a 35° slope. Mice were placed on this ramp one at a time, oriented to face down the slope. Upon placement and release of the mouse, a 60-s timer was started. A successful trial was considered as one in which the mouse turned >90° on the ramp (crossing the plane perpendicular to the original placement in either direction). The time it took for the mouse to perform this movement was recorded. A failed trial was considered one in which mice were unable to change their orientation within 60-s or in which mice lost their footing on the ramp and fell. After a completed trial (either success or failure), each mouse was returned to their home cage. For each mouse, this process was repeated for a total of three trials per mouse, per behavioral timepoint.
## Surface righting reflex
The righting reflex was measured at P7, P9, and P11 [bib_ref] Maturation of Purkinje Cell Firing Properties Relies on Neurogenesis of Excitatory Neurons, Van Der Heijden [/bib_ref] [bib_ref] A Battery of Motor Tests in a Neonatal Mouse Model of Cerebral..., Feather-Schussler [/bib_ref]. In this assay, each mouse was placed on its back on a clean cage without bedding, and then this position was gently held by one finger until timing started. Upon removal of the finger, the time required for the mouse to right itself onto its four paws was recorded. All mice were tested three times at each age. A "failed" trial was defined as one in which the mouse did not right itself within 60 s.
## Ultrasonic vocalizations
Pup vocalizations were recorded at P7, P9, and P11 as described previously [bib_ref] Maturation of Purkinje Cell Firing Properties Relies on Neurogenesis of Excitatory Neurons, Van Der Heijden [/bib_ref] [bib_ref] Vocal Development in Dystonic Rats, Riede [/bib_ref]. Pups were placed in an anechoic, sound-attenuating chamber (Med Associates Inc.) within a round plastic tub that was positioned under a CM16 microphone (Avisoft Bioacoustics) in the center of the chamber. Sound was amplified and digitized using UltraSoundGate 416H at a 250 kHz sampling rate and a bit depth of 16 while Avisoft RECORDER software was used to collect the recordings. The USVs of each pup were monitored for 2 min.
## Statistical analyses
Analyses for this study were performed using MATLAB (Mathworks, United States). We plotted and then quantified statistically significant differences using a repeated-measures ANOVA and quantified the differences between genotypes and time-points using a Tukey Kramer post-hoc analysis. We used an alpha of 0.05 to accept statistical significance.
# Results
## Dystonic postures
We started by confirming the presence of dystonic postures in mice during the second postnatal week (P7-P11 mice show very few coordinated, ambulatory movements, and therefore the ability to accurately distinguish spontaneous dystonic movements is difficult. Nevertheless, when positioned on a flat surface, we observed that the Ptf1a Cre ;Vglut2 fl/fl pups often remained in fixed dystonic postures without initiating specific dedicated movements. To distinguish the postures seen in Ptf1a Cre ;Vglut2 fl/fl mutants compared to control mice, we acquired a series of images to help visualize the different body postures displayed from P7 to P11 . In addition, we also relied on video recordings to examine the full extent and dynamics of the dystonia-associated behaviors (Supplementary Video S1). In the Ptf1a Cre ;Vglut2 fl/fl mice, we frequently observed dystonic postures including: hyper-extension of the hindlimbs, twisted body posturing that resulted in the front-or hind-paws extending on either side of the body, strongly curved spines and lateral positioning, rigid or kinked tail positioning, and splayed digits on the fore-and hind-paws [bib_ref] Cerebellar Dysfunction as a Source of Dystonic Phenotypes in Mice, Brown [/bib_ref]. Combinations of these dystonic postures were evident in all Ptf1a Cre ;Vglut2 fl/fl mice (n = 10) included in our study . These data confirm that the Ptf1a Cre ;Vglut2 fl/fl mutant mice have early-onset dystonia-associated motor impairments that are robust and reproducible from animal to animal.
## Negative geotaxis reflex
Next, we investigated whether these dystonia-associated impairments prevented Ptf1a Cre ;Vglut2 fl/fl mice from properly executing the negative geotaxis reflex ;
Supplementary Video S2). This behavioral reflex naturally arises in mice around P7 [bib_ref] Coloboma Hyperactive Mutant Exhibits Delayed Neurobehavioral Developmental Milestones, Heyser [/bib_ref] and when mice mature, their time to complete the negative geotaxis reflex decreases due to their increasing motor control. We found that at the youngest ages tested (P7 and P9), control Vglut2 fl/fl mice and dystonic Ptf1a Cre ;Vglut2 fl/fl mice took equally long to complete geotaxis reflexes . However, while control Vglut2 fl/fl mice showed a rapid decrease in the time needed to successfully turn by P11, the time taken to turn remained prolonged in P11 dystonic Ptf1a Cre ;Vglut2 fl/fl mice . Similarly, we found that the number of failed trials (trials in which the pup lost grip of the padding and rolled down the slope or did not turn within 60 s) was similar between control and Ptf1a Cre ;Vglut2 fl/fl mice at P7 and P9, but was higher at P11 in Ptf1a Cre ;Vglut2 fl/fl mice. Together, these results indicate that, compared to controls, Ptf1a Cre ;Vglut2 fl/fl mice have delays in the development of the negative geotaxis reflex, which assays the early development of motor control.
## Surface righting reflex
We further investigated early motor control in dystonic Ptf1a Cre ;Vglut2 fl/fl mice by examining their performance of the surface righting reflex ; Supplementary Video S3). Mice usually acquire this reflex around P5 (62), although some variability in reflex onset and reflex time may be observed. We previously studied En1 Cre ;Atoh1 fl/− mice that have impaired neurogenesis of excitatory neurons, which lead to dystonic motor impairments [bib_ref] Maturation of Purkinje Cell Firing Properties Relies on Neurogenesis of Excitatory Neurons, Van Der Heijden [/bib_ref]. These mice also exhibit impairments in the surface righting reflex, which was likely due to abnormal motor control. We therefore tested whether the surface righting was also impaired in dystonic Ptf1a Cre ;Vglut2 fl/fl mice . We found that the Ptf1a Cre ;Vglut2 fl/fl mutant mice required more time to right themselves from a supine position to all four paws at P7, P9, and P11 relative to controls, although we did note that the time they took to right decreased with age. Furthermore, we observed that the number of trials in which the Ptf1a Cre ;Vglut2 fl/fl mice failed to right themselves within 60 s was higher than in control mice at P7 and P9, but not at P11 . Together, these results show that the Ptf1a Cre ;Vglut2 fl/fl mutant mice, which express striking dystonic behaviors including bouts of twisting of the limbs and torso, also have delayed development of the surface righting reflex, likely reflecting impairments in the development of normal motor control.
## Ultrasonic vocalizations
To test whether the circuit disruptions in Ptf1a Cre ;Vglut2 fl/fl mice lead to aberrant function in behavioral domains outside of motor control, we tested whether the Ptf1a Cre ;Vglut2 fl/fl mutant mice had abnormal vocalizations when socially isolated. In the first few postnatal weeks, pups make USVs when separated from the nest and their dam (55) . These vocalizations are thought to be a measure of early social behavior in mice and are often changed in number, duration, and/or frequency in models of neurodevelopmental disability and autism spectrum disorders [bib_ref] Behavioral Tests for Mouse Models of Autism: An Argument for the Inclusion..., Simmons [/bib_ref] [bib_ref] Ultrasonic Vocalizations: a Tool for Behavioural Phenotyping of Mouse Models of Neurodevelopmental..., Scattoni [/bib_ref] [bib_ref] Behavioral Phenotypes of Genetic Mouse Models of Autism, Kazdoba [/bib_ref]. Interestingly, pup vocalizations are also abnormal in mouse models with other cerebellar alterations [bib_ref] Maturation of Purkinje Cell Firing Properties Relies on Neurogenesis of Excitatory Neurons, Van Der Heijden [/bib_ref] [bib_ref] Autistic-like Behaviour and Cerebellar Dysfunction in Purkinje Cell Tsc1 Mutant Mice, Tsai [/bib_ref] [bib_ref] Splitting of the Cerebellar Vermis in Juvenile Rats-Eeffects on Social Behavior, Vocalization..., Al-Afif [/bib_ref] as well as in dystonic rats [bib_ref] Vocal Development in Dystonic Rats, Riede [/bib_ref]. We found that while call duration was not statistically different between control Vglut2 fl/fl mice and the dystonic Ptf1a Cre ;Vglut2 fl/fl mutant mice at P7, P9, or P11 , the number of calls was significantly lower in Ptf1a Cre ;Vglut2 fl/fl mutants compared to control Vglut2 fl/fl mice at P7, P9, and P11 . These results show that Ptf1a Cre ;Vglut2 fl/fl mice that express dystonic motor behaviors also produce fewer vocalizations of normal duration compared to littermate controls, reflecting possible deficits in socialization in addition to the observed deficits in motor control [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref].
# Discussion
In this study, we investigated the expression of early postnatal reflexive behaviors in the Ptf1a Cre ;Vglut2 fl/fl mouse model of early-onset dystonia [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref]. We found that these mice have delayed acquisition of the negative geotaxis and surface righting reflexes compared to their control littermates. We further observed that the dystonic Ptf1a Cre ;Vglut2 fl/fl mice make fewer USVs when separated from their dams, a behavioral impairment that is frequently seen in mouse models of neurodevelopmental disability and autism spectrum disorders. Together, we show that these simple assays can serve as a practical toolkit for investigating delays in the achievement of neurodevelopmental milestones using a circuitbased mouse model of early-onset dystonia.
The behavioral assays described in this study are particularly advantageous as they provide straightforward methods to study the impact of functional neural defects during a postnatal period that is often considered experimentally difficult to assess. The quantification of reflexive behaviors is unbiased, the assays are non-invasive, and the motor assays do not require special equipment. Furthermore, the impairments in these reflexive motor behaviors, as reported here, are not unique to rodent models of dystonia [bib_ref] Behavioral Effects of Neonatal Lesions on the Cerebellar System, Lalonde [/bib_ref]. For example, impairments in surface righting, but not the geotaxis reflex, have also been observed in ataxic and tremoring shaker rats [bib_ref] A Behavioral Study of the Development of Hereditary Cerebellar Ataxia in the..., Wolf [/bib_ref] and ataxic lurcher mice [bib_ref] Neurobehavioral Evaluation of Lurcher Mutant Mice during Ontogeny, Thullier [/bib_ref]. However, these reflexive behaviors are uniquely impaired in models with impaired development. For example, the neurodegeneration that occurs later in life in a mouse model for spinocerebellar ataxia 6 (SCA6) is not accompanied by impairments in either reflexive behavior [bib_ref] Transient Cerebellar Alterations during Development Prior to Obvious Motor Phenotype in a..., Jayabal [/bib_ref]. Due to their non-invasive nature, these postnatal reflex assays can provide quantitative and reliable measures of the relative motor impairments caused by dystonia or other developmental disorders. Such measures could be used to assess the onset of developmental motor disorders and evaluate improvements of motor behavior after providing treatment.
Regarding the behavior of Ptf1a Cre ;Vglut2 fl/fl mice, we confirmed that they exhibit dystonic motor behaviors at early postnatal ages , making this genetic mouse model a powerful tool to study neurological deficits and circuit dysfunction in pediatric-onset dystonia. Furthermore, a recent study has shown that homozygous loss of TSPOAP1 causes pediatric-onset dystonia through synaptic abnormalities in the cerebellum, underscoring that aberrant synaptic function in the cerebellum can also cause dystonia in humans [bib_ref] Bi-allelic Variants in TSPOAP1, Encoding the Active Zone Protein RIMBP1, Cause Autosomal..., Mencacci [/bib_ref]. Adult Ptf1a Cre ;Vglut2 fl/fl mice respond to cerebellar deep brain stimulation targeted to the cerebellar nuclei [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref] , making this mouse a seminal pre-clinical model to predict the efficacy of cerebellar deep brain stimulation in alleviating symptoms in patients with dystonia [bib_ref] Deep Cerebellar Stimulation for Tremor and Dystonia, Horisawa [/bib_ref] [bib_ref] Cerebellar Deep Brain Stimulation for Acquired Hemidystonia, Brown [/bib_ref]. Furthermore, gross cerebellar hemispheric dysfunction is a frequent finding in children born early preterm (75) of whom many suffer from dystonia [bib_ref] Patterns of Motor Disability in Very Preterm Children, Bracewell [/bib_ref]. We therefore propose that even though the specific cerebellar network perturbation in dystonic Ptf1a Cre ;Vglut2 fl/fl mice differs from those induced in other hereditary or acquired pediatric dystonias, the perturbations that impact the "dystonia network" and cause tractable dystonic motor behaviors that can be easily quantified ultimately may be relevant to and potentially predictive of the neural dysfunction(s) that are exhibited in many forms of dystonia.
Of special interest is our finding that dystonic Ptf1a Cre ;Vglut2 fl/fl mice exhibit social deficits in the social isolation/USV assay. The comorbidity of social deficits, dystonia, and delayed early behavioral reflexes supports the hypothesis of a possible shared ontogeny in the domains of social interaction and motor control [bib_ref] Functional Outcomes of Cerebellar Malformations, Gill [/bib_ref] [bib_ref] Abnormal Cerebellar Development in Autism Spectrum Disorders, Van Der Heijden [/bib_ref] [bib_ref] The Cerebellum, Sensitive Periods, and Autism, Wang [/bib_ref]. Intriguingly, in our Ptf1a Cre ;Vglut2 fl/fl mice, the resulting behaviors can be attributed to initial cerebellar dysfunction due to the precision of the genetic manipulation performed in these mice [bib_ref] Genetic Silencing of Olivocerebellar Synapses Causes Dystonia-like Behaviour in Mice, White [/bib_ref]. The observed USV changes further suggest that neurodevelopmental disorders exhibiting deficits across developmental domains, as seen in dystonia, may converge on a shared mechanism of cerebellar dysfunction. Previous work has highlighted the importance of evaluating motor control in mouse models of autism spectrum disorders and cerebellar dysfunction in patients with autism spectrum disorders to uncover the full range of neural dysfunction in these neurodevelopmental disorders [bib_ref] Behavioral Tests for Mouse Models of Autism: An Argument for the Inclusion..., Simmons [/bib_ref] [bib_ref] Cerebellar Dysfunction in Autism Spectrum Disorders: Deriving Mechanistic Insights from an Internal..., Kelly [/bib_ref] [bib_ref] Regulation of Autism-Relevant Behaviors by Cerebellar-Prefrontal Cortical Circuits, Kelly [/bib_ref]. Here, we propose the reverse; in pediatric dystonias and other early-onset movement disorders, social deficits should be taken into consideration and fully evaluated. In conclusion, we postulate that neurodevelopmental disorders that display a spectrum of social and motor deficits may converge at the level of cerebellar dysfunction [bib_ref] Interactions between Purkinje Cells and Granule Cells Coordinate the Development of Functional..., Van Der Heijden [/bib_ref] [bib_ref] Functional Outcomes of Cerebellar Malformations, Gill [/bib_ref] [bib_ref] Abnormal Cerebellar Development in Autism Spectrum Disorders, Van Der Heijden [/bib_ref] [bib_ref] Structure-function Relationships in the Developing Cerebellum: Evidence from Early-Life Cerebellar Injury and..., Stoodley [/bib_ref]. It would be interesting if future studies were to investigate whether the manifestation of motor and social impairments in dystonia is dependent on region-specific dysfunctions [bib_ref] Regulation of Autism-Relevant Behaviors by Cerebellar-Prefrontal Cortical Circuits, Kelly [/bib_ref] , how the resulting dysfunction is related to the underlying genetic mechanisms, and whether the ultimate behavioral abnormalities rely on the timing of genetic or physical insults to the developing cerebellum and its associated brain networks.
|
10.1186/s12911-018-0648-0
|
CCBY
|
6053711
|
30029644
|
s2orc_pubmed_articles
|
Patient facing decision support system for interpretation of laboratory test results
Background: In some healthcare systems, it is common that patients address laboratory test centers directly without a physician's recommendation. This practice is widely spread in Russia with about 28% of patients who visiting laboratory test centers for diagnostics. This causes an issue when patients get no help from the physician in understanding the results. Computer decision support systems proved to efficiently solve a resource consuming task of interpretation of the test results. So, a decision support system can be implemented to rise motivation and empower the patients who visit a laboratory service without a doctor's referral. Methods: We have developed a clinical decision support system for patients that solves a classification task and finds a set of diagnoses for the provided laboratory tests results. The Wilson and Lankton's assessment model was applied to measure patients' acceptance of the solution. Results: A first order predicates-based decision support system has been implemented to analyze laboratory test results and deliver reports in natural language to patients. The evaluation of the system showed a high acceptance of the decision support system and of the reports that it generates. Conclusions: Detailed notification of the laboratory service patients with elements of the decision support is significant for the laboratory data management, and for patients' empowerment and safety.
# Background
In some healthcare systems, it is common that patients address laboratory test centers directly without a physician's recommendation [bib_ref] Development of a clinical decision support system for the patients of a..., Semenov [/bib_ref]. This practice is widely spread in Russia with about 28% of patients who visiting laboratory test centers for diagnostics. This causes an issue when patients get no help from the physician in understanding the results. Patients face a problem when they need to decide how to continue the diagnostics and treatment process. A possible solution to this problem could be that a laboratory test center not only delivers the test results but also their explanation to the patients. This, however, should be done automatically, or at least semi-automatically, to exclude a critical load on the test centers. Clinical decision support systems can become a good technology for an automatic interpretation of test results [bib_ref] The role of standardized data and terminological systems in computerized clinical decision..., Ahmadian [/bib_ref] [bib_ref] Development and evaluation of a smartphone application for managing gestational diabetes mellitus, Jo [/bib_ref]. The experience in implementation of decision support systems for health care professionals shows their efficiency for medical diagnostics. However, patients require a different approach in data presentation and interpretation [bib_ref] Analysis of metrics for the usability evaluation of EHR management systems, Kopanitsa [/bib_ref] [bib_ref] Analysis of metrics for the usability evaluation of electronic health record systems, Kopanitsa [/bib_ref] [bib_ref] Usability of data integration and visualization software for multidisciplinary pediatric intensive care:..., Lin [/bib_ref] [bib_ref] How to improve vital sign data quality for use in clinical decision..., Skyttberg [/bib_ref] [bib_ref] Temporal data representation, normalization, extraction, and reasoning: a review from clinical domain, Madkour [/bib_ref].
Studies [bib_ref] Primary care physician attitudes concerning follow-up of abnormal test results and ambulatory..., Murff [/bib_ref] [bib_ref] I wish I had seen this test result earlier!": dissatisfaction with test..., Poon [/bib_ref] [bib_ref] Assessing the delivery of patient critical laboratory results to primary care providers, Montes [/bib_ref] [bib_ref] Adaption, implementation and evaluation of collaborative service improvements in the testing and..., Litchfield [/bib_ref] have demonstrated that many providers do not have systems that can ensure that the test results are reliably communicated to patients. As shown in [bib_ref] Frequency of failure to inform patients of clinically significant outpatient test results, Casalino [/bib_ref] [bib_ref] Direct reporting of laboratory test results to patients by mail to enhance..., Sung [/bib_ref] normal and abnormal test results are commonly missed, even when a health care system widely uses electronic health records (EHRs), and providers miss 1-10% of abnormal test results. It would not be an exaggeration to say that we do not have sufficient mechanisms to ensure that test results are consistently delivered to patients and understood by them.
As a problem importance is recognized, a number of potential solutions has been studied [bib_ref] Impact of an automated test results management system on patients' satisfaction about..., Matheny [/bib_ref] [bib_ref] Effectiveness of an electronic health record-based intervention to improve follow-up of abnormal..., Laxmisan [/bib_ref] [bib_ref] The PAADRN study: a design for a randomized controlled practical clinical trial..., Edmonds [/bib_ref] [bib_ref] DB4US: a decision support system for laboratory information management, Carmona-Cejudo [/bib_ref] [bib_ref] The ideal laboratory information system DB4US: A decision support system for laboratory..., Sepulveda [/bib_ref]. The first approach originates from the development of computerized decision support systems that support test centers in reviewing results and notifying patients in case of abnormal results [bib_ref] Impact of an automated test results management system on patients' satisfaction about..., Matheny [/bib_ref] [bib_ref] Effectiveness of an electronic health record-based intervention to improve follow-up of abnormal..., Laxmisan [/bib_ref] [bib_ref] The ideal laboratory information system DB4US: A decision support system for laboratory..., Sepulveda [/bib_ref]. Another approach has involved implementation of such testing processes where test centers consistently deliver test results directly to patients. Such systems vary from sending each piece of the test results by mail to complex patient web portals, where they can have access to the history of test results [bib_ref] The PAADRN study: a design for a randomized controlled practical clinical trial..., Edmonds [/bib_ref] [bib_ref] Sharing electronic laboratory results in a patient portal-a feasibility pilot, Wald [/bib_ref].
Interpretation of the test results is a resource consuming task that delays the results and increases costs of each test [bib_ref] Proposed classification of various limit values (guide values) used in assisting the..., Haeckel [/bib_ref] [bib_ref] Problems in interpretation of clinical laboratory test results, Romatowski [/bib_ref]. However, the computer decision support systems proved to solve such tasks efficiently. To increase motivation and support the patients who refer to a test center without a doctor's referral in making better informed decisions, a computer decision support system can be designed and implemented.
The goal of this study is to develop and evaluate a decision support system for patients, which: a) provides a personalized tool to inform patients on the results of the laboratory tests b) empowers patients to form opinions on how to continue or not to continue with a treatment c) prepares patients to have an informed discussion with their doctor.
To support patients, we have implemented and evaluated a decision support system that automatically generates interpretations for laboratory test results:
This paper focuses on the evaluation of correctness and user acceptance of a decision support system for patients of a test center in Saint-Petersburg, Russia.
# Methods
## Implementation
We have developed a clinical decision support system for the patients that solves a classification problem by connecting test results to a list of diagnoses. The decision support is based on a classification algorithm, which produces the following conclusions:
Located a list of diagnoses that can be related to the test results; Found no fitting diagnoses;
To enable a definition of inference rules we have developed a knowledge representation language that is based on the predicate calculus [bib_ref] Pattern recognition as rule-guided inductive inference, Michalski [/bib_ref] and a user interface to allow medical professionals defining the system rules. For the pilot project, we have chosen a limited set of laboratory tests that could be automatically interpreted by the system. We have interviewed 3 laboratory physicians and 3 specialist physicians (gynecologist, urologist and general practitioner) to define the inference rules for the system. The decision support system has been implemented and operating in the Helix laboratory center in Saint-Petersburg, Russia.
The system has been implemented using the following technologies:
1. User interfaces and a back-end are based on the.
NET Core 2.0 2. Data storage is based on PostgreSQL
## Evaluation
## Accuracy of the decision support
To evaluate accuracy of the results produced by the system, we have performed a validation of 1000 randomly generated reports. The reports were generated in a way to allow validating all of 89 decision support algorithms. The reports were given to two independent pathology experts to be reviewed independently. The results of the expert review were used to calculate the following criteria [bib_ref] Sample size in receiver-operating characteristic (ROC) curve analysis, Kawada [/bib_ref] :
1. Error rate as an average classification error 2. Accuracy as an average effectiveness of a classifier 3. Precision ((All terms -Mistakes)/All terms), 4. Recall (ratio of true positives to (true positives + false negatives)), and 5. F-measure (2- recall- precision recallþprecision ).
The reviewers' disagreements were settled by consensus. Cohen's kappa was calculated to rate the disagreement between reviewers [bib_ref] Weighted kappa for multiple raters, Berry [/bib_ref].
## User acceptance
To assess the user acceptance of the system, a Wilson and Lankton's model of patients' acceptance of electronic health solutions was applied [bib_ref] Modeling patients' acceptance of provider-delivered e-health, Wilson [/bib_ref]. The model allowed measuring the following criteria: behavioral intention (BI) to use, intrinsic motivation (IM), perceived ease-of-use (PEOU), and perceived usefulness (PU) of the decision support system. BI represents the intention to utilize the system and to rely on the decision support that it provides; IM represents the willingness to use the system provided that no direct compensation is available; PEOU represents the extent to which the provided reports are clearly presented and comprehended by users; and PU denotes the degree to which the patients believe that the utilization of the decision support system will improve their experience with laboratory tests.
We have applied a Wilson's and Lankton [bib_ref] Modeling patients' acceptance of provider-delivered e-health, Wilson [/bib_ref] revision of the Davis's et al. [bib_ref] User acceptance of computer technology: a comparison of two theoretical models, Davis [/bib_ref] method to measure BI, PEOU, and PU. Intrinsic motivation was measured by utilizing the Davis's et al. method [bib_ref] User acceptance of computer technology: a comparison of two theoretical models, Davis [/bib_ref].
## Questionnaire
We started with detection of possible items for the questionnaire by collecting a large list of acceptance test questions. The questions were collected from preceding internal studies, from the literature and from brainstorming. The list was then reviewed by the study team to eliminate the items that do not help to reach the goals of the study and duplicate questions. The remaining items were simplified and worded as clear to the potential participants as possible.
BI measure consisted of 2 objects whereas IM, PEOU, and PU consisted of 3 objects each. Russian translation of the questionnaires made by the research team was used during the study. To rate each item a Likert scale from 1 (not at all) to 7 (very much) was applied [bib_ref] A comparison of the direction-of-perception technique with the Likert method in the..., Drinkwater [/bib_ref].
## Recruitment
The recruitment of the study participants was done in Saint-Petersburg, Russia. The patients were eligible to be invited if they had experience using the system with a minimum of 5 reports on the test outcomes. The recruitment was done by sending invitations to the 500 eligible patients. Later, we have formed a group of 120 patients based on the first respondedfirst included principal with a recruitment rate of 24%.
Demographic characteristics of the study participants are shown in [fig_ref] Table 1: Demographic characteristics of the study participants [/fig_ref]. We assessed Information technology (IT) literacy of the patients based on how frequently they use smartphones or personal computer. We assessed IT literacy on the scale from beginnerspatients who started using computer or smartphone maximum 6 months before the study begin; intermediatecomputer or smartphone users who do it at least 2 times a week; and advanceddaily users of a computer or a smartphone.
## Data collection and analysis
All the study participants were given individual access to the online questionnaire, which they were asked to fill in (please see Additional file 1 for the questionnaire details). All the patients received a written detailed instruction on how to operate with a questionnaire and the sense of the rating scale.
GNU Octaveversion 4.0.2 was applied to calculate the statistics of the participants' general characteristics and user acceptance measurements.
## Ethics commission approval
The study was approved by the ethics commission of the committee of healthcare of Saint-Petersburg, Russia. All the study participants were informed in written form about the goals of the study and about the meaning of the questionnaires. We assured in written form every participant of their rights to anonymity and confidentiality. Written consent was obtained from every participant. Every participant was informed in written form about a right to withdraw personal data from the study record for up to 3 months after their approval.
# Results
## Implementation
The clinical decision support system consists of the following modules, which provide the main features of the system [fig_ref] Figure 1: Structural scheme of the decision support system [/fig_ref]
## ):
A Data extraction system receives data from external sources, such as hospital or laboratory information systems. It checks the syntax validity of data and sends it to a Database A Data base receives and saves facts from an external laboratory information system. A Knowledge base editor (see section "User acceptance".1) provides an interface to experts to define inference rules that are sent to a Knowledge base. A Knowledge base stores inference rules. Inference engine applies rules from the knowledge base to the facts from the Database to conclude the results and sends them to the explanation system and report generator. Explanation system scrutinizes the sequence of the applied inference rules to demonstrate how the result has been achieved. Report generator creates a readable report from the inference results and sends it to the report storage.
The example of how the system operates is presented in section "Inference example". The knowledge representation language of the system is built upon a first order predicate logic. The scheme of the knowledge base is shown in [fig_ref] Figure 2: Object [/fig_ref].
The main object that the system processes is a laboratory test configuration that consists a laboratory test object and of a list of direct inference rules that can be related to this object.
Laboratory test object is a model that comprises a list of atomic components of the test e.g. a complete blood count test includes 22 atomic components. For each component of a test, we define a list of direct inference rules that have conditions for including this components in the inference. The conditions are represented as comparison operators: =, <>, includes (> = or = <), excludes (> = and = <). Within a rule, the conditions are related by logical operators "and", "or" and "not". For each direct rule, an expert can model a list of exclusion rules to exclude a direct rule from an inference process if the exclusion conditions are met.
An order the object groups laboratory tests reflect the commercial orders that patients actually make.
General inference process is divided into the following steps:
1. When the system receives an order bundle from a laboratory information system, the tests in the order are being analyzed to generate a list of tests, configurations for which are available in the knowledge base. 2. Actual test results are loaded to the Database of the system and become available for an inference process. 3. The inference engine receives the list of tests and selects proper direct inference rules and exclusion rules in the proper sequence, which can be applied to the received facts.
4. If a direct rule has been successfully applied and no exclusion rule is effective, the inference engine adds a text artefact to the resulting json file . 5. After the inference has been completed, the resulting json file is sent to the reporting service to generate a pdf report.
The decision support system has been implemented in the Helix laboratory service in Saint-Petersburg, Russia. The system is in commercial production now generating about 20,000 reports a day.
The system has inference algorithms for the following International Statistical Classification of Diseases and Related Health Problems (ICD) 10 groups:
## Inference example
The full json code of rules and artefacts for the blood sugar testis presented in Additional file 2. The input of the inference is a bundle of resources that has been extracted from a laboratory tests database and were added to the decision support system database [fig_ref] Figure 1: Structural scheme of the decision support system [/fig_ref] : [fig_ref] Figure 4: Inference to the patient [/fig_ref] for the graphical representation of the sequence) based on the available rules from the knowledge base [fig_ref] Figure 1: Structural scheme of the decision support system [/fig_ref].
The inference ends up with the conclusion id = 4785 with the artefact id = 4786. The found artefact is added to the generated report, which is then sent to the report generator to create a human readable pdf file. The resulting rules sequence is being visualized by the explanation system [fig_ref] Figure 1: Structural scheme of the decision support system [/fig_ref].
## User interaction
The decision support system consists of 2 main interfaces: for experts to model knowledge and inference rules and for patients to have access to the test results and their interpretation.
## Expert's interface
We have developed a web knowledge management application that provides the following features:
Create and edit inference rules Group inference rules Create and edit artefacts with doctor's recommendations that form a decision support report as a result of a logical inference.
## Patient's user interface
Patient has access to the test results through a web portal, where a list of available tests is provided. For each test, a patient can have an overview of the results [fig_ref] Figure 7: Personal space for a patient on the online portal [/fig_ref]. The results are presented in the table form with the following columns: Parameter name, My Results and a Reference interval. A patient can click on the "Generate report" button (second button from the left with a doctor icon on the [fig_ref] Figure 7: Personal space for a patient on the online portal [/fig_ref] to open a decision support report [fig_ref] Figure 8: Report [/fig_ref].
## Evaluation correctness
A sample of 1000 reports was independently assessed by two independent pathology experts. The results of assessment for each criterion are shown in [fig_ref] Table 2: Reports' quality evaluation [/fig_ref].
The experts revealed disagreement in the assessment of 2 reports.
## Acceptance
The mean values for BI, IM, PEOU, and PU (5.9, 6.2, 5.7, and 5.9 respectively) showed a high acceptance of the decision support system and the reports that it generates [fig_ref] Table 3: Acceptance criteria Mean Median Max Min Mean Median Max Min Mean Median... [/fig_ref].
# Discussion
The paper describes a development of a patient facing clinical decision support system, which provides interpretation of the test results in the natural language.
## Notification ethics
We need to be very cautious when providing test results to the patients by e-mail or on a web portal. We should assume that the patients may not fully and properly comprehend the interpretation of the results. So, the capability to deliver results and their interpretation in a manner they are understood by a patient is essential for a motivation to refer to a health care professional, particularly when test results are abnormal. Our decision support system only interprets and sends test results that do not need a humane communication according to the standards of the laboratory service. Test results that The decision support system never intended to be prescriptive and communicate a single possible clinical decision to a patient. To follow this approach, we have implemented the reports in a way that they are descriptive and informative rather than prescriptive.
## Correctness
Rules definition process shall be controlled and always reviewed. The evaluation showed that the correctness of the generated reports is high. Seven mistakes out of 1000 analyzed reports were caused by a human factor. The mistakes that were detected by the experts during the assessment were caused by the inaccuracies of the experts when modelling inference rules. This led to a change of rules' definition procedure, where we apply 4 eyes principle [bib_ref] Applying the four-eyes principle to management decisions in the manufacturing sector: are..., Hm Rw [/bib_ref]. Now, when each rule goes to production only after a review and acceptance of a second expert.
## Use acceptance
One of the measures of feasibility was the percent of patients who agreed to take part in the testing of the decision support tool. The rate of 86% of patients who agreed to take part in the study shows high interest and motivation, which is supported by quantitative measurements that were done within the study.
The user acceptance of the system was evaluated after 2 months of operation. Acceptance scores were high, all of them above 5.7 out of 7. Among elder users (60+) the results were a little worse in comparison to the younger users. The maximum difference was 0.7 (10%) for the PU. Elder people felt less motivated about storing their medical data in electronic format (5.0 versus 5.7 for the younger participants). Maximum rates were similar for each statement and age group, and a high median value also indicates a positive attitude to the system. Minimum rates of 4 show an encouragement towards the system, as all the rates were in the positive part of the scale. This is true for all age groups. This indicates high acceptance of the solution and the way the notifications are being delivered. Partial correlations calculated using scales derived for these dimensions suggested that Ease of Use and Usefulness impact one another in a way that enhancements in Ease of Use increase the scores of Usefulness and the other way round. Whereas both Ease of Use and Usefulness steer Satisfaction with Usefulness having comparatively less significant influence. Users are more flexible in their Usefulness scores when they have only reduced experience with a system.
Unfortunately, we could not compare them to the similar studies, as we did not find a patient-oriented decision support system, for which a user acceptance was evaluated. However, we tried to compare the results with similar systems that were not patient oriented.
Acceptance scores were relatively high compared to the results of evaluation results of previous studies [bib_ref] An investigation of the effect of nurses' technology readiness on the acceptance..., Kuo [/bib_ref] [bib_ref] Applying the technology acceptance model to explore public health nurses' intentions towards..., Chen [/bib_ref] [bib_ref] Development and evaluation of theory-based diabetes support services, Guo [/bib_ref] [bib_ref] Development and evaluation of a theory-based physical activity guidebook for breast cancer..., Vallance [/bib_ref] management smartphone app. This can be explained by the fact that most of the study participants had average or above average IT skills. This is in a contrast with the previous studies on the patients' acceptance of decision support tools and can be explained by the increased computer literacy and changing IT habits. Our results mean that the health care providers and EHR developers can move in the direction of electronic notifications of the patients. This will facilitate communication and decrease its costs.
## Implications
The results of our study support other literature suggesting that patients want timely and detailed information and they want to be notified of all laboratory test results, even if they are normal [bib_ref] Communicating the result of breast biopsy by telephone or in person, Campbell [/bib_ref] [bib_ref] Patient preferences for notification of normal laboratory test results: a report from..., Baldwin [/bib_ref]. However, our results contradict the previous ones in regards to the patients preferring phone calls and sealed letters to the web based notification methods [bib_ref] Patient preferences for notification of normal laboratory test results: a report from..., Baldwin [/bib_ref].
The findings of this research have valuable inferences for the design and implementation of patient notification systems. We found that patients in general find the detailed notifications useful, are motivated to use them and don't face significant difficulties to adopt such solutions. It is very important when designing and implementing patients' notification systems to make them valid, easy and simple to use. To achieve this, we advise that a pilot application of the decision support system is tested by the experts for verification and validation of the rules and potential users for the user acceptance, so that corrections can be made during the implementation phase to increase the system's reliability and acceptance. The results of our study suggest that there is a major impact of patients' habits on the test results notification Translation from Russian: Doctor's comments on the results of the laboratory test Clinical Blood Test: blood count, white blood cells, Erythrocyte Sedimentation Rate
## Anemia and a state of erythropoiesis.
During the general blood test, the following parameters were measured: the number of erythrocytes, the level of hemoglobin, erythrocyte indices (the size of the form of erythrocytes and the content of hemoglobin in them). You have no signs of disturbance of erythropoiesis (the process of producing erythrocytes -cells that provide organs and tissues with the necessary amount of oxygen), including the absence of laboratory signs of anemia.
## The state of thrombocytic hemostasis
During the general blood analysis, the following parameters were measured: platelets, platelet indices. On the results of the analysis, you showed signs of a disorder of platelet hemostasis -thrombocytopenia of mild degree (decrease in the level of platelets in the blood). This can be caused by various pathological conditions (bleeding) and diseases including autoimmune thrombocytopenia, many viral infections, in some cases of cirrhosis of the liver. To clarify the reasons for the laboratory signs of thrombocytopenia revealed by the test, you are advised a consultation of a therapist or hematologist. utilization. In addition to the unswerving and natural effect of habit on IT use, a habit also functions as a stowed purpose trail to affect behavior. Promotion of electronic test results notifications still demands major communication effort to strengthen both the stowed intention and its relation to behavior.
## Legal
It is important to mention that in Russia laboratory ervices are legally obliged to provide results of laboratory tests to patients. Providing not only the results but the explanations of their meaning will enhance the notifications and make them more valuable for patients.
## Limitations of the study and future work tool's impact on patients' decision-making
We did not thoroughly gather data on participants' decision to follow up or not laboratory tests especially for the tests with abnormal results. This will become a major part of our next study where we will investigate how the patients decide to follow up or not laboratory tests. A systematic review by Callen et al. found that, across 19 published studies, 6.8-62% of lab tests were not followed up on [bib_ref] Failure to follow-up test results for ambulatory patients: a systematic review, Callen [/bib_ref]. We think that this rate will increase for the patients that receive detailed information about the test results and their possible implications.
Evolving the decision support system
We are evolving the decision support system every day by adding new inference rules and optimizing its architecture. The next steps would be to add a possibility of working with fuzzy rules [bib_ref] Knowledge acquisition in the fuzzy knowledge representation framework of a medical consultation..., Boegl [/bib_ref] to make the inference more flexible. Also, we are redesigning a data storage architecture to move from relational data base to a graph data base, that in our mind is more suitable for modeling knowledge and inference rules.
Mobile application for the patients is also under development now. This can potentially involve younger users to the system.
# Conclusions
The findings of the research provide us with a better understanding of how patients experience detailed notification of laboratory tests without health care professional participating in the process. Detailed notification of laboratory service patients with the elements of decision support is significant for laboratory data The system helps me to make more informed decisions 5.9 5.5 7 4 5.7 5.5 6 4 6.0 6 7 5
The system is reliable and I trust it 6.
[fig] Figure 1: Structural scheme of the decision support system [/fig]
[fig] Figure 2: Object [/fig]
[fig] Figure 5: shows an inference rule creation screen. A rule consists of several conditions connected by logical operators and a resulting artefact, which represents a text [/fig]
[fig] Figure 4: Inference to the patient. The artefacts can be created using an interface fromFig. 6. [/fig]
[fig] Figure 6: Expert's interface for recommendations [/fig]
[fig] Figure 7: Personal space for a patient on the online portal [/fig]
[fig] Figure 8: Report [/fig]
[table] Table 1: Demographic characteristics of the study participants [/table]
[table] Table 3: Acceptance criteria Mean Median Max Min Mean Median Max Min Mean Median Max Min [/table]
[table] Table 2: Reports' quality evaluation [/table]
|
10.3390/medicines6010037
|
CCBY
|
6473744
|
30870993
|
s2orc_pubmed_articles
|
Antiproliferative Activity of (-)-Rabdosiin Isolated from Ocimum sanctum L.
Background: Ocimum sanctum L. (holy basil; Tulsi in Hindi) is an important medicinal plant, traditionally used in India. Methods: The phytochemical study of the nonpolar (dichloromethane 100%) and polar (methanol:water; 7:3) extracts yielded fourteen compounds. Compounds 6, 7, 9, 11, 12, and 13, along with the methanol:water extract were evaluated for their cytotoxicity against the human cancer cell lines MCF-7, SKBR3, and HCT-116, and normal peripheral blood mononuclear cells (PBMCs). Results: Five terpenoids, namely, ursolic acid (1), oleanolic acid (2), betulinic acid(3), stigmasterol (4), and β-caryophyllene oxide (5); two lignans, i.e., (-)-rabdosiin (6) and shimobashiric acid C (7); three flavonoids, luteolin (8), its 7-O-β-D-glucuronide (9), apigenin 7-O-β-D-glucuronide (10); and four phenolics, (E)-p-coumaroyl 4-O-β-D-glucoside (11), 3-(3,4-dihydroxyphenyl) lactic acid (12), protocatechuic acid (13), and vanillic acid (14) were isolated. Compound 6 was the most cytotoxic against the human cancer lines assessed and showed very low cytotoxicity against PBMCs. Conclusions: Based on these results, the structure of compound 6 shows some promise as a selective anticancer drug scaffold.
# Introduction
Indigenous to India and parts of North and Eastern Africa, China, Hainan Island, and Taiwan, Tulsi (Ocimum sanctum L.; syn. Ocimum tenuiflorum L.) is referred to as "the elixir of life" or "the queen of herbs" and is believed to promote longevity [bib_ref] Tulsi-Ocimum sanctum: A herb for all reasons, Cohen [/bib_ref]. Various parts of the plant are used in Ayurveda and Siddha traditional medicine to treat coughs, bronchitis, fever, bile disturbances, and has been also used as an anthelminthic, antiemetic, anticancer, antiseptic, antioxidant, antidiabetic anti-inflammatory, antiulcer, hepatoprotective, cardioprotective, anticoagulant, anticataract, and analgesic agent. Additionally, it has been reported that extracts of the plant can serve as vitalizers and rejuvenators, and are thought to increase life-expectancy and promote disease-free living [bib_ref] Validation of traditional claim of Tulsi, Ocimum sanctum Linn. as a medicinal..., Gupta [/bib_ref] [bib_ref] The Indian holy power plant, Das [/bib_ref] [bib_ref] A reservoir plant for therapeutic applications: An overview, Pattanayak [/bib_ref] [bib_ref] Antinociceptive action of Ocimum sanctum (Tulsi) in mice: Possible mechanisms involved, Khanna [/bib_ref] [bib_ref] In vivo studies on the effect of Ocimum sanctum L. leaf extract..., Babu [/bib_ref] [bib_ref] Biochemical evaluation of antidiabetogenic properties of some commonly used Indian plants on..., Narendhirakannan [/bib_ref] [bib_ref] Ocimum sanctum leaf extracts stimulate insulin secretion from perfusd pancreas, isolated islets..., Hannan [/bib_ref] [bib_ref] Pterocarpus marsupium extract (Vijayasar) prevented the alteration in metabolic patterns induced in..., Grovel [/bib_ref] [bib_ref] Leishmanicidal active constituents from Nepalese medicinal plant Tulsi (Ocimum sanctum L.), Suzuki [/bib_ref].
Despite its wide therapeutic range, special care should be taken in case of the use of Tulsi in conjunction with other prescribed medicines since it exhibits various drug interactions. For example, its concomitant use with anticoagulants, such as heparin, warfarin, aspirin, clopidogrel, etc., is contraindicated due to allergic reactions that may occur. In addition, Tulsi increases the activity of phenobarbital and consequently may stimulate uterine contractions; thus, its use during pregnancy and lactation is not recommended [bib_ref] Polyphenols of Ocimum sanctum L. from Suriname, Skaltsa [/bib_ref].
# Materials and methods
# Plant material
Aerial parts of O. sanctum L. were collected in flowering stage at Suriname, as previously described [bib_ref] Analyse de l'huile essentielle d'Ocimum sanctum L, Skaltsa [/bib_ref]. A voucher specimen (ATHS 093) has been deposited in the Herbarium of the Laboratory of Pharmacognosy, National and Kapodistrian University of Athens.
## General experimental procedures
1 H, [bib_ref] Biochemical evaluation of antidiabetogenic properties of some commonly used Indian plants on..., Narendhirakannan [/bib_ref] C, and 2D NMR spectra were recorded in CDCl [bib_ref] Phytochemical Study of the leaves of Ocimum sanctum L, Skaltsa [/bib_ref] D values were obtained in CHCl 3 or MeOH on a Perkin-Elmer 341 Polarimeter. FT-IR spectra were recorded on a Perkin Elmer PARAGON 500 spectrophotometer. UV spectra were recorded on a Shimadzu UV-160 A spectrophotometer according to . GC-MS analyses were performed on a Hewlett-Packard 5973-6890 system operating in EI mode (70 eV) equipped with a split/splitless injector (220 - C), a split ratio 1/10, using a fused silica HP-5 MS capillary column (30 m x 0.25 mm (i.d.), film thickness: 0.25 µm) with a temperature program for HP-5 MS column from 60 - C (5 min) to 280 - C at a rate of 4 - C/min and helium as a carrier gas at a flow rate of 1.0 mL/min. Vacuum liquid chromatography (VLC): silica gel 60H (Merck, Art. 7736) [bib_ref] The application of Vacuum Liquid Chromatography to the separation of terpene mixtures, Coll [/bib_ref].
## Extraction and isolation
The initial extraction was previously described [bib_ref] Analyse de l'huile essentielle d'Ocimum sanctum L, Skaltsa [/bib_ref]. In brief, the aerial parts of O. sanctum L. (0.40 kg) were air-dried and finely ground, and then extracted at room temperature using dichloromethane and methanol, successively.
Part of the dichloromethane residue (11.9 g) was re-extracted at room temperature with ethyl acetate (EtOAc) and n-BuOH, yielding two fractions (A and B). Fraction A (7.8 g) was fractionated by VLC on silica gel using mixtures of cyclohexane and EtOAc of increasing polarity ( 69.4 mg) was subjected to CC on silica gel as previously described to give 75 fractions; fraction 8 (1.3 mg) was identified as compound 10. Another part of the methanol extract (7.7 g) was redissolved in water and extracted at room temperature with EtOAc and n-BuOH, affording three fractions (MA-MC). MB (eluted with n-BuOH; 5.3 g) was subjected to RP 18 -MPLC using a H 2 O:MeOH gradient system (100% H 2 O→100% MeOH; steps of 10% MeOH) and yielded 11 fractions (MB 1 -MB 10 ). Fraction MB 3 (eluted with H 2 O:MeOH 80:20) was identified as compound 8 (13.6 mg).
It is notable that during the fractionation and isolation procedures, all extracts and subfractions were continuously monitored by analytical TLC and 1 H-NMR. All obtained fractions were concentrated to dryness under vacuum (30 - C) and placed in activated desiccators with P 2 O 5 until their weights were stabilized.
## Cytotoxic effects against cancer cell lines
The cytotoxic activity of the compounds, as well as of the initial methanol extract, were tested against three human cancer cell lines: MCF-7 (breast; estrogen receptor positive (ER+), progesterone receptor (PR)+, and HER2 negative (-)), SKBR3 (breast; ER-, PR-, and HER2+), and HCT-116 (colon). All cell lines were maintained in RPMI-1640, supplemented with 10% heat-inactivated fetal bovine serum (FBS), 10 mM Hepes, 10 U/mL penicillin, 10 U/mL streptomycin, and 5 mg/mL gentamycin (all from Lonza, Cologne, Germany) (thereafter referred to as complete medium) at 37 - C in a humidified 5% CO 2 incubator.
Compounds were prepared at a stock solution of 10.0 mg/mL in DMSO and the extract at 20.0 mg/mL in DMSO. Prior to their use, they were diluted in plain RPMI-1640. Cytotoxicity was evaluated by the MTT reduction assay [bib_ref] Rapid colorimetric assay for cellular growth and survival: Application to proliferation and..., Mosman [/bib_ref] , which determines the effect of treatment with an exogenously added agent on the viability of the cell population. Briefly, cells were plated in 96-well plates (Greiner Bio-One GmbH, Frickenhausen, Germany; 5 × 10 3 cells/well) and incubated at 5% CO 2 and 95% air at 37 - C for 24 h, in order to adhere. Further, cells were incubated with the compounds for 72 h at 37 - C in a 5% CO 2 incubator. The MTT reagent (Sigma-Aldrich, Darmstadt, Germany; 1 mg/mL in phosphate buffered saline (PBS); 100 µL/well) was added during the last 4 h of incubation. The formazan crystals formed were dissolved by adding 0.1 M HCl in 2-propanol (100 µL/well) and absorption was measured using an ELISA reader (Denley WeScan, Finland) at 545 nm with reference filter set at 690 nm. All cultures were set in triplicate, whereas cells incubated in complete medium or in medium containing the equivalent amount of DMSO, as well as cells incubated in the presence of doxorubicin (Sigma-Aldrich) were used as negative and positive controls, respectively. The half maximal inhibitory concentration (IC 50 ) was calculated according to the formula: 100(A 0 − A)/A 0 = 50, where A and A 0 are optical densities of wells exposed to the compounds and control wells, respectively.
The compounds were tested at a concentration range of 200.0 to 6.25 µg/mL and the extract at 750.0 to 1.25 µg/mL. Doxorubicin was used as a standard cytotoxic agent and showed IC 50 values ≤ 0.2 µM in all cell lines tested. All experiments were performed at least three times.
# Flow cytometry analysis
MCF-7, SKBR3 and HCT-116 cells were incubated with compound 6 and analyzed with flow cytometry following staining with annexin V and propidium iodide (PI). Cells were plated into 24-well plates (Greiner Bio-One; 3 × 10 5 /mL; 2 mL/well), let adhere overnight, and incubated with the mean IC 50 value (80 µg/mL) and 40 µg/mL of compound 6 for 72 h. Cells were detached with 2 mM EDTA in Dulbecco's PBS (DPBS), harvested, centrifuged in cold PBS (1500 rpm; 5 min), and stained with the Annexin V-FITC Apoptosis Detection Kit (BioLegend, Fell, Germany; cat# 640914), according to the manufacturers' instructions. In brief, cells were resuspended in binding buffer, then annexin V-FITC (5 µL) and PI (10 µL; 0.03 µg/sample) were added, mixed, and incubated with the cells for 15 min in the dark at room temperature. The volume was adjusted to 500 µL with binding buffer and the cell suspension was immediately analyzed in a FACSCanto II (BD Biosciences, San Diego, CA, USA) using FACSDiva software (V7, BD Biosciences).
## Cytotoxic effect against human peripheral blood mononuclear cells
Compound 6 was additionally assessed for its cytotoxicity against human peripheral blood mononuclear cells (PBMCs) isolated from healthy blood donors' peripheral blood as previously described [bib_ref] Prothymosin α and a prothymosin α-derived peptide enhance TH1-type immune responses against..., Ioannou [/bib_ref]. Prior to blood draw, individuals gave their informed consent according to the regulations approved by the 2nd Peripheral Blood Transfusion Unit and Hemophiliac Centre, "Laikon" General Hospital Institutional Review Board, Athens, Greece. PBMCs were seeded in 24-well plates (5 × 10 5 /mL; 2 mL/well) and exposed to 2 concentrations of compound 6: 80 µg/mL and 40 µg/mL. PBMCs were collected, stained as described in 2.5 and analyzed by flow cytometry.
# Results and discussion
## Secondary metabolites isolated from o. sanctum
The phytochemical study of both nonpolar and polar extracts from O. sanctum aerial parts led to the isolation of 14 compounds identified on the basis of their spectra. More specifically, five terpenoids, i.e., ursolic acid (1) [bib_ref] Pharmacology of oleanolic acid and ursolic acid, Liu [/bib_ref] , oleanolic acid (2), betulinic acid (3) [bib_ref] Stigmasterol reduces plasma cholesterol levels and inhibits hepatic synthesis and intestinal absorption..., Batta [/bib_ref] , stigmasterol (4) [bib_ref] Stigmasterol reduces plasma cholesterol levels and inhibits hepatic synthesis and intestinal absorption..., Batta [/bib_ref] , and β-caryophyllene oxide (5) [bib_ref] Local anaesthetic activity of β-caryophyllene, Ghelardini [/bib_ref] ; two lignans, (-)-rabdosiin [bib_ref] The Indian holy power plant, Das [/bib_ref] [bib_ref] Tetrameric derivative of caffeic acid from Rabdosia japonica, Agata [/bib_ref] [bib_ref] High Rabdosiin and Rosmarinic acid production in Eritrichium sericeum callus cultures and..., Inyushkina [/bib_ref] and shimobashiric acid C (7) [bib_ref] Hyalurodinase inhibitors from Keiskea japonica, Murata [/bib_ref] ; three flavonoids, luteolin (8) [bib_ref] Distribution and biological activities of the flavonoid luteolin, López-Lázaro [/bib_ref] , its 7-O-β-D-glucuronide (9) [bib_ref] Leaf flavonoid glycosides as chemosystematic characters in Ocimum, Grayer [/bib_ref] [bib_ref] Flavonoid variation in the liverwort Conocephalum conicum: Evidence for geographic races, Markham [/bib_ref] [bib_ref] Flavonoid and phenolic glycosides from Salvia officinalis, Lu [/bib_ref] , and apigenin 7-O-β-D-glucuronide (10) [bib_ref] H NMR spectroscopy of flavonoids and their glycosides in DMSO-d6, Markham [/bib_ref] ; and phenolic compounds, (E)-p-coumaroyl 4-O-β-D-glucoside [bib_ref] Antinociceptive action of Ocimum sanctum (Tulsi) in mice: Possible mechanisms involved, Khanna [/bib_ref] [bib_ref] The phenols and prodelphinidins of white cover flowers, Foo [/bib_ref] , 3-(3,4-dihydroxyphenyl) lactic acid [bib_ref] In vivo studies on the effect of Ocimum sanctum L. leaf extract..., Babu [/bib_ref] [bib_ref] Isolation and Characterization of Phenolic Compounds from Coptidis Rhizoma, Yahara [/bib_ref] , protocatechuic acid (13) [bib_ref] New constituents from Ocimum sanctum, Norr [/bib_ref] , and vanillic acid [bib_ref] Ocimum sanctum leaf extracts stimulate insulin secretion from perfusd pancreas, isolated islets..., Hannan [/bib_ref] [bib_ref] New constituents from Ocimum sanctum, Norr [/bib_ref] were isolated. This is the first time that compounds 6, 7, 11, and 12 were isolated from this plant.
According to the literature, the taxonomic description of the genus Ocimum L. is still debatable. It is composed of three subgenera, namely subgenus Ocimum (comprising three sections: Ocimum, Gratissima and Hiantia), subgenus Nautochilus, and subgenus Gymnocimum. The species (O. sanctum L.) under investigation has been located in the subgenus Gymnocimum. This subgenus can be distinguished because of the existence of flavonoid glucuronides, which are found in plants of the subgenera Nautochilus and Ocimum [bib_ref] Distribution and biological activities of the flavonoid luteolin, López-Lázaro [/bib_ref]. Consequently, our work is in agreement with previous studies regarding the chemical profile of the subgenus Gymnocimum. Moreover, it was previously shown that 3-(3,4-dihydroxyphenyl) lactic acid is a precursor of the nonenzymatic synthesis of (S)-(-)-rosmarinic acid and (+)-rabdosiin [bib_ref] A non-enzymatic synthesis of (S)-(-)-rosmarinic acid and a study of a biomimetic..., Bogucki [/bib_ref] , therefore its identification (compound 12) could be related to the biosynthesis of (-)-rabdosiin (6) [bib_ref] Rabdosiin, a new rosmarinic acid dimer with a lignan skeleton, from Rabdosia..., Agata [/bib_ref].
Compound (-)-rabdosiin (6) [fig_ref] Figure 1: Chemical structures of [/fig_ref] is a caffeic acid tetramer connected to a lignan skeleton. Originally, it has been isolated and identified from the stem of Rabdosia japonica, Labiatae [bib_ref] Tetrameric derivative of caffeic acid from Rabdosia japonica, Agata [/bib_ref] , while both enantiomers (-)-rabdosiin and (+)-rabdosiin were later isolated from Macrotomia euchroma, Boraginaceae [bib_ref] Two caffeic acid tetramers having enantiomeric phenyldihydronaphthalene moieties from Macrotomia euchroma, Nishizawa [/bib_ref] and also from other plants of this family such as Lithospermum erythrorhizon [bib_ref] Cafeic acid oligomers in Lithospermum erythrorhizon cell suspension cultures, Yamamoto [/bib_ref] and Eritrichium sericeum [bib_ref] High Rabdosiin and Rosmarinic acid production in Eritrichium sericeum callus cultures and..., Inyushkina [/bib_ref]. Based on the fact that the entire fractionation and isolation procedures were continuously monitored by 1 H-NMR, the active compound 6 was not detected in other fractions (NMR data of 6 are provided as Supplementary Materials, Tables S1 and S2, [fig_ref] Figure 1: Chemical structures of [/fig_ref]. Consequently, being a minor compound of the plant, its activity could derive in synergy with other constituents.
Medicines 2019, 6, x FOR PEER REVIEW 5 of 10 sanctum L.) under investigation has been located in the subgenus Gymnocimum. This subgenus can be distinguished because of the existence of flavonoid glucuronides, which are found in plants of the subgenera Nautochilus and Ocimum [bib_ref] Distribution and biological activities of the flavonoid luteolin, López-Lázaro [/bib_ref]. Consequently, our work is in agreement with previous studies regarding the chemical profile of the subgenus Gymnocimum. Moreover, it was previously shown that 3-(3,4-dihydroxyphenyl) lactic acid is a precursor of the nonenzymatic synthesis of (S)-(-)-rosmarinic acid and (+)-rabdosiin [bib_ref] A non-enzymatic synthesis of (S)-(-)-rosmarinic acid and a study of a biomimetic..., Bogucki [/bib_ref] , therefore its identification (compound 12) could be related to the biosynthesis of (-)-rabdosiin (6) [bib_ref] Rabdosiin, a new rosmarinic acid dimer with a lignan skeleton, from Rabdosia..., Agata [/bib_ref]. Compound (-)-rabdosiin (6) [fig_ref] Figure 1: Chemical structures of [/fig_ref] is a caffeic acid tetramer connected to a lignan skeleton. Originally, it has been isolated and identified from the stem of Rabdosia japonica, Labiatae [bib_ref] Tetrameric derivative of caffeic acid from Rabdosia japonica, Agata [/bib_ref] , while both enantiomers (-)-rabdosiin and (+)-rabdosiin were later isolated from Macrotomia euchroma, Boraginaceae [bib_ref] Two caffeic acid tetramers having enantiomeric phenyldihydronaphthalene moieties from Macrotomia euchroma, Nishizawa [/bib_ref] and also from other plants of this family such as Lithospermum erythrorhizon [bib_ref] Cafeic acid oligomers in Lithospermum erythrorhizon cell suspension cultures, Yamamoto [/bib_ref] and Eritrichium sericeum [bib_ref] High Rabdosiin and Rosmarinic acid production in Eritrichium sericeum callus cultures and..., Inyushkina [/bib_ref]. Based on the fact that the entire fractionation and isolation procedures were continuously monitored by 1 H-NMR, the active compound 6 was not detected in other fractions (NMR data of 6 are provided as Supplementary Materials, Tables S1 and S2, [fig_ref] Figure 1: Chemical structures of [/fig_ref]. Consequently, being a minor compound of the plant, its activity could derive in synergy with other constituents. According to published data, rabdosiin and the similar caffeic acid derivatives have been suggested as potential anti-HIV and antiallergic agents. Moreover, studies showed that rabdosiin is an antioxidant factor (acting as an effective scavenger of reactive oxygen species), as well as a possible inhibitor of hyaluronidase and β-hexosaminidase release [bib_ref] Antiallergic activities of rabdosiin and its related compounds: Chemical and biochemical evaluations, Ito [/bib_ref] [bib_ref] Anti-AIDS agents, 18. Sodium and potassium salts of caffeic acid tetramers from..., Kashiwada [/bib_ref]. Nevertheless, to the best of our knowledge, the antiproliferative activity of rabdosiin is reported for the first time.
## Antiproliferative activityod of secondary metabolites of o. sanctum
Using the MTT dye reduction assay, the methanol:water extract (7:3) and 6 purified secondary metabolites (compounds 6, 7, 9, 11, 12, and 13) were screened for their cytotoxic/cytostatic activity against human breast and colon cell lines. Our results showed that the extract was cytotoxic against all cell lines, with an IC50 range of 45 2.12 to 57 14.14 g/mL [fig_ref] Table 1: In vitro cytotoxicity of the methanol extract and isolated compounds from Tulsi... [/fig_ref]. Based on these data, we further proceeded to the screening of the isolated natural products 6, 7, 9, 11, 12, and 13 against MCF-7 cells which was the mostly affected cell line exposed to the methanol extract of Ο. sanctum L. The IC50 values calculated are presented in [fig_ref] Table 1: In vitro cytotoxicity of the methanol extract and isolated compounds from Tulsi... [/fig_ref]. Among the purified compounds, the most prominent was 6, which was further tested against SKBR3 and HCT-116 cells. Overall, compound 6 demonstrated a considerable cytotoxic activity, with IC50 values 75 2.12, 83 ± 3.54 and 84 ± 7.78 μg/mL against MCF-7, SKBR3, and HCT-116, respectively. According to published data, rabdosiin and the similar caffeic acid derivatives have been suggested as potential anti-HIV and antiallergic agents. Moreover, studies showed that rabdosiin is an antioxidant factor (acting as an effective scavenger of reactive oxygen species), as well as a possible inhibitor of hyaluronidase and β-hexosaminidase release [bib_ref] Antiallergic activities of rabdosiin and its related compounds: Chemical and biochemical evaluations, Ito [/bib_ref] [bib_ref] Anti-AIDS agents, 18. Sodium and potassium salts of caffeic acid tetramers from..., Kashiwada [/bib_ref]. Nevertheless, to the best of our knowledge, the antiproliferative activity of rabdosiin is reported for the first time.
## Antiproliferative activityod of secondary metabolites of o. sanctum
Using the MTT dye reduction assay, the methanol:water extract (7:3) and 6 purified secondary metabolites [fig_ref] Figure 1: Chemical structures of [/fig_ref] were screened for their cytotoxic/cytostatic activity against human breast and colon cell lines. Our results showed that the extract was cytotoxic against all cell lines, with an IC 50 range of 45 ± 2.12 to 57 ± 14.14 µg/mL [fig_ref] Table 1: In vitro cytotoxicity of the methanol extract and isolated compounds from Tulsi... [/fig_ref]. Based on these data, we further proceeded to the screening of the isolated natural products 6, 7, 9, 11, 12, and 13 against MCF-7 cells which was the mostly affected cell line exposed to the methanol extract of O. sanctum L. The IC [bib_ref] Cafeic acid oligomers in Lithospermum erythrorhizon cell suspension cultures, Yamamoto [/bib_ref] values calculated are presented in [fig_ref] Table 1: In vitro cytotoxicity of the methanol extract and isolated compounds from Tulsi... [/fig_ref]. Among the purified compounds, the most prominent was 6, which was further tested against SKBR3 and HCT-116 cells. Overall, compound 6 demonstrated a considerable cytotoxic activity, with IC 50 values 75 ± 2.12, 83 ± 3.54 and 84 ± 7.78 µg/mL against MCF-7, SKBR3, and HCT-116, respectively. To analyze the type of cell death (apoptosis or necrosis) induced by compound 6 on MCF-7, SKBR3, and HCT-116 cells, cells were stained with annexin V which binds phosphatidylserine exposed on the surface of apoptotic cells and PI which intracellulary stains the DNA of necrotic cells. As shown in [fig_ref] Figure 2: Compound 6 induced apoptosis to human cancer cells [/fig_ref] , 80 µg/mL of compound 6 drove ca. 50% of all cells to apoptosis. Specifically, 44.9% of MCF-7 were annexin V+ and 12.3% annexin V+/PI+, suggesting that cells exposed to compound 6 underwent early apoptosis and a small percentage thereof late apoptosis/necrosis. Analogous percentages were obtained for SKBR3 (40.1% early apoptotic; 9.1% late apoptotic/necrotic) and HCT-116 (43.1% early apoptotic; 10.2% late apoptotic/necrotic) cells. When the same cell lines were exposed to 40 µg/mL of compound 6, the percentages of early apoptotic and late apoptotic/necrotic cells were reduced ca. by 50% (13.5-20.1% and 3.9-6.5%, respectively), suggesting that induction of apoptosis by compound 6 is concentration-dependent. To analyze the type of cell death (apoptosis or necrosis) induced by compound 6 on MCF-7, SKBR3, and HCT-116 cells, cells were stained with annexin V which binds phosphatidylserine exposed on the surface of apoptotic cells and PI which intracellulary stains the DNA of necrotic cells. As shown in [fig_ref] Figure 2: Compound 6 induced apoptosis to human cancer cells [/fig_ref] , 80 μg/mL of compound 6 drove ca. 50% of all cells to apoptosis. Specifically, 44.9% of MCF-7 were annexin V+ and 12.3% annexin V+/PI+, suggesting that cells exposed to compound 6 underwent early apoptosis and a small percentage thereof late apoptosis/necrosis. Analogous percentages were obtained for SKBR3 (40.1% early apoptotic; 9.1% late apoptotic/necrotic) and HCT-116 (43.1% early apoptotic; 10.2% late apoptotic/necrotic) cells. When the same cell lines were exposed to 40 μg/mL of compound 6, the percentages of early apoptotic and late apoptotic/necrotic cells were reduced ca. by 50% (13.5-20.1% and 3.9-6.5%, respectively), suggesting that induction of apoptosis by compound 6 is concentration-dependent. Based on the significant cytotoxic activity of compound 6 against cancer cell lines we further tested whether it may also be toxic against normal cells, i.e., PBMCs isolated from two different healthy blood donors. PBMCs were incubated for 24 h with the IC50 and the 1/2 concentration of 6, stained and analyzed by flow cytometry. Interestingly, the IC50 of compound 6 (80 μg/mL) induced Based on the significant cytotoxic activity of compound 6 against cancer cell lines we further tested whether it may also be toxic against normal cells, i.e., PBMCs isolated from two different healthy blood donors. PBMCs were incubated for 24 h with the IC 50 and the 1/2 concentration of 6, stained and analyzed by flow cytometry. Interestingly, the IC 50 of compound 6 (80 µg/mL) induced early and late apoptosis/necrosis in a small percentage of PBMCs (2.8% and 3.0% for donor 1; 4.3% and 3.1% for donor 2, respectively). At half concentration, the percentages were highly reduced and much less early apoptotic and late apoptotic/necrotic cells were detected (1.8% and 1.7% for donor 1; 2.1% and 1.9% for donor 2, respectively) [fig_ref] Figure 3: Compound 6 does not induce apoptosis or necrosis to peripheral blood mononuclear... [/fig_ref].
Medicines 2019, 6, x FOR PEER REVIEW 7 of 10 early and late apoptosis/necrosis in a small percentage of PBMCs (2.8% and 3.0% for donor 1; 4.3% and 3.1% for donor 2, respectively). At half concentration, the percentages were highly reduced and much less early apoptotic and late apoptotic/necrotic cells were detected (1.8% and 1.7% for donor 1; 2.1% and 1.9% for donor 2, respectively) [fig_ref] Figure 3: Compound 6 does not induce apoptosis or necrosis to peripheral blood mononuclear... [/fig_ref]. The good antitumor activity of compound 6 against human cancer cells and the simultaneous marginal cytotoxicity of the same compound when tested against normal human cells (PBMCs), suggest that (-)-rabdosiin may display less toxic side effects when administered in vivo. In support of our results, the few studies carried out in the last decade on the potential anticancer activity of O. sanctum extracts and its essential oil with different human cancer cell lines, clearly suggest that Tulsi may be used as a supplement to enhance anticancer chemotherapy without causing severe damage to normal epithelial cells [bib_ref] Purified essential oil from Ocimum sanctum Linn. triggers the apoptotic mechanism in..., Manaharan [/bib_ref] [bib_ref] Apoptosis Induction by Ocimum sanctum extract in LNCaP prostate cancer cells, Dhandayuthapani [/bib_ref] [bib_ref] Tulsi): An ethnomedicinal plant for the prevention and treatment of cancer, Bhattacharyya [/bib_ref]. Botanical drugs are currently approved in therapy with specific indications and in the last decades, research has focused on the anticancer effect of plant extracts.
Taken altogether, (-)-rabdosiin displays an interesting proapoptotic activity against cancer cell lines and in parallel shows a noticeable selectivity to malignant cells. It is noteworthy that the cytotoxic response of the extract is better compared to the other isolated compounds, including compound 6. As (-)-rabdosiin is a minor compound of the plant, we assume that it contributes to the improved antiproliferative activity of the methanol extract, and that it is probably synergistically with other active metabolites. The good activity of the polar extract, as well as of compound 6 against a series of human cancer cell lines and its marginal cytotoxicity against PBMCs, give evidence toward the effective use of this plant for the prevention of human cancer. Moreover, the core structure of (-)rabdosiin could be considered as drug lead in anticancer drug design.
Supplementary Material: The following are available online at www.mdpi.com/xxx/s1, Table S1: 1 Η-NMR of 6 (CD3OD, 400MHz); : [bib_ref] Biochemical evaluation of antidiabetogenic properties of some commonly used Indian plants on..., Narendhirakannan [/bib_ref] C-NMR of 6 (CD3OD, 400MHz); [fig_ref] Figure 1: Chemical structures of [/fig_ref] : 1 H-NMR spectrum of 6 (CD3OD, 400 Hz); [fig_ref] Figure 2: Compound 6 induced apoptosis to human cancer cells [/fig_ref] : COSY spectrum of 6 (CD3OD, 400 Hz); [fig_ref] Figure 3: Compound 6 does not induce apoptosis or necrosis to peripheral blood mononuclear... [/fig_ref] : 13 C NMR spectrum of 6 (CD3OD, 400 Hz); : HSQC spectrum of 6 (CD3OD, 400 Hz); : HMBC spectrum of 6 (CD3OD, 400 Hz); : Most important HMBC signals of compound 6. The good antitumor activity of compound 6 against human cancer cells and the simultaneous marginal cytotoxicity of the same compound when tested against normal human cells (PBMCs), suggest that (-)-rabdosiin may display less toxic side effects when administered in vivo. In support of our results, the few studies carried out in the last decade on the potential anticancer activity of O. sanctum extracts and its essential oil with different human cancer cell lines, clearly suggest that Tulsi may be used as a supplement to enhance anticancer chemotherapy without causing severe damage to normal epithelial cells [bib_ref] Purified essential oil from Ocimum sanctum Linn. triggers the apoptotic mechanism in..., Manaharan [/bib_ref] [bib_ref] Apoptosis Induction by Ocimum sanctum extract in LNCaP prostate cancer cells, Dhandayuthapani [/bib_ref] [bib_ref] Tulsi): An ethnomedicinal plant for the prevention and treatment of cancer, Bhattacharyya [/bib_ref]. Botanical drugs are currently approved in therapy with specific indications and in the last decades, research has focused on the anticancer effect of plant extracts.
Taken altogether, (-)-rabdosiin displays an interesting proapoptotic activity against cancer cell lines and in parallel shows a noticeable selectivity to malignant cells. It is noteworthy that the cytotoxic response of the extract is better compared to the other isolated compounds, including compound 6. As (-)-rabdosiin is a minor compound of the plant, we assume that it contributes to the improved antiproliferative activity of the methanol extract, and that it is probably synergistically with other active metabolites. The good activity of the polar extract, as well as of compound 6 against a series of human cancer cell lines and its marginal cytotoxicity against PBMCs, give evidence toward the effective use of this plant for the prevention of human cancer. Moreover, the core structure of (-)-rabdosiin could be considered as drug lead in anticancer drug design.
# Supplementary materials:
The following are available online at http://www.mdpi.com/2305-6320/6/1/37/s1, [fig_ref] Table 1: In vitro cytotoxicity of the methanol extract and isolated compounds from Tulsi... [/fig_ref] : 1 H-NMR of 6 (CD 3 OD, 400 MHz); : [bib_ref] Biochemical evaluation of antidiabetogenic properties of some commonly used Indian plants on..., Narendhirakannan [/bib_ref] C-NMR of 6 (CD 3 OD, 400 MHz); [fig_ref] Figure 1: Chemical structures of [/fig_ref] : 1 H-NMR spectrum of 6 (CD 3 OD, 400 Hz); [fig_ref] Figure 2: Compound 6 induced apoptosis to human cancer cells [/fig_ref] : COSY spectrum of 6 (CD 3 OD, 400 Hz); [fig_ref] Figure 3: Compound 6 does not induce apoptosis or necrosis to peripheral blood mononuclear... [/fig_ref] : 13 C NMR spectrum of 6 (CD 3 OD, 400 Hz); : HSQC spectrum of 6 (CD 3 OD, 400 Hz); : HMBC spectrum of 6 (CD 3 OD, 400 Hz); : Most important HMBC signals of compound 6.
[fig] Figure 1: Chemical structures of (-)-rabdosiin (6) isolated from O. sanctum. [/fig]
[fig] Figure 2: Compound 6 induced apoptosis to human cancer cells. MCF-7, SKBR3, and HCT-116 cells were exposed to 40 and 80 μg/mL of compound 6 for 72 h, stained with annexin V and PI, and analyzed by flow cytometry. Control cells were incubated in complete medium supplemented with 0.5% DMSO. Flow cytometry analysis was performed using FACS Diva software. (A). Representative dot plots from cells treated with compound 6. Percentages of early apoptotic (lower right), late apoptotic/necrotic (upper right), and necrotic (upper left) are shown in each quadrat. (B). Histograms of apoptotic and necrotic cells after exposure to compound 6. Blue columns show percentages of early apoptotic, red columns of late apoptotic and green columns of necrotic cells. Mean values ± SD from 3 experiments are shown. *, p < 0.05; **, p < 0.01; ***, p < 0.001, in all cases compared to control after Student's unpaired t-tests. [/fig]
[fig] Figure 3: Compound 6 does not induce apoptosis or necrosis to peripheral blood mononuclear cells (PBMCs). PBMCs were isolated from 2 different donors (1 and 2) and incubated with 40 and 80 μg/mL of compound 6 for 24 h. Other details as in Legend of Figure 2. Representative dot plots from both donors are shown from one experiment performed in duplicate. [/fig]
[table] Table 1: In vitro cytotoxicity of the methanol extract and isolated compounds from Tulsi on human cancer cell lines. [/table]
|
10.3390/ijerph19084434
|
CCBY
|
9030648
|
35457302
|
s2orc_pubmed_articles
|
Laugh before You Study: Does Watching Funny Videos before Study Facilitate Learning?
# Introduction
Emotion is a key factor affecting the process and outcome of learning. Academic settings are abound with emotions such as happiness, enjoyment, hope, pride, anger, anxiety, shame, hope, or boredom. The study of emotion in learning has aroused the interest of many researchers. Many studies have shown that some emotions promote learning while some hinder it. Based on how they promote or hinder learning, emotions can be divided into two broad categories, positive and negative. Happiness, enjoyment, and hope are examples of positive emotions that can boost intrinsic and extrinsic motivation, enhance the use of flexible learning strategies, and assist self-regulation, all of which can improve academic performance under most circumstances [bib_ref] Academic Emotions in Students' Self-Regulated Learning and Achievement: A Program of Qualitative..., Pekrun [/bib_ref]. Negative emotions such as boredom and anger impede students' motivation and learning engagement, thereby hinder learning [bib_ref] Activity Achievement Emotions and Academic Performance: A Meta-analysis, Camacho-Morles [/bib_ref]. Hence, how to arouse students' positive learning emotions is an important prerequisite to achieve the desired learning results.
In the study of emotion in multimedia learning, previous studies have mainly revealed that the positive emotions of instructors in video lectures and the emotional design of learning materials can promote the generation of positive emotions, so as to promote students to achieve better learning results. These studies provide empirical evidence for the integrated cognitive-affective theory of learning with multimedia. However, promoting students' learning through external emotional induction has not been fully studied. Emotions are dynamic and not as stable as mood or affect; they are easily triggered and affected. In multimedia learning, students' emotions can be influenced by three sources. First, learners' own emotions. Students do not start learning without emotions, and their pre-learning emotions are inherent just as their prior knowledge, but they are not as stable and lasting as prior knowledge. The second is the emotion generated by the interaction with the learning material. Shapes, colors, pictures, and emotional attributes of learning materials affect emotions. For example, bright colors tend to give people a bright feeling and make them feel positive, while gray tends to give people a dull feeling and make them feel sad [bib_ref] Relationship between color and emotion: A study of college students, Kaya [/bib_ref]. The third situation involves an instructional video with the teacher's presence. The instructor's vocal feedback, facial expressions, tone of voice, and body posture all convey emotions. From the existing literature, most studies have focused on the influence of the latter two emotional sources on students' learning, while inadequate attention has been paid to the first emotional source-pre-learning emotion.
Students' pre-learning emotion can be induced by various external materials. We often see students viewing funny videos to relieve pressure and regulate emotions while learning. It remains unclear whether funny videos are effective in inducing positive emotion and boosting learning, or how watching funny videos before learning changes students' emotion after learning. Therefore, this study aimed to investigate how learners' pre-learning emotions affect learning outcomes by determining whether the use of funny videos before learning evokes positive emotions and facilitates learning. Learning results were assessed by a retention and a transfer test. Moreover, measures of emotions after learning as well as motivation and satisfaction were used to elucidate how the emotional state might affect learning.
## Literature review
## Emotion and multimedia learning
Emotions are intrinsically motivating and intertwined with cognition, allowing and sustained cognitive activity including key learning mechanisms such as attention and memory [bib_ref] Emotional design in digital media for learning, Plass [/bib_ref]. The integrated cognitive-affective theory of learning with media stipulates that emotions and cognition are inseparable in multimedia learning. Learners experience emotional responses when they perceive auditory and visual information in the learning environment. In the working memory phase, emotions are involved in the choice of texts and images of the material process; in this stage, emotion is a type of mood. In the organization stage, emotions are involved in the visual process of psychological representation of image materials, with interest and motivation influencing each other. Finally, after the integration stage, the emotional schema is formed and stored in the long-term memory.
Considering the key role of emotions in multimedia learning, it has become a research hotspot to influence learners' emotional experience and thus learning results through emotional design of learning materials. Lately, a growing number of studies have focused on emotional design of educational videos that support learning. For example, Mayer and Estrella [bib_ref] Benefits of emotional design in multimedia instruction, Mayer [/bib_ref] redrew the learning material of how a virus causes colds using bright colors such as red or blue, with expressive eyes (registering surprise, fear, and sickness at various stages in the process). The results revealed that the enhanced emotional design markedly improved students' learning performance compared with the black-and-white design. A more detailed comparison of appearance elements is provided in Uzun and Yıldırım [bib_ref] Exploring the effect of using different levels of emotional design features in..., Uzun [/bib_ref]. Not limited to comparing colors and anthropomorphizing features, they added the element of sound. Specifically, they designed teaching materials with four different levels of emotional design: neutral, color, anthropomorphic, and anthropomorphic plus sound. The results revealed that positive emotions typically increased as the amount of emotional design features increased. Besides the shapes, colors, and anthropomorphic features of multimedia elements, the text exhibited an emotional potential, which could be estimated according to the linguistic aspects on how many emotional aspects (e.g., emotional states and emotional situations) are expressed in the text. Stark et al. [bib_ref] Emotional text design in multimedia learning: A mixed-methods study using eye tracking, Stark [/bib_ref] enriched the original text with either positive or negative emotional parentheses to test its impact on learning, establishing that the emotional text designs improved cognitive processing. Although these studies confirmed the impact of emotional design on learning, some studies highlighted that students' emotions actually decreased after learning emotional learning materials. That is, positive emotional design did not increase the learning effect by influencing students' emotions but promoted more effective information selection, organization, and processing by bright colors or eye-catching shapes, thereby improving students' performance in learning tasks [bib_ref] Can emotional design really evoke emotion in multimedia learning?, Li [/bib_ref].
In video lectures with the instructor's presence, the instructor's emotion also attracted many researchers' attention. To promote positive emotional states among students, some studies focused on the instructor's emotion expression. For example, Lawson et al. [bib_ref] Recognizing the emotional state of human and virtual instructors, Lawson [/bib_ref] explored whether students noticed an instructor's emotions during an instructional video, as well as how effectively participants could perceive different emotions depicted by a human and virtual instructor (i.e., animated pedagogical agent) in a video lecture. The results revealed that participants could recognize each of the emotions displayed by the instructor. Furthermore, without the presence of the teacher image, learners could distinguish whether the teacher was in a positive or negative emotion just by listening to the teacher's voice [bib_ref] The Power of Voice to Convey Emotion in Multimedia Instructional Messages, Lawson [/bib_ref]. Moreover, students rated the teacher with a positive voice higher. Regarding learning outcomes, students who studied from teachers with positive emotions attained better learning outcomes than those who studied from teachers with negative and neutral emotions [bib_ref] The positivity principle: Do positive instructors improve learning from video lectures?, Lawson [/bib_ref]. In conclusion, the above mentioned studies demonstrated that instructors' positive emotions can promote students' learning owing to the impact of teachers' emotions on students' emotions. Students' emotions become positive when they learned from teachers with positive emotions, which in turn stimulated students' learning engagement through better learning motivation, and ultimately led to better learning outcomes.
Overall, the studies mentioned above aimed to enhance the learning effect through emotional design of learning materials or teachers' positive emotional expression in the process of learning. However, these studies overlooked students' pre-learning emotions. Learners always have a basic emotion when entering the learning situation. The learners' pre-learning emotions is a crucial personality characteristic that cannot be ignored in teaching, such as students' previous knowledge and experience, and it exerts a more profound impact on the learning process, even affecting how students perceive teachers' emotions and the emotional design of learning materials [bib_ref] The importance of teachers' emotions and instructional behavior for their students' emotions-An..., Becker [/bib_ref]. Thus, this study aimed to investigate whether promoting students' pre-learning emotion can enhance the learning outcome.
## Pre-learning emotion
Emotions are situational; that is, different emotions are triggered in different situations. Thus, the research on emotions should be combined with specific situations. According to different academic scenes, emotions were divided into class-related emotions, learningrelated emotions, and test-related emotions. For each situation, it could be categorized into three parts based on the time period. For example, learning-related emotions could be divided into pre-learning emotions, during-learning emotions, and post-learning emotions. In this study, we largely focused on learning related emotions and more specifically, prelearning emotions.
To date, few empirical studies have focused on learners' pre-learning emotions, and only two studies have investigated this issue. Park et al. [bib_ref] Emotional design and positive emotions in multimedia learning: An eyetracking study on..., Park [/bib_ref] established that by inducing students to have positive pre-learning emotions, their understanding of learning content and transfer performance could be improved; however, they did not compare how negative pre-learning emotions affected learning outcomes. Knörzer et al. [bib_ref] Facilitators or suppressors: Effects of experimentally induced emotions on multimedia learning, Knörzer [/bib_ref] filled this gap by using a combination of music and autobiographic recall to induce learners to produce positive, negative, or neutral emotions in the laboratory environment. The negative emotion group outperformed the positive emotion group, according to the findings. However, they claimed that they could not deduce whether the negative emotional state before learning was beneficial to learning because the negative emotion group was actually in a neutral emotional state in the experiment, and the results should be explained such that the positive emotion exerted an adverse impact on learning. In other studies, the highly active positive emotional state broadened students' scope of attention [bib_ref] Emotions and personality traits as high-level factors in visual attention: A review, Kaspar [/bib_ref] or distract students from learning materials [bib_ref] Irrelevant thoughts, emotional mood states, and cognitive task performance, Seibert [/bib_ref]. The negative or neutral emotional state made students' attention more focused and cognitive processing more detailed. Although the conclusions of the two studies mentioned above are inconsistent, this might be caused by the sample and task difficulty; however, both showed that learners' pre-learning emotions are essential factors that affect cognitive results.
## Emotion induction
Emotion induction and emotion regulation are similar but different terms; both denote the change or influence of emotion. The difference between the two is that emotion induction is primarily influenced by others by certain means and methods. Emotion regulation includes not only the influence and change of others on individual emotions but also the influence and change of oneself on emotions. In this study, we mainly referred to the emotional change of individuals by others, that is, emotion induction. In addition, emotions are the psychological process evoked by a perception of an event, a memory, and specific types of media such as photographs, voice, and words. Siedlecka and Denson [bib_ref] Experimental methods for inducing basic emotions: A qualitative review, Siedlecka [/bib_ref] broadly classified emotion induction techniques into five specific methods: visual stimuli (including static images or videos), listening to music, autobiographical recall, situational procedures, and imagery.
When it comes to inducing students' pre-learning emotions, video may be the best choice. In traditional classroom teaching, we often see or experience that teachers organize students to watch a video before class to stimulate their positive emotions. Learners themselves often unconsciously use short funny videos to regulate their emotions; they watch short videos for entertainment and relaxation after studying for a period, and then re-devote themselves to their studies. In addition to its good ecological validity, video as an emotion inducing material has the following advantages. First, videos convey more information than pictures or sounds alone and thus induce emotions quicker. Second, unlike autobiographical recall and imaginary which relies on individual personality traits, video presents external information to elicit emotion; thus, it induces cleaner emotions [bib_ref] A video is worth a thousand thoughts: Comparing a video mood induction..., Devilly [/bib_ref]. Third, videos save time and manpower when compared with situational procedures, which require creating a social scenario that elicit the desired emotion. Overall, using videos to induce students' emotions is a relatively simple, effective, and scalable method.
## Research questions
Can watching funny videos before study improve learning? This study explored whether using funny videos to induce students' pre-learning emotions can effectively promote their learning. Although two previous studies [bib_ref] Emotional design and positive emotions in multimedia learning: An eyetracking study on..., Park [/bib_ref] [bib_ref] Facilitators or suppressors: Effects of experimentally induced emotions on multimedia learning, Knörzer [/bib_ref] have examined the effects of regulating learners' pre-learning emotions on multimedia learning outcomes, participants learned the material independent of their professional background. In addition, the researchers highlighted that future research should focus on whether inducing students' pre-learning emotions can produce different results when learning materials related to their own learning task; that is, in a conventional learning environment, students' motivation to learn is different from participating in an experiment to learn unrelated learning materials. Motivation is a key driver of learning; therefore, it might lead to conclusions different from those of existing studies. This study explored whether regulating students' pre-learning emotions while learning materials related to their own tasks would affect their learning outcomes. In addition, unlike previous two studies, participants in our experiment were primary school students, and we used funny videos as a means of emotion induction.
The specific research questions addressed in this study are:
RQ1: Can funny videos make students in the experimental group feel more positive before learning? RQ2: How is the change of emotion between the two groups after learning? RQ3: Does the experimental group have higher learning motivation and satisfaction than the control group?
RQ4: Does the experimental group perform better in knowledge retention and knowledge transfer than the control group?
# Materials and methods
## Participants and experimental design
In this study, participants were 5th grade (approx. age: 11 years) students from a primary school in south-central China. A total of 81 students participated in the study. The students here had video learning experience; they often used computers and iPads to surf the Internet to collect learning materials or view videos at school and at home. All participants gave written informed consent of their parents and received a small gift for participating in the study. The Academic Committee of the School of Psychology at CCNU approved the study protocol.
# Materials
# Emotion induction material
We edited two short videos (2-3 min) to induce positive emotions as well as neutral emotions. One video was excerpts from an online program named "light moment", with a child and a dog eating lunch together to induce positive emotions [fig_ref] Figure 1: A screenshot of the positive emotion induction video [/fig_ref]. The other was a multicolored display of moving blue bars to induce neutral emotions [fig_ref] Figure 2: A screenshot of the natural emotion induction video [/fig_ref].
# Learning material
We created an instructional video, the teaching content of which was the elementary science knowledge point "The Composition of Rocks". In order to reduce the influence of emotional design of learning materials on students' emotion in the process of learning, we strictly controlled the emotional design of various multimedia elements to make them present neutral emotional colors. Specifically, the video manuscript, as well as the text and pictures in the slides all did not contain strong emotional attributes. As for the instructor, she kept a gentle facial expression, fluctuated her tone, and faced her body toward the camera. When she used gestures to point to the teaching material, her eyes also looked at the teaching material, and her body turned to the teaching material. [fig_ref] Figure 3: A screenshot of the video lecture [/fig_ref]. We also invited six master-level graduate students and two doctoral students majoring in education to watch the video materials and score the emotional attributes of the video materials. Among them, 1 represents very negative, 4 represents neutral, and 7 represents very positive. The results showed that the affective attribute of learning materials was neutral (M = 4.25, SD = 0.43).
## Measures
The measurements included one demographic questionnaire, three knowledge tests, two emotion questionnaires, one learning motivation scale, and one learning satisfaction scale.
## Demographic questionnaire
Participants were asked to report their sex, age, and year in school.
## Knowledge tests
We invited a science expert, together with the teacher in the video lecture, to design three knowledge tests: (I) a prior knowledge test; (II) a knowledge retention test; and (III) a knowledge transfer test. The pre-knowledge test was designed to understand students' background knowledge about rocks. It included true or false questions (e.g., gold is not a mineral), single-choice questions (e.g., what is true about the composition of rocks? (A) rock is composed of one mineral, (B) rock is composed of three or more minerals, and (C) rock is composed of one or more minerals), and multiple-choice questions (e.g., the following characteristics of rocks are? (A) color, (B) length, and (C) gloss), with a total of 10 points; the higher the score, the higher the level of knowledge. The retention test was designed to measure students' memory of the content. The answers to the questions could be found directly from the video material. It included fill-in-the-blank questions (e.g., the hardest mineral found in nature is ____), true or false questions (e.g., sand is a kind of rock), and picture recognition questions (the pictures included in this section can be found in Supplementary Materials), with a total of 18 points; the higher the score, the better the knowledge retention. The transfer test was designed to test students' understanding of the content and its application to the knowledge [bib_ref] Emotional Design in Multimedia Learning, Um [/bib_ref]. The answers to the questions could not be found directly from the video materials. It required learners to make appropriate reasoning based on their understanding of the learning material. It included true or false questions (e.g., rocks and minerals are the mineral resources of the earth, not the resources that people produce and live in), single choice questions (e.g., the reflection of light on the surface of rock forms the ___ of rock. (A) color, (B) gloss, and (C) transparency), and a short answer question (e.g., name two examples of the use of rocks in your life), with a total of 13 points; the higher the score, the better the knowledge transfer.
## Emotion questionnaires
We adopted the emotion questionnaires from Horovitz and Mayer [bib_ref] Learning with human and virtual instructors who display happy or bored emotions..., Horovitz [/bib_ref] to evaluate students' pre-learning emotions and post-learning emotions.
Before learning, participants rated their emotions on a five-point Likert scale (1 = strongly disagree and 5 = strongly agree) on the following four items: before learning, I feel happy; before learning, I feel content; before learning, I feel frustrated; before learning, I feel bored. After class, participants rated their emotions. Students' post-class emotions also included four questions: after learning, I feel happy; after learning, I feel content. after learning, I feel frustrated. after learning, I feel bored; these four questions used the five-point Likert scale (1 = strongly disagree and 5 = strongly agree).
## Learning motivation
The learning motivation questionnaire was an excerpt from Stull et al. [bib_ref] An eye-tracking analysis of instructor presence in video lectures, Stull [/bib_ref] , containing six items on participants' enjoyment, willingness to learn in this way in the future, understanding of the learning materials, desire to learn more about the content, finding the lesson useful, and motivation to learn the content. The questionnaire used a seven-point Likert scale (1 = strongly disagree and 7 = strongly agree). In this study, Cronbach α for learning motivation was 0.904.
## Learning satisfaction
The learning satisfaction questionnaire contained three items to measure students' satisfaction with the teacher's teaching, teaching content, and learning environment. Each item used a five-point Likert scale (1 = strongly disagree and 5 = strongly agree). In this study, Cronbach α for learning satisfaction was 0.814.
## Procedure
This study was conducted in a multimedia classroom with a projector. All participants came from two parallel classes, that is, there was no significant difference in the students' academic performance in the two classes. Moreover, the two classes had similar sex ratio and age distribution. Therefore, the randomization process was class-based. We randomly assigned one class as the control group and the other as the experimental group. In the control group, there were 22 boys and 18 girls. In the experimental group, there were 23 boys and 18 girls. Both groups were escorted into different multimedia classroom where the study was conducted. The order of measurement implementation may affects the accuracy of the measurement [bib_ref] Test-enhanced learning: Taking memory tests improves long-term retention, Roediger [/bib_ref] ; we fully considered this point. The specific procedure was shown as below:
First, all participants filled out the demographic questionnaire and prior knowledge test. Then, students in the control group watched blue moving bars for 2 min. Meanwhile, students in the experimental group watched funny videos for about 2 min. Second, the control group and the experimental group filled in the pre-learning emotion questionnaire. Rather than taking an emotion test before the prior knowledge test, we avoided pre-learning emotions reported by students from being affected by the prior knowledge. Third, both groups watched the same video to learn the material. Fourth, both groups reported their post-learning emotions. Fifth, both groups did retention and transfer tests. Finally, students were invited to report their motivation and satisfaction with the course.
# Statistical analysis
Different data analysis methods were used in this study to address the research question. For RQ1, RQ3, and RQ4, the independent sample t-test was conducted to compare the differences between the experimental group and the control group on several dependent variables. Our null hypothesis H0 states that there was no significant difference between the experimental group and the control group in pre-learning emotion, motivation, satisfaction, retention score, and transfer score. The formula is µ 1 − µ 2 = 0; accordingly, our alternative hypothesis H0 states that there was a significant difference between the experimental group and the control group in pre-learning emotion, motivation, satisfaction, retention score, and transfer score. The formula is µ 1 − µ 2 = 0. If the significance of t is less than 0.05, the null hypothesis should be rejected; otherwise, the null hypothesis should be accepted. p value was considered statistically significant when it was two-tailed.
For RQ2, the generalized estimation equation (GEE) was used to compare the emotional differences between the two groups at different time points and the emotion differences of each group at different time points. Although repeated measure ANOVA was also applicable to solve this problem, the prerequisite conditions of repeated measure ANOVA are relatively strict such as normality, homogeneity, and particularly sphericity [bib_ref] A Simple and Transparent Alternative to Repeated Measures ANOVA, Grice [/bib_ref]. The generalized estimation equation (GEE) has unique advantages in data analysis of repeated measurements and can support multiple data types and distribution patterns [bib_ref] Generalized Estimating Equations in Longitudinal Data Analysis: A Review and Recent Developments, Wang [/bib_ref]. This study mainly involved two independent variables, time and group, and four dependent variables: happy, content, frustrated, and bored. To answer the research question, we analyzed the interaction effect of time and group with one emotion as the dependent variable. The model of the five correlation structures was carried out and the QIC value was recorded. The structure that obtains the smaller QIC value shows better fit of the model to the data [bib_ref] Akaike's Information Criterion in Generalized Estimating Equations, Pan [/bib_ref]. The QIC results are shown in [fig_ref] Table 1: QIC for different correlation structures [/fig_ref]. It suggests that AR, exchangeable, M-dependent, and unstructured all had a smaller QIC compared to independence. Thus, we could choose one of them except independence for the correlation structure to build the model.
# Results
To adequately interpret the results, it was crucial to determine whether the groups differed significantly on basic characteristics. Thus, we conducted an independent sample t-test between the two groups. As shown in [fig_ref] Table 2: The basic characteristics of the two groups [/fig_ref] , the independent sample t-test showed no significant differences in prior knowledge between the two groups (p = 0.703). To examine whether the funny video promoted the pre-learning emotions of the experimental group, an independent sample t-test was conducted on the pre-learning emotion data of the two groups. A significant difference was noted in pre-learning emotions, as shown in [fig_ref] Table 3: The emotional state of the two groups before learning [/fig_ref]. Students in the experimental group were significantly happier (p < 0.001) and more content (p = 0.002) than students in the control group. In addition, students in the control group were significantly more frustrated (p = 0.003) and more bored (p = 0.001) than students in the experimental group.
## Rq2: how is the change of emotion between two groups after learning?
To examine the change in emotions between the two groups after learning, we conducted generalized estimation equations(GEE) to analyze the data. The tests of model effect results are shown in [fig_ref] Table 4: The tests of model effect [/fig_ref]. It shows that the interaction effect of time and group was not significant for the happy and content emotions, but for the frustrated and bored emotions, the interaction effect of time and group was significant. The estimated marginal mean and pairwise comparison results are shown in [fig_ref] Table 5: Multivariable GEE results [/fig_ref]. The rows 3-6 of the table show the interaction between time and group; it represents the differences in emotion between the experimental group and the control group at different time points. It can be seen that before learning, there were significant differences in happy, content, frustrated, and bored emotions between the experimental group and the control group. Specifically, the experimental group was happier and more content than the control group. The control group was more frustrated and bored than the experimental group. After learning, there were still significant differences in happy, frustrated, and bored emotions between the experimental and control groups. The experimental group was happier than the control group, and the control group was more frustrated and bored than the experimental group. There was no significant difference in content emotion between the two groups.
The last four rows of the [fig_ref] Table 5: Multivariable GEE results [/fig_ref] show the interaction between group and time; it represents the difference in emotion between the experimental group and the control group at different time points. It can be seen that the happy, frustrated, and bored emotions of the experimental group changed significantly before and after learning. Specifically, after learning, the students' happiness decreased and frustration and boredom increased. Students' content emotion also decreased, but there was no significant difference. For the control group, happy, content, frustrated, and bored emotions did not change significantly before and after learning. However, it can be seen from the mean value that the happy and content emotions of the control group showed an upward trend, while the frustrated and bored showed a downward trend. The emotion changes of different groups before and after learning are illustrated in [fig_ref] Figure 4: The emotion changes of the control group [/fig_ref]. The overall emotion changes of the two groups at different time points are shown in [fig_ref] Figure 6: The overall emotion changes of different group at different time points [/fig_ref].
## Rq3: does the experimental group have higher learning motivation and satisfaction than the control group?
To determine any difference between the two groups on motivation and satisfaction, the independent sample t-test was used. The results showed no significant differences on motivation (p = 0.286) and satisfaction (p = 0.520). [fig_ref] Table 6: Dependent variables between the two groups [/fig_ref] presents the descriptive data. To determine any difference between the two groups on retention and knowledge transfer, the independent sample t-test was used. The results showed no significant differences in knowledge retention (p = 0.143) between the two groups. However, the transfer test (p < 0.001) of the experimental group was significantly higher than the control group. [fig_ref] Table 7: Dependent variables between the two groups [/fig_ref] presents the descriptive data.
# Discussion
## This work
This study examined the effect of students' pre-learning emotions on primary students' emotional state, motivation, satisfaction, and learning performance in video lectures. The following is an analysis and discussion of the research results.
## Rq1 can funny videos make students in the experimental group feel more
Positive before Learning?
Students in the experimental group were significantly happier (p < 0.001) and more content (p = 0.002) than students in the control group. Meanwhile, students in the control group were significantly more frustrated (p = 0.003) and more bored (p = 0.001) than students in the experimental group. These findings corroborate previous studies. Abel and Maxwell [bib_ref] Humor and affective consequences of a stressful task, Abel [/bib_ref] reported that viewing a humorous video compared with a nonhumorous video reduced anxiety and improved positive affect under both low and high stress. Moreover, when watching a funny video, one experienced joy and delight [bib_ref] Finding Meaning at Work: The Role of Inspiring and Funny YouTube Videos..., Janicke-Bowles [/bib_ref].
## Rq2 how is the change of emotion between two groups after learning?
After completing video learning, students' emotions in the experimental group became significantly negative, while that of the control group became slightly but not significantly positive. This could be a comparison effect; that is, the comparison between the induction videos and the learning videos resulted in the different trends in emotions of both groups. Regarding the experimental group, the learning videos were not as interesting as the funny videos; thus, the students' emotions became negative after watching the learning videos. Regarding the control group, the learning videos were less monotonous than the color-block videos; thus, after watching the learning video, students' emotions became slightly but not significantly positive.
## Rq3 does the experimental group have higher learning motivation and satisfaction than the control group?
We found no significant difference in motivation and satisfaction between the two groups, which is contrary to a previous study. It was confirmed that students' enjoyment of learning positively correlated with their intrinsic and extrinsic motivation [bib_ref] Emotional Design in Multimedia Learning, Um [/bib_ref] , whereas correlations for boredom with motivation were negative. The lack of significant difference in motivation between both groups is attributable to the way emotions were induced. The funny video materials used in this study were funny fail videos of humans and animals. Per Gable and Harmon-Jones [bib_ref] Approach-motivated positive affect reduces breadth of attention, Gable [/bib_ref] proved that cute animal and human videos triggered emotions of low motivational intensity. Food videos, instead, stimulated high motivational intensity. Thus, the motivation of positive emotions evoked by video clips used in this study was limited. In addition, it correlated with the time of emotion induction. The duration of emotion induction before a formal learning task might not be as long as the duration of emotion induced by the learning environment. Thus, temporary positive emotional experiences might not motivate students to learn.
Regarding satisfaction, we also found no significant difference in learning satisfaction between the two groups. A previous study confirmed that emotions could influence a learner's attitude toward something or somebody [bib_ref] What are emotions? And how can they be measured?, Scherer [/bib_ref]. However, in this study, we did not find a significant difference between the two groups, which could be caused by the comparison, with the control group in a natural emotion rather than a negative emotion. Artino [bib_ref] Motivational beliefs and perceptions of instructional quality: Predicting satisfaction with online training, Artino [/bib_ref] demonstrated that happy emotions positively predicted students' learning satisfaction, while depression negatively predicted students' learning satisfaction. In addition, students' emotions and motivation jointly influenced satisfaction, and they explained a significant portion of variance in satisfaction. Owing to no significant difference in motivation and the emotion difference being small after learning between the two groups, no significant difference was found in satisfaction.
## Rq4 does the experimental group perform better in knowledge retention and
Knowledge Transfer than the Control Group?
We found no significant difference in the retention test, which mainly focused on students' memory of learning material; this is attributable to no significant difference in students' learning motivation. Motivation is the driving force behind cognitive processing, which results in improved learning outcomes. The second reason may due to the level of difficulty of retention tests. Perhaps positive emotions enhance performance on tasks that mostly require divergent thinking (e.g., think about the various uses of the pencil) [bib_ref] Understanding the relationship between mood and creativity: A meta-analysis, Davis [/bib_ref]. In addition, the retention test largely examined the students' knowledge through memory and repetition; it did not require much positive emotion input, but negative emotions impaired students' memory [bib_ref] Negative emotional outcomes impair older adults' reversal learning, Nashiro [/bib_ref]. The third reason is that the activation levels of emotions might influence performance on tasks [bib_ref] A meta-analysis of 25 years of mood-creativity research: Hedonic tone, activation, or..., Baas [/bib_ref]. Finally, the funny video we used in this study was not related to the learning content. A previous study demonstrated that funny videos that were compatible with the course topic boosted student acquisition and content retention [bib_ref] The acute effects of humor and exercise on mood and anxiety, Szabo [/bib_ref].
Regarding the transfer test, it reflected the deep processing of learning materials. We found a significant difference between the two groups in the transfer test, and the experimental group performed better in the transfer test, which is consistent with a previous study. Um et al. [bib_ref] Emotional Design in Multimedia Learning, Um [/bib_ref] showed that external induction of positive emotions did not enhance retention but did improve transfer. Moreover, Politis and Houtz [bib_ref] Effects of Positive Mood on Generative and Evaluative Thinking in Creative Problem..., Politis [/bib_ref] also found that participants who in the positive emotion condition were significantly more fluent when solving creative problem than those who watched the neutral video. Gökçen showed that watching a funny video, especially about a foreign language, affected learning [bib_ref] An exploration of the relationship between watching a funny video and its..., Gökçen [/bib_ref] ; this may be because positive emotions improved the flexibility of thinking, promoted deep processing, and thus improved transfer performance [bib_ref] Emotional modulation of control dilemmas: The role of positive affect, reward, and..., Goschke [/bib_ref]. The second reason could be that positive pre-learning emotions can increase cognitive engagement, which results in better transfer performance [bib_ref] Does speaker's voice enthusiasm affect social cue, cognitive load and transfer in..., Liew [/bib_ref].
## Theoretical and practical contributions
This study extended the literature on emotions in multimedia learning beyond college samples and further considered pre-learning emotion induction. We examined the effects of watching funny videos before learning on primary school students' learning outcomes. This study supported the integrated cognitive-affective theory of learning with media and the understanding that positive emotions promote learners' cognitive processing. When students are in a positive emotion, they are more willing to put more mental work into processing the learning material, thereby producing better learning results. Moreover, the study offered important new insights into emotions. Learners' pre-learning emotions are the real starting point that affects their deep learning.
The practical implication of this study is that the efficacy of funny short videos as a regulation of learning emotions was verified. This study shed light on teaching practices in the following two aspects. First, the appropriate use of short videos can promote students' positive emotions and thus improve their learning performance. Of note, watching funny videos is often used for leisure and entertainment, but few researchers have used it as a means of regulating students' emotions. Teachers can evoke positive emotions of students by using this easy and economic method. Second, teachers should pay attention to students' pre-learning emotions which affect their learning results. It is wise to execute some activities such as singing and playing games to prepare students for positive emotions before learning.
## Limitation and future research
Although meaningful findings were reported in this study, there are three limitations. First, this study only used self-ratings of emotional state; it was subjective data. Although previous studies have highlighted that self-reported arousal correlates with students' regulation of their effort in task and self-reported valence correlates with cognitive regulation processes. Our study lacked some objectivity in this regard. Thus, future research would benefit from more direct measures such as skin conductance response (SCR) and facial muscle electromyogram (EMG). With the combination of multimodal objective (SCR and EMG) and subjective (self-report) data, the learning process can be captured in a new continuous way. The second limitation is generalizability. This study was conducted in a relatively short learning session, and the knowledge type was declarative knowledge with only students from a primary school in China. It was unclear whether the effect of pre-learning positive emotion could last for longer instructional videos. In addition, when it comes to learning procedural knowledge, it may produce different learning results. Finally, do emotions change differently in different cultural contexts for people of different age? Additional studies are needed to examine the efficacy of funny videos to regulate students' pre-learning emotions in a more extended learning session for different knowledge types or in different cultures. The third research limitation is the type of emotion-inducing videos. The video used in this paper was a funny video, which makes students produce positive emotions. However, there are several kinds of videos that can elicit positive emotions such as food videos and singing and dancing videos. Whether these other kinds of videos can produce the same results as this study remains to be further studied.
# Conclusions
This study examined whether watching funny videos before learning can make students' emotions more positive, thereby improving their motivation, satisfaction, and task performance. The findings established that students who watched the funny video actually had more positive pre-learning emotions than students who did not. However, no significant differences were reported in learning motivation, learning satisfaction, and knowledge retention between the two groups. Furthermore, students in the experimental group performed better on the transfer test than those in the control group.
## Institutional review board statement:
The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of the School of Psychology at Central China Normal University (ccnu-irb-202111-003).
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.
# Data availability statement:
The data are not publicly available due to privacy restrictions.
[fig] Figure 1: A screenshot of the positive emotion induction video. [/fig]
[fig] Figure 2: A screenshot of the natural emotion induction video. [/fig]
[fig] Figure 3: A screenshot of the video lecture. [/fig]
[fig] Figure 4: The emotion changes of the control group. [/fig]
[fig] Figure 5: The emotion changes of the experimental group. [/fig]
[fig] Figure 6: The overall emotion changes of different group at different time points. [/fig]
[fig] Supplementary: Materials: The knowledge tests used in this study can be downloaded at: https: //www.mdpi.com/article/10.3390/ijerph19084434/s1. Author Contributions: Conceptualization, M.W. and Z.C.; methodology, M.W.; software, M.W.; validation, M.W.; formal analysis, M.W.; investigation, M.W.; resources, Z.C.; data curation, M.W.; writing-original draft preparation, M.W.; writing-review and editing, M.W. and Z.C.; visualization, M.W.; supervision, Z.C.; project administration, Z.C.; funding acquisition, Z.C. All authors have read and agreed to the published version of the manuscript. [/fig]
[fig] Funding: This research was funded by the National Natural Science Foundation of China (No. 62077022) and National Collaborative Innovation Experimental Base for Teacher Development of Central China Normal University (No. CCNUTEIII 2021-21). [/fig]
[table] Table 1: QIC for different correlation structures. [/table]
[table] Table 2: The basic characteristics of the two groups. [/table]
[table] Table 3: The emotional state of the two groups before learning. [/table]
[table] Table 4: The tests of model effect. [/table]
[table] Table 5: Multivariable GEE results. [/table]
[table] Table 6: Dependent variables between the two groups. [/table]
[table] Table 7: Dependent variables between the two groups. [/table]
|
10.3390/jof6040209
|
CCBY
|
7712195
|
33036295
|
s2orc_pubmed_articles
|
Isavuconazole—Animal Data and Clinical Data
The treatment of invasive fungal infections has deeply evolved in the last years with the inclusion of new antifungals, mainly new azoles (i.e., posaconazole, isavuconazole), to the therapeutic armamentarium. This review focuses on the role of isavuconazole for treating the most important invasive fungal infections both in animals and humans (hematological and non-hematological patients).
# Introduction
In the last years, new antifungal drugs have been commercialized, thus allowing an easier management of invasive fungal diseases (IFD), treatment and outcome. Isavuconazole (ISV) is a new antifungal agent, with a favorable drug-drug interaction profile, reduced drug-related adverse events and efficacy similar to voriconazole for treatment of invasive muld infections, as demonstrated in a non-inferiority trial of IFD treatment [bib_ref] Isavuconazole versus voriconazole for primary treatment of invasive mould disease caused by..., Maertens [/bib_ref]. ISV demonstrated efficacy also against rare fungi and in patients with impaired renal function [bib_ref] Isavuconazole treatment for mucormycosis: A single-arm open-label trial and case-control analysis, Marty [/bib_ref]. Due to the broad spectrum of action and to its favorable interaction profile, ISV has been proposed as a very effective antifungal drug and, mainly for patients with hematological malignancies (HMs), international guidelines strongly recommended its use for the treatment of invasive aspergillosis and mucormycosis [bib_ref] Diagnosis and management of Aspergillus diseases: Executive summary of the 2017 ESCMID-ECMM-ERS..., Ullmann [/bib_ref] [bib_ref] Global guideline for the diagnosis and management of mucormycosis: An initiative of..., Cornely [/bib_ref].
In the present manuscript, we analyzed the literature regarding studies on animal models using ISV and the role of ISV in the treatment (target therapy and prophylaxis) of IFD in humans, stratifying the latter in non-hematological and hematological patients, in order to describe the actual clinical experiences in the use of ISV for treatment of IFDs and discuss further possible clinical applications of this drug.
# Methods
A comprehensive literature search was performed using the Pubmed and Sciencedirect electronic databases. The search was performed until August 1, 2020, limiting the choice to English-language articles. All the papers that regarded "isavuconazole" were reviewed, and the relevant articles were selected. In addition, the references lists from the selected articles were used to obtain further articles not included in the electronic database. The authors reviewed all the publications identified and prepared their observations. After a revision according to the results of the plenary discussion, a summary report was made.
## Isavuconazole in animal models
Studies conducted on animal models using ISV and focusing on dose determination, data on pharmacokinetics (PK) and pharmacodynamics (PD) parameters, and efficacy studies in animal models have demonstrated favorable results, as previously reviewed [bib_ref] Isavuconazole for the treatment of invasive aspergillosis and mucormycosis: Current evidence, safety,..., Natesan [/bib_ref].
More recently, pharmacodynamics and efficacy of ISV have been investigated in animal models of cryptococcal meningitis. ISV to fluconazole and no antifungal treatment in rabbit models of cryptococcal meningoencephalitis and found similar efficacy of the two azoles in terms of reductions in the fungal burden in the brain and cerebrospinal fluid compared to untreated controls. In addition, no dose-dependent response was observed with ISV therapy, nor was a significant difference in efficacy between the two azoles observed in this study. Authors concluded suggesting that ISV could be as effective as fluconazole at least as consolidation and maintenance regimens in high-burden cryptococcal meningoencephalitis [bib_ref] Pharmacodynamics of isavuconazole in a rabbit model of cryptococcal meningoencephalitis, Kovanda [/bib_ref]. Furthermore, Wiederhold at al. compared the efficacy of treatment with oral ISV at two different dosages (120 mg/kg and 240 mg/kg BID) and fluconazole versus an untreated control group of murine models of murine cryptococcal meningitis. They found a significant improvement of survival and reductions in brain fungal burden in both treatment groups compared to controls; based on plasma and brain concentrations of ISV, they also observed better improvements in survival and fungal burden in mice treated with high-dose ISV compared to those treated with low-dose [bib_ref] Isavuconazole is effective for the treatment of experimental cryptococcal meningitis, Wiederhold [/bib_ref]. Guest et al. evaluated the efficacy of ISV in treating a murine model of Aspergillus fumigatus exogenous endophthalmitis. Five routes of ISV administration were performed and compared: oral gavage, intravitreal injections, intravenous injections, intravitreal injection followed by oral gavage, and intravitreal injection followed by intravenous injections. Authors observed a similar and significant improvement of the disease outcome in terms of reduction of ocular fungal burden, preservation of retinal structural integrity and function, and reduction levels of inflammatory cytokines (TNF-α, IL-1β, IL-6) and cellular infiltration in the infected eyes in all ISV administration route groups, suggesting possible beneficial effect of ISV in human ocular infections [bib_ref] Isavuconazole for Treatment of Experimental Fungal Endophthalmitis Caused by Aspergillus fumigatus, Guest [/bib_ref]. Gebremariam et al., who had previously evaluated the treatment with high dose of ISV in mice infected with Rhizopus delemar demonstrating that it was as effective as a high-dose liposomal amphotericin B treatment as reported in the review of Natesan et al. [bib_ref] Isavuconazole for the treatment of invasive aspergillosis and mucormycosis: Current evidence, safety,..., Natesan [/bib_ref] , subsequently investigated a possible beneficial effect of ISV in combination with micafungin for treatment of a murine model of pulmonary mucormycosis due to Mucor. In this study, authors did not find an enhanced survival of mice compared to placebo but no synergism or antagonism between the two antifungal drugs [bib_ref] Monotherapy or combination therapy of isavuconazole and micafungin for treating murine mucormycosis, Gebremariam [/bib_ref]. Finally, the same group assessed the efficacy of prophylaxis (treatment started on Day −2 and continued until mice infection) or continuous treatment (treatment started on Day −2 and ended on Day +2) of ISV, posaconazole and voriconazole in immunosuppressed mice with pulmonary mucormycosis due to R. delemar. In the prophylaxis study, an improvement in survival and fungal burden were observed only in the group of mice treated with ISV, whereas in the continuous therapy study, compared to the placebo group, improved survival and reduction of fungal burden occurred in ISV and posaconazole groups but not in voriconazole groups [bib_ref] Prophylaxis with isavuconazole or posaconazole protects immunosuppressed mice from pulmonary mucormycosis, Gebremariam [/bib_ref].
## Isavuconazole in hematological patients
The efficacy of ISV in hematologic patients with IFD was first documented in the SECURE trial, a phase 3, randomized, double-blind, multicenter, non-inferiority study of ISV versus voriconazole for the primary treatment of invasive mold diseases. In this trial, more than 80% of patients were affected by HMs in both arms [bib_ref] Isavuconazole versus voriconazole for primary treatment of invasive mould disease caused by..., Maertens [/bib_ref]. The study met the primary (non-inferiority of ISV versus voriconazole in the intention to treat population, ITT) and all the secondary objectives (overall response in population with proven-probable invasive mold disease, all-cause mortality at day 84, clinical/mycological/radiological response and safety tolerability). The all-cause mortality at day 42 (primary endpoint) in ITT population was 19% and 20% for ISV and voriconazole, respectively. In the ITT population with proven-probable invasive mold disease at the end of treatment (EOT), the overall response rate (ORR) was similar for ISV (35%) and voriconazole (36%), and the clinical response was 62% and 60% for ISV and voriconazole, respectively. ISV-treated patients had a lower frequency of hepatic, cutaneous and ophthalmologic adverse events. In a post hoc analysis, ISV had comparable efficacy and safety to voriconazole in neutropenic patients with invasive proven-probable aspergillosis [bib_ref] Impact of unresolved neutropenia in patients with neutropenia and invasive aspergillosis: A..., Kontoyiannis [/bib_ref]. Isavuconazole efficacy on IFD caused by rare fungi was also tested in a single-arm open-label trial (VITAL study), which enrolled patients with invasive aspergillosis and renal impairment or with rare IFD. At the EOT, the ORR in 37 (of which 59% were hematologic) patients with mucormycosis was 32%, 36% and 20% for primary treatment, for refractory disease and for intolerant to other antifungal patients, respectively, which was comparable with the response reported for liposomal amphotericin B [bib_ref] Mucormycosis-associated fungal infections in patients with haematologic malignancies, Kara [/bib_ref]. In the same trial, a post hoc analysis showed the activity of ISV on rare fungi (both non-Candida yeasts and other rare molds), which confirmed its efficacy, with an overall treatment success at the EOT of 57.7% [bib_ref] Isavuconazole for treatment of rare invasive fungal diseases, Cornely [/bib_ref]. However, ISV was only marginally active on IFD caused by more than one fungal species (13.3%) [bib_ref] Isavuconazole for treatment of invasive fungal diseases caused by more than one..., Marty [/bib_ref].
Data concerning the real-life use of ISV in HMs are scanty and quite heterogeneous; moreover, the interpretation and comparison between the studies are made difficult by the different response definitions adopted. However, real-life data seem to confirm those reported by clinical trials. Ordaya et al. reported the first small series of patients treated with ISV outside clinical trials; no data about underlying disease were given [bib_ref] Real-life use of isavuconazole in patients intolerant to other azoles, Ordaya [/bib_ref]. Eleven out of 28 patients with probable-proven IFD received ISV because of intolerance to either posaconazole or voriconazole. The ORR was 82% and the overall mortality 18%. In a retrospective cohort study conducted on 91 adult inpatients receiving ISV for both prophylaxis and treatment for possible and proven-probable IFD, including 58 (64%) acute leukemia patients, the ORR was 42/68 (62%) in the evaluable treated patients [bib_ref] Real-world use-Isavuconazole at a large academic medical center, Hassouna [/bib_ref]. It was higher in patients receiving ISV empirically (65%) than in salvage after another antifungal agent (53%) and no breakthrough IFD (b-IFD) were observed. To date, the largest and more specific real-life study on ISV as treatment for IFD in HMs was reported by the Sorveglianza Epidemiologica Infezioni nelle Emopatie (SEIFEM) Group. In this retrospective multicenter study, data about 128 patients, all affected by HMs, including acute leukemia (AL) (67%) and allogeneic stem cell transplantation recipients (alloSCT, 33%) patients, and receiving ISV as first or second-line treatment for possible and proven-probable IFD, were reported. The ORR was 67.2% in 82/122 evaluable patients, which was similar to the clinical response observed in the SECURE trial. The ORR was similar when using ISV as a 1st-or 2nd-line treatment (60.5% vs. 70.9%, respectively, p = 0.24). Among patients receiving ISV as second-line treatment, those on salvage had a lower ORR (53.8%) than patients without IFD refractoriness (70.6%, p = 0.012). The response to ISV was similar in possible and probable IFD (68.6% and 71.2%, respectively) than in proven IFD (41.7%). Both female sex and ISV use during the induction phase of treatment for hematologic disease were predictive of a favorable ISV response. ISV tolerability was excellent in all real-life studies, with a less than 5% of permanent discontinuation. [fig_ref] Table 1: Isavuconazole treatment in hematological malignancies patients with IFD [/fig_ref] summarizes the main results of ISV treatment in hematological malignancies patients with IFD.
Data concerning ISV distribution in central nervous system (CNS) are quite rare but seem to indicate that ISV is active also in this setting of infections. A retrospective analysis of ISV treatment in patients with CNS IFD and participating in VITAL and SECURE studies revealed a clinical response at the EOT in 21/36 patients (58.3%) and a survival rate of 69.4% (25/36) [bib_ref] Isavuconazole for the treatment of patients with invasive fungal diseases involving the..., Schwartz [/bib_ref]. Other case reports showed the efficacy of ISV in hematologic patients affected by CNS IFD [bib_ref] Isavuconazole Diffusion in Infected Human Brain, Rouzaud [/bib_ref] [bib_ref] Isavuconazole Therapy of Disseminated and Encephalic Saprochaete Capitata Infection in an Acute..., Perrone [/bib_ref] -Proven/probable (58%) and possible (42%) IFD -66/71 (93%) aspergillosis ORR: overall response rate (complete+partial response); § Clinical response: complete or partial resolution of all, some or none of the attributable clinical symptoms and physical findings; * evaluable patients; ** 85 received isavuconazole as treatment, 6 as prophylaxis.
ISV was also employed as antifungal prophylaxis in HMs, both affected by AL and undergoing alloSCT, with less convincing results. Cornely et al. reported the results of a phase II dose escalation study designed to assess the safety and tolerability of ISV as antifungal prophylaxis in neutropenia chemotherapy-induced in acute myeloid leukemia (AML) [bib_ref] Safety and pharmacokinetics of isavuconazole as antifungal prophylaxis in acute myeloid leukemia..., Cornely [/bib_ref]. Of 20 patients enrolled, 2 (10%) were considered as failure as they developed a possible fungal infection. A more recent open-label phase II study in AML and myelodysplastic syndrome (MDS) patients undergoing remission-induction therapy reported the development of b-IFD (2 probable-pulmonary aspergillosis-and 8 possible) in 10 (15%) of 75 enrolled patients while on ISV prophylaxis [bib_ref] Isavuconazole (ISAV) As Primary Anti-Fungal Prophylaxis in Acute Myeloid Leukemia or Myelodysplastic..., Bose [/bib_ref]. The possible explanations of these figures may be the advanced age of the population (median: 67 years) and the prolonged neutropenia, due to intensive chemotherapy or to venetoclax-based regimens. Tolerability was excellent with only marginally adverse events. ISV as antifungal prophylaxis was evaluated also in the alloSCT setting in a prospective, single-center study [bib_ref] Open-Label Trial of Isavuconazole Prophylaxis against Invasive Fungal Infection in Patients Undergoing..., Stern [/bib_ref]. In this study, ISV was administered since day +7 and the maximum duration of the study was until +98. The primary end point was prophylaxis failure, defined as discontinuation for proven-probable IFD, toxicity or adverse event, need for systemic antifungal therapy for more than 14 days. Ten (10.7%) of 99 patients discontinued ISV, but only 3 (3.1%) for IFD (all candidemia); however, 4 patients had a diagnosis of possible pulmonary IFD. Seven patients discontinued ISV for toxicity, mainly hepatic. ISV as antifungal prophylaxis has been reported also in two retrospective single-center studies. Bowen et al. reported in 98 hematologic/alloSCT patients and receiving 138 courses, 14 ISV discontinuation, 6 for toxicity, and 8 for possible [bib_ref] Isavuconazole treatment for mucormycosis: A single-arm open-label trial and case-control analysis, Marty [/bib_ref] or proven-probable (2 aspergillosis, 2 candidemia, 1 mucormycosis and 1 fusariosis) IFD [bib_ref] Isavuconazole to prevent invasive fungal infection in immunocompromised adults: Initial experience at..., Bowen [/bib_ref]. Fontana et al. conducted a retrospective review of breakthrough b-IFDs in patients affected by AL or undergoing alloSCT and receiving ISV as antifungal prophylaxis for at least 7 days [bib_ref] Isavuconazole prophylaxis in patients with hematologic malignancies and hematopoietic cell transplant recipients, Fontana [/bib_ref]. Proven-probable b-IFDs (1 candidemia, 7 aspergillosis, 2 fusariosis, 2 mucormycosis) were observed in 12/145 patients (8.3%), mainly acute leukemias undergoing chemotherapy. These figures were higher than historical control with posaconazole and voriconazole, and therefore, the authors decided to replace ISV with posaconazole. Two other studies on ISV prophylaxis in AL patients and alloSCT recipients are ongoing (NCT03149055, NCT03019939).
b-IFDs other than aspergillosis, including rare fungi, have been frequently reported during ISV treatment or prophylaxis. Rausch et al. evaluated 100 patients receiving antifungal prophylaxis and treatment with Isavuconazole [bib_ref] Breakthrough fungal infections in patients with leukemia receiving isavuconazole, Rausch [/bib_ref]. Thirteen (13%) of patients developed a b-IFD; only one due to Aspergillus spp, the remaining due to Candida spp (6, all non-albicans species), Trichosporon asahii (1), Mucorales (4), Fusarium spp (1). A case of mucormycosis and 1 of scedosporidiosis beyond aspergillosis were also reported by Fung et al. in a small series of 5 b-IFDs observed in hematological patients while on ISV for prophylaxis or treatment [bib_ref] Breakthrough Invasive Fungal Infections on Isavuconazole Prophylaxis and Treatment: What Is Happening..., Fung [/bib_ref].
Due to its safety profile, ISV has been immediately proposed as a winning solution in HMs affected by IFD. Beyond safety data concerning clinical trials and real-life studies [bib_ref] Isavuconazole versus voriconazole for primary treatment of invasive mould disease caused by..., Maertens [/bib_ref] [bib_ref] Isavuconazole treatment for mucormycosis: A single-arm open-label trial and case-control analysis, Marty [/bib_ref] , its tolerability has been demonstrated also in 23 acute leukemia patients discontinuing posaconazole for liver or cardiac toxicity and in 50 HMs/alloSCT patients treated for ≥6 months, without any discontinuation due to toxicity, as a confirm of the safety profile also in the "real-life" [bib_ref] Tolerability of isavuconazole after posaconazole toxicity in leukaemia patients, Dipippo [/bib_ref] [bib_ref] Lack of Toxicity with Long-term Isavuconazole Use in Patients with Hematologic Malignancy, Dipippo [/bib_ref]. QTc prolongation has been considered a limiting side-effect of voriconazole and posaconazole in hematologic patients, particularly in those receiving new drugs, such as FLT3-inhibitors and venetoclax. The effect of ISV on QTc interval has been specifically explored in a multicenter study in 26 patients, including also hematologic and alloSCT patients; QTc interval was reduced in all but 2 patients, who showed no changes [bib_ref] Isavuconazole shortens the QTc interval, Mellinghoff [/bib_ref]. Occasional data of electrolyte imbalances have been reported, together with gastroenteric toxicity (nausea/vomiting). ISV excellent plasma bioavailability has been demonstrated by a post hoc analysis of the SECURE trial, where the modest variability in concentrations observed was not associated with any differences in efficacy or safety outcomes [bib_ref] Variability and exposure-response relationships of isavuconazole plasma concentrations in the Phase 3..., Kaindl [/bib_ref]. Based on these observations, ISV therapeutic drug monitoring may be considered less critical than other azoles.
ISV is a moderate inhibitor of CYP3A4 cytochrome and a mild inhibitor of P-glycoprotein efflux pump. Although its drug-drug interaction profile is modest as compared to other azoles, variability in plasma concentration of concomitant drugs which are substrates of CYP3A4 or P-glycoprotein, such as immunosuppressant (i.e., cyclosporine, tacrolimus and sirolimus) or new molecular entities (i.e., ibrutinib, venetoclax and FLT3-inhibitors), may be expected. A pharmacokinetic study in healthy subjects demonstrated an increase in plasma concentration of tacrolimus, sirolimus and cyclosporine by 125%, 84% and 29%, respectively [bib_ref] Pharmacokinetic Assessment of Drug-Drug Interactions of Isavuconazole With the Immunosuppressants Cyclosporine, Mycophenolic..., Groll [/bib_ref]. This concentration increase has been demonstrated also in alloSCT recipients for tacrolimus and sirolimus within the first two week of administration [bib_ref] Effect of isavuconazole on tacrolimus and sirolimus serum concentrations in allogeneic hematopoietic..., Kieu [/bib_ref]. To date, no in vivo pharmacokinetic studies have been reported in patients receiving concomitant new molecular entities. A retrospective study on patients with IFD and treated with ISV while on ibrutinib has been recently reported [bib_ref] Isavuconazole for the treatment of invasive fungal disease in patients receiving ibrutinib, Cummins [/bib_ref]. In this small series of 8 patients, ISV has been proved effective in 7 cases; ibrutinib was prudentially reduced in 5 cases. One patient discontinued ibrutinib for worsening of thrombocytopenia, which was considered potentially related to an increased ibrutinib exposure; both patients with progressive disease received a reduced dose of ibrutinib. These data indicate that further studies are warranted to elucidate pharmacokinetic implications of ISV and new small molecules coadministration.
## Isavuconazole in non-hematological patients
Clinical trials evaluating the efficacy and safety of ISV in treatment of invasive mold diseases included predominantly patients suffering from HMs, and data on patients with underlying diseases other than HMs are not specifically reported [bib_ref] Isavuconazole versus voriconazole for primary treatment of invasive mould disease caused by..., Maertens [/bib_ref]. In the ACTIVE trial, a Phase 3, randomized, double-blind, non-inferiority trial comparing the efficacy and safety of intravenous (IV) ISV followed by oral ISV to IV caspofungin followed by oral voriconazole in the primary treatment of candidemia and invasive candidiasis, a total of 440 patients (400 in modified intention to treat population) were included, 221 in ISV group and 219 in caspofungin group [bib_ref] Isavuconazole Versus Caspofungin in the Treatment of Candidemia and Other Invasive Candida..., Kullberg [/bib_ref]. Although patients' underlying diseases were not specifically reported, neutropenia was present in only 25/221 (11.3%) and 24/219 (11%) patients in ISV and caspofungin group, respectively; therefore, most of the patients probably did not suffer from HMs. At primary efficacy end-point (successful overall response at the end of IV therapy) evaluation, ISV failed to demonstrate non-inferiority to caspofungin for treatment on invasive candidiasis: 60.3% of patients were successfully treated in the ISV arm and 71.1% in the caspofungin arm (adjusted difference −10.8, 95% confidence interval −19.9-−1.8). Several clinical cases have been reported regarding the use of ISV for treatment of IFD in patients suffering from clinical conditions other than HMs. After exclusion of reports published in non-English language or in which clinical and therapeutic data were not available for the purpose of the present review, a total of 41 cases have been included. Clinical and demographic characteristics, details of the antifungal treatment, outcome and adverse effects reported on these patients are summarized in [fig_ref] Table 2: Isavuconazole treatment in non-hematological patients with invasive fungal infections [/fig_ref]. Thirty-three out of 42 (78.5%) patients were male and the mean age was 52 years (range 20-79). The most frequent and most important comorbidity was solid organ transplantation in a total of 6 patients (liver, n = 1; lung, n = 2; kidney, n = 2; heart, n = 1), followed by sarcoidosis (n = 3) and Acquired Immune Deficiency Syndrome (AIDS, n = 3) and Influenza A (n = 2). Type-2 Diabetes Mellitus was present in a total of 10 patients. In addition, the IFD was diagnosed in the context of active tuberculosis in one case and acute respiratory distress syndrome (ARDS) due to SARS-CoV-2 in another patient. The more frequent IFD treated with ISV were predominantly mucormycosis (n = 12), aspergillosis (n = 13), coccidioidal meningitis (n = 9) and invasive infections due to Histoplasma capsulatum (n = 3). In 7 out of 42 patients, ISV was administered as first-line therapy (often because of the presence of contraindications to other antifungals); in the remaining cases, ISV treatment was started after other first-line therapy, as a consequence of evidence of adverse effects and or clinical or microbiological failure. In some cases, ISV was started as prolonged or lifelong maintenance oral therapy. Clinical cure or stable improvement have been reported in 32 out 42 cases, whereas death or clinical and/or microbiological failure occurred in 5 cases; in two cases, treatment was discontinued due to severe adverse effects and in 3 cases the outcome was not reported. ISV treatment was generally well tolerated; in 4 cases, authors specifically stated that patients did not present adverse effects, whereas in 30 cases they were not reported. Adverse events were reported in 9/42 cases (21.4%). In six cases, gastrointestinal symptoms (n = 4) or hepatic injury (n = 2) occurred. ISV treatment was discontinued for adverse events only in 3/42 (7.1%) cases, due to nausea, vomiting, myalgia and lethargy, severe hypomagnesemia in one case, and severe liver injury and alopecia in the other two cases, respectively.
ISV has been evaluated as antifungal prophylaxis in patients diagnosed with clinical conditions other than HMs. Samanta et al. conducted a single-center, retrospective study comparing efficacy and safety of ISV (144 patients) versus voriconazole (156 patients) as antifungal prophylaxis in lung transplant recipients; of note, adjunctive inhaled amphotericin B was also administered to 100% and 41% of patients in ISV and voriconazole groups, respectively [bib_ref] Isavuconazole is as effective as and better tolerated than voriconazole for antifungal..., Samanta [/bib_ref]. At 1-year follow-up, no difference in IFD occurrence was reported between the two groups (7% vs. 8%). Authors also identified red blood cell transfusion >7 units at transplant mold-positive respiratory culture as independent risk factors for both breakthrough IFD and breakthrough invasive mold infections, whereas the African-American race was independently associated specifically with breakthrough IFD and bronchial necrosis >2 cm from anastomosis and basiliximab induction were independent risk factors only for invasive mold infections. ISV was tolerated significantly better than voriconazole (premature discontinuation rates due to adverse events: 11% vs. 36% of patients, respectively). Finally, antifungal prophylaxis prolonged for ≥90 days was associated with a significantly lower rate of IFD at 1-year follow-up (3% vs. 9%). ISV for prophylaxis in a patient undergoing lung transplantation was used also in another patient who had developed QTc prolongation while receiving voriconazole; ISV was well tolerated with normalization of QTc level and no subsequent occurrence of IFD was reported [bib_ref] Use of isavuconazole in a patient with voriconazole-induced QTc prolongation, Trang [/bib_ref]. Finally, ISV was used as occupational post-exposure prophylaxis after a deep cut with a scalpel used for processing a clinical sample contaminated by Rhizopus spp.; prophylaxis was discontinued after two weeks because of side effects (severe nausea and diarrhea), and no subsequent local fungal infection was reported [bib_ref] Post-exposure prophylaxis with isavuconazole after occupational exposure to Rhizopus, Al-Obaidi [/bib_ref].
# Discussion
## Proper summary
Registrative trials demonstrated that ISV has a similar efficacy to voriconazole for the treatment of invasive aspergillosis and to liposomal amphotericin B for the treatment of mucormycosis [bib_ref] Isavuconazole versus voriconazole for primary treatment of invasive mould disease caused by..., Maertens [/bib_ref] [bib_ref] Isavuconazole for treatment of invasive fungal diseases caused by more than one..., Marty [/bib_ref]. Efficacy against rare fungi has also been reported [bib_ref] Isavuconazole for treatment of rare invasive fungal diseases, Cornely [/bib_ref]. On the contrary, ISV did not prove its efficacy on candidemia and invasive candidiasis as it failed to demonstrate the non-inferiority to caspofungin [bib_ref] Successful isavuconazole salvage therapy in a patient with invasive mucormycosis, Ervens [/bib_ref].
The role of ISV in target therapy is well defined both in HMs and non-HMs patients, and data from real life confirm the efficacy of this antifungal agent. Animal models and sporadic reports in humans seem to indicate its efficacy also in CNS infections.
ISV has been reported better tolerated than voriconazole in the SECURE study [bib_ref] Isavuconazole versus voriconazole for primary treatment of invasive mould disease caused by..., Maertens [/bib_ref] , and real life data confirm this safety profile [bib_ref] Real-world use-Isavuconazole at a large academic medical center, Hassouna [/bib_ref]. Considering the lower side effect incidence and reduced drug-drug interactions, a possible application in prophylaxis, use not licensed at present, has been evaluated, although b-IFDs, including rare fungi, have been frequently reported and this phenomenon needs close surveillance.
# Conclusions
Isavuconazole, in conclusion, has proven to be effective as a treatment both against aspergillosis and against other molds, with the advantage of better handling respect of the other azoles. Accordingly, international guidelines indicated ISV as a useful alternative to voriconazole (i.e., the current gold standard) and the other available agents in the treatment of invasive aspergillosis mainly in patients with HMs [bib_ref] Diagnosis and management of Aspergillus diseases: Executive summary of the 2017 ESCMID-ECMM-ERS..., Ullmann [/bib_ref]. In addition, the recent Global Guidelines for Mucormycosis indicated ISV, with liposomal amphotericin B, for the treatment of mucormycosis [bib_ref] Global guideline for the diagnosis and management of mucormycosis: An initiative of..., Cornely [/bib_ref].
## Future perspectives
Although non-registration clinical studies do not attest convincing results in prophylaxis at present, further studies are warranted to explore this type of approach, which would be very attractive in patients receiving as concomitant treatment drugs metabolized via CYP450 and P-glycoprotein efflux pump pathway, given the modest inhibition of the CYP450 system by ISV. Moreover, the efficacy of ISV in treatment of CNS IFDs seems very promising and should be verified, as the choice of ISV in this setting may also offer the possibility of very prolonged treatment without significant toxicities. Another possible future application of ISV could be in combination with other antifungal drugs for the treatment of very aggressive and refractory IFDs or for mixed IFDs; however, at present, there is no evidence that this approach could be a benefit for the treatment of this peculiar subset of IFDs.
Author Contributions: All authors participated to the conceptualization and preparation of the paper. All authors have read and agreed to the published version of the manuscript.
Funding: This work did not receive any fund.
Conflicts of Interest: L.P. was a Board member of Gilead Sciences, MSD, Pfizer, Janssen, Novartis, Cidara and has been a speaker for Gilead Sciences, MSD, Pfizer Pharmaceuticals, Astellas. All other authors declared nothing to disclose.
[table] Table 1: Isavuconazole treatment in hematological malignancies patients with IFD. [/table]
[table] Table 2: Isavuconazole treatment in non-hematological patients with invasive fungal infections. [/table]
|
10.1186/s12864-021-07682-3
|
CCBY
|
8157689
|
34039278
|
s2orc_pubmed_articles
|
Population transcriptomic sequencing reveals allopatric divergence and local adaptation in Pseudotaxus chienii (Taxaceae)
Background: Elucidating the effects of geography and selection on genetic variation is critical for understanding the relative importance of adaptation in driving differentiation and identifying the environmental factors underlying its occurrence. Adaptive genetic variation is common in tree species, especially widely distributed long-lived species. Pseudotaxus chienii can occupy diverse habitats with environmental heterogeneity and thus provides an ideal material for investigating the process of population adaptive evolution. Here, we characterize genetic and expression variation patterns and investigate adaptive genetic variation in P. chienii populations. Results: We generated population transcriptome data and identified 13,545 single nucleotide polymorphisms (SNPs) in 5037 unigenes across 108 individuals from 10 populations. We observed lower nucleotide diversity (π = 0.000701) among the 10 populations than observed in other gymnosperms. Significant negative correlations between expression diversity and nucleotide diversity in eight populations suggest that when the species adapts to the surrounding environment, gene expression and nucleotide diversity have a reciprocal relationship. Genetic structure analyses indicated that each distribution region contains a distinct genetic group, with high genetic differentiation among them due to geographical isolation and local adaptation. We used F ST outlier, redundancy analysis, and latent factor mixed model methods to detect molecular signatures of local adaptation. We identified 244 associations between 164 outlier SNPs and 17 environmental variables. The mean temperature of the coldest quarter, soil Fe and Cu contents, precipitation of the driest month, and altitude were identified as the most important determinants of adaptive genetic variation. Most candidate unigenes with outlier signatures were related to abiotic and biotic stress responses, and the monoterpenoid biosynthesis and ubiquitin-mediated proteolysis KEGG pathways were significantly enriched in certain populations and deserve further attention in other long-lived trees.
Conclusions: Despite the strong population structure in P. chienii, genomic data revealed signatures of divergent selection associated with environmental variables. Our research provides SNPs, candidate unigenes, and biological pathways related to environmental variables to facilitate elucidation of the genetic variation in P. chienii in relation to environmental adaptation. Our study provides a promising tool for population genomic analyses and insights into the molecular basis of local adaptation.
Keywords: Pseudotaxus chienii, Population transcriptome, SNP, Population structure, Genotype-environment association, Local adaptation Background Dissecting the distribution of genetic variation across landscapes helps us to understand the ecological and evolutionary processes under climate change. The influence of natural selection on genetic variation and expression variation in natural populations has received increasing attention in studies on adaptive evolution and molecular ecology. As species are forced to cope with environmental changes, it becomes increasingly important to understand how populations quickly adapt to diverse environments. Long-lived trees with a wide range of natural habitats often show clear adaptation to local environments. Evidence for local adaptation can be detected if there is significant association with the environmental variables at some loci. Individuals growing in different geographical areas will be subject to different selection pressures and therefore adapt to different local environmental conditions. Genetic divergence may be caused by selection imposed by environmental pressures or the influence of genetic drift and limited gene flow when populations are partially isolated. High levels of gene flow and continuous migration have homogenization effects, but natural selection is inferred to drive genetic divergence. Describing spatial isolation and natural selection is essential for disentangling the processes that initiate genetic divergence, including the relative role of adaptation in driving differentiation and the number and identity of its potentially associated genetic targets.
With the development of sequencing technology, nextgeneration sequencing (NGS) has made it possible to obtain genome-wide scale sequence information across populations, greatly promoting the investigation of adaptive evolution and molecular ecology in nonmodel species. Previous studies using anonymous markers (i.e., simple sequence repeat (SSR) and amplified fragment length polymorphism (AFLP)) were unable to assess the degree of linkage and the independence of loci, making them less reliable than other studies. RNA sequencing (RNA-Seq) based on NGS can provide a more accurate estimate of the number of independent loci involved in adaptation and be used to detect potential candidate genes. RNA-Seq can be used to perform gene expression studies in species without genomic sequence information; thus, it is a very promising application in research on adaptation. Expression variation may occur before genetic variation and may be heritable; therefore, expression differences may reflect the early process of adaptive divergence at the population level. In addition to identifying gene expression variations, RNA-Seq data can also allow the development of single-nucleotide polymorphisms (SNPs) on a large scale, which can capture potential sequence variations. These sequence variations and expression variations may be involved in the adaptation of a species to its natural habitat.
Transcriptome sequencing is a powerful tool that represents a cost-effective approach for examining genetic and expression patterns and investigating adaptive divergence at the levels of sequences, genes or biological metabolic pathways among natural populations in nonmodel organisms. For example,sequenced the transcriptomes of 78 Miscanthus lutarioriparius individuals from 10 populations and found genes related to photosynthetic processes and responses to environmental stimuli such as temperature and reactive oxygen species.compared the transcriptomes of Pinus yunnanensis from high-and low-elevation sites and identified 103,608 high-quality SNPs and 321 outlier SNPs based on RNA-Seq to investigate adaptive genetic variation. The 321 outlier SNPs from 131 genes displayed significant divergence in terms of allelic frequency between high-and low-elevation populations and indicated that the flavonoid biosynthesis pathway may play a crucial role in the adaptation of P. yunnanensis to high-elevation environments. These studies provide insights into the patterns of genetic variation and gene expression in natural populations and aid in the exploration of loci involved in adaptation to diverse habitats.
The white-berry yew, Pseudotaxus chienii (W. C. Cheng) W. C. Cheng, is a threatened tertiary relict monotypic gymnosperm in the genus Pseudotaxus (Taxaceae). This species is a dioecious evergreen shrub or tree that grows in the subtropical mountains of China. The distribution of P. chienii covers a relatively large geographical area with abundant environmental variation, in which includes mountain forests of northwestern Hunan, central Guangxi, southwestern Jiangxi, and southern Zhejiang. Significant environmental heterogeneity has been found among most populations of P. chienii. The wide range of natural habitats of P. chienii demonstrates its adaptability to various soils and growth conditions. Populations of P. chienii primarily grow in shallow and acidic soil, in rock crevices or on cliffs. P. chienii can adapt well to diverse habitats with environmental heterogeneityand thus provides an ideal material for investigating the process of population adaptive evolution. Morphological surveys of P. chienii in different geographical areas with different climatic conditions demonstrated that the width of the leaves gradually increases geographically from east to west, providing evidence for local adaptation of the plant phenotype. In plants, a large part of the phenotypic variation can be attributed to divergent selection imposed by environmental variables. Nevertheless, the main environmental variables that drive selection between natural populations are still unknown in most plants. The currently available data cannot provide a comprehensive understanding of the genetic status and adaptive divergence of P. chienii populations, and population genomic data from natural populations of this species are needed to solve these problems.
Adaptive genetic variation is common in tree species, especially widely distributed long-lived species. Candidate loci/genes related to adaptive changes in different environments are increasingly included in investigations of adaptive divergence in trees. In this study, we applied population transcriptome data to detect the genetic basis of local adaptation in P. chienii and determine which environmental variables are essential in driving population genetic differentiation. We detected 13,545 SNPs in 5037 unigenes across 10 populations using RNA-Seq. Population genetics and gene expression variation were explored. We integrated environmental and geographic information and used genetic loci to evaluate the impacts of environmental factors and geographic factors on genetic variation. The outlier SNPs associated with environmental variables and the candidate unigenes that contribute to local adaptation in P. chienii were also identified. The results of our study are expected to improve insights into evolutionary processes and local adaptation in P. chienii.
# Results
## De novo assembly and snp calling
For 108 individuals, we obtained a total of 6336.45 Mbp raw reads with an average of 58.67 Mbp (Additional file 1). After the filtering process, 6258.14 Mbp clean reads representing 938.69 G bases were retained, with an average Q20 of 98.09%. Based on clean reads, 600,273 unigenes with a total of 426.75 Mbp nucleotide bases were assembled de novo. The mean N50 length and the mean length were 891 bp and 711 bp, respectively (Additional file 2). Of these unigenes, 230,731 (38.44%) were 301-500 bp, 172,167 (28.68%) were 501-1000 bp, 77,275 (12.87%) were 1-2 kb and 28,612 (4.77%) were more than 2 kb (Additional file 3). The final 600,273 unigenes from the 108 individuals were used as the reference sequences for P. chienii.
The clean reads of each individual were mapped to the reference sequences, and the mapping rates ranged from 66.48% in LMD_10 to 74.15% in DXG_7 (Additional file 4), indicating ideal mapping. We successfully identified 1,430,611 and 828,372 raw SNPs using GATK and SAMtools, respectively. After filtering steps, SNPs were retained using GATK and SAMtools, respectively. To obtain high-quality SNPs, only SNPs identified by both SAMtools and GATK were retained. Overall, 13,545 SNPs from 5037 unigenes were identified across the 108 individuals from 10 populations.
## Genetic variation and population genetic structure
At the species level, the nucleotide diversity (π) of P. chienii was 0.000701. At the population level, LMD had the lowest π (0.000512), whereas LHS had the highest π (0.000723). The observed heterozygosity (H O ) and expected heterozygosity (H E ) of the 10 populations ranged from 0.383 (ZZB) to 0.493 (ZJJ) and from 0.356 (YSGY) to 0.422 (ZJJ), respectively. Wright's inbreeding coefficient (F IS ) values were positive in all 10 populations. Regarding population differentiation, the F ST value was highest between ZJJ and BJS (0.380), while MS and LMD had the lowest F ST value (0.078) (Additional file 5). Moreover, the pairwise F ST values of ZZB vs. BJS and LMD vs. SMJ were negative, implying that gene flow between these populations was common. We further tested the pairwise F ST values among the four groups (see Methods section). The pairwise F ST values among the four groups ranged from 0.216 (ZJ vs. JX) to 0.361 (HN vs. JX), suggesting that HN and JX had the greatest genetic distance (Additional file 6).
Principal component analysis (PCA) unambiguously revealed four distinct genetic clusters. The first two principal components (PCs), which explained 12.97 and 11.57% of the total genetic variance, respectively, differentiated the four geographically distinct P. chienii groups: Zhejiang (ZJ: SQS, DXG, LMD, MS, and SMJ populations), Jiangxi (JX: BJS and ZZB populations), Guangxi (GX: LHS and YSGY populations), and Hunan (HN: ZJJ population) . These four groups corresponded almost entirely to separate geographic regions. To further explore the population genetic structure of P. chienii, genetic clustering of the 108 individuals was performed using ADMIXTURE, which also indicated that four genetic clusters (K = 4) was optimal with the lowest cross-validation error. With K = 4, individuals of the JX (BJS and ZZB populations), ZJ (LMD, MS, SMJ, and SQS populations), and GX (YSGY and LHS populations) groups clustered into three clusters, and the DXG population of the ZJ group was assigned to an independent cluster. The HN (ZJJ population) group contained a mixture of genetic components of the ZJ, JX and GX clusters. Although K = 4 was the optimal K value, several other K values also showed biologically relevant patterns. When K = 3, DXG was clustered into the ZJ cluster, which was consistent with the geographical distribution of P. chienii and the PCA results.
A phylogeny based on 13,545 genome-wide SNPs showed three lineages, corresponding to ZJ, GX + HN, and JX . The JX lineage was at the most basal position, followed by GX + HN and then ZJ. Although the ADMIXTURE analyses showed that the HN group contained a mixture of genetic components of ZJ, JX and GX, phylogenetic analysis further confirmed that HN was closer to GX than JX or ZJ.
Analysis of molecular variance (AMOVA) of 13,545 SNPs revealed that 74.59% of the overall variation (df = 206; p < 0.0001) was distributed within populations and 25.41% among populations (df = 9; p < 0.0001). AMOVA found significant genetic differentiation among populations (F ST = 0.254; p < 0.0001). The Mantel test detected a statistically significant correlation between pairwise F ST and geographic distance among the 10 populations (r = 0.688, p = 0.001), indicating a significant pattern of isolation by distance (IBD). We also identified a significant pattern of isolation by environment (IBE) (r = 0.602, p = 0.001), and the level of correlation was similar to that of IBD.
## Population gene expression variation
The population gene expression level (E p ) and expression diversity (E d ) were analyzed based on 108 P. chienii individuals from 10 populations. The distribution of E p for 16,225 unigenes was right-skewed and peaked at expression level intervals of 0-10 (Additional file 7a). The quantiles of log 2 E p in each population were similar . The average E p values of the 10 populations ranged from 2.244 (SMJ) to 2.634 (ZJJ). E d also showed a right-skewed distribution with a peak at 0.2-1.3 (Additional file 7b). The quantiles of E d shifted down in LMD and SMJ . The average E d values of the 10 populations ranged from 0.663 (MS) to 0.800 (LMD).
We further analyzed the relationship between E d and π in each population. At the unigene level, the relationship between E d and π in each population except BJS and MS showed a significant negative correlation (r = − 0.075 -− 0.032; p = 6.80 × 10 − 7 -0.031; Additional file 8). However, at the population level, there was no significant difference between the average E p and π among the 10 populations (r = 0.39; p = 0.26; Additional file 9). Expression similarity (E p similarity) was also not significantly correlated with genetic distance (r = − 0.07; p = 0.38; Additional file 10).
## Directional migration rates
The bidirectional relative migration rates (m R ) among the 10 populations/four groups were similar across three measures (Jost's D, G ST , and Nm) of genetic differentiation; therefore, we describe the result based only on the Nm . Among the 10 populations, high relative migration rates were observed in both directions between BJS and ZZB (m R > 0.90) and from LMD to SMJ (m R = 0.77). The relative migration rates between LHS and YSGY (m R = 0.17 for LHS to YSGY; m R = 0.11 for YSGY to LHS) were lower than the migration rates between most populations in the ZJ group (SQS, DXG, LMD, MS, and SMJ) . Among the four groups, the highest relative migration rates (m R = 1 for JX to ZJ) were observed. High relative migration rates were also observed from GX to ZJ (m R = 0.78), from HN to ZJ (m R = 0.69), and from ZJ to JX (m R = 0.62) .
Additionally, the relative migration rates between HN and ZJ were higher than those between HN and GX, despite the closer geographic proximity of HN and GX.
## Ecological niche differences among populations of p. chienii
Ecological niche modelings were constructed for the four groups of P. chienii to predict their current, past and future potential distributions. All Maxent models for the four P. chienii groups had high predictive performance, with area under the receiver operating characteristic curve Under the current climate, the predicted distribution of P. chienii is basically consistent with the actual distribution of each group, although there are a few predicted areas where the species is not found, such as Taiwan. Under the interglacial (LIG) climate, JX, GX, and HN showed considerable contraction in suitable habitats, while clear range expansions were observed for the ZJ group. For the last glacial maximum (LGM) model, clear expansions in suitable habitats were predicted for all groups. The future distribution models showed a loss of suitable habitats for ZJ and JX relative to the current distribution, while the predicted current and future distributions were nearly identical for GX and HN (Additional file 12).
## Identification of outlier snps and unigene annotation
We identified 979 outlier SNPs using BayeScan software with a 0.001 q-value thresholdand 402 were annotated in the Pfam and SwissProt protein databases, respectively. Gene ontology (GO) terms were used to functionally classify the 642 unigenes, which were classified into three main categories: 337 unigenes in "biological process", 381 unigenes in "molecular function", and 216 unigenes in "cellular component" (Additional file 13). The top 15 GO terms of the three main categories identified for these unigenes are shown in Additional file 14. The GO enrichment analysis of 642 unigenes showed that "translation regulator activity" (GO:0045182) and "protein binding" (GO: 0005515) were significantly enriched (q-values < 0.05) (Additional file 15). Based on niche overlap analysis, the ecological differentiations of GX vs. ZJ and HN vs. ZJ were valid. Therefore, we further used selective sweep analysis to identify the unigenes underlying divergent adaptation in the ZJ, GX, and HN groups. Based on the top 5% of F ST values and π ratio cutoffs (F ST > 0.64 and 0.65 and log 2 (π ratio) > 1.85 and 1.70 for GX vs. ZJ and HN vs. ZJ, respectively;, b), we identified 54 and 43 candidate unigenes involved in habitat adaptation in the ZJ group. These two unigene datasets contained 10 duplicated unigenes. Among the 87 candidate unigenes for habitat adaptation in the ZJ group, 56, 57 and 57 unigenes were annotated in the SwissProt, Pfam, and GO databases, respectively (Additional file 16). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analysis of these 87 candidate unigenes revealed one significantly overrepresented KEGG pathway with a q-value < 0.05: "monoterpenoid biosynthesis" (ko00902) (Additional file 17). Based on the top 5% of F ST values and π ratio, we identified three candidate unigenes involved in habitat adaptation in the HN group. The three candidate unigenes encode some proteins, including an AT-rich interactive domain-containing protein 2, an anaphase-promoting complex subunit 13, and the ETS transcription factor family, which is important for habitat adaptation in the HN group (Additional file 18). One significantly overrepresented KEGG pathway, "ubiquitin-mediated proteolysis" (ko04120), was identified (q-values < 0.05) (Additional file. Based on the top 5% of F ST values and π ratio cutoffs (F ST > 0.64 and log 2 (π ratio) > 2.61 for ZJ vs. GX;, we identified 17 candidate unigenes involved in habitat adaptation in the GX group. Among the 17 candidate unigenes, 10, 9 and 9 unigenes were annotated in the SwissProt, Pfam, and GO databases, respectively (Additional file 20). The red points (corresponding to the top 5% of the log 2 (π ratio) distribution and the top 5% of the F ST distribution) are genomic regions under selection in P. chienii. a Distribution of log 2 (π ratio) and F ST values calculated between the Guangxi group (GX) and Zhejiang group (ZJ). b Distribution of log 2 (π ratio) and F ST values calculated between the Hunan group (HN) and the ZJ group. c Distribution of log 2 (π ratio) and F ST values calculated between the ZJ group and HN group. d Distribution of log 2 (π ratio) and F ST values calculated between the ZJ group and GX group
## Association of genomic variation with environmental variables
We utilized the outlier test, redundancy analysis (RDA), and latent factor mixed models (LFMMs) to detect signatures of local adaptation among P. chienii populations and identify unigenes under selection. Forward selection of the environmental variables revealed two sets of eight environmental variables as significantly predictive of genetic variation for all loci and outlier loci (Additional file 21 and. The mean temperature of the coldest quarter, aspect, soil Fe content, precipitation of the driest month, and leaf area index were identified as the most important determinants of genetic variation for all loci, while the mean temperature of the coldest quarter, soil Fe content, soil Cu content, precipitation of the driest month, and altitude were the strongest determinants for outlier loci. The RDA axes were ordered by the amount of variance explained. Eight RDA axes (RDA1 to RDA8) explained 31.51% of the total genetic variance for all loci. The amount of explained variance increased to 64.06% when using only outlier loci as response variables. The permutation tests of the RDA models revealed p-values lower than 0.001 in these two analyses, thus confirming the high significance of the constrained variable effect.
Using all loci and outlier loci, we also carried out variation partitioning analysis to determine the relative contributions of environmental factors and geographic factors to the genetic variation. The models including all parameters ([a + b + c] inshowed a significant effect of these two factors (adjusted R 2 = 0.6484, p = 0.001 for outlier loci; adjusted R 2 = 0.3210, p = 0.001 for all loci). Environmental factors alone [a] (F = 4.0786, adjusted R 2 = 0.0820, p = 0.001) and geographic factors alone [c] (F = 1.8585, adjusted R 2 = 0.0059, p = 0.001) explained 8 and 1% of the variation at all loci, respectively; however, they explained 23% of the genetic variation when considered jointly [b] (adjusted R 2 = 0.2331). Using outlier loci, pure environmental factors [a] explained 11% of the genetic variation (F = 11.815, adjusted R 2 = 0.1130, p = 0.001), and pure geographic factors [c] explained 1% of the genetic variation (F = 3.1993, adjusted R 2 = 0.0078, p = 0.001). Environmental factors and geographic factors together explained 53% of the genetic variation (adjusted R 2 = 0.5276). In summary, the population divergence of P. chienii was strongly shaped by the joint effect of environmental factors and geographic factors, and environmental factors were more important than geography.
To detect candidate outlier loci for local adaptation, we performed LFMM analyses that tested the correlations of single-locus-single-variable. We identified 244 associations between 164 outlier SNPs and 17 environmental variables (Additional file 22). Among the associations, 5 were related to temperature, 43 to precipitation, 65 to ecological factors, 43 to topographic variables, and 88 to soil variables. Only precipitation seasonality (CV) was not found to be associated with any outlier SNP. Of the other environmental variables, the fraction of absorbed photosynthetically active radiation was (29), soil Zn content (25), percent tree cover (21), aspect, and precipitation of the driest month. These 164 outlier SNPs associated with environmental variables resided in 127 unigenes, of which 84, 93, and 35 were annotated in the SwissProt, Pfam, and KEGG databases, respectively (Additional file 23). Ninety-three unigenes were assigned to three main GO categories, including 71 unigenes in "biological process", 83 unigenes in "molecular function", and 46 unigenes in "cellular component". For the biological process category, unigenes involved in "oxidation-reduction process" (GO: 0055114), "transport" (GO:0006810), "ribosome biogenesis" (GO:0042254) and "signal transduction" (GO: 0007165) were highly represented. In the molecular function category, the major GO terms were "protein binding" (GO:0005515), "ATP binding" (GO:0005524), and "zinc ion binding" (GO:0008270). The major terms for cellular component were "membrane" (GO:0016020), "integral component of membrane" (GO:0016021), and "nucleus" (GO:0005634).
# Discussion
The rapid development of sequencing technologies has provided powerful tools with which to investigate the genetic mechanisms in natural populations and new insights into the evolutionary and ecological processes underlying genetic differentiation and species adaptations. We used RNA-Seq, a commonly used NGS approach, to quantify the expression level of each unigene in each individual by mapping clean reads to reference sequences. SNP markers, third-generation molecular markers, have wide applicability, particularly SNP markers from transcriptome sequences, which can effectively reveal functional SNPs at the whole-genome level. In this study, 13,545 high-quality SNPs were identified in 108 individuals of P. chienii based on transcriptome data to explore the driving mechanism of this species' adaptations to its natural habitat. Our results show that very information in natural populations can be obtained from SNP markers of the P. chienii transcriptome.
## Population divergence and structure
Based on the transcriptome data from the populations, P. chienii had a lower nucleotide diversity (π = 0.000701) than other gymnosperms, such as Cupressus chengiana (π = 0.0077), Cupressus duclouxiana (π =0.0031), and Cupressus gigantea (π = 0.0029), suggesting that the sequences of protein-coding genes and functional elements of P. chienii captured by this method are highly conserved.analyzed the genetic diversity of P. chienii populations using cpDNA sequences and nuclear loci and found similarly low nucleotide diversity (π = 0.0009) for cpDNA and more abundant diversity (π = 0.00265) for nuclear loci compared with those detected in this study. The difference may be due to the number of loci used in each analysis. Our estimates were based on 13,545 loci, whereas previous studies used only 14 nuclear loci. Therefore, we believe that the estimates of this study are more accurate than previous estimates. SNPs from RNA-Seq data are much more abundant DNA markers than other markers in plant genomes and have higher reproducibility, higher genotyping efficiency, and easier automation.. Furthermore, the ZJJ and LMD populations appeared to harbor relatively high genetic variation compared with that in other populations, suggesting that the ZJJ and LMD populations have undergone adaptation in response to natural selection. After the genetic admixture event (see below), the mixed population (ZJJ population) may experience its own unique genetic variation, resulting in higher genetic diversity. This phenomenon is common in long-lived gymnosperms, such as C. chengiana. Positive F IS values were observed in all sampled populations, suggesting significant heterozygote deficits. Heterozygote deficits are probably caused by several factors, including inbreeding, linkage disequilibrium, null alleles, recent admixture, and partial clonality. In our study, inbreeding and/or recent admixture seemed to be the most likely driver of the positive F IS values, as investigations of Taxus yunnanensis (F IS = 0.228) and Taxus wallichiana (F IS = 0.290) showed prevalent inbreeding. The genetic differentiation levels among P. chienii populations/groups were generally high (F ST > 0.15), except for a few pairs of populations in geographical proximity. The spatial context of natural selection and the balance between the strength of divergent selection and migration rates between populations/groups are of great significance to genetic differentiation. The high genetic differentiation levels may occur because most P. chienii grows in the understory; therefore, the biological characteristics of wind pollination may not help P. chienii spread pollen over a long distance. Additionally, the seed cone of P. chienii possesses a cup-like, fleshy, white ariland mainly depends on biotic dispersal. Seeds with white arils are less attractive to birds and mammals than brightly colored seeds, such as those of Taxus species, which limits the spread of P. chienii seeds. From ecological and evolutionary perspectives, the species has been exposed to stressors or diverse environments, which has resulted in genetic differentiation and genetic heterogeneity. A few populations in geographical proximity showed relatively low genetic differentiation, possibly due to high migration rates and a high rate of effective pollen and seed dispersal. In this study, geographic distance and environmental distance were correlated with genetic differentiation (IBD: r = 0.688, p = 0.001; IBE: r = 0.602, p = 0.001), suggesting that IBD and IBE were important to the divergence among P. chienii populations. These results further revealed that the genetic differentiation among the P. chienii populations was mainly the result of geographical isolation and habitat heterogeneity.
Generally, a higher level of genetic differentiation indicates a stronger population structure. Genetic differentiation among populations with different geographic distributions was found in the ADMIXTURE and PCA analyses in our study. The ADMIXTURE results suggested the presence of three major genetic clusters and a smaller cluster (DXG) across the species' natural range, whereas three clusters were identified for two chloroplast regions and 14 nuclear loci. Our results show that SNP markers based on transcriptome data are better able to detect fine-scale population structure than classic genetic markers. Here, ADMIXTURE analysis revealed that the ZJJ population was genetically admixed to the GX, ZJ and JX groups. Intuitively, this isolated population ZJJ had little chance of leaving the footprint of admixture introduced by other groups. However, considering that the HN group had experienced multiple expansions and contractions during the Quaternary climate oscillations, it was plausible that genetic admixture was established through gene flow between populations. Then, the specific habitat requirements of this group caused it to persist only in montane regions, and other low-altitude populations might extirpate due to local maladaptation, creating the geographically isolated population ZJJ. Overall, the transcriptome data based on high-throughput sequencing used here provide abundant markers that can contribute to the accurate description of admixture signals.
## Population gene expression variation
Gene expression variation among populations may be due to developmental, environmental, genetic or other biological effects, which are essential for adaptive evolution. Understanding the patterns of genetic variation and gene expression in populations from different habitats can reveal the response of plants to different environments through variations in gene expression and/or allelic characteristics. In our data, we found significant negative correlations between expression diversity and nucleotide diversity in eight populations. This result suggests that when the species adapts to the surrounding environment, gene expression and nucleotide diversity have a reciprocal relationship. This phenomenon might be attributed to gene duplication events occurring in the P. chienii genome during the evolutionary process. Compared with single-copy genes, duplicated genes usually significantly increase the diversity of gene expression, while genetic diversity remains relatively weak. Higher gene expression diversity may have balanced the effects of lower genetic variation, thereby maintaining the stability of the phenotype under long-term natural selection in native habitats. We speculate that genetic variation and expression diversity both played a potential role in local adaptation.
## Evidence for local adaptation
Local adaptation to environmental variables is generally believed to play a major role in the diversification of species, but its contribution relative to those of other evolutionary forces is rarely quantified. Despite the strong population structure in P. chienii, analysis of genomic data revealed signatures of divergent selection associated with environmental variables. The F ST outlier approach, RDA, the LFMM method, and selective sweep analysis were used to detect signatures of local adaptation among P. chienii populations and identify unigenes under selection. Environmental association studies have been more robust in identifying loci under selection and can also provide context for selection forces. Testing for IBD and IBE with Mantel test revealed that environmental and geographic distances were important to the divergence among P. chienii populations. We further applied RDA to dissect the individual roles of environmental factors and geographic factors and their confounding effect on genetic variation. Our RDA showed that population divergence in P. chienii was strongly shaped by the joint effect of environmental factors and geographic factors, and environmental factors were more important than geography, a pattern consistent with local adaptation. Environmental differences among populations may constitute key factors maintaining genetic differentiation despite high relative migration rates between local populations. It follows that environmental differences among populations are closely related to the maintenance of genetic variation and that local adaptation may be the main driving force of these patterns. However, a large part of the variation remains unexplained. This may be due to several factors that cannot be fully resolved. First, although we included a large number of environmental variables in our study, many other unmeasured ecological forces may also play a role. Second, other evolutionary forces that maintain local genetic diversity, such as balancing selection, may weaken the signal of locusenvironment associations. Third, the multivariate environmental association approach models only linear associations, so nonlinear statistical relationships will not be captured.
In this study, we found evidence for local adaptation signals to genetic variation associated with environmental variables. We detected 164 SNPs residing in 127 unigenes as candidate targets of adaptive importance. The GO annotation analysis of 127 unigenes showed that the majority of the unigenes were related to abiotic and biotic stress responses, which is of particular interest for future population genomic research. In the biological process category, oxidation-reduction process, signal transduction, and protein phosphorylation were the most represented GO terms, which is consistent with the findings of adaptability studies in conifers and other plants. These three biological processes were shown to be involved in the drought response in Masson pine (Pinus massoniana). The oxidation-reduction process may have contributed to cold resistance in the mature leaves of tea plants (Camellia sinensis)and adaptation to hypoxia, extreme temperatures, and high ultraviolet (UV) radiation in Kandelia obovata.
Protein phosphorylation is involved in activating cold acclimation. Signal transduction connects sensing mechanisms with genetic responses, which is important for sensitivity to environmental stresses and promotes effective downstream processes in response to these environmental stresses. Within the molecular function category, the term binding, including protein binding, ATP binding, and DNA binding, was generally related to the environmental stress response. The GO term binding was shown to be enriched in upregulated open reading frames (ORFs) associated with the cold response in Chinese yew (Taxus chinensis). Additionally, lines of evidence support that the membrane plays a key role in abiotic stress and plant defense. The membrane can directly or indirectly sense stress through physical properties to initiate signal transduction pathways. Previous studies revealed significant changes in plasma membrane function in response to cold stress in T. chinensis. Although the annotation analysis suggested functional importance for most candidate unigenes, the unannotated unigenes are still promising candidates for future study, as they may be related to adaptive genes or genes of unknown importance.
The allele frequency of SNPs changes under selection pressure, thereby rapidly maximizing adaptability in different environments. Niche differentiation was detected for GX vs. ZJ and HN vs. ZJ; thus, we further used selective sweep analysis to identify the unigenes underlying divergent adaptation in the ZJ, GX, and HN groups. We found 87 candidate unigenes for habitat adaptation in the ZJ group. The KEGG pathway of monoterpenoid biosynthesis was significantly enriched (q-values < 0.05) for 87 candidate unigenes. Terpenoids play an important role in abiotic and biotic stresses and are involved in the defense against pathogens and herbivore attack in conifers. Drought can increase the concentration of monoterpenoids in conifers, such as Picea abies and Pinus halepensis. Monoterpenoid biosynthesis may play an important role in the local adaptation of P. chienii in the ZJ group. Ubiquitin-mediated proteolysis was significantly enriched (q-values < 0.05) for candidate unigenes for habitat adaptation in the HN group. Ubiquitin-mediated proteolysis impacts many aspects of plant growth and development, including plant hormone signal transduction, reproduction, and abiotic stress responses. Some genes involved in ubiquitin-mediated proteolysis show signs of positive selection. The identified candidate unigenes for habitat adaptation in the GX group are directly or indirectly related to biotic or abiotic stress responses. For instance, Cluster-242, 496.98097, encoding a 70 kDa heat shock protein (HSP70), is associated with pathogen and disease resistance and plant responses to high temperatures. These results suggest that these pathways and candidate unigenes may play an important role in local adaptation in the GX, ZJ, and HN groups of P. chienii.
Detecting associations between SNPs and environmental variables helps us recognize the ecological, bioclimatic, and topographic variables that influence genetic variation. We found that the fraction of absorbed photosynthetically active radiation, soil Cu content, soil Zn content, percent tree cover, aspect, and precipitation of the driest month were associated with the most outlier SNPs when using the LFMM method (Additional file 22), suggesting their importance in shaping the genetic variation underlying P. chienii adaptability. The forward selection performed as part of RDA showed that the mean temperature of the coldest quarter, soil Fe content, soil Cu content, precipitation of the driest month, and altitude were important determinants of outlier genetic variation. These variables may be selective factors driving the local adaptation of P. chienii. Aspect, a topographic factor, is a key predictor of differences in forest radiation exposure and has a strong influence on the microclimate. This factor has been found to be related to genetic variation within and among Salix species. Both the LFMM method and RDA revealed a strong signal of divergent selection in relation to precipitation-related variables.highlighted the importance of precipitation of the driest month in shaping the species distribution of P. chienii. Precipitation can directly affect soil water content, thus affecting the absorption and transport of plant water and nutrients. A decrease in precipitation of the driest month is expected over the coming decades (current with a mean of 45.9 mm; 2050 with a mean of 33.2 mm), particularly in the distribution range of the JX group of P. chienii. In the case of rapid climate change, if P. chienii populations cannot adapt to increasing drought, ecological benefits will be greatly damaged. According to a global assessment, due to droughts, high temperatures, and insect outbreaks under climate change, the mortality rate of forest trees may increase. Selecting and planting genotypes adapted to climate change is of great significance to the protection of the endangered species P. chienii. The identified candidate unigenes are directly or indirectly related to biotic or abiotic stress responses. For instance, Cluster-242,496.114750, encoding an LRR receptor-like serine/threonine-protein kinase, is associated with pathogen and disease resistance. Candidate unigenes encoding one methyltransferase, four ubiquitin, and one auxin-responsive protein were detected (Additional file 23). They were associated with multiple environmental variables, including leaf area index, percent tree cover, fraction of absorbed photosynthetically active radiation, altitude, precipitation of the driest month, and soil Mg, Zn, Cu, and Mn contents. These proteins have been reported to be related to loblolly pine adaptability. These detected unigenes with functional annotations provide strong support for adaptive variation in P. chienii. Future climate trends in the distribution range of P. chienii, including increased temperature and decreased precipitation, will pose challenges to P. chienii in terms of its environmental adaptation. Our research provides SNPs and candidate unigenes related to environmental variables to facilitate elucidation of the genetic variation and structure of P. chienii in relation to environmental adaptation.
# Conclusions
We identified 13,545 SNPs to determine genetic and expression variation patterns and local adaptation across 10 populations of P. chienii. Gene expression and nucleotide diversity had a reciprocal relationship when P. chienii adapted to the surrounding environment. Despite the strong population structure in P. chienii, genomic data revealed signatures of divergent selection associated with environmental variables. We identified 244 associations between 164 outlier SNPs and 17 environmental variables. The mean temperature of the coldest quarter, soil Fe and Cu contents, precipitation of the driest month, and altitude were identified as the most important determinants of adaptive genetic variation. Most candidate unigenes with outlier signatures were related to abiotic and biotic stress responses. The results of our study are expected to improve insights into evolutionary processes and local adaptation in P. chienii.
# Methods
## Sample collection and rna isolation
Our sampling work complies with the laws of the People's Republic of China and has a permission letter from Sun Yat-sen University. Voucher specimens were identified by Prof. Zien Zhao (Wuhan Botanical Garden, Chinese Academy of Sciences, Wuhan, Hubei, China) and kept at the Herbarium of Sun Yat-sen University (No: ds-2018-1001-ds-2018-1010).
Fresh and mature needles of 108 individuals from 10 populations of P. chienii were collected from first-year branches across the distribution range in China in May 2018 . A global positioning system (GPS) was used to record the geographic coordinates of the sampling locations. Plants were sampled at intervals of 20 m in each population. All samples were washed with purified water, cut into pieces and then immediately stored in RNAfixer (BioTeke, Shanghai, China). The RNAfixer with samples was stored in a − 20°C freezer until further use.
Total RNA extraction of each individual was performed using the RNAprep Pure Plant Kit following the protocol of the manufacturer (Tiangen, Beijing, China). The purity and integrity of the extracted RNA were detected using a NanoDrop spectrophotometer (Thermo Scientific, DE, USA) and an Agilent 2100 Bioanalyzer (Agilent Technologies, CA, USA), respectively. An RNA integrity number > 6.0 was required for cDNA synthesis and library construction. One microgram of RNA from each individual was used for cDNA library preparation.
## Library construction, sequencing, and assembly
The Illumina library for each individual was constructed using the NEBNext Ultra™ RNA Library Prep Kit (NEB, MA, USA) following the manufacturer's protocol. Poly(A) mRNA was enriched from total RNA using oligo (dT) magnetic beads. Then, the poly(A) mRNA was fragmented into small pieces using divalent cations under elevated temperature in NEBNext First Strand Synthesis Reaction Buffer (5×). The RNA fragments were reverse transcribed into first-strand cDNA using random hexamer primers and M-MuLV Reverse Transcriptase (RNase H-). Subsequently, second-strand cDNA was synthesized using dNTPs, DNA polymerase I, and RNase H. The purified double-strand cDNA was end-repaired and A-tailed, and then Illumina paired-end adapters were ligated. The library fragments were purified using AMPure XP beads (Beckman Coulter, Beverly, USA) to select cDNA fragments with lengths of 250-300 bp. Then, PCR was performed with Phusion High-Fidelity DNA polymerase, universal PCR primer and index (X) primer. After size selection and PCR amplification, qualified cDNA libraries were sequenced on the Illumina NovaSeq platform, generating paired-end reads with a length of 150 bp.
Raw Illumina RNA-Seq reads were filtered via inhouse Perl scripts. Clean reads were obtained by removing the reads with more than 10% ambiguous bases ("N"), adapter reads, and Qphred scores ≤20 bases with more than 50% from the raw reads. Finally, clean reads of 108 individuals were de novo assembled to obtain the final unigenes with Trinity v.2.4.0. The final highquality unigenes of 108 individuals served as the reference sequences for estimating genetic and expression variation among the 10 populations of P. chienii.
## Read mapping and snp calling
Clean reads for each individual were mapped to the reference sequences of P. chienii using Bowtie 2 (http:// bowtie-bio.sourceforge.net/bowtie2/index.shtml) with default parameters. Duplicate reads were excluded by Picard tools ver. 1.96 (http://broadinstitute.github.io/ picard/); then, the reads were sorted and indexed in BAM format using SAMtools ver. 1.4 with default settings. mpileup of SAMtools was used to analyze the alignment results of the reference sequence base sites with the parameters -q 1 -C 50 -t SP -t DP -m 2 -F 0.002. SNP calling was conducted with BCFtools of SAMtools using the following parameters: -Q 20 -d 1 -D 1000 -a 2 -w 3 -W 10. To ensure the accuracy of SNP identification, we also used GATK ver. 3.7to identify the SNPs with the parameters FS < 30.0, QD > 2.0, and DP > 10. All monomorphic SNPs were removed. Only SNPs identified by both SAMtools and GATK were retained. To minimize false-positive SNPs and obtain high-quality SNPs, we filtered the SNP loci using the criteria depth > 2, call rate > 0.5, and minor allele frequency (MAF) > 0.05 and kept only biallelic SNPs. For consistency with the gene expression data, we removed SNPs contained in unigenes with extreme fragments per kilobase of transcript per million mapped reads (FPKM) values (see below).
## Genetic variation and population genetic structure based on snp data
Based on the SNP data, the nucleotide diversity (π) per site was calculated using VCFtools ver. 0.1.11 (https:// vcftools.github.io/index.html) with nonoverlapping 1000bp genomic windows. Pairwise genetic differentiation (F ST ), observed heterozygosity (H O ), and expected heterozygosity (H E ) were calculated using Arlequin ver. 3.5.1.2, and the significance of F ST was determined using 1000 permutations. Wright's inbreeding coefficient (F IS ) was estimated for each population using the basic.stats function in the R package 'hierfstat'.
The population genetic structure of the 108 individuals was examined in ADMIXTURE ver. 1.23using the maximum-likelihood method to identify evolutionary clusters. The number of genetic clusters (K) was set from 2 to 10. Cross-validation error (CV error) was used to determine the most likely number of clusters. The lowest CV error indicated the optimum K value. GCTA ver. 1.93.2 softwarewas used to perform PCA on the P. chienii individuals. The first two components were plotted for P. chienii to explore its genetic structure.
AMOVA was used to assess the extent of genetic structure within and among populations as implemented in Arlequin, and the significance was determined for 1000 permutations.
The best model of nucleotide substitution was identified with PhyML 3.0using Akaike's information criterion (AIC), and GTRGAMMAI was the best-fitting model. A maximum likelihood (ML) tree was constructed using RAxML ver. 8.2.4with 1000 bootstrap replicates.
## Gene expression variation and population differentiation based on fpkm values
The clean reads of each individual were mapped to the reference sequences of P. chienii using Bowtie 2 of the RSEM software (http://bowtie-bio.sourceforge.net/ Bowtie2/index.shtml). The readcount values of each unigene for 108 individuals were obtained. Considering the influence of the gene length and sequence depth on the fragments, all readcounts were normalized to the FPKM values. The FPKM values were calculated using the formula below: FPKM = (10 9 × C)/(N × L), where C is the number of fragments mapped to the transcript, N is the total number of mappable reads, and L is the length of the transcript. The expression level of each unigene in 108 individuals was determined by calculating the FPKM value. To filter the extremely large FPKM values, 1 was added to the FPKM value of each unigene, and then log 2 transformation was performed. Quartile analysis was used to filter out values greater than 1.5 times the interquartile range. In total, 16,225 unigenes with at least half of the individuals with log 2 -transformed FPKM values larger than 0 were retained.
Gene expression variation in populations was evaluated using Xu et al.'s (2015)method. The population gene expression level (E p ) was evaluated as the average FPKM value of the individuals from the population. Expression diversity (E d ) was calculated as the gene expression variation in the population. The formulas for E p and E d were based on Xu et al.'s (2015) method. To estimate the gene expression level relationship among populations (E p similarity), we calculated Pearson's correlation coefficients (r) based on the average correlation coefficients of 108 individuals. The significance of the relationship between genetic distance and E p similarity in populations was tested with a Mantel test using 1000 random permutations. To detect the relationship between nucleotide diversity and expression diversity within populations, Pearson's correlation coefficients (r) of π and E d for 16,225 unigenes in each population were calculated. Additionally, Pearson's correlation analysis was performed between π and E p among populations. The 'HMISC' package in R was used for Pearson's correlation analysis, and the significance was determined for 1000 permutations.
## Directional migration rates
We pooled the populations into four groups (Jiangxi group, JX: BJS and ZZB; Guangxi group, GX: LHS and YSGY; Hunan group, HN: ZJJ; and Zhejiang group, ZJ: SQS, SMJ, MS, LMD, and DXG) based on the results from the PCA, phylogenetic tree, and geographical distribution. To assess directional relative migration rates among the 10 populations/four groups of P. chienii, a putatively neutral dataset SNPs) was performed with the divMigrate function in the R package 'diveRsity'based on three measures of genetic differentiation (Jost's D, G ST , and Nm). This approach is a relative bidirectional measure of gene flow using all available genetic differentiation measures to evaluate the consistency of estimates among measures. The confidence interval (95%) of the relative migration rates was calculated with 1000 bootstrap iterations.
## Environmental variables
The environmental variables include 19 bioclimatic, 25 ecological, and three topographic variables. Bioclimatic data (averaging over the period 1950-2000) comprising 19 bioclimatic variables at a spatial resolution of 2.5 arcmins were obtained from the WorldClim database (http://www.worldclim.org). The ecological variables comprised five ecological factors and 20 soil variables. The five ecological factors included the normalized difference vegetation index, percent tree cover, leaf area index, enhanced vegetation index, and fraction of absorbed photosynthetically active radiation, which were obtained from the Land Process Distributed Active Archive Center (http://lpdaac.usgs.gov, recorded in 2001-2018) based on the moderate resolution imaging spectroradiometer (MODIS) dataset. The annual MODIS layers were generated based on a maximum value composite procedure in the App Store of the ENVI 5.3 package. Then, the layers of different years were averaged to obtain a single layer representing the ecological factor. Twenty soil variables, including pH, electrical conductivity, fresh water content, air-dried water content, organic matter, and total N, P, C, S, Si, K, Ca, Na, Mg, Al, Fe, Mn, Zn, Cu, and Pb, were obtained from published research by our research group. The three topographic variables included altitude, aspect, and slope. Altitude was derived from data recorded during field sampling. Aspect and slope were derived from SRTM elevation data under a digital terrain model with a resolution of 2.5 arc-mins, and the values for sampling sites were calculated in ArcGIS ver. 10.4.1 (http://www. esri.com/software/arcgis/arcgis-for-desktop). To reduce multicollinearity among variables, the variance inflation factors (VIFs) were calculated for the 19 bioclimatic variables, five ecological factors, 20 soil variables, and three topographic variables using the vif function of 'usdm' in R software. Environmental variables with a VIF > 10 were removed. Five bioclimatic variables, four ecological factors, six soil variables, and three topographic variables were retained (Additional file 24).
## Isolation by distance (ibd) and isolation by environment (ibe)
To test for IBD and IBE, we generated geographic and environmental distance matrices. The 18 environmental variables (Additional file 24) were subjected to principal component analysis using the prcomp function in R software. We selected the resulting first five principal component axes, which explained 84.72% of the environmental variance. The five principal component axes were used to calculate environmental distances (Euclidean distance) using the dist function in R software. To obtain geographic distance matrices, we calculated the geographic distance in kilometers among populations.
To investigate the roles of environmental and geographic factors in shaping genetic differentiation, we calculated the associations between pairwise F ST among populations and environmental distance and geographic distance with the Mantel test using the mantel function of the R package 'vegan' (https://github.com/vegandevs/ vegan/), and the significance was determined for 999 permutations.
## Ecological niche modeling
After removing duplicate records, a total of 60 records of P. chienii, including 15 for GX, six for HN, 11 for JX, and 28 for ZJ, were collected from field sampling, published resources, the Chinese Virtual Herbarium (CVH, http://www.cvh.ac.cn/), and the Global Biodiversity Information Facility (GBIF, http://www.gbif.org), which covered most of the distribution range of this species.
Current climatic data (averaging over the period 1950-2000) comprising 19 bioclimatic variables at a spatial resolution of 2.5 arc-mins were obtained from the WorldClim database (http://www.worldclim.org). The VIFs were calculated for the 19 bioclimatic variables to reduce the multicollinearity among the variables. Bioclimatic variables with a VIF > 10 were removed. Five bioclimatic variables were retained, namely, mean temperature of the coldest quarter, precipitation of the wettest month, precipitation of the driest month, precipitation seasonality (CV), and precipitation of warmest the quarter. Maxent ver. 3.3.3 kwas used to predict the current potential distributions of the four groups (ZJ, JX, GX, and HN) of P. chienii with the following parameters: replicates, 10; replicated run type, subsample; maximum iterations, 500; and random test points, 25.
The AUC values were used to predict the performance of the models, with AUC values closer to 1.0 indicating better model performance. ENMTools ver. 1.4.4was used to measure the niche differences between pairs of the four groups using the niche overlap tool. Schoener's D and the standardized Hellinger distance (I) were calculated to measure niche overlap in group pairs. Identity tests of six comparisons (GX vs. JX, GX vs. ZJ, HN vs. GX, HN vs. JX, HN vs. ZJ, and JX vs. ZJ) were performed.
To examine the effects of past and future climatic shifts on the four groups of P. chienii, ecological niche modeling was used to predict potential distribution patterns during the future (2050, average for 2041-2060), the LIG (approximately 130-114 kya), and the LGM (approximately 21 kya). Climate layers for the LGM and future at a spatial resolution of 2.5 arc-mins and LIG at a spatial resolution of 30 arc-secs were obtained from the WorldClim database.
## Detection of candidate unigenes and annotation
To determine if there were unigenes putatively under selection in the 10 populations, we implemented an F ST outlier approach in BayeScan ver. 2.1. It has been reported that BayeScan software has lower false-positive rates than other similar software programs. This method is based on a logistic regression that decomposes genetic variation into a population-specific F ST coefficient (β) shared by all loci and a locus-specific F ST coefficient (α) shared by all the populations. All parameters set in BayeScan software were kept as the default. To reduce the occurrence of false positives, we calculated the qvalues in BayeScan, and SNPs with a q-value lower than 0.001 were considered outlier SNPs. We also calculated the locus-specific F ST coefficient (α), where a positive value indicates diversifying selection, while a negative value indicates purifying/balancing selection. In the analysis of the correlation between genomic variation and environmental variables, we considered only SNPs with positive α values and ignored SNPs with negative α values.
Based on niche overlap analysis, the ecological differentiation of GX vs. ZJ and HN vs. ZJ was valid. Therefore, we further used selective sweep analysis to detect adaptation. We calculated the F ST values and π ratios (π GX /π ZJ , π ZJ /π GX , π HN /π ZJ , and π ZJ /π HN ) for group pairs. The π ratios were subjected to log 2 transformation. The regions with F ST and log 2 (π ratio) values in the top 5% were considered candidate outliers subjected to strong selective sweeps. Then, all outliers were assigned to corresponding unigenes.
We performed functional annotation of the candidate unigenes containing outlier SNPs identified in BayeScan and selective sweep analysis. Hmmscan of HMMER ver. 3.1was used to perform Pfam protein database annotation. Based on the protein annotation information from the Pfam database, GO term annotation was determined using Blast2GOand a custom script. KEGG and Swiss-Prot database annotations were implemented using DIAMOND ver. 0.8.36with an E-value of 1.0 × 10 − 5 . To identify significantly enriched biological functions and pathways, we performed GO and KEGG enrichment analyses. The GO enrichment analysis of candidate unigenes was performed using the 'GOseq' R package based on the Wallenius noncentral hypergeometric distribution. KEGG pathway enrichment was determined using KOBAS (2.0). q-values (adjusted p-values) were used to test the statistical significance of GO and KEGG enrichment, with q-values < 0.05 considered significant.
|
10.1155/2021/6682705
|
CCBY
|
8298177
|
34336332
|
s2orc_pubmed_articles
|
COVID-19: Moral and Ethical Implications for Orthopaedic Spine Surgeons
The rapid spread of COVID-19 has made a significant impact on healthcare systems worldwide, with a large influx of patients prompting the cancellation of elective surgery in order to conserve resources and prevent the risk of exposure to the novel virus. In this case report, we present a 66-year-old male patient, with a history of cerebral palsy and developmental disabilities, exhibiting an increasing loss of function over the course of 10 days amid the COVID-19 pandemic. The patient was initially refused transport to the hospital by emergency medical services and later transported per independent request from his surgeon. Upon admittance to the hospital, the patient was found to have severe spinal cord compression with myelopathic symptoms and underwent an anterior cervical discectomy and fusion. This case highlights the need for more specific guidelines regarding the evaluation of a spinal injury by EMS and the hospital system amid a national crisis.
# Introduction
In December 2019, the first case of coronavirus disease 2019 (COVID-19) caused by the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) was documented in Wuhan, China [bib_ref] Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and coronavirus disease-2019 (COVID-19): the..., Lai [/bib_ref]. Since then, COVID-19 has evolved into a global pandemic, affecting over 12 million people as of July 2020. The large population of affected individuals has put a strain on the global healthcare system. Personal protective equipment, operating rooms, and ventilators are in short supply throughout areas with large numbers of COVID-19 cases. In an effort to conserve resources and reduce spread of the virus, both the ethics and frameworks of our healthcare system have undergone a remodeling.
Amid the pandemic, spine surgeons faced multiple sources of moral distress caused by major changes in healthcare. One such source stemmed from physicians and hospitals recognizing a needed shift away from patientcentered healthcare towards public health-centered care. In order to conserve equipment and space for COVID-19 patients, surgeons were forced to make decisions that would benefit the public's health, rather than benefit a sole patient. This resulted in a reduced ability for a surgeon to respect a patient's choices [bib_ref] Surgeons, ethics, and COVID-19: early lessons learned, Angelos [/bib_ref]. Additionally, many hospitals had set out a mandate for surgeons to remain at home during the pandemic unless they were specifically called upon. This resulted in another source of moral distress, as surgeons were not able to provide care in the midst of a public health crisis [bib_ref] Surgeons, ethics, and COVID-19: early lessons learned, Angelos [/bib_ref].
Although telehealth was implemented in the place of some in-person visits, not all patients could be assessed in this manner. Patients that needed sutures removed for example or patients that required a more hands-on evaluation by the attending surgeon would still stand to benefit most from in-person care [bib_ref] Departmental experience and lessons learned with accelerated introduction of telemedicine during the..., Loeb [/bib_ref]. In this particular case, the use of telemedicine aided in Dr. Verma's assessment of the patient's case as emergent, leading to the patient's prioritization and rapid hospitalization following the telehealth visit. In an emergent case, as described in this case report, teleconsultation can be beneficial in the delivery of timely medical intervention and avoidance of debilitating, permanent consequences. In the midst of a pandemic, telehealth allows the surgeon to determine a baseline evaluation of the degree of urgency regarding a patient's case and the avoidance of unnecessary hospital visits in less emergent cases. At the same time; however, a patient may need to come in for a more accurate evaluation, and a remote visit can be offered as a precursor to determine whether a patient can be managed via telehealth versus whether additional imaging or an in-person visit is necessary [bib_ref] Departmental experience and lessons learned with accelerated introduction of telemedicine during the..., Loeb [/bib_ref].
Orthopaedic spine surgeons were additionally facing the mandated cancellation of elective surgeries in order to reduce the strain on hospitals, conserve resources, and slow down the spread of the virus. Due to these restrictions, spine surgeons were put in a position in which they needed to make difficult decisions regarding which patients were in most urgent need of surgery and which patient's surgical needs could be delayed. These decisions can be made more difficult in instances where postponing surgical spine treatment can lead to poorer recovery or irreversible neurological damage. Risk factors such as the contraction of COVID-19 postoperatively and the possibility of this delaying recovery need to be taken into account; minimizing the risk of a patient contracting COVID-19 in a hospital setting, namely, elderly patients who are most at risk for severe disease or death, should be considered by the surgeon [bib_ref] Triaging spine surgery and treatment during the COVID-19 pandemic, Rizkalla [/bib_ref]. Many hospitals have been developing protocols that stratify the urgency of specific cases. Conditions exhibited by the patient that may denote neurological compromise, such as new or progressing weakness, myelopathy from degenerative disease, and severe pain due to nerve compression, have been considered in need of urgent surgical intervention [bib_ref] Editorial. COVID-19 and spinal surgery, Ghogawala [/bib_ref]. Additionally, patients at risk for permanent neurological damage have been deemed urgent cases [bib_ref] Editorial. COVID-19 and spinal surgery, Ghogawala [/bib_ref].
Typically, orthopaedic surgeries are categorized into urgent inpatient, urgent outpatient, elective inpatient, and elective outpatient categories [bib_ref] Orthopaedic surgical selection and inpatient paradigms during the coronavirus (COVID-19) pandemic, Massey [/bib_ref]. However, the COVID-19 pandemic and restriction of elective surgeries have resulted in a need for a more specific system to categorize cases based on the priority level.
Although there has not yet been a multi-institutional system implemented, a scoring system aimed at categorizing patient cases based on urgency has been developed by 16 spinal surgeons across the United States [bib_ref] Scoring system to triage patients for spine surgery in the setting of..., Sciubba [/bib_ref]. On a scale from -19 to 91, cases are categorized by the degree of progression and impairment caused by their deficit, with higher numbers corresponding to increasing urgency [bib_ref] Scoring system to triage patients for spine surgery in the setting of..., Sciubba [/bib_ref]. Patients that are at greater risk of postoperative complications, more severe neurological deficits, and spinal instability are additionally scored higher [bib_ref] Scoring system to triage patients for spine surgery in the setting of..., Sciubba [/bib_ref]. This scoring system may provide a guide to help in the prioritization of patients during COVID-19; however, the ultimate decision will still be left to the surgeon.
There are also guidelines formulated by the French Spine Surgery Society regarding the management of patients and urgency of surgery [bib_ref] French Spine Surgery Society guidelines for management of spinal surgeries during COVID-19..., Prost [/bib_ref]. This is based on three levels of urgency: urgent surgical indications, surgical indications associated with a potential loss of chance for the patient, and nonurgent surgical indications [bib_ref] French Spine Surgery Society guidelines for management of spinal surgeries during COVID-19..., Prost [/bib_ref]. Patients categorized in level 1 urgent surgical indication would require a procedure; however, patients categorized in levels 2 and 3 will have either a scheduled surgery with reasonable delay or a completely postponed surgery [bib_ref] French Spine Surgery Society guidelines for management of spinal surgeries during COVID-19..., Prost [/bib_ref]. These models can aid in a surgeon's decisions on how to rank the urgency of surgeries.
New York State's largest healthcare provider, Northwell Health, uses similar guidelines. At Northwell, surgeons identified their cases from levels 0-4, with level 4 representing emergent surgeries and level 3 representing urgent surgeries, both of which were prioritized for surgery, whereas levels 2 and lower were postponed [bib_ref] COVID-19 and orthopaedics: recovery after the pandemic surge, Petrone [/bib_ref]. Level 0 indicates an elective surgery, where the timing of surgical intervention would not impact a patient's outcome. These classifications for prioritizing surgery are determined based on how harmful a time-based delay in surgical intervention would be for the patient. Cases that are deemed a level 4 priority indicate that the patient would be at an immediate risk for permanent disability or severe harm without surgery and indicate a need for immediate surgical intervention [bib_ref] COVID-19 and orthopaedics: recovery after the pandemic surge, Petrone [/bib_ref]. An accepted treatment protocol in the United States for spinal cord injury, which would render a case emergent, follows a suggested 8-24hour surgical intervention time window for the best recovery results [bib_ref] Triaging spine surgery and treatment during the COVID-19 pandemic, Rizkalla [/bib_ref]. A level 3 case indicates that a patient has failed previous medical management of their condition and would face a prolonged stay/increased chance of another hospital admission if a delay in surgery occurred. Cases deemed urgent do not currently have a clear exact recommended timing of surgical intervention in the United States, although these cases are considered in need of prompt intervention, more so than levels 0-2 but not as immediate as level 1 cases [bib_ref] Triaging spine surgery and treatment during the COVID-19 pandemic, Rizkalla [/bib_ref]. Finally, level 2 surgeries are considered semiurgent, where a delay past three months would put the patient at a risk of harm, and a case considered level 1 indicates that a delay past six months would not put a patient at risk of harm [bib_ref] COVID-19 and orthopaedics: recovery after the pandemic surge, Petrone [/bib_ref]. The pause of elective surgeries in New York lasted from March 32rd, 2020, to June 28 th , 2020 [bib_ref] Resumption of elective orthopaedic surgery in the US epicenter of COVID-19, Krauss [/bib_ref]. During this time, there were approximately 16,000 surgical cases postponed, with about 1,800 of these cases being orthopaedic, in order to conserve resources during the pandemic [bib_ref] COVID-19 and orthopaedics: recovery after the pandemic surge, Petrone [/bib_ref].
It was up to the surgeon to identify level 3 and 4 cases that were postponed and submit case information to their respective department chair to approve scheduling in the operating room. Then, the Surgical Emergency Operations Committee reviewed all data and provided oversight for all scheduled cases [bib_ref] COVID-19 and orthopaedics: recovery after the pandemic surge, Petrone [/bib_ref]. Once the case was approved, the surgery was scheduled and the patient underwent medical clearance exams and COVID testing. All COVID-positive patients were postponed, unless classified as level 4 [bib_ref] COVID-19 and orthopaedics: recovery after the pandemic surge, Petrone [/bib_ref].
Research done in Germany has revealed that restrictions on elective surgery have led to postponements affecting over 60% of patients that would otherwise be receiving surgery [bib_ref] COVID-19 nonessential surgery restrictions and spine surgery, Mehta [/bib_ref]. Once elective surgeries resume, these delays may lead to increased wait times in receiving treatment as well as higher readmission rates, as the delay in surgery may lead to a higher complication rate [bib_ref] Risk factors for and complications after surgical delay in elective single-level lumbar..., Wagner [/bib_ref]. Delays in surgery, namely, anterior cervical discectomy and fusion surgeries, have been 2
Case Reports in Orthopedics associated with a five-fold increase in hospital stay time and mortality rate as well as a higher rate of returning to the OR [bib_ref] Risk factors for delay in surgery for patients undergoing elective anterior cervical..., Renfree [/bib_ref].
## Case report
A 66-year-old male with a past medical history of cerebral palsy, seizure disorder, and mental retardation with developmental disabilities presented to the North Shore University Hospital emergency department with symptoms of neurologic deterioration. At baseline, the patient was ambulatory with a walker with good functional status in upper and lower extremities.
Prior to admission, the patient's sister and healthcare proxy had contacted our attending orthopaedic spine surgeon, who was informed that 10 days prior the patient experienced multiple falls and progressive weakness, resulting in minimal ambulation. The patient had increasing loss of function of bilateral upper extremities, with right greater than left symptoms, and bilateral lower extremities and essentially became bed bound in the span of 10 days. The patient also experienced progressive difficulty with feeding himself.
Emergency medical services (EMS) had refused initial requests to acquire and transport the patient to the hospital due to concerns of safety and exposure during the COVID-19 pandemic. As a result, the attending surgeon independently contacted EMS and persuaded them to bring the patient to North Shore University Hospital for urgent evaluation and operative intervention.
Clinical examination of the bilateral upper extremities revealed 3/5 motor strength in the deltoids, biceps, and triceps, proximally and distally, and 1-2/5 strength in grip strength. Examination of bilateral lower extremities revealed 3-4/5 motor strength proximally in the hip flexors and knee extensors. Tibialis anterior, extensor hallucis longus, and peroneals were found to be 1-2/5 in motor strength. There was also reduced sensation throughout, in both the upper and lower extremities. The patient also had a positive Babinski sign, as well as positive Hoffman's sign. Increased deep tendon reflexes were also present. The patient did not exhibit any bowel or bladder dysfunction; however, given the signs and symptoms, the patient was given a modified Japanese Orthopaedic Association (mJOA) score of 8, defining the severity of myelopathy as severe [bib_ref] The modified Japanese Orthopaedic Association scale: establishing criteria for mild, moderate and..., Tetreault [/bib_ref].
MRIs of the cervical, thoracic, and lumbar spine were performed which neuro-radiologists interpreted as C2-C3 moderate to severe right neural foraminal narrowing, C3-4 moderate to severe bilateral neural foraminal narrowing, and frank spinal cord compression at C4-C5 without definite abnormal intrinsic cord signal with moderate bilateral neural foraminal narrowing [fig_ref] Figure 1: Sagittal T2-weighted MRI of the cervical spine, demonstrating spinal cord compression at... [/fig_ref].
Neurology and internal medicine specialists were consulted, and after evaluation by these teams, the patient was deemed to have a myelopathic condition. Although the findings on MRI did reveal cord compression but were not conclusive of a myelopathic condition, given the severe progression of symptoms of motor weakness and sensory deficits, the decision was made that the patient would ben-efit from surgical intervention and the patient was prepared and optimized for surgery.
One day after presentation to the hospital, the patient underwent an anterior cervical discectomy and fusion at the level of C4-C5 [fig_ref] Figure 3: Lateral intraoperative radiograph of the cervical spine, status post-C4 and C5 anterior... [/fig_ref]. The patient tolerated the procedure well, and there were no complications.
On postoperative day 1, clinical examination revealed mild improvement in motor strength in bilateral upper and lower extremities; however, the patient remained weak compared to baseline. Sensation improved and remained intact throughout the hospital stay. The patient was discharged from the hospital on postoperative day 7.
During a 3-month outpatient follow-up visit, clinical examination of the patient revealed improvement of neurologic symptoms. The patient was able to walk with assistance and a walker. Motor strength and sensation demonstrated improvement since the hospitalization with minimal pain. Clinical examination of the bilateral upper extremities revealed 4/5 motor strength in the deltoids, biceps, and triceps, proximally and distally, as well as an improved motor 3+/5 in grip strength. Examination of bilateral lower extremities revealed 4+/5 motor strength proximally in the hip flexors and knee extensors, with improvement of tibialis anterior, extensor hallucis longus, and peroneals to 3+/5 motor strength. Most importantly, the patient was able to perform basic activities of daily living, including selffeeding. Overall, the patient was ambulatory with a walker with significant improvement in functional status.
# Discussion/conclusion
In this case report, we outline an event in which a patient who experienced rapid neurological decline with myelopathic symptoms secondary to cervical spinal stenosis and radiographically confirmed myelomalacia was denied initial access to urgent clinical evaluation and treatment, in the setting of the COVID-19 pandemic. This event serves to shed light on the inherent issues surrounding access to care during a global and national health crisis. We describe a situation where a patient's clinical care and surgical treatment were delayed and urged independently by a surgeon due to the hesitancy of action by the system, caused by the fears and Case Reports in Orthopedics prejudices inherent during this pandemic that may have potentially caused irreversible damage and morbidity. We highlight here an issue that may have longer standing implications given the current state of the world and possible recurrences of similar global health crises in the future.
In reflection of the circumstances, several opportunities for education are available for the emergency medical services (EMS). While resource conservation during a medical catastrophe is sometimes necessary, acute neurological decline must be a consistent priority for the first responders. The EMS baseline assessment should include questions regarding changes in functional status, activities of daily life, bowel or bladder function, extremity use, and any fevers or chills associated with neurological decline. Whenever caring for a patient with a history of cognitive or neurological impairments, it is critical to consider the function of the patient in the weeks prior compared to the patient's present status with development of symptoms. If the patient's family is present, obtaining patient history from their perspective is invaluable. Similarly for patients living in extended care facilities, nursing staff and family members should be consulted when assessing baseline function and obtaining a detailed history. Through education of first-line responders, we can ensure vigilance of neurological symptoms and provide emergent care for at-risk patients who may otherwise experience a delay of care.
When EMS transports patients with neurological compromise, especially during a medical catastrophe, consideration must be given to which admitting facility the patient will be transported. During the pandemic, certain hospitals were not able to retain spine surgeons to operate at all hours, nor did they have the appropriate infrastructure to care for these types of patients. Certain facilities, which had limited high-quality imaging necessary for diagnosing myelopathic pathology, would not have been able to accurately diagnose nor care for a patient with neurological compromise. Those institutions without access to adequate critical care infrastructures, such as an ICU and the pertinent staff needed to care for these types of patients, would also not have been able to accommodate the needs of a postoperative spine patient. Our institution includes a Certified Center of Spine Excellence serving the greater New York Metropolitan Area and has the support staff to serve these acute patients. EMS needs to be aware of the services that specific medical centers are able to provide in order to continue to provide emergent care, especially in the midst of a pandemic.
It is also worthwhile to mention the beneficial utilization of telemedicine in the setting of a pandemic, where concerns and risk of transmission are high. In this case report, telehealth aided in the prompt recognition of a patient who needed emergent intervention to prevent further deterioration and permanent neurologic impairment. This case highlights the importance of the utility of teleconsultation, especially in these unique circumstances where direct in- Case Reports in Orthopedics person visits may not be accessible or feasible. Going forward, the development and use of this newer modality of patient evaluation may prove invaluable in medicine. During this pandemic, there was a fear of exhaustion of resources and risk of contagion in hospitals. This was evident in the field triaging of patients deemed too critically ill to transport by EMS [bib_ref] s 911 system is overwhelmed. 'I'm terrified,' a paramedic says, Watkins [/bib_ref]. The use of EMS for another patient more likely to survive due to rapid intervention and transport was the justification in this scenario. In the end, society dictates which patients are appropriate to address with EMS; however, it must be considered that neurological decline may be just as grave as respiratory compromise during a pandemic. Therefore, there should still be a low threshold of suspicion for neurological compromise, in a patient exhibiting such symptoms to be brought in by EMS for further assessment. The cost of resource utilization and risks of spread are outweighed by the benefits of early intervention in the outcomes of patients with spinal pathology and subsequent neurological decline.
In the setting of surgical cancellations amidst a global pandemic, it is imperative that a scoring system is implemented that helps stratify the urgency of surgical management, so that similar situations to the one described do not become a common occurrence. Northwell Health uses a categorical system based on scores of 0-4, with a score of 4 reflecting a case with the highest urgency and need for surgical intervention in order to prevent secondary negative outcomes [bib_ref] COVID-19 and orthopaedics: recovery after the pandemic surge, Petrone [/bib_ref]. Although the patient in the case described in this report had a good overall outcome, there was still a delay in proper surgical intervention that may have resulted in permanent neurological damage and morbidity. Therefore, we strongly support the implementation of a multi-institutional guideline that also outlines the proper steps that must be taken at each level of patient care, including the first responders, surgeons, hospital administrators, and administrative chairs. Having a system that addresses the complicated nature of patient care during a pandemic or similar health crisis is of tantamount importance given the current state of our nation and the imperfect response to such a crisis in the age of such technological advancements in the modern era.
## Consent
No written consent has been obtained from the patients, as there is no patient identifiable data included in the case report.
[fig] Figure 1: Sagittal T2-weighted MRI of the cervical spine, demonstrating spinal cord compression at the level of C4-C5. [/fig]
[fig] Figure 2: Axial T2-weighted MRI of the cervical spine, demonstrating severe canal stenosis with cord compression at the level of C4-C5. [/fig]
[fig] Figure 3: Lateral intraoperative radiograph of the cervical spine, status post-C4 and C5 anterior cervical discectomy and fusion. [/fig]
[fig] Figure 4: Anterior-posterior intraoperative radiograph of the cervical spine, status post-C4 and C5 anterior cervical discectomy and fusion. [/fig]
|
10.3390/cancers6041953
|
CCBY
|
4276952
|
25268164
|
s2orc_pubmed_articles
|
The Future of Glioblastoma Therapy: Synergism of Standard of Care and Immunotherapy
The current standard of care for glioblastoma (GBM) is maximal surgical resection with adjuvant radiotherapy and temozolomide (TMZ). As the 5-year survival with GBM remains at a dismal <10%, novel therapies are needed. Immunotherapies such as the dendritic cell (DC) vaccine, heat shock protein vaccines, and epidermal growth factor receptor (EGFRvIII) vaccines have shown encouraging results in clinical trials, and have demonstrated synergistic effects with conventional therapeutics resulting in ongoing phase III trials. Chemoradiation has been shown to have synergistic effects when used in combination with immunotherapy. Cytotoxic ionizing radiation is known to trigger pro-inflammatory signaling cascades and immune activation secondary to cell death, which can then be exploited by immunotherapies. The future of GBM therapeutics will involve finding the place for immunotherapy in the current treatment regimen with a focus on developing strategies. Here, we review current GBM therapy and the evidence for combination of immune checkpoint inhibitors, DC and peptide vaccines with the current standard of care.
# Introduction
Glioblastoma (GBM) is the most prevalent type of primary brain tumor, with a median survival of approximately one year [bib_ref] Factors influencing survival in high-grade gliomas, Buckner [/bib_ref] [bib_ref] Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma, Stupp [/bib_ref]. Poor median survival has provided an impetus for developing survival-prolonging or curative therapies [bib_ref] Brain tumors, Deangelis [/bib_ref] [bib_ref] Evaluation of BCNU and/or radiotherapy in the treatment of anaplastic gliomas, Walker [/bib_ref] [bib_ref] Randomized comparisons of radiotherapy and nitrosureas for the treatment of malignant glioma..., Walker [/bib_ref]. Despite current treatment regimens, GBM continues to be associated with a <10% 5-year survival rate [bib_ref] Brain tumors, Deangelis [/bib_ref] [bib_ref] Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in..., Parsa [/bib_ref]. A number of recent advances have shown promise for combating resistance to chemotherapy, and combination radiotherapy and chemotherapy regimes have been successful in modestly improving survival [bib_ref] Temozolomideand other potential agents for the treatment of glioblastoma multiforme, Nagasawa [/bib_ref].
The current standard of care for GBM is safe maximal surgical resection followed by radiotherapy and concurrent Temozolomide (TMZ), followed by adjuvant TMZ [bib_ref] Temozolomideand other potential agents for the treatment of glioblastoma multiforme, Nagasawa [/bib_ref]. Retrospective analysis has demonstrated that the extent of glioblastoma resection including residual volume of tumor left in the tumor bed significantly impacts overall survival and recurrence [bib_ref] Establishing percent resection and residual volume thresholds affecting survival and recurrence for..., Chaichana [/bib_ref] [bib_ref] Neurosurgical outcomes in a modern series of 400 craniotomies for treatment of..., Sawaya [/bib_ref] [bib_ref] Glioma extent of resection and its impact on patient outcome, Sanai [/bib_ref]. While TMZ is the mainstay chemotherapy for GBM, carmustine (BCNU) wafers have shown promise in providing localized chemotherapy for high-grade glioma [bib_ref] Temozolomideand other potential agents for the treatment of glioblastoma multiforme, Nagasawa [/bib_ref] [bib_ref] A phase 3 trial of local chemotherapy with biodegradable carmustine (BCNU) wafers..., Westphal [/bib_ref] [bib_ref] Interstitial chemotherapy with biodegradable BCNU (Gliadel) wafers in the treatment of malignant..., Bota [/bib_ref] [bib_ref] Interstitial chemotherapy with carmustine-loaded polymers for high-grade gliomas: A randomized double-blind study, Valtonen [/bib_ref] [bib_ref] The basis for current treatment recommendations for malignant gliomas, Fine [/bib_ref]. As a result of only modest improvements in survival emerging from the traditional triad of surgical excision, radiotherapy, and chemotherapy, there have been intensified research efforts studying immunotherapy as a more effective treatment paradigm [bib_ref] Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in..., Parsa [/bib_ref] [bib_ref] Cancer immunotherapy: Breaking thebarriers to harvest the crop, Pardoll [/bib_ref] [bib_ref] Potential role for STAT3 inhibitors in glioblastoma, Jackson [/bib_ref]. Immunotherapy such as cancer vaccines have shown encouraging results in phase I and II clinical trials, and have even demonstrated synergistic effects with standard therapy [bib_ref] Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant..., Sampson [/bib_ref] [bib_ref] Clinical responsiveness of glioblastoma multiforme to chemotherapy after vaccination, Wheeler [/bib_ref] [bib_ref] The combination of ionizing radiation and peripheral vaccination produces long-term survival of..., Newcomb [/bib_ref] [bib_ref] Synergy between chemotherapy and immunotherapy in the treatment of established murine solid..., Nowak [/bib_ref] [bib_ref] Sensitization of malignant glioma to chemotherapy through dendritic cell vaccination, Liu [/bib_ref]. Immunotherapy is a promising approach to complement and enhance the current chemoradiation regimen.
## Temozolomide: the current chemotherapeutic standard
TMZ is a second-generation DNA alkylating agent that induces thymine mispairing during DNA replication resulting in tumor cell G2/M phase arrest and autophagy [bib_ref] Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma, Stupp [/bib_ref] [bib_ref] Proautophagic drugs: A novel means to combat apoptosis-resistant cancers, with a special..., Lefranc [/bib_ref] [bib_ref] p53 effects both the duration of G2/M arrest and the fate of..., Hirose [/bib_ref]. In a correlative specimen analysis from EORTC 26981/NCIC CE.3, Hegi et al. found that those with a methylated MGMT promoter had markedly increased survival after receiving TMZ compared to those with unmethylated MGMT promoters, suggesting that stratification of individuals based on tumor MGMT methylation status may be a necessary part of future trials and treatment regimens [bib_ref] MGMT gene silencing and benefit from temozolomide in glioblastoma, Hegi [/bib_ref] [bib_ref] Temozolomide-Mediated radiation enhancement in glioblastoma: A report on underlying mechanisms, Chakravarti [/bib_ref]. TMZ is an oral medication that is capable of crossing the blood-brain barrier and accumulates in high concentrations in the brain and in highly angiogenic tumors such as GBM [bib_ref] Temozolomide-Mediated radiation enhancement in glioblastoma: A report on underlying mechanisms, Chakravarti [/bib_ref] [bib_ref] Disposition and pharmacokinetics of temozolomide in rat, Reyderman [/bib_ref] [bib_ref] Plasma and cerebrospinal fluid pharmacokinetics of intravenous temozolomide in non-human primates, Patel [/bib_ref] [bib_ref] A new model for prediction of drug distribution in tumor and normal..., Rosso [/bib_ref].
Initial studies in high grade gliomas showed that TMZ produced an objective response rate of 11% and an objective response of 47% [bib_ref] Multicentre CRC phase II trial of temozolomide in recurrent or progressive high-grade..., Bower [/bib_ref]. Further phase II studies continued to support TMZ's efficacy in providing moderate response rates (23.8%) in GBM with acceptable toxicities, even in those individuals with poor performance status and with relapsing GBM refractory to surgical and radiotherapy [bib_ref] Temozolomide as a second-line systemic regimen in recurrent high-grade glioma: A phase..., Brandes [/bib_ref] [bib_ref] Phase II study of temozolomide in patients with relapsing high grade glioma..., Janinis [/bib_ref] [fig_ref] Table 1: Phase II clinical trials measuring temozolomide efficacy in Glioblastoma [/fig_ref]. Yung and colleagues demonstrated the superiority of TMZ over procarbazine (PCB) in a phase II trial that became the foundation for FDA approval of TMZ as a first-line chemotherapeutic agent for GBM; they found that the 6-month overall survival rate for 112 patients receiving TMZ was 60% compared to 44% for 113 PCB-treated patients (p = 0.019) and the median progression free survival (PFS) with TMZ therapy was significantly improved over PCB weeks vs. 8.32 weeks, p = 0.0063) [bib_ref] A phase II study of temozolomide vs. procarbazine in patients with glioblastoma..., Yung [/bib_ref] [bib_ref] Current therapeutic paradigms in glioblastoma, Quick [/bib_ref]. Later studies built upon these findings, confirming the moderate clinical response and acceptable safety profile of TMZ in various sub-populations, such as the elderly and those at first-relapse [bib_ref] Multicenter phase II trial of temozolomide in patients with glioblastoma multiforme at..., Brada [/bib_ref] [bib_ref] Phase II study of temozolomide without radiotherapy in newly diagnosed glioblastoma multiforme..., Chinot [/bib_ref]. Robust population-based studies have shown that since the adoption of TMZ as a component of standard therapy for GBM, there has been modest but meaningful improvements in survival in the GBM patient population, with 2-year survival increasing from 7% in those cases diagnosed from [bib_ref] Adult glioblastoma multiforme survival in the temozolomide era: A population-based analysis of..., Darefsky [/bib_ref]. Importantly, in Stupp and colleagues definitively showed the superiority of adjunctive TMZ therapy when they found that radiotherapy plus TMZ resulted in a 2-year survival rate of 26.5% vs. [bib_ref] Glioma extent of resection and its impact on patient outcome, Sanai [/bib_ref].4% with radiotherapy alone [bib_ref] Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma, Stupp [/bib_ref]. Nevertheless, while the success of TMZ in clinical trials showed great promise for its overall efficacy, emerging TMZ resistance in the form of enhanced MGMT activity, acquired mismatch repair gene mutations, and the persistence of cancer stem cell sub-populations necessitated examination of resistance pathways and therapeutics for overcoming them [bib_ref] MGMT gene silencing and benefit from temozolomide in glioblastoma, Hegi [/bib_ref] [bib_ref] Overcoming temozolomide resistance in glioblastoma via dual inhibition of NAD+ biosynthesis and..., Goellner [/bib_ref] [bib_ref] MSH6 mutations arise in glioblastomas during temozolomide therapy and mediate temozolomide resistance, Yip [/bib_ref] [bib_ref] Chemotherapy resistance of glioblastoma stem cells, Eramo [/bib_ref] [bib_ref] Chemoresistance of glioblastoma cancer stem cells-Much more complex than expected, Beier [/bib_ref].
## Managing temozolomide resistance
Hegi and colleagues and subsequent studies established that MGMT promoter methylation is an important prognostic indicator of response to TMZ therapy [bib_ref] MGMT gene silencing and benefit from temozolomide in glioblastoma, Hegi [/bib_ref] [bib_ref] Overcoming temozolomide resistance in glioblastoma via dual inhibition of NAD+ biosynthesis and..., Goellner [/bib_ref] [bib_ref] O 6 -methylguanine DNA methyltransferase and p53 status predict temozolomide sensitivity in..., Hermisson [/bib_ref]. Patients with un-methylated MGMT experience decreased survival and responsiveness to TMZ therapy. In vitro studies introduced O 6 -benzylguanine as a candidate for overcoming TMZ resistance in tumors with unmethylated MGMT that is mediated by O 6 -alkylguanine DNA alkyltransferase (AGT), the protein encoded by gene MGMT [bib_ref] Effect of O 6 -Benzylguanine analogues on sensitivity of human tumor cells..., Dolan [/bib_ref] [bib_ref] 3-Aminobenzamide and/or O 6 -benzylguanine evaluated as an adjuvant to temozolomide or..., Wedge [/bib_ref] [bib_ref] Phase I trial of temozolomide plus O 6 -benzylguanine for patients with..., Quinn [/bib_ref]. An AGT substrate, O 6 -benzylguanine inactivates AGT and prohibits its DNA repair activity [bib_ref] Phase I trial of temozolomide plus O 6 -benzylguanine for patients with..., Quinn [/bib_ref]. In vitro and in vivo studies demonstrated that O 6 -benzylguanine-mediated AGT inhibition enhances TMZ cytotoxicity, and in a Phase II clinical trial Quinn et al. established the safety and efficacy of O 6 -benzylguanine in restoring TMZ sensitivity in TMZ-resistant anaplastic glioma with a 16% response rate [bib_ref] Effect of O 6 -Benzylguanine analogues on sensitivity of human tumor cells..., Dolan [/bib_ref] [bib_ref] 3-Aminobenzamide and/or O 6 -benzylguanine evaluated as an adjuvant to temozolomide or..., Wedge [/bib_ref] [bib_ref] Phase I trial of temozolomide plus O 6 -benzylguanine for patients with..., Quinn [/bib_ref] [bib_ref] Phase II trial of temozolomide plus O 6 -Benzylguanine in adults with..., Quinn [/bib_ref] [bib_ref] Role of O 6 -methylguanine-DNA methyltransferase in resistance of human brain tumor..., Bobola [/bib_ref] [bib_ref] Activity of temozolomide in the treatment of central nervous system tumor xenografts, Friedman [/bib_ref]. Other avenues for overcoming MGMT-mediated TMZ resistance in GBM involve dual inhibition of base excision repair and NAD+ biosynthesis; blocking base excision repair results in hyperactivation of poly(ADP-ribose) polymerase as the cell attempts to repair TMZmediated damage without the appropriate excision repair pathways, and the cell ultimately dies from energy depletion induced by the accumulation of repair intermediates [bib_ref] Overcoming temozolomide resistance in glioblastoma via dual inhibition of NAD+ biosynthesis and..., Goellner [/bib_ref]. Goellner and colleagues have found that TMZ-resistant GBM cells with elevated MGMT become sensitized to TMZ therapy after blocking both base excision repair and biosynthesis of the energy source, NAD+ [bib_ref] Overcoming temozolomide resistance in glioblastoma via dual inhibition of NAD+ biosynthesis and..., Goellner [/bib_ref]. Nevertheless, the most recent studies assessing the efficacy of O 6 -benzylguanine have not shown promising results, and the recent phase III RTOG 0525 trial failed to show that dose dense TMZ overcomes MGMTmediated TMZ resistance [bib_ref] RTOG 0525: A randomized phase III trial comparing standard adjuvant temozolomide (TMZ)..., Gilbert [/bib_ref].
Other sources of TMZ resistance include sub-populations of tumor cells with characteristics of cancer stem cells. Chen et al. have identified such a sub-population of endogenous glioma cells with characteristics of cancer stem cells that are responsible for transient bursts in proliferative cell populations and may account for GBM recurrence after standard therapy [bib_ref] A restricted cell population propagates glioblastoma growth after chemotherapy, Chen [/bib_ref]. Others have speculated that abnormalities in the apoptotic pathway in cancer stem cell-like subpopulations lead to little response to therapy and later cancer recurrence [bib_ref] Chemotherapy resistance of glioblastoma stem cells, Eramo [/bib_ref]. Regardless, the behavior of such subpopulations in conditions of chemotherapeutic stress remains a complex phenomenon attributed to both intrinsic and extrinsic factors that contribute to the tumor microenvironment [bib_ref] Chemoresistance of glioblastoma cancer stem cells-Much more complex than expected, Beier [/bib_ref].
## Radiation therapy
While the superiority of radiation plus surgery over radiation alone has been known for decades, more recently it was shown radiation plus TMZ is superior to radiation therapy alone [bib_ref] Current therapeutic paradigms in glioblastoma, Quick [/bib_ref]. Stupp and colleagues demonstrated in their phase III trial that concomitant and adjuvant TMZ with radiotherapy in GBM produced a 2-year survival rate of 26.5% versus 10.4% survival in those who received radiotherapy alone; in so doing they established the current standard of care involving both TMZ and radiotherapy along with surgical resection [bib_ref] Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma, Stupp [/bib_ref] [bib_ref] Promising survival for patients with newly diagnosed glioblastoma multiforme treated with concomitant..., Stupp [/bib_ref]. Population-based studies have shown the most pronounced improvements in median OS after the introduction of TMZ and combined radiation therapy compared to prior periods of TMZ or radiation therapy alone [bib_ref] Adult glioblastoma multiforme survival in the temozolomide era: A population-based analysis of..., Darefsky [/bib_ref].
Subsequent studies provided evidence that TMZ and radiotherapy act synergistically to promote tumor killing through TMZ-mediated radiation enhancement, specifically in those GBMs that have undetectable MGMT expression [bib_ref] Temozolomide-Mediated radiation enhancement in glioblastoma: A report on underlying mechanisms, Chakravarti [/bib_ref] [bib_ref] Novel radiation-enhancing agents in malignant gliomas, Zhang [/bib_ref]. It is suggested that TMZ enhances dsDNA breaks in a radiation-induced pro-apoptotic environment. Those who have MGMT positive tumors can also benefit from the synergism of TMZ and radiotherapy when O 6 -benzylguanine is added concurrently, sensitizing resistant cells to TMZ [bib_ref] Temozolomide-Mediated radiation enhancement in glioblastoma: A report on underlying mechanisms, Chakravarti [/bib_ref]. Interestingly, Rivera et al. suggest that MGMT methylation may also predict response to radiotherapy, with methylated MGMT tumors having an overall improved response to radiotherapy in the absence of alkylating chemotherapy, while unmethylated MGMT tumors actually progress during radiation treatment (71% vs. 42% of individuals stable or response on post radiotherapy scan) [bib_ref] MGMT promoter methylation is predictive of response to radiotherapy and prognostic in..., Rivera [/bib_ref].
While the standard dose of radiation therapy is 60 Gy, studies have shown that a maximum tolerated radiation dose of 75 Gy for locally aggressive tumors is safe and efficacious with concurrent TMZ therapy, resulting in no radiation necrosis and a median overall survival of 20.1 months in one study [bib_ref] Concurrent temozolomide and dose-escalated intensity-modulated radiation therapy in newly diagnosed glioblastoma, Tsien [/bib_ref]. In the elderly, 50 Gy in 28 fractions has been shown to optimize improvement in survival without a concomitant decline in cognition or quality of life (median OS 29.1 weeks vs. 16.9 weeks with supportive care alone) [bib_ref] Radiotherapy for glioblastoma in the elderly, Keime-Guibert [/bib_ref]. Retrospective analyses have shown that concomitant, rather than adjuvant, TMZ with radiotherapy may be the superior regimen, resulting in a median overall survival of 17 months vs. 14.8 months in the EORTC study involving concomitant and adjuvant TMZ [bib_ref] Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma, Stupp [/bib_ref] [bib_ref] Concomitant (without adjuvant) temozolomide and radiation to treat glioblastoma: A retrospective study, Sridhar [/bib_ref]. A recent phase III trial showed that in the elderly with newly diagnosed GBM, standard radiotherapy or hypofractionated radiotherapy did not confer increased survival over standard TMZ chemotherapy; as such, they recommended TMZ alone in the treatment of the elderly with GBM [bib_ref] Temozolomide versus standard 6-week radiotherapy versus hypofractionated radiotherapy in patients older than..., Malmstrom [/bib_ref].
## Current immunotherapies
An intact immune system will imprint, or edit, the tumor with an immunologic signature ultimately impervious to immune activity [bib_ref] Cancer immunoediting: Integrating immunity's roles in cancer suppression and promotion, Schreiber [/bib_ref]. Dunn and colleagues have described this process as the -Three Es‖ of immunoediting: Elimination, Equilibrium, and Escape [bib_ref] The three Es of cancer immunoediting, Dunn [/bib_ref]. Elimination involves immunosurveillance of the tumor, as the immune system destroys tumor cells it recognizes; equilibrium consists of the selection of tumor cells that cannot be destroyed as described above; escape reflects the process of aggressive resistant tumor growth. Facilitated by a tolerogenic tumor microenvironment that suppresses the cytotoxic activity of tumor infiltrating lymphocytes (TILs), poorly immunogenic tumors proliferate rapidly and often outpace even the destructive effects of chemotherapy and radiation. Targeted immunotherapies provide promise for combating tumor growth through immune activation of TILs and decreasing immune tolerance within the tumor microenvironment [bib_ref] Cancer immunotherapy comes of age, Topalian [/bib_ref].
Immunotherapy centers on the principle that the host immune system may destroy tumor if immune effector function is appropriately augmented [bib_ref] Cancer immunoediting: From immuno-surveillance to tumor escape, Dunn [/bib_ref]. The immune system may eliminate tumor cells through recognition and processing of tumor antigens, tumor antigen presentation by antigen presenting cells via major histocompatibility complexes (MHC), or signaling through IFN-γ pathway constituents. In addition, the tumor may establish a locally immunosuppressive state by secreting cytokines such as VEGF and TGF-β, expressing galectin and the enzyme indeolamine 2,3-dioxygenase which depletes the microenvironment of amino acids necessary for lymphocyte anabolism [bib_ref] Cancer immunoediting: Integrating immunity's roles in cancer suppression and promotion, Schreiber [/bib_ref]. Moreover, immunosuppressive cells such as Treg and myeloid derived suppressor cells (MDSCs) inhibit effector lymphocyte function. Tumor associated macrophages aid in tumor invasion and metastasis, and local regulatory B cells may produce pro-tumorigenic humoral responses. In addition to the aforementioned tumor strategies to evade the immune response, gliomas harbor unique immune evasion mechanisms. Such mechanisms include increased levels of immunosuppressive Tregs and decreased levels of effector CD8+ T cells in the glioma microenvironment, the lack or low level expression of co-stimulatory molecules in the brain, the expression of co-inhibitory molecules, such as B7-H1, on gliomas, and the presence of immunosuppressive glioma cancer stem cells that persist despite host anti-tumor immune responses [bib_ref] Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in..., Parsa [/bib_ref] [bib_ref] The role of glioma microenvironment in immune modulation: Potential targets for intervention, Dey [/bib_ref] [bib_ref] The role of human glioma-infiltrating microglia/macrophages in mediating the antitumor immune responses, Hussain [/bib_ref] [bib_ref] Immunobiological characterization of cancer stem cells isolated from glioblastoma patients, Di Tomaso [/bib_ref] [bib_ref] Modulation of T-cell activation by malignant melanoma initiating cells, Schatton [/bib_ref] [bib_ref] Glioblastoma cancer-initiating cells inhibit T cell proliferation and effector responses by the..., Wei [/bib_ref] [bib_ref] Glioma associated cancer-initiating cells induce immune suppression, Wei [/bib_ref].
There are three basic strategies underlying immunotherapy: immune modulating cytokine therapy, passive therapy and active immune therapy including cancer vaccines. These three strategies will be discussed as follows in the context of glioma and non-glioma tumor types, as many of these therapies have yet to be tested in glioma.
## Cytokine therapy
Cytokine therapy utilizes mediators of immune activation and proliferation to broadly induce an anti-tumor immune response. Cytokines that have been studied include gamma-chain cytokines, such as the interleukins IL-2, IL-7, IL-15, and IL-21 [bib_ref] Novel gamma-chain cytokines as candidate immune modulators in immune therapies for cancer, Fewkes [/bib_ref] ; specifically, IL-2 has been approved for metastatic renal cell carcinoma for inducing high rates of regression [bib_ref] Results of treatment of 255 patients with metastatic renal cell carcinoma who..., Fyfe [/bib_ref]. IL-12 is a non-gamma chain cytokine that has also been relevant in cancer immunotherapy. Importantly, while cytokine therapy is effective at immune activation, the immune effects are non-specific and often lead to extensive systemic toxicities, limiting their use [bib_ref] Novel gamma-chain cytokines as candidate immune modulators in immune therapies for cancer, Fewkes [/bib_ref].
Cytokine IL-7 is involved in T-cell development and is expressed by stromal and parenchymal cells [bib_ref] Interleukin-7 receptor expression: Intelligent design, Mazzucchelli [/bib_ref] [bib_ref] Selective expression of the interleukin 7 receptor identifies effector CD8 T cells..., Kaech [/bib_ref]. The effect of IL-7 in restoring the immune response and augmenting peripheral responses has been shown in pre-clinical models [bib_ref] Antitumor effects of interleukin-7 and adoptive immunotherapy on human colon carcinoma xenografts, Murphy [/bib_ref] [bib_ref] Interleukin 7 generates antitumor cytotoxic T lymphocytes against murine sarcomas with efficacy..., Jicha [/bib_ref] [bib_ref] Expression of murine interleukin 7 in a murine glioma cell line results..., Aoki [/bib_ref] [bib_ref] Intratumoral IL-7 delivery by mesenchymal stromal cells potentiates IFNgamma-transduced tumor cell immunotherapy..., Gunnarsson [/bib_ref] [bib_ref] Chemo-immunotherapy with oxaliplatin and interleukin-7 inhibits colon cancer metastasis in mice, Gou [/bib_ref]. A phase I study demonstrated safety of a vaccine against renal cell carcinoma (RCC) containing tumor cells expressing RCC26/IL-7/CD80; while they did not observe increased Th1 specific immune responses, they did find higher levels of Th2 mediated IL-10 expression in patients who responded [bib_ref] Allogeneic gene-modified tumor cells (RCC-26/IL-7/CD80) as a vaccine in patients with metastatic..., Westermann [/bib_ref]. Moreover, another phase I study in refractory solid cancers demonstrated improved circulating CD3+, CD4+ and CD8+ T-lymphocyte profiles with only mild toxicities [bib_ref] Phase I study of recombinant human interleukin-7 administration in subjects with refractory..., Sportes [/bib_ref]. A number of phase I clinical trials in metastatic melanoma, RCC, and other solid tumors are currently underway, but the clinical effect of IL-7 in human glioma remains to be seen.
Unlike IL-7, IL-15 is not a soluble cytokine and is bound via its receptor to cells such as monocytes and DCs [bib_ref] IL-15Ralpha recycles and presents IL-15 in trans to neighboring cells, Dubois [/bib_ref] [bib_ref] T cell-independent interleukin 15Ralpha signals are required for bystander proliferation, Lodolce [/bib_ref] [bib_ref] Interleukin (IL)-15R[alpha]-deficient natural killer cells survive in normal but not IL-15R[alpha]-deficient mice, Koka [/bib_ref] [bib_ref] Coordinate expression and trans presentation of interleukin (IL)-15Ralpha and IL-15 supports natural..., Burkett [/bib_ref] [bib_ref] Cutting edge: Murine dendritic cells require IL-15R alpha to prime NK cells, Koka [/bib_ref]. IL-15 is responsible for natural killer (NK) cell maturation and for the preservation of CD8+ memory T-cells, and this ability of inducing NK and CD8+ differentiation has garnered interest in its role in adoptive immunotherapy. Preclinical studies have shown increased potency of antigen-specific T-cells cultured with IL-15 in melanoma and plasmacytoma [bib_ref] IL-15 enhances the in vivo antitumor activity of tumor-reactive CD8+ T cells, Klebanoff [/bib_ref] [bib_ref] Failed adoptive immunotherapy with tumor-specific T cells: Reversal with low-dose interleukin 15..., Roychowdhury [/bib_ref] [bib_ref] Riviè re, I.; et al. Eradication of systemic B-cell tumors by genetically..., Brentjens [/bib_ref] [bib_ref] Differentiation of CD8+ T cells from tumor-invaded and tumor-free lymph nodes of..., Anichini [/bib_ref]. Phase I trials testing IL-15 are underway in a number of solid tumors, including melanoma, RCC, NSCLC, and squamous head and neck cancers.
IL-21 is produced only by activated CD4+ lymphocytes, but has stimulatory effects on CD4+, CD8+ and NK cells while inhibiting Treg function [bib_ref] IL-21 initiates an alternative pathway to induce proinflammatory T(H)17 cells, Korn [/bib_ref] [bib_ref] Essential autocrine regulation by IL-21 in the generation of inflammatory T cells, Nurieva [/bib_ref] [bib_ref] IL-21 blockade reduces graft-versus-host disease mortality by supporting inducible T regulatory cell..., Bucher [/bib_ref] [bib_ref] IL-21 limits NK cell responses and promotes antigen-specific T cell activation: A..., Kasaian [/bib_ref]. Preclinical studies with IL-21 show inhibition of tumor growth in melanoma and RCC and anti-tumor cytotoxic activity in murine genetically modified IL-21 secreting tumors [bib_ref] IL-21 induces tumor rejection by specific CTL and IFN-gamma-dependent CXC chemokines in..., Carlo [/bib_ref] [bib_ref] IL-21 activates both innate and adaptive immunity to generate potent antitumor responses..., Ma [/bib_ref] [bib_ref] Local IL-21 promotes the therapeutic activity of effector T cells by decreasing..., Kim-Schulze [/bib_ref] [bib_ref] Immunotherapy of neuroblastoma by an Interleukin-21-secreting cell vaccine involves survivin as antigen, Croce [/bib_ref] [bib_ref] Enhancing therapy of B16F10 melanoma efficacy through tumor vaccine expressing GPI-anchored IL-21..., Zhao [/bib_ref]. A recent phase II study in metastatic melanoma treated with IL-21 demonstrated an overall response rate of 22.5%, median overall survival of 12.4 months and median progression free survival of 4.3 months [bib_ref] Interleukin-21 has activity in patients with metastatic melanoma: A phase II study, Petrella [/bib_ref]. Interestingly, given the dual effect of IL-21 in stimulating effector lymphocytes and inhibiting Th17 mediated responses, the combination of IL-21 and anti-PD-1 antibody is currently being studied in a phase I dose escalation study in advanced or metastatic solid cancers, and IL-21 combined with ipilimumab is being studied in a phase I trial in advanced melanoma (NCT01629758, NCT01489059). Preclinical studies of IL-21 are needed in GBM, but the clinical possibilities of combination immunotherapy are promising.
Produced by APCs, IL-12 is responsible for inducing Th1 immune responses and augmenting proliferation of NK and CD8+ T-cells, as well as engendering the production of IFN-γ, a marker of T-cell activation [bib_ref] Interleukin-12: A cytokine at the interface of inflammation and immunity, Trinchieri [/bib_ref] [bib_ref] Identification and purification of natural killer cell stimulatory factor (NKSF), a cytokine..., Kobayashi [/bib_ref]. IL-12 therapy has resulted in increased tumor rejection and local inflammatory responses in mouse models of colon cancer, glioma, and hematologic malignancy [bib_ref] Combination gene therapy of IL-12 and allogeneic MHC class I gene via..., Yoon [/bib_ref] [bib_ref] Paracrine delivery of IL-12 against intracranial 9L gliosarcoma in rats, Dimeco [/bib_ref] [bib_ref] Antitumoral vaccination with granulocyte-macrophage colony-stimulating factor or interleukin-12-expressing DHD/K12 colon adenocarcinoma cells, Lechanteur [/bib_ref] [bib_ref] Interleukin-12 (IL-12) gene therapy of leukemia: Immune and anti-leukemic effects of IL-12-transduced..., Gautam [/bib_ref]. Early clinical trials in ovarian cancer with intraperitoneal IL-12 plasmid vaccination have demonstrated safety and feasibility of delivery via a plasmid vector [bib_ref] Phase I trial of a formulated IL-12 plasmid in combination with carboplatin..., Anwer [/bib_ref] [bib_ref] Phase-I clinical trial of IL-12 plasmid/lipopolymer complexes for the treatment of recurrent..., Anwer [/bib_ref].
While cytokine therapy showed some potential in pre-clinical studies, initial clinical results in tumor regression have been disappointing. This may be due to the multiplicity of mechanisms of tumor immune evasion in the tumor microenvironment, such as immune checkpoint interactions, that continue to suppress immune activity despite the effects of cytokine therapy. Alternatively, local delivery of cytokine may be necessary to deliver a high enough intratumoral dose to be effective.
## Passive therapy: immune checkpoint blockade and adoptive cell therapy
Passive anticancer therapy utilizes antibodies to mediate cancer killing, either by targeting tumor antigen or preventing immune checkpoint-mediated inhibition. Antibody therapy that targets specific tumor antigen may promote tumor killing in one of three ways: 1. Direct receptor-mediated tumor cell apoptosis, 2. Immune-mediated tumor killing through macrophage phagocytosis, complement mediated destruction, or antigen presentation leading to T-cell activation, 3. Antagonism of vascular receptors and the destruction of vascular stromal cells, leading to cell death in the tumor parenchyma [bib_ref] Antibody therapy of cancer, Scott [/bib_ref]. A number of antibodies have been FDA approved for the treatment of cancers, including antibodies conjugated to chemotherapeutic agents or radioisotopes as well as unconjugated antibodies that target pro-oncogenic factors, such as growth receptors [bib_ref] Monoclonal antibodies: Versatile platforms for cancer immunotherapy, Weiner [/bib_ref] [bib_ref] Antibodies in oncology, Pillay [/bib_ref]. One of the obstacles to successful antibody therapy, however, is immune escape of the tumor through the establishment of a locally immunosuppressive tumor microenvironment. A mechanism to overcome this challenge is the development of antibodies that target inhibitory immune pathways, termed immune checkpoints.
Antibody-mediated immune checkpoint blockade is one type of passive therapy that promotes immune effector function. Under normal circumstances, immune checkpoints are responsible for preventing autoimmunity during peripheral inflammatory responses. When bound to their respective ligands, immune checkpoint receptors may down-regulate or up-regulate immune activity through intracellular signal cascades. Malignant tumors may hijack this machinery by expressing ligands to inhibitory immune checkpoints, thereby depressing the activity of immune effector cells that may otherwise eradicate tumor. Examples of such inhibitory immune checkpoints include programmed death-1 (PD-1) and cytotoxic lymphocyte antigen-4 (CTLA-4) [bib_ref] The blockade of immune checkpoints in cancer immunotherapy, Pardoll [/bib_ref].
As immunotherapy gains traction as a putative treatment for glioblastoma, immune checkpoint inhibitors are gaining considerable attention as they show promise for the immunologic elimination of poorly immunogenic tumors. Parsa and colleagues have demonstrated a mechanism for GBM escape through loss of tumor suppressor gene PTEN, which results in overexpression of immunosuppressive B7 homolog 1 (B7-H1), also known as programmed death ligand-1 (PDL1) [bib_ref] Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in..., Parsa [/bib_ref] [bib_ref] Implications for immunotherapy of tumor-mediated T-cell apoptosis associated with loss of the..., Waldron [/bib_ref]. PDL1 is one component of an array of inhibitory immune checkpoint proteins that may be targeted to prevent immune dysregulation during tumor growth [bib_ref] The blockade of immune checkpoints in cancer immunotherapy, Pardoll [/bib_ref]. PDL1 expressed on host tissues binds to PD-1 receptor on activated lymphocytes in the periphery to prevent autoimmunity during inflammatory responses; engagement of PDL1 with PD-1 inhibits T-cell activation by inhibiting kinases via phosphatase SHP2 [bib_ref] Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member..., Freeman [/bib_ref]. PD-1 is also highly expressed in regulatory T-cells (Treg) [bib_ref] The blockade of immune checkpoints in cancer immunotherapy, Pardoll [/bib_ref]. Through overexpression of PDL1 in GBM, tumor cells may exhaust activated T-cells, thereby precluding a robust immune response. Zeng and colleagues have reported that antibody blockade of PD-1 in mice with intracranial glioma combined with stereotactic radiosurgery (SRS) results in significantly improved survival [bib_ref] Anti-PD-1 blockade and stereotactic radiation produce long-term survival in mice with intracranial..., Zeng [/bib_ref]. While PD-1 alone provides modest survival benefit, the addition of SRS is key to eliciting a synergistic anti-tumor response in the mouse glioma model [bib_ref] Anti-PD-1 blockade and stereotactic radiation produce long-term survival in mice with intracranial..., Zeng [/bib_ref]. The underlying mechanism may be that SRS modifies the tumor microenvironment to produce inflammation, augmenting the anti-tumor immune response [bib_ref] Combining radiotherapy and immunotherapy: A revived partnership, Demaria [/bib_ref]. A phase I/II trial of anti-PD-1 therapy in relapsed GBM is currently underway (NCT01952769).
In addition to PD-1, immune checkpoint receptor CTLA-4 has begun to generate interest in its application as a part of combination immunotherapy. Located on T-cells, inhibitory CTLA-4 outcompetes stimulatory T-cell receptor CD28 for binding to their ligands CD80 and CD86, thus inhibiting T-cell effector function [bib_ref] Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member..., Freeman [/bib_ref]. Unlike PD-1, CTLA-4 acts largely on naï ve and resting T-cells. Ultimately, CTLA-4 downregulates helper T-cell activity and augments Treg activity [bib_ref] Engagement of the PD-1 immunoinhibitory receptor by a novel B7 family member..., Freeman [/bib_ref]. Demaria and colleagues tested the effects of CTLA-4 blockade and SRS on breast cancer in mice; they reported that CTLA-4 blockade synergizes with SRS to significantly prolong survival, and that this survival benefit and prevention of distant metastases is largely mediated by CD8 T-cells [bib_ref] Immune-Mediated inhibition of metastases after treatment with local radiation and CTLA-4 blockade..., Demaria [/bib_ref]. Moreover, Wolchok et al. demonstrated that in a clinical trial of patients with melanoma, concurrent PD-1 and CTLA-4 blockade resulted in a 53% response rate with significant tumor regression and a manageable safety profile-this response rate far exceeded anything previously encountered with anti-PD-1 or anti-CTLA-4 monotherapy [bib_ref] Nivolumab plus ipilimumab in advanced melanoma, Wolchok [/bib_ref]. It is important to note, however, that the systemic autoimmune toxicities associated with checkpoint blockade are significant, including enterocolitis and hypophysitis [bib_ref] Nivolumab plus ipilimumab in advanced melanoma, Wolchok [/bib_ref] [bib_ref] Cancer regression and autoimmunity induced by cytotoxic T lymphocyte-associated antigen 4 blockade..., Phan [/bib_ref]. Nevertheless, clinical results in immune checkpoint blockade in melanoma and other tumors are encouraging; these findings support further preclinical study of checkpoint inhibitors in glioblastoma as potential future therapeutics.
A second form of passive immunotherapy is adoptive T-cell therapy, in which tumor-specific T-cells are activated ex vivo and transferred to the patient [bib_ref] Immunotherapy and biologic modifiers for the treatment of malignant brain tumors, Marras [/bib_ref] [bib_ref] T cell adoptive immunotherapy of newly diagnosed gliomas, Plautz [/bib_ref] [bib_ref] Effects of local injection of ex vivo expanded autologous tumor-specific T lym-phocytes..., Tsuboi [/bib_ref]. As the immune cells are activated before coming into contact with tumor, it is less likely that they will succumb to the inhibitory effects of the tumor microenvironment. Transfer of tumor reactive T-cells to patients with melanoma have led to objective responses of 34%-50% [bib_ref] Treatment of patients with metastatic melanoma with autologous tumor-infiltrating lymphocytes and interleukin..., Rosenberg [/bib_ref] [bib_ref] Clinical responses in a phase II study using adoptive transfer of short-term..., Besser [/bib_ref]. Objective response rates up to 30% have been observed in patients with melanoma receiving adoptive T-cell transfers expressing genetically altered T-cell receptors (TCRs) that have higher affinities for MHC-antigen complexes [bib_ref] Gene therapy with human and mouse T-cell receptors mediates cancer regression and..., Johnson [/bib_ref]. A number of phase I and II clinical studies have been completed or are underway testing adoptive T-cell therapies in GBM (NCT00331526, NCT01082926, NCT00693095, NCT01588769, NCT01454596, NCT00730613, NCT01109095).
## Active therapy: vaccines
Vaccine therapy aims to break immune tolerance of tumors by exposing antigen presenting cells (APCs) to immunogenic tumor peptides in the hopes of activating a broad lymphocytic immune response involving both CD4+ and CD8+ T-cells. Recent advances in cancer vaccine therapy have been achieved in hormone-resistant prostate cancer and advanced melanoma [bib_ref] Sipuleucel-T immunotherapy for castration-resistant prostate cancer, Kantoff [/bib_ref] [bib_ref] gp100 peptide vaccine and interleukin-2 in patients with advanced melanoma, Schwartzentruber [/bib_ref].
Cancer vaccines have emerged as a therapy that offers the following unique advantages: vaccine-stimulated immune cells (1) have high specificity for tumor, (2) are able to kill target cells in all stages of the cell cycle, unlike checkpoint inhibitors, (3) can target tumor cells with intrinsic or acquired drug resistance, a major obstacle for current chemotherapeutic regimen [bib_ref] Chemotherapy resistance of glioblastoma stem cells, Eramo [/bib_ref] [bib_ref] Drug resistance in cancer: A perspective, Goldie [/bib_ref] , (4) can confer immunologic memory and durable long-term protection, and (5) can potentially be used in the future for preventative immunization in high-risk populations [bib_ref] Sensitization of malignant glioma to chemotherapy through dendritic cell vaccination, Liu [/bib_ref] [bib_ref] Cancer vaccines: Between the idea and the reality, Finn [/bib_ref]. Immunotherapy can also be dynamic in that the immune response can adapt to changes in antigen expression of the tumor, a phenomenon known as epitope spreading (REF).
## Dendritic cell vaccine
Dendritic cells are the most potent APCs of the immune system with the ability to induce lasting immunity in naive individuals [bib_ref] Dendritic cells and the control of immunity, Banchereau [/bib_ref] [bib_ref] Glioma-associated hyaluronan induces apoptosis in dendritic cells via inducible nitric oxide synthase:..., Yang [/bib_ref] [bib_ref] Dendritic cell vaccines to combat glioblastoma, Wheeler [/bib_ref]. DC vaccines are capable of inducing systemic immunity through a number of different ex vivo and in vivo approaches, including DCs loaded with tumor lysate, MHC I derived peptides, or tumor apoptotic bodies as well as direct intratumoral injection [bib_ref] In vitro immunization of patient T cells with autologous bone marrow antigen..., Coulon [/bib_ref] [bib_ref] Murine dendritic cells pulsed with whole tumor lysates mediate potent antitumor immune..., Fields [/bib_ref] [bib_ref] Immunotherapy with dendritic cells and tumor major histocompatibility complex class I-derived peptides..., Rawson [/bib_ref] [bib_ref] Generation of tumor-specific T lymphocytes by cross-priming with human dendritic cells ingesting..., Hoffmann [/bib_ref]. Because DCs have the capability of priming the immune system through tumor antigen presentation to T-cells, DC vaccines are capable of orchestrating a robust cell-mediated immune response against infiltrating tumor cells. By exposing DCs to tumor antigen either ex vivo or in vivo, DC vaccines facilitate tumor destruction via the body's own adaptive immune response.
In an early study, Mayordomo and colleagues demonstrated that DCs pulsed with MHC I restricted tumor-associated peptides initiated a T-cell immune response that was protective after subsequent tumor challenge and induced 80% regression in established C3 sarcoma and 3LL lung carcinoma mouse models [bib_ref] Bone marrow-derived dendritic cells pulsed with synthetic tumour peptides elicit protective and..., Mayordomo [/bib_ref]. Subsequently, Liau et al. produced prolonged survival in a rat glioma model treated with DCs pulsed with various tumor antigens [bib_ref] Treatment of intracranial gliomas with bone marrow-derived dendritic cells pulsed with tumor..., Liau [/bib_ref]. It was quickly recognized that the primary limitation of DC vaccines is the necessity of direct physical interaction of DCs and tumor antigens, preferably ex vivo as DCs are largely incapable of processing tumor antigens in the immunosuppressive tumor microenvironment [bib_ref] Glioma-associated hyaluronan induces apoptosis in dendritic cells via inducible nitric oxide synthase:..., Yang [/bib_ref] [bib_ref] Cutting edge: Physical interaction between dendritic cells and tumor cells results in..., Celluzzi [/bib_ref].
Nevertheless, a series of phase I and II clinical trials proved promising for DC vaccination as an effective immunotherapy for GBM [fig_ref] Table 2: Phase I and II clinical trials measuring cancer vaccine safety and efficacy... [/fig_ref]. Phase I trials demonstrated enhanced antigen-specific cytoxotic T-cell responses in both peptide-and tumor lysate-pulsed vaccine formats [bib_ref] Vaccination of malignant glioma patients with peptide-pulsed dendritic cells elicits systemic cytotoxicity..., Yu [/bib_ref] [bib_ref] Vaccination with tumor lysate-pulsed dendritic cells elicits antigen-specific, cytotoxic T-Cells in patients..., Yu [/bib_ref] [bib_ref] Dendritic cell vaccination in glioblastoma patients induces systemic and intracranial T-cell responses..., Liau [/bib_ref] [bib_ref] Clinical evaluation of dendritic cell vaccination for patients with recurrent glioma: Results..., Yamanaka [/bib_ref]. Specifically, Yu et al. found that treatment of malignant glioma with tumor lysate-pulsed DC vaccine produced a median survival of 133 weeks, while Liau and colleagues showed that a robust intratumoral cytotoxic T-cell response was elicited after treatment with peptide-pulsed DC vaccination that positively correlated with survival (p = 0.047) [bib_ref] Dendritic cell vaccination in glioblastoma patients induces systemic and intracranial T-cell responses..., Liau [/bib_ref]. Of note, Liau et al. also found that TGF-β2 expression negatively correlated with the degree of intratumoral T-cell infiltrate as well as survival and clinical response, possibly due to its immunosuppressive effects on T-lymphocytes [bib_ref] Dendritic cell vaccination in glioblastoma patients induces systemic and intracranial T-cell responses..., Liau [/bib_ref]. Moreover, Yamanaka and colleagues showed an enhanced overall survival of 480 days in those with grade 4 glioma treated with tumor lysate-pulsed DC vaccine compared to 400 days in the control group, with the most improved survival in those receiving both intradermal and intratumoral vaccination; such differences in survival could be attributed to activation of both central and peripheral immune responses [bib_ref] Clinical evaluation of dendritic cell vaccination for patients with recurrent glioma: Results..., Yamanaka [/bib_ref]. An important step in the development of DC vaccine therapy for GBM occurred in a study by Wheeler et al. which demonstrated correlated immune and clinical responses in those with grade 4 glioma. They found that 53% of GBM patients treated with vaccine exhibited ≥1.5 fold cytokine responses and statistically significant time to progression and time to survival as well as a 2-year survival rate of 41% (p = 0.03) [bib_ref] Vaccination elicits correlated immune and clinical responses in glioblastoma multiforme patients, Wheeler [/bib_ref].
Another potential limitation of DC vaccines was the assumption that multi-antigenic vaccines would result in catastrophic autoimmunity, and that the safest vaccines would utilize single tumor antigens. However, a number of studies have incorporated multi-antigenic DC vaccines without the predicted autoimmunity and have demonstrated clinical anti-tumor effects [bib_ref] Vaccination of malignant glioma patients with peptide-pulsed dendritic cells elicits systemic cytotoxicity..., Yu [/bib_ref] [bib_ref] Vaccination with tumor lysate-pulsed dendritic cells elicits antigen-specific, cytotoxic T-Cells in patients..., Yu [/bib_ref] [bib_ref] An epidermal growth factor receptor variant III-targeted vaccine is safe and immunogenic..., Sampson [/bib_ref] [bib_ref] Greater chemotherapy-induced lymphopenia enhances tumor-specific immune responses that eliminate EGFRvIII-expressing tumor cells..., Sampson [/bib_ref] [bib_ref] Phase I trial of a personalized peptide vaccine for patients positive for..., Terasaki [/bib_ref]. One of these studies is a phase I trial that showed that ICT-107, an autologous multi-antigenic DC vaccine utilizing six GBM peptides, produced a median overall survival of 38.4 months and a progression free survival of 16.9 months; a phase II trial studying ICT-107 is currently underway [bib_ref] Phase I trial of a personalized peptide vaccine for patients positive for..., Terasaki [/bib_ref]. In all, DC vaccines show promising immune and clinical responses in early clinical trials, but phase III trials are needed to confirm the clinical efficacy of such immunotherapy. Moreover, whether the evolving tumor microenvironment overcomes the immunity induced by DC vaccination remains to be seen [bib_ref] The three Es of cancer immunoediting, Dunn [/bib_ref] [bib_ref] Dendritic cell vaccines to combat glioblastoma, Wheeler [/bib_ref] [bib_ref] Dendritic cell immunotherapy: Mapping the way, Figdor [/bib_ref] [bib_ref] Cancer immunotherapy: Moving beyond current vaccines, Rosenberg [/bib_ref].
## Egfrviii vaccine
A promising antigen for peptide vaccination is epidermal growth factor receptor (EGFR), a transmembrane tyrosine kinase receptor that is rearranged and overexpressed in nearly half of malignant gliomas, resulting in structural rearrangements and enhanced oncogenicity [bib_ref] Structural alterations of the epidermal growth factor receptor gene in human gliomas, Wong [/bib_ref] [bib_ref] Increased expression of the epidermal growth factor receptor gene in malignant gliomas..., Wong [/bib_ref] [bib_ref] Epidermal growth factor receptor VIII peptide vaccination is efficacious against established intracerebral..., Heimberger [/bib_ref] [bib_ref] Receptor dimerization is not a factor in the signalling activity of atransforming..., Chu [/bib_ref]. The most commonly observed mutation of EGFR in human neoplasms is variant III (EGFRvIII), contributing to increased tumorogenicity in primary and secondary brain tumors [bib_ref] Receptor dimerization is not a factor in the signalling activity of atransforming..., Chu [/bib_ref] [bib_ref] Monoclonal antibodies against EGFRvIII are tumor specific and react with breast and..., Wikstrand [/bib_ref] [bib_ref] Mutant epidermal growth factor receptor (EGFRvIII) contributes to head and neck cancer..., Sok [/bib_ref] [bib_ref] Evidence of high incidence of EGFRvIII expression and coexpression with EGFR in..., Ge [/bib_ref] [bib_ref] Anti-synthetic peptide antibody reacting at the fusion junction of deletion-mutant epidermal growth..., Humphrey [/bib_ref] [bib_ref] Epidermal growth factor ligand-independent, unregulated, cell-transforming potential of a naturally occurring human..., Batra [/bib_ref] [bib_ref] Mutant epidermal growth factor receptor up-regulates molecular effectors of tumor invasion, Lal [/bib_ref]. Wong and colleagues were the first to describe increased expression of the EGFR gene in gliomas with amplified EGFR as well as characterize the structural rearrangements of EGFR in malignant gliomas [bib_ref] Structural alterations of the epidermal growth factor receptor gene in human gliomas, Wong [/bib_ref] [bib_ref] Increased expression of the epidermal growth factor receptor gene in malignant gliomas..., Wong [/bib_ref]. found that the expression of mutant EGFRvIII in GBM is an independent negative prognostic marker, and tumor specific antibodies to EGFRvIII have been found in breast cancer, head and neck cancers, and other tumors, with the absence of EGFRvIII in normal tissues [bib_ref] Epidermal growth factor receptor VIII peptide vaccination is efficacious against established intracerebral..., Heimberger [/bib_ref] [bib_ref] Monoclonal antibodies against EGFRvIII are tumor specific and react with breast and..., Wikstrand [/bib_ref] [bib_ref] Mutant epidermal growth factor receptor (EGFRvIII) contributes to head and neck cancer..., Sok [/bib_ref] [bib_ref] Evidence of high incidence of EGFRvIII expression and coexpression with EGFR in..., Ge [/bib_ref] [bib_ref] Anti-synthetic peptide antibody reacting at the fusion junction of deletion-mutant epidermal growth..., Humphrey [/bib_ref]. Moreover, a number of studies have established EGFRvIII as an oncogene, enhancing tumorigenicity through increased cell motility, migration, and consequent invasiveness [bib_ref] Epidermal growth factor ligand-independent, unregulated, cell-transforming potential of a naturally occurring human..., Batra [/bib_ref] [bib_ref] Mutant epidermal growth factor receptor up-regulates molecular effectors of tumor invasion, Lal [/bib_ref] [bib_ref] Constitutive EGFR signaling confers a motile phenotype to neural stem cells, Boockvar [/bib_ref]. Increased expression of EGFRvIII promotes cell proliferation and inhibits apoptosis, which has been shown to confer chemo-and radio-resistance in a number of different tumors [bib_ref] A common mutant epidermal growth factor receptor confers enhanced tumorigenicity on human..., Nagane [/bib_ref] [bib_ref] Expression of oncogenic epidermal growth factor receptor family kinases induces paclitaxel resistance..., Montgomery [/bib_ref] [bib_ref] Radiation-induced activation of a common variant of EGFR confers enhanced radioresistance, Lammering [/bib_ref].
Given that EGFRvIII is expressed exclusively in malignant tissue and is responsible for enhanced tumorigenicity, it is a logical target for immunotherapy. Preclinical data in mice showed that peptide vaccination against the mutated segment of EGFRvIII (PEP-3) prevented melanoma tumor development in 70% and that sera from immunized mice was protective in 31% of non-immunized mice implanted with tumor. Moreover, mice treated with DC vaccine pulsed with EGFRvIII produced an antigen-specific immune response, leading to long-term anti-tumor immunity [bib_ref] Epidermal growth factor receptor VIII peptide vaccination is efficacious against established intracerebral..., Heimberger [/bib_ref] [bib_ref] Dendritic cells pulsed with a tumor-specific peptide induce long-lasting immunity and are..., Heimberger [/bib_ref]. Clinical trials have shown EGFRvIII-specific humoral responses are detectable in GBM patients vaccinated with PEP-3 and that anti-EGFVRvIII DC vaccines are safe and efficacious immunotherapeutic alternatives or adjuncts to standard therapy [bib_ref] An epidermal growth factor receptor variant III-targeted vaccine is safe and immunogenic..., Sampson [/bib_ref] [fig_ref] Table 2: Phase I and II clinical trials measuring cancer vaccine safety and efficacy... [/fig_ref]. A recent phase II clinical trial demonstrated a significant increase in median overall survival (26 months, p = 0.0013) after vaccination against EGFRvIII in GBM compared to non-vaccinated historical controls [bib_ref] Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant..., Sampson [/bib_ref]. However, the same trial also showed immunologic escape as 82% of individuals had lost tumoral EGFRvIII expression at GBM recurrence. Currently, a phase III clinical trial (ACT IV) is underway studying the effects of CDX-110 (PEPvIII-KLH vaccine) in newly diagnosed, EGFRvIII positive GBM. This is a randomized, double-blind trial in which patients are randomized to either the CDX-110 treatment arm or the keyhole limpet hemocyanin (KLH) control arm; both arms will receive standard TMZ therapy and will be followed until death. The ReAct trial is a phase II study currently underway examining the effects of CDX-110 in recurrent GBM refractory to bevacizumab therapy. Patients with recurrent GBM who have never been treated with bevacizumab are randomized to receive either CDX-110 or the control KLH, while patients who are refractory to bevacizumab will receive CDX-110 and bevacizumab. Patients are followed for survival.
Preclinical and early clinical data regarding the EGFRvIII vaccine is encouraging. In addition to EGFR, others have raised the possibility of personalized peptide vaccines or even individualized whole tumor vaccines as more targeted alternatives for immunotherapy [bib_ref] Phase I trial of a personalized peptide vaccine for patients positive for..., Terasaki [/bib_ref] [bib_ref] Immunologic evaluation of personalized peptide vaccination for patients with advanced malignant glioma, Yajima [/bib_ref] [bib_ref] Antigenic profiling of glioma cells to generate allogeneic vaccines or dendritic cell-based..., Zhang [/bib_ref] ].
## Heat shock proteins
Heat shock proteins (HSPs) are chaperones responsible for facilitating protein folding and are overexpressed during cellular environmental stress, such as hyperthermia, UV irradiation, or oxidative stress [bib_ref] A new puffing pattern induced by temperature shock and DNP in drosophila, Ritossa [/bib_ref] [bib_ref] HSP82 is an essential protein that is required in higher concentrations for..., Borkovich [/bib_ref] [bib_ref] Interaction of HSP-70 with newly synthesized proteins-Implications for protein folding and assembly, Beckmann [/bib_ref]. Moreover, HSPs are responsible for mediating apoptotic pathways and capable of stimulating antigen-specific cytotoxic T-cell responses through the action of APCs [bib_ref] Selective depletion of inducible HSP70 enhances immunogenicity of rat colon cancer cells, Gurbuxani [/bib_ref] [bib_ref] Heat shock protein vaccines: Form bench to bedside, Binder [/bib_ref] [bib_ref] Heat shock proteins: The -swiss army knife‖ vaccines against cancers and infectious..., Srivastava [/bib_ref] [bib_ref] Heat-shock protein vaccines against cancer, Blachere [/bib_ref] [bib_ref] Cellular-requirements for tumor-specific immunity elicited by heat-shock proteins-Tumor rejection antigen-GP96 primes CD8..., Udono [/bib_ref]. Given that they are highly immunogenic and tumor-specific, HSPs are attractive antigens for vaccine therapy [bib_ref] Heat shock protein vaccines as active immunotherapy against human gliomas, Yang [/bib_ref]. Heat shock protein-96 peptide complex (HSPPC-96) has been developed as an autologous tumor antigen peptide vaccination for GBM, and is currently being tested in phase II clinical trials for those with recurrent and newly diagnosed GBM. Preliminary trial results demonstrate a median survival of those treated with HSPPC-96 of 44 weeks compared to the historical median survival of 26 weeks [bib_ref] Autologous heat shock protein vaccine for patients with newly diagnosed and recurrent..., Parsa [/bib_ref]. Such results are encouraging for the future use of HSPs in potent anti-cancer vaccines alongside standard therapy.
## Combining vaccine immunotherapy with standard of care
The persistence of GBM refractory to chemotherapy and radiation has provided the impetus for exploring other therapeutic avenues, including new antigens for vaccines and multi-modality therapy that combines immunotherapy with the current standard of care. While cytotoxic therapy may decrease tumor volume, survival remains poor as a result of the resilient GBM tumor microenvironment. As described previously, the tumor cell expression of molecules that inhibit immune effector function preclude tumor elimination. Immunotherapy provides promise to overcome such immunologic obstacles [bib_ref] Cancer immunoediting: Integrating immunity's roles in cancer suppression and promotion, Schreiber [/bib_ref] [bib_ref] Cancer immunotherapy comes of age, Topalian [/bib_ref]. The following sections will explore novel antigens for peptide vaccines as well as the promise of combination immunotherapy alongside standard of care.
## Peptide vaccines
## Novel peptide vaccines
Peptide vaccines consist of small peptide antigens that are expressed by the target tumor cells and are injected subcutaneously. Capable of creating complexes with HLA class I antigen, these peptides are digested by host APCs and subsequently presented to cytotoxic T-lymphocytes (CTLs) in the lymph nodes. Once sensitized and after clonal expansion, the CTLs disseminate systemically and destroy tumor upon recognition of the peptide-HLA complex correlate on tumor cells. In this manner, peptide vaccines prime the adaptive immune system to eliminate tumor and promote tumor regression. Use of normal gene products as vaccine targets carries an inherent risk of autoimmune toxicity. However, several antigens that are expressed in immunoprivileged or tissue-specific sites of the body have been characterized as potentially effective antitumor targets [bib_ref] EGFRvIII-targeted vaccination therapy of malignant glioma, Choi [/bib_ref] [bib_ref] The promise of cancer vaccines, Gilboa [/bib_ref]. These new antigens could be targets for future therapies . . Selected tumor antigens targeted in glioma vaccination studies.
## Antigen significance examples of use in glioma vaccines
TRP-2 (tyrosinase-related protein-2)
Human melanoma-derived tissue differentiation antigen present in 50% of GBM-derived cell lines [bib_ref] Identification of tyrosinase-related protein 2 as a tumor rejection antigen for the..., Bloom [/bib_ref]. Overexpression associated with drug and radiation resistance.
Wheeler, et al. [bib_ref] Clinical responsiveness of glioblastoma multiforme to chemotherapy after vaccination, Wheeler [/bib_ref] Liu, et al. [bib_ref] Molecular and functional analysis of tyrosinase-related protein (TRP)-2 as a cytotoxic T..., Liu [/bib_ref] Liu, et al. [bib_ref] Cytotoxic T cell targeting of TRP-2 sensitizes human malignant glioma to chemotherapy, Liu [/bib_ref] HER2 Well-defined oncogenic protein and tumor antigen present with high frequency in breast, ovarian, renal cell carcinoma, and colon cancers. Documented expression in human GBM cells and recognized by cytotoxic T cells [bib_ref] AIM-2: A novel tumour antigen expressed and presented by human glioma cells, Liu [/bib_ref] Phuphanich, et al. [bib_ref] Phase I trial of a multi-epitope-pulsed dendritic cell vaccine for patients with..., Phuphanich [/bib_ref] Yu, et al. [bib_ref] Vaccination with tumor lysate-pulsed dendritic cells elicits antigen-specific, cytotoxic T-Cells in patients..., Yu [/bib_ref] gp100 (human melanoma-associated antigen)
Highly immunogenic antigen in melanoma found to be expressed in GL26 and GL261 glioma cell lines [bib_ref] Immunotherapeutic targeting of shared melanoma-associated antigens in a murine glioma model, Prins [/bib_ref] and recognized by CTLs [bib_ref] Expression of interleukin-13 receptor alpha2 in glioblastoma multiforme: Implications for targeted therapies, Jarboe [/bib_ref] Phuphanich, et al. [bib_ref] Phase I trial of a multi-epitope-pulsed dendritic cell vaccine for patients with..., Phuphanich [/bib_ref] Okada, et al. [bib_ref] Pathways of apoptotic and non-apoptotic death in tumour cells, Okada [/bib_ref] Yu, et al. [bib_ref] Vaccination with tumor lysate-pulsed dendritic cells elicits antigen-specific, cytotoxic T-Cells in patients..., Yu [/bib_ref] MAGE-1 (melanoma-associated antigen) Immune-stimulating testis tumor antigen group with four-fold higher expression in GBM than normal astrocytes. Overexpression associated with chemotherapy resistance in ovarian [bib_ref] Overexpression of MAGE/GAGE genes in paclitaxel/doxorubicin-resistant human cancer cell lines, Duan [/bib_ref] , gastric [bib_ref] Melanoma-associated antigen-A1 expression predicts resistance to docetaxel and paclitaxel in advanced and..., Suzuki [/bib_ref] and medulloblastoma [bib_ref] Expression of MAGE and GAGE genes in medulloblastoma and modulation of resistance..., Kasuga [/bib_ref] cancer cell lines.
Wheeler, et al. [bib_ref] Clinical responsiveness of glioblastoma multiforme to chemotherapy after vaccination, Wheeler [/bib_ref] Reddy, et al. [bib_ref] Dlxin-1, a member of MAGE family, inhibits cell proliferation, invasion and tumorigenicity..., Reddy [/bib_ref] Phuphanich, et al. [bib_ref] Phase I trial of a multi-epitope-pulsed dendritic cell vaccine for patients with..., Phuphanich [/bib_ref] Yu, et al [bib_ref] Vaccination with tumor lysate-pulsed dendritic cells elicits antigen-specific, cytotoxic T-Cells in patients..., Yu [/bib_ref] EGFRvIII Mutant transmembrane tyrosine kinase receptor overexpressed in nearly half of all malignant gliomas. Associated with resistance to chemotherapy and radiation therapy [bib_ref] Epidermal growth factor receptor VIII peptide vaccination is efficacious against established intracerebral..., Heimberger [/bib_ref] [bib_ref] Monoclonal antibodies against EGFRvIII are tumor specific and react with breast and..., Wikstrand [/bib_ref] [bib_ref] Mutant epidermal growth factor receptor (EGFRvIII) contributes to head and neck cancer..., Sok [/bib_ref] [bib_ref] Evidence of high incidence of EGFRvIII expression and coexpression with EGFR in..., Ge [/bib_ref] [bib_ref] Anti-synthetic peptide antibody reacting at the fusion junction of deletion-mutant epidermal growth..., Humphrey [/bib_ref] [bib_ref] Epidermal growth factor ligand-independent, unregulated, cell-transforming potential of a naturally occurring human..., Batra [/bib_ref] [bib_ref] Mutant epidermal growth factor receptor up-regulates molecular effectors of tumor invasion, Lal [/bib_ref] [bib_ref] Constitutive EGFR signaling confers a motile phenotype to neural stem cells, Boockvar [/bib_ref] [bib_ref] A common mutant epidermal growth factor receptor confers enhanced tumorigenicity on human..., Nagane [/bib_ref] [bib_ref] Expression of oncogenic epidermal growth factor receptor family kinases induces paclitaxel resistance..., Montgomery [/bib_ref] [bib_ref] Radiation-induced activation of a common variant of EGFR confers enhanced radioresistance, Lammering [/bib_ref] Heimberger, et al. [bib_ref] Epidermal growth factor receptor VIII peptide vaccination is efficacious against established intracerebral..., Heimberger [/bib_ref] Heimberger, et al. [bib_ref] Dendritic cells pulsed with a tumor-specific peptide induce long-lasting immunity and are..., Heimberger [/bib_ref] Sampson, et al. [bib_ref] Immunologic escape after prolonged progression-free survival with epidermal growth factor receptor variant..., Sampson [/bib_ref] IL13Rα2 Cell surface receptor overexpressed by a subset of high grade gliomas [bib_ref] Expression of interleukin-13 receptor alpha2 in glioblastoma multiforme: Implications for targeted therapies, Jarboe [/bib_ref] [bib_ref] Receptor for interleukin 13 is a marker and therapeutic target for human..., Debinski [/bib_ref]. May be overexpressed by treatment refractory glioma stem cells, rendering them susceptible to targeted CTLs [bib_ref] Glioma IL13Rα2 is associated with mesenchymal signature gene expression and poor patient..., Brown [/bib_ref] Phuphanich, et al. [bib_ref] Phase I trial of a multi-epitope-pulsed dendritic cell vaccine for patients with..., Phuphanich [/bib_ref] Pollack, et al. [bib_ref] Peptide vaccine therapy for childhood gliomas, Pollack [/bib_ref] AIM-2 (Antigen isolated from Immunoselected Melanoma-2)
Tumor antigen expressed in melanoma, as well as neuroectodermal, breast, ovarian and colon cancer cells [bib_ref] Immunotherapy and chemotherapy-A practical partnership, Lake [/bib_ref]. Overexpressed in human glioma cells and recognized by CTLs [bib_ref] AIM-2: A novel tumour antigen expressed and presented by human glioma cells, Liu [/bib_ref].
Phuphanich, et al. [bib_ref] Phase I trial of a multi-epitope-pulsed dendritic cell vaccine for patients with..., Phuphanich [/bib_ref] An overview of major antigenic peptides used in glioma antitumor vaccines.
## Trp-2
Tyrosine-related protein-2 (TRP-2) is a normal tissue differentiation antigen from human melanoma that is recognized by T cells [bib_ref] Lineage-specific mechanism of drug and radiation resistance in melanoma mediated by tyrosinase-related..., Pak [/bib_ref]. A study by Liu et al. demonstrated TRP-2 was present in over 50% of GBM-derived cell lines. Furthermore, the antigen was not only expressed on the glioma cell surface, but also processed by its major histocompatibility complexes (MHCs) and presented to cytotoxic T cells [bib_ref] Molecular and functional analysis of tyrosinase-related protein (TRP)-2 as a cytotoxic T..., Liu [/bib_ref]. Wheeler et al. observed that patients who received chemotherapy after active TRP-2-bearing dendritic cell vaccination experienced slower progression and longer survival than patients who received either chemotherapy or vaccination alone [bib_ref] Clinical responsiveness of glioblastoma multiforme to chemotherapy after vaccination, Wheeler [/bib_ref]. A subsequent human study by the same group established an association between the presence of TRP-2 specific cytotoxic T cell (CTL) activity post-vaccination and significantly reduced TRP-2 expression in resected glioma cells. Furthermore, these cells demonstrated higher sensitivity to carboplatin and TMZ [bib_ref] Cytotoxic T cell targeting of TRP-2 sensitizes human malignant glioma to chemotherapy, Liu [/bib_ref]. Overexpression of TRP-2 has been associated with drug and radiation resistance through activation of the extracellular signal regulated kinase (ERK)/mitogen-activated protein kinase (MAPK) pathway [bib_ref] Lineage-specific mechanism of drug and radiation resistance in melanoma mediated by tyrosinase-related..., Pak [/bib_ref] [bib_ref] Radiation resistance of human melanoma analysed by retroviral insertional mutagenesis reveals a..., Pak [/bib_ref]. Thus, the study by Liu et al. provides compelling evidence that vaccine-mediated depletion of TRP-2 expressing glioma cells could undermine tumor resistance to the current standard of care. Moreover, a recent phase I trial of a multi-epitope DC vaccine containing TRP-2 for newly diagnosed GBM has demonstrated prolonged overall survival and progression free survival [bib_ref] Phase I trial of a multi-epitope-pulsed dendritic cell vaccine for patients with..., Phuphanich [/bib_ref].
## Mage-1
The melanoma-associated antigen (MAGE) family of peptides belongs to the cancer/testis tumor antigen group, which are normally expressed in immunoprivileged testicular tissue. Originally identified as an immune stimulating antigen found in 50% of melanomas, MAGE-1 has been shown to have four-fold higher expression in glioblastoma than in normal astrocytes [bib_ref] Vaccination with mage-3A1 peptide-pulsed mature, monocyte-derived dendritic cells expands specific cytotoxic T..., Thurner [/bib_ref] [bib_ref] MAGE-E1, a new member of the melanoma-associated antigen gene family and its..., Sasaki [/bib_ref]. MAGE-1 was another target of dendritic cell vaccines that was found by Wheeler et al. to sensitize tumors to subsequent chemotherapy [bib_ref] Clinical responsiveness of glioblastoma multiforme to chemotherapy after vaccination, Wheeler [/bib_ref]. At present, there are no studies that implicate MAGE-1 in glioma drug resistance. However, MAGE and GAGE gene overexpression has been associated with resistance to paclitaxel and doxorubicin in human ovarian cancer cell lines [bib_ref] Melanoma-associated antigen-A1 expression predicts resistance to docetaxel and paclitaxel in advanced and..., Suzuki [/bib_ref] ; docetaxel and paclitaxel in advanced and recurrent gastric cancer [bib_ref] Expression of MAGE and GAGE genes in medulloblastoma and modulation of resistance..., Kasuga [/bib_ref] ; and cisplatin and etoposide in medulloblastoma cell lines and specimens [bib_ref] Dlxin-1, a member of MAGE family, inhibits cell proliferation, invasion and tumorigenicity..., Reddy [/bib_ref]. In contrast, a study by Reddy et al. demonstrated that overexpression of Dixin-1, a MAGE family antigen, inhibited proliferation, invasion, and tumorigenicity in highly-chemoresistant glioma stem cells [bib_ref] Overexpression of MAGE/GAGE genes in paclitaxel/doxorubicin-resistant human cancer cell lines, Duan [/bib_ref]. Thus, the role of MAGE-1 vaccines in drug-resistant tumors is unclear and requires further investigation.
## Standard of care and immunotherapy: opportunities for synergism
## Vaccines and chemotherapy
Chemotherapy and immunotherapy have often been regarded as independent or antagonistic treatment modalities. This assumption is based on two considerations. First, cytotoxic chemotherapy is associated with severe lymphopenia, due to non-specific cell death of proliferating cells [bib_ref] Immunotherapeutic targeting of shared melanoma-associated antigens in a murine glioma model, Prins [/bib_ref] [bib_ref] Melanoma-Reactive CD8+ T cells recognize a novel tumor antigen expressed in a..., Mamoru [/bib_ref] [bib_ref] Apoptosis: A basic biological phenomenon with wideranging implications in tissue kinetics, Kerr [/bib_ref]. Second, chemotherapy-induced cell death was presumed to occur through a non-inflammatory apoptotic process or by induction of immune tolerance [bib_ref] Immunotherapeutic targeting of shared melanoma-associated antigens in a murine glioma model, Prins [/bib_ref] [bib_ref] Melanoma-Reactive CD8+ T cells recognize a novel tumor antigen expressed in a..., Mamoru [/bib_ref] [bib_ref] Apoptosis: A basic biological phenomenon with wideranging implications in tissue kinetics, Kerr [/bib_ref]. In the first case, lymphopenia could theoretically curtail immunotherapy's effectiveness by depleting the peripheral pool of immune cells available for mounting an antitumor response. In the second case, chemotherapy's effects would be either a passive lack of immune activation and proliferation or an active suppression of antigen presentation by tumor infiltrating T cells [bib_ref] Melanoma-Reactive CD8+ T cells recognize a novel tumor antigen expressed in a..., Mamoru [/bib_ref] [bib_ref] Combining immunotherapy with chemotherapy to treat cancer, Van Der Most [/bib_ref]. However, new evidence has called these assumptions into question.
In a 2008 case study by immune activation was demonstrated after concomitant treatment with TMZ and EGFRvIII vaccine [bib_ref] Immunological responses in a patient with glioblastoma multiforme treated with sequential courses..., Heimberger [/bib_ref]. In addition, notable was the lack of CD4 and CD8 T cell count decline after sequential courses of chemoradiation and vaccination. The authors suggest that chemotherapy is not necessarily counterproductive to immunotherapy as long as it is administered outside of the effector window, during which vaccine-induced CTLs are most responsive and active against the tumor. Outside this effector stage, the lymphodepleting effects of TMZ may actually be desirable, as it can reduce immunosuppressive lymphocytes such as Tregs, and thereby improve the local immune profile [bib_ref] Immunological responses in a patient with glioblastoma multiforme treated with sequential courses..., Heimberger [/bib_ref] [bib_ref] Cross-coupling of AP-1 and intracellular hormone receptors, Lamph [/bib_ref]. A subsequent study by found that low-dose metronomic TMZ administration in a rat glioma model resulted in decreased circulating Tregs and reduced tumor progression, though the latter did not reach statistical significance [bib_ref] Treg depletion with a low-dose metronomic temozolomide regimen in a rat glioma..., Banissi [/bib_ref]. These studies serve to emphasize the delicate immune balance that can ultimately determine the success of a treatment protocol. In practice, clinicians may either administer vaccines concurrently with chemoradiation, or use an interdigitated alternating regimen as a strategy to minimize CTL depletion and maximize Treg inhibition.
In addition to harnessing chemotherapy's immune-activating effects, immunotherapy may also be used as chemo-sensitizing agents that render drug-resistant tumors more amenable to standard therapeutic options [bib_ref] Sensitization of malignant glioma to chemotherapy through dendritic cell vaccination, Liu [/bib_ref]. In 1982, van Pel et al. demonstrated that tumors with apparently no transplantation immunogenicity still displayed enough foreign antigens to elicit a syngeneic rejection response [bib_ref] Protection against a nonimmunogenic mouse leukemia by an immunogenic variant obtained by..., Van Pel [/bib_ref]. Subsequent studies of brain tumors have focused on identifying immunologically susceptible GBM antigen profiles that may be exploited as targets for vaccines.
Ever since the feasibility and safety of synthetic peptide vaccines were demonstrated in clinical trials [bib_ref] Immunologic and therapeutic evaluation of a synthetic peptide vaccine for the treatment..., Rosenberg [/bib_ref] [bib_ref] Human tumor antigens for cancer vaccine development, Wang [/bib_ref] , concerted efforts have been made to identify MHC-restricted GBM antigens that can be delivered in the form of isolated peptides or peptide-bearing APCs. These antigens are often characterized as a product of either normal gene overexpression (i.e., TRP-2, MAGE-1) or a mutated somatic gene (EGFRvIII) [bib_ref] EGFRvIII-targeted vaccination therapy of malignant glioma, Choi [/bib_ref] [bib_ref] The promise of cancer vaccines, Gilboa [/bib_ref]. Furthermore, certain antigens have been associated with a drug or radiation resistant phenotype, thereby providing a target to overcome these treatment barriers. Antigens that are suspected to play a role in synergy between drug and vaccines therapies, including TRP-2, MAGE-1, and EGFRvIII have been discussed previously (refer to Section 5.1: Peptide Vaccines).
## Vaccines and radiation therapy
RT is commonly considered to be an immunosuppressive agent that non-selectively targets cells undergoing rapid division [bib_ref] Combination approaches to immunotherapy: The radiotherapy example, Gough [/bib_ref]. T lymphocytes, in particular, have been shown to be extremely sensitive to ionizing radiation [bib_ref] The molecular and cellular basis of radiosensitivity: Implications for understanding how normal..., Rosen [/bib_ref]. Thus, it is natural to assume that radiotherapy is counterproductive to immunotherapy, as it systematically eliminates key mediators of the antitumor response. In a study by Grossman et al., patients treated with radiation and TMZ for high grade gliomas experienced a CD4 count nadir at 2 months after treatment initiation. Treatment-induced CD4 count suppression was found to be not only long-lasting, but also associated with earlier death secondary to tumor progression [bib_ref] Immunosuppression in patients with high-grade gliomas treated with radiation and temozolomide, Grossman [/bib_ref]. Similar drops in CD4 counts were seen after treatment with steroids and radiotherapy only [bib_ref] Primary brain tumors treated with steroids and radiotherapy: Low CD4 counts and..., Hughes [/bib_ref] , further associating radiation with local and systemic immune compromise.
However, in addition to its cytocidal effects, RT has been shown to induce significant cellular and stromal changes by triggering -danger‖ signals that activate the innate and acquired immune system [bib_ref] A sense of danger from radiation, Mcbride [/bib_ref] [bib_ref] Sensors of ionizing radiation effects on the immunological microenvironment of cancer, Demaria [/bib_ref] [bib_ref] Tolerance, danger, and the extended family, Matzinger [/bib_ref]. Irradiated apoptotic tumor cells are phagocytosed by potent APCs such as DC's, which process and present tumor-specific antigens on MCH class 1 complexes, thereby activating CD8+ CTLs through the endogenous pathway [bib_ref] Dendritic cells acquire antigen from apoptotic cells and induce class I-restricted CTLs, Albert [/bib_ref] [bib_ref] Consequences of cell death: Exposure to necrotic tumor cells, but not primary..., Sauter [/bib_ref] [bib_ref] Immune modulation by ionizing radiation and its implications for cancer immunotherapy, Friedman [/bib_ref]. Immune activation secondary to apoptotic cell death suggests a synergistic role for radiation therapy and immunotherapy. When used in conjunction with anti-CTLA-4 antibodies, RT has been associated with an immune-mediated inhibition of tumor cells outside the irradiated field-also known as the abscopal effect [bib_ref] Fractionated but not single-dose radiotherapy induces an immune-mediated abscopal effect when combined..., Dewan [/bib_ref] [bib_ref] Ipilimumab and radiation therapy for melanoma brain metastases, Silk [/bib_ref]. A recent study by showed that combined use of anti-PD1 antibodies and localized RT increased expression of MHC class I, CXCL16, and ICAM molecules in vitro, and resulted in a survival benefit in orthotopic GBM mouse models [bib_ref] Anti-PD-1 blockade and stereotactic radiation produce long-term survival in mice with intracranial..., Zeng [/bib_ref]. These findings strongly suggest a proinflammatory effect of radiotherapy that may synergize with therapeutic immune modulators.
MHC downregulation is a well-described phenomenon in invading glioma cells [bib_ref] A functional role of HLA-G expression in human gliomas: An alternative strategy..., Wiendl [/bib_ref] [bib_ref] Downregulation of major histocompatibility complex antigens in invading glioma cells: Stealth invasion..., Zagzag [/bib_ref]. RT increases expression of MHC molecules, thereby counteracting a principal strategy for immune evasion by GBM. Increased antigen presentation has an added benefit of providing a natural target for vaccine-induced antitumor immunity. Using GL261 glioma mouse models, Newcomb et al. observed in vivo and upregulation of β2-microglobulin light chain subunit of the MHCI complex on the glioma cells, with a concomitant increase in CTL and helper T cell infiltration after whole body radiation therapy (WBRT). Furthermore, administration of both WBRT and allogeneic GL261 vaccination resulted in a survival advantage compared to WBRT alone, with superior long-term survival from 40% to 80% after combination therapy and 0% to 10% with monotherapy [bib_ref] The combination of ionizing radiation and peripheral vaccination produces long-term survival of..., Newcomb [/bib_ref].
The synergistic effects of radiation-induced tumor necrosis and DC therapy have also been reported. Exposure to necrotic cells has been shown to induce DC maturation and enhanced host antitumor response secondary to cross-priming with effector lymphocytes [bib_ref] Generation of tumor-specific T lymphocytes by cross-priming with human dendritic cells ingesting..., Hoffmann [/bib_ref] [bib_ref] Consequences of cell death: Exposure to necrotic tumor cells, but not primary..., Sauter [/bib_ref]. The survival benefits of combined RT and DC vaccination have been described in several tumor models including melanoma [bib_ref] Mechanisms involved in radiation enhancement of intratumoral dendritic cell therapy, Teitz-Tennenbaum [/bib_ref] , prostate cancer [bib_ref] In situ vaccination with CD204 gene-silenced dendritic cell, not unmodified dendritic cell,..., Guo [/bib_ref] , liver metastasis [bib_ref] A restricted cell population propagates glioblastoma growth after chemotherapy, Chen [/bib_ref] , and mammary tumors [bib_ref] Immunotherapy of radioresistant mammary tumors with early metastasis using molecular chaperone vaccines..., Weng [/bib_ref]. In a study by intratumoral injection of either immature DCs or irradiated necrotic glioma cells (IR-GCs) alone had no effect on survival in glioma mouse models. However, combined administration of DCs and IR-GCs resulted in synergistic antitumor effects and prolonged survival [bib_ref] Intratumoral injection of dendritic and irradiated glioma cells induces anti-tumor effects in..., Kikuchi [/bib_ref].
Several studies have delved into the synergistic effects of EGFR vaccines and RT. As stated previously, EGFRvIII overexpression is a poor prognostic indicator. Radiation exposure has been shown to result in robust stimulation of the EGFRvIII mutant receptor, leading to increased tumor survival and expansion [bib_ref] Radiation-induced activation of a common variant of EGFR confers enhanced radioresistance, Lammering [/bib_ref] [bib_ref] Inhibition of the type III epidermal growth factor receptor variant mutant receptor..., Lammering [/bib_ref]. The mechanism for enhanced clonogenic survival after RT may be EGFR-mediated activation of the phosphatidylinositol 3-kinase (PI3K)/Akt-1 pathway, leading to accelerated repair of double-stranded DNA breaks, thereby abrogating RT cytotoxicity [bib_ref] Inhibition of the type III epidermal growth factor receptor variant mutant receptor..., Lammering [/bib_ref]. Gefitinib (an EGFR inhibitor) and LY294002 (PI3K inhibitors) have been shown to attenuate repair of double-stranded breaks [bib_ref] EGFRvIII and DNA double-strand break repair: A molecular mechanism for radioresistance in..., Mukherjee [/bib_ref].
As of yet, there are no studies that specifically examine the synergistic effects of the EGFRvIII vaccination and RT. However, the existing evidence suggests that there may be a role for peptide vaccines in the radiosensitization of malignant gliomas.
## Conclusions and future directions
Despite aggressive treatment with the current standard of care (maximal surgical resection with adjuvant radiotherapy and TMZ), 5-year survival for GBM remains at a dismal <2% [bib_ref] Loss of tumor suppressor PTEN function increases B7-H1 expression and immunoresistance in..., Parsa [/bib_ref]. Immunotherapy has potential as a future adjunct to standard of care; cancer vaccines such as the dendritic cell (DC) and epidermal growth factor receptor (EGFR) vaccines have shown encouraging results in phase I and II clinical trials and raise the possibility of personalized peptide vaccines or even individualized whole tumor vaccines as more targeted alternatives for immunotherapy. Synergy between cancer vaccines and conventional chemotherapy and radiation therapy has also demonstrated a potential role for immunotherapy in multi-modal treatment paradigms. Vaccines against tumor-associated antigens such as TRP-2 and MAGE-1 and ongogenic mutations such as EGFRvIII may facilitate host antitumor immunity and prolonged survival. Additional studies advocate a role for peptide vaccines in the radiosensitization of malignant gliomas. Further research on the interactions between chemotherapy, radiation, and immunotherapy is needed to incorporate cancer vaccines into the current standard of care, and to maximize the antitumor potential of each treatment modality.
# Author contributions
Mira A. Patel and Jennifer E. Kim contributed equally to the production of this work. Jacob Ruzevick, Gordon Li and Michael Lim reviewed this work.
[table] Table 1: Phase II clinical trials measuring temozolomide efficacy in Glioblastoma. [/table]
[table] Table 2: Phase I and II clinical trials measuring cancer vaccine safety and efficacy in Glioblastoma. Selected major studies included that report clinical efficacy of cancer vaccines for GBM. Major findings refers to significant clinical responses reported in each study. 1 Phase I trial. 2 Phase I/II trial. 3 Phase II trial. DC = Dendritic Cell. OS = Overall Survival. TTS = Time to Survival. [/table]
|
10.3390/molecules24112187
|
CCBY
|
6600630
|
31212606
|
s2orc_pubmed_articles
|
One-Step Regioselective Synthesis of Benzofurans from Phenols and α-Haloketones
Reported here is the direct synthesis of naphthofurans and benzofurans from readily available phenols and α-haloketones promoted by titanium tetrachloride which combines Friedel-Crafts-like alkylation and intramolecular cyclodehydration into one step. This simple protocol allows for the formation of a variety of high value naphthofurans and benzofurans within which a series of cyclic and acyclic groups are readily incorporated. This process demonstrates the advantages of high levels of regioselectivity, broad substrate scope, and moderate to excellent yields.Abstract: Reported here is the direct synthesis of naphthofurans and benzofurans from readily available phenols and α-haloketones promoted by titanium tetrachloride which combines Friedel-Crafts-like alkylation and intramolecular cyclodehydration into one step. This simple protocol allows for the formation of a variety of high value naphthofurans and benzofurans within which a series of cyclic and acyclic groups are readily incorporated. This process demonstrates the advantages of high levels of regioselectivity, broad substrate scope, and moderate to excellent yields.
# Introduction
Benzofuran derivatives, especially naphthofurans, constitute a valuable class of heterocyclic compounds due to their natural occurrence and remarkable biological activities . Currently, more than 30 drugs bearing a benzofuran moiety have been approved by the United States Food and Drug Administration (USFDA) [bib_ref] Bioactive benzofuran derivatives: An insight on lead developments, radioligands and advances of..., Radadiya [/bib_ref]. Furthermore, naphthofurans have attracted significant attention in recent years owing to their powerful paradigm in the development and design of potential anticancer drugs [bib_ref] Synthesis of 1-(3 ,4 ,5 -trimethoxy) phenyl naphtho[2,1b]furan as a novel anticancer..., Srivastava [/bib_ref] , dual inhibitors of Alzheimer's disease , inhibitors of human protein kinase , and regulators of the nuclear receptor [bib_ref] Identification of small molecule regulators of the nuclear receptor HNF4α based on..., Guével [/bib_ref] , as well as other bioactivities [bib_ref] Design, synthesis, and biological evaluation of novel benzofuran derivatives bearing N-aryl piperazine..., Ma [/bib_ref]. Several representative bioactive compounds possessing a naphthofuran or benzofuran skeleton are listed in [fig_ref] Figure 1: Representative drugs containing benzofuran core [/fig_ref] [bib_ref] Xylarianaphthol-1, a novel dinaphthofuran derivative, activates p21 promoter in a p53-independent manner, Kotoku [/bib_ref]. Therefore, the development of novel synthetic methods for their direct preparation from readily accessible materials is very important.
# Introduction
Benzofuran derivatives, especially naphthofurans, constitute a valuable class of heterocyclic compounds due to their natural occurrence and remarkable biological activities . Currently, more than 30 drugs bearing a benzofuran moiety have been approved by the United States Food and Drug Administration (USFDA) [bib_ref] Bioactive benzofuran derivatives: An insight on lead developments, radioligands and advances of..., Radadiya [/bib_ref]. Furthermore, naphthofurans have attracted significant attention in recent years owing to their powerful paradigm in the development and design of potential anticancer drugs [bib_ref] Synthesis of 1-(3 ,4 ,5 -trimethoxy) phenyl naphtho[2,1b]furan as a novel anticancer..., Srivastava [/bib_ref] , dual inhibitors of Alzheimer's disease , inhibitors of human protein kinase , and regulators of the nuclear receptor [bib_ref] Identification of small molecule regulators of the nuclear receptor HNF4α based on..., Guével [/bib_ref] , as well as other bioactivities [bib_ref] Design, synthesis, and biological evaluation of novel benzofuran derivatives bearing N-aryl piperazine..., Ma [/bib_ref]. Several representative bioactive compounds possessing a naphthofuran or benzofuran skeleton are listed in [fig_ref] Figure 1: Representative drugs containing benzofuran core [/fig_ref] [bib_ref] Xylarianaphthol-1, a novel dinaphthofuran derivative, activates p21 promoter in a p53-independent manner, Kotoku [/bib_ref]. Therefore, the development of novel synthetic methods for their direct preparation from readily accessible materials is very important. Because of the aforementioned importance, numerous approaches have been reported for the preparation of these scaffolds through transition-metal catalysis, Lewis or Brønsted acid catalyzed, or base-promoted cyclizations [bib_ref] A new pathway for the synthesis of a new class of blue..., Moldoveanu [/bib_ref] [bib_ref] Synthesis of new 2-arylbenzo[b]furan derivatives via palladium-catalyzed Suzuki cross-coupling reactions in aqueous..., Chen [/bib_ref]. Many of these methods rely on harsh conditions, expensive transition metals, or substrates that are difficult to obtain. The strategy using phenols and α-haloketones as starting materials to obtain benzofurans is one of the most convenient routes. 3-Substituted benzo[b]furan 4 can be easily synthesized by a stepwise [bib_ref] Enantioselective synthesis of chiral β-aryloxy alcohols by ruthenium-catalyzed ketone hydrogenation via dynamic..., Bai [/bib_ref] [bib_ref] Preparation of 2-, 3-, 4-and 7-(2-alkylcarbamoyl-1-alkylvinyl)benzo[b]furans and their BLT1 and/or BLT2 inhibitory..., Ando [/bib_ref] [bib_ref] Titanium tetrachloride promoted cyclodehydration of aryloxyketones: Facile synthesis of benzofurans and naphthofurans..., Zhang [/bib_ref] or a one-step method [bib_ref] OTf)3-mediated synthesis of substituted benzofurans, Wang [/bib_ref] which involves o-alkylation of simple phenols with α-haloketone followed by intramolecular cyclization (Scheme 1a). However, there are seldom reports concerned with the synthesis of 2-substituted benzo[b]furans using α-haloketone and phenols as starting materials [bib_ref] The chemistry of α-haloketones and their utility in heterocyclic synthesis, Erian [/bib_ref]. Recently, reported that 2-aryl benzo[b]furan 5 can be obtained with excellent regioselectivity under refluxing temperature using neutral alumina as a promoter and xylene as a solvent [bib_ref] Regioselective preparation of benzo[b]furans from phenols and α-bromoketones, Arias [/bib_ref]. Nevertheless, the scope of α-haloketone is limited to only aryl ketone, without any alkyl ketone being employed (Scheme 1b). To continue our research [bib_ref] Titanium tetrachloride promoted cyclodehydration of aryloxyketones: Facile synthesis of benzofurans and naphthofurans..., Zhang [/bib_ref] , we report here that 2-alkyl benzo[b]furan 6 can be regioselectively formed directly from α-haloketones and phenols in the presence of titanium tetrachloride (Scheme 1c). Because of the aforementioned importance, numerous approaches have been reported for the preparation of these scaffolds through transition-metal catalysis, Lewis or Brønsted acid catalyzed, or base-promoted cyclizations [bib_ref] A new pathway for the synthesis of a new class of blue..., Moldoveanu [/bib_ref] [bib_ref] Synthesis of new 2-arylbenzo[b]furan derivatives via palladium-catalyzed Suzuki cross-coupling reactions in aqueous..., Chen [/bib_ref]. Many of these methods rely on harsh conditions, expensive transition metals, or substrates that are difficult to obtain. The strategy using phenols and αhaloketones as starting materials to obtain benzofurans is one of the most convenient routes. 3-Substituted benzo[b]furan 4 can be easily synthesized by a stepwise [bib_ref] Enantioselective synthesis of chiral β-aryloxy alcohols by ruthenium-catalyzed ketone hydrogenation via dynamic..., Bai [/bib_ref] [bib_ref] Preparation of 2-, 3-, 4-and 7-(2-alkylcarbamoyl-1-alkylvinyl)benzo[b]furans and their BLT1 and/or BLT2 inhibitory..., Ando [/bib_ref] [bib_ref] Titanium tetrachloride promoted cyclodehydration of aryloxyketones: Facile synthesis of benzofurans and naphthofurans..., Zhang [/bib_ref] or a one-step method [bib_ref] OTf)3-mediated synthesis of substituted benzofurans, Wang [/bib_ref] which involves o-alkylation of simple phenols with α-haloketone followed by intramolecular cyclization (Scheme 1a). However, there are seldom reports concerned with the synthesis of 2substituted benzo[b]furans using α-haloketone and phenols as starting materials [bib_ref] The chemistry of α-haloketones and their utility in heterocyclic synthesis, Erian [/bib_ref]. Recently, Arias et al. have reported that 2-aryl benzo[b]furan 5 can be obtained with excellent regioselectivity under refluxing temperature using neutral alumina as a promoter and xylene as a solvent [bib_ref] Regioselective preparation of benzo[b]furans from phenols and α-bromoketones, Arias [/bib_ref]. Nevertheless, the scope of α-haloketone is limited to only aryl ketone, without any alkyl ketone being employed (Scheme 1b). To continue our research [bib_ref] Titanium tetrachloride promoted cyclodehydration of aryloxyketones: Facile synthesis of benzofurans and naphthofurans..., Zhang [/bib_ref] , we report here that 2-alkyl benzo[b]furan 6 can be regioselectively formed directly from α-haloketones and phenols in the presence of titanium tetrachloride (Scheme 1c). Scheme 1. Synthesis of benzofurans from α-haloketones and phenols.
# Results and discussion
In order to develop a concise approach to naphthofuran, 2-naphthol (1a) and 2-chloroacetone (2a) were selected as model substrates [fig_ref] Table 1: Model reaction optimization a [/fig_ref]. To our delight, the reaction took place smoothly and proceeded to completion in ten hours when titanium tetrachloride was used in the presence of trifluoroethanol (TFE). The desired product 6a [fig_ref] Table 1: Model reaction optimization a [/fig_ref] was formed regioselectively, without any other isomer being detected (entry 8).
As a matter of fact, no conversion to the desired product was observed when commonly used Brønsted acids or other Lewis acids were tested [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entries 1-4). When TMSOTf or BF3.Et2O was used, the reaction produced numerous by-products and finally provided only a few furan products [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entries 5 and 6). Using reaction conditions reported in the literature [bib_ref] Regioselective preparation of benzo[b]furans from phenols and α-bromoketones, Arias [/bib_ref] , the reaction did take place but produced an inseparable mixture of 4a and 6a [fig_ref] Table 1: Model reaction optimization a [/fig_ref]. Moreover, it was discovered that increasing or decreasing the amount of titanium tetrachloride led to lower reaction efficiency [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entries 9-11). Note that slightly higher reaction temperature is beneficial to both reaction rate and efficiency. Actually, only a trace amount of 6a was detected by TLC when the reaction mixture was stirred at room temperature overnight, and the reaction turned out to be complex if prolonging the reaction time [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entry 13). The reaction could be carried out in several conventional solvents (CH2Cl2 and toluene) in addition to TFE, although resulting in significantly diminished conversions and a longer reaction time [fig_ref] Table 1: Model reaction optimization a [/fig_ref]. Other solvents (CH3CN, Et2O and THF) were also screened at their refluxing temperatures, but no new product could be detected after stirring overnight [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entry 16). The effect of catalyst amount and reaction temperature in this reaction was then investigated. Scheme 1. Synthesis of benzofurans from α-haloketones and phenols.
# Results and discussion
In order to develop a concise approach to naphthofuran, 2-naphthol (1a) and 2-chloroacetone (2a) were selected as model substrates [fig_ref] Table 1: Model reaction optimization a [/fig_ref]. To our delight, the reaction took place smoothly and proceeded to completion in ten hours when titanium tetrachloride was used in the presence of trifluoroethanol (TFE). The desired product 6a [fig_ref] Table 1: Model reaction optimization a [/fig_ref] was formed regioselectively, without any other isomer being detected (entry 8).
As a matter of fact, no conversion to the desired product was observed when commonly used Brønsted acids or other Lewis acids were tested [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entries 1-4). When TMSOTf or BF 3 .Et 2 O was used, the reaction produced numerous by-products and finally provided only a few furan products [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entries 5 and 6). Using reaction conditions reported in the literature [bib_ref] Regioselective preparation of benzo[b]furans from phenols and α-bromoketones, Arias [/bib_ref] , the reaction did take place but produced an inseparable mixture of 4a and 6a (Table 1, entry 7). Moreover, it was discovered that increasing or decreasing the amount of titanium tetrachloride led to lower reaction efficiency (Table 1, entries 9-11). Note that slightly higher reaction temperature is beneficial to both reaction rate and efficiency. Actually, only a trace amount of 6a was detected by TLC when the reaction mixture was stirred at room temperature overnight, and the reaction turned out to be complex if prolonging the reaction time [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entry 13). The reaction could be carried out in several conventional solvents (CH 2 Cl 2 and toluene) in addition to TFE, although resulting in significantly diminished conversions and a longer reaction time [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entries 14 and 15). Other solvents (CH 3 CN, Et 2 O and THF) were also screened at their refluxing temperatures, but no new product could be detected after stirring overnight [fig_ref] Table 1: Model reaction optimization a [/fig_ref] , entry 16). The effect of catalyst amount and reaction temperature in this reaction was then investigated. With an optimal set of catalysis conditions selected, we were then poised to test the one-pot process and evaluate the substrate scope of this reaction. When the reaction was conducted in refluxing TFE in the presence of titanium tetrachloride, we were delighted to find that both 1-and 2naphthols functioned efficiently in the reaction with 2-chloroacetone, with nearly single isomer being isolated [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6a-6d, 99:1 rr). The yields and reaction rates for 1-naphthol, in general, were a little better than those of 2-naphthol. Bromo-substituted naphthols were also highly effective regardless of the position of the bromo group on the phenyl ring [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6c and 6d, 89% and 66% yields, respectively). Additionally, simple 3-chloro-2-butanone was also highly effective in the current protocol [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6e and 6f, 74% and 76% yields, respectively).
Regioisomers were obtained in the reactions of other acyclic α-haloketones [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6g-6l). The phenomenon of isomerization was particularly obvious for the reaction of 2-chloro-3-pentanone stirred at room temperature [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6g, 2:1 rr). However, the problem caused by isomerization was readily overcome by slightly raising the reaction temperature and dropping α-haloketones into the reaction mixture. By employing the above-mentioned procedures, the desired products were afforded with high regioselectivity and good yields (Table 2, 6g, 9:1 rr; 6h, 10:1 rr). With an optimal set of catalysis conditions selected, we were then poised to test the one-pot process and evaluate the substrate scope of this reaction. When the reaction was conducted in refluxing TFE in the presence of titanium tetrachloride, we were delighted to find that both 1-and 2-naphthols functioned efficiently in the reaction with 2-chloroacetone, with nearly single isomer being isolated [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6a-6d, 99:1 rr). The yields and reaction rates for 1-naphthol, in general, were a little better than those of 2-naphthol. Bromo-substituted naphthols were also highly effective regardless of the position of the bromo group on the phenyl ring [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6c and 6d, 89% and 66% yields, respectively). Additionally, simple 3-chloro-2-butanone was also highly effective in the current protocol [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6e and 6f, 74% and 76% yields, respectively).
Regioisomers were obtained in the reactions of other acyclic α-haloketones [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6g-6l). The phenomenon of isomerization was particularly obvious for the reaction of 2-chloro-3-pentanone stirred at room temperature [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6g, 2:1 rr). However, the problem caused by isomerization was readily overcome by slightly raising the reaction temperature and dropping α-haloketones into the reaction mixture. By employing the above-mentioned procedures, the desired products were afforded with high regioselectivity and good yields (Table 2, 6g, 9:1 rr; 6h, 10:1 rr). With an optimal set of catalysis conditions selected, we were then poised to test the one-pot process and evaluate the substrate scope of this reaction. When the reaction was conducted in refluxing TFE in the presence of titanium tetrachloride, we were delighted to find that both 1-and 2naphthols functioned efficiently in the reaction with 2-chloroacetone, with nearly single isomer being isolated [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6a-6d, 99:1 rr). The yields and reaction rates for 1-naphthol, in general, were a little better than those of 2-naphthol. Bromo-substituted naphthols were also highly effective regardless of the position of the bromo group on the phenyl ring [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6c and 6d, 89% and 66% yields, respectively). Additionally, simple 3-chloro-2-butanone was also highly effective in the current protocol [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6e and 6f, 74% and 76% yields, respectively).
Regioisomers were obtained in the reactions of other acyclic α-haloketones [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6g-6l). The phenomenon of isomerization was particularly obvious for the reaction of 2-chloro-3-pentanone stirred at room temperature [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] , 6g, 2:1 rr). However, the problem caused by isomerization was readily overcome by slightly raising the reaction temperature and dropping α-haloketones into the reaction mixture. By employing the above-mentioned procedures, the desired products were afforded with high regioselectivity and good yields (Table 2, 6g, 9:1 rr; 6h, 10:1 rr). We next examined the scope of cyclic α-chloroketone which finally produced furans with four cycles [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref]. It was gratifying to find that these reactions were completed in 3-10 h to afford the corresponding tetracyclic products with moderate to excellent yields. Importantly, the transformation is not limited to six-membered cyclic α-chloroketone, as five-, seven-, and eightmembered cyclic α-chloroketones are competent substrates. Interestingly, both reaction rate and yield for six-membered cyclic α-chloroketone [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref] , 7b, 7e, 7i, and 7m) were better than those of other cyclic α-chloroketones. Intriguingly, all the reactions of 1-naphthol with cyclic α-chloroketones proceeded to completion in 3 h, offering products with excellent yields [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref] , 7e-7g). However, the reactions of bromo-substituted naphthols, such as 6-bromo or 7-bromo-2-naphthol, required longer reaction times (10-24 h) and offered only moderate yields of naphthofurans 7h, 7j-7l, and 7n-7o [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref]. [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref]. Reaction of naphthols with different cyclic α-haloketones a, b . a The molar ratio between naphthol 1, haloketone 2 and TiCl 4 is 1:1.2:1. b Isolated yield. c The ratio of regioisomers was determined by crude 1 H NMR (see Supplementary Materials). d The products were inseparable when purified by chromatography on silica gel. e Using 1-napthol as starting material.
We next examined the scope of cyclic α-chloroketone which finally produced furans with four cycles [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref]. It was gratifying to find that these reactions were completed in 3-10 h to afford the corresponding tetracyclic products with moderate to excellent yields. Importantly, the transformation is not limited to six-membered cyclic α-chloroketone, as five-, seven-, and eight-membered cyclic α-chloroketones are competent substrates. Interestingly, both reaction rate and yield for six-membered cyclic α-chloroketone [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref] , 7b, 7e, 7i, and 7m) were better than those of other cyclic α-chloroketones. Intriguingly, all the reactions of 1-naphthol with cyclic α-chloroketones proceeded to completion in 3 h, offering products with excellent yields [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref] , 7e-7g). However, the reactions of bromo-substituted naphthols, such as 6-bromo or 7-bromo-2-naphthol, required longer reaction times (10-24 h) and offered only moderate yields of naphthofurans 7h, 7j-7l, and 7n-7o [fig_ref] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b [/fig_ref]. To further extend the reaction scope, we carried out the reaction with phenols. First, we used 2chlorocyclohexanone to examine the reactivity of various substituted phenols. Gratifyingly, all alkylor alkoxy-substituted phenols reacted successfully with 2-chlorocyclohexanone to produce the desired benzofurans 9a-9h [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] with excellent yields. Additionally, it was found that the substituent patterns (ortho-, meta-and para-) on the benzene ring showed no observed effects on the To further extend the reaction scope, we carried out the reaction with phenols. First, we used 2chlorocyclohexanone to examine the reactivity of various substituted phenols. Gratifyingly, all alkylor alkoxy-substituted phenols reacted successfully with 2-chlorocyclohexanone to produce the desired benzofurans 9a-9h [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] with excellent yields. Additionally, it was found that the substituent patterns (ortho-, meta-and para-) on the benzene ring showed no observed effects on the a The molar ratio between naphthol 1 and haloketone 2 is 1:1.2. b Isolated yield. c Using 1-napthol as starting material.
To further extend the reaction scope, we carried out the reaction with phenols. First, we used 2-chlorocyclohexanone to examine the reactivity of various substituted phenols. Gratifyingly, all alkylor alkoxy-substituted phenols reacted successfully with 2-chlorocyclohexanone to produce the desired benzofurans 9a-9h [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] with excellent yields. Additionally, it was found that the substituent patterns (ortho-, meta-and para-) on the benzene ring showed no observed effects on the reaction [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9b-9d). Note that the reaction exhibits sensitivity to steric constraints on the phenol substrate, that is, the Friedel-Crafts-like alkylation occurs preferentially at the less hindered position, which can be demonstrated by the formation of a single isomer . Unfortunately, phenols incorporating an electron-withdrawing group did not react under these conditions. For example, no new spot was detected by TLC when 4-nitrophenol was employed to react with 2-chlorocyclohexane for 24 h. Moreover, phenols bearing a strong electron-donating substituent, such as methoxyl, failed to give better yields [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9e and 9f, 72% and 73% yields, respectively), although the reaction rates were faster than that of non-substituted phenol. On the other hand, the reactions of phenols with acyclic α-haloketones were also examined, which proceeded smoothly [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9i-9k, 77-81% yields). . Note that the reaction exhibits sensitivity to steric constraints on the phenol substrate, that is, the Friedel-Crafts-like alkylation occurs preferentially at the less hindered position, which can be demonstrated by the formation of a single isomer . Unfortunately, phenols incorporating an electron-withdrawing group did not react under these conditions. For example, no new spot was detected by TLC when 4-nitrophenol was employed to react with 2-chlorocyclohexane for 24 h. Moreover, phenols bearing a strong electron-donating substituent, such as methoxyl, failed to give better yields [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9e and 9f, 72% and 73% yields, respectively), although the reaction rates were faster than that of non-substituted phenol. On the other hand, the reactions of phenols with acyclic α-haloketones were also examined, which proceeded smoothly [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9i-9k, 77-81% yields). To gain insight into the reaction mechanism, we carried out the reaction with unsymmetrically substituted haloketone 2m (Scheme 2), under the optimized conditions. Pleasingly, both the desired furan 6m and isomer 6f (Scheme 2) were isolated in nearly equal amounts. However, reactions for αhalo aromatic ketones, such as 2-bromoacetophenone and 2-bromo-1-phenylpentanone, failed to occur. These facts suggested that oxy-allyl cation may be one of the key intermediates for the reaction between α-halo alkyl ketones and phenols.
## Scheme 2. mechanism research.
Molecules 2019, 24, x FOR PEER REVIEW 6 of 14 reaction outcomes [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9b-9d). Note that the reaction exhibits sensitivity to steric constraints on the phenol substrate, that is, the Friedel-Crafts-like alkylation occurs preferentially at the less hindered position, which can be demonstrated by the formation of a single isomer [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9c). Unfortunately, phenols incorporating an electron-withdrawing group did not react under these conditions. For example, no new spot was detected by TLC when 4-nitrophenol was employed to react with 2-chlorocyclohexane for 24 h. Moreover, phenols bearing a strong electron-donating substituent, such as methoxyl, failed to give better yields [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9e and 9f, 72% and 73% yields, respectively), although the reaction rates were faster than that of non-substituted phenol. On the other hand, the reactions of phenols with acyclic α-haloketones were also examined, which proceeded smoothly [fig_ref] Table 4: Reaction of phenols with different α-haloketones a, b [/fig_ref] , 9i-9k, 77-81% yields). To gain insight into the reaction mechanism, we carried out the reaction with unsymmetrically substituted haloketone 2m (Scheme 2), under the optimized conditions. Pleasingly, both the desired furan 6m and isomer 6f (Scheme 2) were isolated in nearly equal amounts. However, reactions for αhalo aromatic ketones, such as 2-bromoacetophenone and 2-bromo-1-phenylpentanone, failed to occur. These facts suggested that oxy-allyl cation may be one of the key intermediates for the reaction between α-halo alkyl ketones and phenols. To gain insight into the reaction mechanism, we carried out the reaction with unsymmetrically substituted haloketone 2m (Scheme 2), under the optimized conditions. Pleasingly, both the desired furan 6m and isomer 6f (Scheme 2) were isolated in nearly equal amounts. However, reactions for α-halo aromatic ketones, such as 2-bromoacetophenone and 2-bromo-1-phenylpentanone, failed to occur. These facts suggested that oxy-allyl cation may be one of the key intermediates for the reaction between α-halo alkyl ketones and phenols. To gain insight into the reaction mechanism, we carried out the reaction with unsymmetrically substituted haloketone 2m (Scheme 2), under the optimized conditions. Pleasingly, both the desired furan 6m and isomer 6f (Scheme 2) were isolated in nearly equal amounts. However, reactions for αhalo aromatic ketones, such as 2-bromoacetophenone and 2-bromo-1-phenylpentanone, failed to occur. These facts suggested that oxy-allyl cation may be one of the key intermediates for the reaction between α-halo alkyl ketones and phenols.
## Scheme 2. mechanism research.
It is reported that azepinium ions can be generated by the ether cleavage reaction of 2-methoxy-2H-azepine derivatives with titanium tetrachloride as a Lewis acid [bib_ref] Electrophilic behavior of the π delocalized azepinium ion: Friedel-Crafts reactions with benzenes..., Kubota [/bib_ref] [bib_ref] Synthesis of a delocalized azepinium ion and investigation of its electrophilic character, Satake [/bib_ref]. Furthermore, titanium tetrachloride is also a powerful dehydrating agent and demonstrates a prominent effect in the condensation reaction of triketones to yield furans [bib_ref] Paal-Knorr furan synthesis using titanium tetrachloride as dehydrating agent: A concise furan..., Luo [/bib_ref].Apart from the reaction paths reported in the literature [fig_ref] Figure 1: Representative drugs containing benzofuran core [/fig_ref] [bib_ref] Regioselective preparation of benzo[b]furans from phenols and α-bromoketones, Arias [/bib_ref] , another reaction route for the one-pot synthesis of benzofuran was proposed in Scheme 3 [bib_ref] Efficient catalytic-free method to produce alpha-aryl cycloalkanones through highly chemoselective coupling of..., Luo [/bib_ref] [bib_ref] Addition of indoles to oxyallyl cations for facile access to α-indole carbonyl..., Tang [/bib_ref]. First, oxy-allyl cation I, evolved from 2m with the aid of titanium It is reported that azepinium ions can be generated by the ether cleavage reaction of 2-methoxy-2H-azepine derivatives with titanium tetrachloride as a Lewis acid [bib_ref] Electrophilic behavior of the π delocalized azepinium ion: Friedel-Crafts reactions with benzenes..., Kubota [/bib_ref] [bib_ref] Synthesis of a delocalized azepinium ion and investigation of its electrophilic character, Satake [/bib_ref]. Furthermore, titanium tetrachloride is also a powerful dehydrating agent and demonstrates a prominent effect in the condensation reaction of triketones to yield furans [bib_ref] Paal-Knorr furan synthesis using titanium tetrachloride as dehydrating agent: A concise furan..., Luo [/bib_ref]. Apart from the reaction paths reported in the literature (Scheme 1a,b) [bib_ref] Regioselective preparation of benzo[b]furans from phenols and α-bromoketones, Arias [/bib_ref] , another reaction route for the one-pot synthesis of benzofuran was proposed in Scheme 3 [bib_ref] Efficient catalytic-free method to produce alpha-aryl cycloalkanones through highly chemoselective coupling of..., Luo [/bib_ref] [bib_ref] Addition of indoles to oxyallyl cations for facile access to α-indole carbonyl..., Tang [/bib_ref]. First, oxy-allyl cation I, evolved from 2m with the aid of titanium tetrachloride, reacts with 1b to produce Friedel-Crafts type intermediate II or III (Scheme 3). Then, due to the powerful dehydration ability of titanium tetrachloride, intramolecular cyclodehydration of the intermediate II or III easily takes place to obtain benzofuran 6m or 6f (Scheme 3).
## Experimental section
## General information
Nuclear magnetic resonance spectra ( 1 H and 13 C) were recorded on 400 and 600 MHz spectrometers (Bruker, Karlsruhe, Germany) with tetramethylsilane (TMS) as an internal standard.The splitting patterns are designated as singlet (s), doublet (d), triplet (t), quartet (q), dd (doublet of doublets), m (multiplets), etc. All first-order splitting patterns were assigned on the basis of the appearance of the multiplet. Splitting patterns that could not be easily interpreted were designated as multiplet (m) or broad (br). High resolution mass spectral analysis (HRMS) was performed on an ESI-QTOP mass spectrometer (Bruker Solari XFT-ICR-MS system). Purification was done by column chromatography and preparative TLC using silica gel. TLC analyses were performed on commercial glass plates (Qingdao Haiyang Chemical Co., Ltd, Qingdao, China) bearing a 0.25mmlayer of silica gel GF254.Visualization was performed using a UV lampor chemical stains like KMnO4 and I2. Commercially available materials were used as received.
All reactions were carried out under nitrogen atmosphere. Dehydrated solvents were purchased from commercial suppliers (Alfa Aesar, Ward Hill, MA, USA; Adamas, Shanghai, China) and stored under nitrogen atmosphere. Unless otherwise noted, materials were obtained from commercial suppliers and used without further purification. Some of the α-chloroketones and bromoketones were prepared using literature methods [bib_ref] Mechanism of the reaction of hexachloroacetone with enamines. A new, convenient synthesis..., Laskovics [/bib_ref] [bib_ref] A mild and efficient procedure for alpha-bromination of ketones using N-bromosuccinimide catalysed..., Tanemura [/bib_ref].
## General procedure for the reaction between phenol and α-haloketone
To a 25 mL two-necked flask equipped with a reflux condenser, fresh distilled 2,2,2trifluoroethanol (1.0 mL), phenol (1.0 mmol), and titanium tetrachloride (1.0 mmol) were added under nitrogen atmosphere. Then, a mixture of α-haloketone (1.2 mmol) in 2,2,2-trifluoroethanol (1.0 mL) was dropped into the reaction mixture under refluxing temperature. After completion of the reaction (monitored by TLC), the mixture was quenched with a saturated aqueous solution of NH4Cl (20 mL). After filtration of the mixture, the water layer was extracted by dichloromethane (3 × 10 mL) and dried with anhydrous sodium sulphate. The organic mixture was concentrated under reduced pressure, and separated by silica-gel column chromatography using ethyl acetate−hexane as eluent in increasing polarity to yield the desired furan compound.
3.2.1. Characterizations of Naphthofuran 6 [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] 2-Methylnaphtho[2,1-b]furan (6a) [bib_ref] Direct synthesis of 2-methylbenzofurans from calcium carbide and salicylaldehyde p-tosylhydrazones, Fu [/bib_ref]. The title compound was obtained as white solid (76%), mp: 39-40 °C, and the analytical data are consistent with those in the literature.
## Experimental section
## General information
Nuclear magnetic resonance spectra ( 1 H and 13 C) were recorded on 400 and 600 MHz spectrometers (Bruker, Karlsruhe, Germany) with tetramethylsilane (TMS) as an internal standard.The splitting patterns are designated as singlet (s), doublet (d), triplet (t), quartet (q), dd (doublet of doublets), m (multiplets), etc. All first-order splitting patterns were assigned on the basis of the appearance of the multiplet. Splitting patterns that could not be easily interpreted were designated as multiplet (m) or broad (br). High resolution mass spectral analysis (HRMS) was performed on an ESI-QTOP mass spectrometer (Bruker Solari XFT-ICR-MS system). Purification was done by column chromatography and preparative TLC using silica gel. TLC analyses were performed on commercial glass plates (Qingdao Haiyang Chemical Co., Ltd, Qingdao, China) bearing a 0.25-mmlayer of silica gel GF 254 .Visualization was performed using a UV lampor chemical stains like KMnO 4 and I 2 .
Commercially available materials were used as received.
All reactions were carried out under nitrogen atmosphere. Dehydrated solvents were purchased from commercial suppliers (Alfa Aesar, Ward Hill, MA, USA; Adamas, Shanghai, China) and stored under nitrogen atmosphere. Unless otherwise noted, materials were obtained from commercial suppliers and used without further purification. Some of the α-chloroketones and bromoketones were prepared using literature methods [bib_ref] Mechanism of the reaction of hexachloroacetone with enamines. A new, convenient synthesis..., Laskovics [/bib_ref] [bib_ref] A mild and efficient procedure for alpha-bromination of ketones using N-bromosuccinimide catalysed..., Tanemura [/bib_ref].
## General procedure for the reaction between phenol and α-haloketone
To a 25 mL two-necked flask equipped with a reflux condenser, fresh distilled 2,2,2-trifluoroethanol (1.0 mL), phenol (1.0 mmol), and titanium tetrachloride (1.0 mmol) were added under nitrogen atmosphere. Then, a mixture of α-haloketone (1.2 mmol) in 2,2,2-trifluoroethanol (1.0 mL) was dropped into the reaction mixture under refluxing temperature. After completion of the reaction (monitored by TLC), the mixture was quenched with a saturated aqueous solution of NH 4 Cl (20 mL). After filtration of the mixture, the water layer was extracted by dichloromethane (3 × 10 mL) and dried with anhydrous sodium sulphate. The organic mixture was concentrated under reduced pressure, and separated by silica-gel column chromatography using ethyl acetate−hexane as eluent in increasing polarity to yield the desired furan compound. [fig_ref] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c [/fig_ref] 2-Methylnaphtho[2,1-b]furan (6a) [bib_ref] Direct synthesis of 2-methylbenzofurans from calcium carbide and salicylaldehyde p-tosylhydrazones, Fu [/bib_ref]. The title compound was obtained as white solid (76%), mp: 39-40 - C, and the analytical data are consistent with those in the literature. [fig_ref] Figure 1: Representative drugs containing benzofuran core [/fig_ref] 2-Benzyl-1-phenylnaphtho furan (6i). The title compound was obtained as yellow oil (70%), and the analytical data are consistent with those in the literature. 2-Benzyl-3-phenylnaphtho furan (6j) [bib_ref] InCl 3 -catalyzed propargylation of indoles and phenols with propargylic acetates: Application..., Liu [/bib_ref]. The title compound was obtained as yellow oil (74%), and the analytical data are consistent with those in the literature. [bib_ref] Diversified synthesis of furans by coupling between enols/1,3-dicarbonyl compounds and nitroolefins: Direct..., Ghosh [/bib_ref]. The title compound was obtained as yellowish oil (86%), and the analytical data are consistent with those in the literature. [bib_ref] Dehydrative C-H alkylation and alkenylation of phenols with alcohols: Expedient synthesis for..., Lee [/bib_ref].The title compound was obtained as yellowish oil (92%), and the analytical data are consistent with those in the literature.
[fig] Figure 1: Representative drugs containing benzofuran core. [/fig]
[fig] Scheme 2: Mechanism research. a The molar ratio between phenol 8 and haloketone 2 is 1:1.2. b Isolated yield. [/fig]
[fig] Molecules 2019 ,: 24, x FOR PEER REVIEW 6 of 14 [/fig]
[fig] Scheme 3: Proposed mechanism. [/fig]
[fig] 3 9: furan (6b)[31]. The title compound was obtained as colorless oil (80%), and the analytical data are consistent with those in the literature.1 H NMR (600 MHz, CDCl 3 ) δ 8.27 (d, J = 8.2 Hz, 1H), 7.91 (d, J = 8.2 Hz, 1H), 7.74-7.51 (m, 3H), 7.45 (d, J = 7.b]furan (6c) [32]. The title compound was obtained as white solid (89%), mp: 91-92 • C, and the analytical data are consistent with those in the literature. 1 H NMR (600 MHz, CDCl 3 ) δ 8.06 (d, J = 1.8 Hz, 1H), 7.90 (d, J = 8.7 Hz, 1H), 7.60 (dd, J = 8.8, 1.8 Hz, 2H), 7.53 (d, J = 8.9 Hz, 1H), 6.81 (s, 1H), 2.55 (s, 3H); 13 C NMR (150 MHz, CDCl 3 ) δ 155.b]furan (6d). The title compound was obtained as white solid (66%), mp: 93-94 • C. 1 H NMR (600 MHz, CDCl 3 ) δ 8.20 (d, J = 1.9 Hz, 1H), 7.77 (d, J = 8.7 Hz, 1H), 7.59 (s, 2H), 7.52 (dd, J = 8.7, 2.0 Hz, 1H), 6.80 (s, 1H), 2.55 (d, J = 0.8 Hz, 3H); 13 C NMR (150 MHz, CDCl 3 ) δ 155.2, 152.3, 130.3, 128.6, 128.5, 127.5, 125.9, 123.6, 123.5, 120.0, 112.5, 101.6, 14.2; HRMS (ESI) calcd for C 13 H 10 BrO (M + H) + : 260.9910. Found: 260.9909. 1,2-Dimethylnaphtho[2,1-b]furan (6e) [33]. The title compound was obtained as yellowish oil (74%), and the analytical data are consistent with those in the literature. 1 H NMR (600 MHz, CDCl 3 ) δ 8.40 (d, J = 8.3 Hz, 1H), 7.96 (d, J = 8.1 Hz, 1H), 7.65 (d, J = 8.8 Hz, 1H), 7.63-7.53 (m, 2H), 7.53-7.38 (m, 1H), 2.57 (s, 3H), 2.49 (s, 3H); 13 C NMR (150 MHz, CDCl 3 ) δ 151.3, 149.9, 130.7, 128.9, 128.7, 125.8, 124.0, 123.8, 123.2, 123.0, 112.1, 111.7, 11.8, 11.4. 2,3-Dimethylnaphtho[1,2-b]furan (6f) [30]. The title compound was obtained as yellowish solid (71%), mp: 201-202 • C, and the analytical data are consistent with those in the literature. 1 H NMR (400 MHz, CDCl 3 ) δ 8.26 (d, J = 8.3 Hz, 1H), 7.92 (d, J = 8.2 Hz, 1H), 7.64 (d, J = 8.5 Hz, 1H), 7.61-7.51 (m, 2H), 7.51-7.36 (m, 1H), 2.50 (s, 3H), 2.24 (s, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ 149.8, 148.9, 130.9, 128.3, 126.0, 125.7, 124.3, 122.5, 121.1, 119.7, 117.9, 110.9, 12.00, 8.1. 2-Ethyl-3-methylnaphtho[1,2-b]furan (6g). The title compound was obtained as yellowish oil (88%). 1 H NMR (600 MHz, CDCl 3 ) δ 8.27 (d, J = 8.2 Hz, 1H), 7.91 (d, J = 8.2 Hz, 1H), 7.63 (d, J = 8.4 Hz, 1H), 7.54 (t, J = 7.5 Hz, 2H), 7.43 (t, J = 7.5 Hz, 1H), 2.86 (q, J = 7.6 Hz, 2H), 2.25 (s, 3H), 1.36 (t, J = 7.6 Hz, 3H); 13 C NMR (150 MHz, CDCl 3 ) δ 154.HRMS (ESI) calcd for C 15 H 15 O (M + H) + : 211.1117. Found: 211.1118.2-Ethyl-1-methylnaphtho[2,1-b]furan (6h)[26]. The title compound was obtained as yellowish oil (80%), and the analytical data are consistent with those in the literature.1 H NMR (400 MHz, CDCl 3 ) δ 8.41 (d, J = 8.3 Hz, 1H), 7.96 (d, J = 8.1 Hz, 1H), 7.74-7.53 (m, 3H), 7.53-7.41 (m, 1H), 2.87 (q, J = 7.6 Hz, 2H), 2.59 (s, 3H), 1.36 (t, J = 7.6 Hz, 3H); 13 C NMR (100 MHz, CDCl 3 ) δ 155.0, 151.3, 130.7, 129.0, 128.9, 125.8, 124.0, 123.7, 123.3, 123.1, 112.2, 110.8, 19.6, 13.2, 11.3. [/fig]
[fig] 3 1: Tetrahydro-7H-cyclohepta[b]naphtho[2,1-d]furan (7f)[38]. The title compound was obtained as white solid (84%), mp: 38-39 • C, and the analytical data are consistent with those in the literature.1 H NMR (400 MHz, CDCl 3 )δ 8.25 (d, J = 8.3 Hz, 1H), 7.91 (d, J = 8.2 Hz, 1H), 7.63 (d, J = 8.11,12-Hexahydrocycloocta[b]naphtho[2,1-d]furan (7g)[38]. The title compound was obtained as colorless oil (87%), and the analytical data are consistent with those in the literature.1 H NMR (400 MHz, CDCl 3 ) δ 8.29 (d, J = 8.3 Hz, 1H), 7.93 (d, J = 8.2 Hz, 1H), 7.65 (d, J = 8.5 Hz, 1H), 7.57 (t, J = 8.1 Hz, 2H), 7.45 (t, J = 7.4 Hz, 1H), 3.18-2.99 (m, 2H), 2.99-2.82 (m, 2H), 2.03-1.72 (m, 4H), 1.68-1.44 (m, 4H); 13 C NMR (100 MHz, CDCl 3 ) δ 154.10-dihydro-8H-cyclopenta[b]naphtho[1,2-d]furan (7h). The title compound was obtained as white solid (61%), mp: 36-37 • C. 1 H NMR (600 MHz, CDCl 3 ) δ 8.06 (s, 1H), 7.93 (d, J = 8.7 Hz, 1H), 7.60 (dd, J = 18.5, 8.8 Hz, 2H), 7.51 (d, J = 8.9 Hz, 1H), 3.06 (t, J = 6.8 Hz, 2H), 3.00-2.89 (m, 2H), 2.77-2.56 (m, 2H); 13 C NMR (150 MHz, CDCl 3 ) δ 162.7, 157.3, 131.5, 130.4, 128.9, 125.9, 125.8, 122.1, 122.0, 121.8, 117.9, 114.0, 27.9, 25.0, 23.8; HRMS (ESI) calcd for C 15 H 12 BrO (M + H) + : 287.0066. Found: 287.0068. 3-Bromo-8,9,10,11-tetrahydronaphtho[2,1-b]benzofuran (7i). The title compound was obtained as white solid (77%). 1 H NMR (400 MHz, CDCl 3 ) δ 8.04 (dd, J = 9.2, 5.4 Hz, 2H), 7.68-7.53 (m, 2H), 7.49 (d, J = 8.9 Hz, 1H), 3.00 (t, J = 4.8 Hz, 2H), 2.83 (t, J = 5.0 Hz, 2H), 2.04-1.84 (m, 4H); 13 C NMR (100 MHz, CDCl 3 ) δ 153.9, 151.5, 131.8, 130.7, 128.8, 126.8, 125.2, 122.7, 122.6, 117.5, 114.1, 113.3, 23.7, 23.0, 23.0, 22.5; HRMS (ESI) calcd for C 16 H 14 BrO (M + H) + : 301.0223. Found: 301.0220. 3-Bromo-9,10,11,12-tetrahydro-8H-cyclohepta[b]naphtho[1,2-d]furan (7j). The title compound was obtained as yellowish oil (65%). 1 H NMR (400 MHz, CDCl 3 ) δ 8.29 (d, J = 9.0 Hz, 1H), 8.06 (d, J = 2.0 Hz, 1H), 7.56 (ddd, J = 25.HRMS (ESI) calcd for C 17 H 16 BrO (M + H) + : 315.0379. Found: 315.0381.3-Bromo-8,9,10,11,12,13-hexahydrocycloocta[b]naphtho[1,2-d]furan (7k). The title compound was obtained as white solid (68%), mp: 36-37 • C. 1 H NMR (600 MHz, CDCl 3 ) δ 8.20 (s, 1H), 8.07 (s, 1H), 7.60 (s, 2H), 7.52 (s, 1H), 3.13 (s, 2H), 2.98 (s, 2H), 1.93 (s, 2H), 1.78 (s, 2H), 1.58 (s, 2H), 1.45 (s, 2H); 13 C NMR (150 MHz, CDCl 3 ) δ 155.5, 151.4, 132.0, 130.9, 128.9, 127.0, 124.7, 122.8, 122.7, 117.4, 115.5, 113.4, 29.5, 28.0, 26.6, 26.0, 25.7, 23.3; HRMS (ESI) calcd for C 18 H 18 BrO (M + H) + : 329.0536. Found: 329.0534. [/fig]
[table] Table 1: Model reaction optimization a . [/table]
[table] Table 2: Reaction of naphthols with different acyclic α-haloketones a, b, c . The mole ratio between 1a and 2a is 1:1.2. b Isolated yields. c Solvents: ACN, Et 2 O, THF. d An inseparable mixture of 4a and 6a was obtained. [/table]
[table] Table 3: Reaction of naphthols with different cyclic α-haloketones a, b . The molar ratio between naphthol 1 and haloketone 2 is 1:1.2. b Isolated yield. c Using 1-napthol as starting material. [/table]
[table] Table 4: Reaction of phenols with different α-haloketones a, b . [/table]
[table] 4: Hz, 3H);13 C NMR (100 MHz, CDCl 3 ) δ 154.4, 151.6, 130.7, 129.1, 128.4, 125.9, 124.1, 123.7, 123.2, 122.5, 116.2, 112.3, 30.9, 27.3, 26.0, 23.4, 22.5, 14.1, 14.0; HRMS (ESI) calcd for C 19 H 23 O (M + H) + : 267.1743. Found: 267.1741. 2-Butyl-3-propylnaphtho[1,2-b]furan (6l). The title compound was obtained as yellowish oil (84%). 1 H NMR (400 MHz, CDCl 3 ) δ 8.26 (d, J = 8.3 Hz, 1H), 7.90 (d, J = 8.2 Hz, 1H), 7.57 (ddd, J = 17.6, 16.0, 8.2 Hz, 3H), 7.50-7.36 (m, 1H), 2.82 (dt, J = 9.9, 7.5 Hz, 2H), 2.68 (dd, J = 15.8, 7.9 Hz, 2H), 1.91-1.72 (m, 2H), 1.67 (dd, J = 15.7, 7.6 Hz, 2H), 1.51-1.33 (m, 2H), 0.98 (ddt, J = 16.3, 8.9, 7.4 Hz, 6H); 13 C NMR (100 MHz, CDCl 3 ) δ 154.1, 149.1, 130.8, 128.3, 125.9, 125.0, 124.3, 122.3, 121.2, 119.8, 118.3, 115.8, 23.5, 23.4, 22.6, 22.5, 22.1, 14.1, 13.9; HRMS (ESI) calcd for C 19 H 23 O (M + H) + : 267.1743. Found: 267.1744.2-Ethylnaphtho[1,2-b]furan (6m) [36].The title compound was obtained as yellowish oil (40%). 1 H NMR (600 MHz, CDCl 3 ) δ 8.27 (d, J = 8.3 Hz, 1H), 7.91 (d, J = 8.2 Hz, 1H), 7.65-7.51 (m, 3H), 7.50-7.38 (m, 1H), 6.51 (d, J = 0.9 Hz, 1H), 2.92 (qd, J = 7.5, 0.9 Hz, 2H), 1.40 (t, J = 7.6 Hz, 3H). 13 C NMR (150 MHz, CDCl 3 ) δ 160.3, 149.8, 130.9, 128.3, 126.1, 124.4, 124.4, 122.9, 121.2, 119.7, 119.4, 102.1, 21.9, 12.2.3.2.2. Characterizations of Naphthofuran7(Table 3)9,10-Dihydro-8H-cyclopenta[b]naphtho[1,2-d]furan (7a). The title compound was obtained as white solid (72%), mp: 125-126 • C. 1 H NMR (400 MHz, CDCl 3 ) δ 8.05 (d, J = 8.1 Hz, 1H), 7.90 (d, J = 8.1 Hz, 1H), 7.59 (s, 2H), 7.51 (t, J = 7.4 Hz, 1H), 7.43 (t, J = 7.5 Hz, 1H), 3.16-2.99 (m, 2H), 2.91 (t, J = 7.2 Hz, 2H), 2.73-2.52 (m, 2H); 13 C NMR (100 MHz" CDCl 3 ) δ 162.0, 157.3, 130.4, 128.5, 127.5, 125.8, 124.2, 124.1, 123.1, 122.2, 121.7, 113.0, 27.9, 25.0, 23.9; HRMS (ESI) calcd for C 15 H 13 O (M + H) + : 209.0961. Found: 209.0960. 8,9,10,11-Tetrahydronaphtho[2,1-b]benzofuran (7b) [/table]
|
10.3390/cancers14215342
|
CCBY
|
9658866
|
36358760
|
s2orc_pubmed_articles
|
Latent Microsporidia Infection Prevalence as a Risk Factor in Colon Cancer Patients
Citation: Redondo, F.; Hurtado-Marcos, C.; Izquierdo, F.; Cuéllar, C.; Fenoy, S.; Sáez, Y.; Magnet, Á.; Galindo-Regal, L.; Uribe, N.; López-Bañeres, M.; et al. Latent Microsporidia Infection Prevalence as a Risk Factor in Colon Cancer Patients. Cancers 2022, 14, 5342.
# Introduction
Many risk factors have been linked to Colon Cancer (CC), including hereditary, environmental, and inflammatory syndromes that affect the gastrointestinal tract [bib_ref] Regional copy number-independent deregulation of transcription in cancer, Stransky [/bib_ref]. Loss of genomic stability in epithelial cells of the intestine is one of these mechanisms since it facilitates the appearance of mutations in genes that encode proteins functionally associated with the control of chromosomal stability, repair of defective DNA, cell cycle control or tumor suppression. It is outstanding that p53 gene alteration is one of the most frequent genetic mutations in human cancer [bib_ref] Molecular origins of cancer: Molecular basis of colorectal cancer, Markowitz [/bib_ref].
It is also essential to know the factors that set in motion the molecular processes that lead to the malignant transformation of cells. In this sense, it has been identified that specific infectious agents can induce cancer in humans. Helicobacter pylori, hepatitis B and C, papilloma and, Epstein-Barr viruses have been linked to the development of stomach, liver, and nasopharyngeal cancer, respectively. Human immunodeficiency virus (HIV) and human herpesvirus 8 (HHV-8) are linked to Kaposi's sarcoma and lymphomas [bib_ref] Cancers attributable to infection in the UK in 2010, Parkin [/bib_ref] [bib_ref] Infections and cancer: Established associations and new hypotheses, Martel [/bib_ref]. Among parasites, Trichomonas vaginalis has been associated with cervical and prostate cancers; Toxoplasma gondii is related to ocular tumors, meningiomas, leukemias, and lymphomas; and Plasmodium spp. have been associated with Burkitt's lymphomas [bib_ref] Parasites and malignancies, a review, with emphasis on digestive cancer induced by..., Benamrouz [/bib_ref] [bib_ref] Trichomonas vaginalis infection-associated risk of cervical cancer: A meta-analysis, Yang [/bib_ref] [bib_ref] Importance of nonenteric protozoan infections in immunocompromised people, Barratt [/bib_ref]. The intracellular intestinal parasite Cryptosporidium parvum has been shown to induce digestive adenocarcinoma, even with low doses of inoculum, in an experimental murine model. It has also been described that cryptosporidiosis is associated with colon carcinomas in patients with human immunodeficiency virus [bib_ref] Cryptosporidium parvum infection in SCID mice infected with only one oocyst: qPCR..., Benamrouz [/bib_ref] [bib_ref] Opportunistic intestinal infections and risk of colorectal cancer among people with AIDS, Shebl [/bib_ref]. Schistosoma haematobium is associated with urinary bladder cancer and Opisthorchis viverrini and Clonorchis sinensis with cholangiocarcinoma. Infection with Schistosoma japonicum is particularly interesting in our study as it has been associated with the development of colorectal cancer. The tumorigenic mechanisms that have been studied involve chronic inflammation, mutations of p53 by the action of toxins, and inhibition of the activity of the immune system, promoting the accumulation of myeloid-derived suppressor cells in lymphoid tissues and the tumor microenvironment [bib_ref] Intratumoral Immunotherapy-Update, Hamid [/bib_ref]. Recently, an increased risk of opportunistic Blastocystis sp. infection was associated with colorectal cancer [bib_ref] Colorectal cancer and Blastocystis sp. infection, Sulżyc-Bielicka [/bib_ref].
Microsporidia are obligate intracellular parasites that cause severe infections in immunosuppressed subjects. Nevertheless, it has also been detected in healthy immunocompetent subjects [bib_ref] Human microsporidiosis, Ruf [/bib_ref]. Pathogenic effects range from diarrheal processes to ocular disease, sinusitis, myositis, encephalitis, and disseminated infection. The presence of intestinal microsporidia has been investigated in patients with cancer undergoing chemotherapy, both in children and adults. Several authors have found a prevalence of 4% to 21% [bib_ref] Incidence of microsporidia in cancer patients, Lono [/bib_ref] [bib_ref] Genetic identification of intestinal microsporidia species in immunocompromised patients in Tunisia, Chabchoub [/bib_ref] [bib_ref] The prevalence of Microsporidium among patients given a diagnosis of cancer, Karaman [/bib_ref] [bib_ref] Intestinal microsporidiosis in Iran: Infection in immune-compromised and immunocompetent patients, Nooshadokht [/bib_ref] [bib_ref] Molecular Phylodiagnosis of Enterocytozoon bieneusi and Encephalitozoon intestinalis in Children with Cancer:..., Ghoyounchi [/bib_ref] , and only one study found a very high prevalence of 69.9% in cancer patients undergoing chemotherapy treatment.
In previous research carried out by our group, it was observed that Encephalitozoon microsporidia modulated the immune response by regulating the apoptosis induction pathway and the cell cycle, inhibiting the activation of the apoptotic protein Caspase-3 and the transcription of the universal protective tumor suppressor protein, p53 [bib_ref] Encephalitozoon microsporidia modulates p53-mediated apoptosis in infected cells, Del Aguila [/bib_ref]. On the other hand, it has been described that Encephalitozoon intestinalis infection increased host cellular mutation frequency in mice, indicating a possible connection between microsporidiosis and cancer induction [bib_ref] Encephalitozoon intestinalis infection increases host cell mutation frequency, Leonard [/bib_ref].
Furthermore, previous studies have discussed the relationship between the etiopathogenesis of Crohn's disease (CD) and microsporidia. In Crohn's, a deficiency of CD3+CD8+γδ T cells has been described, located in the mucous epithelium. These cells are essential for the defense against infections and tumors [bib_ref] Deficit of gammadelta T lymphocytes in the peripheral blood of patients with..., Andreu-Ballester [/bib_ref] [bib_ref] Microsporidia and its relation to Crohn's disease. A retrospective study, Andreu-Ballester [/bib_ref]. These studies are critical and relevant since individuals with CD are at risk of developing CC [bib_ref] Cancer risk in patients with inflammatory bowel disease: A populationbased study, Bernstein [/bib_ref]. Moreover, patients with newly diagnosed treatment-naïve infiltrating colonic adenocarcinoma have a significant decrease in γδ T cells compared to healthy control volunteers [bib_ref] Differences in circulating γδ T cells in patients with primary colon cancer..., Andreu-Ballester [/bib_ref].
Based on the above considerations, the aim of this study was to investigate the presence of microsporidia in tissues of patients with CC without previous specific anti-tumor treatment.
# Materials and methods
# Ethics statement
The Hospital Research Ethics Committee approved the study, and each patient signed an informed consent document. This epidemiological survey was carried out in compliance with fundamental ethical principles, including those reflected in the Charter of Fundamental Rights of the European Union and the European Convention on Human Rights and its supplementary protocols. All participants attested to their involvement in this clinical research study through written informed consent. The consent was evaluated and approved by the Research Ethics Committee of the Arnau de Vilanova Hospital (CEIm), following the recommendations of the Spanish Bioethics Committee, the Spanish legislation on Biomedical Research (Law 14/2007, of July 3), and Personal Data Protection (Spanish Law 3/2018 and European Law UE676/2018). These laws define that access to the clinical record for judicial, epidemiological, public health, research, or educational purposes carries an obligation to keep the patient's identification data separated from clinical and healthcare data; therefore, anonymity is ensured.
## Subjects of study and clinical samples
Eighty-seven newly diagnosed colon cancer patients were recruited at the Arnau de Vilanova Hospital in Valencia (Spain). Colon biopsies were obtained from both tumor and normal peritumoral tissue through a surgical procedure from 87 patients, [fig_ref] Table 1: Characteristics of patients with Colon Cancer [/fig_ref]. Tissues were also collected from 25 control subjects with no pathological findings in the colon cancer screening program. Tissue samples were separated into two equal parts. One was kept at −80 - C to extract the DNA, and the other part was paraffinized to perform the histological sections analyzed by IFAT. For this reason, before performing DNA extraction of the sample, tissue imprints were made on new slides. These imprints subsequently served to complete the immunofluorescence tests. For the serological study, seventy-two serum samples were obtained from CC patients and 25 from healthy controls.
Finally, the following exclusion criteria were taken into account, both in patients with CC and in controls: absence of infectious, inflammatory, or autoimmune disease, known immunodeficiency, and a lack of specific anti-tumor or immunosuppressive treatment in CC patients; and the application of some vaccine for at least one year.
## Immunofluorescence antibody test (ifat) 2.3.1. deparaffination and imprints
Deparaffination of tissue sections was carried out in several dissolvents for ten minutes in each one of the following steps: xylol, xylol-ethanol, ethanol absolute, ethanol 90%, ethanol 70%, and distilled water [bib_ref] Microsporidia and its relation to Crohn's disease. A retrospective study, Andreu-Ballester [/bib_ref]. Both deparaffined tissues and imprints were aired dry and fixed with methanol-acetone. The contours of the sample were marked with a DAKOPEN ® marker (Biolabs) to contain the IFAT solutions.
## Assay
For the assay, two primary monoclonal antibodies (MAb) were used. The MAb 2C2 (murine origin and isotype IgG2a) recognizes the wall's exospore and the developing states of the Encephalitozoon species [bib_ref] Production and characterization of monoclonal antibodies against Encephalitozoon intestinalis and Encephalitozoon sp...., Izquierdo [/bib_ref]. The MAb 6E52D9 (murine origin and isotype IgG2a) recognizes the exospore of the wall of Enterocytozoon bieneusi [bib_ref] Production of monoclonal antibodies directed against the microsporidium Enterocytozoon bieneusi, Accoceberry [/bib_ref]. Monoclonal antibodies (undiluted supernatant) were added to each well (tissue or imprint) on each slide and incubated at 37 - C for 60 min in a wet chamber. Slides were washed three times in distilled water and then were air-dried at room temperature.
The second antibody Fluorescein Isothiocyanate conjugated (Sigma, St. Louis, MO, USA), was diluted following the manufacturer's instructions. The slides were incubated again at 37 - C for 60 min, washed three times in distilled water, and air dried at room temperature in the dark. The mounting liquid (PVA/Dabco) was added to the slides, covered by a cover slip, and examined at 40× with a Nikon immunofluorescent microscope. Positive controls were included using species homologous to microsporidia spores of each monoclonal antibody used. The samples and the positive controls were stored in the dark at 4 - C until the reading.
## Encephalitozoon cuniculi antigen and determination of specific antibodies (elisa)
E. cuniculi antigen was obtained from E. cuniculi (USP-A1) spores following the Aguila et al. protocol [bib_ref] Seroprevalence of anti-Encephalitozoon antibodies in Spanish immunocompetent subjects, Del Aguila [/bib_ref]. Briefly, spores were disrupted using glass beads, 2.5% SDS, and 2% mercaptoethanol. Soluble antigens were obtained from the supernatant after centrifugation. Protein content was determined by the Bradford method and adjusted to 0.8 mg/mL to coat ELISA plates. Duplicate dilutions of human serum at 1/100 in PBS-Tween, containing 0.1% BSA, were added and incubated. Horseradish peroxidase (HRP) conjugated goat antihuman total immunoglobulins (Ig's), IgG, IgM, or IgA (Biosource International, Camarillo, CA) were used. For IgE determination, test serum was added in duplicate at a 1/2 dilution. A murine monoclonal antibody against a human IgE chain (IgG1k, E21A11, INGENASA, Madrid, Spain) was added and incubated, followed by a goat anti-mouse IgG1 (gamma) HRP conjugate (CALTAG Laboratories, Burlingame, CA, USA) [bib_ref] Immunoglobulins anti-Anisakis simplex in patients with gastrointestinal diseases, Gutiérrez [/bib_ref] [bib_ref] Gastro-allergic anisakiasis as a consequence of simultaneous primary and secondary immune response, Daschner [/bib_ref].
## Dna extraction and detection of microsporidia by multiplex real-time pcr
DNA from the 174 colon biopsies (87 patients) was extracted using the Dneasy Blood and Tissue kit, Qiagen ® . Real-Time PCR was carried out with the SYBR Green reagent following the protocol of Polley et al. [bib_ref] Detection and species identification of microsporidial infections using SYBR Green real-time PCR, Polley [/bib_ref] , modified by Andreu-Ballester et al. [bib_ref] Encephalitozoon intestinalis infection increases host cell mutation frequency, Leonard [/bib_ref]. The following primers were utilized: MsRTF (5 -CAGGTTGATTCTGCCTGACG-3 ) and MsRTR (5 -CCATCTCTCAGGCTCCCTCT-3 ), which ensure amplification of Encephalitozoon intestinalis, Encephalitozoon cuniculi, Encephalitozoon hellem, and Enterocytozoon bieneusi SSU-rRNA sequences. Forward and reverse primers (10 pmol each) were used in 20 µL reactions containing 5 µL template DNA. The cycling conditions were 95 - C for 10 s, 60 - C for 20 s, and 72 - C for 20 s. The assay was set up in duplicate for each sample and run for 35 cycles.
# Data analysis
A Mann-Whitney U test was used to study the quantitative variables. Odds Ratio's (OR) were used to study the relation between positive and negative microsporidia with CC patients and healthy controls. The level of significance was taken as a P-value of less than 0.05 (bilateral contrast). Data were analyzed using the statistical software SPSS, version 19.
# Results
## Prevalence of microsporidia in tissues
Thirty-six (41.3%) patients of the 87 studied with CC (by PCR and IFAT) had microsporidia in their tissues [fig_ref] Table 2: Microsporidia species detection in CC patient tissue [/fig_ref] and. Twenty-six (30%) of these patients with CC were positive by PCR vs. none in the tissues of control subjects (25 samples with negative PCR), Odds Ratio (OR) = 36.1 (IC 95% = 2.1-613.0) p < 0.013. Thus, the probability of finding microsporidia in colon tissue from patients with CC is over 36 times higher than in tissues from healthy subjects.
## Determination of specific antibodies
The levels of anti-E. cuniculi Ig's, IgG, IgM, and IgA were determined in the serum of 72 patients with CC, and in 64 sera for IgE. A significant increase in IgE and IgG anti-E. cuniculi was observed in patients with CC vs healthy subjects (p < 0.01 and p < 0.05, respectively).
Serum that presented an optical density (OD) above the mean plus the standard deviation of the control group was considered as positive when comparing the CC and control groups. Firstly, 42.2% of the serum from patients with CC showed a positive result for IgE, followed by IgG (25%) and IgA (22.2%). Significant differences among percentages in CC and controls were observed in the case of IgG and IgE isotypes (p < 0.05). The lowest levels were found in the cases of IgM and Ig's with percentages of 9.7% and 11.1% of positive serum, respectively.
Of the 72 serums from CC patients studied, 40 (55.6%) were positive for some of the specific isotypes analyzed by ELISA. Of the 40 CC sera (55.6%) that were positive for some class of immunoglobulin, 35 were analyzed in tissue samples. Of these 35 CC patients, 12 (34.5%) were positive by PCR or IFAT in some tissue.
If we consider the patients who had a positive result on PCR and/or IFAT, 58.3% of them were positive for some of the isotypes studied by ELISA. On the one hand, from the serum of the patients who had a positive result in PCR, 55.6% of them showed anti-E. cuniculi antibodies. On the other hand, of the serum samples that had a positive result on IFAT, 60% of them showed anti-E. cuniculi antibodies. Finally, in the patients that were both positive via PCR and IFAT, 62.5% of them showed some positive antibody isotype by ELISA against E. cuniculi. Furthermore, a significant increase in specific IgE in patients with CC and a positive PCR result in tissues versus patients with CC and a negative PCR results in tissues (p = 0.007) was found.
# Discussion
This report is the first study investigating the relationship of microsporidia with CC and examining the presence of this parasite in surrounding healthy and tumor tissues. A prevalence of 41.3% of microsporidia in the colon tissue of patients with CC was found, compared to 0% of microsporidia in the colon tissues of healthy subjects. The prevalence rate of microsporidia infection in immunocompetent subjects ranges from 0% to 50% in different countries, although most studies are based on the detection of spores in feces with a few in urine [bib_ref] Intestinal microsporidiosis in Iran: Infection in immune-compromised and immunocompetent patients, Nooshadokht [/bib_ref] [bib_ref] Latent microsporidial infection in immunocompetent individuals-A longitudinal study, Sak [/bib_ref] [bib_ref] Encephalitozoon cuniculi infection among immunocompromised and immunocompetent humans in Egypt, Abu-Akkada [/bib_ref] [bib_ref] Identification of opportunistic enteric parasites among immunocompetent patients with diarrhoea from Northern..., Ghoshal [/bib_ref] [bib_ref] Molecular detection and species identification of Enterocytozoon bieneusi isolated from immunocompetent Orang..., Ashikin [/bib_ref].
To date, only one study has analyzed a control group of small bowel biopsies from healthy subjects, finding a 0% prevalence of microsporidia [bib_ref] Microsporidia and its relation to Crohn's disease. A retrospective study, Andreu-Ballester [/bib_ref] , similar to the findings of our current study. Additionally, finding microsporidia in feces or urine does not imply that the parasite is present in the tissues, causing infection and disease.
There are about 1400 described species of microsporidia, of which 14 have been shown to infect humans. In the present study, primers that detect Encephalitozoon sp. (E. hellem, E. intestinalis, and E. cuniculi) and Enterocytozoon bieneusi were used. For this reason, there could be some species present in the tissues that we were unable to detect, so the prevalence of microsporidia in patients with CC could be higher.
Research looking into the molecular basis of the possible origin of cancer due to microsporidia infection showed that the parasite was capable of inhibiting apoptosis through different mechanisms. In this "in vitro" study, the mechanisms by which microsporidia regulate and modulate apoptosis were studied. It was observed that the cells infected with Encephalitozoon sp. did not form the exclusion product caspase-3 (which is an apoptosis pathway). This fact would indicate that microsporidia can modulate and regulate this activation pathway, suggesting that the parasite could modulate, or even inhibit, the apoptotic response of the host cell [bib_ref] Encephalitozoon microsporidia modulates p53-mediated apoptosis in infected cells, Del Aguila [/bib_ref].
On the other hand, it has been described in patients with intestinal infections that the levels of hydrogen peroxide, advanced protein oxidation products, and malondialdehyde are higher than in uninfected people [bib_ref] Elevated levels of urinary hydrogen peroxide, advanced oxidative protein product (AOPP) and..., Chandramathi [/bib_ref]. This result would indicate more significant oxidative stress in the organism and, therefore, a greater probability of damage to the cell's genetic material, thus increasing the risk of cancer [bib_ref] DNA damage checkpoints in stem cells, ageing and cáncer, Sperka [/bib_ref]. Another study revealed that microsporidia infection increased the frequency of nuclear mutations and that the increase in these mutations depended on the viable spores. The conclusion of this study showed that microsporidia infection is genotoxic to the host cell. This means that, either indirectly or directly, it would induce damage to cellular DNA, producing mutations that could result in cancer development in the patient in the long term [bib_ref] Encephalitozoon intestinalis infection increases host cell mutation frequency, Leonard [/bib_ref].
Concerning the present work and specifically the detection of microsporidia spores in the analyzed tissue samples, the combination of immunological and molecular diagnostic techniques was shown to be a helpful tool that allows for the obtainment of highly reliable results. Due to the complexity of the diagnosis of this group of parasites, the complementarity of different techniques is necessary to reach a coherent and accurate result. Therefore, in the present work it was decided to use immunological diagnostic methods using monoclonal antibodies (IFAT) and molecular approaches to reinforce the results obtained. Both techniques have advantages and disadvantages, but the combination of both has allowed for consistent interpretation and justification of the obtained results.
The results obtained in the tissue samples studied by IFAT and real-time PCR reflected four different sets of results. The first group includes the samples that were positive in both IFAT and PCR. The positivity of both techniques would show a correct correlation, confirming the result obtained. The second group of samples with double negativity in IFAT and PCR would reflect the absence of the parasite in the samples studied. The third group with a negative IFAT and a positive PCR would be due to the lack of sensitivity of the IFAT and/or low parasite load of the sample that the PCR has detected due to its greater sensitivity. Finally, the fourth group with a positive IFAT and a negative PCR would be a result related to the possible presence of inhibitors in the sample, degradation of the genomic material of the sample, or the extrusion of the spores with the loss of parasite DNA. It should be noted that the two monoclonal antibodies used in the present work have allowed for the detection of the four most prevalent species in humans, as well as the possibility of identifying foci of infection or intracellular states of the parasite indicative of an active infection in the patient.
Regarding the study of seroprevalence and characterization of specific antibodies, the increase in anti-Encephalitozoon IgE in patients with CC compared to healthy subjects is noteworthy. Also, the significant increase in patients with CC and positive PCR vs. patients with negative PCR should be considered. The relationship of microsporidia with Crohn's disease through PCR has already been proven, as well as the increase in anti-Encephalitozoon IgE for the same patients [bib_ref] Microsporidia and its relation to Crohn's disease. A retrospective study, Andreu-Ballester [/bib_ref].
Moreover, Crohn's disease could be a precursor to CC. Both pathologies have another common factor: the deficiency of γδ T cells and high microsporidia prevalence [bib_ref] Microsporidia and its relation to Crohn's disease. A retrospective study, Andreu-Ballester [/bib_ref] [bib_ref] Differences in circulating γδ T cells in patients with primary colon cancer..., Andreu-Ballester [/bib_ref]. Therefore, there are a number of factors that can justify its relationship with the appearance of colon cancer. The deficiency of γδ T cells would favor infection by microsporidia, which in turn, by different mechanisms [bib_ref] Encephalitozoon microsporidia modulates p53-mediated apoptosis in infected cells, Del Aguila [/bib_ref] , would induce malignant transformation, which exacerbates the deficient defense against the tumor by the γδ T cells [bib_ref] Differences in circulating γδ T cells in patients with primary colon cancer..., Andreu-Ballester [/bib_ref]. According to this hypothesis, the sequence of events includes first a γδ T cell deficit, then infection by microsporidia, and lastly the development of cancer. It seems essential to study the relationship of microsporidia infection with lymphocyte populations in CC patients. This study could give a clue as to who is the inducer of the tumor process.
Despite this, the high levels of specific IgG found in CC patients would indicate continuous and previous contact with the parasite. This could be due to successive exposures or a chronic infection due to an immune deficit. These results are consistent when compared with the results obtained for IgM levels. In this case, very low levels were obtained that would indicate low rates of primary infection in these patients. These values would confirm that the presence of the parasite predates the cancer diagnosis and would result in the microsporidia-cancer association. In the case of IgA anti-Encephalitozoon antibodies, slightly high levels were obtained that would confirm the continuous exposure to the parasite antigens. This could stimulate the production of this isotype, associated with the defense of the intestinal mucosa and characteristic of the response against enteric parasites.
Finally, it is necessary to highlight the high levels of IgE observed that reach a positivity rate close to 50%. This result shows that the parasite is active within the tumor lesion, continuously stimulating the immunoglobulin class change with its antigens. In this sense, it is necessary to mention that the production of this immunoglobulin is associated with the extrusion mechanism of the spores [bib_ref] Recognition profiles of microsporidian Encephalitozoon cuniculi polar tube protein 1 with human..., Furuya [/bib_ref]. The presence of specific IgE was independent of the microsporidia species, appearing in patients in whom PCR had detected the presence of the parasite in both tumor and healthy tissue. This shows that anti-Encephalitozoon IgE is a highly sensitive and specific marker for microsporidia in tissues.
# Conclusions
Microsporidium infection was found in the tissues of more than 40% of CC patients compared to 0% of healthy subjects. Significantly higher levels of IgE and IgG anti-E.cuniculi were observed in CC patients with significantly higher rates of positivity compared to the healthy control group. The high seroprevalence and prevalence of microsporidia in tissues of patients with CC would suggest a relationship between microsporidia infection and the etiopathogenesis of CC. These laws define that access to the clinical record for judicial, epidemiological, public health, research, or educational purposes carry an obligation to keep the patient's identification data separated from clinical and healthcare data; therefore, anonymity is ensured.
# Informed consent statement:
Written informed consent has been obtained from the patient(s) to publish this paper.
[fig] Figure 1: (A) IFAT of microsporidia species. A: Positive control of MAb 2C2 with purified spores of E. cuniculi (40×); B: Positive control of MAb 6E52D9 with spores of E. bieneusi (100×); C: Spores of Encephalitozoon sp. in positive biopsies by IFAT with MAb 2C2 (40×); D and E: Spores of Encephalitozoon sp. in positive biopsies by IFAT with MAb 2C2 (100×); F: Spore of E. bieneusi in positive biopsies by IFAT with MAb 6E52D9 (40×). G and H: Spores of E. bieneusi in positive biopsies by IFAT with MAb 6E52D9 (100×). (B) Anti-E. cuniculi antibody levels (O.D. = optical density) measured by ELISA in colon cancer patients (CC, N = 87) and healthy controls (H, N = 25). Ig's: total immunoglobulins. (C) Percentage of serum samples considered as positive against E. cuniculi in colon cancer patients (CC) compared to healthy controls (HC) obtained by ELISA. Ig´s: total immunoglobulins. Double T bars denote standard deviation (*** p < 0.001, ** p < 0.01). A Mann-Whitney U test was used. [/fig]
[fig] Author: Contributions: Conceptualization, C.H.-M., F.I., C.D.Á. and J.C.A.-B.; methodology, F.R, Y.S., C.H.-M., F.I., C.D.Á., C.C. and J.C.A.-B.; software, J.C.A.-B.; validation, C.H.-M., F.I., C.D.Á. and J.C.A.-B.; investigation, C.H.-M., F.I., C.D.Á. and J.C.A.-B.; resources, L.G.-R., A.I.J., S.F., Á.M., N.U. and M.L.-B.; data curation, L.G.-R.; writing-original draft preparation, C.H.-M., F.R., F.I., C.D.Á., C.C. and J.C.A.-B.; writing-review and editing, C.H.-M., F.I., C.D.Á. and J.C.A.-B.; visualization, C.H.-M., F.I., C.D.Á. and J.C.A.-B.; supervision, A.L.-C.; project administration, C.H.-M., F.I., C.D.Á. and J.C.A.-B.; funding acquisition, C.H.-M., F.I. and C.D.Á. All authors have read and agreed to the published version of the manuscript. Funding: Banco Santander Foundation-CEU Foundation. Research Project: "Microsporidia as a risk factor in colorectal cancer pathology." Reference Project: FUSP-BS-PPC11/2014. Cell Culture Unit. USPCEU.Refrence: SAI-UCC. Institutional Review Board Statement: The Hospital Research Ethics Committee approved (14/2015) the study, and each patient signed an informed consent document. This epidemiological survey was carried out in compliance with fundamental ethical principles, including those reflected in the Charter of Fundamental Rights of the European Union and the European Convention on Human Rights and its Supplementary Protocols. All participants attested to their involvement in this clinical research through written informed consent. Consent was evaluated and approved by the Research Ethics Committee of the Arnau de Vilanova Hospital (CEIm), following the recommendations of the Spanish Bioethics Committee, the Spanish legislation on Biomedical Research (Law 14/2007, of Jul 3) and Personal Data Protection (Spanish Law 3/2018 and European Law UE676/2018). [/fig]
[table] Table 1: Characteristics of patients with Colon Cancer (CC) (N = 87). [/table]
[table] Table 2: Microsporidia species detection in CC patient tissue (N = 87). [/table]
|
10.1007/s13238-018-0510-2
|
CCBY
|
6160385
|
29497989
|
s2orc_pubmed_articles
|
SCFβ-TRCP E3 ubiquitin ligase targets the tumor suppressor ZNRF3 for ubiquitination and degradation
HeLa IB: ZNRF3 IB: Tubulin IB: p27 HeLa -2 5 10 MLN4924 (µM) IB: ZNRF3 IB: Tubulin IB: p27 -20 30 40 MG132 (µM) D E Time after Wnt3a treatment (hr) shGFP shTRCP1 0 0.25 0.5 1 2 IB: β-catenin IB: Tubulin HeLa IB: β-TRCP1 Hours after Wnt3a 0 0.25 0.5 1 2 Relative β-catenin protein level F G . ZNRF3/RNF43 interacts with TRCP1 (A-B) IB analysis of WCL derived from HeLa cells treated with MG132 or MLN4924 different does for 12 h before harvesting.
## (c-d) ib analysis of ip and wcl derived from hek293t cells transfected with
indicated constructs and were treated with MG132 for 12 h before harvesting. (E) IB analysis of WCL derived from HeLa cells infected with lentivirus encoding control (sh-Scr) or multiple independent shRNAs against TRCP (sh-TRCP). Infected cells were selected with 1 g/mL puromycin for 72 hr to eliminate non-infected cells before harvesting. (F-G) HeLa cells were lentivirally infected with shRNA against β-TRCP1 and selected with puromycin (1 g/mL) for 5 days, and the resulting cells were stimulated with wnt3a proteins for different time point and subjected to IB analysis (F). The relative β-catenin protein levels were quantified with tubulin and normalized with time=0 in G. with ZNRF3-WT or SSG-mut ZNRF3 retrovirus, and selected with hygromycin (200 g/mL) for 5 days. The resulting cells were stimulated with or without wnt3a and subjected for IB analysis.
[fig] Figure S1, Figure S2: ZNRF3CKI, but not CKII, promotes the degradation of ZNRF Figure S3. β-TRCP does not promote the degradation of RNF43 [/fig]
[fig] Figure: S4. β-TRCP promotes the degradation of ZNRF3 in a degron dependent manner [/fig]
|
10.1186/s12915-021-01065-6
|
CCBY
|
8212472
|
34144713
|
s2orc_pubmed_articles
|
Publisher Correction to: MATISSE: a method for improved single cell segmentation in imaging mass cytometry
Publisher Correction to: MATISSE: a method for improved single cell segmentation in imaging mass cytometryPublisher Correction to: MATISSE: a method for improved single cell segmentation in imaging mass cytometry
P U B L I S H E R C O R R E C T I O N Publisher Correction to: BMC Biol 19, 99 (2021)
The original article can be found online at https://doi.org/10.1186/s12915-021-01043-y. Combining fluorescence microscopy with multiplex IMC data of colorectal tissue advances quality of single cell segmentation. a Cartoon describing MATISSE, a novel pipeline adding microscopic imaging to multiplex IMC analysis and downstream segmentation. In short: tissue sections on slides were stained using isotope-conjugated primary antibodies, DNA intercalator, and DAPI. The tissue was first scanned using a fluorescent microscope and then processed with IMC. Data produced by both techniques is aligned using the nuclear staining. Nuclear and membranous pixel probability maps are produced based on the fluorescent images and IMC data respectively. These probability maps are used to generate a segmentation map, where all detected cells are included. b Representative images of DNA intercalator on a colorectal tissue section analyzed by Ir193 labeling and IMC (left) or DAPI labeling and fluorescent microscopy . c IMC-only (IMC) and MATISSE cell segmentation (MATISSE) were performed, and shown are the different predicted outlines on a representative image of Ir193 labeling. Arrows indicate areas with cell fragmentation. d Display of a large region of interest showing an overlay of the predicted cell outlines (pink) upon IMC or MATISSE segmentation on a representative IMC image of DNA-Ir193 labeling of colorectal tissue. Highlighted in yellow is the approximate position of the basement membrane surrounding the epithelial monolayer. Scale bar 25 μm. e Cell density was calculated as the number of cells within a radius of 10 μM from the center of each single cell . This number is displayed with a color code for each cell in the representative image
|
10.1093/nar/gkaa751
|
CCBY
|
7641730
|
32986843
|
s2orc_pubmed_articles
|
Structural basis of DNA replication origin recognition by human Orc6 protein binding with DNA
[fig_ref] Figure S4: The 1 H, 15 N HSQC NMR spectrum of full length HsOrc6... [/fig_ref] [bib_ref] Jalview Version 2--a multiple sequence alignment editor and analysis workbench, Waterhouse [/bib_ref]
## Figure. s3
The 1 H, 15 N HSQC NMR spectra of (A) HsOrc6-N (residue 1-94) titrated with HsOrc6-M (residue 95-187) and (B) HsOrc6-N+M (residue 1-187) titrated with Orc6-C (residue 188-252). Spectra are colored as: free protein (red), with ratio 1:2 (green) and with ration 1:5 (blue). [fig_ref] Figure. S11: Superposition of 1 H, 15 N-HSQC spectra of HsOrc6-DBD [/fig_ref] were plotted as a function of the HsOrc6-DBD:DNA molar ratios. The dissociation constants (KD) were obtained by fitting the experimental data as described in Material and Methods.
## Figure. s15
The HADDCOK model of full length HsOrc6/DNA complex. The secondary structure of each domain is colored as green in HsOrc6-N, magenta in HsOrc6-M and in purple for HsOrc6-C, while the loops are colored gray. The region containing Arg198-Lys199-Arg200-Lys201 is shown in red.
[fig] Figure.: S2 (A) Differences in chemical shifts of individual HsOrc6 domains with full length HsOrc6 are plotted versus residue number. (B) Differences in chemical shifts of the HsOrc6 construct containing residues 1-207 with full length HsOrc6 are plotted versus residue number. [/fig]
[fig] Figure S4: The 1 H, 15 N HSQC NMR spectrum of full length HsOrc6 with assignment. [/fig]
[fig] Figure: S5 { 1 H}-15 N heteronuclear NOE values of full length HsOrc6 in apo form measured at 298 K and pH 6.5. [/fig]
|
10.1371/journal.pone.0155518
|
CCBY
|
4981393
|
27513473
|
s2orc_pubmed_articles
|
SRCIN1 Suppressed Osteosarcoma Cell Proliferation and Invasion
SRCIN1 (SRC kinase signalling inhibitor 1) is a new tumor suppressor gene. Previous studies showed that SRCIN1 played a tumor suppressor role in the development of lung cancer and breast cancer. However, the role of SRCIN1 in osteosarcoma is still unknown. In this study, we demonstrated that SRCIN1 was downregulated in osteosarcoma cell lines compared with osteoblastic cell line. Moreover, SRCIN1 was downregulated in osteosarcoma tissues compared with the adjacent tissues. Further investigation revealed that overexpression of SRCIN1 inhibited the osteosarcoma cell line MG-63 proliferation. This effect was confirmed by measuring the ki-67 and PCNA expression. SRCIN1 overexpression promoted E-cadherin expression and suppressed N-cadherin, Vimentin and Snail expression, suggesting that SRCIN1 overexpression inhibited EMT of the osteosarcoma cell. In addition, ectopic expression of SRCIN1 inhibited the MG-63 cell colony formation and invasion. These data suggested that SRCIN1 acted as a tumor suppressor gene in the development of osteosarcoma.
# Introduction
Osteosarcoma is one of the most common primary bone malignancies in young adults and adolescents, with an estimated 5.6 per million children suffering from osteosarcoma yearly [bib_ref] Overexpression of metadherin mediates metastasis of osteosarcoma by regulating epithelial-mesenchymal transition, Tang [/bib_ref] [bib_ref] MicroRNA-34c Inversely Couples the Biological Functions of the Runt-related Transcription Factor RUNX2..., Van Der Deen [/bib_ref] [bib_ref] The tumor suppressor role of miR-124 in osteosarcoma, Geng [/bib_ref]. It occurs mostly in long extremity bone and around regions with active bone growth [bib_ref] Prognostic value of the microRNA-29 family in patients with primary osteosarcomas, Hong [/bib_ref] [bib_ref] A polymorphism site in the premiR34a coding region reduces miR34a expression and..., Lv [/bib_ref] [bib_ref] Serum levels of microRNA-133b and microRNA-206 expression predict prognosis in patients with..., Zhang [/bib_ref]. Despite the advances in treatment strategies of osteosarcoma, the 5-year survival rate of osteosarcoma patients is still poor [bib_ref] MicroRNA-34c Inversely Couples the Biological Functions of the Runt-related Transcription Factor RUNX2..., Van Der Deen [/bib_ref] [bib_ref] microRNA-377 suppresses the proliferation of human osteosarcoma MG-63 cells by targeting CDK6, Wang [/bib_ref] [bib_ref] MicroRNA-153 Inhibits Osteosarcoma Cells Proliferation and Invasion by Targeting TGF-beta2, Niu [/bib_ref] [bib_ref] microRNA25 promotes osteosarcoma cell proliferation by targeting the cellcycle inhibitor p27, Wang [/bib_ref]. Thus, it is important to find the molecular mechanisms underlying the development of osteosarcoma and to identify novel biomarkers for the treatment, diagnosis and prognosis of this tumor [bib_ref] MicroRNAs in osteosarcoma, Zhang [/bib_ref] [bib_ref] miR-23a suppresses proliferation of osteosarcoma cells by targeting SATB1, Wang [/bib_ref] [bib_ref] Tumor-suppressive microRNA-let-7a inhibits cell proliferation via targeting of E2F2 in osteosarcoma cells, Iwasaki [/bib_ref] [bib_ref] Tumour biology: the journal of the International Society for Oncodevelopmental Biology and..., Miao [/bib_ref]. SRCIN1 (SRC kinase signalling inhibitor 1), also named as p140 Cas-associated protein (p140CAP), contains two regions of highly charged amino acids, two proline-rich regions and two coiled-coil domains [bib_ref] The adaptor proteins p140CAP and p130CAS as molecular hubs in cell migration..., Stefano [/bib_ref] [bib_ref] p140Cap suppresses the invasive properties of highly metastatic MTLn3-EGFR cells via impaired..., Damiano [/bib_ref] [bib_ref] p140Cap dual regulation of E-cadherin/EGFR cross-talk and Ras signalling in tumour cell..., Damiano [/bib_ref]. Previous studies demonstrated that SRCIN1 played an important role in Src inactivation and acted as a tumor suppressor gene in cancers [bib_ref] p140Cap suppresses the invasive properties of highly metastatic MTLn3-EGFR cells via impaired..., Damiano [/bib_ref] [bib_ref] SNIP/p140Cap mRNA expression is an unfavourable prognostic factor in breast cancer and..., Kennedy [/bib_ref]. For example, Cao et al [bib_ref] miR-150 promotes the proliferation and migration of lung cancer cells by targeting..., Cao [/bib_ref]. demonstrated that miR-150 acted as an oncogene through repressing SRCIN1 translation in lung cancer. Sharma et al [bib_ref] Identification of two regions in the p140Cap adaptor protein that retain the..., Sharma [/bib_ref]. showed that SRCIN1 inhibited tumor growth and impaired invasive properties of cancer cells by regulating the tyrosine kinase Src or E-cadherin/EGFR signaling pathways. In addition, the expression of SRCIN1 was inversely correlated with tumor malignancy in breast cancer. Damiano et al [bib_ref] p140Cap suppresses the invasive properties of highly metastatic MTLn3-EGFR cells via impaired..., Damiano [/bib_ref]. showed that SRCIN1 inhibited the highly metastatic breast carcinoma cells invasion by repressing cortactin-dependent cell motility. However, the role of SRCIN1 in osteosarcoma is still unknown.
In our study, we demonstrated that SRCIN1 was downregulated in the osteosarcoma cell lines and tissues. Overexpression of SRCIN1 inhibited the osteosarcoma cell proliferation and EMT. In addition, ectopic expression of SRCIN1 inhibited the MG-63 cell colony formation and invasion.
# Materials and methods
## Clinical specimens and cell line cultured and transfection
Thirty osteosarcoma and adjacent nontumor bone tissues were collected from patients who underwent surgery in our hospital. Samples were immediately snap-frozen in liquid nitrogen. All of the patients have written informed consent to participate in this study. This study was approved by the Ethics Committee of The Fourth Hospital of Harbin Medical University and complied with the Declaration of Helsinki. Human osteosarcoma cell lines (U2OS, MG63, SAOS-2 and SOSP-9607) and one osteoblastic cell line (hFOB) were obtained from the ATCC (American Tissue Culture Collection). Lipofectamine 2000 (Dharmacon, TX, USA) was used to perform to cell transfection.
## Rna extraction and qrt-pcr
Total RNA from tissues and cells were isolated by using Trizol reagent (Invitrogen, Calsbad, CA, USA). Relative expression levels of SRCIN1 mRNA were measured by real-time PCR on the iQ5 Real-Time PCR Detection System (Bio-Rad, California, USA). Quantitative PCR primer for SRCIN1 was forward: 5'-AGCCCCGACAAAAGCAAAC-3' and reverse: 5'-CCAAAGGAAGTCAATACAGGGATAG-3'; GAPDH was forward: 5'-AATGGGCAGCCGTTAG GAAA-3' and reverse: 5'-TGAAGGGGTCATTGATGGCA-3'. GAPDH was used to as an endogenous control.
## Cell proliferation and colony formation
Cell Counting Kit-8 (CCK-8)(Dojindo; Kumamoto, Japan) was performed to measure the cell proliferation. Cells were cultured in 96-well plates for 1, 2or 3 days. The absorbance was detected at a wave length of 450 nm. For cell colony analysis, the cells were cultured at 1000cells/plate density and continue to seed for 2 weeks.
## Cell invasion
For cell invasion analysis, cells were seeded on the top side of transwell chambers coated with Matrigel (BD Bioscience). Mediumin the lower chambers containing 10% FBS acted as the chemo-attractant. After 24 h, the cells moving to the lower side of the membrane were fixed, stained with crystal violet and then counted by a microscope (BX51 Olympus, Japan).
## Western blot
Western blot analysis was measure according to previous studies. Proteins were isolated and detected using the BCA kit (Thermo Scientific, Rockford, IL). Proteins were separated by 10% SDS-PAGE and then transferred to PVDF membranes (Millipore, Danvers, MA). Membranes were cultured with primary antibodies GAPDH (Santa Cruz, CA) or SRCIN1 (Cell Signaling Technology). GAPDH was used as a loading control.
# Statistical analysis
Data was shown as means±SD. Statistics was measured using ANOVA or Student's t-test using SPSS17.0. Statistical significance was shown asP<0.05.
# Results
## Srcin1 was downregulated in the osteosarcoma cells
One representative patient was diagnosed as osteosarcomausing HE staining [fig_ref] Fig 1: The expression of SRCIN1 was downregulated in the osteosarcoma cells [/fig_ref]. The mRNA expression of SRCIN1 was lower in the osteosarcoma cell lines (MG63, U2OS, SAOS-2 and HOS) than in the one osteoblastic cell line (hFOB) [fig_ref] Fig 1: The expression of SRCIN1 was downregulated in the osteosarcoma cells [/fig_ref]. In line with this, the protein expression of SRCIN1 was lower in the osteosarcoma cell lines (MG63, U2OS, SAOS-2 and HOS) than in hFOB [fig_ref] Fig 1: The expression of SRCIN1 was downregulated in the osteosarcoma cells [/fig_ref].
SRCIN1 expression was reduced in the osteosarcoma tissues SRCIN1 expression was also lower in the osteosarcoma tissues than in the adjacent nontumor tissues [fig_ref] Fig 2: SRCIN1 expression was reduced in the osteosarcoma tissues [/fig_ref]. Meanwhile, SRCIN1 expression was downregulated in 29 patients (29/35, 82%) compared withthe adjacent tissues. We also confirmed that the protein expression of SRCIN1 was downregualted in the osteosarcoma tissues [fig_ref] Fig 2: SRCIN1 expression was reduced in the osteosarcoma tissues [/fig_ref].
## Srcin1 suppressed the osteosarcoma cell proliferation
As shown in [fig_ref] Fig 3: SRCIN1 suppressed the osteosarcoma cell proliferation [/fig_ref] , SRCIN1 expression was increased after transfected with SRCIN1 vector. Overexpression of SRCIN1 suppressed cell proliferation in the osteosarcoma cell line MG63 [fig_ref] Fig 3: SRCIN1 suppressed the osteosarcoma cell proliferation [/fig_ref]. As shown in [fig_ref] Fig 3: SRCIN1 suppressed the osteosarcoma cell proliferation [/fig_ref] , the expression of ki-67 was decreased after transfected with SRCIN1 vector. Overexpression of SRCIN1 inhibited the PCNA expression in the MG-63 cell [fig_ref] Fig 3: SRCIN1 suppressed the osteosarcoma cell proliferation [/fig_ref].
## Srcin1 inhibited epithelial-mesenchymal transition of the osteosarcoma cell
Overexpression of SRCIN1 enhanced E-cadherin mRNA expression while suppressed N-cadherin, Vimentin and Snail mRNA expression [fig_ref] Fig 4: SRCIN1 inhibited epithelial-mesenchymal transition of the osteosarcoma cell [/fig_ref]. Moreover, we also showed that SRCIN1 overexpression promoted the expression of E-cadherin and inhibited the expression of N-cadherin, Vimentin and Snail [fig_ref] Fig 4: SRCIN1 inhibited epithelial-mesenchymal transition of the osteosarcoma cell [/fig_ref].
## Srcin1 repressed the osteosarcoma cell colony formation and invasion
Ectopic expression of SRCIN1 inhibited the MG-63 cell colony formation [fig_ref] Fig 5: SRCIN1 repressed the osteosarcoma cell colony formation and invasion [/fig_ref]. Moreover, overexpression of SRCIN1 suppressed the MG-63 cell invasion [fig_ref] Fig 5: SRCIN1 repressed the osteosarcoma cell colony formation and invasion [/fig_ref].
# Discussion
In this study, we demonstrated that SRCIN1 was downregulated in the osteosarcoma cell lines compared with osteoblastic cell line. Moreover, we showed that SRCIN1 expression was downregulated in osteosarcoma tissues compared with the adjacent tissues. Overexpression of SRCIN1 inhibited the osteosarcoma cell line MG63 proliferation. Furthermore, this effect was confirmed by measuring the ki-67 and PCNA expression. SRCIN1 overexpression promoted E-cadherin expression and suppressed N-cadherin, Vimentin and Snail expression. This result suggested that SRCIN1 overexpression inhibited EMT of the osteosarcoma cell. In Previous studies showed that SRCIN1 actedas a crucial role in Src inactivation and acted as a tumor suppressor gene in a lot of tumors [bib_ref] p140Cap suppresses the invasive properties of highly metastatic MTLn3-EGFR cells via impaired..., Damiano [/bib_ref] [bib_ref] p140Cap dual regulation of E-cadherin/EGFR cross-talk and Ras signalling in tumour cell..., Damiano [/bib_ref] [bib_ref] SNIP/p140Cap mRNA expression is an unfavourable prognostic factor in breast cancer and..., Kennedy [/bib_ref]. For example, Cao et al [bib_ref] miR-150 promotes the proliferation and migration of lung cancer cells by targeting..., Cao [/bib_ref].showed that miR-150 acted as an oncogene by inhibiting SRCIN1 translation in lung cancer. Sharma et al [bib_ref] Identification of two regions in the p140Cap adaptor protein that retain the..., Sharma [/bib_ref]. demonstrated that SRCIN1 suppressed tumor growth and impaired invasive properties of cancer cells through inhibiting the tyrosine kinase Src or E-cadherin/EGFR signaling pathways. Moreover, the expression of SRCIN1 was inversely associated with tumor malignancy in breast cancer. Damiano et al [bib_ref] p140Cap suppresses the invasive properties of highly metastatic MTLn3-EGFR cells via impaired..., Damiano [/bib_ref]. demonstrated that SRCIN1 suppressed the highly metastatic breast carcinoma cells invasion by inhibiting cortactin-dependent cell motility. However, the role of SRCIN1 in osteosarcoma is still uncovered. In this study, we showed that SRCIN1 was downregulated in the osteosarcoma cell lines compared with osteoblastic cell line. Furthermore, we measured the SRCIN1 expression in 35 osteosarcoma patients's tissues. We demonstrated that the expression of SRCIN1 was lower in the osteosarcoma tissues than in the adjacent nontumor tissues. Meanwhile, SRCIN1 was downregulated in 29 patients (29/35, 82%) compared with the adjacent tissues. We further investigated the role of SRCIN1 in osteosarcoma cell. Overexpression of SRCIN1 inhibited the osteosarcoma cell line MG63 proliferation. This effect was further confirmed by measuring the ki-67 and PCNA expression. SRCIN1 overexpression promoted the expression of E-cadherin while it suppressed N-cadherin, Vimentin and Snail expression. This result suggested that SRCIN1 overexpression inhibited EMT of the osteosarcoma cell. In addition, ectopic expression of SRCIN1 suppressed the MG-63 cell colony formation and invasion. These data suggested that SRCIN1 acted as a tumor suppressor gene in the development of osteosarcoma.
In conclusion, SRCIN1 may serve as a tumor suppressor gene in the development and metastasis of osteosarcoma. Given that ectopic expression of SRCIN1 suppresses cell proliferation and metastasis in the osteosarcoma, SRCIN1 may be a potent marker for the development of therapeutic strategies for patients with osteosarcoma.
# Author contributions
Conceived and designed the experiments: PW HW XL CZ DZ.
Performed the experiments: PW HW XL YL CZ DZ.
[fig] Fig 1: The expression of SRCIN1 was downregulated in the osteosarcoma cells. (A) One representive patient was diagnosed as osteosarcoma using HE staining. (B) The mRNA expression of SRCIN1 was measured by using qRT-PCR. (C) The protein expression of SRCIN1 was measured by using western blot. doi:10.1371/journal.pone.0155518.g001 [/fig]
[fig] Fig 2: SRCIN1 expression was reduced in the osteosarcoma tissues. (A) The mRNA expression of SRCIN1 was detected by using qRT-PCR in the osteosarcoma tissues and adjacent normal tissues. (B) SRCIN1 expression was downregulated in 29 patients (29/35, 82%) compared to the adjacent tissues. (C) The protein expression of SRCIN1 was measured by using western blot in the osteosarcoma tissues and adjacent normal tissues. doi:10.1371/journal.pone.0155518.g002 SRCIN1 Suppressed Osteosarcoma Cell Proliferation and Invasion PLOS ONE | DOI:10.1371/journal.pone.0155518 August 11, 2016addition, ectopic expression of SRCIN1 inhibited the MG-63 cell colony formation and invasion. These data suggested that SRCIN1 acts as a tumor suppressor gene in the development of osteosarcoma. [/fig]
[fig] Fig 3: SRCIN1 suppressed the osteosarcoma cell proliferation. (A) The protein expression of SRCIN1 was measured by using western blot in the MG-63 cells after treated SRCIN1 vector. (B) The mRNA expression of SRCIN1 was measured by using qRT-PCR in the MG-63 cells after treated SRCIN1 vector. (C) The cell proliferation was measured by CCK-8 analysis in the MG-63 cells. (D) The mRNA expression of ki-67 was measured by using qRT-PCR. (E) The protein expression of ki-67 was measured by using western blot in the MG-63 cells after treated SRCIN1 vector.(F) The mRNA expression of PCNA was measured by using qRT-PCR. (G) The protein expression of PCNA was measured by using western blot. *p<0.05, **p<0.01 and ***p<0.001. doi:10.1371/journal.pone.0155518.g003 SRCIN1 Suppressed Osteosarcoma Cell Proliferation and Invasion PLOS ONE | DOI:10.1371/journal.pone.0155518 August 11, 2016 [/fig]
[fig] Fig 4: SRCIN1 inhibited epithelial-mesenchymal transition of the osteosarcoma cell. (A) The mRNA expression of E-cadherin, N-cadherin, Vimentin and Snail was measured by using qRT-PCR in the MG-63 cells after treated SRCIN1 vector. (B) The protein expression of E-cadherin, N-cadherin, Vimentin and Snail was detected by using western blot in the MG-63 cells after treated SRCIN1 vector. doi:10.1371/journal.pone.0155518.g004 [/fig]
[fig] Fig 5: SRCIN1 repressed the osteosarcoma cell colony formation and invasion. (A) Ectopic expression of SRCIN1 inhibited the MG-63 cell colony formation. (B) Overexpression of SRCIN1 suppressed the MG-63 cell invasion. ***p<0.001. doi:10.1371/journal.pone.0155518.g005 SRCIN1 Suppressed Osteosarcoma Cell Proliferation and Invasion PLOS ONE | DOI:10.1371/journal.pone.0155518 August 11, 2016 [/fig]
|
10.3390/nano13101628
|
CCBY
|
10220528
|
37242043
|
s2orc_pubmed_articles
|
All-Atom Molecular Dynamics Simulations of the Temperature Response of Poly(glycidyl ether)s with Oligooxyethylene Side Chains Terminated with Alkyl Groups
# Introduction
Thermoresponsive polymers with a lower critical solution temperature (LCST) are attracting significant research interest. These polymers exhibit structural changes in response to external temperature change; thus, they are exploited for many applications, including drug delivery. Among the LCST-type thermoresponsive polymers, including poly(N-isopropylacrylamide) (PNIPAM) [bib_ref] Characterization of the LCST Behaviour of Aqueous Poly(N-Isopropylacrylamide) Solutions by Thermal and..., Boutris [/bib_ref] [bib_ref] N-Isopropylacrylamide): Experiment, Theory and Application, Schild [/bib_ref] , poly(ethylene glycol) (PEG) [bib_ref] Upper and Lower Critical Solution Temperatures in Poly, Saeki [/bib_ref] , poly[poly(ethylene glycol) methyl ether methacrylate] [bib_ref] Polymerization of Oligo(Ethylene Glycol) (Meth)Acrylates: Toward New Generations of Smart Biocompatible Materials, Lutz [/bib_ref] [bib_ref] Ethylene Glycol)-Methacrylate-Based Polymers and Microgels, Hu [/bib_ref] [bib_ref] LCST-Type Phase Separation of Poly[Poly(Ethylene Glycol) Methyl Ether Methacrylate]s in Hydrofluorocarbon, Koda [/bib_ref] , poly(glycidyl ether) (PGE) [bib_ref] Thermal Response of Poly(Ethoxyethyl Glycidyl Ether) Grafted on Gold Surfaces Probed on..., Inoue [/bib_ref] [bib_ref] Smart Hydrogels Based on Double Responsive Triblock Terpolymers, Reinicke [/bib_ref] [bib_ref] Controlled Polymerization of Glycidyl Methyl Ether Initiated by Onium Salt/Triisobutylaluminum and Investigation..., Labbé [/bib_ref] [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] , polyisocyanates [bib_ref] Synthesis of Water-Soluble Polyisocyanates with the Oligo(Ethylene Glycol) Side-Chain as New Thermoresponsive..., Sakai [/bib_ref] , poly(N,Ndiethylacrylamide) [bib_ref] Thermosensitivity of Aqueous Solutions of Poly(N,N-Diethylacrylamide), Idziak [/bib_ref] , poly(N-vinylcaprolactam) [bib_ref] Hydration and Phase Behavior of Poly(N-Vinylcaprolactam) and Poly(N-Vinylpyrrolidone) in Water, Maeda [/bib_ref] , and poly ethyl methacrylate] [bib_ref] Effect of Composition of PDMAEMA-b-PAA Block Copolymers on Their PHand Temperature-Responsive Behaviors, Han [/bib_ref] , PGE is known for its low cost, handling ease, and nontoxicity; thus, many PGE derivatives are commercially available. Additionally, PGE is widely employed in biomedicine for drug delivery [bib_ref] Allyl Glycidyl Ether)-Block-Poly(Ethylene Oxide): A Novel Promising Polymeric Intermediate for the Preparation..., Hrubý [/bib_ref] [bib_ref] Biocompatibility and Characterization of Polyglycerol-Based Thermoresponsive Nanogels Designed as Novel Drug-Delivery Systems..., Gerecke [/bib_ref] and as a scaffold material for tissue engineering [bib_ref] Glycidyl Ether) Brushes on Gold: Surface Engineering Parameters and Their Implication for..., Heinen [/bib_ref] [bib_ref] Fast and Solvent-Free Microwave-Assisted Synthesis of Thermoresponsive Oligo(Glycidyl Ether)s, Stöbener [/bib_ref]. Studies have shown that the LCST of PGE is controlled by its molecular weight [bib_ref] Controlled Polymerization of Glycidyl Methyl Ether Initiated by Onium Salt/Triisobutylaluminum and Investigation..., Labbé [/bib_ref] and molecular structure [bib_ref] Smart Hydrogels Based on Double Responsive Triblock Terpolymers, Reinicke [/bib_ref] [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] , as well as the balance between hydrophilic and hydrophobic side chains [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] [bib_ref] Novel Thermosensitive Polyethers Prepared by Anionic Ring-Opening Polymerization of Glycidyl Ether Derivatives, Aoki [/bib_ref].
For example, Watanabe et al. reported that poly(glycidyl methyl ether), poly(ethyl glycidyl ether), and poly(ethoxyethyl glycidyl ether) exhibited LCST-type phase transitions Nanomaterials 2023, 13, 1628 2 of 19 from 14.6 - C to 57.7 - C when the hydrophobicity of their side chains was reduced [bib_ref] Novel Thermosensitive Polyethers Prepared by Anionic Ring-Opening Polymerization of Glycidyl Ether Derivatives, Aoki [/bib_ref]. In addition, we previously synthesized PGEs with various oligooxyethylene side chains terminated with an alkyl group and established that the LCSTs of poly(glycidyl methyl ether), poly(ethyl glycidyl ether), poly(2-methoxyethyl glycidyl ether), poly(2-ethoxyethyl glycidyl ether), and poly(2-(2-ethoxyethyl)ethyl glycidyl ether) phase transitions increased from 10.3 - C to 91.6 - C with a decrease in the hydrophobicity of their oligooxyethylene side chains [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref]. However, despite the application prospects of PGE as a promising thermoresponsive material, the relationship between its temperature responsivity and polymer structure has not been adequately studied.
Molecular dynamics (MD) simulations have been employed to understand the response mechanisms of some thermoresponsive polymers, including PNIPAM [bib_ref] Role of Solvation Dynamics and Local Ordering of Water in Inducing Conformational..., Deshmukh [/bib_ref] [bib_ref] Molecular Dynamics Study of the LCST Transition in Aqueous Poly(N-n-Propylacrylamide), De Oliveira [/bib_ref] [bib_ref] Molecular Dynamics Study of Coil-to-Globule Transition in a Thermo-Responsive Oligomer Bound to..., Consiglio [/bib_ref] , PGE [bib_ref] Modes of Interaction in Binary Blends of Hydrophobic Polyethers and Imidazolium Bis(Trifluoromethylsulfonyl)Imide..., Bentley [/bib_ref] , PEG [bib_ref] On the Thermally Reversible Dynamic Hydration Behavior of Oligo(Ethylene Glycol) Methacrylate-Based Polymers..., Sun [/bib_ref] , and poly(N-vinylcaprolactam) (PVCL) [bib_ref] Atomistic Molecular Dynamics Simulations of the LCST Conformational Transition in Poly(N-Vinylcaprolactam) in..., Zhelavskyi [/bib_ref] [bib_ref] Atomistic Molecular Dynamics Simulations of the Lower Critical Solution Temperature Transition of..., Sun [/bib_ref] [bib_ref] Coil-to-Globule Transition of Thermo-Responsive γ-Substituted Poly (ε-Caprolactone) in Water: A Molecular Dynamics..., Koochaki [/bib_ref]. In addition, there is a research on the hydrophobic behavior of carbon nanoparticles using MD simulation in the study of their water solubility. Conventionally, PNIPAM undergoes a coil-globule transition by the recombination of the polymer-water hydrogen bonds with the intrapolymer hydrogen bonds [bib_ref] Role of Solvation Dynamics and Local Ordering of Water in Inducing Conformational..., Deshmukh [/bib_ref] [bib_ref] Molecular Dynamics Study of the LCST Transition in Aqueous Poly(N-n-Propylacrylamide), De Oliveira [/bib_ref] [bib_ref] Molecular Dynamics Study of Coil-to-Globule Transition in a Thermo-Responsive Oligomer Bound to..., Consiglio [/bib_ref]. Conversely, the mechanism of the coil-globule transition in PGE has not yet been clarified. From a thermodynamic standpoint, hydrogen bonds between polymer chains and water molecules contribute favorable enthalpies to the free energy of mixing, whereas bonds between water molecules and polymer chains, particularly hydrophobic hydration shells formed around hydrophobic groups, promote the ordering of water molecules. Therefore, they contribute negatively to the mixing entropy. At relatively high temperatures, the entropy term, TS, becomes predominant, and the free energy of mixing becomes positive, which accounts for the insolubility of the polymer. The coil-globule structural transition in PNIPAM has been studied using MD simulations [bib_ref] Role of Solvation Dynamics and Local Ordering of Water in Inducing Conformational..., Deshmukh [/bib_ref] [bib_ref] Molecular Dynamics Study of the LCST Transition in Aqueous Poly(N-n-Propylacrylamide), De Oliveira [/bib_ref] [bib_ref] Molecular Dynamics Study of Coil-to-Globule Transition in a Thermo-Responsive Oligomer Bound to..., Consiglio [/bib_ref]. Additionally, and Graziano reported that the solvent exclusion volume effect can be an indicator of the LCST phase transition [bib_ref] On the Effect of Sodium Salts on the Coil-to-Globule Transition of Poly(N-Isopropylacrylamide), Pica [/bib_ref] [bib_ref] Effect of Sodium Thiocyanate and Sodium Perchlorate on Poly(N-Isopropylacrylamide) Collapse, Pica [/bib_ref]. PGE, which is actively studied in biomedicine, requires strict temperature control. If the LCST temperature can be predicted by MD simulations, the polymer structure can be designed according to the target temperature.
In this study, we performed MD simulations on six different types of side-chain structures investigated by Isono and Satoh et al. [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] and examined the effect of the number of oxyethylene units on the LCST. We found that the hydrophobic shell intensity of the sidechain terminal alkyl groups well-correlated with LCST.
# Computational method
## Initial structure of the polymers
Models of poly(glycidyl methyl ether) (poly(MeGE)), poly(ethyl glycidyl ether) (poly(EtGE)), poly(2-methoxyethyl glycidyl ether) (poly(MeEOGE)), poly(2-ethoxyethyl glycidyl ether) (poly(EtEOGE)), poly(2-(2-methoxyethoxy)ethyl glycidyl ether) (poly(MeEO 2 GE)), and poly(2-(2-ethoxyethyl)ethyl glycidyl ether) (poly(EtEO 2 GE)) were prepared by the random addition of monomers with d and l structures, mimicking the tacticity of the whole chain [fig_ref] Figure 1: Chemical structures of [/fig_ref]. These polymer structures were optimized by the molecular mechanics 2 (MM2) method included in the chem3D 19.0 package of PerkinElmer (Waltham, MA, USA). All the polymer molecular weights (MW) were set at ca. 2500, and 5000.
## Atomic charges and force fields
The atomic partial charges for the six polymer species were obtaine AM1-BCC protocol [bib_ref] Efficient Generation of High-Quality Atomic Charges. AM1-BCC Model: I. Method, Jakalian [/bib_ref] [bib_ref] Efficient Generation of High-Quality Atomic Charges. AM1-BCC Model: II. Parameterization and Validation, Jakalian [/bib_ref] implemented in the antechamber program [bib_ref] Development and Testing of a General Amber Force Field, Wang [/bib_ref] All other force-field parameters for the polymers were acquired using the g force field (GAFF) version 1.7 [bib_ref] Development and Testing of a General Amber Force Field, Wang [/bib_ref]. The GAFF was chosen because it ha strated to be accurate for the simulation of most organic molecules. The det parameters are provided in gaff.dat [bib_ref] Development and Testing of a General Amber Force Field, Wang [/bib_ref].
[formula] = , , , , /2 1 cos ∅ , ′ ′ 4 , [/formula]
## Solvent water addition into a simulation box
Each polymer was centered in a cubic simulation box under periodic ditions. Over 2000 TIP3P-modeled water molecules were added with a m along each dimension in [fig_ref] Table 1: Calculated principal R g peaks and experimental T CLP [/fig_ref] of Supplementary Materials.
## Energy minimization and temperature equilibration
The whole system consisted of a polymer, and the water molecules we imized and equilibrated with the sander program implemented in the A before the MD simulations. The energy minimization was performed in t the solvent molecules were energy-minimized using the steepest decent m steps and the conjugate gradient method for 1500 steps while keeping the so
## Atomic charges and force fields
The atomic partial charges for the six polymer species were obtained through the AM1-BCC protocol [bib_ref] Efficient Generation of High-Quality Atomic Charges. AM1-BCC Model: I. Method, Jakalian [/bib_ref] [bib_ref] Efficient Generation of High-Quality Atomic Charges. AM1-BCC Model: II. Parameterization and Validation, Jakalian [/bib_ref] implemented in the antechamber program [bib_ref] Development and Testing of a General Amber Force Field, Wang [/bib_ref] of AMBER 14. All other force-field parameters for the polymers were acquired using the general AMBER force field (GAFF) version 1.7 [bib_ref] Development and Testing of a General Amber Force Field, Wang [/bib_ref]. The GAFF was chosen because it has been demonstrated to be accurate for the simulation of most organic molecules. The detailed force field parameters are provided in gaff.dat [bib_ref] Development and Testing of a General Amber Force Field, Wang [/bib_ref].
[formula] V AMBER = n bonds ∑ i b i (r i − r i,eq ) 2 + n angles ∑ i a i (θ i − θ i,eq ) 2 + n dihedrals ∑ i n i,max ∑ n (V i, n /2)[1 + cos(n∅ i − γ i, n )] + n atoms ∑ i<j A ij r 12 ij − B ij r 6 ij + n atoms ∑ i<j q i q j 4πε 0 r ij ,(1) [/formula]
## Solvent water addition into a simulation box
Each polymer was centered in a cubic simulation box under periodic boundary conditions. Over 2000 TIP3P-modeled water molecules were added with a margin of 10 Å along each dimension in [fig_ref] Table 1: Calculated principal R g peaks and experimental T CLP [/fig_ref] of Supplementary Materials.
## Energy minimization and temperature equilibration
The whole system consisted of a polymer, and the water molecules were energyminimized and equilibrated with the sander program implemented in the AMBER system before the MD simulations. The energy minimization was performed in two steps. First, the solvent molecules were energy-minimized using the steepest decent method for 1000 steps and the conjugate gradient method for 1500 steps while keeping the solute molecular chains restrained with a force of 500 kcal mol −1 Å −2 . Second, the restraint was eliminated, and the entire system, including the polymer chain, was energy-minimized using the steepest decent method for 1000 steps and the conjugate gradient method for 1500 steps as in the first step.
Subsequently, the energy equilibration was performed in two steps. First, the system was gradually heated to an assigned temperature, 278, 300, 323, 343, and 368 K, under constant volume boundary conditions for 10,000 steps, with a time step of 2 fs while maintaining the 10 kcal mol −1 Å −2 constraint on the solute polymer chains. In the second step, the restraint was removed, and the entire system was equilibrated under periodic boundary conditions at a constant pressure of 1 atm for 50,000 steps, with a time step of 2 fs while maintaining the specified temperature. The Langevin method was employed for temperature control, and isotropic scaling was employed for pressure control. The SHAKE algorithm was applied to maintain constant hydrogen atom bond lengths. A cutoff length of 9.0 Å and particle mesh Ewald (PME) sums were applied for the electrostatic interactions.
## Md simulation
After confirming the equilibrium of the whole system, MD simulations were performed for 20 ns or 100 ns, with a time step of 2 fs using a pmemd program of AMBER 14. All the MD simulations were conducted in NPT (number of molecules (N), pressure (P), and temperature (T) are conserved) ensembles to capture the dynamics and structural evolution of both the polymer and water molecules at five different temperatures (278, 300, 323, 343, and 368 K). The simulations were carried out at atmospheric pressure (1 atm). The trajectories were outputted at 2.0 ps intervals.
# Analysis method
## Radius of gyration
The structural evolution of each polymer was studied along the time course of the stored atomic trajectories by analyzing the radius of gyration (R g ). R g , which is a measure of the extension degree of a polymer chain, was calculated using the equation below:
[formula] R 2 g = 1 N ∑ n (R n − R c ) 2 ,(2) [/formula]
where R c is the center of mass of the polymer chain, and N is the number of atoms included in the polymer chain. The sum for all the atoms in the chain was obtained.
## Radial distribution function
The radial distribution function (RDF), g(r), represents the number density of the particles existing in the spherical shells between distances r and r + dr from a given reference particle. It can be expressed as:
[formula] g(r) = n(r) 4πr 2 drρ(3) [/formula]
where n(r) indicates the number of particles of the spherical shell at distances of r and r + dr from the given reference particle. ρ is the average number density of the particles in the bulk condition. The g(r) function is zero when the interparticle distance, r, approaches zero because repulsive force prevents the atoms from closely approaching each other. Additionally, when distance r is infinite, the local density in the shell is equivalent to the mean bulk density averaged over the whole volume; therefore, g(r) is standardized to 1. The RDF was calculated to determine the difference in the structural arrangements of the polymer atoms and water molecules at 278, 300, 323, 343, and 368 K. The RDFs around the carbon atom of the side chain and for the water-oxygen atom were carefully investigated for six different kinds of polymers.
# Results and discussion
## Temperature dependence of r g
In PNIPAM, the LCST-type phase transition and coil-globule structural transitions are known to be correlated. The occurrence of the coil-globule structural transition can be determined using the R g , which represents the degree of chain extension. Therefore, the temperature dependence of R g was analyzed to observe the structural transition behaviors of poly(MeGE), poly(EtGE), poly(MeEOGE), poly(EtEOGE), poly(MeEO 2 GE), and poly(EtEO 2 GE). Five polymers other than poly(MeEO 2 GE) have been reported to exhibit phase transitions [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref]. In this study, MD simulations were performed with molecular weights of 2500 and 5000 for all polymers. The degrees of polymerization with molecular weights of 2500 for poly(MeGE), poly(EtGE), poly(MeEOGE), poly(EtEOGE), poly(MeEO 2 GE), and poly(EtEO 2 GE) were [bib_ref] Coil-to-Globule Transition of Thermo-Responsive γ-Substituted Poly (ε-Caprolactone) in Water: A Molecular Dynamics..., Koochaki [/bib_ref] [bib_ref] Modes of Interaction in Binary Blends of Hydrophobic Polyethers and Imidazolium Bis(Trifluoromethylsulfonyl)Imide..., Bentley [/bib_ref] [bib_ref] Fast and Solvent-Free Microwave-Assisted Synthesis of Thermoresponsive Oligo(Glycidyl Ether)s, Stöbener [/bib_ref] [bib_ref] Biocompatibility and Characterization of Polyglycerol-Based Thermoresponsive Nanogels Designed as Novel Drug-Delivery Systems..., Gerecke [/bib_ref] [bib_ref] Hydration and Phase Behavior of Poly(N-Vinylcaprolactam) and Poly(N-Vinylpyrrolidone) in Water, Maeda [/bib_ref] , and 15, respectively. Additionally, the degrees of polymerization with molecular weights of 5000 for poly(MeGE), poly(EtGE), poly(MeEOGE), poly(EtEOGE), poly(MeEO 2 GE), and poly(EtEO 2 GE) were 55, 47, 37, 33, 27, and 29, respectively. The averaged trajectories obtained from two MD simulations were employed for the analysis. [fig_ref] Figure 2: Rg distribution for poly [/fig_ref] and Figures S1 and S2 in Supplementary Materials show the R g distributions for six different kinds of polymers having molecular weights 2500 obtained with their time courses for 20 ns. The R g distributions of poly(EtGE), poly(EtEOGE), and poly(EtEO 2 GE) with ethyl terminal groups were relatively unimodal across the temperature range, whereas those of poly(MeGE) and poly(MeEOGE) with methyl terminal groups were multimodal or broadened in the low-to-medium temperature range. [fig_ref] Table 1: Calculated principal R g peaks and experimental T CLP [/fig_ref] summarizes the peak values of the R g distribution at each temperature and the corresponding experimental cloud point (CLP) values. The experimental CLP temperatures (T CLP ) are referenced from the results obtained using the molecular weight of ca. 5000 [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref]. [fig_ref] Figure 3: R g peaks plotted versus temperatures for poly [/fig_ref] shows the R g peak values versus the temperatures in [fig_ref] Table 1: Calculated principal R g peaks and experimental T CLP [/fig_ref]. No significant change was observed in the main peak values of R g above and below T CLP for any of the polymers. This suggests that the coil-globule structural transition was not a prerequisite for the LCST-type phase transition in the PGEs with oxyethylene units. For the standard deviation (SD), the PGE with side-chain ethyl groups tended to have a smaller SD value than those with side-chain methyl group. This implies that the PGE with ethyl groups at the side-chain ends are bulkier than those with methyl groups, and thus, the movement of the main chain structure should be suppressed. Notably, the difference in the bulkiness of the terminal structure affected the distribution of R g . [fig_ref] Figure 3: R g peaks plotted versus temperatures for poly [/fig_ref] , Supplementary Materials shows snapshots of each polymer corresponding to the main peak values of R g at temperatures below and above T CLP . No clear coil-globule structural transitions were observed above and below T CLP for any of the polymers, as evidenced by the analysis of R g . It was difficult to observe structural transitions from the R g distributions and snapshots for six different kinds of polymers having molecular weights of 5000 obtained with their time courses for 100 ns in [fig_ref] Figure 5: R g distribution for poly [/fig_ref] and [fig_ref] Figure 4: Rg peaks plotted versus temperatures for poly [/fig_ref] , Supplementary Materials. The R g distributions of polymers with the molecular weight 5000 tended to have larger width than those with molecular weight 2500, which should be attributed to the greater fluctuations caused by the doubling of the molecular weight.
## Degree of chain extensions and their fluctuations
To investigate whether the polymer is in the coil or globule state, the ratio of the maximum length of the main chain to the time average length obtained from
## Degree of chain extensions and their fluctuations
To investigate whether the polymer is in the coil or globule state, the ratio of the maximum length of the main chain (L max ) to the time average length L obtained from the MD simulation performed with a molecular weight of 2500 was calculated. The time dependence and distribution of L for the six polymers are shown in [fig_ref] Figure 1: Chemical structures of [/fig_ref] , Supplementary Materials. For freely jointed chains or ideal chains with the degree of polymerization, N, the following equation holds:
[formula] L L max = 1 √ N .(4) [/formula]
If the ratio is larger than 1/ √ N, the polymer can be roughly considered to be in the coil state, and if it is smaller, it can be considered to be in the globule state. [fig_ref] Table 2: Calculated L/L max at each temperature [/fig_ref] summarizes the values of L/L max and 1/ √ N for all the polymers. For poly(MeGE) (28-mers) and poly(EtGE) (24-mers), which had relatively large degrees of polymerization, the values of L/L max were smaller than the corresponding 1/ √ N values in almost all temperature ranges, suggesting that they consistently adopted globule structures. Poly(MeEOGE) (19-mers) exhibited larger values of L/L max than 1/ √ N in all temperature ranges; thus, it could be considered to consistently be in the coil state. Conversely, poly(EtEOGE) (17-mers) exhibited smaller values of L/L max than 1/ √ N at 300, 323, and 343 K; thus, it was expected to be in the globule state. Poly(MeEO 2 GE) (14-mers) and poly(EtEO 2 GE) (15-mers) with relatively small degrees of polymerization exhibited larger values of L/L max than 1/ √ N; thus, they were expected to be in the coil state in all temperature ranges. The coefficient of variation (CV) was calculated to compare the fluctuation of L/L max for the six polymers without the effect of different degrees of polymerization. The CV is the SD normalized by the mean value. It is a dimensionless number used to evaluate the relative relationship of the SDs among the data with different mean values. When the CV values between poly(MeGE) and poly(EtGE), poly(MeEOGE) and poly(EtEOGE), and poly(MeEO 2 GE) and poly(EtEO 2 GE) were compared, the polymer with the methyl group at the side-chain ends tended to exhibit higher values than those for the polymer with the ethyl group at the sidechain ends. According to the R g measurement results, the magnitude of fluctuation was different between the methyl and ethyl side-chain ends; however, it was unclear whether the difference was due to the main chain or the side chain. This suggests that the polymer main chain structure with the methyl group fluctuated more than that with the ethyl group. shows the ratio of the maximum length of the main chain (L max ) to the temperature average length ( L temp ), SD, and CV for the six polymers at five temperatures. Resultantly, except for poly(MeEO 2 GE), the CV values of the polymers with terminal methyl groups were larger than those of the polymers with terminal ethyl groups. This indicates that the main-chain structural fluctuation of the polymer with a terminal methyl group on the side chain should be larger than that of the polymer with a terminal ethyl group on the side chain. It has been reported that the T CLP of poly(MeEOGE) decreases as the molecular weight increases [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref]. The L/L max values of poly(MeEOGE) for molecular weights of 1250 and 5000 were calculated, and the polymer-chain extension was investigated. [fig_ref] Table 4: Calculated L/L max of poly [/fig_ref] summarizes the results of the L/L max and 1/ √ N calculations for poly(MeEOGE) with molecular weights of 1250, 2500, and 5000. As the molecular weight increased, the value of L/L max tended to decrease. For the molecular weight of 1250, the values of L/L max were larger than the corresponding 1/ √ N values in all temperature regions except for 278 K. This indicated that poly(MeEOGE) with a molecular weight of 1250 exhibited a coil structure. For the molecular weight of 5000, the values of L/L max were smaller than the 1/ √ N values in all temperature regions, indicating that the poly(MeEOGE) with a molecular weight of 5000 had a globule structure. A similar tendency was observed for poly(EtEOGE) [fig_ref] Table 2: Calculated L/L max at each temperature [/fig_ref] , Supplementary Materials). These results suggested that the smaller the degree of polymerization of PGE, the lower the probability that it exhibits the coil structure, and the larger the degree of polymerization, the higher the probability that it exhibits the globule structure. The larger the degree of polymerization of poly(MeEOGE), the higher its tendency to adopt the globule structure. This implies that the change from the coil to the globule state with increasing molecular weight was related to the change in the T CLP . This relation will be explained by RDF analysis.
## The side-chain lengths and their fluctuations
The results of the L/L max calculations suggested that the main chain structure of PGE with methyl terminal groups fluctuated more than that of the PGE with ethyl terminal groups at the side-chain ends. To further investigate if the fluctuation of the side-chain structure affects the polymer fluctuation, the side-chain lengths (i.e., the lengths between the main-chain carbon atom and the side-chain terminal carbon atom) of six polymers with molecular weights of 2500 were obtained using the time-averaged values of the sidechain length in the MD simulations. [fig_ref] Table 5: Time-averaged side chain lengths and their fluctuations [/fig_ref] summarizes the time-averaged side-chain lengths, and their temperature-averaged values, SDs, and CVs, as well as the calculated side-chain lengths at each temperature, are shown in , Supplementary Materials. When comparing the CV values, when the side chain was short, as in poly(MeGE) and poly(EtGE), no difference in the magnitude of the fluctuations was observed. However, when the side chain was long, as in poly(MeEOGE), poly(EtEOGE), poly(MeEO 2 GE), and poly(EtEO 2 GE), the PGE with the methyl group exhibited a higher magnitude of fluctuation in the side-chain structure than those with ethyl group. This suggested that the fluctuation of the side-chain structure affected the fluctuation of the polymer structure.
## Rdf of hydrophobic solvation
Since the structural transition corresponding to the LCST-type phase transition was not observed by R g analysis, an alternative physical quantity that reflects CLP was investigated. The RDF is utilized to explore the local order of water molecules surrounding polymers. Therefore, the RDF for the oxygen atoms of waters with polymers having carbon atoms at the side-chain ends was investigated. All the results are based on the average values obtained from two MD simulations conducted with a molecular weight 2500. [fig_ref] Figure 6: RDFs of [/fig_ref] shows the water-oxygen RDF measured using the carbon atoms at all the sidechain ends of poly(MeGE), poly(EtGE), poly(MeEOGE), poly(EtEOGE), poly(MeEO 2 GE), and poly(EtEO 2 GE) at 278, 300, 323, 343, and 368 K, respectively. The first peaks in the water-oxygen RDF spectra were observed at around 3.6 Å for all polymers; thus, the first hydrophobic hydration shells are formed at 3.6 Å from carbon atoms at the side-chain ends. Thereafter, the temperature dependence of the peak intensity in the first hydrophobic hydration shell was investigated. As shown in , the peak intensity of the hydrophobic hydration shell decreased during the heating process. Furthermore, the crossing point temperature (T CRP ) at which the peak intensity decreased below g (r) = 1.0 was analyzed for the six types of polymers. The T CRP values are summarized together with the T CLP values reported in the experiments [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] in [fig_ref] Table 6: Calculated T CRP vs [/fig_ref]. The T CRP values for poly(MeGE), poly(EtGE), poly(MeEOGE), poly(EtEOGE), and poly(EtEO 2 GE) appeared at 326, 291, 349, 334, and 348 K, respectively, and they were close to the experimental T CLP values. These results indicated that the hydrophobic hydration shell strength at the end of the polymer side chain correlated with T CLP . By examining the side-chain end structures, we established that the T CRP obtained by simulation for poly(MeGE) and poly(MeEOGE) with methyl-side-chain ends were lower than the experimentally obtained T CLP values. Conversely, the T CRP obtained by simulation for poly(EtGE), poly(EtEOGE), and poly(EtEO 2 GE) with ethyl-sidechain ends were higher than the experimentally obtained T CLP values. In this study, MD simulations were performed with a single polymer chain, and the T CRP should be higher than the experimental T CLP because the polymer concentration in the MD simulation is lower than that in the actual experiment. In the experiments, T CLP tends to increase as the concentration of the polymer solution decreases. This is because the collision frequency between the polymers is reduced, and aggregation is less likely to occur. Contrarily, for the PGE with methyl groups at the side-chain ends, T CRP was lower than T CLP . This relationship is plotted in [fig_ref] Figure 8: Plots of T CRP and T CLP vs [/fig_ref]. The PGE with methyl terminal groups exhibited large fluctuations in the main-chain and side-chain structures, as evidenced by the results of R g , L/L max , and time-averaged side-chain length measurements. Therefore, it is expected that water molecules enter the gaps between the adjacent side chains of the PGE with hydrophobic methyl groups, while the proportion of the side-chain end occupying the polymer surface decreases, and the proportion of the hydrophilic ether part occupying the polymer surface increases [fig_ref] Figure 9: Schematic illustration of [/fig_ref]. T CRP only reflects the aqueous environment at the end of the side chain and not the entire polymer surface. This suggests that the T CRP of the PGE with methyl groups at the side-chain ends was lower than the T CLP . terminal groups exhibited large fluctuations in the main-chain and side-chain structures, as evidenced by the results of Rg, / , and time-averaged side-chain length measurements. Therefore, it is expected that water molecules enter the gaps between the adjacent side chains of the PGE with hydrophobic methyl groups, while the proportion of the sidechain end occupying the polymer surface decreases, and the proportion of the hydrophilic ether part occupying the polymer surface increases [fig_ref] Figure 9: Schematic illustration of [/fig_ref]. only reflects the aqueous environment at the end of the side chain and not the entire polymer surface. This suggests that the of the PGE with methyl groups at the side-chain ends was lower than the . and , respectively. Red arrows mean the is lower than the , and blue arrows mean the is higher than the . Open and filled circles correspond to T CLP and T CRP , respectively. Red arrows mean the T CRP is lower than the T CLP , and blue arrows mean the T CRP is higher than the T CLP . and , respectively. Red arrows mean the is lower than the , and blue arrows mean the is higher than the . In addition, to investigate if the molecular weight affects the RDF, the T CRP values for poly(MeEOGE) and poly(EtEOGE) with molecular weights of 1500, 2500, and 5000 were analyzed. [fig_ref] Figure 1: Chemical structures of [/fig_ref] shows that as the molecular weight increased, T CRP decreased. It has been reported that the phase-transition temperature of poly(MeEOGE) tends to decrease with increasing molecular weight [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] , and a similar result was observed in our T CRP analysis in the MD calculations. This suggests that as the molecular weight increased, the hydrophobicity of the polymer surface increased because of the hydrophobic side chain, resulting in a decrease in T CLP . In addition, to investigate if the molecular weight affects the RDF, the values for poly(MeEOGE) and poly(EtEOGE) with molecular weights of 1500, 2500, and 5000 were analyzed. [fig_ref] Figure 1: Chemical structures of [/fig_ref] shows that as the molecular weight increased, decreased. It has been reported that the phase-transition temperature of poly(MeEOGE) tends to decrease with increasing molecular weight [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] , and a similar result was observed in our analysis in the MD calculations. This suggests that as the molecular weight increased, the hydrophobicity of the polymer surface increased because of the hydrophobic side chain, resulting in a decrease in .
## Numbers of hydrogen bonds between the polymer and water molecules
In the previous section, the hydrophobic hydration shells were analyzed, after which the hydrogen bonds between the polymer oxygen atoms and water molecules were investigated to determine the contribution of the hydrophilic ether oxygen to the LCTS-type phase transition. The average number of hydrogen bonds between the oxygen atoms in the main chains or the side chains and the hydrogen atoms in water are listed in [fig_ref] Table 7: Average numbers of hydrogen bonds between polymer and water calculated over 20... [/fig_ref]. [fig_ref] Figure 1: Chemical structures of [/fig_ref] show the average numbers of hydrogen bonds between the polymer and water molecules in the main-chain oxygen atoms or the side-chain oxygen atoms per oxygen in the polymer residue, 〈 〉), or 〈 〉 , versus temperature.
The average numbers of hydrogen bonds between the polymer and water molecules in the main-chain oxygen atoms tend to decrease as the temperature increases for all polymers, and similar results were obtained for the side-chain oxygen atoms. In addition, the average number of hydrogen bonds between the water and oxygen atoms in the side chain tended to be higher than that of the bonds between the water and oxygen atoms in the main chain. Furthermore, a relatively large decrease in the number of hydrogen bonds around or was observed. Although not shown in [fig_ref] Table 7: Average numbers of hydrogen bonds between polymer and water calculated over 20... [/fig_ref] , the average number of hydrogen bonds between the intrapolymer was almost zero at any temperature for all polymers. The average numbers of hydrogen bonds within an intrapolymer are shown in [fig_ref] Table 4: Calculated L/L max of poly [/fig_ref] , Supplementary Materials. This suggests that the recombination of intrapolymer hydrogen bonds and polymer-water hydrogen bonds for the OEO ether group of the PGE side chain had not occurred. N-substituted acrylamide-based polymers, such as PNIPAM, have amide groups with a proton donor and acceptor, and they promote the recombination of hydrogen bonds between an intrapolymer and between polymer and water molecules [bib_ref] Change in Hydration State during the Coil-Globule Transition of Aqueous Solutions of..., Maeda [/bib_ref] [bib_ref] On the Molecular Origin of the Cooperative Coil-to-Globule Transition of Poly(N-Isopropylacrylamide) in..., Tavagnacco [/bib_ref]. However, since PGE has ether groups with only a proton acceptor, the
## Numbers of hydrogen bonds between the polymer and water molecules
In the previous section, the hydrophobic hydration shells were analyzed, after which the hydrogen bonds between the polymer oxygen atoms and water molecules were investigated to determine the contribution of the hydrophilic ether oxygen to the LCTS-type phase transition. The average number of hydrogen bonds between the oxygen atoms in the main chains or the side chains and the hydrogen atoms in water are listed in [fig_ref] Table 7: Average numbers of hydrogen bonds between polymer and water calculated over 20... [/fig_ref]. [fig_ref] Figure 1: Chemical structures of [/fig_ref] show the average numbers of hydrogen bonds between the polymer and water molecules in the main-chain oxygen atoms or the side-chain oxygen atoms per oxygen in the polymer residue, ( N main chain HB ), or ( N side chain HB ), versus temperature. The average numbers of hydrogen bonds between the polymer and water molecules in the main-chain oxygen atoms tend to decrease as the temperature increases for all polymers, and similar results were obtained for the side-chain oxygen atoms. In addition, the average number of hydrogen bonds between the water and oxygen atoms in the side chain tended to be higher than that of the bonds between the water and oxygen atoms in the main chain. hydrogen bonds between the intrapolymer should be weak; therefore, no hydrogen bond recombination was observed. In addition, it has been reported that hydrogen bonding does not affect the LCST-type transition in poly(oligo(ethylene glycol)methyl ether methacrylate) with an ether group [bib_ref] The Role of Hydrophobic Hydration in the LCST Behaviour of POEGMA300 by..., Dalgakiran [/bib_ref]. This suggests that the weak hydrogen bonding in the PGE with ether groups led to the absence of a clear coil-globule structural transition. Furthermore, a relatively large decrease in the number of hydrogen bonds around T CLP or T CRP was observed. Although not shown in [fig_ref] Table 7: Average numbers of hydrogen bonds between polymer and water calculated over 20... [/fig_ref] , the average number of hydrogen bonds between the intrapolymer was almost zero at any temperature for all polymers. The average numbers of hydrogen bonds within an intrapolymer are shown in [fig_ref] Table 4: Calculated L/L max of poly [/fig_ref] , Supplementary Materials. This suggests that the recombination of intrapolymer hydrogen bonds and polymer-water hydrogen bonds for the OEO ether group of the PGE side chain had not occurred. N-substituted acrylamide-based polymers, such as PNIPAM, have amide groups with a proton donor and acceptor, and they promote the recombination of hydrogen bonds between an intrapolymer and between polymer and water molecules [bib_ref] Change in Hydration State during the Coil-Globule Transition of Aqueous Solutions of..., Maeda [/bib_ref] [bib_ref] On the Molecular Origin of the Cooperative Coil-to-Globule Transition of Poly(N-Isopropylacrylamide) in..., Tavagnacco [/bib_ref]. However, since PGE has ether groups with only a proton acceptor, the hydrogen bonds between the intrapolymer should be weak; therefore, no hydrogen bond recombination was observed. In addition, it has been reported that hydrogen bonding does not affect the LCST-type transition in poly(oligo(ethylene glycol)methyl ether methacrylate) with an ether group [bib_ref] The Role of Hydrophobic Hydration in the LCST Behaviour of POEGMA300 by..., Dalgakiran [/bib_ref]. This suggests that the weak hydrogen bonding in the PGE with ether groups led to the absence of a clear coil-globule structural transition.
# Conclusions
In this study, we performed MD simulations of PGE with six different oligooxyethylene side chains terminated with alkyl groups in aqueous solutions to investigate the effects of side-chain structures on LCST-type phase transition. To investigate if the coil-globule structural transition occurs, we analyzed the R g , which indicates the degree of polymer chain extension, and the L/L max , which indicates the degree of polymer shrinkage. Comparing the SD of R g and the CV values of L/L max and the side-chain lengths for the six polymers, the PGE with a methyl group at the end of the side chain revealed larger values than the PGE with an ethyl group, suggesting that the PGE with terminal methyl groups on the side chain has larger fluctuations both in the main and side chains than the PGE with terminal ethyl groups on the side-chain. Although we did not observe clear coil-globule structural transition in any PGEs, we found that, as the degree of polymerization increased, the globule structure became easier to obtain. To investigate if the coil-globule transition occurs in more detail, we will analyze the eigenvalues of the radius of the gyration tensor in our future study.
Next, we focused on the RDF as another physical quantity that indicates the distribution of water molecules around the polymer and calculated the RDF for the oxygen atoms of waters measured from carbon atoms at the side-chain ends of the polymers. The result showed that hydrophobic hydration shells were formed around 3.6 Å from the carbon atoms at the end of the polymer side chains in all six polymers. We investigated the temperature dependence of the peak intensity of the hydrophobic hydration shells and found that the crossing point temperatures (T CRP ) at which the peak intensity was below g (r) = 1.0 were close to the CLP temperatures (T CLP ) obtained in the experiments [bib_ref] Design and Synthesis of Thermoresponsive Aliphatic Polyethers with a Tunable Phase Transition..., Isono [/bib_ref] for all polymers. The phase-transition temperature of the PGE was correlated with the intensity of the primary hydrophobic hydration shell of the side-chain terminal alkyl carbon. Furthermore, the T CRP of the PGE with terminal methyl groups on the side chain was lower than the experimental T CLP , while the T CRP for the PGE with terminal ethyl groups on the side chain was higher than the experimental T CLP . In this work, we performed MD simulations with a single polymer chain; therefore, the T CRP should be higher than the experimental T CLP because the polymer concentration is lower than that in the actual experiment. We consider that the magnitude of the polymer fluctuation affects the relationship between the T CRP and the T CLP . In other words, since the PGE with terminal methyl groups on the side chain tends to have larger polymer fluctuations than the PGE with terminal ethyl groups on the side chain, as evidenced by the results of R g , L/L max , as well as time-averaged side-chain length measurements, we assume that water molecules entered the gaps between the adjacent side chains of PGE with hydrophobic methyl groups. Consequently, the proportion of the hydrophilic ether part occupying the polymer surface increased. Thus, the T CRP of PGE with methyl groups at the side-chain ends was lower than the T CLP value. In the future, we will expand our research to encompass multiple polymer chains to investigate interpolymer interactions.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/nano13101628/s1, [fig_ref] Table 1: Calculated principal R g peaks and experimental T CLP [/fig_ref] : Total number of water molecules and box length; [fig_ref] Table 2: Calculated L/L max at each temperature [/fig_ref] : Calculated L/L max of poly(EtEOGE) for molecular weights of 1250, 2500, and 5000; : Calculated side chain lengths at each temperature; [fig_ref] Table 4: Calculated L/L max of poly [/fig_ref] : Average numbers of hydrogen bonds within an intrapolymer calculated over 20 ns at 278 K, 300 K, 323 K, 343 K, and 368 K; [fig_ref] Figure 1: Chemical structures of [/fig_ref] : R g distribution inserted for poly(MeEOGE) 2.5k (left) and poly(EtEOGE) 2.5k (right). The inset is the corresponding time dependence of R g .; [fig_ref] Figure 2: Rg distribution for poly [/fig_ref] : R g distributions for poly(MeEO 2 GE) 2.5k (left) and poly(EtEO 2 GE) 2.5k (right). The inset is the corresponding time dependence of R g ; [fig_ref] Figure 3: R g peaks plotted versus temperatures for poly [/fig_ref] : Snapshots of (a) poly(MeGE) 2.5k , (b) poly(EtGE) 2.5k , (c) poly(MeEOGE) 2.5k , (d) poly(EtEOGE) 2.5k , and (e) poly(EtEO 2 GE) 2.5k at the end of 20 ns below and above T CLP . The carbon, oxygen, and hydrogen atoms of polymers are shown in green, red, and grey colors, respectively. Solvent water molecules have been omitted for clarity; [fig_ref] Figure 4: Rg peaks plotted versus temperatures for poly [/fig_ref] : R g distributions for poly(MeEOGE) 5k (left) and poly(EtEOGE) 5k (right). The inset is the corresponding time dependence of R g ; [fig_ref] Figure 5: R g distribution for poly [/fig_ref] : R g distributions for poly(MeEO 2 GE) 5k (left) and poly(EtEO 2 GE) 5k (right). The inset is the corresponding time dependence of R g ; [fig_ref] Figure 6: RDFs of [/fig_ref] : Snapshots of (a) poly(MeGE) 5k , (b) poly(EtGE) 5k , (c) poly(MeEOGE) 5k , (d) poly(EtEOGE) 5k , and (e) poly(EtEO 2 GE) 5k at the end of 20 ns below and above T CLP . The carbon, oxygen, and hydrogen atoms of polymers are shown in green, red, and grey colors, respectively. Solvent water molecules have been omitted for clarity; : Time dependence of L for poly(MeGE) 2.5k (left) and poly(EtGE) 2.5k (right); [fig_ref] Figure 8: Plots of T CRP and T CLP vs [/fig_ref] : Time dependence of L for poly(MeEOGE) 2.5k (left) and poly(EtEOGE) 2.5k (right); [fig_ref] Figure 9: Schematic illustration of [/fig_ref] : Time dependence of L for poly(MeEO 2 GE) 2.5k (left) and poly(EtEO 2 GE) 2.5k (right); [fig_ref] Figure 1: Chemical structures of [/fig_ref] : L distribution for poly(MeGE) 2.5k (left) and poly(EtGE) (right) 2.5k ; [fig_ref] Figure 1: Chemical structures of [/fig_ref] : L distribution for poly(MeEOGE) 2.5k (left) and poly(EtEOGE) 2.5k (right); [fig_ref] Figure 1: Chemical structures of [/fig_ref] : L distribution for poly(MeEO 2 GE) 2.5k (left) and poly(EtEO 2 GE) 2.5k (right).
[fig] Figure 1: Chemical structures of (a) poly(MeGE), (b) poly(MeEOGE), (c) poly poly(EtGE), (e) poly(EtEOGE), and (f) poly(EtEO2GE). Red marker shows methyl marker shows ethyl group. [/fig]
[fig] *: Cloud point temperature observed by the turbidity measurements[10]. The LCST-type phase transition was not observed at temperatures up to 374 K[10]. [/fig]
[fig] Figure 2: Rg distribution for poly(MeGE)2.5k (left) and poly(EtGE)2.5k (right). The inset is the corresponding time dependence of Rg. [/fig]
[fig] Figure 3: R g peaks plotted versus temperatures for poly(MeGE) 2.5k , poly(MeEOGE) 2.5k , and poly(MeEO 2 GE) 2.5k with error bars. [/fig]
[fig] Figure 4: Rg peaks plotted versus temperatures for poly(EtGE)2.5k, poly(EtEOGE)2.5k, and poly(EtEO2GE)2.5k with error bars. [/fig]
[fig] Figure 5: R g distribution for poly(MeGE) 5k (left) and poly(EtGE) 5k (right). The inset is the corresponding time dependence of R g . [/fig]
[fig] Figure 6: RDFs of (a) poly(MeGE)2.5k, (b) poly(EtGE)2.5k, (c) poly(MeEOGE)2.5k, (d) poly(EtEOGE)2.5k, (e) poly(MeEO2GE)2.5k, and (f) poly(EtEO2GE)2.5k for O atoms of water measured from C atoms at the side-chain ends. [/fig]
[fig] Figure 7, Figure 8: Temperature response of the first peak intensity in the RDFs of O atoms (water) observed from C atoms at the side-chain ends, (a) poly(MeGE) 2.5k and poly(EtGE) 2.5k , (b) poly(MeEOGE) 2.5k and poly(EtEOGE) 2.5k , and (c) poly(MeEO 2 GE) 2.5k , and poly(EtEO 2 GE) 2.5k . The solid line shows T CLP and the broken line shows T CRP . Nanomaterials 2023, 13, Plots of and vs. number of oxyethylene units for methyl-terminated poly(Me(EO)nGE) (a) and ethyl-terminated poly(Et(EO)nGE) (b). Open and filled circles correspond to [/fig]
[fig] Figure 9: Schematic illustration of (a) poly(Me(EO)nGE) and (b) poly(Et(EO)nGE) in water. [/fig]
[fig] Figure 8: Plots of T CRP and T CLP vs. number of oxyethylene units for methyl-terminated poly(Me(EO) n GE) (a) and ethyl-terminated poly(Et(EO) n GE) (b). [/fig]
[fig] Author: Contributions: The manuscript was written through the contributions of all authors. Conceptualization: E.T. and S.S.; Investigation: E.T.; Data curation: E.T.; Validation: S.S.; Writing-original draft: E.T.; Writing-review and editing: S.S., T.K., T.Y., T.S. and T.I. All authors have read and agreed to the published version of the manuscript. [/fig]
[fig] Funding: The authors are grateful for the financial support from Grants-in-Aid for Scientific Research C (17K05025) and Grant-in-Aid for Challenging Exploratory Research (22K19929029) from JSPS. [/fig]
[table] Table 1: Calculated principal R g peaks and experimental T CLP . [/table]
[table] Table 2: Calculated L/L max at each temperature.Table 3. Calculated L temp /L max with their SD and CV. [/table]
[table] Table 4: Calculated L/L max of poly(MeEOGE) for molecular weights of 1250, 2500, and 5000. [/table]
[table] Table 5: Time-averaged side chain lengths and their fluctuations. [/table]
[table] Table 6: Calculated T CRP vs. experimental T CLP . [/table]
[table] Table 7: Average numbers of hydrogen bonds between polymer and water calculated over 20 ns at 278, 300, 323, 343, and 368 K. [/table]
|
10.1111/ina.13118
|
CCBY
|
9828560
|
36305066
|
s2orc_pubmed_articles
|
SARS‐CoV‐2 indoor environment contamination with epidemiological and experimental investigations
## | introduc ti on
Increasing scientific evidence indicates the dominance of short-and long-range airborne transmission of SARS-CoV-2. [bib_ref] Scientific evidence supports aerosol transmission of SARS-COV-2, Macintyre [/bib_ref] [bib_ref] Ten scientific reasons in support of airborne transmission of SARS-CoV-2, Greenhalgh [/bib_ref] [bib_ref] Covid-19 has redefined airborne transmission, Tang [/bib_ref] [bib_ref] Airborne transmission of respiratory viruses, Wang [/bib_ref] [bib_ref] COVID-19 rarely spreads through surfaces. So why are we still deep cleaning?, Lewis [/bib_ref] [bib_ref] Viral load of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in respiratory..., Coleman [/bib_ref] The observed transmission risks have been higher indoors than outdoors, [bib_ref] Outdoor transmission of SARS-CoV-2 and other respiratory viruses: a systematic review, Bulfone [/bib_ref] and discussion on precautions for hospital and home environments has been intense.
In a study that aerosolized SARS-CoV-2 under laboratory conditions, aerosols' infectivity was retained for up to 16 h, [bib_ref] Persistence of severe acute respiratory syndrome coronavirus 2 in aerosol suspensions, Fears [/bib_ref] while another study estimated the half-life in aerosols to be approximately 1.1-1.2 h (95% CI 0. . [bib_ref] Aerosol and surface stability of SARS-CoV-2 as compared with SARS-CoV-1, Van Doremalen [/bib_ref] Outside of the laboratory, signs of viable SARS-CoV-2 in the air have been detected, [bib_ref] Viable SARS-CoV-2 in the air of a hospital room with COVID-19 patients, Lednicky [/bib_ref] [bib_ref] Isolation of SARS-CoV-2 from the air in a car driven by a..., Lednicky [/bib_ref] [bib_ref] Detection and isolation of airborne SARS-CoV-2 in a hospital setting, Rufino De Sousa [/bib_ref] and the virus has also been cultured from exhaled air. [bib_ref] Infectious SARS-CoV-2 in exhaled aerosols and efficacy of masks during early mild..., Adenaiye [/bib_ref] A direct link between SARS-CoV-2 viral load, emission, and airborne concentration was recently demonstrated by Buonanno et al. [bib_ref] Link between SARS-CoV-2 emissions and airborne concentrations: closing the gap in understanding, Buonanno [/bib_ref] A few studies have detected SARS-CoV-2 RNA in the air at home-environment. [bib_ref] Are the portable air cleaners (PAC) really effective to terminate airborne SARS-CoV-2?, Rodríguez [/bib_ref] [bib_ref] SARS-CoV-2 in residential rooms of two self-isolating persons with COVID-19, Shankar [/bib_ref] In hospitals, PCR-based studies have found SARS-CoV-2 RNA in room air, [bib_ref] Aerodynamic analysis of SARS-CoV-2 in two Wuhan hospitals, Liu [/bib_ref] [bib_ref] Aerosol and surface contamination of SARS-CoV-2 observed in quarantine and isolation care, Santarpia [/bib_ref] [bib_ref] Detection of air and surface contamination by SARS-CoV-2 in hospital rooms of..., Chia [/bib_ref] [bib_ref] Hidden hazards of SARS-CoV-2 transmission in hospitals: a systematic review, Ribaric [/bib_ref] [bib_ref] Airborne SARS-CoV-2 surveillance in hospital environment using high-flowrate air samplers and its..., Ang [/bib_ref] [bib_ref] Characterization of hospital airborne SARS-CoV-2, Stern [/bib_ref] [bib_ref] Evidence of air and surface contamination with SARS-CoV-2 in a Major Hospital..., Da Silva [/bib_ref] [bib_ref] Risk of air and surface contamination of SARS-CoV-2 in isolation wards and..., Wang [/bib_ref] [bib_ref] Environmental SARS-CoV-2 contamination in hospital rooms of patients with acute COVID-19, Nagle [/bib_ref] as well as from air conditioning filters located over 50 m from the patient room. [bib_ref] Long-distance airborne dispersal of SARS-CoV-2 in COVID-19 wards, Nissen [/bib_ref] Even though previous studies have mainly used long collection times or high flow rates, challenge is that only a proportion of the air present in a room can be analyzed.
Additionally, indoor turbulence highly affects local concentrations. [bib_ref] Highresolution large-eddy simulation of indoor turbulence and its effect on airborne transmission..., Auvinen [/bib_ref] Thus, questions remain about the risk of infection during shorter meetings or in rooms with a larger air space, and whether the findings would be similar in the home environment. As environmental sampling is highly demanding and sample sizes rather small, more patient data are also needed to draw further conclusions in future systematic reviews.
According to laboratory studies, the stability of SARS-CoV-2 on surfaces varies depending on the surface type and environmental conditions. [bib_ref] Aerosol and surface stability of SARS-CoV-2 as compared with SARS-CoV-1, Van Doremalen [/bib_ref] [bib_ref] Stability of SARS-CoV-2 and other coronaviruses in the environment and on common..., Aboubakr [/bib_ref] [bib_ref] Increasing temperature and relative humidity accelerates inactivation of SARS-CoV-2 on surfaces, Biryukov [/bib_ref] [bib_ref] Stability of SARS-CoV-2 in different environmental conditions, Chin [/bib_ref] [bib_ref] Survival of SARS-CoV-2 on clothing materials, Virtanen [/bib_ref] However, its ability to sustain infectivity on surfaces outside laboratory conditions is largely unknown. [bib_ref] Survival of enveloped and non-enveloped viruses on inanimate surfaces, Firquet [/bib_ref] SARS-CoV-2 RNA has been found, for example, on high-touch surfaces, floors, and toilets, [bib_ref] Aerosol and surface contamination of SARS-CoV-2 observed in quarantine and isolation care, Santarpia [/bib_ref] [bib_ref] Detection of air and surface contamination by SARS-CoV-2 in hospital rooms of..., Chia [/bib_ref] [bib_ref] Aerosol and surface distribution of severe acute respiratory syndrome coronavirus 2 in..., Guo [/bib_ref] [bib_ref] Environmental contamination with SARS-CoV-2 in nursing homes, Mody [/bib_ref] and there are a few possibly positive culture findings of SARS-CoV-2 from the surfaces. [bib_ref] Severe acute respiratory syndrome coronavirus 2 environmental contamination in hospital rooms is..., Warren [/bib_ref] [bib_ref] Detection of SARS-CoV-2 on surfaces in households of persons with COVID-19, Marcenac [/bib_ref] [bib_ref] Detection and quantification of infectious severe acute respiratory coronavirus-2 in diverse clinical..., Lin [/bib_ref] The effect of age and neutralizing antibodies (NAbs) on the spread of SARS-CoV-2 has been speculated, there is a lack of clear evidence for the role of patient-related factors.
This study sought to increase knowledge of SARS-CoV-2 transmission in different environments by analyzing air, surface, and patient samples from a COVID-19 cohort ward in Helsinki University Hospital (HUS), Finland, and from patients' homes. The aims were to determine whether SARS-CoV-2 RNA or viable virus could be found in the home and hospital environments, and which patient-and environment-related factors affect the risk of environmental contamination. A team consisting of researchers from HUS, the University of Helsinki, the Finnish Meteorological Institute, and the Finnish Institute of Occupational Health was established to enable a multidisciplinary approach to the above research questions.
## | me thods
## | index patients and safety measures
Patients were voluntary participants with a qRT-PCR-confirmed symptomatic COVID-19 infection between 1.7.2020 and .
None of the participants had been vaccinated. As infectivity has been observed to be highest in early disease, the patient with the most recent onset of symptoms was selected as the index patient, [bib_ref] SARS-CoV-2, SARS-CoV, and MERS-CoV viral load dynamics, duration of viral shedding, and..., Cevik [/bib_ref] except for collection 13, where all the patients in the room had been symptomatic for over 10 days and the patient with the freshest positive PCR result (P26) was selected . Environmental measurements were performed in the vicinity of the index patients.
Saliva samples were also collected from other patients who were in the ward at the same time and who agreed to the study in 4 collections (named as "other"). Family members of the home patients were examined for infection and seroconversion. Environmental sampling was performed twice with patients P2 and P3, and P2 was considered as an index due to the more recent start of the symptoms; however, some personal items from both were sampled. The sampling process is presented in a flow chart in .
All research personnel conducting the sampling followed aerosol safety protocols and precautions and no infections were detected.
All procedures that involved human participants, including environmental sampling, were conducted in accordance with the ethical standards of the institutional or national research committee and the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. The Ethics Committee of Helsinki University Hospital approved the study protocol (HUS/1701/2020). All respondents provided written informed consent prior to their participation.
## | infection prevention protocols in the hospital
The infection prevention and control protocols on the COVID ward included hand hygiene, universal masking for staff (FFP2/3 for ICU and surgical masks for the COVID ward), guidance on social distancing (2 m), and personal protecting equipment (PPE) following droplet precautions. The patients did not use face masks. The ward and
## Practical implications
- The finding of SARS-CoV-2 RNA from the air in the absence of aerosol generating procedures (AGP) and in the absence of respiratory symptoms emphasizes the use of respiratory protection and airborne precautions also in situations where AGPs are not performed and regardless of the patients' symptoms.
- Families that used respiratory protection were able to prevent further infections.
- Air and surface contamination was detected in both homes and hospital even though the day from the start of the symptoms was later in hospital measurements. This may follow from more severe disease and increased viral loads which were associated with older age. Infection control measures should be used in both environments to prevent further infections.
ICU were cleaned twice a day between 9 to 10 am and 4 to 5 pm.
The sample collections were done between cleaning around 11 am to 9 pm and thus reflect quite reliably patients' infection status of the collection day. The specific cleaning protocol is presented in the supplement.
## | cell lines
Vero E6 cells (VE6) and their TMPRSS2-expressing clone VE6-TMPRSS2-10 (VE6T) [bib_ref] A generic, scalable, and rapid time-resolved förster resonance energy transfer-based assay for..., Rusanen [/bib_ref] were grown as previously described. [bib_ref] Kinetics of neutralizing antibodies of COVID-19 patients tested using clinical D614G, B.1.1.7,..., Virtanen [/bib_ref] To inhibit fungal growth, 0.205 μg/ml of amphotericin B (Fungizone, Thermo Scientific) was added to the medium of the cells that were taken to the hospital for aerosol collections. The used VE6 cell line is originally from ATCC (American type Culture Collection 45 ), and the VE6T cell line has been modified from the original line according to the previous study. [bib_ref] A generic, scalable, and rapid time-resolved förster resonance energy transfer-based assay for..., Rusanen [/bib_ref]
## | sampling protocols for air sampling
Seven different air collection methods were used. Details of the collections and samples are presented in . Sampling times and air volumes varied between different sampling methods depending on the expected optimal collection time for each device according to the manufacturer and previous studies, and the knowledge gathered during the study. A Dekati PM10 cascade impactor (20 L/min air flow, model PMS-420) with three stages (>10.0 μm, 2.5-10 μm, and 1.0-2.5 μm), including a backup filter for particles <1 μm, was used in 11 collections. The three impaction stages were fitted with 25-mm-diameter cellulose acetate membrane filters (CA filter, GE Healthcare Life Sciences) and the backup plate with a 40-mm CA filter. Analyzing the three stages and backup filter, particle distribution according to aerodynamic size (PM10, PM2.5, and PM1) can be ascertained. The collector was placed within 1-2 m from the patient, and particles were collected for 2-4 h. After sampling, filters were immediately placed in 2 ml (25-mm filter) or 3 ml (40-mm filter) of minimal essential Eagle's medium (MEM, Sigma-Aldrich).
The BioSpot 300p bioaerosol sampler prototype (Aerosol Devices Inc.) has a flow rate of 8 L/min and a mechanism that allows water to condense on aerosol particles from as small as 5-10 nm to 20 μm in diameter and minimize the stress when the sample is impacted onto the surface with the collection medium. To increase the sample collection rate, the biosampler is equipped with eight wicking tubes fitted with three nozzle jets to secure gentle transfer of the sample. This sampler was used in 8 collections for 1.5-4 h within a distance of 1-2.5 m from the patient, and the sample was collected in 1-2 ml of MEM.
As a more portable solution for personal area air sampling, a standard 25-mm gelatin (Sartorius Stedim Biotech) or mixed cellulose ester (MCE) filter equipped in the Button sampler with a Gilian F I G U R E 1 Patient inclusion and sampling process and additional analysis performed in the study 5000 air sampling pump, 4 L/min air flow, and a porous curved surface inlet was used in 9 collections. The Button sampler collects particles smaller than 100 μm. [bib_ref] Performance characteristics of the button personal inhalable aerosol sampler, Aizenberg [/bib_ref] The stability of SARS-COV-2 on two filter materials was compared under laboratory conditions to select the more optimal filter type and to optimize the collection time (details in Supplementary Material). Samples were collected for 10-30 min from patients' breathing area. Depending on the health status, a conversation was prompted to increase the output of aerosols. The collection filter was removed into 3 ml of MEM immediately after collection ended.
Three Andersen cascade impactors (400 W pump and 28.3 L/min flow rate) were used simultaneously in six collections. The impactors consist of six stages with size cut points of (1) >7 μm, (2) 4.7-7.0 μm, An additional inlet was used during measurements limiting the upper limit of the particle size to 12 μm. To ensure the correct volume flow rate, each Andersen impactor was fitted with a TSI flow meter. To evaluate the real-time particle number concentration during the hospital collections and to gather additional air samples, a Dekati eFilter was used in two collections. The eFilter monitors changes in real-time particle concentration by utilizing a small diffusion charger powered by an inner chargeable battery. The charge changes were automatically translated into a signal, which was recorded on a data card. When postprocessing the data, the raw charge signal was further converted to represent particle number concentrations using a conversion factor (411 cm −3 fA −1 ) provided by the manufacturer. A count median diameter (CMD) of 60 nm and a geometric standard deviation (GSD) of 1.5 were assumed. [bib_ref] Ultrafine particle concentrations in a hospital, Riesenfeld [/bib_ref] In addition, the eFilter simultaneously collected samples on a 47-mm gelatin filter using an external pump. After sample collection, the gelatin filter was transferred into 6 ml of MEM. The eFilter was fitted with the same EPA-designed inlets as the Andersen cascade impactors. The Living cells were transported to the laboratory in a warm environment with heat accumulators warmed to 37°C. One plate was used as a negative control to ensure that the cells survived the transport. Other samples were transported with cold accumulators and handled during the same or next day.
## | sampling protocols for surface sampling
Altogether, 252 surface samples in 26 collections for qRT-PCR testing were taken from surfaces in possible direct or indirect contact with the patient
## | other sampling protocols
Saliva samples were taken from 26 index patients either with a Dacron swab (collections 5-9) or by spitting into a Falcon tube (from collection 10 onwards). Ten additional saliva samples were collected from other patients from the ward in four collections and from seven healthy family members of home-treated patients. If possible, patients were asked to rinse their mouth before sampling. In collection 23, the index patient and a healthy family member also took followup saliva samples until 12 days from the start of the patient's symptoms. In collection 26, follow-up saliva samples were taken from patients until Days 14-17 from the start of symptoms.
Nasopharyngeal samples from consenting patients were taken and sent to HUSLAB for a fresh diagnostic PCR. [bib_ref] Comparison of two commercial platforms and a laboratory-developed test for detection of..., Mannonen [/bib_ref] [bib_ref] Real-life clinical sensitivity of SARS-CoV-2 RT-PCR test in symptomatic patients, Kortela [/bib_ref] Serum samples from consenting patients were taken within a day from sampling and tested for SARS-CoV-2 IgG antibodies with two different tests. [bib_ref] Performance of six SARS-CoV-2 immunoassays in comparison with microneutralisation, Jääskeläinen [/bib_ref] Serum samples (dilutions 1:10 to 1:640) were studied with the microneutralization assay. [bib_ref] Serological and molecular findings during SARS-CoV-2 infection: the first case study in..., Haveri [/bib_ref] Blood lymphocyte and eosinophil counts, and plasma CRP from consenting patients were measured within a day of sampling, and plasma ferritin, ALP, ALT, D-dimer, and fibrinogen levels within 3 days. The respiration rate and SpO2 levels were measured during the same day .
Since the first cases caused by variants of concern (VoC) were detected in Finland at the end of December 2020, they were determined from all patients as a part of routine diagnostics. This information was used to compare the results between VoC strains (mainly alpha in Finland) and non-VoC strains. Virus strains of collections 1-22 (P1-P45) were considered as non-VoC, as they were collected before the first cases were reported in Finland. were analyzed with the thermocycler software Rotor-Gene 6.0.31 (Qiagen) using similar criteria as with other samples described above.
## | culturing protocols
Samples were initially cultured in VE6 cells (collections 1-18), which were changed to VE6T cells after reports of these being more sensitive (collections [bib_ref] Detection of air and surface contamination by SARS-CoV-2 in hospital rooms of..., Chia [/bib_ref] [bib_ref] Hidden hazards of SARS-CoV-2 transmission in hospitals: a systematic review, Ribaric [/bib_ref] [bib_ref] Airborne SARS-CoV-2 surveillance in hospital environment using high-flowrate air samplers and its..., Ang [/bib_ref] [bib_ref] Characterization of hospital airborne SARS-CoV-2, Stern [/bib_ref] [bib_ref] Evidence of air and surface contamination with SARS-CoV-2 in a Major Hospital..., Da Silva [/bib_ref] [bib_ref] Risk of air and surface contamination of SARS-CoV-2 in isolation wards and..., Wang [/bib_ref] [bib_ref] Environmental SARS-CoV-2 contamination in hospital rooms of patients with acute COVID-19, Nagle [/bib_ref] [bib_ref] Long-distance airborne dispersal of SARS-CoV-2 in COVID-19 wards, Nissen [/bib_ref] [bib_ref] Highresolution large-eddy simulation of indoor turbulence and its effect on airborne transmission..., Auvinen [/bib_ref] [bib_ref] Stability of SARS-CoV-2 and other coronaviruses in the environment and on common..., Aboubakr [/bib_ref] [bib_ref] Increasing temperature and relative humidity accelerates inactivation of SARS-CoV-2 on surfaces, Biryukov [/bib_ref] [bib_ref] Stability of SARS-CoV-2 in different environmental conditions, Chin [/bib_ref]. [bib_ref] Ultrafine particle concentrations in a hospital, Riesenfeld [/bib_ref] Air samples that were collected directly on cells were cultured as such, and the rest of the air and surface samples and 75 μl of saliva were used for culturing in 6-well plates. Medium was added to the final volume of 3 ml (saliva) or 2 ml (other samples). Efilter samples were cultured in two wells (3 ml/well). Samples were cultured at 37°C for 10-14 days and checked for cytopathic effect (CPE).
A 200μl sample of culture medium was taken from those samples that had unclear results based on microscopic observation or possible CPE and tested with N Charité qRT-PCR. Culturing was considered positive if CPE was detected and the Ct value of qRT-PCR performed from the culture media was under 20. If Ct value was higher, it was judged to be caused by original (possibly noninfectious) virus in the sample instead of virus growth. All virus culturing was performed in a BSL3 laboratory.
Optimization of the culturing protocols is described in more detail in the Supplementary Material.
## | statistical tests and design
Statistical tests were carried out with SPSS IBM Statistics version 27.
When comparing means between two independent groups, data were first tested for normality with the Shapiro-Wilk test before testing them either with the independent-samples t-test or a non-parametric test (independent-samples Mann-Whitney U-test for two groups and independent-samples Kruskal-Wallis test for more than two groups
## | re sults
## | patient characteristics and collection surroundings
We
## | sars-cov-2 rna in air
Overall, 258 air samples were obtained from 29 air collections .
The samples were divided into actively and passively collected samples . Estimated copy numbers varied between 1.04 × 10 3 copies/ml and 2.05 × 10 7 copies/ ml. All air samples were cultured, but no viable viruses were observed.
The protocols and tests used to optimize the culturing protocol are . On-line particle con- , but distances from the patient were similar (p = 0.398, . In total, 12 deposition sample (11.5%) in 8 (57.1%) collections were positive for SARS-CoV-2 RNA . There was no statistically significant difference in the proportions of qRT-PCR-positive samples (p = 0.333) or collections (p = 1.000) between home and hospital . .
## | sars-cov-2 rna on surfaces
## | effects of patient factors on environmental contamination
Positive air samples were found even when the index patient did not report any respiratory symptoms (2/3, 66.6%). However, there was a statistically significant connection between low oxygen saturation (SpO2) levels and SARS-CoV-2 RNA findings from surfaces, and a possible but nonsignificant connection between low SpO2 levels and RNA findings from the air (surface: p = 0.026, air: . Toilet surfaces were qRT-PCR positive in 33.3%
[formula] p = 0.098, [/formula]
(3/9) of cases when the index patient had GI symptoms and 0% of cases (0/9) when the index patient did not report any GI symptoms (p = 0.229, . No positive environmental samples were obtained if the saliva sample from the index patient was negative with both qRT-PCRs. Positive surface samples were detected more often when there were multiple COVID-19 patients in the ward/house during the sampling (p = 0.018, . However, no statistically significant difference was detected for air collections (p = 0.845) . Possible but statistically nonsignificant associations were observed between positive environmental samples and an earlier symptom day, as well as an older age and S1c).
No statistically significant connections were found between air and
## | sars-cov-2 in saliva
## | sars-cov-2 antibodies in serum samples
## | transmission of covid-19 to family members
The spread of COVID-19 within the family was examined by collecting saliva samples from family members of the five home-treated patients and analyzing qRT-PCR results and SARS-CoV-2 antibody levels. In two families that used protective measures, including respiratory protection (surgical mask or respirator) and intensified cleaning, no further infections were detected. One of these families used masks in common areas, but not in their own rooms behind closed doors. In another family, the bedroom was shared, and masks were used all the time. However, in three families that did not apply
## | discuss ion
This study detected considerable SARS-CoV-2 RNA contamination from both home and hospital environments. The virus was found in the air in particle size ranges of 0.65-4.7 μm, 7.0-12.0 μm, >10 μm, and < 100 μm in diameter , supporting existing literature. [bib_ref] Aerodynamic analysis of SARS-CoV-2 in two Wuhan hospitals, Liu [/bib_ref] [bib_ref] Detection of air and surface contamination by SARS-CoV-2 in hospital rooms of..., Chia [/bib_ref] [bib_ref] Hidden hazards of SARS-CoV-2 transmission in hospitals: a systematic review, Ribaric [/bib_ref] [bib_ref] Characterization of hospital airborne SARS-CoV-2, Stern [/bib_ref] Our findings also support discoveries that normal respiratory activities generate infective particles even in the absence of AGPs, 2,6,64,65 and respiratory symptoms. Additionally, low oxygen saturation showed a connection with a higher possibility of SARS-CoV-2 surface findings and a potential connection with air findings, which could follow from increased particle generation due to respiratory stress. Most (83%, 15/18) of our positive air samples with known particle size were in particles smaller than 4.7 μm, which supports the findings that at least 85% of the viral load is emitted in aerosols smaller than 5 μm. [bib_ref] Viral load of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) in respiratory..., Coleman [/bib_ref] [bib_ref] Hidden hazards of SARS-CoV-2 transmission in hospitals: a systematic review, Ribaric [/bib_ref] [bib_ref] The size and culturability of patient-generated SARS-CoV-2 aerosol, Santarpia [/bib_ref] This is in line with the fact that particle generation produces a distribution which form depends on the activity that is causing the particles. In human respiratory activities, generated particles are mainly small, under 5 μm in dry size distribution. [bib_ref] Aerosol emission of adolescents voices during speaking, singing and shouting, Murbe [/bib_ref] [bib_ref] Comparing aerosol concentrations and particle size distributions generated by singing, speaking and..., Gregson [/bib_ref] A previous study showed that a high sampling flow rate increases the success rate in detecting SARS-CoV-2 from air samples. [bib_ref] Airborne SARS-CoV-2 surveillance in hospital environment using high-flowrate air samplers and its..., Ang [/bib_ref] only qRT-PCR findings, support concerns that a shorter exposure time should be considered, at least for close contacts. [bib_ref] Implementation and evolution of mitigation measures, testing, and contact tracing in the..., Mack [/bib_ref] Overall, the exposure risk is cumulating with time and no limit to zero risk can be determined. The risk for infection depends on the concentration exposed to (depending on ventilation and produced quanta) as well as persons immunity.when there were only two patients. Overall, larger spaces are considered safer than small ones due to the larger air volume per person. [bib_ref] A guideline to limit indoor airborne transmission of COVID-19, Bazant [/bib_ref] However, it seems that also larger indoor spaces may form a risk environment if occupied by an infected person for a prolonged time period. [bib_ref] Epidemiologic evidence for airborne transmission of SARS-CoV-2 during church singing, Katelaris [/bib_ref] [bib_ref] Indirect virus transmission in cluster of COVID-19 cases, Cai [/bib_ref] [bib_ref] Possible indirect transmission of COVID-19 at a squash court, Brlek [/bib_ref] In our study, all patients in the ward were COVID patients. However, in many countries, COVID-positive patients have been separated from COVID-negative ones with just curtains and distance. As hospital-acquired infections have been a significant part of overall infections and deaths, [bib_ref] deaths from hospital acquired covid are everyone's problem, Oliver [/bib_ref] [bib_ref] The contribution of hospital-acquired infections to the COVID-19 epidemic in England in..., Knight [/bib_ref] [bib_ref] Hospital-acquired SARS-Cov-2 infections in patients: inevitable conditions or medical malpractice?, Barranco [/bib_ref] it is important to reduce the risk of infections in hospital wards. It should be noted that the infective aerosol particles may still generate an infection risk even when larger space allows more dilution with increasing distance, as shown by Karan et al. [bib_ref] The risk of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) transmission from..., Karan [/bib_ref] Our findings were mainly from a close distance similar to a previous study that saw higher probability to environmental findings inside 2-meter range, 24 even though the risk for infection especially in prolonged exposure remains also further away. [bib_ref] Highresolution large-eddy simulation of indoor turbulence and its effect on airborne transmission..., Auvinen [/bib_ref] It is interesting that the proportion of the positive samples was similar in hospital and home even when the ventilation was more efficient in the hospital and patients have later symptom day. This may be due to more patients in the same room or higher overall viral load which has been associated with more severe disease and higher age in previous studies. [bib_ref] The relationship between COVID-19 viral load and disease severity: a systematic review, Dadras [/bib_ref] [bib_ref] Exhaled aerosol increases with COVID-19 infection, age, and obesity, Edwards [/bib_ref] We observed a trend for an older age being associated with a higher viral load and a larger number of positive surface samples but confirming this would require further studies with a larger sample sizes. In earlier studies, higher viral loads have been associated with an increased probability of viral transmission. [bib_ref] Quantifying the relationship between SARS-CoV-2 viral load and infectiousness, Marc [/bib_ref] [bib_ref] Risk factors for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA environmental..., Seo [/bib_ref] Possible reasons for the relationship between age and infectivity include reduced saliva production, differences in mucus viscosity and salivary immunoglobulins, [bib_ref] Aging-related changes in quantity and quality of saliva: where do we stand..., Xu [/bib_ref] Other more frequently qRT-PCR-positive surfaces included highly touched personal items, hospital equipment, and the floor, which is in line with the previous findings. [bib_ref] Aerosol and surface contamination of SARS-CoV-2 observed in quarantine and isolation care, Santarpia [/bib_ref] [bib_ref] Detection of air and surface contamination by SARS-CoV-2 in hospital rooms of..., Chia [/bib_ref] [bib_ref] Aerosol and surface distribution of severe acute respiratory syndrome coronavirus 2 in..., Guo [/bib_ref] Even though RNA may persist on surfaces for some time, RNA findings most likely result from contamination on the same day due to daily cleaning.
The building body of evidence supports airborne route predominance for SARS-CoV-2 transmission, [bib_ref] Scientific evidence supports aerosol transmission of SARS-COV-2, Macintyre [/bib_ref] [bib_ref] Ten scientific reasons in support of airborne transmission of SARS-CoV-2, Greenhalgh [/bib_ref] [bib_ref] Covid-19 has redefined airborne transmission, Tang [/bib_ref] [bib_ref] Airborne transmission of respiratory viruses, Wang [/bib_ref] [bib_ref] Link between SARS-CoV-2 emissions and airborne concentrations: closing the gap in understanding, Buonanno [/bib_ref] and an animal study indicates that aerosol inoculation is a more efficient route and causes more severe pathology and higher viral loads. [bib_ref] SARS-CoV-2 disease severity and transmission efficiency is increased for airborne compared to..., Port [/bib_ref] Fomites have not been proven to serve as the sole or primary vehicle of transmission. [bib_ref] Bringing transmission of SARS-CoV-2 to the surface: is there a role for..., Katona [/bib_ref] The probability for surface transmission is estimated to be likely rare, generally less than 1 in 10 000, and the disease manifestation milder.The environmental samples that commonly presented infectious virus in previous studies were mainly in direct contact with infected patients' mucus membranes, or saliva or sputum secretions (e.g., nasal prongs, nasal canula, used tissue, patients mask, and endotracheal tube). [bib_ref] Environmental SARS-CoV-2 contamination in hospital rooms of patients with acute COVID-19, Nagle [/bib_ref] [bib_ref] Detection and quantification of infectious severe acute respiratory coronavirus-2 in diverse clinical..., Lin [/bib_ref] In this study, families that took protective measures (including isolation of the infected family member) and respiratory protection (surgical masks or FFP2 respirators)
were able to prevent further infections even when qRT-PCR-positive samples were collected from both surfaces and air. However, in a household where all surfaces were cleaned many times a day but no respiratory protection was used, all family members became infected. This supports the importance of air hygiene, including also portable air cleaners as a supportive method as shown in previous studies, [bib_ref] Are the portable air cleaners (PAC) really effective to terminate airborne SARS-CoV-2?, Rodríguez [/bib_ref] [bib_ref] Highresolution large-eddy simulation of indoor turbulence and its effect on airborne transmission..., Auvinen [/bib_ref] and also encourages control of infection spread in homes. Similar findings supporting the use of masks, isolation with closed door, and opening windows in home environment were found to lower the risk of contamination in the work of Picard et al. [bib_ref] Home quarantine: a numerical evaluation of SARS-CoV-2 spread in a single-family house, Picard [/bib_ref] Overall our results indicate that transmission may happen through several transmission routes as supported also in previous systematic review. [bib_ref] Hidden hazards of SARS-CoV-2 transmission in hospitals: a systematic review, Ribaric [/bib_ref] Infection control is even more important with VoC strains that feature a higher rate of household transmission. [bib_ref] Increased household secondary attacks rates with variant of concern severe acute respiratory..., Buchan [/bib_ref] To better understand the infectivity and state of the infection compared to the environmental findings, we collected saliva and serum samples. SARS-CoV-2 was cultured from saliva during symptom Days 2-11. 42 SARS-CoV-2 RNA was detected in the saliva of patients who had already formed IgG and NAbs, which align with previous findings of prolonged RT-PCR-positivity. [bib_ref] Duration and key determinants of infectious virus shedding in hospitalized patients with..., Van Kampen [/bib_ref] [bib_ref] Clinical characteristics, cause analysis and infectivity of COVID-19 nucleic acid repositive patients:..., Li [/bib_ref] [bib_ref] Correlation between 3790 quantitative polymerase chain reaction-positives samples and positive cell cultures,..., Jaafar [/bib_ref] In addition, the saliva of P16 on symptom Day 11 was still positive in virus culture, even though the patient had NAbs. Moreover, we obtained positive air and surface samples when the index patient had a positive IgG result and NAbs, which agrees with the findings of Lei et al. [bib_ref] SARS-CoV-2 environmental contamination associated with persistently infected COVID-19 patients, Lei [/bib_ref] This contradicts the suggestion that NAbs solely could be a reliable marker for non-infectivity In the view of infection control, we agree with Lei et al., [bib_ref] SARS-CoV-2 environmental contamination associated with persistently infected COVID-19 patients, Lei [/bib_ref] Tan et al. [bib_ref] Air and surface contamination by SARS-CoV-2 virus in a tertiary hospital in..., Tan [/bib_ref] and Wölfel et al. [bib_ref] Virological assessment of hospitalized patients with COVID-2019, Wölfel [/bib_ref] [bib_ref] Revisiting SARS-CoV-2 environmental contamination by patients with COVID-19: the omicron variant does..., Glinert [/bib_ref] suggesting that the results from the earlier variants can still be considered to provide valid information for the current situation.
Our study also has some limitations. Even though the overall number of our samples is quite high compared to previous studies, it is still limited in the statistical aspect and only able to detect major differences and associations. As environmental sampling is time consuming and resource intensive, making it challenging to achieve a statistically large enough sample size, it is important to combine findings from several different studies for more detailed analysis.
We only conducted environmental sampling at a single time point. In the future, a longitudinal examination could enable a more accurate examination of the effects of the course of disease for environmental contamination. The mean time from the onset of symptoms until sampling varied between homes and hospital and may affect the results. However, as the infection requiring hospital treatment is more severe, the viral loads may stay high longer, [bib_ref] Trajectory of viral RNA load among persons with incident SARS-CoV-2 G614 infection..., Stankiewicz Karita [/bib_ref] and accordingly, no major differences were detected in viral load in saliva in our study between homes and hospital patients. This would provide rather similar expectation for environmental contamination as Buonanno et al examined. 14 Also, patients generally arrive to the hospital at the later stage of the disease (excluding hospital-acquired infections that were not detected in this dataset) which makes our dataset suitable to represent the real situation between homes and hospitals.
In addition, we only measured the IgG and NAb response, but viral secretion from mucus membranes can continue if the IgA response is weak. [bib_ref] IgA dominates the early neutralizing antibody response to SARS-CoV-2, Sterlin [/bib_ref] The IgA immune response should thus be examined further in upcoming studies. The qRT-PCR results might include some uncertainty due to the differences in the texture and fluidity of saliva and should be considered as estimates. As many samples were collected from a large patient hall, it is possible that some observed viruses might have originated from other than the index patient.
However, most of the surface samples were from patient-specific surfaces, and aerosols are known to concentrate near the source, [bib_ref] Modelling aerosol transport and virus exposure with numerical simulations in relation to..., Vuorinen [/bib_ref] indicating that most of the positive samples are expected to be produced by the index patient. Particle size cutoffs in Andersen samplers might be slightly higher than estimated, as the amount of liquid used in the sampling was slightly smaller than recommended due to practical reasons. Finally, we strongly suggest developing a new, more sensitive methodology for assessing the virus viability to better assess transmission mechanisms.
## | con clus ions
## Ack n owled g m ents
We thank Esa Pohjalainen for assistance in the laboratory, Emma
Klemetti for the illustrations of
## Co n fli c t o f i nte r e s t
The authors declare no competing interests.
[fig] 2. 7 |: RNA extraction and PCR protocols for air, saliva, and culture medium samples Trizol (Invitrogen) was used to extract RNA from all saliva samples and from air and culture medium samples of collections according to the manufacturers' instructions. A 200μl sample was added to 800 μl of Trizol reagent, and a resuspension volume of 50 μl was used. RNA was extracted from air and culture medium samples of collections 24 onwards with a QIAcube HT system and QIAamp 96 Virus QIAcube HT kit (QIAgen) using offboard lysis.All samples were tested with two different qRT-PCRs, N Charité,53 and N1 US CDC, 54 using TaqMan Fast Virus 1-Step Master Mix (ThermoFisher), a 20μl reaction volume, 45 cycles/ run, and fast cycling mode (annealing temperatures 55°C (N1 US CDC) and 58°C (N Charité)). The primer and probe concentrations of N Charité were according to the original publication, 53 and those of N1 US CDC were 500 nM of both primers and 125 nM of probe (Table S7). qRT-PCRs were performed using a Stratagene Mx3005P instrument (Agilent Technologies) with a Ct cutoff value of 0.04. The results were considered positive if both qRT-PCRs were positive with a Ct value under 40 or if one qRT-PCR was positive with a Ct value under 38. Comparable cutoff limits have been used in previous studies. 18,55,56 Samples with Ct values over 38 in one qRT-PCR and no Ct with the other one were treated as negative, even though the possibility of them being very weak positives could not be excluded. RNA extracted from the Fin/20 strain 52 culture was used as a positive control and nuclease-free water as a negative control. Limit of detection in different laboratories has been around 5 copies/reaction for N1 US CDC and up to 50 copies/reaction for N Charité. 57 The N gene transcript for qRT-PCR was prepared as follows: The target region (352-712, 360 bp) was amplified from SARS-CoV-2 RNA, Wuhan strain, and cloned into pGEM-T cloning vector (Promega) under control of the SP6 promoter. The presence of the insert was verified by sequencing and restriction enzyme analysis. After linearization of the plasmid by digestion with AscI (Thermo Fisher Scientific), RNA was generated using the RiboMAX™ Large Scale RNA production system with SP6 polymerase (Promega) according to the manufacturer's instructions. The transcribed RNA was then treated with DNAse I and purified with the RNeasy Mini Kit (QIAGEN). Finally, RNA was quantified by spectrophotometry, and the RNA copy number was calculated based on its concentration, length, and molecular weight. qRT-PCR was performed with N Charité PCR by including a transcript dilution series from 10 to 10 9 copies/reaction in triplicate and samples in duplicate. Copy numbers as copies/ml of saliva were calculated by tracing back from copies/ reaction reported by MxPro software (standard curve equation: Y = −3.570 × LOG[X] + 43.98, RSq = 0.984). Quantitating air samples in a similar way was unsuccessful due to weak positive samples and repeated freeze-and-thaw cycles and copy numbers for those were estimated based on the Ct values from initial qRT-PCR and above equation from a different run but the same protocol. [/fig]
[fig] 2. 8 |: RNA extraction and PCR protocols for surface samples RNA was extracted with the NucliSENS miniMAG kit (Biomerieux).Process control virus (mengovirus) was added to at least half of the samples. Tubes containing PBS and swabs were mixed by vortexing, and swabs were moved to 1 ml of high pH tris-glycine-beef extract buffer (TGBE, pH 9.5). The tubes were vortexed again and agitated at250 rpm for 5 min in an orbital shaker (IKAKS 2060 basic, Patterson Scientific), and the swabs were moved into a tube with 4 ml of lysis buffer, vortexed and agitated at 250 rpm for 10 min. PBS, TGBE, and lysis buffer were then combined, vortexed, and incubated for 10 min. PBS without process control virus was included as a negative control and PBS with process control virus as a positive control. The rest of the extraction was carried out according to the NucliSENS miniMAG kit instructions. The samples were further treated with the OneStep PCR Inhibitor Removal Kit (Zymo Research) according to the manufacturer's instructions. Samples were tested for SARS-CoV-2 with modified versions of N Charité 53 and N1 US CDC qRT-PCRs 54 and for process control virus. 58 The RT-qRT-PCR was carried out using a QuantiTect Probe RT-PCR kit (Qiagen). Reaction mixes included 10 μl of 2X QuantiTect Probe RT-PCR Master Mix, 0.2 μl of QuantiTect RT mix, 0.6 μM of forward and 0.8 μM of reverse primer, 0.2 μM of probe for N Charité qRT-PCR primers, and 5 μl of RNA template, and the volume was adjusted to 20 μl with water. For US CDC qRT-PCR, final concentrations of 0.5 μM for both primers and 0.2 μM for the probe were used. For mengovirus qRT-PCR, 1 μM of both primers and 0.2 μM of probe were used. N Charité and N1 US CDC runs included one 10 −4 dilution of SARS-CoV-2 RNA extracted from cell-grown virus as a standard positive control and one or two blanks as a standard negative control, and the reactions were performed in duplicate whenever the sample amount was sufficient. A Rotor Gene 3000 (Qiagen) realtime PCR cycler was used. The cycling conditions were reverse transcription for 30 min at 53°C, a denaturation step at 95°C for 15 min, followed by 45 cycles of amplification/denaturation at 95°C for 15 s, annealing at 58°C for 45 s, and extension at 72°C for 45 s. The results [/fig]
[fig] samples and 8: /10 of the samples from other patients were qRT-PCR positive. RNA copy numbers varied between 1.65 × 10 3 and 5.13 × 10 7 copies/ml (mean 3.55 × 10 6 copies/ml [SD 1.10 × 10 7 ]). Six of the qRT-PCR-positive samples taken between symptom days 2 and 11 were also positive in virus culture (five of which were index patients). Culture-positive samples had lower Ct values than culture-negative samples (p < 0.001 (N1 US CDC), p = 0.019 (N Charité), Figure S3). Age showed a trend of positive correlation with copy number, but it was not statistically significant (Spearman's rho = 0.339, p = 0.106, Figure S5a). The mean copy number in the saliva of the index patients was 9.37 × 10 5 copies/ml (SD 7.57 × 10 5 ) in collections that had qRT-PCR-positive air samples and 7.74 × 10 6 copies/ml (SD 7.26 × 10 6 ) in collections where all air samples were qRT-PCR negative (p = 0.536) (Figure S5e). The respective figures for surface collections were 5.61 × 10 6 copies/ml (SD 1.52 × 10 7 ) in positive and 1.54 × 10 5 copies/ml (SD 1.85 × 10 5 ) in negative collections (p = 0.291) (Figure S5e). No connections were observed between saliva culturing results and PCR from the environment: PCR-positive air samples were detected in 40% (2/5) of the collections when the saliva of the index patient was culture positive on a collection day and in 44% (8/18) of the collections when saliva was culture negative. In surface collections, the same numbers were 60% (3/5) and 65% (11/17). [/fig]
[fig] When 50 L: /min air samplers were used, no positive samples were detected, but when sampling flowrate was raised to 150 L/min 72% of samples were positive. 21 In our study, SARS-CoV-2 was detected from the air with a minimum collection period of 10 min (Andersen's impactor) and a minimum air volume of 72 L (Button sampler). With an average respiratory rate of 14/min and volume of 0.5 L/breath, this would mean exposure times of 40 min (Andersen) and 10 min (Button) for the examined virus variants (alpha and undetermined VoC (Andersen), as well as non-VoC (Button)). Current safety guidelines use 15 min exposure time regarding contact tracing. 69 Our results, although limited due to the low number of collections and [/fig]
[fig] Figure 2 ,: Sampo Oksanen for the help with the data tables and Dekati Ltd for lending measurement equipment. This study was financially supported by Business Finland Corona Co-Creation funding 40988/31/2020, the Scientific Advisory Board for Defense (MATINE) grant number VN/627/2020-PLM-9 (materials and equipments), Academy of Finland funding number 309570 and Academy of Finland COVID-19 special funding number 335681, as well as Helsinki University Hospital research grant M7100COVID, Jalmari ja Rauha Ahokkaan Säätiö-grant and Jane and Aatos Erkko foundation. [/fig]
|
10.1371/journal.ppat.1008458
|
CCBY
|
7205318
|
32339205
|
s2orc_pubmed_articles
|
Hyd ubiquitinates the NF-κB co-factor Akirin to operate an effective immune response in Drosophila
The Immune Deficiency (IMD) pathway in Drosophila melanogaster is activated upon microbial challenge with Gram-negative bacteria to trigger the innate immune response. In order to decipher this nuclear factor κB (NF-κB) signaling pathway, we undertook an in vitro RNAi screen targeting E3 ubiquitin ligases specifically and identified the HECT-type E3 ubiquitin ligase Hyperplastic discs (Hyd) as a new actor in the IMD pathway. Hyd mediated Lys 63 (K63)-linked polyubiquitination of the NF-κB cofactor Akirin was required for efficient binding of Akirin to the NF-κB transcription factor Relish. We showed that this Hyd-dependent interaction was required for the transcription of immunity-related genes that are activated by both Relish and Akirin but was dispensable for the transcription of genes that depend solely on Relish. Therefore Hyd is key in NF-κB transcriptional selectivity downstream of the IMD pathway. Drosophila depleted of Akirin or Hyd failed to express the full set of genes encoding immune-induced anti-microbial peptides and succumbed to immune challenges. We showed further that UBR5, the mammalian homolog of Hyd, was also required downstream of the NF-κB pathway for the activation of Interleukin 6 (IL6) transcription by LPS or IL-1β in cultured human cells. Our findings link the action of an E3 ubiquitin ligase to the activation of immune effector genes, deepening our understanding of the involvement of ubiquitination in inflammation and identifying a potential target for the control of inflammatory diseases.
[formula] a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 [/formula]
Introduction During evolution, metazoans developed strategies to effectively protect themselves from microbial threats. Since the molecular pathways mediating the innate immune response in insects and mammals are conserved, the fruit fly Drosophila melanogaster is a relevant model to explore the immune response [bib_ref] Inflammation and cancer: role of nuclear factor-kappaB activation, Maeda [/bib_ref]. In Drosophila [bib_ref] The Drosophila systemic immune response: sensing and signalling during bacterial and fungal..., Ferrandon [/bib_ref] , the defense against microbes is executed mainly through the production of antimicrobial peptides (AMPs) under the control of two NF-κB transcription factors: Dorsal-related Immunity Factor (DIF) and Relish, respectively acting downstream of Toll and IMD pathways and homologues of mammalian RelB and p50 transcription factors.
Posttranslational regulation of proteins by the ubiquitin pathway is key for proper immune response [bib_ref] The ubiquitin system in immune regulation, Park [/bib_ref]. The conjugation of ubiquitin polymers to target proteins by an ubiquitin ligase is a key mechanism for controlling the activity, localization, or stability of the targets. Lysine (Lys) residues of proteins can be modified by a polymer of ubiquitin (polyubiquitin) linked through Lys 48 (K48) or Lys 63 (K63) of ubiquitin molecules. Whereas K48-linked polyubiquitin mainly triggers degradation of proteins by the proteasome, K63-linked polyubiquitin mainly regulates the activity and the subcellular localization of proteins by modifying their proteinprotein interactions [bib_ref] Ubiquitin modifications, Swatek [/bib_ref]. In both mammals and Drosophila, ubiquitination is involved at various levels of the NF-κB pathways [bib_ref] The Drosophila ubiquitinspecific protease dUSP36/Scny targets IMD to prevent constitutive immune signaling, Thevenon [/bib_ref]. Furthermore, deregulation of ubiquitin ligases is implicated in inflammatory pathologies [bib_ref] Ubiquitin-Modifying Enzymes and Regulation of the Inflammasome, Kattah [/bib_ref] and tumor progression . HECT-domain E3 ligases directly attach ubiquitin to a substrate, conversely to RING domain E3 ubiquitin ligases . To achieve selectivity and specificity toward their substrates, HECT ubiquitin E3 ligases are tightly regulated, thus the identification of their substrates and regulators is critical in developing targets for drug discovery in the treatment of HECT E3 ubiquitin ligases related diseases .
In Drosophila, Death-associated inhibitor of apoptosis 2 (Diap2) is the only E3 ubiquitin ligase identified thus far as a positive regulator of the IMD pathway [bib_ref] Inhibitor of apoptosis 2 and TAK1-binding protein are components of the Drosophila..., Kleino [/bib_ref] [bib_ref] An RNA interference screen identifies Inhibitor of Apoptosis Protein 2 as a..., Gesellchen [/bib_ref]. IMD protein functions as an adaptor protein in the IMD signaling pathway. The activation of pathogen recognition receptors (PRRs) by bacterial infection initiates the IMD cascade that transduces immunity signals notably through Diap2 [bib_ref] A recessive mutation, immune deficiency (imd), defines two distinct control pathways in..., Lemaitre [/bib_ref] [bib_ref] Caspase-mediated cleavage, IAP binding, and ubiquitination: linking three mechanisms crucial for Drosophila..., Paquette [/bib_ref]. This protein is involved in the formation and activation of protein complexes organized around the IMD protein. To deepen our understanding of NF-κB pathway regulation by the ubiquitin system, we focused on identifying Drosophila ubiquitin ligases that are required for the activity of the IMD pathway through a RNAibased screen in cultured Drosophila S2 cells.
Several E3 ubiquitin ligases emerged from our screen as positive or negative regulators of the IMD pathway. We focused on Hyperplastic discs (Hyd) because it was the only HECTtype E3 ubiquitin ligase identified in this screen and it had also emerged as a potential IMD pathway regulator in a parallel pilot screen [bib_ref] Landscape of protein-protein interactions in Drosophila immune deficiency signaling during bacterial challenge, Fukuyama [/bib_ref]. Our results showed that Hyd was required in vivo for flies to survive an immune challenge with Gram-negative bacteria. Genetic epistasis analysis revealed that Hyd acted at the level of the NF-κB co-factor Akirin, which orchestrates the activation of a subset of NF-κB target genes in combination with the SWI/SNF chromatin remodeling complex [bib_ref] Akirins are highly conserved nuclear proteins required for NF-kappaB-dependent gene expression in..., Goto [/bib_ref] [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] [bib_ref] Akirin2 is critical for inducing inflammatory genes by bridging IkappaB-zeta and the..., Tartey [/bib_ref].
We showed that Hyd decorated Akirin with K63-polyubiquitin chains that were required for Akirin binding to the NF-κB homolog Relish. Furthermore, we observed that the human ortholog of Hyd, UBR5 (also known as EDD1) [bib_ref] Identification of a human HECT family protein with homology to the Drosophila..., Callaghan [/bib_ref] , played a conserved role in NF-κB signaling in human cell lines. Similarly to human-AKIRIN2 (AKIRIN2), UBR5 was required for the activation of only a subset of NF-κB target genes after stimulation by LPS or IL-1β in cultured human cells. Thus, upon immune challenge, ubiquitin chains are instrumental to bridge NF-κB and its co-factor Akirin to activate an effective immune response.
# Results
## The e3 ubiquitin ligase hyd is required for activation of the imd pathway
To identify E3 ubiquitin ligases that modulate the IMD pathway, we screened a library of 174 double-strand RNAs (dsRNAs) targeting Drosophila proteins classified as putative E3 ubiquitin ligases in Flybase [bib_ref] FlyBase at 25: looking to the future, Gramates [/bib_ref]. We used stably transfected Drosophila S2 cells expressing the Attacin-A-luciferase gene, a reporter of activation of the IMD pathway upon immune challenge with Gram-negative bacteria [bib_ref] Toll-related receptors and the control of antimicrobial peptide expression in Drosophila, Tauszig [/bib_ref]. We evaluated the ability of dsRNA targeting each putative E3 ubiquitin ligase to interfere with the luciferase reporter upon stimulation of cells with heatkilled Escherichia coli (HKE), a general IMD pathway agonist (S1 .
Diap2 is an E3 ubiquitin ligase that positively regulates the pathway by binding, polyubiquitinating and activating Death-related ced-3/Nedd2-like caspase (Dredd) which is required for Relish-mediated induction of antimicrobial peptides [bib_ref] The Drosophila IMD pathway in the activation of the humoral immune response, Kleino [/bib_ref] [bib_ref] Ubiquitylation of the initiator caspase DREDD is required for innate immune signalling, Meinander [/bib_ref]. Knockdown of Diap2 resulted in a strong decrease of the luciferase reporter induction upon immune stimulation relative to the control GFP-targeting dsRNA (dsGFP) that does not target any transcripts present in the cells [fig_ref] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway [/fig_ref] , providing proof of concept for the screen. Knockdown of six E3-ubiquitin ligases (M-cup, Mkrn1, CG2926, CG31807, Mura, and CG12200) resulted in a strong increase in reporter activity upon immune stimulation. Therefore these E3 ubiquitin ligases behave as negative regulators of the IMD pathway. Conversely, knockdown of two Really Interesting New Gene (RING) domain E3-ubiquitin ligases, Bon and CG5334, or of the homologous to the E6-AP carboxyl terminus (HECT) domain E3 ubiquitin ligase Hyd resulted in an important decrease in reporter activity [fig_ref] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway [/fig_ref]. This suggests that Bon, CG5334, and Hyd are positive regulators of the IMD pathway. We decided to focus on Hyd because it was the only HECT domain E3-ubiquitin ligase we identified.
To validate the reporter assay, we transfected Drosophila S2 cells with dsRNA targeting either the NF-κB factor Relish, the Relish cofactor Akirin, Hyd, or E3 ubiquitin ligases chosen randomly among those that presented in the screen the strongest increase or decrease of Attacin-A induction (Bon, Diap2, M-cup, Mkrn1 and Mura). Then, we stimulated the cells with HKE and monitored endogenous Attacin-A mRNA by quantitative reverse transcription PCR (RT-qPCR). Reducing the abundance of Relish, Akirin, or Hyd significantly decreased HKEmediated Attacin-A induction, compared to cells treated with dsGFP [fig_ref] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway [/fig_ref]. We observed that the RING-domain E3 ubiquitin ligases Bon, M-cup, Mkrn1, and Mura were required for the normal activation of Attacin-A expression and that the HECT E3 ubiquitin ligase Hyd acted as a positive regulator of Attacin-A expression in Drosophila S2 cells [fig_ref] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway [/fig_ref]. Drosophila E3 ubiquitin ligases were knocked down in S2 cells harboring the Attacin-A-luciferase reporter gene. Cells were transfected with individual dsRNAs targeting each E3 ligase before the IMD pathway was induced by stimulating the cells with heat-killed E. coli (HKE). Luciferase activity is expressed as an induction percentage compared to control cells treated with dsRNA targeting GFP. Three independent experiments were performed. Genes indicated by a gray bar were analyzed in [fig_ref] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway [/fig_ref] (B) Quantitative RT-PCR of Attacin-A mRNA from HKE-stimulated S2 cells transfected with dsRNA targeting GFP (negative control), Relish or Akirin (positive controls), and a subset of E3 ubiquitin ligases. After the ratio of stimulated over unstimulated values for each condition was determined, statistical significance was calculated by comparing genes knockdown to GFP dsRNA control. (C) Genetic epistasis experiment to place Hyd within the IMD pathway in S2 cells. The IMD pathway was induced by HKE stimulation or by transfecting the cells with Pgrp-LC, Imd-V5, or Rel-HA expressing plasmids. Cells were also transfected with dsRNA targeting Akirin or Hyd. Statistical significance was established by comparing values from the different conditions with cells treated with empty vector alone. Data are represented as mean ± standard deviation of three independent experiments (B-C). � P-value < 0.05; �� P-value < 0.01; ��� P-value < 0.001.
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response
In order to identify at which level of the IMD pathway Hyd is required, we undertook an epistasis analysis. Drosophila S2 cells were treated by dsRNA targeting Hyd or Akirin as a control and the IMD pathway was activated at different levels by transfecting either a truncated form of PeptidoGlycan Receptor Protein-Long Chain a (Pgrp-LCa), Imd or the 68kD activeform of Relish (Rel68) [bib_ref] Akirins are highly conserved nuclear proteins required for NF-kappaB-dependent gene expression in..., Goto [/bib_ref]. Measurement of Attacin-A expression by RT-qPCR assessed activation of the IMD pathway. We show that Hyd was required at the same level of Relish [fig_ref] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway [/fig_ref] to exert its positive regulation on IMD pathway activation.
## Hyd acts at the level of akirin to trigger full activation of the imd pathway
Downstream of the IMD pathway, Relish target genes are divided into two subsets: genes that depend only on Relish for their expression (including Attacin-D, Pgrp-LB, Pirk and the majority of negative regulators) and ones requiring Akirin, and the deposition of an acetyl group on the lysine 4 of the histone 3 (H3K4ac), in addition to Relish (including Attacin-A, Attacin-C, CecA1 and the majority of effectors) [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. Upon immune challenge in S2 cells, using RT-qPCR, we observed that Hyd depletion recapitulated the immune phenotype of cells depleted for Akirin [fig_ref] Fig 2: Hyd is required for the full activation of IMD response [/fig_ref]. Consequently, Hyd is acting on Akirin-dependent NF-κB transcriptional selectivity in vitro.
We next investigated if Akirin and Hyd were similarly required for NF-κB transcriptional selectivity in vivo, using RNAi. As Drosophila embryonic development is impaired in absence of Akirin, we used the C564-Gal4 transgene [bib_ref] Gene expression systems in Drosophila: a synthesis of time and space, Mcguire [/bib_ref] to express RNAi constructs targeting Akirin, Hyd and Relish in the adult fat body, the main immune organ of Drosophila [bib_ref] The Drosophila systemic immune response: sensing and signalling during bacterial and fungal..., Ferrandon [/bib_ref]. Flies depleted of Akirin (C564-Gal4/UAS-RNAi-akirin), Relish (C564-Gal4/UAS-RNAi-relish) or Hyd (C564-Gal4/UAS-RNAi-hyd1 or C564-Gal4/UAS-RNAi-hyd2) displayed an impaired survival following E. coli infection when compared to control flies (C564-Gal4/UAS-RNAi-GFP) or to PBS pricking [fig_ref] Fig 2: Hyd is required for the full activation of IMD response [/fig_ref].
Following immune challenge by E. coli, expression of Attacin-A, but not of Attacin-D, was reduced in the absence of Akirin or Hyd, when compared to control flies (C564-Gal4/UAS-RNAi-GFP) [fig_ref] Fig 2: Hyd is required for the full activation of IMD response [/fig_ref].
Our results suggest that Hyd is required at the level of Relish to activate the Akirin-dependent subset of Relish target genes during the immune response, allowing Drosophila to survive a Gram-negative bacterial challenge.
## Hyd-mediated k63-polyubiquitination of akirin is critical for akirin binding to relish
We next investigated if Akirin could be a bona fide target for the E3 ubiquitin-ligase Hyd. Coimmunoprecipitation assay in S2 cells showed that V5-tagged Hyd (Hyd-V5) [bib_ref] Hyperplastic discs differentially regulates the transcriptional outputs of hedgehog signaling, Wang [/bib_ref] binds to endogenous Akirin [fig_ref] Fig 3: Fig 3 [/fig_ref]. By contrast, V5-tagged Hyd-CS (Hyd-CS-V5), which displays a mutated HECT domain by conversion of the catalytic cysteine at position 2854 to serine [bib_ref] Hyperplastic discs differentially regulates the transcriptional outputs of hedgehog signaling, Wang [/bib_ref] , is unable to bind to Akirin [fig_ref] Fig 3: Fig 3 [/fig_ref]. As a control we confirmed that Diap2, the E3-ubiquitin ligase acting upstream of Akirin in the IMD signaling cascade [bib_ref] Inhibitor of apoptosis 2 and TAK1-binding protein are components of the Drosophila..., Kleino [/bib_ref] [bib_ref] An RNA interference screen identifies Inhibitor of Apoptosis Protein 2 as a..., Gesellchen [/bib_ref] , does not interact with Akirin (S3 .
Protein extracts from cells transfected with a tagged version of Akirin (Akirin-V5) were immunoprecipitated with an anti-V5 antibody. Using cellular fractioning, we showed that upon immune challenge, polyubiquitinated Akirin-V5 accumulated in the nucleus of Drosophila S2 cells [fig_ref] Fig 4: Hyd/UBR5 is necessary for NF-κB target genes activation [/fig_ref] Western-blot experiments with antibodies targeting K63-polyUb chains showed that Akirin was K63-polyubiquitinated 1h and 3h after immune challenge with HKE [fig_ref] Fig 3: Fig 3 [/fig_ref]. This immune-induced post-translational modification of Akirin was attenuated upon knockdown of Hyd [fig_ref] Fig 3: Fig 3 [/fig_ref].
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response According to literature, Hyd is suspected to also deposit K48 polyubiquitin chains on its target proteins. Here we observed that Akirin-V5 was also decorated by K48-polyubiquitin chains, but independently of Hyd (S5 .
Collectively, these data suggest that upon immune challenge, Hyd physically interacts with Akirin through its catalytic HECT domain to decorate Akirin with K63-polyUb chains. We previously published that Akirin physically bridges the NF-κB factor Relish and Bap60, a core member of the SWI/SNF chromatin-remodeling complex [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. To understand whether Akirin K63-polyubiquitination is instrumental for the interaction of Akirin with Relish or Bap60, we performed co-immunoprecipitation experiments in S2 cells depleted for Hyd and transfected with Akirin-V5 and Rel68-HA or BAP60-HA [fig_ref] Fig 3: Fig 3 [/fig_ref]. As previously reported [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] , Akirin-V5 co-precipitated either with the active form of the NF-κB homolog Relish (Rel68-HA) or with Bap60 (Bap60-HA) [fig_ref] Fig 3: Fig 3 [/fig_ref]. However, in the absence of Hyd, the interaction between Akirin-V5 and Rel68-HA was weakened [fig_ref] Fig 3: Fig 3 [/fig_ref]. Of note the interaction between Akirin-V5 and Bap60-HA is independent of Hyd [fig_ref] Fig 3: Fig 3 [/fig_ref]. Upon immune challenge, the transcriptional kinetic of Akirin-dependent and Akirin-independent genes was different. Akirin-independent genes (AttD, Pgrp-LB) were strongly transcribed after 1 hour, whereas Akirin-dependent genes (AttA, AttC) were strongly detectable after 3 hours (S6 Interestingly, the delayed activation of Akirin-dependent genes correlated with the robust K63-polyUb chains detection on Akirin at 3 hours after immune challenge [fig_ref] Fig 3: Fig 3 [/fig_ref].
These results suggest that, upon immune challenge, Hyd is required to deposit K63-polyUb chains on Akirin for subsequent binding to the NF-κB factor Relish, and efficient transcription of Akirin-dependent genes.
## Ubr5, the human ortholog of hyd, is required for nf-κb transcriptional selectivity during the inflammatory response
The Akirin-dependent molecular mechanism underlying the selective activation of NF-κB target genes is well conserved from Drosophila to mammals [bib_ref] Akirins are highly conserved nuclear proteins required for NF-kappaB-dependent gene expression in..., Goto [/bib_ref] [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] [bib_ref] Akirin2 is critical for inducing inflammatory genes by bridging IkappaB-zeta and the..., Tartey [/bib_ref]. It is established that i) in human, AKIRIN2 is the functional homologue of Drosophila Akirin [bib_ref] Akirins are highly conserved nuclear proteins required for NF-kappaB-dependent gene expression in..., Goto [/bib_ref] [bib_ref] Akirin2 is critical for inducing inflammatory genes by bridging IkappaB-zeta and the..., Tartey [/bib_ref] , and ii) downstream of NF-κB pathways, AKIRIN2 operates a dichotomy between the AKIRIN2-dependent and -independent genes to be transcribed [bib_ref] Akirin2 is critical for inducing inflammatory genes by bridging IkappaB-zeta and the..., Tartey [/bib_ref]. UBR5 is the human ortholog of the Drosophila E3-ubiquitin ligase Hyd [bib_ref] Identification of a human HECT family protein with homology to the Drosophila..., Callaghan [/bib_ref]. Comparison between UBR5 and Hyd revealed the presence of similar functional domains, as well as 39,9% of sequence identity and 56,5% of sequence similarity [fig_ref] Fig 4: Hyd/UBR5 is necessary for NF-κB target genes activation [/fig_ref]. Therefore, we addressed the potential requirement of UBR5 in NF-κB selective transcriptional response mediated by AKIRIN2 during the human inflammatory response. A previous study using human HeLa cells and mouse macrophages has identified AKIRIN2-dependent genes (such as IL6, Ifit1, and, IL12β), and AKIRIN2-independent genes (such as IL8, IL1α and TNF) [bib_ref] Akirin2 is critical for inducing inflammatory genes by bridging IkappaB-zeta and the..., Tartey [/bib_ref]. Here, we depleted THP1 (a human monocytic cell line) or HeLa cell lines for either NF-κB1, AKIRIN2 or UBR5 by siRNA (using scrambled siRNA as controls). We first confirmed that, as in HeLA cells, the AKIRIN2-dependent genes (IL6, Ifit1, and IL12β) and AKIRIN2-independent genes (IL8, IL1α and TNF) conserved their
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response transcriptional activation dichotomies in THP1 cells [fig_ref] Fig 4: Hyd/UBR5 is necessary for NF-κB target genes activation [/fig_ref]. Lacking NF-κB1 in these human cell lines impaired NF-κB target genes activation upon LPS or IL-1β stimulation [fig_ref] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway [/fig_ref] and HeLa cells depleted for AKIRIN2 or UBR5 showed a decreased level of AKIRIN2-dependent target gene (IL6, Ifit1, and, IL12β) activation when compared to control [fig_ref] Fig 4: Hyd/UBR5 is necessary for NF-κB target genes activation [/fig_ref]. This result suggests a conserved function of UBR5 in the selective transcription of NF-κB target genes mediated by AKIRIN2. However, the precise mechanisms by which UBR5 impacts the transcription of AKIRIN2-dependent target genes remains to be explored.
Taken altogether, our results show that the HECT-type E3 ubiquitin ligase Hyd/UBR5 is involved in NF-κB pathway regulation in Drosophila and mammals. In fruit fly, Hyd deposits K63-polyUb chains on Akirin and is required to bridge Akirin and the NF-κB factor Relish. This interaction is necessary for the transcription of an essential subset of NF-κB target genes, downstream of the IMD pathway [fig_ref] Fig 5: Schematic model of Hyd involvement in the IMD signaling pathway in Drosophila [/fig_ref].
# Discussion
Using Drosophila genetics, we describe here a function for the HECT E3 ubiquitin ligase Hyd in the innate immune response. Using human cell lines (HeLa and THP1), we could also show that this function of Hyd downstream of the NF-κB pathway is conserved in humans.
In both Drosophila and humans, NF-κB dependent signaling pathways are among the bestknown examples of the role of ubiquitin linkage to target proteins in signal transduction [bib_ref] The ubiquitin system in immune regulation, Park [/bib_ref] [bib_ref] Study of pathway cross-talk interactions with NF-kappaB leading to its activation via..., Ghosh [/bib_ref] , ubiquitination being involved at every level of the NF-κB pathway, from membrane receptors to chromatin-associated proteins. In order to identify new E3-ubiquitin ligases involved in the Drosophila innate immune response, we conducted a RNAi-based screen. We showed, in addition to Diap2 already known to be a bona fide member of the IMD pathway [bib_ref] Inhibitor of apoptosis 2 and TAK1-binding protein are components of the Drosophila..., Kleino [/bib_ref] [bib_ref] An RNA interference screen identifies Inhibitor of Apoptosis Protein 2 as a..., Gesellchen [/bib_ref] , that other RING-domain E3 ubiquitin ligases (CG5334, bon) were involved in the activation of the IMD pathway in Drosophila S2 cells after immune challenge. In addition, our results suggested that other RING-domain E3 ubiquitin ligases such as M-cup, Mkrin1 and Mura down-regulate IMD pathway target genes activation. This screen also suggested that a HECT E3-ubiquitin ligase, namely Hyd, is involved in the innate immune response.
In Drosophila, Hyd is reported to be located in the nuclear and in the cytoplasmic fraction of cells to participate in various phenomena during development such as cellular proliferation [bib_ref] The ubiquitin ligase Hyperplastic discs negatively regulates hedgehog and decapentaplegic expression by..., Lee [/bib_ref] [bib_ref] Genetic and molecular analysis of hyperplastic discs, a gene whose product is..., Mansfield [/bib_ref]. More precisely, Hyd shapes hedgehog signaling by differentially restraining the transcriptional activity of Cubitus interuptus via selective association with respective promoters [bib_ref] Hyperplastic discs differentially regulates the transcriptional outputs of hedgehog signaling, Wang [/bib_ref]. More recently, Hyd and its orthologue UBR5 were reported to act at the level of Wnt signaling target gene promoter to enable gene transcription. Here we identified the HECT E3 ubiquitin-ligase Hyd in Drosophila as responsible for the ubiquitination of Akirin and its subsequent binding to the NF-κB transcription factor Relish. In addition, identification of protein interactors for AKIR-1, the Caenorhabditis elegans homologue of Drosophila Akirin, also
## Hyd mediated-ubiquitination of akirin is necessary for interaction with relish. (a) co-immunoprecipitation assay between over-expressed akirin and
Hyd in S2 cells. The cells were transiently transfected with Akirin, Gal4, Hyd-V5 and/or Hyd-CS-V5 expressing plasmids. Cell lysates were immunoprecipitated with anti-Akirin coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-V5 or anti-Akirin antibodies. (B) Immunoprecipitation assay of K63-polyUb chains on Akirin before and after immune challenge. S2 cells were transiently transfected with Akirin-V5 expressing plasmid. Cell lysates were immunoprecipitated with anti-K63-polyUb coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-V5 antibodies. (C) Immunoprecipitation assay of Akirin after immune challenge. S2 cells were transiently transfected with Akirin-V5 expressing plasmid and dsRNA targeting GFP or Hyd. Cell lysates were immunoprecipitated with anti-V5 coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-K63-polyUb and anti-V5 antibodies. (D) Co-immunoprecipitation assay between over-expressed Akirin and Relish or Bap60 in S2 cells. The cells were transiently transfected with Akirin-V5 and Rel-HA or Bap60-HA expressing plasmids and dsRNA targeting GFP or Hyd. Cell lysates were immunoprecipitated with anti-V5 coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-HA or anti-V5 antibodies. Data are representative of 2 independent experiments. https://doi.org/10.1371/journal.ppat.1008458.g003
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response revealed Hyd/UBR5 [bib_ref] Evolutionary plasticity in the innate immune function of Akirin, Polanowska [/bib_ref]. Altogether these results point to a conserved function of the HECT E3 ubiquitin ligase Hyd/UBR5 as nuclear selector for gene activation.
Downstream of the IMD pathway, the NF-κB transcription factor Relish target genes could be divided into two subgroups, Akirin-dependent and Akirin-independent genes [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. Targeting Hyd by RNAi in Drosophila S2 cells dampened the activation of Akirin-dependent genes upon immune stimulation. Depleting Hyd from Drosophila fat-body prevents fly survival to immune challenge with the Gram-negative bacteria E. coli, demonstrating the biological relevancy of its function. Co-immunoprecipitation experiments showed that Hyd interacts with Akirin through its catalytic domain to deposit K63-polyUb chains. Of note, the
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response K63-polyubiquitination of Akirin by Hyd is performed only after immune challenge, suggesting that an immune-triggered signal governs this event and remains to be explored. Because our observations are based on overexpressed proteins, it would be of interest to evaluate if endogenous Akirins are K63-polyubiquitinated by Hyd upon immune challenge. Additionally, it is still unclear how the K63-polyubiquitin chains on Akirin physically interact with Relish to set a bridge, as no Ubiquitin Binding Domain (UBD) have been described for Relish. During development, Akirin is also known to be able to link another transcription factor, Twist [bib_ref] Akirin links twist-regulated transcription with the Brahma chromatin remodeling complex during embryogenesis, Nowak [/bib_ref]. It remains to be investigated if Hyd affects Akirin-Twist dependent transcriptional response.
The HECT Ubiquitin ligase family is known in mammals and Drosophila to regulate biological phenomena . We found that the mammalian ortholog of Hyd, UBR5 [bib_ref] Identification of a human HECT family protein with homology to the Drosophila..., Callaghan [/bib_ref] is involved in NF-κB transcriptional selective response in human cell lines as well. This suggests a conserved role for Hyd/UBR5 on AKIRIN2, even though we do not know if AKIRIN2 is ultimately ubiquitinated. A dedicated study of UBR5 role in NF-κB pathways is needed to completely assess it. It is known that UBR5 inhibits the TNF receptor associated factor 3 (Traf3) [bib_ref] The p90 ribosomal S6 kinase-UBR5 pathway controls Toll-like receptor signaling via miRNA-induced..., Cho [/bib_ref] , an inhibitor of the NF-κB pathway [bib_ref] TRAF3 and its biological function, He [/bib_ref]. Thus, the role of UBR5 might be indirect.
When Hyd or UBR5 was attenuated, only a subset of NF-κB target genes is expressed, diminishing the intensity of the innate immune response in Drosophila and inflammatory response in mammals, similarly to the inactivation of Akirin [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] [bib_ref] Akirin2 is critical for inducing inflammatory genes by bridging IkappaB-zeta and the..., Tartey [/bib_ref]. The link between excessive activation of NF-κB signaling pathways during e.g chronic inflammation and cancer progression or appearance is now on the spotlight [bib_ref] NF-kappaB, inflammation, immunity and cancer: coming of age, Taniguchi [/bib_ref]. Uncontrolled activation of NF-κB due to deregulation of ubiquitin-ligases has been reported in many diseases [bib_ref] Diverse roles of the ubiquitin system in NF-kappaB activation, Iwai [/bib_ref] and UBR5 has been described to be involved in several types of cancer in humans [bib_ref] Functional Roles of the E3 Ubiquitin Ligase UBR5 in Cancer, Shearer [/bib_ref]. Our findings point to the HECT E3-ubiquitin ligase UBR5 as an interesting drug target to modulate NF-κB signaling, control the development of inflammatory diseases and potentially improve treatments for cancer.
# Material and methods
## Cell culture
S2 cells were cultured at 25˚C in Schneider's medium (Biowest) supplemented with 10% fetal (vol/vol) calf serum (FCS), penicillin/streptomycin (50 μg/ml of each) and 2 mM glutamax. HeLa cell line (gift from IGBMC, Illkirch, France) was cultured in DMEM containing 10% (vol/vol) FCS, 40 μg/mL gentamycin. THP1 cell line (ATCC-TIB-202) was cultured in RPMI containing 10% (vol/vol) FCS and penicillin/streptomycin (50 μg/ml of each).
## Dsrna e3 ubiquitin ligases screen
A list containing 174 E3 ubiquitin ligases in the Drosophila genome, consisting predominantly of HECT, RING, and U-box proteins was curated manually by GO-and protein domain-term search in Flybase FB2012_06 Dmel Release 5.48 [bib_ref] FlyBase at 25: looking to the future, Gramates [/bib_ref]. Based on this list, a Drosophila E3 ligase dsRNA library was generated in Michael Boutros's laboratory [bib_ref] Design and evaluation of genome-wide libraries for RNA interference screens, Horn [/bib_ref] , listed in S2 [fig_ref] Table: Induction of Attacin-A after knockdown of the luciferase screen candidates inDrosophila S2... [/fig_ref] The screen experiments were performed using 1F3 cells stably expressing AttA firefly luciferase [bib_ref] Landscape of protein-protein interactions in Drosophila immune deficiency signaling during bacterial challenge, Fukuyama [/bib_ref]. Two days after transfection with an Actin renilla luciferase construct, cells were collected and distributed into 96-well screening plates at a density of 4.5 x 10 4 cells per well. Cells were transfected with 3 μg of each dsRNA in the Drosophila E3-ubiquitin ligase dsRNA library in triplicate by bathing method as previously described [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. At day 5 post-transfection, cells were stimulated for 48 h with heat-killed E. coli (40:1) before being lysed. Luciferase activity was quantified in a luminometer (Mithras LB940, Berthold) after addition of the substrate (Dual luciferase assay kit, Promega).
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response
## Rna interference
The double-strand RNAs for the knockdown experiments in Drosophila cells were prepared according to [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. Fragments for the different genes were generated from genomic DNA templates using oligonucleotides designed for use with Genome-RNAi libraries [bib_ref] GenomeRNAi: a database for cell-based and in vivo RNAi phenotypes, 2013 update, Schmidt [/bib_ref] and are listed in S3 [fig_ref] Table: Induction of Attacin-A after knockdown of the luciferase screen candidates inDrosophila S2... [/fig_ref] The small interfering RNAs used for the knockdown experiment in mammalian cell lines cells were purchased from Ambion (S4 .
## Plasmid constructs
pAC-Akirin, pAC-Akirin-V5, pAC-Pgrp-LC, pAC-Imd, pMT-Rel-HA, pMT-Bap60-HA and pAC-Gal4 constructs were described previously [bib_ref] Akirins are highly conserved nuclear proteins required for NF-kappaB-dependent gene expression in..., Goto [/bib_ref] [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. pUAS-Hyd-V5 and pUAS-Hyd-CS-V5 were kindly provided by Xinhua Lin laboratory [bib_ref] Hyperplastic discs differentially regulates the transcriptional outputs of hedgehog signaling, Wang [/bib_ref].
## Cell transfection
Drosophila S2 cells were transfected with double-strand RNAs using the bathing method described in [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] or with plasmids using the Effectene transfection kit (Qiagen). HeLa cells were transfected with siRNA using Lipofectamine RNAiMax Transfection Reagent (Invitrogen). THP1 cells were transfected with siRNA using Lipofectamine 3000 Transfection Reagent (Invitrogen).
## Rna extraction and quantification
For the in vitro experiments, RNA was extracted from cells and treated with DNAse, using RNA Spin kit (Macherey Nagel). For the in vivo experiments, the procedure was done accordingly to [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. Similarly, reverse-transcription and quantitative RT-PCR were performed as indicated in [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref]. The levels of expression of genes of interest were normalized against the measured level of the RNA coding determined in each sample for ribosomal protein-49 in the case of Drosophila experiments and for GAPDH for HeLa and THP1 cells. Primers used for quantitative RT-PCR are listed in S5 Table.
## Isolation of nuclear and cytoplasmic cell fractions
Nuclear/cytoplasmic fractionation was performed using the NE-PER Nuclear and Cytoplasmic Extraction Reagents kit (Thermo Scientific), following the manufacturer's instructions. The cell pellet was suspended in 500 μl of cytoplasmic extraction reagent I. After a 10 min incubation on ice, 11 μl of cytoplasmic extraction reagent II was added. The supernatant fraction was then collected after a 5 min centrifugation at 16 000 g. The nuclear fraction was collected following the suspension of the nuclear pellet in 250 μl of Nuclear extraction reagent, incubation for 40min on ice and centrifugation for 10 min at 16 000g.
## Immunoprecipitation and western blot
Cells were treated for the indicated times with heat-killed E. coli (40:1) at 25˚C. The cells were harvested, washed in PBS and lysed in 200 μl of Pierce IP Lysis Buffer (Thermo Scientific), with protease inhibitor cocktail (Roche). Immunoprecipitations were performed overnight at 4˚C based on [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] [bib_ref] Optimising methods for the preservation, capture and identification of ubiquitin chains and..., Emmerich [/bib_ref] , with either rabbit polyclonal anti-Akirin [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] or anti-ubiquitin Lys63 specific antibodies (Millipore 05-1308), or anti-V5 (Merck V8137) coupled with Dynabeads Protein G (Invitrogen) and anti-V5 or anti-HA antibodies coupled to agarose beads (Sigma). Proteins were detected by Western blotting using anti-Akirin, anti-ubiquitin Lys63 specific, anti-ubiquitin Lys48 specific (Merck 05-1307), anti-Ubiquitin (SantaCruz Biotechnology
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response P4D1), anti-H3 histone (Abcam ab1791), anti-RpS15 (Abcam ab157193), anti-V5 (Invitrogen r96025) and anti-HA (Abcam ab9110) antibodies.
## Fly strains
Stocks were raised on standard cornmeal-yeast-agar medium at 25˚C with 60% humidity. To generate conditional knockdown in adult flies, we used the GAL4-GAL80 ts system [bib_ref] Gene expression systems in Drosophila: a synthesis of time and space, Mcguire [/bib_ref]. Fly lines carrying a UAS-RNAi transgene targeting relish (108469), akirin (109671), and hyd (44675, named here Hyd1; 44676, named here Hyd2) were obtained from the Vienna Drosophila RNAi Center (http://stockcenter.vdrc.at/control/main). Fly line carrying a UAS-RNAi transgene against GFP (397-05) was obtained from the Drosophila Genetic Resource Center (Kyoto, Japan; http://www.dgrc.kit.ac.jp/index.html). UAS-RNAi flies were crossed with Actin-GAL4/ CyO; Tub-GAL80ts flies at 18˚C. Emerged adult flies were then transferred to 29˚C to activate the UAS-GAL4 system for 6-7 days. Nine-day-old flies were used-three batches of twenty females for survival assays and ten males for quantitative RT-PCR experiments.
## Immune challenge
Drosophila S2 cells were stimulated with heat-killed E. coli (40:1) [bib_ref] The Drosophila serpins: multiple functions in immunity and morphogenesis, Reichhart [/bib_ref]. HeLa cells were stimulated with recombinant human IL-1β (10 ng/ml) for 4h. THP1 cells were stimulated with Lipopolysaccharide (LPS) Solution (1 μg/ml) for 4h. IL-1β and LPS were purchased from Invitrogen. Microbial challenges were performed by pricking adult flies with a sharpened tungsten needle dipped into PBS or concentrated E. coli strain DH5aGFP bacteria solution at 25˚C, for either several days (for survival assays) or 6h (for quantitative RT-PCR experiments) [bib_ref] Akirin specifies NF-kappaB selectivity of Drosophila innate immune response via chromatin remodeling, Bonnay [/bib_ref] [bib_ref] The Drosophila serpins: multiple functions in immunity and morphogenesis, Reichhart [/bib_ref]. Bacteria were grown in Luria broth (LB) at 37˚C.
## Bioinformatic analysis of hyd and ubr5
Amino acid sequences of Drosophila melanogaster HyD (ID: P51592) and Homo sapiens UBR5 (ID: O95071) were retrieved from UniProt (uniport.org). Predicted functional domains were annotated using the InterPro database (v77.0) from the European Molecular Biology Laboratory [bib_ref] InterPro in 2019: improving coverage, classification and access to protein sequence annotations, Mitchell [/bib_ref]. In order to calculate identity and similarity percentages, sequences were aligned to each other using the EMBOSS NEEDLE tool [bib_ref] A general method applicable to the search for similarities in the amino..., Needleman [/bib_ref] , also from the European Molecular Biology Laboratory, under default settings. Graphical visualization of the similarity percentage between HyD and UBR5 alongside the amino acid sequence was performed with the LALNVIEW software [bib_ref] LALNVIEW: a graphical viewer for pairwise sequence alignments, Duret [/bib_ref] after sequence alignment with the SIM-Local similarity program [bib_ref] A time-efficient, linear-space local similarity algorithm, Huang [/bib_ref] under default settings.
# Statistical analysis
P values for quantitative RT-PCR were calculated using the two-tailed unpaired Student t test using GraphPad Prism version 6.0c for Mac, GraphPad Software, San Diego, California USA. Log-rank analyses of survival assay were performed using RStudio version 1.
## Plos pathogens
The HECT E3 ubiquitin-ligase Hyd is required for Drosophila immune response
[fig] Fig 1: E3-ubiquitin ligases screen identified in vitro Hyd as involved in IMD pathway. (A) Induction of the IMD pathway measured by luciferase activity. 174 [/fig]
[fig] Fig 2: Hyd is required for the full activation of IMD response. (A) Quantitative RT-PCR of Attacin-A, Attacin-C, Cecropin-A1, Attacin-D, Pgrp-LB and Pirk mRNA from HKE-stimulated S2 cells transfected with dsRNA targeting GFP (negative control), Relish or Akirin (positive controls), and Hyd. (B) In vivo survival experiments performed on C564-Gal4/UAS-RNAi females Drosophila. They were infected with E. coli by septic injury (with PBS pricking as control). (C) Quantitative RT-PCR of Attacin-A and Attacin-D mRNA from C564-Gal4/UAS-RNAi Drosophila males infected with E. coli by septic injury. Data are represented as mean ± standard deviation of three independent experiments (A-C).After the ratio of stimulated over unstimulated values for each condition was determined, statistical significance was established by comparing genes knockdown with GFP dsRNA or RNAi control (B-C). � Pvalue < 0.05; �� P-value < 0.01; ��� P-value < 0.001.https://doi.org/10.1371/journal.ppat.1008458.g002 [/fig]
[fig] Fig 3: Fig 3. Hyd mediated-ubiquitination of Akirin is necessary for interaction with Relish. (A) Co-immunoprecipitation assay between over-expressed Akirin and Hyd in S2 cells. The cells were transiently transfected with Akirin, Gal4, Hyd-V5 and/or Hyd-CS-V5 expressing plasmids. Cell lysates were immunoprecipitated with anti-Akirin coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-V5 or anti-Akirin antibodies. (B) Immunoprecipitation assay of K63-polyUb chains on Akirin before and after immune challenge. S2 cells were transiently transfected with Akirin-V5 expressing plasmid. Cell lysates were immunoprecipitated with anti-K63-polyUb coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-V5 antibodies. (C) Immunoprecipitation assay of Akirin after immune challenge. S2 cells were transiently transfected with Akirin-V5 expressing plasmid and dsRNA targeting GFP or Hyd. Cell lysates were immunoprecipitated with anti-V5 coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-K63-polyUb and anti-V5 antibodies. (D) Co-immunoprecipitation assay between over-expressed Akirin and Relish or Bap60 in S2 cells. The cells were transiently transfected with Akirin-V5 and Rel-HA or Bap60-HA expressing plasmids and dsRNA targeting GFP or Hyd. Cell lysates were immunoprecipitated with anti-V5 coupled agarose beads. Immunoprecipitates were analyzed by Western blotting with anti-HA or anti-V5 antibodies. Data are representative of 2 independent experiments. [/fig]
[fig] Fig 4: Hyd/UBR5 is necessary for NF-κB target genes activation. (A) Graphical representation of the predicted functional domains of Drosophila melanogaster Hyd and Homo sapiens UBR5. This representation highlights a conserved Ubiquitin-associated (UBA) domain, poly(A)-binding protein C-terminal (PABC) domain, Homologous to the E6-AP Carboxyl Terminus (HECT) domain and UBR type Zinc Finger (UBR ZF) domain. Annotation based InterPro software. Sequence similarity percentages calculated using the EMBOSS NEEDLE tool (default settings). (B) Quantitative RT-PCR of IL6, Ifit1, IL12β, IL8, Il1α and TNF mRNA from LPS-stimulated THP1 cells. They were transfected with scrambled siRNA (negative control) or siRNA targeting NF-κB1, AKIRIN2 (positive controls) or UBR5. Data are represented as mean ± standard deviation of three independent experiments. After the ratio of stimulated over unstimulated values for each condition was determined, statistical significance was established by comparing genes knockdown with scrambled siRNA control. � P-value < 0.05; �� P-value < 0.01; ��� P-value < 0.001. https://doi.org/10.1371/journal.ppat.1008458.g004 [/fig]
[fig] Fig 5: Schematic model of Hyd involvement in the IMD signaling pathway in Drosophila.Model showing the role of Hyd in the expression of the Akirin-dependent genes in the IMD pathway. After activation of the pathway, allowed by the K63-polyUb chains deposition on the complexes IMD and DREDD by the E3-ubiquitin ligase Diap2, Relish is translocated. Hyd is necessary for the K63-polyUb of the NF-κB co-factor Akirin and its interaction with the NF-κB factor Relish. This interaction is crucial for the expression of Akirin-dependent genes (like Attacin-A and Attacin-C), necessary for an adequate innate immune response. DAP-PGN: mesodiaminopimelic acid-type peptidoglycan of Gram-negative bacteria, P-tag: phosphorylation marks.https://doi.org/10.1371/journal.ppat.1008458.g005 [/fig]
[fig] S3: Fig. Interaction assay between Diap2 and Akirin. (DOCX) S4 Fig. Ubiquitinated Akirin accumulates in nuclear cell fraction after immune challenge. (DOCX) S5 Fig. Akirin is K48-polyubiquitinated independently of Hyd. (DOCX) S6 Fig. Kinetic of immune induced genes at different hours after stimulation. (DOCX) S7 Fig. Drosophila HyD and Human UBR5 share a high sequence similarity. (DOCX) S8 Fig. Knockdown efficiency in HeLa cells of the small interfering RNA used in mammalian cell lines. (DOCX) S9 Fig. Hyd/Ubr5 is necessary for activation of a subset of NF-κB target genes in HeLa cells. (DOCX) S1 [/fig]
[table] Table: Induction of Attacin-A after knockdown of the luciferase screen candidates inDrosophila S2 cells. (DOCX) S2Table. Double strand RNA sequences used in the luciferase screen in Drosophila S2 cells. (XLSX) S3 Table. Oligonucleotides used to generate double strand RNA in Drosophila S2 cells. (DOCX) S4 Table. References of small interfering RNA used in mammalian HeLa cells. (DOCX) S5 Table. Oligonucleotides used for quantitative real-time PCR. (DOCX) [/table]
|
10.1155/2015/285919
|
CCBY
|
4588348
|
26451389
|
s2orc_pubmed_articles
|
Clinical Outcomes in Men and Women following Total Knee Arthroplasty with a High-Flex Knee: No Clinical Effect of Gender
While it is generally recognized that anatomical differences exist between the male and female knee, the literature generally refutes the clinical need for gender-specific total knee prostheses. It has been found that standard, unisex knees perform as well, or better, in women than men. Recently, high-flex knees have become available that mechanically accommodate increased flexion yet no studies have directly compared the outcomes of these devices in men and women to see if gender-based differences exist. We retrospectively compared the performance of the high-flex Vanguard knee (Biomet, Warsaw, IN) in 716 male and 1,069 female knees. Kaplan-Meier survivorship was 98.5% at 5.6-5.7 years for both genders. After 2 years, mean improvements in Knee Society Knee and Function scores for men and women (50.9 versus 46.3; 26.5 versus 23.1) and corresponding SF-12 Mental and Physical scores (0.2 versus 2.2; 13.7 versus 12.2) were similar with differences not clinically relevant. Postoperative motion gains as a function of preoperative motion level were virtually identical in men and women. This further confirms the suitability of unisex total knee prostheses for both men and women.
# Introduction
Morphometric differences exist between the male and female knee populations. Such differences include smaller size [bib_ref] Anatomy revisited: morphometry in the light of sex-specific total knee arthroplasty, Dargel [/bib_ref] [bib_ref] Gender-specific design in total knee arthroplasty, Greene [/bib_ref] , larger Q-angle [bib_ref] Anatomy revisited: morphometry in the light of sex-specific total knee arthroplasty, Dargel [/bib_ref] [bib_ref] The female knee: anatomic variations, Conley [/bib_ref] [bib_ref] Parameters and comparisons of the quadriceps angle of college-aged men and women..., Woodland [/bib_ref] , smaller observable prominence of the anterior condyle [bib_ref] Anatomy revisited: morphometry in the light of sex-specific total knee arthroplasty, Dargel [/bib_ref] , smaller medial-lateral (ML) to anterior-posterior (AP) femoral condyle aspect ratio [bib_ref] Anatomy revisited: morphometry in the light of sex-specific total knee arthroplasty, Dargel [/bib_ref] [bib_ref] Gender-specific design in total knee arthroplasty, Greene [/bib_ref] [bib_ref] The female knee: anatomic variations, Conley [/bib_ref] [bib_ref] Gender differences in distal femoral morphology and the role of gender specific..., Guy [/bib_ref] , and thinner patella [bib_ref] Anthropometric measurements of the human knee: correlation to the sizing of current..., Hitt [/bib_ref] in female knees compared to male knees. This has led to the thought that standard total knee implants in women may have a tendency to overstuff the patellofemoral compartment leading to a reduced range of motion and increased overhang with subsequent lateral and medial knee pain due to soft tissue irritation [bib_ref] Will genderspecific total knee arthroplasty be a better choice for women? A..., Xie [/bib_ref]. Genderspecific knee designs attempt to address these concerns through design modifications to better accommodate the female femoral condyle such as modifying the anterior flange to include a recessed sulcus and reduced anterior condyle height, reducing the ML : AP aspect ratio, and increasing the angle of the trochlear groove [bib_ref] The female knee: anatomic variations, Conley [/bib_ref] [bib_ref] Will genderspecific total knee arthroplasty be a better choice for women? A..., Xie [/bib_ref]. The most studied genderspecific knee system is the Zimmer Gender Solutions NexGen Knee (Zimmer Inc., Warsaw, IN). Such investigations have included unilateral [bib_ref] Early results of gender-specific posterior stabilized total knee arthroplasty without patella resurfacing, Roth [/bib_ref] [bib_ref] Hi-flexion and gender-specific designs fail to provide significant increases in range of..., Song [/bib_ref] and bilateral [bib_ref] Hi-flexion and gender-specific designs fail to provide significant increases in range of..., Song [/bib_ref] [bib_ref] Do we need a gender-specific total knee replacement? A randomised controlled trial..., Thomsen [/bib_ref] [bib_ref] Comparison of standard and gender-specific posterior-cruciate-retaining high-flexion total knee replacements: a prospective,..., Kim [/bib_ref] [bib_ref] Comparison of outcomes after bilateral simultaneous total knee arthroplasty using gender-specific and..., Song [/bib_ref] studies in women as well as Indian [bib_ref] Gender-specific high-flexion knee prosthesis in Indian women: a prospective randomised study, Singh [/bib_ref] and Tai [bib_ref] The early results of gender-specific total knee arthroplasty in Thai patients, Tanavalee [/bib_ref] patients. The literature, however, provides little support for gender-specific prostheses. Following a systematic review, Merchant et al. [bib_ref] The female knee: anatomic variations and the female-specific total knee design, Merchant [/bib_ref] concluded that the apparent anatomic differences between male and female knees were due to the smaller height and size of women and not due to gender, per se. Their review also indicated that rather than women having poorer total knee arthroplasty (TKA) results than men using a standard prosthesis, female outcomes are actually as good or better. A systematic review and meta-analysis by Xie et al. [bib_ref] Will genderspecific total knee arthroplasty be a better choice for women? A..., Xie [/bib_ref] also found no evidence to support the need for gender-specific knees. In addition to studies showing that outcomes are similar between the sexes with standard knees [bib_ref] The female knee: anatomic variations and the female-specific total knee design, Merchant [/bib_ref] [bib_ref] Analysis of the outcome in male and female patients using a unisex..., Dalury [/bib_ref] [bib_ref] The John Insall award: gender-specific total knee replacement: prospectively collected clinical outcomes, Macdonald [/bib_ref] [bib_ref] Gender influence on the outcome of an unisex total knee arthroplasty system, Muenzberg [/bib_ref] [bib_ref] The clinical effect of gender on outcome of total knee arthroplasty, Ritter [/bib_ref] , it has also been shown that outcomes are not substantially different in women whether they receive a standard or gender-specific knee [bib_ref] Early results of gender-specific posterior stabilized total knee arthroplasty without patella resurfacing, Roth [/bib_ref] [bib_ref] Do we need a gender-specific total knee replacement? A randomised controlled trial..., Thomsen [/bib_ref] [bib_ref] Comparison of standard and gender-specific posterior-cruciate-retaining high-flexion total knee replacements: a prospective,..., Kim [/bib_ref] [bib_ref] Comparison of outcomes after bilateral simultaneous total knee arthroplasty using gender-specific and..., Song [/bib_ref] [bib_ref] Gender-specific high-flexion knee prosthesis in Indian women: a prospective randomised study, Singh [/bib_ref].
Recently, there has been interest in high-flex knees [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] [bib_ref] High flexion total knee arthroplastymid-term follow up of 5 years, Endres [/bib_ref] [bib_ref] Three-to six-year follow-up results after high-flexion total knee arthroplasty: can we allow..., Cho [/bib_ref] , that is, those that can mechanically accommodate flexion 2
The Scientific World Journal in excess of 125 ∘ (ASTM F2083); however their ability to equally serve males and females has not been established. Only one small intraoperative study (40 patients) [bib_ref] Hi-flexion and gender-specific designs fail to provide significant increases in range of..., Song [/bib_ref] and a few short-term small cohort (<50 patients) [bib_ref] Do we need a gender-specific total knee replacement? A randomised controlled trial..., Thomsen [/bib_ref] [bib_ref] Comparison of outcomes after bilateral simultaneous total knee arthroplasty using gender-specific and..., Song [/bib_ref] or intermediate cohort (up to 138 patients) [bib_ref] Early results of gender-specific posterior stabilized total knee arthroplasty without patella resurfacing, Roth [/bib_ref] [bib_ref] Comparison of standard and gender-specific posterior-cruciate-retaining high-flexion total knee replacements: a prospective,..., Kim [/bib_ref] [bib_ref] Gender-specific high-flexion knee prosthesis in Indian women: a prospective randomised study, Singh [/bib_ref] studies have compared a high-flex knee with a gender-specific knee, finding no clear advantage of the latter. The flexion achieved by a given patient will be dependent upon the amount that can be accommodated by the design of the knee and anatomical features of the patient including soft tissue restraint that is often a limiting factor. In the case of high-flex knees, the limiting effect of patient anatomy may be more evident than would be the case for standard knees.
The purpose of this study was to compare the midterm functional outcomes and survivorship of large male and female cohorts receiving the same cruciate-retaining (CR) high-flex knee. Our hypothesis was that there would be no difference in these metrics between the genders with this prosthesis.
# Materials and methods
The Vanguard knee (Biomet, Warsaw, IN) is a high-flex design as it can mechanically accommodate up to 145 ∘ of flexion although that achieved clinically may be less due to the soft tissue restraint of the patient [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] [bib_ref] The relationship between pre-and postoperative range of motion utilizing three cruciate-retaining total..., Stormont [/bib_ref]. From September 2004 to April 2013, a consecutive series of 1,328 patients received 1,785 cruciate-retaining (CR) knees of this type via a medial parapatellar approach using cemented fixation, including 802 women (1,069 knees) and 526 men (716 knees). All patellae were resurfaced. Standard overlay templating was used to determine proper implant size until 2011, after which digital templating was employed. There are ten nominal sizes of femoral components, that is, 55-75 in increments of 2.5, plus 80, and nine nominal tibial component sizes, that is, 59-91 in increments of 4 in this system. All femoral and tibial component sizes are compatible with each other throughout their respective ranges. The tibial component was sized for best fit on the cut surface without AP or ML overhang. The femoral component was sized using the posterior referencing femoral sizer that was part of the system. All surgeries were performed by the senior author at a single site following Institutional Review Board approval and signed consent from each patient. Mean follow-up for the male and female patients was 2.4 years (range: 0.8-6.9 years) and 2.4 years (range: 0.7-7.2 years), respectively.
# Survivorship analysis.
Kaplan-Meier survivorship analysis was performed for each gender (716 male knees and 1,069 female knees), including 95% confidence limits, with the endpoint defined as revision of any component for any reason. Final survivorship intervals were chosen to correspond to those at which 20 knees remained at risk to avoid the instability that can result when the remaining population becomes too small [bib_ref] Survivorship analysis in the evaluation of joint replacement, Dorey [/bib_ref]. This final interval was 5.6 years for men and 5.7 years for women.
# Functional analysis.
Functional analysis was performed on only those knees with a minimum of 2 years of complete clinical follow-up. Clinical assessment consisted of preop and final postop Knee Society Score (KSS) [bib_ref] Rationale of the knee society clinical rating system, Insall [/bib_ref] and SF-12 [bib_ref] A shorter form health survey: can the SF-12 replicate results from the..., Jenkinson [/bib_ref]. Three hundred thirteen male patients (397 knees) and 463 female patients (574 knees) had complete preop and minimum 2-year postop clinical assessment, with mean final follow-up of 2.9 years (range: 2.0-6.9 years) and 2.9 years (range: 2.0-7.2 years), respectively.
Preoperative and ≥2-year postoperative passive range of motion (ROM) and passive peak flexion (PF) data was available for 462 male and 671 female knees. Motion results were stratified by preoperative motion range, that is, <95 ∘ , 95-105 ∘ , and >105 ∘ .
# Statistical analysis.
Interval data means between the genders (patient age, body mass index, length-of-stay, length of follow-up, KSS, SF12, ROM, and PF) were compared using the pooled -test. Preop to postop changes in outcomes for a given gender were compared by the paired -test. Nominal data mean differences (proportion of right knees and distribution of primary diagnoses) were compared by the Chi-square test. A value of < 0.05 was chosen for statistical significance. [fig_ref] Table 1: Patient demographics [/fig_ref] summarizes the patient demographics for the total cohorts as well as those limited to a minimum of 2-year complete follow-up. No significant gender difference was seen in primary diagnosis ( = 0.409), proportion of right knees ( = 0.271), or length of follow-up ( = 1.0). However, the differences in patient age ( = 0.032), body mass index ( = 0.009), and length-of-stay ( < 0.001) for males compared to females were significant. A total of 37 patients died for reasons unrelated to their knee procedure (13 men with 18 knees and 15 women with 19 knees) with all implants in place at the time of death. [fig_ref] Table 2: Knee Society Score and SF-12 outcomes summaries after a minimum of 2... [/fig_ref] presents the KSS and SF-12 results. Preoperatively, men and women had similar SF-12 Physical scores while women had greater KS Knee scores and men had greater KS Function and SF-12 Mental scores. Postoperatively, men and women had similar KS Knee scores, with men having greater KS Function and SF-12 (both components) scores. For both men and women there was a significant preop to postop increase ( < 0.0001) in both components of the KSS and the SF-12, with the exception of the SF-12 Mental score for men. [fig_ref] Table 3: Comparison of preop to postop score changes [/fig_ref] lists the means of the preop to postop differences (Δscores) for the men and women. While the gender differences were not pronounced, they did reach statistical significance, with greater KS Knee and Function and SF-12 Physical Δscores for men and greater SF-12 Mental Δscore for women.
# Results
Tables 4 and 5 stratify the male and female ROM and PF results, respectively, by preop motion with two principal observations apparent. First, the improvement in motion (ΔROM and ΔPF) was inversely related to the preop motion of the patient, that is, knees with less motion prior to surgery tended to achieve a greater increase after surgery than did knees initially presenting with a high degree of motion. Second, there were no significant differences in the pre-to The Scientific World Journal 3 postop change in motion between men and women with the exception of a 2-degree PF differential in favor of men for the mid-functioning preop group. The same proportion of male and female knees (81-83%) achieved ≥120 ∘ of ROM and PF with no significant differences between the genders (ROM:
# Discussion
There is little doubt that morphometric differences exist between female and male knees [bib_ref] Anatomy revisited: morphometry in the light of sex-specific total knee arthroplasty, Dargel [/bib_ref] [bib_ref] Gender-specific design in total knee arthroplasty, Greene [/bib_ref] [bib_ref] The female knee: anatomic variations, Conley [/bib_ref] [bib_ref] Parameters and comparisons of the quadriceps angle of college-aged men and women..., Woodland [/bib_ref] [bib_ref] Gender differences in distal femoral morphology and the role of gender specific..., Guy [/bib_ref] [bib_ref] Anthropometric measurements of the human knee: correlation to the sizing of current..., Hitt [/bib_ref] [bib_ref] Intraoperative measurements of male and female distal femurs during primary total knee..., Chin [/bib_ref]. This has led to the hypothesis that standard (unisex) knees designed without regard to gender differences could produce inferior outcomes in women and that a gender-specific knee would be required to address this issue [bib_ref] The female knee: anatomic variations, Conley [/bib_ref] [bib_ref] Will genderspecific total knee arthroplasty be a better choice for women? A..., Xie [/bib_ref]. Systematic reviews and metaanalysis of the literature, however, suggest the opposite; that is, women obtain equivalent, if not better, outcomes than men using standard knees [bib_ref] Will genderspecific total knee arthroplasty be a better choice for women? A..., Xie [/bib_ref] [bib_ref] The female knee: anatomic variations and the female-specific total knee design, Merchant [/bib_ref]. Other studies that collectively compared 8,700 female knees with 5,927 male knees, both receiving standard implants, came to the same conclusion [bib_ref] Analysis of the outcome in male and female patients using a unisex..., Dalury [/bib_ref] [bib_ref] The John Insall award: gender-specific total knee replacement: prospectively collected clinical outcomes, Macdonald [/bib_ref] [bib_ref] Gender influence on the outcome of an unisex total knee arthroplasty system, Muenzberg [/bib_ref] [bib_ref] The clinical effect of gender on outcome of total knee arthroplasty, Ritter [/bib_ref]. In recent years, high-flex knees have been developed to provide increased flexion potential which is especially useful for high-demand patients [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] [bib_ref] High flexion total knee arthroplastymid-term follow up of 5 years, Endres [/bib_ref] [bib_ref] Three-to six-year follow-up results after high-flexion total knee arthroplasty: can we allow..., Cho [/bib_ref]. It is possible that gender-related differences in outcomes might become manifest with the use of such high-performance 4 The Scientific World Journal prostheses since such devices could potentially intensify the influence of knee morphology on clinical performance. Four studies collectively examined 308 bilateral female patients, each having one knee replaced with a genderspecific high-flex knee and the other knee receiving a unisex knee (high-flex or non-high-flex) [bib_ref] Do we need a gender-specific total knee replacement? A randomised controlled trial..., Thomsen [/bib_ref] [bib_ref] Comparison of standard and gender-specific posterior-cruciate-retaining high-flexion total knee replacements: a prospective,..., Kim [/bib_ref] [bib_ref] Comparison of outcomes after bilateral simultaneous total knee arthroplasty using gender-specific and..., Song [/bib_ref] [bib_ref] Gender-specific high-flexion knee prosthesis in Indian women: a prospective randomised study, Singh [/bib_ref]. While there was no apparent advantage of the gender-specific high-flex design in women, these studies did not directly address the need for gender-specific high-flex knees since no male patients were included with which to compare outcomes. Ours was the first study to compare the use of a unisex high-flex knee in both male and female cohorts. The high-flex CR knee design used in this study was chosen to address this issue, in part, because there have been recent large-cohort, midterm studies published on this system to provide a baseline to which our results may be compared [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] [bib_ref] Early experience with the Vanguard complete total knee system: 2-7 years of..., Kievit [/bib_ref].
[formula] b Postop ROM ( ∘ ) a ΔROM ( ∘ ) a,b Postop ROM ( ∘ ) a ΔROM ( ∘ ) a,bb Postop PF ( ∘ ) a ΔPF ( ∘ ) a,b Postop PF ( ∘ ) a ΔPF ( ∘ ) a,b [/formula]
In our study, the total population of 1,785 high-flex cruciate-retaining knees resulted in only seven revisions. Stratifying survivorship by gender yielded Kaplan-Meier survivorship estimates of 98.5% for both men and women at 5.6 years and 5.7 years, respectively. Other investigators have found similar Kaplan-Meier survivorship for the same knee, that is, 97.8% at 7.0 years reported by Schroer et al. [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] (957 knees; 85.0% posterior-stabilized, 15.0% cruciate-retaining; 36.5% male; 63.5% female) and 98.6% at 6.0 years reported by Kievit et al. [bib_ref] Early experience with the Vanguard complete total knee system: 2-7 years of..., Kievit [/bib_ref] (807 knees; 51.3% posterior-stabilized, 48.7% cruciate-retaining; 35.8% male, 64.2% female).
With the exception of the KS Knee score, the preoperative condition of the male knees was better than that of the female knees, which was significant ( < 0.01) for all but the physical component of the SF-12. Others have reported similar findings likely because women tend to present later for surgery, with lower function and pain scores than men [bib_ref] Analysis of the outcome in male and female patients using a unisex..., Dalury [/bib_ref] [bib_ref] The John Insall award: gender-specific total knee replacement: prospectively collected clinical outcomes, Macdonald [/bib_ref] [bib_ref] The clinical effect of gender on outcome of total knee arthroplasty, Ritter [/bib_ref] [bib_ref] Women recover faster than men after standard knee arthroplasty, Liebs [/bib_ref]. Both men and women showed significant increases ( < 0.0001) in both components of the KSS and SF-12 with the exception of the SF-12 Mental component for males. Schroer et al. [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] observed similar improvement in KSS for a combined male-female population receiving this knee.
As regards the KSS and SF-12 outcome comparisons between the men and women, it is best to compare the Δscores since the preop scores were different. The Δscores were significantly greater for males than females for KS Knee score (50.9 versus 46.3, = 0.002), KS Function score (26.5 versus 23.1, = 0.026), and the SF-12 Physical score (13.7 versus 12.2, = 0.031), while females were ahead with the SF-12 Mental score (2.2 versus 0.2, = 0.003). Consequently, the statistical comparisons of the score improvements did not show a consistent advantage for either gender.
In general, the change in motion following TKA is inversely related to the preoperative value, that is, patients presenting with restricted motion tend to gain much motion after surgery while those with a high degree of motion initially tend to stay about the same, or perhaps lose a small amount of motion, after surgery [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] [bib_ref] The relationship between pre-and postoperative range of motion utilizing three cruciate-retaining total..., Stormont [/bib_ref]. While this "regression toward the mean" is well documented, a potential gender effect has not been previously investigated. When stratifying motion outcomes by their preop values, that is, <95 ∘ , 95 ∘ -105 ∘ , and >105 ∘ , we found virtually identical ΔROM and ΔPF values for both men and women, that is, ΔROM: 32 ∘ -34 ∘ , 18 ∘ , and 4 ∘ and ΔPF: 23 ∘ -24 ∘ , 15 ∘ -17 ∘ , and 4 ∘ , respectively. As such, no gender influence on motion was apparent in general or as a function of preop motion. By way of comparison, Schroer at al. [bib_ref] Sevenyear survivorship and functional outcomes of the high-flexion Vanguard complete knee system, Schroer [/bib_ref] documented similar Δ PF in a combined male-female population (two-thirds female) receiving the same knee (627 knees; 509 posterior-stabilized and 118 cruciate-retaining) of 23.6 ∘ , 19.3 ∘ , and 1.8 ∘ in these three preop categories, respectively.
Taken in aggregate, the survivorship, KSS, SF-12, ROM, and PF values suggest similar performance, overall, of this CR high-flex knee in both men and women. While there were some statistically significant differences in some outcomes between the genders, these differences were small and were of the magnitude obtained by others who did not ascribe clinical relevancy to them [bib_ref] Do we need a gender-specific total knee replacement? A randomised controlled trial..., Thomsen [/bib_ref] [bib_ref] Comparison of standard and gender-specific posterior-cruciate-retaining high-flexion total knee replacements: a prospective,..., Kim [/bib_ref] [bib_ref] Comparison of outcomes after bilateral simultaneous total knee arthroplasty using gender-specific and..., Song [/bib_ref] [bib_ref] Gender-specific high-flexion knee prosthesis in Indian women: a prospective randomised study, Singh [/bib_ref] [bib_ref] Analysis of the outcome in male and female patients using a unisex..., Dalury [/bib_ref] [bib_ref] The John Insall award: gender-specific total knee replacement: prospectively collected clinical outcomes, Macdonald [/bib_ref] [bib_ref] Gender influence on the outcome of an unisex total knee arthroplasty system, Muenzberg [/bib_ref] [bib_ref] The clinical effect of gender on outcome of total knee arthroplasty, Ritter [/bib_ref].
Gender considerations notwithstanding, current trends in high-flex knee design, include reducing the posterior radius of curvature which increases the contact area between the posterior femoral condyle and the tibial insert [bib_ref] Design may be counterproductive for optimizing flexion after TKR, Ranawat [/bib_ref]. This increase in contact area, however, may not be sufficient to effectively distribute the high forces developed during deep squatting which can potentially lead to increased polyethylene wear. Extreme flexion may increase patellofemoral joint stress and disrupt patellar-trochlear groove congruity, leading to other complications such as pain, patellar fracture, and patellar loosening [bib_ref] Design may be counterproductive for optimizing flexion after TKR, Ranawat [/bib_ref]. In our study of 1,785 high-flex knees, a total of seven (0.39%) revisions were performed, including three (0.17%) for infection, three (0.17%) for aseptic loosening, and one (0.06%) for dislocation. As such, no particular sequelae associated with the high-flex knee design were evident.
So what does this all mean? First, the literature has generally concluded that standard, unisex knee designs are equally suitable for both men and women. Second, our study helped fill a void in the literature by comparing the same unisex highflex knee design in both men and women, thereby extending the results of others while reaching the same conclusions. Third, the complete compatibility of the entire range of femoral and tibial component sizes of the studied knee with each other may have allowed sufficient latitude to address patient needs regardless of gender. In other words, knee gender differences can be largely addressed through implant size rather than implant design considerations.
There were limitations in our study that should be considered. First, this was a retrospective study so there may be inherent biases that could have influenced the results. Despite this, the knee populations were large which may have partially mitigated this limitation. Second, only one type of high-flex knee was studied. As such, these results cannot be directly extended to other high-flex knee designs. Third, only midterm survivorships were reported. Long-term survivorships of at least 10 years will be required to fully document genderrelated outcomes differences that may exist with this knee design.
In summary, we found this knee to be highly effective in both men and women as evidenced by significant improvement in KSS and SF-12, similar ROM and PF outcomes, and high mid-term survivorship of 98.5%, confirming our study hypothesis.
# Conclusion
It is important that surgeons have the necessary information available to make an informed decision about treatment options for their patients. This is particularly true in joint replacement where there are a plethora of implant types and design philosophies. The contention that women have inferior outcomes following standard TKA, which is the raison d'être for gender-specific designs, has been dispelled by the literature. We have further solidified and extended this position by showing that one particular high-flex knee design had comparable clinical effectiveness in men and women.
## Conflict of interests
Dr. Nassif is a consultant to Biomet Inc. Dr. Pietrzak has no competing interests to disclose.
[fig] = 0 605 ,: PF: = 0.423). A total of seven revisions were performed including four male knees at 0.65 years (aseptic loosening), 0.73 years (infection), 2.01 years (infection), and 2.13 years (aseptic loosening), and three female knees at 2.51 years (infection), 4.18 years (aseptic loosening), and 5.25 years (dislocation). All components were replaced in four revisions, only the tibial component and liner in two revisions, and the femoral component, tibial component, and liner in one revision.Kaplan-Meier survivorship was identical in both cohorts, that is, 98.5% (95% CI: 97.8-99.2%) at 5.6 years for males and 98.5% (95% CI: 96.5-100%) at 5.7 years for females. [/fig]
[table] Table 1: Patient demographics. [/table]
[table] Table 2: Knee Society Score and SF-12 outcomes summaries after a minimum of 2 years. [/table]
[table] Table 3: Comparison of preop to postop score changes (Δscores). [/table]
[table] Table 4: Range of motion (ROM) comparisons after a minimum of 2 years. [/table]
[table] Table 5: Peak flexion (PF) comparisons after a minimum of 2 years. [/table]
|
10.2903/j.efsa.2017.4899
|
CCBYND
|
7009939
|
32625560
|
s2orc_pubmed_articles
| |
10.1186/1475-2840-12-160
|
CCBY
|
4176757
|
24188631
|
s2orc_pubmed_articles
|
Effects of combination therapy with vildagliptin and valsartan in a mouse model of type 2 diabetes
Background: Dipeptidyl peptidase-4 (DPP-4) inhibitors modulate incretin hormones and exert anti-diabetic effects in type 2 diabetes mellitus. Treatment with angiotensin II type 1 receptor blockers (ARB) is a proven successful intervention for hypertension with type 2 diabetes. The present study investigated the combined effects of the DPP-4 inhibitor vildagliptin and the ARB valsartan in a mouse model of type 2 diabetes.Methods: C57BL/6 J mice fed with high-fat diet (HFD) or db/db mice were treated with placebo, phloridzin (PHZ), vildagliptin alone (ViL), valsartan alone (VaL) or ViL with VaL (ViLVaL) for 8 weeks.Results: Glucose metabolism was improved in response to PHZ, ViL and ViLVaL in both HFD and db/db mice. Upon glucose challenge, ViLVaL showed the greatest suppression of blood glucose excursions, with increased insulin secretion, in db/db mice. ViLVaL treatment also showed an improvement of insulin sensitivity in db/db mice. Serum inflammatory cytokines were significantly decreased, and adiponectin was highest, in the ViLVaL group. ViLVaL improved insulin signaling and attenuated stress signaling in liver with amelioration of hepatic steatosis due to activated fatty acid oxidation in db/db mice. Furthermore, immunohistochemical analysis of the pancreas revealed that the combination treatment resulted in an increased expression of insulin and PDX-1, and increased insulin content. Conclusions: The combination therapy of ViL and VaL improves both pancreatic beta-cell function and insulin sensitivity, with a reduction of the inflammatory and cell stress milieu in mouse models of T2DM. Our results suggest that this combination therapy exerts additive or even synergistic benefits to treat T2DM.
# Background
Type 2 diabetes is a major global health problem associated with high morbidity and excessive mortality. Because the prevalence of type 2 diabetes is rapidly increasing and current treatments are not sufficient to satisfactorily control the disease in the majority of patients, better treatments are still required [bib_ref] Pathophysiology of prediabetes and treatment implications for the prevention of type 2..., Bergman [/bib_ref]. Pancreatic beta-cell dysfunction and insulin resistance are the two major pathophysiological features of type 2 diabetes [bib_ref] The relative contributions of insulin resistance and beta-cell dysfunction to the pathophysiology..., Kahn [/bib_ref]. Hence, to maintain normoglycemia, the most desirable treatment for type 2 diabetes should target improvement of beta-cell dysfunction and insulin resistance simultaneously.
Diabetic micro-and macrovascular complications are at least partly dependent on dysglycemia itself, which has two main components: chronic sustained hyperglycemia and post-prandial glycemic fluctuations. Both components lead to diabetic complications through several major mechanisms, including activation of oxidative stress and increased chronic inflammatory activity [bib_ref] Biochemistry and molecular cell biology of diabetic complications, Brownlee [/bib_ref]. Oxidative stress has been highly and positively correlated with glycemic variability over a daily period as assessed from the mean amplitude of glycemic excursions (MAGE). In this context, tight glycemic control, as well as suppression of chronic inflammation and cellular stress, is required to achieve ideal treatment of type 2 diabetes.
Glucagon-like peptide 1 (GLP-1) is a gut-derived incretin hormone that stimulates insulin and suppresses glucagon, inhibits gastric emptying, and reduces appetite and food intake. Because the action of incretin is attenuated in type 2 diabetics [bib_ref] Reduced incretin effect in type 2 (non-insulin-dependent) diabetes, Nauck [/bib_ref] , exogenous GLP-1 administration improves glycemic control [bib_ref] Normalization of fasting hyperglycaemia by exogenous glucagon-like peptide 1 (7-36 amide) in..., Nauck [/bib_ref]. Therapeutic approaches for enhancing incretin action include degradation-resistant GLP-1 receptor agonists and inhibitors of dipeptidyl peptidase-4 activity [bib_ref] The incretin system: glucagon-like peptide-1 receptor agonists and dipeptidyl peptidase-4 inhibitors in..., Drucker [/bib_ref]. DPP-4 inhibitors including vildagliptin have been reported to amplify betacell mass and improve its function in a number of in vitro and in vivo studies [bib_ref] Incretin receptors for glucagon-like peptide 1 and glucose-dependent insulinotropic polypeptide are essential..., Flock [/bib_ref] [bib_ref] The DPP-4 inhibitor vildagliptin increases pancreatic beta cell mass in neonatal rats, Duttaroy [/bib_ref] [bib_ref] Dipeptidyl peptidase IV inhibitor treatment stimulates beta-cell survival and islet neogenesis in..., Pospisilik [/bib_ref] [bib_ref] Enhanced beta cell function and anti-inflammatory effect after chronic treatment with the..., Omar [/bib_ref]. A significant and sustained increase in active GLP-1 with suppression of glucagon in response to vildagliptin has been reported in type 2 diabetes [bib_ref] Effects of vildagliptin twice daily vs. sitagliptin once daily on 24-hour acute..., Marfella [/bib_ref]. Vildagliptin improves blood glucose stability in a dose-independent manner [bib_ref] Pharmacodynamics of vildagliptin in patients with type 2 diabetes during OGTT, He [/bib_ref]. Furthermore, vildagliptin may be superior as a means to control MAGE and mean 24 h blood glucose compared with other DPP-4 inhibitors [bib_ref] Comparison of vildagliptin twice daily vs. sitagliptin once daily using continuous glucose..., Sakamoto [/bib_ref]. Blunting glucose fluctuations contributes to reducing oxidative stress and inflammation in diabetics [bib_ref] Reduction of oxidative stress and inflammation by blunting daily acute glucose fluctuations..., Rizzo [/bib_ref]. These results together suggest that, among DPP-4 inhibitors, vildagliptin may have the best potential to achieve total glycemic control and minimize glucose fluctuations.
Patients with type 2 diabetes frequently experience complications such as hypertension. Inhibition of the reninangiotensin system (RAS), especially through the use of angiotensin II type 1 receptor blockers (ARB) such as valsartan, was reported to have a preventive effect on type 2 diabetes [bib_ref] Mechanisms of protective effects induced by blockade of the renin-angiotensin system: novel..., Leung [/bib_ref] [bib_ref] ACE inhibitors and angiotensin receptor antagonists and the incidence of new-onset diabetes..., Aguilar [/bib_ref]. In particular, a lower incidence of newly diagnosed diabetes was confirmed in response to valsartan treatment [bib_ref] The valsartan antihypertensive long-term use evaluation (VALUE) trial: outcomes in patients receiving..., Julius [/bib_ref]. Recent studies have suggested that valsartan increases glucose-stimulated insulin secretion (GSIS) and insulin sensitivity in normotensive subjects with impaired glucose tolerance (IGT) [bib_ref] Valsartan improves {beta}-cell function and insulin sensitivity in subjects with impaired glucose..., Van Der Zijl [/bib_ref]. The prospective NAVIGATOR trial showed that treatment with valsartan reduced type 2 diabetes incidence by 14% in subjects with IGT [bib_ref] Effect of valsartan on the incidence of diabetes and cardiovascular events, Mcmurray [/bib_ref]. ARB treatment was shown to improve GSIS and insulin biosynthesis, and delay the onset of diabetes, in db/db mice [bib_ref] Angiotensin II type 1 receptor blockade improves beta-cell function and glucose tolerance..., Chu [/bib_ref]. These observations suggest that a combination of the DPP-4 inhibitor vildagliptin with the ARB valsartan may exert beneficial effects in the treatment of type 2 diabetes through complementary actions.
# Materials and methods
## Animals
All animals, obtained from CLEA (CLEA Japan Inc., Tokyo, Japan), were housed in the Center for Animal Resources and Development of Kumamoto University. The experimental procedures were approved by the Animal Experimentation Ethics Committee of Kumamoto University (B23-178, B24-128).
Diet and treatment protocol C57BL/6 mice were fed High Fat Diet 32 (HFD group) (CLEA), and db/db mice were fed Rodent Diet CE-2 (db/db group) (CLEA). We had a total of 66 4-week-old db/db mice and 66 6-week-old C57BL/6 J mice (2 weeks of HFD in advance). The first 60 mice of each strain were randomly divided into the following five groups, and the remaining 6 mice of each strain were then randomly assigned to the final 3 groups. All mice received their respective treatments for 8 weeks:
(1) CT: placebo treatment (n = 12), (2) PHZ: Phloridzin treatment (0.5% in diet) (n = 12), (3) ViL: Vildagliptin treatment (1 mg/kg/day in drinking water) (n = 14), (4) VaL: Valsartan treatment (10 mg/kg/day in drinking water) (n = 14), (5) ViLVaL: Vildagliptin (1 mg/kg/day) and Valsartan treatment (10 mg/kg/day) (n = 14). The levels of ViL and VaL used in this study were chosen based on previous studies [bib_ref] Acute and chronic effects of the incretin enhancer vildagliptin in insulin-resistant rats, Burkey [/bib_ref] [bib_ref] Combination of the dipeptidyl peptidase IV inhibitor LAF237 [(S)-1-[(3-hydroxy-1-adamantyl)ammo] acetyl-2-cyanopyrrolidine] with the..., Cheng [/bib_ref] and represent relevant physiological doses for a human clinical situation.
## Measurements of biochemical markers
Blood glucose was measured using a Glutest Neo (Sanwa Chemical Co., Nagoya, Japan). The concentration of insulin was measured using an insulin ELISA kit (Shibayagi, Gunma, Japan). Serum C-reactive protein (CRP; Life Diagnostics Inc., West Chester, PA), tumor necrosis factor (TNF)-α (Invitrogen, Carlsbad, CA), monocyte chemoattractant protein (MCP)-1 (Invitrogen), interleukin (IL)-6 (R&D Systems Inc., Minneapolis, MN), leptin (Crystal Chem Inc., Chicago, IL) and adiponectin (Otsuka Pharmaceutical, Tokyo, Japan) levels were measured using the appropriate ELISA kits.
## Intra peritoneal glucose tolerance test (gtt) and intraperitoneal insulin tolerance test (itt)
After 8 weeks of treatment, GTT (2.0 g/kg) and ITT (1 U/kg for HFD or 2 U/kg for db/db) were performed following an intra-peritoneal injection of glucose or insulin, as previously described [bib_ref] Mild electrical stimulation with heat shock ameliorates insulin resistance via enhanced insulin..., Morino [/bib_ref].
## Pancreatic insulin content
The pancreas was rapidly removed, homogenized, and extracted in acid ethanol overnight at 4°C. The insulin content was measured using a Mouse Insulin ELISA Kit (Shibayagi) [bib_ref] Hyperthermia with mild electrical stimulation protects pancreatic beta-cells from cell stresses and..., Kondo [/bib_ref].
In situ measurement of nitrotyrosine Liver tissue was dissected and immediately frozen in liquid nitrogen. After homogenization, the content of nitrotyrosine was measured using NWLSS™ Nitrotyrosine ELISA test kits (Northwest Life Science Specialties, LLC).
# Western blot analysis
The liver and muscle tissues were dissected 5 min after injection with or without insulin (5 U/mice) via the inferior vena cava and immediately frozen in liquid nitrogen. Western blotting was performed as described previously [bib_ref] Hyperthermia with mild electrical stimulation protects pancreatic beta-cells from cell stresses and..., Kondo [/bib_ref] using antibodies purchased from Cell Signaling Technology (CST; Beverly, MA, USA), except for antiα-tubulin clone DM1A, which was purchased from Millipore (Billerica, MA, USA).
## Immunohistochemistry
Immunohistochemical analysis was performed as described previously [bib_ref] Hyperthermia with mild electrical stimulation protects pancreatic beta-cells from cell stresses and..., Kondo [/bib_ref]. Anti-insulin antibody (diluted 1:200, Santa Cruz (SC) Biotechnology Inc., Santa Cruz, CA), antiglucagon antibody (diluted 1:100, SIGMA-ALDRICH), anti-PDX-1 antibody (diluted 1:100, CHEMICON International Inc., Temecula, CA USA), anti-IRS2 antibody (diluted 1:100, SC), anti-NF-κB p65 antibody (diluted 1:100, CST) and anti-HSP72 antibody (diluted 1:100, Stressgen Biotechnologies, Victoria, BC, Canada) were used.
# Morphometric analysis
The area of insulin positive beta-cells relative to the total area of pancreatic tissue was calculated using Image J image analysis software (Version 1.61; http://rsbweb.nih.gov/ij/), using 18 sections (i.e., six sections from three different areas of the pancreas) for each group of mice.
# Statistical analysis
All data are shown as the mean ± S.E.M. Comparisons of multiple means were performed using two-way repeated measures analysis of variance (for homogeneous variance) as indicated. Multiple comparisons between groups were performed using an analysis of variance followed by Tukey's post-hoc test. A value of p < 0.05 was considered statistically significant.
# Results
## Combination treatment with vildagliptin and valsartan decreases blood glucose
After 8 weeks of treatment, food intake and body weight were unaltered in all groups. Fasting blood glucose levels were significantly decreased in PHZ (HFD: -33.3%, p = 0.027; db/db: -35.7%, p = 0.019), ViL (HFD: -31.6%, p = 0.041; db/db: -39.8%, p = 0.032) and ViLVaL (HFD: -35.8%, p = 0.012; db/db: -42.6%, p = 0.028) interventions compared with the corresponding CT groups (HFD: CT: 140.3 ± 6.3 mg/dL, db/db: CT: 229.5 ± 9.6 mg/dL) after 8 weeks of treatment . There were similar changes in random-fed glucose levels in the PHZ (HFD: -27.8%, p = 0.044; db/db: -36.0%, p = 0.021), ViL (HFD: -30.1%, p = 0.040; db/db: -33.6%, p = 0.018) and ViLVaL (HFD: -34.9%, p = 0.029; db/db: -55.5%, p = 0.005, groups compared with CT (HFD: CT: 212.5 ± 6.8 mg/dL, db/db: CT: 523.2 ± 29.8 mg/dL).
The VaL group did not show a significant decrease in either fasting or random-fed glucose levels and D). The most profound suppression of random glucose levels was observed in ViLVaL-treated db/db animals compared with ViL-treated db/db animals (−33.0%, p = 0.036), suggesting that the combination therapy may have additional effects on glucose lowering.
After 8 weeks of treatment, homeostatic model assessment insulin resistance (HOMA-IR) of HFD mice was significantly decreased in PHZ (CT: 6.8 ± 1.1 PHZ: -29.0%, p = 0.033), ViL (−21.2%, p = 0.041) and ViLVaL (−26.0%, p = 0.015) compared with CT mice . HOMA-IR of db/db mice was significantly reduced in all treatment groups compared with the control group (CT: 22.7 ± 2.1. PHZ: -31.7%, p = 0.046; ViL: -21.0%, p = 0.032; VaL: -30.0%, p = 0.028; ViLVaL: -28.2%, p = 0.024) .
HOMA-beta in HFD showed a significant increase in the PHZ (CT: 91.9 ± 19.2 PHZ: + 168.1%, p = 0.02), ViL (+170.0%, p = 0.031) and ViLVaL (+228.3%, p = 0.018) groups compared with CT animals . HOMAbeta in db/db was also upregulated in PHZ (CT: 86.5 ± 14.2. PHZ: +109.1%, p = 0.016), ViL (+190.9%, p = 0.003) and ViLVaL (+202.7%, p = 0.007) groups compared with the CT group . These data indicate that PHZ, ViL and ViLVaL treatment improved blood glucose, insulin resistance and basal insulin secretion in both groups of mice.
## Combination treatment with vildagliptin and valsartan improves both insulin resistance and insulin secretion
To further examine glucose homeostasis in these animals, we performed GTT with measurements of insulin secretion, and ITT. After 8 weeks of treatments, GTT in HFD mice showed that the blood glucose levels in all treatment groups 30 min after injection were significantly lower than those in the CT group . Glucose levels in the ViL and ViLVaL-treated groups at 60 and 90 min were also significantly lower than those in the CT group .
(See figure on previous page.) Combination treatment with vildagliptin and valsartan decreases blood glucose in HFD and db/db mice. Weekly food intake (A), body weight increments (B), fasting (C) and random-fed blood glucose (D) were measured during each treatment in HFD or db/db mice. HOMA-IR and HOMA-beta were calculated using fasting blood glucose and insulin levels after 8 weeks of treatment (E and F). Values represent mean ± SEM, n = 12 -14. *p < 0.05, **p < 0.01 compared with the CT group. †p < 0.05 compared with the ViL group.
GTT in db/db mice showed that the blood glucose levels in ViL and ViLVaL-treated groups 30, 60, 90 and 120 min after injection were significantly lower than those in the CT group . At 120 min after injection, ViL and ViLVaL treatment groups displayed a significant decrease in blood glucose levels compared with the VaL group .
Serum insulin responses during GTT in db/db mice were improved in PHZ, ViL and ViLVaL mice at 15 min and in ViLVaL mice at 30 min . Insulinogenic index (I.I.) during GTT in db/db was also improved in the PHZ, ViL and ViLVaL groups . I.I. in the ViLVaL-treated group was significantly increased even compared with ViL monotherapy .
The insulin sensitivity estimated by ITT in HFD mice was unchanged in all treated groups . In db/db mice, only ViLVaL-treated mice showed significantly lower glucose levels at 15 and 30 min after insulin injection .
Combination ViLVaL treatment decreases CRP, TNF-α, IL-6 and MCP-1, and increases adiponectin levels Dysregulation of inflammatory cytokines and adipokines is an important pathophysiological feature of type 2 diabetes, and this was investigated 8 weeks after treatment in HFD and db/db mice [fig_ref] Table 1: Plasma inflammatory cytokines and adipokines in HFD and db/db mice treated with... [/fig_ref].
In HFD mice, plasma CRP levels were significantly decreased in ViL, VaL and ViLVaL-treated mice compared with CT mice. TNF-α levels were also significantly reduced in ViL, VaL and ViLVaL-treated mice compared with CT mice. IL-6 was significantly decreased only in ViLVaL compared with CT mice. Plasma MCP-1 levels were significantly lower in the VaL and ViLVaL groups compared with the CT group. Interestingly, ViLVaL showed additional suppression of MCP-1 levels compared with ViL monotherapy. Plasma leptin levels were unaffected by any treatment. Plasma adiponectin levels were significantly increased in ViL and ViLVaL mice compared with CT mice, being highest in the ViLVaL group.
In db/db mice, similar trends were observed in inflammatory cytokines and adipokines. As a result, ViLVaL treatment group showed a strong improvement in inflammatory markers both in HFD and db/db mice.
## Combination therapy attenuates insulin resistance with alleviation of er stress and oxidative stress in liver
To elucidate the molecular mechanism of whole-body insulin sensitization in mice treated with ViL and/or VaL, insulin signaling in db/db mouse liver and muscle was examined. Upon insulin stimulation in vivo, phosphorylation of Akt in the liver was increased in response to ViL (2.03 fold), VaL (2.21 fold) and ViLVaL (2.83 fold) treatment , and was also increased in response to ViLVaL in muscle (1.68 fold, . In order to clarify the improved insulin signaling in liver, endoplasmic reticulum (ER) stress signaling and oxidative stress markers were assessed. Phosphorylation of PKR-like ER kinase (PERK) was significantly decreased in the VaL (−40.7 ± 5.6%, p = 0.019) and ViLVaL (−47.6 ± 6.8%, p = 0.006) treatment groups compared with CT mice . Phosphorylation of eukaryotic translation initiation factor 2A (eIF2α) was also decreased in the VaL (−50.0 ± 7.6%, p = 0.005) and ViLVaL (−49.4 ± 2.3%, p = 0.006) treatment groups .
The oxidative stress marker nitrotyrosine was significantly decreased in VaL and ViLVaL treatment groups compared with the CT and PHZ groups (CT: 6.40 ± 1.10 nM; PHZ: +3.9%, p = 0.999; ViL: -16.7%, p = 0.818; VaL: -52.0%, p = 0.025 (p = 0.014 vs. PHZ); ViLVaL: -54.0%, p = 0.024 (p = 0.014, vs. PHZ: . VaL and ViLVaL treatment groups showed improved insulin signaling accompanied by alleviation of ER stress and oxidative stress.
## Combination therapy improves hepatic lipid accumulation
Hepatic steatosis was examined by oil red O staining in liver of db/db mice. ViL or VaL alone showed a slight improvement of fatty liver, and there was a marked reduction of lipid accumulation in response to ViLVaL compared with the CT or PHZ groups .
To further evaluate the molecular mechanisms of hepatic lipid reduction, the expression of heat shock protein (HSP)72, peroxisome proliferator-activated receptor (PPAR)α and sirtuin1 (Sirt1) were evaluated by western blotting, because these molecules are involved in insulin sensitivity and lipid metabolism in liver [bib_ref] Mild electrical stimulation with heat shock ameliorates insulin resistance via enhanced insulin..., Morino [/bib_ref] [bib_ref] HSP72 protects against obesity-induced insulin resistance, Chung [/bib_ref] [bib_ref] An acylic polyisoprenoid derivative, geranylgeranylacetone protects against visceral adiposity and insulin resistance..., Adachi [/bib_ref] [bib_ref] Lack of SIRT1 (Mammalian Sirtuin 1) activity leads to liver steatosis in..., Xu [/bib_ref]. HSP72, PPARα and Sirt1 were significantly increased in the ViLVaL group compared with all other groups , which was consistent with the decrease in lipid accumulation in liver.
In order to evaluate fatty acid beta-oxidation in liver, mRNA levels of peroxisomal acyl-CoA oxidase 1 (Acox1) and medium chain acyl-CoA dehydrogenase (Mcad) were measured. Acox1, the first enzyme of the betaoxidation pathway, was 1.54 times higher compared with control (p = 0.041, . Mcad, which is responsible for initial dehydrogenation step in the beta-oxidation cycle, was also increased 1.61 times compared with control (p = 0.03, .
## Combination therapy contributes to maintenance of islet morphology and increases the pancreatic insulin content
To identify the molecular mechanism of improved insulin secretion in db/db mice treated with ViLVaL, we analyzed the morphology of pancreatic islets using immunohistochemistry. After 8 weeks of treatments, pancreatic insulin content was significantly increased in the ViLVaL treatment group compared with the CT group (CT: 82.47 ± 20.76 ng/mg; PHZ: +56.4%, p = 0.667; ViL: +77.6%, p = 0.366; VaL: +30.9%, p = 0.946; ViLVaL: +118.8%, p = 0.032: .
The beta-cell area relative to whole pancreas was also significantly increased in the ViL and ViLVaL groups compared with the CT group (CT: 0.64 ± 0.18% vs. PHZ: 1.55 ± 0.36%, p = 0.081; ViL: 1.73 ± 0.19%, p = 0.025; VaL: 0.89 ± 0.16, p = 0.944; ViLVaL: 2.10 ± 0.34, p = 0.0018: . The combination therapy group also showed a significant increase in beta-cell area compared with ViL monotherapy (p = 0.011, .
Morphological analysis showed that immuno-stained insulin was low in CT islets, and this was increased in the ViL and ViLVaL groups . Nuclear localization of pancreatic and duodenal homeobox (PDX)-1 was only sparsely detectable in CT islets, but was increased in the ViL and ViLVaL groups . Anti-apoptotic insulin receptor substrate (IRS)-2 positive area in islets was relatively small in CT mice and was increased in each of the treatment groups . Inflammatory signal detected by nuclear factor (NF)-κB p65 subunit nuclear accumulation was activated in CT islets, and was clearly lower in ViLVaL islets (CT: 9.53 ± 2.25/islet vs. PHZ: 7.33 ± 2.02/islet, p = 0.04; ViL: 2.60 ± 0.95/islet, p = 0.002; VaL: 2.93 ± 1.06/islet, p = 0.003; ViLVaL: 2.20 ± 0.98/islet, p = 0.0001: . The inducible HSP72 (See figure on previous page.) Combination therapy attenuates insulin resistance with alleviation of ER stress and oxidative stress in liver. Liver (A) or muscle (B) lysates were extracted 8 weeks after initiation of each treatment in db/db mice with or without 5 units of insulin stimulation through the inferior vena cava, and were analyzed by western blotting. PERK (C) and eIF2α phosphorylation (D) were also analyzed in liver samples without insulin stimulation. The amount of nitrotyrosine (E) was measured in liver of db/db mice. Values represent mean ± SEM, n = 6 -8. *p < 0.05, **p < 0.01 significantly different from the CT group. † †p < 0.05 compared with the PHZ group.
is also associated with cell protection in islets [bib_ref] Hyperthermia with mild electrical stimulation protects pancreatic beta-cells from cell stresses and..., Kondo [/bib_ref]. Immuno-positive HSP72 was low in CT islets and was increased in ViL, VaL and ViLVaL islets . These results indicate that the combination therapy protects beta-cells from inflammation and degradation in db/db mice.
# Discussion
The main finding of this study is that a combination therapy of vildagliptin and valsartan significantly magnified the beneficial effect of either monotherapy in diabetic mice.
## Insulin sensitizing effects
The ViLVaL combination improved insulin signaling mainly in the liver, with attenuation of ER stress and oxidative stress. As ViL alone shows only slight insulin sensitization, the additional effect of VaL in ameliorating stress signals in liver could explain this additional benefit. The RAS cascade is activated in obesity, metabolic syndrome and type 2 diabetes. RAS activation inhibits insulin signaling through multiple mechanisms, including IRS-1 serine phosphorylation, an increase of reactive oxygen species (ROS) and activation of c-jun N-terminal kinase (JNK), or TNF-α activation [bib_ref] Angiotensin II and the development of insulin resistance: implications for diabetes, Olivares-Reyes [/bib_ref]. Indeed, AT1R blockade with valsartan has been shown to suppress nicotinamide adenine dinucleotide phosphate (NA(D)PH) oxidase p22 phox and gp91 phox [bib_ref] The synergistic effect of valsartan and LAF237 [(S)-1-[(3-hydroxy-1-adamantyl)ammo]acetyl-2-cyanopyrrolidine] on vascular oxidative stress..., Shen [/bib_ref] , ER stress [bib_ref] Valsartan protects against ER stress-induced myocardial apoptosis via CHOP/Puma signaling pathway in..., Wu [/bib_ref] and NF-κB activation [bib_ref] Angiotensin II receptor blocker valsartan suppresses reactive oxygen species generation in leukocytes,..., Dandona [/bib_ref] , resulting in reduced oxidative stress and inflammation. In this study, valsartan ameliorated the inflammatory milieu in liver, resulting in enhancement of insulin action.
HSP72 is a major inducible HSP against several stresses such as heat, ischemia or hypoxia, which protects cells through JNK inhibition [bib_ref] HSP72 protects against obesity-induced insulin resistance, Chung [/bib_ref]. The HSP72 level is decreased in type 2 diabetes and restoration of HSP72 improves insulin resistance and glucose homeostasis in mice [bib_ref] Mild electrical stimulation with heat shock ameliorates insulin resistance via enhanced insulin..., Morino [/bib_ref] [bib_ref] Hyperthermia with mild electrical stimulation protects pancreatic beta-cells from cell stresses and..., Kondo [/bib_ref] [bib_ref] HSP72 protects against obesity-induced insulin resistance, Chung [/bib_ref] [bib_ref] An acylic polyisoprenoid derivative, geranylgeranylacetone protects against visceral adiposity and insulin resistance..., Adachi [/bib_ref]. We observed significant HSP72 up-regulation in liver in response to the ViLVaL combination, suggesting this also contributes to insulin sensitization.
Vildagliptin alone modulates the expression of genes important for lipid metabolism [bib_ref] Incretin receptors for glucagon-like peptide 1 and glucose-dependent insulinotropic polypeptide are essential..., Flock [/bib_ref]. GLP-1 signaling stimulates PPARα and Sirt1 in liver, which in turn enhances fatty acid oxidation and attenuates inflammatory signals. The improvement of hepatic steatosis and presumably inflammation is likely due to activation of GLP-1R on hepatocytes, because others have shown that exenatide stimulates hepatocyte expression of PPARα and PPARγ that improve hepatic fatty acid oxidation, lipid export, and insulin sensitivity [bib_ref] Glucagon-like peptide-1 receptor is present on human hepatocytes and has a direct..., Gupta [/bib_ref]. We observed a significant increase of HSP72, PPARα and Sirt1 with amelioration of hepatic steatosis, particularly in the combined treatment group, suggesting that not only vildagliptin, but also valsartan may participate in activating this pathway. Moreover, hepatic fatty acid beta-oxidation, evaluated by assessment of Acox1 and Mcad RNA levels, was enhanced by the combination treatment, although the lipogenic pathway was not altered (data not shown).
## Beta-cell protecting effects
Cellular stresses such as ER stress and/or oxidative stress are key components in the pathophysiology of type 2 diabetes. In fact, pancreatic beta-cell specific C/EBPbeta transgenic mice exhibit diabetes and decreased beta-cell mass associated with increased apoptosis, decreased proliferation, and aggravated ER stress, and these can be restored by oral administration of vildagliptin [bib_ref] DPP4 inhibitor vildagliptin preserves beta cell mass through amelioration of ER stress..., Shimizu [/bib_ref]. Treatment of rats with vildagliptin significantly increased insulin content and decreased inflammatory markers such as nitric oxide and TNF-α [bib_ref] Chronic DPP-IV inhibition with PKF-275-055 attenuates inflammation and improves gene expressions responsible..., Akarte [/bib_ref] [bib_ref] Vildagliptin selectively ameliorates GLP-1, GLUT4, SREBP-1c mRNA levels and stimulates beta-Cell proliferation..., Akarte [/bib_ref]. Chronic administration of vildagliptin in IRS-2 KO mice improved glucose metabolism and suppressed beta-cell apoptosis, suggesting that vildagliptin may also activate beta-cell proliferation independent of IRS-2 axis [bib_ref] Impact of the dipeptidyl peptidase-4 inhibitor vildagliptin on glucose tolerance and beta-cell..., Sato [/bib_ref]. HFDinduced glucose intolerance and beta cell inflammation were both prevented by chronic treatment with vildagliptin [bib_ref] Enhanced beta cell function and anti-inflammatory effect after chronic treatment with the..., Omar [/bib_ref]. We observed a beta-cell protective effect of ViLVaL in the islets of db/db mice. ViL alone increased the proportion of beta-cells but there was not a significant increase in insulin content. The ViLVaL combination significantly increased both insulin content and the proportion of beta-cells, as well as up-regulating PDX-1 and HSP72, and attenuating NF-κB. We also found that IRS-2 was up-regulated in the treated islets. This effect may also contribute to the protection of beta-cells from apoptosis. HSP72 can attenuate NF-κB nuclear translocation by inhibiting the inhibitor of κB (IκB)αphosphorylation [bib_ref] HSP72 protects against obesity-induced insulin resistance, Chung [/bib_ref]. This indicates that the combination therapy attenuates pancreatic local inflammation, thus leading to beta-cell protection.
## Attenuation of chronic inflammation
We observed a significant reduction of circulating inflammatory markers such as CRP, TNF-α, IL-6 and MCP-1, and an increase of adiponectin, in both HFD (See figure on previous page.) ViLVaL combination therapy improves hepatic lipid accumulation. Hepatic steatosis was evaluated by Oil Red O staining of liver tissues of db/db mice. Scale bars indicate 100 μm at × 200 magnification (A). Liver extracts were subjected to western blotting for HSP72, PPARα and Sirt1 (B). Hepatic lipid oxidation was evaluated by measuring Acox1 and Mcad mRNA expression. Values represent mean ± SEM, n = 6 -8. *p < 0.05, **p < 0.01 compared with the CT group (C). and db/db mice. The ViLVaL combination significantly and profoundly improved these markers, indicating that the systemic chronic inflammatory milieu is alleviated. These cytokines are mainly produced by adipose tissues and macrophages. The RAS blockade decreases adipocyte size and adipose tissue inflammatory markers, and improves glucose homeostasis in rodents [bib_ref] Angiotensin II type 1 receptor blockade improves beta-cell function and glucose tolerance..., Chu [/bib_ref] [bib_ref] Valsartan protects pancreatic islets and adipose tissue from the inflammatory and metabolic..., Cole [/bib_ref] [bib_ref] Blockade of the renin-angiotensin system decreases adipocyte size with improvement in insulin..., Furuhashi [/bib_ref]. Valsartan treatment also reduces abdominal adipocyte size and macrophage infiltration in IGT subjects. These findings suggest that interventions with ViLVaL may attenuate adipose tissue inflammation, thereby contributing to the improvement of whole body glucose homeostasis [bib_ref] Valsartan improves adipose tissue function in humans with impaired glucose metabolism: a..., Goossens [/bib_ref].
Serum DPP-4 levels correlate with adipocyte size and adipocytes potentially represent an important source of DPP-4 in obesity, which impairs insulin signaling [bib_ref] Dipeptidyl peptidase 4 is a novel adipokine potentially linking obesity to the..., Lamers [/bib_ref]. Inhibiting DPP-4 by vildagliptin may directly influence the insulin resistant effect of circulating DPP-4 itself.
We have demonstrated anti-inflammatory benefits of ViL in diabetic mice, however some researchers have reported that ViL may not have cardio-protective effects in a rat model of myocardial infarction [bib_ref] Early and late effects of the DPP-4 inhibitor vildagliptin in a rat..., Yin [/bib_ref]. An underlying diabetic phenotype might be necessary to observe any cardio-protective effects of ViL through its antiinflammatory potency. This possibility needs to be elucidated in further studies.
## The benefits of valsartan in combination with vildagliptin
The combination therapy of vildagliptin and valsartan in db/db mice leads to improvement of islet ROS production, apoptotic events, beta-cell mass, and whole body glucose homeostasis [bib_ref] Combination of the dipeptidyl peptidase IV inhibitor LAF237 [(S)-1-[(3-hydroxy-1-adamantyl)ammo] acetyl-2-cyanopyrrolidine] with the..., Cheng [/bib_ref]. The expression of NAD(P)H oxidase subunits is also significantly decreased by a similar combination resulting in decreased ROS production with GLP-1 up-regulation [bib_ref] The synergistic effect of valsartan and LAF237 [(S)-1-[(3-hydroxy-1-adamantyl)ammo]acetyl-2-cyanopyrrolidine] on vascular oxidative stress..., Shen [/bib_ref]. The decrease of vascular oxidative stress and inflammatory reactions in response to the combination therapy is also more prominent than monotherapy with valsartan or vildagliptin analogues [bib_ref] The synergistic effect of valsartan and LAF237 [(S)-1-[(3-hydroxy-1-adamantyl)ammo]acetyl-2-cyanopyrrolidine] on vascular oxidative stress..., Shen [/bib_ref]. These data support our findings, and we have further described the attenuation of systemic and local inflammation, hepatic steatosis, ER stress and HSP72 induction. We also found beneficial effects on glucose metabolism in the PHZ group, but the beneftts on insulin sensitization, cell stress attenuation and inflammation were relatively small, suggesting that the combination of a DPP-4 inhibitor with an ARB may have advantageous effects in addition to just lowering glucose, which is the case with PHZ.
Although this combined therapy reduced inflammatory cytokines and increased adiponectin, attenuated ER stress, and increased insulin content and proportion of beta-cells, the effect of this combination on glucose metabolism was relatively modest in GTT. Because a DPP-4 inhibitor works in the food-ingested state, an intraperitoneal glucose challenge may have represented a lesser stimulation than food ingestion, and thus led to relatively smaller effects. It is also possible that a longer treatment period with ViLVaL may have led to more pronounced glucose improvement effects, because the differences in glucose levels between treatment groups appeared to increase with time.
In conclusion, the combination of vildagliptin and valsartan shows additive or even synergistic effects in insulin action, hepatic lipid oxidation and beta-cell protection mediated by attenuation of cellular stresses and systemic/ local inflammation. This combination may be useful to ameliorate insulin resistance and beta-cell failure simultaneously, and could further delay vascular complications in type 2 diabetes patients.
Abbreviations DPP-4: Dipeptidyl peptidase-4; ARB: Angiotensin II type 1 receptor blockers; HFD: High-fat diet; MAGE: Mean amplitude of glycemic excursions; GLP-1: Glucagon-like peptide 1; RAS: Renin-angiotensin system; GSIS: Glucose-stimulated insulin secretion; IGT: Impaired glucose tolerance; CRP: C-reactive protein; TNF-α: Tumor necrosis factor-α; MCP-1: Monocyte chemoattractant protein-1.
## Competing interests
The authors declare that they have no competing interests.
Authors' contributions KM generated data, analyzed the data and wrote the manuscript. TK designed the research, generated data, wrote the manuscript, and is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. RG, RM, KO, SK, and SK generated data. MI, JK, TM, and HM contributed to discussions and reviewed the manuscript. EA is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript.
(See figure on previous page.) ViLVaL combination therapy contributes to maintenance of islet morphology and increases the insulin content of the pancreas. Insulin content was measured in pancreatic extracts (A). The proportion of beta-cells to the whole pancreas was calculated using insulin-stained pancreatic sections (B). Sections of pancreas were immunostained for insulin with glucagon (C to G), PDX-1 (H to L), IRS-2 (with DAPI; M to Q), NF-κB p65 subunits (R to U) or HSP72 (V to Z). Scale bars indicate 20 μm at × 400 magnification. Values represent mean ± SEM, n = 6 -8. *p < 0.05, **p < 0.01 compared with the CT group. †p < 0.05 compared with the ViL group.
[table] Table 1: Plasma inflammatory cytokines and adipokines in HFD and db/db mice treated with corresponding interventions [/table]
|
10.1107/s1600577520014319
|
CCBY
| null |
33399576
|
s2orc_pubmed_articles
|
20 µs-resolved high-throughput X-ray photon correlation spectroscopy on a 500k pixel detector enabled by data-management workflow.
The performance of the new 52 kHz frame rate Rigaku XSPA-500k detector was characterized on beamline 8-ID-I at the Advanced Photon Source at Argonne for X-ray photon correlation spectroscopy (XPCS) applications. Due to the large data flow produced by this detector (0.2 PB of data per 24 h of continuous operation), a workflow system was deployed that uses the Advanced Photon Source data-management (DM) system and high-performance software to rapidly reduce area-detector data to multi-tau and two-time correlation functions in near real time, providing human-in-the-loop feedback to experimenters. The utility and performance of the workflow system are demonstrated via its application to a variety of small-angle XPCS measurements acquired from different detectors in different XPCS measurement modalities. The XSPA-500k detector, the software and the DM workflow system allow for the efficient acquisition and reduction of up to $10 9 area-detector data frames per day, facilitating the application of XPCS to measuring samples with weak scattering and fast dynamics. research papers J. Synchrotron Rad. (2021). 28, 259-265 Qingteng Zhang et al. High-throughput XPCS 265
# Introduction
X-ray photon correlation spectroscopy (XPCS) provides information about the dynamic properties of condensed matter across a range of time scales that spans length scales from the atomic in metallic glasses to near micrometres at fluid interfaces. Many of these applications are summarized in recent reviews. Partially driven by XPCS-applicable area detectors with increasingly high frame ratesbut also by emerging near-diffraction-limited storage rings (DLSRs), a growing area of interest is XPCS applied to rapidly fluctuating and often weakly scattering soft materials.
In the case of measurements exhibiting stationary dynamics, the signal-to-noise ratio (SNR) of an equilibrium XPCS measurement is proportional to I 0 t e (N p N f ) 1/2, with I 0 , t e , N p and N f corresponding to the scattering rate of the sample (photons per pixel per second), the exposure time of an individual frame, the number of nominally equivalent pixels in a frame, i.e. within a narrow band of Q, and the number of frames in an acquisition sequence, respectively. Area detectors allow multiple intensity-time series to be measured simultaneously, improving the equilibrium g 2 SNR by (N p ) 1/2 , while also providing concurrent measurements of the Q dependence of g 2 , thereby improving the efficiency of measurements. In the case of two-time measurements where the dynamics evolve over time, the SNR does not improve by increasing N f since the correlation is calculated between two frames, making the large N p provided by an area detector the only way to maintain the SNR of the correlation without sacrificing the time resolution indicated by t e .
Recent developments in fast XPCS-suitable area detectors provide access to increasingly short delay times and a large dynamic range that is relevant to measuring dynamics in soft materials but also exacerbates the challenge of quickly and straightforwardly managing and reducing large amounts of time-series data to correlation functions. For instance,shows g 2 measured from 20 nm-diameter octadecyl-grafted silica nanoparticles suspended in decalin at a volume fraction of 20% (hereinafter referred to as D20) for varying numbers of frames.shows g 2 determined from a single sequence of 100 000 frames acquired at 52 kHz using the Rigaku XSPA-500k detector (described further below) whileis the average of 300 repeats or 'batches' of this same measurement. Evidently, the quality of the result inis much better but achieving this efficiently requires a highthroughput workflow that can keep up with the unprecedented data accumulation rate of the XSPA-500k to provide near real-time human-in-the-loop experiment feedback. In particular, for the example shown, near real-time feedback amounts to our system acquiring, reducing and managing data from approximately 10 13 pixel intensity-time values in 20 min.
In the remainder of this article we describe the performance of the Rigaku XSPA-500k detector and then document the architecture and performance of the workflow and datareduction tools on representative small-angle XPCS measurements using different detectors.
## Xspa-500k detector
The XSPA-500k (X-ray Seamless Pixel Array) detector is a photon-counting detector with a pixel array size of 1024 Â 512, a pixel size of 76 mm and a maximum frame rate of 52 kHz (signal depth is 2 bits in this mode of operation). In the context of XPCS workflow and data management and reduction, we present one extreme of XSPA-500k operation, namely the detector running at its highest frame rate where acquisition is limited to short measurements of 2 s separated by 2 s of sparsification and data movement off the detector control computer.
The XSPA-500k comprises 16 UFXC moduleseach with 256  128 pixels. A single 320 mm-thick silicon sensor spans all 16 modules. As illustrated schematically in , a unique feature of this detector is that the sensor is bonded to the readout module with the sensor pixels displaced from the readout modules' pixels in a manner that results in seamless readout of the entire sensor area of 78 mm  39 mm. Specifically, shows a typical sensor and tiled readout pixelarray detector (PAD) design and the XSPA-500k design with displaced sensor and readout pixels. The red rectangles and squares are sensor pixels, the black squares are readout module pixels, the black circles are the connections (a) g 2 measured from D20 via a single batch of 100 000 frames. The contour of the ROI is shown in [fig_ref] Figure 3: Input arguments are provided by the user and/or assembled by research papers... [/fig_ref] (black dashed lines) and corresponds to Q = 0.0026 Å À1 with a total of 643 pixels. (b) g 2 averaged from 300 batches.
## Figure 2
Schematic perspective views of tiled PADs, (a) without and (b) with displacement between the sensor pixels and readout module pixels. Red rectangles and squares are sensor pixels, black squares are readout module pixels, black circles are the bonding locations between the sensor and readout module pixels, and the cross-hatched blue squares in (a) represent the gaps between tiled PAD readout modules. The XSPA-500k uses the scheme illustrated in (b) to seamlessly span the gap regions between readout modules with (square) pixel sizes that are independent of position on the sensor. between the sensor and readout module pixels, and the crosshatched blue squares in represent the gaps between the readout modules. Notice that the sensor pixels at the gaps in , a typical PAD design, are rectangles, while the sensor pixels in are all squares and are displaced from the readout module pixels to cover the gap regions.
Example data from the XSPA-500k are presented in Figs. 1 and 3. [fig_ref] Figure 3: Input arguments are provided by the user and/or assembled by research papers... [/fig_ref] is the 2D time-averaged small-angle X-ray scattering (SAXS) pattern of D20 determined from a single data-acquisition sequence of 100 000 frames, while [fig_ref] Figure 3: Input arguments are provided by the user and/or assembled by research papers... [/fig_ref] is the azimuthally averaged scattering pattern obtained by averaging the 2D pattern around the origin of reciprocal space.is g 2 calculated from this same sequence andis the average of 300 such sequences or batches. The SAXS and XPCS analyses were performed using the same method as documented by, except that the Q range within which the SAXS intensity is averaged is logarithmically spaced instead of linearly spaced. Note that, due to the limitation of the 19 ms time resolution, g 2 from D20 can only be accurately evaluated for Q < 0.008 Å À1 , although [fig_ref] Figure 3: Input arguments are provided by the user and/or assembled by research papers... [/fig_ref] shows coverage up to Q = 0.016 Å À1 thanks to the large field of view of the XSPA-500k. More efficient use of the large field of view of the XSPA-500k for XPCS studies can therefore be achieved by operating the XSPA-500k at a detector distance much larger than the current 4 m on beamline 8-ID-I at the Advanced Photon Source (APS), Argonne. An example is the up to 16 m setup on the 11-ID beamline (coherent hard X-ray scattering, CHX) at NSLS-II (Brookhaven, New York, USA).
A data acquisition sequence of 100 000 frames can be triggered from the beamline control system as frequently as every 4 s. Two seconds are used for continuous data acquisition, with the remaining time used for sparsification on the detector control computer, transfer and augmentation of the detector data with metadata describing the beamline and experiment configurationto the beamline storage system, and purging the detector memory and readying it for the next acquisition sequence. Sparsification is performed by reading sequences of detector frames into RAM and converting them into sparse data using parallel C# code running on all 20 threads of an i9-9900 CPU with 128 GB RAM. Each photon event is recorded as a 64 bit little-endian number with bits 64-41, 36-17 and 11-1 corresponding to the frame number, pixel index and counts, respectively. Unused bits are reserved to accommodate future detector developments, such as identifying multiple XSPA-500k modules, identifying the acquisition mode or accommodating larger dynamic ranges. Running at full speed, this operation mode of the XSPA-500k can record 0.2 PB of data every 24 h. This should be compared with other state-of-the-art detectors such as 0.03 PB per 24 h for the two-million-pixel ePix10k (vanor 2.1 PB per 24 h for the one-million-pixel AGIPD[both primarily used for free-electron laser (FEL) applications; data rate estimated assuming 2 bytes per pixel and continuous operation over 24 h], while flagship products from Dectris and X-Spectrum are in the petabyte per 24 h range. The large data rate from detectors at next-generation X-ray sources therefore makes automated high-throughput workflow an operationcritical component for effective beamline operation.
## Automated high-throughput data-management workflow
A data-management (DM) workflow has been developed at APS and implemented on many beamlines there, including those dedicated to tomography where it was first deployed. The DM system is a software suite that provides a common framework and a set of services and tools that beamlines can use to tailor their data acquisition, cataloging, storage, distribution and reduction processes. Of particular relevance for this work is the DM processing service that provides support for managing user-defined workflows and for submitting, running and monitoring processing jobs based on those workflows. A workflow definition is kept in the beamline workflow database and is represented by a set of processing steps (stages) that are executed in a defined order. Each stage is associated with a command or script with an arbitrary number of input arguments and output variables.is calculated.
the DM system based on the beamline operating environment and the results of previous steps in the workflow. The execution engine can run multiple processing jobs in parallel and automatically parallelize processing-stage execution over multiple input files. These features enable optimal usage of the storage, network and high-performance computing (HPC) resources available to the beamline. Here, we describe the workflow on 8-ID-I that provides automated high-throughput data reduction with near real-time human-in-the-loop visualization and feedback on XPCS measurements.
The major steps in the XPCS workflow and data reduction are illustrated inand are as follows:
(1) Data are acquired by the detector and stored in local memory.
(2) Raw detector data are sparsified in the memory of the detector control computer.
(3) Sparse data are transferred from storage on the detector control computer to beamline-local networked storage.
(4) Sparse data are combined with metadata and transferred from beamline-local storage to an HPC cluster or workstation via high-bandwidth Globus transfer tools (https:// www.globus.org).
(5) Data are reduced (turned into correlation functions) using newly developed OpenMP-enabled correlation code.
(6) Calculated results and metadata are assembled in an HDF5 file and copied to local networked storage and a Globus endpoint for examination and visualization by and distribution to users.
The metadata capture parameters required for XPCS analysis and also information about the measurement to facilitate interpretation of the measurements. They are assembled from beamline EPICS (Experimental Physics and Industrial Control System, https://epics-controls.org/) process variables and information provided by the user, such as sample identification, and stored in an HDF5 file that accompanies the raw data (compressed detector data are currently kept in a separate but associated binary file so that they can be used in a sparse format throughout processing, a feature not currently supported in HDF5). One example of metadata is the region of interest (ROI) on the detector. Another example can be indices attached to detector pixel maps that define groups of pixels with nominally equivalent Q values for azimuthally averaged SAXS and XPCS analysis.
For XSPA-500k measurements, the typical data-reduction time is approximately 5 s per acquisition, although the exact time can vary depending on the size of the file generated by the measurement. Since the data reduction for each acquisition is distributed and performed in parallel across many nodes, overall data reduction roughly keeps up with measurement time so users can examine the calculated and averaged correlation functions and initiate further local processing of the batch results during the experiment. At the end of the experiment, all user data, including sparsified detector frames, correlation functions and user-defined local processing routines, are transferred to a user-specific Globus endpoint located outside the APS firewall for distribution. Of the DM workflow steps, Step (2) typically leads to a hundred-fold reduction in data volume due to the low photon occupancy of individual frames. Steps (3) and (4) are enabled using Globus high-bandwidth command-line interface (CLI) data-transfer tools. The maximum network transfer speeds for Steps (3) and (4) are 10 and 40 Gb s À1 , respectively, but the actual speed is limited by the speed of writing to the beamline storage hard disk and this is currently 5 Gb s À1 .
Step (5) is performed in newly developed multi-threaded software that expands on the capabilities previously developed. This software reads the sparsified data directly and converts them to a sparsified matrix format without expansion, saving considerable processor memory. The matrix format and the sparse operations have been implemented in C++ to improve performance. The final operation in the g 2 calculation, namely the division of the numerator by the denominator, is performed using the vectorization method from the Eigen libraryfor optimal speed and memory use. Step (6) assembles the calculated results and extensive metadata aggregated from the experiment control and DM systems in an HDF5 file and sends this back to local networked storage for further examination and visualization by users using a graphical user interface (GUI) developed in MATLAB. The GUI allows the time-averaged and time-dependent batch results to be examined individually, overlayed or merged, for example, to improve SNR.
## Xpcs data reduction
Two of the most commonly used XPCS reduction algorithms are multi-tauand two-time. The multi-tau reduction is suitable for XPCS studies where the time scale of the dynamics does not vary significantly during the measurement. In multi-tau, correlations within a time series are averaged over pairs of frames with the same time spacing, resulting in a 1D array. A key feature of the multi-tau correlation routine is using progressively collapsed averages of frames when correlating across successively larger time separations. This feature improves the SNR in calculated correlation functions at long delay times, reducing the computational complexity (N log N versus N 2 ) and providing a practical way of calculating correlation functions over a large dynamic range. The result from a multi-tau reduction of a single XSPA-500k acquisition sequence from a stationary sample (D20) is shown in, where a batch of measurements is performed and reduced as illustrated in. Users can then load XPCS results from individual batches into the MATLABbased XPCS GUIand generate the averaged XPCS result shown infor further analysis.
For dynamics where the time scales vary significantly over time, such as dynamics driven by external stimuli or dynamics displayed during relaxation towards equilibrium, it is necessary to express the correlation as a function of both the measurement time since the impulse and the time separation between pairs of frames. This results in a 2D map of correlation where the x and y axes are the measurement times of the frames and the value of the element at those indices corresponds to the correlation. This is the so-called two-time correlation.is an example of a two-time correlation function at Q = 0.026 Å À1 showing the aging of a 'soft' glass after a rheological shear event. The glass is formed of a concentrated suspension of charged silica nanoparticles in water. The measurements were performed using the Lambda detector running at 10 Hz. Details regarding the rheo-XPCS experimental setup and beam parameters can be found in the work of. As illustrated in, elapsed time since the shear event is obtained from the lower-left to upperright diagonal in the figure, while vertical or horizontal cuts extending from this diagonal show the decay of the correlation as the elapsed time increases. The aging or slowing down of dynamics after the rheological shear event is indicated by the widening of the correlation with increased time after the shear event. The same information can be obtained from, which shows selected cuts ofalong the delay-time cut directions as a function of time after shear. With the DM workflow, different types of XPCS data reduction on data collected from different detectors can be performed quickly and automatically via commands in an execution script, allowing for near real-time human-in-the-loop feedback on the experiment.
## Performance
The performance of the DM workflow system and datareduction system for the two use cases described above is documented in . A significant difference between the two detectors is the ability to sparsify data concurrently with the Lambda detector as opposed to the XSPA-500k detector. This is largely a consequence of the far higher peak frame rate provided by the XSPA-500k. Concurrent sparsification with the Lambda detector is accomplished using the EPICS area-Detector plugin .
For a more equivalent comparison of workflow and data reduction between the two detectors, the two rightmost columns of also document the performance when measuring a 1 mm-thick aerogel static reference. With a 4 m sample-to-detector distance, the Lambda sees a scattering rate of 10 photons per pixel per second with a 55 mm pixel size, while the XSPA-500k sees a scattering rate of 20 photons per pixel per second with a 76 mm pixel size. The performance with the Lambda operating at 1 kHz and the XSPA-500k at 52 kHz.
We note that the run time of the data-reduction step [
Step (5)] is typically more than ten times longer than the rest of the steps combined, indicating that the biggest throughput gains can be achieved by significantly improving the performance of this step. In practice, however, since the reduction occurs on a single node of an HPC cluster and can happen concurrently with the other steps, the effective average time per job can be reduced proportionately by running multiple jobs on different nodes of the cluster. Also, since the bottleneck of the correlation calculation is the arithmetic operation between two equal-length slices of the 1D array recording the scattering intensity at different frames in time, which is evaluated independently at each pixel, significant speed-up can be achieved by parallelizing the pixel-wise computation on a GPU, which typically has a hundred times more cores than a generic CPU.
# Conclusions
We have demonstrated the new Rigaku XSPA-500k detector that acquires data at 52 kHz from a 500k pixel gapless sensor. The data output rate of 0.2 PB per 24 h of continuous operation motivated the application and further optimization of an automated and efficient workflow system coupled to XPCS data reduction on an HPC cluster. We have also demonstrated successful integration of the workflow with various state-of-the-art XPCS-suitable area detectors. The capabilities we have described ease the operational burden of performing XPCS measurements and facilitate human-in-the-loop experiment feedback. Our work has found immediate applications in soft-matter studies on beamline 8-ID-I, typically examining fast dynamics from weakly scattering materials, where automation and high-repetition shortburst measurements are often key to progress.
Looking forward, we see greatly increased data flow arising from XPCS end stations either being discussed or planned for near-diffraction-limited synchrotron sources like MAX-IV. We believe that the infrastructure we have described here will become an essential component for XPCS experiments at future light sources. Measurement types and their performance using the workflow and data-reduction system. (3) Transfer from detector controller to local storage 10 s 1.5 s 0.5 s 0.2 s Step (4) Transfer from local storage to HPC 10 s 1.5 s 0.5 s 0.2 s Step (5) Data reduction (single node, OpenMP) 60 s 60 s 60 s 5 s
Step (6) Results and metadata assembled and returned to the user 1 s 1 s 1 s 1 s
[fig] Figure 3: Input arguments are provided by the user and/or assembled by research papers J. Synchrotron Rad. (2021). 28, 259-265 Qingteng Zhang et al. High-throughput XPCS 261 [/fig]
[table] table documents: research papers J. Synchrotron Rad. (2021). 28, 259-265 Qingteng Zhang et al. High-throughput XPCS 263 [/table]
|
10.1155/2016/8516748
|
CCBY
|
4736813
|
26881045
|
s2orc_pubmed_articles
|
Interplay between Superoxide Dismutase, Glutathione Peroxidase, and Peroxisome Proliferator Activated Receptor Gamma Polymorphisms on the Risk of End-Stage Renal Disease among Han Chinese Patients
Background. Single nucleotide polymorphisms (SNPs) of antioxidants, including superoxide dismutase 2 (SOD2) and glutathione peroxidase 1 (GPX1), play an important role in the risk for cancer and metabolic disorders. However, little is known regarding the effect of antioxidant SNPs on renal events. Methods. We prospectively enrolled multicenter patients with end-stage renal disease (ESRD) and those without chronic kidney disease (CKD) of Han Chinese origin, with SOD2 (Val16Ala), GPX1 (Pro197Leu), and PPAR-(Pro12Ala, C161T) genotyped. Multiple regression analyses were conducted to evaluate the significant risk determinants for ESRD. Results. Compared to ESRD patients, non-CKD subjects were more likely to have T allele at SOD2 Val16Ala ( = 0.036) and CC genotype at PPAR-Pro12Ala ( = 0.028). Regression analysis showed that TT genotype of SOD2 Val16Ala conferred significantly lower ESRD risk among patients without diabetes (odds ratio 0.699; = 0.018). GPX1 SNP alone did not alter the risk. We detected significant interactions between SNPs including PPAR-Pro12Ala, C161T, and GPX1 regarding the risk of ESRD. Conclusion. This is the first and largest study on the association between adverse renal outcomes and antioxidant SNPs among Han Chinese population. Determination of SOD2 and PPAR-SNPs status might assist in ESRD risk estimation.
# Introduction
Patients with persistent renal failure are increasing worldwide, and the number is expected to rise even higher as the population age with multimorbidity. This creates tremendous burden on the existent healthcare resources. In USA, the annual expenditure for the care of end-stage renal disease (ESRD) patients reaches nearly 50 billion dollars in 2011, with an incremental range of 2.8 to 3.4% per year. Given this unavoidable trend, researches focusing on the risk factors and mortality determinants for ESRD assume increasing importance.
Oxidative stress is proposed to play an important role in the pathogenesis of diseases including cardiometabolic 2 Oxidative Medicine and Cellular Longevity diseases, neurodegenerative diseases, neoplasm formation, and, most important of all, chronic kidney disease (CKD) [bib_ref] Increased prevalence of oxidant stress and inflammation in patients with moderate to..., Oberg [/bib_ref] [bib_ref] Endothelial dysfunction, oxidative stress, and risk of cardiovascular events in patients with..., Heitzer [/bib_ref] [bib_ref] Oxidative stress and its significant roles in neurodegenerative diseases and cancer, Thanan [/bib_ref]. Surge of oxidative stress is frequently linked to reactive oxygen species (ROS) formation and/or lowering antioxidant capacity, incurring damage to the surrounding tissues and organ injury. Defending against oxidative stress, antioxidants could be categorized as exogenous (carotene, tocopherol, ascorbate, etc.) and endogenous, especially thiol group-containing macromolecules and enzymes [bib_ref] Targeting antioxidants for cancer therapy, Glasauer [/bib_ref]. There are reports suggesting that prooxidant levels increase and antioxidant enzyme activities decrease stepwise as renal function deteriorates [bib_ref] Effect of different stages of chronic kidney disease and renal replacement therapies..., Tbahriti [/bib_ref]. Moreover, ROS accumulation could not only perpetuate immunogenic glomerular/interstitial injury, but also contribute to endothelial dysfunction over the long run, further deranging the intraglomerular and systemic hemodynamics [bib_ref] Endothelial dysfunction and hypertension, Brandes [/bib_ref].
Among the endogenous antioxidants, superoxide dismutase (SOD) and glutathione peroxidase (GPX) are particularly important players. Several studies have investigated the risk of CKD conferred by the genetic heterogeneity of these two genes. Both SOD1 and SOD2 single nucleotide polymorphisms (SNPs) are found to be associated with 2to 3-fold higher risk of developing diabetic nephropathy in type 1 and 2 diabetes mellitus (DM) patients [bib_ref] The polymorphism of manganese superoxide dismutase is associated with diabetic nephropathy in..., Nomiyama [/bib_ref] [bib_ref] Multiple superoxide dismutase 1/splicing factor serine alanine 15 variants are associated with..., Al-Kateb [/bib_ref]. CKD patients carrying SOD2 Ala16Val genotype also had greater decline of estimated glomerular filtration rate (eGFR) during follow-up [bib_ref] Glutathione peroxidase, superoxide dismutase and catalase genotypes and activities and the progression..., Crawford [/bib_ref]. Similarly, plasma GPX activity also correlated with annual eGFR change in CKD patients [bib_ref] Glutathione peroxidase, superoxide dismutase and catalase genotypes and activities and the progression..., Crawford [/bib_ref]. However, very few studies focused on Asian population, and none of the past reports addressed the influence of SNPs of these antioxidants, especially SOD and GPX, on the risk of dialysisdependent ESRD patients. Previously, we have identified that peroxisome proliferator activated receptor-(PPAR-) SNPs, Pro12Ala, and C161T significantly modified the risk of developing ESRD and prognosis of ESRD patients of Han origin [bib_ref] Sequence variants of peroxisome proliferator-activated receptor-gamma gene and the clinical courses of..., Chao [/bib_ref]. In the current study, we aimed to determine the relationship between SOD2, GPX1, PPAR-genotypes, and the risk of developing ESRD among patients of Han origin, using a large multicenter cohort.
# Material and methods
## Study population.
In this multicenter study, we identified ESRD patients ( = 671) who initiated long-term hemodialysis or peritoneal dialysis from 3 hospitals and 9 clinics in the northern Taiwan between 2002 and 2003 [bib_ref] Serum vitamin D levels are positively associated with varicella zoster immunity in..., Chao [/bib_ref] [bib_ref] Advanced age affects the outcome-predictive power of RIFLE classification in geriatric patients..., Chao [/bib_ref] [bib_ref] Simple self-report FRAIL scale might be more closely associated with dialysis complications..., Chao [/bib_ref] [bib_ref] The severity of initial acute kidney injury at admission of geriatric patients..., Chao [/bib_ref]. Another 780 patients without chronic kidney disease (CKD) were recruited from the health checkup services during the same time period. Blood samples were collected from each subject at the time of enrollment, and individuals were prospectively followed up. The current study was approved by the institutional review boards of all participating institutions.
All subjects completed a questionnaire about their sociodemographic status (e.g., birthday, sex, and education), comorbidity history (e.g., hypertension and diabetes), and lifestyle factors (e.g., smoking and alcohol consumption) with the assistance of general practitioners. Diabetes was determined based on a history of physician-diagnosed diabetes or 2 separate episodes of fasting blood glucose levels ≥126 mg/dL (7.0 mmole/L). The causes of ESRD and the vintage were recorded for all ESRD cases at enrollment. Diabetic nephropathy as the cause of ESRD was recognized if the patient had a long-standing diabetes history (at least 5 years) before dialysis start, presence of diabetic retinopathy, and/or proteinuria (≥100 mg/dL by urinalysis) without other causes. . Genomic DNA was extracted from peripheral blood leukocytes using the Chemagic DNA Blood kits (Chemagen AG, Baesweiler, Germany). Genotypes of the SNPs were determined by the polymerase chain reaction-restriction fragment length polymorphism (PCR-RFLP). Fragments encompassing the SNPs (V16A of SOD2, rs1050450 of GPX1, P12A, and C161T of PPAR-) were generated from genomic DNA by PCR using the primer pairs shown in [fig_ref] Table 1: Primers and restriction enzymes for PCR-RFLP [/fig_ref]. The PCR products were then digested with restriction enzymes (New England BioLabs, Beverly, MA, USA), electrophoresed on a 2.5% agarose gel, and stained with ethidium bromide. The size of the digested PCR products for each SNP variant was also summarized in [fig_ref] Table 1: Primers and restriction enzymes for PCR-RFLP [/fig_ref].
## Laboratory assays
# Statistical analysis.
Data analysis was performed using the SAS, 9.1.3 software (SAS Institute Inc., Cary, NC, USA) and the R, 2.11.1 software (R Foundation for Statistical Computing, Vienna, Austria). In statistical testing, two-sided value ≤ 0.05 was considered statistically significant. The distributional properties of continuous variables were expressed as mean ± standard deviation (SD), whereas categorical variables were presented as frequency and percentage.
The potential risk factors for ESRD were examined in univariate analysis using chi-square test, Fisher's exact test, two-sample -test, Wilcoxon rank-sum test, and log-rank test, as appropriate. Multivariate analysis was subsequently conducted using weighted logistic regression model to estimate the effects of risk factors on the probability of long-term dialysis.
To ensure the quality of analysis results, basic modelfitting techniques for (1) variable selection, (2) goodness-of-fit (GOF) assessment, and (3) regression diagnostics were used in our regression analyses. Specifically, the stepwise variable selection procedure (with iterations between the forward and backward steps) was applied to obtain the candidate final regression model. All the univariate significant and nonsignificant relevant covariates listed in [fig_ref] Table 3: Characteristics of patients with ESRD and those without CKD [/fig_ref] and their interaction variables were selected and the significance levels for entry (SLE) and for stay (SLS) were set to 0.15 or larger (up to 0.35). The best final regression model was identified by reducing the significance levels to 0.05. Both the GOF measures, including the percentage of concordant pairs, estimated area under the receiver operating characteristic (ROC) curve, and adjusted generalized 2 , and the GOF tests, including the deviance GOF test, Pearson chi-squared GOF test, and Hosmer-Lemeshow GOF test (for logistic regression model), and, if feasible, Grønnesby-Borgan GOF test (for Cox's proportional hazards model), were examined to assess the GOF of the fitted regression models. The generalized additive models (GAM) were applied to detect nonlinear effects of continuous covariates. [fig_ref] Table 2: Comparisons of the minor allele frequencies of the four SNPs and examinations... [/fig_ref]. Both SOD2 and GPX1 SNPs satisfied the HWE in non-CKD and ESRD patients, while PPAR-Pro12Ala among ESRD cases showed deviation from HWE ( = 0.02). This suggests a potential association between PPAR-Pro12Ala and the presence of ESRD, as demonstrated in our previous study [bib_ref] Sequence variants of peroxisome proliferator-activated receptor-gamma gene and the clinical courses of..., Chao [/bib_ref].
# Results
## Clinical features of
The genotype distributions and the other clinical characteristics were compared between the enrolled ESRD and non-CKD patients in [fig_ref] Table 3: Characteristics of patients with ESRD and those without CKD [/fig_ref]. The mean serum creatinine (mg/dL) was significantly higher in ESRD patients (non-CKD versus ESRD, 1.0 ± 0.3 versus 11.2 ± 2.5; < 0.001). In addition, the proportion of patients with DM was significantly higher in ESRD patients (non-CKD versus ESRD, 10.6% versus 37.7%; < 0.001). The distributions of the SOD2 exon 2 and PPARexon B genotypes were significantly different between the two groups. TT genotype of SOD2 exon 2 was predominant in non-CKD patients ( = 0.036), while CC genotype of PPARexon B occurred more frequently in non-CKD patients ( = 0.028). There was no significant heterogeneity regarding GPX1 and PPAR-exon 6 genotypes between ESRD and non-CKD patients.
We further stratified the ESRD patients according to their ESRD etiology as DM or non-DM related, since PPARpolymorphism is reportedly associated with development of DM nephropathy (DMN) [fig_ref] Table 4: Characteristics of ESRD patients due to DMN and non-DMN related ESRD [/fig_ref] [bib_ref] Peroxisome proliferatoractivated receptor polymorphism Pro12Ala is associated with nephropathy in type 2..., Zhang [/bib_ref]. About 27% patients had DMN-related ESRD. No significant difference in genotype distributions of SOD2, GPX1, and PPAR-was noted between those with or without DMN.
## Risk of esrd associated with snps of sod2 and gpx1.
We first investigated the influence of SOD2 and GPX1 genetic polymorphism on developing ESRD. As shown in [fig_ref] Table 5: Multivariate analysis * of the predictors of dialysis by fitting multiple logistic... [/fig_ref] , multiple logistic regression analysis incorporating factors in [fig_ref] Table 3: Characteristics of patients with ESRD and those without CKD [/fig_ref] identified five significant predictors of ESRD developing among the entire cohort. Patients with DM and SOD2 exon 2 CC genotype had significantly higher risk
## Risk of esrd associated with snps of sod2, gpx1
, PPAR-Pro12Ala, and C161T. We further combined the previously reported PPAR-SNPs (i.e., Pro12Ala and C161T) and SOD2 as well as GPX1 SNPs in statistical models for a more in-depth analysis. Regression analysis identified multiple significant predictors of developing ESRD. Individuals without DM but with SOD2 exon 2 TT genotype were still at lower risk of ESRD (OR 0.69, 95% CI 0.51-0.93; = 0.014). Additionally, we discovered that PPAR-SNPs and GPX1 SNPs interacted with regard to their effect on ESRD risk. Patients with PPAR-exon 6 TT genotype and exon B GG genotype displayed a high risk of ESRD ( < 0.001), while exon 6 CC genotype and exon B CC genotype were protective against ESRD (OR 0.78, 95% CI 0.61-0.99; = 0.04). Patients with simultaneous GPX1 exon 2 CC genotype and PPARexon B GG genotype had significantly lower ESRD risk ( < 0.001).
# Discussion
In the current study, we addressed the issue whether antioxidative enzymes (SOD2 and GPX1) SNPs affect the risk of developing ESRD, in a large cohort of patients of Han origin, and their potential interactions with other known ESRD riskmodifying SNPs (PPAR-). We discovered that risk conferred by SOD2 and PPAR-SNPs in Han Chinese population is modified in the presence of DM, and interactions exist among different SNPs. Particularly, PPAR-exon B GG genotypes increased risk of ESRD in patients without DM and those with exon 6 TT genotype. In addition, the influence of SOD2 SNPs on the risk of ESRD differed between patients with and without DM. These findings suggested a renewed explanation for the associations between both antioxidants, PPAR-SNPs and DM, non-DM related ESRD in the literature.
The SOD2 SNPs, rs4880, substitute a C to T at position 2734 and change amino acid coding from alanine to valine at position 16 (Ala16Val). This alteration leads to mRNA instability and reduced SOD2 expression. SOD2 Ala16Val SNPs have been found to correlate variably with different cancer susceptibility. A meta-analysis concluded that Ala allele conferred a borderline elevated risk of breast cancer [bib_ref] Commonly studied single-nucleotide polymorphisms and breast cancer: results from the breast cancer..., Consortium [/bib_ref]. Other researchers identified a higher risk of prostate cancer (odds ratio, 1.16) and increased aggressiveness among Ala alleles of SOD2 exon 2 SNP carriers [bib_ref] Relationships between single nucleotide polymorphisms of antioxidant enzymes and disease, Crawford [/bib_ref]. Similarly, carriers of SOD2 Ala16Val exhibited a higher risk of bladder cancer, particularly among concurrent smokers [bib_ref] Genetic polymorphisms of MPO, COMT, MnSOD, NQO1, interactions with environmental exposures and..., Hung [/bib_ref]. On the other hand, SOD2 Ala16Val SNPs might also play a role in the susceptibility of multiple metabolic disorders. Ala allele has been linked with elevated risk of diabetic macular edema and diabetic retinopathy in Asian population [bib_ref] Association of manganese superoxide dismutase gene polymorphism (V16A) with diabetic macular edema..., Lee [/bib_ref] [bib_ref] Polymorphism of the manganese superoxide dismutase gene but not of vascular endothelial..., Kangas-Kontio [/bib_ref]. Moreover, SOD2 Ala16Val SNPs also serve as an important risk factor Oxidative Medicine and Cellular Longevity 5 for atherosclerotic burden, cardiac events, and nonalcoholic steatohepatitis [bib_ref] The signal sequence polymorphism of the MnSOD gene is associated with the..., Kakko [/bib_ref] [bib_ref] Manganese superoxide dismutase dimorphism relationship with severity and prognosis in cardiogenic shock..., Charniot [/bib_ref] [bib_ref] Polymorphisms of microsomal triglyceride transfer protein gene and manganese superoxide dismutase gene..., Namikawa [/bib_ref]. These vascular and metabolic abnormalities among SOD2 Ala16Val carriers, coupling with the susceptibility to urologic malignancies, might pave the way for incident CKD development, progression, and ESRD [bib_ref] Association of non-alcoholic fatty liver disease with chronic kidney disease: a systematic..., Musso [/bib_ref]. Indeed, subsequent study revealed that Val allele of SOD2 exon 2 carriers had a 3-4-fold higher risk of rapid renal function decline compared to the others, but this was established only in Caucasians [bib_ref] Glutathione peroxidase, superoxide dismutase and catalase genotypes and activities and the progression..., Crawford [/bib_ref]. Based upon a large cohort of Han Chinese patients, we discovered that CC genotype of SOD2 exon 2 portended significantly higher risk of ESRD among those with DM, while TT genotype decreased the risk among non-DM ones [fig_ref] Table 5: Multivariate analysis * of the predictors of dialysis by fitting multiple logistic... [/fig_ref]. Although our identified at-risk allele (Ala) in Han Chinese population differs from that of Caucasians (Val) in most studies, both findings underline the importance of SOD2 exon 2 SNPs in modifying the risk of CKD progression as well as ESRD. GPX1, a selenoenzyme that catalyzes the breakdown of various peroxides into O 2 and H 2 O, is produced by nearly all tissues and serves as a cellular barrier against oxidative stress. Past studies focusing on the role of GPX1 SNPs mostly included GPX1 SNP rs1050450, which causes a C to T mutation [bib_ref] Relationships between single nucleotide polymorphisms of antioxidant enzymes and disease, Crawford [/bib_ref]. GPX1 SNP rs1050450 is found to influence the risk of lung cancer and bladder cancer [bib_ref] Do genetic factors protect for early onset lung cancer? A case control..., Rosenberger [/bib_ref]. Risk of vascular calcification and atherosclerosis is also affected by Leu allele of GPX1 SNPs [bib_ref] Polymorphisms of microsomal triglyceride transfer protein gene and manganese superoxide dismutase gene..., Namikawa [/bib_ref] [bib_ref] Genetic association of glutathione peroxidase-1 with coronary artery calcification in type 2..., Nemoto [/bib_ref]. However, available evidence did not agree that GPX1 SNP rs1050450 plays a significant role in CKD progression or renal allograft dysfunction [bib_ref] Glutathione peroxidase, superoxide dismutase and catalase genotypes and activities and the progression..., Crawford [/bib_ref] [bib_ref] Lack of association of C599T polymorphism in the glutathione peroxidase (GPX1) gene..., Dutkiewicz [/bib_ref]. We similarly did not detect significant influences of GPX1 Pro197Leu SNPs on the risk of ESRD in Han Chinese population [fig_ref] Table 5: Multivariate analysis * of the predictors of dialysis by fitting multiple logistic... [/fig_ref]. This is compatible with the literature and further supports the finding that GPX1 SNPs alone did not affect the risk of ESRD in most reports. However, a subgroup of patients with GPX1 CC genotype did demonstrate lower risk of ESRD, that is, those with concurrent PPAR-exon B GG. A previously undetected interaction between antioxidants SNPs and PPAR-SNPs might exist and warrants further investigation.
The interactions between DM/antioxidant SNPs and within different antioxidant SNPs on the risk of ESRD are interesting. From our findings, we propose that DM and CC genotype might act synergistically to elevate the risk of subsequent ESRD or that CC genotype might promote the development of diabetic nephropathy [fig_ref] Table 5: Multivariate analysis * of the predictors of dialysis by fitting multiple logistic... [/fig_ref]. However, past studies failed to disclose any significant effect of SOD2 Ala16Val SNPs on the presence of DM, but, on the contrary, TT genotype carrier of SOD2 exon 2 was more likely to be found in Caucasian and Japanese patients with DM nephropathy than those without [bib_ref] The polymorphism of manganese superoxide dismutase is associated with diabetic nephropathy in..., Nomiyama [/bib_ref] [bib_ref] Association of manganese superoxide dismutase gene polymorphism (V16A) with diabetic macular edema..., Lee [/bib_ref] [bib_ref] The V16A polymorphism in SOD2 is associated with increased risk of diabetic..., Möllsten [/bib_ref]. It is likely that SOD2 exon 2 SNPs, specifically CC genotypes, contribute to the risk of DM nephropathy among Han Chinese population. Furthermore, we also discovered that patients without DM were at significantly lower risk of ESRD if they carried TT genotypes of SOD2 exon 2 [fig_ref] Table 5: Multivariate analysis * of the predictors of dialysis by fitting multiple logistic... [/fig_ref]. This could be a circumstantial evidence supporting the detrimental effect of CC genotypes on ESRD, while it alternatively could suggest a protective effect brought by TT genotypes against non-DM ESRD. Finally, our findings that PPAR-Pro12Ala interacted with non-DM status on the effect of risk for ESRD could also be a token of gene-environmental interaction, as PPAR-Pro12Ala has been found to act together with body mass index or dietary fatty acid intake to increase cardiometabolic risk [bib_ref] Meta-analysis of gene-environment interaction: joint estimation of SNP and SNP × environment..., Manning [/bib_ref] [bib_ref] Interaction of PPARG Pro12Ala with dietary fat influences plasma lipids in subjects..., Alsaleh [/bib_ref].
Several lines of evidence suggested that interactions between different SNPs could modify the risk of CKD and its progression. FIND (Family Investigation of Nephropathy and Diabetes) study group found that -actinin-4 polymorphism could interact with APOL1 SNPs with regard to the risk of non-DM related ESRD in African Americans [bib_ref] Relevance of the ACTN4 gene in African-Americans with non-diabetic endstage renal disease, Bostrom [/bib_ref]. The same group subsequently utilized a large cohort study to identify podocin (NPHS2) and APOL1 SNPs might jointly promote nondiabetic nephropathy in African Americans, although the mechanisms of synergism still remained undetermined [bib_ref] Genetic association and gene-gene interaction analyses in African American dialysis patients with..., Bostrom [/bib_ref]. Importantly, PPAR-activation has been reported to regulate renal mesangial inflammation and modulate renal fibrosis through p65 and NF-B pathways [bib_ref] Opposite action of peroxisome proliferator-activated receptor-in regulating renal inflammation: functional switch by..., Wen [/bib_ref]. Furthermore, experimental evidence also suggested that PPAR-participates in regulating oxidative stress and acute kidney injury [bib_ref] Interaction of oxidative stress, nitric oxide and peroxisome proliferator activated receptor in..., Yousefipour [/bib_ref]. Consequently, we could expect that PPAR-SNPs and even antioxidant SNPs might interact with regard to their influence on the risk of ESRD. In this study, we discovered that PPAR-SNPs, Pro12Ala, C161T, and GPX1 SNPs might interact with each other, lending support to our hypothesis that there could be interplay between different SNPs concerning their clinical effects.
Our study has its strength and limitations. The case number was large, and our investigation on the reciprocal relationship between different SNPs among a population receiving less attention, Han Chinese, made this study more appealing. However, the current study was also limited by its clinical setting and the limited inclusion of antioxidant SNPs. Further research utilizing more comprehensive methods for SNPs survey might be needed in the future.
# Conclusion
Using a large Han Chinese-based cohort, we studied the modifying effect of SOD2, GPX1, and PPAR-SNPs on the risk of ESRD. We found that CC genotype of SOD2 elevated the risk of ESRD in patients with DM, while TT genotype of SOD2 decreased the risk in patients without DM. In addition, interactions between PPAR-exon B/exon 6 and PPAR-exon B/GPX1 exon 2 as ESRD risk determinants were also detected. These findings pave the way for a better understanding of the genetic influences on the susceptibility to ESRD among the Han Chinese, a population with high ESRD prevalence.
[fig] *: Multiple logistic regression model: = 1451, adjusted generalized 2 = 0.304, estimated area under the receiver operating characteristic (ROC) curve = 0.769, and the Hosmer and Lemeshow goodness-of-fit chi-square test = 0.0047 (df = 8). # Variables with odds ratio (OR) of extreme values were listed below: DM × SOD2 exon 2 CC, OR 2.7 × 10 6 ( < 0.001); non-DM × PPAR-exon B GG, OR 3.8 × 10 13 ( < 0.001); PPAR-exon 6 TT × PPAR-exon B GG, OR 8.2 × 10 11 ( < 0.001); GPX1 exon 2 CC × PPAR-exon B GG, OR 1.9 × 10 −7 ( < 0.001). [/fig]
[table] Table 1: Primers and restriction enzymes for PCR-RFLP. [/table]
[table] Table 2: Comparisons of the minor allele frequencies of the four SNPs and examinations of the Hardy-Weinberg equilibrium among the entire cohort. [/table]
[table] Table 3: Characteristics of patients with ESRD and those without CKD. 30%) * Continuous variables were tested by Kruskal-Wallis test, whereas categorical variables were tested by Fisher's exact test. [/table]
[table] Table 4: Characteristics of ESRD patients due to DMN and non-DMN related ESRD. [/table]
[table] Table 5: Multivariate analysis * of the predictors of dialysis by fitting multiple logistic regression model with the stepwise variable selection method, including SOD2 and GPX1 polymorphisms # . [/table]
|
10.1186/1472-6831-10-10
|
CCBY
|
2867770
|
20429938
|
s2orc_pubmed_articles
|
Periodontal disease-associated micro-organisms in peri-menopausal and post-menopausal women using or not using hormone replacement therapy. A two-year follow-up study
Background: Despite conflicting results on the use of hormone replacement therapy (HRT) there is no doubt that many women benefit from it. Women using HRT are known to be more health conscious in general with putative positive implications in the mouth. However, we observed recently in our cohort hardly any difference in oral health status between HRT-users and non-users. There are only a few studies about HRT and oral microbiota. We hypothesized that counts of periodontal micro-organisms are lower in health-conscious HRT-users than non-users.Methods:Two-year open follow-up study was conducted on originally 200 HRT-users and 200 non-users from age cohorts of 50-58 years. After clinical examination pooled subgingival plaque samples were taken for polymerase chain reaction analyses. The results of finally 135 women meeting the inclusion criteria were analyzed with cross-tabulation and chi-square test. Explanatory factors were studied by step-wise logistic regression analysis.Results:In HRT group, the numbers of positive samples for Porphyromonas gingivalis (P. gingivalis, p < 0.07), Prevotella intermedia (P. intermedia, p < 0.05)and Tannerella forsythia (T. forsythia, p < 0.01) decreased in women with ≥ 4-mm-deep pockets. Respectively in HRT users with ≥ 6-mm-deep pockets the numbers of positive samples for P. gingivalis (p < 0.05) and T. forsythia (p < 0.01) were decreased. No corresponding differences were observed in the non-HRT group. In logistic regression, the existence of deep periodontal pockets explained the majority of cases harboring specific microorganisms in both groups.Conclusion:Although use of HRT did not correlate with periodontal health status, HRT led to decreasing numbers of positive samples of the periodontal pathogens P. gingivalis and T. forsythia. Further studies with longer observation time are needed to observe the clinical relevance of the results.
# Background
Women use hormone replacement therapy (HRT) to cope with estrogen deficiency-induced vasomotor and urogenital symptoms. Women do not only take HRT to avoid climacteric symptoms but also to maintain quality of life and to prevent osteoporosis [bib_ref] The epidemiology of coronary heart disease and estrogen replacement in postmenopausal women, Grodstein [/bib_ref]. The use of HRT has received much publicity, however, when the results from large randomized controlled trials were published [bib_ref] Effects of estrogen replacement on the progression of coronary-artery atherosclerosis, Herrington [/bib_ref] [bib_ref] Randomized trial of estrogen plus progestin for secondary prevention of coronary heart..., Hulley [/bib_ref] [bib_ref] for the Women's Health Intiative Investigators: Conjugated equine estrogens and coronary heart..., Hsia [/bib_ref] [bib_ref] for the Women's Health Intiative Investigators: Estrogen plus progestin and the risk..., Manson [/bib_ref]. Recent experimental and clinical studies have indicated that effects of HRT depend on the estrogen and progesterone/progestin formulation, dosage, mode of administration, patient's age, associated diseases, and duration of treatment [bib_ref] Hormonal replacement therapy (HRT) in postmenopause: a reappraisal, Caufriez [/bib_ref] [bib_ref] Estrogen replacement in menopausal women: recent and current prospective studies, the WHI..., Harman [/bib_ref]. However, there is no doubt that many women clearly benefit from the use of HRT, which may also have implications in the oral cavity [bib_ref] The physiology, medical management and oral implications of menopause, Friedlander [/bib_ref].
Chronic infections are involved in the etiopathogenesis of many systemic diseases by releasing pro-inflammatory mediators that may cause endothelial damage and initiate, for example, cholesterol plaque attachment (for review, see [bib_ref] Oral health, atherosclerosis and cardiovascular disease, Meurman [/bib_ref]. Periodontal disease being highly prevalent in the populations presents a marked inflammatory burden in this regard (for review, see [bib_ref] Periodontal diseases, Pihlstrom [/bib_ref].
There is little knowledge about the prevalence of periodontal microbiota among peri-menopausal and postmenopausal women [bib_ref] Periodontal status of women taking postmenopausal estrogen supplementation, Norderyd [/bib_ref]. However, sex hormones have long been considered to play an influential role on periodontal tissues and periodontal disease progression and therefore the role of sex hormones has been of interest in further investigations [bib_ref] Hormonal influences on gingival tissue: relationship to periodontal disease. Review Article, Sooriyamoorthy [/bib_ref] [bib_ref] Influence of sex hormones on the periodontium. Review Paper, Mascarenhas [/bib_ref] [bib_ref] Bacterial species in subgingival plaque and oral bone loss in postmenopausal women, Brennan [/bib_ref] [bib_ref] Gingival changes during pregnancy: II. Influence of hormonal variations on the subgingival..., Carrillo-De-Albornoz [/bib_ref]. We thought it interesting to study if differences occur in harboring certain periodontal bacteria, namely Porphyromonas gingivalis (P. gingivalis), Tannerella forsythia (T. forsythia), Aggregatibacter actinomycetemcomitans (A. actinomycetemcomitans), Treponema denticola (T. denticola), Prevotella intermedia (P. intermedia) and Prevotella nigrescens (P. nigrescens), between pre-and postmenopausal women using or not using HRT. The follow-up time was two years and the present report is a continuation of our study on oral health of the cohort of peri-menopausal and postmenopausal women in the age groups of 50-58 years [bib_ref] Oral health in perimenopausal and early post menopausal women from baseline to..., Tarkkila [/bib_ref]. In the earlier study we did not observe any difference in the clinical oral health status between the groups. However, taking into account the known associations between periodontal bacteria and systemic diseases [bib_ref] Oral health, atherosclerosis and cardiovascular disease, Meurman [/bib_ref] and the interactions between HRT use and general health [bib_ref] for the Women's Health Intiative Investigators: Estrogen plus progestin and the risk..., Manson [/bib_ref] we thought that HRT might affect periodontal infections as reflected in the occurrence of specific bacteria. We thus assumed that the more health-conscious HRT users harbor less frequently periodontal pathogens when compared with non-users at the end of this 2-year study.
# Methods
## Patients and their examination
Details of our patient material are given in our earlier contribution [bib_ref] Oral health in perimenopausal and early post menopausal women from baseline to..., Tarkkila [/bib_ref]. Patients were chosen from original sample of 3173 women at age cohorts 50, 52, 54, 56, and 58 years, who had participated in our questionnaire study on oral symptoms. Of those five age cohorts 40 women reporting the use of HRT and 40 reporting not to use HRT at the time of the questionnaire were randomly selected and invited to attend our clinical study. Hence, 200 HRT users and 200 non-HRT users received our invitation letter. One recall letter was sent to those who did not respond to the first invitation.
For the baseline study altogether 249 women attended, with a response rate of 62%. After two years all those who attended the baseline clinical study were invited to a follow-up study. For the follow-up study 193 women attended (78%). However, for statistical analyses of this follow-up study the women who had changed their HRT protocol (using/not-using HRT) between the baseline and 2-year follow-up were excluded (n = 32). Also women who had reported regular menstruation were excluded from the final analyses (n = 11). Finally the results from 161 were included in our analyses but complete microbi-ological results were available for only 135 subjects. The study profile is given in [fig_ref] Figure 1: Flow-chart of the inclusion of subjects [/fig_ref].
Ethical approval of the study had been originally given by the City of Helsinki Health Department (ethical permit #53 1.4.1997). All subjects attending signed an informed consent according to the Declaration of Helsinki.
Use and duration of prescribed medications including HRT as well as illnesses diagnosed by physician were asked with open questions with a structured questionnaire. Use of intrauterine contraception was not considered as HRT. For the statistical analyses a woman was included into the HRT group if she had used HRT at least six months at the time of the questionnaire.
Both the periodontal examinations and microbiological samplings, at baseline and two years later, were made by one dentist (author L.T). Oral examinations were made according to the WHO guidelines [bib_ref] Development of the World Health Organization (WHO) Community periodontal index of treatment..., Ainamo [/bib_ref]. Panoramic jaw radiographs were available of each subject at the clinical examination. Periodontal probing depths were determined at all mesial, distal, vestibular (buccal, labial), and oral (lingual, palatal) tooth surfaces for scoring the Community Periodontal Index of Treatment Needs (CPITN) [bib_ref] Development of the World Health Organization (WHO) Community periodontal index of treatment..., Ainamo [/bib_ref]. The periodontal pocket depths were measured with a WHO probe (tip diameter 0.5 mm) to the nearest milli- meter (mm) from the gingival margin to the bottom of the pocket. "Periodontitis" was diagnosed if at least one sextant had a CPITN -score 3 or higher, thus indicating at least ≥ 4 mm deep pockets in the subject. Individual pocket depths were not recorded separately. However, the number of ≥ 6 mm deep pockets was counted separately for each patient.
## Microbial sampling methods and analyses
Subgingival plaque from the four deepest periodontal sites was sampled in every patient. If no ≥ 4 mm pockets existed or there were less than four ≥ 4 mm pockets in the subject the rest of the plaque samples were taken from the first premolars. Before sampling the supragingival sites, plaque was gently removed with a cotton swab and the site of collection was isolated with cotton rolls and dried. Thereafter, the subgingival plaque samples were taken with a sterile curette and placed into a micro-centrifuge tube containing 0.5 ml distilled water. The pooled samples were then centrifuged and 5 μl aliquots of supernatant were added to the PCR reaction mixtures. The PCR method has been described in detail by Wahlfors et al. [bib_ref] Simultaneous detection of Actinobacillus actinomycetemcomitans and Porphyromonas gingivalis by a rapid PCRmethod, Wahlfors [/bib_ref] and Meurman et al. [bib_ref] Identification of Bacteroides forsythus in subgingival dental plaque with the aid of..., Meurman [/bib_ref]. Periodontal bacteria A. actinomycetemcomitans, P. gingivalis, T. forsythia, T. denticola, P. intermedia and P. nigrescens were analyzed.
# Statistical methods
Women using or not using HRT were analyzed separately and then the groups were compared in the 2-year followup scheme. Only those who were examined both at baseline and then followed-up for two years were included in the final analyses. Cross-tabulations and chi-square tests were used in the SPSS for Windows statistical program (SPSS Inc. Chicago, Ill., USA, version 13.0). Stepwise logistic regression analyses were made to study the explanatory factors for the positive periodontal microbial findings. The explanatory factors used in the backward stepwise logistic regression model were age, prevalence of cardiovascular disease, number of teeth, use of HRT, WHO diseased, missed, filled tooth index -score, existence of ≥ 4 mm deep and ≥ 6 mm deep periodontal pockets using the CPITN-score data, and self-assessed poor oral health. The effect of systemic diseases and medications on the results was analyzed separately for HRT and non-HRT groups using median values of the total number of concomitant illnesses and drugs used daily; both at baseline and two years later.
# Results
The reasons for drop-outs and exclusion from the study were mainly change in the original protocol by either starting HRT by women originally assigned to the non-HRT group or vice versa, or lack of interest. The demographic characteristics of the drop-outs and those excluded did not differ from the women who were eligible for the 2-year follow-up analyses. As expected, women not using HRT reported more often climacteric symptoms than HRT users: 35% at baseline vs. 13% at the 2year examination, respectively (p < 0.001).
At baseline the percentage of women with periodontitis (defined as at least one sextant with CPITN score-3) was 79% in HRT group and 80% in non-HRT group. Two years later the respective figures were 71% and 76%. The mean numbers of ≥ 6 mm deep periodontal pockets were 0.9 ± 1.7 at baseline vs. 1.1 ± 2.1 two years later in the HRT group, and 1.0 ± 1.7 vs. 1.2 ± 1.9, respectively, in the non-HRT group. These differences were not statistically significant. More detailed data of the dental health parameters are described in [fig_ref] Table 1: Descriptive data of the study subjects divided in groups according to the... [/fig_ref]. [fig_ref] Figure 2: Periodontal micro-organisms detected at baseline and two years later in the women... [/fig_ref] gives the frequencies of the positive periodontal microbial results of the subjects with ≥ 4 mm and ≥ 6 mm deep periodontal pockets, respectively, as assessed using the CPITN scoring. In general, at baseline the periodontal pathogen P. gingivalis was detected in approximately 20% among all the women, while A. actinomycetemcomitans was detected in less than 5%. T. forsythia and T. denticola were detected in 30-50% of the women. There was no statistically significant difference in the frequency of positive microbial findings between the groups at baseline. However, two years later significantly less positive cases of P. intermedia (p < 0.05) and T. forsythia (P < 0.01) were detected in the samples from the ≥ 4 mm deep pockets of the HRT users while no such difference was observed in the non-HRT group. Of the HRT users the decrease in the number of positive samples of P. gingivalis almost reached statistical significance in the ≥ 4 mm deep pockets (P = 0.069). Correspondingly, the 2year follow-up results of the samples from the ≥ 6 mm deep pockets of the women using HRT showed a significant decrease in the number of positive cases of P. gingivalis (p < 0.05) and T. forsythia (p < 0.05) while no difference was observed in the non-HRT group.
When the background variables for the prevalence of periodontal micro-organisms in general were analyzed in the stepwise logistic regression, the existence of deep periodontal pockets (data based on CPITN score 3 or higher) explained the majority of the cases. As regards the effect of reported systemic diseases on the microbiological results, A. actinomycetemcomitans was found to be more prevalent in women with cardiovascular disease (Odds ratio [OR 5.48], 95% confidence interval [CI] 1.04 -29.77; p < 0.05). No other bacterial species investigated associated statistically with any systemic disease reported. However, in general the number of these cases was only a few individuals, so the finding on A. actinomycetemcomitans probably has no clinical relevance.
Similarly to the results of the prevalence of reported systemic diseases vs. positive bacterial samples analyses of the associations between the use of drugs and the microbiological results showed that in addition to hormones in the HRT group other medication was rare in the subjects. At baseline the mean number of daily used drugs was 0.7 in all the women while two years later the respective figures were 0.5 drugs in the HRT group and 0.7 drugs in the non-HRT group. At the 2-year examination, women of the non-HRT group reported more often use of neurological drugs than those of the HRT group (16% vs. 5% p < 0.01). In the non-HRT group T. forsythia was detected more often among women not using any systemic drugs (73.7% vs. 26.3%, p < 0.05). Again, however, the number of the women in these subgroups analyzed was small and, subsequently, the findings may not have clinical relevance.
In both groups the main explanatory factor in the model for the occurrence of P. gingivalis, P. intermedia, T. forsythia and T. denticola was the existence of deep peri-odontal pockets. The results of the stepwise logistic regression analyses are given in detail in [fig_ref] Table 2: Odds ratios [/fig_ref].
# Discussion
We observed that the number of T. forsythia-positive samples decreased significantly during the 2-year followup in both study groups while the decrease in the number of P. gingivalis -positive samples was significant only in the HRT users, in particular in women with ≥ 6 mm deep periodontal pockets. A significant decrease was also observed in the P. intermedia-positive samples from women of the HRT group with ≥ 4 mm deep periodontal pockets. No such differences were observed in the non-HRT group. These findings partly support our study hypothesis, namely that HRT use may affect periodontal infections, since a decrease was observed specifically in the number of positive samples for the pathogens T. forsythia and P. gingivalis.
Because our follow-up time was only two years no significant difference was expected in the clinical periodontal status between the women using or not using HRT. As earlier reported this was indeed the case in the present study cohort [bib_ref] Oral health in perimenopausal and early post menopausal women from baseline to..., Tarkkila [/bib_ref]. The strongest explanatory factor for harboring the bacteria investigated was the existence of deep periodontal pockets in both study groups. This result came as no surprise since periodontal pockets and furcation lesions are obviously the characteristic sites harboring periodontal pathogens. Similarly, self-assessed poor oral health status associated with harboring the bacteria investigated but in the present study only the less harmful P. nigrescens -species linked statistically with this parameter. P. nigrescens has been associated with gingivitis [bib_ref] Occurrence of Prevotella intermedia and Prevotella nigrescens in relation to gingivitis and..., Lie [/bib_ref] , but it is also frequently isolated from periodontally healthy sites similar to the P. intermedia species [bib_ref] Bacterial diversity in human subgingival plaque, Paster [/bib_ref]. Since we used pooled plaque samples there are no data from individual periodontal pockets. Our aim was rather to assess the overall effect of HRT on periodontal microbiota than monitoring single pockets in this respect. In the retrospect, it might have been interesting to monitor the periodontal bacteria at individual pockets. Taken together, however, more long-term investigations are needed for confirming the present results and for the final evaluation of the effect of HRT on periodontal micro-organisms.
It is worth a notion that oral micro-organisms have been shown to metabolize in vitro steroid hormones [bib_ref] Metabolism of 17 beta-estradiol by oral Streptococcus mutans, Streptococcus sanguis, Bacillus cereus..., Ojanotko-Harri [/bib_ref] [bib_ref] Targets for steroid hormone mediated actions of periodontal pathogens, cytokines and therapeutic..., Soory [/bib_ref]. The metabolism of progesterone, for example, has been shown to increase in inflamed gingival tissue [bib_ref] A: Metabolism of progesterone by healthy and inflamed human gingiva in vitro, Ojanotko-Harri [/bib_ref]. Of the bacterial species analyzed in our study, T. denticola has been shown to utilize host-derived steroids as growth factors which phenomenon may link to its virulence [bib_ref] The metabolism of cholesterol and certain hormonal steroids by Treponema denticola, Clark [/bib_ref]. Hence, in light of these metabolic findings one might have expected increased T. denticola counts during the 2-year observation which is contrary to our observation. The present results illustrate the complexity of the interactions between steroid hormones, periodontal tissue and micro-organisms. There is a need for more in-depth studies also in this area.
Earlier investigations have shown that in menopause the risk for tooth loss is smaller in women using HRT [bib_ref] Post-menopausal hormone use and tooth loss: a prospective study, Grodstein [/bib_ref]. The circulating levels of estrogen influence alveolar bone density so that more periodontal lesion sites showed loss in alveolar bone density if serum estradiol level was suboptimal [bib_ref] The association between estrogen status and alveolar bone density changes in postmenopausal..., Payne [/bib_ref]. In our study the women using HRT had used it at least for 6 months already in the beginning of the investigation so they were in the steady state regarding hormone balance. However, Pilgram et al. [bib_ref] Relationships between clinical attachment level and spine and hip bone mineral density:..., Pilgram [/bib_ref] investigated 135 women in a randomized controlled trial with estrogen replacement for three years and did not find statistically significant changes in clinical attachment of teeth and bone mineral density of lumbar spine. A study from Spain, on the other hand, showed improvement in periodontal probing depths and tooth mobility in 190 women randomized to receive HRT for one year [bib_ref] Periodontal aspects in menopausal women undergoing hormone replacement therapy, Lopez-Marcos [/bib_ref]. However, there are no systematic studies on the effect of different hormone combinations on periodontal microorganisms and oral health parameters. It can be assumed that different hormone combinations may also affect periodontal tissues and microbiota in a highly specific way.
To our knowledge, the present study was the first follow-up study where periodontal micro-organisms in general have been investigated in clearly defined age groups of peri-menopausal and postmenopausal women. Norderyd and coworkers reported earlier that postmenopausal women taking estrogen supplementation had more frequent gingival bleeding than control group, but they did not find significant differences in the levels of periodontal pathogens examined except for Capnocytophaga -species which are usually implicated in the initiation of puberty gingivitis [bib_ref] Periodontal status of women taking postmenopausal estrogen supplementation, Norderyd [/bib_ref] [bib_ref] Influence of sex hormones on the periodontium. Review Paper, Mascarenhas [/bib_ref] [bib_ref] Microbial changes associated with the development of puberty gingivitis, Mombelli [/bib_ref]. The strength of the present study was the fact that the subjects were originally recruited from a large group of women invited for free mammography examination in age groups of 50-58 years [bib_ref] Oral health in perimenopausal and early post menopausal women from baseline to..., Tarkkila [/bib_ref]. Since the examination was free-of-charge it can be anticipated that selection bias was minimal. However, there is always the possibility that women less interested in their health did not come, but the same holds true for both those using and not using HRT. The weakness of the study was the fairly short 2-year follow-up period. Hence big changes in oral health parameters were not to be expected. For practical reasons the protocol had to be time-restricted and a 2-year longitudinal study was possible as also conducted earlier by Grodstein et al. [bib_ref] Post-menopausal hormone use and tooth loss: a prospective study, Grodstein [/bib_ref]. Another weakness was the drop-out of subjects which was mainly due to the reason that women originally not using HRT had started the therapy and those in the HRT group had ceased using hormones during the 2-years of follow-up. They were then excluded from the statistical analyses.
Finally, the observation that certain periodontal pathogens decreased in the HRT group during the 2-year follow-up might also be due to increased health consciousness of the participants. The women might have improved their oral hygiene habits after oral hygiene information given at the baseline clinical examination. Thus an unavoidable bias caused by trial effect might modify the results in this respect.
# Conclusion
To conclude, we observed a decreased number of P. gingivalis-and T. forsythia-positive samples in women of the HRT group during the 2-year follow-up. The findings partly support our study hypothesis, namely that HRT use may affect periodontal infections. Clinical relevance of the results needs to be assessed in future studies with longer observation time, however.
[fig] Figure 1: Flow-chart of the inclusion of subjects. [/fig]
[fig] Figure 2: Periodontal micro-organisms detected at baseline and two years later in the women with (HRT) or without (non-HRT) hormone replacement therapy. Graph A shows the results of women with the Community Periodontal Index for Treatment Need (CPITN) score 3, indicating the existence of ≥ 4 mm deep periodontal pockets, and graph B of those with CPITN score 4, respectively, indicating ≥ 6 mm periodontal pockets. A.a. = Aggregatibacter actinomycetemcomitans, P.g. = Porphyromonas gingivalis, P.i. = Prevotella intermedia, P.n. = Prevotella nigrescens, T.f. = Tannerella forsythia, T.d. = Treponema denticola. The asterisks show statistically significant differences between baseline and follow-up samples (*p < 0.05, **p < 0.01; #p = 0.07). [/fig]
[table] Table 1: Descriptive data of the study subjects divided in groups according to the use of hormone replacement therapy (HRT). [/table]
[table] Table 2: Odds ratios (OR) with 95% confidence intervals (95% CI) of the statistically significant explanatory factors after backward elimination for the prevalence of periodontal micro-organisms analyzed by stepwise logistic regression.#A.a. = Aggregatibacter actinomycetemcomitans; P.g. = Porphyromonas gingivalis; P.i. = Prevotella intermedia; P.n. = Prevotella nigrescens; T.f. = Tannerella forsythia; T.d. = Treponema denticola.§Periodontal pockets: >4 mm deep pockets; Poor oral health = self-assessed poor oral health; § §HRT = hormone replacement therapy § § §Includes diuretics, antihypertensive agents, nitrates, digitalis and antiarrhythmic agents [/table]
|
10.1155/2021/6622398
|
CCBY
|
8024069
|
33860044
|
s2orc_pubmed_articles
|
Bioelectrical Impedance Analysis and Manual Measurements of Neck Circumference Are Interchangeable, and Declining Neck Circumference Is Related to Presarcopenia
Purpose. Preventive medicine is important in an aging society. Presarcopenia is the preliminary stage of sarcopenia. Recent advances in bioelectrical impedance analysis (BIA) devices have enabled automatic estimation of neck circumference (NC). However, the agreement between and interchangeability of NC measured manually and that calculated with BIA have not been evaluated. We performed these analyses in the context of health checkups and investigated their associations with presarcopenia. Methods. We enrolled 318 participants who underwent anthropometric measurements, including NC measured manually and by BIA; assessment of physical function; and blood testing. We used Bland-Altman analysis to calculate the agreement between and interchangeability of NC measurements by BIA and by the manual method. We then statistically compared normal participants and those with presarcopenia. Using multivariable analysis, we subsequently investigated significant risk factors for presarcopenia. We defined presarcopenia according to the appendicular skeletal muscle index (aSMI; the ratio of arm and leg skeletal muscle mass to height 2 ). Results. Bland-Altman analysis showed that bias (BIA-manual) was negative overall (−1.07), for male participants (−1.23), and for female participants (−0.96). This finding suggests that BIA measurement is an underestimate in comparison with manual measurement. NC measurement by BIA was found to be interchangeable with that by manual methods, inasmuch as the percentage error was less than 5% overall (4.38%), for male participants (3.81%), and for female participants (4.58%). Univariable analysis revealed that NC was significantly smaller in the participants with presarcopenia than in those without. Multivariable analysis, adjusted for confounding factors, revealed that a decrease in NC was significantly correlated with presarcopenia. Conclusions. BIA measurements of NC are interchangeable within about 95% with manual measurements. The decrease in NC measured by BIA was significantly associated with presarcopenia in both genders. NC measurement can be used for early detection of presarcopenia.
# Introduction
Sarcopenia, a syndrome characterized by progressive and generalized loss of skeletal muscle mass and strength, is associated with adverse outcomes such as physical disability, poor quality-of-life (QOL), and death [bib_ref] Sarcopenia: European consensus on definition and diagnosis: report of the European Working..., Cruz-Jentoft [/bib_ref] [bib_ref] Sarcopenia and physical function are associated with inflammation and arteriosclerosis in community-dwelling..., Hida [/bib_ref]. To measure muscle mass, the Asia Working Group for Sarcopenia (AWGS) recommended using the skeletal muscle index, defined by the ratio of appendicular skeletal muscle mass to height squared [bib_ref] Sarcopenia in Asia: consensus report of the Asian working group for sarcopenia, Chen [/bib_ref]. The European Working Group on Sarcopenia in Older People (EWGSOP) defined low muscle mass only as "presarcopenia" [bib_ref] Prevalence and temporal trends of presarcopenia metrics and related body composition measurements..., Li [/bib_ref].
Presarcopenia is the preliminary stage of sarcopenia. To establish better public health policy and devise prevention strategies, the prevalence and temporal trends of presarcopenia and related body composition measurements must be understood in relation to sex, age, and race. Evidence-based prevention strategies for sarcopenia and musculoskeletal diseases must be based on the incidence of disability, risk factors, and other epidemiological data [bib_ref] Incidence and predictors of sarcopenia onset in community-dwelling elderly Japanese women: 4-year..., Kim [/bib_ref]. Few such strategies have been developed, however, and the risk factors for presarcopenia that are associated with anthropometric markers in middle-aged and elderly people remain unclear [bib_ref] Predictors of presarcopenia in community-dwelling older adults: a 5-year longitudinal study, Kobayashi [/bib_ref] [bib_ref] Diagnosis of presarcopenia using body height and arm span for postmenopausal osteoporosis, Ono [/bib_ref].
Neck circumference (NC), an anthropometric marker for detecting risk for metabolic disorders , can also reflect upper body fat deposition and thereby help identify individuals at high risk for these disorders [bib_ref] Neck circumference as a novel measure of cardiometabolic risk: the Framingham Heart..., Preis [/bib_ref]. Increased NC can reflect elevated blood pressure, insulin resistance, lipid abnormalities, and the presence of metabolic syndrome [bib_ref] Relationship of neck circumference to cardiovascular risk factors, Ben-Noun [/bib_ref].
Bioelectrical impedance analysis (BIA), a method of easily measuring water content and body fat mass, is commonly performed during general medical examinations [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref] [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref]. Recent advances in BIA devices have enabled clinicians to quickly estimate not only NC but also the circumferences of the chest, abdomen, and hips [bib_ref] Bioelectrical impedance analysis-part I: review of principles and methods, Kyle [/bib_ref]. Nevertheless, BIA and manual measurements of NC have not been validated, particularly with regard to interrater reliability, agreement, and interchangeability.
Therefore, the purpose of this study was to validate the measurement of NC by both BIA and manual methods and to evaluate whether these two measurements are interchangeable. We also investigated whether NC measurement by BIA is correlated with presarcopenia according to gender during general health checkups in a large, prospective population.
# Materials and methods
2.1. Participants. The study participants were volunteers who underwent health checkups supported by the local government of Yakumo, Japan, in 2019. Of Yakumo's population of approximately 17,000 people, 28% are older than 65 years. A sizable proportion of its residents are engaged in agriculture and fishery. Since 1982, health checkups have been conducted annually in Yakumo; they consist of voluntary orthopedic and physical function examinations, internal medical examinations, and psychological tests, and a health-related QOL survey is administered as well [bib_ref] Staged decrease of physical ability on the locomotive syndrome risk test is..., Imagama [/bib_ref] [bib_ref] Risk factors for neuropathic pain in middle-aged and elderly people: a fiveyear..., Imagama [/bib_ref] [bib_ref] Back muscle strength and spinal mobility are predictors of quality of life..., Imagama [/bib_ref] [bib_ref] Influence of sagittal balance and physical ability associated with exercise on quality..., Imagama [/bib_ref] [bib_ref] Differences of lumbopelvic sagittal parameters among community-dwelling middle-age and elderly individuals: relations..., Machino [/bib_ref]. In our participants, we obtained blood samples; conducted anthropometric measurements, including manual and BIA measurements of NC; and assessed physical function, in that order. On the basis of previous reports, participants underwent these evaluations after overnight fasting [bib_ref] Bioelectrical impedance analysis-part II: utilization in clinical practice, Kyle [/bib_ref].
We excluded from this study subjects with neck masses (e.g., goiter or cervical lymphadenopathy), thyroid diseases, a history of spine and limb joint surgery, severe knee injury, severe osteoarthritis, a history of hip or spine fracture, neurological disorders, severe mental illness, diabetes, kidney or heart disease, and severe disability in walking or standing or any dysfunction of the central or peripheral nervous system. We also excluded data from participants who did not fast before testing. In addition, we excluded patients with sarcopenia (defined as reduced muscle mass and either reduced muscle strength or performance) and severe sarcopenia (defined as reduced muscle mass, reduced strength, and reduced performance) in order to focus on presarcopenia. Of the 537 individuals who underwent the health checkups, 318 participants (125 men and 193 women) met the selection criteria.
The study protocol was approved by the ethics committee of human research and the institutional review board of our university (approval no. 2014-0207). All participants provided written informed consent to participate before the study. The study procedures were carried out in accordance with the principles of the Declaration of Helsinki.
## Anthropometric measurements.
Through BIA, we collected anthropometric data: weight, body mass index (BMI), percentage of body fat (PBF), appendicular skeletal muscle index (aSMI) to represent muscle mass, and NC. We used the InBody 770 body composition and body water analyzer (InBody, Seoul, Republic of Korea), a BIA unit, to differentiate tissues (such as fat, muscle, and bone) according to their electrical impedance [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref] [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref]. Measurements by this device in various contexts have been reported to be accurate [bib_ref] Comparison between dualenergy X-ray absorptiometry and bioelectrical impedance analyses for accuracy in..., Lee [/bib_ref] [bib_ref] Reliability and agreement of various InBody body composition analyzers as compared to..., Mclester [/bib_ref]. Electrodes are embedded in the handles of the analyzer, which were grasped by each participant, and in the platform, on which the soles of the participant's feet rested; two electrodes were in contact with each foot and hand. BMI was calculated as body weight (in kilograms) divided by body height squared (in meters). PBF was calculated as fat mass (in kilograms) divided by body weight ðin kilogramsÞ × 100. The aSMI was calculated as arm and leg skeletal muscle mass (in kilograms) divided by body height squared (in meters) [bib_ref] Appendicular skeletal muscle mass: measurement by dual-photon absorptiometry, Heymsfield [/bib_ref]. NC was calculated twice automatically by the InBody 770 BIA device, and mean data were adopted [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref].
## Definition and diagnosis of presarcopenia.
Presarcopenia is characterized by a reduction in muscle mass that does not affect muscle strength or physical performance, according to the EWGSOP [bib_ref] Sarcopenia: European consensus on definition and diagnosis: report of the European Working..., Cruz-Jentoft [/bib_ref]. The cutoff aSMI values (7.0 kg/m 2 for men and 5.7 kg/m 2 for women, calculated with BIA) were based on the diagnostic criteria described by the AWGS [bib_ref] Sarcopenia in Asia: consensus report of the Asian working group for sarcopenia, Chen [/bib_ref] [bib_ref] Deterioration of presarcopenia and its risk factors following kidney transplantation, Nanmoku [/bib_ref].
## Manual measurements of neck circumference.
While the participants stood with the head positioned in the Frankfort horizontal plane and the shoulders relaxed, we used a nonstretchable plastic tape to manually measure NC from the level just below the laryngeal prominence perpendicular to the long axis of the neck; measurements were performed twice by two independent investigators, and mean data were recorded in centimeters and rounded to the nearest millimeter [bib_ref] Neck circumference as an anthropometric indicator of central obesity in patients with..., Anothaisintawee [/bib_ref] [bib_ref] Neck circumference is associated with adipose tissue content in thigh skeletal muscle..., Tellez [/bib_ref] [bib_ref] Associations between abdominal obesity indices and diabetic complications: Chinese visceral adiposity index..., Wan [/bib_ref]. Intraobserver and interobserver variations regarding the measurement of NC by manual methods were confirmed using the reliability statistics by intraclass correlation coefficient (ICC) by two independent observers, and the mean intra-and interobserver ICCs were 0.94 and 0.92, respectively. As a result, this measurement was considered to be reasonable. [bib_ref] Predictors of presarcopenia in community-dwelling older adults: a 5-year longitudinal study, Kobayashi [/bib_ref] [bib_ref] The relationship between neuropathic pain and spinal alignment: independent risk factors for..., Imagama [/bib_ref]. Participants were in the standing position, and both hands were tested once; the average value was used as the participant's grip strength. To measure back muscle strength (the maximal isometric strength of the trunk muscles), we used a digital back muscle strength meter (T.K.K.5402; Takei Scientific Instruments Co., Niigata, Japan) while participants were in a standing position in 30°o f lumbar flexion [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref] [bib_ref] Influence of global spine sagittal balance and spinal degenerative changes on locomotive..., Machino [/bib_ref]. To evaluate mobility, participants performed two tasks: (1) they walked a straight 10 m course one time at their fastest pace, and the time necessary to complete the course was recorded as the 10 m gait time [bib_ref] Predictors of presarcopenia in community-dwelling older adults: a 5-year longitudinal study, Kobayashi [/bib_ref] [bib_ref] The relationship between neuropathic pain and spinal alignment: independent risk factors for..., Imagama [/bib_ref] , and (2) they rose from a standard chair (46 cm seat height from the ground), walked a distance of 3 m, turned around, walked back to the chair, and sat down (the 3 m timed upand-go test (3-m TUG)), and the time necessary to accomplish this was measured twice, and the mean of the two measurements was recorded [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref] [bib_ref] Influence of global spine sagittal balance and spinal degenerative changes on locomotive..., Machino [/bib_ref].
2.6. Blood Tests. We analyzed venous blood samples from each participant for levels of albumin (a marker of nutritional status), total cholesterol and triglycerides (which can indicate the presence of metabolic syndrome), and C-reactive protein (a marker of inflammation). Biochemical analyses of the blood samples were performed with the use of an autoanalyzer (JCA-RX20; Nihon Denshi, Tokyo, Japan) [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref] [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref].
# Statistical analysis.
All data were analyzed with SPSS statistical software (version 25.0; SPSS Statistics, IBM Corp., Armonk, NY, USA). We calculated continuous variables as means and standard deviations (SDs) and categorical variables as percentages. We used the Mann-Whitney U test and the chi-square test to evaluate between-group differences, as appropriate for the data distribution. We examined correlations between manual and BIA measurements of NC by using the Spearman r and ICC (absolute agreement, two-way random, and single measures). To interpret Spearman correlations, cutoff values of less than 0.20 were considered very weak; 0.20 to 0.39, as weak; 0.40 to 0.59, as moderate; 0.60 to 0.79, as strong; and 0.80 to 1.0, as very strong [bib_ref] Neck circumference is associated with adipose tissue content in thigh skeletal muscle..., Tellez [/bib_ref] [bib_ref] User's guide to correlation coefficients, Akoglu [/bib_ref]. To interpret the ICC, cutoff values of less than 0.20 were considered slight; 0.20 to 0.39, as fair; 0.40 to 0.59, as moderate; 0.60 to 0.79, as substantial; and 0.80 to 1.0, as almost perfect [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref]. To examine the level of agreement between the manual and BIA measurements, we used the Bland-Altman analysis [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref]. The mean of the difference between measurements (BIA versus manual) was defined as bias, and SD was also used to calculate 95% confidence limits of agreement (bias: ±1.96 SD). The Bland-Altman plots graphically displayed the mean of the two measured values (NC measurement with BIA and the manual method) on the x-axis and the difference (BIA versus manual) between measured values on the y-axis. To determine whether BIA measurements were interchangeable with manual measurements, we used a percentage error (the ratio of 1.96 SD to the mean value of the manual method) as 8% or less [bib_ref] Statistical methods for assessing agreement between two methods of clinical measurement, Bland [/bib_ref] [bib_ref] A meta-analysis of studies using bias and precision statistics to compare cardiac..., Critchley [/bib_ref]. To determine the factors associated with presarcopenia among the variables that exhibited differences (p < 0:01) in the univariable analyses, logistic regression analysis using a stepwise method was performed using the aforementioned variables as covariables. A p value of < 0.05 was considered significant in all analyses.
# Results
The average age of the 318 participants was 63.4 years (range, 40-87 years; SD, 10.0 years), the average BMI was 23.7 kg/m 2 , and the average PBF was 29.4%. [fig_ref] Table 1: Demographic, anthropometric, physical function, blood test, and presarcopenia prevalence data of the... [/fig_ref] lists the demographic data, anthropometric measurements, physical function results, blood test results, and presarcopenia prevalence. Men and women exhibited significant differences for all variables except blood test data. Both manual and BIA measurements of NC were smaller in women than in men. The prevalence of presarcopenia was higher in women than in men. [fig_ref] Table 2: Correlation coefficient [/fig_ref] lists Spearman r, ICC, and Bland-Altman analysis results for the NCs as measured by the two methods. BIA measurements were found to be interchangeable with manual measurements, inasmuch as the percentage error was less than 5% (4.38% overall, 3.81% for men, and 4.58% for women). The Bland-Altman plots showed that BIA measurements of NC were in almost perfect agreement with manual measurements, according to ICC [fig_ref] Table 2: Correlation coefficient [/fig_ref].
BIA measurements of NC were very strongly correlated manual measurements of NC overall (Spearman r = 0:90, p < 0:0001), for men (Spearman r = 0:81, p < 0:0001), and for women (Spearman r = 0:83, p < 0:0001; . The Bland-Altman analysis showed that bias (BIA versus manual) was negative overall (−1.07), for men (−1.23), and for women (−0.96), which suggested that BIA measurements were underestimates in comparison with manual measurements . [fig_ref] Table 3: Comparison between the normal group and the presarcopenia group according to sex [/fig_ref] lists the results of the comparisons between the normal participants and those with presarcopenia, by gender. Overall, all variables differed significantly between the two groups, except for age and blood test data. Both men and women with presarcopenia had lower body weights and smaller NCs according to manual and BIA measurements than did the normal participants. The physical function of both men and women with presarcopenia was inferior to that of the normal men and women [fig_ref] Table 3: Comparison between the normal group and the presarcopenia group according to sex [/fig_ref].
The results of the logistic regression model for presarcopenia in all participants are listed in [fig_ref] Table 4: Logistic regression model for presarcopenia in all the participants [/fig_ref]. BIA measurement of NC (p < 0:001), body weight (p < 0:001), grip strength (p < 0:001), back muscle strength (p < 0:001), PBF (p = 0:003), and female gender (p = 0:025) were significantly associated with presarcopenia. The results of the logistic regression model according to sex are listed in [fig_ref] Table 5: Logistic regression model for presarcopenia according to sex [/fig_ref]. BIA measurement of NC was significantly associated with presarcopenia in both men and women.
BIA measurement of NC was significantly and very strongly positively correlated with aSMI overall (r = 0:89, p < 0:0001), for men (r = 0:80, p < 0:0001), and for women (r = 0:80, p < 0:0001; [fig_ref] Figure 3: Scatterplot of neck circumference [/fig_ref]. From these results, we found that the decline in BIA measurements of NC was significantly associated with presarcopenia, which indicates a decrease in aSMI.
# Discussion
Sarcopenia is an important barometer of disability and frailty in elderly people, inasmuch as it exacerbates poor general health or frailty [bib_ref] Sarcopenia and physical function are associated with inflammation and arteriosclerosis in community-dwelling..., Hida [/bib_ref] [bib_ref] Incidence and predictors of sarcopenia onset in community-dwelling elderly Japanese women: 4-year..., Kim [/bib_ref]. Presarcopenia is the preliminary stage of sarcopenia [bib_ref] Predictors of presarcopenia in community-dwelling older adults: a 5-year longitudinal study, Kobayashi [/bib_ref]. The causes of sarcopenia are varied and complex; they include disuse of muscles as a result of malnutrition, vitamin D deficiency, cerebral infarction, heart failure, and osteoarthritis; age-related changes in levels of hormones such as testosterone, estrogen, insulin-like growth factor 1, and insulin; apoptosis, denervation, inflammation, and changes in immunity involving interleukin-(IL-) 1, IL-6, and tumor necrosis factor-α; and social and mental causes, such as decline in cognitive function or decreased social activity [bib_ref] Role of apoptosis in sarcopenia, Leeuwenburgh [/bib_ref] [bib_ref] Apoptosis-inducing factor regulates skeletal muscle progenitor cell number and muscle phenotype, Armand [/bib_ref] [bib_ref] Cytokines, insulin-like growth factor 1, sarcopenia, and mortality in very old community-dwelling..., Roubenoff [/bib_ref] [bib_ref] Myosteatosis and myofibrosis: relationship with aging, inflammation and insulin resistance, Zoico [/bib_ref] [bib_ref] Proinflammatory cytokines, aging, and age-related diseases, Michaud [/bib_ref]. Loss of appendicular skeletal muscle mass is a predictor of mortality in elderly people [bib_ref] Sarcopenia in Asia: consensus report of the Asian working group for sarcopenia, Chen [/bib_ref] [bib_ref] Ultrasound measurement of thigh muscle thickness for assessment of sarcopenia, Hida [/bib_ref]. Intervention at the presarcopenia stage is necessary to prevent sarcopenia. Therefore, indicators of presarcopenia must be established.
Interest in NC as a clinical measurement has increased substantially [bib_ref] Neck circumference is associated with adipose tissue content in thigh skeletal muscle..., Tellez [/bib_ref]. Unlike waist circumference, the measurement of which differs widely in location, and the location of NC measurement is standard and straightforward. Meals, respiration, and position have no effect on NC measurement [bib_ref] Neck circumference as an anthropometric indicator of central obesity in patients with..., Anothaisintawee [/bib_ref]. NC can be measured without the need to remove clothing and can also be measured in pregnant and ascitic patients [bib_ref] Neck circumference is associated with adipose tissue content in thigh skeletal muscle..., Tellez [/bib_ref].
In general, NC measurement is performed manually. In large-scale health checkups, manual measurements of many patients within a limited time may lead to interrater error because of the increasing numbers of clinicians who perform the measurements. To solve such problems, a device that can be used to perform automatic measurements in a short time must yield measurements that correspond to manual measurements. Portable BIA devices have been used to measure body composition and anthropometric markers [bib_ref] Bioelectrical impedance analysis-part II: utilization in clinical practice, Kyle [/bib_ref]. Clinicians also use them in the context of health checkups. Because BIA can account for a large amount of data from one measurement in a short period of time, the labor, time, and interrater errors involved are reduced in comparison with those associated with manual measurement [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref]. As a result of technological advances, BIA devices that can BioMed Research International automatically measure anatomical circumferences, including NC, have been developed [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref]. Nevertheless, BIA has not been validated with regard to manual measurement values.
Our study was the first in which these two methods of measuring NC were validated in a large-scale prospective population of patients undergoing health checkups. Moreover, BIA devices have been used to measure PBF and aSMI in various studies, and so we hypothesized that additional evidence could be accumulated [bib_ref] Comparison between dualenergy X-ray absorptiometry and bioelectrical impedance analyses for accuracy in..., Lee [/bib_ref] [bib_ref] Reliability and agreement of various InBody body composition analyzers as compared to..., Mclester [/bib_ref]. aSMI is a well-accepted measure for the screening of presarcopenia [bib_ref] Predictors of presarcopenia in community-dwelling older adults: a 5-year longitudinal study, Kobayashi [/bib_ref]. The use of BIA as a noninvasive and easy-to-use method for evaluation of sarcopenia has increased [bib_ref] Sarcopenia and physical function are associated with inflammation and arteriosclerosis in community-dwelling..., Hida [/bib_ref]. Our results also demonstrated that BIA measurement of NC was independently associated with aSMI.
With regard to accuracy, BIA and manual measurements of NC must be in agreement because manual measurement can introduce errors. If the agreement is high, it is possible that the measurement modalities are interchangeable. In general, Spearman correlations and ICC are used to validate the values obtained with the two measurement methods. However, correlation analysis concerns the relationship between two different events, and the value of the correlation coefficient cannot reveal differences between or variations in measurements [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref]. Therefore, these assessments alone are insufficient for evaluating interchangeability. The Bland-Altman analysis, a method of measuring two methods and examining the differences between the measurements, is also necessary. We evaluated the interchangeability of BIA and manual measurements on the basis of past reports [bib_ref] Sarcopenia and physical function are associated with inflammation and arteriosclerosis in community-dwelling..., Hida [/bib_ref] [bib_ref] Waist circumference measured by bioelectrical impedance analysis is interchangeable with manual measurement:..., Tanaka [/bib_ref]. For men and women, BIA and manual measurements of NC were strongly correlated and thus demonstrated the interchangeability of the two methods. This result is very Furthermore, because we confirmed that BIA measurement of NC is interchangeable with manual measurement, we examined the results of a previous report in which NC was not measured with BIA and presarcopenia as the preliminary stage of sarcopenia was not considered [bib_ref] Declining neck circumference is an anthropometric marker related to frailty in middle-aged..., Tanaka [/bib_ref]. The univariable analysis finding in this study was that BIA measurements of NC were significantly smaller in participants with the presarcopenia group overall and by gender. Moreover, even if statistical adjustments were made for other factors in logistic regression analysis, a decline in NC as measured with BIA was a significant risk factor for presarcopenia. This study allowed us to build upon the previous report and accumulate further evidence. Declining NC will serve as a predictive factor for presarcopenia in the future, and it may be used to evaluate therapeutic effect. The results of this study indicate that additional research on sarcopenia is warranted.
This study had some limitations. First, data from only one race of people in a single center were analyzed, and their demographic characteristics did not reflect those of people dwelling in the general community. The subjects were healthy middle-aged and elderly people who lived in a relatively rural area, and many had jobs in agriculture or fishing; therefore, they did not represent people living in an urban environment [bib_ref] The relationship between neuropathic pain and spinal alignment: independent risk factors for..., Imagama [/bib_ref] [bib_ref] Influence of global spine sagittal balance and spinal degenerative changes on locomotive..., Machino [/bib_ref]. Second, this study was a crosssectional study. A longitudinal larger-scale study is necessary to identify the causes of presarcopenia. Third, BIA devices from different manufacturers may yield different measurements. Therefore, standardization of technology and cross- BioMed Research International calibration of electrical resistance should be addressed. It is necessary to further increase the number of participants and verify the justification in future studies. The progress of BIA technology has been remarkable; depending on the device, it is possible to measure NC, and a large amount of data can be obtained in a short time from one measurement. The results of this study indicate that the relationship between NC and various diseases and conditions might be investigated on a large scale. We hope that the simple measurement of NC will play an important role in screening for presarcopenia and that early intervention in the form of muscle-strengthening exercises may prevent the conversion to sarcopenia.
# Conclusion
BIA measurements of NC are interchangeable within about 95% with manual measurements. The decline in NC as measured by BIA was significantly associated with presarcopenia in both genders. NC measurement may play a role in the early detection of presarcopenia.
## Data availability
The cohort data used to support the findings of this study are restricted by the Institutional Review Board of Nagoya Uni-versity Graduate School of Medicine in order to protect the privacy of subjects in the Yakumo study.
# Ethical approval
The study protocol was approved by the Institutional Review Board of Nagoya University Graduate School of Medicine. Moreover, the study protocol was approved by the Committee on Ethics in Human Research of our university, and the study procedures were carried out in accordance with the principles of the Declaration of Helsinki.
## Consent
All participants provided written informed consent.
## Conflicts of interest
The authors report no conflict of interest except for the national grant.
[fig] Figure 1, Figure 2: Scatterplot of neck circumference (NC) by bioelectrical impedance analysis (BIA) and by the manual method. Measurement of NC by BIA was significantly and very strongly positively correlated with that by the manual method. (a) Total (r = 0:90; p < 0:0001), (b) in male participants (r = 0:81; p < 0:0001), and (c) in female participants (r = 0:83; p < 0:0001). Mean of BIA and manual (cm) Difference (BIA-manual) Bland-Altman plot of difference in neck circumference (NC)-measurement with bioelectrical impedance analysis minus manual measurement-against the mean of two measurements. The middle line denotes bias (mean difference between the two measurements), and the dashed lines denote 95% limits of agreement (1.96 standard deviation from the difference). (a) Total (bias: −1.07; 95% limits of agreement (LOA): −4.02 to 1.89), (b) in male participants (bias: −1.23; 95% LOA: −3.96 to 1.50), and (c) in female participants (bias: −0.96; 95% LOA: and in future research, investigators can focus on the use of BIA not only for NC but also for other anatomical circumferences. [/fig]
[fig] Figure 3: Scatterplot of neck circumference (NC) by bioelectrical impedance analysis (BIA) and by appendicular skeletal muscle index (aSMI). Measurement of NC by BIA was significantly and very strongly positively correlated with that by aSMI. (a) Total (r = 0:89; p < 0:0001), (b) in male participants (r = 0:80; p < 0:0001), and (c) in female participants (r = 0:80; p < 0:0001). [/fig]
[table] Table 1: Demographic, anthropometric, physical function, blood test, and presarcopenia prevalence data of the study participants.Evaluated using the Mann-Whitney U test and chi-square test. Parameter values are shown as means (standard deviations) or numbers. Bold values indicate significant difference. PBF, aSMI, and NC by BIA were measured using InBody 770 BIA unit. BMI: body mass index; PBF: percent body fat; aSMI: appendicular skeletal muscle index; NC: neck circumference; BIA: bioelectrical impedance analysis; TUG: timed up-and-go; CRP: C-reactive protein. [/table]
[table] Table 2: Correlation coefficient (r), ICC, and Bland-Altman analysis in NC measured by two methods: manual and BIA. NC: neck circumference; BIA: bioelectrical impedance analysis; SD: standard deviation; LOA: limits of agreement. [/table]
[table] Table 3: Comparison between the normal group and the presarcopenia group according to sex. Evaluated using the Mann-Whitney U test and chi-square test. Parameter values are shown as means (standard deviations). Bold values indicate significant difference. BMI: body mass index; PBF: percent body fat; aSMI: appendicular skeletal muscle index; NC: neck circumference; BIA: bioelectrical impedance analysis; TUG: timed up-and-go; CRP: C-reactive protein. [/table]
[table] Table 4: Logistic regression model for presarcopenia in all the participants.All variables (p < 0:01) that showed a certain degree of difference in univariate analysis were used as covariates. The dependent variable was presarcopenia. Covariates were age, sex, body weight, PBF, NC by BIA, grip strength, and back muscle strength. Bold values indicate significant difference. β: partial regression coefficient; CI: confidence intervals; NC: neck circumference; BIA: bioelectrical impedance analysis; PBF: percent body fat. [/table]
[table] Table 5: Logistic regression model for presarcopenia according to sex.All variables (p < 0:01) that showed a certain degree of difference in univariate analysis were used as covariates. The dependent variable was presarcopenia. Covariates were body weight, PBF, NC by BIA, grip strength, and back muscle strength. Italicized values indicate significant difference. β: partial regression coefficient; CI: confidence intervals; NC: neck circumference; BIA: bioelectrical impedance analysis; PBF: percent body fat. [/table]
|
10.3390/ijerph20146386
|
CCBY
|
10380075
|
37510618
|
s2orc_pubmed_articles
|
Exploring the Synergy of Music and Medicine in Healthcare: Expert Insights into the Curative and Societal Role of the Relationship between Music and Medicine
Citation: Hennenberg, J.; Hecking, M.; Sterz, F.; Hassemer, S.; Kropiunigg, U.; Debus, S.; Stastka, K.; Löffler-Stastka, H. Exploring the Synergy of Music and Medicine in Healthcare: Expert Insights into the Curative and Societal Role of the Relationship between Music and Medicine. Int. J. Environ. Res. PublicHealth 2023, 20, 6386. https:// Abstract: Our study aimed to investigate the correlation between medicine, health perception, and music as well as the role of music in the healthcare setting. To gain insights into the dynamics between these two fields, we gathered opinions from attendees and presenters at an international conference on music medicine, musicians' health, and music therapy. A team of six interviewers conducted a total of 26 semi-structured interviews. The interview guide focused on four predetermined themes: (1) "music in medicine", (2) "performing arts medicine", (3) "music for the individual", and (4) "music for society". The responses were analyzed using grounded theory methods as well as thematic and content analysis. To enhance the analytical strength, investigator triangulation was employed. Within the predefined themes, we identified several subthemes. Theme 1 encompassed topics such as "listening and performing music for treating diseases and establishing non-verbal relationships", "the value of music in specific disorders, end-of-life care, and pain management", and "the design of sound spaces". Theme 2 explored aspects including the "denial and taboo surrounding physical and mental health issues among musicians", "the importance of prevention", and an antithesis: "pain and suffering driving creativity". Theme 3 addressed the "mental role of music in ordinary and extraordinary life" as well as "music's ability to enable self-conditioning". Lastly, Theme 4 examined the role of music in "cultural self-identification" and "development and education for children". Throughout the interviews, participants expressed a lack of knowledge and awareness regarding interdisciplinary research and the fields of music and medicine. Our findings affirm the significance of music therapy and performing arts medicine as well as the broader relationship between music and medicine. They highlight the potential benefits of perception and experiential pathways for individuals and, consequently, for human society.
# Introduction
Throughout the history of medicine, music has played a prominent and intertwined role. For instance, the Assyrians in 2000 BC documented the use of music to counteract the influence of evil spirits [bib_ref] Music for healing: From magic to medicine, Conrad [/bib_ref]. In ancient Greece, Apollo was venerated as the god of both music and healing [bib_ref] Plato's attitude to poetry and the fine arts, and the origins of..., Leszl [/bib_ref]. Plato eloquently expounded on the salutary effects of music on health in his Socratic dialogue "Politeia" [bib_ref] Music, medicine, healing, and the genome project, Meymandi [/bib_ref]. Additionally, the Old Testament recounts King Saul finding solace from an "evil spirit" through the soothing sounds of David's lyre. Music's significance persisted through the Middle Ages, as it formed a fundamental part of medical education in Europe [bib_ref] The Evolution of Performing Arts Medicine, Harman [/bib_ref]. In the late 19th century, recognition of music's therapeutic potential led to extensive research endeavors [bib_ref] The Economic Impact of Music in Europe, Moore [/bib_ref] and its integration into medical practice.
This deep integration of music within medicine gave rise to performing arts medicine, an occupational medicine discipline that addresses the etiology, treatment, and prevention of health issues encountered by performing artists [bib_ref] Grounded theory research: A design framework for novice researchers, Chun Tie [/bib_ref]. As early as the 18th century, the physiological impact of music-making was observed, with 19th-century publications detailing injuries among keyboard and violin players as well as the occurrence of musician's cramp (focal dystonia) [bib_ref] Overview of qualitative research, Grossoehme [/bib_ref]. Notably, performing arts medicine extends beyond empathy, as music contributed a substantial 81.9 billion EUR to the European Union's gross domestic product in 2018 [bib_ref] Qualitative Content Analysis, Mayring [/bib_ref].
Moreover, the connection between music and medicine transcends mere curative aspects, as medicine can permeate the realm of music through the lens of health perception for the listener. The experience of music from performers or composers who may exhibit health problems or push the boundaries of their physical well-being differs from attending a performance where musicians appear to be in good health. Throughout the centuries and across genres, numerous renowned artists have openly shared their struggles with health issues and mental challenges while projecting an image of perfect health and ideal aging during their performances, an aspect intricately interwoven with the overall impact of their artistry.
As authors of this analysis, we acknowledge that we do not claim expertise in this relationship solely based on formal training. Rather, our objective was to deepen our understanding of this connection, exploring the potential value of music in medicine and the reciprocal influence of medicine and overall health on music creation and experiences. To achieve this, we conducted a qualitative interview study at the "Science and Sounds" international conference, dedicated to "music medicine, musicians' health, and music therapy". Our aim was to utilize a thematic analysis of expert opinions to address the current knowledge gap regarding the intricate relationship between music and medicine, while identifying potential areas for future research and funding.
# Materials and methods
## Conference details and study participants
The "Science and Sounds" international conference was held 8-10 September 2022, in Hamburg, Germany. It was a joint project between the University of Music and Drama, Hamburg, and the University Medical Center, Hamburg-Eppendorf, Germany, and was conducted in collaboration with the International Society for Music in Medicine [bib_ref] Zielgeführte Evaluation von Programmen-Ein Leitfaden, Beywl [/bib_ref]. The main presentations were given by international experts from the fields of musicians' health, music medicine, and music therapy. The main focus of the conference was on the domain of music medicine and music therapy, with prominent international experts in the field [bib_ref] Consolidated criteria for reporting qualitative research (COREQ): A 32-item checklist for interviews..., Tong [/bib_ref]. Invited presenters and attendants of this conference were randomly invited (i.e., without specific selection criteria) to be interviewed for the present study. No repeat interviews were conducted.
## Ethical considerations, approval, and consent to participate
This study was carried out in accordance with relevant national and international guidelines and regulations. The Ethics Committee of the Medical University of Vienna determined that ethics approval was not required according to national regulations, and due to the fact that the relationship between interviewers and interviewees was one of peers rather than, for example, a relationship involving patient-doctor dependence (communication with the Ethics Committee will be made available upon request). All study participants gave their written informed consent to participate in this study, and all interviews were anonymized before the analysis.
## Investigators
The interviewers comprised six co-authors of the present article (J.H., S.H., K.S., H.L.-S., F.S. and M.H.). Members of this team ranged from 26 to 66 years old, and included four men and two women (self-declared), who conducted all but one interview on a one-to-one basis with the interviewees.
## Interview setting
The "Science and Sounds" conference took place at two venues: the University of Music and Drama, Hamburg [bib_ref] Mozart, music and medicine, Pauwels [/bib_ref] , one of the larger universities of music in Germany; and the University Medical Center, Hamburg-Eppendorf. Semi-structured face-to-face single and group interviews were predominantly scheduled during the intermissions of the "Science and Sounds" conference, and were conducted at the University of Music and Drama and the University Clinic. In specific, group interviews in this setting allowed for the exploration of shared perspectives, facilitated dynamic discussions, and stimulated collective thinking. However, the presence of a group setting may have introduced social influence or conformity, potentially influencing participants' responses and limiting the expression of individual viewpoints. None of the investigators had previous familiarity with the meeting venue, and the interviews were conducted in the most convenient remote areas of the venue building.
## Interview guide
A semi-structured interview guide (see Appendix B) was developed with consensus of the researchers. Specifically, authors SH and MH made independent suggestions, which were discussed, adopted, and finalized among the group of six interviewers. To address potential bias stemming from rigid interview structures, participants agreed to employ a flexible interview outline that could be adapted based on their responses. This approach aimed to promote a more organic and nuanced exploration of the topic. Additionally, establishing rapport and considering diverse perspectives within the research team helped ensure a comprehensive and unbiased analysis of the collected data. German was the preferred interview language, although English was used when necessary. Investigators initially introduced themselves to the participants, and gave a short explanation of the research objectives. After asking the participants for some central demographic information, the main aim was to explore possible interdependencies of music and medicine, in four pre-specified themes: (1) "music in medicine", (2) "performing arts medicine", (3) "music for the individual", and (4) "music for society". Participants were encouraged to openly express their views, and anonymity was guaranteed in all aspects. To gain a comprehensive understanding, valuable insights of the interviewees' personal experiences were obtained by asking, for example, about their personal limitations. This was conducted with full respect for the interviewees' autonomy and comfort, as disclosure of such limitations during the interviews was voluntary and not compulsory. The decision to forgo the typical development of the semi-structured interview guide within focus groups was driven by participant availability, and a desire to capture diverse individual perspectives more comprehensively, and individual interviews were chosen as an alternative approach to ensure efficient data collection and optimize the depth of understanding.
## Recruitment goal, data transcription and translation, and analysis
Our goal was to recruit up to 30 participants, depending on how soon data saturation would be reached. The interviewers discussed whether saturation had been reached three times between the interview rounds. At the third discussion time-points, when 26 interviews had been completed, the co-authors (particularly J.H., H.L.-S., and F.S.) voiced that the responses they were obtaining were increasingly overlapping and that it would be sensible to refrain from recruiting more interviewees. No specific person was assigned to recruitment. Instead, the congress president was asked to announce, introduce, and spread word-of-mouth information of our plans at the beginning of the conference sessions. No willing participant was excluded. Interviews were audio-recorded and transcribed using basic standard transcription systems supplying dialogue lists and documents [bib_ref] Anatomically distinct dopamine release during anticipation and experience of peak emotion to..., Salimpoor [/bib_ref]. The study investigators performed translation into English.
While striving for data saturation and diverse perspectives, the recruitment goal of up to 30 participants may have introduced complexities that necessitate meticulous attention, such as the possibility of variations in participant selection and representation. It is crucial to acknowledge and address these factors in the discussion section, as they can potentially influence the interpretation of the findings and the generalizability of the study results.
Seeking data saturation was essential to ensure a comprehensive understanding of the research topic, as it signifies that no new information or themes were emerging from additional interviews or data sources, enhancing the credibility and robustness of the study findings.
# Data analysis
The analytical coding process used in this study followed a mixed method approach, applying procedural rules from grounded theory [bib_ref] Music-selective neural populations arise without musical training, Boebinger [/bib_ref] [bib_ref] The rewards of music listening: Response and physiological connectivity of the mesolimbic..., Menon [/bib_ref] , as well as principles from qualitative content and thematic analysis [bib_ref] Familiarity mediates the relationship between emotional arousal and pleasure during music listening, Van Den Bosch [/bib_ref]. After author UK held an instruction session, the anonymized interview transcripts were thoroughly coded by authors J.H., F.S., and M.H., and were then assigned to subthemes within four pre-specified themes using a text-sorting technique (TST) for qualitative data analysis [bib_ref] Musical Breaks-Live Music in a Hemodialysis Setting-A Qualitative Study on Patient, Nurse,..., Bro [/bib_ref] , based on the Microsoft Word program. Authors J.H., F.S., and M.H. worked individually on these subthemes and arranged four joint coding sessions in which they verified that all subthemes represented essential patterns and relationships within the data set. Next, two coder triangulation sessions were arranged with author UK, as well as a support team triangulation session with authors S.H., H.L.-S. and S.D. The Consolidated Criteria for Reporting Qualitative Research (COREQ) was utilized to enhance the transparency, rigor, and comprehensive reporting of qualitative research methods and findings in this study [bib_ref] Effect of music intervention during hemodialysis: A comprehensive meta-analysis, Wu [/bib_ref].
# Results
# Descriptive results
Here, we present the results from 26 interviews conducted with participants of the "Science and Sounds" international conference held in Germany in September 2022 [bib_ref] Effect of instrumental music on anxiety and depression among hemodialysis patients: A..., Imani [/bib_ref]. In one-to-one interviews, we asked open questions. The responses revealed that additional resources are needed to raise awareness for and to support music therapy as well as to address musicians' medical needs. Analysis of the responses also revealed issues related to the use of music in medicine and, vice versa, to diseases related to performing music. [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] shows the demographics and personal characteristics of the 26 interviewed study participants. Among the participants, 18 self-identified as women and 6 as men; 23 resided in Germany; 22 stated they were of Caucasian ethnicity; and 20 indicated that their native language was German. Most of the participants (n = 10) were 21-30 years of age. While 4 interviewees did not self-identify as musicians, 6 reported that they spent >10 h per week with active music making, 13 reported 2-10 h per week, and 3 reported <2 h per week. Six interviewees worked as music therapists, six were students or in training, and four were medical professionals (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] for further details). The collection of participants was made in an effort to encompass a diverse cohort and obtain a range of perspectives. ≤2 years 0 Not asked No active music making Table A1 (see Appendix A) presents the themes and subthemes. The four themes were pre-specified based on our interview guide (see Section 2 and Appendix B). Our thematic analysis (carried out by authors J.H., F.S. and M.H.) revealed up to six subthemes for each theme. Some of the identified subthemes re-appeared throughout the four themes. For example, participants expressed that they experienced a popular lack of knowledge and awareness of the impact of music in the medical field (e.g., within theme 1, e.g., subtheme 6: They are unaware that music therapy exists) and of the importance of medicine for music and musicians (within theme 2, e.g., subtheme 2: I still find it frightening how other instrumentalists, in particular, treat their bodies). Likewise, subtheme 2 from theme 3 ("mental role of music in ordinary and extraordinary life") was somewhat similar to subtheme 2 from theme 4 ("music for cultural self-identification") and subtheme 5 from theme 1 ("overcoming loneliness").
When we further condensed all subthemes from the four pre-specified themes, we found that music itself was perceived as a common language outside of human speech, rooted in childhood, with importance for cultural self-identification, individual development, and education. Participants emphasized that music reduces stress and enables self-conditioning, albeit with variations in form and activity (active to passive). Two participants stated that music had to be "politically correct" (subtheme 5, theme 4). Interviewees also stated that the act of listening and/or performing aids in medical treatment, establishes a non-verbal relationship with the caretaker, can be applied in various (physical) environments (e.g., see subtheme 4, theme 1), and has special importance at the end of life. Based on their experience, interviewees stated that taking advantage of medical treatment was a taboo for most musicians. Nevertheless, the role of medicine for performers was considered important for the prevention of serious illness, while requiring interaction between multiple medical disciplines. As an interesting antithesis, at least two participants stated that medical treatment ran the risk of counteracting artistic development, based on the notion that creativity came from suffering (subtheme 5, theme 2: The absence of medicine as a creative boost; Illness and, above all, pain makes the "I" awaken to itself. Pain is a great awareness maker. Or also the crisis. In this respect, it is also a motor for creativity, without there being an inevitable link between disorder and artistic creation. It is a drive.).
# Discussion
The findings were divided into four pre-specified themes, based on our interview guide (see Appendix B), which will be discussed in greater detail below. Within those four themes, we further identified 21 subthemes. All things considered, our interviewees voiced that there is an enormous demand for physicians and musicians to make more and better use of their corporate influence, not only for their own benefit but also to promote further growth of society in the field of music-medicine.
## Theme 1: music in medicine
Regarding the first pre-specified theme, "music in medicine", our study participants unanimously stated that both listening to music and performing music in a medical setting can benefit a patient's progress in treatment (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subthemes 1-2, theme 1). The act of making music together was described as helpful for establishing a personal relationship between a patient and their therapist or physician-enabling patients to "open up", and fostering their emotional stability with different perceptions through experiential learning and new learning pathways. Interviewees also voiced that a "pleasant environment" supports the development of the patient's relationship with a therapist and/or caretaker (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 1, theme 1). These opinion-based subthemes are supported by research findings. Similar to rewards that activate the striatal dopaminergic system, intangible stimuli, such as music, can cause euphoria, cravings, and emotional arousal [bib_ref] Effectiveness of Music to Improve Anxiety in Hemodialysis Patients: A Systematic Review..., Burrai [/bib_ref] [bib_ref] Music-based interventions for pain relief in patients undergoing hemodialysis: A PRISMA-compliant systematic..., Cheng [/bib_ref] [bib_ref] Individualised music for people living with dementia and the experiences and perceptions..., Gaviola [/bib_ref] [bib_ref] What carers and family said about music therapy on behaviours of older..., Tuckett [/bib_ref]. Such processes can be measured with functional magnetic resonance imaging, and do not depend on the individual having prior musical training [bib_ref] Music-based interventions for pain relief in patients undergoing hemodialysis: A PRISMA-compliant systematic..., Cheng [/bib_ref] [bib_ref] Individualised music for people living with dementia and the experiences and perceptions..., Gaviola [/bib_ref].
During their interviews, our participants mentioned various specific medical fields in which the presence of music was particularly helpful in patients' recuperation (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 2-5, theme 1): [ . . . ] Pregnant women, women who may have had a miscarriage and are now afraid to get pregnant again or who already are, premature babies, children in orthopedics, in oncology, patients with eating disorders, people in the very last phase of life, in palliative care and until the day they die. Overall, music intervention was said to be associated with an improvement in patients' health-related quality of life (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subthemes 1-3, theme 1). These statements are in line with numerous research studies showing that music intervention in various medical settings is a safe and affordable method of improving the course of a disease. Studies have shown that music improves patients' physical and psychological outcomes during treatment, e.g., for hemodialysis [bib_ref] Music-based interventions in neurological rehabilitation, Sihvonen [/bib_ref] [bib_ref] Music listening after stroke: Beneficial effects and potential neural mechanisms, Särkämö [/bib_ref] [bib_ref] Reduction of seizure occurrence from exposure to auditory stimulation in individuals with..., Bodner [/bib_ref] [bib_ref] Music in epilepsy: Predicting the effects of the unpredictable, Rafiee [/bib_ref] ; in the hospice setting [bib_ref] Music therapy for depression, Aalbers [/bib_ref] [bib_ref] Effects of music therapy on depression: A meta-analysis of randomized controlled trials, Tang [/bib_ref] [bib_ref] Music interventions for improving psychological and physical outcomes in cancer patients, Bradt [/bib_ref] [bib_ref] The effectiveness of music therapy for patients with cancer: A systematic review..., Li [/bib_ref] ; in the fields of neurology [bib_ref] Effectiveness of Music to Improve Anxiety in Hemodialysis Patients: A Systematic Review..., Burrai [/bib_ref] [bib_ref] The Impact of Music Therapy on Anxiety in Cancer Patients Undergoing Simulation..., Rossetti [/bib_ref] [bib_ref] Music interventions can alleviate cancer-related fatigue: A metaanalysis. Support. Care Cancer Off, Qi [/bib_ref] [bib_ref] The effects of a music and singing intervention during pregnancy on maternal..., Wulff [/bib_ref] [bib_ref] Singing for respiratory health: Theory, evidence and challenges, Gick [/bib_ref] [bib_ref] Use of Singing for Lung Health as an alternative training modality within..., Kaasgaard [/bib_ref] , psychiatry [bib_ref] Singing for Lung Health" as a therapy for individuals with chronic obstructive..., Lewis [/bib_ref] [bib_ref] Meta-analysis evaluating music interventions for anxiety and pain in surgery, Kühlmann [/bib_ref] , and oncology [bib_ref] Music as an aid for postoperative recovery in adults: A systematic review..., Hole [/bib_ref] [bib_ref] Music interventions for preoperative anxiety, Bradt [/bib_ref] [bib_ref] The impact of perioperative music on abdominal surgery patients' experience of postoperative..., Dale [/bib_ref] [bib_ref] The effect of music on postoperative recovery in older patients: A systematic..., Van Der Wal-Huisman [/bib_ref] ; during pregnancy [bib_ref] The benefits of perioperative music interventions for patients undergoing neurosurgery: A mixed-methods..., Ng Kee Kwong [/bib_ref] ; or in COPD [bib_ref] Effect of Music Therapy on Pain After Orthopedic Surgery-A Systematic Review and..., Lin [/bib_ref] [bib_ref] The effect of music on preoperative anxiety in day surgery, Cooke [/bib_ref] [bib_ref] Effects of non-pharmacological interventions on preoperative anxiety and postoperative pain in patients..., Tola [/bib_ref].
Also related to theme 1, interviewees often reported the impact of music on pain management: I have to mention pain as the area with the best study record (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 2, theme 1). The literature includes many studies of music in perioperative settings for reducing anxiety and pain [bib_ref] A systematic review of music-based interventions for procedural support, Yinger [/bib_ref] [bib_ref] The Effects of Music Therapy on Patients with Coronary Artery Disease Before..., Oyur Çelik [/bib_ref] [bib_ref] A role for music in cataract surgery: A systematic review, Dahshan [/bib_ref] [bib_ref] Favorite Music Mediates Pain-related Responses in the Anterior Cingulate Cortex and Skin..., Antioch [/bib_ref] [bib_ref] Effect of Genre and amplitude of music during laparoscopic surgery, Yang [/bib_ref] [bib_ref] Influence of musical background on surgical skills acquisition, Sun [/bib_ref] [bib_ref] Effects of background music on concentration of workers, Huang [/bib_ref] [bib_ref] The effect of music on cognitive performance: Insight from neurobiological and animal..., Rickard [/bib_ref] [bib_ref] Exploring the effect of sound and music on health in hospital settings:..., Iyendo [/bib_ref] as well as during invasive procedures in patients with coronary heart disease, where music has positively affected blood pressure, respiration rate, anxiety, pain level, and the amount of sedative medication required [bib_ref] Performing arts as a health resource? An umbrella review of the health..., Mccrary [/bib_ref]. Another study showed that music reduced the need for benzodiazepines during cataract surgery. Although the medical pain pathway includes the limbic system and is linked to emotion, it remains unknown exactly how music affects the pathway activation and, therefore, to what extent music alters pain reception.
Participants emphasized the impact of music, not only on patients but also on employees in the medical environment, such as physicians, especially surgeons listening to their favorite music during an intervention (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subthemes 3 & 4, theme 1). A previous study showed that playing music during surgery improved intervention time, accuracy, and precision during laparoscopic procedures [bib_ref] A Painful Symphony: The Presence of Overuse Syndrome in Professional Classical Musicians, Marić [/bib_ref]. According to the literature, it seems important not to select background music that is too strongly liked or disliked, as this might also negatively affect concentration [bib_ref] Work postures and neck-shoulder pain among orchestra musicians, Nyman [/bib_ref]. Nevertheless, in line with another previous study [bib_ref] Medical problems of musicians, Lockwood [/bib_ref] , our participants voiced that music exposure could enhance human cognitive performance (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 3, theme 1): That this is good for me when operating. Specifically, I notice that with classical music I can concentrate much better in the manual-practical activity and then also dissect much more targeted and quickly.
The last notable aspect related to theme 1 is that interviewees reported positive impacts of music played in different locations of the medical environment, not only preoperatively but also in a hospital's entrance hall or in the outpatient clinic. In such environments, music is considered to reduce stress and anxiety. Our interviewees most frequently mentioned classical music: When you bring music in as a feel-good factor in a hospital atmosphere that can be rather sterile and scary, you can break down a lot in patients up front. In line with subtheme 4 of theme 1 (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , previous research has shown that creating soundscapes can improve the experience of the hospital environment for both patients and providers. One interviewee claimed that the use of compositions explicitly written for the entrance hall of a hospital created a comfortable and welcoming place for patients, and that they had already implemented this strategy in one of their city hospitals (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 3, theme 1).
## Theme 2: performing arts medicine
The predefined Theme 2, "performing arts medicine", encompassed the impact of medicine in the world of performing arts. The most frequent issue voiced by interviewees was the importance of advertising and promoting the use of targeted performing arts medicine (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 2, theme 2). It was also emphasized that there are various different diseases that affect musicians. Interviewees stated that musicians need medical support for their daily life that is specific to their art. They also suggested that medical professionals and musicians could improve their collaboration by developing interdisciplinary "boards" for particular diseases, such as those that exist in oncology. Participants proposed that such a collaboration should also involve music teachers, therapists, and those research professionals (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 4, theme 2). These results and subthemes are interesting in light of the achievements that have already been made in performing arts medicine. Performing arts clinics have now been established around the world, and collaboration exists between medical professionals with expertise in treating musicians, including orthopedics; neurology; ear, nose, and throat medicine; and psychology [bib_ref] Lifetime prevalence of upper-body musculoskeletal problems in a professional-level symphony orchestra: Age,..., Abreu-Ramos [/bib_ref]. Likewise, research is being conducted and presented at meetings to further develop modern and preventative approaches. Our results might indicate that these efforts and studies are currently being ignored. It has previously been shown that there is a great need to advertise and promote performing arts medicine [bib_ref] Neuromuscular and musculoskeletal problems in instrumental musicians, Lederman [/bib_ref] , and interdisciplinary interaction in performing arts medicine [bib_ref] The treatment of overuse syndrome in musicians. Results in 175 patients, Fry [/bib_ref]. Albrecht Lahme stated "No overuse without misuse", meaning that the severity of medical conditions among musicians is linked to and caused by the amount of practicing hours [bib_ref] Overuse syndrome of the hand and wrist in musicians: A systematic review, Betzl [/bib_ref]. Our participants actually made several statements referring to overuse syndrome [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 2, theme 2), which is common among professional musicians [bib_ref] Overuse syndrome in musicians, Bird [/bib_ref]. Musicians' intense focus on their instruments may affect their physical and mental health [bib_ref] Toward Better Prevention of Injuries Among Performing Artists, Manchester [/bib_ref] [bib_ref] Imagery-Based Interventions for Music Performance Anxiety: An Integrative Review, Finch [/bib_ref]. For example, enduring in an non-ergonomic position may result in musculoskeletal diseases [bib_ref] Current Approaches for Management of Music Performance Anxiety: An Introductory Overview, Zhukov [/bib_ref] [bib_ref] Treatment and prevention of music performance anxiety, Spahn [/bib_ref] [bib_ref] The Alexander Technique and musicians: A systematic review of controlled trials, Klein [/bib_ref]. Such musculoskeletal injuries may eventually necessitate medical consultation for treatment [bib_ref] Music performance anxiety-Part 1. A review of its epidemiology, Brugués [/bib_ref] [bib_ref] Music performance anxiety-Part 2. A review of treatment options, Brugués [/bib_ref] [bib_ref] Treatment of music performance anxiety via psychological approaches: A review of selected..., Nagel [/bib_ref]. Similar opinions were given throughout our interviews: The circumstances. I still find it terrifying how other instrumentalists, in particular, deal with their bodies, the suffering they go through, and what they believe is necessary to reach a certain level. How the lack of education at the universities also guarantees that no awareness is developing.
It has previously been expressed that "a sixteenth note of prevention is worth a whole note of cure" [bib_ref] Education through music'-The model of the Musikkindergarten Berlin, Uibel [/bib_ref]. Interviewees clearly felt that there was a serious need for preventive measures benefiting musicians. Importantly, the participants also questioned, for example, why screenings and check-ups are not yet being promoted for musicians, as there is a high demand [bib_ref] Music and Brain Development, Fernandez [/bib_ref] (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 6, theme 1 and subtheme 4, theme 2).
The final topic identified in theme 2 is that the interviewees voiced that there is a need to improve the present lack of knowledge about prospects for improving musicians' physical health (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subthemes 1 and 2, theme 2). Interviewees also felt that psychological support must be offered to a greater extent, such as when dealing with performance anxiety, which interviewees felt could result in physical disorders (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 2, theme 2). These perceptions are important because psychological support for performance anxiety may be among the best-researched areas in performing arts medicine, as indicated by the number of reviews and meta-analyses on this subject [bib_ref] What you learn & when you learn it: Impact of early bilingual..., Vaquero [/bib_ref] [bib_ref] From music making to speaking: Engaging the mirror neuron system in autism, Wan [/bib_ref] [bib_ref] Music playschool enhances children's linguistic skills, Linnavalli [/bib_ref] [bib_ref] Music training and child development: A review of recent findings from a..., Habibi [/bib_ref] [bib_ref] Suggestive but not conclusive: An independent meta-analysis on the auditory benefits of..., Román-Caballero [/bib_ref] [bib_ref] Memory modulations through musical pleasure, Ferreri [/bib_ref] [bib_ref] Music affects functional brain connectivity and is effective in the treatment of..., Speranza [/bib_ref]. However, the interviewees' opinions on this topic indicated that stage fright was not adequately being managed, despite the available therapeutic options.
## Theme 3: music for the individual
The majority of interviewees discussed the importance of music in their everyday lives. Most participants had experienced the presence of music in their own childhood starting from a very early age (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] and explained that music had become an essential part of their personal lives growing up, giving them joy and stability. Herbert von Karajan and Daniel Barenboim are perhaps among the most famous musicians who have claimed that playing an instrument as a child can alter the brain and neural pathways in a beneficial way and the long term [bib_ref] Autobiographical memory, the ageing brain and mechanisms of psychological interventions, Allen [/bib_ref] [bib_ref] The cognitive characteristics of music-evoked autobiographical memories: Evidence from a systematic review..., Kaiser [/bib_ref]. Evidence for their claims has been published [bib_ref] Music meets surgery: Two sides to the art of "healing, Moris [/bib_ref] [bib_ref] Enhancing surgical performance by adopting expert musicians' practice and performance strategies, Rui [/bib_ref] [bib_ref] The use of Music Therapy in the treatment of Mental Illness and..., Wang [/bib_ref] [bib_ref] The use of arts interventions for mental health and wellbeing in health..., Jensen [/bib_ref] and was also reinforced by our interviewees (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 1, theme 3, and subtheme 3, theme 4), despite some degree of published controversy [bib_ref] Group singing improves both physical and psychological wellbeing in people with and..., Campbell [/bib_ref]. Many interviewees stated that music could influence personal well-being and mood. Furthermore, our interviewees described that listening to a song or composition that one had listened to in the past might bring back memories that were created during that time of life (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subthemes 1 to 5, theme 3). Music and memory are another recurring and important research topic in the scientific literature [bib_ref] A chorus of voices: Social singing and health promotion, Fahey [/bib_ref] [bib_ref] What works for wellbeing? A systematic review of wellbeing outcomes for music..., Daykin [/bib_ref] [bib_ref] Me and us: Cultivating presence and mental health through choir singing, Damsgaard [/bib_ref] , especially regarding individuals with dementia [bib_ref] Group singing in bereavement: Effects on mental health, self-efficacy, self-esteem and well-being, Fancourt [/bib_ref] [bib_ref] Brain correlates of music-evoked emotions, Koelsch [/bib_ref].
Perhaps, as a result of their personal experience, and consistent with Barenboim's musical kindergarten project in Berlin, Germany [bib_ref] The cognitive characteristics of music-evoked autobiographical memories: Evidence from a systematic review..., Kaiser [/bib_ref] , many interviewees emphasized the need to implement active music making at different educational levels, such as in primary school courses, making it available to everyone starting from a very young age. Participants made the following statements: I think that children should be introduced to music much earlier. [ . . . ] Children today are overwhelmed with the world. The methods of education go in many directions. You can't teach everything that is necessary these days. I can imagine that music can calm children down and get them down, at least [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref].
An interesting aspect and additional subtheme of theme 3 was the perception of music as a way of self-conditioning. This process was explained through different statements made by several of our interviewees. The participants expressed that music could be helpful in a broad range of tasks, from simplifying household chores to making one more focused and concentrated during manual-practical tasks (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 4, theme 3). This subtheme of self-conditioning is somewhat related to reports in the literature describing the benefits that surgeons may gain from adopting the practice and performance strategies of expert musicians [bib_ref] Music as therapy in early history, Thaut [/bib_ref] [bib_ref] Music and the child in society, Masserman [/bib_ref]. Notably, this subtheme shows some overlap with subtheme 3 from theme 1.
Additionally, within theme 3, interviewees emphasized that active singing had particular importance. Since singing outside of a professional setting is possible without needing to learn to play an instrument, it had a special role during the interview sessions [bib_ref] Brain networks mediating the influence of background music on selective attention, Fernandez [/bib_ref]. Everyone has a voice, and can therefore use it for oneself and or with others. Melodies that are stuck in people's minds throughout the day may influence their way of thinking (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 5, theme 3). Furthermore, it was stated that music could achieve stress reduction through the same pattern as pain relief, as mentioned above (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 4, theme 4). "There is stress relief via the rhythm and establishing a relationship with oneself". Listening to and performing music may shift one's thoughts, take one away from a one-sided thinking pattern, and even help a person develop more stable mental health . A person can merely focus on the sound of music, thereby putting other thoughts in the background for an amount of time, such that one may have different ideas on a subject afterwards.
## Theme 4: music for society
Many interviewees stated that music could facilitate the building of relationships within society in various possible ways (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] and subthemes 1 and 2, theme 4). Music was described as embodying a sense of belonging, a community that can generate resources, harmony of a society. [One should make it] accessible to all population groups. It connects many people with each other, on a beautiful emotional level, not dependent on one's origin, income, occupation ( . . . ) It is an important bridge (this citation is not included [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref]. This perception is in line with previous reports that include the notion that music plays a role in societal cohesion. Interviewees also explained that music is a product of culture: It has something to do with down-to-earthiness, anchoring, and identity (see [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 2, theme 4). On the other hand, music has also been perceived as constitutive of people's youth and being associated with the evolution of social groups during young adolescence.
We also identified some criticism related to topics in theme 4. One such criticism concerned concert tours and traveling, which are typically a part of musicians' schedules: I practice a beautiful art and delight people worldwide in concert halls, but what happens? Entire orchestras are flown across the globe. What happens to the waste I am creating? (See [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 5, theme 4). Additionally, interviewees voiced concerns that music, especially learning to perform, is not affordable for everybody, although it is perceived as such an important factor for human development: For myself, I will go with the direction of social justice. What happens in society? Why are there so many refugees? There's so much social injustice. People with no privilege, homeless shelters. People don't have a place to live. We should think about community music therapy. This is the concept. Music therapy that works in refugee camps. That is something that needs to be established long-term. [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref] , subtheme 5, theme 4).
One interviewee also stated that people can listen to music via headphones as a means of dissociating from their surroundings, such as in public transport. This perception is concordant with the idea that music can be seen as a tool of self-identification and distancing through different music genres experienced by the same individual.
# Conclusions
A literature search conducted in January 2023 using the PubMed online database yielded approximately 6000 citations on the topic of "music and medicine". However, to the best of our knowledge, a comprehensive, systematic description of the overall relationship between music and medicine has not yet been established. Despite the wealth of clinical research suggesting the evident effects of music on health, there remains a need to explore the underlying mechanisms that contribute to these effects.
In our qualitative study examining the perceptions of music and medicine, we found it easier to identify themes related to the impact of music on medicine (specifically in music therapy and music medicine) compared to the impact of medicine on music. Recognizing that the health status of performers and composers is intertwined with the music they create, we aimed to elucidate the interdependence between music and medicine. We sought the perspectives of attendees and invited speakers at a music and medicine conference, exploring not only what music could do for medicine but also what medicine could do for music.
The responses we received covered a wide range of perspectives, with interviewees overwhelmingly acknowledging the positive impact of music on individual development and society as a whole. They emphasized the urgent need for increased awareness and recognition of the synergistic relationship between music and medicine, suggesting that community campaigns could play a vital role in achieving this goal.
One important limitation of our study is the potential selection bias resulting from conducting interviews at the "Science and Sounds" conference. The pre-selected sample of participants may not be fully representative of the general population, potentially influencing the findings and generalizability of the study. To mitigate this bias, we made efforts to interview a diverse range of individuals across different ages and specializations.
Furthermore, although our recruitment goal was aimed at achieving data saturation and diverse perspectives, careful consideration must be given to potential selection bias. Future work is necessary to further explore these perceptions and their implications. We believe that the insights gathered from this study provide valuable information for funders and researchers, emphasizing the importance of increasing public visibility of their work to improve the lives of musicians and non-musicians, particularly those with health issues but also those without. Institutional Review Board Statement: Ethical review and approval were waived for this study due to no patients were included. The study was conducted according the data protection committee regulations.
# Informed consent statement:
Written informed consent has been obtained from the participants to publish this paper. Data Availability Statement: Data is given in Appendix A; further data is unavailable due to privacy or ethical restrictions.
# Acknowledgments:
The authors wish to thank all study participants for sharing their insightful experiences. Thank you, Judith Platter and team (www.sprachdschungel.com/de, accessed on 10 July 2023), for excellent and rapid interview transcriptions. [fig_ref] Table 1: Demographics and personal characteristics [/fig_ref]. Themes and subthemes with representative quotes.
## Conflicts of interest
## Music in medicine
## Subtheme 1: listening and performing to treat disease and establish a non-verbal relationship
[I have seen] what one can do with music, how one can help people, and not just in general with mood, but also specifically how one can see physical progress.
For some, it [music] was an icebreaker on the relationship level. They saw that they didn't have to talk to a lady at all, but that they could also talk instrumentally with music. Then I also took up an instrument.
## Subtheme 2: value of music in specific disorders, at the end of life, and in pain management
"I was able to experience in a very touching way how music can simply carry them through these times: Pregnant women, women who may have had a miscarriage and are now afraid to get pregnant again or who already are, premature babies, children in orthopedics, in oncology, patients with eating disorders, people in the very last phase of life, in palliative care and until the day they die".
I believe that the focus is reduced from one's own pain, because the attention is being split. Of course, it depends on whether the music was chosen by the patient. // I have to mention pain as the area with the best study evidence.
## Subtheme 3: application of music in different environments
Bringing music in as a feel-good factor in a hospital atmosphere, which can be rather sterile and scary, can break down a lot [of negative factors] in patients beforehand.
Z is doing a lot with music in the operating room, but I'm not that involved there yet. But in the internal medicine ward, where I am right now, I would like to do that [turn up the radio . . . ]. For example, I brought my violin. Some of the patients started crying.
## Subtheme 4: design of sound spaces
One tries to change the atmosphere of medical rooms and institutions. [ . . . ] The music does not have to be loud. One can change atmospheres in medical institutions through music.
Not only the decorative picture of Mirot on the wall, but also sound decoration, sound design in social spaces. That can also mean more silence. That doesn't have to mean plastering everything with kitsch music.
## Subtheme 5: overcoming loneliness
What about the music would it have been? S2: Not to be alone. Especially when it's medically borderline, someone is about to be treated, or there's an examination, everything is very medical and organized, and no one really has a deeper human contact with the patients. Subtheme 6: Need for more awareness in the medical field S1: So, you would have to do campaigning? S2: Yes. The inclusion specialist in our school asked "What does music therapy do? I know what ergotherapy does, but tell me what music therapy does". S1: So, there is too little effect known? S2: Yes.
Every medical professional should know something about the effect of music therapy. [ . . . ] They don't know that music therapy exists. Neither does my sister. They think it has nothing to do with medicine. They ask, "Why do they teach you medical subjects or psychoanalysis?" She has no idea. That hurts a little bit. Drug use occurs, especially alcohol. This also almost always has to do with anxiety disorders or controlling stage fright or gaining control over it. That's a difficult topic, especially one that is very taboo. Almost no one goes there. But you hear it in the background.
How did those who came to you know that [performing arts medicine] existed? S2: Word got around. People came secretly. Some of the lecturers who work here also came, and I am still in contact with them today. They claimed that they had a nutrition problem, for example, but that wasn't it.
## Subtheme 2: lack of knowledge and awareness
"The circumstances. I still find it terrifying how other instrumentalists, in particular, deal with their bodies, the suffering they go through, and what they believe is necessary to reach a certain level. How the lack of education at the universities also guarantees that awareness is not developing".
## Subtheme 3: importance of prevention
When you also hear how many drugs are taken, what kind of psychological pressure is behind it all, and how many are on beta blockers. I think that's where medicine can still make a big step in prevention among people who play music, especially in teaching.
As I said, I would make a prevention offer, above all, because we see many people who are stuck in a very strict apparatus. If a complaint is added to that, then they collapse.
## Subtheme 4: interaction between disciplines
[ . . . ] when one takes medicine broadly, and also includes physiotherapy and everything that revolves around staying healthy, [ . . . ].
[ . . . ] tell a doctor, try to look at what's happening with the MRI, and why. Where is the difference between a physiotherapist and a music therapist when we talk about movement. [ . . . ] Collaboration with medical doctors, also with biologists-what happens in the blood when I get music therapy regularly. As well as in psychology. [ . . . ] working in the field of psychology, I would explore with the psychotherapist how we can further work on this feeling that I have aroused in the patient through music.
## Subtheme 5: antithesis-pain and suffering driving creativity
My first impulse when asked what medicine could do for music was to say "actually nothing". For me, music is something very free, creative, arising out of the human being. A great deal of good music and art, I would say, arose from bad conditions of the artists. The absence of medicine as a creative boost.
Illness and, above all, pain makes the "I" awaken to itself. Pain is a great awareness maker. Or also the crisis. In this respect, it is also a motor for creativity, without there being an inevitable link between disorder and artistic creation. It is a drive. Often, even on a smaller scale, it is always part of family celebrations for me. There is no family celebration that does not end with notes being handed out and eight-part singing or all have their instruments with them. I used to play in a very special orchestra. [ . . . ] It is an orchestra comprising people with or without mental illness. The conductor himself lives with bipolar disorder. The orchestra manager has anxiety disorder. I was playing in this orchestra in a very special role because I don't have a diagnosis but there are a lot of patients playing and a psychiatrist. We all work together and play music together.
## Subtheme 4: music enabling self-conditioning
[ . . . that this is good for me during surgery. Specifically, I notice that with classical music, I can concentrate much better in the manual-practical activity and then also prepare much more purposefully and quickly.
I just sat down in my room at the piano, closed the door, and just regulated myself through the music. I could give myself something through music. Yes, of course. Music plays a very big role, but so does singing. When a singing lesson has gone well, you sometimes feel like you're floating on clouds.
## Music for the society
## Subtheme 1: a common language outside of verbal speech
It is an incredibly important other language that is not a language of words or gestures. Actually, everyone knows it too. Why does everyone turn on the radio and try to get into different moods through music? But we have to be careful that we continue to take it seriously enough. Not only the hyped musician on stage must be important, but also the students must be taught: how good that you exist and want to bring music into the world, whether as a music teacher, or in small or large orchestras. Everything must be important.
I often notice abroad that you can connect with each other very quickly through music and form a deep bond, even if you don't speak the other language at all.
## Subtheme 2: music for cultural self-identification
It's an important point that we have cultures, whether it's music, painting, or theater. It's a way to get to know yourself better emotionally and to get rid of or work on your own emotions.
It's something cultural, especially when you go somewhere else and start living anew. What is their music? It has something to do with down-to-earthness, anchoring, and identity.
## Subtheme 3: development and education for children
I think that children should be introduced to music much earlier.
[ . . . ] Children today are overwhelmed with the world. The methods of education go in many directions. You can't teach everything that is necessary these days. I can imagine that music can calm children down and get them down, at least.
If the family can't pick that up either, the kids just stay there and either explode or get all sad. S1: And your hypothesis is that music could be a help? S2: Yes. I have also experienced myself in the private sphere that it helps me tremendously, even without verbalizing anything, to get something stuck flowing nonverbally.
## Subtheme 4: stress reduction
What is the status of music in society? S2: Many aspects. First of all, stress reduction, because we are all stressed out.
There is a lot of manipulation via the music, the arrangement and the tempo. There is a lot of unloading; there is stress relief via the rhythm and establishing a relationship with oneself. You can see it as a pure hedonistic event, or also note that the rhythm is the component of entering into relationship and feeling oneself, because you can surrender to this rhythm.
## Subtheme 5: political correctness
"I practice a beautiful art and delight people worldwide in concert halls, but what happens? Entire orchestras are flown across the globe. What happens to the waste I am creating?" For myself, I will go with the direction of social justice. What happens in society? Why are there so many refugees? There's so much social injustice. People with no privilege, homeless shelters. People don't have a place to live. We should think about community music therapy. This is the concept. Music therapy that works in refugee camps. That is something that needs to be established long-term. Music and social justice.
# Appendix b
# Interview guide introduction
Thank you for agreeing to participate in this interview. We have prepared a note for your declaration of consent. For your information: Everything will be evaluated anonymously. May we also ask you about your personal information? (Read out demographics.) With this interview study, we want to administer a questionnaire that should help us understand the importance of music in medicine and, vice versa, the importance of medicine in music. To this end, we would also like to ask personal questions.
Here is a summary of the interview conditions:
## -
The interview will be digitally recorded. The recording will be transcribed verbatim, and themes will be extracted from your answers in an anonymous fashion.
## -
You are welcome to use names, but we will remove all names from the document. - There are no right or wrong answers, we are only interested in your opinion. - Your participation is voluntary. If you want to take a break during the interview, want to end the whole thing, or something similar, please let me know.
## -
The interview should not last longer than an hour, ideally only half an hour, depending on how we get through and how long you want to talk. - Do you have questions?
If you don't mind, I will start the interview now. Opening
## 1.
Have you already had your own personal experiences with music in medicine? Do you have any health problems yourself? 2.
Do you feel limited in any way in your daily life? If there are no limitations now, have there been any in the past?
## Impact of music in medicine
## 3.
What are the 2-3 most important experiences/circumstances that you would like to report, in which music played a role in the course of a medical disease? These can be about yourself, but also about others. Why are these the most important experiences/circumstances? What consequences did these experiences have for you? (For example: physical, psychological, social, lifestyle, or financial consequences) 4.
When you think about your daily life, in which areas does music play a role? And how? (For example: in routine activities, like washing, dressing, and eating; work; interactions with family/friends/others; or leisure/spare time)
## Impact of medicine in music
## 5.
What are the 2-3 most important experiences/circumstances that you would like to report, in which medicine played a role in music? These can be about yourself, but also about others. Why are these the most important experiences/circumstances? What consequences did these experiences have for you? (For example: physical, psychological, social, lifestyle, or financial consequences) 6.
When you think about your daily life, in which areas does medicine play a role? And how? (For example: in routine activities like washing, dressing, and eating; work; interactions with family/friends/others; or leisure/spare time)
## Role of music for society
## 7.
What are the 2-3 most important experiences/circumstances that you would like to report, in which you think music played a role in society? This can be about yourself, but also about others. Why are these the most important experiences/circumstances? What consequences did these experiences have for you? (For example: physical, psychological, social, lifestyle, or financial consequences) 8.
When you think about your impressions of society, in which areas does music play a role? And how?
## 9.
How does one recognize the role/value/importance of music in society?
## What to study and what to fund
10. If you had €12 million available for music and medicine, what would you finance and why? 11. What can music improve for medicine? (active/receptive?) 12. What can medicine improve for music? Close 13. Do you have any suggestions on how music can improve the treatment of people with diseases? 14. What potential do you see in broad-based and outpatient/ambulatory offers? 15. Is there anything you would like to add?
Thank you again for your willingness to participate in the interview. If you don't mind, I'll email you a copy of the preliminary results.
[fig] Author: Contributions: Conceptualization, M.H. and F.S.; methodology, S.H.; software, M.H.; validation, H.L.-S., J.H. and S.H.; formal analysis, U.K.; investigation, K.S., J.H., S.H., F.S. and H.L.-S.; resources, M.H. and S.D.; data curation, M.H.; writing-original draft preparation, J.H.; writing-review and editing, H.L.-S.; visualization, J.H.; supervision, H.L.-S.; project administration, M.H.; funding acquisition, M.H. and S.D. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. [/fig]
[fig] Subtheme 2: Mental role of music in ordinary and extraordinary life Then, of course, in everyday life, I sing a lot around home, dance and listen very much overall. Otherwise, just at festivals, parties, weddings, funerals. // So, it's a de-stressing and concentration-enhancing effect? S2: Exactly. Maybe I also have a need for it by now. It's just part of it, it's conditioning for the day, [ . . . ] I think about the fact that [ . . . ] in moments when they [people] become insecure or go through difficult situations, [music] is one of the first things they resort to. In which emotions are also processed, whether it's active or passive music making. I think that's a great outlet for a lot of people. Subtheme 3: Variability of music listening and active music making Classical music, even if it's jazz, is like a bubble for me, a soft pillow. For light music or anything else, I'm really open and have a wide repertoire on my music accounts, [ . . . ]. On the weekend or on vacation, I sometimes listen to other playlists, [ . . . ] [/fig]
[fig] Subtheme 5: Special role of singing "Melodies in my head" is very present with me. A keyword comes up and I think directly of a piece. I myself listen a lot and sing a lot. You can incorporate singing a lot in everyday life, you can sing on the road, hum or whatever. [/fig]
[table] Table 1: Demographics and personal characteristics. [/table]
[table] Table A1: Cont. [/table]
|
10.1038/s41598-021-91986-7
|
CCBY
|
8211801
|
34140544
|
s2orc_pubmed_articles
|
Production of basil (Ocimum basilicum L.) under different soilless cultures
The main aim of this paper was to investigate the possibility of growing basil under three soilless systems (aeroponic, hydroponic and peatmoss slab systems). A model was developed to predict the nutrients consumption by basil plants. Shoot and root height, fresh and dry mass of whole plant, nutrients uptake, and oil content were studied during the growth period (after 4 and 7 weeks from transplanting). The results indicated that the shoot lengths of basil plants were 71.67 ± 2.89, 65.67 ± 1.15 and 62.33 ± 2.31 cm at the end of growth period for aeroponic, hydroponic and peatmoss slabs, respectively. The highest value of root height of basil plants was 37.67 ± 6.66 cm for aeroponic system. The dry mass of shoot of basil plants ranged from 28.48 ± 0.91 to 44.77 ± 0.97 and 72.98 ± 0.83 to 117.93 ± 1.40 g plant −1 after 4 and 7 weeks from transplanting, respectively. The highest values of the N, P, K, Ca and Mg uptakes were 753.99 ± 5.65, 224.88 ± 3.05, 449.75 ± 4.59, 529.12 ± 6.63 and 112.44 ± 1.67 mg plant −1 at the end of experimental period, respectively. The basil oil content ranged from 1.129 (1.11%) to 2.520 (1.80%) and 2.664 (1.42%) to 6.318 (1.44%) g plant −1 after 4 and 7 weeks from transplanting, respectively at the same pervious order. The production costs of basil plant were 2.93, 5.27 and 6.24 EGP kg −1 of plant. The model results were in a reasonable agreement with the experimental ones.There is an increasing interest recently for growing sweet basil (Ocimum basilicum L.) in greenhouse soilless culture, which offer a suitable condition for maximization of production 1,2 . It is cultivated commonly in an open field with a variability in productivity and quality 3 .Basil has high nutritional contents with low caloric values. It is used as a pharmacological raw material. Also, it contains vitamins A, B 6 and C as well as carotene besides calcium, potassium, phosphorus, magnesium, iron. Therefore, it needs a warm climate and high temperature and soil should be fecundity 4,5 .The advantages of the soilless culture are the earlies growth and higher yield compared to traditional culture. Also, this system assures an equal supply of nutrient solution, so it can obtain a homogeneous crop. The mineral elements concentration and composition are well adjusted. Also, the buffer capacity of nutrient solution is low. pH and mineral composition of solution are easily changeable. Soilless culture decreases the time of adjusting solution 6 .Soilless cultivation systems provide plant management under controlled water and minerals supply of the nutrient solution with or without medium. There are three systems of soilless cultivation namely, system with solid medium, in a liquid medium and aerated medium 7,8 .Hydroponic system is a way plants without soil in water having a nutritional solution. The soil is used in traditional cultivation as a medium to add water and minerals in it, this soil is not needed in hydroponic because the minerals are added directly to water where the plants grow. It is more efficient to control water which can be reused after adjustment. It decreases the use of pesticides. It is used for many crops such as beets, radishes, carrots, potatoes, cereal crops, fruits, ornamentals and seasonal flowers can be grown on inert supporting medium instead of soil 9-11 . QI 12 reported that the aeroponic system is a type of growing plants in air or mist environment without using any soil. In hydroponic, plant's roots are growing in water with nutrients. But for aeroponic, the nutrients are added through mist spray by sprinkles to plant's roots. The aeroponic system consist of a pump, nozzles, and OPEN
www.nature.com/scientificreports/ growing chamber. There are a few types of aeroponic like low pressure type, high pressure type and commercial system. Basil is used as fresh and dried leaves a medicinal herb [bib_ref] Comparative essential oil composition of flowers, leaves and stems of basil (Ocimum..., Chalchat [/bib_ref] [bib_ref] Evaluation of rhizobacteria of some medicinal plants for plant growth promotion and..., Ahmed [/bib_ref] for its diuretic and stimulating properties and also used in perfume compositions [bib_ref] Potassium rate alters the antioxidant capacity and phenolic concentration of basil (Ocimum..., Nguyen [/bib_ref]. Basil is growing better in soilless systems than conventional systems and many studies have used basil as aquaponic or hydroponic crop [bib_ref] Inoculation effect of Azospirillum brasilense on basil grown under aquaponics production system, Mangmang [/bib_ref].
The most severe problem in the hydroponic system and soilless is the root rot which is due to the low oxygen level in the nutrient solution, therefore, proper aeration is required to overcome this problem. Aeroponic system is the proper solution to provide the plant with the required oxygen and nutrients. Besides, demand of organic production is increasing day after day. Therefore, this study aimed to improve the basil production under three soilless systems.
# Materials and methods
The experiment was conducted at Agricultural and Bio-Systems Engineering Department, Faculty of Agriculture Moshtohor, Benha University, Egypt (latitude 30° 21′ N and 31° 13′ E), during the period of May to July, 2019 season under the university guidelines and legislation. Basil seedlings were sown in the plastic cups (7 cm diameter and 7 cm height) filled with peat moss. The cups were irrigated daily using water with nutrient solution (Ca(NO 3 ) 2 , 236 g L −1 , KNOCulture systems description.,b show the experimental setup. It shows the system which consists of hydroponic system, aeroponic system, soilless substrate, solution system and pumps.
The hydroponic system (Deep Water Culture (DWC)) consists of three rectangular polyethylene tanks that used for basil plants culture. Dimensions of each tank are 80 cm long, 40 cm wide and 30 cm high. The slope of hydroponic tanks was 2% and stand 1 m high above the ground. The hydroponic tanks were covered with foam boards to support the plants. Each hydroponic tank provided with an air blower (Model NS 780-Flow Rate 850 L h −1 -Head 1.5 m-Power 15 W, China) to increase dissolved oxygen concentrations. The solution was circulated by a pump (Model First QB60-Flow Rate 30 L min −1 -Head 25 m-Power 0.5 hp, China) from the solution tank to the upper ends of the hydroponic tanks. Small tubes (16 mm) were used to provide tanks with solution in a closed system.
Aeroponic system consists of three rectangular polyethylene tanks that used for basil plants culture. Dimensions of each tank are 80 cm long, 40 cm wide and 50 cm high. The aeroponic tanks were established 1 m above the ground. Each aeroponic tank was divided into two parts, the lower part was made from polyethylene and the upper part was made from wood. The aeroponic tanks were covered with foam boards to support the plants. Each aeroponic tank was provided with two fog nozzles (Model M3MNWT5M -Orifice 2 mm -Discharge 8 L h −1 , India) located at the bottom of the tank sprayed nutrient solution into the tank in order to keep the roots wet. Small tubes (16 mm) were used to provide aeroponic tank with solution in a closed system.
Soilless substrates consist are placed in three rows are 2 m long. Each row consists standard peat moss slabs (1.00 m × 0.20 m × 0.075 m). Basil plants were placed on row peat moss slabs with a drip irrigation system. There were three plants per slab giving a mean density of 9.0 plant m −2 . Each plant was fed by a single drip.
The circular polyethylene tank of the nutrient solution system 500 L capacity was used for collecting the drained solution by gravity from the ends of the three systems. The nutrient solutions were prepared manually once per ten days 17,18 by dissolving appropriate amounts of Ca(NO 3 ) 2 , 236 g L −1 , KNO 3 , 101 g L −1 , K 2 SO 4 , 115 g L −1 , KH 2 PO 4 , 136 g L −1 , MgSO 4 246 g L −1 and chelates for trace elements into preacidified groundwater (from the following ppm concentration are achieved in this formulation: N = 210, P = 31, K = 234, Ca = 200, Mg = 48, S = 64, Fe = 14, Mn = 0.5, Zn = 0.05, Cu = 0.02, B = 0.5, Mo = 0.01). pH and Electrical Conductivity (EC) were further adjusted to 6.5-7.0 and 1.4-1.8 dS m −1 , respectively, after salt addition. The average air ambient temperature was 25.97 ± 4.37 °C and the average water temperature was 24.03 ± 3.92 °C. The average relative humidity was 65.4% and the light intensity was 338.55 ± 40.06 W m −2 .
Measurements. Three plants sample were taken during the vegetative and flowering stages (four and seven weeks after transplanting, respectively) for growth measurement and chemical analysis. Plant height, root length and the fresh and dry weight of leaves, stems and roots were determined. After measuring fresh mass, the plants were oven dried at 65 °C until constant weight was reached [bib_ref] Temperature effect on drying and phytochemicals of basil leaves, Parmar [/bib_ref]. Total content of macro elements was evaluated after being digested 20 . Nitrogen was determined by Kjeldahl digestion methods [bib_ref] Nitrogen-total, Bremmer [/bib_ref]. Potassium, Calcium and magnesium were determined by Photofatometer (Model Jenway PFP7-Range 0-160 mmol L −1 , USA) and phosphorus (P) was determined colorimetrically method [bib_ref] A modified single solution method for determination of phosphate in natural waters, Murphy [/bib_ref]. The content of oil was determined in different organs: leaves, stems and inflorescences according to [bib_ref] Rosmarinic acid content in basil plants grown in vitro and in hydroponics, Kiferle [/bib_ref].
Water samples were taken, at inlet and outlet of the culture units for measuring nitrogen (N), phosphorus (P), potassium (K), calcium (Ca) and magnesium (Mg) were measured every week at 10 am during the experimental period.
Total production cost. The cost calculation based on the following parameters was also performed: where I n is the interest, EGP year −1 . i n is the interest as compounded annually, decimal (12%). Shelter, taxes and insurance costs (Si). Shelter, taxes and insurance costs were assumed to be 3% of the purchase price of the automatic feeder (Pm). Then:
## Fixed costs (fc). depreciation costs (d c )
## Variable (operating) costs (vc). repair and maintenance costs (r m ):
(1) Total costs (T c ). [fig_ref] Table 1: The input parameters of calculate total production costs of basil plants grown... [/fig_ref] shows the input parameters of calculate total production costs of basil plants grown in different soilless systems.
[formula] D c = P d − S r L d (2) I n = P d + S r 2 × i n( [/formula]
Nutrients consumption rate. The Nutrients consumption rate were calculated as the differences between the nutrients at inlet and outlet of culture units by the following formula [bib_ref] Utilization of effluent fish farms in tomato cultivation, Khater [/bib_ref] :
where C Nc is the nutrients consumption rate, mg day −1 plant −1 . Nc in is the nutrients at inlet of the hydroponic unit, mg L −1 . Nc out is the nutrients at outlet of the hydroponic unit, mg L −1 . Q is the discharge, L h −1 .
## Model development of nutrient consumption. model assumptions:
- N, P, K, Ca and Mg are the nutrients used in study.
- The plants are uniformity distributed in the solution, so they work as a uniform sink for water and minerals with space at any time. - The root systems are uniformly dispersed in the solution with uniform root length density at any time.
- The whole root system uptake characteristics are uniform.
- Water losses by evaporation are negligible.
The simplest nutrient consumption models relate the nutrient consumption to the concentration gradient using some sort of proportionality factor such as root permeability or conductivity [bib_ref] The concept of a root demand coefficient, Nye [/bib_ref] [bib_ref] Uptake of aluminum into root cytoplasm: predicted rates for important solution complexes, Akeson [/bib_ref]. The nutrient consumption was determined by using the following equation:
where NC is the nutrient consumption, mg plant −1 day −1 . ∆C is the concentration gradient, mg plant −1 day −1 . a NC is the proportionality factor, dimensionless.
A similar model of nutrient consumption takes into consideration the differing effects caused by variations in root growth stage. Assuming that growth follows a first order differential equation and assuming that the root growth is exponential 27 , then Eq. (11) can be derived. This equation is presented in similar form to Eq. (10) and use the following equation:
where C planto is the concentration of the nutrients in the plant at time t 0 , mg plant −1 . A r is the root surface area at time t, cm 2 plant −1 . A r0 is the root surface area at time t 0 , cm 2 plant −1 .
Root surface area was calculated from root length and mean root radius using the following equation:
The root length increment using the following equation [bib_ref] Soil and plant factors affecting evapotranspiration, Ritchie [/bib_ref] :
where ∆L r is the root length increment, cm day −1 . ∆DW root is the daily amount of root dry mass increment, g day −1 . v is the ratio of root length and mass of roots, cm g −1 .
The daily amount of dry weight of roots is calculated from the following equation [bib_ref] Modeling root growth and the soil-plant-atmosphere continuum of cotton crops, Coelho [/bib_ref] :
(4) R m = 100% deprecation cost/hour of use per year www.nature.com/scientificreports/ where LAI is the leaf area index. Leaf area index was changed in the same proportions as root length density to maintain a constant ratio between roots and shoots. The leaf area index is calculated from the following equation [bib_ref] Simulation study of nutrient uptake by plants from soilless cultures as affected..., Silberbush [/bib_ref] :
[formula] (5) E = EC × EP (6) L a = Salary of one worker × No. of workers (7) Variable costs = Rm + E + La (8) Total costs = Fixed costs + Variable costs (9) C Nc = Nc in − Nc out Number of plants × Q × 24 (10) NC = a NC · C (11) NC = � � C plant − C plant0 � A r − A r0 � · ln � A r A r0 � t − t 0 .A r (12) A r = 2πr 0 L r (13) L r = DW root v [/formula]
where LAI max is the maximum leaf area index. K 2 and k 1 are the coefficients of the growth functions.
All computational procedures of the model were carried out using Excel spreadsheet. The computer program was devoted to mass balance for predicting the nutrients consumption. The differences between the predicted and measured values were evaluated using RMSE indicator (root means square error) which is calculated using the following equation:
The parameters used in the model that were obtained from the literature are listed in [fig_ref] Table 2: The parameters used in the model [/fig_ref]. [fig_ref] Figure 2: Flow chart of nutrients consumption rate [/fig_ref] shows flow chart of the model.
## Statistical analysis. three replicates of each treatment were allocated in a randomize complete block
Design (RCBD) in the system. Data were analyzed one-way ANOVA (analysis of variance) using statistical package for social sciences (spss v21). Means were separated using New Duncan Multiple Range Test (DMRT). Data presented are mean ± standard division (SD) of four replicates.
# Results and discussion
Shoot length. [fig_ref] 3: Fixed cost = D c + I n + 0 [/fig_ref] shows the shoot length of basil plants grown in different soilless systems (Aeroponic, hydroponic and peatmoss slabs) at the vegetative stage (4 weeks after transplanting) compared to the flowering stage (7 weeks after transplanting). The results indicate that the shoot in aeroponic was taller than those of hydroponic system and peatmoss slabs at the vegetative and flowering stages. It could be seen that the shoot length of basil plants were 62.00 ± 2.65, 57.83 ± 7.42 and 48.77 ± 2.89 cm for aeroponic, hydroponic and peatmoss slabs, respectively, after 4 weeks from transplanting. Meanwhile, they were 71.67 ± 2.89, 65.67 ± 1.15 and 62.33 ± 2.31 cm for aeroponic, hydroponic and peatmoss slabs, after 7 weeks from transplanting at the same previous. These results agreed with those obtained by 33 whose found that the plants grown aeroponically were twice as high as those in hydroponics and 4 times taller than those grown in sand. These previous results may be due that the roots of aeroponics systems are hanged in mid-air inside containers or chambers at 100% humidity and fed up a fine mist of nutrient solutions. This pervious system stimulates absorption of roots to much needed oxygen and nutrients, those increasing metabolism and rate of growth compared with soil 34 .
[formula] (14) �DW root = 5LAI [/formula]
for LAI ≤ 0.5 2.5 + 23.9(LAI-0.5) for LAI > 0.5 www.nature.com/scientificreports/
[formula] (15) LAI = LAI max 1 + K 2 e (−k1t) (16) RMSE = (Predicted − Measured) 2 n [/formula]
The statistical analysis showed that the differences between the obtained data of shoot length due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was significant. Root surface area at time t www.nature.com/scientificreports/ Root length. [fig_ref] Figure 4: The root length of basil plants grown in different soilless systems [/fig_ref] shows the root length of basil plants grown in different soilless systems (Aeroponic, hydroponic and peatmoss slabs) at the vegetative stage (4 weeks after transplanting) compared to the flowering stage (7 weeks after transplanting). The results of measurements of root of the plants grown in aeroponic system were taller than those of hydroponic system and peatmoss slabs at the vegetative and flowering stages. It could be seen that the highest value of root length of basil plants was [bib_ref] Growth responses and root characteristics of lettuce grown in aeroponics, hydroponics, and..., Li [/bib_ref].67 ± 6.66 cm for aeroponic system, while, the lowest value of root length of basil plants was [bib_ref] Using williams equation to evaluate nutrient uptake rate by intact plants, Silberbush [/bib_ref].67 ± 0.58 cm was found with peatmoss slabs. The root length for basil plants grown in aeroponic system were 1.68 and 2.12 times taller than those grown in peatmoss slabs after 4 and 7 weeks from transplanting, respectively. These results agreed with those obtained by. Also, many studies showed that the aeroponic system enhance the rates of plants growth by promoting the root aeration because of the root system is grown totally suspended at the air, giving the plant stem and roots systems access to 100% of the available oxygen at the air [bib_ref] Evaluation of aeroponics for clonal propagation of Caralluma edulis, Leptadenia reticulata and..., Mehandru [/bib_ref]. These results are in agreement with findings which were reported by [bib_ref] Growth responses and root characteristics of lettuce grown in aeroponics, hydroponics, and..., Li [/bib_ref] that they showed that plant root length, area, volume of aeroponic system were significantly exceeded the hydroponic and substrate systems.
The statistical analysis showed that the differences between the obtained data of root length due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was significant.
Fresh and dry mass of shoot. [fig_ref] Figure 5: Fresh and dry mass of shoot of basil plants, [/fig_ref] ,b show the fresh and dry mass of shoot of basil plants grown in different soilless systems (Aeroponic, hydroponic and peatmoss slabs) at the vegetative stage (4 weeks after transplanting) compared to the flowering stage (7 weeks after transplanting). The results indicate that the fresh and dry of shoot grown in aeroponic system were better than those of hydroponic system and peatmoss slabs at the vegetative and flowering stages. It could be seen that the fresh and dry mass of shoot of basil plants were 140.00 ± 13.76, 139.02 ± 10.19 and 102.06 ± 35.54 g plant −1 and 44.77 ± 0.97, 32.36 ± 0.68 and 28.48 ± 0.91 g plant −1 for Aeroponic, hydroponic and peatmoss slabs, respectively, after 4 weeks from transplanting. Meanwhile, the results also indicate that the fresh and dry mass of shoot of basil plants were 438.61 ± 42.61, 229.33 ± 10.30 and 187.99 ± 24.84 g plant −1 and 117.93 ± 1.40, 77.85 ± 0.77 and 72.98 ± 0.83 g plant −1 for aeroponic, hydroponic and peatmoss slabs, respectively, after 7 weeks from transplanting. We can explain those that the aeroponic system enhance the rates of plants growth by promoting the root aeration because of the root system is grown totally suspended at the air, giving the plant stem and roots systems access to 100% of the available oxygen at the air [bib_ref] Methods of pre-basic seed potato production with special reference to aeroponics-a review, Buckseth [/bib_ref].
The statistical analysis showed that the differences between the obtained data of fresh mass of shoot due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was significant. Also, the statistical analysis showed that the differences between the obtained data of dry mass of shoot due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was non-significant.
Fresh and dry mass of root. [fig_ref] Figure 6: Fresh and dry mass of root of basil plants, [/fig_ref] ,b show the fresh and dry mass of root of basil plants grown in different soilless systems (Aeroponic, hydroponic and peatmoss slabs) at the vegetative stage (4 weeks after transplanting) compared to the flowering stage (7 weeks after transplanting). The results indicate that the fresh and dry of root grown in aeroponic system were better than those of hydroponic system and peatmoss slabs at the vegetative and flowering stages. It could be seen that the fresh and dry mass of root of basil plants were The statistical analysis showed that the differences between the obtained data of fresh mass of root due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was significant. Also, the statistical analysis showed that the differences between the obtained www.nature.com/scientificreports/ data of dry mass of root due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was non-significant.
Nutrients uptake. [fig_ref] Table 3: The nutrients uptake of basil plants grown in different soilless systems [/fig_ref] shows the nitrogen, phosphorus, potassium, calcium and magnesium uptake of basil plants grown in different soilless systems (Aeroponic, hydroponic and peatmoss slabs) at the vegetative stage (4 weeks after transplanting) compared to the flowering stage (7 weeks after transplanting). The results indicate that the uptake of nitrogen, phosphorus, potassium, calcium and magnesium by the basil plants were higher in aeroponic system compared those of hydroponic system and peatmoss slabs at the vegetative and flowering stages. It could be seen that the nitrogen uptake of .88 ± 1.10 mg plant −1 after 4 and 7 weeks from transplanting, respectively, were found with peatmoss slabs. These results agreed with those obtained by 37 they reported that the nutrients uptake of both aeroponic and hydroponic were higher than that in substrate cultivated.
The statistical analysis showed that the differences between the obtained data of nutrients uptake due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was significant as shown in [fig_ref] Table 3: The nutrients uptake of basil plants grown in different soilless systems [/fig_ref]. [fig_ref] Figure 7: The basil oil content grown in different soilless systems [/fig_ref] shows the basil oil content in different soilless systems (aeroponic, hydroponic and peatmoss slabs) at the vegetative stage (4 weeks after transplanting) compared to the flowering stage (7 weeks after transplanting). The results indicate that the basil oil content higher in aeroponic system compared to those of hydroponic system and peatmoss slabs at the vegetative and flowering stages. It could be seen that the basil oil content values were 2.520 (1.80%), 1.722 (1.24%) and 1.129 (1.11%) g plant −1 for aeroponic, hydroponic and peatmoss slabs, respectively, after 4 weeks from transplanting. Meanwhile, the results also indicate that the basil oil content were 6.318 (1.44%), 4.359 (1.90%) and 2.664 (1.42%) g plant −1 for aeroponic, hydroponic and peatmoss slabs, after 7 weeks from transplanting at the same previous. These results are in agreement with findings which were reported by [bib_ref] Essential oil composition of five Basil cultivars (Ocimum basilicum) from Albania, Cheliku [/bib_ref]. The statistical analysis showed that the differences between the obtained data of basil oil content due to the effect of culture system (A) and plant age (B) were significant. The analysis showed also that the interaction between both AB was significant.
## Content of oil.
Production costs. [fig_ref] Table 4: The total production costs of basil plants grown in different soilless systems [/fig_ref] shows the total production costs of basil plants grown in different soilless systems (aeroponic, hydroponic and peatmoss slabs) at the end growing period. It could be seen that the results indicate that the production costs of basil plant were 2.93, 5.27 and 6.24 EGP kg −1 of plant. The total production costs of basil plants grown in hydroponic system were 1.8 times higher than those basil plants grown in aeroponic system, also the total production costs of basil plants grown in peatmoss slabs were 2.1 times higher than those basil plants grown in aeroponic system. Besides it is considered as an organic product which is safe for the human health. The best fit for the relationship between the predicted and the measured values of nutrients consumption was in the following form:
where NC P is the predicted nutrients consumption, mg plant −1 day −1 . NC M is the measured nutrients consumption, mg plant −1 day −1 .
The constants of these equation and coefficient of determination are listed in [fig_ref] Table 5: The constants a, b and coefficient of determination for nutrients consumption [/fig_ref].
# Conclusions
An experiment study was conducted to investigate the possibility of growing basil under three soilless systems (aeroponic, hydroponic and peatmoss slabs). The vegetative parameters, nutrient uptake and oil content were studied. A mathematical model for mass balance of the system was developed successively for predicted the nutrients consumption by basil plant. It is concluded that the aeroponic system recorded the highest values of vegetation parameters (roots, shoots and leaves) and essential oil content. Meanwhile, it consumed the highest values of nutrients (N, P, K, Ca and Mg) and recorded the lowest costs (2.93 EGP kg −1 of plant). The model results were in a reasonable agreement with the experimental ones. www.nature.com/scientificreports/
[fig] 3: Fixed cost = D c + I n + 0.03 P m / hour of use per year [/fig]
[fig] Figure 1: (a) The experimental setup. (b) Images of system. [/fig]
[fig] Figure 2: Flow chart of nutrients consumption rate. [/fig]
[fig] Figure 4: The root length of basil plants grown in different soilless systems. Scientific Reports | (2021) 11:12754 | https://doi.org/10.1038/s41598-021-91986-7 [/fig]
[fig] Figure 5: Fresh and dry mass of shoot of basil plants, (a) at vegetative stage, and (b) at flowering stage. [/fig]
[fig] Figure 6: Fresh and dry mass of root of basil plants, (a) at vegetative stage and (b) at flowering stage. [/fig]
[fig] 17: NC P = aNC M + b [/fig]
[fig] Figure 7: The basil oil content grown in different soilless systems. [/fig]
[fig] Figure 8, Figure 9: The predicted and the measured nutrients consumption by basil plants during the whole growth period. (a) N, (b) P, (c) K, (d) Ca, (e) Mg. The comparison between the predicted and the measured nutrients consumption by basil plants during the whole growth period. (a) N, (b) P, (c) K, (d) Ca, (e) Mg. [/fig]
[table] Table 1: The input parameters of calculate total production costs of basil plants grown in different soilless systems. [/table]
[table] Table 2: The parameters used in the model. [/table]
[table] Table 4: The total production costs of basil plants grown in different soilless systems. https://doi.org/10.1038/s41598-021-91986-7 [/table]
[table] Table 5: The constants a, b and coefficient of determination for nutrients consumption. [/table]
|
10.3389/fonc.2022.937403
|
CCBY
|
9691441
|
36439444
|
s2orc_pubmed_articles
|
Four circadian rhythm-related genes predict incidence and prognosis in hepatocellular carcinoma
Circadian dysregulation can be involved in the development of malignant tumors, though its relationship with the progression of hepatocellular carcinoma is not yet fully understood. We identified genes related to circadian rhythms from the Cancer Genome Atlas (TCGA), measured gene expression, and conducted genomic difference analysis to construct a circadian rhythm-related signature. The resulting prognosis model proved to be an effective biomarker, as demonstrated by Kaplan-Meier survival analysis for both the training (n = 370, P = 2.687e-10) and external validation cohorts (n = 230, P = 1.45e-02). Further, we found that patients considered 'high risk', with an associated poor prognosis, displayed elevated levels of immune checkpoint genes and immune filtration. We also conducted functional enrichment, which indicated that the risk model showed a significant positive correlation with certain malignant phenotypes, including G2M checkpoint, MYC targets, and the MTORC1 signaling pathway. In summary, we identified a novel circadian rhythm-related signature allowing assessment of prognosis for hepatocellular carcinoma patients, and further can be used to predict immune infiltration sensitivity. (2022) Four circadian rhythm-related genes predict incidence and prognosis in hepatocellular carcinoma.
# Introduction
Hepatocellular carcinoma, the most common form of liver cancer, is one of the most common causes of cancer-related deaths worldwide, and is associated with substantial costs [bib_ref] Hepatocellular carcinoma, Villanueva [/bib_ref]. Potential treatments at early stages include liver resection, ablation, and transplantation, though most HCC patients are diagnosed at later stages, limiting treatment options [bib_ref] A global view of hepatocellular carcinoma: Trends, risk, prevention and management, Yang [/bib_ref]. Treatments for advanced-stage HCC are primarily systemic therapy approaches, including immune checkpoint inhibitors and targeted agents [bib_ref] Targeted therapy for hepatocellular carcinoma, Huang [/bib_ref].
Despite these available treatments, HCC patients generally experience recurrence and chemoresistance, resulting in overall poor prognosis. Empowered by innovations in multi-omics profiling, contemporary studies have provided prognostic candidates with clinical applications. In order to further the field, it is critical to identify strong molecular biomarkers that correlate with outcomes for HCC patients [bib_ref] Identification and validation of ubiquitin-specific proteases as a novel prognostic signature for..., Ni [/bib_ref].
The molecular circadian clock is an evolutionarily conserved system that coordinates internal time with the external environment. The circadian clock is integral in homeostasis and regulation, and manages many biological activities and b e h a v i o r s . W i t h i n t h e h y p o t h a l a m u s , t h e suprachiasmatic nucleus (SCN) has a central clock that regulates daily rhythms through neural and humoral coordination of additional clocks found in peripheral tissues and organs, and these clocks control physiological functions including sleep/wake cycles and hormone secretion. The World Health Organization lists circadian clock disruption as a putative carcinogen, as evidenced by population and laboratory-based findings [bib_ref] Carcinogenicity of shift-work, painting, and fire-fighting, Straif [/bib_ref]. Thus, research into the relationship between the expression of circadian-related genes and tumor development has been a focus. A previous study demonstrated that circadianrelated genes are correlated with the incidence and progression of NSCLC [bib_ref] Research on circadian clock genes in non-small-cell lung carcinoma, Qiu [/bib_ref]. Further research is required to fully explore the relationship between circadian genes and the prognosis for LIHC patients [bib_ref] Identification of a circadian gene signature that predicts overall survival in lung..., Gao [/bib_ref].
In the current study, we investigate the predictive role of circadian genes in LIHC patients using The Cancer Genome Atlas (TCGA) data from the NCI Genomic Data Commons, including clinical characteristics and the SNV, CNV, methylation, and mRNA expression profiles of tumor and tumor-adjacent normal tissues. We use these data to establish a prognostic multigene signature with differentially expressed circadian clock genes and validate this signature with the ICGC dataset. Essential molecular mechanisms are explored by gene set enrichment analysis (GSEA) using Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways (10).
# Materials and methods
## Data collection and processing
Multi-omics data profiles for the 14 circadian genes [bib_ref] Identification of a circadian gene signature that predicts overall survival in lung..., Gao [/bib_ref] [bib_ref] Circadian clock genes and the transcriptional architecture of the clock mechanism, Cox [/bib_ref] [bib_ref] Circadian rhythmassociated clinical relevance and tumor microenvironment of non-small cell lung cancer, Li [/bib_ref] [bib_ref] Circadian clock genes promote glioma progression by affecting tumour immune infiltration and..., Wang [/bib_ref] from 374 HCC patients together with their clinical data were obtained from TCGA LIHC datasets (https://cancergenome.nih. gov/). Heatmap was completed to visualize expression levels of the 14 circadian genes in HCC and of the normal tissues in the TCGA LIHC cohort. RNA-seq data of 232 HCC cases were also obtained from the ICGC cohort (https://dcc.icgc.org/projects/ LIRI-JP). Clinical data for the TCGA and ICGC cohorts are displayed in [fig_ref] TABLE 1: Demographical and clinical characteristics [/fig_ref]. Transcriptome RNA sequencing data in the count and FPKM format and corresponding clinical data were acquired from TCGA (https://portal.gdc.cancer.gov/) and used as the training dataset. Genes with average expression < 0.5 were omitted from the current study. The EdgeR package in R was applied for genomic difference detection with |logFC| > 1 and adjusted P < 0.05 filtering. Volcano plot was created with the ggplot2 package to visualize the difference analysis.
# Survival analysis
The survival package in R was used for Kaplan-Meier survival analysis with log-rank test, and for univariate and multivariate Cox regression. The glmnet R package was used for Lasso, and 10fold cross-validation was implemented. Calibration plots were created with the rms package. The survivalROC package was used to create the time-dependent ROC curves.
Univariate Cox regression test was implemented to evaluate the correlation of circadian-related genes with clinical outcomes for HCC patients. Circadian-related genes with P<0.05 were included in further analysis with the Univariate Cox regression test with the least absolute shrinkage and selection operator (LASSO) algorithm with the "glmnet" package in R. A 4 gene predictive signature was screened using the minimum criteria. Risk score for each patient was computed using the normalized expression level of each gene as well as its coefficients. The formula was the following: coefficient (gene i) derived from the Cox proportional hazards regression analysis. The TCGA LIHC cohort was split into two sub-groups on the basis of the median value of the risk score.
## Gsea and gsva
Gene ontology (GO) was employed to investigate potential functions based on the differential gene profiles of the two groups (|log2FC| ≥ 1, FDR <0.05). The hallmark gene sets v. 7.4 were obtained from MSigDB and used as the reference dataset. GSVA was carried out with the GSVA package of R, and parameters were the following: min. size = 10, max. size =500, verbose = Ture, and parallel. size = 1. Difference detection was conducted with Limma, with the filtering threshold |logFC| > 0.1 and adjusted P < 0.05. GSEA was conducted with GSEA software (version 4.1.0), with the number of permutations as 1000 [bib_ref] oncoPredict: An r package for predicting in vivo or cancer patient drug..., Zeng [/bib_ref]. The gene sets with nominal P < 0.05 and FDR q <0.05 were considered statistically significant. An interaction network of 14 circadian-related genes was recapitulated with STRING (http://string-db.org).
## Chemotherapeutic effectiveness analyses
The oncoPredict R package was used to assess the chemotherapeutic sensitivity of LIHC patients in the TCGA-LIHC cohort (15). Transcriptome data and IC50 values of 20 different cell lines were obtained from the GDSC database (https:// www.cancerrxgene.org/) in order to corroborate the predictive value of CRRS for response to chemotherapy. Microarray RNA expression data retrieved from the GDSC dataset was normalized using Robust Multi-Array Average (RMA).
# Statistical analysis
R software (version 4.1.0) was used for statistical analysis. The Wilcoxon Signed-rank test was used for continuous variables from the bioinformatical analyses. Violin diagrams and boxplots were both created with the ggplot2 package. Chisquare test was employed for categorical variables, and results were visualized using the ggplot2 package.
# Results
## Expression profile of circadian clock genes and clinical features in tcga lihc cohort
We included circadian clock genes (CCGs), including both core clock genes and clock control genes that were found in previous studies in our analysis [bib_ref] Circadian clock genes and the transcriptional architecture of the clock mechanism, Cox [/bib_ref]. SNVs and CNVs related to CCGs were investigated. Samples with mutated or CNVs CSNK1D, PER2, NR1D1, CUL1, and BTRC accounted for 1% of all samples from the TCGA dataset . Additionally, the CNV profile of core CCGs could explain the abnormal expression of the CCGs. Expression of CSNK1D was increased in tumor CNV amplification samples relative to nonapplication samples and normal tissues, which corresponds with the CNV profile of the CCGs. The expression levels of CUL1, NR1D1, and PER2 were lower in tumor deletion samples than in other tumor and normal tissue samples . Methylation and expression profiles of the CCGs between normal liver tissue and hepatocellular carcinoma samples is shown in the heatmap . High methylation was observed for CLU1, PER3, PER3, RORA, and PER1, but was not associated with RNA expression; these genes showed high expression in the LIHC datasets. Low expression was observed for FBXL21 and RORB, but the methylation levels were not elevated. For FBXL21, only the partial methylation site was elevated. No significant relationship was observed between the methylation and gene expression of CCGs in LIHC.
## Construction of the circadian gene based signature in the tcga lihc cohort
To investigate the predictive value of these differentially expressed CCGs, we first carried out univariate analysis of the standardized expression of the CCGs from the LIHC datasets in the training sets to detect prognostic CCGs. In the TCGA LIHC cohort, 9 circadian genes were differentially expressed between tumor and tumor-adjacent normal tissues. Three candidate genes were significantly associated with OS, as detected by univariate Cox proportional regression analysis . In addition, we performed LASSO Cox proportional hazard regression to screen the ultimate CCGs for the risk-score models, and found that expression of 4 CCGs correlated with OS in LIHC patients . Therefore, we constructed a risk-score model for predicting the prognosis of LIHC patients using the calculation formula described in the Methods. We identified PER1, CRY2, CSNK1D, and FBXL21 as risk genes in the LIHC risk-score model . Survival analysis revealed that there was a significant difference between the high-risk group and the low-risk group for OS, and being in the high-risk group significantly correlated with an poorer prognosis (LIHC: p < 0.0001) .
## Validating the signature in the icgc cohort
We confirmed the accuracy and stability of the prognostic risk-score model using the testing sets, which included the LIRI-JP cohorts obtained from ICGC. OS was used as the indicator for comparison of the groups, and samples were split into low and high risk-score groups according to the calculated risk score. The formula was as described in the Methods. For the testing set of the LIRI type, 230 samples were separated into low and high risk-score groups. Survival analysis showed that there was a significant difference between the high and low risk-score groups (p < 0.0001, HR: 1.493, 95% CI: 1.248 to 1.787) . To summarize, our results confirmed that the risk-score model based on CCGs signature was stable and accurate in predicting the prognosis of patients .
Identification of the circadian-based signature as an independent prognostic factor for HCC We then investigated the accuracy and stability of the risk score for use as a clinical indicator. We analyzed seven variables, including age, gender, tumor stage, and risk-score using the TCGA LIHC and ICGC LIRI groups, with multivariate analysis and Cox proportional hazard regression, respectively . Our analysis showed that risk-score was an independent prognostic indicator in both LIHC and LIRI patients. We then constructed ROC curves for the different variables to evaluate the risk-score as classifiers, and the AUC was calculated and used as the basis for evaluation (LIHC: AUC 0.71; LUSC: AUC 0.76) . Our results indicated that the risk-score showed accuracy and predictability for comparing other clinical characteristics in liver cancer samples. The protein expression level of these genes in liver hepatocellular carcinoma samples was also detected via immunohistochemistry (IHC), as shown in .
## Functions and pathways correlated with the circadian gene-based signature
In the TCGA dataset, genes that were differentially expressed in the high-risk and low-risk groups were subjected to GSEA to evaluate the hallmark, KEGG, and GO pathways . The results indicated that tumorigenesis pathways related to G2M checkpoint, E2F target, MYC targets, mitotic spindle, and mTORC1 signal were enriched in the high-risk group, while metabolism pathways related to xenobiotic, fatty acid, bile acid, adipogenesis, coagulation, oxidative phosphorylation, and peroxisome were enriched in the low-risk group. . In the TCGA dataset, KEGG enrichment analysis based on GSEA analysis suggested that CCGs modulate the cAMP signaling pathway, Epstein−Barr virus infection, human papillomavirus infection, neuroactive ligand−receptor interaction, spinocerebellar ataxia, and viral carcinogenesis. Additionally, GO analysis based on GSEA analysis suggested that CCGs modulate carbohydrate-binding, leukocyte apoptotic process, lipoprotein metabolic process, modulation of the process of another organism, response to acid chemical, response to heat, and response to nutrient, and the differential expression of genes between the high-risk and low-risk groups also supported that conclusion.
To identify the key player in LIHC, we developed the gene coexpression network in the TCGA-LIHC and ICGC-LIRI datasets. We investigated the correlation between risk score-related CCGs and gene modules. The heatmap revealed the correlations between the modules and risk score genes in the TCGA-LIHC and ICGC-LIRI dataset . Further, we found that CRY2 was negatively correlated with MYOGENESIS and GLYCOLYSIS, while others were positively correlated. Additionally, CRY2 and PER1 were negatively correlated with MITOTIC_SPINDLE and G2M_CHECKPOINT, while CSNK1D and FBXL21 were positively correlated .
## Drug response and signature pathway profile based on risk score group
We also predicted drug sensitivity according to riskScore. Entospletinib, KU−55933, PF−4708671, Ribociclib, and WZ4003 showed a positive correlation with risk score, indicating their potential role as drugs targeting liver hepatocellular carcinoma, both in TCGA-LIHC and ICGC-LIRI . We next assessed the association between risk score and signature pathway in the TCGA dataset. We used the ssgsea algorithm, and found that higher risk score correlated with a higher arachidonic acid metabolism, cardiolipin biosynthesis, Hexosamine Biosynthesis, and m6A molecular cancer genesets, while lower risk score was associated with higher Valine Leucine and Isoleucine Biosynthesis . These findings were validated in the other datasets supporting a difference in metabolism between the high-risk and low-risk samples.
# Discussion
Liver cancer diagnosis rates have tripled in the past three decades, and mortality has also increased [bib_ref] Cancer statistics, Siegel [/bib_ref]. Thus, is it critical to identify effective biomarkers that allow creation of accurate predictive models for HCC patient OS. Ongoing investigation of metabolomics has revealed the role of metabolism in the incidence and development of cancer [bib_ref] Non-coding RNAs rewire cancer metabolism networks, Lin [/bib_ref]. The circadian clock affects and drives many biological processes, and disruption to the circadian clock is implicated in breast and colorectal cancers [bib_ref] Circadian rhythm disruption promotes lung tumorigenesis, Papagiannakopoulos [/bib_ref] [bib_ref] Circadian gene clock contributes to cell proliferation and migration of glioma and..., Samuelsson [/bib_ref] , with dysregulation of CCGs correlated with cancer progression [bib_ref] Deregulated expression of the clock genes in gliomas, Chen [/bib_ref]. In the current study, we established a predictive model and demonstrated that a dysregulated circadian clock correlated with hepatocellular carcinoma. The risk score model incorporating the four genes CRY2, CSNK1D, FBXL21, and PER1, based on the TCGA-LIHC dataset, was shown to be robust. These four genes have been demonstrated in previous studies to be associated with tumor development and cancer metastasis [bib_ref] CRY2 missense mutations suppress P53 and enhance cell growth, Chan [/bib_ref] [bib_ref] Structure, regulation, and (patho-)physiological functions of the stress-induced protein kinase CK1 delta..., Xu [/bib_ref] [bib_ref] PER1 as a tumor suppressor attenuated in the malignant phenotypes of breast..., Liu [/bib_ref] [bib_ref] Upregulation of stressinduced protein kinase CK1 delta is associated with a poor..., Liu [/bib_ref]. CRY2 codes for a flavin adenine dinucleotide-binding protein that acts as a regulator of the circadian clock, and may promote growth of tumor cells [bib_ref] Upregulation of stressinduced protein kinase CK1 delta is associated with a poor..., Liu [/bib_ref]. CSNK1D encodes an enzyme that has many roles, including in circadian rhythm, response to DNA damage, the cell cycle, functions of the cytoskeleton, and others (27, 28). FBXL21 is a gene that plays a role in circadian clock oscillation by mediating ubiquitination and stabilization of cryptochromes (29). PER1 is important in the maintenance of the circadian rhythm in cells, and its down-regulation can enhance tumor growth [bib_ref] PER1 as a tumor suppressor attenuated in the malignant phenotypes of breast..., Liu [/bib_ref]. Based on these genes, the current study showed that patients with a high-risk score had significantly reduced OS. Our predictive model was robust and accurate in correlating clinicopathological data and outcomes with higher risk scores.
We used GSEA analysis to examine the biological mechanisms driving outcomes for the two patient groups. We found that the top-upregulated HALLMARK gene sets in patients in the high-risk group were primarily associated with cell cycle-related pathways, which form the canonical signaling pathways involved in the initiation, invasion, and metastasis of tumor cells. In addition to the three common signal pathways "DNA-Repair", "Myc-Targets-V1", and "Myc-Targets-V2", two other significantly enriched immune-related pathways were identified in patients in the low-risk group. The mechanistic target of the rapamycin (mTOR) signaling pathway is closely related to immune and inflammatory effects and plays an important role in activation and differentiation.
We also that predicted Entospletinib, KU−55933, PF−4708671, Ribociclib, and WZ4003 may be potential drugs for high-risk LIHC patients. Notably, a previous study supported Entospletinib monotherapy in patients with relapsed or refractory chronic lymphocytic leukemia previously treated with B-cell receptor inhibitors: the results of a phase 2 study [bib_ref] Entospletinib monotherapy in patients with relapsed or refractory chronic lymphocytic leukemia previously..., Awan [/bib_ref] showed that Ribociclib was cytotoxic and reduced cell proliferation rate. The effect on cell viability was enhanced when Ribociclib was combined with progesterone and/or mitotane. Therefore, the risk score model can also be applied to predict potential drugs for treatment of LIHC. Collectively, these results indicate the accuracy and potential wide application of our model based on 4 genes for assessing the association between circadian dysregulation and cancer incidence. There are several limitations to this study. It is a bioinformatic study without sample verification or in vitro and in vivo experiments. Moreover, the current study only focuses on gene transcription regulation. Other regulatory layers such as epigenetic, post-transcription, translation efficiency, and posttranslation modification also affect the chronophysiology and pathology [bib_ref] In silico integrative analysis of multi-omics reveals regulatory layers for diurnal gene..., Jiang [/bib_ref]. Future investigations on experiments and prospective trials are needed to assess the predictive score of gene signature and its mechanisms.
# Conclusion
This study examined the potential role of the expression of circadian-related genes in the development of HCC. We found that 4 genes could be used to accurately predict the OS for HCC patients in the TCGA LIHC and ICGC (LIRI-JP) cohorts. Comparison of drug response and signature score between high-risk and low-risk groups. Further, we identified one of the four genes, CRY2, as a possible molecular target for HCC incidence and progression. These findings provide a theoretical basis for ongoing study of circadian-related genes for predicting and treating HCC.
# Data availability statement
The original contributions presented in the study are included in the article/Supplementary Material. Further inquiries can be directed to the corresponding author.
[table] TABLE 1: Demographical and clinical characteristics. [/table]
|
10.3390/s22134870
|
CCBY
|
9269240
|
35808366
|
s2orc_pubmed_articles
|
Biogenic VOCs Emission Profiles Associated with Plant-Pest Interaction for Phenotyping Applications
Citation: Valencia-Ortiz, M.; Marzougui, A.; Zhang, C.; Bali, S.; Odubiyi, S.; Sathuvalli, V.; Bosque-Pérez, N.A.; Pumphrey, M.O.; Sankaran, S. Biogenic VOCs Emission Profiles Associated with Plant-Pest Interaction for Phenotyping Applications. Sensors 2022, 22, 4870.
# Introduction
Climate change strongly influences pest population dynamics that may increase crop damage and new strategies are needed to face challenges from pests and pathogens.
In recent years, researchers have focused on volatile organic compounds (VOCs) as a tool to protect plants from stress and boost crop production [bib_ref] Exploiting Plant Volatile Organic Compounds (VOCs) in Agriculture to Improve Sustainable Defense..., Brilli [/bib_ref]. These VOCs play essential roles in many ecological functions [bib_ref] The Network of Plants Volatile Organic Compounds, Vivaldo [/bib_ref] , such as plant defenses to biotic and abiotic stresses, providing information about crop status, mutualists, competitors [bib_ref] Plant Volatiles, Baldwin [/bib_ref] , and promoting plant growth [bib_ref] Microbial Volatiles as Plant Growth Inducers, Fincheira [/bib_ref]. From an integrated pest management perspective, crop disease resistance is an essential component to protect crop yields against pests and pathogens.
In general, plants under attack activate signaling pathways leading to the expression of plant resistance mechanisms. Gene expression responses and leaf metabolic potential can modify the proportion of VOC biogenesis and emission levels [bib_ref] Metabolic and gene expression controls on the production of biogenic volatile organic..., Monson [/bib_ref]. The VOC biogenesis can be either constitutive or induced [bib_ref] Abiotic Stresses and Induced BVOCs, Loreto [/bib_ref]. The emission of constitutive VOCs occurs regardless of whether the plant is under stress, while the emission of induced VOCs only occurs under conditions of stress [bib_ref] Quantitative Patterns between Plant Volatile Emissions Induced by Biotic Stresses and the..., Niinemets [/bib_ref]. For example, the release of biogenic VOCs in the grapevine by exogenous stimuli suggests a plant defense response against pathogens, resistance induction, and biomarker use [bib_ref] Biogenic Volatile Organic Compounds in the Grapevine Response to Pathogens, Beneficial Microorganisms,..., Lazazzara [/bib_ref]. Therefore, biogenic VOCs can serve as biomarkers to inform about the plant defense mechanisms and resistance levels. Given the potential to detect VOCs rapidly and non-invasively, biomarkers can also assist as a robust phenotyping tool. However, the application of biogenic VOCs for phenotyping is still limited due to their inherent reactivity and low concentrations [bib_ref] Methods in Plant Foliar Volatile Organic Compounds Research, Materić [/bib_ref]. There is a need for further evaluation of such concepts to assess the applicability of phenotyping methods to study crop resistance to pests and pathogens. The present study was designed to evaluate the variability in VOC emission profiles as phenotypes using two specific studies involving potato and wheat plants.
In potato production systems, nematodes limit the crop yield and tuber quality worldwide, where annual losses can be estimated to be around 78 billion USD or 10-15 percent of total yield [bib_ref] Nematodes affecting potato and sustainable practices for their management, Lima [/bib_ref]. Columbia Root-knot nematode (Meloidogyne chitwoodi) causes significant damage to potato tubers, affecting the market value in the fresh pack and processing industries. Solanum bulbocastanum, a wild potato species, was identified as the first source of genetic resistance to M. chitwoodi (Mc1) [bib_ref] Interspecific Somatic Hybridization between Solanum tuberosum L. and S. bulbocastanum Dun. as..., Austin [/bib_ref]. The histological analysis of nematode resistance from S. bulbocastanum (SB22) introgressed into cultivated potato (PA99N82-4, an advanced breeding clone) indicated restriction in nematode feeding site formation. This resistant response is associated with reactive oxygen species (ROS) accumulation and hypersensitive responses; where salicylic acid, polyamines, and suberin are important resistant mediators [bib_ref] Transcriptome Profiling of Resistance Response to Meloidogyne Chitwoodi Introgressed from Wild Species..., Bali [/bib_ref]. A previous study on PA99N82-4 resistance response found that calcium plays an important role in the hypersensitive response mechanism against M. chitwoodi [bib_ref] Calcium Is Involved in the R Mc1 (Blb)-Mediated Hypersensitive Response against Meloidogyne..., Davies [/bib_ref]. However, a resistance-breaking pathotype of Mc1, Roza was identified in Prosser, WA. This newly emerged pathotype could successfully penetrate and establish feeding sites in SB22 plants and thus complete its life cycle. In the first study, SB22 plants were inoculated with Mc1 and its pathotype Roza to compare the VOCs emission profiles resulting from resistant and susceptible responses.
In the second study, Hessian fly (Mayetiola destructor) infestation in wheat (Triticum aestivum) was assessed in resistant ('Hollis') and susceptible ('Alturas') cultivars. The Hessian fly is a destructive insect affecting wheat in the United States and other wheatproducing regions worldwide [bib_ref] Hessian Fly, Mayetiola destructor (Say) (Diptera: Cecidomyiidae), Shukle [/bib_ref]. The impact is through canopy and grain losses, and deterioration of the grain quality. The larvae feed on the stem and sap, resulting in low canopy vigor. Moreover, the insect population can grow intensively as several life cycles are supported within a year, thereby making control measures challenging [bib_ref] Hessian Fly Larval Feeding Triggers Enhanced Polyamine Levels in Susceptible but Not..., Subramanyam [/bib_ref]. Thus, Hessian fly resistance in wheat cultivars has proved to be an effective management strategy [bib_ref] Impact of Tillage Practices on Hessian Fly-Susceptible and Resistant Spring Wheat Cultivars, Castle Del Conte [/bib_ref]. Sadeghi et al.reported a significant concentration of jasmonic acid in resistant wheat (cv. Hollis) compared to susceptible wheat (cv. Alturas) infested with M. destructor. Similarly, Subramanyam et al. [bib_ref] Hessian Fly Larval Feeding Triggers Enhanced Polyamine Levels in Susceptible but Not..., Subramanyam [/bib_ref] reported enhanced production of free polyamines, putrescine, spermidine, and spermine in susceptible cultivars upon Hessian fly infestation in comparison to resistant cultivars and the respective controls. Some of these compounds may be associated with the changes in VOC emission profiles, which need to be evaluated further.
Although the above-mentioned studies highlight the potential of VOCs as biomarkers, VOC emission profile detection has not been studied in the context of phenotyping and the onset of resistance. The research described in this paper aimed to evaluate the variability in VOC emission profiles in potato and wheat plants resulting from different responses to the pests. Gas chromatography-flame ionization detection (GC-FID) analyses were performed to capture the VOC emission profiles from diverse samples.
# Materials and methods
## Potato-nematode study
SB22 plants were grown from 4-week-old tissue culture seedlings. Eighteen plantlets were transferred from tissue culture media to pre-sterilized 1:1 sand-soil mix and left to grow for 3-weeks in a greenhouse (24 - C/18 - C with 16 h photoperiod) with regular watering. A total of six plants were inoculated with 1000 J2s' of Mc1 (resistant), another six with 1000 J2s' of Roza (susceptible), and six plants were used as control (inoculated with sterilized distilled water). The VOCs emission profiles were sampled and analyzed with a GC-FID system at 2, 5, 10, and 25 days after inoculation (DAI). The J2 penetration was confirmed using microscopic evaluation. Details on nematode egg extraction, hatching, and inoculation can be found in Bali et al. [bib_ref] Transcriptome Profiling of Resistance Response to Meloidogyne Chitwoodi Introgressed from Wild Species..., Bali [/bib_ref]
## Wheat-insect study
The initial plant growth and infestation were performed at the University of Idaho, Moscow, Idaho. Two wheat cultivars, Hollis (resistant) and Alturas (susceptible), were used in this experiment, and plants were infested with Hessian fly. Wheat seeds (15 seeds per pot) were planted in sixteen 10 cm pots filled with soil mixture, where eight pots (4 pots/cultivar × 2 cultivars) were used for infestation with Hessian flies and another eight (4 pots/cultivar × 2 cultivars) were used as control. In addition, two pots containing no plants (were used as blank controls during VOCs emission profiles analysis to eliminate the background noise from other materials). Wheat seedlings were grown at 24 - C with a 16 h photoperiod and irrigated with nutrient water regularly in transparent plexiglass cages (53-cm length × 51-cm width × 51-cm height), which were used to prevent the escape of Hessian flies during the infestation. Adult Hessian flies (five females and four males) were introduced in each cage 11 days after planting eight pots of the wheat seedlings, intended for eggs to hatch and infest the plants. Wheat seedlings were sampled to evaluate VOC emission profiles and analyzed with the GC-FID system at about 18 and 26 days after planting. A more detailed procedure of Hessian fly establishment and infestation for resistant screening can be found in Schotzko and Bosque-Pérez [bib_ref] Relationship between Hessian fly infestation density and early seedling growth of resistant..., Schotzko [/bib_ref].
## Sampling and data collection
The sampling of VOCs in both studies was performed using headspace sampling using a solid-phase micro-extraction (SPME) fiber. The SPME fiber was made of 0.65 µm polydimethylsiloxane/divinylbenzene (Supelco Co., Bellefonte, PA, USA). For the potato plants, 0.5 g of root tissue was added to 325 g/mL sodium chloride solution, prior to the sampling of the headspace. The method details can be seen in Iyer et al. [bib_ref] Study of the Early Events Leading to Cassava Root Postharvest Deterioration, Iyer [/bib_ref] , also described in Marzougui et al. [bib_ref] Evaluation of Biogenic Markers-Based Phenotyping for Resistance to Aphanomyces Root Rot in..., Marzougui [/bib_ref]. Data were collected at 2, 5, 10, and 25 DAI. Dynamic headspace sampling with 35 mL/min airflow was used for sampling wheat plants. The analysis was also performed on control (blank) pots without plants. Data were collected at about 18 and 26 days after planting (7 and 15 days after infestation). Since some of the replicates did not have a successful infestation, the datasets across time points were combined to perform a meaningful statistical assessment. Both studies incorporated 50 min sampling time with SPME fibers. The GC-FID system, Hewlett Packard 5890 Series II (Agilent Technologies, Wilmington, DE, USA), was used for analysis. Inlet and detector temperatures were set at 200 - C, respectively. The settings were 33 - C start temperature (5-min hold), ramp rate 2 - C/min to 50 - C and followed by ramp rate 5 - C/min to 225 - C (5-min hold). The GC-FID analysis protocol was similar to those described in Marzougui et al. [bib_ref] Evaluation of Biogenic Markers-Based Phenotyping for Resistance to Aphanomyces Root Rot in..., Marzougui [/bib_ref] and Sangjan et al. [bib_ref] Identification of Volatile Biomarkers for High-Throughput Sensing of Soft Rot and Pythium..., Sangjan [/bib_ref].
# Data analysis
GC-FID data were preprocessed in MATLAB (2021a, The MathWorks, Natick, MA, USA) using a preprocessing protocol developed by Marzougui et al. [bib_ref] Evaluation of Biogenic Markers-Based Phenotyping for Resistance to Aphanomyces Root Rot in..., Marzougui [/bib_ref] , which includes signal extraction, peak alignment, data matrix reduction, background signals removal (with the help from blank samples), and identification of retention time of peaks that were present in at least two replicates of each condition/treatment. The VOC peaks that occurred only once were not considered for further analysis. R software (release 4.1.1, http://www.r-project.org/ (accessed on 10 January 2022) was then used to present the data in a Venn diagram to visualize the number of common and unique peaks between different treatments within a crop. Common peaks at specific retention time (RTs) refer to the VOC peaks present in both treatments (control and infested), while unique peaks with specific RTs refer to the VOC peaks present only in infected/infested samples. Common and unique peak RTs were extracted using 'unique' and 'setdiff' functions and presented using the ggplot2 package in the R program. Unpaired t-tests were used to compare mean differences in peak intensities of common RTs among treatments. One-way ANOVA and post-hoc Tukey's test were also used to evaluate the peak intensity comparisons across DAIs. Finally, Python (Version 3.8.0, interpreter-Spyder) was used to arrange the peak intensity and RTs by treatment to display the heatmap of averaging peak intensities across RTs and treatment.
# Results and discussion
## Voc profiles from potato plants
Plants produce constitutive VOC emissions regardless of stress conditions, while induced VOC emissions occur only under stress conditions [bib_ref] Quantitative Patterns between Plant Volatile Emissions Induced by Biotic Stresses and the..., Niinemets [/bib_ref]. The Venn diagram [fig_ref] Figure 1: Venn diagrams showing common and unique peaks represented by different rete times... [/fig_ref] developed using GC-FID data provided quantitative data (number of peaks) on constitutive (common peak RTs) and induced volatiles (unique peak RTs). The total number of common and unique peaks RTs representing VOC emission profiles associated with Roza inoculated, Mc1 inoculated, and control samples across different DAI after data preprocessing was 38. The largest number of unique peaks were found at 2 DAI for SB22 inoculated with Mc1 (resistance response) [fig_ref] Figure 1: Venn diagrams showing common and unique peaks represented by different rete times... [/fig_ref]. This early time point (2 DAI) is critical for SB22 resistance response to Mc1, as, at this time point, the nematode penetrates the root system to try and establish a feeding site. The VOCs emission profiles may indicate the activation of immune responses of the SB22 plants, which subsequently restricts the feeding site formation. Bali et al. [bib_ref] Transcriptome Profiling of Resistance Response to Meloidogyne Chitwoodi Introgressed from Wild Species..., Bali [/bib_ref] reported large differential gene expression in PA99N82-4 (introgression line with nematode resistance from SB22) at 48 h after Mc1 inoculation (also at 7, 14, and 21 DAI), and the VOCs could be associated with the change in the gene expression that majorly represents the defense responses. The histological analysis in the study reported that PA99N82-4 plants were able to restrict the establishment of the feeding site 48 h after inoculation.
Most common peaks were found at 5 and 10 DAI [fig_ref] Figure 1: Venn diagrams showing common and unique peaks represented by different rete times... [/fig_ref] for samples inoculated with both Mc1 and Roza. After combining the VOC peaks data across multiple time points, 23 common peak RTs were detected between Mc1 inoculated, Roza inoculated, and control plants [fig_ref] Figure 2: Venn diagram showing common and unique peaks represented by different times extracted... [/fig_ref]. Out of 23 common peaks RTs, two peaks with RT of 25.8 min and 43.6 min showed significant differences in intensity between Mc1 inoculated and control plants at 25 DAI [fig_ref] Figure 3: Peak intensity of three common retention times across different days after inocu... [/fig_ref] ,c, p-value < 0.05). In addition, the ANOVA of Mc1 VOC peak intensity (25.8 min) indicated significant differences based on the DAIs, where the average peak intensity increased significantly across DAIs [fig_ref] Figure 3: Peak intensity of three common retention times across different days after inocu... [/fig_ref]. ANOVA also showed some significant difference in VOC peak intensity (25.8 min) on Roza; however, this average peak intensity did not differ significantly from the control. The average peak intensity of Mc1 plants was 9.5 and 1.6 times those of control at 25.8 min and 43.6 min RTs, respectively. These findings suggest a further increase in constitutive VOC peak intensities as a sign of SB22 defense response to the Mc1 attack. In the case of Roza inoculated plants, out of 23 common peaks RTs, only one peak (43.6 min RT) showed significant differences at 25 DAI [fig_ref] Figure 3: Peak intensity of three common retention times across different days after inocu... [/fig_ref] , p-value < 0.05) with average peak intensity 1.4 times that of control plants. In contrast, there was a significant reduction (2.2 times, p-value < 0.05) in average peak intensity (27.4 min RT) of Roza inoculated plants than those from the control plants at 5 DAI [fig_ref] Figure 3: Peak intensity of three common retention times across different days after inocu... [/fig_ref]. The reduction of this VOC peak may potentially be related to the susceptibility of SB22 plants to the Roza and may be associated with the post-infection immune response to nematode attack. Most common peaks were found at 5 and 10 DAI [fig_ref] Figure 1: Venn diagrams showing common and unique peaks represented by different rete times... [/fig_ref] for samples inocu with both Mc1 and Roza. After combining the VOC peaks data across multiple points, 23 common peak RTs were detected between Mc1 inoculated, Roza inocul and control plants [fig_ref] Figure 2: Venn diagram showing common and unique peaks represented by different times extracted... [/fig_ref]. Out of 23 common peaks RTs, two peaks with RT o min and 43.6 min showed significant differences in intensity between Mc1 inoculated control plants at 25 DAI [fig_ref] Figure 3: Peak intensity of three common retention times across different days after inocu... [/fig_ref] ,c, p-value < 0.05). In addition, the ANOVA of Mc1 peak intensity (25.8 min) indicated significant differences based on the DAIs, wher average peak intensity increased significantly across DAIs [fig_ref] Figure 3: Peak intensity of three common retention times across different days after inocu... [/fig_ref]. ANOVA showed some significant difference in VOC peak intensity (25.8 min) on Roza; how this average peak intensity did not differ significantly from the control. The average intensity of Mc1 plants was 9.5 and 1.6 times those of control at 25.8 min and 43.6 Heatmap [fig_ref] Figure 4: Heat-map showing the peak intensity extracted from GC-FID data of samples from... [/fig_ref] of average peak intensities of VOCs across the combined dataset shows variations in VOCs emission profiles between control, Mc1 inoculated, and Roza inoculated SB22 plants. In general, VOC peaks at 21.5, 23.1, and 25.9 RTs showed the highest peak intensities across the samples, although the differences were not statistically significant. The literature suggests that VOCs can be indicators of pest/pathogen defense mec anisms. In soybean, Lin et al. [bib_ref] An (E,E)-α-Farnesene Synthase Gene of Soybean Has a Role in Defence against..., Lin [/bib_ref] reported the sesquiterpene (E,E)-α-farnesene as a ma VOC released during nematode infestation. Similarly, increase in α-farnesene and α-be gamotene sesquiterpenes was reported by Castorina et al. [bib_ref] Characterization of the Biogenic Volatile Organic Compounds (BVOCs) and Analysis of the..., Castorina [/bib_ref] in Vitis vinifera during nem atode Xiphinema index (Dagger nematode) attack. In Arabidopsis thaliana, the number nematodes that penetrated roots was reduced by the sesquiterpene nootkatone [bib_ref] The Plant Sesquiterpene Nootkatone Efficiently Reduces Heterodera Schachtii Parasitism by Activating Plant..., Habash [/bib_ref]. Oth compounds such as ascaridole and citronellal can also act as toxic compounds again Meloidogyne incognita [bib_ref] Medicinal Plant Volatiles Applied against the Root-Knot Nematode Meloidogyne incognita, De Freitas Silva [/bib_ref].
## Voc profiles from wheat plants
The number of common and unique VOC peaks from Hessian fly-infested and co trol plants and resistant (Hollis) and susceptible (Alturas) wheat plants are shown Venn diagrams [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. Infested Alturas plants did not show unique peaks, while i fested Hollis plants released three unique peaks [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref] , which suggests these thr unique RTs can be used as biomarkers to detect resistance to Hessian fly. In our rece study [bib_ref] Evaluation of Biogenic Markers-Based Phenotyping for Resistance to Aphanomyces Root Rot in..., Marzougui [/bib_ref] , we evaluated VOCs emission profile to assess pea plant responses (resista and susceptible cultivar) to Aphanomyces root rot. Similar results were found with mo peaks found in infected than control samples. In another wheat study [bib_ref] Volatile Cues Influence the Response of Rhopalosiphum padi (Homoptera: Aphididae) to Barley..., Jiménez-Martínez [/bib_ref] , a higher co centration of volatiles was observed in wheat cultivar (Lambert) susceptible to the Bar yellow dwarf luteovirus in comparison to transgenic-resistant genotype and control plan which could be associated with plant responses to aphid vector (Rhopalosiphum padi L.) The literature suggests that VOCs can be indicators of pest/pathogen defense mechanisms. In soybean, Lin et al. [bib_ref] An (E,E)-α-Farnesene Synthase Gene of Soybean Has a Role in Defence against..., Lin [/bib_ref] reported the sesquiterpene (E,E)-α-farnesene as a major VOC released during nematode infestation. Similarly, increase in α-farnesene and αbergamotene sesquiterpenes was reported by Castorina et al. [bib_ref] Characterization of the Biogenic Volatile Organic Compounds (BVOCs) and Analysis of the..., Castorina [/bib_ref] in Vitis vinifera during nematode Xiphinema index (Dagger nematode) attack. In Arabidopsis thaliana, the number of nematodes that penetrated roots was reduced by the sesquiterpene nootkatone [bib_ref] The Plant Sesquiterpene Nootkatone Efficiently Reduces Heterodera Schachtii Parasitism by Activating Plant..., Habash [/bib_ref]. Other compounds such as ascaridole and citronellal can also act as toxic compounds against Meloidogyne incognita [bib_ref] Medicinal Plant Volatiles Applied against the Root-Knot Nematode Meloidogyne incognita, De Freitas Silva [/bib_ref].
## Voc profiles from wheat plants
The number of common and unique VOC peaks from Hessian fly-infested and control plants and resistant (Hollis) and susceptible (Alturas) wheat plants are shown by Venn diagrams [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. Infested Alturas plants did not show unique peaks, while infested Hollis plants released three unique peaks [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref] , which suggests these three unique RTs can be used as biomarkers to detect resistance to Hessian fly. In our recent study [bib_ref] Evaluation of Biogenic Markers-Based Phenotyping for Resistance to Aphanomyces Root Rot in..., Marzougui [/bib_ref] , we evaluated VOCs emission profile to assess pea plant responses (resistant and susceptible cultivar) to Aphanomyces root rot. Similar results were found with more peaks found in infected than control samples. In another wheat study [bib_ref] Volatile Cues Influence the Response of Rhopalosiphum padi (Homoptera: Aphididae) to Barley..., Jiménez-Martínez [/bib_ref] , a higher concentration of volatiles was observed in wheat cultivar (Lambert) susceptible to the Barley yellow dwarf luteovirus in comparison to transgenic-resistant genotype and control plants, which could be associated with plant responses to aphid vector (Rhopalosiphum padi L.).
In this study, interestingly, both plant cultivars exhibited the same four common peaks [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. The greatest average peak intensity (p-value < 0.05) was found at 22.4 min RT where Hollis infested samples showed 4.3 times higher average peak intensity than Alturas infested samples [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. Since this peak was also found to be higher in Hollis control samples, further investigations may be needed. Heatmap shows that the three induced VOC peaks from Hollis-infested samples were at 21.4, 23.0, and 35.9 RTs [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. In this study, interestingly, both plant cultivars exhibited the same four common peaks [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. The greatest average peak intensity (p-value < 0.05) was found at 22.4 min RT where Hollis infested samples showed 4.3 times higher average peak intensity than Alturas infested samples [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. Since this peak was also found to be higher in Hollis control samples, further investigations may be needed. Heatmap shows that the three induced VOC peaks from Hollis-infested samples were at 21.4, 23.0, and 35.9 RTs [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref].
# Conclusions
In both potato and wheat studies, interestingly the resistant cultivars released a greater number of VOCs (peak RTs). For example, at 2 DAI, SB22 plants infested with Mc1 showed a higher number of induced VOCs. Additionally, at 25 DAI, constitutive VOCs of SB22 plants inoculated with Mc1 displayed a higher average peak intensity between 1.6- Comparison of (a) number of peaks from Hollis infested and control wheat data, (b) number of peaks from Alturas infested and control wheat data, and (c) peak intensity of four common retention times between both cultivars and treatments. * Significant differences from t-test analysis (p-value < 0.05). In this study, interestingly, both plant cultivars exhibited the same peaks [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. The greatest average peak intensity (p-value < 0.05) wa min RT where Hollis infested samples showed 4.3 times higher average than Alturas infested samples [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref]. Since this peak was also found Hollis control samples, further investigations may be needed. Heatmap three induced VOC peaks from Hollis-infested samples were at 21.4, 23.0 [fig_ref] Figure 5: Venn diagram showing common and unique peaks represented by different retention times... [/fig_ref].
# Conclusions
In both potato and wheat studies, interestingly the resistant cultiv greater number of VOCs (peak RTs). For example, at 2 DAI, SB22 plants inf showed a higher number of induced VOCs. Additionally, at 25 DAI, consti SB22 plants inoculated with Mc1 displayed a higher average peak intensit 9.5 times than those of the control samples. Similarly, samples from Holli
# Conclusions
In both potato and wheat studies, interestingly the resistant cultivars released a greater number of VOCs (peak RTs). For example, at 2 DAI, SB22 plants infested with Mc1 showed a higher number of induced VOCs. Additionally, at 25 DAI, constitutive VOCs of SB22 plants inoculated with Mc1 displayed a higher average peak intensity between 1.6-9.5 times than those of the control samples. Similarly, samples from Hollis only showed induced VOCs in infested samples (induced VOCs were absent in Alturas). The reported studies highlight the differences in VOC emission profiles between healthy (control) and infected/infested samples, as well as profile differences between resistant and susceptible cultivars. With this potential, VOC emission profiles can serve as a phenotyping tool to screen the plant materials non-invasively, especially at early time points. Further studies are required to determine changes in VOC emission profiles associated with diverse genes providing resistance to Hessian fly in multiple wheat cultivars.
The major benefit would be the screening of the same plant materials across several time points. To identify the compounds associated with peak RTs, analyses of samples using gas chromatography-mass spectrometer, and comparison with the National Institute of Standards and Technology (NIST) database is needed. Even so, the identification of compounds associated with VOC peak RTs remains challenging (e.g., [bib_ref] Evaluation of Biogenic Markers-Based Phenotyping for Resistance to Aphanomyces Root Rot in..., Marzougui [/bib_ref] [bib_ref] Identification of Volatile Biomarkers for High-Throughput Sensing of Soft Rot and Pythium..., Sangjan [/bib_ref]. In this regard, the VOC peaks or VOC emission profiles [bib_ref] Trends and Applications in Plant Volatile Sampling and Analysis, Tholl [/bib_ref] can be used as biomarkers. Several high-throughput sensing techniques such as e-noses and IMS-based systems can further increase the throughput in sampling, data acquisition, and analysis [bib_ref] Trends and Applications in Plant Volatile Sampling and Analysis, Tholl [/bib_ref]. Our previous study investigated a high-throughput VOC sensing system, a field asymmetric ion mobility spectrometer, with successful results for post-harvest potato rot detection for applications in storage [bib_ref] Feasibility of Volatile Biomarker-Based Detection of Pythium Leak in Postharvest Stored Potato..., Kothawade [/bib_ref] [bib_ref] Asymmetric Ion Mobility Spectrometry for Pre-Symptomatic Rot Detection in Stored Ranger Russet..., Kothawade [/bib_ref]. In our future studies, such high-throughput VOC sensing systems will be explored with an increase in the number of samples, replicates, and cultivars under both treatment conditions.
# Data availability statement:
The data presented in this study are available on request from the corresponding author.
[fig] Figure 1: Venn diagrams showing common and unique peaks represented by different rete times extracted from GC-FID data of wild potato species, SB22 plants infected with M. chi Race 1 (Mc1), and its pathotype Roza across different days after inoculation (DAI). (a) 2 DAI DAI, (c) 10 DAI, and (d) 25 DAI. [/fig]
[fig] Figure 2: Venn diagram showing common and unique peaks represented by different times extracted from GC-FID data of SB22 plants infected with M. chitwoodi Race 1 (Mc pathotype Roza by combining all time points (2, 5, 10, and 25 DAI). [/fig]
[fig] Figure 3: Peak intensity of three common retention times across different days after inocu SB22 plants with M. chitwoodi Race 1 (Mc1) and its pathotype Roza. Retention time: (a) 25.8 43.6 min, and (c) 27.4 min. * p-value < 0.05 from t-test analysis. Different letters in each co nematode race denote significant differences in peak intensities across DAIs at p < 0.05 usin test. [/fig]
[fig] Figure 4: Heat-map showing the peak intensity extracted from GC-FID data of samples from potato SB22 plants infected with M. chitwoodi Race 1 (Mc1) and its pathotype Roza by combining all time points (2, 5, 10, and 25 DAI). [/fig]
[fig] Figure 5: Venn diagram showing common and unique peaks represented by different retention times extracted from GC-FID data of samples from resistant and susceptible wheat infested with Hessian fly. Comparison of (a) number of peaks from Hollis infested and control wheat data, (b) number of peaks from Alturas infested and control wheat data, and (c) peak intensity of four common retention times between both cultivars and treatments. * Significant differences from t-test analysis (p-value < 0.05). [/fig]
[fig] Figure 6: Heat-map showing the peak intensity extracted from GC-FID data of samples from resistant and susceptible wheat infested with Hessian fly. [/fig]
[fig] Author: Contributions: Conceptualization, S.S., V.S., N.A.B.-P. and M.O.P.; methodology, M.V.-O., A.M., C.Z., S.B., S.O., V.S., N.A.B.-P., M.O.P. and S.S.; formal analysis, M.V.-O., A.M. and S.S.; resources, S.S., V.S., N.A.B.-P. and M.O.P.; data collection, A.M., C.Z., S.B. and S.O.; data curation, M.V.-O.; writing-original draft preparation, M.V.-O. and S.S.; writing-review and editing, M.V.-O., A.M., C.Z., S.B., S.O., V.S., N.A.B.-P., M.O.P. and S.S.; visualization, M.V.-O.; project administration, S.S., V.S., N.A.B.-P. and M.O.P.; funding acquisition, S.S., V.S., N.A.B.-P. and M.O.P. All authors have read and agreed to the published version of the manuscript. [/fig]
[fig] Funding: This activity was funded in part by the U.S. Department of Agriculture-National Institute for Food and Agriculture (USDA-NIFA) Agriculture and Food Research Initiative (AFRI) Competitive Project WNP06825 (accession number 1011741), Hatch Project WNP00011 (accession number 1014919), and Washington State University's College of Agricultural, Human, and Natural Resource Sciences Emerging Research Issues project. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. [/fig]
|
10.1038/s41598-018-29072-8
|
CCBY
|
6060137
|
30046139
|
s2orc_pubmed_articles
|
Leukocyte telomere length correlates with hypertrophic cardiomyopathy severity
[fig] Supplemental: Figure 1|LTL is negatively associated with HOCM severity. HOCM patients were stratified by tertiles of telomere length (A). Patients in the first tertile demonstrated significantly higher LVOT grad. max (D) and LVPWD (C) values, while LVEDD (B) was significantly lower in comparison to patients in tertile 3. The results are [/fig]
|
10.1016/j.amsu.2019.03.010
|
CCBY
|
6449703
|
30992989
|
s2orc_pubmed_articles
|
Lymphangiomatosis of the ileum with perforation: A case report and review of the literature
Lymphangiomatosis is a benign proliferation of lymph vessels. Lymphatic diseases can vary from small lymphangioma to generalized lymphangiomatosis, which is a rare condition and can have several clinical manifestations. The gastrointestinal tract may be affected, but the incidence in the intestinal wall is very low. We propose in our study a case of ileal lymphangiomatosis presenting with perforation, in which the diagnosis was made after the pathological analysis of the resected intestinal tract. Although rare and not described in the literature, intestinal lymphangiomatosis could manifest itself with acute abdomen and could be a surgical urgency. This disease should be considered when intestinal perforation is observed.
# Introduction
Lymphangiomatosis is a benign proliferation of lymph vessels. Lymphatic diseases can vary from small lymphangioma to generalized lymphangiomatosis, which is a rare condition and can have several clinical manifestations [bib_ref] Abdominal lymphangiomatosis with intestinal lymphangiectasia diagnosed by magnetic resonance lymphangiography: a case..., Valakada [/bib_ref] [bib_ref] Abdominal lymphangiomatosis in a 38-year-old female: case report and literature review, Lin [/bib_ref] [bib_ref] A case of primary intestinal lymphangiectasia, Won [/bib_ref] [bib_ref] Generalized lymphangiomatosis: radiologic findings in three pediatric patients, Yang [/bib_ref] [bib_ref] Intestinal lymphangiectasia: evaluation by CT and scintigraphy, Puri [/bib_ref] [bib_ref] Lymphangiomatosis: clinical overview, Blei [/bib_ref]. Although the underlying pathogenetic mechanism is unknown, it is generally considered as congenital malformation of the lymphatic system associated with alterations in the circulatory dynamics of the lymph [bib_ref] A case report with lymphangiomatosis of the colon, Jung [/bib_ref]. It can occur anywhere in the body and also abdominal lymphangiomatosis is reported, but in many cases it involves the mesentery, omentum, mesocolon and retroperitoneum. The gastrointestinal (GI) tract may be affected, but the incidence in the intestinal wall is very low [bib_ref] For the SCARE Group. The SCARE 2018 statement: updating consensus Surgical CAse..., Riaz [/bib_ref] [bib_ref] A clinical evaluation of lymphangioma of the large intestine: a case presentation..., Matsuda [/bib_ref]. In the few cases described in the literature, the symptomatology was characterized mainly by abdominal pain and bleeding. Aggressive surgery should be avoided in symptomless cases, because it is now known that these lesions are benign [bib_ref] A case report with lymphangiomatosis of the colon, Jung [/bib_ref].
We propose in our study a case of ileal lymphangiomatosis presenting with perforation, in which the diagnosis was made after the pathological analysis of the resected intestinal tract. In addition, the relevant medical literature on intestinal lymphangiomatosis was reviewed. This work has been reported in line with the SCARE criteria [bib_ref] For the SCARE Group. The SCARE 2018 statement: updating consensus Surgical CAse..., Riaz [/bib_ref].
## Presentation of case
A 41-year-old male presented to the Emergency Department with significant diffuse abdominal pain, nausea, vomiting and inability to pass gas or stool (constipation); the patient reported that these symptoms were present for 10 hours. He denied any significant family history of disease and he did not refer to major diseases or prior surgical interventions in his own medical history. During the physical examination, the subject was in discrete general conditions, collaborative, the sensory was intact, the decubitus indifferent, the breath eupneic and the pulse rhythmic. The vital parameters were preserved, and temperature was within normal limits. The abdominal examination revealed a flat abdomen, tender to palpation, painful to deep palpation on all quadrants, liver size appears within normal limits, Murphy's sign was negative, Blumberg's sign was positive, bowel sounds were absent. Rectal exploration indicated nothing significant. Laboratory values upon admission reported hemoglobin value of 15.5 g/dL, White Blood Cells 12.000/mmc, PCR 0.04 mg/dl. Upright abdominal films, then confirmed with computed tomography (CT) enhancement scan, revealed bowel distension and the presence of multiple gas-fluid levels. This framework suggested the presence of small bowel obstruction. As the etiology of the obstruction remained unidentified, the decision was made to perform a diagnostic laparoscopy. On entering the peritoneal cavity, the small bowel was examined: the intestinal loops were vital https://doi.org/10.1016/j.amsu.2019.03.010 Received 14 February 2019; Accepted 24 March 2019 and vascularized. At approximately 80 cm from the ileocaecal valve, a volvulus was identified, and in close proximity there was a long tubular structure, which proved to be a Meckel's diverticulum, with adhesions to parietal peritoneum. A lysis of adhesions was performed, the Meckel's diverticulum was divided at the base using a linear stapler (45 mm) and a surgical drain was placed. On the first postoperative day, the patient showed an acute abdomen: untreatable-tensely distended abdomen, painful to deep palpation on all quadrant; Murphy's sign was negative, Blumberg's sign was positive; bowel sounds were absent. Surgical drain took out enteric material and blood. The body temperature was 38°C. Laboratory values reported hemoglobin value of 12.7 g/dL, White Blood Cells 7.000/mmc, PCR 10.22 mg/dl. The patient again underwent urgent surgery. The intervention was an exploratory laparoscopy, then converted to open ileal resection for a microperforation of the bowel. The perforation site seemed to be distant above the previous diverticulum. About 20 cm of ileal resection was performed and side-to-side mechanical anastomosis was made. Two surgical drains were also placed. The postoperative course was regular, and the patient was discharged one the 10th post-operative day.
Histological examination showed stratification of fibrin and granulocytes on the serosa and presence of diffuse lymphangiomatosis (positive immunohistochemistry of D2-40 marker) involving the submucosa and, in some parts, the full-thickness muscular wall (Figs. 1 and 2).
# Discussion
Lymphangiomatosis is a multisystemic disorder, characterized by congenital malformation of the lymphatic system with channels and cystic spaces of varying size. It may affect all the areas of the body; therefore, symptoms and complications are related to localization. Usually in the course of the disease patients are asymptomatic at first, but then the abnormally proliferating lymphatic channels are capable of massive expansion and infiltration into surrounding tissues. Abdominal lymphangiomatosis is quite often reported, but in many cases it arises in the mesentery, omentum, mesocolon and retroperitoneum. The GI tract may be affected, but the incidence of lymphangioma of the intestinal wall is very low, and it is even rarer in the small bowel (less than 1%) [bib_ref] A clinical evaluation of lymphangioma of the large intestine: a case presentation..., Matsuda [/bib_ref] [bib_ref] Caecal lymphangioma: a rare cause of gastrointestinal blood loss, Rai [/bib_ref]. To the best of our knowledge, all cases reported in the literature with GI tract involvement are summarized in [fig_ref] Table 1: Papers about lymphangiomatosis-related disease [/fig_ref] [bib_ref] Abdominal lymphangiomatosis with intestinal lymphangiectasia diagnosed by magnetic resonance lymphangiography: a case..., Valakada [/bib_ref] [bib_ref] Abdominal lymphangiomatosis in a 38-year-old female: case report and literature review, Lin [/bib_ref] [bib_ref] A case report with lymphangiomatosis of the colon, Jung [/bib_ref] [bib_ref] Caecal lymphangioma: a rare cause of gastrointestinal blood loss, Rai [/bib_ref] [bib_ref] Cavernous mesenteric lymphangiomatosis mimicking metastasis in a patient with rectal cancer: a..., Hwang [/bib_ref] [bib_ref] Primary intestinal lymphangiomatosis of the ileum in an adult-the role of surgical..., Ilhan [/bib_ref] [bib_ref] Colonic lymphangiomatosis associated with anemia, Chung [/bib_ref] [bib_ref] Colonic lymphangiomatosis resolved after excisional biopsy, Lee [/bib_ref] [bib_ref] Small intestinal hemolymphangioma with bleeding: a case report, Fang [/bib_ref] [bib_ref] Abdominal kaposiform hemangioendothelioma associated with lymphangiomatosis involving mesentery and ileum: a case..., Dong [/bib_ref] [bib_ref] Lymphangiomatosis of the sigmoid colon -a rare cause of lower gastrointestinal bleeding:..., Lu [/bib_ref] [bib_ref] Huge lymphangiomatosis of the esophagus, Xue [/bib_ref].
There is only one case report of lymphangiomatosis involving the oesophagus [bib_ref] Huge lymphangiomatosis of the esophagus, Xue [/bib_ref] and one case report involving fundus of the stomach [bib_ref] Abdominal lymphangiomatosis in a 38-year-old female: case report and literature review, Lin [/bib_ref]. Lymphangiomatosis of the colon is described in four papers [bib_ref] A case report with lymphangiomatosis of the colon, Jung [/bib_ref] [bib_ref] Colonic lymphangiomatosis associated with anemia, Chung [/bib_ref] [bib_ref] Colonic lymphangiomatosis resolved after excisional biopsy, Lee [/bib_ref] [bib_ref] Small intestinal hemolymphangioma with bleeding: a case report, Fang [/bib_ref]. In these patients, the main symptoms were abdominal discomfort, bleeding and anaemia.
Lymphangiomatosis of small bowel is described in 6 works [1, [bib_ref] Caecal lymphangioma: a rare cause of gastrointestinal blood loss, Rai [/bib_ref] [bib_ref] Cavernous mesenteric lymphangiomatosis mimicking metastasis in a patient with rectal cancer: a..., Hwang [/bib_ref] [bib_ref] Primary intestinal lymphangiomatosis of the ileum in an adult-the role of surgical..., Ilhan [/bib_ref] [bib_ref] Small intestinal hemolymphangioma with bleeding: a case report, Fang [/bib_ref] [bib_ref] Abdominal kaposiform hemangioendothelioma associated with lymphangiomatosis involving mesentery and ileum: a case..., Dong [/bib_ref]. Patients had abdominal pain, bleeding, melena, anaemia, pedal edema, weakness and weight loss, but no case of perforation is reported.
In the majority of patients described in the literature, diagnosis was suggested by colonoscopy and biopsy, which mainly showed protruding submucosal lesions. The diagnostic workup was sometimes integrated by contrast-enhanced CT, showing marked thickening of the walls of the bowel loops.
In five papers [bib_ref] Caecal lymphangioma: a rare cause of gastrointestinal blood loss, Rai [/bib_ref] [bib_ref] Cavernous mesenteric lymphangiomatosis mimicking metastasis in a patient with rectal cancer: a..., Hwang [/bib_ref] [bib_ref] Primary intestinal lymphangiomatosis of the ileum in an adult-the role of surgical..., Ilhan [/bib_ref] [bib_ref] Small intestinal hemolymphangioma with bleeding: a case report, Fang [/bib_ref] [bib_ref] Abdominal kaposiform hemangioendothelioma associated with lymphangiomatosis involving mesentery and ileum: a case..., Dong [/bib_ref] , patients were treated with surgical resection of the affected part of intestine, whereas one patient improved on conservative management and was put on low-fat and high-protein diet [bib_ref] Abdominal lymphangiomatosis with intestinal lymphangiectasia diagnosed by magnetic resonance lymphangiography: a case..., Valakada [/bib_ref]. In all reported cases, follow-up was of short duration. In two cases patients continued to have symptoms after surgery [bib_ref] Abdominal lymphangiomatosis in a 38-year-old female: case report and literature review, Lin [/bib_ref] or after endoscopic mucosectomy [bib_ref] Colonic lymphangiomatosis associated with anemia, Chung [/bib_ref].
The histological features of the lymphangiomatosis are non-specific, and the definitive diagnosis requires the demonstration of an hyperproliferation of normal lymphatic vessels with normal endothelium, predominantly in the context of submucosa, with disruption of the muscular layer and sometimes of the serosa. This condition creates a locus minoris resistentiae, and this may explain the pathogenesis of the perforation, reported in our case. The involvement can be continuous or, more frequently, segmental. The impairment of the muscular layer and of the submucosal nerve plexus could also have contributed to the development of the intestinal obstruction, which was the symptom because our patient came to our attention. This suggests that also lymphangiomatosis should be taken into account among other rare causes of intestinal obstruction [bib_ref] Idiopathic intramural hematoma of sigmoid colon. A case report, De Santis [/bib_ref] [bib_ref] Mesenterical lymphangiomatosis causing volvulus and intestinal obstruction, De Vries [/bib_ref].
# Conclusion
In conclusion, lymphangiomatosis of the small bowel is a rare disease that has no specific clinical features, so it is easy for a clinician to make a misdiagnosis or to miss diagnosis. In some cases, surgical resection may be required to provide definitive histological diagnosis, as occurred in our cases. We want to share our experience about this, because, although rare and not described in the literature, intestinal lymphangiomatosis could manifest itself with an acute abdomen and surgical urgency. This disease should be considered when intestinal perforation is observed. In particular, the pathologist should keep it in mind in the differential diagnosis, when he analyses a case of perforation whose cause is not very clear and specified.
# Ethical approval
Ethical approval was not required.
# Sources of funding
# No funding
Author contribution Antonio Giuliani, Lucia Romano: Writing. Gino Coletti, Mohammad Walid A Fatayer, Giuseppe Calvisi: Images and contribution to the text.
Francesco Maffione, Chiara Muolo, Vincenzo Vicentini: Data collection.
Mario Schietroma, Francesco Carlei: Study design and review.
## Conflicts of interest
No conflict of interest.
## Research registration number
None.
## Guarantor
Prof. Francesco Carlei.
## Consent
Written informed consent was obtained from the patient for publication of this case report and accompanying images. A copy of the written consent is available for review by the Editor-in-Chief of this journal on request.
## Provenance and peer review
Not commissioned, externally peer reviewed.
[fig] Figure 1: Microscopic findings. (A) Hematoxylin and eosin staining of numerous dilated lymphatic vessels (4× magnification). (B) Immunohistochemical D2-40 expression (brown color) in dilated lymphatic vessels of the submucosa (10× magnification). (C) Subserous dilated lymphatic vessels with discontinuity of the muscular layer and serositis (H&E, 10× magnification). (For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) [/fig]
[fig] Figure 2: (A) D2-40 immunostaining shows positive reactivity for lining endothelial cells of lymphatic spaces in the muscular layer (10× magnification). (B) Numerous submucosal dilated lymphatic vessels (H&E, 10× magnification). (C) Lymphatic vessels that interrupt muscular layer (D2-40 stain, 10× magnification). [/fig]
[table] Table 1: Papers about lymphangiomatosis-related disease. [/table]
|
10.3390/s20216394
|
CCBY
|
7665147
|
33182460
|
s2orc_pubmed_articles
|
Use of Functional Linear Models to Detect Associations between Characteristics of Walking and Continuous Responses Using Accelerometry Data
Various methods exist to measure physical activity. Subjective methods, such as diaries and surveys, are relatively inexpensive ways of measuring one's physical activity; however, they are prone to measurement error and bias due to self-reporting. Wearable accelerometers offer a non-invasive and objective measure of one's physical activity and are now widely used in observational studies. Accelerometers record high frequency data and each produce an unlabeled time series at the sub-second level. An important activity to identify from the data collected is walking, since it is often the only form of activity for certain populations. Currently, most methods use an activity summary which ignores the nuances of walking data. We propose methodology to model specific continuous responses with a functional linear model utilizing spectra obtained from the local fast Fourier transform (FFT) of walking as a predictor. Utilizing prior knowledge of the mechanics of walking, we incorporate this as additional information for the structure of our transformed walking spectra. The methods were applied to the in-the-laboratory data obtained from the Developmental Epidemiologic Cohort Study (DECOS).
# Introduction
Use of wearable accelerometers has become increasingly common in studies of physical activity, aging, and obesity [bib_ref] Obesity history and daily patterns of physical activity at age 60-64 years:..., Cooper [/bib_ref] [bib_ref] Physical activity, sedentary behavior and the risk of overweight and obesity in..., Keane [/bib_ref] [bib_ref] Performance on fast-and usual-paced 400-m walk tests in older adults: are they..., Lange-Maia [/bib_ref] [bib_ref] Accelerometer-derived sedentary and physical activity time in overweight/obese adults with type 2..., Healy [/bib_ref] [bib_ref] Accelerometer assessment of physical activity in active, healthy older adults, Copeland [/bib_ref] [bib_ref] Use of accelerometry to measure physical activity in older adults at risk..., Pruitt [/bib_ref] [bib_ref] Assessment of free-living daily physical activity in older claudicants: Validation against the..., Gardner [/bib_ref] [bib_ref] Walking energetics, fatigability, and fatigue in older adults: The study of energy..., Richardson [/bib_ref]. Self-reported measures, such as questionnaires, have been widely used to assess physical activity (PA) previously [bib_ref] Use of accelerometry to measure physical activity in older adults at risk..., Pruitt [/bib_ref] [bib_ref] Physical activity as a novel outcome of total knee replacement: Comparing self-report..., Losina [/bib_ref]. One reason why we care about using accelerometry over self-reporting is because there are some populations whose self-reported measures can be inaccurate [bib_ref] Singh-Manoux, A. Physical activity and adiposity markers at older ages: Accelerometer vs..., Sabia [/bib_ref]. One such population is older adults. Questionnaires require individuals to recall their daily activities, which can be extremely difficult, particularly for older individuals [bib_ref] Validity and reliability of self report measures of physical activity: An information-processing..., Baranowski [/bib_ref]. Schrack et al. [bib_ref] Assessing the "physical cliff": Detailed quantification of age-related differences in daily patterns..., Schrack [/bib_ref] showed there are changes in the daily patterns and amount of PA as people age. Accelerometry is also an important tool that can be applied to the general population, because walking is the most popular form of aerobic physical activity.
Accelerometers offer a non-invasive and objective alternative to self-reporting methods. Advancements in data processing allow for the analysis of specific gait characteristics, such as cadence and asymmetry [bib_ref] Estimation of gait cycle characteristics by trunk accelerometry, Moe-Nilssen [/bib_ref]. Likewise, advancements in statistical methodology for analysis of high dimensional data have opened up new paths for analyzing more complex and potentially more informative summaries of accelerometry data. Accelerometers are electro-mechanical devices that measure acceleration along three orthogonal axes. They are often worn on a person's waist or wrist, and they provide high frequency, high-throughput data represented by three time series of acceleration measurements [bib_ref] Prediction of sustained harmonic walking in the free-living environment using raw accelerometry..., Urbanek [/bib_ref] [bib_ref] Accelerometer output and its association with energy expenditure in persons with multiple..., Sandroff [/bib_ref]. The data are typically collected at the sub-second level (usually between 10 to 100 observations per second); however, most studies aggregate the data over one minute epochs, or windows, and often, using a user specified threshold, the data are characterized into activity counts per minute. While thresholding methods are useful in describing the timing and duration for certain levels of PA, many nuances of the data are lost. For example, it is not possible to evaluate how a person is walking from activity count summaries. Especially in certain populations, such as older or obese populations, this raises the question as to whether a more detailed quantification of the walking signal can provide additional information. For example, if an older person's legs are bothering them, we may detect signs of limping that could help explain the low levels of PA. Our strategy is to extract detailed information from the raw accelerometry signal during periods of walking. We transform this information into useful quantities, and then we build regression models to associate characteristics of walking with continuous responses.
To illustrate the complex nature of the data collected, [fig_ref] Figure 1: Triaxial accelerometer data from the 400 m walk for a single individual [/fig_ref] shows the raw data collected from a triaxial accelerometer (Actigraph GT3X+) for a single individual during an in-laboratory 400 m walk. The top left panel of [fig_ref] Figure 1: Triaxial accelerometer data from the 400 m walk for a single individual [/fig_ref] presents the entire 400 m walk, where each axis is shown in a different color. With just over 5 min worth of data, the characteristics of the signal are nearly impossible to visualize. The top right panel of [fig_ref] Figure 1: Triaxial accelerometer data from the 400 m walk for a single individual [/fig_ref] shows a 10 s window of the same data. At this scale, we can discern a fairly periodic signal. In the bottom row of [fig_ref] Figure 1: Triaxial accelerometer data from the 400 m walk for a single individual [/fig_ref] , we present the vector magnitude of the same data presented in the top row. We can see that the periodic nature of the data are preserved while information about the three dimensional direction is lost. However, in free-living data collection, it is difficult to control the orientation of the device when participants are able to remove the device. For this reason, the magnitude of the signal is sufficient and relatively stable to capture the gait characteristics described in this manuscript. The periodic characteristics of walking naturally lend themselves to a frequency analysis approach for quantifying the features of walking. By utilizing the methods described in Urbanek et al. [bib_ref] Prediction of sustained harmonic walking in the free-living environment using raw accelerometry..., Urbanek [/bib_ref] , we can extract estimates of cadence (steps per second) and average magnitude from windows of raw data. In addition to these more common features, we also utilize the spectra obtained from the local fast Fourier Transform (FFT) as a functional predictor for modeling the association of walking with continuous responses such as age and body mass index (BMI). By incorporating the walking spectra as a predictor, we gain additional information about the characteristics of a person's gait which may be associated with the response of interest. For example, an individual with a very smooth walking stride would result in most of the energy from walking concentrated around the frequencies near the cadence. However, an individual with an interrupted stride (e.g., a limp) would result in energy being more dispersed through higher frequencies.
Several methods exist for fitting a scalar-on-regression function, such as y = I W(s)β(s)ds + , where W(·) is a functional predictor and y is a scalar response variable. Several methods for estimating β(·) are based on the eigenfunctions associated with some covariance operator defined by the predictors [bib_ref] Structured penalties for functional linear models-partially empirical eigenvectors for regression, Randolph [/bib_ref]. Due to the periodic nature of walking, we have strong reason to believe the vast majority of information contained in the walking spectra will be located around the harmonics centered at multiples of the dominant frequency. The PEER method developed by Randolph et al. [bib_ref] Structured penalties for functional linear models-partially empirical eigenvectors for regression, Randolph [/bib_ref] allows for the incorporation of presumed structure directly into the estimation process and is preferable to a purely empirical estimator. This particular method has been widely used in other areas of research, for example, in heritability and evolution studies [bib_ref] Inflorescence characteristics as function-valued traits: Analysis of heritability and selection on architectural..., Kulbaba [/bib_ref] [bib_ref] Variation and evolution of function-valued traits, Gomulkiewicz [/bib_ref] and in microbiome analysis [bib_ref] Kernel-penalized regression for analysis of microbiome data, Randolph [/bib_ref]. However, to the best of our knowledge, this is a novel use of the PEER method in the analysis of the raw accelerometry data.
In this manuscript, we propose a novel application of recently developed statistical methods for the analysis of accelerometry data by associating continuous responses, such as age and BMI, with the Fourier spectrum of walking. The purpose of this manuscript is to serve as a proof of concept for researchers seeking to utilize more information from the accelerometry data in modeling associations between characteristics of walking and health related outcomes. We show how this additional information can be combined with scalar predictors in a linear regression model framework. The remainder of this manuscript is structured as follows. In Section 2, we describe the data collection and pre-processing procedures. In Section 3, we describe the functional linear model used to fit the data and how the estimation is performed. In Section 4, we apply the proposed model to data collected in the laboratory from a study of an aging adult population. In Section 5, we conclude with a discussion.
## Data collection and pre-processing
Eighty-nine community-dwelling older adults were recruited from the Pittsburgh, Pennsylvania area for the National Institute on Aging, Aging Research Evaluating Accelerometry (AREA) project, part of the Developmental Epidemiologic Cohort Study (DECOS) [bib_ref] Performance on fast-and usual-paced 400-m walk tests in older adults: are they..., Lange-Maia [/bib_ref]. AREA was a cross-sectional methodological initiative designed to examine the impact of accelerometry wear location on assessment of physical activity and sedentary behavior among 89 older adults enrolled between March and May of 2010 [bib_ref] Validation of gait characteristics extracted from raw accelerometry during walking against measures..., Urbanek [/bib_ref]. The report included data from 51 healthy participants (25 men and 26 women) who had complete the "in-the-lab" (N = 46) or "in-the-wild" (N = 48) accelerometry data. Individuals were excluded from the DECOS study for the following reasons: hip fracture, stroke in the past 12 months, cerebral hemorrhage in the past 6 months, heart attack, angioplasty, heart surgery in the past 3 months, chest pain during walking the past 30 days, current treatment for shortness of breath or a lung condition, usual aching, stiffness, or pain in their lower limbs and joints, and bilateral difficulty in bending or straightening the knees fully [bib_ref] Physical activity and change in long distance corridor walk performance in the..., Lange-Maia [/bib_ref]. This paper focuses on the N = 46 participants that completed the "in-the-lab" fast-paced 400 m walk. Data were collected with an Actigraph GT3X+ worn at the right hip. The devices collected raw accelerometry data along three orthogonal axes at a sampling frequency of 80 Hz. A summary of the demographic data for all N = 46 participants is provided in [fig_ref] Table 1: DECOS participant characteristics [/fig_ref].
The first step in pre-processing the data is to split the observed triaxial signal from the 400 m walk into 10 s non-overlapping windows. For each window, we transform the raw triaxial signal into vector magnitude (V M), where V M is defined as the root sum of squares of the three axes, i.e.,
[formula] vm(t) = x 1 (t) 2 + x 2 (t) 2 + x 3 (t) 2 . The vector magnitude count (V MC) is then calculated as the mean absolute deviation of the V M: v τ (t) = 1 τ t+τ/2 ∑ u=t−τ/2 |vm(u) − vm|,(1) [/formula]
where τ is the window size expressed as number of sampling points [bib_ref] Prediction of sustained harmonic walking in the free-living environment using raw accelerometry..., Urbanek [/bib_ref]. We then transform the V M from the time domain into the frequency domain using the local FFT, or short time Fourier transform (STFT). Similarly to Urbanek et al. [bib_ref] Prediction of sustained harmonic walking in the free-living environment using raw accelerometry..., Urbanek [/bib_ref] , we define the STFT at time t of the vm(t) as
[formula] X(t, f ; τ) = [t+τ/2] ∑ u=[t−τ/2] vm(u)h(u)e −i2π f u/τ ,(2) [/formula]
where f is the frequency index and τ is a tuning parameter specifying the number of observations in the interval centered at t. The Hanning weights, defined as,
[formula] h(u; τ) = 0.5[1 − cos{2πu/(τ − 1)}] [/formula]
are used to avoid a blurring of the obtained spectra which can happen as a result of the windowing process. The spectrum is then defined as the absolute value of the STFT, |X(t, f ; τ)|. For each spectrum obtained, we then identify the fundamental frequency (cadence) as the location of the largest peak in the spectrum. Since the reported frequency of walking is between 1.4 and 2.5 Hz [bib_ref] Frequency and velocity of people walking, Pachi [/bib_ref] , which corresponds to 1.4 to 2.5 steps per second, we look for the cadence in a conservative range of 1.2-4.0 Hz to be consistent with Urbanek et al. [bib_ref] Prediction of sustained harmonic walking in the free-living environment using raw accelerometry..., Urbanek [/bib_ref]. The frequency axis used is from 0 to 39.9 Hz sampled every 0.1 Hz which ensures every individual's spectra will contain at least 10 multiples of their dominant frequency, or cadence. In the top left panel of [fig_ref] Figure 2: Pre-processing data [/fig_ref] , we display all spectra for a single participant. Although these spectra appear similar, it is evident that there is variability between spectra obtained in different windows. Therefore, once the spectra are all obtained, and we have identified their fundamental frequencies, it is imperative that we align each spectrum for aggregation.
In order to align all spectra at their fundamental frequency, we further transform each spectrum from the frequency domain into the order domain by scaling the frequency axis by the fundamental frequency for each spectrum. Linear interpolation is then used to place each spectrum back on the same sampling grid. This ensures that all spectra are aligned and sampled at equally spaced points. The top right panel of [fig_ref] Figure 2: Pre-processing data [/fig_ref] illustrates how the realigned spectra for a single participant. As can be seen, all spectra are perfectly aligned at the dominant frequency.
However, each spectrum is sampled discretely, therefore, further harmonics may be slightly misaligned in the order domain. To compensate, we average the spectra across all windows for each participant in order to obtain a global estimate of walking features for each individual. Each spectrum is restricted to 546 points between 0.3 and 5.75 in the order domain to avoid modeling signal noise at the beginning and end of the spectra. The bottom left panel of [fig_ref] Figure 2: Pre-processing data [/fig_ref] shows the averaged spectra for all N = 46 participants. The peaks of the average spectra for each individual are now nearly perfectly aligned in the order domain at multiples of the fundamental frequency.
Finally, we scale each individual's average spectrum by the magnitude of the spectrum at the cadence. By scaling the spectra in this way, the magnitude at each harmonic can be interpreted as a ratio to the magnitude at the cadence. This process is illustrated in bottom right panel of [fig_ref] Figure 2: Pre-processing data [/fig_ref] , and this entire process is fully detailed in Algorithm 1 . These steps for pre-processing the raw accelerometry data are essential in order to properly fit the statistical model described in Section 3.
## Algorithm 1:
Steps for pre-processing the raw accelerometry data.
Input : x(t)-accelerometry signal, f s -sampling frequency, s min = 1.2 Hz, s max = 4.0 Hz Output : FFT-scaled average FFT spectrum, V MC-average VMC, Cadence-average cadence
# Statistical methods
Frequently, the methods applied to analyze the data arising from the raw accelerometry signal rely on the discrete features extracted from such data, leading to possible loss of information. In contrast, we take the full time series signal into account and work with its continuous properties, namely, the spectrum obtained from the walking portion of the accelerometry signal. We summarize the approach taken below.
Let W i (·) denote a functional predictor (e.g., a scaled average FFT spectrum) from the ith study participant where [fig_ref] Figure 1: Triaxial accelerometer data from the 400 m walk for a single individual [/fig_ref]. We will assume that each observed predictor is obtained as a discretized version of an idealized function at p equally-spaced points, s 1 , . . . , s p , as can be seen in the walking spectra in [fig_ref] Figure 2: Pre-processing data [/fig_ref]. We let w i := [w i (s 1 ), . . . , w i (s p )] T be the p × 1 vector of values sampled from W i (·). Then our observed data take the form {y i ; x i ; w i } where y i is a scalar response, x i is a K × 1 vector of measurements from K scalar predictors (e.g., sex or average cadence), and w i is the functional predictor from the ith participant. We denote the true coefficient function by β(·), and then, the functional regression model of interest is given by
[formula] y i = x T i γ + I W i (s)β(s)ds + i(3) [/formula]
where i ∼ N(0, σ 2 ). Here x T i γ is the linear effect from K scalar predictors and I W i (s)β(s)ds is the functional effect.
## Estimation of parameters
Several approaches can be used to estimate the association between the scalar x i and functional w i (·) predictors with the outcome y i . In our work, we utilize the approach proposed in Randolph et al. [bib_ref] Structured penalties for functional linear models-partially empirical eigenvectors for regression, Randolph [/bib_ref] , which incorporates functional structure into the estimation of β(·). Specifically, the properties of the estimated spectra, i.e., their continuity, smoothness, and common behavior, are taken into the estimation procedure explicitly. To represent our model in a compact form, we combine the data from all N participants and express Equation (3) as
[formula] y = Xγ + Wβ +(4) [/formula]
where y = [y 1 , . . . , y N ] T is an N × 1 vector of responses; X = [x T 1 , . . . , x T N ] T is an N × K design matrix corresponding to the scalar predictors with coefficient vector γ; W = [w T 1 , . . . , w T N ] T is an N × p design matrix corresponding to the functional predictors with functional coefficient vector β.
Given the periodic nature of the walking behavior, if the walking spectral properties are associated with the outcomes, the relevant information contained in the walking spectra is localized around the harmonics at multiples of the dominant frequency. We thus want to estimate β while imposing this prior information on its functional structure. We achieve this by using the penalty operator, L [bib_ref] Structured penalties for functional linear models-partially empirical eigenvectors for regression, Randolph [/bib_ref] , which is created from the basis functions in the right panel of [fig_ref] Figure 3: Pre-processed walking spectra [/fig_ref]. The penalized estimates of γ and β are obtained as the solution to the following criterion
[formula] [γ,β λ,L ] T = arg min γ,β {||y − Xγ − Wβ|| 2 + λ||Lβ|| 2 L 2 },(5) [/formula]
where we only penalize the functional coefficient vector β.
Given some prior knowledge about the structure of our functional predictor, the penalty is defined utilizing a subspace containing this information [bib_ref] Structured penalties for functional linear models-partially empirical eigenvectors for regression, Randolph [/bib_ref]. We define this informative space, Q, to be a span of basis functions (right panel of [fig_ref] Figure 3: Pre-processed walking spectra [/fig_ref] emphasizing the relevant features of β(·) and consider the orthogonal projection P Q = QQ + . As described in Randolph et al. [bib_ref] Structured penalties for functional linear models-partially empirical eigenvectors for regression, Randolph [/bib_ref] , we define the decomposition-based penalty as
[formula] L ≡ L Q = a(I − P Q ) + P Q(6) [/formula]
for some a > 0. When a > 1 the estimate is penalized more in the non-informative space orthogonal to Q. When a = 1, the estimate is simply an ordinary ridge regression estimate. Therefore, a generalized ridge estimate of γ and β can be obtained as
[formula] [γ,β] T = (X T o X o + λL T o L o ) −1 X T o y,(7) [/formula]
where X o = [X W] and L o = blockdiag{0, L T Q L Q }. The tuning parameter, λ, is estimated in a principled way via a linear mixed model equivalence, as described in Ruppert et al. [bib_ref] Semiparametric Regression, Ruppert [/bib_ref]. Specifically, the optimization criterion (5) is written in an equivalent linear mixed model form with the coefficients γ being fixed and the coefficients β being random with a distribution β ∼ N(0, σ 2 β ). The estimate of the tuning parameter, λ, is then simply the ratio of the variances σ 2 and σ 2 β .
## Decos example
We applied the methods discussed in Section 3 to the data described in Section 2 to study the associations of walking spectra obtained from the fast-paced 400 m walk with age and BMI [bib_ref] Performance on fast-and usual-paced 400-m walk tests in older adults: are they..., Lange-Maia [/bib_ref]. The fast-paced 400 m walk is often used in epidemiological studies of older adults to assess aerobic fitness [bib_ref] Estimating cardiorespiratory fitness in well-functioning older adults: Treadmill validation of the long..., Simonsick [/bib_ref]. The most common protocol implemented for the fast-paced 400 m walk is the long distance corridor walk (LDCW) [bib_ref] Measuring fitness in healthy older adults: The Health ABC Long Distance Corridor..., Simonsick [/bib_ref]. The pre-processed walking spectra described in Section 2 were each sampled at k = 546 distinct sampling points within 0.3 and 5.75 of the order domain. This range was chosen because there is little energy contained in the spectra beyond 13.5 Hz. Assuming an average cadence of 2.0 Hz, this range sufficiently covers the relevant features of walking. There were N = 46 participants that completed the LDCW. In addition to each participant's scaled average walking spectra, an estimate of their average cadence was used as a predictor in the proposed models to control for participant-specific walking speeds. In addition, V MC was used to control for the energy magnitude each participant produced. For example, an individual with a very controlled and smooth walking style would have shown lower magnitude than an individual with a heavy stomp in their walk. We also adjusted each model for gender differences.
In order to use our prior knowledge about the structure of the walking spectra, we define a penalty L Q as given in Equation (6) (with a = 2). We define our basis functions as normal density functions centered at multiples of the cadence from 0.5 × cadence to 5.5 × cadence using steps of 0.5 × cadence. We chose a standard deviation such that the distributions were nearly orthogonal. Scaled average walking spectra and basis functions are displayed in [fig_ref] Figure 3: Pre-processed walking spectra [/fig_ref].
Following the general formulation of the functional regression model (3), we fit the following model to these data:
[formula] y i = γ 0 + Male i * γ 1 + Cadence i * γ 2 + V MC i * γ 3 + I Spectrum i (s)β(s)ds + i(8) [/formula]
where y i is either age or BMI, Male i is a binary variable, and Cadence i and V MC i are the cadence and vector magnitude count for participant i, respectively. Spectrum i (·) is the scaled average walking spectrum for participant i as described in Section 2. We assume that i ∼ N(0, σ 2 ). Regression coefficients γ and the regression function β(s) are estimated via the procedure described in Section 3.1 using the peer() function from the refund package in R. Thus, the scalar outcome y i is predicted via a weighted sum of the products of the collected data (indicator variable of male sex, cadence, VMC, and a spectrum) and their respective estimated regression coefficientsγ's andβ(·). [fig_ref] Figure 4: Estimates of the coefficient function,β, [/fig_ref] displays the estimates of β(·) along with the pointwise 95% confidence bands for the two models described in Equation [bib_ref] Walking energetics, fatigability, and fatigue in older adults: The study of energy..., Richardson [/bib_ref]. These figures show that the estimated regression function is different from zero at different multiples of the cadence. The regression function for age shows that the coefficient function,β, is negative at the multiples 1.5 and 3.5 and positive at the multiples 4, 4.5, and 5, whereas for the other multiples the estimated coefficients are not significantly different from zero. These results indicate that younger individuals have larger magnitude in the lower harmonics relative to the magnitude at their cadence which indicates a heavier stomp component and controlled walking motion. Older individuals have larger magnitudes in the higher harmonics relative to the magnitudes of their cadence, which indicates a less controlled compensatory walking motion (e.g., a limp). These results seem to be consistent with findings of prior research [bib_ref] The effect of age on variability in gait, Gabell [/bib_ref] [bib_ref] Basic gait parameters: Reference data for normal subjects, 10-79 years of age, Öberg [/bib_ref] [bib_ref] Age-related differences in walking stability, Menz [/bib_ref] [bib_ref] Accelerometry: A technique for quantifying movement patterns during walking, Kavanagh [/bib_ref] [bib_ref] Characteristic gait patterns in older adults with obesity-Results from the Baltimore Longitudinal..., Ko [/bib_ref] [bib_ref] Body mass index affects knee joint mechanics during gait differently with and..., Harding [/bib_ref] [bib_ref] Lower limb sagittal gait kinematics can be predicted based on walking speed,..., Moissenet [/bib_ref]. However additional research is needed to validate our conclusions. The regression results for BMI (bottom) show that the coefficient function,β, is positive at the multiple 2.5 and negative at the multiples 4 and 5, whereas for the other multiples the estimated coefficients are not significantly different from zero. These results could indicate that overweight or obese individuals tend to walk with a heavier stomp component, resulting in higher magnitudes of the lower harmonics than the normal weight individuals. Leaner individuals walk with a lighter stomp component, resulting in walking characteristics with lower magnitudes in the lower harmonics and higher magnitudes in the higher harmonics. Further discussion and possible limitations of this interpretation are provided in Section 5.
Both results described above show that the information provided by the penalty reduces the number of spurious findings and at the same time emphasizes the signal content of the scaled Fourier spectra.
# Discussion
In this paper we proposed a novel application of existing functional linear model methods to the study of physical activity data collected by accelerometers. We proposed an algorithm for pre-processing the raw data collected from accelerometers to quantify the characteristics of walking in a more detailed manner than is typically used with activity count summaries. By utilizing the periodic characteristics of walking, we were able to reduce the dimensionality of the raw data into a form that retained some details of the original signal while allowing us to use existing statistical methods for analyses. We applied these methods to the in-the-laboratory data collected from a study of an older adult population.
While FFT has been widely used for pre-processing accelerometry data, the features extracted from such methods have been applied to the problem of classification of activity types as opposed to associating characteristics of walking to continuous response variables [bib_ref] Prediction of sustained harmonic walking in the free-living environment using raw accelerometry..., Urbanek [/bib_ref] [bib_ref] Physical activity classification using the GENEA wrist-worn accelerometer, Zhang [/bib_ref] [bib_ref] A comparison of feature extraction methods for the classification of dynamic activities..., Preece [/bib_ref] [bib_ref] Activity recognition using a single accelerometer placed at the wrist or ankle, Mannini [/bib_ref]. To our knowledge, this is the first proposed application of functional linear regression techniques to model the association of walking spectra with continuous responses. Due to the periodic characteristics of walking, the proposed method naturally lends itself to this application, wherein we can inform the penalty operator of where the relevant information is contained in the spectra. This method is not limited to the cross-sectional setting, as demonstrated in this paper, and it is easily extended to responses collected longitudinally [bib_ref] Longitudinal functional models with structured penalties, Kundu [/bib_ref]. In addition to walking speed, this more detailed quantification of walking may provide additional information as to how certain degenerative diseases (e.g., Parkinson's disease and multiple sclerosis) affect a person's ability to walk over the progression of disease. Reuter et al. [bib_ref] Effects of a flexibility and relaxation programme, walking, and nordic walking on..., Reuter [/bib_ref] showed that certain walking programs can actually improve gait characteristics of individuals with Parkinson's disease over the course of a 6-month study. Gait characteristics were measured on a specialized treadmill outfitted with specialized sensors to accurately measure foot-ground contact. The application of these proposed methods could alleviate any financial restrictions of such studies to allow for much larger randomized prospective studies to determine whether these exercise therapies actually slow down the progression of such diseases. Utility of these methods can only be assessed with the inclusion of accelerometers in such studies, and they are being increasingly used.
We acknowledge that there are limitations in our analyses. For example, we did not collect data from the 3D motion capture or ground force reaction (GFR) measurements to validate the findings of our analyses. This manuscript is meant to demonstrate how researchers can utilize existing statistical methodology to analyze finer features of the raw accelerometry data retaining additional information about characteristics and changes in gait that are usually lost due to over-aggregation of the raw data collected. It was beyond the scope of this study to obtain information from specialized laboratory equipment, but we do feel that further research is warranted to validate our findings and draw associations between the additional characteristics of gait observed in our analyses with gait measurements obtained in a laboratory. Ko et al. [bib_ref] Characteristic gait patterns in older adults with obesity-Results from the Baltimore Longitudinal..., Ko [/bib_ref] found that older adults with obesity modify their gait patterns compared to normal weight counterparts while walking at normal and fast speeds. For example, they found that obese participants had lower mechanical work expenditure (MWE) in the ankle and significantly higher MWE in the knee and hip compared to normal weight participants. Paired with the raw accelerometry data, these findings could identify the specific mechanisms validating the additional associations we found in our analyses. Menz et al. [bib_ref] Age-related differences in walking stability, Menz [/bib_ref] observed that older participants exhibited a more conservative gait pattern characterized by reduced velocity, shorter step length, and increased step timing variability which could be contributing to the portion of the signals observed at the higher frequencies of the walking spectra in our study.
In conclusion, we acknowledge that age and BMI are easy to measure, but the strength of this study is that it serves as a proof of concept for how researchers can utilize the extracted walking characteristics in the presence of more relevant health related outcomes, e.g., fatigability or neurocognitive function. In addition, we studied only relatively healthy elderly individuals. Thus, generalizability of the findings to either healthy younger individuals or unhealthy older individuals needs to be studied. Given the small sample size of our study and utilization of the data from the laboratory experiment only, additional research is needed to establish similar associations between the health outcomes and the free-living walking data characteristics.
[fig] Figure 1: Triaxial accelerometer data from the 400 m walk for a single individual (top left) and a zoomed 10 s window (top right). Vector magnitude from the 400 m walk for same individual (bottom left) and zoomed 10 s window (bottom right). [/fig]
[fig] Figure 2: Pre-processing data. Observed FFT spectra for one participant as described in step 4 of Algorithm 1 (top left). Observed spectra realigned into order domain for the same participant as described in step 6 of Algorithm 1 (top right). Average realigned spectra for all participants as described in step 7 of Algorithm 1 (bottom left). Scaled average spectra for all participants as described in step 9 of Algorithm 1 (bottom right). [/fig]
[fig] Figure 3: Pre-processed walking spectra (top) and basis functions used for modeling (bottom). The x-axis represents multiples of the frequency of the cadence. [/fig]
[fig] Figure 4: Estimates of the coefficient function,β, (with 95% point-wise confidence band) for the association of walking with age and BMI, as described in Section 4. The x-axis represents multiples of the frequency of the cadence. [/fig]
[fig] Author: Contributions: W.F.F. conceptualization, formal analysis, investigation, writing, and visualization; J.K.U. writing and resources; N.W.G. writing and resources; J.H. conceptualization, writing, resources, and supervision. All authors have read and agreed to the published version of the manuscript. [/fig]
[fig] Funding: This research was supported by Pittsburgh Claude D. Pepper Older Americans Independence Center Research Registry and Developmental Pilot Grant (PI: Glynn)-NIH P30 AG024827 and a National Institute on Aging Professional Services Contract HHSN271201100605P to Dr. Glynn. The project was also supported, in part, by the Intramural Research Program of the National Institute on Aging. Doctors Harezlak and Fadel were supported in part by the NIMH grant R01 MH108467 and with support from the Indiana Clinical and Translational Sciences Institute Design and Biostatistics Pilot Grant, funded, in part, by grant UL1TR001108, from the National Institutes of Health, National Center for Advancing Translational Sciences, Clinical and Translational Sciences Award. [/fig]
[table] Table 1: DECOS participant characteristics (N = 46). [/table]
[bib_ref] Performance on fast-and usual-paced 400-m walk tests in older adults: are they..., Lange-Maia [/bib_ref] [bib_ref] Accelerometer-derived sedentary and physical activity time in overweight/obese adults with type 2..., Healy [/bib_ref] [bib_ref] Accelerometer assessment of physical activity in active, healthy older adults, Copeland [/bib_ref] [bib_ref] Use of accelerometry to measure physical activity in older adults at risk..., Pruitt [/bib_ref] [bib_ref] Assessment of free-living daily physical activity in older claudicants: Validation against the..., Gardner [/bib_ref] [bib_ref] Walking energetics, fatigability, and fatigue in older adults: The study of energy..., Richardson [/bib_ref] [bib_ref] Physical activity as a novel outcome of total knee replacement: Comparing self-report..., Losina [/bib_ref]
|
10.7759/cureus.19380
|
CCBY
|
8654642
|
34925983
|
s2orc_pubmed_articles
|
Effectiveness of the Internet-Based Versus Face-to-Face Interaction on Reduction of Tobacco Use Among Adults: A Meta-Analysis
Literature reported the effectiveness of internet-based interventions over face-to-face interaction on tobacco quitting; however, limited sample size reinforces to integrate and analyze these studies' findings to reach a single conclusion. Therefore, we evaluated the effectiveness of the internet as an intervention approach versus face-to-face interaction on reducing tobacco use among adults. A systematic search was performed through various electronic databases such as Medline, PsychInfo, PubMed, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), ResearchGate, Google Scholar, and Academia. Reference lists of the eligible articles were also screened. Full-text articles were included as per eligibility criteria (PICO framework). No ethnicity restriction was applied. A total of 13 studies were selected for meta-analysis, with 3852 and 3908 participants in intervention and control groups, respectively. Forest plot favours the intervention group at one month follow up for tobacco quitting (OR: 2.37, CI: 1.86-3.02, P=0.00001, I2=0%), at three months (OR: 1.88, CI: 1.48-2.40, P=0.00001, I2=42%) at six months (OR: 2.02, CI: 1.64-2.50, P=0.00001, I2=38%) and at one year of follow-up (OR: 1.43, CI: 1.18-1.74, P=0.00001, I2=36%) comparing to control group. Conclusively, internet and web-based interventions are highly useful in tobacco quitting at one month, three months, six months, and one year of follow-up compared to face-to-face interaction or no intervention, although the level of evidence was moderate. Additionally, limited trials in developing countries, arising need for research on internet use for tobacco control in developing countries.
# Introduction and background
Tobacco use is the leading cause of avertible and premature deaths worldwide. The burden of tobaccorelated disease is increasing in developed and developing countries as well. Interestingly, the deaths are declining in developed countries, and the burden is shifting to developing countries. However, tobacco consumption pattern varies across gender; male vs. female, domicile; rural vs. urban, regions, cultural practices, and family income [bib_ref] Smoking prevalence and cigarette consumption in 187 countries, Ng [/bib_ref]. Men are more frequently (23%) indulging in tobacco use than their counterparts (3%) [bib_ref] Tobacco usage in Uttarakhand: a dangerous combination of high prevalence, widespread ignorance,..., Grills [/bib_ref]. Quitting any form of smoking is challenging and involves physiological, psychological, and many other factors, including social and environmental milieu to become successful [bib_ref] Cognitivebehavioral treatment with behavioral activation for smoking cessation: Randomized controlled trial, Martínez-Vispo [/bib_ref]. In the case of smoking cessation, the best use of positive and negative reinforcements helps alleviate the withdrawal symptoms, and the role of behavioral approaches in smoking cessation cannot be denied [bib_ref] Declining alternative reinforcers link depression to young adult smoking, Audrain-Mcgovern [/bib_ref].
Over the years, many innovative forms of internet-based approaches have been tried to quit tobacco use globally. The use of health communication and internet-based interventions like tailored computerized programs, text messages, mobile or telephone, and WhatsApp for reminder or call, app-based intervention, chat-based instant messaging, video assistance using the website and mobile [bib_ref] Cognitivebehavioral treatment with behavioral activation for smoking cessation: Randomized controlled trial, Martínez-Vispo [/bib_ref] and use of social media, has been vividly used in recent decades to quit smoking among different age groups [bib_ref] Effectiveness of a computer-tailored smoking cessation program: a randomized trial, Etter [/bib_ref]. Although there is ample research and data regarding the potential influence of the media [bib_ref] Effectiveness of a computer-tailored smoking cessation program: a randomized trial, Etter [/bib_ref] , face to face health education, cognitive behavior therapy, motivational influences, and nurses-assisted counseling [bib_ref] Tobacco usage in Uttarakhand: a dangerous combination of high prevalence, widespread ignorance,..., Grills [/bib_ref] , on behavioral changes among smokers, there are scanty reports on the internet use or behavioral interventions. They are neither planned nor conducted rigorously to indicate firm evidence of any encouraging effects on health outcomes.
Interestingly, the internet and other electronic platforms are abundantly present in this era and have almost become part and parcel of the health care system. A medical expert with just a computer device and internet access, and some necessary handling skills can reach many people and communicate inexpensively. Though the effectiveness of internet-based and face-to-face interventions on quitting smoking are very well reported in the literature, every study carries one or another limitation in methodology and limited sample size. Therefore, it is required to integrate and analyze these studies' findings to reach a single conclusion. This study was planned to assess the effectiveness of the internet versus face-to-face interactions on reducing tobacco use among adults.
# Review methods
## Data sources and search strategy
The electronic databases, such as Medline, PsychInfo, PubMed, Embase, Cochrane Central Register of Controlled Trials (CENTRAL), Google Scholar, ResearchGate, and Academia, were explored. Reference lists of the eligible articles were also screened. All relevant studies available on the topic were included irrespective of time duration. The systematic search was restricted to studies published in the English language. The keywords were "smoker or smokers OR smoking," "tobacco" OR cigarette OR nicotine OR smoking cessation OR "tobacco consumption OR cessation, OR abstain* OR quit* OR stop* OR computer OR computer-aided design, OR internet, OR computer, OR networks, OR media, OR cellular phone OR mobile, OR text OR message* OR SMS, OR web, OR electronic mail OR Chat, OR video recording.
## Eligibility criteria (pico framework) for participants
Inclusion criteria were adults aged more than 18 years who use the internet or face-to-face interventions to reduce or quit tobacco use. No ethnicity restrictions were applied. Exclusion criteria were Cochrane studies that compare the internet to face-to-face interventions with other interventions.
Intervention: Internet interventions such as Phone, mobile, WhatsApp, Facebook, Online network group, Online Support group, text messaging, other internet media.
Comparator: Face-to-face interventions or no intervention in the comparator group. Face-to-face interventions include counseling, cognitive behavior therapy, or health education forms with control or routine care.
Outcome: Post-intervention tobacco quitting -number of participants quitting tobacco after the intervention (internet use).
Study design: This study is based on randomized controlled trials.
Time frame: No restriction to the time frame was applied
## Screening of eligible studies
A systematic search was done by two reviewers independently. After searching, studies were screened with titles and abstracts of respective studies. All selected studies were imported to Rayyan, a free web-based software. Two reviewers (PY and RK) screened the full text of articles based on eligibility criteria determined as per review protocol. Any relevant discrepancy has been resolved by consensus with the help of a third reviewer (MB). We adhered to the guidelines of Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) 2009 [bib_ref] Preferred reporting items for systematic reviews and metaanalyses: the PRISMA statement, Moher [/bib_ref]. PRISMA flow chart displays all the steps followed in the inclusion and exclusion of studies
## Figure 1: flow chart (prisma)
## Prisma: preferred reporting items for systematic reviews and meta-analyses
Eligible studies were exported to RevMan software 5.4 (The Cochrane Collaboration, London, UK)for data analysis. Forest plots have been created to present the results with Odds ratio (OR), confidence interval (CI), and effect size.
The GRADE approach was also followed to explore the quality of evidence on high, moderate, and low levels. RevMan files were exported to the GRADE Profiler to assess the quality of studies and create a "Summary of Findings"
## Data extraction
Two reviewers (PY and RK) extracted the data from the full text of eligible studies. Corresponding authors of included studies were contacted for the relevant data. Data excel sheet was prepared to note the characteristics of selected studies. It includes the author's name with publication year, country, sample size, the mean age of participants, male to female ratio, baseline tobacco consumption, and follow-up period after the intervention (
## Figure 2: risk of bias graph and summary
The reviewers independently assessed the quality of included studies [bib_ref] Happy ending: a randomized controlled trial of a digital multi-media smoking cessation..., Brendryen [/bib_ref] [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref] [bib_ref] Internet-based photoaging within Australian pharmacies to promote smoking cessation: randomized controlled trial, Burford [/bib_ref] [bib_ref] Comparative effectiveness of an internet-based smoking cessation intervention versus clinic-based specialty care..., Calhoun [/bib_ref] [bib_ref] Preventing smoking relapse via Web-based computer-tailored feedback: a randomized controlled trial, Elfeddali [/bib_ref] [bib_ref] Smoking cessation via the internet: a randomized clinical trial of an internet..., Japuntich [/bib_ref] [bib_ref] The RealU online cessation intervention for college smokers: a randomized, An [/bib_ref] [bib_ref] An internet-based smoking cessation program for Korean Americans: results from a randomized..., Mcdonnell [/bib_ref] [bib_ref] Efficacy and use of an internet-delivered computertailored lifestyle intervention, targeting saturated fat..., Oenema [/bib_ref] [bib_ref] A cluster randomized trial in general practice with referral to a group-based..., Pisinger [/bib_ref] [bib_ref] Effectiveness of a web-based multiple tailored smoking cessation program: a randomized controlled..., Smit [/bib_ref] [bib_ref] A randomised control study of a fully automated internet based smoking cessation..., Swartz [/bib_ref]
# Data analysis
Review Manager software 5.4 version was used for meta-analysis. The fixed-effects model and effect measures were calculated as the OR with P-value < 0.05 considered statistically significant. I2 statistics with 25%, 50%, and 75% were measured to compute statistical heterogeneity in low, moderate, and high gradestabulated data presented in a forest plot .
## Figure 3: forest plot comparing internet intervention with the control group
Tobacco quit at one month follow up [bib_ref] Happy ending: a randomized controlled trial of a digital multi-media smoking cessation..., Brendryen [/bib_ref] [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref]
## Figure 5: forest plot comparing internet intervention with the control group
Tobacco quit at six months follow up [bib_ref] Happy ending: a randomized controlled trial of a digital multi-media smoking cessation..., Brendryen [/bib_ref] [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref] [bib_ref] Internet-based photoaging within Australian pharmacies to promote smoking cessation: randomized controlled trial, Burford [/bib_ref] [bib_ref] Smoking cessation via the internet: a randomized clinical trial of an internet..., Japuntich [/bib_ref] [bib_ref] The RealU online cessation intervention for college smokers: a randomized, An [/bib_ref] [bib_ref] Effectiveness of a web-based multiple tailored smoking cessation program: a randomized controlled..., Smit [/bib_ref]
## Figure 6: forest plot comparing internet intervention with the control group
Tobacco quit at one year follow up [bib_ref] Happy ending: a randomized controlled trial of a digital multi-media smoking cessation..., Brendryen [/bib_ref] [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref] [bib_ref] Preventing smoking relapse via Web-based computer-tailored feedback: a randomized controlled trial, Elfeddali [/bib_ref] [bib_ref] An internet-based smoking cessation program for Korean Americans: results from a randomized..., Mcdonnell [/bib_ref] [bib_ref] A cluster randomized trial in general practice with referral to a group-based..., Pisinger [/bib_ref] The funnel plots have also been created to assess the publication bias across studies. It measures an effect estimate against its standard error for an outcome [fig_ref] FIGURE 7: Funnel plot [/fig_ref]. Tobacco quitting among participants has been analyzed at one, three, six, and twelve months of follow-up and presented in a forest plot.
# Results
A total of 13 articles were found suitable for meta-analysis, with 3852 and 3908 participants in intervention and control groups [bib_ref] Happy ending: a randomized controlled trial of a digital multi-media smoking cessation..., Brendryen [/bib_ref] [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref] [bib_ref] Internet-based photoaging within Australian pharmacies to promote smoking cessation: randomized controlled trial, Burford [/bib_ref] [bib_ref] Comparative effectiveness of an internet-based smoking cessation intervention versus clinic-based specialty care..., Calhoun [/bib_ref] [bib_ref] Preventing smoking relapse via Web-based computer-tailored feedback: a randomized controlled trial, Elfeddali [/bib_ref] [bib_ref] Smoking cessation via the internet: a randomized clinical trial of an internet..., Japuntich [/bib_ref] [bib_ref] The RealU online cessation intervention for college smokers: a randomized, An [/bib_ref] [bib_ref] An internet-based smoking cessation program for Korean Americans: results from a randomized..., Mcdonnell [/bib_ref] [bib_ref] Efficacy and use of an internet-delivered computertailored lifestyle intervention, targeting saturated fat..., Oenema [/bib_ref] [bib_ref] A cluster randomized trial in general practice with referral to a group-based..., Pisinger [/bib_ref] [bib_ref] Effectiveness of a web-based multiple tailored smoking cessation program: a randomized controlled..., Smit [/bib_ref] [bib_ref] A randomised control study of a fully automated internet based smoking cessation..., Swartz [/bib_ref]. All studies revealed data with a sample size ranging from 160 [bib_ref] Internet-based photoaging within Australian pharmacies to promote smoking cessation: randomized controlled trial, Burford [/bib_ref] to 2159 [bib_ref] Efficacy and use of an internet-delivered computertailored lifestyle intervention, targeting saturated fat..., Oenema [/bib_ref]. Baseline characteristics of included studies have been described in [fig_ref] TABLE 1: Summary of findings table [/fig_ref] [bib_ref] Smoking cessation via the internet: a randomized clinical trial of an internet..., Japuntich [/bib_ref] [bib_ref] A cluster randomized trial in general practice with referral to a group-based..., Pisinger [/bib_ref]. Calhoun et al. had the majority of male participants in the intervention (85%) and control group (84%) [bib_ref] Comparative effectiveness of an internet-based smoking cessation intervention versus clinic-based specialty care..., Calhoun [/bib_ref].
Two studies measured the outcome at four steps: one, three, six months, and one year [bib_ref] Happy ending: a randomized controlled trial of a digital multi-media smoking cessation..., Brendryen [/bib_ref] [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref]. Two studies followed up the participants only for one month [bib_ref] Efficacy and use of an internet-delivered computertailored lifestyle intervention, targeting saturated fat..., Oenema [/bib_ref] [bib_ref] A randomised control study of a fully automated internet based smoking cessation..., Swartz [/bib_ref]. Two studies measured the outcome at six months only [bib_ref] Internet-based photoaging within Australian pharmacies to promote smoking cessation: randomized controlled trial, Burford [/bib_ref] [bib_ref] The RealU online cessation intervention for college smokers: a randomized, An [/bib_ref]. Calhoun et al. measured the outcome at three months and twelve months of internet intervention and telehealth medication clinic unite with a telehealth medication clinic for nicotine replacement therapy [bib_ref] Comparative effectiveness of an internet-based smoking cessation intervention versus clinic-based specialty care..., Calhoun [/bib_ref]. Even four studies assessed the outcome of tobacco use at one year of different web or internet-based interventions [bib_ref] Preventing smoking relapse via Web-based computer-tailored feedback: a randomized controlled trial, Elfeddali [/bib_ref] [bib_ref] An internet-based smoking cessation program for Korean Americans: results from a randomized..., Mcdonnell [/bib_ref] [bib_ref] A cluster randomized trial in general practice with referral to a group-based..., Pisinger [/bib_ref].
Subgroup analysis with tobacco quitting outcomes at one, three, six months, and one-year follow-up further lowers the heterogeneity across studies. Sensitivity analysis was done to find a better result with a random effect model. We observed similar results with the random effect model also. Pike et al. have been removed from the analysis due to the massive difference in the number of participants in both groups, creating heterogeneity [bib_ref] American Cancer Society's QuitLink: randomized trial of Internet assistance, Pike [/bib_ref].
The forest plot favors the intervention group (OR: 2.37, CI: 1.86-3.02, P=0.00001, I2=0%) in comparison to the control group for quitting tobacco at a one-month follow-up . The forest plot also favors the intervention group compared to the control group (OR: 1.88, CI: 1.48-2.40, P=0.00001, I2=42%) for quitting tobacco at three months follow up [fig_ref] FIGURE 4: Forest plot comparing internet intervention with the control group Tobacco quit at... [/fig_ref]. The forest plot also favors the intervention group compared to the control group (OR: 2.02, CI: 1.64-2.50, P=0.00001, I2 =38%) for quitting tobacco at six months follow up . The forest plot also favors the intervention group compared to the control group (OR: 1.43, CI: 1.18-1.74, P-0.00001, I2 = 36%) for quitting tobacco at a one-year follow-up . The forest plot suggests significantly higher tobacco quitting events in the internet intervention group at one, three, six, and twelve months of follow-up of participants with moderate heterogeneity across the studies.
Risk of bias has been assessed and created a risk of bias graph and summary of included studies under the heads of selection bias, performance bias, detection bias, attrition bias, reporting bias, and any other bias observed across the studies. It depicts that there was no serious risk of bias across the studies [fig_ref] 2: Baseline characteristics of included studiesRisk of Bias AssessmentTwo reviewers [/fig_ref]. A funnel plot has been created to estimate the effect against its standard error for included studies in each outcome [fig_ref] FIGURE 7: Funnel plot [/fig_ref].
# Discussion
Over the year, many innovative forms of internet-based approaches, i.e., tailored computerized programs, text messages, mobile or telephone, and WhatsApp for reminder or call, app-based intervention, chat-based instant messaging, video assistance using the website and mobile and use of social media [bib_ref] Preferred reporting items for systematic reviews and metaanalyses: the PRISMA statement, Moher [/bib_ref] , have been practiced commonly to quit tobacco in different age group population. Although, various methodological issues reduce the ability to estimate the effects of internet-based approaches.
This study evaluated the impact of the internet approaches versus face-to-face interaction on reducing tobacco use in the adult population. Results suggest significantly higher tobacco quitting events in the internet intervention group than the control group at one month, three months, six months, and one year of follow-up of participants with moderate heterogeneity across the studies. Happy ending, a digital multimedia smoking cessation intervention consisting of more than 400 contacts through emails, interactive voice response, Web pages, and short message service compared with self-help booklet, reported higher point abstinence rates in the treatment group in the long-term effect of the intervention [bib_ref] Happy ending: a randomized controlled trial of a digital multi-media smoking cessation..., Brendryen [/bib_ref] [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref].
A written list of internet resources for smoking cessation was found more helpful than written self-help material to quit smoking for a long-term period of one year. Internet-based self-help smoking cessation program, interactive, individual advice, multiple computer-tailored smoking cessation internet interventions, and a video-based internet site presented strategies for motivational materials and smoking cessation found no effect at six months of intervention but the significant effect at 12 months of follow up [bib_ref] An internet-based smoking cessation program for Korean Americans: results from a randomized..., Mcdonnell [/bib_ref] [bib_ref] A cluster randomized trial in general practice with referral to a group-based..., Pisinger [/bib_ref] [bib_ref] Effectiveness of a web-based multiple tailored smoking cessation program: a randomized controlled..., Smit [/bib_ref] [bib_ref] A randomised control study of a fully automated internet based smoking cessation..., Swartz [/bib_ref]. Personalized smoking cessation through an online life magazine in the young population enhanced smoking cessation at the end of 12 months [bib_ref] The RealU online cessation intervention for college smokers: a randomized, An [/bib_ref].
Internet use and telehealth medication clinic combined with a telehealth medication clinic for nicotine replacement therapy reported no significant difference (17% vs. 12%) in comparison to clinical-based smoking cessation after three months of intervention [bib_ref] Comparative effectiveness of an internet-based smoking cessation intervention versus clinic-based specialty care..., Calhoun [/bib_ref]. However, Burford et al. compared a computergenerated photoaging intervention with no treatment group and reported a higher (27.5%) incidence of smoking quit than the control group (6.3%) at six months follow up [bib_ref] Internet-based photoaging within Australian pharmacies to promote smoking cessation: randomized controlled trial, Burford [/bib_ref]. [bib_ref] Comparing internet assistance for smoking cessation: 13-month follow-up of a six-arm randomized..., Rabius [/bib_ref] [bib_ref] Evaluation of an Internet-based smoking cessation program: lessons learned from a pilot..., Feil [/bib_ref]. Findings were also reinforced by the researchers that the participants' loss inevitably influences research on the internet for health purposes [bib_ref] Issues in evaluating health websites in an Internet-based randomized controlled trial, Eysenbach [/bib_ref]. After the quit attempts, web-based interventions could be more effective in preventing relapse in the long term, which requires adherence to the intervention for its effectiveness [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref].
Additionally, the approach to a website supporting smoking abstinence is not related to smoking cessation [bib_ref] A digital smoking cessation program delivered through internet and cell phone without..., Brendryen [/bib_ref]. Civljak et al. reported the strong effect of uniting tailored materials with nicotine replacement therapy on tobacco cessation and a significant positive impact of tailored materials among precontemplators.
The internet services should be based on their preference and easily accessible to those who want to quit smoking and seek related information through the internet, need to utilize the internet services for the same. Presently, internet interventions' incremental cost is less than other modalities, facilitating and evaluating online programs' effectiveness [bib_ref] Development and implementation cost analysis of telephone-and Internet-based interventions for the maintenance..., Meenan [/bib_ref]. Online interventions also can access smokers and support them in quitting tobacco, which is also firmly associated with the total and physical quality of life among adults [bib_ref] Quality of life after quitting smoking and initiating aerobic exercise, Bloom [/bib_ref].
# Strength and limitations
Subgroup analysis explored and discussed the possibility of tobacco quitting in the adult population at different time points. Sensitivity analysis strengthened the evidence by exploring possible alternate findings.
Although there was a lack of uniformity of internet-based approaches in included trials, they had different internet approaches, which have also been discussed. The risk of included bias in the individual trial also contributed towards the limitation of meta-analysis [fig_ref] 2: Baseline characteristics of included studiesRisk of Bias AssessmentTwo reviewers [/fig_ref].
This article is available on preprint server Research Square (https://www.researchsquare.com/article/rs-318627/v1).
# Conclusions
This meta-analysis pooled the data of randomized controlled trials with a limited sample size. It winded up that internet use is highly effective in tobacco quitting at one, three, six, and twelve months of follow-up of participants compared to face-to-face intervention or no intervention with moderate heterogeneity across the studies. A moderate level of evidence supports the findings. Further studies are required to explore internet interventions' durable adherence among the adult population who spared their maximum time with the internet in any form. Additionally, limited availability of trials in developing countries requires research of internet use in developing countries to quit tobacco. Findings provide evidence to policymakers to utilize the internet as an effective instrument for tobacco control in their countries. Conclusively, this metaanalysis adds to the evidence for the promising approach of the internet-based intervention in modifying behavior, reducing tobacco use, and enhancing positive health practices among adults.
# Additional information disclosures
## Conflicts of interest:
In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work.
[fig] 2: Baseline characteristics of included studiesRisk of Bias AssessmentTwo reviewers (PY and MB) independently assessed the quality of included studies. Risk of bias graph and summary has been created in Review Manager software 5.4 version under the heads of random sequence generation (selection bias), blinding of participants and personnel (performance bias), allocation concealment (selection bias), blinding of outcome assessment (detection bias), selective reporting (reporting bias), incomplete outcome data (attrition bias), and other bias[9] (Figure 2). [/fig]
[fig] FIGURE 4: Forest plot comparing internet intervention with the control group Tobacco quit at three months follow up[11,12,15,17,23] [/fig]
[fig] FIGURE 7: Funnel plot: shows publication bias across studies for each outcome (a) Tobacco quitting at one month, (b) tobacco quitting at three months, (c) tobacco quitting at six months, and (d) tobacco quitting at one year [/fig]
[table] TABLE 1: Summary of findings table [/table]
|
10.1364/OE.465125
|
CCBY
| null |
36558675
|
s2orc_pubmed_articles
|
Unidirectional mode selection in bistable quantum cascade ring lasers.
Ideal ring resonators are characterized by travelling-wave counterpropagating modes, but in practice travelling waves can only be realized under unidirectional operation, which has proved elusive. Here, we have designed and fabricated a monolithic quantum cascade ring laser coupled to an active waveguide that allows for robust, deterministic and controllable unidirectional operation. Spontaneous emission injection through the active waveguide enables dynamical switching between the clockwise and counterclockwise states of the ring laser with as little as 1.6% modulation of the electrical input. We show that this behavior stems from a perturbation in the bistable dynamics of the ring laser. In addition to switching and bistability, our novel coupler design for quantum cascade ring lasers offers an efficient mechanism for outcoupling and light detection. arXiv:2205.07156v1 [physics.optics] 15 May 2022
# Introduction
Ring lasers have been and continue to be investigated for their rich dynamics in a variety of geometries and material platforms. Numerous phenomena based on ring lasers have been numerically studied and experimentally observed, such as unidirectional and bidirectional operation, alternate oscillations, bistability and chaos [bib_ref] Unidirectional bistability in semiconductor waveguide ring lasers, Sorel [/bib_ref] [bib_ref] Operating regimes of GaAs-AlGaAs semiconductor ring lasers: experiment and model, Sorel [/bib_ref] [bib_ref] Bistability and all-optical switching in semiconductor ring lasers, Pérez [/bib_ref] [bib_ref] Twodimensional phase-space analysis and bifurcation study of the dynamical behaviour of a..., Der Sande [/bib_ref] [bib_ref] Random optical pulse generation with bistable semiconductor ring lasers, Sunada [/bib_ref] [bib_ref] Fast random bits generation based on a single chaotic semiconductor ring laser, Nguimdo [/bib_ref]. These phenomena result from the nonlinear interaction of the clockwise (CW) and counterclockwise (CCW) fields. These two fields, which in an ideal resonator do not couple to each other, give rise to the travelling-wave nature of the total field in a ring laser. This is in contrast to Fabry-Perot (FP) lasers, where the electric field exhibits a standing-wave pattern leading to spatial hole burning (SHB) of the gain medium. In lasers where the gain recovery time is fast, such as quantum cascade lasers (QCLs), SHB effects are dominant and lead to multimode instabilities [bib_ref] Multimode regimes in quantum cascade lasers: From coherent instabilities to spatial hole..., Gordon [/bib_ref]. Although ring resonators seem to evade SHB, pure travelling waves are only possible in an ideal cavity; in a realistic cavity, even small imperfections will perturb the rotational invariance of the counter-propagating modes and pin them to a standing-wave pattern similar to FP lasers. Thus, the only realistic way to obtain travelling waves in a ring laser is by ensuring unidirectional operation where only one of the two modes, CW or CCW, is allowed to lase at a time. Unidirectional operation of ring QCLs has been shown to be necessary for the demonstration of active mode locking and mid-infrared, ring-based frequency combs [bib_ref] External ring-cavity quantum cascade lasers, Malara [/bib_ref] [bib_ref] Mid-infrared frequency comb from a ring quantum cascade laser, Meng [/bib_ref]. Previous work on QCLs aimed directly at mode selection between CW and CCW modes has relied on an "S-shaped" waveguide design to favor one mode [bib_ref] A unidirectional quantum cascade ring laser, Nshii [/bib_ref] or external cavities where the gain medium is placed within a free-space-coupled ring cavity [bib_ref] External ring-cavity quantum cascade lasers, Malara [/bib_ref] [bib_ref] Active mode locking of quantum cascade lasers in an external ring cavity, Revin [/bib_ref].
In this work, we show robust and deterministic unidirectional operation from a monolithic racetrack QCL. Instead of a specialized waveguide design of the ring laser itself, which permanently favors one mode, we rely on an active waveguide evanescently coupled to the racetrack to perform mode selection and switching, which allows for controllable operation in both directions. This in turn, may enable denser integration and greater functionality of mid-infrared photonic integrated circuits.
Switching between the two modes of a semiconductor ring laser has also been used to demonstrate random number generation [bib_ref] Random optical pulse generation with bistable semiconductor ring lasers, Sunada [/bib_ref]. The fast carrier lifetime of QCLs (∼ ps) may be used in such applications to improve the speed of this operation. In addition, we show that our ring laser exhibits a region of bistability, a dynamical regime that has not yet been explored in mid-infrared QCLs. The bistable nature of the ring allows for full switching between the two modes with small input perturbations.
Our QCL operates at λ ≈ 8 µm, a wavelength where the fabrication of an evanescent coupler becomes challenging because the active region is buried deep in the QC layers thus requiring a complex etching process. Unlike previous work, which focuses on short wavelengths (λ < 5 µm), [bib_ref] A unidirectional quantum cascade ring laser, Nshii [/bib_ref] [bib_ref] Homogeneous photonic integration of mid-infrared quantum cascade lasers with low-loss passive waveguides..., Jung [/bib_ref] we fabricate an evanescent coupler that does not rely on any specialized growth or processing techniques.
Lastly, ring QCLs have recently been studied as emerging platforms for frequency combs [bib_ref] Mid-infrared frequency comb from a ring quantum cascade laser, Meng [/bib_ref] [bib_ref] Frequency combs induced by phase turbulence, Piccardo [/bib_ref] [bib_ref] Dissipative kerr solitons in semiconductor ring lasers, Meng [/bib_ref]. In these studies, light extraction is done by sensing the weak scattering off the waveguides due to bending losses. While this technique aids in minimizing backscattering between the two counter-propagating modes, it may pose a limitation on the practical use of these devices. In addition to controllable mode selection and switching, the active waveguide in our system also serves as an efficient outcoupling mechanism for the racetrack laser.
## Design and fabrication
We have fabricated a racetrack resonator coupled to a bus waveguide on a conventional λ ≈ 8.8 µm QCL wafer based on a double-phonon resonance design [bib_ref] Room-temperature continuouswave quantum cascade lasers grown by MOCVD without lateral regrowth, Liu [/bib_ref]. The active bus waveguide differs from the more commonly utilized passive bus waveguide in two key ways. First, it has intersubband transitions at the same wavelength that the racetrack emits; thus, to overcome absorption, the waveguide must be pumped to transparency. Second, due to the same transitions and depending on the pumping level, it has its own spontaneous emission which is coupled into the racetrack and is responsible for the switching properties we will discuss here. The schematic of our system is shown in [fig_ref] Figure 1: Device schematic showing the racetrack laser ridge and the evanescently-coupled waveguide with... [/fig_ref]. The two arms of the waveguide have separate metal contacts and are thus controlled independently. The coupling mechanism in this system is evanescent coupling where a small gap is deep etched between the ridges. To determine the size of the gap as well as the length of the coupler we use coupled mode theory (CMT) simulations based on the computed fundamental mode. Our CMT simulations show that significant coupling can only occur for relatively thin ridges. In our case we choose a 5 µm ridge width because the mode is less confined laterally thereby expanding outside the waveguide core and enhancing evanescent coupling. For ease of fabrication in the following processing steps, we adiabatically taper our waveguide width to 10 µm outside of the coupling region. This also helps to limit loss due to weak confinement and proximity of the mode to the lossy metal on the sidewalls. The detailed design of the coupler is shown in [fig_ref] Figure 1: Device schematic showing the racetrack laser ridge and the evanescently-coupled waveguide with... [/fig_ref]. We have considered several combinations of gap and coupling length and their corresponding coupling strength is given in the table in [fig_ref] Figure 1: Device schematic showing the racetrack laser ridge and the evanescently-coupled waveguide with... [/fig_ref]. The chosen physical dimensions are compatible with conventional lithography techniques. The most challenging step in the fabrication is deep etching of the gap with an aspect ratio ≥ 5 while still maintaining vertical sidewalls and minimal roughness. We utilize a multistep inductively coupled plasma -reactive ion etching (ICP-RIE) process based on Cl 2 and BCl 3 . The ICP-RIE process consists of many parameters, which we optimize for minimal sidewall roughness. [fig_ref] Figure 2: Scanning electron microscope [/fig_ref] shows scanning electron microscope (SEM) images of various elements of the device to especially showcase the etch characteristics. To reduce bending losses, all bend radii are designed to be larger than 450 µm. To avoid feedback from reflection, the waveguide arms are angled 17 - with respect to the cleaved facet which lowers their reflectivity to ≈ 0.01 [bib_ref] Enhanced light output power of quantum cascade lasers from a tilted front..., Ahn [/bib_ref] [bib_ref] High peak power (≥10 mW) quantum cascade superluminescent emitter, Aung [/bib_ref]. We route the waveguides to the same side of the chip for ease of measurement, while ensuring both arms have practically the same path length to match their optical properties. Finally, an important design consideration when working with a system of two or more elements on the same wafer is electrical crosstalk due to a non-negligible ground impedance. To mitigate this issue, instead of the more conventional common, substrate-side ground used in QCLs, we use separate top ground pads for the racetrack laser and each waveguide arm.
## Temporal dynamics and mode switching
The racetrack laser supports two counter-propagating modes which in an ideal cavity are best described by travelling waves. Although we optimize the fabrication process, even minimal sidewall roughness will lead to backscattering and coupling between the modes. One way to preserve travelling waves in the cavity is if only one of the counter-propagating modes is allowed to lase at a time. A mechanism which allows us to select the dominant mode accurately and to efficiently switch between them is a useful tool to achieving controllable unidirectional lasing in the racetrack. To distinguish the two modes experimentally, in our measurements we use a thermoelectrically (TE)-cooled HgCdTe (MCT) detector with a 1x1 mm active area along with a pair of ZnSe lenses of 1.5" focal length. By translating the detector across the length of the device, we are able to fully resolve the two peaks emitted from both waveguide arms, which correspond to the CW and CCW modes. All measurements are performed in pulsed mode using 300 ns pulses with 50 kHz repetition rate. The device is thermally stabilized at room temperature with a TE cooler.
We are able to detect the signal from the racetrack laser exiting the waveguide facets at ports 3 and 4 (shown in the top schematic of even without any pump applied to the waveguide arms, although the signal is attenuated due to significant absorption along each arm. The racetrack threshold current is ≈ 1.25 A or 2.89 kA/cm 2 . Our measurements show that the racetrack laser is unidirectional; however, at each current pulse, it selects between the CW and CCW mode randomly. Under pulsed measurements, the output thus fluctuates between the two modes.
We show that a small amount of amplified spontaneous emission injected into one of the two arms favors one of the modes. In our experimental demonstration of mode selection we differentially pump the two waveguide arms taking advantage of the fact that spontaneous light emitted by each arm only couples to one of the two possible modes. Similarly, each mode will couple to a different arm of the active waveguide which allows us to distinguish them with our MCT detector. The schematic in illustrates these mode interactions with color-coded arrows. In our experiment we pump the two waveguide arms differentially by applying a variable current I = I 0 ± ∆I to each arm. This way the average applied current, I 0 , at any point during our sweep remains constant. With the external MCT detector we then capture the signal at ports 3 and 4 (marked in orange in the schematic). The racetrack signal exiting the facets at 3 and 4 has interacted with the waveguide arms which provide pump-dependent absorption or amplification. An experimental measurement of this interaction is shown on the bottom right panel of which depicts the CCW output at port 3 as a function of the current applied on the left waveguide arm in the same schematic (which favors CCW operation). For up to ≈ 400 mA of applied current, absorption is the dominant effect; however, beyond this point, gain in the waveguide takes over and the output is amplified with increasing current. Measuring this characteristic gain curve allows us to calculate the output right as it exits the racetrack (schematically depicted by the turquoise ports 1 and 2). We use a Softplus function to fit the gain curve and then use this fit to calculate what the measured signal exiting ports 3 and 4 looks like at ports 1 and 2. This allows us to investigate and display the mode selection mechanism more clearly without the additional effects of absorption or amplification in the waveguide. : Top: System schematic. Bottom left: CW and CCW normalized intensity given as a function of injected current imbalance between the two waveguide arms as illustrated in the system schematic. The CW (CCW) mode shown in dotted red (blue) dominates for positive (negative) ∆I. This is the signal right as it exits the racetrack, represented by the turquoise markers at ports 1 and 2 in the schematic. In solid red (blue) we show the measured probability that the mode is CW (CCW) from 15 consecutive measurements. Bottom right: CCW power measured by an MCT photodetector as it exits the waveguide (noted by an orange marker at ports 3) along with a Softplus fit.
We have shown on the bottom left panel of the measured CW and CCW signals corrected using the fit discussed above as a function of the current differential between the two waveguide arms. This is the output immediately after it is outcoupled from the racetrack before it travels through the length of the waveguide arms. The lasing mode (CW or CCW) is only dependent on which arm has a higher pump. Following notation in the schematic of , for ∆I > 0, the right arm has a higher spontaneous emission, favoring the CW mode. Conversely, for ∆I < 0, the left arm's stronger emission favors the CCW mode. For a small region around ∆I = 0, the mode is chosen at random. In the same plot we show the probability that the mode is CW (red, solid) or CCW (blue, solid) as computed from a representative set of 15 consecutive measurements for any given applied current. The intensity of the dominant mode remains relatively constant as we sweep ∆I. Any light injected into the racetrack from the waveguide arms in these measurements is due to spontaneous emission. In fact, the two waveguide arms form a FP cavity and therefore can achieve lasing if pumped together. However, in any of these measurements, the bus waveguide is well below lasing because we only pump one arm at a time which leads to significant loss in the bus cavity. Together with the 17 - facets, this loss leads to a high lasing threshold (≈ 7.84 kA/cm 2 or 3.36 A if both arms are pumped simultaneously).
A mode selection mechanism based on spontaneous emission injection is consistent with our measurements. We have also considered the role that feedback of the racetrack signal from the facets back into the racetrack may play in our system; however, we have found this effect to be weak considering the combined effect of the weak reflection from the facets and a double pass through the 15% coupler.
Having shown that we are able to select one of the two modes based on the current differential ∆I, we now examine the effectiveness of switching between modes. We measure the amount of current imbalance needed to switch for various average currents I 0 applied to the waveguide arms. To this end, we assess the probability for each mode to dominate by capturing a representative train of 15 consecutive measurements from the MCT detector as shown schematically in [fig_ref] Figure 4: Top [/fig_ref]. We only observe two possible responses from the photodetector. For example, if we are observing at the CW port, we see a high photodetector response if the mode is CW or zero response if the mode is CCW and never in between. The probability P CW (CCW ) is then defined as the number of pulses measured in CW (CCW) mode divided by the total number of pulses, in our case 15. The results of this analysis are shown in [fig_ref] Figure 4: Top [/fig_ref] for three different values of I 0 . Note the crossover happens around ∆I ≈ 0 mA. It is always slightly shifted to one side, which we explain based on a slight imbalance in the lengths of the two waveguide arms. Another key detail is the span of the bistable region (where probabilities of each mode are comparable). To determine this span we fit our data using the error function and calculate the width going from 10% to 90% probability. For I 0 = 620 mA this region spans ≈ 73.2 mA. As I 0 increases to 900 mA, the span of this region decreases to 13.8 mA.
## Bistability mechanism
Thus far we have shown that we can select the direction in which the racetrack laser emits by modulating the current differential into the waveguide arms. However, a question arises on the mechanism that enables this phenomenon. Is it necessary to continuously pump a particular arm of the waveguide if we wish to maintain the directionality of the racetrack emission or is it sustained by the racetrack dynamics itself? To answer this question we conduct the experiment shown in [fig_ref] Figure 5: Top [/fig_ref]. The setup is described by the upper panels of the figure.
A 300 ns pulse of ≈ 1.81 A is applied to the racetrack. At this current, the racetrack is 1.4 times above the lasing threshold. Midway through the racetrack pulse, we apply a short current pulse of width ≈ 30 ns and current up to 1.15 A to the waveguide arm that favors CCW operation. In the opposite arm we apply a 300 ns pulse of ≈ 480 mA just above the necessary current input for absorption mitigation. Observing the photodetector response during each pulse before and after the short 30 ns pulse input is one way to address the question we have posed above. On the lower left plot we show the output on the CCW side (port 3) measured on a TE-cooled MCT detector when the input CCW spike peak (in blue) is lower than the longer CW pulse (in red). We plot 15 consecutive measurements which reveal a seemingly random selection of the lasing direction. When the laser's CCW mode dominates, we measure a pulsed signal on the detector; conversely, when the CW mode dominates, no signal is measured, as shown in [fig_ref] Figure 5: Top [/fig_ref]. On the lower-right plot we show photodetector measurements when the spike injected into the arm favoring CCW operation is higher than the CW background injected into the other arm. We observe that this input perturbation causes the racetrack laser to switch to the CCW mode deterministically, regardless of the initial lasing mode. This can be seen for the section of the pulse highlighted in gray, where all of the measurements show a high signal on the detector, corresponding to the CCW mode. Moreover, the laser remains fixed at the CCW mode, even after the perturbation that caused the switch dissipates. This sheds some light on the mode selection mechanism.
While pumping the external waveguide arm favors one particular racetrack mode to dominate, the selected mode is self-sustaining thereafter. This demonstrates the bistable nature of the racetrack dynamics which allows for unidirectional mode selection.
## Spectral characteristics
The spectral properties of the racetrack laser and the impact that pumping of the waveguide arms has on them are summarized in [fig_ref] Figure 6: Spectral characteristics of the device [/fig_ref]. We measure the signal exiting one of the waveguide arms using a Fourier Transform Infrared Spectrometer (FTIR) with 0.125 cm −1 spectral resolution. The pure racetrack spectrum with no current applied to either waveguide arm is shown in the left panel and it reveals a series of longitudinal modes spanning ≈ 14 cm −1 with a measured free-spectral range (FSR) of ≈ 0.70 cm −1 which is an excellent match for the calculated value 1/nL = 0.701 cm −1 , where L = 4.318 mm is the designed racetrack cavity length and n = 3.3 is the effective refractive index of the ridge waveguide. On the right panel of [fig_ref] Figure 6: Spectral characteristics of the device [/fig_ref] we show the CW spectra as a function of current imbalance between the arms ∆I as well as average current I 0 which is marked at the top. In this case, a positive ∆I means higher current is applied to the waveguide arm favoring CCW operation which explains why the spectra within each panel are generally weaker for ∆I > 0. The FSR, once some current is applied to the waveguide arms, remains the same as for the pure racetrack spectrum, although there is a small red shift in the center frequency which can be explained by increased local temperature of the chip due to the applied current 2I 0 . For low values of I 0 , despite the red shift, the spectral envelope that we measure remains similar to the racetrack spectrum without injected current. As I 0 increases this envelope changes significantly with only a few modes becoming dominant as can be seen for I 0 = 790 mA and I 0 = 900 mA. For these high values of I 0 we are able to recover the pure racetrack spectrum around ∆I = 0 mark. This is most visible for I 0 = 790 mA where even though only a couple of modes dominate for negative ∆I, around the 0 mark, we recover the original envelope. Hence, lower values of current applied to the waveguide arms do not alter the spectral envelope of the racetrack laser; however, for higher values the envelope changes significantly. When the current differential between the arms is relatively small, the resulting spectra are also similar to the pure racetrack spectrum envelope.
# Conclusion
In this paper we have shown the operation of a novel monolithic QCL-based system comprising of a racetrack laser and an evanescently-coupled bus waveguide with independently-controlled arms based on a specially-designed tapered coupler. We demonstrate deterministic and controllable selection between the two counter-propagating racetrack modes via spontaneous emission coupling from the differentially pumped waveguide arms. As little as 14 mA of current imbalance, which represents ≈ 1.6% of the average input current into the waveguides, can trigger a full and permanenet switch between the CW and CCW modes. By further investigating the mechanism of mode selection, we have demonstrated the possibility of bistable operation of our device. Because of the bistable nature of the ring, the waveguide arms do not need constant pumping in order to maintain a particular directional mode. In fact, a small perturbation is sufficient to select the mode and the racetrack laser will maintain this mode even after the input perturbation dissipates.
Finally, we have demonstrated an efficient outcoupling mechanism by measuring up to ≈ 4 mA of optical power coupled out of the racetrack laser. These results bring us closer towards practical applications of controllable unidirectional ring or racetrack QCLs.
Funding This work was supported by the Eric and Wendy Schmidt Transformative Technology Fund. S.K. acknowledges support from the Yan Huo *94 Graduate Fellowship.
[fig] Figure 1: Device schematic showing the racetrack laser ridge and the evanescently-coupled waveguide with facets tilted at 17 • . The dotted purple box shows a magnified view of the taper and s-bend employed on the ridges in order to enhance their coupling. The blue and red arrows illustrate how light is coupled in and out of the racetrack. Yellow regions represent a schematic of the top and ground metal contact pads. The bottom left panel shows the calculated mode intensity of a 5 µm ridge waveguide. The bottom right panel offers a few combinations of gap width and coupling length leading to particular coupling strengths as calculated by CMT. The devices discussed here correspond to the last set of listed parameters. [/fig]
[fig] Figure 2: Scanning electron microscope (SEM) images of the device, focusing on the coupler. Top left: A top-view of the coupling region and taper. Top right: Cross-sectional view along the taper. The darker contour along the ridges is the insulating SiNx layer. The coupling region is kept free of metal. Bottom left: Cross-sectional view of the ridge waveguide elsewhere in the device with a dielectric window and Ti/Au metal layer for electrical contact. Bottom right: Magnified view showing the etch profile and minimal, highly sub-wavelength roughness on the waveguide sidewalls. [/fig]
[fig] Figure 4: Top: Schematic of a received pulse train used to measure the probability of each mode appearing. In this example, the probability P CW = P CCW = 0.5. Bottom: Probability of mode selection for the CW (red) and CCW (blue) modes. The solid line is an error function fit of the raw data (displayed as dots). The mean current increases from left to right from 620 mA to 900 mA. Required switching current, calculated as the width of the 10%-90% interval of each fit, decreases from 73.2 mA to 13.8 mA from left to right. The racetrack current in these measurements is held constant at 1.81 A (4.19 kA/cm 2 ). [/fig]
[fig] Figure 5: Top: Input pulses into the racetrack and each waveguide arm. (a) The spike coupling to the CCW mode is smaller than the CW background. (b) The spike coupling to the CCW mode is larger than the CW background. (c), (d) 15 superimposed photodetector traces at port 3 corresponding respectively to pumping scenarios in (a), (b). Note the difference in the region shaded in gray between plots (c) and (d). When the short pulse exceeds pumping in the opposite arm, the CCW mode always dominates. (e) Device schematic along with mode interactions. [/fig]
[fig] Figure 6: Spectral characteristics of the device. The left panel corresponds to the spectrum of the racetrack when no current is applied to the waveguide arms. For the other panels, the average applied current I 0 is given at the top and the spectra are given as a function of ∆I where negative (positive) ∆I favors the CW (CCW) mode. The spectra in each panel are normalized to the maximum intensity point of that panel. [/fig]
|
10.1039/d2ra00971d
|
CCBY
|
8988269
|
35424978
|
s2orc_pubmed_articles
|
Comparison of mesoporous fractal characteristics of silica-supported organocatalysts derived from bipyridine-proline and resultant effects on the catalytic asymmetric aldol performances
Experimental sectionMaterialsIonic cetyltrimethylammonium bromide (CTAB, 99% of purity) was purchased from Sinopharm Chemical Reagent Beijing Co., Ltd. Tetraethyl orthosilicate (TEOS, 98%) was purchased from Bailingwei Technology Beijing Co., Ltd. Non-ionic surfactant Pluronic tri-block copolymer (P123, Mn: 5800) was supplied by Sigma-Aldrich company. Zinc acetate dihydrate (99%), p-nitrobenzaldehyde (99.8%), 2naphthaldehyde (99.8%), 9-anthracenecarboxaldehyde (99.8%), and 1pyrenecarboxaldehyde (99.8%) were supplied by Shanghai Aladdin Biochemical Technology Co., Ltd. Dimethyl sulfoxide (DMSO, 99.9%), petroleum ether (60-90 o C), ethyl acetate (EtOAc, 99.9%), dichloromethane (CH 2 Cl 2 , 99.9%), trifluoroacetic acid (TFA, 99.9%), and cyclohexanone (99.9%) were provided by Tianjin Fuchen chemical reagents factory. Anhydrous methanol (99.9%) and ethanol (99.9%), anhydrous Na 2 SO 4 (99.5%), and ammonium hydroxide (25%, NH 3 ·H 2 O) were obtained from Beijing chemical works. 0.25 mm SDS silica gel coated glass plates (60F254) and silica gel (200-300 mesh) was purchased from Qingdao Haiyang Chemical Co., Ltd. All the chemicals were purchased as reagent grade and used without further purification. All of the solvents and reagents were of A.R. grade. Deionized water was used in all experiments.CharacterizationsThe SAXS experiments were carried out using synchrotron radiation at the 1W2A Xray beamline at Beijing Synchrotron Radiation Facility (BSRF). The wavelength of the incident X-ray was 0.154 nm. The sample-to-detector distance for SAXS was 1.59 m, calibrated with the diffraction ring of a standard sample. The scattering vector Electronic Supplementary Material (ESI) for RSC Advances. This journal is
# Materials
Ionic cetyltrimethylammonium bromide (CTAB, 99% of purity) was purchased from Sinopharm Chemical Reagent Beijing Co., Ltd. Tetraethyl orthosilicate (TEOS, 98%) was purchased from Bailingwei Technology Beijing Co., Ltd. Non-ionic surfactant Pluronic tri-block copolymer (P123, Mn: 5800) was supplied by Sigma-Aldrich company. Zinc acetate dihydrate (99%), p-nitrobenzaldehyde (99.8%), 2naphthaldehyde (99.8%), 9-anthracenecarboxaldehyde (99.8%), and 1pyrenecarboxaldehyde (99.8%) were supplied by Shanghai Aladdin Biochemical Technology Co., Ltd. Dimethyl sulfoxide (DMSO, 99.9%), petroleum ether (60-90 o C), ethyl acetate (EtOAc, 99.9%), dichloromethane (CH 2 Cl 2 , 99.9%), trifluoroacetic acid (TFA, 99.9%), and cyclohexanone (99.9%) were provided by Tianjin Fuchen chemical reagents factory. Anhydrous methanol (99.9%) and ethanol (99.9%), anhydrous Na 2 SO 4 (99.5%), and ammonium hydroxide (25%, NH 3 ·H 2 O) were obtained from Beijing chemical works. 0.25 mm SDS silica gel coated glass plates (60F254) and silica gel (200-300 mesh) was purchased from Qingdao Haiyang Chemical Co., Ltd. All the chemicals were purchased as reagent grade and used without further purification. All of the solvents and reagents were of A.R. grade. Deionized water was used in all experiments.
## Characterizations
The SAXS experiments were carried out using synchrotron radiation at the 1W2A Xray beamline at Beijing Synchrotron Radiation Facility (BSRF). The wavelength of the incident X-ray was 0.154 nm. The sample-to-detector distance for SAXS was 1.59 m, calibrated with the diffraction ring of a standard sample. The scattering vector Electronic Supplementary Material (ESI) for RSC Advances. This journal is © The Royal Society of Chemistry 2022 magnitude q ranged from 0.08 to 3.05 nm -1 for the experiment reported in this paper. The intensity of scattering, I(q) was measured as a function of q, where q was considered as 4πsinθ/λ. The sample was loaded into a sample cell and sealed with scotch tape on a groove. The thickness of the sample cell was approximately 1 mm. The scattering image was collected with an exposure time of 5 min by the single-frame mode with a 'multi-read' of 2 times. The two-dimensional SAXS images were transferred to one-dimensional data by using the Fit2D software (http://www.esrf.eu/computing/scientific/FIT2D),and further processed with the S program package.In this study, a slit-collimated beam was used, for which the mass fractals (D m ) or surface fractals (D s ) were calculated according to the power law. Besides, the powder XRD patterns were recorded on a D6 Advance X-ray diffractometer (Beijing Pu Analysis General Instrument Co., Ltd, Beijing, China) using Cu-Kα radiation (λ = 0.154056 nm, 36 kV, 20 mA) at a scanning speed of 1 o and 4 o per min in the 1 -10 o and 10 -50 o 2θ range for the experiments, respectively. The textural properties of related samples were determined from N 2 sorption isotherms measured at 77 K using JWGB JW-BK300 analyzer. Prior to each adsorption measurement the samples were outgassed under vacuum at 353 K for at least 5 h. The specific surface area was calculated using the Brunauer-Emmett-Teller (BET) method and the pore size distribution was calculated from the isotherm using the Barrett-Joyner-Halenda (BJH) model. Besides that, the total pore volume was determined from the N 2 adsorbed amount at a relative vapor pressure (P/P 0 ) of approximately 0.99. The SEM images were performed on a Hitachi field-emission scanning electron microscope (S-4800) with an acceleration voltage of 15.0 kV. The morphology and elemental composition of the materials were acquired with a TEM microphotograph (JEOL, JEM-2100F, Japan) coupled with EDX spectroscopy. The TGA was carried out on a PerkinElmer Simultaneous Thermal Analyzer (STA-8000) instrument from 30 to 900 o C at a heating rate of 10 o C/min under the N 2 atmosphere with a flow rate of 20 mL/min. The content of nitrogen was measured by vario MICRO cube elemental analyzer and the metal Zn concentrations were determined by ICP-OES using Optima 8300, PerkinElmer. The FT-IR spectra were measured recorded on a Bruker Tensor 2 spectrometer (Bruker Optik GmbH, Germany) via the KBr tablet method, in which the spectral resolution was 4 cm -1 , and 32 scans were recorded for each spectrum. The UVvis DR spectra were carried out in the wavelength range of 200-800 nm with a Shimadzu UV-2600 spectrophotometer (Shimadzu, Japan). The dr and ee values were determined by HPLC analyses on a Agilent Technologies 1200 system equipped with a photodiode array detector, using Chiralpak AD-H column (25 cm × 0.46 cm) by Daicel Chemical Ind., Ltd. Besides that, the chemical identity of representative chiral ligands (Z 1 , Z 2 , and Z 3 ) and all aldol products has been confirmed by nuclear magnetic resonance ( 1 H-or 13 C-NMR) in our previous report. 3 Based on the experimental results of this study and our group previous literature data,major advantages for these heterogeneous catalysts (Z 1 -, or Z 2 -, or Z 3 ZnBMMs, Z 1 -, or Z 2 -, or Z 3 ZnMCM-41, and Z 1 -, or Z 2 -, or Z 3 ZnSBA-15) by comparison with that of the hybrid catalyst (Z 0 -BMMs), which were prepared by Tang et al. via grafting Z 0 onto the surface of bimodal mesoporous silicas (BMMs). Taking Z 2 ZnBMMs as an example, as shown in [fig_ref] Table S2: Comparison of catalytic activity [/fig_ref].
# Additional results
## Porod plots
## Edx spectra
## Icp-oes and elemental analysis
## Ft-ir and uv-vis dr spectra
## The comparison between current research and literature
In particular, for the purpose of comparison, Z 2 grafted BMMs was prepared without Zn by post treatment in this work. As can be seen in [fig_ref] Table S2: Comparison of catalytic activity [/fig_ref] , from the perspective of their catalytic activity, Z 2 ZnBMMs as catalyst showed an excellent reusability with high activity (yield of 85%, [fig_ref] Table S2: Comparison of catalytic activity [/fig_ref] , entries 4), however, Z-BMMs, if reused more than 2 times, showed rapid catalytic deactivation (yield from 80 to 10%, [fig_ref] Table S2: Comparison of catalytic activity [/fig_ref] , entries 3 and 5).
In addition, on the basis of the N elemental data, Z 2 -grafted amounts could be roughly estimated in Z 2 -BMMs and Z 2 ZnBMMs, corresponding to 5.54 and 12.50 wt% [fig_ref] Table S2: Comparison of catalytic activity [/fig_ref] , entries 1 and 2), respectively. Obviously, further confirming that the Z 2 were successfully immobilized on the mesoporous surfaces.
In summary, we found that the immobilization of Z 2 on BMMs by coordination bonding [fig_ref] Table S2: Comparison of catalytic activity [/fig_ref] , entries 2) was more stable than that of hydrogen bonding without Zn [fig_ref] Table S2: Comparison of catalytic activity [/fig_ref] , entries 1). (2)
## Ton and tof results
[formula] = 1 2 [/formula]
Where n 1 denote the moles of the product molecules, n 2 denote the moles of the involved in catalysis Z, m 1 denote the weight of the used catalyst, and t denotes the whole catalytic reaction time, respectively. The turnover frequency (TOF) and turnover number (TON) can be described in detail as Equations 1 and 2.
As shown in [fig_ref] Table S3: Comparison of TON and TOF under the catalysis of Z 1 -,... [/fig_ref] , taking Z 1 -, Z 2 -, Z 3 ZnBMMs-100 as an example, comparison of TON and TOF under above catalytic systems. The TOF values presented in [fig_ref] Table S3: Comparison of TON and TOF under the catalysis of Z 1 -,... [/fig_ref] almost remained the same as around 0.14 -1.54 [(gh) -1 ], and the TON values in the range of 0.50 -4.85. In addition, the aldol reaction with electron-donating substituent pmethoxybenzaldehyde as a reactant by comparison with p-nitrobenzaldehyde were carried out, but the results were not good presumably due to the inappropriate catalyst and substrate. Unfortunately, the bands of aldol products were almost not reflected on TLC plates, as shown in [fig_ref] Figure S6: Comparison of the results of TLC separation on the asymmetric aldol reaction... [/fig_ref].
## Comparative experiments for asymmetric aldol reaction
## References
[fig] Figure S1: ln[q 4 I(q)] ~ q 2 curves of (A) BMMs, (B) ZnBMMs, (C) Z 1 ZnBMMs-100, (D) Z 2 ZnBMMs-100, (E) Z 3 ZnBMMs-100, (F) MCM-41, (G) ZnMCM-41, (H) Z 1 ZnMCM-41-100, (I) Z 2 ZnMCM-41-100, (J) Z 3 ZnMCM-41-100, (K) SBA-15, (L) ZnSBA-15, (M) Z 1 ZnSBA-15-100, (N) Z 2 ZnSBA-15-100, (O) Z 3 ZnSBA-15-100, and 3 rd recycled (P) Z 2 ZnBMMs-100, (Q) Z 2 ZnMCM-41-100, (R) Z 2 ZnSBA-15-100, in which, (a) Porod plot data with deviation, (b) deviation corrected Porod plot data. [/fig]
[fig] Figure S2: (A) EDX spectra and results of Z 2 ZnSBA-15-100; (a) TEM image and its corresponding EDX mapping spectra of (b) Si, (c) O, (d) C, (e) S, (f) Zn. [/fig]
[fig] Figure S3: FT-IR spectra of (A) MCM-41-, and (B) SBA-15-based samples. (a) pure mesoporous materials, (b) Zn-modified samples, (c) Z 1 -immobilized samples, (d) Z 2 -immobilized samples, and (e) Z 3 -immobilized samples, in which, the molar ratios of Z/Zn for all samples were around 1:1. [/fig]
[fig] Figure S4: UV-vis DR spectra (A) MCM-41-, and (B) SBA-15-based samples. (a) pure mesoporous materials, (b) Zn-modified samples, (c) Z 1 -immobilized samples, (d) Z 2 -immobilized samples, and (e) Z 3 -immobilized samples, in which, the molar ratios of Z/Zn for all samples were around 1:1. 2.5. Particle size distribution of Z 2 ZnMCM-41-100 Fig. S5. SEM image and particle size distribution (inset) of Z 2 ZnMCM-41-100. [/fig]
[fig] Figure S6: Comparison of the results of TLC separation on the asymmetric aldol reaction between the different aldehyde and cyclohexanone in the presence of (a) Z 1 ZnBMMs-100, (b) Z 2 ZnBMMs-100, (c) Z 3 ZnBMMs-100, (d) Z 2 , (e) Z 2 ZnBMMs-100, in which, p-nitrobenzaldehyde as a reactant for (a), (b), and (c), respectively; and p-methoxybenzaldehyde as a reactant for (d) and (e), respectively. [/fig]
[table] Table S1: Collection of elemental composition and Zn concentrations of all related samples. The molar ratios of Z/Zn for all samples were around 1:1. b The values of the loading of Z were calculated by the nitrogen content. c Determined by TGA results. d Determined by ICP-OES. e The recycled catalyst after three runs. [/table]
[table] Table S2: Comparison of catalytic activity (yield) according to the reported hybrid catalysts (Z 0 -BMMs) on the asymmetric aldol reaction of p-nitrobenzaldehyde with cyclohexanone, as well as leaching of active species. [/table]
[table] Table S3: Comparison of TON and TOF under the catalysis of Z 1 -, Z 2 -, Z 3 ZnBMMs-100. [/table]
|
10.1186/1617-9625-8-7
|
CCBY
|
2914043
|
20609260
|
s2orc_pubmed_articles
|
Determinants of resilience to cigarette smoking among young Australians at risk: an exploratory study
Background: Numerous researchers studied risk factors associated with smoking uptake, however, few examined protective factors associated with smoking resilience. This study therefore aims to explore determinants of smoking resilience among young people from lower socioeconomic backgrounds who are at risk of smoking.Methods:Overall, 92 out of 92 vocational education students accepted invitation to participate in this exploratory study. The Adelaide Technical and Further Education (TAFE) Arts campus was chosen for the study given the focus on studying resilience in young people of lower socioeconomic status i.e. resilient despite the odds. A self-report questionnaire comprising a measure of resilience: sense of coherence, sense of humour, coping styles, depression, anxiety and stress, and family, peers and community support, was distributed among participants aged 15 to 29. Additional factors researched are parental approval and disapproval, course type, and reasons for not smoking. Using the Statistical Package for the Social Sciences (SPSS, version 13.0), analyses were undertaken using frequencies, means, standard deviations, independent sample t-tests, correlations, analysis of variance, logistic regression, and chi-square test.Results: Twenty five (27%) out of 92 students smoked. Young people with peer support tended to smoke (p < .05). A relationship between daily smoking and depression, anxiety and stress was also found (p < .05). When both mothers and fathers disapproved of their children smoking, it had a greater influence on females not smoking, compared with males. The majority of students chose 'health and fitness' as a reason for not smoking. Students in the Dance course tended to not smoke.Conclusions:The current study showed that most students chose 'health and fitness' as the reason for not smoking. Single anti-smoking messages cannot be generalised to all young people, but should recognise that people within different contexts, groups and subcultures will have different reasons for choosing whether or not to smoke. Future studies should use larger samples with a mixed methods design (quantitative and qualitative).
# Background
Cigarette smoking is the single biggest cause of preventable morbidity and mortality in Australia and worldwide, responsible for 19,000 deaths per annum in 1998 in Australia [bib_ref] The quantification of drug-caused mortality and morbidity in Australia, Ridolpho [/bib_ref]. Young people have the highest rate of smoking prevalence. In 2004, 17.4% of the Australian population aged 14 years and over reported smoking daily, whereas for the [bib_ref] Association between cigarette smoking and anxiety disorders during adolescence and early adulthood, Johnson [/bib_ref] [bib_ref] Smoking is associated with first-ever incidence of mental disorders: A prospective population-based..., Cuijpers [/bib_ref] [bib_ref] Alcohol, cannabis and tobacco use among Australians: A comparison of their associations..., Degenhardt [/bib_ref] [bib_ref] The structure and properties of the sense of coherence scale, Antonovsky [/bib_ref] [bib_ref] Personality, stress, and the decision to commence cigarette smoking in adolescence, Byrne [/bib_ref] [bib_ref] Emotion, plasticity, context, and regulation: Perspectives from affective neuroscience, Davidson [/bib_ref] [bib_ref] Adolescent resilience: A framework for understanding healthy development in the face of..., Fergus [/bib_ref] [bib_ref] Changes in social influence at adolescence: Who induces cigarette smoking?, Krosnick [/bib_ref] [bib_ref] Comparing two self-report measures of coping: The sense of coherence scale and..., Sammallahti [/bib_ref] year old age group the prevalence was at 34.2% [bib_ref] National drug strategy household survey: First results, Institute [/bib_ref].
Numerous researchers studied risk factors associated with smoking uptake, such as lower socioeconomic status and young age, however, few have sought to examine the various protective factors associated with smoking resilience. The majority of resilience research is limited to investigations of single variables and their relationship to resilience. The current exploratory study has made an attempt to bring together variables identified as promoting resilience, and to examine the strength of the relationships.
Positive psychologists define resilience as the interplay between personal characteristics and environmental fac-tors resulting in the ability to overcome adversity [bib_ref] Resilience in the face of adversity: protective factors and resistance to psychiatric..., Rutter [/bib_ref]. Antonovsky's salutogenic model explores factors assisting individuals in being resilient despite exposure to major life stressors. Rather than seeking the causes of disease, Antonovsky recommended focusing on the causes of successful and positive individual health. Resilience, or resistance to adverse outcomes, has been associated with protective factors both within and outside the individual. Factors within the individual are: sense of coherence (an important internal resource assisting individuals in dealing with stress that is applicable to all cultures and demographic contexts) [bib_ref] Validity of Antonovsky's sense of coherence scale: A systemic review, Erikson [/bib_ref] , sense of humour, positive coping style, and mental health. External factors in the social environment are: family, peers, and community.
In particular, sense of humour is recognised as a buffer that improves the capacity to cope and experience positive feelings, despite adversity [bib_ref] Vulnerability and competence: a review of research on resilience in childhood, Luthar [/bib_ref] [bib_ref] Individual differences in uses of humor and their relation to psychological well-being:..., Martin [/bib_ref] [bib_ref] Sense of humor when coping with life stress, Overholser [/bib_ref] [bib_ref] The Psychobiology of depression and resilience to stress: Implications for prevention and..., Southwick [/bib_ref] [bib_ref] Development and validation of a multidimensional sense of humor scale, Thorson [/bib_ref]. It can promote resilience within both interpersonal and intrapersonal contexts. It is therefore plausible that those who do not use humour are less resilient and more likely to smoke and suffer from tobacco related illnesses.
Research also shows a strong association between coping style and smoking status. A study comparing smokers with those who have never smoked found that the latter group used active problem solving to cope with stress (confrontative coping) whereas smokers used avoidant coping such as distraction and substance abuse [bib_ref] Are psychosocial factors related to smoking grade-6 students?, Pederson [/bib_ref].
People without mental illness are also more likely to exhibit smoking resilience. South Australian data showed that 39.4% of individuals with a self-reported mental illness smoked, compared to 22.7% for people with no mental illness. In contrast with the progressive decline of tobacco smoking rates in the general population, smoking rates for individuals with depression continue to increase. As with depression, low levels of anxiety were also associated with smoking resilience. Johnson, et al. [bib_ref] Association between cigarette smoking and anxiety disorders during adolescence and early adulthood, Johnson [/bib_ref] tested this association in a sample of late adolescents and young adults and concluded that smoking could moderately predict the development of anxiety disorders. Subsequent research confirmed these results [bib_ref] Smoking is associated with first-ever incidence of mental disorders: A prospective population-based..., Cuijpers [/bib_ref] [bib_ref] Alcohol, cannabis and tobacco use among Australians: A comparison of their associations..., Degenhardt [/bib_ref]. Stress, a milder form of anxiety, can also contribute to smoking initiation [bib_ref] The structure and properties of the sense of coherence scale, Antonovsky [/bib_ref] [bib_ref] Personality, stress, and the decision to commence cigarette smoking in adolescence, Byrne [/bib_ref]. Those who successfully manage stress and quickly recover from stressful situations have increased smoking resilience [bib_ref] Emotion, plasticity, context, and regulation: Perspectives from affective neuroscience, Davidson [/bib_ref].
Moreover, it is widely accepted that family support and in particular parental support, is highly important in building resilience in young people [bib_ref] Adolescent resilience: A framework for understanding healthy development in the face of..., Fergus [/bib_ref]. Young peers are another strong influence. Krosnick and Judd [bib_ref] Changes in social influence at adolescence: Who induces cigarette smoking?, Krosnick [/bib_ref] suggested that parental influence does not decrease in adolescence, but rather, the influence of peers increases. Peers can both promote and deter the use of tobacco.
In addition to positive social support from family and peers, resilience in young people can also be influenced by community support. The emphasis on community support is consistent with resilience research stressing the importance of the socioecological context in helping young people to avoid the negative effects of risk factors [bib_ref] Adolescent resilience: A framework for understanding healthy development in the face of..., Fergus [/bib_ref].
The current study chose to explore the problem of resilience in the context of smoking among young people. In particular, the study focused on young people of lower socioeconomic status who, despite the odds, do not smoke. The vocational education students from TAFE (Technical and Further Education) have particularly high smoking rates, 44.6%compared with the average of 21.9% in the adult population of 15 years and over. For this reason and due to the fact that many TAFE students come from low socioeconomic backgrounds, this group was targeted for the current study. The main aim of this study was to identify factors contributing to smoking resilience among young people at risk.
# Methods
## Participants and setting
A total of 92 questionnaires were handed out, and 92 completed questionnaires were returned (100% response rate). The sample consisted of 37 male and 55 female students. The majority did not smoke and were of lower socioeconomic status. [fig_ref] Table 1: Demographic Summary of Student Characteristics [/fig_ref] presents a summary of selected demographic statistics for the sample. Participants were students from the TAFE Adelaide College of the Arts campus (vocational education) from classes across all year levels (year 1, 2 and 3) in Bachelor of Dance Performance, Advanced Diploma of Arts, and Bachelor of Visual Arts and Applied Design. Eligibility was limited to young people, defined as 15-29 years of age. Compared with other TAFE campuses in the metropolitan region, the Adelaide College of the Arts had the highest rate of students who had or were eligible for concession fees, and who were thus classified as being of lower socioeconomic status. This campus was therefore chosen given the current study's focus on researching resilience in young people of lower socioeconomic status, i.e. resilient despite the odds. Eligibility for concession fees required that students or their caregivers held a valid concession card, for example, the Health Care Card. Eligibility is assessed by Centrelink (an Australian commonwealth department supporting people in financial need) who assess the socioeconomic status of applicants. The survey asked students to indicate whether they were on or eligible for concession. Given that the majority of students (79.3%) were on concession, the sample was representative of a campus regarded as being of lower socioeconomic status.
## Course type and smoking status
Of the 92 students, 17.39% (n = 16) were enrolled in the Advanced Diploma of Arts (Diploma of Arts), 53.26% (n = 49) in the Bachelor of Visual Arts and Applied Design (Visual Arts), and 29.35% (n = 27) in the Bachelor of Dance Performance (Dance). Of the students in Diploma of Arts, 37.5% (n = 6) were smokers, and 62.5% (n = 10) were non smokers. For Visual Arts, 32.7% (n = 16) were smokers, and 67.3% (n = 33) were non smokers, and in Dance 11.1% (n = 3) were smokers, and 88.9% (n = 24) were non smokers (See. The above descriptive statistics indicate that Bachelor of Dance Performance students have a tendency to not smoke. Of the Dance students, 26 out of 27 were male. Dance students tended not to smoke.
# Materials
Participants received a nine page self-report questionnaire consisting of 101 questions (see Additional file 1, Questionnaire) divided into sections on sense of humour, style of coping, social support (family, peers, and community), sense of coherence, and depression, anxiety and stress (DASS). Background information included questions on smoking behaviour, for example, whether or not participants smoked, parental approval and disapproval, course type, and students' reported reasons for not smoking. Due to low numbers of daily smokers, participants were labelled as smokers if they smoked daily, weekly, or monthly. All scales except DASS asked participants to answer questions based on the past month, and DASS for the past week.
## Sense of coherence scale
Sense of coherence (SOC) was based on Antonovsky'sOrientation to Life Questionnaire, more commonly known as the 'Sense of Coherence Scale'. The shortened 13 item version was used. Five items were reverse scored. Wording was slightly modified for the sample. Total range of possible scores were 13-65, with higher scores (45-65) indicating higher SOC. The SOC scale has been widely used and validated [e.g. [bib_ref] The structure and properties of the sense of coherence scale, Antonovsky [/bib_ref] [bib_ref] Validity of Antonovsky's sense of coherence scale: A systemic review, Erikson [/bib_ref] [bib_ref] Comparing two self-report measures of coping: The sense of coherence scale and..., Sammallahti [/bib_ref] ]. In 124 studies using the SOC-29, the Cronbach's alpha ranged from 0.70 to 0.95, and in 127 studies using the SOC-13 it ranged from 0.70 to 0.92, supporting claims of validity. The scale is adaptable to all demographics. The Cronbach's alpha in the present study was 0.65.
## Sense of humour scale
The scale for sense of humour was constructed by the researcher for this study. A 5-point rating scale was used, ranging from 1 for 'never' to 5 for 'always.' Higher scores (3 to 5) indicated a higher sense of humour. The scale contains three items, two addressing the individuals' own perceptions of their sense of humour, and one regarding other people's perceptions. The latter was included to increase the scale's validity. To test for internal consistency, the Cronbach's alpha for this scale was calculated, showing 0.68.
## Coping style for adults
The scale for coping style was based on the 'Coping Style for Adults' (CSA) by Frydenberg and Lewis, and was modified for consistency with Holahan and Moos' [bib_ref] Personal and contextual determinants of coping strategies, Holahan [/bib_ref] theory of avoidant and confrontative coping. Two items measured avoidant coping and two measured confrontative coping. The wording was slightly modified for a younger population. A 5-point rating scale ranged from 1 for 'never' to 5 for 'always'. Moderate to high scores were High scores for questions 1 ("How often did you try to cope with problems by talking about them to other people?") and 3 ("How often did you reflect on problems, plan solutions, and tackle problems systematically?") indicated a moderate to high confrontative coping style, whereas high scores for questions 2 ("How often did you try to cope with problems by keeping them to yourself?") and 4 ("How often did you consciously block out problems?") indicated a moderate to high avoidant coping style. Frydenberg and Lewisclaimed validity for the CSA scale based on analysis of five studies reporting significant relationships between desired outcomes (stressfree) and productive coping strategies. Similarly, Holahan and Moos [bib_ref] Personal and contextual determinants of coping strategies, Holahan [/bib_ref] enjoy widespread support and acceptance for their theory on avoidant and confrontative coping. The Cronbach's alpha for the current study's modified scale was 0.62.
## Depression, anxiety and stress scale (dass)
Depression, anxiety and stress were measured using the 42 item 'Depression, Anxiety, and Stress Scale' (DASS) developed by Lovibond and Lovibond. The scale ranged from 0 for 'did not apply to me at all,' to 3 for 'applied to me very much or most of the time.' Higher scores indicated higher levels of depression, anxiety and stress. Moderate to high levels of depression were represented by 14-28+, for anxiety 10-20+, and stress 19-34+. The DASS is widely used with strong claims of validity [e.g. [bib_ref] Psychometric properties of the 42-item and 21-item versions of the Depression Anxiety..., Antony [/bib_ref] ]. In particular, Lovibond and Lovibond'ssupport for the validity of the scale is based on a study using predominantly students, thereby arguably increasing validity for subsequent studies also using students. The Cronbach's alpha for this scale was 0.96. The scale can be used with participants as young as twelve.
## Multidimensional support scale
The scale on social support for family, peers and community was based on the 'Multidimensional Support Scale' (MDSS) developed by Winefield, Winefield and Tiggeman [bib_ref] Social support and psychological well-being in young adults: The Multi-Dimensional Support Scale, Winefield [/bib_ref]. Minor modifications were made, for example, the Likert scale was reduced from a 7-point to a 5-point scale for consistency, ranging from 1 for 'never' to 5 for 'always.' Family and peer support measures consisted of six items each, and three for community support. Higher scores indicated higher support levels. Validity of the MDSS was addressed by Winefield et al. [bib_ref] Social support and psychological well-being in young adults: The Multi-Dimensional Support Scale, Winefield [/bib_ref] by conducting multiple linear regressions, showing that support measures significantly increased the amount of explained variance in well-being measures. The authors removed one item from the questionnaire due to poor validity (How often do you tell jokes and chatter?). This item was not included in the current study. The MDSS is flexible and adaptable for use with different populations. The Cronbach's alpha for the current study's modified scale was 0.87 for family and peers, and 0.91 for community.
The additional factors: parental approval and disapproval, course type, and students' reasons for not smoking, were researched using information obtained from the Background Information section of the survey. Parental approval and disapproval was measured by asking students to tick a box (either 'approves' or 'disapproves') for the following questions: "How does your mother/female caregiver feel about smoking" and "How does your father/ male caregiver feel about smoking." Information on course type was obtained by asking for the full name of the course the students were studying. Students' reasons for not smoking were measured by asking, "If you haven't taken up smoking (daily) why?" They were asked to tick one box, choosing the main reason from the following options: influence of friends, influence of family, influence of people in the community, health or fitness reasons, it's not cool/people would look down on me, other.
## Procedure
Questionnaires were handed to students during class. The teacher notified students at the start of class that a research student would arrive towards the end of class inviting voluntary participation in a research study. The researcher commenced with a brief presentation on the purpose and nature of the study, including information on voluntary participation and management of privacy and confidentiality. Data were collected over a two week period in six classes. The researcher chose to present the questionnaires personally in order to answer any questions, and to maximise interest and subsequent response rate.
# Analysis
Categorical variables are presented as frequencies and percentages. Continuous variables are presented as means and standard deviations. Independent sample ttests (two-sided) were conducted to compare smokers and non-smokers with respect to normally distributed continuous variables of interest. To examine the extent to which the predictor variables were able to classify people as smokers or non-smokers, a logistic regression was conducted using 'do you currently smoke' as the dependent variable (recoded 0 = non-smoker, 1 = smoker) with peer support levels as the predictor. An exploratory analysis comparing depression, anxiety and stress scores in relation to how often a person smoked (i.e. daily, weekly, or monthly) was conducted with one-way analysis of variance (ANOVA). In order to determine which groups differed from each other we used the Tukey's honestly significant difference (HSD) test. Pearson's correlation was performed to observe relationships between parental approval/disapproval and smoking status.
## Ethics
Approval for the study was given by the University of Adelaide Human Research Ethics Committee.
# Results
## Smoking status and psychological and social variables
No significant differences were found for smoking status in relation to sense of coherence, sense of humour, coping style, family support and community support. A relationship was found between peers and smoking status [fig_ref] Table 2: Predictor of smoking status [/fig_ref]. Additional exploratory analyses compared depression, anxiety and stress scores in relation to how often a person smoked (daily, weekly or monthly). A summary of these findings, including the results for a one-way ANOVA is provided in [fig_ref] Table 3: Levels of Depression, Anxiety, and Stress according to Smoking Frequency [/fig_ref]. Post-hoc tests (Tukey's HSD) showed that scores for all three measures were significantly higher in the daily smoking group than in the other two groups. However, when all smokers and all non-smokers were compared on depression, anxiety and stress no difference was detected (p > .05).
## Peer support and smoking status
Smokers reported significantly higher peer support (M = 4.00, SD = .54) than non smokers (M = 3.67, SD = .71), t (90) = 2.10, p < .05 (Cohen's d = .44). The Cohen's d indicates a moderate effect size.
A logistic regression analysis [fig_ref] Table 2: Predictor of smoking status [/fig_ref] showed that, for each unit increase in peer support scores, a student was 2.15 times more likely to be a smoker. The percentage of cases correctly classified was 73% indicating that the model was substantially better than chance in its ability to classify cases.
## Parental approval/disapproval, and smoking status
Smoking status in relation to parental approval and disapproval was examined to determine whether parents potentially influenced their children's smoking status. No young people reported smoking if their fathers approved and mothers disapproved of their smoking (0 out of 10) and only 2 out of 12 smoked when their fathers disapproved and their mother approved. When analyses were confined only to those who smoked, a strong correlation was found between parental approval (both parents) and the frequency of smoking (coded as separate categorical variables in point bi-serial correlations). If there was parental approval, then students tended to be less frequent smokers: (r (22) = -.90, p < .01).
## Family support and smoking status
A very strong positive correlation was found between weekly smoking, and the item, 'family really tried to listen and understand' (r (22) = .95, p < .05). A very strong posi-tive correlation was also found for weekly smoking and the item 'using family as an example of solving problems' (r (79) = .95 p < .05)
## Smoking status according to parental approval/ disapproval and gender
Gender differences were also considered in relation to the relationship between smoking status and parental approval. For males whose parents did not approve, there were 16 non smokers and 10 smokers, whereas for females, there were 34 non smokers, 9 smokers if parents disapproved. There was no significant relationship between gender and smoking status in this subsample (p > .05) as indicated by a chi-squared test, but there was a trend toward a relatively higher number of smokers in the male group. [fig_ref] Table 4: Frequency scores for Reasons Given for Not Smoking [/fig_ref] presents the reasons given by students for not smoking. Health and fitness were reported the most frequent reasons these students did not smoke.
## Students' reported reasons for not smoking
# Discussion
The aims of the current exploratory study were to bring together a number of variables related to resilience and smoking status and study them in the one investigation. The study sought to identify factors contributing to smoking resilience among young people at risk, i.e. those who despite the odds do not smoke.
Previous studies have indicated relationships between smoking status and the following factors: sense of coherence, sense of humour, coping style, depression, anxiety, stress, and social support (family, peers, and community). The present study found a relationship between peers and smoking status (smoking or non smoking), but other expected differences were not obtained. Other analyses revealed a relationship between smoking status with smoking frequency, parental approval, course type, and health and fitness.
Although no association was found between smokers and non smokers for depression, anxiety and stress, an association was found for frequency of smoking (daily, weekly, monthly). Scores for depression, anxiety and stress were higher for daily smokers. This is consistent with previous studies that reported depression, anxiety and stress using daily measures for smoking [bib_ref] The psychopharmacology of tobacco dependence, Balfour [/bib_ref]. Given the limited research on smoking frequency, it would be useful to replicate the current study's measures for daily, weekly and monthly smoking. In addition, questions regarding the type and strength of cigarettes smoked could provide useful insights.
The current study found an association between family support and smoking status. Very strong positive correlations were found between weekly smoking and the items 'how often did your family really try to listen when you talked about your problems or worries?' and 'how often could you use them as examples of how to deal with problems or worries?' This indicated that students who smoked weekly perceived their family as trying to listen to them more, and more frequently regarded them as examples of how to deal with problems. Progression to weekly smoking may feasibly have captured the attention and concern of parents or other family members (whether they directly noticed the smoking or other accompanying changes in behaviour), which could explain the results. However, previous research reported that daily smokers tended to believe that the adults who were most significant to them would most likely approve of them smoking [bib_ref] Smoking among a sample of Australian teenagers: Perceptions of social and health..., O'callaghan [/bib_ref]. Perhaps these smokers had conflicting feelings about their smoking status, and in an effort to reduce the discomfort (cognitive dissonance) they decided that others approved of their smoking. Differences between daily, weekly, and monthly smoking warrant further attention as this may inform smoking assessment and prevention programmes. Exploratory analyses showed that students in the Dance course had a tendency to not smoke. Smoking is an incompatible choice given that exercise is an integral and ongoing component of the course. Braverman's [bib_ref] Research on resilience and its implications for tobacco prevention, Braverman [/bib_ref] study of social status suggested that boys avoided smoking and gained social status through participation in sport. As with Braverman's study, young people in the Dance course were also predominantly exposed to non smokers, and therefore in regards to smoking resilience they were associating with peers of positive influence. This is an important finding. It shows that given the right motivation, young people will avoid smoking uptake. A literature review showed that overall, young people did not respond to messages about the health consequences of smoking if their motivation for smoking was stronger and more overpowering, such as fulfilling a need for social intimacy or lubricating social interactions [bib_ref] Cigarette smoking in relation to depression: Historical trends from the Stirling county..., Murphy [/bib_ref] [bib_ref] Family and friends' influences on the uptake of regular smoking from mid-adolescence..., West [/bib_ref]. Furthermore, given that most students in Dance were male (26 out of 27), future research could explore these gender differences.
Consistent with this, the majority of students chose 'health and fitness' as the reason for not smoking. This provides an opportunity for future research to expand on health and fitness as a motivating factor to not smoke. A number of students also ticked 'other' among the options presented. Future research could ideally measure this item using a qualitative design.
The majority of research has tried to predict smoking but very little has been done on issues of resilience despite the odds. Given that resilience can be taught, an understanding of smoking resilience predictors can have implications for interventions. Single anti-smoking messages cannot be generalised to all young people, but should recognise that different contexts, groups and subcultures will have different reasons for choosing whether or not to smoke. The findings also highlight the importance of utilising a positive approach in understanding human behaviour. For example, young people with a commitment to and enjoyment of Dance can view this activity as a strength, which might not only contribute to smoking resilience but could also be transferred to numerous other domains such as resilience against bullying, and moral resilience. This supports results of previous studies that emphasised the importance of individual strengths and virtues as valuable resources [bib_ref] Positive psychology: An introduction, Seligman [/bib_ref]. If viewed from a socioecological perspective, the findings illustrate the interplay between the individual and the social environment. Therefore, by creating supportive environments specific to the needs of a particular group type (such as the Dance students) healthy non smoking lifestyles may be successfully promoted. The design of the study is limited by inherent problems of self report measures. The validity of self-reported smoking in particular is often questioned due to the belief that smokers tend to deny smoking [bib_ref] The validity of smoking self-reports by adolescents: A re-examination of the bogus..., Murray [/bib_ref]. This could account for the lack of significant differences found between smokers and non smokers. In addition, due to the low number of reported daily smokers, the scores for weekly and monthly smokers were added to form a combined score. A larger sample would yield higher daily smokers, thereby increasing the power of comparison across all factors. Results are limited to the sample of TAFE students and cannot be generalised to the population. However, the study highlights the importance of recognising that different individuals within different settings will yield unique results for smoking resilience.
Future studies should also use a mixed methods design (quantitative and qualitative) and could be longitudinal, ideally enlisting students at commencement of their course through to completion to identify factors that affect smoking resilience.
# Conclusions
Young people with peer support tended to smoke. A relationship between daily smoking and depression, anxiety and stress was also found. When both mothers and fathers disapproved of their children smoking, then fewer children smoked, but this influence was greater on females compared with males. The majority of students chose 'health and fitness' as a reason for not smoking. Single anti-smoking messages cannot be generalised to all young people, but should recognise that people within different contexts, groups and subcultures will have different reasons for choosing whether or not to smoke. Future studies should use larger samples with a mixed methods design (quantitative and qualitative).
# Additional material
[fig] Figure 1 &RXUVH: (QUROOPHQW RI 1RQ 6PRNHUV %DFKHORU RI YLVXDO DUWV DSSOLHG GHVLJQ %DFKHORU RI GDQFH SHUIRUPDQFH $GYDQFHG GLSORPD RI DUWV represented by responses of 3 to 5 on individual items. [/fig]
[table] Table 1: Demographic Summary of Student Characteristics [/table]
[table] Table 2: Predictor of smoking status [/table]
[table] Table 3: Levels of Depression, Anxiety, and Stress according to Smoking Frequency (Daily, Weekly, Monthly) [/table]
[table] Table 4: Frequency scores for Reasons Given for Not Smoking [/table]
|
10.1186/1756-0500-5-217
|
CCBY
|
3416577
|
22559742
|
s2orc_pubmed_articles
|
Ectopic ERK expression induces phenotypic conversion of C10 cells and alters DNA methyltransferase expression
Background: Many lung carcinogens activate mitogen activated protein kinase (MAPK) pathways and DNA methyltransferases (DNMTs) are under investigation as therapeutic targets for lung cancer. Our goal is to determine whether C10 type II alveolar epithelial cells are a sensitive model to investigate ERK-dependent transformation and DNMT expression patterns in experimental lung cancer.Findings: Ectopic expression of an extracellular signal regulated kinase (ERK)-green fluorescent protein (ERK1-GFP) induces acquisition of growth in soft agar that is selectively associated with latent effects on the expression of DNA methyl transferases (DNMT1 and 3b), xeroderma pigmentosum complementation group A (XPA), DNA-dependent protein kinase catalytic subunit (DNA-PKcs), increased phosphatase activity and enhanced sensitivity to 5-azacytidine (5-azaC)-mediated toxicity, relative to controls.Conclusions: Ectopic expression of ERK alone is sufficient to promote phenotypic conversion of C10 cells associated with altered DNMT expression patterns and sensitivity to DNMT inhibitor. This model may have applications for predicting sensitivity to DNMT inhibitors.
## Findings
Many lung carcinogens activate the extracellular signal regulated kinase (ERK) [bib_ref] Nicotine, lung and cancer, Grozio [/bib_ref] and in some model systems constitutive overexpression of ERK can induce transformation [bib_ref] Constitutive activation of the MEK/ERK pathway mediates all effects of oncogenic H-ras..., Zhang [/bib_ref] [bib_ref] Constitutive MAP kinase activation in hematopoietic stem cells induces a myeloproliferative disorder, Chung [/bib_ref] [bib_ref] Effects of active MEK1 expression in vivo, Scholl [/bib_ref] [bib_ref] Dual role of MEK/ERK signaling in senescence and transformation of intestinal epithelial..., Boucher [/bib_ref] [bib_ref] Transfection of constitutively active mitogen-activated protein/extracellular signal-regulated kinase kinase confers tumorigenic and..., Welch [/bib_ref]. However, it is unclear whether ERK alone can modulate cell transformation responses in lung type II alveolar epithelial cells. In addition, cell transformation induced by ERK overexpression does not correlate with ERK activity [bib_ref] Mek1 phosphorylation site mutants activate Raf-1 in NIH 3 T3 cells, Alessandrini [/bib_ref] , suggesting an important role for secondary regulatory events. Murine C10 type II alveolar epithelial cells have been used as an in vitro model to investigate molecular determinants of lung cell physiology and pathophysiology [bib_ref] Inhaled particulate matter causes expression of nuclear factor (NF)-kappaB-related genes and oxidant-dependent..., Shukla [/bib_ref] [bib_ref] Hyperoxia stimulates an Nrf2-ARE transcriptional response via ROS-EGFR-PI3K-Akt/ ERK MAP kinase signaling..., Papaiahgari [/bib_ref] [bib_ref] Increased localization and substrate activation of protein kinase C delta in lung..., Lounsbury [/bib_ref] [bib_ref] Different accumulation of activated extracellular signal-regulated kinases (ERK 1/2) and role in..., Buder-Hoffmann [/bib_ref]. C10 type II alveolar epithelial cells are a non-tumorigenic cell line derived from normal BALB/c mouse lung tissue and do not contain native Ras mutations [bib_ref] Mouse lung epithelial cell lines-tools for the study of differentiation and the..., Malkinson [/bib_ref] [bib_ref] Transcriptional regulation of basal cyclooxygenase-2 expression in murine lung tumor-derived cell lines..., Wardlaw [/bib_ref]. Type II features include the presence of lamellar bodies, the biosynthesis of surfactant, proliferation that is contact inhibited and anchorage-dependent growth [bib_ref] Development and characterization of type 2 pneumocyte-related cell lines from normal adult..., Smith [/bib_ref]. Here we ectopically expressed an ERK1-GFP chimera in C10 cells using retroviral technology as previously described [bib_ref] Efficient screening of retroviral cDNA expression libraries, Kitamura [/bib_ref] and asked whether ectopic ERK expression induced phenotypic conversion. Thus, our use of the terms "transformation or phenotypic conversion" are constrained to observable changes in cell behavior linked to carcinogenesis in vitro, such as loss of cell densitydependent growth arrest, anchorage-independent growth and morphological changes associated with an epithelial to mesenchymal transition. ERK1-GFP protein expression was confirmed by Western blot , panel A bottom). The phosphorylation of both ERK1-GFP and endogenous ERKs was increased in response to growth factor treatment (10% FBS; 10 min), relative to respective quiescent controls . ERK1-GFP translocated to the nuclear compartment following growth factor stimulation , panel B), consistent with nuclear translocation of ERK upon activation [bib_ref] Growth factors induce nuclear translocation of MAP kinases (p42mapk and p44mapk) but..., Lenormand [/bib_ref]. ERK is exported from the nucleus by the chromosome region maintenance protein 1 (CRM1) exporter [bib_ref] RhoA influences the nuclear localization of extracellular signal-regulated kinases to modulate p21Waf/Cip1..., Zuckerbraun [/bib_ref] and treatment with a CRM1 inhibitor (20 nM LB; 30 min) resulted in nuclear retention of ERK1-GFP. Therefore, ERK1-GFP displays expected regulatory patterns in C10 cells.
ERK1-GFP transduced cells exhibit a morphological change upon prolonged passaging [fig_ref] Figure 2: Phenotypic conversion of C10 cells induced by ERK1-GFP [/fig_ref] , compare early and late passage cells) which often accompanies cell transformation in vitro [bib_ref] The epithelial-mesenchymal transition (EMT) and colorectal cancer progression, Bates [/bib_ref]. Total cell number was increased by approximately 11 fold in 5 day postconfluent cultures of late passage ERK1-GFP cells, relative to postconfluent vector control or early passage ERK1-GFP cells [fig_ref] Figure 2: Phenotypic conversion of C10 cells induced by ERK1-GFP [/fig_ref] , indicating loss of growth inhibition by cell-cell contact. Late passage ERK1-GFP cells grow in soft agar, while vector controls do not show significant Panel A: Expression of ERK1-GFP chimera was confirmed by Western blot analysis (bottom). Phosphorylation of both ERK1-GFP and endogenous ERKs is increase by growth factor stimulation (FBS), relative to quiescent controls (top). Panel B: Epifluorescent microscopy confirmed ERK1-GFP nuclear translocation following stimulation with FBS, relative to unstimulated ERK1-GFP cells. Blocking nuclear export with leptomycin B (30 min treatment) resulted in retention of ERK1-GFP in the nucleus as expected. Similar results were observed in two independent experiments. Methods for retroviral expression of ERK1-GFP can be found in [bib_ref] Efficient screening of retroviral cDNA expression libraries, Kitamura [/bib_ref]. Late passage cells exhibit a qualitative change in their capacity to grow in soft agar, while vector controls show marginal anchorage-independent growth responses. Methods for the soft agar assay can be found in [bib_ref] AP-1, NF-kappa-B, and ERK activation thresholds for promotion of neoplastic transformation in..., Suzukawa [/bib_ref]. Similar results were observed in two independent experiments. anchorage-independent growth potential [fig_ref] Figure 2: Phenotypic conversion of C10 cells induced by ERK1-GFP [/fig_ref] , defined as previously described [bib_ref] Modulation of JB6 mouse epidermal cell transformation response by the prostaglandin F2alpha..., Weber [/bib_ref]. Collectively, late passage ERK1-GFP cells display multiple phenotypic alterations that suggest they have transformed to a malignant state.
ERK can regulate DNMT expression [bib_ref] Role of the ras-MAPK signaling pathway in the DNA methyltransferase response to..., Deng [/bib_ref] [bib_ref] Procainamide inhibits DNA methyltransferase in a human T cell line, Scheinbart [/bib_ref] which could impact epigenetic programming. Altered epigenetic programming is an attractive candidate in carcinogenesis because alterations in methylation of DNA are heritable and can lead to transcriptional dysregulation linked to neoplastic cellular changes. Upon examination of DNMT expression patterns in our model, we observed a marked increase in DNMT1 and 3b isoforms in late passage ERK1-GFP cells, relative to early passage cells and vector controls . DNMT3a was not detected by Western blot in our experiments (data not shown). DNMT1 was consistently characterized by the appearance of multiple bands immunoreacting with anti-DNMT1 antibody that were absent in early passage ERK1-GFP cells and vector controls. At present we do not know whether these bands represent alternative splice variants, degradation products, post-translational modifications or some combination. Thus, increased DNMT expression is latent in ERK transduced C10 cells , suggesting that ERK is not directly regulating DNMT expression in this model, or that compensatory mechanisms prevent significant increases in DNMT expression patterns in early passage cells.
In previous studies we have used xeroderma pigmentosum complementation group A (XPA) as a loading control for nuclear extracts because the expression of this protein generally showed little change under a variety of experimental conditions. However, we observed a marked increase in the expression of XPA in late passage ERK1-GFP cells, relative to early passage cells and vector controls . We subsequently defined the expression of DNA-dependent protein kinase catalytic subunit (DNA-PKcs) in late passage ERK1-GFP cells as an additional index for DNA damage signaling which was also increased . Lamin a/c levels were not increased under these conditions and served as loading control. The combined results of three independent experiments are illustrated in . The reason for increased expression of DNA repair proteins is unclear. One possible interpretation may relate persistent ERK activation to genomic instability, which is a common feature of human cancers [bib_ref] Genetic instabilities in human cancers, Lengauer [/bib_ref]. Genomic instability encompasses a broad array of chromosomal rearrangements and DNA damage events [bib_ref] Non-targeted and delayed effects of exposure to ionizing radiation: II. Radiation-induced genomic..., Morgan [/bib_ref] that could generate signals leading to the regulation of repair proteins such as XPA and DNA-PKcs. ERK regulates NADPH oxidase activity [bib_ref] Molecular regulation of NADPH oxidase 5 via the MAPK pathway, Pandey [/bib_ref] , which is associated with a significant generation of oxygen free radicals [bib_ref] Expression and modulation of an NADPH oxidase in mammalian astrocytes, Abramov [/bib_ref] and chronic oxidative stress can induce genomic instability [bib_ref] Persistent oxidative stress in chromosomally unstable cells, Limoli [/bib_ref]. Alternatively, DNA-PKcs is hypothesized to play an important role in maintaining genomic stability [bib_ref] MEK/ERK inhibitor U0126 increases the radiosensitivity of rhabdomyosarcoma cells in vitro and..., Marampon [/bib_ref] and the increase in DNA-PKcs may reflect effort to maintain stability in an unstable environment.
DNMTs possess HDAC binding domains [bib_ref] Potential of DNMT and its Epigenetic Regulation for Lung Cancer Therapy, Tang [/bib_ref] and DNMT/HDAC systems are believed to be interdependent [bib_ref] DNMT3B interacts with hSNF2H chromatin remodeling enzyme, HDACs 1 and 2, and..., Geiman [/bib_ref]. Therefore, we surveyed whether HDAC activity was altered under these conditions as an additional index for altered epigenetic programming that may be directly influenced by HDAC binding domains on DNMTs [bib_ref] Potential of DNMT and its Epigenetic Regulation for Lung Cancer Therapy, Tang [/bib_ref] [bib_ref] DNMT3B interacts with hSNF2H chromatin remodeling enzyme, HDACs 1 and 2, and..., Geiman [/bib_ref]. HDAC activity was significantly increased in late passage ERK1-GFP cells, relative to early passage cells Panel A: DNMT1, DNMT3b, XPA, and DNA-PKcs levels were increased in nuclear extracts prepared from late passage ERK1-GFP cells, relative to early passage ERK1-GFP cells and vector controls. DNMT1 consistently exhibited multiple bands immunoreacting with anti-DNMT1 antibody in late passage ERK1-GFP cells. Lamin a/c was used as a loading control. Panel B: Graph illustrates pooled results from three independent experiments for early (white) and late (black) passage ERK1-GFP cells, relative to vector control (referenced as fold change value of 1). Values represent mean ± se (n = 3). *Significantly different from vector control. General methods for Western blot analysis can be found in [bib_ref] Basic fibroblast growth factor regulates persistent ERK oscillations in premalignant but not..., Weber [/bib_ref]. and vector control [fig_ref] Figure 4: HDAC activity in nuclear extracts from vector control, early and late passage... [/fig_ref]. The abrupt increase in HDAC activity in late passage, but not early passage ERK1-GFP cells, is consistent with increased DNMT expression patterns and coupled HDAC regulation [bib_ref] Potential of DNMT and its Epigenetic Regulation for Lung Cancer Therapy, Tang [/bib_ref]. There is precedence for an ERK-DNMT-HDAC linkage in fear conditioning [bib_ref] Epigenetic alterations are critical for fear memory consolidation and synaptic plasticity in..., Monsey [/bib_ref] and we hypothesize that these activities may also be aligned in carcinogenesis. Additional studies are required to dissect the relative amount of HDAC activity that is dependent on DNMTs in late passage cells.
To determine if increased DNMT expression was linked to cell fate regulation, we asked whether vector control, early and late passage ERK1-GFP cells were differentially sensitive to a DNMT inhibitor (5-azaC). Cells were treated with 0.5-50 μM 5-azaC for 7 days and cell viability was determined using a neutral red assay as previously described [bib_ref] PGE2-mediated cytoprotection in renal epithelial cells: evidence for a pharmacologically distinct receptor, Weber [/bib_ref]. Cell viability was reduced in a dose-dependent manner by 5-azaC in late passage ERK1-GFP cells, but not in vector controls [fig_ref] Figure 5 5: -azaC toxicity in ERK1-GFP transduced cells [/fig_ref]. Early passage ERK1-GFP cells displayed a small reduction in cell viability at the highest concentrations of 5-azaC (25-50 μM) employed. DNMT's are under investigation as therapeutic targets for lung cancer [bib_ref] Potential of DNMT and its Epigenetic Regulation for Lung Cancer Therapy, Tang [/bib_ref]. Biomarkers that can predict when DNMT inhibitors may exhibit high efficacy could significantly aid in this effort. Because the C10 model developed here is sensitive to DNMT inhibitors, it may provide insight into molecular features that may serve as biomarkers, to the extent that such features are conserved in human cancers.
We consistently observed that late passage ERK1-GFP cells exhibited a marked reduction in phospho-ERK (P-ERK) levels, but not total ERK protein levels, relative to vector controls and early passage ERK1-GFP cells [fig_ref] Figure 6: Evidence for increased phosphatase activity in late passage ERK1-GFP cells [/fig_ref]. Treatment of serum starved cells (0.1% FBS) with 10% FBS for 5 min resulted in increased P-ERK levels in vector controls and early passage ERK1-GFP cells, which is the expected response to serum stimulation. Lack of P-ERK levels in late passage cells could result from either a general lack of signal transduction leading to ERK activation or an increase in phosphatase activity. We treated late passage ERK1-GFP cells with 1 mM sodium orthovanadate (Na 3 VO 4 ) to determine whether a broad spectrum phosphatase inhibitor could restore P-ERK levels. P-ERK levels were restored within minutes of Na 3 VO 4 treatment [fig_ref] Figure 6: Evidence for increased phosphatase activity in late passage ERK1-GFP cells [/fig_ref] , suggesting that the decrease in P-ERK levels associated with late passage ERK1-GFP cells was due to increased phosphatase activity. Vector control (square), early passage ERK1-GFP (circle) and late passage ERK1-GFP (triangle) were maintained in media supplemented with 5-azaC at the indicated concentrations for 7 days at which time cell viability was measured using a neutral red assay as described [bib_ref] PGE2-mediated cytoprotection in renal epithelial cells: evidence for a pharmacologically distinct receptor, Weber [/bib_ref]. Values represent the mean ± se (n = 3). *Significantly different from vector control. Similar results were observed in three independent experiments.
# Conclusions
Ectopic expression of ERK alone is sufficient to induce phenotypic conversion of C10 cells and this model may provide insight into the underlying molecular determinants of this response. The window between early and late passage ERK-transduced variants that encompasses phenotypic conversion (approximately 15 passages) is a reasonable time frame to enable interrogative studies to define molecular determinants. Our expectation is that causal molecular processes will precede the appearance of the transformed phenotype and will be observed in early passage cells. At present, we have characterized changes in DNMT, DNA damage recognition and repair proteins and phosphatase activities that are selectively altered in late, but not early passage cells, suggesting they are secondary to transformation. Additional studies, perhaps with a more global screening approach, may provide insight into those molecular processes perturbed by ERK overexpression in early passage cells. Alternatively, because DNMTs are under investigation as therapeutic targets for lung cancer, the C10 model may provide insight into the molecular processes that confer sensitivity to DNMT inhibitors and regulate their aberrant expression. Evidence for increased phosphatase activity in late passage ERK1-GFP cells. Panel A: Western blot analysis of P-ERK levels in serum starved (0.1% FBS) vs 10% FBS stimulated cells. Treatment of cells with 10% FBS for 5 min results in increased P-ERK levels in vector control and early passage ERK1-GFP cells, as expected, while late passage ERK1-GFP cells display diminished P-ERK, but not total ERK levels. Panel B: Treatment of late passage ERK1-GFP cells with a broad spectrum phosphatase inhibitor (1 mM Na 3 VO 4 ) for 5 or 60 min resulted in reappearance of P-ERK levels, suggesting that the decrease in P-ERK levels is due to increased phosphatase activity. Similar results were observed in two independent experiments
[fig] Figure 2: Phenotypic conversion of C10 cells induced by ERK1-GFP. Panel A: Late passage ERK1-GFP cells exhibit rounded morphology, relative to early passage ERK1-GFP cells and vector controls. Panel B: The proliferation of late passage ERK1-GFP cells (L) is not inhibited by cellcell contact at confluence, while proliferation of early passage ERK1-GFP cells (E) and vector controls (V) are contact inhibited. *Significantly different from vector control. Values represent the mean ± se (n = 3). Similar results were observed in three independent experiments. Panel C: [/fig]
[fig] Figure 4: HDAC activity in nuclear extracts from vector control, early and late passage ERK1-GFP cells. Increased relative fluorescence units (RFU) is an index for increased HDAC activity, which was measured using the Fluor de Lys -Green kit (Enzo Life Sciences, Plymouth Meeting, PA) according to manufacturer's directions. HDAC activity was increased in late passage ERK1-GFP cells, relative to early passage ERK1-GFP cells and vector control. * Significantly different from vector control. Values represent the mean ± se (n = 3). Similar results were observed in three independent experiments. [/fig]
[fig] Figure 5 5: -azaC toxicity in ERK1-GFP transduced cells. [/fig]
[fig] Figure 6: Evidence for increased phosphatase activity in late passage ERK1-GFP cells. Panel A: Western blot analysis of P-ERK levels in serum starved (0.1% FBS) vs 10% FBS stimulated cells. Treatment of cells with 10% FBS for 5 min results in increased P-ERK levels in vector control and early passage ERK1-GFP cells, as expected, while late passage ERK1-GFP cells display diminished P-ERK, but not total ERK levels. Panel B: Treatment of late passage ERK1-GFP cells with a broad spectrum phosphatase inhibitor (1 mM Na 3 VO 4 ) for 5 or 60 min resulted in reappearance of P-ERK levels, suggesting that the decrease in P-ERK levels is due to increased phosphatase activity. Similar results were observed in two independent experiments [/fig]
|
10.3389/fneur.2021.630301
|
CCBY
|
7905208
|
33643207
|
s2orc_pubmed_articles
|
Safety and Efficacy of Tirofiban for Acute Ischemic Stroke Patients With Large Artery Atherosclerosis Stroke Etiology Undergoing Endovascular Therapy
Objective: To investigate the safety and efficacy of tirofiban in acute ischemic stroke (AIS) patients with large artery atherosclerosis (LAA) stroke etiology receiving endovascular therapy (EVT).Methods: In this multi-center prospective study, patients who were considered to have an indication received a low dose intra-arterial bolus (0.25-1 mg) of tirofiban. The safety and efficacy outcomes at 90-day follow-ups included symptomatic intracranial hemorrhage (sICH), recanalization rate, functional outcome, and mortality.Results: Among the 649 AIS patients with LAA, those in the tirofiban group (n = 244) showed higher systolic blood pressure (BP) and NIHSS score on admission, puncture-to-recanalization time, lower frequency of intravenous thrombolysis and intra-arterial thrombolysis, higher frequency of antiplatelet, heparinization, mechanical stent retrieval, aspiration, balloon angioplasty, and more retrieval times compared with those in the non-tirofiban group (n = 405) (all P < 0.05). Tirofiban was found to be associated with superior clinical outcomes in anterior circulation stroke and major stroke patients [adjusted odds ratio (OR) = 2.163, 95% confidence interval (CI) = 1.130-4.140, P = 0.02 and adjusted OR = 2.361, 95% CI = 1.326-4.202, P = 0.004, respectively] and a lower risk of mortality at 90-day follow-ups (adjusted OR = 0.159, 95% CI = 0.042-0.599, P = 0.007 and adjusted OR = 0.252, 95% CI = 0.103-0.621, P = 0.003, respectively). There was no significant difference in sICH between the two groups.Conclusions: Tirofiban in AIS patients with LAA undergoing EVT is safe and may benefit the functional outcomes in anterior circulation and major stroke patients and showed a trend for reduced mortality.
# Introduction
The non-peptide platelet GP IIb/IIIa receptor inhibitor, tirofiban, has been increasingly applied as a rescue therapy, by either intraarterial or intravenous route during endovascular treatment (EVT). Tirofiban can selectively and efficiently block the final pathway of platelet aggregation and subsequent thrombus formation in atherosclerotic lesions. Recent metaanalysis studies have reported that the safety profile and efficacy of tirofiban may make it a potential choice for treatment in patients with acute ischemic stroke (AIS). It has also been reported to be more feasible and effective in AIS patients with large artery atherosclerosis (LAA) compared to those with cardioembolic stroke etiology. However, the treatment results were inconsistentand a study reported an increased risk of symptomatic intracranial hemorrhage (sICH) and a poor outcome in patients treated with tirofiban during mechanical thrombectomy. Moreover, to the best of our knowledge, there are no reports on which stratified population may benefit the most from rescue tirofiban therapy.
To address this issue, we explored the safety and efficacy of rescue tirofiban treatment in AIS patients with LAA stroke etiology and evaluated which stratified population gained the most benefit from rescue tirofiban in a large multi-center cohort study in China.
# Methods
## Patient selection and data collection
This multi-center nationwide prospective study of an Acute Ischemic Stroke Cooperation group in the Endovascular Treatment (ANGEL) registry recruited 917 Chinese patients with AIS to evaluate EVT delivery and improve EVT. The study protocol was similar to our previous research. The present study was approved by the ethics committee at each participating center, and informed consent was obtained from all participants prior to commencing the study.
Patient's baseline data, such as age, gender, systolic blood pressure (SBP), the National Institutes of Health Stroke Scale (NIHSS) score, Alberta Stroke Program Early CT Score (ASPECTS), time intervals [onset-to-door (OTD), door-topuncture (DTP), puncture-to-recanalization (PTR), onset-topuncture (OTP), and onset-to-recanalization (OTR)], were recorded within 24 h after admission. Vascular risk factors included atrial fibrillation, diabetes mellitus, history of previous stroke, hypertension, smoking, and drinking. The data related to the peri-procedural anti-thrombotic and anticoagulation therapies, such as administration of antiplatelets, bridging intravenous thrombolysis (IVT), and heparin, were recorded as along with the procedural techniques.
AIS patients undergoing EVT were divided into tirofiban and non-tirofiban groups. All EVT procedures were performed by neurointerventionalists with extensive experience in neurovascular intervention.
## Dose and indication of rescue tirofiban
Rescue tirofiban with low-dose intra-arterial bolus (0.25-1 mg) is suggested when there are the following indications: (1) severe residual stenosis or instant re-occlusion requiring emergency stenting or balloon angioplasty; (2) stent retrieval times > 3 passes for presumed vascular endothelial injury or instant reocclusion; and (3) severe degree of in situ atherosclerosis with a tendency to early re-occlusion. Low dose rescue tirofiban followed by intravenous continuous infusion (0.1 µg/kg/min) for 12-24 h is suggested when there is no indication of post-operative intracranial hemorrhage following a CT examination.
## Clinical efficacy and safety outcomes
SICH, which was defined by the European Cooperative Acute Stroke Study III (ECASS-III) trial as evidence of hemorrhage on a CT or MRI, was considered a primary safety endpoint. The primary efficacy endpoints were the functional independence (mRS 0-2) and mortality at 90 day follow-ups. A successful recanalization, which was defined as modified Thrombolysis in Cerebral Infarction (mTICI), is considered the secondary efficacy endpoint in the present study.
# Statistical analysis
The baseline characteristics of patients were compared between the tirofiban and non-tirofiban groups. The χ 2 test or Kruskal-Wallis test was used to compare the baseline characteristics and safety and efficacy outcomes at 90 days between the tirofiban and non-tirofiban groups. The logistic regression model was used to evaluate the odds ratios (OR)/hazard ratio (HR) with a 95% confidence interval (CI) of safety and efficacy endpoints (sICH), mTICI grade 2b-3, complete reperfusion (mTICI 3), functional independence (mRS 0-2), and mortality with or without use of tirofiban. The multivariate models were adjusted for some potential confounders with P < 0.05 in univariate analysis, which included SBP, NIHSS, atrial fibrillation, smoking history, anterior and posterior circulation, occlusion of the M2 or M3 segment of the middle cerebral artery (MCA) M2/3 segment, vertebral artery (VA), basilar artery (BA), posterior cerebral artery (PCA), antiplatelet, bridging IVT during EVT, general anesthesia, MT stent retrieval and aspiration, balloon angioplasty and intraarterial thrombolysis, and retrieval times. A P-value < 0.05 was considered statistically significant. All statistical analyses were conducted using SPSS 20.0 software (IBM, Armonk, NY, USA).
# Results
## Baseline characteristics of patients
Two of the 917 patients were excluded from the data analysis due to missing baseline data. Subsequently, 266 patients with embolic stroke etiology were also excluded. Finally, 649 patients with large vessel atherosclerosis who underwent EVT with or without receiving tirofiban were analyzed., the median age of patients was 63 (55-71) years, 464 (71.5%) patients were male, and 244 (37.6%) had received tirofiban. In the tirofiban group, SBP and NIHSS on admission were relatively higher and smoking history was more frequent, while atrial fibrillation was less obvious than those in the non-tirofiban group (all P < 0.05). Rescue tirofiban was used more in the posterior circulation (particularly VA, BA, and PCA), but less in the anterior circulation group (particularly MCI M23 segment). In the tirofiban group, general anesthesia, stent retrieval, MT aspiration, and balloon angioplasty were more frequently performed as compared to the non-tirofiban group (45.5 vs. 26.2%, P < 0.001), (76.2 vs. 59.8%), P < 0.001), (10.2 vs. 2.7%), P < 0.001), and (17.6 vs. 10.4%), P = 0.008)), respectively. Moreover, anti-platelet therapy was administered more in the tirofiban group (38.1 vs. 13.6%), P < 0.001). Meanwhile, the proportions of bridging IVT and intra-arterial thrombolysis were less in the tirofiban group compared to the non-tirofiban group (20.1 vs. 30.9%, P = 0.003) and (11.5 vs. 30.6%, P < 0.001).
## As shown in
There was no significant difference in age, gender, other vascular risk factors (diabetes mellitus, previous stroke, hypertension, and drinking), other occlusion sites (ICA, MCA M1, or ACA), time workflow (OTD, DTP, PTR, OTP, and OTR), heparinization during EVT, and stent angioplasty between the tirofiban and non-tirofiban groups (all P > 0.05).
## Safety and efficacy outcomes
The safety and efficacy outcomes are shown in
## .
Overall, 27 (4.2%) patients developed sICH within 24 h post-EVT, and no significant difference was noted in the sICH incidence between the tirofiban group and the non-tirofiban group (P > 0.05). Tirofiban was not correlated with the incidence of sICH (adjusted HR 0.998; 95% CI 0.021-46.825; P = 0.999) even after adjusting for some potential confounders. Similar results were demonstrated when the population was stratified into anterior/posterior circulation and minor (NIHSS 0-5)/major (NIHSS > 5) stroke (all P > 0.05).
At 90 day follow-ups, excellent outcome (mRS0-1) and functional independence (mRS0-2) could be achieved in 295 (45.5%) and 182 (44.9%) patients, respectively. However, 87 (13.4%) patients had died (mRS 6) by the three-month followup . A slightly higher rate of superior clinical outcomes and a lower risk of mortality were found in patients who received tirofiban. Moreover, tirofiban was associated with excellent outcomes and functional independence after adjusting for several potential confounders (adjusted OR, 1.819; 95%CI, 1.064-3.110; P = 0.029 and OR, 1.849; 95%CI, 1.065-3.212; P = 0.029, respectively). Further analysis showed a strong association of tirofiban with favorable functional outcomes in the anterior circulation (adjusted OR 2.163; 95%CI, 1.130-4.140; P = 0.02) and NIHSS > 5 (adjusted OR 2.361; 95% CI, 1.326-4.202; P = 0.004). Furthermore, tirofiban was significantly correlated with a lower risk of mortality (adjusted HR 0.2; 95% CI, 0.079-0.507; P = 0.001) even after adjusting for potential factors. This strong association was significantly demonstrated in the anterior circulation (adjusted OR 0.159; 95% CI, 0.042-0.599; P = 0.007), posterior circulation (adjusted OR 0.001; 95% CI, 0.000-0.188; P = 0.009), and NIHSS > 5 (adjusted OR 0.252; 95% CI, 0.103-0.621; P = 0.003).
# Discussion
The present study showed that rescue tirofiban offers a safe outcome for the risk of sICH in AIS patients with LAS who received EVT. In the LAS population, rescue tirofiban showed superior clinical outcomes in patients with an AC stroke and NIHSS > 5. Rescue tirofiban may lower the mortality risk in these stratified patients as well as those with a PC stroke.
The clinical benefit of tirofiban remains controversial in AIS patients who received recanalization therapy. Previous studies have reported the feasibility and effectiveness of tirofiban, and suggested tirofiban use in failed mechanical thrombectomy. In contrast, another study reported no clinical benefit and also highlighted safety concerns of tirofiban. These conflicting results might be attributed to the small sample size, various treatment strategies, and uncontrolled study design in these preliminary studies. Thus, special caution is needed when interpreting these results. However, the majority of these studies shared similar indications that tirofiban is more beneficial for LAA patients. Moreover, a recent meta-analysis indicated that tirofiban use is safe and appears to be effective in treating AIS patients. Since we compared tirofiban and nontirofiban use only in patients with LAA, the clinical benefit of rescue tirofiban was more significant in this study. Interestingly, our results demonstrated that patients with an AC stroke and a major stroke received more clinical benefit either through functional outcome or mortality risk from rescue tirofiban, while no significant clinical benefit was found in patients with PC stroke and minor stroke. Despite no functional benefit in those with a PC stroke, rescue tirofiban was advantageous in lowering the mortality rate in the study.
This study was in agreement with previous findings that showed rescue tirofiban did not affect recanalization. However, the clinical benefit of rescue tirofiban in LAA patients is that it prevents subsequent ischemic events and the mechanisms have been well-described. Tirofiban has anti-inflammatory effects and may stabilize inflamed stenotic lesions and maintain blood flow, which is helpful in preventing ischemic events caused by inflammation and platelet aggregation. In addition, this rescue therapy might benefit cases with stent retrieval times > 3, which are prone to vascular endothelial injury or instant re-occlusion. Moreover, it is recommended to use tirofiban in patients with no history of anti-platelet, as it has more significant dose-dependent blockade effects on platelet aggregation and thrombosis. Tirofiban is a highly selective platelet antagonist that can block fibrinogen, and its mechanical effect is usually maintained for 20 min after administration.
The current study showed that not all LAA patients may receive clinical benefit from rescue tirofiban, including those with a PC stroke or a minor stroke. Accordingly, we assumed that the dosage of tirofiban may account for the clinical benefits in different stratified populations. Based on previously reported medication regimes of tirofiban in AIS patients undergoing EVT, we adopted an intra-arterial administration of < 1 mg and an intravenous infusion of 0.1 µg/kg/min for 12-24 h in patients refractory to recanalization. The present study demonstrated that this low-dose rescue tirofiban was effective in cases of AC stroke and major stroke. Nevertheless, since tirofiban was administered within the dosage range in our study, it might have different treatment effects in AIS patients under certain circumstances and may confound the therapeutic effects at a particular dose. Thus, further study with doseescalation methods is needed for verification. In addition, the present study demonstrated that the use of tirofiban had more favorable outcome in anterior circulation strokes than in posterior circulation strokes. The possible postulated mechanisms attributed to this result may be due to the pathologic mechanisms of stroke and the fact that treatment modalities were significantly different in anterior and posterior circulation, which affect their clinical outcome. Posterior circulation stroke patients often presented severe preoperative symptoms and required longer emergency procedures, leading to poor neurological function recovery. In addition, the goal for rescue tirofiban is mainly to maintain blood flow and prevent acute occlusion. However, this issue remains uncertain and needs further large prospective trials or randomized controlled trials for verification.
This study had several limitations. First, an uneven proportion between the tirofiban and non-tirofiban groups may cause a bias. Second, the EVT and several other rescue therapies were undertaken at individual discretion, which might affect the treatment results. However, the indications triggering the use of rescue tirofiban were in accordance with standard clinical practice. Third, as the patients enrolled in this study were from China, the results cannot be generalized to the global population. Nonetheless, a strength of the current study was the relatively large sample size compared to previous studies. However, further randomized controlled trials are needed for verification.
# Conclusions
Low-dose rescue tirofiban is safe in AIS patients with LAA, may provide clinical benefit to those with AC stroke or major stroke, and had a tendency to reduce the risk of mortality. However, large cohort or randomized controlled trials with dose-escalation are urgently needed for further verification.
# Data availability statement
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.
# Ethics statement
The studies involving human participants were reviewed and approved by ethics committee of Beijing Tiantan Hospital. The patients/participants provided their written informed consent to participate in this study.
# Author contributions
ZM, YlW, and YjW conceived and led the project. DM, FG and NM supervised and performed quality control for the study. AW performed statistical analysis, XH and R acquired the data and co-wrote the manuscript with input from all co-authors. All authors contributed to the article and approved the submitted version.
|
10.1186/s12889-015-2409-7
|
CCBY
|
4615338
|
26482904
|
s2orc_pubmed_articles
|
Association of hypoadiponectemia with smokeless/dipping tobacco use in young men
Background: Low levels of adiponectin, an adipocytokine with anti-diabetic, antiatherogenic and cardioprotective properties, is associated with increased risk of coronary disease in young men. Previous studies have demonstrated that smokeless tobacco is linked with a reduction of plasma adiponectin levels. However, the influence of smokeless tobacco (dipping tobacco) on plasma adiponectin levels still remains unknown. This study was conducted to assess the plasma adiponectin levels in young men who were using dipping tobacco. Methods: This was a community based study, which consisted of 186 young lean healthy males aged 20 to 35 years. Among these, 96 men were dipping tobacco users (BMI = 23.07 ± 2.68) and 90 were non-dipping tobacco users (BMI = 23.67 ± 1.46). Serum adiponectin levels were assessed by Enzyme Linked ImmunoSorbent Assay (ELISA). Results: A statistically significant difference in the mean adiponectin level between tobacco dipper and non-dipper groups was observed (p = 0.0001). A significant difference between the two groups was also observed in baseline parameters including triglyceride and random blood sugar levels (p < 0.05). However, no significant difference was observed between the two groups in other clinical parameters.Conclusions: Findings of this study suggest that dipping tobacco use was significantly associated with low level of adiponetin in community dwelling young males. This emphasizes the importance of developing community intervention to reduce the use of dipping tobacco, which will reduce the tobacco associated disease burden in the community and will improve public health.
# Background
The adipocyte derived plasma protein adiponectin, also called ARCP30, ADIpoQ, apM1 or GBP28 is a 247 amino acid peptide that accounts for about 0.05 % of total serum proteins [bib_ref] AdipoQ is a novel adipose-specific gene dysregulated in obesity, Hu [/bib_ref] [bib_ref] Plasma adiponectin levels and risk of myocardial infarction in men, Pischon [/bib_ref] [bib_ref] A novel serum protein similar to C1q, produced exclusively in adipocytes, Scherer [/bib_ref]. The normal concentration of adiponectin in plasma ranges from 5 to 30 μg/ml in males, whereas a higher concentration is found in females. Adiponectin is an important molecule shown to be involved in obesity, metabolic syndrome, cardiovascular disease, lipid metabolism, and hypertension [bib_ref] Plasma concentrations of a novel, adipose-specific protein, adiponectin, in type 2 diabetic..., Hotta [/bib_ref] [bib_ref] Adiponectin and lipid metabolism in skeletal muscles, Lee [/bib_ref] [bib_ref] Adiponectin and metabolic syndrome, Matsuzawa [/bib_ref] [bib_ref] Hypoadiponectinemia in obesity and type 2 diabetes: close association with insulin resistance..., Weyer [/bib_ref]. Lower levels of circulating adiponectin have recently been shown to be associated with a number of conditions in humans. Prominent among them are coronary artery disease, myocardial infarction, atherosclerosis, re-stenosis after percutaneous coronary intervention, type 2 diabetes mellitus as well as hypertension [bib_ref] Plasma adiponectin levels and risk of myocardial infarction in men, Pischon [/bib_ref] [bib_ref] Plasma concentrations of a novel, adipose-specific protein, adiponectin, in type 2 diabetic..., Hotta [/bib_ref] [bib_ref] Hypoadiponectinemia is an independent risk factor for hypertension, Iwashima [/bib_ref] [bib_ref] Association of hypoadiponectinemia with coronary artery disease in men, Kumada [/bib_ref] [bib_ref] Implications of plasma concentrations of adiponectin in patients with coronary artery disease, Nakamura [/bib_ref] [bib_ref] Adiponectin, an adipocyte-derived plasma protein, inhibits endothelial NF-κB signaling through a cAMP-dependent..., Ouchi [/bib_ref] [bib_ref] Significance of changes in plasma adiponectin concentration after the implantation of stents..., Sako [/bib_ref]. Recent studies suggest that adiponectin has anti-atherogenic and anti-inflammatory properties. It is involved in the lowering of many anti-inflammatory cytokines such C-reactive protein, tumor necrosis factor-α and interleukin-6 [bib_ref] Association between adiponectin and mediators of inflammation in obese women, Engeli [/bib_ref] [bib_ref] Inflammatory markers, adiponectin, and risk of type 2 diabetes in the Pima..., Krakoff [/bib_ref] [bib_ref] Inflammation in atherosclerosis, Libby [/bib_ref] [bib_ref] Decreased plasma adiponectin concentrations in women with low-grade C-reactive protein elevation, Matsubara [/bib_ref] [bib_ref] Reciprocal association of C-reactive protein with adiponectin in blood stream and adipose..., Ouchi [/bib_ref].
Tobacco use is a significant contributing risk factor for non-communicable diseases leading to approximately six million deaths worldwide every year. The increasing trend of tobacco use in developing countries is alarming, where 80 % of deaths are expected to occur in next few decades due to toabcco use. . In south Asian countries, the prevalence of tobacco use was estimated to be 72.3, 60.0, 47.3, 34.7, 34.1, 33.6 and 31.6 % in Indonesia, Bangladesh, Maldives, Cambodia, India and Pakistan, respectively [bib_ref] Smoking and smokeless tobacco use in nine South and Southeast Asian countries:..., Sreeramareddy [/bib_ref].
Similar to that of smoking tobacco, smokeless tobacco use was also found equally harmful in a variety of diseases. reported that coronary artery disease (CAD) was more prevalent in patients with hypoadiponectemia (less than 4 μg/mL) irrespective of other known risk factors like diabetes mellitus, dyslipidemia, hypertension, smoking habit, and BMI in male subjects [bib_ref] Association of hypoadiponectinemia with coronary artery disease in men, Kumada [/bib_ref]. Huhtasaari et al., reported smokeless tobacco as a possible risk factor for myocardial infarction in a population based study performed on middle-aged men [bib_ref] Smokeless tobacco as a possible risk factor for myocardial infarction: a populationbased..., Huhtasaari [/bib_ref]. Smokeless tobacco was also found to be associated with hypertension in rural Indians [bib_ref] Plasma adiponectin levels and risk of myocardial infarction in men, Pischon [/bib_ref]. A significant increase in both systolic and diastolic blood pressure was observed in young male population using smokeless tobacco compared to non-users [bib_ref] Association of exclusive smokeless tobacco consumption with hypertension in an adult male..., Pandey [/bib_ref]. Wolk et al., reported that nicotine of spit tobacco induced catecholamine release from the adrenal medulla which in turn causes increase in heart rate and blood pressure [bib_ref] Hemodynamic and autonomic effects of smokeless tobacco in healthy young men, Wolk [/bib_ref]. Smokeless tobacco (Naswar) use and its association with cancer is well documented. The data from Pakistani hospitals specialized in treating cancer patients showed that 9.9 % of all cancer patients were suffering from oropharyngeal cancer. In addition, with 23.6 % being smokers and 37.4 % being smokeless tobacco users [bib_ref] Smokeless tobacco use in Pakistan and its association with oropharyngeal cancer, Bile [/bib_ref].
Smokeless tobacco has not gained the attention of the scientific research community in comparison to smoking, though its use is common all over the world. Majority of dipping tobacco users believe that it is less harmful and that its use can go unnoticed. Dipping tobacco and other forms of smokeless tobacco are widely used in some populations. Grinded tobacco leaves in moist powdered form make dipping tobacco, which is kept between the lower lip and gum and discarded after about 15-30 min. In the local language (Pashto), it is called Naswar (dipping tobacco/moist snuff ). It is available locally as loose meshed ground tobacco or as sachets in the developed world.
Peshawar is the capital city of Khyber Pakhtunkhwa, where the vast majority of ethnic Pashtoons reside where 43.79 % of males and 4.4 % of females use dipping tobacco [bib_ref] Prevalence and pattern of tobacco use in rural area of Peshawar, Shah [/bib_ref]. Several studies have suggested the association of smoking with low levels of adiponectin in plasma, but data regarding dipping tobacco is lacking. Therefore, a case control study was conducted in community dwelling young males to determine if there was any correlation of dipping tobacco usage and levels of plasma adiponectin.
# Methods
## Study design
This was a case control study. The study population consisted of 186 young, lean and apparently healthy male adults. The study participants were divided in to two groups: dipping tobacco users (n = 96) and non-dipping tobacco users (n = 90). Inclusion criteria were: healthy individuals between 20 and 35 years of age and body mass index (BMI) in the range of 20-25 kg/m 2 . The cases (dipping tobacco users) were those who had history of taking dipping tobacco on a daily basis for a minimum of 5 years regularly with an average of one packet (approximately 50-60 g/day). The controls (non-dippers) were selected from similar matched socio-demographic background and were young healthy subjects with no history of using dipping tobacco or tobacco use by other means. In addition, these control participants were matched to case participants in terms of age and BMI. All the cases and controls selected were young, healthy, lean subjects to remove the possibility of obesity induced hypoadiponectemia or risk of aging associated derangements in physical and biochemical parameters. A structured questionnaire was used in each face to face interview to obtain the relevant information. The study participants were evaluated by medical history for use of alcohol, by physical examination for obesity (BMI >25 kg/m 2 ), and routine clinical laboratory investigations to exclude individuals with type 2 diabetes, anemia, liver diseases, chronic renal failure and cardiovascular diseases. Past dipping tobacco users, and subjects with BMI more than 25 were excluded from the study. Study participant on drug(s) which alter hormone levels were also excluded from the study. The prevalence of dipping tobacco among women was very low around 4.4 % and the access to women was difficult due to traditional and cultural restrictions in this area, therefore, women were also excluded from the study [bib_ref] Prevalence and pattern of tobacco use in rural area of Peshawar, Shah [/bib_ref] [bib_ref] Smoking status is associated with serum high molecular adiponectin levels in community-dwelling..., Kawamoto [/bib_ref]. The study was approved by the institution ethical committee of Khyber Medical University, Pakistan. After the nature of the study was explained in detail, informed consent was obtained from all participants.
## Anthropometric parameters
Anthropometric data (weight & height) of the study subjects were collected and BMI was calculated as kg/m 2 . Resting blood pressure was measured using a standard mercury sphygmomanometer. Two blood pressure readings were taken with the participant resting for 10 min, and the mean of the two readings was calculated and used for data analysis.
## Laboratory measurement
Non-fasting venous blood samples (10 ml divided into two 5 ml tubes) were collected in sterile syringes using aseptic technique. The samples were immediately transported to the laboratory (within 2 h) in ice cooled containers for further examination. Serum was separated after centrifugation at 12,000 × g and stored at −20°C for subsequent analysis. Total cholesterol (TC), triglyceride (TG), high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), random blood sugar, serum creratinin were measured in both cases and control groups with a standard protocol using semi-automated chemistry analyzer COBAS C111 (Roche) and kits from the same company.
## Adiponectin assay
Serum low molecular weight adiponectin concentration was determined by ELISA method using the kits supplied by Glory Science Co., Ltd. Research (Del Rio, TX 78840, USA). The kit uses double-antibody sandwich ELISA to assess the level of human adiponectin in serum samples. The wells were pre-coated with human adiponectin monoclonal antibodies. The secondary antibodies consisted of adiponectin antibodies labelled with biotin, and combined with streptavidine-HRP to form an immune complex. After washing completely, 3,3′,5,5′-tetramethylbenzidine (TMB) liquid substrate was added. The TMB substrate changed a blue color when HRP enzyme catalyzed the reaction. The reaction was terminated by the addition of a stop solution and the color change was measured at a wavelength of 450 nm. All the steps of the assay were done according to the manufacturer's instructions. The results were analyzed by three independent researchers, blinded to the clinico-demographic data of study participants to remove subjective bias. The concentration of adiponectin in the sample was determined by comparing the optical densities of the samples from the standard calibration curve with negative controls. Adiponectin samples were run in duplicate and mean values were calculated for each sample.
# Statistical analysis
Means or proportions of clinical characteristics were analyzed for each group in the study. All data was expressed as means ± standard deviation. Differences between groups were examined for statistical significance using the student's t-test, p ≤ 0.05 was considered statistically significant. Unpaired t-test (independent sample t test) was used to examine the differences in adiponectin levels between the two groups. All analyses were performed using SPSS software, version 19 "SPSS (IBM, Armonk, NY, USA)."
# Results
The study participants were primarily of Pashtoon ancestry, living in the north western areas of Pakistan with the majority using dipping tobacco (about 64 %). The subjects in this study comprised of a total of 186 young, lean, male adults, who were apparently healthy. Among these subjects 96 were dipping tobacco users (28.84 ± 4.77 years old), 90 were non-dipping tobacco users (28.41 ± 4.99 years old), with age range from 20 to 35 years. [fig_ref] Table 1: Baseline characteristics of young adults in dipping tobacco users and nondippersAbbreviations [/fig_ref] summarizes the baseline characteristics of each participant. Significant changes in some clinical parameters were observed between cases and control groups while in others, no significant variations were observed. BMI was 23.07 ± 2.68 kg/m 2 in the dipping tobacco users while it was 23.67 ± 1.46 kg/m 2 in the nonusers. Systolic and diastolic blood pressure in tobacco dippers and non-dippers:119.8 ± 9.67 and 118.61 ± 8.51; 80.36 ± 9.60 and 79.5 ± 8.74 mmHg respectively showing no significant difference. Serum creatinin was 0.78 ± 0.09 in dippers as compared to non-dippers 0.78 ± 0.08 mg/dl (p = 0.94). The values of triglycerides were 127.39 ± 33.43 in dippers and 113.81 ± 28.57 mg/dl in non-dippers (p = 0.003) showing highly significant difference between the two study groups. Similarly, the values of HDL-C were 48.28 ± 3.74 and 47.22 ± 6.48 mg/dl in dippers and non-dippers respectively (p = 0.171). The level of random blood sugar in dippers was 101.8 ± 15.46 and 96.54 ± 10.06 mg/dl in non-dippers (p = 0.013). The level shows the scatter plots of the difference in the level of random blood sugar and triglycerides between the dipping tobacco users and non-dippers. It is clear that the level of random blood sugar significantly changed (p = 0.013) in dippers as compared to non-dippers . Similarly, the values of triglycerides were also significantly changed (p = 0.003) in both groups . [fig_ref] Figure 2: Serum adiponectin concentrations in dipping tobacco users and non-users [/fig_ref] shows adiponectin levels in dippers and non-dippers. The most promising finding in our study was the low level of adiponectin in dippers compared to those of non-dippers. The mean serum adiponectin level was 3.23 ± 2.07 in dippers, while it was 12.94 ± 10.50 in non-dippers (p = 0.0001, independent sample t test). Thus a statistically significant lower level of serum adiponectin was observed in dippers compared to non-dippers. No association of adiponectin was observed with other clinical parameters, e.g. hypertension, HDL-C and raised triglyceride in both tobacco dippers and non-dippers.
# Discussion
This study was designed to assess the effect of dipping tobacco use on serum adiponectin concentration. Our results showed a significantly lower serum adiponectin level in healthy young males who were using dipping tobacco for a minimum of 5 years regularly with an average of one packet (approximately 50-60 g) daily. In contrast to the published literature we did not find any association of lower adiponectin with hypertension (both systolic and diastolic), serum creatinin, total cholesterol, LDL-C and HDL-C levels [bib_ref] Smoking status is associated with serum high molecular adiponectin levels in community-dwelling..., Kawamoto [/bib_ref] [bib_ref] Serum adiponectin is associated with smoking status in healthy Korean men, Sull [/bib_ref]. The discrepancies in these parameters might probably be due to the inclusion of young lean adults in the current study. The present study is therefore unique because we considered lean young individuals who were in the highly productive and highly resistant ages of their lives compared to previous studies which included subjects with age above 40 years [bib_ref] Smoking status is associated with serum high molecular adiponectin levels in community-dwelling..., Kawamoto [/bib_ref] [bib_ref] Serum adiponectin is associated with smoking status in healthy Korean men, Sull [/bib_ref] [bib_ref] The relationship among smoking, plasma adiponectin, leptin, inflammatory markers and insulin resistance, Shousha [/bib_ref]. The clinical parameters including blood pressure, serum creatinin, total cholesterol, LDL-C and HDL-C were not significantly different in dipping tobacco users relative to those of age matched non-dipping control group [fig_ref] Table 1: Baseline characteristics of young adults in dipping tobacco users and nondippersAbbreviations [/fig_ref]. The level of adiponectin is reported to increase in nephritic and renal failure [bib_ref] Review article: Adiponectin: its role in kidney disease, Shen [/bib_ref]. Therefore, we measured serum creatinin in order to exclude any renal dysfunction in our study groups. Astonishingly, we did not detect any significant change in the levels of HDL-C, LDL-C, serum creatinine and total cholesterol in the present study.
The levels of triglyceride were significantly different between tobacco dippers and non-dippers [fig_ref] Table 1: Baseline characteristics of young adults in dipping tobacco users and nondippersAbbreviations [/fig_ref] shown in bold face letter). Our results support the previous literature where adiponectin was found to be inversely associated with triglyceride levels [bib_ref] Young men with high-normal blood pressure have lower serum adiponectin, smaller LDL..., Kazumi [/bib_ref] [bib_ref] Decreased plasma adiponectin concentrations in women with dyslipidemia, Matsubara [/bib_ref].
It has been reported previously that cigarette smoking is associated with low plasma adiponectin levels [bib_ref] Hypoadiponectinemia is an independent risk factor for hypertension, Iwashima [/bib_ref] [bib_ref] Adipocyte derived plasma protein, adiponectin, is associated with smoking status in patients..., Miyazaki [/bib_ref] , however, to the best of our knowledge, no study has explored the relationship between smokeless tobacco use and serum adiponectin level in young adult males. Therefore, we hypothesized that similar to smoking tobacco; smokeless tobacco might be associated with low level of adiponectin in community-dwelling young adults. Our results support the previous literature on tobacco induced hypo-adiponectemia. Kawamoto et al., reported significantly lower adiponectin level in community dwelling Japanese men aged 20 to 89 years using smoking tobacco [bib_ref] Smoking status is associated with serum high molecular adiponectin levels in community-dwelling..., Kawamoto [/bib_ref]. An interesting study showed a significant difference Scatter plots of serum random blood sugar (a) and triglyceride (b) (individual's data for each subject) in smokeless tobacco dippers (n = 96) and non-dippers (n = 90). The horizontal lines represent the means between the tobacco dippers and non-dippers. P values represent the significance of the effect of random blood sugar (a) triglyceride (b) levels between the two study groups in adiponectin level in current smokers, those who have stopped smoking and those who had never smoked [bib_ref] Hypoadiponectinemia is an independent risk factor for hypertension, Iwashima [/bib_ref]. Importantly, some studies have shown an increase in plasma adiponectin level following cessation of smoking [bib_ref] Smoking cessation is associated with increased plasma adiponectin levels in men, Otsuka [/bib_ref] [bib_ref] Changes of plasma adiponectin levels after smoking cessation, Won [/bib_ref]. In a study by Otsuka et al., cessation of smoking was associated with increased plasma adiponectin level in 47 non smokers and 25 current smokers with a 6 month follow-up period [bib_ref] Smoking cessation is associated with increased plasma adiponectin levels in men, Otsuka [/bib_ref]. reported that low level of adiponectin (≤4.5 μg/mL) is associated with adverse cardiovascular health [bib_ref] Adiponectin: an emerging cardiovascular risk factor. The reference study, Barrios [/bib_ref] , though the mean age and BMI of our study participants was less than than the participants of that study.
These findings have led us to explore new possibilities of the association between dipping tobacco and plasma adiponectin levels. Adiponectin is a major contributor in insulin sensitivity and in decreasing the risk of diabetes and coronary heart diseases. Dipping tobacco is commonly used in ethnic Pashtoon areas in Pakistan and Afghanistan. Our study showed an inverse relationship of dipping tobacco with adiponectin level. Clearly, this association would lead to adverse cardiovascular health. Thus this novel finding could help the general public and health authorities understand the association of dipping tobacco with adverse cardiovascular health. In addition, this study will also help designing future studies on exploring the underlying molecular mechanisms responsible for hypoadiponectemia.
Our study has certain limitations. In this study we were unable to compare our results with individuals using smoking tobacco. The inclusion of smoking tobacco group would have made the results of this study more informative. This would have provided us with evidence on adiponectin level in three groups: dipping tobacco users; smoking tobacco users; and both dipping and smoking tobacco users. However, due to limited funding we were not able to accomplish this part of the study but we plan to explore this aspect in a future study. In addition, the smokeless tobacco was not biochemically verified. Adiponectin may lower C-reactive proteins as well as a variety of other inflammatory biomarkers as previously reported [bib_ref] Association between adiponectin and mediators of inflammation in obese women, Engeli [/bib_ref] [bib_ref] Inflammatory markers, adiponectin, and risk of type 2 diabetes in the Pima..., Krakoff [/bib_ref] [bib_ref] Inflammation in atherosclerosis, Libby [/bib_ref] [bib_ref] Decreased plasma adiponectin concentrations in women with low-grade C-reactive protein elevation, Matsubara [/bib_ref] [bib_ref] Reciprocal association of C-reactive protein with adiponectin in blood stream and adipose..., Ouchi [/bib_ref]. However, we were not able to measure the levels of these inflammatory biomarkers in our study. Another limitation of our study was the non-fasting venous blood sampling due to restricted access to individuals and to convince them for providing fasting blood samples. We were also unable to assess the dose dependent response of dipping tobacco in changing the biochemical parameters. Thus to explore the aforementioned associations future studies are recommended.
# Conclusions
This study showed that the serum adiponectin levels were significantly lower in the dipping tobacco users as compared to non-users in adult lean males. These findings support a critical association of hypoadiponectemia with dipping tobacco use in community-dwelling youth. Consequently, the results of this study provide evidence that will increase the awareness of the general public regarding the adverse effects associated with dipping tobacco use. This will also hopefully help the public health authorities in deciding what measures to take to reduce the use of dipping tobacco and reduce the associated health burden. Moreover, as the level of adiponectin is significantly lowered in dipping tobacco users than non-dippers, it might aid in the early diagnosis of clinical situations (e.g. cardiovascular health).
[fig] Figure 2: Serum adiponectin concentrations in dipping tobacco users and non-users. Individual's data for each subject in smokeless tobacco dippers (n = 96) and non-dippers (n = 90). The horizontal lines represent the means between the tobacco dippers and non-dippers. P values represent the significance of the effect of adiponectin concentration between dipping tobacco users and non-users [/fig]
[table] Table 1: Baseline characteristics of young adults in dipping tobacco users and nondippersAbbreviations: BMI Body Mass Index, TG Triglycerides, HDL High Density Lipoproteins, LDL Low Density Lipoproteins. For these variables, data are presented as mean ± SD and student's t-test was applied. The letters in bold shows the significant difference between the two study groups [/table]
|
10.1186/s12934-018-0858-2
|
CCBY
|
5776760
|
29357936
|
s2orc_pubmed_articles
|
The effect of metabolic stress on genome stability of a synthetic biology chassis Escherichia coli K12 strain
Background: Synthetic organism-based biotechnologies are increasingly being proposed for environmental applications, such as in situ sensing. Typically, the novel function of these organisms is delivered by compiling genetic fragments in the genome of a chassis organism. To behave predictably, these chassis are designed with reduced genomes that minimize biological complexity. However, in these proposed applications it is expected that even when contained within a device, organisms will be exposed to fluctuating, often stressful, conditions and it is not clear whether their genomes will retain stability.Results:Here we employed a chemostat design which enabled us to maintained two strains of E. coli K12 under sustained starvation stress: first the reduced genome synthetic biology chassis MDS42 and then, the control parent strain MG1655. We estimated mutation rates and utilised them as indicators of an increase in genome instability. We show that within 24 h the spontaneous mutation rate had increased similarly in both strains, destabilizing the genomes. High rates were maintained for the duration of the experiment. Growth rates of a cohort of randomly sampled mutants from both strains were utilized as a proxy for emerging phenotypic, and by association genetic variation. Mutant growth rates were consistently less than rates in non-mutants, an indicator of reduced fitness and the presence of mildly deleterious mutations in both the strains. In addition, the effect of these mutations on the populations as a whole varied by strain.Conclusions:Overall, this study shows that genome reductions in the MDS42 did not stabilize the chassis under metabolic stress. Over time, this could compromise the effectiveness of synthetic organisms built on chassis in environmental applications.
are increasingly being forced to confront as the discipline matures [bib_ref] Scaling up synthetic biology: do not forget the chassis, Danchin [/bib_ref]. In highly controlled laboratory conditions it has been shown that reduced genome chassis can help with this transition [bib_ref] Unbottling the genes, Editorial [/bib_ref] [bib_ref] Considerations for the design and construction of a synthetic platform cell for..., Foley [/bib_ref] in two respects. First, a reduced genome eases the metabolic burden of both replicating and expressing a large number of genes thus allowing more energy to be directed towards endogenous gene expression [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref] [bib_ref] Low-mutation-rate, reduced-genome Escherichia coli: an improved host for faithful maintenance of engineered..., Csorgo [/bib_ref] [bib_ref] Pseudomonas 2.0: genetic upgrading of P. putida KT2440 as an enhanced host..., Martinez-Garcia [/bib_ref] [bib_ref] The energetics of genome complexity, Lane [/bib_ref] [bib_ref] Reduced evolvability of Escherichia coli MDS42, an IS-less cellular chassis for molecular..., Umenhoffer [/bib_ref] [bib_ref] Genome reduction boosts heterologous gene expression in Pseudomonas putida, Lieder [/bib_ref]. Second, by deleting mobile genetic elements, the genome becomes more stable and reproducible, with a reduced likelihood of both mobile-element driven mutagenesis and unwanted byproducts [bib_ref] Low-mutation-rate, reduced-genome Escherichia coli: an improved host for faithful maintenance of engineered..., Csorgo [/bib_ref] [bib_ref] Reduced evolvability of Escherichia coli MDS42, an IS-less cellular chassis for molecular..., Umenhoffer [/bib_ref]. The benefits of reduced genomes were conceived of with highly controllable white-biotechnologies in mind [bib_ref] Unbottling the genes, Editorial [/bib_ref] [bib_ref] Considerations for the design and construction of a synthetic platform cell for..., Foley [/bib_ref]. However, it remains a moot point as to whether these benefits hold in biotechnologies that routinely experience external perturbations. So, for example, in bioenergy, environmental sensing or resource recovery from waste products it would be naive to expect these engineered organisms to live stress-free. This begs the question; does reducing a genome remain beneficial in the face of, inevitable, environmental stress?
Whilst a myriad of different environmental stressors exist, here we take one by way of an example: starvation. Engineered microorganisms deployed in the environment, in for example, a sensing device would need to 'live off the land". Almost all environmental bacterial, almost all of the time, live in oligotrophic conditions; typical carbon concentration of 1-5 mg/L as compared to more-than 2 g/L in typical lab media [bib_ref] Bioavailability of energy and its relationship to growth and starvation survival in..., Morita [/bib_ref] [bib_ref] Survival strategies of bacteria in the natural environment, Roszak [/bib_ref]. These long periods of starvation equate to chronic metabolic stress [bib_ref] Survival strategies of bacteria in the natural environment, Roszak [/bib_ref]. In lab studies and in natural microbial populations, starvation causes a decrease in growth rate and an increase in transient mutation rates termed "stressinduced mutagenesis" [bib_ref] The influence of cellular physiology on the initiation of mutational pathways in..., Notley-Mcrobb [/bib_ref] [bib_ref] Stress-induced mutagenesis in bacteria, Bjedov [/bib_ref]. This phenomenon can promote the emergence of hypermutable subpopulations, which can become temporarily advantageous and contribute to persistence of the entire population [bib_ref] Evolutionary significance of stressinduced mutagenesis in bacteria, Tenaillon [/bib_ref]. However, for a population of engineered organisms with reduced genomes that are intended to remain stable and perform reliably in a biotechnology, the rapid emergence of mutants introduces an element of unwanted unpredictability, where the resultant loss or gain of function could ultimately compromise the technology.
In this study, the E. coli reduced genome strain MDS42 was used as an example of a chassis that was engineered for stability. It was created by systematically deleting non-essential genes, mobile DNA elements and cryptic prophages from the MG1655 strain, while retaining optimal growth properties, and has been described and extensively characterized elsewhere [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref] [bib_ref] Low-mutation-rate, reduced-genome Escherichia coli: an improved host for faithful maintenance of engineered..., Csorgo [/bib_ref] [bib_ref] Reduced evolvability of Escherichia coli MDS42, an IS-less cellular chassis for molecular..., Umenhoffer [/bib_ref] [bib_ref] Indispensability of horizontally transferred genes and its impact on bacterial genome streamlining, Karcagi [/bib_ref]. Its genome is 14.30% smaller than MG1655 and has lost, inter alia, all insertion sequence (IS) elements [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. A fluctuation analysis showed that the spontaneous background mutation rate was 2.4-fold lower in the MDS42 strain than in the parent MG1655 strain, evidence in support of this engineered stability [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. We questioned whether targeted genome reductions that promote stability would offer an advantage over a nonreduced genome in a stressed environment. Concordant with 'streamlining theory' (for example: [bib_ref] Implications of streamlining theory for microbial ecology, Giovannoni [/bib_ref] , a reduced genome increases the efficiency of a microorganism, enabling it to persist in a depleted environment. This ability to persist despite limited resources could translate into reduced instances of stress-induced mutagenesis. In contrast, genome reductions might eliminate compensatory pathways, increasing the overall effect of mutations on the organism's genome. To test our hypothesis, we grew and maintained populations of both the reducedgenome (MDS42) and their parent strain (MG1655) under sustained metabolic stress while we quantified and compared mutation rates and their effect on fitness and stability.
Triplicate chemostats were utilized, which enabled us to simulate sustained starvation conditions via glucoselimitation [bib_ref] Bacterial physiology, regulation and mutational adaptation in a chemostat environment, Ferenci [/bib_ref] [bib_ref] rpoS mutations and loss of general stress resistance in Escherichia coli populations..., Notley-Mcrobb [/bib_ref] [bib_ref] Hungry bacteria-definition and properties of a nutritional state, Ferenci [/bib_ref] over a period of 21 days (504 h or 73 generations). Importantly, it is thought that steady state growth, only achievable in chemostats, is closer to the state in which microbes grow in their natural environment [bib_ref] Continuous-culture chemostat systems and flowcells as methods to investigate microbial interactions, Drake [/bib_ref]. Therefore we grew and maintained steady populations of E. coli K12 multiple deletion series (MDS) strain MDS42 [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. Then, in identical conditions, we grew and maintained steady populations of the parent K12 MG1655 strain, which served as a control. We quantified spontaneous mutation accumulation in both sets of populations periodically. We show that within 24 h both strains accumulated mutations at a similar rate. Furthermore, these mutation rates were dynamic in strains from both populations and increased overall, showing that this engineered stability was insufficient when strains underwent metabolic stress. Maximum growth rates of mutants isolated from both cultures were consistently less than those of the non-mutants suggesting that the mutations were mildly deleterious in both strains, however their effect in each population was different. High rates of emerging genetic variation introduce a level of unpredictability that could ultimately compromise the function of engineered organisms and adversely affect the biotechnology they are embedded in.
# Results
Growth rates (denoted as 'η' in the current study), and hence generation times, of both the MG1655 and the MDS42 strains have been previously compared and have been shown to be unequal [bib_ref] Indispensability of horizontally transferred genes and its impact on bacterial genome streamlining, Karcagi [/bib_ref]. We confirmed these findings via batch growth in the minimal glucosedepleted medium that was utilized in all experiments in our study. Doubling times were 38 (1.1 h −1 ) min for the MG1655 strain and 53 (0.79 h −1 ) min for the MDS42 strain (Additional file 1: . Unequal generation times have a bearing on mutation rates thus utilizing a chemostat environment enabled us to fix the growth rate (η) and generation time [bib_ref] Bacterial physiology, regulation and mutational adaptation in a chemostat environment, Ferenci [/bib_ref] [bib_ref] Experiments with the chemostat on spontaneous mutations of bacteria, Novick [/bib_ref] in both strains, allowing for a direct comparison. In addition, controlling dilution rates and the carbon source (substrate) concentrations can allow for bacterial growth under a constant selective pressure, which in this experiment was hunger and starvation [bib_ref] Bacterial physiology, regulation and mutational adaptation in a chemostat environment, Ferenci [/bib_ref] [bib_ref] rpoS mutations and loss of general stress resistance in Escherichia coli populations..., Notley-Mcrobb [/bib_ref] [bib_ref] Hungry bacteria-definition and properties of a nutritional state, Ferenci [/bib_ref]. Triplicate populations (N = 3) of both strains of E. coli were grown in continuous culture in the chemostats that were fed a substrate of minimal glucosedepleted media. A dilution rate of 0.1 h −1 ensured that the substrate in the chemostat was depleted and yielded a generation time of 6.9 h. Previous work with derivatives of the E. coli K12 strain that have not been engineered to remain stable reported that they acquire mutations readily in glucose-limited chemostats [bib_ref] The influence of cellular physiology on the initiation of mutational pathways in..., Notley-Mcrobb [/bib_ref]. Therefore, it was expected that mutations would accumulate in the triplicate MG1655 chemostats, and hence they were utilized here as controls. Populations of both strains reached a steady state within 20 h, maintaining an average density of 1.6·10 8 colony forming units (CFU) per mL for the MG1655 strain and 1.1·10 8 CFU*mL −1 for the MDS42 strain, with no significant difference between the two (P = 0.128; .
Spontaneously occurring mutations are extremely rare and are typically observed when they produce a detectable phenotype, usually observed via a selective screen. Fortuitously in E. coli, spontaneous mutations produce a detectable phenotype change when they occur at either the RNA polymerase subunit (rpoB) locus or the cycloserine A (cycA) locus. Specifically, mutations at the rpoB locus render an organism resistant to the antibiotic Rifampicin (rif R ) [bib_ref] Use of the rpoB gene to determine the specificity of base substitution..., Garibyan [/bib_ref] and mutations at the cycA locus confer resistance to d-cycloserine (cyc R ) [bib_ref] Characterization of cycA mutants of Escherichia coli. An assay for measuring in..., Feher [/bib_ref]. Therefore antibiotic-containing selector plates can be used to screen for these specific mutants. Although mutations at different loci and/or silent mutations are excluded here, this is a commonly used system with the cycA locus especially being used to estimate background mutation rates in E. coli (see Ref. [bib_ref] Characterization of cycA mutants of Escherichia coli. An assay for measuring in..., Feher [/bib_ref] for example). Indeed the cycA locus was used for the first estimate of background mutation rates in the MDS strains [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. Here we exploited this system by sampling the continuous culture populations periodically and plating these samples on selector plates to quantify spontaneous mutant accumulation over time. Samples were also plated on non-selector plates for a colony count. Rif R and cyc R mutants accumulated in both the MG1655 and MDS42 strains, with marked variation between runs . Mutation rates (denoted as 'μ' in the current study) were estimated using a simple linear mutation accumulation model, that reports the number of mutations per cell per generation and is suitable for chemostats [bib_ref] Methods for determining spontaneous mutation rates, Foster [/bib_ref] ; . Mutation rate estimates (μ) in the MDS42 strain were 1.4-fold higher than in MG1655 by 24 h, increasing to approximately 20-fold higher by 168 h (7 days) . These were then surpassed by the MG1655 strain 504 h into the experiment, where the average estimated mutation rate reached 1.03·10 −3 mutations cell −1 generation −1 , which is approximately 30-fold higher than in the MDS42 strain ; .
The spontaneous mutation rate in both strains had previously been estimated at the cycA locus via a fluctuation test [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. Working under the assumption that these were an estimate of a background mutation rate, we compared the rates acquired in our chemostats at the cycA locus only , with these previous estimates, and found an approximately eight-fold increase in just 24 h in the MDS42 strain and a two-fold increase in the MG1655 strain (Additional file 1: . This fold-difference increased further in both strains in comparison with this background rate estimate.
In a chemostat, growth rate is an important measure of fitness. The mean growth rate is held constant by controlling the dilution rate. This means that if any portion of the population gains an advantage it is at the expense of the remainder. So, faster growing mutants will ultimately dominate and slower growers will ultimately wash out of the population. Thus, growth rate was used as an index to phenotypic variation. We reasoned that in a clonal population, a wider distribution of growth rates was indicative of increased phenotypic, and hence, by association, emerging genetic variation. Previously published work [bib_ref] Clonal adaptive radiation in a constant environment, Maharjan [/bib_ref] [bib_ref] The multiplicity of divergence mechanisms in a single evolving population, Maharjan [/bib_ref] has shown that phenotype screening is an effective way of assessing the emergence of genetic diversity in chemostat populations. Therefore, populations of mutants from two chemostat runs-one for each strain-where the mutant fraction increased monotonically over time were chosen for further investigation. Forty-eight cyc R mutant colonies were chosen at random from each time point with the exception of the MDS42 mutants sampled at 24 h; a total of 20 colonies was available here. Each mutant was grown individually in batch culture using the same minimal glucose-depleted medium as used in the chemostat experiments. Twentyfour non-mutants from both strains were also grown in the same conditions for comparison. These non-mutant populations were streaked from fresh glycerol stocks and were thus assumed to have no mutations. To check that tradeoffs between growth and yield did not confound our assertion that growth rate is a reasonable measure of fitness, we estimated biomass yield in the batch experiments and showed that the average yield of mutant strains was lower (MG1655) or the same as (MDS42) the non-mutant strains (Additional file 1: [fig_ref] Figure 3: A comparison of mean cyc R mutant growth rates of batch monocultures... [/fig_ref]. The observed growth rate distributions were different between strains . Of the MDS42 mutants, 83-100% sampled through time were able to grow in isolation in batch. The overall mean mutant growth rate for the MDS42 mutants was 1.2-fold slower than non-mutant growth under the same conditions in batch [fig_ref] Figure 3: A comparison of mean cyc R mutant growth rates of batch monocultures... [/fig_ref]. Seven days into the experiment, we observed a wide range of growth rates, which expanded further by day 14 ; a small percentage of mutant growth rates also approached the mean rate of non-mutant growth [fig_ref] Figure 3: A comparison of mean cyc R mutant growth rates of batch monocultures... [/fig_ref]. Following this, the distribution of growths for mutants sampled on day 21 had a smaller range and a lower mean .
In comparison, 4-75% of the MG1655 mutants grew in isolation. For the MG1655 mutants sampled we observed a similar broad range of growths after day 7 . Here too mean mutant growth rates were much slower than the non-mutant populations [fig_ref] Figure 3: A comparison of mean cyc R mutant growth rates of batch monocultures... [/fig_ref]. The rate of decline in the mean and variance of growth rate of mutants in batch was far greater in the MG1655 strain. By day 14 only three of the mutants grew in pure batch culture, which dropped to only two by day 21 . This was unexpected because the mutant fraction in samples from the chemostat for this particular experimental run was very high (5.03% by day 21). We questioned whether this observation was the result of individuals in the population with cross-feeding polymorphisms. Cross-feeders are individuals that only grow well as part of a group and have been known to emerge in glucose-limited chemostats [bib_ref] Clonal adaptive radiation in a constant environment, Maharjan [/bib_ref]. If this were the case then reconstituting the mutant population would change the mean growth rate. We tested this with one of the mutant populations and found that when we mixed the mutants, we observed a 4.8-fold increase in growth rate compared to when each was grown individually .
The observation of a relative increase in mutant growth rate, when grouped, led us to investigate whether this was a general phenomenon that occurred in all the chemostats. A chemostat environment ensures that cell density is controlled by the dilution rate (steady state). Although cell density remained fairly stable over time , there were still some fluctuations observed that were quantified as deviations from the mean. An increase in the growth rate in a portion of the population would be expected to yield a temporary rise in density until the system equilibrates back to the original density and mean growth rate. We tested whether this fluctuation from 'steady state' was related to mutant accumulation (mutant CFU/mL) using Kendell's tau (τ). We found a strong trend towards a correlation (τ = 1.0000; P = 0.0833) between mutant CFU and the variation in cell density of A description of mean steady-state growth, mutant accumulation, and mutation rate for all chemostat runs: n = 3 per strain. The vertical scale is logarithmic (Log 10 ) and the horizontal scale is in hours, for all three sections of this figure. a The relationship between mean cell density measured as colony forming units per milliliter (CFU/mL) and time measured in hours. Cell density for all runs reached steady state in approximately 20 h and stayed that way till the end of the sampling period. The diamond shape and dotted line represents the mean cell density for the MG1655 strain. The circle and the solid line represent mean cell density for the MDS42 strain. Horizontal error bars represent the standard deviation above and below the mean. b The relationship between the mean of the total number of mutant colony forming units (cyc R , rif R ) for each chemostat run and time (in hours). The diagonally stripped boxes represent the means for the runs containing the MG1655 strain and the solid grey boxes represent the means for runs containing the MDS42 strain. Standard deviations are not shown, as some were negative values, which cannot be plotted on graphs that utilise a logarithmic scale. Instead the horizontal error bars represent the distance between each mean and the maximum and minimum number of mutant colonies quantified at each time point. c The mean mutation rate (μ) measured as mutations per cell per generation, that was estimated at four time points [bib_ref] Indispensability of horizontally transferred genes and its impact on bacterial genome streamlining, Karcagi [/bib_ref] , 504 h) during the experiment. The horizontal axis is scaled to generations, where the generation rate was 0.14 generations per hour for a growth rate (η) of 0.1 hr −1 . The grey squares represent the MG1655 strains and the open circles represent the MDS42 strains. The horizontal error bars represent the standard deviation above and below the mean mutation rate the MG1655 population, but not the MDS42 population (τ = 0.3333, P = 0.7500).
# Discussion
A major motivation for adopting minimal genome chassis organisms in applications of synthetic biology centers on minimising mutations and genetic diversity and thus enhancing stability and reproducibility. These are desirable properties in biotechnologies and there is promising lab-scale evidence [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref] [bib_ref] Pseudomonas 2.0: genetic upgrading of P. putida KT2440 as an enhanced host..., Martinez-Garcia [/bib_ref] [bib_ref] Reduced evolvability of Escherichia coli MDS42, an IS-less cellular chassis for molecular..., Umenhoffer [/bib_ref] [bib_ref] Genome reduction boosts heterologous gene expression in Pseudomonas putida, Lieder [/bib_ref] that minimal chassis can deliver them for white-biotechnologies, where the environment can be tightly controlled [bib_ref] Unbottling the genes, Editorial [/bib_ref] [bib_ref] Considerations for the design and construction of a synthetic platform cell for..., Foley [/bib_ref]. However, for many proposed applications of synthetic biology, such as in environmental sensors, organisms will be exposed to environmental stress, such as metabolic stress. Evidence from both lab and natural bacterial populations show that stresses increase mutation rates [bib_ref] Stress-induced mutagenesis in bacteria, Bjedov [/bib_ref] [bib_ref] Evolutionary significance of stressinduced mutagenesis in bacteria, Tenaillon [/bib_ref]. An increase in stress-induced mutagenesis in an engineered organism would have potentially detrimental effects on the organism's stability and persistence.
Our results show that even in as little as 24 h of metabolic stress, the MDS42 strain accumulated mutations just as quickly as MG1655, a strain that was not engineered for genetic stability. Despite marked variation within the runs, the estimated mutation rates rose in both strains suggesting that minimising a genome does not offer the benefits of stability during prolonged metabolic stress. Moreover, the prevalence of slow growth rates for mutants from both strains at all time points were indicative of a drop in fitness. The average biomass yield for both MDS42 and MG1655 mutants were either similar to or lower than yields for non-mutants (Additional file 1: [fig_ref] Figure 3: A comparison of mean cyc R mutant growth rates of batch monocultures... [/fig_ref] suggesting mutants were unlikely to be adaptive. For the MG1655 mutants we also showed that whilst mixed cultures of mutants grew faster than monocultures their cumulative growth rate did not exceed rates observed in non-mutants. Therefore, taken together the data suggest that the accumulation of mutants is more likely attributable to a continued supply of slowgrowing new mutants rather than a few fast growers. Slow growing mutants are more likely to be washed out of the chemostat and therefore our samples are likely to underestimate mutation accumulation. So, the true rate of mutation in our stressed populations is expected to be even higher. Indeed, to demonstrate just how conservative our estimate of mutation rate is we considered the probability that the average mutant is washed out of the chemostat. Suppose that mutants randomly appear and then grow at a rate η m . Letting N(t) be the number of individuals of a particular mutant, t hours after they first appeared. Then assuming N(t) is continuous, where Q is the flowrate into and out of the chemostat and V is the volume and hence, Q/V is the dilution rate, which fixes the mean growth rate. So, replacing Q/V with η the growth rate of the non-mutant majority and solving we get where N 0 = N(t = 0). Given that in our case N 0 = 1 if we assume that births and loss occur randomly in the period t we can approximate the probability, P, that the mutant population has left by P = 1 − N(t) and so, Therefore, we can estimate the time taken for the probability of washout of the mutant subpopulation to be P as
[formula] (1) dN (t) dt = η m N (t) − Q V N (t),(2)N (t) = N 0 e (η m −η)t ,(3)1 − P = e (η m −η)t . (4) t = ln(1 − P) (η m − η) [/formula]
## Table 1 estimates of mutation rates (μ) in mutations/cell/ generation for the mg1655 and mds42 strains
Mutation rates for each strain (italics) were estimated separately at each locus, and then as a total. The time at which each sample was quantified is indicated in hours (italics) at the top of each column, with the number of generations, a scaled measure of time, indicated in the brackets below. (See figure on next page.) The distribution of individual growth rates, ascertained via batch culture, for the isolated cyc R mutants sampled from intervals starting at 24 h (day 1), and then seven, 14, and 21 days respectively. Forty-eight mutants were isolated from each time point from each strain except for the 24-h interval for MDS42, where only 20 mutants were available. Each mutant was grown individually in batch culture using the same growth medium as used in the chemostat cultures. The percentage of mutants that had a growth rate that was greater than zero is indicated in the top-right corner of each panel
In our case the dilution rate and hence growth rate (η) is 0.1 h −1 . We assume that the mutant, with lower growth rate, does not become abundant enough to affect the overall population growth rate and hence η remains constant. The average mutant in the MDS42 strain grows half as fast as a non-mutant in the chemostat and therefore, η m = 0.05 h −1 . So with probability, P = 0.99, the mutants will have washed out in t = 92.1 h. Thus, when mutants grow half as quickly as the general population, then we can be 99% confident that those that we see in a sample appeared in the population within the past 92 h (Additional file 1: for the full distribution). So in our case when we see 6.52 × 10 7 mutants (the average of the three MDS42 chemostat runs at 21 days) in the population, a mutant has appeared on average every 0.005 s.
It is generally thought that new mutations are usually neutral or deleterious and beneficial mutations are very rare [bib_ref] Rates of spontaneous mutation, Drake [/bib_ref]. Moreover if the mutation has occurred in a gene that only makes a small contribution to a particular phenotype, the mutation might appear 'silent' , with little or no discernible effect over the (non-mutation driven) major transcriptional changes that make large contributions to the overall phenotype under investigation. Indeed, metabolic stress and fluctuating reactor conditions elicit transcriptional changes in chemostat populations of E coli in a matter of seconds, leading to a new 'steady state' that is presumably reached by all cells within these populations [bib_ref] Engineering E. coli for large-scale production-strategies considering ATP expenses and transcriptional responses, Loffler [/bib_ref] [bib_ref] Transcriptional response of Escherichia coli to ammonia and glucose fluctuations, Simen [/bib_ref]. It is reasonable to assume that this has occurred in both the MDS42 and MG1655 chemostat populations in the present study. However, our screen enabled us to select individual mutants from these populations, and their growth rate phenotypes varied, suggesting that we were able to observe additional changes to this phenotype that deviates over and above the new steady state. Although our growth rate distributions suggested a high turnover of phenotypes, brought on by an elevated mutation rate, the fitness effect of mutations, and whether they could lead to adaptation were not clear. For the MDS42 mutants the first 14 days saw a wide distribution of slow growth rates, an indicator that mutations were likely to be deleterious, but insufficiently so to see them washed out of the chemostats quickly. Beyond 14 days a smaller range of growth rates was observed, which could be indicative of either a decrease in the accumulation of new mutants, or the possibility that the sustained high mutation rates led to the acquisition of increasingly deleterious phenotypes. This is plausible given that IS elements have been deleted from the MDS42 strains. Previous reports have suggested that the systematic deletion of IS elements hinders the ability of these multiple deletion strains to evolve in that acquired mutations, even with a detrimental effect on growth, are not fatal to the cell [bib_ref] Reduced evolvability of Escherichia coli MDS42, an IS-less cellular chassis for molecular..., Umenhoffer [/bib_ref] [bib_ref] Indispensability of horizontally transferred genes and its impact on bacterial genome streamlining, Karcagi [/bib_ref]. Therefore, even in instances where high mutation rates are sustained, new mutations of varying effects continue to accumulate rapidly in MDS42 populations, which over time could make the population more susceptible to genetic drift or purifying selection events.
In contrast, the range and mean of the distribution of mutant growth rates in the MG1655 strain appear to fall after just 7 days in batch monocultures. In addition, by day 14, 94% of mutants would not grow in monocultures and those that did, grew very slowly, an indication of deleterious mutations. Yet surprisingly, mutant accumulation in the chemostats continued to rise over the 21 days. Furthermore, in mixed batch cultures, mutants appeared to thrive, with a growth rate approximately four times faster than when grown individually. So it appears Analyses of the cross feeding (CF) phenotype that was detected amongst the MG1655 cyc R mutants. Mean growth rates of 48 cyc R mutants that were grown in batch monocultures (dotted) and as a group (grey) that for the strain with no genome reductions, MG1655, the mutants that survive in the chemostat are no longer exploiting the same niche as the non-mutants; they do not grow in the original media. Their ability to thrive as a cohort suggests that in these strains, acquired mutations have delivered cross-feeding phenotypes. Cross-feeders metabolise by-products from other individuals in the population [bib_ref] Experimental evidence for sympatric ecological diversification due to frequency-dependent competition in Escherichia..., Friesen [/bib_ref] [bib_ref] Microbial evolution in a simple unstructured environment: genetic differentiation in Escherichia coli, Rosenzweig [/bib_ref]. Therefore, in a chemostat, a limited but renewed source of glucose could establish a hierarchy whereby a percentage of the population utilises this glucose, and produces byproducts that could, in turn, feed other members of the population, allowing cross feeders to adapt. Multiple byproducts could open up a multitude of niches, leading to the adaptation of several putative beneficial mutations. This could explain the weak, but present, correlation between mutant accumulation and deviation from steady state in the MG1655 chemostats, which suggests accumulating mutations were causing an increase in cell density. This correlation was absent in the MDS42 reactors.
Overall, our study showed that stress increased mutation rates in both reduced and non-reduced E. coli strains. The MDS42 strain has been engineered for stability and was intended for a closed and controlled biotechnological application [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. Although the current study shows that the elevated mutation rate was the result of metabolic stress that is typical of an open environment, it stands to reason that closed systems would behave similarly when under any stress. Previously documented evidence from lab evolution studies using E. coli show that similar to any other stress, glucose-limitation in a chemostat elicits the general stress response affecting genes such as rpoS, a master regulator of this stress response [bib_ref] rpoS mutations and loss of general stress resistance in Escherichia coli populations..., Notley-Mcrobb [/bib_ref]. This stress-response can also down-regulate the DNA mismatch-repair machinery via mutations in mutS and mutY which ultimately leads to an increase in background mutation rates [bib_ref] Stress-induced mutagenesis in bacteria, Bjedov [/bib_ref] [bib_ref] Bacterial physiology, regulation and mutational adaptation in a chemostat environment, Ferenci [/bib_ref]. Although not specifically assayed in the current study, these mutations are very likely to have occurred in our populations too. Moreover, chromosome replication times have been observed to change in stressed chemostat populations in association with altered expression of genes involved in DNA replication, repair, and recombination [bib_ref] Environmental stress speeds up DNA replication in Pseudomonas putida in chemostat cultivations, Lieder [/bib_ref]. Here too, these genes are triggered by and control a multitude of processes involving DNA integrity and hence could lead to an increase in mutation rates if mis-expressed.
Given that stress can lead to mutations via so many different genetic pathways, it is reasonable to assume that a stress of any type, even in a closed system could lead to an elevated mutation rate. This study also found that the adaptive effect of mutations was difficult to predict. We observed a different effect in populations from each strain, neither of which would be ideal for a biotechnology. For example, the cross-feeding mutants that emerged in the MG1655 populations could produce unwanted byproducts. Moreover, if embedded in an environmental biotechnology, the adaptations could lead to the engineered microorganisms becoming bio contaminants. In the MDS42 strain, the deletions of IS elements appeared to initially 'dampen' the effect of mutations a potential advantage for a biotechnology, provided that the process was completed in a few days. For longer-term applications, or for proposed environmental biotechnologies, further chassis modifications would be required.
# Methods
## Bacterial strains
Strains of bacteria used in this study are both derivatives of E. coli K12. The MG1655 (F-lambda-ilvG-rfb-50, rph-1) substrain is the E. coli reference strain (GI: 556503834) and was purchased from the DSMZ culture collection (Braunsweig, Germany). The multiple deletion series (MDS) MDS42 substrain (GI: 471332236) was created by deleting IS elements, cryptic prophages, and degenerate genes from MG1655 and has been described in detail in Posfai et al. [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. It was purchased from Scarab Genomics (Wisconsin, USA).
## Chemostats and culture conditions
Strains were streaked onto fresh LB-agar plates from glycerol stocks and growth overnight at 37 °C. A single colony was then inoculated into a 100 mL culture of freshly prepared minimal M9 mediasupplemented with 10 mM glucose and 0.2% casamino acids and grown till it reached an optical density of 0.1. Eight millilitres of this culture was used to inoculate a 500 mL computer controlled-chemostat (modified miniBio ® 500, Applikon Biotechnology, Delft, NL). The continuous culture was maintained at a volume of 350 mL with constant stirring at a dilution rate of 0.1 hr −1 at a temperature of 37 °C and a pH of 7.00 for 504 h (21 days). Chemostat cultures were maintained in a glucose-depleted medium, which was made by supplementing a minimal M9 medium, with 1 mM glucose and 0.2% casamino acids. Amino acids were utilised as a result of repeat test growth experiments with both strains in batch culture (data not shown). We found that growth for both strains was more consistent across tests when the medium was supplemented with amino acids. The optical density was monitored both continuously using the bugLab ® BE 2100 biomass monitor (Applikon Biotechnology, Delft, NL) and periodically, every 24 h, via subsampling and measurement on the Infinite ® m200 Pro automated microplate reader (Tecan Group Ltd., Männedorf, Switzerland). In these conditions, the OD600 reached 0.1 in about 20 h (steady state) and remained that way for the duration of the experiment.
## Mutant fraction and mutation rate estimates
At each sampling period [bib_ref] Indispensability of horizontally transferred genes and its impact on bacterial genome streamlining, Karcagi [/bib_ref] , and, 504 h) ~ 2-5 mL of the reactor liquor was subsampled using sterile conditions and techniques. Of this volume, 1 mL was utilised to make a glycerol stock, while 1 mL each was plated onto 5 LB-d-cycloserine (100 μg/mL) plates and 5 LB-Rifampicin (50 μg/mL) plates. All plates were freshly prepared. An additional 1 mL was diluted and plated onto LB-agar plates for a colony count. All plates were incubated overnight at 37 °C. Spontaneous mutations occurred and accumulated in populations from both strains and were scored via acquired resistance to either d-Cycloserine (cyc R ) or Rifampicin (rif R ). Mutation rate (μ) was estimated using a mutation accumulation model [bib_ref] Methods for determining spontaneous mutation rates, Foster [/bib_ref] , where N is the constant cell number in a chemostat, and r(t) is the number of mutations measured at time (t) and t 1 < t 2 are two discrete time points. The term λ is generations per unit time (generations hr −1 ) [bib_ref] Mutation in continuous cultures. I. Dependence of mutational response upon growth-limiting factors, Kubitschek [/bib_ref] , where and η is the growth rate in the chemostat of 0.1 hr −1; therefore, λ equals 0.14 generations hr −1 . Time, in 'hours, ' is eliminated from Eq. (5), leaving generations, which is a scaled measure of time commonly used in genetics to report mutation rates. Therefore using Eq. (5), mutation rate (μ) is expressed as mutations per cell per generation (mutations cell −1 generation −1 ).
## Mutant monoculture batch growth conditions
Up to 48 randomly picked colonies were batch cultured individually at 37 °C overnight in minimal media supplemented with 10 mM glucose and 0.2% casamino acids. Each culture was then diluted 100-fold into fresh M9 supplemented with 1 mM glucose and 0.2% casamino acids. Cultures were then grown (in batch mode) at 37 °C and optical density at 600 nanometres (OD 600 ) was measured every 20 min in the Infinite ® m200 Pro microplate reader (Tecan Group Ltd., Männedorf, Switzerland) for a period of 17 h. Blank measurements were also obtained and subtracted from OD 600 at each time point. Growth rate (h −1 ) was estimated as the slope of the straight-line portion of the plot of the natural log of each blank-corrected OD 600 measurement (Ln[OD t -blank]) versus time. Yield was equivalent to the final OD600 measurement at the end (17 h) of each batch culture experiment.
[formula] (5) µ = 1 N r(t 2 ) − r(t 1 ) t 2 − t 1 ,(6) [/formula]
= η ln (2) ,
## Additional file
Additional file 1: . Mean growth rates of 24 individual colonies each of the MG1655, and the MDS42 strains from batch monocultures. . Fold-difference between mutation rates estimated from the current study (stress) and [bib_ref] Emergent properties of reduced-genome Escherichia coli, Posfai [/bib_ref]. [fig_ref] Figure 3: A comparison of mean cyc R mutant growth rates of batch monocultures... [/fig_ref]. Average biomass yield for mutants and non-mutants of the MG1655 and MDS42 strains. . Wash out time distributions for isolated cyc R mutants from both the MG1655 and MDS42 strains at different time points during the experiments.
[fig] Figure 3: A comparison of mean cyc R mutant growth rates of batch monocultures of the MG1655 (stripped bars) and MDS42 strains (grey bars). The two horizontal lines represent the mean growth rate for non-mutant strains of MG1655 (dashed line; 1.1 hr −1 ) and MDS42 (dotted-dashed line; 0.79 hr −1 ) Fig. 4 [/fig]
|
10.1155/2014/382439
|
CCBY
|
4132342
|
25147847
|
s2orc_pubmed_articles
|
δ-Cut Decision-Theoretic Rough Set Approach: Model and Attribute Reductions
Decision-theoretic rough set is a quite useful rough set by introducing the decision cost into probabilistic approximations of the target. However, Yao's decision-theoretic rough set is based on the classical indiscernibility relation; such a relation may be too strict in many applications. To solve this problem, a -cut decision-theoretic rough set is proposed, which is based on the -cut quantitative indiscernibility relation. Furthermore, with respect to criterions of decision-monotonicity and cost decreasing, two different algorithms are designed to compute reducts, respectively. The comparisons between these two algorithms show us the following: (1) with respect to the original data set, the reducts based on decision-monotonicity criterion can generate more rules supported by the lower approximation region and less rules supported by the boundary region, and it follows that the uncertainty which comes from boundary region can be decreased; (2) with respect to the reducts based on decision-monotonicity criterion, the reducts based on cost minimum criterion can obtain the lowest decision costs and the largest approximation qualities. This study suggests potential application areas and new research trends concerning rough set theory.
# Introduction
Decision-theoretic rough set (DTRS) was proposed by Yao et al. in the early 1990s [bib_ref] A decision-theoretic rough set model, Yao [/bib_ref] [bib_ref] A decision theoretic framework for approximating concepts, Yao [/bib_ref]. Decision-theoretic rough set introduces Bayesian decision procedure and loss function into rough set. In decision-theoretic rough set, the pair of thresholds and , which are used to describe the tolerance of approximations, can be directly calculated by minimizing the decision costs with Bayesian theory. Following Yao's pioneer works, many theoretical and applied results related to decision-theoretic rough set have been obtained; see [bib_ref] Minimum cost attribute reduction in decision-theoretic rough set models, Jia [/bib_ref] [bib_ref] Non-monotonic attribute reduction in decision-theoretic rough sets, Li [/bib_ref] [bib_ref] Cost-sensitive classification based on decision-theoretic rough set model, Li [/bib_ref] [bib_ref] Cost-sensitive three-way decision: a sequential strategy, Li [/bib_ref] [bib_ref] Triangular fuzzy decision-theoretic rough sets, Liang [/bib_ref] [bib_ref] Systematic studies on three-way decisions with interval-valued decision-theoretic rough sets, Liang [/bib_ref] [bib_ref] A multiple-category classification approach with decision-theoretic rough sets, Liu [/bib_ref] [bib_ref] Incorporating logistic regression to decision-theoretic rough sets for classifications, Liu [/bib_ref] [bib_ref] Multigranulation decision-theoretic rough sets, Qian [/bib_ref] [bib_ref] An automatic method to determine the number of clusters using decision-theoretic rough..., Yu [/bib_ref] [bib_ref] Multi-class decision-theoretic rough sets, Zhou [/bib_ref] for more details.
In decision-theoretic rough set, Pawlak's indiscernibility relation is a basic concept [bib_ref] A novel cognitive system model and approach to transformation of information granules, Xu [/bib_ref] [bib_ref] Multi-granulation rough sets based on tolerance relations, Xu [/bib_ref] [bib_ref] On multigranulation rough sets in incomplete information system, Yang [/bib_ref] [bib_ref] Test cost sensitive multigranulation rough set: model and minimal cost selection, Yang [/bib_ref] [bib_ref] Hierarchy on multigranulation structures: a knowledge distance approach, Yang [/bib_ref] , and it is an intersection of some equivalence relations in knowledge base. It should be noticed that, in [bib_ref] Data analysis based on discernibility and indiscernibility, Zhao [/bib_ref] , Zhao et al. have made a further investigation about indiscernibility relation and proposed another two indiscernibility relations, which are referred to as weak indiscernibility and -cut quantitative indiscernibility relations, respectively. Correspondingly, Pawlak's indiscernibility relation is called the strong indiscernibility relation. By comparing such three binary relations, it is proven that thecut quantitative indiscernibility relation is a generalization of both strong and weak indiscernibility relations. Therefore, it is interesting to construct -cut decision-theoretic rough set based on -cut quantitative indiscernibility relation. This is what will be discussed in this paper.
Furthermore, attribute reduction is one of the most fundamental and important topics in rough set theory and has drawn attention from many researchers. As far as attribute reduction in decision-theoretic rough set, the properties of nonmonotonicity and decision cost should be concerned.
(1) On the one hand, as we all know, in Pawlak's rough set model, the positive region is monotonic with respect to the set inclusion of attributes. However, the monotonicity property of the decision regions with respect to the set inclusion of 2
The Scientific World Journal attributes does not hold in the decision-theoretic rough set model [bib_ref] Decision region distribution preservation reduction in decision-theoretic rough set model, Ma [/bib_ref] [bib_ref] A note on attribute reduction in the decision-theoretic rough set model, Zhao [/bib_ref]. To fill such a gap, Yao and Zhao proposed the definition of decision-monotonicity criterion based attribute reduction [bib_ref] Attribute reduction in decision-theoretic rough set models, Yao [/bib_ref] ; (2) on the other hand, decision cost is a very important notion in decision-theoretic rough set model; to deal with the minimal decision cost, Jia et al. proposed a fitness function and designed a heuristic algorithm [bib_ref] On an optimization representation of decision-theoretic rough set model, Jia [/bib_ref].
As a generalization of decision-theoretic rough set, in our -cut decision-theoretic rough set, we conduct the attribute reductions from above two aspects. Firstly, we introduce the notion of decision-monotonicity criterion into attribute reduction and design a significance to measure attributes; secondly, to deal with the minimum decision cost problem, we regard it as an optimization problem and apply the generic algorithm to obtain a reduct with the lowest decision cost.
To facilitate our discussions, we present the basic knowledge, such as Pawlak's rough set, -cut quantitative rough set, and Yao's decision-theoretic rough set in Sections 2 and 3. In Section 4, we propose a new -cut decision-theoretic rough set and present several related properties. In Section 5, we discuss the attribute reductions by considering two criterions. The paper ends with conclusions in Section 6.
## Indiscernibility relations and rough sets
## Strong indiscernibility
Relation. An information system is a pair = ( , ), in which universe is a finite set of the objects;
is a nonempty set of the attributes, such that for all ∈ , and is the domain of . For all ∈ , ( ) denotes the value of on . Particularly, when = ∪ and ∩ = 0 ( is the set of conditional attributes and is the set of decisional attributes), the information system is also called decision system. Each nonempty subset ⊆ determines a strong indiscernibility relation ( ) as follows:
[formula] ( ) = {( , ) ∈ 2 : ( ) = ( ) , ∀ ∈ } . (1) [/formula]
A strong indiscernibility relation with respect to is denoted as ( ). Two objects in satisfy ( ) if and only if they have the same values on all attributes in ; it is an equivalence relation.
( ) partitions into a family of disjoint subsets / ( ) called a quotient set of :
[formula] ( ) = {[ ] : ∈ } ,(2) [/formula]
where [ ] denotes the equivalence class determined by with respect to ; that is,
[formula] [ ] = { ∈ : ( , ) ∈ ( )} .(3) [/formula]
Definition 1. Let be an information system, let be any subset of , and let be any subset of . The lower approximation of denoted as ( ) and the upper approximation of denoted as ( ), respectively, are defined by
[formula] ( ) = { ∈ : [ ] ⊆ } ; ( ) = { ∈ : [ ] ∩ ̸ = 0} .(4) [/formula]
The pair [ ( ), ( )] is referred to as Pawlak's rough set of with respect to the set of attributes .
## Weak indiscernibility relation.
In the definition of strong indiscernibility relation, we can observe that two objects in satisfy ( ) if and only if they have the same values on all attributes in ; such case may be too strict to be used in many applications. To address this issue, Zhao and Yao proposed a notion which is called weak indiscernibility relation. The semantic interpretation of weak indiscernibility relation is that two objects are considered as indistinguishable if and only if they have the same values on at least one attribute in . In an information system , for any subset of , a weak indiscernibility relation can be defined as follows [bib_ref] Data analysis based on discernibility and indiscernibility, Zhao [/bib_ref] :
[formula] ( ) = {( , ) ∈ 2 : ( ) = ( ) , ∃ ∈ } . (5) [/formula]
From the description of the weak indiscernibility relation we can find that a weak indiscernibility relation ( ) with respect to only requires that two objects have the same values on at least one attribute in . A weak indiscernibility relation is reflexive and symmetric, but not necessarily transitive. Such a relation is known as a compatibility or a tolerance relation.
where [ ] = { ∈ : ( , ) ∈ ( )} is the set of objects, which are weak indiscernibility with in terms of set of attributes .
## -cut quantitative indiscernibility
Relation. The strong and weak indiscernibility relations represent the two extreme cases, which include many levels of indiscernibility. With respect to a nonempty set of attributes ⊆ , a -cut quantitative indiscernibility relation is defined as a mapping from × to the unit interval [0, 1].
Definition 3 (see [bib_ref] Data analysis based on discernibility and indiscernibility, Zhao [/bib_ref]. Let be an information system; for all ⊆ , the -cut quantitative indiscernibility relation ( ) is defined by
[formula] ( ) = {( , ) ∈ 2 : { ∈ : ( ) = ( )} | | ≥ } ,(7) [/formula]
where | ⋅ | denotes the cardinality of a set.
By the definition of -cut quantitative indiscernibility relation, we can obtain the lower and upper approximations as in the following definition.
where [ ] = { ∈ : ( , ) ∈ ( )} is the set of objects, which are -cut indiscernibility with in terms of set of attributes .
## Decision-theoretic rough set
The Bayesian decision procedure deals with making a decision with minimum risk based on observed evidence. Yao and Zhou introduced a more general rough set model called a decision-theoretic rough set (DTRS) model [bib_ref] Probabilistic rough set approximations, Yao [/bib_ref] [bib_ref] Three-way decision: an interpretation of rules in rough set theory, Yao [/bib_ref] [bib_ref] Naive bayesian rough sets, Yao [/bib_ref]. In this section, we briefly introduce the original DTRS model. According to the Bayesian decision procedure, the DTRS model is composed of two states and three actions. The set of states is given by Ω = { , ∼ } indicating that an object is in or not, respectively. The probabilities for these two complement states can be denoted as (
[formula] | [ ] ) = | ∩ [ ] |/|[ ] | and (∼ | [ ] ) = 1 − ( | [ ] ) [/formula]
. The set of actions is given by A = { , , }, where , , and represent the three actions in classifying an object , namely, deciding that belongs to the positive region, deciding that belongs to the boundary region, and deciding that belongs to the negative region, respectively. The loss functions are regarding the risk or cost of actions in different states. Let , , and denote the cost incurred for taking actions , , and , respectively, when an object belongs to , and let , , and denote the cost incurred for taking the same actions when an object belongs to ∼ .
According to the loss functions, the expected costs associated with taking different actions for objects in [ ] can be expressed as follows:
[formula] = ( | [ ] ) = ⋅ ( | [ ] ) + ⋅ (∼ | [ ] ) ; = ( | [ ] ) = ⋅ ( | [ ] ) + ⋅ (∼ | [ ] ) ; = ( | [ ] ) = ⋅ ( | [ ] ) + ⋅ (∼ | [ ] ) .(9) [/formula]
The Bayesian decision procedure leads to the following minimum-risk decision rules: ; that is to say, the loss of classifying an object belonging to into the positive region is no more than the loss of classifying into the boundary region, and both of these losses are strictly less than the loss of classifying into the negative region. The reverse order of losses is used for classifying an object not in . We further assume that a loss function satisfies the following condition:
[formula] ( ) if ≤and( − ) ⋅ ( − ) > ( − ) ⋅ ( − ) .(10) [/formula]
Based on the above two assumptions, we have the following simplified rules:
[formula] ( 1) if ( | [ ] ) [/formula]
≥ , then this decides that belongs to the positive region;
( 1) if < ( | [ ] ) < , then this decides that belongs to the boundary region;
( 1) if ( | [ ] ) ≤ , then this decides that belongs to the negative region,
[formula] where = − ( − ) + ( − ) ; = − ( − ) + ( − ) ,(11) [/formula]
with 1 ≥ ≥ ≥ 0. Using these three decision rules, for all ⊆ and for all ⊆ , we get the following probabilistic approximations:
## -cut decision-theoretic rough set
As the discussion in Section 3, we can observe that the classical decision-theoretic rough set is based on the strong indiscernibility relation which is too strict since it requires that the two objects have the same values on all attributes. In this section, we introduce the concept of -cut indiscernibility relation into the decision-theoretic rough set model. The pair [ ( , ) ( ), ( , ) ( )] is referred to as a -cut decision-theoretic rough set of with respect to the set of attributes .
## Definition of -cut decision-theoretic rough set
After obtaining the lower and upper approximations, the probabilistic positive, boundary, and negative regions are defined by
[formula] POS ( , ) ( ) = ( , ) ( ) ; BND ( , ) ( ) = ( , ) ( ) − ( , ) ( ) ; NEG ( , ) ( ) = − POS ( , ) ( ) ∪ BND ( , ) ( ) = − ( , ) ( ) .(14) [/formula]
Let be a decision system and let = { 1 , 2 , . . . , } be a partition of the universe , which is defined by the decision attribute , representing classes. By the definition of quantitative decision-theoretic rough set, the lower and upper approximations of the partition can be expressed as follows:
[formula] ( , ) ( ) = ( ( , ) ( 1 ) , ( , ) ( 2 ) , . . . , ( , ) ( )) ; ( , ) ( ) = ( ( , ) ( 1 ) , ( , ) ( 2 ) , . . . , ( , ) ( )) .(15) [/formula]
For this -classes problem, it can be regarded as twoclass problems; following this approach, the positive region, boundary region, and negative region of all the decision classes can be expressed as follows:
[formula] POS ( , ) ( ) = ⋃ =1 POS ( , ) ( ) ; BND ( , ) ( ) = ⋃ =1 BND ( , ) ( ) ; NEG ( , ) ( ) = − POS ( , ) ( ) ∪ BND ( , ) ( ) .(16) [/formula]
Based on the notions of the three regions in -cut decision-theoretic rough set model, three important rules should be concerned, that is, positive rule, boundary rule, and negative rule. Similar to Yao's decision-theoretic rough set, when > , for all ∈ , we can obtain the following decision rules, that is, tie-break:
[formula] ( -) if ( | [ ] ) ≥ , then this decides that ∈ POS ( , ) ( ); ( -) if < ( | [ ] ) < , then this decides that ∈ BND ( , ) ( ); ( -) if ( | [ ] ) ≤ , then this decides that ∈ NEG ( , ) ( ). [/formula]
Let be a decision system, ∈ (0, 1]; for all ∈ , the Bayesian expected costs of decision rules can be expressed as follows:
[formula] (i) ( -) cost: ∑ ∈ ∑ ∈POS( ) ( ⋅ ( | [ ] )+ ⋅ (∼ | [ ] )); (ii) ( -) cost: ∑ ∈ ∑ ∈NEG( ) ( ⋅ ( | [ ] ) + ⋅ (∼ | [ ] )); (iii) ( -) cost: ∑ ∈ ∑ ∈BND( ) ( ⋅ ( | [ ] )+ ⋅ (∼ | [ ] )). [/formula]
Considering the special case where we assume zero cost for a correct classification, that is, = = 0, the decision costs of rules can be simply expressed as follows:
[formula] (i) ( -1) cost: ∑ ∈ ∑ ∈POS( ) ⋅ (∼ | [ ] ); (ii) ( -1) cost: ∑ ∈ ∑ ∈NEG( ) ⋅ ( | [ ] ); (iii) ( -1) cost: ∑ ∈ ∑ ∈BND( ) ( ⋅ ( | [ ] ) + ⋅ (∼ | [ ] )). [/formula]
For any subset of conditional attributes, the overall cost of all decision rules can be denoted as COST( ), such that
[formula] COST ( ) = COST POS + COST NEG + COST BND = ∑ ∈ ∑ ∈POS( ) ⋅ (∼ | [ ] ) + ∑ ∈ ∑ ∈NEG( ) ⋅ ( | [ ] ) + ∑ ∈ ∑ ∈BND( ) ( ⋅ ( | [ ] ) + ⋅ (∼ | [ ] )) .(17) [/formula]
# Related properties
## Proposition 6. let be an information system; if
[formula] = = 1 and = = = = 0, ∀ ⊆ , one has ( , ) ( ) = ( ) ; ( , ) ( ) = ( ) .(18) [/formula]
Proof. In this proposition, we suppose that there is a unit misclassification cost if an object in is classified into the negative region or if an object in ∼ is classified into the positive region; otherwise there is no cost; that is, = = 1 and = = = = 0. By the computational processes of and , we have = 1 and The Scientific World Journal 5 = 0 and by the definition of -cut decision-theoretic rough set, we can observe that
[formula] ( , ) ( ) = { ∈ : ( | [ ] ) ≥ 1} = { ∈ : ∩ [ ] [ ] ≥ 1} = { ∈ : [ ] ⊆ } = ( ) .(19) [/formula]
Similarly, it is not difficult to prove ( , ) ( ) = ( ).
## Proposition 7.
Let be an information system; for all ⊆ , for all ⊆ , one has
[formula] ( , ) ( ) ⊇ ( ) ; ( , ) ( ) ⊆ ( ) .(20) [/formula]
Proof. For all ∈ ( ) and by Similarly, it is not difficult to prove ( , ) ( ) ⊆ ( ).
Propositions 6 and 7 show the relationships betweencut decision-theoretic rough set and classical -cut quantitative rough set. The details are given as follows: the classical -cut quantitative indiscernibility lower approximation is included into the -cut decision-theoretic lower approximation and the -cut decision-theoretic upper approximation is included into the classical -cut quantitative indiscernibility upper approximation. Particularly, with some limitations, the -cut decision-theoretic rough set can degenerate to the classical -cut quantitative rough set. As the discussion above, we can observe that the -cut decision-theoretic rough set is a generalization of classical -cut quantitative rough set, and it can increase lower approximation and decrease upper approximation.
Proof. It is not difficult to prove this proposition by Definitions 3 and 5 and the definition of decision-theoretic rough set.
Proposition 8 shows the relationships between -cut decision-theoretic rough set and Yao's decision-theoretic rough set. The details are the following: if we set the value of with 1, the lower and upper approximations based on our decision-theoretic rough set are equal to those based on Yao's decision-theoretic rough set. By Proposition 8 we can observe that our decision-theoretic rough set is also a generalization of Yao's decision-theoretic rough set.
## Attribute reductions in quantitative
Decision-Theoretic Rough Set
## Decision-monotonicity criterion based reducts. in
Pawlak's rough set theory, attribute reduction is an important concept which has been addressed by many researchers all around the world. In classical rough set, the reduct is a minimal subset of attributes which is independent and has the same power as all of the attributes. The positive region, the boundary region, and the negative region are monotonic with respect to the set inclusion of attributes in classical rough set theory. However, in decision-theoretic rough set model, the monotonicity property of the decision regions with respect to the set inclusion of attributes does not hold. To solve such a problem, Yao and Zhao have proposed a decision-monotonicity criterion [bib_ref] Attribute reduction in decision-theoretic rough set models, Yao [/bib_ref]. The decision-monotonicity criterion requires two things. Firstly, the criterion requires that by reducing attributes a positive rule is still a positive rule of the same decision. Secondly, the criterion requires that by reducing attributes a boundary rule is still a boundary rule or is upgraded to a positive rule with the same decision. Following their work, it is not difficult to introduce the decision-monotonicity criterion into our -cut decision-theoretic rough set. The detailed definition is shown in Definition 9 as follows.
Definition 9. Let = ( , ∪ ) be a decision system, ∈ (0, 1], and let be any subset of conditional attributes; is referred to as a decision-monotonicity reduct in if and only if is the minimal set of conditional attributes, which preserves ( , ) ( ) ⊆ ( , ) ( ), for each ∈ .
Let be a decision system, ∈ (0, 1], and let be any subset of conditional attributes and ∈ ; we define the following coefficients: sig in ( , , )
[formula] = ∑ ∈ ( ( , ) ( ) ⊙ − { } ( , )( )) [/formula]
⋅ ; sig out ( , , )
[formula] = ∑ ∈ ( ( , ) ( ) ⊙ ∪ { } ( , )( )) [/formula]
⋅ ,
where and are the numbers of objects and decision classes, respectively, and Step 6.
[formula] ⊙ = { | − | ⊆ , − | − | otherwise.(23) [/formula]
= .
Algorithm 1: Heuristic algorithm for attribute reduction based on decision-monotonicity criterion.
## Input:
Decision system = ( , ∪ ), threshold ; Output: A optimal cost reduct .
Step 1. Create an initial random population (number = 40);
Step 2. Evaluation the population;
Step 3. While Number of generations < 100 do Select the fittest chromosomes in the population; Perform crossover on the selected chromosomes to create offspring; Perform mutation on the selected chromosomes; Evaluate the new population;
## End
Step 4. Selected the fittest chromosome form current population and output it as .
Algorithm 2: Genetic algorithm for attribute reduction based on cost minimum criterion.
Based on these measures, we can design a heuristic algorithm to compute the decision-monotonicity reduct; the details are shown as in Algorithm 1.
## Cost minimum criterion based reducts.
Cost is one of the important features of the -cut decision-theoretic rough set. In Section 4.1 we have discussed the cost issue of our -cut decision-theoretic rough set. However, in the reduction process, from the viewpoint of cost criterion, we want to obtain a reduct with smaller or smallest cost. Similar to the decision-monotonicity criterion, it is not difficult to introduce the cost criterion into our rough set model.
## Definition 10. let
= ( , ∪ ) be a decision system, ∈ (0, 1], and let be any subset of conditional attributes; is referred to as a cost reduct in if and only if is the minimal set of conditional attributes, which satisfies COST( ) ≤ COST( ), and, for each set ⊂ , COST( ) > COST( ).
In this definition, we want to find a subset of conditional attributes so that the overall decision cost will be decreased or unchanged based on the reduct. In most situations, it is better for the decider to obtain a smaller or smallest cost in the decision procedure. We propose an optimization problem with the objective of minimizing the cost values; the minimum cost can be denoted as follows [bib_ref] Minimum cost attribute reduction in decision-theoretic rough set models, Jia [/bib_ref] :
[formula] min COST ( ) .(24) [/formula]
Then the optimization problem is described as finding a proper attributes set to make the whole decision cost minimum. Therefore, in the following, we will present a genetic algorithm to compute cost minimum based reducts. The details of genetic algorithm are described in Algorithm 2.
The Scientific World Journal 7
## Experimental analyses.
In this subsection, by experimental analyses, we will illustrate the differences between Algorithms 1 and 2. All the experiments have been carried out on a personal computer with Windows 7, Intel Core 2 DuoT5800 CPU (4.00 GHz), and 4.00 GB memory. The programming language is Matlab 2012b. We download four public data sets from UCI Repository of Machine Learning Databases, which are described in [fig_ref] Table 1: Data sets description [/fig_ref]. In the experiment, 10 different groups of loss functions are randomly generated. Tables 2, 3, 4, and 5 show the experimental results of ( ) rules, ( ) rules, and ( ) rules. The number of these rules is equivalent to the number of objects in positive region, boundary region, and negative region, respectively. This is mainly because each object in positive/boundary/negative region can induce a ( )/( )/( ) decision rule.
Based on these four tables, it is not difficult to draw the following conclusions.
(1) With respect to the original data set, decisionmonotonicity reducts can generate more ( ) rules; this is mainly because the condition of decisionmonotonicity reducts requires that, by reducing attributes, a positive rule is still a positive rule, or a boundary rule is upgraded to a positive rule. This mechanism not only keeps the original ( ) rules unchanged, but also increases the ( ) rules. (2) With respect to the original data set, decisionmonotonicity reducts can generate less ( ) rules; this is mainly because the second condition of decisionmonotonicity reducts requires that, by reducing attributes, a boundary rule is still a boundary rule or is upgraded to a positive rule; that is to say, the number of ( ) rules may be equal to or less than those of original data set.
In order to compare the differences between decisionmonotonicity criterion based reducts and cost minimum criterion based reducts, we conduct the experiments from three aspects, that is, decision costs, approximation qualities, and running times. On the one hand, [fig_ref] Figure 1: Reducts' costs comparison between decision-monotonicity and cost minimum criterions [/fig_ref] shows the costs comparisons between these two attribute reduction algorithms; on the other hand, Tables 6, 7, 8, and 9 show the differences between decision-monotonicity criterion based reducts and cost minimum criterion based reducts in approximation qualities and running times, respectively.
In [fig_ref] Figure 1: Reducts' costs comparison between decision-monotonicity and cost minimum criterions [/fig_ref] , each subfigure is corresponding to a data set. In each subfigure, the -coordinate pertains to different 8 The Scientific World Journal 307 ± 0 3 0 7 ± 0 0 ± 0 0 ± 0 9 2 1 ± 0 9 2 1 ± 0 1.0 307 ± 0 3 0 7 ± 0 0 ± 0 0 ± 0 9 2 1 ± 0 9 2 values of , whereas the -coordinate concerns the values of costs. Through an investigation of [fig_ref] Figure 1: Reducts' costs comparison between decision-monotonicity and cost minimum criterions [/fig_ref] , it is not difficult to observe that, in all the ten used values of , the decision costs of cost minimum criterion based reducts are the same or lower than those obtained by decisionmonotonicity criterion based reducts. [fig_ref] Table 6: The comparison between decision-monotonicity criterion based reducts and cost based reducts [/fig_ref] show the differences between decisionmonotonicity criterion based reducts and cost minimum criterion based reducts in approximation qualities and running times, respectively. It is not difficult to note that, from the viewpoint of approximation qualities, the approximation qualities of decision-monotonicity criterion based reducts are The Scientific World Journal larger than those of cost minimum criterion based reducts at times. However, in most cases, the approximation qualities of cost minimum criterion based reducts are larger than those of decision-monotonicity criterion based reducts. From the point of running times, it is easy to observe that the run times of genetic algorithm are greater than those of heuristic algorithm.
To sum up, we can draw the following conclusions.
(1) From the viewpoint of decision monotonicity, our heuristic algorithm based on decision-monotonicity criterion can generate more ( ) rules and less ( ) rules with respect to the original data set. Such approach not only increases the certainties which are expressed by ( ) rules and ( ) rules, but also decreases the uncertainty coming from ( ) rules. (2) From the viewpoint of decision costs, the generic algorithm based on cost minimum criterion can obtain the lowest decision costs and the largest approximation qualities by comparing with heuristic algorithm based on decision-monotonicity criterion. However, such approach loses the property of decision monotonicity and it wastes larger running times than heuristic algorithm.
# Conclusion
In this paper, we have developed a generalized framework of decision-theoretic rough set, which is referred 10
The Scientific World Journal to as a -cut decision-theoretic rough set. Different from Yao's decision-theoretic rough set model, our model is constructed based on -cut quantitative indiscernibility relation, and it can degenerate to Yao's decision-theoretic rough set with some limitation. Based on the proposed model, we discussed the attribute reductions from two criterions; the experiments show that, on the one hand, the decision-monotonicity criterion based reducts can generate more positive rules and less boundary rules; on the other hand, the cost minimum criterion based reducts can obtain the lowest decision costs with high approximation qualities.
The Scientific World Journal
## 11
The present study is the first step towards -cut decisiontheoretic rough set. The following are challenges for further research.
(1) -cut decision-theoretic rough set approach to complicated data type, such as interval-valued data, is one of the challenges; incomplete data may be an interesting topic.
(2) The threshold learning of in this paper is also a serious challenge.
[fig] Definition 2: Let be an information system; for all ⊆ , for all ⊆ , the lower and upper approximations of based on weak indiscernibility relation, denoted as ( ) and( ), respectively, are defined by [/fig]
[fig] .: Let be an information system; for all ⊆ , for all ⊆ , the -cut quantitative indiscernibility based lower and upper approximations are denoted by ( ) and ( ), respectively: [/fig]
[fig] Definition 5: Let be an information system; for all ⊆ , for all ⊆ , the decision-theoretic lower and upper approximations based on the - [/fig]
[fig] Proposition 8: Let be an information system; if = 1, then, for all ⊆ , for all ⊆ , one has ( , ) ( ) = , ( ) ; ( , ) ( ) = , ( ) . [/fig]
[fig] Figure 1: Reducts' costs comparison between decision-monotonicity and cost minimum criterions. [/fig]
[table] Table 1: Data sets description. [/table]
[table] Table 2: The decision rules between raw data and decision-monotonicity criterion based reducts (Annealing). [/table]
[table] Table 3: The decision rules between raw data and decision-monotonicity criterion based reducts (Dermatology). [/table]
[table] Table 4: The decision rules between raw data and decision-monotonicity criterion based reducts (Soybean). [/table]
[table] Table 6: The comparison between decision-monotonicity criterion based reducts and cost based reducts (Annealing). [/table]
[table] Table 7: The comparison between decision-monotonicity criterion based reducts and cost based reducts (Dermatology). [/table]
[table] Table 8: The comparison between decision-monotonicity criterion based reducts and cost based reducts (Soybean). [/table]
[table] Table 9: The comparison between decision-monotonicity criterion based reducts and cost based reducts (Zoo). [/table]
|
10.21203/rs.3.rs-2683913/v1
|
CCBY
|
10055658
|
36993740
|
s2orc_pubmed_articles
|
Typical Diagnostic Reference Levels of Common Indications for Computed Tomography Scans Among Adult Patients in Uganda: a Cross-sectional Study
Background
# Abstract
Background Medical exposure to ionizing radiation has increased due to an increase in the number of computerized tomography (CT) scan examinations performed. The International Commission on Radiological Protection (ICRP) recommends indication-based diagnostic reference levels (IB-DRLs) as an effective tool that aids in optimizing CT scan radiation doses. In many low-income settings, there is a lack of IB-DRLs to support optimization of radiation doses.
## Objective
To establish typical DRLs for common CT scan indications among adult patients in Kampala, Uganda.
# Methodology:
A cross sectional study design was employed involving 337 participants enrolled from three hospitals using systematic sampling. The participants were adults who had been referred for a CT scan. The typical DRL of each indication was determined as the median value of the pooled distribution of CTDIvol (mGy) data and the median value of the pooled distribution of total DLP (tDLP)(mGy.cm) data from three hospitals. Comparison was made to anatomical, and indication based DRLs from other studies.
Results 54.3% of the participants were male. The following were typical DRLs for: acute stroke (30.17mGy and 653mGy.cm); head trauma (32.04mGy and 878mGy.cm); interstitial lung diseases/ high resolution chest CT scan (4.66mGy and 161mGy.cm); pulmonary embolism (5.03mGy and 273mGy.cm); abdominopelvic lesion (6.93mGy and 838mGy.cm) and urinary calculi (7.61mGy and 975mGy.cm).
Indication based total Dose Length Product (tDLP) DRLs was lower than tDLP DRLs of a whole anatomical region by 36.4% on average.
Most of the developed typical IB-DLP DRLs were lower or comparable to values from studies in Ghana and Egypt in all indications besides urinary calculi while they were higher than values in a French study in all indications besides acute stroke and head trauma.
# Conclusion
Typical IB-DRLs is a good clinical practice tool for optimization of CT doses and therefore recommended for use to manage CT radiation dose. The developed IB-DRLs varied from international values due to differences in selection of CT scan parameters and standardization of CT imaging protocols may narrow the variation. This study can serve as baseline for establishment of national indication-based CT DRLs in Uganda.
# Background
The technological advancement in Computerized Tomography (CT) has increased its clinical applications with multidetector CT (MDCT) scans worldwide . CT scanners contribute the highest (43%) collective effective dose of radiation to the population among all medical imaging modalities . This increases the long-term risk for some cancers by 2% [bib_ref] Projected cancer risks from computed tomographic scans performed in the United States..., De González [/bib_ref]. The International Commission on Radiological Protection (ICRP) among other international bodies has issued recent calls for development and utilization of CT indication based-DRLs (IB-DRLs) for better optimization of CT doses per examination within or among CT facilities in preference to CT anatomical based DRLs . This added onto the Bonn Call for action by IAEA and WHO to strengthen radiation dose optimization through DRL establishment and use within a decade from 2017. The IB-DRL is superior to a single anatomical DRL that cannot logically optimize doses for all the different indications with different image quality requirements in a body region as indicated by ICRP, IAEA and European Society of Radiology . However, IB-DRLs are yet to be considered a priority by the Atomic Energy Councils in low Middle-Income Countries (LMICs) as evidenced by limited published data on IB-DRLs from some studies conducted in Ghana, Egypt and Nigeria [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref] [bib_ref] Clinical indication-based diagnostic reference levels for paediatric head computed tomography examinations in..., Zira [/bib_ref]. This current study was conducted in Uganda, a low-resource setting, but with increasing use of CT in patient care due to the availability, relative affordability, high resolution images and fast image acquisition of CT equipment [bib_ref] An audit of registered radiology equipment resources in Uganda, Kiguli-Malwadde [/bib_ref]. Currently, in many LMICs, there is limited published data on DRLs [bib_ref] Diagnostic reference levels in low-and middle-income countries: early "ALARAm" bells?, Meyer [/bib_ref]. Some of such studies that have been conducted to establish anatomical CT DRLs in adults and pediatrics [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref]. However, CT IB-DRLs are lacking at national level and for smaller groups of hospitals in Uganda. DRLs are recommended as one of the tools that can be employed to minimize ionizing radiation used in medical exposures to as low as reasonably achievable levels (ALARA) (6). Therefore, this study sought to establish typical IB-DRLs for common CT scan indications among adult patients in Kampala, Uganda to aid CT dose optimization since national indication based-DRLs (IB-DRLs) are lacking.
Findings from the study can be used by many other LMIC settings to establish their own IB-DRLs.
# Methods and materials
## Study design
A cross-sectional design was used.
## Selection of the participating ct facilities
At the time of the study, the central region of Uganda where Kampala is located, had 13 out of the 25 CT scanners in the country. Since the study`s target was to establish DRLs using less than 10 CT scanner rooms/hospitals, according to the available nancial and time resources, the type of DRL it was to develop was a typical DRL as de ned in ICRP publication 135(6). For ethical reasons the hospitals were anonymized and coded with alphabetical letters. Five tertiary hospitals with a high bed capacity, wide range of radiological services and functional CT scanners within Kampala were selected randomly to represent the public sector (Hospital C), private not for pro t sector (Hospital A) and private for-pro t sector (Hospitals B, D and E). A survey was done at the selected hospitals to ascertain the monthly number of adult head, chest and abdominopelvic CT scan examinations performed and their clinical indications with the results presented in [fig_ref] Table 1: CT examinations performed among adult patients and their indications at ve hospitals... [/fig_ref]. Hospitals with the highest number of head CT examinations were A, B and C. Hospitals with the highest number of chest CT examinations were A, D and E. Hospitals that performed the highest number adult CT examinations in most body regions were selected as the study sites and these included Hospitals A, B and C. Permission was sought from the administrations of the three study sites, each of which had one CT scanner.
## Selection of the most common adult ct scan indications
The two indications most frequently examined for among head CT scans, chest CT scans and abdominopelvic CT scans at each of the ve hospitals were identi ed and are presented in [fig_ref] Table 1: CT examinations performed among adult patients and their indications at ve hospitals... [/fig_ref]. The indications that were most frequently examined for across all the ve hospitals or across most of the hospitals were selected as the most common indications for adult CT scans and these included head trauma and acute stroke among head CT scans, pulmonary embolism (PE) and interstitial lung diseases/high resolution chest CT scan (ILD/HRCT) among chest CT scans and abdominopelvic lesion (ABDPL) and urinary calculi (UC) among abdominopelvic CT scans.
Patients who had been referred to the hospital CT units for examination were screened to check for eligibility which included: age of 18 years and above (adult); a common CT scan indication, informed consent; weight of 50-90kg for those with the indication of either interstitial lung disease, pulmonary embolism, abdominopelvic lesion or urinary calculi as the size of the chest and abdomen varies with body weight; and any body weight for head trauma and acute stroke as the size of the head does not vary substantially with change in weight (6). CT examinations with incomplete raw data on the CT console and those with mixed indications (e.g., stroke or brain mass) were excluded.
Since data was not collected from large electronic data bases, and was collected for a few participants using paper forms, it was important to restrict the weight range to minimize variation in the CT doses of indications within the chest and abdomen regions whose size varies with body weight. Therefore DRLs of PE, ILD/HRCT, ABDPL and UC were developed for a standardized adult of 50-90 kg as suggested by ICRP (6).
Sampling and sample size estimation There was need to correct the CT doses displayed on the consoles into the actual emitted doses. The actual radiation dose output from the CT scanners was recorded as X and the dose displayed on the CT console was recorded as Y for head CT examinations and abdomen CT examinations. A formula for calculating the actual radiation dose emitted by the CT scanner during an examination was then developed as (Y)/conversion factor. The conversion factor being (Y)/(X). The conversion factors for head CT examinations and abdomen CT examinations were interpolated for each CT scanner to get the speci c conversion factor for chest CT examinations.
These conversion factors were used on the collected displayed CT doses as there were no in-country Siemens biomedical engineers at the time of the study to go on with CT machine calibration using the conversion factors. The IAEA recommends use of conversion factors to verify the displayed CT radiation doses (4).
## Data collection
Among the patients who had been referred for a CT examination at the hospital CT unit, those who met the study's eligibility criteria by age and CT indication were approached to obtain informed consent. The consented participants with the indications of acute stroke and head trauma were enrolled. The consented participants with the indications of PE, ILD/HRCT, ABDPL and UC were weighed and those within 50-90 kg were enrolled into the study. The age, gender, weight and CT scan indication of the participants were recorded. After the CT examination had been performed by a radiographer, the CT scan parameters (which included mAs, kVp, pitch, rotation time (Trot), slice thickness, scan length, number of scan sequences and contrast use) and the radiation dose output from the CT scanners (CTDIvol and DLP) were recorded from the CT console. The quality of all CT images of the recruited participants was assessed by a radiologist at each hospital, using a scale adopted from the IAEA CT dose data collection tool, as acceptable, higher than needed or unacceptable and was recorded too. All data was recorded onto a paper form using a CT data collection tool adapted from the international atomic energy agency (IAEA).
## Study variables
The study variables were kVp, total mAs, effective mAs, slice thickness, scan length, contrast use, number or range of scan sequences, total DLP, CTDIvol.
Image quality was assessed using a 3-point scale adapted from the International Atomic Energy Agency (IAEA) CT dose data collection tool. Every image was scored for overall quality by the radiologist on a 3-point scale which included: 1 = Acceptable, 2 = Higher than needed and 3 = Unacceptable.
## Data management and analysis
Data was de-identi ed, checked daily for completeness and kept under lock and key. It was entered into excel for cleaning, error checks, editing and storage. Using excel, the recorded CT dose (Y) was converted into the actual radiation dose emitted by the CT scanner during examination of a participant using the formula; actual radiation dose emitted (A) = (Y)/conversion factor.
The actual CT doses and other data were then exported into and analyzed using R and R studio version 4.10. The Shapiro test was used to test for normality of the data and the data was found to be skewed. Descriptive analysis of the data was performed to include frequencies, the median and interquartile range.
The typical IB-DRLs were determined following the method recommended in ICRP publication 135 (6). The typical DRL for each indication was determined as the median value of the average CTDIvol of the whole examination and the median value of the total DLP using dose data of 60 participants combined from all three hospitals, for all indications besides urinary calculi. For urinary calculi, the typical DRL was developed using data from a total of only 30 participants: 20 from Hospital B and 10 from Hospital A. Only 7 participants were recruited at Hospital C during the study period, therefore these were very few to be included in the calculation of the typical DRL for urinary calculi as the minimum number of participants that can be included in the calculation of a DRL for most examinations generally maybe 10 (4, 6).
In addition, an overall IB-DRL for each indication was calculated as the 75th percentile of the median CTDIvol values and the median total DLP values from the three hospitals for comparison to national DRLs from other studies including anatomical based national DRLs in Uganda where the study was conducted from and IB-DRLs from international studies in Ghana, Egypt and Ghana [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref] [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref] [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref].
# Results
The purpose of this study was to establish typical DRLs for common indications of CT scans among adult patients.
## Characteristics of the sample population
There were 154 (45.74%) females. The overall median age was 53.5(36.25-67.75) years with the youngest participants being those with head trauma at 35(28-45) years and the eldest being those with stroke at 57.5(44.5-70) years. The overall median weight for all indications in the study excluding head trauma and stroke was 74(63.6-79.7) kg. The distribution of the study sample by age, weight, and gender in each CT indication at all hospitals combined is presented in [fig_ref] Table 2: Distribution of sample population used to determine CT doses by age, weight,... [/fig_ref]. CT scanner characteristics All the study hospitals possessed CT scanners manufactured by Siemens with that of Hospital C being a 16-slice scanner and the other characteristics of the CT scanners are presented in [fig_ref] Table 3: Characteristics of the CT Scanners [/fig_ref].
## Scanning parameters per ct indication
All CT dose data for each indication was collected from images of acceptable quality as subjectively assessed by radiologist. The median x-ray tube kilovoltage of 130 kVp was used to acquire CT images for each indication unlike for pulmonary embolism (PE) in which 110 (110-114.25) kVp was used. Contrast material was only used in the indications of PE, ABDPL and UC. The rest of the scan parameters that were used to perform the CT examinations for each indication in the study are presented in [fig_ref] Table 4: Scanning parameters used to the acquire CT images per indication For number... [/fig_ref]. Typical IB-DRLs
The typical CTDIvol DRLs of indications within the head, chest and abdomen were comparable with a varifying factor (vf) of 1.1. The typical DLP DRL of head trauma was 1.34-fold higher than for acute stroke. The typical DLP DRL for PE was 1.7-fold higher than for ILD/HRCT. The typical DLP DRL for UC was higher to that of ABDPL by a vf of 1.2-fold. The developed typical DRLs at 50th percentile plus the overall DRLs at the 75th percentile are presented in [fig_ref] Table 5: The typical DRLs at 50th percentile and the overall DRLs at the... [/fig_ref]. Comparison of the developed overall DRLs at 75th percentile (P) to published anatomical based national DRLs (AB-NDRLs) in Uganda.
The CTDIvol DRLs of all indications were lower than the AB-CTDIvol DRLs of the corresponding anatomical regions. A comparison of the current study's DRLs for the various indications to the Ugandan AB-NDRLs of the corresponding anatomical regions revealed that the DLP DRLs were only lower than the AB-national DLP values in the indications of head trauma (lower by 27.4%), acute stroke (lower by 42.4%), and ILD/HRCT (lower by 39.4%) (17) as presented in [fig_ref] Table 6: Comparison of the developed overall IB-DRLs at 75th P with national anatomical... [/fig_ref]. Comparison of the overall IB-DRLs at 75th percentile in the current study to some of the published national IB-DRLs at 75th percentile.
The CTDIvol DRLs were lower than Ghana's (10) in all indications (acute stroke, head trauma, PE, ABDPL and UC) and lower than Egypt's [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref] in ILD/HRCT as presented in [fig_ref] FiguresFigure 1: Research reported in this publication was supported by the Fogarty International Center... [/fig_ref]. The CTDIvol DRLs were lower than France's [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] in acute stroke and head trauma, comparable to France's [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] in ABDPL and UC and only higher than France's [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] in ILD/HRCT as presented in [fig_ref] FiguresFigure 1: Research reported in this publication was supported by the Fogarty International Center... [/fig_ref].
The DLP DRLs were mostly lower than those of Ghana (10) in the indications of acute stroke, head trauma and PE, but higher than Ghana's (10) only in UC and only comparable to Ghana's [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] in ABDPL as presented in . The DLP DRL of ILD/HRCT was also lower than Egypt's (11) as presented in . The current study's DLP DRLs were mostly higher than those of France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] in the indications ILD/HRCT, PE, ABDPL and UC and were only lower than France's [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] values in the indications of acute stroke and head trauma as presented in .
# Discussion
The purpose of this study was to establish typical DRLs for common CT indications among adult patients in Uganda. In this study, IB-DRLs for common indications of CT examinations among adults were developed as typical DRLs set at the 50 th percentile of the pooled distribution of dose data from 12% (3/25) of the CT scanners in Uganda, a low-income country. The common CT scan indications were similar to those in a study within Ghana and in Europe [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] [bib_ref] Diagnostic reference levels and median doses for common clinical indications of CT:..., Bos [/bib_ref].
The overall median weight for all indications excluding head trauma and stroke was 74(63.6-79.7) kg with pulmonary embolism having the most heavy patients with 78(71.5-85) kg and interstitial lung diseases having the least heavy patients with 69.5(60-77) kg. This overall median weight was within that of a standard adult population de ned by ICRP with 70+/-20kg. These ndings were also similar to those of other studies in which heavier participants suffered more recurrent episodes of venous thromboembolism and pulmonary embolism [bib_ref] Overweight, Obesity, and the Risk of Recurrent Venous Thromboembolism, Eichinger [/bib_ref] and lower weight was associated with disease progression in interstitial lung diseases [bib_ref] Weight loss as a predictor of mortality in patients with interstitial lung..., Pugashetti [/bib_ref].
CT scanning parameters:
X-ray tube voltage (measured in peak kilovoltage, kVp): The tube voltage of 130kVp that was used for all indications besides pulmonary embolism was higher than values used in other studies for example higher than 118 (±8.3) to 121.8(±7.4) in Ghana (10), 120 kVp in Egypt [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref] and (100 to 120) kVp in France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref]. The tube voltage of 110 (110-114.25) kVp for pulmonary embolism was similar to trends in some studies for example (100-120)kVp in France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] while this kVp for PE was lower than values in other studies for example 117.8(±4.0) in Ghana. There is potential to reduce CT radiation doses by lowering kVp for patients of smaller body weight, especially for ILD/HRCT by setting a speci c kVp for a particular weight category on the CT machines at the hospitals as kVp in the recent CT machine models is xed for an examination and is not modulated automatically during an examination to suit body size like it happens for x-ray tube current. There is hope though that newer CT scanner models may bring further improvements in the range of available tube voltages and more advanced automatic tube voltage selection tools which can automatically alter kVp according to patient size to achieve ALARA doses (25).
X-ray tube current-time product (measured in milliampere per second, mAs): Indications that did not use contrast material like ILD/HRCT generally had a lower effective mAs and a lower total mAs and compared to those that used contrast material like ABDPL, because of the lower attenuation in the absence of iodinated contrast material in the noncontrasted examinations. Indications within the chest generally had a lower total mAs than those in the abdomen and head because of the lower density/attenuation of the chest tissues for which tube current modulation software automatically reduces the mAs during the scan to minimize radiation dose. This trend was similar to the trend in the mAs used in a study within Ghana (10).
Scan length: The scan length for head trauma was longer than that for acute stroke due to the inclusion of a longer part of the neck to rule out concomitant cervical spine injury for early management to mitigate complications similar to a studies in Ghana and Uganda [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref].
The scan length for PE was shorter than that for ILD/HRCT due to the exclusion of the most peripheral chest parts in which emboli are not generally seen within the terminal pulmonary arterioles. This trend was similar to that in a study within Ghana [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. This trend of scan length was largely within the recommended limits of the anatomical extent for PE examinations which extends from the aortic arch to the base of the hear (26).
The scan length for UC was shorter than that for ABDPL probably due the focus on calculi within the upper urinary system as per the scan request form with less inclusion of the most distal parts of the pelvis after the urinary bladder This was dissimilar to ndings in a study within Ghana [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] in which both ABDPL and kidney stones had similar scan lengths. This was also dissimilar to the recommendation by ESR that advises the scan length for UC to extend longer than that of ABDPL (appendicitis) i.e., from inferior margin of T10 to lower edge urinary bladder (approximately at lower edge of pubis symphysis) for UC and from inferior margin of T10 to superior border of pubis symphysis (26).
Scan length depends on the height of the participant and the extent of the anatomy that has to be demonstrated therefore it is important to keep it within limits that answer the clinical question for the CT scan.
Number of scan sequences: Only a precontrast scan sequence was used to examine head trauma, acute stroke and ILD/HRCT as the tissues provide adequate natural contrast to allow visualization of the questioned pathology in these indications similar to a study within Ghana (10).
In the chest, the scan sequences for PE were more than for ILD/HRCT partly due to the higher image quality requirements with contrast use and partly due to the inclusion extra sequences at some hospitals for example a precontrast HRCT phase to rule COVID-19 pneumonia during the pandemic and a delayed phase to rule emboli in pulmonary veins. These ndings differed from other studies in Ghana and France that used less sequences (1-2) [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref].
In the abdomen, a higher-than-expected number of scan sequences were used for UC contrary to a single noncontrast sequence which is usually adequate as calculi provide high contrast to visualize them easily. Many postcontrast scan sequences were used in UC and ABPL mainly due to absence of standardized scan protocols optimized for these indications and these ndings differed from other studies that used one noncontrast scan sequence for UC and one postcontrast scan sequence for ABDPL in Ghana, France and Europe [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] [bib_ref] Diagnostic reference levels and median doses for common clinical indications of CT:..., Bos [/bib_ref].
There is potential to reduce CT doses by lowering the number of scan sequences for PE, ABDPL and UC through adherence to the developed indication-based examination protocols where they exist or development of the optimized indication-based scan protocols where they are absent.
Slice thickness: The acquisition slice thickness was similar for all indications across all body regions even for pulmonary embolism and HRCT/ILD which require thinner slice acquisition due to the need for higher image quality. The slice thickness for indications in the head(head trauma and acute stroke) and for abdominopelvic indications (ABPL and UC) was expectedly comparable as the indications in both regions do not necessarily require very thin slice thickness to have diagnostic images similar to ndings in a study within Ghana [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. The slice thickness nding for PE differed from ndings of a study within Ghana in which the slice thickness for PE was lower than that of other indications in the chest region [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. It is important to note that image quality in the current study was assessed subjectively and was adequate for PE similar to the high image quality in the cited study [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] that assessed the quality objectively. There is need to assess image quality objectively in future studies to better assess if a slice thickness greater than 1mm provides good enough image quality for PE assessment.
## The typical drls
The CTDIvol DRLs were observed to be comparable for different indications within the same body regions similar to a study in Ghana [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. The CTDIvol DRLs in the head region (acute stroke and head trauma) and abdomen region (abdominopelvic lesion and urinary calculi) were similar due to use of similar CT scan parameters (effective mAs). The CTDIvol DRLs in the indications within the chest region (ILD and PE) ended up being comparable due to use of a combination of CT scan parameters that raise and reduce CTDIvol i.e., in ILD, the lower pitch raises the CTDIvol while the lower effective mAs reduces the CTDIvol while in PE, the higher effective mAs raises the CTDIvol while the higher pitch reduces the CTDIvol.
The DLP DRLs were observed to differ among indications with the same body region due to the differences in image quality needs and use of different scan parameters. The DLP DRL for head trauma being higher than for acute stroke can be explained by a longer scan length due to the need to rule out cervical spine injury similar to a study within Ghana [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. The DLP DRL for PE being higher than for ILD/HRCT can be explained by the higher image quality need that requires contrast use and therefore a higher total mAs, plus the use of a higher-than-expected number of scan sequences. This trend of CT doses within the chest ndings were similar those in a study within France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref]. The DLP DRL for UC being unexpectedly higher to that for ABDPL can be explained by the absence or less frequent utilization of an optimized scan protocol for UC which allowed room for use of postcontrast sequences which were high in number than those for ABDPL. This nding was different from that in other studies in which the DLP DRL for UC was lower than for ABDPL due to use of a single noncontrast scan sequence for UC in an optimized protocol [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref] [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref].
Comparison of the IB-DRLs to anatomical based national DRLs (AB-NDRLs) in Uganda.
The CTDIvol DRLs of all indications being lower than corresponding anatomical based values was probably due to use of a lower total mAs. The IB-DLP DRLs for head trauma, acute stroke and ILD/HRCT being lower than AB-DLP DRLs for head CT and chest CT scans respectively can be explained mainly using a lower total mAs. The need for lower image quality requirements without need for contrast media may have further contributed to the DLP DRLs of acute stroke, head trauma and ILD/HRCT being lower. Some of the current study's indication based DLP DRLs were lower than DRLs of an anatomical region by 36.4% on average similar to ndings in a studies within Finland (28) and Switzerland (29) that found IB-DRLs lower by 20% and by 21-32% respectively.
However, the DLP DRL for PE was higher than for chest CT scans due to use of thinner acquisition slice thickness and generally a higher need for higher image quality similar to another study in Switzerland [bib_ref] Local clinical diagnostic reference levels for chest and abdomen CT examinations in..., Brat [/bib_ref].
The DLP DRL for ABDPL in the current study ended up being comparable to the DLP value for abdomen CT scans [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref] due to a combination of CT scan parameters that raise the DLP including a longer scan length of 49.12 (46.29-53.09)cm and a thinner slice thickness of 3 (1-5) mm in ABDPL examinations compared to a scan length of 41.25 (32.0-63.3) cm and to a slice thickness of 3.75mm slice thickness that were used in abdomen CT scans [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref] , and due to use of a lower total mAs of 4783 (3496-7088) mAs which reduces the DLP compared to 6115.5 (3619-9869) mAs of abdomen CT scans [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref]. The nding of the DLP DRL of an ABDPL being comparable to the DLP of abdomen CT scans (17) differed from ndings in a study within Switzerland(29) that found the DLP DRL for appendicitis to be lower than for abdomen CT scans because this cited study in Switzerland used optimized protocols for appendicitis with probably fewer scan sequences.
The DLP DRL for UC in the current study ended up being unexpectedly comparable to the high DLP value for abdomen CT scans (17) due to a combination of CT scan parameters that raise the DLP including many scan sequences that included many postcontrast phases (0-6), a longer scan length of 46.88 (42.83-49.13) cm and a thinner slice thickness of 3 (1-5) mm in UC examinations compared to 41.25 (32.0-63.3) cm scan length and to 3.75mm slice thickness that were used in abdomen CT scans [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref] , and due to use of a lower total mAs of 4888 (3324-6221) mAs which reduces the DLP compared to 6115.5 (3619-9869) mAs of abdomen CT scans [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref]. The nding of the DLP DRL of UC being comparable to the DLP of abdomen CT scans [bib_ref] Adult Computed Tomography examinations in Uganda: Towards determining the National Diagnostic Reference..., Erem [/bib_ref] differed from ndings in a study within Switzerland(29) that found the DLP DRL for kidney stones to be lower than for abdomen CT scans because this cited study in Switzerland used optimized protocols for kidney stones with probably fewer scan sequences.
Comparison of the overall IB-DRLs at 75th percentile to some of the published national IB-DRLs at 75th percentile Acute Stroke
The CTDIvol and DLP DRLs of acute stroke being much lower than values in Ghana (10)was mainly due to use of a lower effective mAs of 169-(160-203) compared to the higher tube loading of 238.0 (± 80) mAs in Ghana [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. The reasons for the CTDIvol and DLP DRLs for acute stroke being lower than France's values [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] were not ascertained because very few CT scan parameters were mentioned in the cited study within France which limited more comparative analysis.
## Head trauma
The CTDIvol and DLP DRLs of head trauma being much lower than values in Ghana (10) was mainly due to use of a lower effective mAs of 181(168-206) compared to the mAs used in Ghana (229.4 ± 73.5 mAs) [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. The reasons for the CTDIvol DRL for head trauma being lower than, and the DLP DRL for head trauma being comparable to, the corresponding values in France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] were not ascertained as very few scan parameters were mentioned in the cited study in France for more comparative analysis.
## Ild/hrct
The CTDIvol and DLP DRLs of ILD/HRCT being much lower than the values in Egypt [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref] was probably due to use of a lower effective mAs of 61.5 compared to the mAs used in Egypt within a range of (100 minimum mAs to 430 maximum mAs) [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref].
The CTDIvol and DLP DRLs for ILD/HRCT being higher than values in France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] was due to use of a higher kilovoltage of 130 compared to 100 kVp used in most examinations (61 %) the cited study within France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref].
## Pe
The CTDIvol and DLP DRLs of PE being lower than values in Ghana (10) was probably due to use of a wider slice thickness of 3 (1-5) mm compared to Ghana's (2.20 ± 1.7)mm, a lower peak tube kilovoltage of 110 (110-114.25) kVp compared to Ghana's (117.8 ± 4.0)kVp, and a lower effective mAs of 98 compared to Ghana's tube loading of (167.5 ± 92.9) mAs [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. Even though the current study's DRLs for PE were lower than Ghana's, the quality of PE examination images was of acceptable quality to make a diagnosis as assessed by radiologists.
The reason for the CTDIvol DRL of PE being lower than the value in France [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] was unknown as the cited study did not mention its scan parameters which limited comparative analysis.
The corresponding DLP DRL for PE was instead higher than France's value [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] due to use of a higher tube peak kilovoltage of 110 (110-114.25) compared to 100kVp used in most examinations (55%) in the French study [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] , probably due to frequent use of unoptimized examination protocols with a higher number of scan sequences [bib_ref] Projected cancer risks from computed tomographic scans performed in the United States..., De González [/bib_ref] and probably due to a longer scan length compared to the usually optimized PE protocols in European countries(26).
## Abdpl
The CTDIvol DRL being lower than Ghana's value (10) was due to using a lower effective mAs of 98(81.33) compared to Ghana's tube loading mAs of (137.0 ± 91.4)(10).
In a comparison to the cited study in Ghana (10), it was observed that most of the scan parameters used in the current study to examine an ABDPL were higher than those in Ghana for example, the current study used a higher number of postcontrast scan sequences(1-3), longer scan length 49.12 (46.29-53.09)cm, higher kilovoltage(130kVp), and thinner slice thickness 3 (1-5)mm compared to Ghana's single postcontrast scan sequence(1), 45.99(±4.3)cm scan length, 118.7(±7.5)kVp and 5.40(±2.6)mm slice thickness respectively [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref]. For this reason, the DLP DRL of ABDPL in the current study was expected to be much higher than Ghana's value but was instead comparable due to use of a lower effective mAs of 98 compared to Ghana's tube loading mAs of (137.0 ± 91.4)(10).
The reason for the CTDIvol DRL for ABDPL being comparable to France's value [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] was not ascertained as the cited study did not mention much about its scan parameters which limited comparative analysis .
The corresponding DLP DRL for ABDPL being higher than France's value [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] was due to use of a higher tube peak kilovoltage of 130 compared to 100/120 kVp used for most examinations (59%) in the French study [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] and probably due to frequent use of unoptimized examination protocols with a higher number of scan sequences (2-4) compared to the usually optimized protocols for ABDPL in European countries(26).
## Uc
The CTDIvol DRL of UC being lower than Ghana's value (10) was due to use of a lower effective mAs of 105 compared to Ghana's (138.4 ± 98.4) mAs [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref].
The DLP DRL for UC being higher than Ghana's value was probably due to use of a higher kVp (130), thinner slice thickness 3 (1-5) mm and a higher total number of (1-7) scan sequences compared to Ghana's 118(±8.3) kVp, 5.3(±2.5) mm and only one noncontrast scan sequence in the examination respectively (10).
The reason for the CTDIvol DRL of UC being comparable to France's value [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] was not ascertained because the French study did not mention much about their CT scan parameters which limited comparative analysis.
The DLP DRL for UC being higher than France's value [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] was due to use of a higher tube peak kilovoltage of 130 compared to 100 kVp used in most examinations (69%) in this cited French study [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref] and probably due to frequent use of unoptimized examination protocols with a higher number of scan sequences (1-7) which included (0-6) postcontrast sequences compared to the optimized UC protocol with only one noncontrasted scan sequence in the French study [bib_ref] Patient dose evaluation in computed tomography: a French national study based on..., Geryes [/bib_ref].
Like in other published studies, the overall DRLs of some indications at 75th P varied signi cantly from national IB-DRLs in other countries mainly due to the difference in the scan parameters chosen for use in the examinations including kilovoltage, acquisition slice thickness, effective mAs and number of scan sequences plus the use of protocols that are less optimized for indications [bib_ref] National indication-based diagnostic reference level values in computed tomography: Preliminary results from..., Botwe [/bib_ref] [bib_ref] Establishing national diagnostic reference levels (DRLs) for computed tomography in Egypt, Salama [/bib_ref] [bib_ref] Diagnostic reference levels and median doses for common clinical indications of CT:..., Bos [/bib_ref] [bib_ref] International variation in radiation dose for computed tomography examinations: prospective cohort study, Smith-Bindman [/bib_ref].
## Study strengths and limitations
The strengths of this study include having performed QC tests on the CT scanners, use of actual CT scanner output radiation doses to develop the IB-DRLs following corrections of the doses and the development of DRLs using the required minimum of 20 participants per CT scanner room for most indications as recommended by ICRP.
The main study limitation was having selected three (3) out of 13 (23%) CT facilities in Kampala and central region of Uganda, the developed typical IB-DRLs may be less representative of the doses and practices used during examinations of the selected CT indications country-wide. However, ndings from the study still give an indication that locally determining typical DRLs can optimize the use of CT equipment. Further studies should be conducted using more CT facilities to develop IB-DRLs for the common CT indications that are more representative of other settings.
# Conclusion
The typical IB-DRLs determined in this study were 30.17mGy and 653mGy.cm for acute stroke, 32.04mGy and 878mGy.cm for head trauma, 4.66mGy and 161mGy.cm for interstitial lung diseases/ high resolution chest CT scan, 5.03mGy and 273mGy.cm for pulmonary embolism, 6.93mGy and 838mGy.cm for abdominopelvic lesion and 7.61mGy and 975mGy.cm for urinary calculi.
The developed typical IB-DRLs are recommended for use to optimize CT radiation doses among adults. Most of the developed typical IB-DLP DRLs were lower or comparable to DRLs from studies in Ghana and Egypt while they were higher than DRLs in the French study due to differences in selection of CT scan parameters. Standardized indication-based examination protocols for the common CT indications should be developed for indications where they do not exist, and their use strengthened to minimize variation in the DRLs in comparison to international DRL values. This can also optimize the use of CT equipment especially in low resource settings where it is not yet widely available. Declarations Ethics approval and consent to participate: Ethical clearance was obtained from the Makerere University School of Medicine Research and Ethics Committee (SOMREC) (Reference number: Mak-SOMREC-2021-233). Administrative clearance was also obtained from the administrations of Hospital A, Hospital B and Hospital C. The actual names of the participating hospitals have not been declared on their request. Informed consent was obtained from participants who had been referred by medical workers for a CT examination for their CT dose data to be used in the study. The study was also carried out following relevant guidelines and regulations according to the Helsinki declaration.
## List of abbreviations
## Consent for publication: not applicable
Availability of data and materials: All the necessary original data and materials used in this study have been included. Comparison of the current study's overall IB-CTDIvol DRLs at 75th P to published national IB-CTDIvol DRLs at 75th P
[fig] FiguresFigure 1: Research reported in this publication was supported by the Fogarty International Center of the National Institutes of Health, US Department of State's O ce of the US Global AIDS Coordinator and Health Diplomacy (S/GAC), and President's Emergency Plan for AIDS Relief (PEPFAR) under Award Number 1 R25TW011213. The content is solely the responsibility of the authors and does not necessarily represent the o cial views of the National Institutes of Health. Authors' contributions: K.N, E.N, J.M.K.M, F.K, G.E and H.K contributed to proposal development, research methodology, manuscript writing and reviewing manuscript. E.N, K.N and J.M.K.M analyzed the data. K.H was the overall supervisor in the study. A.G.M and the [/fig]
[table] Table 1: CT examinations performed among adult patients and their indications at ve hospitals in Kampala during a survey. [/table]
[table] Table 2: Distribution of sample population used to determine CT doses by age, weight, and gender in each indication [/table]
[table] Table 3: Characteristics of the CT Scanners [/table]
[table] Table 4: Scanning parameters used to the acquire CT images per indication For number of scan sequences, the topogram sequence and monitoring phases (e.g., in the cases of PE) were not included as it is the situation for some established DRLs. Mn-Minimum, Mx-Maximum. [/table]
[table] Table 5: The typical DRLs at 50th percentile and the overall DRLs at the 75th percentile. [/table]
[table] Table 6: Comparison of the developed overall IB-DRLs at 75th P with national anatomical DRLs in Uganda at 75th percentile (Units for CTDIvol are mGy and for tDLP are mGy.cm) [/table]
|
10.1590/S1806-83242010000500009
|
CCBY
| null |
20857076
|
s2orc_pubmed_articles
|
Promotion of Oral Health in the public and private context.
Symposium DiscussionsDrDrDrDrDrSymposium Discussions
031CB7A0044973048351FC27BA54AC0B
Good morning to all and thank you for coming.The purpose of this symposium is to provide a forum for the discussion of the topics presented during our event.Participant: Prof. Cassius, why is it so difficult to put the health promotion policies, such as those dealing with oral health and tobacco smoking cessation, into practice?Cassius Torres: Generally speaking, people are familiar with the health determinants.As Prof. Kowalski mentioned during his presentation, there are various reasons why people do not seek a professional when they have a problem: "I was foolish not to go to the doctor" or they think "it won't happen to me".Patients cannot be dealt with in a simplistic manner, in terms of oral cavity prevention, nor can we disregard their context or environment.Our approach must be professional, but this is not usually a relationship we invest too much in.Dealing with this duality, which is part of what is so difficult about working in the health sector, means taking the person as a whole into consideration.It is important to understand if the environment the person lives in puts them at risk, what is their lifestyle and if there is any way to promote change.It is also necessary to understand that at times it is impossible to convince a person to change their lives because they need to make the decision themselves.We need to understand the addiction mechanisms in order to stage an intervention, and we are not trained to do this.We need to understand that addicts, whether they are addicted to tobacco or crack, are solely responsible for the decision to quit their habit, and sometimes the solution to the problem is out of our hands.Participant: Is it true that the two oral health indicators were taken out of the 2010 health pact?How are the issues of the first appointment and the collective procedures being dealt with?And how does the Ministry of Health see the sustainability of government policy in transforming these actions into public policies?Professor Nilce, how can social control mechanisms support our initiatives?How can we act as a mechanism for social control in health, whether within institutions, health councils or social movements?For example, how can we ensure universal access when there is such a large discrepancy among health professionals, when dentists are outnumbered by doctors?And lastly, taking the case of oral health, how are we going to make sure that oral health promotion is included in the national policy for the promotion of health?
Gilberto Pucca: We currently do not have any agreement upon oral health indicators.In 2003, when the Smiling Brazil ("Brasil Sorridente") program was launched, we agreed upon two indicators: supervised brushing and the first appointment.It is worth noting that the purpose of indicators is to monitor the impact of the implemented measures and to follow up on the results achieved, the outputs and effects.In dentistry, however, there is no single indicator that is capable of summing up the care actions (those performed directly between the professional and the patient) and promotion actions (those with an educational character).That is why we have chosen the indicator of supervised brushing in order to monitor the health promotion actions.With the second indicator, our purpose was to measure the care provided, the clinical act itself, as well as the behavior inside the office, and that is why we chose the first appointment.This indicator was chosen based on the rationale that the patient would be registered in our system every time they came in on a planned basis.This same patient would not be registered, therefore, if the appointment was an emergency one.We have invested in planned actions that must be interdependent in order to have an epidemiological impact.It is critical that the service be planned, rather than be based on the model in which the dentist waits for the patient in his office.There is a need for the professional to be part of a team effort.The first appointment indicator is intended to reflect entry in a planned manner.In Brazil, we have a long tradition of providing pre-natal care and individual care in an unplanned way.There are a myriad of indicators for the health fields in which the indicators are agreed upon in a tripartite way (Unified Health System -SUS, State and Municipalities) and there is a movement within the Ministry of Health to clean them up because, with so many indicators, we have lost sight of quality control.Thus, in the negotiation process, it was decided that there would be only one indicator for the oral health field.Since the Ministry of Health was investing in the implementation of new services and in setting up a care network, we made the decision to include the care indicator in order to measure the pace at which the network was growing.The logic behind the orga-nization of the services emphasizes the Oral Health Teams, but they should work within the basic care guidelines and that is not always the case.Setting up an Oral Health Team (OHT) is the responsibility of the municipality.The Ministry of Health transfers the funds, and the municipality then decides whether or not they will set up an OHT, but this does not imply the reorganization of basic care, which is the most important aspect.The fact that OHTs are implemented in the sphere of the family health strategies does not automatically entail planning; that is to say, the municipalities have the autonomy to reorganize the basic care system.In the view of the National Council of the Health Secretariats (CONASS) and the National Council of Municipal Health Secretariats (CONASEMS), the oral health indicator should be the number of OHTs, but I do not share that view.Given the salutary tension that took root, it was not possible to include the oral health indicator, but we will renegotiate this aspect and we will probably include the number of teams, but I believe we still have to further discuss what it means to promote health.
Nilce Tomita: In regard to the issue of social control, we need to understand what the people perceive to be their needs, their own view on their oral health and what responses the health system must provide in light of these.Another important issue is to understand the historical reasons that led to the discrepancy between the number of professionals in a certain sector -medicine -and that observed in another sector -dentistry.There are historical reasons why the Oral Health Teams turned to a traditional care model without incorporating the precepts of Oral Health Care.However, more important than statistics is the issue of work organization.I can imagine that there is a corresponding number of OHAs (oral health auxiliaries) and OHTCs (oral health technicians) working alongside these dentists and that they operate according to the logic of the Oral Health Teams by reorganizing the care provided, offering high-quality services and with a coverage in proportion to the number of professionals.In regard to formal social control, the role of the councils is critical.It is important to make sure dentists participate in these councils in order to advocate for the issues involving oral health, with epidemiological indicators that portray the health of the population in view of the services available.As far as the means of "informal" social control, we have the media and the press which provide a forum for the users to be heard, and translate the demands of society in order to induce discussion.
Rui Oppermann: In reference to the issue that was raised about the discrepancy between the two categories, I believe that the central aspect is to make the dentist work in a productive way, which is something that entails knowing how to delegate tasks, the ability to plan and to work in a goal-oriented way.In public health, we need more doctors than we need dentists.I am not undermining the importance of the field of dental care, but I am looking at it from a different perspective.
Marisa Maltz: Dr. Pucca, until now, we have only been discussing the service indicators.But how is the discussion about problem resolution being carried out in the Ministry?Wouldn't it be more appropriate to work with indicators that emphasize the short-term rather than to take stock of the situation in ten-year increments?Wouldn't this allow us to better assess the service?
Gilberto Pucca: Indicators are a powerful and indispensable evaluation tool and they are useful in evaluating policies, in particular public policies.But the indicators need to be considered in light of the Brazilian perspective because they measure trends, and that is something we should bear in mind.We live in a country comprised by 26 states and one Federal District and which has over 5,000 municipalities.We cannot expect indicators to paint an exact picture of reality.Which trends are we going to measure?There are indicators specifically designed for each of the distinct levels.We need to design a set of indicators that will provide information on the state and municipal systems.Since there is no indicator pertaining to oral health in the SIAB (Basic Care Information System), we came up with a file that has recently undergone public scrutiny and that we hope will be available by the end of this year.
Lilian Marly de Paula: Is there any proposal currently under consideration by the Basic Care Secretariat for a National Network for Research in Ba-sic Care along the same lines as the National System of Clinical Research?I would also like to note that an oral health indicator serves no purpose if it is not linked to the basic care system.
Gilberto Pucca: There is.This is another aspect in which we are making significant progress, which is the production of knowledge in oral health policy.However, we need the funds to accomplish our goals.The Ministry of Health, by means of the Oral Health Coordination, is sponsoring research in Oral Health which was previously funded exclusively by the CNPq (National Council for Scientific and Technological Development).There have been two grant calls already and we already have plans for a third one, and the oral health indicators are one of the main topics for funding.
Participant: I am concerned about the prevalence of oral cancer among patients that had a bone marrow transplant for one reason or another.Therefore, I would like to know what is the prevalence exactly, and for how long should we monitor these patients?
Cassius Torres: I don't have an exact number to give you.But a depressed immunological function is a risk factor for a malignant neoplasm.The patient, who is already vulnerable due to a bloodrelated malignant neoplasm, is submitted to aggressive treatment (radiotherapy, chemotherapy and bone marrow transplant).These are highly relevant chronic factors.The first bone marrow transplant in Latin America was performed in Curitiba in 1979.In our case history, we have the patients with the highest survival rates, but we do not have the data that could provide an answer to your question.
Participant: One of the greatest problems we have been dealing with is the absence of a computerized system to record patient information and to keep track of the procedures performed.This has led to data loss, difficulties in scheduling follow-up appointments and problems in gathering epidemiological data.Is there any effort currently undertaken to computerize the basic care units and the hospitals that provide oral care?Also, in your opinion, what has yet to be achieved in the oral health sector?
Gilberto Pucca: The Oral health Policy reflects how a society is organized and which are its pri-orities.As an example of this relationship we have the prison health policy, for which we are already setting up multi-professional teams comprised of doctors, nurses, dentists and assistants, and that is co-funded by the Ministries of Health and Justice.The health of the native population is also another complex problem.We have approximately 500,000 Indians living in our country, and the provision of health services to them is difficult because there is no central command.What we have is a great number of NGOs (non-governmental organizations) and therefore the governability of Indian health is low.We created a Secretariat for the Health of Indians in the Secretariat for Healthcare, organizing it according to the SUS logic, because it was a parallel system.In regard to oral cancer, it was made one of our priorities.And in the CEOs (Centers for Dental Specialties), one of the areas that must be necessarily implemented is stomatology for oral diagnosis.Another important advance was offering dental care services in the CACONS (High Complexity Oncologic Care Centers) and UNACONS (High Complexity Oncologic Care Units).We are on the verge of finalizing a consensual National Oral Cancer Policy within the Ministry of Health and with the participation of the INCA (National Cancer Institute).It is the first step to effectively deal with oral cancer, but we have great potential given that we are putting together a network.What we now have to do is organize the service.There is an effort to computerize, starting with the CEOs, all of the forty thousand basic healthcare units throughout the country, which is a highly complex undertaking because they are spread out.
Nilce Tomita: Even though there have been significant and worthy accomplishments, we need to provide an answer to a highly challenging question: "What have we yet to achieve?"Despite the growing numbers, we have not yet reorganized the work process of the dentist within a team comprised of at least one oral health auxiliary and one oral health technician.There is limited availability of technicallevel professionals to work in the oral health teams.Given this strategy, but also beyond it, how can we bring together the efforts of the Oral Health Coordination and the SGTES (Secretariat for Labor & Education Management in the Health System) in order to train these professionals?
Lilian Marly de Paula: In addition to the problem of the OHA and the OHTC, in organizing the Brazilian Unified Health System (SUS), we also have to deal with the issue of the future professionals and the reallocation of the professionals that are currently in the network.Despite the efforts of the Oral Health Coordination and the SGTES (just remember the PRÓ-SAÚDE model) to show the universities that their curricula should be changed, we still have a long way to go before we are able to achieve these goals.This is not an effort that will yield results in the short-term, but rather it is contingent on the local partnerships and on the pressure exerted by the professionals for a higher quality service.The management is not compatible with the current model and the public network professionals work as if they were in the private sector.This is an effort that is the responsibility of the health professionals.Nothing will happen if the professional does not believe in what he is doing.We have to rethink the professionals in the network and this is something that can only be done internally and locally.
Cassius Torres: In some municipalities, this movement has already taken hold.When we understand how these actions are organized, we can see that a good manager in Brazil today is capable of grasping how the system was devised and set up, and immediately putting into practice very interesting actions, even though for the oral health public policy we lack solid evidence to make certain decisions.In regard to self-examination, I consider it to be a questionable strategy: is it worthwhile to go out prematurely, without sufficient evidence and prepare instructional material to generate cases?But I know it is possible, with a family health strategy, to plan actions, and even with a rudimentary resource such as the SIAB (Basic Care Information System) we can identify the target audience to promote an active search.Sometimes I think there is a disproportionate expectation in regard to what the Ministry should provide.About the supplementary health program, until the second to last review of the list of procedures, there was no charge for the anatomic pathology tests.There are subtle modifications.
Participant: In reference to the promotion of health in the private context, it is the first time we have brought up the ANS (National Supplementary Health Agency).There are health providers that operate under the philosophy of health promotion and there are providers that have always covered anatomic pathology tests.I would like to know if the private sector acknowledges all of the actions of the Ministry of Health developed with the intention of improving oral health, however, with the understanding that the new list of procedures actually entails a transfer of responsibilities.When patients cannot solve their health problems within the public sector, they subscribe to health plans offered by the health providers and the ANS is passing along to them what would actually fall under the responsibility of the SUS.The 1988 Constitution dictates, in a way, that the promotion, the recovery and the rehabilitation of patients is the responsibility of the SUS.Prosthetics are part of the rehabilitation process and it is a way of reintegrating an individual into society, but this responsibility is being passed along to the providers.Is that advisable?It is.But I would like to know what role the Ministry of Health had in arriving at this solution, whether the Ministry of Health had any influence or if this was solely an initiative taken by the National Supplementary Health Agency.
Gilberto Pucca: On the topic of the training of OHTCs and OHAs, beginning in 2003, the Secretariat for Labor & Education Management in the Health System was created.It was created as a result of an understanding in the Ministry of Health regarding the need to improve training and to invest in this field.But the discussion surrounding the topic of training is one of the greatest challenges.We have to attempt to bring the academic world and the realm of knowledge closer to the world of service provision.This is a daily challenge we face.And the speed at which the construction and incorporation of new knowledge is made by academics is not enough to meet the needs of the service sector.The service sector is more pragmatic and has to provide answers at a faster pace, but in training, the process is undertaken with a distinct logic.This has to be respected, of course, but it also has to be called into question so that both sides start moving at equal paces, or at least at less unequal paces.The SGTES, by means of the PRÓ-SAÚDE, has invested in a curricular reform.To review curricular guidelines is a daunting task.It is like the old story of arriving at a consensus.Everybody agrees in theory, but when the time comes to put the plans into action, it is a different story, but we are dealing with it.We are setting up an oral health policy very quickly, but we do not have professionals with adequate training to act within it.It is complicated, but we cannot expect the universities to change their curricula to fit the strategy.Therefore, it is a thorny problem that has to be dealt with in light of the Brazilian perspective and characteristics.This is the challenge we must face on a daily basis.In regard to the issues pertaining to the ANS, I think that the procedure list is an advance taking into consideration the situation that existed previously.But I fully agree with you.The Constitution says that the private sector is complementary.This must be taken into consideration and everything must be built within this logic, but we have an enormous historical deficit.I am aware that the complementary sector does not fall under the responsibility of the Ministry of Health, but rather of the Ministry of Justice, and not in the punitive sense, but in the sense that they must decide on these issues instead of exempting the government from the investments it is obligated to make.The private sector has an important role to fulfill, which is complementary and that cannot prevail over the public sector, much less in the realm of regulation, which is the sole responsibility of the public sector.Lastly, I would like to ask Rui to comment on the recently published statement that people who do not brush their teeth at least two times a day are 70% more likely to suffer a heart attack.
Rui Oppermann: Since most of the population brushes their teeth at least three times a day, we are outside the risk zone.We should be cautious in relation to the information we derive from the epidemiological analyses.It has been a while now that the theory that those people who do not floss are more likely to suffer a myocardial infarction has been established.But what is relevant is the fact that the person who flosses is likely to have other associated healthy habits and so the indicator is not the cause.Epidemiological associations are rarely causal.In order to determine causality, we need a randomized clinical trial including everything else that is necessary to provide evidence.The second aspect is that the associations also have to be relativized.I am not familiar with the odds ratio of brushing.If it is seventy percent, then the association is 1.7.Professor Kowalsky shared with us a host of associations between eating habits and head and throat cancer.I cannot go into details, but the associations are statistically very weak.In a 1.7 association, the seven stands for the seventy percent risk that the epidemiologist claims exists.But the bottom line is that for an association of this nature to have any chance to affect the outcome, such as cardiac events, the odds ratio should be higher than 3.And above 3, in terms of cardiac events, is found exclusively with smoking, sedentary lifestyle, serological lipids count and diabetes.People who brush their teeth religiously are sure to take better care of themselves and are more concerned about their health, and therefore will also take care of their heart.In regard to the topic of the curriculum and the PRÓ-SAÚDE, I would like to mention that we were in the process of reviewing the curricular guidelines in our school when the PRÓ-SAÚDE was launched.So we did not change our curriculum because of the PRÓ-SAÚDE program, but rather PRÓ-SAÚDE came to help us by providing us with resources.The first PRÓ-SAÚDE referred to the academic units, but the PRÓ-SAÚDE II was institutional: the university submitted a proposal for a teacher-centered care district that is providing care to almost 600 thousand people in the city of Porto Alegre.This changes the situation completely because the health system as a whole is committed.In addition to this, it became very clear to us that there was a need to encourage the network work closely with the service providing sector.And that is where the PET-SAÚDE came in.This program provides a stipend to students and teachers and to one dentist in the network, the so-called preceptor.This is an important aspect because it serves as motivation.In August we will open a dentistry course that will be offered at night, and I believe it will be the first in the federal university system.This is a course with very evident social connotation.We received a generous budget from the Ministry of Education and we are waiting for the resources from the Ministry of Health.But now we need funding for this network.There is a considerable issue, which is that, as Cassius described, we are a federation that is highly dependent on the federal sphere of government.But the government has its own policies and its own way of stimulating the discussion of certain issues.However, in regard to the implementation of certain health policies, we have to deal with the municipaland state-level policies in the state of Rio Grande do Sul.Even though we would like to work with the national-level issues, the local policy determines what will happen.In the city of Porto Alegre, the number of oral health teams in the family health program is minimal, and the dentistry schools focus their curricula on the PSF (Family Health Program).It is therefore necessary to look at the issue as a whole.We have the important federal policies, which are provided in the Constitution.If we don't have the corresponding municipal and state policies that allow us to apply the federal policies, a very difficult situation is created.Take the example of the state of Paraná, for instance.Despite the several governments in power, they have been able to deal with the issue in a quick, dynamic way that is more attuned to the issues that are important to the people.We also have to act locally in this direction.
Marisa Maltz: I believe that this issue of the relationship between the three spheres of government is also very important when we consider the resources for research.The serious problem of academic research conducted with the service provision sector is that the researcher is considered the sole responsible for the research outcomes.The network has to be responsible and must be evaluated in the use made of these resources and for the final outcome of the research in the same degree as its academic partners.
Participant: In the recent past, in a daily practice situation, the access indicator used was the first dental appointment.When a patient is submitted to a full dental treatment, several appointments are required to complete the treatment because he has many previously unattended needs to be filled.And if this patient is seen several times by the dentist, then the access indicators will be lower.But if we distribute tickets, the patients will be entitled to one or two appointments only, and then they will have to take another ticket in order to continue their treatment.And so the access indicators in this context tend to be higher.What is the best solution in this case?And how should we deal with this in terms of risk groups?
Marisa Maltz: I would like to add to your question.In the state of Rio Grande do Sul, during a study involving academic and service sectors, it was found that there was a high number of follow-up appointments, which gave rise to an order establishing that each patient could be seen at most 3 times.
Gilberto Pucca: This has to do with what is considered to be important, at the central level of the SUS, to the Ministry of Health.What you have stated is pertinent, but we have to think about this issue at the local level.Because at the level of the Ministry of Health, what we need to measure and understand about public policies is the level of access.For example, how many people had access to the service?Going into this level of detail is strategic and we need an indicator that will be discussed at the local level.That is why it is important, in light of these indicators that we will have or already have, that the States, Municipalities and Federal government agree upon oral health indicators that are clear for the local sphere of management.The problem with this first appointment indicator that began in 2003 is that it was not clear.A patient would present himself as being a first-time patient, but that was not exactly the case.Our first effort was to redefine this indicator.
Participant: Wouldn't it be wise to appoint auditors at least in the capital cities in order to monitor the application of the resources granted to the units located in the States and Municipalities?How is the spending by the units monitored?
Gilberto Pucca: Yes, I agree that it is indeed important.Civil society must take ownership of the monitoring mechanisms.The municipal councils are one of the most salient instances of social control.But it is a new experience for oral health as well.Three years ago, we began a project with DENASUS (National Department of Auditing of the Unified Health System) to create an audit instrument rather than a monitoring instrument, and last year we audited 20 Brazilian municipalities drawn by a lottery system.It is still incipient but it is already working.As a result of this growth we need auditors, whether they are dentists or not, in order to audit the three spheres: State, Federal and Municipal.
Participant: How do you view the design, management and implementation of the SB Brazil 2010 [National Oral Health Survey]?The backdrop to this question is the following: in the SB Brazil 2003, 50 municipalities were listed by macro regions and drawn by a lottery.But the capitals were not; they were placed in context.And when they are not drawn, we violate the basic principle of biostatistics by which each basic care unit must have the possibility of being drawn and should be listed in a group of municipalities.Since this was not done, the biostatistical methods that should have been applied were not.What is now being done in Brazil 2010 to change this?The second question has to do with the oral health of the Indian populations.We always deal with the oral health of Indian populations as a sub-system of oral health.I would actually like to make a request because we have 34 sanitary districts for the Indian populations and little under 500 thousand Indians in their own communities and, according to data from the IBGE (Brazilian Institute of Geography and Statistics), there are slightly over two hundred thousand Indians in urban areas and we have a healthcare system for Indians that includes a module of oral health which has been operational since 2007 and, yet, this population is invisible from the epidemiological point of view.
Gilberto Pucca: The SB 2010 is different.What is important is that we, oral health professionals, are building an epidemiological survey from the SUS perspective and not from the perspective of the Oral Health National Coordination.The SB 2010, ever since it was first conceived, was based on the close interaction between the Municipalities and the States.To be clear, it is the responsibility of the SUS.In the Municipalities, the stakeholders are discussing the SB, they would like to expand it.It is important not only for dentists but also for health person-nel in general.The idea is that we would be able to include the national epidemiological surveys within the framework of health surveillance policies.The survey and the SB would then become instruments and components of the National Health Policy, which is the health policy at the national level that has an oral health component.It would therefore be seen under a different light.It would become indeed a policy for health surveillance and that, in my view, is a breakthrough.We reexamined the sample and the methodology.The SB 2003 itself was submitted to criticism, which is also good, and was also the topic of discussions and scientific papers.It was a process by which it earned its stripes.The SB, in the Brazilian model, includes over 110,000 people.I don't think there is an epidemiological survey along the same lines anywhere else in the world.And we are talking about technology.We have been perfecting it.For the SB 2010, we have put together an advisory board for the sample.This sample was reviewed and redefined.
Cassius Torres: For all of the capital cities to be included, which was indeed a precondition, the sample calculation was changed as well as the lottery principle for the other municipalities.Therefore, the lottery system is drawing many more municipalities than would otherwise be necessary.
Participant: Will all of the SB 2010 data be available?
Rui Oppermann: The government service provision sector and the people responsible for the implementation of the oral health policies have expressed an interest in using the survey as a guide for policies and for their evaluation.There is also an academic interest in this data, not necessarily for the purpose of public health, but to know, for example, if two brushings a day aid in prevention.There is an academic interest in the SB database so we can also derive other types of associations that are not necessarily of immediate interest to the manager, but which are of academic interest.As an academic and a researcher, I would like to be granted access to this database in order to identify a series of correlations and associations that are of value to further the scientific knowledge about the distribution of oral diseases among the population.This has been confirmed.
Participant: 150 municipalities were drawn in the lottery but the 27 capitals did not take part in the lottery, so we have two reference points: one at the macro regional level, in which we drew 30 municipalities per macro region.This allows us to make two inferences: one macro regional inference and other inferences for the capitals.
Nilce Tomita: I would like to congratulate you, Dr. Pucca, for setting up the centers in the Northeastern part of Brazil as you described yesterday during your lecture and that work with these indicators and in oral health surveillance.It is both important and necessary in our country.I must say that I was a calibration instructor in a municipality in the State of São Paulo and that I have witnessed what takes place in certain municipalities, where the patient files indicate that fewer examinations have been performed than the target.It seems to me that this has also occurred in other states.If the sample is smaller than the one estimated according to the parameters listed in this first draw, what will be the political decision in regard to the survey?
Gilberto Pucca: Everything that has to do with the sample, especially the aspects you have raised, really does occur in certain places.These situations, when they are found in practice, will have to be dealt with and there is no way around it.This has already been included in our planning agenda.If it has an impact on the final object, it will have to be reviewed.
Marisa Maltz: But, Dr. Pucca, in the survey plan, you also include the non-answer, which must already have been computed.
Gilberto Pucca: Yes, it has already been computed.
Participant: What are the perspectives, the possibilities and the difficulties in developing a vaccine against dental caries?Marisa Maltz: A vaccine against dental caries was widely discussed during the 80s.The prevalence of dental decay was drastically reduced during the 70s, 80s and 90s.Research indicates that given the lower prevalence of dental decay and the possibilities to control it by other means, a vaccine would not have a significant impact on its epidemiological control.
Participant: Given the explanation that was given about the link between mouthwash and oral cancer, I would like to know what guidance the panel has to give about the use of mouthwash.
Rui Oppermann: Both Dr. Kowalsky and Dr. Sheila broached the topic of alcohol-based mouthwashes.There is a wide array of oral antiseptic solutions currently available on the market that, when used in conjunction with brushing, are highly effective in reducing biofilm and gingivitis.We recommend that certain patients use mouthwashes as a means to aid in oral hygiene, especially those people with recurrent gingivitis and that cannot control the problem with brushing alone.The issue with alcohol was made clear by Dr. Kowalsky, who said that there is no scientific evidence that alcohol-based mouthwashes have a role in oral cancer.I do not agree with Dr. Kowalsky's recommendation, however, in the sense that if there is a lack of evidence, then it is better not to recommend its use.One study established a link between rinsing and head and throat cancers, in particular oral cancer.This study showed that people with cancer declared that they rinsed more frequently than those people who did not have cancer.Nevertheless, Kowalsky himself later demonstrated that when people have oral ulcers they tend to rinse their mouths with homemade tea or an over-the-counter drug.It is not surprising, therefore, that the correlation was made.Secondly, this study did not infer whether the mouthwash was alcohol-based or not.Another study performed a meta-analysis of 1500 published works and concluded that there is not sufficient evidence of a correlation between rinsing with an alcohol-based mouthwash and oral cancer.I reaffirm that the fact that there is no evidence does not mean that the correlation does not exist.We will wait for this evidence to be produced.In the meantime, we should make use of these mouthwashes that are recommended by the dental health organizations.The American Dental Association, for example, has been recommending its use for over one hundred years.Therefore, I believe we have a resource that can be clinically used to improve the hygiene standards when there is a problem with gingivitis, and, when indeed gingivi-tis is a problem, it can be recommended.There are signs that point to the fact that there might be a correlation, but the researchers need to produce better evidence of this association.
Cassius Torres: If you want a guide for clinical conduct, the recommendation is to always recommend its use according to the clinical symptoms, and the treatment should last long enough to help resolve the patient's issues.In the case of gingivitis, it should be used as long as the problem is present and in association with other therapies.I have seen patients who overuse antiseptic mouthwashes.I have seen mouthwashes being offered in bars and restaurants, in dispensers located in the bathrooms.We need to ascertain whether the patient is overusing the product.If this is indeed the case, then we have to deal with this issue differently.
Participant: Would you recommend the partial removal of carious tissue in a public service facility?If the answer is affirmative, then how would we monitor patients?
Marisa Maltz: Any treatment indication is valid for both the public and the private sectors.It does not matter where the treatment will be performed.Secondly, I believe that the removal of carious tissue is worthless in and of itself.We are talking about decay that reaches the deeper layers of the teeth, but where?If I perform a total removal of the tissue, I will have problems related to pulp exposure and we have already discussed that if the pulp is exposed, with decayed tissue surrounding it, the prognosis is bad.Therefore, in principle, I would never recommend total removal in the case of deep carious lesions.There are many studies dating back to the seventies and eighties that show that the probability of exposing the pulp is very large.So the alternative is a treatment that is already considered classic, which is the removal of the decayed tissue in two steps, which is known as the expectant treatment.With the expectant treatment, which has been performed and has many reported cases in the literature, when we reopen the lesion, we find that the lesion has not evolved.There are studies about the hardness of the dentin and microbiological studies in support of our findings.They show that the carious tissue does not evolve when isolated from the oral environ-ment.What we have today in terms of evidence and systematic review is one ten-year study on partial removal with very good results and that concluded that there is no unequivocal evidence that partial removal is sufficient.On the other hand, there is not enough evidence to support complete removal.I have shared with you long-term studies we have performed demonstrating a very good prognosis, even better than complete removal.Our randomized clinical work, funded by the SUS, has shown that after two years, the partial removal of decayed tissue has better results than those of the expectant treatment among a sample of patients in Brasília and Porto Alegre.In the study conducted with the service, we found that when we perform an expectant treatment and we use the temporary filling, we assume the patient will come back.However, that does not always happen because the patient no longer feels pain.Therefore, the results are actually worse.When we evaluate the patients that do indeed come back, we find that the expectant treatment produces similar results to those produced by the partial removal.Thus, the conclusion is that the partial removal of carious tissue in deep lesions can and should be used as treatment both by the public and private services because there is evidence that points to the fact that the prognosis is very good.But there is an associated cost.In the study we performed, we focused on the cost analysis and attempted to determine how much a basic health unit stood to gain if this type of treatment was performed.In addition to the benefits for the patient and the benefits that we will derive from the procedure, if we perform an endodontic therapy, the restorative treatment would be more complex and expensive; the treatment itself has a much lower cost.
Participant: I took part of a two-year study, published in the Revista Gaúcha de Odontologia, of the Conceição Hospital Stomatology service, and one of the issues that was raised was that the Ministry places an emphasis on oral cancer.In a sample of 435 patients, the prevalence of oral cancer was 4%, but if we added up the endodontic and periodontic diseases and other lesions, the sum of all of these would be much higher than the prevalence of oral cancer.I would like you to comment on this.
Cassius Torres: I will read the results of your study.To avoid a discussion on the topic of prevalence concepts, it is better to refer to this data, of the specific service, of "frequency of lesions".It is only natural in a specialized service to which cases are referred, especially if it is a hospital-based service.So I am not sure if you are surprised by the 4% prevalence.Don't be.In our outpatient department, within the university, it is 2.5%.But the characteristics that we find in regard to what you described as being the case among your patients are similar to those of other services like yours.
Participant: The central authorities emphasize cancer and sometimes we forget to deal with other lesions and even referral and counter-referral criteria.We have to work on this.
Cassius Torres: The CEO directive refers to this as "oral diagnosis with an emphasis on the early detection of oral cancer".Indeed, it was overemphasized in relation to other diseases.And there is a very clear reason for this.The morbidity and mortality statistics mention the prevalence, but we should think about the related costs.During the rehabilitation of the oral cancer patient, we do not see the logic that is found in treating fissures, at which we are much more competent.
Participant: As mentioned by Dr. Pucca, we do not have professionals to work with the Family Strategy and Dr. Lilian also mentioned two other problems: the lack of professionals with this profile to work with the Family Health Program and the training of the generalists.I would also include the training of the management.How are the managers being trained so that they are prepared to work with the teams in the field?How are the cases of oral cancer recorded in the system?Would they be included with other diseases, such as oropharyngeal diseases, or is there a way to select these cases to provide a more exact idea in regard to prevalence and incidence?
Cassius Torres: There are two main systems where we can collect this information: the first one is a system that collects data on mortality, the SIM (Mortality Information System).The other is the system on tumor pathology.They do not take into consideration all of the samples.They actually omit important information.For example, there is a record of a death caused by sepsis resulting from osteomyelitis in a recently operated mandible.However, the fact that it is a result of a previous problem and is related to the morbidity of oral cancer is not mentioned.The notification of cases of oral cancer is not mandatory and some systems are lacking in this sense.For our purposes, the inclusion of oropharyngeal diseases is useful.The head and neck cancers all have a very similar background.It is worth grouping them together, we should not separate them.That is the source of the data that you find in the yearly estimates of the INCA, which might have some problems, but are actually quite good.
Gilberto Pucca: The more general and relevant problem pertains to training.We are lacking in this field.We are only now beginning to have professionals with this profile because the Brazilian dental system was set up in light of a specific market niche, and the educational sector conformed to this logic over ten years ago.The market did not actually have a place for them.The perspective we are referring to was one of setting up a private practice.There was a very fast change in the organization of the market and in professional opportunities.Today the educational sector is beginning to adapt.The opportunities open to those students who are just now beginning their education are very different.But to do these things with professionals that have not yet been adequately trained requires many things to happen simultaneously.The training of managers and of professionals has to conform to the local needs.This is critical.We need to figure out how to combine the needs at the local levels with the availability at the central levels.The local levels should plan their needs and present their proposals for training to the SUS.As the coordination level, we should advocate for the training proposals presented by the services so that they can be present in the SGTES/Ministry of Health or in the State Secretariat.And there are resources that can be invested in training.We need more joint efforts and more project proposals in order to optimize these efforts.
Cassius Torres: There should be studies to evaluate the impact of these training efforts currently being offered.The manager considers himself to be ready because he took part in the training, but he did not necessarily change his practice.Offering the training is a positive step, but it is not enough.We need to consider what kind of training is being offered.There are two kinds of professionals.There is the professional who is already in the network, with a completely different language.And then there are the differences between the municipalities.
Participant: In the treatment with partial removal and in the expectant treatment, is the oral cavity cleaned with chlorhexidine or other antimicrobial agents?
Participant: Will the SB 2010 give us a specific idea by municipality of the influence of fluoridated water in cavity prevention?
Gilberto Pucca: The sample only allows us to do this at the national level.Some municipalities that were drawn by the lottery expanded the sample and in these municipalities we can make this inference.In the SB 2010, the database will be made public and access will be granted to all researchers.
Marisa Maltz: In response to the question of cleaning the oral cavity, a partial removal or an expectant treatment is performed when there is no spontaneous pain, but rather only provoked pain; otherwise the prognosis is not good.And there is no need to use an antiseptic, which would only prolong the treatment and make it more expensive.
Lilian Marly de Paula: The main purpose of the present symposium was the establishment of a debate concerning oral health in the public and private practices in Brazil.For the first time, a broad discussion was held on health promotion in a more integral way.Besides, other topics as oral cancer were approached.
An open debate between the public, mainly constituted by health providers, lecturers, students, and management agents, raised questions concerning the different problems, policies and management of oral health.
Definitely, this debate has opened a new communication between the different actors of the complex system of the SUS (National Unified Health System), and will contribute substantially for the improvement of the oral health system.However, this initial step needs to be maintained and we hopefully ex-pect that this essential aspect of oral health service will continue to be discussed in future ABOPREV meetings.
Evidence-based data concerning the health-disease process, including the definition of risk factors and their repercussion in the individual's general health need to be further discussed.Also, education practices that consider social disparities and cultural regional aspects are necessary in order to better impact behavioral general health changes, individually and in the population as a whole.
§ Summary of the discussions held at the "Promotion of Oral Health in the Public and Private Context" National Symposium, held at the 15 th Congress of the Brazilian Association for Oral Health Promotion (ABOPREV), May 27-29, 2010, Brasilia, DF, Brazil.
|
10.1371/journal.pone.0143518
|
CCBY
|
4670075
|
26637123
|
s2orc_pubmed_articles
|
Enhanced Phospholipase A2 Group 3 Expression by Oxidative Stress Decreases the Insulin-Degrading Enzyme
Oxidative stress has a ubiquitous role in neurodegenerative diseases and oxidative damage in specific regions of the brain is associated with selective neurodegeneration. We previously reported that Alzheimer disease (AD) model mice showed decreased insulindegrading enzyme (IDE) levels in the cerebrum and accelerated phenotypic features of AD when crossbred with alpha-tocopherol transfer protein knockout (Ttpa -/-) mice. To further investigate the role of chronic oxidative stress in AD pathophysiology, we performed DNA microarray analysis using young and aged wild-type mice and aged Ttpa -/mice. Among the genes whose expression changed dramatically was Phospholipase A2 group 3 (Pla2g3); Pla2g3 was identified because of its expression profile of cerebral specific up-regulation by chronic oxidative stress in silico and in aged Ttpa -/mice. Immunohistochemical studies also demonstrated that human astrocytic Pla2g3 expression was significantly increased in human AD brains compared with control brains. Moreover, transfection of HEK293 cells with human Pla2g3 decreased endogenous IDE expression in a dose-dependent manner. Our findings show a key role of Pla2g3 on the reduction of IDE, and suggest that cerebrum specific increase of Pla2g3 is involved in the initiation and/or progression of AD.
# Introduction
Alzheimer disease (AD) is the most common age-related neurodegenerative disease characterized by the deposition of amyloid-beta (Aβ), accumulation of hyperphosphorylated Tau containing neurofibrillary tangles, reactive astrocytes and loss of synapses and neurons [bib_ref] Alzheimer's disease: genes, proteins, and therapy, Selkoe [/bib_ref]. While there are several kinds of rare familial early-onset AD caused by gene mutations, the majority are sporadic cases with various etiologies. However, accumulation of Aβ is the initial event of AD according to the Aβ hypothesis [bib_ref] Amyloid deposition as the central event in the aetiology of Alzheimer's disease, Hardy [/bib_ref] [bib_ref] The amyloid hypothesis of Alzheimer's disease: progress and problems on the road..., Hardy [/bib_ref]. The strongest risk factor for AD is aging so its prevalence increases exponentially with age [bib_ref] Advances in Alzheimer's disease, Katzman [/bib_ref] [bib_ref] National estimates of the prevalence of Alzheimer's disease in the United States, Brookmeyer [/bib_ref]. Aβ accumulation is seen even in the preclinical AD brains; however, it is more robust in mild cognitive impairment (MCI) brains and is most severe in AD brains [bib_ref] Toward defining the preclinical stages of Alzheimer's disease: recommendations from the National..., Sperling [/bib_ref]. One of AD model mouse strains that overexpresses the Swedish mutant form of human Aβ precursor protein (APPsw) also shows age-related Aβ accumulation in the brain [bib_ref] Correlative memory deficits, Abeta elevation, and amyloid plaques in transgenic mice, Hsiao [/bib_ref] [bib_ref] Age-dependent changes in brain, CSF, and plasma amyloid (beta) protein in the..., Kawarabayashi [/bib_ref].
Another early pathological event in AD is increased oxidative stress [bib_ref] Neurodegenerative diseases and oxidative stress, Barnham [/bib_ref] [bib_ref] Oxidative stress and neurodegeneration, Moreira [/bib_ref] [bib_ref] Neuroinflammation, oxidative stress and the pathogenesis of Alzheimer's disease, Agostinho [/bib_ref] [bib_ref] Mitochondrial dysfunction: common final pathway in brain aging and Alzheimer's disease-therapeutic aspects, Müller [/bib_ref] [bib_ref] Alzheimer's disease: diverse aspects of mitochondrial malfunctioning, Santos [/bib_ref] , which contributes to membrane damage, deterioration of synaptic plasticity and neuronal cell death [bib_ref] Lipid peroxidation and protein oxidation in Alzheimer's disease brain: potential causes and..., Butterfield [/bib_ref] , suggesting that oxidative stress promotes the initiation and/or progression of the disease. We previously generated α-tocopherol transfer protein (α-TTP) knockout (Ttpa -/-) mice [bib_ref] Delayed-onset ataxia in mice lacking alpha -tocopherol transfer protein: model for neuronal..., Yokota [/bib_ref]. Ttpa -/mice show marked lipid peroxidation in the brain because of a lack of α-tocopherol and are highly valuable as a model of chronic oxidative stress to the brain. By crossing AD transgenic model mice with Ttpa -/mice, we obtained double mutant APPsw/Ttpa -/mice, which showed increased Aβ deposits in the cerebrum and enhanced cognitive dysfunctions to be compared with APPsw mice. The accumulation of Aβ and cognitive deficits in the double mutant mice were ameliorated with α-tocopherol supplementation [bib_ref] Depletion of vitamin E increases amyloid beta accumulation by decreasing its clearances..., Nishida [/bib_ref] [bib_ref] Deletion of vitamin E enhances phenotype of Alzheimer disease model mouse, Nishida [/bib_ref]. Together with the fact that lipid peroxidation may be a major cause for aging of the brain [bib_ref] Elevated thiobarbituric acid-reactive substances and antioxidant enzyme activity in the brain in..., Lovell [/bib_ref] , oxidative stress and aging are major contributors to AD pathogenesis.
Although many studies of oxidative stress on neurons and glial cells have been done using a cell culture system, the precise molecular and cellular mechanisms underlying AD pathology in vivo have not been fully elucidated. The goal of this study is to clarify the molecular mechanisms of how aging and/or oxidative stress accelerates AD pathophysiology. The region specific pathophysiological changes, Aβ deposits and neurofibrillary tangles in cerebral cortex but not in cerebellum, were well characterized using the postmortem AD brains [bib_ref] Alzheimer's disease: genes, proteins, and therapy, Selkoe [/bib_ref]. APPsw mice also showed the distribution of Aβ mainly in cerebrum but not in cerebellum [bib_ref] A quantum dot probe conjugated with aβ antibody for molecular imaging of..., Feng [/bib_ref]. This lesion specific increase of Aβ accumulation in AD brains let us suppose a hypothesis that some molecules, which alter its expression only in the cerebrum but not in the cerebellum, have important role on AD pathophysiology. We performed DNA microarray analysis using Ttpa -/mice and found Pla2g3 expression was significantly induced only in the cerebral cortex of Ttpa -/mice but not in the cerebellum. Chronic oxidative stress is thought to be highly relevant to the slow progression of AD pathophysiology, thus, specificity of Pla2g3 induction in chronic oxidative condition was confirmed using Ttpa -/mice as chronic oxidative stress model and acute oxidative stress model mice induced by traumatic brain injury and stroke. In this study, we thus provide evidence that Pla2g3 plays a role in brain region-specific changes in AD pathology under chronic oxidative stress.
# Ethical statement
The Institutional Review Board (IRB) of Tokyo Medical and Dental University approved this study. We obtained written informed consent for research use of human paraffin sections from relatives at the time of autopsy. The Animal Experiment Committee of Tokyo Medical and Dental University approved animal experiments. The Ethics Committee of Tokyo Metropolitan Institute of Gerontology approved human study.
## Materials and methods animals
Generation of Ttpa -/mice was previously described [bib_ref] Delayed-onset ataxia in mice lacking alpha -tocopherol transfer protein: model for neuronal..., Yokota [/bib_ref]. Wild-type and Ttpa -/mice were fed on a normal (36 mg of α-tocopherol/kg), α-tocopherol-supplemented (600 mg of α-tocopherol/kg), or α-tocopherol-deficient diet until the time of use. For the ischemic mouse model, male C57Bl/6 (18 weeks old) mice were used for the surgical procedure for left permanent middle cerebral artery occlusionand used for the histological examinations on day 7 after surgery. For the traumatic brain injury model, postnatal day 2 (P2) mice were subjected to cryogenic injury as described previously [bib_ref] Enhancement of neuroblast migration into the injured cerebral cortex using laminin-containing porous..., Ajioka [/bib_ref]. Briefly, P2 pups were anesthetized by spontaneous inhalation of 2% isoflurane, and the parietal skull was exposed through a scalp incision. A metal probe (1.5-mm diameter) cooled by liquid nitrogen was placed on the right skull (0.5-mm anterior and 1.2-mm lateral to the bregma) for 30 seconds. Following this procedure, the scalp was immediately sutured and the pups were returned to the home cage and the pups were fixed after 3 days. The Animal Experiment Committee of Tokyo Medical and Dental University approved animal experiments.
## Rna isolation
RNA was isolated using mirVana mRNA isolation kit (Ambion) according to the manufacturer's protocol. RNA concentration was quantified with NanoDrop Spectrometer (NanoDrop technologies) at a wavelength of 260nm and an Agilent 2100 Bioanalizer was used to evaluate its integrity.
## Microarray
Gene expression in cerebral cortex and cerebellum of mice were determined using Agilent chips. To ensure higher quality results in gene expression data, we conducted microarrays on 4 mice per group. Young mice were 2 months old and the other aged mice were 29 months old at the time of use. Data were standardized using global normalization and processed by R-program. An absolute fold change threshold of greater than 1.5 was required to be considered for further analyses. Expression values were in log2 scale. The original data analysed in this publication have been deposited in NCBI's Gene Expression Omnibus and are accessible through GEO series accession number GSE75047.
## Cell culture
Culture conditions of HEK293 and TR-AST cell lines were described previously [bib_ref] Interchangeable binding of Bcl10 to TRAF2 and cIAPs regulates apoptosis signaling, Yui [/bib_ref] [bib_ref] Acidic amino acid transport characteristics of a newly developed conditionally immortalized rat..., Tetsuka [/bib_ref]. Primary cultured astrocytes were prepared from postnatal day 0 Wistar rats. Culture preparation was performed as described previously [bib_ref] The increased activity of TRPV4 channel in the astrocytes of the adult..., Butenko [/bib_ref]. Primary astrocytes were cultured for 2 weeks before the oxidative stress examination. HEK293 cells were used for transient transfection of human Pla2g3 expression vector (Origene). For the oxidative stress examinations, TR-AST cells were treated with 2 mM hydrogen peroxide for 6 hours, and primary astrocytes were treated with 0.4 mM hydrogen peroxide for 6 hours, otherwise, indicated in the figure legend. The culture plates of the cells were placed on ice at the harvest and rinsed with ice-cold PBS then followed by RNA extraction procedure.
## Qrt-pcr
Reverse transcription of RNA to cDNA was performed with the Roche Reverse transcription kit using the following conditions: 25°C for 10 min, 37°C for 50 min, and 85°C for 5 min. Quantitative PCR was performed subsequently on the resulting DNA using Roche SYBR green master mix and the primers. All the primers were designed using Roche universal design tool and pairs of primers met with highest criteria score were selected. The primer sequences are provided in S1 The PCR thermal cycling conditions are follows: denaturing step at 95°C for 5 min, followed by 45 cycles of 95, 60, 72°C for 10, 10 and 10 sec, respectively. Specificity of each primer set was checked by the melting curve analysis upon quantitative PCR. The delta-Ct for mouse Pla2g3 and mouse GAPDH were subjected to simple t-test to yield the estimation of delta-delta-Ct. The relative expression changes were calculated with amplification efficiency approximating 2.
## Western blotting
The temporal lobe was dissected and homogenized in lysis buffer containing 1% Triton X-100, 0.1% SDS, 20 mM Tris pH 7.5, 150 mM NaCl, 1 mM EDTA and protease inhibitors cocktail (Roche). Tissue was then sonicated using Poly Tron 2100 (Kinematica AG), and insoluble material was removed from the protein extracts by centrifugation at 10,000rpm for 20 min at 4°C. Protein content in the supernatant was quantified with the BCA assay (Thermo). Equivalent amounts of total protein were separated in 4-20% gradient SDS-PAGE gels (Wako) and transferred to nitrocellulose membranes (Millipore). Membranes were blocked with Tris-buffer saline containing 0.1% Tween 20 and 5% nonfat milk for 1 hour, then incubated with primary antibody to Pla2g3 (1:1000; Origene), alpha-tubulin (1:2000; MBL), and were then visualized with appropriate HRP-conjugated secondary antibodies.
## Histology
Mice were intracardially perfused with ice-cold PBS followed by 4% (w/v) ice-cold paraformaldehyde (PFA) in PBS. The dissected brains were then post-fixed overnight with 4% PFA at 4°C. Three mice were used in each group until otherwise stated in the figure legend. Human paraffin sections of superior frontal gyrus (Brodmann area 9) were provided from Tokyo Metropolitan Geriatric Hospital and Institute of Gerontology under the material transfer agreement. We obtained written informed consent for research use from relatives at the time of autopsy. The Ethics Committee of Tokyo Metropolitan Institute of Gerontology approved human study.
## Immunohistochemistry and antibodies
For DAB staining, the fixed brains were paraffin embedded, coronally sectioned, and analyzed using Olympus BX53 microscope. For immunofluorescent staining, vibratome sections were analyzed by confocal microscopy (LSM-510 META; Zeiss) [bib_ref] TrkA gene ablation in basal forebrain results in dysfunction of the cholinergic..., Sanchez-Ortiz [/bib_ref]. DAB and immunofluorescent staining were performed as previously described [bib_ref] Conditional ablation of brain-derived neurotrophic factor-TrkB signaling impairs striatal neuron development, Li [/bib_ref]. Antibodies used for immunostaining were as follows: anti-Pla2g3 (1:250; Origene) and anti-GFAP (1:500; Sigma).
## Quantification of pla2g3-immuno reactive cells
To quantify Pla2g3-immunoreactive (IR) cells, six images of superior frontal gyrus including the molecular and external granular layers were captured per paraffin section. Images were taken at × 200 magnification and analyzed with ImageJ software to quantify Pla2g3-IR cells. The result of quantification was expressed as number of cells/450 × 325 μm.
# Statistical analysis
All data values are presented as the mean ± S.E.M. At least four mice per genotype were used per experiments. Student's t tests were applied to data with two groups of samples. One-way ANOVAs were used for comparisons of data with more than two groups and were followed by the Turkey-Kramer post hoc test for significance. A p value of <0.05 was considered statistically significant. Data were analyzed using GraphPad Prism for Windows.
The Institutional Review Board (IRB) of Tokyo Medical and Dental University approved this study.
# Results
## Dna microarray analysis in mice under oxidative stress
The causes of Aβ accumulation in sporadic AD are not fully understood, but oxidative stress is involved in this process. To investigate the role of chronic oxidative stress on AD pathophysiology in vivo, we employed Ttpa -/mouse line that has marked oxidative stress due to vitamin E deficiency in the brain. Our previous reports showed the accelerated accumulation of Aβ in the cerebrum of APPsw/Ttpa -/double mutant mice [bib_ref] Depletion of vitamin E increases amyloid beta accumulation by decreasing its clearances..., Nishida [/bib_ref] , and the double mutant mice showed earlier and more severe cognitive dysfunction than APPsw mice [bib_ref] Deletion of vitamin E enhances phenotype of Alzheimer disease model mouse, Nishida [/bib_ref]. Consequently, Ttpa -/mouse is thought to be suitable to study the role of oxidative stress in AD pathophysiology.
To find the candidate genes implicated in the acceleration of AD pathophysiology, we performed microarray analysis using Ttpa -/mouse cerebrum and cerebellum [fig_ref] Fig 1: Microarray analysis of cerebrum specific genes regulated by oxidative stress and aging [/fig_ref]. First, we compared 29-month-old Ttpa -/mouse cerebrum with age-matched wild-type mouse cerebrum and found 273 up-regulated genes and 267 down-regulated genes by oxidative stress. Next, we compered 29-month-old wild-type mouse cerebrum with 2-month-old wild-type mouse cerebrum and found 879 up-regulated genes and 509 down-regulated genes by aging in the cerebrum. Among those genes, 13 genes were up-regulated both by oxidative stress and by aging in the cerebrum and 6 genes were down-regulated in the same manner. Further, we compered 29-month-old wild-type mouse cerebellum with 2-month-old wild-type mouse cerebellum and found 1655 up-regulated genes and 573 down-regulated genes by aging in the cerebellum. We excluded 454 genes that were up-regulated by aging both in the cerebrum and in the cerebellum and 51 genes that were down-regulated in the same manner from this study. Then, we finally found nine genes which were differently expressed by oxidative stress and aging only in the cerebrum but not in the cerebellum. Five genes, Fcgbp, Gm3448, Mybpc2, Pla2g3 and Tcp10c, are significantly up-regulated in the cerebrum both by normal aging and by oxidative stress without any significant changes in the cerebellum by normal aging [fig_ref] Fig 1: Microarray analysis of cerebrum specific genes regulated by oxidative stress and aging [/fig_ref]. Also, four genes, Car8.1, Car8.2, Gm2252 and Tsga10, are significantly down-regulated in the cerebrum both by normal aging and by oxidative stress without any significant changes in the cerebellum by normal aging [fig_ref] Fig 1: Microarray analysis of cerebrum specific genes regulated by oxidative stress and aging [/fig_ref]. Among those genes with significantly changed expression by array analysis, Pla2g3 is the only one that has been reported to be involved in Alzheimer's disease and we first focused on this gene.
## Induction of pla2g3 expression in cerebrum by oxidative stress
We confirmed that Pla2g3 mRNA was significantly increased in the cerebrum of wild-type and Ttpa -/mice fed with vitamin E deficient diet compared to wild-type mice fed with normal diet by qRT-PCR [fig_ref] Fig 2: Induction of Pla2g3 expression in cerebrum by oxidative stress [/fig_ref]. In contrast, the expression of Pla2g3 mRNA was not affected under oxidative stress in cerebellum [fig_ref] Fig 2: Induction of Pla2g3 expression in cerebrum by oxidative stress [/fig_ref]. The increased Pla2g3 protein expression in cerebral cortex by oxidative stress was also confirmed by western blot [fig_ref] Fig 2: Induction of Pla2g3 expression in cerebrum by oxidative stress [/fig_ref]. Thus, Pla2g3 could be one of candidate genes involved in this region specific acceleration of Aβ accumulation by oxidative stress and we studied its role further in the following experiments.
## The expression of pla2g3 in astrocytes is dramatically increased by oxidative stress
It has been reported that Pla2g3 is expressed in variety of cell types in many organs including brain [bib_ref] Group III secreted phospholipase A2 regulates epididymal sperm maturation and fertility in..., Sato [/bib_ref]. To characterize the expression pattern of Pla2g3 in the brain by oxidative stress, we carried out immunofluorescence study. Pla2g3 signals were weak and disperse and were mostly co-localized with astrocyte marker, GFAP, in the cerebral cortex of wild-type mice [fig_ref] Fig 3: Astrocytic expression of Pla2g3 by chronic oxidative stress [/fig_ref]. On the other hand, the active astrocytes were observed in the equivalent area of Ttpa -/- mice cerebral cortex and the intense Pla2g3 signals were induced in those active astrocytes [fig_ref] Fig 3: Astrocytic expression of Pla2g3 by chronic oxidative stress [/fig_ref]. Pla2g3 induction was not observed in an acute oxidative stress accompany with the gliosis by ischemia [fig_ref] Fig 3: Astrocytic expression of Pla2g3 by chronic oxidative stress [/fig_ref] or traumatic brain injury [fig_ref] Fig 3: Astrocytic expression of Pla2g3 by chronic oxidative stress [/fig_ref] to the cerebral cortex, suggesting that the chronic oxidative stress is a key condition to induce Pla2g3 expression in astrocytes in vivo.
## Overexpression of pla2g3 reduces ide expression
We previously reported that the clearance of Aβ from brain was decreased in Ttpa -/mice compared with wild-type mice, mainly due to the decreased expression level of IDE [bib_ref] Deletion of vitamin E enhances phenotype of Alzheimer disease model mouse, Nishida [/bib_ref]. Surprisingly, IDE expression was significantly decreased in TR-AST cells after exposure to hydrogen peroxide in both a time-dependent and dose-dependent manner [fig_ref] Fig 4: Overexpression of Pla2g3 reduces IDE expression [/fig_ref] and IDE expression of rat primary astrocytes also showed extremely high sensitivity to hydrogen peroxide [fig_ref] Fig 4: Overexpression of Pla2g3 reduces IDE expression [/fig_ref].
To further study the possible role of Pla2g3 in the accumulation of Aβ, we overexpressed Pla2g3 using a cell culture system to examine whether Pla2g3 affects the endogenous expression level of IDE. Because of low transfection efficiency of primary astrocytes and TR-AST cells, we employed HEK293 cells to yield high transfection efficiency that also express IDE endogenously. In HEK293 cells, IDE expression was significantly reduced by the expression of Pla2g3 dose-dependently after 48 hours [fig_ref] Fig 4: Overexpression of Pla2g3 reduces IDE expression [/fig_ref]. This indicates that Pla2g3 might be implicated in the accumulation of Aβ through the suppression of IDE expression.
## Increased expression of pla2g3 in ad patients
We next examined the expression of Pla2g3 in AD patients. DAB staining using anti-Pla2g3 antibody showed only weak and diffused staining in frontal cortex of normal control brains [fig_ref] Fig 5: Increased expression of Pla2g3 in frontal cortex of AD patients [/fig_ref]. On the other hand, in AD brains, the intense Pla2g3-immunoreactivity was observed in the external granular layer (Layer II) and the signals were spread to the deeper layers. The quantitative analysis of Pla2g3-immunoreactive cells demonstrated significant increase in cerebral cortex of AD [fig_ref] Fig 5: Increased expression of Pla2g3 in frontal cortex of AD patients [/fig_ref]. Immunofluorescence studies revealed only disperse staining of Pla2g3 in normal controls [fig_ref] Fig 5: Increased expression of Pla2g3 in frontal cortex of AD patients [/fig_ref] as compared to the intensive Pla2g3 staining co-localized with GFAP signals observed in AD patients [fig_ref] Fig 5: Increased expression of Pla2g3 in frontal cortex of AD patients [/fig_ref] , indicating that Pla2g3 expression is promoted mostly in astrocytes of AD patients. Together with the earlier findings, Pla2g3 might be involved in the initiation and/or progression of AD through IDE suppression in astrocytes.
# Discussion
The lipids of the brain, such as phosphatidylcholine, are integrated in the phospholipid membrane of brain cells and at one time were believed to have only a structural role. PLA2s are a diverse family of enzymes that catalyze the cleavage of fatty acids at the sn-2 position of glycerophospholipids to generate lysophospholipids and free fatty acids including arachidonic acids, which are important second messengers in signal transduction [bib_ref] Diversity of group types, regulation, and function of phospholipase A2, Dennis [/bib_ref] [bib_ref] The growing phospholipase A2 superfamily of signal transduction enzymes, Dennis [/bib_ref]. PLA2 are categorized into three groups, Ca 2+ -dependent cytosolic PLA2 (cPla2), Ca 2+ -independent PLA2 (iPla2), and Ca 2+ -dependent secretory PLA2 (sPla2) which includes Pla2g3. It has been established that brain lipids participate both in the function and in structure of brain cells and PLA2s play a key role not only in physiological events but also in pathological events such as neurogenesis [bib_ref] Rapid regulation of neurite outgrowth and retraction by phospholipase A2-derived arachidonic acid..., Smalheiser [/bib_ref] [bib_ref] Secretory phospholipases A2 induce neurite outgrowth in PC12 cells through lysophosphatidylcholine generation..., Ikeno [/bib_ref] [bib_ref] Neuronal expression and neuritogenic action of group X secreted phospholipase A2, Masuda [/bib_ref] and neuronal cell death [bib_ref] Differential roles of phospholipases A2 in neuronal death and neurogenesis: implications for..., Schaeffer [/bib_ref] [bib_ref] Human group IIA secretory phospholipase A2 induces neuronal cell death via apoptosis, Yagami [/bib_ref] [bib_ref] Human group IIA secretory phospholipase A2 potentiates Ca2+ influx through L-type voltage-sensitive..., Yagami [/bib_ref] [bib_ref] Cytosolic phospholipase A2 mediates neuronal apoptosis induced by soluble oligomers of the..., Kriem [/bib_ref] [bib_ref] Phospholipase A2 reduction ameliorates cognitive deficits in a mouse model of Alzheimer's..., Sanchez-Mejia [/bib_ref] [bib_ref] Group IIA secretory phospholipase A2 (GIIA) mediates apoptotic death during NMDA receptor..., Chiricozzi [/bib_ref].
In the present study, we demonstrated that Pla2g3 expression is increased in cerebral cortex but not in cerebellum by chronic oxidative stress with vitamin E deficiency. Pla2g3 is expressed both in neurons and astrocytes but oxidative stress-induced Pla2g3 expression is predominantly in astrocytes in the cerebrum. It is particularly noteworthy that this increased Pla2g3 expression is absent in the astrogliosis induced by the ischemia or traumatic brain injury to the cerebral cortex, indicating that not the acute oxidative stresses but the chronic oxidative stress is the essential factor for induction of Pla2g3 expression in vivo. Moreover, Pla2g3 expression is not associated with the formation of reactive astrocytes. This brain region-specific astrocytic induction of Pla2g3 may be due to the regional differences of vulnerability to the accumulation of oxidative stress during aging. In fact, cerebellum has been reported to be more resistant to oxidative stress than hippocampus and frontal cortex of human brain [bib_ref] Elevated oxidative stress and decreased antioxidant function in the human hippocampus and..., Venkateshappa [/bib_ref] , which provides a rational explanation for the cerebral cortex specific induction of Pla2g3 we observed. Additionally, Pla2g5 and iPla2 did not increase significantly and Pla2g2e was not detectable by oxidative stress in vivo despite a marked simultaneous increase in TR-AST cells after hydrogen peroxide treatment (S1 suggesting that the regulation of PLA2s gene expression varies among the family to support the complex lipids metabolism in the brain and each PLA2 might execute distinct cellular processes.
PLA2s participate in a variety of physiological processes, including remodeling of cellular membranes, signal transductions and host defense. Pla2g3 is expressed in a wide variety of organs but is most abundant in testis and brain [bib_ref] Diversity of group types, regulation, and function of phospholipase A2, Dennis [/bib_ref]. Human Pla2g3 has been suggested to be involved in atherosclerosis in apo-E deficient mice [bib_ref] Analyses of group III secreted phospholipase A2 transgenic mice reveal potential participation..., Sato [/bib_ref] , and its overexpression causes spontaneous skin inflammation in human Pla2g3-transgenic mice [bib_ref] Group III secreted phospholipase A2 transgenic mice spontaneously develop inflammation, Sato [/bib_ref]. Recent study has shown that Pla2g3 plays a key role in maturation of mast cells in vivo [bib_ref] Mast cell maturation is driven via a group III phospholipase A2-prostaglandin D2-DP1..., Taketomi [/bib_ref] , indicating Pla2g3 modulates gene expression profile in certain cells. However, the function of Pla2g3 in the brain has not been fully understood. Bee venom Pla2g3 has been demonstrated to cause apoptosis in rat primary cortical neurons [bib_ref] Group III secreted phospholipase A2 causes apoptosis in rat primary cortical neuronal..., Decoster [/bib_ref] although human Pla2g3 has pro-survival effects on PC12 cells and promotes its neuron-like differentiation [bib_ref] Differential roles of phospholipases A2 in neuronal death and neurogenesis: implications for..., Schaeffer [/bib_ref]. In the current study, human Pla2g3 transfected HEK293 cells also did not show any apoptotic signs after 48 hours of transfection (data not shown). Moreover, despite the profound induction of apoptosis observed in the region of ischemia and traumatic brain injury site of the mice, no induction of Pla2g3 expression was observed in this time course. Together with previous reports, our results indicate that Pla2g3 is not directly involved in apoptosis so that there may be a functional difference between bee venom and human Pla2g3. It is of note we found that human Pla2g3 is capable of reducing IDE expression in vitro. Expression of many genes including proteases, cytokines and chemokines are affected by Pla2g3 and those changes are considered to be mediated by bioactive lipid mediators [bib_ref] Insulin-degrading enzyme rapidly removes the beta-amyloid precursor protein intracellular domain (AICD), Edbauer [/bib_ref] , thus, IDE expression could be influenced by the lipid mediator pathways acting directly downstream of Pla2g3. Although other PLA2s are possibly capable of reducing IDE expression because of the same enzymatic activities they possess, they are less likely to be involved in AD pathophysiology since either their expression was undetectable by qRT-PCR or unchanged by chronic oxidative stress (S1 AD is a progressive neurodegenerative disorder, which leads to the most common causes of dementia. The alteration of amyloid precursor protein (APP) processing, the aggregation and/ or clearance of Aβ can be considered crucial in the progression of AD. We previously reported that the production of Aβ was not changed but the clearance of Aβ from brain was prominently decreased in APPsw/Ttpa -/double mutant mice causing accelerated Aβ accumulation compared with APPsw mice, indicating that oxidative stress has a profound effect on Aβ clearance. One of enzymes involved in Aβ homeostasis is IDE, which degrades Aβ and plays a role on Aβ efflux from brain [bib_ref] Secreted phospholipase A2 and mast cells, Murakami [/bib_ref]. In Ttpa -/mice, IDE expression is markedly decreased compared to wild type mice. We demonstrated that overexpression of human Pla2g3 suppresses IDE expression in a dose-dependent manner. Therefore, we presume that Pla2g3 is one of the responsible factors for the acceleration of Aβ accumulation through decreased IDE expression. As a robust IDE expression in human astrocytes has been also reported [bib_ref] The effect of amyloid associated proteins on the expression of genes involved..., Mulder [/bib_ref] , induction of Pla2g3 expression may reduce IDE expression in astrocytes although previous studies of IDE regulation in AD gave inconsistent results, which showed either reduction [bib_ref] Degradation of soluble amyloid beta-peptides 1-40, 1-42, and the Dutch variant 1-40Q..., Pérez [/bib_ref] [bib_ref] Reduced hippocampal insulin-degrading enzyme in late-onset Alzheimer's disease is associated with the..., Cook [/bib_ref] [bib_ref] Insulin degrading enzyme activity selectively decreases in the hippocampal formation of cases..., Zhao [/bib_ref] or induction of IDE in AD [bib_ref] Differential cerebral deposition of IDE and NEP in sporadic and familial Alzheimer's..., Dorfman [/bib_ref]. We could not exclude the possibility that not only suppression of IDE, but also human Pla2g3-induced inflammation [bib_ref] Group III secreted phospholipase A2 transgenic mice spontaneously develop inflammation, Sato [/bib_ref] and neuronal cell death [bib_ref] PLA2G3, a gene involved in oxidative stress induced death, is associated with..., Martínez-García [/bib_ref] would be closely related to the progression of AD pathophysiology. Crossing APPsw mice with Pla2g3 knockout mice could be another possible study to comprehend the further role of Pla2g3 involved in AD pathology.
In summary, our data indicate that Pla2g3 plays a pivotal role in suppression of IDE. Thus, increased Pla2g3 expression in the astrocytes by chronic oxidative stress may disrupt Aβ homeostasis, which ultimately leads to the initiation and/or progression of AD. Consequently, the Pla2g3 and IDE pathway could be a suitable target for the development of novel treatment strategies for AD.
Supporting Information S1 ARRIVE Checklist. Completed "The ARRIVE Guidelines Checklist" for reporting animal data in this manuscript. (PDF) S1 [fig_ref] Fig 2: Induction of Pla2g3 expression in cerebrum by oxidative stress [/fig_ref] expression in TR-AST and mouse cerebral cortex under oxidative stress. A, Quantitative PCR results of PLA2s expression in TR-AST cells treated with 2mM hydrogen peroxide for 6 hours compared with non-treated controls. Fold changes to non-treated controls are indicated. B-D, Quantitative PCR results of indicated PLA2s in mouse cortex. Fold changes to the aged wild-type mice on normal diet are indicated. n = 4 in each group. Abbreviation used; WT normal; wild type 29 months old mice fed on normal diet, WT def; 29 months old wild type mice fed on Vitamin E deficient diet, ttpKO def; 29 months old Ttpa -/mice fed on Vitamin E deficient diet. (TIF) S1 Nucleotide sequences of oligonucleotides used in this study. The sequences are shown in 5' to 3' direction. (TIF)
[fig] Fig 1: Microarray analysis of cerebrum specific genes regulated by oxidative stress and aging. A, Each circle is indicating the numbers of either up or down regulated genes by oxidative stress, aging in the cerebrum and in the cerebellum. Five genes were up-regulated and four genes were down regulated by both oxidative stress and aging in the cerebrum without significant changes in the cerebellum. B, The heat maps of five genes up-regulated and C, four genes down-regulated by oxidative stress and aging. Red indicates upregulated genes and blue indicates down-regulated genes, while white indicates no significant change in expression. The color gradients indicate the intensity of expression. The data shows the results of four mice per each group. Abbreviation used; OS; oxidative stress. doi:10.1371/journal.pone.0143518.g001 [/fig]
[fig] Fig 2: Induction of Pla2g3 expression in cerebrum by oxidative stress. A, B, Quantitative RT-PCR results of Pla2g3 in cerebrum and cerebellum are shown. Fold changes to the aged wild-type mice on normal diet are indicated. n = 4 in each group. C, Western blots for Pla2g3 and alpha-tubulin are shown. Abbreviation used; WT normal; wild type 29 months old mice fed on normal diet, WT deficient; 29 months old wild type mice fed on vitamin E deficient diet, ttpKO deficient; 29 months old Ttpa -/mice fed on vitamin E deficient diet. *p<0.05. doi:10.1371/journal.pone.0143518.g002 [/fig]
[fig] Fig 3: Astrocytic expression of Pla2g3 by chronic oxidative stress. Double immunostaining of Pla2g3 (green) and GFAP (red) in 29 months old wild type mouse (A, B), 29 months old Ttpa -/mouse (C, D), Ischemic site (E, F) and traumatic injury site (G, H) of cerebral cortex of wild type mice. Arrow heads indicate the strong Pla2g3 expression in astrocytes of Ttpa -/cortex. Scale bar: 50 μm. doi:10.1371/journal.pone.0143518.g003 [/fig]
[fig] Fig 4: Overexpression of Pla2g3 reduces IDE expression. A, B, Quantitative RT-PCR result of IDE in TR-AST cells and rat primary astrocytes. TR-AST cells and primary astrocytes were treated with hydrogen peroxide in indicated conditions. Fold changes to the non-treated cells as controls are indicated. C, Quantitative PCR results of IDE in Pla2g3 transfected HEK293 cells. Human Pla2g3 was transiently expressed and cells were harvested after 48 hours of transfection. Fold changes to the mock control are indicated. *p<0.05, ***p<0.001. doi:10.1371/journal.pone.0143518.g004 [/fig]
[fig] Fig 5: Increased expression of Pla2g3 in frontal cortex of AD patients. A, Pla2g3 expression in normal control brain and AD brain. The images are representative of 6 cases in each group. Frontal cortex was stained with anti-Pla2g3 antibody. B, Analysis of Pla2g3-positive cell numbers in frontal cortex. *p<0.05 C-H, Double immunostaining of Pla2g3 and GFAP in normal control brain (C-E) and AD brain (F-H). Arrowheads indicate the intensive staining of Pla2g3 in astrocytes of AD brain. Scale bars: A, 300 μm; C, 50 μm. Abbreviation used; NC; normal control, AD; Alzheimer's disease patient, I; Molecular layer, II; External granular layer.doi:10.1371/journal.pone.0143518.g005 Pla2g3 Decreases IDE PLOS ONE | DOI:10.1371/journal.pone.0143518 December 4, 2015 [/fig]
|
10.1186/1477-5956-7-30
|
CCBY
|
2744661
|
19715574
|
s2orc_pubmed_articles
|
Proteomic profiling of L-cysteine induced selenite resistance in Enterobacter sp. YSU
Background: Enterobacter sp. YSU is resistant to several different heavy metal salts, including selenite. A previous study using M-9 minimal medium showed that when the selenite concentration was 100,000 times higher than the sulfate concentration, selenite entered Escherichia coli cells using two pathways: a specific and a non-specific pathway. In the specific pathway, selenite entered the cells through a yet to be characterized channel dedicated for selenite. In the non-specific pathway, selenite entered the cells through a sulfate permease channel. Addition of L-cystine, an L-cysteine dimer, appeared to indirectly decrease selenite import into the cell through the non-specific pathway. However, it did not affect the level of selenite transport into the cell through the specific pathway.Results: Growth curves using M-9 minimal medium containing 40 mM selenite and 1 mM sulfate showed that Enterobacter sp. YSU grew when L-cysteine was present but died when it was absent. Differential protein expression analysis by two dimensional gel electrophoresis showed that CysK was present in cultures containing selenite and lacking L-cysteine but absent in cultures containing both selenite and L-cysteine. Additional RT-PCR studies demonstrated that transcripts for the sulfate permease genes, cysA, cysT and cysW, were down-regulated in the presence of L-cysteine.Conclusion: L-cysteine appeared to confer selenite resistance upon Enterobacter sp. YSU by decreasing the level of selenite transport into the cell through the non-specific pathway.
# Introduction
Selenium is an important cofactor in some mammalian and bacterial enzymes [bib_ref] Molecular biology of selenium with implications for its metabolism, Burk [/bib_ref] [bib_ref] Selenoproteins and the metabolic features of the archaeal ancestor of eukaryotes, Foster [/bib_ref] [bib_ref] The bacterial response to the chalcogen metalloids Se and Te, Zannoni [/bib_ref]. It is found in mammalian glutathione peroxidase [bib_ref] Identification of the catalytic site of rat liver glutathione peroxidase as selenocysteine, Forstrom [/bib_ref] and bacterial formate dehydrogenase [bib_ref] Nucleotide sequence and expression of the selenocysteine-containing polypeptide of formate dehydrogenase (formate-hydrogen-lyase-linked)..., Zinoni [/bib_ref] in the form of selenocysteine. A series of Escherichia coli (E. coli) proteins, SelA, SelB, SelC and SelD, incorporate selenium into selenocysteine which is then inserted into proteins. The mechanisms that transport selenite into the cell are not well understood. Once it enters the cell, it may be reduced to selenide by glutathione [bib_ref] Selenium metabolism in Escherichia coli, Turner [/bib_ref] or thioredoxin [bib_ref] Selenite assimilation into formate dehydrogenase H depends on thioredoxin reductase in Escherichia..., Takahata [/bib_ref]. Then, SelD, a selenophosphate synthetase, catalyzes a reaction with selenide and ATP to make selenophosphate [bib_ref] Escherichia coli mutant SELD enzymes. The cysteine 17 residue is essential for..., Kim [/bib_ref] [bib_ref] In vitro synthesis of selenocysteinyl-tRNA(UCA) from seryl-tRNA(UCA): involvement and characterization of the..., Leinfelder [/bib_ref]. SelA, a selenocysteine synthetase, uses selenophosphate to convert ser-ine to selenocysteine [bib_ref] Selenocysteine synthase from Escherichia coli, Forchhammer [/bib_ref] , which is carried by SelC, a special tRNA [bib_ref] Gene for a novel tRNA species that accepts L-serine and cotranslationally inserts..., Leinfelder [/bib_ref]. Finally, SelB, a translation factor, inserts selenocysteine at a UGA stop codon. The mRNAs of proteins that encode selenocysteine also contain a special stem loop structure or selenocysteine insertion sequence (SECIS) which stalls the ribosome and allows SelB to insert selenocysteine at the UGA stop codon [bib_ref] The bulged nucleotide in the Escherichia coli minimal selenocysteine insertion sequence participates..., Li [/bib_ref] [bib_ref] Revised Escherichia coli selenocysteine insertion requirements determined by in vivo screening of..., Sandman [/bib_ref].
Although selenium is an important cofactor, too much can be toxic. Selenium inhibits the growth of E. coli when it is present at overabundant concentrations as selenite, but non-toxic when it is present as elemental selenium [bib_ref] Active oxygen species generation and cellular damage by additives of parenteral preparations:..., Terada [/bib_ref]. Selenite reacts with glutathione and other thiol containing proteins to produce highly reactive superoxides [bib_ref] Active oxygen species generation and cellular damage by additives of parenteral preparations:..., Terada [/bib_ref] [bib_ref] Involvement of superoxide dismutases in the response of Escherichia coli to selenium..., Bebien [/bib_ref] [bib_ref] Active oxygen generation as a possible mechanism of selenium toxicity, Seko [/bib_ref] [bib_ref] Selenium toxicity: cause and effects in aquatic birds, Spallholz [/bib_ref] , which may kill the cells by damaging their DNA and lipids [bib_ref] Active oxygen generation as a possible mechanism of selenium toxicity, Seko [/bib_ref] [bib_ref] The genotoxicity of selenium, Shamberger [/bib_ref].
Selenite and selenate resistant bacteria appear to remove excess selenium by reducing it to elemental selenium or methylating it [bib_ref] The bacterial response to the chalcogen metalloids Se and Te, Zannoni [/bib_ref]. A strain of Stenotrophomonas maltophilia (S. maltophilia) isolated from a selenium-contaminated drainage pond reduces selenate and selenite to elemental selenium and deposits it near the cell surface and in the surrounding growth medium [bib_ref] Transformations of selenate and selenite by Stenotrophomonas maltophilia isolated from a seleniferous..., Dungan [/bib_ref]. In addition, Cupriavidus metallidurans (C. metallidurans) CH34, a multi-metal resistant strain [bib_ref] Alcaligenes eutrophus CH34 is a facultative chemolithotroph with plasmid-bound resistance to heavy..., Mergeay [/bib_ref] from a zinc contaminated site, is resistant to selenate and selenite and reduces both oxyanions to elemental selenium and alkyl selenide under aerobic conditions [bib_ref] Mobilization of selenite by Ralstonia metallidurans CH34, Roux [/bib_ref] [bib_ref] Chemical forms of selenium in the metalresistant bacterium Ralstonia metallidurans CH34 exposed..., Sarret [/bib_ref]. A C. metallidurans CH34 protein, DedA, which was identified by transposon mutagenesis, may be involved in selenite uptake [bib_ref] A probable link between the DedA protein and resistance to selenite, Ledgham [/bib_ref]. A set of E. coli proteins, YgfK, YgfM and YgfN, which were also identified by transposon mutagenesis, act as an enzyme that reduces selenate to elemental selenium [bib_ref] Involvement of a putative molybdenum enzyme in the reduction of selenate by..., Bebien [/bib_ref]. Finally, some bacterial strains also methylate selenite and selenate by encoding a thiopurine methyltransferase, which produces volatile dimethyl selenide and dimethyl diselenide [bib_ref] Freshwater bacteria can methylate selenium through the thiopurine methyltransferase pathway, Ranjard [/bib_ref] [bib_ref] Methylation of inorganic and organic selenium by the bacterial thiopurine methyltransferase, Ranjard [/bib_ref].
Understanding how selenite enters the cell is important to understanding the mechanisms of selenite resistance. An early study in E. coli shows that selenite and sulfate follow a Michaelis-Menten model when they are transported into the cell and compete competitively with each other [bib_ref] Single transporter for sulfate, selenate, and selenite in Escherichia coli K-12, Lindblow-Kull [/bib_ref]. Using a low sulfate, M-9 minimal medium containing 10 μM sulfate and 75 Se-labeled selenite mixed with 1 M unlabeled selenite, Muller et al define two pathways for selenium incorporation in E. coli: (1) specific and (2) nonspecific [bib_ref] The path of unspecific incorporation of selenium in Escherichia coli, Muller [/bib_ref] [bib_ref] Direct detection of potential selenium delivery proteins by using an Escherichia coli..., Lacourciere [/bib_ref]. In the specific pathway, selenite enters the cell through a pathway dedicated for selenite and becomes incorporated into selenocysteine intentionally. In the non-specific pathway, selenite enters the cell through the sulfate transport system and becomes incorporated unintentionally to produce selenocysteine instead of L-cysteine. When 10 μg/ml of L-cystine, an Lcysteine dimer, is added to the medium, the wild type E. coli strain incorporates selenium randomly into cellular proteins, presumably via the non-specific pathway. However, when 60 μg/ml L-cystine is added to the medium, the amount of randomly incorporated selenium in wild type E. coli decreases dramatically. These results suggest that feedback inhibition blocks both the biosynthesis pathway for L-cysteine and the accidental incorporation of selenium into proteins. In support of this hypothesis, cysK mutants, which lack the ability to add sulfide to Oacetylserine to generate L-cysteine, fail to randomly incorporate radioactive selenite into protein through the nonspecific pathway [bib_ref] The path of unspecific incorporation of selenium in Escherichia coli, Muller [/bib_ref] [bib_ref] Direct detection of potential selenium delivery proteins by using an Escherichia coli..., Lacourciere [/bib_ref]. Thus, under most conditions when selenite concentrations are lower than sulfate concentrations, selenite is excluded from the non-specific pathway. However, when selenite concentrations are higher than sulfate concentrations, selenite is transported into the cell through the non-specific pathway instead of sulfate.
Enterobacter sp. YSU is a multi-metal resistant strain that grows in the presence of mercury, cadmium, zinc, gold, silver, arsenic and selenium [bib_ref] Comparison of two multimetal resistant bacterial strains: Enterobacter sp. YSU and Stenotrophomonas..., Holmes [/bib_ref]. It does not grow in M-9 minimal medium containing 1 mM sulfate and 40 mM selenite. However, growth inhibition is relieved by the addition of L-cysteine. Here, we define L-cysteine-dependent selenite resistance in Enterobacter sp. YSU using growth curves, 2-D gel electrophoresis and reverse-transcriptasepolymerase chain reactions (RT-PCR).
# Results
## Requirement of l-cysteine for selenite resistance
Enterobacter sp. YSU was grown in four M-9 minimal medium cultures. During early log phase after 1.5 hours of growth, selenite or an equal volume of water was added to give the following growth conditions: no L-cysteine and no selenite (NCNS), no L-cysteine and selenite (NCS), Lcysteine and no selenite (CNS), and L-cysteine and selenite (CS). Every 45 minutes, culture samples were diluted and plated to follow viable cell count versus time . Each experiment was performed four times, and each plotted point was the average of cell counts from at least three of the four experiments. Both conditions without selenite (NCNS and CNS) demonstrated normal growth curves. The bacteria in the culture lacking Lcysteine and containing selenite (NCS) were killed by the selenite, and 3.5 hours after adding selenite, the number of viable cells decreased on average by 98%. The culture containing both L-cysteine and selenite (CS), on the other hand, demonstrated a normal growth curve, and 3.5 hours after adding selenite, the number of viable cells increased on average by 890%. These results clearly showed that without L-cysteine, 40 mM selenite is toxic to Enterobacter sp. YSU.
## Growth curves used for proteomic analysis
Two stationary phase cultures of Enterobacter sp. YSU grown in M-9 minimal medium containing and lacking 40 μg/ml L-cysteine were diluted 1:20 into fresh medium containing and lacking L-cysteine. Growth was followed by measuring turbidity every 30 minutes. After 2.5 hours of growth, the NCNS and CNS samples were taken for protein analysis. The cultures were split into equal volumes, and selenite or water was added to give the four conditions. One hour after selenite was added, the NCS and CS samples were taken for protein analysis.
These growth curves also showed that L-cysteine played a role in selenite resistance . The two positive control cultures, NCNS and CNS, demonstrated a typical growth curve and increased on average to final cell densities of 160 ± 19 and 138 ± 38 Klett units, respectively. The cultures that lacked L-cysteine and contained selenite, NCS, increased in cell density on average by only12 Klett units to 46 ± 13 Klett units 3.5 hours after selenite was added. However, the cultures that contained L-cysteine and selenite (CS) were able to grow. This culture increased in cell density to 193 ± 81 Klett units 3.5 hours after selenite was added.
## Two-dimensional gel electrophoresis (2dge)
Protein samples from all 4 culture conditions were analyzed by 2DGE between the pI ranges of 4 and 7 [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref]. Differences in protein expression were detected by comparing the size and intensity of the same spot on each gel. Spots that appeared with equal intensities or higher intensities compared to other spots in the same location on other gels were excised, digested with trypsin and analyzed by mass spectrometry (Nano-LC/MS/MS analysis). Analysis of the peptides from each spot using the Mascot software package [bib_ref] Probability-based protein identification by searching sequence databases using mass spectrometry data, Perkins [/bib_ref] identified significant matches to known proteins. For a match to be considered significant, it contained at least two different peptide fragments matching part of an entire known sequence.
## Proteins in spots that appeared at equivalent intensities in gels of samples prepared from cells grown under all four conditions
Two landmark spots, which are identified by blue arrows [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref] , appeared at equal intensities in all four gels and were selected to demonstrate accuracy. Spot 1 from all four gels contained peptides that matched to an OmpF Viable cell count growth curves Viable cell count growth curves. Overnight cultures were diluted 1:20 in fresh M-9 minimal medium and grown at 37°C. After 1.5 hours of growth, selenite or water was added to each culture. Samples were diluted and spread on plates containing LB medium every 45 minutes. Values at each time point are the average of at least 3 different experiments, and error was calculated using the student t test at a 95% confidence level. Turbidity growth curves. Overnight cultures were diluted 1:20 in fresh M-9 minimal medium and grown at 37°C. After 2.5 hours of growth, selenite or water was added to each culture. Turbidity was measured every 30 minutes using a Klett colorimeter. Values at each time point are the average of 6 different experiments, and error was calculated using the student t test at a 95% confidence level.
## Turbidity growth curves
porin from Enterobacter cloacae with Mascot scores ranging from 496-520 and sequence coverages ranging from 22-23% [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref]. Spot 2 from all four gels contained peptides that matched to a protein chain elongation factor EF-Ts from Salmonella typhimurium (S. typhimurium) with Mascot scores ranging from 654-1090 and sequence coverages ranging from 44-60%.
## Proteins in a spot that appeared at a higher intensity in gels of samples prepared from cells grown in the absence of l-cysteine
Spot 3, which is identified by a purple arrow [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref] , appeared at a higher intensity in the gels of samples prepared from cells grown in the absence of L-cysteine (NCNS and NCS). This spot contained peptides that matched to a conserved hypothetical lipobinding protein from Erwinia pyrifoliae with a Mascot score of 153 and a sequence coverage of 7% [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref]. Spot 3 from the NCS gel also contained peptides that matched to two other proteins: a branched-chain amino-acid aminotransferase from Yersinia pestis with a Mascot score of 277 and a sequence coverage of 25% and an outer membrane pro-tein II from Enterobacter aerogenes (E. aerogenes) with a Mascot score of 104 and a sequence coverage of 9%.
## Proteins in spots that appeared at higher intensities in gels of samples from cells grown in the absence of selenite
Three spots, which are identified by dark red arrows [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref] , appeared at higher intensities on gels that contained samples prepared from cells grown in the absence of selenite (NCNS and CNS). Spot 4 from the NCNS and CNS gels was unique. This spot contained peptides that matched to an EF-Tu protein from S. typhimurium with Mascot scores of 1143 and 1467 and sequence coverages of 64% and 52%, respectively [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref]. In addition, spots 5 and 6 were more intense in gels containing samples prepared from cells grown in the absence of selenite (NCNS and CNS). They contained peptides that matched to EF-Tu from S. typhimurium with Mascot scores of 611 and 666 and sequence coverages of 25% and 30%, respectively. Spot 6 also contained peptides that matched to 5-methyltetrahydropteroyltriglutamate-homocysteine methyltransferase (MetE) from Erwinia carotovora with a Mascot score of 173 and a sequence coverage of 10%.
## Proteins in spots that appeared at higher intensities in gels of samples prepared from cells grown in the presence of selenite
Five spots, which are identified by red arrows [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref] , appeared at higher intensities in gels of samples prepared from cells grown in the presence of selenite (NCS and CS). Spot 7 was the largest in the CS gels. In the no L-cysteine, selenite gel (NCS), it contained peptides that matched to 4 proteins [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref] : an outer membrane protein II from E. aerogenes with a Mascot score of 921 and a sequence coverage of 86%, OmpA from E. sakazakii with a Mascot score of 700 and a sequence coverage of 40%, a putative membrane component hydrogenase from S. typhimurium with a Mascot score of 561 and a sequence coverage of 25% and a CysK protein from E. coli with a Mascot score of 245 and a sequence coverage of 16%. In the CS gels, it contained a similar set of peptides for 3 proteins: an outer membrane protein II from E. aerogenes with a Mascot score of 960 and a sequence coverage of 73%, OmpA from E. sakazakii with a Mascot score of 696 and a sequence coverage of 38% and a putative membrane component hydrogenase from S. typhimurium with a Mascot score of 549 and a sequence coverage of 25%. It lacked peptides that matched to CysK.
Under selenite sensitive condition (NCS), spots 8 and 9 were unique [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref]. They did not appear in gels of samples prepared from cells grown under the other three conditions. Spot 8 contained peptides that matched to 4 proteins [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref] : a putative tellurium resistance protein C from E. coli [bib_ref] Iha: a novel Escherichia coli O157:H7 adherence-conferring molecule encoded on a recently..., Tarr [/bib_ref] with a Mascot score of 516 and a sequence coverage of 52%, a protein similar to GroES Negative images of Enterobacter sp. YSU total protein sepa-rated by 2DGE over a pI range of 4-7 and with L-Cysteine and Selenite (CS) were harvested after 3.5 hours of growth. Spots identified with blue arrows appeared with equal intensities under all 4 conditions. The spot identified by a purple arrow appeared at a higher intensity in gels containing samples from cells grown in the absence of L-cysteine. Spots identified by dark red arrows appeared at higher intensities in gels containing samples from cells grown in the absence of selenite. Spots identified by red arrows appeared at higher intensities in gels containing samples from cells grown in the presence of selenite. The spot numbers correspond to the spot numbers in [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref].
## Detection of cysk, cyse and sulfate permease transcripts in cells grown in the presence and absence of l-cysteine
The absence of CysK in protein samples of cells grown in the presence of L-cysteine and selenite (CS) suggested that the addition of L-cysteine down-regulated the expression of the L-cysteine synthesis and the sulfate permease genes.
To verify this hypothesis, Enterobacter sp. YSU was grown in M-9 minimal medium in the presence and absence of L-cysteine. The untreated samples (NCNS and CNS) were harvested immediately before selenite was added, and the selenite treated samples (NCS and CS) were harvested one hour after selenite was added. Equal amounts of purified RNA were converted to cDNA. Then, PCR reactions using equal amounts of cDNA and primers specific for cysK, cysE, cysT, cysA and cysW were analyzed on a 1% agarose gel [fig_ref] Figure 4: Detection of the cysK, cysA, cysT, cysW and cysE transcripts using RT-PCR [/fig_ref]. All PCR products were between 730 and 930 bp in length, and each experiment was repeated 3 times. As expected for E. coli [bib_ref] Biosynthesis of cysteine, Kredich [/bib_ref] , the cysE transcript, which served as an internal control, was expressed equally under all 4 conditions whether L-cysteine was present or absent [fig_ref] Figure 4: Detection of the cysK, cysA, cysT, cysW and cysE transcripts using RT-PCR [/fig_ref]. The cysK transcript was sometimes but not always expressed at higher levels in the absence of Lcysteine (Lanes 1-2) than in the presence of L-cysteine (Lanes 3-4). The transcripts for cysA and cysT were consistently expressed at higher levels in the absence of Lcysteine (Lanes 1-2) than in its presence [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref]. Finally, the transcripts for cysW were not always detectable but when they were visible, they were only observed in samples from cells grown in the absence of L-cysteine (Lanes 1-2) and not in samples from cells grown in the presence of L-cysteine [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref]. Thus, it appeared that the addition of L-cysteine repressed the expression of the sulfate permease transcripts, cysA, cysT and cysW.
# Discussion
A previous proteomic study on selenite tolerant E. coli cells used M-9 minimal medium that lacked L-cysteine [bib_ref] Involvement of superoxide dismutases in the response of Escherichia coli to selenium..., Bebien [/bib_ref]. When two derivative K12 strains, MC4100 and GC4468, growing at logarithmic phase were pretreated with 0.25 mM selenite for 60 minutes and then exposed to 25 mM selenite, they were able to tolerate the increased selenite concentrations and were even able to grow for one hour. After this point, the cells entered into a stationary phase and began to die. This increased tolerance was attributed to higher expression levels of superoxide dismutase in response to selenite induced oxidative stress. The current work carried these E. coli studies one step further by demonstrating that Enterobacter sp. YSU was also sensitive to selenite in M-9 minimal medium, but the selenite inhibition was relieved by supplementing the medium with L-cysteine .
The role of L-cysteine in the observed selenite resistance/ sensitivity phenotypes was investigated using proteomics. Interestingly, spot 7 from cells grown under the selenite sensitive conditions (NCS) contained peptides for CysK [fig_ref] Figure 3: Negative images of Enterobacter sp [/fig_ref] [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref] , whereas the same spot from cells grown under selenite resistant conditions (CS) did not. This result supported the hypothesis that L-cysteine conferred selenite resistance to Enterobacter sp. YSU by preventing selenite uptake through the non-specific pathway or sulfate permease system. The presence of L-cysteine probably caused feedback inhibition of CysE, which converted serine to O-acetylserine [bib_ref] The path of unspecific incorporation of selenium in Escherichia coli, Muller [/bib_ref] [bib_ref] Biosynthesis of cysteine, Kredich [/bib_ref]. Then, lower levels of the O-acetylserine, the derivative of N-acetylserine, which acted as an inducer for the L-cysteine synthesis genes, resulted in decreased expression of cysK and the sulfate transport genes. Although the RT-PCR experiments did not conclusively show that the addition of L-cysteine to the growth medium reduced the level of cysK transcripts, they did suggest that L-cysteine indirectly repressed the expression of the sulfate transport genes. By decreasing the level of uptake through the sulfate permease system, L-cysteine allowed the cells to grow even when the selenite concentrations were 40 times higher than the sulfate concentrations.
Under the selenite sensitive conditions (NCS), Enterobacter sp. YSU appeared to respond to selenite by expressing a putative tellurium resistance protein C [bib_ref] The bacterial response to the chalcogen metalloids Se and Te, Zannoni [/bib_ref] [bib_ref] Iha: a novel Escherichia coli O157:H7 adherence-conferring molecule encoded on a recently..., Tarr [/bib_ref] in spot 8. Since tellurium is located directly under selenium in the periodic table of elements and is chemically similar to selenium, this protein may have been expressed nonspecifically in response to selenite but was not successful in conferring resistance. The association of a GroES-like protein with the tellurium resistance protein and the appearance of the small heat shock protein in unique spot 9 suggested that these cells were probably experiencing selenite-induced oxidative stress [bib_ref] Involvement of superoxide dismutases in the response of Escherichia coli to selenium..., Bebien [/bib_ref].
The function of other identified proteins in spots 7, 10 and 11, which were present at higher intensities in cultures grown in the presence of selenite (NCS and CS) than in cultures grown in the absence of selenite (NCNS and CNS), was not clear. They all contained peptides for 3 proteins: outer membrane protein II (OmpA) from E. aerogenes [bib_ref] Primary structure of major outer membrane protein II (OmpA protein) of Escherichia..., Chen [/bib_ref] , OmpA from E. sakazakii and a putative component membrane hydrogenase of OmpA. Basic Local Alignment Search Tool (BLAST) analysis and the references associated with the BLAST results did not provide any additional information about these polypeptides. However, an anaerobic strain of Clostridium pasteurianum reduced selenite using a hydrogenase during anaerobic respiration, and other Gram negative bacteria such as Stenotrophomonas maltophilia and Enterobacter sp. SLD1a-1 reduced selenite and selenate and deposited elemental selenium just inside or outside the cell surface [bib_ref] Transformations of selenate and selenite by Stenotrophomonas maltophilia isolated from a seleniferous..., Dungan [/bib_ref] [bib_ref] Reduction of selenite to elemental selenium by Enterobacter cloacae SLD1a-1, Dungan [/bib_ref].
Since the three proteins in spots 7, 10 and 11 shared many of the same peptides [fig_ref] Table 1: Identification of select protein spots that appeared under all four conditions [/fig_ref] , they may form a single OmpA-like protein that reduced selenite to elemental selenium using the hydrogenase component. Further studies are needed to understand the role that they played in selenite resistance.
# Conclusion
Viable cell count and turbidometric growth curves in M-9 minimal medium showed that Enterobacter sp. YSU required L-cysteine to be resistant to 40 mM selenite. Selenite can enter E. coli through a specific, undefined pathway and a non-specific sulfate permease pathway [bib_ref] The path of unspecific incorporation of selenium in Escherichia coli, Muller [/bib_ref]. Proteomic and RT-PCR analysis of Enterobacter sp. YSU cultures grown in the absence of L-cysteine and presence of selenite (NCS) and in the presence of L-cysteine and selenite (CS) suggested that L-cysteine conferred selenite resistance by feedback inhibition of the synthesis of N-acetylserine. This intermediate in L-cysteine synthesis acted as an inducer for cysK and the sulfate permease genes, cysA, cysT, and cysW. The lower levels of inducer decreased the expression of sulfate permease and may have limited selenite transport into the cells through the non-specific pathway, allowing the bacteria survive. This work linked studies on selenite tolerance in M-9 medium lacking L-cysteine [bib_ref] Involvement of superoxide dismutases in the response of Escherichia coli to selenium..., Bebien [/bib_ref] with research on selenite transport in E. coli [bib_ref] The path of unspecific incorporation of selenium in Escherichia coli, Muller [/bib_ref].
# Methods
## Bacterial strain and media
M-9 minimal mediumwas described previously, and 5X M-9 Salts were obtained from Becton, Dickinson and Company (Sparks, MD). When required, M-9 medium was supplemented with 0.04 mg/ml L-cysteine (Fisher Scientific, Fair Lawn, NJ) and 40 mM sodium selenite (MP Biomedicals, Aurora, OH). Luria-Bertani (LB) mediumand agar were obtained from Fisher Scientific. Enterobacter sp. YSU was described previously [bib_ref] Comparison of two multimetal resistant bacterial strains: Enterobacter sp. YSU and Stenotrophomonas..., Holmes [/bib_ref].
## Viable cell count growth curves
Four Enterobacter sp. YSU cultures, two containing Lcysteine and two lacking L-cysteine, were grown in M-9 minimal overnight at 37°C and diluted 1:20 into corresponding fresh M-9 minimal medium containing or lacking L-cysteine. These four new cultures were grown at 37°C with shaking. After 1.5 hours of growth, selenite or an equal volume of water was added to give the following culture conditions: no L-cysteine and no selenite (NCNS), no L-cysteine and selenite (NCS), L-cysteine and no selenite (CNS), and L-cysteine and selenite (CS). Samples from each culture were removed every 45 minutes, serially diluted and plated on LB-agar medium in triplicate. Plates were incubated overnight at 37°C, and colony forming units (CFUs) were counted.
Detection of the cysK, cysA, cysT, cysW and cysE transcripts using RT-PCR
## Proteomic analysis growth curves
Two Enterobacter sp. YSU cultures, one containing Lcysteine and the other lacking L-cysteine, were grown in M-9 minimal overnight at 37°C and diluted 1:20 into corresponding fresh M-9 minimal medium containing or lacking L-cysteine. These new cultures were grown at 37°C with shaking, and turbidity was measured every 0.5 hour using a Klett Colorimeter with a KS-54 filter. After 2.5 hours of growth, the two cultures were divided into equal volumes. Sodium selenite or an equal volume of water was added to give the four NCNS, NCS, CNS and SC growth conditions. Immediately before and one hour after the addition of sodium selenite, samples were harvested by centrifugation at 5,000 × g and 4°C for 10 minutes, and cell pellets were stored at -80°C.
## Protein extraction
Cells were thawed and resuspended in lysing buffer con- [bib_ref] A rapid and sensitive method for the quantitation of microgram quantities of..., Bradford [/bib_ref] [bib_ref] Protein profiling of the dimorphic, pathogenic fungus, Penicillium marneffei, Chandler [/bib_ref] was used to determine protein concentrations before 2DGE analysis.
## Two-dimensional gel electrophoresis (2dge)
Isoelectric focusing [bib_ref] High resolution two-dimensional electrophoresis of proteins, O'farrell [/bib_ref] was carried out using a Bio-Rad Protean IEF Cell (Bio-Rad). A total of 150 μg of protein was separated using an 11 cm IPG Ready Strip (Bio-Rad) with a fixed pH range of 4-7. After active rehydration at 50 V and 20°C for 12 hours, the sample was focused, starting at 0 V and ending at 8,000 V for a total of 40,000 volt hours.
[fig] Figure 3: Negative images of Enterobacter sp. YSU total protein separated by 2DGE over a pI range of 4-7. Cultures were grown as in Fig 2. Cultures grown with No L-Cysteine and No Selenite (NCNS) and with L-Cysteine and No Selenite (CNS) were harvested after 2.5 hours of growth. The cultures grown with No L-Cysteine and Selenite (NCS) [/fig]
[fig] Figure 4: Detection of the cysK, cysA, cysT, cysW and cysE transcripts using RT-PCR. Cells were grown and harvested as in Fig 3. Equal volumes of cDNA synthesized from 0.5 μg of total RNA were used in PCR reactions containing primers specific for each gene, and 10 μl of each PCR reaction were analyzed by agarose gel electrophoresis. Lanes: (1) No L-Cysteine, No Selenite (NCNS); (2) No L-Cysteine, Selenite (NCS); (3) L-Cysteine, No Selenite (CNS); (4) L-Cysteine, Selenite (CS). The gene, cysE, was used as an internal control. [/fig]
[table] Table 1: Identification of select protein spots that appeared under all four conditions (NCNS, NCS, CNS and CS) a . [/table]
|
10.1074/jbc.M115.710632
|
CCBY
| null |
26786098
|
s2orc_pubmed_articles
|
Identification of Kaposi Sarcoma Herpesvirus (KSHV) vIRF1 Protein as a Novel Interaction Partner of Human Deubiquitinase USP7*
Viral interferon regulatory factor 1 (vIRF1), a Kaposi sarcoma herpesvirus protein, destabilizes p53 by inhibiting p53 acetylation and Hdm2 phosphorylation. This leads to increased ubiquitination and degradation of p53 by Hdm2, which cripples the cellular p53-mediated antiviral response. Ubiquitin-specific protease 7 (USP7) deubiquitinates p53 and Hdm2 and regulates their stability. We identified an EGPS consensus sequence in vIRF1, which is identical to that found in Epstein-Barr virus nuclear antigen 1 (EBNA1) that interacts with the N-terminal domain of USP7 (USP7-NTD). GST pulldown assays demonstrated that vIRF1 interacts with USP7-NTD via its EGPS motif. NMR heteronuclear single quantum correlation (HSQC) analysis revealed chemical perturbations after titration of USP7-NTD with vIRF1 44 SPGEGPSGTG 53 peptide. In contrast, these perturbations were reduced with a mutant vIRF1 peptide, 44 SPG-EGPAGTG 53 . Fluorescence polarization analysis indicated that the vIRF1 peptide interacted with USP7-NTD with a K d of 2.0 M. The crystal structure of the USP7-NTD⅐vIRF1 peptide complex revealed an identical mode of binding as that of the EBNA1 peptide to USP7-NTD. We also showed that USP7 interacts with vIRF1 in U2OS cells. Decreased levels of p53, but not Hdm2 or ataxia telangiectasia-mutated (ATM), were seen after expression of vIRF1, but not with a vIRF1 mutant protein. Our results support a new role for vIRF1 through deregulation of the deubiquitinating enzyme USP7 to inhibit p53-mediated antiviral responses.Human herpesviruses (HHVs) 2 are double-stranded DNA viruses, classified into ␣, , or ␥ subfamilies (1). To establish latency, HHVs suppress the host immune response to evade the crossmark
immune system [bib_ref] Kaposi's sarcoma-associated herpesvirus and innate immunity, West [/bib_ref]. In the latent state, herpesviruses express only a small number of proteins essential for suppressing the host immune system [bib_ref] Kaposi's sarcoma-associated herpesvirus and innate immunity, West [/bib_ref]. HHVs have evolved various mechanisms for host immune evasion including inhibition of cellular senescence and apoptosis as well as promoting cell proliferation [bib_ref] Implication of human herpesviruses in oncogenesis through immune evasion and suppression, Alibek [/bib_ref]. Another strategy to evade host immune surveillance is through expression of viral homologues of genes that are the host's first line of defense against viral infection such as interferons and interferon regulatory factors (IRFs) and therefore sabotage the function and regulation of the cellular proteins [bib_ref] The viral interferon regulatory factors of KSHV: immunosuppressors or oncogenes?, Jacobs [/bib_ref]. Deregulation of cellular proteins involved in growth control by HHVs has led to recognition of some HHVs as underlying agents of cancer [bib_ref] Identification of herpesvirus-like DNA sequences in AIDS-associated Kaposi's sarcoma, Chang [/bib_ref].
Kaposi sarcoma herpesvirus (KSHV), HHV-8, is the causative agent of Kaposi sarcoma, primary effusion lymphoma, and multicentric Castleman disease, which are especially prevalent in immunocompromised patients [bib_ref] The viral etiology of AIDS-associated malignancies, Angeletti [/bib_ref]. The KSHV vIRF1 protein is encoded by ORF K9 and is believed to have been acquired through molecular piracy [bib_ref] Antiviral activity of tumor-suppressor pathways: clues from molecular piracy by KSHV, Moore [/bib_ref]. vIRF1 contains two domains, an N-terminal DNA binding domain (DBD) and a C-terminal IRF interaction domain (see [fig_ref] FIGURE 1: vIRF1 interacts with USP7-NTD in vitro [/fig_ref] [bib_ref] The crystal structure of the DNAbinding domain of vIRF-1 from the oncogenic..., Hew [/bib_ref]. The vIRF1 DBD has ϳ40% sequence similarity to the DBDs of human IRF3 and IRF7 and contains a helix-turn-helix motif, which is common in IRFs and DNA-binding proteins [bib_ref] The crystal structure of the DNAbinding domain of vIRF-1 from the oncogenic..., Hew [/bib_ref]. vIRF1-mediated deregulation of IRF3 and IRF7 leads to disruption of cellular antiviral activity. vIRF1 is a potent inhibitor of the histone acetyltransferase activity of p300. It leads to hypoacetylation of histones and alteration of the chromatin structure, reducing expression of IFNs [bib_ref] Inhibition of p300 histone acetyltransferase by viral interferon regulatory factor, Li [/bib_ref]. vIRF1 also directly interacts with p53 and inhibits its acetylation by p300 [bib_ref] Inhibition of p53 tumor suppressor by viral interferon regulatory factor, Nakamura [/bib_ref]. Furthermore, vIRF1 has an inhibitory effect on ataxia telangiectasia-mutated (ATM) kinase [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref]. ATM activation leads to the phosphorylation of Ser 15 on p53 and Ser 395 on Hdm2. These phosphorylations disrupt the Hdm2 ubiquitination of p53, leading to its stabilization [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref] [bib_ref] Phosphorylation of Ser-20 mediates stabilization of human p53 in response to DNA..., Chehab [/bib_ref]. Also, phosphorylation of Ser 15 is a signal for p300 acetylation of p53, which is important for its stability and activation [bib_ref] Phosphorylation of p53 serine 15 increases interaction with CBP, Lambert [/bib_ref]. Thus, inhibition of ATM activity by vIRF1 increases ubiquitination and degradation of p53 by Hdm2 [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref].
Human ubiquitin-specific protease 7 (USP7), also known as herpes-associated ubiquitin-specific protease (HAUSP), is a deubiquitinating enzyme originally identified as an interacting protein with HSV-1 immediate early protein, ICP0 [bib_ref] Herpes simplex virus type 1 immediate-early protein Vmw110 binds strongly and specifically..., Meredith [/bib_ref]. USP7 harbors a TRAF-like domain in its N terminus (USP7-NTD), a central catalytic domain, and five ubiquitin-like folds in its C terminus (USP7-CTD) [fig_ref] FIGURE 1: vIRF1 interacts with USP7-NTD in vitro [/fig_ref] [bib_ref] Structure of the p53 binding domain of HAUSP/USP7 bound to Epstein-Barr nuclear..., Saridakis [/bib_ref] [bib_ref] Crystal structure of a UBP-family deubiquitinating enzyme in isolation and in complex..., Hu [/bib_ref] [bib_ref] Mechanism of USP7/HAUSP activation by its C-terminal ubiquitin-like domain and allosteric regulation..., Faesen [/bib_ref] (P/A/E)XXS motif [bib_ref] Structure of the p53 binding domain of HAUSP/USP7 bound to Epstein-Barr nuclear..., Saridakis [/bib_ref] [bib_ref] Molecular recognition of p53 and MDM2 by USP7/HAUSP, Sheng [/bib_ref] [bib_ref] Further insight into substrate recognition by USP7: structural and biochemical analysis of..., Sarkari [/bib_ref] [bib_ref] Structural basis of competitive recognition of p53 and MDM2 by HAUSP/USP7: implications..., Hu [/bib_ref]. USP7-NTD harbors a 164 DWGF 167 motif in its binding pocket where the Asp and Trp residues are essential for protein interaction [bib_ref] Molecular recognition of p53 and MDM2 by USP7/HAUSP, Sheng [/bib_ref]. The p53 tumor suppressor protein and its negative regulator, the E3 ligase, Hdm2, are substrates of USP7 and interact with the 164 DWGF 167 motif of USP7-NTD through their (P/A)XXS motifs [bib_ref] Molecular recognition of p53 and MDM2 by USP7/HAUSP, Sheng [/bib_ref] [bib_ref] Structural basis of competitive recognition of p53 and MDM2 by HAUSP/USP7: implications..., Hu [/bib_ref] [bib_ref] Deubiquitination of p53 by HAUSP is an important pathway for p53 stabilization, Li [/bib_ref]. Because Hdm2 has a higher affinity for USP7 than p53, reduction in cellular levels of USP7 leads to instability of p53, whereas its removal results in destabilization of Hdm2, which in turn stabilizes p53 [bib_ref] Loss of HAUSP-mediated deubiquitination contributes to DNA damage-induced destabilization of Hdmx and..., Meulmeester [/bib_ref] [bib_ref] Tumour suppression: disruption of HAUSP gene stabilizes p53, Cummins [/bib_ref]. USP7 also regulates the stability of HdmX, a homologue of Hdm2 [bib_ref] Loss of HAUSP-mediated deubiquitination contributes to DNA damage-induced destabilization of Hdmx and..., Meulmeester [/bib_ref]. USP7 is an essential component of the p53-Hdm2-HdmX pathway that maintains balance in the cellular levels of these proteins [bib_ref] MDM2 and MDMX: alone and together in regulation of p53, Shadfan [/bib_ref].
We identified a 47 EGPS 50 consensus sequence in vIRF1 that is identical to a motif reported in EBNA1 responsible for mediating its interaction with USP7-NTD. This led us to investigate whether vIRF1 interacts with USP7. We characterized the interaction between USP7-NTD and KSHV vIRF1 using GST pulldown and fluorescence polarization assays, and further mapped the binding interface by two-dimensional NMR HSQC. We also determined the crystal structure of a vIRF1 peptide with USP7-NTD. We confirmed that these two proteins interact in vivo and determined that expression of wildtype vIRF1 but not the 47 EGPS 50 deletion mutant, ⌬vIRF1, decreased cellular levels of p53, but not Hdm2 or ATM.
## Experimental procedures
Protein Expression-N-terminal hexahistidine-tagged WT and D164A/W165A mutant USP7-NTD were expressed from the pET15b vector as described previously [bib_ref] Molecular recognition of p53 and MDM2 by USP7/HAUSP, Sheng [/bib_ref]. His-tagged USP7-CTD was expressed from p15TV-L vector [bib_ref] Crystal structure of USP7 ubiquitin-like domains with an ICP0 peptide reveals a..., Pfoh [/bib_ref]. Fulllength His-USP7 in pFastBac was expressed in Spodoptera frugiperda (Sf9) cells as described previously [bib_ref] Protein interaction domains of the ubiquitin-specific protease, USP7/HAUSP, Holowaty [/bib_ref]. Full-length vIRF1 was synthesized by GenScript. ⌬vIRF1 (deletion of residues 45 PGEGPS 50 ) was made by ACGT Corp. (Toronto, Canada). The N-terminal GST fusion constructs were generated by PCR amplification corresponding to residues 1-90 of WT or ⌬vIRF1 and inserted between the BamHI and NdeI sites of the pGEX2-TK vector.
Protein Purification-Constructs were expressed in Escherichia coli BL21 (DE3) cells in Terrific Broth (BioShop) (except 15 N USP7-NTD, which was expressed in M9 medium with 0.7 g/liter 15 NH 4 Cl as the sole nitrogen source) with overnight 0.4 mM isopropyl -D-1-thiogalactopyranoside induction at 16°C. The cells were harvested, resuspended in 50 mM Tris, pH 7.5, 500 or 150 mM NaCl, 5 mM imidazole, 1ϫ protease inhibitor cocktail (1 mM benzamidine and 0.5 mM PMSF), and 1ϫ protease inhibitor tablet (Roche Applied Science cOmplete ULTRA Tablets), and lysed using sonication. The lysate was cleared and allowed to interact with nickel-nitrilotriacetic acid beads (Qiagen) for 1 h. Beads were washed extensively with 50 mM Tris, pH 7.5, 500 or 150 mM NaCl, and 20 mM imidazole. Proteins were eluted with the addition of 50 mM Tris, pH 7.5, 500 or 150 mM NaCl, and 250 mM imidazole.
Peptide Synthesis-Wild-type ( 44 SPGEGPSGTG 53 ) and mutant ( 44 SPGEGPAGTG 53 ) vIRF1 peptides were synthesized by CanPeptide Inc. (Montreal, Canada) with both N-terminal acetylation and C-terminal amidation to mimic the native pep-tides. FITC-labeled WT (FITC-Acp-44 SPGEGPSGTG 53 -NH 2 ) and mutant (FITC-Acp-44 SPGEGPAGTG 53 -NH 2 ) vIRF1 peptides were also synthesized by CanPeptide Inc. (Montreal, Canada).
Fluorescence Polarization Binding Assay-Both wild-type and mutant USP7-NTD were further purified by size-exclusion chromatography using a HiLoad 16/60 Superdex 200 (GE Healthcare) on an ÄKTApurifier 10 UPC system (GE Healthcare) in 150 mM NaCl and 50 mM Tris, pH 8.0. FITC-labeled WT (FITC-Acp-44 SPGEGPSGTG 53 -NH 2 ) and mutant (FITC-Acp-44 SPGEGPAGTG 53 -NH 2 ) vIRF1 peptides were initially dissolved in dimethyl sulfoxide to a final concentration of 10 mM. A 400 nM working stock of each peptide was prepared in assay buffer (50 mM Tris, pH 8.0, 150 mM NaCl, and 0.01% Triton X-100). 500 M of each protein was serially diluted and incubated with 40 nM of each peptide. 10 l of the above mixtures were transferred into a 384-well plate (Corning), and fluorescence polarization was measured on a Synergy H4 microplate reader (BioTek) with ex ϭ 485 nm and em ϭ 520 nm. Polarization values were analyzed by GraphPad Prism 5.0 using a one-site binding model to obtain the equilibrium dissociation constant (K d ). Data were calculated based on four individual experiments, and the standard deviation was calculated.
NMR Spectroscopy-NMR spectra were acquired at 25°C on a Bruker 700-MHz NMR spectrometer equipped with a triple resonance cryoprobe. Interaction of USP7-NTD with WT ( 44 SPGEGPSGTG 53 ) and mutant ( 44 SPGEGPAGTG 53 ) vIRF1 peptides was monitored by analyzing 1 H-15 N HSQC spectra. Briefly, 15 N-labeled USP7 was incubated with thrombin to cleave the hexahistidine fusion tag followed by dialysis into NMR buffer (200 mM NaCl, 25 mM sodium phosphate, pH 7.0, and 10 mM DTT). Up to 0.4 mM unlabeled wild type or mutant vIRF1 peptide was titrated in 0.2 mM 15 N-labeled USP7 (containing 10% D 2 O) up to 2:1 peptide:USP7 molar ratio. Spectra were processed with TopSpin 3.2 and analyzed with the SPARKY program [bib_ref] Add to subtract": a simple method to remove complex background signals from..., Ye [/bib_ref] [bib_ref] Rapid analysis of protein backbone resonance assignments using cryogenic probes, a distributed..., Monleón [/bib_ref].
GST Pulldown Assay-Cells expressing GST-tagged fusion proteins were lysed using sonication in 50 mM Tris, pH 7.5, 150 mM NaCl, 1ϫ protease inhibitor cocktail (1 mM benzamidine and 0.5 mM PMSF) and 1ϫ protease inhibitor tablet (Roche Applied Science cOmplete ULTRA Tablets). Lysate was cleared and allowed to interact with glutathione-Sepharose beads (GE Healthcare) for 1 h. The beads were washed extensively with 50 mM Tris, pH 7.5, and 150 mM NaCl. GST-tagged WT and deletion mutant vIRF1 1-90 were kept bound to the glutathione-Sepharose beads. Prior to the GST pulldown assays, USP7 proteins were dialyzed against 100 mM NaCl, 50 mM Tris, pH 8.0, 5% glycerol, 5 mM -mercaptoethanol and 1ϫ protease inhibitor tablet. 15 l of GST-tagged wild type and deletion mutant vIRF1 1-90 bound to glutathione-Sepharose resin were incubated with 5 nmol of purified full-length USP7, wild-type USP7-NTD, mutant USP7-NTD, or USP7-CTD for 2 h at 4°C. 5 nmol of GST alone bound to glutathione-Sepharose beads were used as a negative control. The mixtures were then transferred to micro-columns and washed extensively with assay buffer. The bound proteins were eluted with 20 mM reduced glutathione and detected by Coomassie Blue staining following SDS-PAGE.
Crystallization-Prior to setting up crystal trials, purified USP7-NTD was incubated with thrombin to remove the hexahistidine N-terminal fusion tag. USP7-NTD was further purified by size-exclusion chromatography using a HiLoad 26/60 Superdex 75 (GE Healthcare) on an ÄKTApurifier 10 UPC (GE Healthcare) in 500 mM NaCl and 20 mM Hepes, pH 7.5. USP7-NTD (100 mg/ml) was co-crystalized with at least 5-fold molar excess of vIRF1 ( 44 SPGEGPSGTG 53 ) peptide at 4°C using the hanging-drop vapor diffusion method. Rod-shaped crystals appeared after 4 days following one round of micro-seeding using USP7-NTD⅐UbE2E1 peptide crystal seeds in 30% PEG 4000, 0.1 M Tris, pH 8.5, and 0.2 M LiSO 4 [bib_ref] Ubiquitin-specific protease 7 is a regulator of ubiquitin-conjugating enzyme UbE2E1, Sarkari [/bib_ref].
X-ray Data Collection and Structure Determination-X-ray data were collected at the Advanced Photon Source (APS). Diffraction data were integrated and scaled using the autoPROC software [bib_ref] Data processing and analysis with the autoPROC toolbox, Vonrhein [/bib_ref]. The structure was determined by molecular replacement employing USP7-NTD (Protein Data Bank (PDB) ID 1YY6, without the EBNA1 peptide) as search model using CNS 1.3 [bib_ref] Version 1.2 of the Crystallography and NMR system, Brunger [/bib_ref]. The electron density was visualized and the vIRF1 peptide model was built using Coot [bib_ref] Coot: model-building tools for molecular graphics, Emsley [/bib_ref]. CNS was used for refinement and water picking at 1.5 Å resolution. The data collection and refinement statistics are shown in [fig_ref] TABLE 1 X: -ray data collection and refinement parametersNumbers in brackets refer to the highest... [/fig_ref]. [fig_ref] FIGURE 3: Crystal structure of the UPS7-NTD⅐vIRF1 peptide complex [/fig_ref] was prepared using PyMOL.
Structure Deposition-Coordinates and structure factors for the USP7-NTD⅐vIRF1 peptide complex have been deposited in the Research Collaboratory for Structural Bioinformatics (RCSB) under PDB ID 4YSI.
Cell Culture and Antibodies-Human osteosarcoma U2OS cells were grown in McCoy's medium supplemented with 10% FBS and 1 mg/ml penicillin-streptomycin. The antibodies were rabbit polyclonal against USP7 (Bethyl Laboratories, A300-033A), mouse monoclonal against USP7 (Millipore, 05-1946), mouse monoclonal against Myc (Millipore, 05-724), rabbit polyclonal against FLAG tag (Bethyl Laboratories, A190-102A), mouse monoclonal against FLAG tag (Sigma, F3165), mouse monoclonal against p53 (Santa Cruz, sc-126), rabbit polyclonal against phospho-p53 (Ser 15 ) (Cell Signaling, 9284), rabbit monoclonal against ATM (Cell Signaling, 2873), rabbit monoclonal against phospho-ATM (Ser 1981 ) (Cell Signaling, 5883), mouse monoclonal against MDM2 (Santa Cruz, sc-965), rabbit polyclonal against phospho-MDM2 (Ser 166 ) (Cell Signaling, 3521), and mouse monoclonal against GAPDH (Santa Cruz, sc-47724). To detect proteins of interest, HRP-conjugated antimouse IgG (Jackson ImmunoResearch Laboratories, 115-035-166) and anti-rabbit IgG (Jackson ImmunoResearch Laboratories, 111-035-003) antibodies were used.
Co-immunoprecipitation-Cells were transfected with pCMV/ N-Myc USP7 and pcDNA3.1/FLAG vIRF1 (wild type or EGPS deletion mutant) vectors (total of 10 g of DNA per 15-cm tissue culture plate) using the PolyJet transfection reagent according to the manufacturer's protocol (SignaGen Laboratories). Cells were harvested 48 h after transfection and lysed in radioimmunoprecipitation buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Nonidet P-40, 20% glycerol, 1ϫ protease inhibitor cocktail (Roche Applied Science)) followed by 2 s of sonication at 10% amplitude. Cell lysates were incubated with either rabbit polyclonal anti-USP7 or anti-FLAG primary antibodies overnight at 4°C followed by the addition of pre-cleared protein A/G PLUS-Agarose beads (Santa Cruz, sc-2003) for 1 h. Immunoprecipitates were washed with radioimmunoprecipitation buffer five times and then boiled in SDS sample buffer for 5 min at 95°C. Samples were resolved on 10% SDS-polyacrylamide gels and immunoblotted using the antibodies described above. The co-immunoprecipitation experiment for endogenous USP7 was carried out using the same protocol.
Immunoblotting-Cells were transfected with pcDNA3.1/FLAG empty vector and pcDNA3.1/FLAG vIRF1 (wild type or EGPS deletion mutant) vectors (1 g of DNA per 10-cm tissue culture plate) using the PolyJet transfection reagent according to the manufacturer's protocol (SignaGen Laboratories). Cells were harvested 24 h after transfection and lysed in radioimmunoprecipitation buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 0.5% Nonidet P-40, 1ϫ protease inhibitor cocktail (Roche Applied Science) and 1ϫ phosphatase inhibitor cocktail (Cell Signaling)) followed by 2 s of sonication at 20% amplitude. Supernatants were boiled in SDS sample buffer for 5 min at 95°C and resolved on SDS-polyacrylamide gels. Immunoblotting was performed using antibodies described above. Bands were quantified and normalized using ImageJ.
# Results
Identification of an EGPS Sequence in KSHV vIRF1-It is well established that many USP7 substrates including Hdm2, p53, and HdmX contain (P/A)XXS motifs for interaction with USP7-NTD through the highly conserved 164 DWGF 167 motif with dissociation constant (K d ) values ranging from 10 to . KSHV proteins including vIRF4 and LANA interact with the USP7-NTD 164 DWGF 167 motif through (P/A)XXS sequences [bib_ref] Bilateral inhibition of HAUSP deubiquitinase by a viral interferon regulatory factor protein, Lee [/bib_ref] [bib_ref] The ubiquitin-specific protease USP7 modulates the replication of Kaposi's sarcoma-associated herpesvirus latent..., Jäger [/bib_ref]. EBV EBNA1 also interacts with USP7-NTD, although with an unusual motif, EGPS, instead of (P/A)XXS [bib_ref] Structure of the p53 binding domain of HAUSP/USP7 bound to Epstein-Barr nuclear..., Saridakis [/bib_ref]. A recent study revealed that ORF45, a KSHV immediate early protein, also interacts with USP7-NTD through an EGPS sequence [bib_ref] A survey of the interactome of Kaposi's sarcoma-associated herpesvirus ORF45 revealed its..., Gillen [/bib_ref]. We utilized ScanProsite to search for viral proteins containing the DPGEGPST sequence of EBNA1 and identified 44 SPGEGPSG 51 from KSHV vIRF1 [bib_ref] ScanProsite: detection of PROSITE signature matches and ProRule-associated functional and structural residues..., De Castro [/bib_ref]. Alignment of EBV EBNA1, KSHV ORF45, and KSHV vIRF1 proteins showed conservation of the EGPS interaction motif, and therefore we hypothesized that vIRF1 may interact with USP7-NTD [fig_ref] FIGURE 1: vIRF1 interacts with USP7-NTD in vitro [/fig_ref].
## Usp7-ntd and virf1
Interact in Vitro-To test the binary interaction between vIRF1 and USP7, a series of GST pulldown assays were performed using GST-tagged vIRF1 1-90 . GST-vIRF1 1-90 interacted with both full-length USP7 and USP7-NTD but not USP7-CTD, confirming that vIRF1 binding is mediated through USP7-NTD [fig_ref] FIGURE 1: vIRF1 interacts with USP7-NTD in vitro [/fig_ref]. Neither full-length USP7 nor USP7-NTD was retained by GST alone, indicating that the interaction with GST-vIRF1 1-90 was specific. Deletion of residues 45 PGEGPS 50 (GST-⌬vIRF1 1-90 ) abolished its interaction with both full-length USP7 and USP7-NTD, confirming that vIRF1 interacts with USP7-NTD through its EGPS sequence, which is similar to the USP7 binding motif found in EBNA1 [fig_ref] FIGURE 1: vIRF1 interacts with USP7-NTD in vitro [/fig_ref]. We also performed GST pulldowns using USP7-NTD DW , a double mutant containing alanine mutations at residues Asp 164 and Trp 165 within the 164 DWGF 167 motif. As expected, USP7-NTD DW did not interact with vIRF1 [fig_ref] FIGURE 1: vIRF1 interacts with USP7-NTD in vitro [/fig_ref]. These pulldown assays demonstrated that USP7-NTD interacts specifically with vIRF1 in vitro.
Interaction between USP7-NTD and vIRF1-Fluorescence polarization was used to determine the dissociation constant between USP7-NTD and vIRF1 peptides. FITC-labeled WT ( 44 SPGEGPSGTG 53 ) and mutant ( 44 SPGEGPAGTG 53 ) vIRF1 peptides containing the potential USP7 interaction site were used in these assays. A fixed concentration of 40 nM FITClabeled vIRF1 peptide was titrated with increasing concentrations of USP7-NTD (up to 500 M) to detect the fluorescence polarization changes during formation of the protein⅐peptide complex. The dissociation constant of WT vIRF1 peptide was calculated to be 2.0 Ϯ 0.1 M, whereas the K d for the mutant peptide was calculated to be 46.1 Ϯ 4 M [fig_ref] FIGURE 2: The vIRF1 EGPS sequence is essential for interaction with USP7-NTD [/fig_ref]. Interaction was not observed between USP7-NTD DW and either WT or mutant vIRF1 peptides [fig_ref] FIGURE 2: The vIRF1 EGPS sequence is essential for interaction with USP7-NTD [/fig_ref]. The K d value of 2 M for the interaction between vIRF1 and USP7 compares well with that of EBNA1 and corroborates that KSHV vIRF1 interacts with USP7-NTD.
NMR Analysis of the USP7-NTD and vIRF1 Peptide Interaction-To further investigate the interaction between USP7-NTD and the vIRF1 peptide, two-dimensional NMR HSQC spectra of 15 N-labeled USP7-NTD were analyzed in the presence of unlabeled WT ( 44 SPGEGPSGTG 53 ) or mutant ( 44 SPGEGPAGTG 53 ) vIRF1 peptides. [fig_ref] FIGURE 2: The vIRF1 EGPS sequence is essential for interaction with USP7-NTD [/fig_ref] displays the overlay of USP7-NTD 1 H-15 N HSQC spectra for USP7-NTD alone (in black) and USP7-NTD titrated with 0.4 mM of wild-type (in red) or mutant (in cyan) vIRF1 peptides (2:1 peptide:USP7-NTD ratio). Strong perturbations were observed in the USP7-NTD resonances in the presence of WT vIRF1 peptide, whereas these disturbances were notably decreased in the presence of mutant vIRF1 peptide. Using previously assigned USP7-NTD spectra (15) (Biological Magnetic Resonance Data Bank (BMRB) Entry 6939), we were able to observe that upon vIRF1 peptide binding, disturbances in USP7-NTD spectra were mostly in -strand 7 residues such as Asp 164 , Trp 165 (shift was observed in indole side chain), Gly 166 , Phe 167 , Ser 168 , and Met 171 [fig_ref] FIGURE 2: The vIRF1 EGPS sequence is essential for interaction with USP7-NTD [/fig_ref]. Resonances of a few residues from 3, 4, and 6 were also affected including Met 100 , Phe 118 , and Ser 155 . These observations are very similar to those of USP7-NTD⅐EBNA1 interaction. However, in contrast to EBNA1, spectra of residues such as Asn 169 and Phe 170 from 7 and residues Met 102 and Phe 117 did not show any change upon the addition of vIRF1 peptide, suggesting a slightly different mode of interaction with USP7-NTD.
Crystal Structure of the USP7-NTD⅐vIRF1 Peptide Complex-To elucidate the molecular basis of this interaction in vitro, we co-crystalized USP7-NTD with the vIRF1 peptide, [bib_ref] The nuclear location of PML, a cellular member of the C3HC4 zinc-binding..., Maul [/bib_ref] SPGEG-PSGTG 53 . The structure of the USP7-NTD⅐vIRF1 peptide complex was determined using molecular replacement and refined to 1.5 Å resolution. USP7-NTD residues 54 -62 and 106 -111 are disordered and were not built in the final model. As shown previously, USP7-NTD forms a TRAF domain similar to that of tumor necrosis factor receptor-associated factor (TRAF) 2, which consists of an eight-stranded antiparallel -sandwich fold with a shallow groove on the surface [fig_ref] FIGURE 3: Crystal structure of the UPS7-NTD⅐vIRF1 peptide complex [/fig_ref] [bib_ref] Structural basis for self-association and receptor recognition of human TRAF2, Park [/bib_ref]. The conserved 164 DWGF 167 motif is found in -strand 7, which also contains a -bulge essential for its interaction with binding partners. The electron density allowed building of vIRF1 residues [bib_ref] The nuclear location of PML, a cellular member of the C3HC4 zinc-binding..., Maul [/bib_ref] 3C). Residues 52 and 53 of the vIRF1 peptide did not have interpretable electron density, suggesting that they do not make contact with USP7-NTD. There are several polar and non-polar interactions that occur between USP7-NTD and the vIRF1 peptide [fig_ref] FIGURE 3: Crystal structure of the UPS7-NTD⅐vIRF1 peptide complex [/fig_ref]. Ser 50 of vIRF1 forms H-bonds through its side chain hydroxyl and backbone amide group with Asp 164 of USP7-NTD -strand 7. It also forms a H-bond with Arg 104 of -strand 3. The Glu 47 side chain of vIRF1 forms a water-mediated H-bond with Trp 165 of USP7. Comparison between the EBNA1 and the vIRF1 peptides revealed that they are superimposable and make very similar contacts with USP7-NTD [fig_ref] FIGURE 3: Crystal structure of the UPS7-NTD⅐vIRF1 peptide complex [/fig_ref].
USP7 and vIRF1 Interact in Vivo-We investigated the vIRF1 interaction with USP7 in vivo. The ability of vIRF1 to interact with USP7 was examined by transfecting U2OS cells with Myctagged USP7 and FLAG-tagged WT vIRF1. After immunoprecipitation of the lysate with a USP7 antibody, immunoblotting with anti-FLAG led to identification of FLAG-tagged vIRF1 . Lysate incubated with rabbit IgG served as a negative control and did not show any interaction. We also examined whether the PGEGPS vIRF1 deletion mutant (⌬vIRF1) could be co-immunoprecipitated with USP7. U2OS cells were transfected with Myc-tagged USP7 and FLAG-tagged ⌬vIRF1. The lysate was immunoprecipitated with a USP7 antibody and immunoblotted with anti-FLAG; however, ⌬vIRF1 could not be detected . These results indicated that USP7 was interacting with vIRF1 but not ⌬vIRF1 in U2OS cells.
We also performed the reciprocal experiments in which immunoprecipitation of the lysate with anti-FLAG rather than anti-USP7 readily identified USP7 in complex with vIRF1, whereas ⌬vIRF1 was not able to pull down USP7 . Combined, these results confirmed that the EGPS residues in vIRF1 are essential for interaction with USP7 in vivo.
Further, we examined the interaction of vIRF1 with endogenous USP7. After transfection of U2OS cells with FLAG-tagged vIRF1, USP7 was immunoprecipitated with a USP7 antibody. Blotting for anti-FLAG readily identified vIRF1 . In a reciprocal experiment, immunoprecipitation of FLAG-tagged vIRF1 from transfected U2OS lysate also successfully led to detection of endogenous USP7. In each case, lysate incubated with rabbit IgG served as negative control that did not show any interaction.
Effect of vIRF1 on p53 Stability-USP7 stabilizes cellular levels of the tumor suppressor, p53, and its negative regulator, Hdm2, through deubiquitination. Reduction in cellular levels of USP7 leads to instability of p53 [bib_ref] Deubiquitination of p53 by HAUSP is an important pathway for p53 stabilization, Li [/bib_ref] , whereas deletion of USP7 leads to degradation of Hdm2, resulting in stabilization of p53 [bib_ref] Tumour suppression: disruption of HAUSP gene stabilizes p53, Cummins [/bib_ref]. We hypothesized that interaction between vIRF1 and USP7-NTD at the same binding site that is known to interact with p53 and Hdm2 should decrease the availability of USP7 and therefore lead to instability of p53. To test the effect of vIRF1 on p53, U2OS cells were transfected with expression vectors for FLAG-tagged vIRF1 or ⌬vIRF1. Endogenous levels of p53, Ser 15 -phosphorylated p53, and USP7 were detected by immunoblotting after transfection. As shown in , D and F, transfection of cells with vIRF1 resulted in a significant decrease in the level of p53 (about 50%) when compared with levels in cells transfected with empty vector (p Ͻ 0.01) or ⌬vIRF1.
Effect of vIRF1 on Hdm2 Stability-Considering that USP7 deubiquitinates and stabilizes Hdm2, we monitored Hdm2 as well as phospho-Hdm2 (Ser 166 ) in cells transfected with vIRF1 or ⌬vIRF1 to assure that changes in the levels of p53 were not a result of change in the stability of Hdm2. Phosphorylation at Ser 166 stabilizes Hdm2 and stimulates p53 ubiquitination [bib_ref] Stabilization of Mdm2 via decreased ubiquitination is mediated by protein kinase B/Akt-dependent..., Feng [/bib_ref]. To test our hypothesis, U2OS cells were transfected with expression vectors for FLAG-tagged vIRF1 or ⌬vIRF1. As shown in , expression of vIRF1 or ⌬vIRF1 had no effect on the endogenous levels of USP7, Hdm2, or Ser 166 -phosphorylated Hdm2 when compared with control. This observation suggests that change in the stability of p53 is not caused by changes in the stability of Hdm2 or USP7.
Effect of vIRF1 on ATM Activity-It was previously reported that vIRF1 expression inhibits auto-phosphorylation of ATM on Ser 1981 , which is important for ATM activation. Decreased ATM activity resulted in reduced Ser 15 phosphorylation of p53 [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref]. p53 phosphorylation decreases the Hdm2 affinity for p53 and leads to the stability of p53 [bib_ref] Requirement of ATM in phosphorylation of the human p53 protein at serine..., Nakagawa [/bib_ref]. As shown in , we observed decreased levels of Ser 1981 -phosphorylated ATM as well as Ser 15 -phosphorylated p53 in cells transfected with vIRF1 or ⌬vIRF1, suggesting that under our experimental conditions, ⌬vIRF1, which is incapable of binding to USP7, is still able to exert an inhibitory effect on ATM activation and Ser 15 phosphorylation of p53. This indicates that the reduced p53 levels observed with ⌬vIRF1 when compared with vIRF1 are mainly due to the interaction between vIRF1 and USP7. Therefore, we show that interaction of vIRF1 with USP7 leads to destabilization of p53.
# Discussion
Members of the herpesvirus family such as herpes simplex virus type 1, Epstein-Barr virus, Kaposi sarcoma herpesvirus, and cytomegalovirus have evolved to interfere with the USP7-p53-Hdm2 pathway by competitively binding to USP7 and hindering its interaction with cellular substrates or binding proteins [bib_ref] Herpes simplex virus type 1 immediate-early protein Vmw110 binds strongly and specifically..., Meredith [/bib_ref] [bib_ref] Structure of the p53 binding domain of HAUSP/USP7 bound to Epstein-Barr nuclear..., Saridakis [/bib_ref] [bib_ref] Crystal structure of USP7 ubiquitin-like domains with an ICP0 peptide reveals a..., Pfoh [/bib_ref] [bib_ref] Bilateral inhibition of HAUSP deubiquitinase by a viral interferon regulatory factor protein, Lee [/bib_ref] [bib_ref] The ubiquitin-specific protease USP7 modulates the replication of Kaposi's sarcoma-associated herpesvirus latent..., Jäger [/bib_ref] [bib_ref] Proteomic profiling of the human cytomegalovirus UL35 gene products reveals a role..., Salsman [/bib_ref]. HSV-1 ICP0 is an E3 ubiquitin ligase that ubiquitinates and causes degradation of host proteins [bib_ref] The herpes simplex virus type 1 (HSV-1) regulatory protein ICP0 interacts with..., Boutell [/bib_ref]. ICP0 interaction with USP7 leads to its rescue from self-ubiquitination and proteasomal degradation [bib_ref] Reciprocal activities between herpes simplex virus type 1 regulatory protein ICP0, a..., Boutell [/bib_ref]. UL35, a human CMV latent protein, interacts with USP7 and alters its subcellular localization [bib_ref] Proteomic profiling of the human cytomegalovirus UL35 gene products reveals a role..., Salsman [/bib_ref]. UL35 also co-localizes with and disrupts promyelocytic leukemia (PML) nuclear bodies. Other USP7-interacting viral proteins such as ICP0 and EBNA1 also disrupt PML nuclear bodies and induce degradation of PML proteins [bib_ref] The nuclear location of PML, a cellular member of the C3HC4 zinc-binding..., Maul [/bib_ref] [bib_ref] Epstein-Barr nuclear antigen 1 contributes to nasopharyngeal carcinoma through disruption of PML..., Sivachandran [/bib_ref]. PML nuclear bodies are primarily composed of PML protein and are important mediators of critical cellular processes including apoptosis, DNA repair, and intrinsic response to viral infection [bib_ref] Proteomic profiling of the human cytomegalovirus UL35 gene products reveals a role..., Salsman [/bib_ref] [bib_ref] Genome-wide screen of three herpesviruses for protein subcellular localization and alteration of..., Salsman [/bib_ref] [bib_ref] Structure, dynamics and functions of promyelocytic leukaemia nuclear bodies, Bernardi [/bib_ref]. PML nuclear bodies also recruit proteins such as p53 and p300, which interact with vIRF1 [bib_ref] Viral interferon regulatory factor 1 of Kaposi's sarcoma-associated herpesvirus binds to p53..., Seo [/bib_ref]. In KSHV-infected BCBL-1 cells, vIRF1 was shown to localize with PMLs during both lytic and latent cycles [bib_ref] Short duration of elevated vIRF-1 expression during lytic replication of human herpesvirus..., Pozharskaya [/bib_ref]. USP7 also associates with PML nuclear bodies [bib_ref] The nuclear location of PML, a cellular member of the C3HC4 zinc-binding..., Maul [/bib_ref] [bib_ref] The herpesvirus associated ubiquitin specific protease, USP7, is a negative regulator of..., Sarkari [/bib_ref].
Through analysis of viral protein sequences followed by biochemical, structural, and in vivo studies, we identified KSHV vIRF1 as a novel USP7-binding protein. Fluorescence polarization indicated that the vIRF1 peptide, 44 SPGEGPSGTG 53 , interacts with USP7-NTD with a K d value of 2.0 M when compared with substrates such as Hdm2 and p53 that bind with K d values varying between 10 and 45 M, respectively. A single point mutation of Ser to Ala in the vIRF1 peptide decreased the FIGURE 4. vIRF1, but not ⌬vIRF1, interacts with USP7 in vivo and destabilizes p53. A and B, overexpressed USP7 interacts with ectopically expressed vIRF1 but not EGPS deletion mutant vIRF1 (⌬vIRF1). U2OS cells were transfected with Myc-USP7 and FLAG-vIRF1 or FLAG-⌬vIRF1. Immunoprecipitation (IP) was carried out by incubating the lysate with USP7 (A) or FLAG (B) antibodies using rabbit IgG as the negative control. Immunoblotting (IB) was performed with antibodies against Myc and FLAG tags. C, endogenous USP7 interacts with vIRF1 in vivo. U2OS cells were transfected with FLAG-vIRF1. Immunoprecipitation was performed using USP7 or FLAG antibodies followed by immunoblotting using antibodies against USP7 and FLAG. D, effect of vIRF1 on ATM and p53. U2OS cells were transfected with FLAG-vIRF1 or FLAG-⌬vIRF1. After 24 h, immunoblotting was performed using antibodies against ATM, phospho-ATM (Ser 1981 ) (p-ATM (Ser1981)), USP7, p53, phospho-p53 (Ser 15 ) (p-p53 (Ser15)), and FLAG tag. Cells transfected with empty vector were used as negative control. GAPDH levels were detected as control for equal loading. E, effect of vIRF1 on Hdm2. The levels of USP7, Hdm2, and Ser 166 -phosphorylated Hdm2 (p-Hdm2 (Ser166)) were monitored after expression of vIRF1 or ⌬vIRF1 in U2OS cells. Cells were incubated for 24 h following transfection with FLAG-tagged vIRF1 and ⌬vIRF1. immunoblotting was performed on whole cell lysates using USP7, Hdm2, phospho-Hdm2 (Ser 166 ), and FLAG tag antibodies. F, -fold change was determined by normalizing p53 and Hdm2 ratios to the vector control ratio. A paired t test was performed to evaluate statistical significance of the changes in p53 and Hdm2 levels after the introduction of vIRF1 or ⌬vIRF1. Error bars indicate S.E. and are from three independently transfected U2OS lysates. FIGURE 5. A new role for vIRF1 in hijacking USP7 and degrading p53. vIRF1 disrupts the p53 signaling pathway during viral infection by 1) inhibiting ATM phosphorylation, 2) inhibiting p53 transcription activation, and 3) inhibiting p300 acetylation of p53. Our results suggest that vIRF1 also binds USP7 and decreases the availability of USP7 for deubiquitinating and stabilizing p53. dissociation constant ϳ25-fold, indicating the importance of Ser 50 in mediating interaction with the USP7-NTD binding site. The interaction between USP7-NTD DW and the vIRF1 peptide was completely abolished, confirming that the vIRF1 47 EGPS 50 sequence interacts with the USP7-NTD 164 DWGF 167 motif.
The vIRF1 peptide binds to a groove on the surface of USP7-NTD, forming identical interactions with USP7-NTD as those seen with the USP7-NTD⅐EBNA1 peptide complex structure. The HSQC NMR analysis revealed perturbations in USP7-NTD residues from -strands 3, 4, 6, and 7 upon the addition of the vIRF1 peptide. These residues correlate well with the residues seen in the crystal structure analysis, especially those found in -strand 7. The side chains of residues from -strands 3 and 7 form polar contacts with the vIRF1 peptide, whereas the remaining residue side chains, from -strands 4 and 6, are in close proximity to the vIRF1 peptide but do not make any polar contacts. Along with EBNA1 and ORF45, vIRF1 contains a negatively charged glutamic acid residue at the position of P/A in the (P/A)XXS USP7 binding motif. Glu rather than Pro or Ala in that position is thought to increase the affinity of EBNA1 for USP7-NTD, therefore allowing it to effectively compete with USP7 cellular substrates [bib_ref] Molecular recognition of p53 and MDM2 by USP7/HAUSP, Sheng [/bib_ref]. Comparing the interaction with USP7-NTD made by Glu rather than Pro or Ala shows an increased number of H-bonds formed with Glu to USP7-NTD. This unique mode of interaction achieved through the substitution of a Glu residue for Ala or Pro is advantageous for vIRF1, ORF45, and EBNA1 in that it provides them with a higher binding affinity. This empowers the virus to sufficiently disrupt USP7 deubiquitination of cellular substrates such as p53. Identification of yet another USP7-NTD-interacting herpesvirus protein suggests that these viruses target USP7 at least in part because it is a critical cellular regulator of p53.
Our in vivo data in U2OS cells indicate that vIRF1, but not ⌬vIRF1, leads to a decrease in the levels of cellular p53. Therefore, vIRF1 binds USP7-NTD through its EGPS sequence and decreases the availability of USP7 for binding and deubiquitinating p53. vIRF1 was previously reported to directly interact with p53 through its central region (residues 152-360), prevent p53 acetylation by p300, and thus inhibit p53 transcriptional activation [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref] [bib_ref] Viral interferon regulatory factor 1 of Kaposi's sarcoma-associated herpesvirus binds to p53..., Seo [/bib_ref]. It was also reported that vIRF1 led to increased ubiquitination and degradation of p53 [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref]. Our data suggest that increased ubiquitination and degradation of p53 can in part be attributed to vIRF1 hijacking USP7, thus decreasing the ability of USP7 to deubiquitinate and stabilize p53.
Because Hdm2, a p53 negative regulator, is also a substrate of USP7, we monitored cellular levels of Hdm2 to investigate whether the vIRF1-mediated decrease in the availability of USP7 also destabilized Hdm2. We were never able to detect any significant changes in the level of Hdm2, suggesting that the USP7-vIRF1 interaction preferably disrupted USP7 deubiquitination and stabilization of p53.
It was previously reported that vIRF1 inhibits Ser 1981 autophosphorylation (activation) of ATM, which prevents ATMmediated Ser 15 phosphorylation of p53 and leads to p53 degradation [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref]. We monitored ATM and Ser 1981 -phosphorylated ATM but did not observe any changes in their levels between vIRF1-and ⌬vIRF1-transfected cells; however, both showed lower levels when compared with mock-transfected cells, indicating that vIRF1 and ⌬vIRF1 are still able to interact with ATM and inhibit its activity. As it was previously reported that vIRF1 interacts with ATM through its CTD, whereas the USP7 binding motif that we identified is located N-terminal to its DBD, we were not expecting changes in ATM levels [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref].
High levels of vIRF1 expression in KSHV-infected BCBL-1 cells were only observed transiently during lytic infection [bib_ref] Short duration of elevated vIRF-1 expression during lytic replication of human herpesvirus..., Pozharskaya [/bib_ref]. However, low levels of vIRF1, localized with PML nuclear bodies, were reported during latency [bib_ref] Short duration of elevated vIRF-1 expression during lytic replication of human herpesvirus..., Pozharskaya [/bib_ref]. In KSHV-infected cells, p53 was only detectable in PML nuclear bodies during the lytic cycle, whereas latently infected cells did not show detectable levels of p53 [bib_ref] Short duration of elevated vIRF-1 expression during lytic replication of human herpesvirus..., Pozharskaya [/bib_ref]. These observations suggest that vIRF1 may prevent USP7 deubiquitination of p53 during latency as all three proteins are localized to PML nuclear bodies. However, it is also important that vIRF1-mediated disruption of USP7 regulation of p53 is only one of the many mechanisms used by vIRF1 and other KSHV-expressed proteins to weaken the cellular antiviral response [bib_ref] The viral interferon regulatory factors of KSHV: immunosuppressors or oncogenes?, Jacobs [/bib_ref].
KSHV vIRF4, ORF45, and LANA interact with and inhibit multiple members of the USP7-p53-Hdm2 pathway, indicating the importance of p53 degradation for KSHV to establish and maintain life-long latency in host cells [bib_ref] Inhibition of the ATM/p53 signal transduction pathway by Kaposi's sarcoma-associated herpesvirus interferon..., Shin [/bib_ref] [bib_ref] Bilateral inhibition of HAUSP deubiquitinase by a viral interferon regulatory factor protein, Lee [/bib_ref] [bib_ref] The ubiquitin-specific protease USP7 modulates the replication of Kaposi's sarcoma-associated herpesvirus latent..., Jäger [/bib_ref] [bib_ref] Distinct p53, p53:LANA, and LANA complexes in Kaposi's sarcoma-associated herpesvirus lymphomas, Chen [/bib_ref]. vIRF1 also uses multiple redundant mechanisms to combat the cell's antiviral response by inhibiting multiple targets of the USP7-p53-Hdm2 pathway. Under normal (unstressed) cellular conditions, USP7 predominantly stabilizes Hdm2, which permits p53 ubiquitination and degradation as low levels of p53 are required for normal cellular homeostasis . Upon cellular stress (DNA damage or viral entry into the nucleus), ATM kinase is activated, which leads to phosphorylation of p53, Hdm2, and HdmX. ATM-mediated Hdm2 phosphorylation leads to p53 stabilization. ATM-mediated phosphorylation of p53 decreases its affinity for Hdm2 and signals its acetylation and transactivation by p300, as well as its stabilization by USP7. vIRF1 is a potent inhibitor of ATM kinase activity, p300 acetyltransferase activity, and p53 transcriptional activity . We have shown that vIRF1 interaction with USP7 decreases p53 levels, suggesting that it also blocks USP7 deubiquitination and stabilization of p53. Our novel finding that vIRF1 also targets USP7 fits well with the mechanism that KSHV must destroy p53 for a successful lifelong infection.
Author Contributions-S. C. designed and performed biochemical and cell biological experiments. I. K. L. analyzed data and prepared the mechanism. O. E. performed initial cell biology experiments. S. F. and Y. S. performed and analyzed NMR experiments. V. S. conceived, designed, supervised, the study, determined the crystal structure, and wrote the manuscript with input from other authors. All authors analyzed the results and approved the final version of the manuscript.
[fig] FIGURE 1: vIRF1 interacts with USP7-NTD in vitro. A, schematic representation of USP7 and vIRF1 domain organization. USP7 contains an N-terminal substrate binding TRAF domain, a catalytic domain, and a C-terminal domain that is composed of five ubiquitin-like (Ubl) sub-domains. vIRF1 contains a DBD and an interferon association domain (IAD). The arrow indicates the location of identified EGPS residues in vIRF1 in an unstructured region close to the DBD. B, alignment of the USP7 binding motif of vIRF1 with the previously identified USP7 binding motifs of EBV EBNA1 and KSHV ORF45. Asterisks represent conserved residues, and the period represents semi-conserved substitution. C, GST pulldown assay with GST-vIRF1 1-90 , GST-⌬vIRF1 1-90 (deletion of 45 PGEGPS 50 ), or GST alone as negative control. Full-length USP7, USP7-NTD, USP7-CTD, and USP7-NTD DW were used as prey. L, load; E, elution. [/fig]
[fig] FIGURE 2: The vIRF1 EGPS sequence is essential for interaction with USP7-NTD. A, fluorescence polarization of FITC-labeled wild-type ( 44 SPGEGPSGTG 53 in black) or mutant ( 44 SPGEGPAGTG 53 in red) vIRF1 peptides with USP7-NTD and USP7-NTD DW . mP, milliPolarization. Error bars indicate means Ϯ S.D. B, dissociation constants (K d values) for the USP7-NTD interaction with FITC-labeled vIRF1 peptides. Average values with S.D. for three or more experiments are shown. C, overlay of two-dimensional 1 H-15 N HSQC correlation spectra of 15 N-labeled USP7-NTD (black) titrated with 1:2 molar ratio of unlabeled wild-type ( 44 SPGEGPSGTG 53 in red) or mutant ( 44 SPGEGPAGTG 53 in cyan) vIRF1 peptides. D, individual residues from USP7-NTD -strand 7 that showed the largest shifts are identified. [/fig]
[fig] FIGURE 3: Crystal structure of the UPS7-NTD⅐vIRF1 peptide complex. A, electrostatic surface representation of USP7-NTD showing the vIRF1 peptide 44 SPGEGPSGTG 53 in stick form (yellow). B, ribbon diagram of USP7-NTD (green) showing the vIRF1 peptide in stick form (yellow). C, electron density prior to the addition of peptide with the final peptide model (yellow) contoured at 1 . D, interactions formed between USP7-NTD (green) and the vIRF1 peptide (yellow). E, superposition of the vIRF1 ( 44 SPGEGPSGTG 53 , yellow) and EBNA1 ( 441 DPGEGPSTGP 450 , fuchsia) peptides. [/fig]
[table] TABLE 1 X: -ray data collection and refinement parametersNumbers in brackets refer to the highest resolution shell, 1.023-1.02 Å. sym ϭ ͚ ͉I Ϫ ͗I͉͘/͚I where I is the observed intensity and ͗I͘ is the average intensity from multiple observations of symmetry-related reflections. [/table]
|
10.1017/S0033291717001301
|
CCBY
|
6421839
|
28528584
|
s2orc_pubmed_articles
|
Autism risk following antidepressant medication during pregnancy
Background.-Previous studies have examined if maternal antidepressant medication during pregnancy increase the risk of autism spectrum disorder (ASD) in the offspring, but the results have been conflicting.Methods.-In a population-based cohort of 179 007 children born in 2006 and 2007 and followed through 2014 when aged 7 and 8, we estimated relative risks (RRs) of ASD and 95% confidence intervals (CIs) from Cox regression in children exposed to any antidepressant medication during pregnancy, and nine specific antidepressant drugs. Analyses were adjusted for potential confounders and were conducted in the full population sample, and in a clinically relevant sub-sample of mothers with at least one diagnosis of depression or anxiety during life.Results.-The adjusted RR of ASD in children of mothers who used antidepressant medication during pregnancy was estimated at 1.23 (95% CI 0.96-1.57), and at 1.07 (95% CI 0.80-1.43) in women with a history of depression or anxiety. Analyses of specific antidepressants initially revealed increased RRs of offspring ASD confined to citalopram and escitalopram (RR: 1.47; 95% CI 0.92-2.35) and clomipramine (RR: 2.86; 95% CI 1.04-7.82).Conclusion.-Medication with antidepressants during pregnancy does not appear to be causally associated with an increased risk of ASD in the offspring. Instead, the results suggest that the association is explained by factors related to the underlying susceptibility to psychiatric disorders. Based on these findings, the risk of ASD in the offspring should not be a consideration to withhold treatment with commonly used antidepressant drugs from pregnant women. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence, which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited. Full sample, N= 179 007 Clinical sub-sample b , N= 18 551 Forest plot Medication Exposed children Exposed children with ASD (%) Relative risk (95% CI) p value Bonferroni-Holm corrected p value c Exposed children Exposed children with ASD (%) Relative risk (95% CI) p value Bonferroni-Holm
# Introduction
It has been suggested that antidepressant medication in pregnant women may increase the risk of autism spectrum disorder (ASD) in the offspring. This hypothesis has been examined in several studies, yet the results have been mixed; some studies have observed an increased risk [bib_ref] Antidepressant use during pregnancy and childhood autism spectrum disorders, Croen [/bib_ref] [bib_ref] Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum..., Rai [/bib_ref] [bib_ref] In utero exposure to selective serotonin reuptake inhibitors and risk for autism..., Gidaya [/bib_ref] [bib_ref] Prenatal SSRI use and offspring with autism spectrum disorder or developmental delay, Harrington [/bib_ref] [bib_ref] Antidepressant use during pregnancy and the risk of autism spectrum disorder in..., Boukhris [/bib_ref] , and others have not [bib_ref] Use of selective serotonin reuptake inhibitors during pregnancy and risk of autism, Hviid [/bib_ref] [bib_ref] Antidepressant exposure in pregnancy and risk of autism spectrum disorders, Sorensen [/bib_ref] [bib_ref] Prenatal antidepressant exposure is associated with risk for attentiondeficit hyperactivity disorder but..., Clements [/bib_ref] [bib_ref] Absence of evidence for increase in risk for autism or attention-deficit hyperactivity..., Castro [/bib_ref] [bib_ref] Gestational exposure to selective serotonin reuptake inhibitors and offspring psychiatric disorders: a..., Malm [/bib_ref].
Pharmacotherapy plays a central role in the management of the depressive illness, and untreated depression has been associated with poor health outcomes in both mothers and offspring [bib_ref] The burden of depression and anxiety in general medicine, Lecrubier [/bib_ref] [bib_ref] Psychiatric disorders among offspring of depressed mothers: associations with paternal psychopathology, Marmorstein [/bib_ref] [bib_ref] Postpartum depression: current status and future directions, O'hara [/bib_ref] [bib_ref] The long-term psychiatric and medical prognosis of perinatal mental illness, Meltzer-Brody [/bib_ref]. With the introduction of selective serotonin re-uptake inhibitors (SSRIs), the last two decades have seen a substantial increase in prescriptions of antidepressant drugs [bib_ref] National patterns in antidepressant medication treatment, Olfson [/bib_ref] , also in pregnant women [bib_ref] Use of antidepressant medications during pregnancy: a multisite study, Andrade [/bib_ref] [bib_ref] Increase in use of selective serotonin reuptake inhibitors in pregnancy during the..., Bakker [/bib_ref]. Although SSRI antidepressants are typically well tolerated [bib_ref] Comparative efficacy and acceptability of 12 new-generation antidepressants: a multiple-treatments meta-analysis, Cipriani [/bib_ref] , the risk of adverse effects on the developing fetus has not been resolved.
An association between maternal antidepressant medication during pregnancy and offspring ASD could either be due to: (1) a direct effect of the drug or (2) factors that confound the antidepressant treatment and the outcome studied.
Previous reported associations between maternal antidepressant medication during pregnancy and ASD in the offspring have been limited to studies of antidepressant medications as a whole [bib_ref] Antidepressant use during pregnancy and childhood autism spectrum disorders, Croen [/bib_ref] [bib_ref] Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum..., Rai [/bib_ref] [bib_ref] Antidepressant exposure in pregnancy and risk of autism spectrum disorders, Sorensen [/bib_ref] [bib_ref] Prenatal antidepressant exposure is associated with risk for attentiondeficit hyperactivity disorder but..., Clements [/bib_ref] [bib_ref] Absence of evidence for increase in risk for autism or attention-deficit hyperactivity..., Castro [/bib_ref] , or SSRI antidepressants as a whole [bib_ref] Use of selective serotonin reuptake inhibitors during pregnancy and risk of autism, Hviid [/bib_ref] [bib_ref] In utero exposure to selective serotonin reuptake inhibitors and risk for autism..., Gidaya [/bib_ref] [bib_ref] Prenatal SSRI use and offspring with autism spectrum disorder or developmental delay, Harrington [/bib_ref] [bib_ref] Antidepressant use during pregnancy and the risk of autism spectrum disorder in..., Boukhris [/bib_ref] [bib_ref] Gestational exposure to selective serotonin reuptake inhibitors and offspring psychiatric disorders: a..., Malm [/bib_ref]. The risk of offspring ASD associated with specific antidepressant drugs is still unknown. If the risk is confined to certain drugs, the proportion of mothers treated with these drugs in prior studies may explain the varying results. Furthermore, recent studies indicate that psychiatric disorders share genetic determinants [bib_ref] Common genetic determinants of schizophrenia and bipolar disorder in Swedish families: a..., Lichtenstein [/bib_ref] [bib_ref] Genetic architectures of psychiatric disorders: the emerging picture and its implications, Sullivan [/bib_ref] [bib_ref] Parental psychiatric disorders and autism spectrum disorders, Jokiranta [/bib_ref]. Pregnant women suffering from a mental illness during pregnancy, for which antidepressants may be prescribed, likely carry genetic susceptibility that could be inherited by the child [bib_ref] Parental psychiatric disorders and autism spectrum disorders, Jokiranta [/bib_ref]. As such, the genetic susceptibility, rather than the medication, may completely or partially explain the increased risk of ASD in children of mothers using antidepressants during pregnancy. Yet, previous studies have not thoroughly examined to what extent the risk of offspring ASD may be influenced by underlying susceptibility to mental illness.
The aim of the study was to examine the association between maternal antidepressant medication during pregnancy and ASD in the offspring, investigating any antidepressant medication, and specific anti-depressant drugs. We performed an analysis using one of the most comprehensive databases for offspring ASD and parental disorder and medication data available. This allowed us to investigate specific antidepressant drugs used during pregnancy, and use detailed adjustment for parental psychiatric diagnoses. Moreover, we investigated both the full cohort and a clinically relevant sub-sample of mothers with any diagnosis of depression and/or anxiety in their lifetime.
# Methods population
A birth cohort based on all live-born children conceived from July 1, 2005 and born in 2006 and 2007 was established by linkage of Swedish National registers using the unique individual Swedish national registration number [bib_ref] The Swedish personal identity number: possibilities and pitfalls in healthcare and medical..., Ludvigsson [/bib_ref]. Offspring and mothers were identified in the Swedish Medical Birth Register that covers 99% of all births nation-wide since 1973 and provides information on gestational age at birth that were used to calculate the beginning of pregnancy [bib_ref] A quality study of a medical birth registry, Cnattingius [/bib_ref]. In Sweden, 95% of all pregnant women receive early second trimester ultrasonography, which provide the gestational age of the fetus with an error margin of ±7 days [bib_ref] Early dating by ultrasound and perinatal outcome. A cohort study, Hogberg [/bib_ref]. The fathers were identified using the Multi-Generation Register [bib_ref] The Swedish multi-generation register, Ekbom [/bib_ref]. To be included, the children had to have complete information on gestational age at birth and the identity of the father. The study was approved by the Regional Ethics Committee in Stockholm, Sweden.
## Exposures
The Swedish Prescribed Drug Register holds information on all dispensed prescription drugs in Sweden since July 1, 2005 along with drug name, prescription-and dispensation dates, and the Anatomical Therapeutic Chemical Classification System (ATC) code [bib_ref] The new Swedish Prescribed Drug Registeropportunities for pharmacoepidemiological research and experience from..., Wettermark [/bib_ref]. We identified dispensations of all psychotropic drugs prescribed in Sweden, including antidepressants, anxiolytics, stimulants, mood stabilizers, antipsychotics, and sedatives (online . The offspring were classified as unex-posed to antidepressants if they were born to mothers without any dispensation of an antidepressant with a medication period overlapping the pregnancy. Since the number of antidepressant dispensations was reduced during pregnancy, compared with prior or after (online , it was assumed that a single anti-depressant dispensation overlapping pregnancy could represent medication that was halted prior the pregnancy. Therefore, antidepressant-exposed offspring were grouped into those born to a mother with either: (1) one single medication dispensation overlapping the pregnancy or (2) two or more dispensations of antidepressants with medication periods overlapping pregnancy (online .
## Ascertainment of asd
The children were followed from birth through 2014 when aged 7 or 8. A clinically ascertained diagnosis of offspring ASD was identified in the Swedish Patient Register [bib_ref] External review and validation of the Swedish national inpatient register, Ludvigsson [/bib_ref] [bib_ref] Autism risk associated with parental age and with increasing difference in age..., Sandin [/bib_ref]. This register includes all inpatient psychiatric admissions since 1973 and all outpatient specialist admissions since 2001, and provides admission dates along with the main and eight secondary diagnosis codes in accordance with the International Classification of Disease (ICD). Autism spectrum disorder was defined by having at least one in-or out-patient specialist care admission between birth and end of follow-up at December 31, 2014 with an ICD-10 code according to: F84.0, F84.1, . The ASD diagnoses in the Swedish Patient Register has previously been validated, and published on extensively [bib_ref] Autism risk associated with parental age and with increasing difference in age..., Sandin [/bib_ref].
## Covariates
To adjust for potential temporal trends, the birth date of the offspring was included as number of days from January 1, 2005 to the birth date of the child. Maternal and paternal age at childbirth was categorized into below 20, 20-29, 30-39, and above 40 years of age. Maternal and paternal susceptibility to mental illness were ascertained based on having at least one psychiatric diagnosis in the Swedish Patient Register within several psychiatric disorder sub-groups at any time in life (online . The father's medication with any psychotropic drugs overlapping the pregnancy was also included, as well as mother's dispensations of other psychotropic medication that overlapped the pregnancy.
# Statistical analysis
Relative risks (RRs) of ASD and the associated Wald-type two-sided 95% confidence intervals (CIs) were estimated by the hazard ratios from Cox regression models. The Cox regression models were fitted using days since birth as the underlying time scale. Each child was followed from birth until a diagnosis of ASD, death as identified in the Statistics Sweden register of vital statistics, or end of follow-up at December 31, 2014 -whichever came first. The RR of ASD was calculated in: (1) offspring born to mothers with a singledrug dispensation with a medication period overlapping the pregnancy and (2) offspring born to mothers with at least two dispensations with medication periods overlapping the pregnancy, compared with offspring born to mothers without a dispensation with a medication period overlapping the pregnancy.
Analyses were conducted in: (A) the complete sample to provide a public health perspective and in (B) a clinically relevant sub-sample of children born to mothers with at least one diagnosis of depression or anxiety in their lifetime, since most individuals who receive treatment with antidepressants also suffer from those disorders. In this sub-sample, the control group corresponds to mothers who are more likely to be considered for antidepressant treatment, and may share a similar genetic susceptibility of mental illness as the medicated mothers.
First, we examined children of mothers medicated during pregnancy with any type of antidepressant. The RRs of ASD were calculated in a sequence of models with increasing degree of adjustment for potential confounding factors according to: model 1 crude analyses without covariate adjustment; model 2 analyses adjusted for including offspring birthdate, mother's and father's dispenses of other psychotropic medications during pregnancy, and maternal and paternal age; model 3 analyses with additional adjustment for any depressive diagnosis in the mother's life time; and model 4 analyses further adjusted for any diagnosis of specific psychiatric disorder sub-groups in either the mother and/or father's life time. The psychiatric disorder sub-groups included depression, anxiety disorders, schizophrenia, bipolar disorder, substance use disorder, compulsive disorder, attention-deficit hyperactive disorder, ASD, intellectual disability, and any other psychiatric diagnosis (online .
Secondly, we further calculated the RR of ASD among children of mothers treated exclusively with one of the nine most prevalent antidepressants in the sample: sertra-line, citalopram and escitalopram, fluoxetine, venlafaxine, paroxetine, clomipramine, amitriptyline and nortriptyline, duloxetine, and mirtazapine, compared with children of mothers not treated with antidepressants during pregnancy. These analyses were adjusted for all included covariates, corresponding to model 4 in the analyses of any antidepressant. For the analyses examining specific drugs, to protect against an inflated error rate as a result of performing many statistical tests, we additionally present multiplicity-corrected p values using the Bonferroni-Holm procedure [bib_ref] A simple sequentially rejective multiple test procedure, Holm [/bib_ref].
All tests of statistically significance were done at the nominal 5% level of significance. Data management and statistical analyses were done using SAS 9.4 and STATA/IC 14, respectively.
## Sensitivity analyses
The proportional hazards assumption for the Cox regression models was examined using Schoenfeld residuals [bib_ref] Proportional hazards tests and diagnostics based on weighted residuals, Grambsch [/bib_ref]. To account for potential within-family correlations in the data due to multiple births from the same parents, we used bootstrap techniques. The analyses were separately repeated for the outcome autistic disorder (ICD-10 F84.0). Potential sex specificity of associations was tested by analyses of male and female off-spring separately. Analyses were also conducted separately for SSRI antidepressants, non-SSRI antidepressants, and non-antidepressant psychotropic drugs. The role of socioeconomic status was examined by including an analysis adjusted for education length as a potential confounder. To further examine the effect of the underlying disorder susceptibility, we compared: (a) children born to mothers treated with antidepressants during pregnancy (N = 3982), to (b) children born to mothers with no psychotropic medication but diagnosed with: (1) at least one of type psychiatric disorder, (2) at least two psychiatric disorders, or (3) at least three psychiatric disorders according to online . presents descriptive data for the cohort. Among the 180 444 children conceived from July 1, 2005 and born up until December 31, 2007 in the Medical Birth Register, 1437 (0.7%) did not have complete data and were excluded from the statistical analyses. presents descriptive statistics for the 179 007 included children and their parents. At least one diagnosis of any psychiatric illness was observed in 47 204 (13.2%) parents. Autism spectrum disorder was observed in 1641 (0.9%) of the children, and among those, 1004 (61.0%) had a diagnosis of autistic disorder. Among the offspring, 2379 (1.3%) were born to a mother with a single antidepressant dispensation with a medication period overlapping pregnancy, and 3982 (2.2%) were born to a mother with at least two antidepressant dispensations with medication periods overlapping pregnancy.
# Results
Due to the ambiguity of drug exposure in children to mothers with only a single antidepressant dispensation overlapping pregnancy, the results focus on the children with at least two dispensations overlapping pregnancy, as compared with children without exposure to antidepressants. Results in the offspring of mothers with only a single dispensation overlapping pregnancy are presented in the online supplement (online .
## Risk of asd
In the full population sample, the crude RR of ASD in children of mothers with at least two dispensations of antidepressants overlapping pregnancy compared with unexposed children was estimated at 2.46 (95% CI 1.97-3.05). With adjustment for all included potential confounders, the RR was reduced to 1.23 (95% CI 0.96-1.57). In analyses confined to children of mothers with at least one diagnosis of depression or anxiety in their lifetime, the crude RR of ASD in children of mothers with at least two dispensations of antidepressants overlapping pregnancy was estimated at 1.30 (95% CI 0.99-1.71). With adjustment for potential confounders, the RR was reduced to 1.07 (0.80-1.43) .
## Risk of asd associated with specific medications
In the full population sample, the adjusted RRs of ASD in children of mothers with at least two dispensations of specific antidepressants overlapping pregnancy compared with unexposed children was estimated at 1.35 (95% CI 0.87-2.08) for sertraline, 1.71 (95% CI 1.16-2.51) for citalopram and escitalopram, 1.04 (95% CI 0.53-2.02) for fluoxetine, 1.22 (95% CI 0.54-2.75) for venlafaxine, 1.40 (95% CI 0.52-3.76) for paroxetine, 3.27 (95% CI 1.33-8.00) for clomipramine, 0.59 (95% CI 0.08-4.20) for amitriptyline and nortriptyline, and 1.53 (95% CI 0.38-6.23) for mirtazapine .
Within the sub-sample where both the medicated and non-medicated women had at least one diagnosis of depression or anxiety in their lifetime, the adjusted RRs of ASD in children of mothers with at least two dispensations of specific antidepressants overlapping pregnancy compared with unexposed children was estimated at 1.17 (95% CI 0.71-1.95) for sertraline, 1.47 (95% CI 0.92-2.35) for citalopram and escitalopram, 1.08 (95% CI 0.53-2.21) for fluoxetine, 0.88 (95% CI 0.32-2.38) for venlafaxine, 1.21 (95% CI 0.38-3.80) for paroxetine, 2.86 (95% CI 1.04-7.82) for clomipramine, and 1.00 (95% CI 0.14-7.24) for mirtaza-pine .
None of the analyses examining specific drugs, in the full sample and in the clinically relevant sub-sample, revealed statistically significant p values after multiplicity correction using the Bonferroni-Holm procedure [bib_ref] A simple sequentially rejective multiple test procedure, Holm [/bib_ref].
## Sensitivity analyses
Inspection of the Schonfeld residuals did not suggest any violation of the proportional hazards assumption (online . Analyses confined to specifically SSRI antidepressants, non-SSRI antidepressants, and non-antidepressant psychotropic drugs revealed results quantitatively similar to the analyses of any antidepressant (online . Confidence intervals of selected point estimates (RRs) calculated using bootstrap revealed close to identical results as the parametric Wald estimates (online . Analyses confined to autistic disorder showed results similar to that of ASD (online . Analyses of male and female offspring separately did not reveal any sex-specific association (online . The complementary analysis adjusting for education as a marker of socioeconomic status did not affect our results (online . Analyses with comparison groups without psychotropic medication, but with increasing number of diagnosed psychiatric diagnoses revealed that the RR of ASD in the offspring was closely correlated with the number of different psychiatric disorders diagnosed in the mothers (online Supplementary Figs. S11 and S12).
# Discussion
In this population-based, prospective cohort study of 179 007 children and their parents, we observed an increased RR of ASD in offspring of mothers treated with antidepressant medication during pregnancy compared with offspring of mothers not treated with antidepressants during pregnancy. However, detailed adjustments for confounding by parental psychiatric liability attenuated this risk. Moreover, when the analyses were restricted to mothers ever diagnosed with depression or anxiety, a likely target group for anti-depressant medication, the association between antidepressants in pregnancy and ASD were attenuated even further and close to none.
To our knowledge, this is the first study to examine the RR of offspring ASD associated with specific types of antidepressant drugs compared with children unex-posed to antidepressants. Among the nine studied drugs, only the SSRI antidepressants citalopram and escitalopram, and the tricyclic antidepressant clomipramine displayed an increased RR of ASD in the off-spring. While these findings could suggest that these particular medications may have a causal effect associated with increased risk of ASD in the offspring, these positive associations could also be due to residual confounding. Citalopram and escitalopram have similar mechanisms of action as sertraline and paroxetine, yet only citalopram and citalopram displayed an elevated RR of ASD in the offspring. Furthermore, this increase in RR was modest, not statistically significant in the clinically relevant subsample, and not statistically different from the results of the other specific antidepressants studied. The results of the present study strongly suggest that the associations between antidepressant treatment during pregnancy and offspring ASD are gradually attenuated with increasing covariate adjustment. Although the present study adjusts for a broad set of factors that may confound the association, the ability to capture confounding is not complete. Mothers treated with different antidepressant medications may be different in other aspects that the current study cannot adjust for. This is exemplified by clomipramine that is not recommended as a first line of treatment of depression, which therefor may be given to women with a more severe or complex form of depression. Further analysis of this treatment group indeed revealed an increased prevalence of both compulsive disorder and schizophrenia compared with women treated with other antidepressants (online .
To further investigate the role of psychiatric disorders in the association between antidepressant treatment during pregnancy and ASD in the offspring, we compared children exposed to antidepressants to children of mothers not using psychotropic medication during pregnancy, but with increasing mean number of psychiatric disorders (online . These analyses revealed that a higher mean number of different psychiatric disorders diagnosed during the mothers' lifetime was correlated with the RR of ASD in the offspring. This was also observed in analyses of specific antidepressant drugs (online Nevertheless, the relationship between treatment with specific antidepressants during pregnancy and the risk of ASD in the offspring should be examined in additional samples to shed more light on the association and the specific underlying disorders.
Prior research into the association between maternal antidepressant medication during pregnancy and off-spring ASD has delivered mixed and inconclusive results [bib_ref] Antidepressant use during pregnancy and childhood autism spectrum disorders, Croen [/bib_ref] [bib_ref] Use of selective serotonin reuptake inhibitors during pregnancy and risk of autism, Hviid [/bib_ref] [bib_ref] Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum..., Rai [/bib_ref] [bib_ref] Antidepressant exposure in pregnancy and risk of autism spectrum disorders, Sorensen [/bib_ref] [bib_ref] In utero exposure to selective serotonin reuptake inhibitors and risk for autism..., Gidaya [/bib_ref] [bib_ref] Prenatal SSRI use and offspring with autism spectrum disorder or developmental delay, Harrington [/bib_ref] [bib_ref] Prenatal antidepressant exposure is associated with risk for attentiondeficit hyperactivity disorder but..., Clements [/bib_ref] [bib_ref] Antidepressant use during pregnancy and the risk of autism spectrum disorder in..., Boukhris [/bib_ref] [bib_ref] Absence of evidence for increase in risk for autism or attention-deficit hyperactivity..., Castro [/bib_ref] [bib_ref] Gestational exposure to selective serotonin reuptake inhibitors and offspring psychiatric disorders: a..., Malm [/bib_ref].
The present findings may resolve this earlier ambiguity. Our findings of attenuated associations with incremental covariate adjustment are in line with several earlier studies [bib_ref] Use of selective serotonin reuptake inhibitors during pregnancy and risk of autism, Hviid [/bib_ref] [bib_ref] Antidepressant exposure in pregnancy and risk of autism spectrum disorders, Sorensen [/bib_ref] [bib_ref] Prenatal antidepressant exposure is associated with risk for attentiondeficit hyperactivity disorder but..., Clements [/bib_ref] [bib_ref] Absence of evidence for increase in risk for autism or attention-deficit hyperactivity..., Castro [/bib_ref] [bib_ref] Gestational exposure to selective serotonin reuptake inhibitors and offspring psychiatric disorders: a..., Malm [/bib_ref]. The present findings are, however, at odds with number of previous studies that report statistically sign-ificant associations. A study of 1054 Canadian children with ASD [bib_ref] Antidepressant use during pregnancy and the risk of autism spectrum disorder in..., Boukhris [/bib_ref] reported an unadjusted RR of ASD following SSRI treatment (RR = 2.27; 95% CI 1.48-3.46) that were partially attenuated after adjustments for factors, including maternal history of depression (RR = 1.75; 95% CI 1.03-2.97). Similarly sized RRs have also been reported in a Californian sample of 298 children with ASD (RR = 2.2; 95% CI 1.2-4.3) [bib_ref] Antidepressant use during pregnancy and childhood autism spectrum disorders, Croen [/bib_ref] , in a Danish sample of 5215 children with ASD (adjusted RR = 1.8; 95% CI 1.4-2.3) [bib_ref] In utero exposure to selective serotonin reuptake inhibitors and risk for autism..., Gidaya [/bib_ref] , in a US study of 421 male cases with ASD (adjusted RR = 2.91; 95% CI 1.07-7.93) [bib_ref] Prenatal SSRI use and offspring with autism spectrum disorder or developmental delay, Harrington [/bib_ref] , and in a Swedish study of 4429 cases with ASD (RR = 3.34; 95% CI 1.50-7.47) [bib_ref] Parental depression, maternal antidepressant use during pregnancy, and risk of autism spectrum..., Rai [/bib_ref].
The discrepancies between the results of previous studies and the present study could be due to several factors. If the present findings of different psychiatric disorder profiles in mothers treated with specific anti-depressants (online Supplementary Figs. S11 and S12) was comparable in previous studies, the proportion of these specific antidepressant drugs could explain the mixed results in previous studies. However, the limited adjustment for the parents' mental illnesses in previous studies may also explain the discrepancies.
Strengths of the study include a large prospective population-based sample of children and their parents with close to complete healthcare data coverage. Any estimate of prevalence of ASD is highly dependent on the birth year, follow-up time, and sex distribution in the sample, still our estimates agree well with other studies in the Nordic countries and in the USA [bib_ref] Autism risk associated with parental age and with increasing difference in age..., Sandin [/bib_ref]. Data from a healthcare system with equal access limit the risk of selection biases. Inclusion of children born during a limited period reduces potential confounding by factors that may vary over time. Examination of any diagnosis of mental illness in the parents' lifetime allowed detailed adjustment for confounding due to, e.g., genetic liability. The requirement of at least two dispensations overlapping the pregnancy further increased the specificity of the exposure. Furthermore, supplementary analyses of antidepressants divided into specifically SSRI anti-depressants and non-SSRI antidepressants, and nonantidepressant psychotropic drugs all revealed similar findings, lending further support to the interpretation that the mental illness confounds the association between maternal antidepressant medication during pregnancy and increased risk of ASD in offspring.
However, the findings should also be interpreted in light of some limitations. Overall, our study summarizes the results of a multitude of statistical tests, which always increase the risk of erroneous rejection of a specific hypothesis. When applying the Bonferroni-Holm procedure in the comparison of specific drugs compared with 'no treatment of antidepressants', no drugs were at a statistically significantly higher risk. Although drugs recorded in the Prescribed Drug Register have been prescribed and collected, we cannot be entirely sure to what extent the medication was actually consumed. To address this limitation, we divided the exposed children into those born to a mother with only a single dispensation, and those with at least two dispensations overlapping the pregnancy. The study focused on those with at least two dispensations, as a continuous dispensation pattern was assumed to better reflect an ongoing treatment. Although we adjusted for both parents' education level at childbirth, we could not adjust for the income of the family. While education level had no effect on the association in our Swedish sample, this may not be the case in cohorts from other countries. Moreover, the Patient Register does not provide information from the primary care, and as such, any diagnoses confined to a primary care setting will go undetected. Still, this will not affect the study's ability to detect children with an ASD, as they are referred to a specialist that is covered by the Patient Register. The lack of primary care data may, however, affect the study's ability to capture psychiatric diagnoses as measure of mental illness among the parents, and consequently the ability to adjust for potential confounding due to underlying mental illness diagnosed in a primary care setting only. Antidepressant prescriptions are accompanied with a diagnosis in Sweden, yet supplemental analyses show that about 25% of mothers treated with an antidepressant during pregnancy lack a psychiatric diagnosis in the Patient Register (online , which could indicate residual confounding. Moreover, the severity of the psychiatric disorders investigated was not known, nor if the mothers experienced ongoing episodes of the lifetime diagnosed disorders. It is likely that mothers treated with antidepressants during pregnancy experienced a more severe disorder than unmedicated mothers with a lifetime diagnosis of depression or anxiety. Finally, the current study cohort was restricted to children born in 2006 and 2007, and therefore is underpowered to perform family-control analyses, which can be valuable to address potential residual confounding due to shared familial and genetic factors.
# Conclusion
Medication with antidepressants during pregnancy does not appear to be causally associated with an increased risk of ASD in the offspring. Instead, the results suggest that the association is explained by factors related to the underlying susceptibility to psychiatric disorders. Based on these findings, the risk of ASD in the offspring should not be a consideration to withhold treatment with commonly used antidepressant from pregnant women.
# Supplementary material
Refer to Web version on PubMed Central for supplementary material. ASD, autism spectrum disorder; CI, confidence interval.
a
The sample consists of 179,007 children born during 2006 and 2007, of which 1,641 had been diagnosed with ASD. The figure presents relative risks of ASD and two-sided 95% confidence intervals in children of mothers with at least two dispensations of a specific antidepressant drug overlapping the pregnancy, compared with unexposed children. The analyses are adjusted for all included covariates, as in model 4 in . b
The clinical sub-sample consists of 18 551 children, of which 361 had been diagnosed with autism spectrum disorder. All mothers, both medicated and non-medicated, had at least one diagnosis of depression or an anxiety disorder in their lifetime (online . Thereby, the offspring of mothers with medication during pregnancy is contrasted with offspring of mothers that may share similar underlying factors.
c Analyses not adjusted for covariates. d Analyses adjusted for birthdate, maternal and paternal age, the father's psychotropic medication that overlapped the pregnancy, and the mother's one-time dispensations of psychotropic medication that overlapped the pregnancy.
e Analyses adjusted for the factors listed in footnote d , and for any diagnosis of depression in the mother's lifetime (yes/no) (see online Supplementary for specific diagnosis codes).
f Analyses adjusted for the factors listed in footnote d , and for any diagnosis of specific psychiatric disorder sub-groups in either the mother and/or father's life time (yes/no), including depression, anxiety disorders, substance use disorder, bipolar disorder, compulsive disorder, attention deficit hyperactivity disorder, autism spectrum disorder, intellectual disability, schizophrenia, and 'other psychiatric diagnosis' (see .
b
The clinical sub-sample consists of 18 551 children, of which 361 had been diagnosed with ASD. All mothers, both medicated and non-medicated, had at least one diagnosis of depression or an anxiety disorder in their lifetime (online . Thereby, the offspring of mothers with medication during pregnancy is contrasted with offspring of mothers that may share similar underlying factors.
|
10.1098/rsif.2011.0577
|
CCBY
|
3350721
|
22072449
|
s2orc_pubmed_articles
|
How to identify water from thickener aqueous solutions by touch
Water detection is one of the most crucial psychological processes for many animals. However, nobody knows the perception mechanism of water through our tactile sense. In the present study, we found that a characteristic frictional stimulus with large acceleration is one of the cues to differentiate water from water contaminated with thickener. When subjects applied small amounts of water to a glass plate, strong stick-slip phenomena with a friction force of 0.46 + 0.30 N and a vertical force of 0.57 + 0.36 N were observed at the skin surface, as shown in previous studies. Surprisingly, periodic shears with acceleration seven times greater than gravitational acceleration occurred during the application process. Finite-element analyses predicted that these strong stimuli could activate tactile receptors: Meissner's corpuscle and Pacinians. When such stimuli were applied to the fingertips by an ultrasonic vibrator, a water-like tactile texture was perceived by some subjects, even though no liquid was present between the fingertip and the vibrator surface. These findings could potentially be applied in the following areas: materials science, information technology, medical treatment and entertainment.How to identify water by touch Y. Nonomura et al. 1217
# Introduction
How do we recognize water through our five senses? For several decades, researchers have attempted to address this question [bib_ref] Has water a specific taste, Zotterman [/bib_ref] [bib_ref] Taste of water in cat: effect on sucrose preference, Bartoshuk [/bib_ref] [bib_ref] Cellular identification of water gustatory receptor neurons and their central projection pattern..., Inoshita [/bib_ref] [bib_ref] Drosophila hygrosensation requires the TRP channels water witch and nanchung, Liu [/bib_ref] [bib_ref] The molecular basis for water taste in Drosophila, Cameron [/bib_ref] [bib_ref] Innate recognition of water bodies in echolocating bats, Greif [/bib_ref] because obtaining water is one of life's most important activities. Recent neurophysiological studies have shown that humans perceive water as a sweet taste [bib_ref] A TAS1R receptor-based explanation of sweet 'water-taste', Galindo-Cuspinera [/bib_ref]. Almost all studies have focused on water perception by taste; however, humans can also recognize water by their tactile sense. Kajimoto and co-workers [bib_ref] Tactile perception of a water surface: contributions of surface tension and skin..., Sato [/bib_ref] found that mechanical stress on skin hair plays a major role in the perception of a liquid surface. We showed that water caused a stick-slip feel when a small amount was rubbed using a fingertip on artificial skin that mimicked the structure of human skin [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] [bib_ref] How do we recognize water and oil through our tactile sense?, Nonomura [/bib_ref]. The results of frictional analyses predicted that this stick-slip feel was caused by a drastic change in frictional resistance. Such stick-slip phenomena and an increase of friction force have been observed on wet skin surfaces [bib_ref] The skin and friction: deviations from Amonton's laws, and the effects of..., Comaish [/bib_ref] [bib_ref] Frictional properties of skin, Highley [/bib_ref] [bib_ref] Frictional properties of skin: proposal of a new approach, Koudine [/bib_ref] [bib_ref] Coefficient of friction: tribological studies in man-an overview, Sivamani [/bib_ref] [bib_ref] Tribology of skin, Sivamani [/bib_ref] [bib_ref] In vivo friction study of human skin: influence of moisturizers on different..., Ramalhoa [/bib_ref] [bib_ref] Friction and lubrication of human skin, Adams [/bib_ref] [bib_ref] Tribology of human skin and mechanical skin equivalents in contact with textiles, Derler [/bib_ref] [bib_ref] Skin friction blistering: computer model, Xing [/bib_ref] [bib_ref] Influence of epidermal hydration on the friction of human skin against textiles, Gerhardt [/bib_ref] [bib_ref] Friction and deformation behaviour of human skin, Kwiatkowska [/bib_ref] [bib_ref] Friction of human skin against smooth and rough glass as a function..., Derler [/bib_ref] [bib_ref] Influence of surface roughness, material and climate conditions on the friction of..., Hendriks [/bib_ref] [bib_ref] Fingertip moisture is optimally modulated during object manipulation, André [/bib_ref] [bib_ref] Effect of skin hydration on the dynamics of fingertip gripping contact, André [/bib_ref] [bib_ref] Understanding the friction mechanisms between the human finger and flat contacting surfaces..., Tomlinson [/bib_ref].
Living beings obtain important information from tactile stimuli through active movement. By means of active touch, much of the surrounding environment can be perceived in the absence of vision [bib_ref] Observations on active touch, Gibson [/bib_ref]. During active tactile sensation by rodents, whisker movements across surfaces generate complex whisker micro-motions that carry information about surface properties [bib_ref] Sparse temporal coding of elementary tactile features during active whisker sensation, Jadhav [/bib_ref]. To illustrate the mechanisms of tactile detection, not only forces on the skin, but also tactile behaviour or movement velocity must be evaluated because tactile sense depends on these factors [bib_ref] Hand movements: a window into haptic object recognition, Lederman [/bib_ref] [bib_ref] Tactual discrimination of softness, Srinivasan [/bib_ref]. Furthermore, activation of sensory receptors induced by external stimuli is important because all tactile sensation originates from perceptual information conveyed by these receptors. In human skin, there are four tactile receptors: Merkel's discs, Meissner's corpuscles, Ruffini endings and Pacinians [bib_ref] The roles and functions of cutaneous mechanoreceptors, Johnson [/bib_ref] [bib_ref] Mammalian somatosensory mechanotransduction, Tsunozaki [/bib_ref]. Each of these four receptors mediates specific portions of the overall threshold-frequency range [bib_ref] Four channels mediate the mechanical aspects of touch, Bolanowski [/bib_ref].
In the present study, 10 subjects identified whether the liquids on glass plates were water or thickener aqueous solutions based on their tactile sense. During the identification process, the movement velocity of the fingertips was evaluated by a high-speed camera (figure 1a). On the other hand, using strain gauges on two leaf springs, the friction metre measured friction and vertical forces when subjects applied liquid samples to the glass plate [bib_ref] How do we recognize biological materials by touch?, Miyashita [/bib_ref] [bib_ref] Relationship between tribological characteristics and perceived texture when humans touch artificial skin..., Kamikawa [/bib_ref]. These experimental evaluations would show the mechanical stimuli on skin surface and fingertip movement when the subjects touched water. Next, we simulated the stress distribution around the tactile receptors that occurred when the subjects applied water to a glass substrate with a fingertip. The strain energy density estimated by the finite-element model correlates with the frequency of nerve impulses at tactile receptors [bib_ref] An investigation of the mechanics of tactile sense using two-dimensional models of..., Srinivasan [/bib_ref] [bib_ref] 3-D finite-element models of human and monkey fingertips to investigate the mechanics..., Dandekar [/bib_ref] [bib_ref] Recent advances in biodynamics of human hand-arm system, Dong [/bib_ref]. We also confirmed that the strain energy density obtained by the present model is proportional to the frequency of nerve impulses in the previous study [bib_ref] Relationship between the structure of human finger tissue and the location of..., Maeno [/bib_ref] [bib_ref] Modeling of tactile texture recognition mechanism, Shirado [/bib_ref]. Furthermore, the tactile texture of water on the glass plates was displayed using a tactile display system equipped with an ultrasonic vibrator. Suitable conditions for the tactile display would reflect the characteristics of the tactile texture of water. Shiokawa et al. [bib_ref] Hybrid display of realistic tactile sense using ultrasonic vibrator and force display, Shiokawa [/bib_ref] , Winfield et al. [bib_ref] T-PaD: tactile pattern display through variable friction reduction, Winfield [/bib_ref] and Biet et al. [bib_ref] Discrimination of virtual square gratings by dynamic touch on friction based tactile..., Biet [/bib_ref] showed that a vibrating plate with a squeeze film is suitable for a haptic interface. The water-like tactile texture can arise from the mechanical stimulus because the characteristic friction phenomena are the most important factor of the tactile texture of water [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] [bib_ref] How do we recognize water and oil through our tactile sense?, Nonomura [/bib_ref].
# Material and methods
# Materials
Water was purified using a DX-15 demineralizer (Kurita Water Industries Ltd., Tokyo, Japan). The thickener Polyquaternium-10 (O-(2-hydroxy-3-(trimethylammonio) propyl) hydroxy cellulose chloride, POIZ C-60H) was obtained from Kao Co. (Tokyo, Japan). The aqueous samples, 0.15, 0.5, 1 and 2 wt% thickener solutions, were prepared using a Vortex Genie 2 mixer from Scientific Industries Inc. (NY, USA). These evaluation items were selected based on a preliminary test performed by three professionals: one was a researcher who had engaged in tactile evaluations for more than 10 years, and the other two were students who had studied tactile sense. The professionals stated that these thickener solutions were suitable for the items because the addition of a small amount of the thickener induced a drastic change on the tactile texture of water. Samples were heated at 608C to dissolve the Polyquaternium-10. The thickener ingredient was checked for human safety. The viscosities of the aqueous samples are as follows: 0.9 (water), 4.5 (0.15 wt% thickener solution), 110 (0.5 wt% thickener solution), 730 (1 wt% thickener solution) and 3400 mPa s (2 wt% thickener solution). The viscosity range of these thickener solutions covers many liquids that we touch in our daily life; for example, ethyl alcohol ¼ 1.2 mPa s at 208C; olive oil ¼ 84 mPa s at 208C; machine oil ¼ 661 mPa s at 168C, 127 mPa s at 388C; glycerin ¼ 954 mPa s at 258C.
## Tactile evaluations
Tactile evaluations and friction evaluations were carried out simultaneously as follows [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] [bib_ref] How do we recognize water and oil through our tactile sense?, Nonomura [/bib_ref]. Similarity with a standard sample (water), and seven tactile factors, including 'a cool feel', 'a fresh feel', 'a slippery feel', 'a sticky feel', 'a slimy feel', 'a rough feel' and 'a stick-slip feel' were evaluated for water and thickener aqueous solutions when subjects applied them to the glass plate installed on a friction evaluation system [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] [bib_ref] How do we recognize water and oil through our tactile sense?, Nonomura [/bib_ref]. The tactile evaluations were rated on a seven-point scale, where score 7 means 'exactly the same texture as water', while score 1 means 'exactly the opposite texture from water'. All evaluations were conducted according to the principles expressed in the Declaration of Helsinki. The subjects included five male students and five female students ranging in age from 20 to 23 years. The evaluations were carried out in a quiet room at 258C + 18C after the subjects washed both hands with commercial liquid hand soap. The relative humidity was 50 + 3% in the room. Subjects used their forefingers to rub 0.1 ml of a standard sample (water) followed by 0.1 ml of one of the evaluated samples on a glass plate. After filling in a questionnaire, the subjects washed their hands with water again. The unit process (i.e. applying the aqueous samples, filling in the questionnaire and washing the hands) was repeated for all samples. The order of the samples was random to eliminate order effects. During the evaluation, the composition of the samples was not revealed to the subjects. The subjects touched the liquid samples on the glass plate through a black-out curtain. The content of the test was announced previously. The subjects decided for themselves whether they would join our evaluation test.
## Frictional evaluation
In the present study, we used a friction evaluation system that simultaneously evaluated tactile sensation and friction properties [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] [bib_ref] How do we recognize water and oil through our tactile sense?, Nonomura [/bib_ref] [bib_ref] How do we recognize biological materials by touch?, Miyashita [/bib_ref] [bib_ref] Relationship between tribological characteristics and perceived texture when humans touch artificial skin..., Kamikawa [/bib_ref]. This device measured the friction and vertical forces using strain gauges on two leaf springs. In the present paper, the friction coefficient is defined as the ratio friction force/vertical force. The detection limits of the friction and vertical forces were 0.20 N and 0.08 N, respectively. The maximum measurable load of the device was 5 N, with a time resolution of 0.5 ms.
# Fingertip movement analysis
We observed fingertip movements with a high-speed camera, as shown in [fig_ref] Figure 1: Movement behaviours of a fingertip when a subject tested water or thickener... [/fig_ref]. The high-speed images were taken using an EX-F1 high-speed video camera (Casio, Tokyo, Japan) with a frame rate of 600 frames s 21 and a space resolution of about 200 mm pixel 21 . Fifteen black dots of 1 mm in diameter were plotted in oil-based ink at intervals of approximately 4 mm to follow the movement. The skin surfaces on the glass plates were illuminated with a video light VL-G151 (lamp: halogen 150 W; colour temperature: 3075 K, LPL Co., Tokyo, Japan). Movement distance, velocity and acceleration of the fingertips were analysed using the two-dimensional movement analysis software Move-tr/2D v. 7.0 (Library Co., Tokyo, Japan) by the centre of gravity method. In the analysis, the values of movement for the centred dot were selected as representative values because significant differences were not observed between the movements of the 15 dots. The measurement error of the movement analysis reflects on the noises of velocity and acceleration. The noises of the velocity and acceleration were about 0.025 ms 21 and 20 ms 22 , respectively.
## Simulations using the finite-element method
The stress distribution around tactile receptors in the skin was simulated using a previously reported method [bib_ref] Relationship between the structure of human finger tissue and the location of..., Maeno [/bib_ref] [bib_ref] Modeling of tactile texture recognition mechanism, Shirado [/bib_ref]. We analysed finger deformation using the finite-element analysis software MARC (MSC Software Co., CA, USA). Electronic supplementary material, figure S1 shows a mesh model of a finger section that mimics the structure of an index finger. The plane strain element is used because the deformation outside the modelled plane is negligible nodes at the surfaces of the nail and bone. The finger skin consists of stratum corneum, epidermis, dermis and subcutaneous tissue. In the model, the stratum corneum is divided into two layers, i.e. a soft inner layer and a hard outer layer to reflect the heterogeneity of the stratum corneum because the inner layer is more hydrated than the outer layer [bib_ref] Evaluation of the skin surface hydration in vivo by electrical measurement, Tagami [/bib_ref]. Some measurements predicted that the hydration softens the stratum corneum layers [bib_ref] Water increases the fluidity of intercellular membranes of stratum corneum: correlation with..., Alonso [/bib_ref]. There are papillae at the interface of the epidermis and dermis underneath the epidermal ridges. The nail and bone were not modelled because their Young's moduli are large compared with that of the skin. The four symbols in the figure represent the nodes where the four tactile receptors are located. The physical factors of these biological tissues are shown in electronic supplementary material, table S1; for example, longitudinal elastic moduli of the outer stratum corneum, inner stratum corneum, epidermis, dermis and subcutaneous tissue were 0.816, 0.408, 0.136, 0.080 and 0.034 MPa, respectively. These factors were determined based on the measured values of skins for human and guinea [bib_ref] Relationship between the structure of human finger tissue and the location of..., Maeno [/bib_ref] [bib_ref] Modeling of tactile texture recognition mechanism, Shirado [/bib_ref]. In the present simulations, the process consisted of two steps; in the first step, the finger model was moved 0.5 mm vertically towards the solid substrate for 0.125 s, whereas in the second step, it was moved horizontally for 0.875 s.
In the present simulations, the input factors that change for water and four thickener solutions were the friction coefficient and the movement velocity, which were determined from experimental results. On the basis of the experimental data, the friction coefficients were 0.84, 0.54, 0.29, 0.20 and 0.13 for water, and 0.15, 0.5, 1and 2 wt% for thickener solutions, respectively. The movement velocities of horizontal and vertical motions are shown in electronic supplementary material, . The velocity profiles were sinusoidal in acceleration/deceleration processes. Strain energy densities were filtered in consideration of the frequency characteristics of the tactile receptors. Partial fraction decomposition and inverse Laplace transform were carried out for the transfer functions of each receptor H(s) to obtain the time-varying function h(t). The filtered strain energy densities were obtained by the convolution. The filtering properties used for each tactile receptors were obtained from a previous study [bib_ref] Modeling of tactile texture recognition mechanism, Shirado [/bib_ref].
## Tactile display of water texture
Mechanical stimuli were applied to human skin by a tactile display system equipped with a Langevintype ultrasonic vibrator (SEDECO Co., Tokyo, Japan), an analogue I/O terminal (AIO-160802AY-USB, CONTEC Co., Osaka, Japan), a differential function generator (Wave Factory WF1946A, NF Co., Yokohama, Japan) and power amplifer (HSA4011, NF Co., Yokohama, Japan; electronic supplementary material, [bib_ref] Hybrid display of realistic tactile sense using ultrasonic vibrator and force display, Shiokawa [/bib_ref]. The frequency and maximum amplitude were 28.20 kHz and 20 mm, respectively. In our previous studies, the mechanical stimuli under these conditions were suitable to raise some tactile feels: bumpy/flat, rough/fine and hard/soft feels [bib_ref] Hybrid display of realistic tactile sense using ultrasonic vibrator and force display, Shiokawa [/bib_ref]. The width of the contact surface was 30 mm. As shown in electronic supplementary material, , the mechanical stimuli were applied on skin surface intermittently to reflect stick-slip phenomena on the glass. The amplitude was controlled with a pick-up coil sensor. The position of a finger was evaluated by an infrared (IR) marker, and an IR sensor was used to control the ultrasonic vibrations. The ultrasonic vibrator was switched on and off in conjunction with the moving distance determined with an IR marker [bib_ref] Hybrid display of realistic tactile sense using ultrasonic vibrator and force display, Shiokawa [/bib_ref]. We studied the effects of the term T, which was a total movement distance in an on-period and an off-period, and the duty ratio t on the tactile texture when (i) duty ratio t ¼ 0; (ii) T ¼ 1 mm; t ¼ 0.5; (iii) T ¼ 10 mm, t ¼ 0.5; (iv) t ¼ 1. If the velocity of the finger is assumed to be about 0.1 ms 21 based on the results of the fingertip movement analysis (electronic supplementary material, table S2), the friction forces change at 10 or 100 ms intervals. This interval roughly agrees with the interval of stick-slip motion when a subject touched water. Ten subjects evaluated the similarity of the ultrasonic vibrator oscillating under various conditions with standard samples (water and 2 wt% thickener aqueous solution). The tactile evaluations were rated on a seven-point scale: a score of seven indicated 'exactly the same texture as the standard sample', whereas a score of 1 indicated 'exactly the opposite texture of the standard sample'.
# Results
## Effects of friction and fingertip movement on tactile feels
Similarity with a standard sample (water), and seven tactile factors were evaluated for water and 0.15, 0.5, 1 and 2 wt% thickener solutions, when 10 subjects applied them to the glass plate. Electronic supplementary material, [fig_ref] Figure 4: Tactile stimulation of a fingertip by an ultrasonic vibrator [/fig_ref] shows the similarity scores of the five aqueous samples. The evaluation value of water was 6.60 + 0.49, which was the highest among the five samples. Here, the values following + are the standard deviations. The correlations between the similarity score and the other tactile evaluations were analysed based on the correlation coefficients, r. Of the seven factors, the stick-slip feel had the highest value, r ¼ 0.841, while the sticky feel showed a strong negative correlation, r ¼ 20.824. As shown in electronic supplementary material, [fig_ref] Figure 4: Tactile stimulation of a fingertip by an ultrasonic vibrator [/fig_ref] , the score of stick-slip feel (slimy feel) decreased (increased) with the thickener concentration. These results indicate that the stick-slip feel is the characteristic property of the texture of water on a glass plate [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] [bib_ref] How do we recognize water and oil through our tactile sense?, Nonomura [/bib_ref]. Live images of subjects applying water or the thickener aqueous solutions were obtained with a high-speed video camera. Slow-motion images (600 frames s 21 , 20 times slower than normal speed) are shown in the electronic supplementary material (videos S1 and S2). In the case of water, a regular pattern consisting of 'stick periods', when the fingertip remained stationary and 'slip periods', when it moved rapidly, was repeated regardless of the subject's motion. [fig_ref] Figure 1: Movement behaviours of a fingertip when a subject tested water or thickener... [/fig_ref] shows the examples of movement distance, velocity and acceleration of the fingertip. The profile of the movement distance (x) was a staircase pattern in which x increased discontinuously at several tens of millisecond intervals. During the slip periods, the average velocity and acceleration of the 10 subjects were 0.231 + 0.076 m s 21 and 66.0 + 16.4 m s 22 , respectively (electronic supplementary material, table S3). Surprisingly, the acceleration was about seven times greater than gravitational acceleration. During the slip periods, frictional stimuli with a friction force of 0.46 + 0.30 N and friction coefficient of 0.84 + 0.29 were applied to the fingertips at several tens of millisecond intervals (electronic supplementary material, . These stimuli caused a stick-slip feel, i.e. an intermittent friction feel, in eight of 10 subjects. In our knowledge, it is the first report on the high-speed observation when subjects identify water on solid substrates based on information from tactile stimuli through active movement.
Such stick-slip motion of the fingertips was not observed when the subjects applied aqueous solutions containing thickener (electronic supplementary material, video S2); instead, the fingertips slid on the glass plate smoothly without a change in acceleration (figure 1b). Electronic supplementary material, table S3 shows the effect of thickener concentration on the times and acceleration of stick-slip motions. These observed values (183 + 58 and 66.0 + 16.4 m s 22 , respectively, for water) decreased with increasing thickener concentration. The friction properties of human skin have been studied in vivo. As mentioned in §1, the friction depends on the hydration condition of human skin [bib_ref] The skin and friction: deviations from Amonton's laws, and the effects of..., Comaish [/bib_ref] [bib_ref] Frictional properties of skin, Highley [/bib_ref] [bib_ref] Frictional properties of skin: proposal of a new approach, Koudine [/bib_ref] [bib_ref] Coefficient of friction: tribological studies in man-an overview, Sivamani [/bib_ref] [bib_ref] Tribology of skin, Sivamani [/bib_ref] [bib_ref] In vivo friction study of human skin: influence of moisturizers on different..., Ramalhoa [/bib_ref] [bib_ref] Friction and lubrication of human skin, Adams [/bib_ref] [bib_ref] Tribology of human skin and mechanical skin equivalents in contact with textiles, Derler [/bib_ref] [bib_ref] Skin friction blistering: computer model, Xing [/bib_ref] [bib_ref] Influence of epidermal hydration on the friction of human skin against textiles, Gerhardt [/bib_ref] [bib_ref] Friction and deformation behaviour of human skin, Kwiatkowska [/bib_ref] [bib_ref] Friction of human skin against smooth and rough glass as a function..., Derler [/bib_ref] [bib_ref] Influence of surface roughness, material and climate conditions on the friction of..., Hendriks [/bib_ref] [bib_ref] Fingertip moisture is optimally modulated during object manipulation, André [/bib_ref] [bib_ref] Effect of skin hydration on the dynamics of fingertip gripping contact, André [/bib_ref] [bib_ref] Understanding the friction mechanisms between the human finger and flat contacting surfaces..., Tomlinson [/bib_ref]. On human skin, a glass slider was observed to exhibit a stick-slip motion that may be attributed to the accumulation of a water film, which becomes more pronounced with increasing sliding velocity [bib_ref] Friction and lubrication of human skin, Adams [/bib_ref].
## The stress distribution around tactile receptors
To show the effects of friction stimuli with acceleration on neural systems, the stress distribution around the tactile receptors in the skin was simulated using the finiteelement method. The strain energy density obtained by the simulation reflected the firing frequency. Amplitudes greater than a specific value produced one impulse every cycle [bib_ref] A model accounting for effects of vibratory amplitude on responses of cutaneous..., Freeman [/bib_ref]. The strain energy distribution images show a model finger press a rigid object 0.5 mm vertically for 0.125 s and then move it horizontally for 0.875 s (figure 2, electronic supplementary material, videos S3 and S4). For water, the energy density profile showed the spatio-temporal asymmetric properties with the strain energy concentrated in the direction of forward motion and periodic changes at several hundred millisecond intervals. In contrast, the energy density profile showed the symmetric properties for 2.0 wt% thickener aqueous solution. These energy profiles were filtered based on the response characteristics of each tactile receptor [bib_ref] Relationship between the structure of human finger tissue and the location of..., Maeno [/bib_ref] [bib_ref] Modeling of tactile texture recognition mechanism, Shirado [/bib_ref]. shows temporal changes in strain energies on the tactile receptors when subjects touched water or thickener aqueous solutions. Contact with these liquids excited all tactile receptors, and unusual patterns were induced on Meissner's corpuscles and Pacinians following the application of water. Namely, the strain energy on Meissner's corpuscles was 4 Â 10 25 J m 23 at 0.2 s and decreased with periodic changes at approximately 100 ms intervals. Additionally, the energy on Pacinians was 4 Â 10 26 J m 23 at 0.15 s and changed at intervals between several tens of millisecond and 100 ms. These characteristics of the strain energy profiles were abolished by increasing the thickener concentration. Although the strain energy was distributed on Merkel's discs and Ruffini endings, an unusual profile was not observed for water. These results predict that the acceleration, which was seven times greater than gravitational acceleration, could activate Meissner's corpuscles and Pacinians in predictable patterns. In our knowledge, this is the first report on the effects of mechanical stimuli induced by the contact with wet substrates on neural systems.
## Display of water-like tactile texture
Mechanical stimuli were then applied to human skin using a tactile display equipped with an ultrasonic How to identify water by touch Y. vibrator to mimic the texture of water. In this system, the friction force between the fingertip and the contact surface changed with the oscillation amplitude of the ultrasonic vibrator [bib_ref] Hybrid display of realistic tactile sense using ultrasonic vibrator and force display, Shiokawa [/bib_ref] [bib_ref] T-PaD: tactile pattern display through variable friction reduction, Winfield [/bib_ref] [bib_ref] Discrimination of virtual square gratings by dynamic touch on friction based tactile..., Biet [/bib_ref] ; the friction force on the oscillating vibrator was lower than that on the resting vibrator. The squeeze effect of ultrasonic vibration decreases the friction force. The similarity score with water was 2.3 + 1.5 when the vibrator did not oscillate, and 4.5 and 4.6 when the vibrator oscillated at a term T of 1 mm or 10 mm, respectively (electronic supplementary material, [fig_ref] Figure 4: Tactile stimulation of a fingertip by an ultrasonic vibrator [/fig_ref]. The highest similarity score was obtained when the vibrator oscillated intermittently. In addition, the similarity score for the 2 wt% thickener solution was highest when the duty ratio t was 1, i.e. when the vibrator was oscillating continuously. These results demonstrated that a water-like tactile texture could be achieved by applying shear force with a strong acceleration per several tens of millisecond. The similarity score of 4.6, however, was not statistically significant. This dissociation may arise from the fact that thermal sensation and tenderization of the skin and irregularity of the stick-slip pattern were neglected in the present system [bib_ref] The perceptions of liquidity, semiliquidity and solidity, Sullivan [/bib_ref] [bib_ref] Sweating on paws and palms: what is its function, Adelman [/bib_ref] (figure 4).
# Discussion
On the basis of the earlier mentioned results, we propose that the detection mechanisms that humans distinguish water from thickener aqueous solutions by the stick-slip motions with large acceleration per several tens millisecond and the firing of Meisner's corpuscles and Pacinians. The present mechanism is supported comprehensively by the results of three experiments and simulations: the frictional evaluation, movement analysis, finite-element analysis and tactile display. For example, the stick-slip motion with large acceleration was observed by both the frictional evaluations and fingertip movement analysis. The results of tactile display backed up the importance of the stick-slip motion to detect water. The experimental results of the frictional evaluation and movement analysis also contributed to show realistic stress distribution on tactile receptors in the finite-element analysis.
The application of a shear force with acceleration seven times greater than gravitational acceleration was characteristic for water and was abolished by the addition of a small amount of thickener. The quantitative analysis of finger movement on wet glass plate is achieved by our high-speed observation. This intermittent stimulation is caused by the stick-slip phenomenon, which is well known in the field of tribology. It is characterized by the phenomenon transitions between a static (solidlike) state and a kinetic (liquid-like) state and is observed when lower sliding velocity induces a larger frictional resistance. Water swells human skin and increases the contact area and friction force between skin and solid surfaces [bib_ref] Friction and lubrication of human skin, Adams [/bib_ref] [bib_ref] Influence of epidermal hydration on the friction of human skin against textiles, Gerhardt [/bib_ref]. Recently, André et al. [bib_ref] Effect of skin hydration on the dynamics of fingertip gripping contact, André [/bib_ref] reported that the skin hydration level markedly affected the dynamics of the contact encapsulated in the course of evolution from sticking to slipping. In contrast, stickslip motion is inhibited by thickener in contaminated water [bib_ref] Frictional properties of skin, Highley [/bib_ref]. Adams et al. showed that this intermittent motion may be attributed to the accumulation of a water film, which becomes more pronounced with increasing sliding velocity [bib_ref] Friction and lubrication of human skin, Adams [/bib_ref] [bib_ref] Understanding the friction mechanisms between the human finger and flat contacting surfaces..., Tomlinson [/bib_ref]. Such accelerations are common during tactile interactions; Maeno et al. [bib_ref] Control of grasping force by detecting stick/slip distribution at the curved surface..., Maeno [/bib_ref] proposed a method for controlling a grasping force using the strain distribution in relation to the stick/slip information at the surface of the elastic finger. Shao et al. [bib_ref] Finite element simulations of static and sliding contact between a human fingertip..., Shao [/bib_ref] predicted stress oscillations during sliding over a textured surface. Is our model that water is perceived by the firing of Meissner's corpuscles and Pacinians reasonable? Meissner's corpuscles are responsible for the perception of events that produce low-frequency, low-amplitude skin motion and detect microscopic skin motions. Pacinians are responsible for the perception of events transmitted to the hand as high-frequency vibrations and are sensitive to acceleration. It is reasonable that Pacinians respond to the stick-slip motion per several tens of millisecond because they are sensitive to vibrations of several hundred hertzs. The argument that Meissner's corpuscles receptors are slip detectors is based on the experiments by Srinivasan et al. [bib_ref] Tactile detection of slip: surface microgeometry and peripheral neural codes, Srinivasan [/bib_ref] , who showed that small surface features are required for slip detection and that the responses of Meissner's corpuscles account for the limits of slip detection. In this vein, the idea that water is perceived by the firing of Meissner's corpuscles and Pacinians is justified.
In the present study, we show that humans have a sophisticated system to recognize water from thickener aqueous solutions by touch: humans distinguish water from thickener aqueous solutions by the stick-slip motions with large acceleration and the firing of Meisner's corpuscles and Pacinians. These findings could be generalized when humans recognize water from other liquids that have a similar viscosity and water containing specific solutes. In previous papers, we showed that subjects can differentiate water from silicone oil, surfactant aqueous solutions and ethanol aqueous solutions whose viscosities were almost similar to water [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] [bib_ref] How do we recognize water and oil through our tactile sense?, Nonomura [/bib_ref]. This finding adds to our understanding of the perception mechanisms for water, which is the main component of the human body and an essential material for life. In day-to-day life, we frequently perceive water through our tactile senses. Of course, in our real-life situations, the stick-slip motion is one of the cues to identify water from other liquid materials. For example, although tactile discrimination of water from aqueous salt/alcohol solutions is difficult [bib_ref] Tactile texture perception of water on human skin, Nonomura [/bib_ref] , we can distinguish them with olfactory or visual cues. These findings will be useful when designing virtual reality systems to mimic the sensation of the texture of liquids. The ultrasonic vibrator oscillated under the following conditions: (i) no oscillation; (ii) wavelength T ¼ 1 mm, duty ratio t ¼ 0.5; (iii) T ¼ 10 mm, t ¼ 0.5; and (iv) t ¼ 1. Error bars denote standard deviations. The statistical significance was measured between similarity scores under four conditions (i) -(iv). The symbols single asterisk (*), triple asterisks (***) and n.s. mean p , 0.01, p , 0.001 and non-significant in t-tests, respectively.
[fig] Figure 1: Movement behaviours of a fingertip when a subject tested water or thickener solutions. (a) A photograph of an observation system with a high-speed camera; examples of (b) movement distance, (c) velocity and (d ) acceleration for water (red lines) and 2 wt% thickener solution (blue lines). [/fig]
[fig] Figure 3, Figure 2: (a) Strain energy density at Meissner's corpuscles, (b) Merkel discs, (c) Ruffini ending and (d ) Pacinians when a subject applied water and thickener solutions on glass. (a) Distribution of strain energy density under the skin surface when a subject applied water or (b) 2 wt% thickener solution. Scale bars, 1.5 mm. [/fig]
[fig] Figure 4: Tactile stimulation of a fingertip by an ultrasonic vibrator. (a) Photograph of the ultrasonic vibrator. (b) Similarity of the tactile texture induced by the ultrasonic vibrator to that of water or (c) 2 wt% thickener solution. [/fig]
|
10.1038/s41467-023-38659-3
|
CCBY
|
10250317
|
37291111
|
s2orc_pubmed_articles
|
Author Correction: Genome-wide association study of serum liver enzymes implicates diverse metabolic and liver pathology
## Supplementary information:
Genome-wide association study of serum liver enzymes implicates diverse metabolic and liver pathology 3.67e-13 ++ CHR:POS, chromosome:position. EA, effect allele. OA, other allele. EAF, effect allele frequency. P_het, P value for heterogeneity between UK BioBank and BioBank Japan. Gene tags: (e) exonic, (i) intronic, (u) upstream, (d) downstream, (inter) intergenic, (UTR) untranslated region. Dir, direction: + indicates that the variant increases the liver enzyme while -indicates that the variant decreases the liver enzyme. Direction of effect in UK BioBank is shown first, followed by effect in BioBank Japan. 1.06e-04 +-CHR:POS, chromosome:position. EA, effect allele. OA, other allele. EAF, effect allele frequency. P_het, P value for heterogeneity between UK BioBank and BioBank Japan. Gene tags: (e) exonic, (i) intronic, (u) upstream, (d) downstream, (inter) intergenic, (UTR) untranslated region. Dir, direction: + indicates that the variant increases the liver enzyme while -indicates that the variant decreases the liver enzyme. Direction of effect in UK BioBank is shown first, followed by effect in BioBank Japan. TMEM150B (e) TMEM150B 6.53E-05 ++ CHR:POS, chromosome:position. EA, effect allele. OA, other allele. EAF, effect allele frequency. P_het, P value for heterogeneity between men and women. Gene tags: (e) exonic, (i) intronic, (u) upstream, (d) downstream, (inter) intergenic, (UTR) untranslated region. Dir, direction: + indicates that the variant increases the liver enzyme while -indicates that the variant decreases the liver enzyme. Direction of effect in UK BioBank male is shown as effect in men followed by effect in women. Significantly positive GCP implies that the metabolic trait is causal for the liver enzyme, and negative GCP implies that the liver enzyme is causal for the metabolic trait. Rho represents the estimated genetic correlation between the liver enzyme and the metabolic trait. SE, standard error. .69 GCP, genetic causality proportion. Significantly positive GCP implies that the metabolic trait is causal for the liver enzyme, and negative GCP implies that the liver enzyme is causal for the metabolic trait. Rho represents the estimated genetic correlation between the liver enzyme and the metabolic trait. SE, standard error. Significantly positive GCP implies that the metabolic trait is causal for the liver enzyme, and negative GCP implies that the liver enzyme is causal for the metabolic trait. Rho represents the estimated genetic correlation between the liver enzyme and the metabolic trait. SE, standard error. 1.20E-01 Effect of each rank unit of polygenic risk score for a specific liver enzyme on cirrhosis and hepatic steatosis. Effects are represented as odds ratio (95% confidence interval).
## Supplementary
## Supplementary
## Supplementary
## Supplementary
## Supplementary
## Supplementary
Supplementary : Effects of liver enzyme-increasing variants on serum/plasma metabolites.
Associations between variants associated with alanine aminotransferase (ALT), aspartate aminotransferase (AST), or alkaline phosphatase (ALP) and serum/plasma metabolites. Data on associations between genetic variants and serum/plasma metabolite concentrations are from [bib_ref] Exome sequencing links corticospinal motor neuron disease to common neurodegenerative disorders, Novarino [/bib_ref]. Red indicates that the liver enzyme-increasing allele increases metabolite concentration, blue that it decreases it, and white that there is no significant association. A Bonferroni correction for 123 metabolites and 378 genetic variants (p < 1.1 x 10 -6 or |Z| > 4.88) was used. Hierarchical clustering of genetic variants was performed using Z scores for variant-metabolite associations as a distance matric. * rs58542926-C (TM6SF2) had opposite directions on ALT and ALP.
Supplementary : Associations between aspartate aminotransferase polygenic risk score and cirrhosis and steatosis.
(A-B) Association between percentile of aspartate aminotransferase polygenic risk score on (A) cirrhosis or (B) steatosis. All results are depicted as odds ratios for cirrhosis or steatosis relative to individuals in the 0-10th percentile of polygenic risk score, adjusted for sex, age, age 2 , and principal components 1-10.
[fig] 1 24 Supplementary, Figure 4: Descriptive summary of phenotypes in UK BioBank .................................................................................................................. 2 Supplementary Table 2: Genomic control parameters ...................................................................................................................................................... 3 SupplementaryTable 3: Alanine aminotransferase-altering alleles with heterogeneity between UK BioBank & BioBank Japan ................................ 4 Supplementary Table 4: Aspartate aminotransferase-altering alleles with heterogeneity between UK BioBank & BioBank Japan ............................. 5 Supplementary Table 5: Alkaline phosphatase-altering alleles with heterogeneity between UK BioBank & BioBank Japan........................................ 6 Supplementary Table 6: Alanine aminotransferase-altering alleles: BioBank Japan only................................................................................................ 7 Supplementary Table 7: Aspartate aminotransferase-altering alleles: BioBank Japan only ............................................................................................ 8 Supplementary Table 8: Alanine aminotransferase-altering alleles with heterogeneity between men and women in UK BioBank ............................ 9 Supplementary Table 9: Aspartate aminotransferase-altering alleles with heterogeneity between men and women in UK BioBank ...................... 10 Supplementary Table 10: Alkaline phosphatase-altering alleles with heterogeneity between men and women in UK BioBank ............................... 11 Supplementary Table 11: Phenome-wide association studies of alanine aminotransferase-increasing alleles ........................................................... 12 Supplementary Table 12: Phenome-wide association studies of aspartate aminotransferase-increasing alleles ....................................................... 14 Supplementary Table 13: Phenome-wide association studies of alkaline phosphatase-increasing alleles .................................................................. 16 Supplementary Table 14: Latent causal variable analysis of alanine aminotransferase and metabolic traits .............................................................. 17 Supplementary Table 15: Latent causal variable analysis of aspartate aminotransferase and metabolic traits .......................................................... 18 Supplementary Table 16: Latent causal variable analysis of alkaline phosphatase and metabolic traits ..................................................................... 19 Supplementary Table 17: Effect of all liver enzyme-increasing alleles on primary biliary cholangitis .......................................................................... 20 Supplementary Table 18: Effects of polygenic risk scores on cirrhosis and hepatic steatosis in Michigan Genomics Initiative ................................. 21 Supplementary Figure 1: Quantile-quantile plots ............................................................................................................................................................ 22 Supplementary Figure 2: Effects of liver enzyme-increasing variants on metabolic traits. ........................................................................................... 23 Supplementary Figure 3: Effects of liver enzyme-increasing variants on serum/plasma metabolites. ......................................................................... Associations between aspartate aminotransferase polygenic risk score and cirrhosis and steatosis. ................................ 25 Supplementary References ................................................................................................................................................................................................ 26 Supplementary [/fig]
[table] Table of Contents: Supplementary [/table]
[table] Table 1: Descriptive summary of phenotypes in UK BioBank [/table]
[table] Table 2: Genomic control parameters Genomic control parameters (lambda_GC) and intercept estimates for the UK BioBank, BioBank Japan, and the meta-analysis. Lambda-GC estimates are from METAL and intercept from LDpred. [/table]
[table] Table 3: Alanine aminotransferase-altering alleles with heterogeneity between UK BioBank & BioBank Japan [/table]
[table] Table 5: Alkaline phosphatase-altering alleles with heterogeneity between UK BioBank & BioBank Japan [/table]
[table] Table 6: Alanine aminotransferase-altering alleles: BioBank Japan only POS, chromosome:position. EA, effect allele. OA, other allele. EAF, effect allele frequency. Gene tags: (e) exonic, (i) intronic, (u) upstream, (d) downstream, (inter) intergenic, (UTR) untranslated region. [/table]
[table] Table 7: Aspartate aminotransferase-altering alleles: BioBank Japan only POS, chromosome:position. EA, effect allele. OA, other allele. EAF, effect allele frequency. Gene tags: (e) exonic, (i) intronic, (u) upstream, (d) downstream, (inter) intergenic, (UTR) untranslated region. [/table]
[table] Table 8: Alanine aminotransferase-altering alleles with heterogeneity between men and women in UK BioBank [/table]
[table] Table 9: Aspartate aminotransferase-altering alleles with heterogeneity between men and women in UK BioBank [/table]
|
10.1136/bmjopen-2018-025486
|
CCBY
|
6429860
|
30852546
|
s2orc_pubmed_articles
|
Assessment of breath volatile organic compounds in acute cardiorespiratory breathlessness: a protocol describing a prospective real-world observational study
# Abstract
Introduction Patients presenting with acute undifferentiated breathlessness are commonly encountered in admissions units across the UK. Existing blood biomarkers have clinical utility in distinguishing patients with single organ pathologies but have poor discriminatory power in multifactorial presentations. Evaluation of volatile organic compounds (VOCs) in exhaled breath offers the potential to develop biomarkers of disease states that underpin acute cardiorespiratory breathlessness, owing to their proximity to the cardiorespiratory system. To date, there has been no systematic evaluation of VOC in acute cardiorespiratory breathlessness. The proposed study will seek to use both offline and online VOC technologies to evaluate the predictive value of VOC in identifying common conditions that present with acute cardiorespiratory breathlessness. Methods and analysis A prospective real-world observational study carried out across three acute admissions units within Leicestershire. Participants with self-reported acute breathlessness, with a confirmed primary diagnosis of either acute heart failure, communityacquired pneumonia and acute exacerbation of asthma or chronic obstructive pulmonary disease will be recruited within 24 hours of admission. Additionally, school-age children admitted with severe asthma will be evaluated. All participants will undergo breath sampling on admission and on recovery following discharge. A range of online technologies including: proton transfer reaction mass spectrometry, gas chromatography ion mobility spectrometry, atmospheric pressure chemical ionisationmass spectrometry and offline technologies including gas chromatography mass spectroscopy and comprehensive two-dimensional gas chromatography-mass spectrometry will be used for VOC discovery and replication. For offline technologies, a standardised CE-marked breath sampling device (ReCIVA) will be used. All recruited participants will be characterised using existing blood biomarkers including C reactive protein, brain-derived natriuretic peptide, troponin-I and blood eosinophil levels and further evaluated using a range of standardised questionnaires, lung function testing, sputum cell counts and other diagnostic tests pertinent to acute disease. Ethics and dissemination The National Research Ethics Service Committee East Midlands has approved the study protocol (REC number: 16/LO/1747). Integrated Research Approval System (IRAS) 198921. Findings will be presented at academic conferences and published in peer-reviewed scientific journals. Dissemination will be facilitated via a partnership with the East Midlands strengths and limitations of this study ► A pragmatic real-world, prospective, observational study across three admission units that focuses on the systematic discovery and replication of volatile organic compound (VOC) in acutely breathless patients using both online and offline technologies. ► The proposed study is the largest of its kind in acute disease to characterise VOC with a range of additional assessments that will build a comprehensive phenotype of acute cardiorespiratory exacerbations. ► The proposed study will build an infrastructure for research and subsequent evaluation of VOC in interventional trials within acute cardiorespiratory exacerbations. ► Prior acute treatment exposure will need to be accounted for when evaluating potential discriminative biomarkers. ► VOC technologies are not currently suited for deployment in patients that are of high clinical acuity.
# Introduction
Breathlessness is a common symptom of cardiorespiratory illnesses that has a significant direct impact on patients' well-being as well as a substantial economic burden on healthcare systems. [bib_ref] Clinical and economic burden of dyspnea and other COPD symptoms in a..., Stephenson [/bib_ref] Although its aetiologies can be variable, exacerbations of common complex chronic cardiorespiratory conditions account for approximately 70% of acute presentations with breathlessness, namely exacerbations of asthma and chronic obstructive pulmonary disease (COPD), acute heart failure and community acquired pneumonia. [bib_ref] Top differential diagnoses in family medicine Dyspnoea can fam physician, Ponka [/bib_ref] Moreover, moderate and severe breathlessness is significantly associated with all-cause, cardiovascular and COPD mortality. [bib_ref] Dyspnea severity, changes in dyspnea status and mortality in the general population:..., Figarska [/bib_ref] As a consequence, symptomatic breathlessness warrants rapid evaluation and targeted diagnostics at presentation. Diagnostic evaluation of acute breathlessness is heavily reliant on blood-based biomarkers for example, C reactive protein (CRP), brain-derived natriuretic peptide (BNP), troponin and on occasions blood eosinophil levels. These biomarkers have clinical utility primarily in patients with single pathologies but have poor discriminatory power in patients with multifactorial presentations of acute breathlessness. [bib_ref] An official American Thoracic Society statement: update on the mechanisms, assessment, and..., Parshall [/bib_ref] There is therefore an unmet need for the development of sensitive and specific biomarkers that differentiate acute breathlessness from its recovery and the common cardiorespiratory conditions that present with acute breathlessness.
CRP plays an important role in diagnosing breathlessness caused by an underlying bacterial pneumonia, [bib_ref] BTS guidelines for the management of community acquired pneumonia in adults: update, Lim [/bib_ref] as well as predicting mortality in patients with chronic obstructive pulmonary disease (COPD). [bib_ref] C-reactive protein level predicts mortality in COPD: a systematic review and meta-analysis, Leuzzi [/bib_ref] BNP is routinely used in acute settings to support the diagnosis of acute heart failure. [bib_ref] B-type natriuretic peptide in heart failure, Moe [/bib_ref] The European Society of Cardiology recommends BNP threshold values of <100 pg/mL to rule out acute congestive cardiac failure and values >500 pg/mL as diagnostic of acute exacerbations of heart failure. [bib_ref] ESC Guidelines for the diagnosis and treatment of acute and chronic heart..., Ponikowski [/bib_ref] The role of peripheral blood eosinophil count in airway inflammation was poorly understood up until the second half of the 19th century when Paul Ehrlich, a German physician and Nobel prize winner, introduced eosin in his technique for white cell differentiation in 1879. [bib_ref] Charcot leyden crystals and crushmann spirals, Sakula [/bib_ref] Considerable advances in the field of airway inflammation and the role of eosinophils have taken place since. [bib_ref] The role of eosinophils in the pathophysiology of asthma, Calhoun [/bib_ref] [bib_ref] Eosinophilic phenotypes of airway disease, Pavord [/bib_ref] [bib_ref] Role of eosinophils in airway inflammation of chronic obstructive pulmonary disease, Tashkin [/bib_ref] More recently, Bafadhel et al [bib_ref] Blood eosinophils to direct corticosteroid treatment of exacerbations of chronic obstructive pulmonary..., Bafadhel [/bib_ref] suggested that peripheral blood eosinophil count can be used to direct corticosteroid therapy during COPD exacerbations in single-centre study.
Currently, blood biomarkers together with clinical, physiological and imaging parameters are used in diagnosing the cause of acute breathlessness. Blood biomarkers may be less specific as they originate far from the target organs of interest (the heart and the lungs in cardiorespiratory disease). Sputum, although potentially a more definitive lung-specific matrix, is comparatively difficult to obtain particularly in acutely unwell patients, limiting its use in acute disease and highlighting the need for better biomarkers. Ideally, these biomarkers would have the following characteristics: (1) they would originate from the target organ of interest, (2) they would significantly add value to conventional risk scoring and diagnostic algorithms in acute breathlessness, (3) they would be minimally invasive and suitable for rapid point of care diagnosis in emergency rooms and acute admissions units and (4) they would have diagnostic value in patients with multifactorial acute breathlessness.
Exhaled breath contains thousands of volatile organic compounds (VOCs) that reflect biological processes occurring in the host both locally in the airways and systematically offering the potential to develop more effective biomarkers in acutely breathless patients (figure 1).
The proposed programme of research will use a combination of offline and online technologies to identify and evaluate the diagnostic and prognostic value of VOC in patients with acute cardiorespiratory-related breathlessness (figure 2).
Exhaled breath analysis of volatile chemicals offers the potential to develop more effective biomarkers in acutely breathless patients. The use of breath analysis for disease diagnostics dates back to ancient Greeks where physicians used exhaled breath to diagnose different diseases. Breath odours allow correct associations to certain diseases. For example, the sweet smell of diabetic ketoacidosis, the fishy smell of breath associated to liver illness, the urinelike odour of kidney disease and the smell of the breath Open access of patients with lung abscesses caused by the proliferation of anaerobic bacteria. [bib_ref] Biopsy and noninvasive methods to assess progression of nonalcoholic fatty liver disease, Bedossa [/bib_ref] [bib_ref] Exhaled breath analysis by electronic nose in airways disease. Established issues and..., Fens [/bib_ref] [bib_ref] Volatile organic compounds as diagnostic biomarkers in gastrointestinal and liver diseases, Probert [/bib_ref] [bib_ref] Breath analysis in critically ill patients: potential and limitations, Schubert [/bib_ref] More recently, exhaled breath analysis has demonstrated early proof of concept in the diagnosis of acute heart failure, and ventilator associated pneumonia. [bib_ref] Surveillance for lower airway pathogens in mechanically ventilated patients by metabolomic analysis..., Fowler [/bib_ref] The validity of breath analysis has also been demonstrated in breathless children. [bib_ref] Metabolomics pilot study to identify VOC of childhood asthma in exhaled breath, Gahleitner [/bib_ref] This population is likely to prefer breath-based tests, as these are minimally invasive. Importantly, a variety of point-of-care sensors are now available to evaluate potential exhaled breath biomarkers in emergency care settings.
A study by Van Berkel et al 20 demonstrated the ability to distinguish COPD subjects from controls solely based on the presence of VOCs in breath, suggesting that analysis of VOC might be highly relevant for diagnosis of COPD. This established the basis of further studies of VOC in COPD [bib_ref] Exhaled volatile organic compounds for phenotyping chronic obstructive pulmonary disease: a cross-sectional..., Basanta [/bib_ref] [bib_ref] Exhaled volatile organic compounds discriminate patients with chronic obstructive pulmonary disease from..., Besa [/bib_ref] [bib_ref] Integrated statistical learning of metabolic ion mobility spectrometry profiles for pulmonary disease..., Hauschild [/bib_ref] [bib_ref] Study of 5 volatile organic compounds in exhaled breath in chronic obstructive..., Jareño-Esteban [/bib_ref] [bib_ref] Exhaled ethane, a marker of lipid peroxidation, is elevated in chronic obstructive..., Paredi [/bib_ref] recommending larger studies for validation.
Several other studies found that VOC profiling in diagnosing asthma is potentially feasible. [bib_ref] Profiling allergic asthma volatile metabolic patterns using a headspace-solid phase microextraction/gas chromatography..., Caldeira [/bib_ref] [bib_ref] Allergic asthma exhaled breath metabolome: a challenge for comprehensive two-dimensional gas chromatography, Caldeira [/bib_ref] [bib_ref] Volatile organic compounds in exhaled breath as a diagnostic tool for asthma..., Dallinga [/bib_ref] [bib_ref] An electronic nose in the discrimination of patients with asthma and controls, Dragonieri [/bib_ref] [bib_ref] Exhaled breath profiling enables discrimination of chronic obstructive pulmonary disease and asthma, Fens [/bib_ref] [bib_ref] Non-invasive phenotyping using exhaled volatile organic compounds in asthma, Ibrahim [/bib_ref] [bib_ref] Diagnostic performance of an electronic nose, fractional exhaled nitric oxide, and lung..., Montuschi [/bib_ref] This, however, has been done in relatively small numbers in stable disease.
Despite the novelty of non-invasive sampling technology and the growing interest in exhaled breath analysis, there remains a disappointing level of comparability across studies due to the lack of standardisation and appropriate data analysis methods. A recent systemic review by Anders Christiansen et al 33 compared 11 publications reporting very heterogeneous designs, methods, patient group sizes, data analytics and, consequently, quite varying results.
To our knowledge, no other large studies exploring the use of breath biomarkers in profiling acute breathlessness have been completed. Several studies have explored the use of electronic nose (eNose) in stable disease with good discriminatory power in COPD, 34 pneumonia 35 and heart failure 36 with relatively small sample size. While eNose has now been widely used in detecting various VOC patterns, gas chromatography mass spectroscopy (GC-MS), a largely validated methodology, remains the gold standard technique for detecting VOCs in exhaled breath. The focus of the current research study will be to evaluate acutely breathless cardiorespiratory patients using a combination of 'discovery' and near-patient care breath sampling technologies.
Medical Research Council (MRC) and Engineering and Physical Sciences Research Council have commissioned a series of molecular pathology nodes aimed at developing molecular signatures relevant to disease diagnosis and progression. This was triggered by the clear need for alliance between academic institutions, industry and National Health Service partners to enhance the benefits of stratified medicine for patients.University of Leicester and Loughborough University were awarded a joint molecular pathology node East Midlands Breathomics Pathology Node (EMBER), which this study forms a key part of.
## Methods and analysis study design
A prospective real-world observational study across three acute admissions units within Leicestershire (two adult admissions units and one children's assessment unit). The acute units routinely assess and treat cardiorespiratory admissions due to breathlessness in adults and children.
Participants with self-reported acute breathlessness, either requiring admission or a change in baseline treatment, will be screened for the study. Informed consent will be obtained in all participants following a clinical review by a senior decision maker within 24 hours of acute admission (figure 3).
## Open access
objectives Primary objective ► To evaluate the sensitivity, specificity, positive and negative predictive values of exhaled breath VOC biomarkers to differentiate acute breathlessness in cardiorespiratory patients.
Secondary objectives ► To replicate selected breath VOC biomarkers identified in acute breathlessness. ► To discover and replicate breath VOC biomarkers that differentiate the common cardiorespiratory conditions that cause acute breathlessness, specifically: (1) acute heart failure, (2) community-acquired pneumonia, (3) adult exacerbations of asthma and chronic obstructive pulmonary disease (COPD) and age-matched adults that do not have cardiorespiratory disease or breathlessness. ► To quantify the level of clinical uncertainty in the primary diagnosis using a 100 mm visual analogue scale (VAS) and independent clinical adjudication of case notes blinded to the following blood biomarkers:
(1) CRP, (2) BNP, (3) troponin-I and (4) blood eosinophils but not clinical history and acute presentation nor chest X-ray imaging. Potential discriminatory breath VOC biomarkers will be adjusted for clinical uncertainly in statistical models. ► To identify and replicate exhaled breath VOC biomarkers in school-age children treated in hospital for severe asthma attacks and compare these with age-matched healthy controls.
Exploratory endpoints (where applicable) ► To evaluate the dynamic profile of selected breath VOC between the acute state and the recovery state postexacerbation. ► To evaluate the relationship between exhaled VOC biomarkers and clinical outcomes including: [bib_ref] Clinical and economic burden of dyspnea and other COPD symptoms in a..., Stephenson [/bib_ref] hospital readmission at 30 days and 60 days postevent and (2) all-cause mortality over a 2-year period postadmission. ► To evaluate the relationship between breath VOC biomarkers and functional measures, for example, physical performance and activity. ► To explore potential breath VOC biomarkers of multifactorial acute breathlessness. ► To evaluate the relationship between diet, lifestyle and environment on breath VOC biomarkers.
sample size estimation Preliminary data were used to conduct sample size estimates from a cohort of acutely breathless patients admitted to acute admissions units over a 6-month period (February 2017-August 2017). One hundred and twelve adult participants (asthma: 46, community-acquired pneumonia: 26 and COPD: 22) and 18 healthy controls were used for the analysis. A panel of 10 prespecified aldehydes, based on literature search, [bib_ref] Non-invasive phenotyping using exhaled volatile organic compounds in asthma, Ibrahim [/bib_ref] were extracted from breath using GC-MS. The aldehydes were normalised to a common internal standard and were not background-subtracted.
A closed formula from Hsieh et al, [bib_ref] A simple method of sample size calculation for linear and logistic regression, Hsieh [/bib_ref] relating sample size to observable effect size, was used to calculate sample size Study flow chart. Figure outlines the patient journey from admission through to discharge and follow-up. Participants with self-reported acute breathlessness presenting to University Hospitals of Leicester are recruited within 24 hours following a senior decision maker review. Breath sampling is carried out on the first visit, at recruitment, and the second visit, up to 6 months after discharge. Patients are admitted through the standard operational emergency medical streaming and care pathways at the University Hospitals of Leicester National Health Service Trust. Early outcomes (hospital readmission) are measured at 30 days and 60 days and late outcomes (mortality) including respiratory and all-cause mortality are measured at 2 years. Assessments carried out at each time point are summarised in table 1. APCI-MS, atmospheric pressure chemical ionisation-mass spectrometry; GC-IMS, gas chromatography ion mobility spectrometry; GC-MS, gas chromatography mass spectroscopy; GC×GC-MS, two-dimensional gas chromatography-mass spectrometry.
Open access from logistic regression models of the ten aldehydes with acute breathlessness as the outcome measure. The sample size estimates are also relevant to acute class comparisons versus the sum of other acute classes.
Based on the sample size estimates, we would have an 80% power, with a type 1 error rate of 5%, to detect an OR of association of 1.2 between two disease classes with 55 patients per class. Given the fact that study seeks to discover and replicate breath VOC among five adult disease classes (community-acquired pneumonia, heart failure, COPD, asthma and healthy aged-matched subjects), we would require 110 adult patients per class-550 patients across the programme to achieve these aims.
The closed formulae by Hajian-Tilaki et al [bib_ref] Sample size estimation in diagnostic test studies of biomedical informatics, Hajian-Tilaki [/bib_ref] were also used to understand the discriminatory power that the samples sizes above would provide with respect to biomarker sensitivity and specificity; the following assumptions were made: ► That a sensitivity of 80% with a precision of 5% would provide a useful biomarker capable of 'ruling out' an acute class. The same target was applied to specificity. ► We assume a prevalence of acute breathlessness of 80% as the recruitment campaign uses acute breathlessness as the initial stratification tool for recruitment and 1:5 patients recruited will be non-breathless healthy controls. ► We aim to balance group sizes across classes equally. For a type 1 error rate of 0.05 and a 95% CI: N sensitivity =307. N specificity =1230. For a type 1 error rate of 0.05 and a 90% CI: N sensitivity =218. N specificity =871. For a type 1 error rate of 0.05 and an 85% CI: N sensitivity =166. N specificity =664. For a type 1 error rate of 0.05 and an 80% CI: N sensitivity =131. N specificity =524. Therefore, we are powered to identify sensitive biomarkers (≥80%) of acute breathlessness with a maximum marginal error in the estimate for sensitivity not exceeding 5% with 95% confidence. Similarly, we are powered to identify specific biomarkers (≥80%) of acute breathlessness with a maximum marginal error in the estimate for specificity not exceeding 5% with 80% confidence.
For the primary analysis, the outcome will be treated as a nominal variable with levels: (1) acute heart failure, (2) community-acquired pneumonia; (3) adult exacerbations of asthma and COPD; and (4) acute exacerbations in school-age children treated in hospital for severe asthma attacks.
The relationship between the primary outcome and the exhaled breath VOC biomarkers will be modelled using multinomial logistic regression. In addition to metabolomics markers, the following independent variables will be included in the model: clinical uncertainty score on a 100 mm VAS scale, age and a validated comorbidity score (the Charlson comorbidity score). Receiver operator analyses will be used to generate reciver operating characteristic curves for individual and multiple panels VOC predictors in the primary analysis.
To understand the dynamic profile of breath biomarkers during (1) the acute state and (2) in the chronic state up to 6-month postexacerbation, a repeated measures model with a random intercept and random effect for time will be fitted; the random effects will be fitted for each patient. For the repeated measures mixed model, an unstructured covariance will be assumed. To evaluate the relationship between breath biomarkers and hospital readmission at 30 and 60 days, Cox proportional hazards and frailty models will be used. [bib_ref] Fisher information for two gamma frailty bivariate Weibull models, Bjarnason [/bib_ref] Analysis of multivariate survival data, competing risk models and joint models will be fitted. [bib_ref] Joint modeling of survival and longitudinal non-survival data: current methods and issues...., Gould [/bib_ref] Relationship between death and breath biomarkers will be evaluated using a logistic regression model. Changes in outcome measures will be measured appropriately for each variable (eg, paired t-test, Mann-Whitney and repeated measures analysis). Tables of descriptive statistics will be compiled for all key variables.
All analysis will be performed using R V.3.5.0 (https:// www. r-project. org/). discovery and replication studies Specific indicator conditions have been selected for targeted recruitment according to their high prevalence and unmet need, their high morbidity and mortality and the need to develop better diagnostic and prognostic algorithms in acute care pathways.
The indicator diagnoses of interest are: (1) exacerbations of adult asthma and COPD, (2) community-acquired pneumonia, (3) acute heart failure; and (4) exacerbation in school age children treated in hospital for severe asthma attacks.
Patient-level clinicopathological and outcome data (spanning the entire acute pathway) will be collected in parallel to breath sampling. In addition, breath samples will be acquired in the stable state post exacerbation .
Age-matched healthy volunteers will be recruited where possible at separate visits. For the purposes of this study, healthy volunteers will be defined as participants who have no prior history of asthma, COPD, heart failure and have not been admitted to hospital with community acquired pneumonia within 6 weeks of the baseline study visit. For acute admission, the study team will approach the spouse, parent or sibling of the index case and seek informed consent for study assessments. All healthy subjects will undergo two assessments separated by a duration of 8-16 weeks to match the acute and recovery time points elapsed in their index case/partner/spouse/ sibling/child. Additional healthy volunteers will be identified from local recruitment databases and via advertising
## Open access
Discovery phase (project months 1-24) The aim of the discovery phase is to discover putative discriminatory breath VOC, using both offline and online technologies.
Preplanned recruitment of acutely breathless patients will be enriched into the following disease strata following senior clinical decision maker assessment and within 24 hours of acute admission.
Acute adult heart failure (n=55), adult community-acquired pneumonia (n=55), adult exacerbations of asthma (n=55) and COPD (n=55) in addition to acute severe asthma attacks in school-age children (n=50).
Additional age-matched healthy volunteers (n=55 adults and 50 children) will be identified as a non-disease reference group [fig_ref] Table 1: Table summarising recruitment targets for both adult and paediatric groups [/fig_ref].
## Replication phase (years 3-4)
The aim of the replication phase is to replicate putative discriminatory breath VOC/VOC signatures identified in the discovery phase.
Similar to the discovery phase, recruitment of acutely breathless patients will be enriched into the following disease strata following senior clinical decision maker assessment and within 24 hours of acute admission.
Acute adult heart failure (n=55), adult community acquired pneumonia (n=55), adult exacerbations of asthma (n=55) and COPD (n=55) in addition to acute severe asthma attacks in school age children (n=25) (table 1).
Additional age-matched healthy volunteers (n=55 adults and 25 children) will be identified as a non-disease reference group.
schedule of assessments A schedule of acute assessments is outlined below and aligns to the movement of acute patients through the clinical care pathway and the overall aim of developing a complete phenotypic picture of acutely breathless patients.
Defining acute breathlessness At presentation (within 24 hours of admission) to one of three acute admissions units, potentially eligible patients will be identified following confirmation of acute breathlessness, identified as: (1) patient-defined acute breathlessness and/or (2) 1 unit increase above patient reported baseline in the extended MRC (eMRC) dyspnoea score and at least one of the indicator diagnoses identified as the primary clinical diagnosis by a senior clinical decision maker. eMRC will be completed by all patients and healthy volunteers at each research visit.
## Informed consent
Patients meeting the prespecified definition of acute breathlessness will be approached for informed consent to the breath VOC biomarker study. Only patients that are eligible to give full written informed consent will be recruited.
## Collection of blood-based pathology markers
Collection of the blood biomarkers CRP, BNP, troponin-I and blood eosinophil count will be performed both acutely and following recovery, when not taken as part of clinical care pathway. These are currently used in profiling acutely breathless patients in clinical practice [fig_ref] Table 2: Type of analyser and methodology used for blood biomarker calculation [/fig_ref].
## Breath voc sampling
Offline breath sampling using GC-MS and comprehensive GC×GC-MS coupled with a standardised and CE-marked breath sampler ReCIVA 48 will be performed. Gas chromatography is considered a gold standard technique in detecting VOCs and as such its sampling will be prioritised. Additionally, the following online technologies, proton transfer reaction mass spectroscopy (PTR-MS), gas chromatography ion mobility spectrometry (GC-IMS) and atmospheric pressure chemical ionisation-mass spectrometry (APCI-MS) will be evaluated according to the sampling strategy outlined in and table 3.
## Collection of additional samples for future biomarker campaigns
Collection of additional biomarkers for future biomarker discovery campaigns including: (1) a urine sample, (2) blood samples, up to 85 mL for DNA, RNA, plasma and serum and peripheral blood cell flow cytometry in selected subjects and (3) spontaneous or induced sputum samples (plugs and supernatants) will be carried out .
All samples will be collected at time points 1 and 2 (figure 3). The additional samples will be used for future omics analyses, these may include detailed analysis of the metagenome in sputum and proteomics applied to urine and serum samples.
## Physiological characterisation
Physiological measures of lung function will be performed in acutely ill participants and at recovery including: (1) handheld forced oscillation technique: an easily accessible measure of lung function. Patients favour this to spirometry as it is effort independent, unlike spirometry, and requires less than a minute of [bib_ref] Recommendations for chamber quantification: a report from the American Society of Echocardiography's..., Lang [/bib_ref] will be used to acquire two-dimensional, colour and Doppler images in conventional parasternal long-axis, short-axis and apical 4-chamber, 2-chamber and 3-chamber views. Left ventricular ejection fraction will be calculated using the biplane method of discs formula (Simpson's rule) to derive left ventricular volume indices. All participants are encouraged to report any testing-related discomfort or concerns to the research team to terminate the sampling process.
Recovery follow-up ► Patient recovery will be defined as:
i. Patient-reported recovery from the acute exacerbation spell and back to their baseline-extended MRC score or clinician-defined recovery from the acute exacerbation spell. ii. At least 6 weeks postexacerbation event (up to 6 months). Patients who readmit to hospital between visits 1 and 2 can have additional visit 1 assessments. Visit 2 will be taken as recovery following the subsequent admission. If a patient is admitted to hospital after visit 2, then they will be eligible to be recruited as a new study participant.
The schedule of assessments at the recovery visit is outlined in table 3.
## Clinical adjudication
In an effort to reduce data variability and minimise bias, an independent panel consisting of two senior acute clinicians (SS and NG) will review all pertinent clinical and diagnostic source documentation while blinded to admission blood biomarkers and clinical diagnosis.
All acutely breathless adult patients' notes will be sequentially divided for adjudication. The panel will independently determine the primary diagnosis of highest probability from a list of the four potential acute indicator diagnoses and mark their level of clinical certainty on a 100 mm VAS. The panel members will be able to review imaging, ECGs and other relevant information but not admission blood-based pathology tests.
In a subset of patients, adjudication will be validated by separate panel member to ensure between observer agreement using Bland-Altman analysis and inter-rater agreement of the primary diagnosis using Kohen's kappa via repeated evaluation of a subset of cases (see statistical methods outlined).
## Clinical informatics
Clinical data collection will be undertaken using a securely hosted bespoke database system (ADD) developed within the National Institute for Health Research (NIHR) Leicester Biomedical Research Centre -Respiratory (BRC). The system links acute admission episodes to hospital pathology records; historical respiratory physiology tests; and demographic information. The system provides functionality to validate data entry, manually verify records and highlight incomplete records. A custom VOC 'module' has been be created to support data collection within the study visits (1 and 2) and standardise diagnoses and medications through the use of clinical ontologies as well as linking hospital records/tests to patient visits.
Non-ADD-based clinical data (eg, hospital admissions, readmissions and mortality) will be extracted from the hospital data warehouse using identifiable patient Open access Summary of baseline and follow-up assessments. The table summarises key assessments carried out at different time points during the study. The participants may undertake any combination of the investigations listed at any of these time points identifiers and subsequently pseudonymised prior to integration.
An informatics pipeline will be created to facilitate the transfer of chemometric data from remote computers to the data repository. This will include tools to: (1) enforce the correct labelling of data sets (eg, study number, visit and type/source of sample) prior to automated validated transfer to the repository; (2) record information about the sample process; and (3) search and extract data sets from the repository for subsequent analysis. Prior to analysis, clinical and chemometric data will be integrated using the study number and any potentially identifiable information will be removed.
## Breath profiling
The technologies used in the VOC study during discovery and replication phases are: Offline technologies: ► ReCIVA+GC-MS. ► ReCIVA+GC ×x GC-MS. Online technologies:
[formula] ► GC-IMS. ► PTR-MS. ► APCI-MS. [/formula]
Offline technologies will underpin the discovery analyses owing to their ability to identify chemical identity and their recognition as the analytical gold standard in exhaled breath VOC analysis.In contrast, online technologies will be used for VOC biomarker replication and at the recovery visits owing to their portability and potential for future point of care testing [fig_ref] Figure 4: Multi-instrument use in breath sampling [/fig_ref]. A brief description of the core VOC platforms is provided below.
A CE-marked breath sampling device (ReCIVA) developed by Owlstone Medical will be used to sample breath onto two adsorbent Tenax tubes. Participants will be asked to breathe through the ReCIVA face mask for a maximum of 900 seconds, aiming for collection of ≥80% of the target sample volume of 1 L, after which the Tenax tubes will be transferred to the laboratory for analysis. This effectively allows decoupling of the breath sampling from the breath sensor and analysis platforms in selected patients that are not able to mobilise to a real-time breath sampling device. The Owlstone ReCIVA sampler will be used in breath collection for offline technologies namely GC-MS and GC×GC-MS. The ReCIVA sampler is capable of entraining oxygen and is therefore suitable for patients with mild respiratory failure requiring low flow rates of oxygen to maintain target oxygen saturations. [bib_ref] ABSTRACT: breathe free: open source development of a breath sampler by a..., Kitchen [/bib_ref] gas chromatography and mass spectroscopy GC-MS is a commonly applied methodology used to accurately measure trace gases in complex mixtures such as exhaled air.Preconcentrating breath volatiles by various means and subsequent analysis constitute a reliable and sensitive method for VOC analysis. [bib_ref] Human exhaled air analytics: biomarkers of diseases, Buszewski [/bib_ref] Despite its high sensitivity, it is, however, a time-consuming technique Open access and carries a risk of contamination at the preconcentration step. It is also not suitable for online and multiple measurements limiting its use as a point-of-care testing technology for VOC. [bib_ref] Applications of breath gas analysis in medicine, Amann [/bib_ref] The instrument used will be an Agilent 7890A gas chromatogram with a 5977a quadrupole mass spectrometer (Agilent Technologies Ltd, Stockport, UK), interfaced with a Markes Unity 2 thermal desorptionunit (Markes International Ltd, Llantrisant, UK).
Comprehensive two-dimensional gas chromatography-mass spectrometry (gC×gC-Ms) GC×GC-MS is an advanced analytical technique for the analysis of complex organic matrices; its main advantage is the unparalleled separation power it affords over conventional one-dimensional chromatographic techniques. [bib_ref] Comprehensive twodimensional gas chromatography in metabolomics, Almstetter [/bib_ref] Previous research, although sparse, has demonstrated the potential of GC×GC-MS for breath analysis with the number of VOC detected exceeding those detected by conventional GC-MS. GC×GC-MS of breath metabolites has been used for the identification of biomarkers related to glucose metabolism, 59 60 tuberculosis 61 and radiation response. [bib_ref] Breath biomarkers of whole-body gamma irradiation in the Göttingen minipig, Phillips [/bib_ref] This has generated interest within the breath research community; however, such studies were conducted on a small scale (<50 patients) and involved the use of expensive detectors and modulators. Method development and analysis of the data-rich GC×GC chromatograms, however, can be time-consuming and require specialist knowledge.
The instrument used will be an Agilent 7890A gas chromatogram, fitted with a G3486A CFT flow modulator and a three-way splitter plate coupled to a flame ionisation detector and a HES 5977B quadrupole mass spectrometer (Agilent Technologies Ltd, Stockport, UK), interfaced with a Markes TD-100xr thermal desorption autosampler (Markes International Ltd, Llantrisant, UK).
Proton Transfer Reaction Time-of-Flight Mass Spectromery (PTR-ToF-MS) is a real-time technique, capable of simultaneously measuring the evolution of multiple gas metabolites from a single breath. It has been used for the identification of potential useful VOC biomarkers for diagnosis of a variety of diseases including various cancers, 63-65 liver disease and respiratory disease. [bib_ref] Comparison of proton transfer reaction-mass spectrometry and gas chromatography-mass spectrometry in analysis..., Tomasz Ligor [/bib_ref] It has several advantages in clinical settings, such as the speed of sampling, the instant result achieved and the lack of need for sample storage or shipping. However, owing to the lack of preconcentration or chromatographic separation, sensitivity and definitive compound identification can be somewhat limited when compared with GC-MS.
Two breath sampling devices will be used. The first device is a Loccioni SOFIA GSI-S; the subject is required to exhale a single breath, five times (three if providing five samples proves too difficult) into a sterile mouthpiece connected to an electrostatic bacterial/viral filter while wearing a nose clip (all CE marked). Flow from the mouthpiece passes into a gas sampling interface capnograph (Loccioni GSI-S -CE marked), and real-time user feedback of flow is provided on screen, allowing the regulation of the breath sampling rate. The gas sampling interface acts to simultaneously trigger the acquisition of the Proton Transfer Reaction Time-of-Flight Mass Spectromery (PTR-ToF-MS) data and the exhaled breath travels through the capnograph down a heated sample line into the ion source of the PTR-ToF-MS.
The second breath sampling device is a ReCIVA breath sampler (Owlstone) with one of the adsorbent Tenax tubes replaced with an outlet tube adapted for online sampling. The exhaled breath is transferred to the PTR-ToF-MS via a heated transfer line connected to the outlet tube, continuously drawn at a constant flow rate by the PTR-ToF-MS. The online adaptation of the consumable adsorbent tube does not affect the CE mark of the ReCIVA sampling device.
Once the breath sample reaches the PTR-ToF-MS, via either breath sampler, the breath mixes with protonated water (H 3 O + ) inducing proton transfer to the target VOCs present, resulting in their ionisation. Sample ions are then guided into the time of flight mass spectrometer, and mass spectra, showing the abundance and mass of the VOCs present, collected throughout the exhalation. Following sampling, mouthpieces, filters and nose clips are disposed of, and all patient-contacted surfaces were wiped down with antiseptic cleaning wipes in preparation for the next patient.
The instrument used will be a Kore Series II high performance proton transfer reaction time of flight-mass spectrometer (Kore Technology Ltd, Cambridge, UK).
gas chromatography -Ion Mobility spectrometry (gC-IMs) (b&s Analytiks) GC-IMS allows the detection of VOCs down to ultratrace level (µg/L -to pg/L -range). For years, IMS has been used to discover potential discriminatory breath VOC in lung cancer, chronic obstructive pulmonary disease (COPD) and asthma. [bib_ref] A mobile instrumentation platform to distinguish airway disorders, Schivo [/bib_ref] Sampling takes place using a Spiroscout spirometer. The patients exhale through a disposable mouth piece connected to a Teflon tube. A piezoelectric pressure sensor is used to monitor the breathing profile; this opens the sampling valve at the appropriate point in the breath profile to collect end-tidal breath in a sample loop of 10 mL volume. After filling this loop, the collected sample air is then transferred to a multicapillary column for a chromatographic separation, which is achieved in 12 min. The separated molecules are then transferred into the IMS, ionised and then separated according to their mobility in a weak electric field.
The technology's multiple advantages of ultrasensitivity, portability, online sampling and short analysis time (typical analysis time of 10 min) with real-time detection brings a promise to provide immediate and potentially reliable results for point of care breath diagnostics. Another concept with IMS devices is that once the required breath signatures have been discovered using GC-MS, IMS offers the potential to be 'tuned' for selective detection of VOC.
The instrument used will be a BioScout a multicapillary column GC-IMS with a 63 Ni ion source, interfaced with Open access a SpiroScout breath sampler (BS Analytik, Dortmund, Germany).
## Apci-ms semiportable compact version (advion)
APCI-MS is one of less sensitive but more affordable versions of mass spectrometers released to the commercial market in recent years. The device uses APCI to produce ions. Although the most common use of APCI-MS systems is the detection in liquid chromatography applications, the technique has proven to be a valuable tool for direct measurement of VOC in air, food and breath. Recently, the technique has shown potential for online, real-time profiling of pseudometabolites in exhaled breath [bib_ref] Real-time monitoring of exhaled volatiles using atmospheric pressure chemical ionization on a..., Heaney [/bib_ref] with sensitivity comparable with other techniques. By combining miniaturised mass spectrometry technology with APCI techniques, adequate quality of on-site, real-time measurements with minimal or no sample preparation requirement can be provided. This is a desirable outcome as it overcomes main limitation of using standard breath analysis method in clinical setting, which is a need for breath sample collection followed by desorption and time-consuming laboratory analysis.
Preconcentrating breath gas by various means and subsequent analysis by means of GC-MS constitute a reliable and sensitive set of methods for VOCs analysis.
There remains an overall lack of standardisation and rigour across these technologies that hindered previous advancements in breath discovery; something we intend to minimise.
The instrument used will be an Advion Compact Mass Spectrometer Express, with atmospheric pressure chemical ionisation, interfaced with a heated breath sampling line (Advion, New York, USA).
ChEMoMEtrIC proCEssIng And dAtA AnAlysIs GC-MS breath data will be aligned, deconvoluted and the features for each participant will be extracted. The extracted features will be grouped and classified by retention index and mass spectrum. The registered and aligned data will be linked to participant metadata to generate a breath matrix. Data handling and analysis will be performed by a senior statistician.
The breath matrix is a × matrix where n is the number of subjects and p is the number of VOC. The breath matrix is high dimensional with ≫ and many potentially correlated VOC. In view of this, we will employ sparse partial least squares discriminant analysis [bib_ref] Sparse PLS discriminant analysis: biologically relevant feature selection and graphical displays for..., Cao [/bib_ref] to investigate which of the VOC can identify breathlessness. We will also investigate which of the VOC can discriminate between the different disease states including acute exacerbations of asthma and COPD and pneumonia. In addition to the supervised methods, unsupervised methods will be explored, specifically sparse principle component analysis. [bib_ref] Sparse principal component analysis, Zou [/bib_ref] Extracted VOC will also be investigated. Relationships between VOC and patient-reported acute breathlessness will be analysed using logistic regression model. VOC associated with patient reported acute breathlessness will be incorporated into multinomial logistic regression models in conjunction with CRP, BNP, blood eosinophils and troponin-I, pathology biomarkers currently in use for diagnosing undifferentiated breathlessness. In addition to the conventional binary and multinomial logistic regression models. [bib_ref] Regularization and variable selection via the elastic net, Zou [/bib_ref] EthICs And dIssEMInAtIon Publications will be prepared according to the MRC-EMBER consortium agreement and the University of Leicester publications policy. All intended publications will be submitted to the EMBER executive board for review and comments within 60 days of journal submission. Authorship will be according to contribution and internationally recognised guidance on journal authorship. patient and public involvement A series of consultations have taken place with our patient involvement team within the NIHR Biomedical Research Centre (Respiratory Theme) and across the wider BRC patient and public involvement group. Representations from the paediatrics team were also present. This group was sent copies of the participant documentation for review and discussion. Various revisions have been made following on from these discussions.
[fig] Figure 1: Relationship between lung proximity and degree of invasiveness of different lung matrices. The figure plots the level of invasiveness of various lung matrices in relation to their proximity to the lung. Given their pathological relevance, the degree of invasiveness of bronchoalveolar lavage (BAL) and lung biopsy makes them less favourable in diagnosing respiratory diseases. [/fig]
[fig] Figure 2: Multi-instrument use in breath sampling. Figure illustrates the various combinations of offline and online devices used in breath sampling and the relevant pros and cons. Offline and online technologies are used for the discovery and validation phases of the study, respectively. [/fig]
[fig] Figure 4: Multi-instrument use in breath sampling. Operational space of the analytical technologies used in EMBER for the analysis of volatile organic compounds in exhaled breath, including proton transfer reaction mass spectrometry (PTR-MS), atmospheric pressure chemical ionisation-mass spectrometry (CMS), gas chromatography ion mobility spectrometry (GC-IMS), gas chromatography mass spectrometry (GC-MS) and two-dimensional gas chromatography-mass spectrometry (GC×GC-MS). Comparing the typical molar mass range detectable; selectivity in detection owing to the type of ionisation involved and the proton affinity of analytes; and the inclusion of a chromatographic separation affecting total time of analysis. The online technologies involving chemical ionisation (PTR-MS, CMS and GC-IMS) can be used in-clinic owing to short analysis times but only detect lower molar mass molecules with a proton affinity higher than 697 KJ/mol. Offline chromatographic techniques (GC-MS and GC×GC-MS) detect a wider range of compounds independent of proton affinity; however, the techniques have longer analysis times and involve sample transportation and storage. EMBER, East Midlands Breathomics Pathology Node. [/fig]
[table] Table 1: Table summarising recruitment targets for both adult and paediatric groups. Total combined sample size of the discovery and replication phases = 700 participants. [/table]
[table] Table 2: Type of analyser and methodology used for blood biomarker calculation [/table]
|
10.3389/fcimb.2020.00031
|
CCBY
|
7025567
|
32117805
|
s2orc_pubmed_articles
|
Murine and Human Cathelicidins Contribute Differently to Hallmarks of Mastitis Induced by Pathogenic Prototheca bovis Algae
Prototheca bovis (formerly P. zopfii genotype-II) is an opportunistic, achlorophyllous alga that causes mastitis in cows and skin disease in cats and dogs, as well as cutaneous lesions in both immunocompetent and immunosuppressed humans. Antifungal medications are commonly ineffective. This study aimed to investigate innate immune responses contributed by cathelicidins to P. bovis in the mammary gland using a mastitis model in mice deficient in the sole murine cathelicidin (Camp). We determined P. bovis caused acute mastitis in mice and induced Camp gene transcription. Whereas, Camp −/− and Camp +/+ littermates had similar local algae burden, Camp +/+ mice produced more pro-inflammatory cytokines, TNF-α, and Cxcl-1. Likewise, Camp +/+ bone marrow-derived macrophages were more responsive to P. bovis, producing more TNF-α and Cxcl-1. Human cathelicidin (LL-37) exhibited a different effect against P. bovis; it had direct algicidal activity against P. bovis and lowered TNF-α, Cxcl-1, and IL-1β production in both cultured murine macrophages and mammary epithelial cells exposed to the pathogenic algae. In conclusion, cathelicidins were involved in protothecosis pathogenesis, with unique roles among the diverse peptide family. Whereas, endogenous cathelicidin (Camp) was key in mammary gland innate defense against P. bovis, human LL-37 had algicidal and immunomodulatory functions.
# Introduction
Prototheca species are unicellular achlorophyllous algae, 3-30 µm in diameter, that lack a specific glucosamine cell wall or chloroplasts. Their reproduction is asexual, with endospores being released from sporangia. Prototheca spp. are widely distributed in the environment, particularly in organic matter with a high moisture content. Prototheca spp. are also pathogenic and provoke a variety of maladies in animals. In dogs and cats, infection with Prototheca spp. causes either cutaneous lesions or systemic disease with hemorrhagic enteritis and progressive retinal degradation. In humans, Prototheca spp. causes rare chronic skin or articular infectionsor disseminated infections mostly in immunodeficient patients.
Among Prototheca species, P. bovis has been proposed as a major cause of chronic mastitis in cattle. Traditional taxonomic studies on Prototheca species identified two genotypes (GT): P. zopfii GT-I and P. bovis (formerly P. zopfii GT-II). P. bovis was isolated from the milk of cows with mastitis and identified as the causative agent of bovine mastitis. Prototheca zopfii GT-I is commonly isolated from the environment and occasionally causes granulomatous lesions in experimentally infected bovine uddersand protothecosis in humans. Clinical signs of protothecal mastitis in cattle can include fever, pain, edema, anorexia, and lethargy, although it is most commonly subclinical, causing decreased milk production. Bovine mastitis caused by P. bovis is refractory to most therapeutic agentsand has a very low rate of spontaneous remission. Virulence factors, including intracellular heat shock proteins (Hsp70) and proteins that augment peroxisome metabolism to resist harsh environments (e.g., macrophage phagolysosomes) are mostly present in pathogenic P. bovis. From an epidemiological perspective, cattle infected with Prototheca spp. are a common source of Prototheca spp. for humans, with immunocompromised farmers at highest risk. Since Prototheca spp. may survive chlorination by forming biofilms, and be returned to the environment via sewage effluent and household waste, cows with mastitis caused by P. bovis are a risk factor in epidemiology of protothecosis.
Little is known about innate mammary gland defenses during P. bovis infection. However, cultured mammary epithelial cells from cattle and mice undergo oxidative stress and apoptosis when exposed to P. bovis. Moreover, bovine mammary epithelial cells respond to P. bovis infection with upregulated expression of Toll Like Receptors (TLRs)−2 and−4, pro-inflammatory cytokines (TNF-α, IL-1β, IL-8) and β-defensin-5. To further understand innate mammary defenses in protothecosis, we focused on cathelicidins. Cathelicidins are cationic amphipathic peptides with an N-terminal domain containing a signal peptide, a well-conserved central cathelin domain, and a variable Cterminal domain. The C-terminal domain is cleaved to produce the peptide with antimicrobial and antifungal activity. In humans, a single cathelicidin gene (cathelicidin antimicrobial peptide, Camp) yields this C-terminal active peptide termed leucineleucine with 37 amino acid residues (LL-37). In mice, the functional homologous peptide is cathelicidin-related-antimicrobial-peptide (Cramp) encoded by the gene Camp. Cathelicidins are abundantly expressed in mammalian cells, mainly neutrophils and epithelial cells and thus, are present in various organs/tissues, including skin, eyes, mouth, lungs, intestine, and mammary gland. Cathelicidins have been involved in various host responses against bacterial and parasitic intracellular pathogens through activation of cytokines/chemokine secretion. In sheep and murine mammary glands and in human milk, cathelicidins also have antimicrobial and anti-inflammatory effects. Moreover, cathelicidins increased in bovine mammary epithelial cells exposed to a main mastitis pathogen, Staphylococcus aureus. Thus, we hypothesized that cathelicidins produced by the mammary gland contribute to host innate defense against P. bovis infection. Using genetically mutant mice that lack cathelicidin (Camp −/− ) and cultured murine macrophages and mammary epithelial cells, we demonstrated that endogenous cathelicidins regulate mammary gland inflammation and macrophage activity against P. bovis, whereas exogenous human cathelicidins (LL-37) downregulated epithelial and macrophage responses and had algicidal activities.
# Materials and methods
# Ethics statement
Animal experiments were conducted in accordance with Canadian Guidelines for Animal Welfare (CGAW) and the University of Calgary Animal Care Committee (Animal Protocol AC16-0061).
## Prototheca bovis
A Prototheca spp. was isolated from the milk of a cow with clinical mastitis at the College of Veterinary Medicine, China Agricultural University, Beijing, China. This Prototheca spp. was characterized as P. bovis by several methods. First, we characterized its cellular fatty acid pattern to confirm that P. bovis had, as expected, more eicosadienoic acid (C20: 2) compared to the reference P. zopfii GT-I. Second, we determined 18S rDNA sequences using genotype-specific PCR. For this, P. bovis was purified from a single colony of P. bovis isolated from Sabouraud dextrose agar (Sigma-Aldrich) and amplified in Sabouraud in dextrose broth (Sigma-Aldrich) (1 mL) (DP302, TIANamp DNA, Tiangen). DNA was quantified (NanoDrop ND-1000 Spectrophotometer, Thermo Fisher Scientific) and stored at −20 - C. The Prototheca (450 bp) fragment internal amplification control was detected using Proto18-4f (GACATGGCGAGGATTGACAGA) and Proto18-4r (AGCACACCCAATCGGTAGGA) sequences. The P. bovis specific amplicon (165 bp) was detected with primers Proto18-4f (GACATGGCGAGGATTGACAGA) and PZGT-II/r (GTC GGCGGGGCAAAAGC).
The P. bovis genotype was further confirmed by restriction fragment length polymorphism (RFLP) analysis targeting the cytb gene fragment (599-668 bp). For this, a PCR mix (25 µL) containing cytb-F1 (5 ′ GyGTwGAAC AyATTATGAGAG-3 ′ ) and cytb-R2 (5 ′ -wACCCATAArAArTA CCATTCwGG-3 ′ ) primers (10 µM each primer), DNA template (1 µL), and 2x EasyTaq PCR supermix (TransGen Biotech, AS111-11; 12.5 µL) was amplified under specific conditions (2 min at 95 - C, followed by 35 cycles of 30 s at 95 - C, 30 s at 50 - C, and 30 s at 72 - C, with final extension of 5 min at 72 - C). The PCR products depicted a 644 base pair (bp) product compatible with P. zopfii as visualized by agarose gel electrophoresis (1%, wt/vol) and stained with ethidium bromide. The amplified cytb gene products (644-bp) were digested by RsaI and TaiI digesting enzymes (FastDigest Enzymes, Thermo Fisher Scientific). The total mixture (30 µL) containing 10x restriction enzyme buffer (3 µL), PCR product (10 µL), enzymes (1.5 µL each) and PCR water (16.5 µL) was digested by RsaI (5 min at 37 - C) followed by TaiI (5 min at 65 - C). The restriction products visualized on 4% agarose gels, stained with ethidium bromide, and exposed to UV light showed DNA fragments of 200 and 450 bp after RSaI/TaiI digestion, compatible with P. bovis (Supplementary. Taken together, we confirmed a P. bovis genotype in the isolate clinically recovered from a case of mastitis in cows.
For experimentation, P. bovis was streaked on Sabouraud dextrose agar (S3181, Sigma-Aldrich) (37 - C for up to 48 h) to produce typical single creamy-white, yeast-like colonies and replicated in Sabouraud dextrose broth (S3306, Sigma-Aldrich) (37 - C for up to 72 h).
## Experimental induction of murine mastitis by prototheca bovis
C57BL/6 (6-8 wk. old) lactating female wild-type Camp +/+ and cathelicidin-null Camp −/− C57BL/6 mice (B6.129X1-Camp tm1Rlg/J ; The Jackson Laboratory) were housed in a specific pathogen-free environment with ad libitum access to feed and water (University of Calgary). In these Camp −/− mice, there is deletion of exons 3 and 4 of the cathelicidin gene Camp; whereas some gene portions could be detected, they do not produce functional cathelicidin peptides. Camp −/− and Camp +/+ mice were infected intramammary (10-14 days after parturition) with either P. bovis (50 µL containing 1 × 10 5 colony forming units (CFU)/mL) or an equal volume of phosphate buffered saline (PBS) (control) in the left 4th and right 4th mammary glands (L4 and R4) (n: 4 per group). Mice were euthanized by carbon dioxide followed by cervical dislocation at 4 days post-infection when they had indications of clinical mastitis. This termination point was chosen based on preliminary studies where mice become lethargic and before humane end points associated with longer intervals.
Immediately after euthanasia, all mammary glands were excised, weighed, and collected as follows: A portion was incubated into Trizol (Invitrogen) for gene quantification and another fixed in formalin (10%) solution, embedded in paraffin wax, sectioning (5 µm) and stained with hematoxylin and eosin (H&E) or toluidine blue for histological examination. Prototheca organisms were identified by specific staining, using periodic acid-Schiff (PAS) and Grocott methenamine silver (GMS).
## Identification of macrophages and neutrophils in murine mammary glands
Fixed mammary gland sections were deparaffinized and dehydrated, permeabilized with PBS/Triton X-100 (0.25%, v/v) (PBS-T) buffer with donkey serum (1%; 017-000-121, Jackson ImmunoResearch) [room temperature (RT), 10 min]. Slides were blocked with PBS-T containing donkey serum [10% (v/v)] and bovine serum albumin (BSA) (1% (v/v); CA97061-416, Sigma-Aldrich) (RT, 2 h). After washing with PBS, sections were incubated with primary antibodies against murine F4/80 (macrophages) (4316835, BD Pharmingen) and Ly-6G (neutrophils) antigens (127609, Biolegend) (1:1,000 in PBS-T plus 1% BSA) for 16 h at 4 - C. Following washing with PBS-T, slides were incubated with secondary antibodies (Alexa Fluor R 488-conjugated affinipure goat anti-rat IgG, 135205, Jackson Immune Research) (1:1,000 in PBS-T plus 1% BSA) (RT, 1 h), washed again with PBS-T and incubated with DAPI (4 ′ , 6-diamidino-2phenylindole) (Invitrogen) (RT, 20 min). Slides were examined with an immunofluorescence microscope (Axio Imager M2, Zeiss).
## Determination of prototheca bovis in infected mammary tissue
Prototheca infection in mammary tissues was assessed by histological counting of Prototheca sporangia and Prototheca culture. For counting, sporongia of Prototheca spp. (7-30 µM in diameter) were counted in 10 images of the mammary gland per mouse, individually and randomly captured at 40x magnifications. For culturing, the entire mammary gland was homogenized in sterile tubes containing 1 mL of 0.025% Triton X-100 in PBS. 10-fold homogenate dilutions were spread onto duplicate plates of Sabouraud dextrose agar (Sigma-Aldrich). Number of colonies, typically with a granular serrated shape, grayish white and with a central protrusion, was determined after 2-3 days of incubation (Gogoi-Tiwari et al., 2017).
## P. bovis infection in cultured murine bone marrow-derived macrophages and immortalized mammary epithelial cells and macrophages
Bone marrow-derived macrophages (BMDMs) were collected from femurs of Camp −/− and Camp +/+ mice and plated into RPMI containing 10% fetal bovine serum (Hyclone), macrophage colony stimulating factor (M-CSF, Peprotech) or granulocytemacrophage colony-stimulating factor (GM-CSF, Peprotech) (10 ng/mL) and antibiotic mixture (penicillin-streptomycin, 1%, P4333, Sigma-Aldrich). BMDMs were cultivated for 7-10 days in 12 well plates and challenged with P. bovis (1 × 10 5 CFU/mL re-suspended in RPMI) (37 - C with 5% CO 2 ) for up to 8 h. Challenge doses of P. bovis were chosen based on previous in vitro experiments with bovine mammary epithelial cells infected with P. bovisand a murine model of P. bovis mastitis .
Mammary epithelial and macrophage responses, relevant during mastitis, were modeled using the murine mammary epithelial cell line (HC11), a prolactin responsive clone COMMA-1D derived from mammary tissue of BALB/c mice in mis-pregnancy; and a murine phagocytic monocyte J774A.1 (ATCC R TIB-67 TM ) derived from BALB/c mice, conventionally used as a macrophage model. To assess the role of exogenous cathelicidin in protothecosis, murine macrophages and mammary epithelial cells were challenged with P. bovis (1 × 10 5 CFU/mL resuspended in DMEM/F12) ± synthetic human cathelicidin LL-37 amide (H-Leu-Leu-Gly-Asp-Phe-Phe-Arg-Lys-Ser-Lys-Glu-Lys-Ile-Gly-Lys-Glu-Phe-Lys-Arg-Ile-Val-Gln-Arg-Ile-Lys-Asp-Phe-Leu-Arg-Asn-Leu-Val-Pro-Arg-Thr-Glu-Ser-NH2 trifluoroacetate salt) (10 µg/mL; H-6224; Bachem) for up to 8 h (37 - C with 5% CO 2 ).
Transcriptional Expression of TNF-α, IL-1β and Cxcl-1, and Secretion of Cxcl-1 and TNF-α in Mammary Tissues and Cultured Cells
Gene mRNA transcription was quantified by quantitative realtime polymerase chain reaction (RT qPCR). Total RNA from murine mammary glands, BMDMs, mammary epithelial cells and macrophages was extracted with Trizol reagent (Invitrogen) and converted into cDNA by reverse transcription (101414-098, VWR). The RNA and cDNA quality was evaluated by the A260/A280 absorbance ratio (NanoVue Spectrophotometer, GE Healthcare Bio-Sciences). The pre-designed primers (RT 2 qPCR Primer Assay, Qiagen) were specific for murine tumor necrosis factor alpha (TNF-α) (PPM03113G), interleukin 1 beta (IL-1β) (PPM03109F), C-X-C motif chemokine ligand 1 (Cxcl-1) (PPM03058C), cathelicidin (Camp) (PPM25023A), defensin-4 (PPM29171A) and housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (forward 5 ′ AAATGGTGAAGGTCGGTGTG and reverse 3 ′ TGAAGGGGTCGTTGATGG). All primers were verified for specificity and efficiency (>95%) to ensure amplification of a single product of the correct size, as indicated in MIQE guidelines. The total reaction mixture (10 µL) included 2 µL of cDNA, 0.5 µM of forward and reverse primers and 1x SsoAdvanced Universal SYBR Green Supermix (BioRad) (CFX-96 real-time PCR system). All reactions were performed in triplicate. Target gene mRNA values were corrected relative to the normaliser, GAPDH. Data were analyzed using the 2 − CT method and reported as mean fold change of target transcript levels in challenged vs. uninfected control groups.
## Secreted cxcl-1 and tnf-α protein quantification
TNF-α and Cxcl-1 proteins produced and secreted by the mammary gland were quantified by respective ELISAs (DY453-05 and DY410-05, R&D Systems) in supernatants collected from the whole murine mammary gland homogenized with 1 mL of 0.025% Triton X-100 in PBS and centrifuged (20,000 g; 10 min).
## Direct algicidal activity of cathelicidin against p. bovis
For killing assays, P. bovis grown into Sabouraud in dextrose broth (Sigma-Aldrich) (37 - C, 2 days on a rotary shaker) was poured into 96 wells plates (1 × 10 5 CFU/mL in logarithmic growth phase) and simultaneously incubated with LL-37 peptides (H-6224; Bachem) (1 and 2 µM diluted in distilled water) (37 - C, up to 1 day). Final Prototheca spp. concentration was determined by CFU/mL and plotted after conversion to logarithm (log 10 ) using a standard curve. P. bovis and media only were used as controls.
## Statistical analyses
Normality was assessed using D'Agostino & Pearson omnibus normality or Shapiro-Wilk (Royston) tests. Analytical data represented as histograms were recorded as mean values with bars representing standard errors of the mean (SEM) from a minimum of two independent experiments, with data obtained in triplicate, unless otherwise stated. All statistical comparisons were performed using one-way analysis of variance (ANOVA) with a post hoc Bonferroni correction for multiple group comparisons (Graph Pad Prism, 5.0). A p-value was assigned to each experimental group with reference to a control group. A p-value of <0.05 was considered significant. Microscopically, uninfected control mice had normal mammary gland architecture whereas both Camp +/+ and Camp −/− infected mice developed severe acute mastitis. Mammary epithelial lobules, interstitial fibrovascular tissue and interlobular adipose tissue in both Camp +/+ and Camp −/− infected mice had infiltration of a mixed population of inflammatory cells, including lymphocytes, plasma cells, neutrophils, macrophagesand rare mast cells. More macrophages infiltrated the mammary in P. bovis infected Camp −/− mice (p < 0.05,. Neutrophils were numerous in P. bovis infected mice, but there was no difference between Camp −/− and Camp +/+ mice (p > 0.05,.
# Results
## P. bovis-challenged
Mammary infection with P. bovis was confirmed microscopically in Camp +/+ and Camp −/− mice 4 days post-infection. Using PAS and GMS staining, round to oval sporangia with internal divisions compatible with Prototheca spp. were visible both free within alveolar lumen and throughout the mammary gland interstitium. The grade of infection with P. bovis did not differ between Camp +/+ and Camp −/− mice at 4 days post-infection, as determined by histological counting of sporangia and Prototheca spp. culture.
The general process of mastitis is characterized by upregulated synthesis of certain pro-inflammatory cytokines, such as TNFα and IL-1β. P. bovis induced mRNA synthesis and protein secretion of TNF-α in mammary glands from both Camp +/+ and Camp −/− mice infected with P. bovis, with the highest concentration in Camp +/+ mice (p < 0.05,.
Cxcl-1 chemoattracts leukocytes to sites of infection; recruitment of neutrophils and monocytes/macrophages from blood to milk compartments is a critical defense mechanism in bovine mastitis . In our study, Cxcl-1 mRNA was upregulated in P. bovis infected mammary glands of Camp +/+ relative to Camp −/− mice (p < 0.05,. Secretion of Cxcl-1 increased in mammary glands from both Camp +/+ and Camp −/− infected mice relative to non-infected controls but, the highest level was in Camp −/− infected mammary glands (p < 0.05,. Transcriptional IL-1β gene levels were increased after P. bovis challenge, but not different between Camp +/+ and Camp −/− infected mammary glands (p > 0.05,. Transcriptomic gene expression of host defense peptides in mammary tissue revealed that β-defensin-4 was not increased after P. bovis infection in mammary glands of Camp −/− and Camp +/+ mice (p > 0.05,, whereas Camp mRNA was increased in infected mammary glands of Camp +/+ mice compared to uninfected Camp +/+ mice (p < 0.05,. No Camp gene expression was detected in Camp −/− mice (data not shown). Taken together, P. bovis provoked acute mastitis in experimentally challenged mice. Furthermore, endogenous production of cathelicidins, induced by the algae, was associated with increased synthesis of pro-inflammatory TNF-α and Cxcl-1.
## Bone marrow-derived macrophages deficient in cathelicidin (camp −/− ) had impaired pro-inflammatory cytokine response to p. bovis infection
Macrophages prevailed in mastitis induced by P. bovis in mice. Furthermore, when exposed to inflammatory stimuli (e.g., Prototheca spp), macrophages secrete a vast array of pro-inflammatory cytokines. To determine if cathelicidins modulate macrophage function in protothecal mastitis, BMDMs from Camp +/+ and Camp −/− mice were challenged with P. bovis. P. bovis induced early (2 h) TNF-α and IL-1β transcription in BMDMs, with higher levels in Camp +/+ BMDMs (p < 0.05). Such TNF-α and IL-1β overexpression lasted for at least 8 h post-infection. Prototheca bovis also induced early (2 h) increased Cxcl-1 mRNA expression in Camp +/+ BMDMs compared to Camp −/− BMDMs (p < 0.05,. Prototheca bovis did not induce β-defensin-4 mRNA (p > 0.05,whereas P. bovis increased Camp gene expression in Camp +/+ BMDMs with an expected non-expression in Camp −/− BMDMs (p < 0.05,. Thus, endogenous cathelicidin contributed to production of pro-inflammatory cytokines in macrophages exposed to P. bovis.
## Human cathelicidin ll-37 inhibited growth of p. bovis and decreased pro-inflammatory responses in infected macrophages and mammary epithelial cells
To mechanistically explore the role of cathelicidin in P. bovis mastitis, we studied whether exogenous cathelicidin exert algicidal or immunomodulatory effects in key cellular components in mastitis: macrophages and mammary epithelium. In regard to direct killing, synthetic human cathelicidin LL-37 (1 and 2 µM) decreased in vitro growth of P. bovis up to 8 h, with some inhibitory activity still present at 24 h (p < 0.05, . In reference to the immunomodulatory role of LL-37, P. bovis upregulated mRNA expression of TNF-α, Cxcl-1, and IL-1β in murine phagocytic J774.A1 cells (at 2 and 8 h post-infection), whereas co-stimulation with LL-37 decreased concentrations of TNF-α, Cxcl-1, and IL-1β mRNA expression (p < 0.05, . Next, the immunoregulatory role of LL-37 was assessed in murine mammary epithelial cells (HC11) capable of producing milk casein protein. Challenge of HC11 cells with P. bovis induced early (2 h) TNF-α, Cxcl-1 and IL-1β gene expression (p < 0.05, , whereas costimulation with LL-37 lowered these mRNA TNF-α, Cxcl-1, and IL-1β responses. Addition of LL-37 to the mammary epithelium stimulated transcriptional expression of IL-1β and TNF-α at a later point (8 h post-infection) (p < 0.05, . Thus, human cathelicidin LL-37 reduced synthesis of proinflammatory cytokines in murine macrophages and mammary epithelium, whereas it had some direct killing effects on P. bovis.
# Discussion
Although P. bovis is a major cause of insidious and chronic (>30 days) mastitis in cattle | Synthetic human cathelicidin LL-37 inhibited in vitro Prototheca bovis. Prototheca bovis (1 × 10 5 CFU/mL in logarithmic growth phase) was incubated with synthetic human cathelicidin LL-37 peptides (up to 2 µM for up 12 h at 37 - C) into Sabouraud dextrose broth and algae concentration were determined using a standard curve, with only P. bovis and expressed as log CFU/mL. Histogram of remaining P. bovis after peptide treatment. Data are shown as means ± SEM (n = 3 independent experiments done in triplicate). *p < 0.05, **p < 0.01 (one-way ANOVA post hoc Bonferroni correction) was considered significant.
2013;, the udder innate immune response to pathogenic Prototheca spp. remains poorly understood. In this study, P. bovis triggered severe acute mastitis in mice and naturally occurring cathelicidins aided in establishing local inflammation by promoting synthesis of pro-inflammatory cytokines. Two main lineages of Prototheca spp. have been relevant in public health: a dominant species typically associated with dairy cattle, namely P. ciferrii (formerly P. zopfii GT-I), P. blaschkeae, and P. bovis (formerly P. zopfii GT-II) and others human-associated Prototheca spp. (i.e., P. wickerhamii, P. cutis, P. miyajii). In this study, we used a Prototheca spp. identified as P. bovis following a taxonomic approach commonly accepted for Protothecaand a cytb-based genotyping used for unambiguous Prototheca spp. identificationbased on the protothecal phylogeny.
Although Camp −/− mice had increased infiltration of macrophages in infected mammary glands, they also produced less local expression and secretion of TNF-α and Cxcl-1. Moreover, BMDMs isolated from Camp −/− mice had lower in vitro synthesis of pro-inflammatory TNF-α, IL-1β, and Cxcl-1 in response to P. bovis. That Cxcl-1 secretion in infected Camp −/− mice ended up higher than in wild type mice denoted a lack of association between Cxcl-1 mRNA and protein kinetics. Such differences in cytokine mRNA/protein expressions are usually due to post-transcriptional mechanismsor perhaps, Cxcl-1 gene transcription is augmented only at early points by cathelicidins or the Cxcl-1 protein half-life is longer compared with mRNA. Alternatively, increased Cxcl-1 in infected Camp −/− mice may correspond to an overwhelming and generalized inflammation, higher than any cathelicidin effect in the wild type. Overall, the observed role of endogenous cathelicidin promoting Cxcl-1 in protothecal mastitis could be critical in attracting leukocytes early to sites of infection and retain macrophages into the udder during the innate immune response against pathogenic algae. In agreement, murine cathelicidin Cramp attracted in vitro T-cells, macrophages, neutrophils, eosinophils and mast cells via an FPRL-1 receptorwhereas LL-37 chemoattracted leukocytes via G-protein coupled receptors. Cathelicidins could also chemoattract neutrophils indirectly, by inducing chemokines. In this regard, cathelicidins induced CXCL8 transcription alone and in synergy with TNFα though Src family kinase pathways downstream of P2X7R in keratinocytesand gingival fibroblastsand in an EGFR-dependent manner in skinand airway epithelial cells. Whereas, the influx of leukocytes in the mammary gland during P. bovis infection could be functionally regulated by cathelicidin, other cathelicidin-independent chemoattractive effects cannot be disregarded. For instance, cell-surface glycoproteins involved in cell-cell interaction, such as CD44, which mediates specific adhesion of bovine blood polymorphonuclears to mammary epithelial cells, could initiate recruitment of inflammatory cells during mastitis.
Our study demonstrated that a cathelicidin can exert an immunosuppressive effect in another species; in particular, human LL-37 mitigated production of pro-inflammatory TNFα, IL-1β, and Cxcl-1 in murine macrophages and mammary epithelial cells challenged with P. bovis. This immunomodulatory role of human cathelicidin LL-37 was not surprising, as LL-37 reduced secretion of IL-1β, IL-6, IL-8, and TNF-α in human and murine neutrophils exposed to heat-inactivated Pseudomonas aeruginosa or Staphylococcus aureus. Likewise, human or murine macrophage-like cells had reduced expression of TNF-α and nitric oxide (NO) in the presence of LL-37 when stimulated with bacterial lipooligosaccharide or LPS. Mechanistically, LL-37 could directly bind LPS in macrophages, thereby suppressing IL-6, IL-1β, and TNF-α synthesis.
In terms of algicidal effects, there was no differences between Camp +/+ and Camp −/− infected mice in P. bovis burden in mammary glands at an early point of infection (4 days). In contrast, human cathelicidin LL-37 in vitro directly reduced number of P. bovis in a dose-dependent manner. These killing properties of cathelicidins, LL-37 in this case, are related with the cationic peptide capacity to bind and disrupt negatively charged microbial membranes, leading to cell death due to destabilization of plasma membranes and efflux of ATP and proteins. Cathelicidins can also cross membranes and disrupt intracellular processes, including RNA and DNA synthesis and protein degradation. Although the role of cathelicidins in fungal/algal infectious is not fully explored, there is increasing evidence of protective effects. Human LL-37 has anti-fungus and algae activity against Candida albicans, Malassezia furfur, FIGURE 6 | Synthetic human cathelicidin LL-37 mitigated production of pro-inflammatory cytokines in murine macrophages and mammary epithelia infected with Prototheca bovis. Murine phagocytic (J774.A1) macrophages (A-C) and mammary epithelial (HC-11) cells (D-F) were challenged with P. bovis (1 × 10 5 cfu/mL) ± synthetic human cathelicidin LL-37 (2 µM) (37 - C with 5% CO 2 ). Transcriptomic expression of (A,D) TNF-α, (B,E) Cxcl-1 and (C,F) IL-1β were determined at 2 and 8 h post-infection. mRNA synthesis was quantified using RT qPCR. Data are shown as means ± SEM (n = 3 independent experiments done in triplicate). *p < 0.05, **p < 0.01 (one-way ANOVA post hoc Bonferroni correction) was considered significant. Ns, not significant.
Trichophyton rubrum and Trichophyton. mentagrophytes. Cattle have a vast repertoire of cathelicidins, with at least 8 naturally occurring and direct killing effects on P. bovis observed in cathelicidins from cattle origin. Bovine myeloid antimicrobial peptide (BMAP)-28 displayed quick anti-Prototheca activity (<1 h), with extensive surface blebbing and release of intracellular material due to cell permeabilizationwhereas other bovine host defense peptides (bactericin 5 and lingual anticrobial peptide; LAP) also had anti Prototheca activity, albeit through non-lytic mechanisms. Conversely, a different susceptibility of P. bovis to cathelicidins of diverse origins agrees with studies in bacteria. Human LL-37 and murine Cramp had dissimilar minimum inhibitory concentrations (MICs) against Citrobacter rodentium, Pseudomonas aeruginosa, Streptococcus pyogenes, Staphylococcus aureus, Escherichia coli, Salmonella spp. and Helicobacter pylori. In addition, endogenous Cramp had a pro-inflammatory role in murine mastitis induced by P. bovis, whereas in contrast, LL-37 reduced synthesis of proinflammatory cytokines in macrophages and mammary epithelia. Although human LL-37 and murine Cramp are considered homologous based on broadly similar and interspecies functions (antimicrobial and immunomodulatory)and structural properties (α-helical and net charge of +6), they actually share <70% sequence identity in their active peptide. This may explain singular activities for each peptide and indeed, human LL-37 binded dsRNA and activated TLR3 signaling in early endosomes but murine Cramp did not. Likewise, LL-37 was more effective in reducing S. aureus proliferation in vitro when compared to Cramp. In our study, murine cathelicidin Cramp may not have killing properties against P. bovis or perhaps acts later on the pathogen burden, when immune defenses are developed and more Cramp is available. Alternatively, Cramp may have indirect antimicrobial effects, promoting neutrophil killing by increasing production of reactive oxygen species (ROS) and enhancing phagocytic activity. In all cases, cathelicidin LL 37 appeared to be a good model of cathelicidins and potential target molecule for developing therapeutics, although comparative studies with various cathelicidins and their derivatives may offer distinctive immunomodulatory advantages.
Mast cells in mammary glands infected with P. bovis were an unexpected finding. Mast cells are immune effectors that degranulate pro-inflammatory granules rich in histamine and heparin upon activation. LL-37 stimulated degranulation of mast cellsthrough MAP kinases p38 and ERK phosphorylation, increasing vascular permeability in skin. Thus, such release of histamine, proteoglycans, serotonin and serine proteases from degranulated mast cells, likely regulated by cathelicidin, may be important in pathogenesis of P. bovis and other mastitis pathogens (e.g., Staphyloccocus aureus, Pseudomonas) by augmenting inflammation (e.g., increasing permeability of capillaries to permit leukocyte extravasation).
In summary, pathogenesis of P. bovis, a globally ubiquitous and environmental (grass, water, trees) algaecapable of infecting various vertebrate hostsremains poorly understood. Therapeutic control for human and animal protothecosis is on demand. Whereas, a human cathelicidin (LL-37) could kill P. bovis, endogenous cathelicidin (Cramp) contributed to the inflammatory response during P. bovisinduced mastitis, likely recruiting early leukocytes, thereby aiding pathogen control. Whereas, we focused on early steps in protothecal mastitis, cathelicidins could have other effects through later stages of inflammation due to known interactions with many cell types, including endothelial cells, epithelial cells, mast cells and macrophages. Cathelicidins were pleitrophic, either preventingor augmenting inflammatory reactions, e.g., skin with rosacea. Therefore, this study represented a proof-of-concept regarding the role of innate immune responses contributed by cathelicidins in protothecal mastitis; therefore, we propose LL-37 has potential as a therapeutic compound for controlling pathogenic algae. More studies are needed to decipher the kinetics of myeloid and non-myeloid cells and cytokine milieu that regulate specific functions of cathelicidins in mastitis.
# Data availability statement
The datasets generated for this study are available on request to the corresponding author.
# Ethics statement
The animal study was reviewed and approved by Animal experiments were conducted in accordance with Canadian Guidelines for Animal Welfare (CGAW) and the University of Calgary Animal Care Committee.
# Author contributions
MS, PC, and JG performed the in vitro and in vivo experiments. CK conducted the histopathologic analysis. MS and JG performed the analysis of data and prepared the figures. MS, BH, HB, and EC conceived this research, designed the experiments, and wrote the manuscript. All authors reviewed the manuscript.
|
10.1038/s41588-022-01121-z
|
CCBY
|
9355874
|
35835912
|
s2orc_pubmed_articles
|
Genome-wide analyses of 200,453 individuals yield new insights into the causes and consequences of clonal hematopoiesis
T he pervasive effects of ageing and somatic mutation shape the landscape of human disease in later life 1 . A ubiquitous feature of ageing is the development of somatic mutation-driven clonal expansions in aged tissues 2,3 . In blood, somatic mutations that enhance cellular fitness of individual hematopoietic stem cells (HSCs) and their progeny give rise to the common age-related phenomenon of CH 4-7 . CH becomes increasingly prevalent with age 4-6 and is associated with an increased risk of hematological cancers4,5,8,9and some nonhematological conditions 5,10,11 . However, our understanding of the biological basis for these associations remains limited, as does our ability to explain how CH driver mutations promote clonal expansion of mutant HSCs 12 . In fact, whilst CH is defined by its association with somatic mutations, its development is influenced by nonmutation factors 13-16 and by the heritable genome 17,18 , in ways that remain poorly understood.Insights into the causes and consequences of CH are confounded by its intimate relationship with ageing. Moreover, even when robust associations are identified, their causality can be difficult to establish. Here, we perform a comprehensive investigation of the genetic and phenotypic associations of CH in 200,453 UK Biobank (UKB) participants, yielding a step change in our understanding of CH pathogenesis. Our study reveals multiple new germline loci associated with CH, including several that interact with specific CH subtypes; uncovers causal links between CH and diverse pathological states across organ systems; and provides evidence for causal associations between smoking and telomere length and CH risk, amongst a series of insights.ResultsOverall and gene-specific prevalence of CH by age and sex. To identify individuals with CH, we analyzed blood whole-exome sequencing (WES) data from 200,453 UKB participants of diverse ancestry 19 aged 38-72 yr (Extended DataFig. 1a-c). We called somatic mutations in 43 CH genes (SupplementaryTable 1)and filtered these against a predefined list of CH driver variants(Supplementary Tables 2 and 3). This identified 11,697 mutations (SupplementaryTable 4) in 10,924 individuals (UKB prevalence: 5.45%), displaying patterns in line with previous reports 4,5,17 (Fig. 1aand Extended DataFig. 1d-h). Interestingly, the age-related rise in CH prevalence differed between driver genes (Fig. 1band Extended DataFig. 2a-c), for example, DNMT3A prevalence rose earlier in life compared with SF3B1 and SRSF2, consistent with what we now know about the lifelong behavior of these CH subtypes 20 . Females and males were similarly affected overall (Extended DataFig. 2d); however, there were significant gene-level differences between sexes(Fig. 1c), reflecting the sex-specific differences in prevalence of these gene-level frequencies in myeloid malignancies 21 .Associations between CH and traits prevalent at baseline. To identify associations between CH and traits or diseases prevalent at the time of enrollment to the UKB, we performed logistic regression Clonal hematopoiesis (CH), the clonal expansion of a blood stem cell and its progeny driven by somatic driver mutations, affects over a third of people, yet remains poorly understood. Here we analyze genetic data from 200,453 UK Biobank participants to map the landscape of inherited predisposition to CH, increasing the number of germline associations with CH in European-ancestry populations from 4 to 14. Genes at new loci implicate DNA damage repair (PARP1, ATM, CHEK2), hematopoietic stem cell migration/homing (CD164) and myeloid oncogenesis (SETBP1). Several associations were CH-subtype-specific including variants at TCL1A and CD164 that had opposite associations with DNMT3A-versus TET2-mutant CH, the two most common CH subtypes, proposing key roles for these two loci in CH development. Mendelian randomization analyses showed that smoking and longer leukocyte telomere length are causal risk factors for CH and that genetic predisposition to CH increases risks of myeloproliferative neoplasia, nonhematological malignancies, atrial fibrillation and blood epigenetic ageing.
analyses with CH as the outcome in the cohort of 200,453 individuals. We found that age increased the risk of CH by 6.7% per year and that prevalent hypertension, but not obesity or type 2 diabetes (T2D), was associated with CH status and . We also found that individuals with CH were more likely to be current or former smokers, an association that held true for different forms of CH and was strongest for ASXL1-mutant CH and Supplementary . Analyses of complete blood count and biochemical parameters identified both known and previously unreported associations with overall CH and CH subtypes and [fig_ref] Figure 6 |: IVW MR forest plots with CH traits as outcomes [/fig_ref]. We also found that CH status was associated with lower prevalent levels of total and low-density lipoprotein cholesterol, most marked for JAK2 and splicing factor-mutant CH and .
Associations between CH and incident disease. We next performed a phenome-wide association study (PheWAS) of incident disease in the UKB considering CH at baseline as the exposure. This identified strong associations with myeloid malignancies and associated sequelae (Extended Data [fig_ref] Figure 3 |: Cell-type-specific enrichment of the CH polygenic signal [/fig_ref] and Supplementary . Analyses for selected phenotypes (Supplementary also identified a high incidence of myeloid malignancies with all forms of CH and and increased risks of other hematological and nonhematological neoplasia, including lymphoma, lung and kidney cancers and . Notably, associations with lung and other cancers were also observed in self-reported never smokers (Extended Data [fig_ref] Figure 3 |: Cell-type-specific enrichment of the CH polygenic signal [/fig_ref] and . Unlike previous reports linking CH with ischemic cardiovascular disease (CVD) 5,10,22 , we did not find a significant association between CH and ischemic CVD, including coronary artery disease (CAD) and stroke; but we did find an association with heart failure and atrial fibrillation, and a composite of all CVD conditions in CH with large clones in multivariable regression models and . While CH was associated with CAD and ischemic stroke in unadjusted analyses, adjusting for age led to these associations attenuating to the null, demonstrating the impact of age as a confounder (Extended Data [fig_ref] Figure 3 |: Cell-type-specific enrichment of the CH polygenic signal [/fig_ref] and . Finally, we also found that CH increased the risk of death from diverse causes and Supplementary .
Heritability of CH and cell-type-specific enrichment. To identify heritable determinants of CH risk, we performed a genome-wide association study (GWAS) on the 184,121 individuals with genetically inferred European ancestry to identify common (minor allele frequency (MAF) > 1%) germline genetic variants predisposing to CH. In the GWAS, we compared 10,203 individuals with CH with 173,918 individuals without CH, after quality control (QC) of the germline genotype data. Linkage disequilibrium score regression (LDSC) [bib_ref] LD score regression distinguishes confounding from polygenicity in genome-wide association studies, Bulik-Sullivan [/bib_ref] showed little evidence of inflation in test statistics due to population structure (intercept = 1.009; lambda genomic control factor = 0.999). The narrow-sense (additive) heritability of CH was estimated at 3.57% (s.e. = 0.85%). We partitioned the heritability across four major histone marks observed in 10 cell-type groups aggregated from 220 cell-type-specific annotations [bib_ref] Partitioning heritability by functional annotation using genome-wide association summary statistics, Finucane [/bib_ref] and identified strong enrichment of the polygenic CH signal in histone marks enriched in hematopoietic cells (P = 5.9 × 10 −5 ; [fig_ref] Figure 3 |: Cell-type-specific enrichment of the CH polygenic signal [/fig_ref] and . Next, we partitioned the heritability of CH across open chromatin state regions in various hematopoietic progenitor cells and lineages [bib_ref] Partitioning heritability by functional annotation using genome-wide association summary statistics, Finucane [/bib_ref] [bib_ref] Lineage-specific and single-cell chromatin accessibility charts human hematopoiesis and leukemia evolution, Corces [/bib_ref]. Previous work on other traits [bib_ref] Lineage-specific and single-cell chromatin accessibility charts human hematopoiesis and leukemia evolution, Corces [/bib_ref] [bib_ref] Heritability enrichment of specifically expressed genes identifies disease-relevant tissues and cell types, Finucane [/bib_ref] has established that trait heritability tends to be enriched in transcriptionally active open chromatin regions in trait-relevant cell types, helping implicate specific cell types as key mediators of the GWAS signal. Consistent with this, we found CH heritability enrichment in accessible chromatin regions in HSCs, common lymphoid and myeloid progenitors, multipotent and erythroid progenitors, and B cells [fig_ref] Figure 3 |: Cell-type-specific enrichment of the CH polygenic signal [/fig_ref] and . Overall, these findings endorse the intuitive assumption that CH associations exert their greatest biological effect on HSC/progenitor populations.
Germline genetic loci associated with overall CH risk. Linkage disequilibrium (LD)-based clumping of 10,013,700 common autosomal and X chromosomal variants identified seven independent (r 2 < 0.05) genome-wide significant loci (lead variant . We replicated the only previously published risk locus associated with DNMT3A-CH in European-ancestry populations at 14q32.13-TCL1A. The overall CH loci at 5p15.33-TERT (signals with lead variants rs2853677, rs13156167 and rs7705526), 3q25.33-SMC4, 6q21-CD164 and 11q22.3-ATM were also genome-wide significant for DNMT3A-CH. We also found two new loci for DNMT3A-CH marked by lead variants rs138994074 at 1q42.12-PARP1 and rs8088824 at 18q12.3-SETBP1 [fig_ref] Figure 4 |: Manhattan plots displaying genome-wide associations between common germline genetic variants and each... [/fig_ref] and Supplementary . The three TET2-CH-associated loci included the lead variant rs2736100 at 5p15.33-TERT, which was moderately correlated (r 2 = 0.44) with the overall CH lead variant rs2853677 in the same region. The other two risk loci, both new for TET2-CH, were at lead variants rs10131341 (14q32.13-TCL1A) and rs79633204 (7q32.2-TMEM209; [fig_ref] Figure 4 |: Manhattan plots displaying genome-wide associations between common germline genetic variants and each... [/fig_ref] and . Notably, the A allele of rs10131341 had opposite associations with TET2-CH (odds ratio (OR) = 1.28, P = 6.8 × 10 −10 ) versus DNMT3A-CH (OR = 0.87, P = 6.4 × 10 −8 ).
[formula] R D W P D W P C T P L T W B C H L R R E T N E M O L Y R B C N R B C E O H T H G B M C H M C V C Y S P H O S G G T A L T A S T A P O A H D L L D L D C H O L H B A 1 C SRSF2 +SF3B1JAK2 [/formula]
A trend for opposite effects at 14q32.13-TCL1A was also observed in a previous study [bib_ref] Inherited causes of clonal haematopoiesis in 97,691 whole genomes, Bick [/bib_ref] , but did not achieve genome-wide significance for TET2-CH. When comparing 4,049 individuals with large or 6,154 individuals with small clones against the 173,918 controls without CH, we found that the overall CH loci at 5p15.33-TERT and 3q25.33-SMC4 were associated at genome-wide significance with large clone CH [fig_ref] Figure 4 |: Manhattan plots displaying genome-wide associations between common germline genetic variants and each... [/fig_ref] and Supplementary Table 20), while 5p15.33-TERT and 6q21-CD164 were associated with small clone CH. For small clone CH risk, we also identified a previously unreported locus marked by rs72755524 at 5p13.3 in a region with several long non-coding RNAs (lncRNAs) [fig_ref] Figure 4 |: Manhattan plots displaying genome-wide associations between common germline genetic variants and each... [/fig_ref] [bib_ref] Germ line variants predispose to both JAK2 V617F clonal hematopoiesis and myeloproliferative..., Hinds [/bib_ref] [bib_ref] Advancing human genetics research and drug discovery through exome sequencing of the..., Szustakowski [/bib_ref] [bib_ref] The longitudinal dynamics and natural history of clonal haematopoiesis, Fabre [/bib_ref] [bib_ref] Male predominance in AML is associated with specific preleukemic mutations, De-Morgan [/bib_ref] revealed that in addition to 14q32.13-TCL1A, the lead alleles at 6q21-CD164 also had opposite effects on DNMT3A-versus TET2-CH. The lead variants at 6q21-CD164 and 5p13.3-LINC02064 were associated with small, but not large, clones while the association at 7q32.2-TMEM209 was highly specific to TET2-CH. The lead variants at 1q42.12-PARP1 and 3q25.33-SMC4 had greater effects on large than small clone CH. At the whole-genome level, we estimated the genetic correlation (r g ) between DNMT3A-CH and TET2-CH as −0.48 (s.e. = 0.33, P = 0.15) and large and small clone CH as 0.37 (s.e. = 0.18, P = 0.018) using high-definition likelihood inference [bib_ref] High-definition likelihood inference of genetic correlations across human complex traits, Ning [/bib_ref]. Finally, we also performed a focused scan to explore rare variant (MAF: 0.2-1%) associations with the three CH traits with largest case numbers (overall, DNMT3A and small clone CH; each compared with 173,918 controls). This identified one new locus at 22q12.1-CHEK2 where the T allele (frequency = 0.3%) of lead variant rs62237617 was perfectly correlated (r 2 = 1) with the 1100delC CHEK2 protein-truncating allele (rs555607708) and conferred a large increase in risk of DNMT3A mutation-associated CH (OR = 4.1, 95% confidence interval (95% CI): 2.7-6.1, P = 6.3 × 10 −12 ).
Replication of genome-wide significant associations. Replication was undertaken using independent somatic mutation calling and germline association analysis pipelines on data from 221,285 European-ancestry individuals in the UKB, for whom WES was performed after our UKB discovery set. We focused on DNMT3A and/or TET2 mutation carriers (n = 9,386) in the replication sample, stratified by these two genes and clone size, and evaluated the 20 unique lead variants identified in the discovery GWAS (representing 26 distinct overall/subtype-specific CH associations). Eighteen of 20 variants were replicated at P < 0.05, with 16 replicating at P < 0.0025 (accounting for testing 20 variants), and 19 showing consistent directionality (Supplementary . Variants rs13130545 (overall CH; 4q35.1-ENPP6) and rs72755524 (small clone CH; 5p13.3-LINC02064) were not associated at P < 0.05 in replication analysis. Notably, we confirmed our observation that lead alleles at TCL1A and CD164 had opposite effects on DNMT3Aand TET2-CH, and replicated the CHEK2 association.
## Blood chromosomal mosaicism and ch due to gene mutation.
It is not known whether the germline genetic architecture underlying predisposition to CH due to individual gene mutations is similar to that underlying CH due to mosaic chromosomal alterations (mCAs). We used data from a recent blood mCA GWAS [bib_ref] Hematopoietic mosaic chromosomal alterations increase the risk for diverse types of infection, Zekavat [/bib_ref] to answer this and found that 13 of 19 unique lead variants identified for the five gene-mutant CH traits were associated with hematological mCA risk (P < 10 −4 ; . Notably, for our lead variants rs2296312 (14q32.13-TCL1A) and rs8088824 (18q12.3-SETBP1), the alleles conferring increased DNMT3A-CH risk reduced hematological mCA risk . We found a correlation between overall CH and mCAs (r g = 0.44, s.e. = 0.21, P = 0.037) using LDSC [bib_ref] LD score regression distinguishes confounding from polygenicity in genome-wide association studies, Bulik-Sullivan [/bib_ref]. This germline genetic correlation together with enrichment of the CH GWAS signal in common lymphoid and myeloid progenitors [fig_ref] Figure 3 |: Cell-type-specific enrichment of the CH polygenic signal [/fig_ref] supports the recent finding that gene-mutant CH and mCAs have overlapping biology that leads them to confer risk of both lymphoid and myeloid malignancies [bib_ref] Distinction of lymphoid and myeloid clonal hematopoiesis, Niroula [/bib_ref]. Further, a phenome-wide scan [bib_ref] PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations, Kamat [/bib_ref] [bib_ref] PhenoScanner: a database of human genotype-phenotype associations, Staley [/bib_ref] showed that several newly identified lead variants in our analyses were associated with multiple blood cell counts/traits .
## Gene-level associations and network analyses.
We used two complementary methods to perform gene-level association tests for each of our five CH traits: multi-marker analysis of genomic annotation (MAGMA) and a transcriptome-wide association study using blood-based cis gene expression quantitative trait locus (eQTL) data on 31,684 individuals 35 and summary-based Mendelian randomization (SMR) coupled with the heterogeneity in dependent instruments (HEIDI) colocalization test [bib_ref] Integration of summary data from GWAS and eQTL studies predicts complex trait..., Zhu [/bib_ref]. Both approaches converged on a new locus at 6p21.1, associated at gene-level genome-wide significance (P MAGMA < 2.6 × 10 −6 , P SMR < 3.2 × 10 −6 ) with DNMT3A-CH and marked by CRIP3 (P MAGMA = 3.4 × 10 −7 , P SMR = 6.6 × 10 −7 ; [fig_ref] Figure 5 |: Gene-level association and PPI network analyses [/fig_ref] and . While CRIP3 was the only 6p21.1 gene to reach gene-level genome-wide significance in both MAGMA and SMR, we did find subthreshold evidence for association between SRF or ZNF318 in the same region and DNMT3A-CH [fig_ref] Figure 5 |: Gene-level association and PPI network analyses [/fig_ref]. Notably, SRF encodes the serum response factor known to regulate HSC adhesion [bib_ref] The transcription factor Srf regulates hematopoietic stem cell adhesion, Ragu [/bib_ref] while ZNF318 is an occasional CH somatic driver [bib_ref] Clonal hematopoiesis, with and without candidate driver mutations, is common in the..., Zink [/bib_ref]. More globally, protein-protein interaction (PPI) network analysis 39 , using proteins encoded by the 57 genes with P MAGMA < 0.001 in the overall CH analysis (Supplementary as 'seeds' , identified the largest subnetwork [fig_ref] Figure 5 |: Gene-level association and PPI network analyses [/fig_ref] as encompassing 13 of 57 proteins with major hub nodes highlighted as TERT, PARP1, ATM and SMC4. This was consistent with the emerging theme that potential trait-associated genes at subthreshold GWAS loci are often part of interconnected biological networks [bib_ref] Network propagation: a universal amplifier of genetic associations, Cowen [/bib_ref]. The subthreshold genes identified by MAGMA that encoded protein hubs in this network included FANCF (DNA repair pathway) and PTCH1 (hedgehog signaling; [fig_ref] Figure 5 |: Gene-level association and PPI network analyses [/fig_ref] , both implicated in acute myeloid leukemia pathogenesis [bib_ref] Bi-allelic silencing of the Fanconi anaemia gene FANCF in acute myeloid leukaemia, Tischkowitz [/bib_ref] [bib_ref] Integration of Hedgehog and mutant FLT3 signaling in myeloid leukemia, Lim [/bib_ref] , and GNAS, a CH somatic driver [bib_ref] The GNASR201C mutation associated with clonal hematopoiesis supports transplantable hematopoietic stem cell..., Ostrander [/bib_ref]. The CH subnetwork was significantly enriched for several pathways including DNA repair, cell cycle regulation, telomere maintenance and platelet homeostasis .
## Functional target gene prioritization at ch risk loci.
To prioritize putative functional target genes at P lead-variant < 5 × 10 −8 loci identified by our GWAS of five CH traits, we combined gene-level genome-wide significant results from MAGMA and SMR with five other lines of evidence: PPI network hub status (Supplementary ; variant-to-gene searches of Open Targets 45 for lead variants; and overlap between fine-mapped variants [bib_ref] PICS2: next-generation fine mapping via probabilistic identification of causal SNPs, Taylor [/bib_ref] [bib_ref] Genetic and epigenetic fine mapping of causal autoimmune disease variants, Farh [/bib_ref]. The genes nominated by the largest number of approaches, representing the most likely targets, were SMC4, ENPP6, TERT, CD164, ATM, PARP1, TCL1A, SETBP1 and TMEM209 .
Among the newly identified loci, lead variant rs138994074 at 1q42.12 was strongly correlated (r 2 = 0.93) with rs1136410, a missense germline mutation in PARP1 wherein the G allele, which is protective for DNMT3A-CH, leads to a missense variant (p.Val762Ala) in the catalytic domain of its protein product associated with reduced Poly(ADP-ribose) polymerase 1 activity 53 . While SETBP1 was the only gene nominated at 18q12.3 (by only one approach, Open Targets 45 ), its nomination is strengthened by the fact that somatic SETBP1 mutations are recognized drivers of myeloid malignancies [bib_ref] Somatic SETBP1 mutations in myeloid malignancies, Makishima [/bib_ref] [bib_ref] Recurrent SETBP1 mutations in atypical chronic myeloid leukemia, Piazza [/bib_ref]. We also evaluated the 'druggability' of the prioritized genes in the context of known therapeutics (yielding support for TERT and PARP1) and ongoing drug development (yielding limited support for SMC4, ATM, Known (previously published) and new loci are indicated by cytoband and target gene (based on the prioritization exercise described in the text). Since there were multiple independent loci at 5p15.33 (LD r 2 < 0.05), we also label the 5p15.33 signals using the lead variant rs number for each signal. Our prioritization exercise was focused on protein coding genes near each lead variant and since there were no protein coding genes within 1 Mb of the lead variant at 5p13.3, we labeled this association using the nearest noncoding RNA. The CH traits corresponding to each Manhattan plot are: a, Overall CH. b, CH with mutant DNTM3A. c, CH with mutant TET2. d, CH with large clones. e, CH with small clones. We used independent (r 2 < 0.001) variants associated with overall, DNMT3A, TET2, and large and small clone CH at P < 10 −5 as genetic instruments for each of these traits and assessed their associations with outcomes . Since more variants were available at P < 5 × 10 −8 for overall and DNMT3A-CH, we also examined the consistency of associations when using genome-wide (GWS; P < 5 × 10 −8 ) and sub-genome-wide significant (sub-GWS; P < 10 −5 ) instruments for these two traits. Using the sub-GWS instrument, genetic liability to overall CH had the largest associations with myeloproliferative neoplasms (MPN) risk 48 (OR = 1.99, 95% CI: 1.23-3.23, P = 5.4 × 10 −3 ), intrinsic epigenetic age acceleration 64 (which represents a core characteristic of HSCs 67 ; beta = 0.39, 95% CI: 0.08-0.69, P = 0.01) and the blood-based Hannum epigenetic clock 64 (beta = 0.27, 95% CI: 0.04-0.49, P = 0.02) and even larger associations were observed when using the GWS instrument. Genetic liability to CH conferred increased risks of lung 68 , prostate [bib_ref] Association analyses of more than 140,000 men identify 63 new prostate cancer..., Schumacher [/bib_ref] , ovarian [bib_ref] Identification of 12 new susceptibility loci for different histotypes of epithelial ovarian..., Phelan [/bib_ref] , oral cavity/pharyngeal 71 and endometrial cancers 72 and . MR analyses did not support causal risk-conferring associations between genetic liability to CH and CAD 73 , ischemic stroke 74 and heart failure 75 , with similar lack of evidence across gene-specific and clone size-specific CH, and GWS instrument analyses and . However, we did uncover an association between genetic liability to overall CH or DNMT3A-CH and atrial fibrillation 76 risk (OR = 1.09, 95% CI: 1.04-1.15, P = 4.9 × 10 −4 for overall CH with the GWS instrument; . Among cytokines/growth factors [bib_ref] Genome-wide association study identifies 27 loci influencing concentrations of circulating cytokines and..., Ahola-Olli [/bib_ref] , genetic liability to overall CH was associated with elevated circulating stem cell growth factor beta (beta = 0.19; 95% CI: 0.07-0.30, P = 1.1 × 10 −3 ). MR analyses also revealed bidirectional associations between CH phenotypes and several blood cell counts/traits 29 , suggesting a shared heritability [fig_ref] Figure 6 |: IVW MR forest plots with CH traits as outcomes [/fig_ref]
[formula] S R F C U L 9 D N P H 1 C R I P 3 Z N F 3 1 8 A B C C 1 0 [/formula]
Genes on chromosome 6 within 250 kb of CRIP3 The HEIDI test is a test of heterogeneity of Wald ratio estimates. b, Largest subnetwork of genes/proteins associated with overall CH risk identified by the NetworkAnalyst tool. NetworkAnalyst uses a 'Walktrap' random walks search algorithm to identify the largest first-order interaction network. All genes (n = 57) with P MAGMA < 0.001 in the overall CH MAGMA analysis were mapped to proteins and used as 'seeds' for network construction which was done by integrating high-confidence PPIs from the STRING database. The largest subnetwork constructed contained 13 of the 57 seed proteins and included 210 nodes and 231 edges. The colored nodes indicate seed proteins that interact with at least two other proteins in this subnetwork with the intensity of redness increasing with number of interacting proteins. Seed proteins that interact with six or more other proteins in the subnetwork are named above their corresponding node.
[formula] D L K 2 T J A P 1 P O L R 1 C Y I P F 3 X P O [/formula]
Tables 37 and 39). We found little evidence to support an association between genetic liability to CH and LTL . Finally, we also performed an MR-PheWAS evaluating associations between genetic liability to overall or DNMT3A-CH and 1,434 disease/trait outcomes in the UKB. Reassuringly, the strongest associations involved blood cell counts/traits and hematopoietic cancers, but we also uncovered new associations such as with malignant skin cancers . Results of MR sensitivity analyses using the weighted median 77 and MR-Egger 78 methods are provided in . (1) standard deviation unit for continuous exposures (alcohol use in drinks per week, BMI, waist-to-hip ratio adjusted for BMI (WHRadjBMI) (a); LTL, two epigenetic aging traits, and red cell, white cell and platelet counts (b); and five circulating lipid traits (c)) and (2) log-odds unit for binary exposures (smoking initiation (ever having smoked regularly) and genetic liability to T2D (a)). IVW regression was used for all MR analyses, and results were not adjusted for multiple comparisons. Details of units are provided in Supplementary
# Discussion
We present an observational and genetic epidemiological analysis of CH in 200,453 individuals in the UKB and report a series of insights into the causes and consequences of this common aging-associated phenomenon. We increase the number of germline associations with CH in European-ancestry populations from 4 (ref. [bib_ref] Inherited causes of clonal haematopoiesis in 97,691 whole genomes, Bick [/bib_ref] to 14, reveal heterogeneity of associations by CH driver gene and clone size, and implicate putative new CH susceptibility genes, including CD164, ATM and SETBP1, through functional annotation. We also demonstrate that the CH GWAS signal is enriched at epigenetic marks specific to the hematopoietic system, particularly in open chromatin regions of hematopoietic stem/progenitor cells. The robustness of our GWAS is supported by replication of the vast majority of associations in an additional set of 221,285 individuals from the UKB and further affirmed by our replication of previous European-ancestry-specific CH associations [bib_ref] Inherited causes of clonal haematopoiesis in 97,691 whole genomes, Bick [/bib_ref] , the consistency of our estimates of CH heritability MR with overall CH as the exposure MR with DNMT3A CH as the exposure
## Fig. 7 | ivw mr forest plots with ch traits as exposures.
Forest plots with OR markers (for cancers and cardiovascular/metabolic traits) or exponentiated beta coefficient (exp(beta)) markers (for blood cell traits, lipids, adiposity measures and epigenetic aging indices). ORs/exp(betas) are represented as per log-odds unit increase in genetic liability to overall CH (a) or DNMT3A-CH (b). OR/exp(beta) markers with corresponding P < 0.05 are represented by filled circles. IVW regression was used for all MR analyses, and results were not adjusted for multiple comparisons. Symbols represent OR markers and error bars represent 95% CIs. Red symbols and error bars represent results using genetic instruments comprised exclusively of genome-wide significant (P < 5 × 10 −8 ) variants. Black symbols and error bars represent results when using genome-wide significant and sub-GWS (P < 10 −5 ) variants in the genetic instrument. Large effect size estimates (ORs/exp(betas)) are shown in the lower panels. Sample sizes for all genome-wide association datasets used are provided in . Full results, including from sensitivity analyses, are presented in . IS, ischemic stroke.
with previous reports [bib_ref] Inherited causes of clonal haematopoiesis in 97,691 whole genomes, Bick [/bib_ref] and the fact that many of our lead variants are associated with related traits 29,31,60,80 . At 14q32.13-TCL1A, we replicate the reported association with DNMT3A-CH (ref. [bib_ref] Inherited causes of clonal haematopoiesis in 97,691 whole genomes, Bick [/bib_ref] and identify a new genome-wide significant association with TET2-CH. Strikingly, however, we found that the association operates in the opposite direction for TET2-CH, versus DNMT3A-CH. This inverse relationship, also supported by our finding of a suggestive negative genetic correlation between TET2-and DNMT3A-CH, is tantalizing in light of recent observations that ageing has different effects on the dynamics of these two forms of CH, resulting in TET2-CH becoming more prevalent than DNMT3A-CH in those over 80 yr . Also notable in this light is the finding of an association at 6q21-CD164 with DNMT3A-CH, and a trend in the opposite direction for TET2-CH that was confirmed in the replication analysis. As CD164 is expressed in the earliest HSCs 82 and encodes a key regulator of HSC adhesion 83,84 , this proposes that HSC migration and homing may play important roles in CH pathogenesis. The reciprocal relationship of both TCL1A and CD164 with the two main CH subtypes suggests that their expression must be tightly regulated to prevent the development of one or other CH subtype, making these loci important targets for hijack by the effects of somatic mutations. In fact, a recent study 85 suggests that this may be how TET2 and ASXL1 mutations interact with a TCL1A promoter variant associated with clonal expansion rate. TCL1A is not expressed in normal or DNMT3A-mutated HSCs and the authors show that the locus becomes susceptible to activation in the presence of TET2 or ASXL1 mutations only when harboring the reference allele at the promoter variant, leading to faster clonal expansion. This type of interaction may operate for CD164 and other CH risk loci, or alternative models of interaction between the germline and somatic genome may exist.
New CH risk loci included the PARP1 coding variant rs1136410, where the G allele is protective for DNMT3A-CH and associated with reduced catalytic activity 53 suggesting that this most common form of CH may be vulnerable to PARP inhibition, in keeping with the observed synergy between PARP and DNMT inhibitors 86 . We also identified three lead variants at the TERT locus for which CH risk alleles were associated with longer LTL, a finding corroborated by our MR results linking increased LTL to CH. Interestingly, a recent study found deleterious rare germline TERT variants associated with shorter telomeres in patients with myelodysplastic syndromes 87 . However, compared with conventional myelodysplastic syndromes, these cases displayed a paucity of somatic mutations in DNMT3A (2 of 41) and TET2 (3 of 41 cases), suggesting that evolutionary paths may differ between cases with long versus short telomeres.
The rich phenotypic data captured by the UKB, coupled with our genetic analysis of CH and external GWAS datasets, enabled us to explore associations of CH using multivariable regression and interrogate, at scale, potential causal relationships between CH and its putative risk factors and consequences using MR. This highlighted that smoking and longer telomere length are causal risk factors for CH. These associations were valid across multiple CH subtypes and, in the case of smoking, corroborated by observational estimates. We also reveal that not only is genetic predisposition to CH causally associated with MPN risk, but it also increases the risk of lung, prostate, ovarian, oral/pharyngeal and endometrial cancers. In these analyses, the use of two-sample MR protected against potential reverse causality arising from cancer therapy-induced selection pressure on hematopoietic clones 88 . These MR results suggest that genetic liability to CH may be a biomarker for development of cancer elsewhere in the body, analogous to the link between genetic predisposition to Y chromosome loss in blood and solid tumor risk 89 .
We investigated the recently identified association of CH with blood-based epigenetic clocks 90 , using bidirectional MR, and show that this association is likely to be causal in the direction from CH to epigenetic age acceleration. We also showed that genetic predisposition to CH was associated with elevated circulating levels of stem cell growth factor beta, a secreted sulfated glycoprotein that regulates primitive hematopoietic progenitor cells 91 . Finally, we unraveled a previously unreported association between genetic liability to CH and atrial fibrillation risk, which was also supported by our observational analysis. However, unlike previous reports based on smaller sample sizes 5,10,22 , we did not find evidence in observational and MR analyses to support an association between CH and CAD or ischemic stroke. However, our MR analyses indicated that higher BMI and circulating apolipoprotein B levels were associated with TET2 and large clone CH risks, respectively, with apolipoprotein B being the key causal lipid risk factor for CAD [bib_ref] Evaluating the relationship between circulating lipoprotein lipids and apolipoproteins with risk of..., Richardson [/bib_ref]. We also demonstrated the impact of age, in particular, as a strong confounder of the CH-CAD/ischemic stroke associations. These results raise the possibility that reported associations of CH with CAD/stroke risks may suffer from residual confounding. Moreover, many of the cohorts that reported these associations are enriched in participants at high cardiovascular risk [bib_ref] Clonal hematopoiesis and risk of atherosclerotic cardiovascular disease, Jaiswal [/bib_ref] , in contrast to the UKB, where participants may be healthier, and potentially have lower epigenetic aging. Recent findings suggest that CH is associated with CAD/stroke only on a background of epigenetic aging 90 , offering a plausible mechanistic explanation for the absence of an association in our study.
Collectively, our findings substantially illuminate the landscape of inherited susceptibility to CH and provide insights into the causes and consequences of CH with implications for human health and ageing.
## Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41588-022-01121-z. Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/. © The Author(s) 2022
# Methods
Study population and WES data. The UKB resource was approved by the North West Multi-centre Research Ethics Committee under reference number 21/NW/0157 and all participants provided written, informed consent to participate. Participants in the UKB are volunteers and not compensated for participation. Data from the UKB resource were accessed under approved application numbers 56844, 29202 and 26041 for this study. The UKB is a prospective longitudinal study containing in-depth genetic and health information from half a million UK participants. For this study, we have selected 200,453 individuals (200k) who had WES data available (age range: 38-72, median age: 58; 55% females). WES was generated in two batches, the first of approximately 50,000 samples (50k) [bib_ref] Exome sequencing and characterization of 49,960 individuals in the UK Biobank, Van Hout [/bib_ref] and the second comprising an additional 150,000 samples (150k) [bib_ref] Advancing human genetics research and drug discovery through exome sequencing of the..., Szustakowski [/bib_ref]. Exomes were captured using the IDT xGen Exome Research Panel v.1.0 including supplemental probes; a different IDT v.1.0 oligo lot was used for each batch. Multiplexed samples were sequenced with dual-indexed 75 × 75-base-pair paired-end reads on the Illumina NovaSeq 6000 platform using S2 (50k samples) and S4 (150k samples) flow cells. The 50k samples were first computed using FE protocol and reprocessed later to match the second batch of 150k sequences which were processed using a new improved unified OQFE pipeline. As the initial 50k samples were sequenced on S2 flow cells and with a different IDT v.1.0 oligo lot from the remaining 150k samples, which were sequenced on S4 flow cells, we included the WES batch as a covariate in downstream analyses.
Sequence data processing, CH mutation calling and filtering. CRAM files generated by the OQFE pipeline were obtained from UKB [fig_ref] Figure 1 |: Characterization of CH in the uKB [/fig_ref]. Variant calling on WES data from 200,453 individuals was performed using Mutect2, Genome Analysis Toolkit (GATK) v.4.1.8.1 (ref.. Briefly, Mutect2 was run in 'tumor-only' mode with default parameters, over the exons of 43 genes previously associated with CH (Supplementary . To filter out potential germline variants we used a population reference of germline variants generated from the 1000 Genomes Project (1000GP) 95 and the Genome Aggregation Database (gnomAD) [bib_ref] The mutational constraint spectrum quantified from variation in 141,456 humans, Karczewski [/bib_ref]. All resources were obtained from the GATK Best practices repository (gs:// gatk-best-practices/somatic-hg38). Raw variants called by Mutect2 were filtered out with FilterMutectCalls using the estimated prior probability of a reading orientation artifact generated by LearnReadOrientationModel (GATK v.4.1.8.1). Putative variants flagged as 'PASS' using FilterMutectCalls or flagged as 'germline' if present at least two times with the 'PASS' flag in other samples were selected for filtering. Gene annotation was performed using Ensembl Variant Effect Predictor (VEP) (v.102) [bib_ref] The Ensembl variant effect predictor, Mclaren [/bib_ref]. We required variants with a minimum number of alternate reads of 2, evidence of the variant on both forward and reverse strands, a minimum depth of 7 reads for single nucleotide variants (SNVs) and 10 reads for short indels and substitutions, and a MAF lower than 0.001 (according to 1000GP phase 3 and gnomAD r2.1). For new variants, not previously described in the Catalogue of Somatic Mutations in Cancer (COSMIC; v.91) [bib_ref] COSMIC: the Catalogue Of Somatic Mutations In Cancer, Tate [/bib_ref] nor in the Database of Single Nucleotide Polymorphisms (dbSNP; build 153) [bib_ref] dbSNP: the NCBI database of genetic variation, Sherry [/bib_ref] , we used a minimum allele count per variant of 4, and a MAF lower than 5 × 10 −5 . From resulting variants, we selected those that: (1) are included in a list of recurring hotspot mutations associated with CH and myeloid cancer (Supplementary ; (2) have been reported as somatic mutations in hematological cancers at least seven times in COSMIC; or (3) met the inclusion criteria of a predefined list of putative CH variants [bib_ref] Inherited causes of clonal haematopoiesis in 97,691 whole genomes, Bick [/bib_ref]. We included previous variants flagged as germline by FilterMutectCalls if: (1) the number of cases in the cohort flagged as germline was lower than the ones flagged as PASS; and (2) at least one of the cases had a P < 0.001 for a one-sided exact binomial test, where the null hypothesis was that the number of alternative reads supporting the mutation was 50% of the total number of reads (95% for copy number equal to one), except for hotspot mutations which were all included. For the final list, we excluded all variants not present in COSMIC or in the list of hotspots that had a MAF equal to or higher than 5 × 10 −5 and either the mean VAF of all cases was higher than 0.2 or the maximum VAF was lower than 0.1. Frameshift, nonsense and splice-site mutations not present in COSMIC or in the hotspot list were further excluded if for each variant none of the cases had a P < 0.001 for a one-sided exact binomial test. A complete list of filtered variants is provided in .
Trait selection and modeling for observational analyses. Phenotypes were downloaded in December of 2020 and individual traits were pulled out from the whole phenotype file. Cancer, metabolic and CVD traits were generated, combining individual traits and diagnosis dates based on disease definitions (Supplementary . For each definition of disease, the first diagnosis event that occurred in each trait was selected. Baseline was defined as the date of sample collection when the individuals attended the assessment centers. The prevalent cases are those identified before the baseline, while incident cases were defined as the events that occurred after the baseline. Unless specified, all regression models included age, sex, smoking status, WES batch and the first ten ancestry principal components as covariates and all analyses were adjusted for multiple comparisons using the false discovery rate computed by the Benjamin-Hochberg procedure implemented in the p.adjust function (R stats package v.4.0.2). Blood cell counts and biochemical traits were log 10 transformed and analyzed using a logistic regression model with overall and gene-specific CH as outcomes, including the assessment center as covariate and, in the case of cholesterol and cholesterol species, the use of cholesterol-lowering medication as an additional covariate. Individuals with myeloid malignancies or hematological neoplasms at baseline (that is, with a cancer diagnosis date before the date they attended the assessment centers) were excluded from the analysis. For cancer, CVD and death risk, we performed a time-to-event regression analysis. In the case of cancer and CVD, we performed a competing risk analysis, using the date of death by other cause as the competing event, while for the risk of death we used the Cox proportional hazards model. The cancer/CVD/death event was used as an outcome and CH was considered as the exposure in these analyses. Individuals without the event who died before the end of the follow-up were censored at the time of death, while the rest were censored at the end of the follow-up. For CVD and death risk analyses, we also included BMI, high-density lipoprotein cholesterol, low-density lipoprotein cholesterol, triglycerides, T2D status and hypertension status as covariates. Individuals with myeloid or other malignant neoplasms at baseline were excluded from all aforementioned analyses. The proportional hazards assumption for the Cox and competing risk models was assessed by examining the Schoenfeld residuals. For the phenome-wide association analysis between International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10) codes as outcomes and CH status, logistic regression models were used including age, sex, WES batch and the first ten genetic ancestry principal components as covariates. Analyses were performed over 11,787 selected ICD-10 codes corresponding to disease conditions (A to N), symptoms, signs, and abnormal clinical and laboratory findings (R), and factors influencing health status (Z). All analyses were performed using glm (R stats package v.4.0.2), coxph (R survival package v.3.2-11) and crr (R cmprsk package v.2.2-10) functions.
Genome-wide association analyses. Germline genotype data used were from the UKB release that contained the full set of variants imputed into the Haplotype Reference Consortium 100 and UK10K + 1000GP (ref. reference panels and genotyped on the UK BiLEVE Axiom Array or UKB Axiom Array 101 . Derivation of the analytic sample for UKB individuals of European ancestries followed the QC protocol of Astle et al. [bib_ref] The allelic landscape of human blood cell trait variation and links to..., Astle [/bib_ref] and included the following steps: after filtering genetic variants (call rate ≥ 99%, imputation quality info score > 0.9, Hardy-Weinberg equilibrium P ≥ 10 −5 ) and participants (removal of genetic sex mismatches), we excluded participants having non-European ancestries (self-report or inferred by genetics) or excess heterozygosity (>3 s.d. from the mean), and included only one of each set of related participants (third-degree relatives or closer). After QC, we were left with 10,203 individuals with CH and 173,918 individuals without CH. The subset with CH included 5,185 and 2,041 individuals with DNMT3Aand TET2-mutant CH, respectively, and 4,049 and 6,154 individuals with large (VAF ≥ 0.1) and small (VAF < 0.1) clone size CH, respectively. Association analyses were performed for autosomal and X chromosomal variants using noninfinitesimal linear mixed models implemented in BOLT-LMM [bib_ref] Efficient Bayesian mixed-model analysis increases association power in large cohorts, Loh [/bib_ref] (v.2.3.6) with age at baseline, sex and first ten genetic principal components included as covariates.
Statistically independent lead variants for each CH phenotype were defined using LD-based clumping with an r 2 threshold of 0.05 applied across all genotyped and imputed variants, with P < 5 × 10 −8 , imputation quality score > 0.6 and MAF > 1%. This was implemented using the FUMA pipeline (v.1.3.6b) (ref. [bib_ref] Functional mapping and annotation of genetic associations with FUMA, Watanabe [/bib_ref]. For the rare variant association scan, we used more stringent cut-offs of P < 10 −9 and imputation quality score > 0.8 to define lead variants but did not require LD-clumping since only one such association was identified. Approximate conditional analysis conditioning on the common (MAF > 1%) lead variants was performed using the --cojo-cond flag in the Genome-wide Complex Trait Analysis (GCTA) v. .
We also evaluated associations of the lead variants for overall CH risk in the 505 individuals with CH and 11,893 controls (retained after the QC steps described above), comprising the ancestrally diverse (non-European) subcohort of the 200k UKB cohort, using logistic regression and adjusting for age, sex, WES batch and 40 genetic ancestry principal components.
Replication of genome-wide significant associations. Replication analysis was performed using 221,285 unrelated UKB individuals of European ancestry (age range: 39-73, mean age: 57; 53% females), for whom WES was performed subsequent to the initial 200k, using the same protocol. Alignment to the GRCh38 genome reference with Illumina DRAGEN Bio-IT Platform Germline Pipeline v.3.0.7 and QC were performed as detailed by Wang et al. [bib_ref] Rare variant contribution to human disease in 281,104 UK Biobank exomes, Wang [/bib_ref]. Somatic variant calling was performed with GATK's Mutect2 (v.4.2.2.0) using a panel of normals to remove recurrent artifacts, and subsequent filtering was performed with FilterMutectCalls, including the filtering of read orientation artifacts using priors generated with LearnReadOrientationModel. Putative somatic variants were identified from Mutect2 'PASS' calls in DNMT3A and TET2 based on (1) matching the list of putative somatic mutations identified in the discovery cohort, or (2) any DNMT3A or TET2 protein-truncating variants as predefined by Wang et al. [bib_ref] Rare variant contribution to human disease in 281,104 UK Biobank exomes, Wang [/bib_ref]. Sample sizes for DNMT3A-, TET2-and large and small clone DNMT3A-or TET2-mutant CH are provided in . Replication association statistics were calculated on the 221,285 replication exomes using the imputed genotype data with logistic regression, adopting age, sex and the first four genetic ancestry principal components as covariates.
Heritability, cell-type enrichment and genetic correlation. We used LDSC (v.1.0.1) [bib_ref] LD score regression distinguishes confounding from polygenicity in genome-wide association studies, Bulik-Sullivan [/bib_ref] to estimate the narrow-sense heritability of CH on the liability scale assuming the population prevalence of CH to be 10% (based on the prevalence of CH in the UKB '200k' cohort as shown in [fig_ref] Figure 1 |: Characterization of CH in the uKB [/fig_ref] and constraining the LDSC intercept to 1. The intercept, which in its unconstrained form protects from bias due to population stratification, was constrained to 1 to provide more precise estimates given that there was little evidence of inflation in test statistics due to population structure in unconstrained analysis (unconstrained intercept estimated as 1.009 (s.e. = 0.0067) and lambda genomic control factor of 0.999). We used the pre-computed 1000 Genomes phase 3 European ancestry reference panel LD score dataset for heritability estimation. We used the same LD scores and the --rg flag in LDSC to estimate the genetic correlation between the CH and mCA GWAS summary statistics [bib_ref] Hematopoietic mosaic chromosomal alterations increase the risk for diverse types of infection, Zekavat [/bib_ref]. Cell-type group partitioned heritability analysis was performed using LD scores partitioned across 220 cell-type-specific annotations that were divided into 10 groups 24 : central nervous system, cardiovascular, kidney, adrenal/pancreas, gastrointestinal, connective/bone, immune/hematopoietic, skeletal muscle, liver and other. Each of the ten groups contained cell-type-specific annotations for four histone marks: H3K9ac, H3K27ac, H3K4me1 and H3K4me3 (ref. [bib_ref] Partitioning heritability by functional annotation using genome-wide association summary statistics, Finucane [/bib_ref]. We also used LD scores annotated based on open chromatin state (assay for transposase-accessible chromatin using sequencing (ATAC-seq)) profiling by Corces et al. [bib_ref] Lineage-specific and single-cell chromatin accessibility charts human hematopoiesis and leukemia evolution, Corces [/bib_ref] [bib_ref] Heritability enrichment of specifically expressed genes identifies disease-relevant tissues and cell types, Finucane [/bib_ref] in various hematopoietic progenitor cells and lineages at different stages of differentiation. To estimate the genetic correlation between DNMT3Aand TET2-CH and between large and small clone CH we used the high-definition likelihood (HDL; v.1.4.0) 30 inference approach to improve power given the low sample size in each subtype-specific CH GWAS.
## Gene-level association and network analyses.
We undertook genome-wide gene-level association analyses using two complementary approaches. First, we used MAGMA (v.1.08 implemented in FUMA v.1.3.6b) which involves mapping germline variants to the genes they overlap, accounting for LD between variants and performing a statistical multi-marker association test [bib_ref] MAGMA: generalized gene-set analysis of GWAS data, De Leeuw [/bib_ref]. Second, we performed a transcriptome-wide association study using blood-based cis gene eQTL data on 31,684 individuals from the eQTLGen consortium [bib_ref] Large-scale cis-and trans-eQTL analyses identify thousands of genetic loci and polygenic scores..., Võsa [/bib_ref] and SMR coupled with the HEIDI colocalization test to identify germline genetic associations with CH risk mediated via the transcriptome [bib_ref] Integration of summary data from GWAS and eQTL studies predicts complex trait..., Zhu [/bib_ref]. The gene-level genome-wide significance threshold in the MAGMA analyses was set at P = 2.6 × 10 −6 to account for testing 19,064 genes and for SMR was set at P = 3.2 × 10 −6 after adjustment for testing 15,672 genes. Further, only genes with SMR P < 3.2 × 10 −6 and HEIDI P > 0.05 were declared genome-wide significant in the SMR analyses since the HEIDI P > 0.05 strongly suggests colocalization of the GWAS and eQTL signals for a given gene [bib_ref] Integration of summary data from GWAS and eQTL studies predicts complex trait..., Zhu [/bib_ref]. NetworkAnalyst 3.0 (ref. [bib_ref] NetworkAnalyst 3.0: a visual analytics platform for comprehensive gene expression profiling and..., Zhou [/bib_ref] was used for network analysis. All genes with P < 10 −3 in each MAGMA analysis for overall, DNMT3A-and TET2-mutant, and large and small clone CH were used as input. The protein-protein interactome selected was STRING v.10 (ref. [bib_ref] STRING v10: protein-protein interaction networks, integrated over the tree of life, Szklarczyk [/bib_ref] with the recommended parameters (confidence score cut-off of 900 and requirement for experimental evidence to support the PPI). The largest possible network was constructed from the seed genes/proteins and the interactome proteins [bib_ref] NetworkAnalyst 3.0: a visual analytics platform for comprehensive gene expression profiling and..., Zhou [/bib_ref]. Hub nodes were defined as nodes with degree centrality ≥ 10 (that is, a node with at least 10 edges or connections to other proteins in the network as a measure of its importance in the network and consequently its biology). Pathway analysis of this largest network was conducted using the enrichment tool built into NetworkAnalyst and with the Reactome pathway repository therein 108 .
Fine-mapping and target gene prioritization. We fine-mapped the lead variant signals identified by the FUMA LD-clumping pipeline using the Probabilistic Identification of Causal Single Nucleotide Polymorphisms (PICS2; v.2.1.1) algorithm [bib_ref] PICS2: next-generation fine mapping via probabilistic identification of causal SNPs, Taylor [/bib_ref] [bib_ref] Genetic and epigenetic fine mapping of causal autoimmune disease variants, Farh [/bib_ref] to identify candidate causal variants most likely to underpin each association. The PICS2 algorithm computes the likelihood that each variant in LD with the lead variant is the true causal variant in the region by leveraging the fact that for variants associated merely due to LD, the strength of association scales asymptotically with correlation to the true causal variant [bib_ref] PICS2: next-generation fine mapping via probabilistic identification of causal SNPs, Taylor [/bib_ref]. We only retained variants with a PICS2 probability of 1% or more in our final list of fine-mapped candidate causal variants. We overlapped these fine-mapped variants with gene body annotations 48 using GENCODE release 33 (ref. [bib_ref] GENCODE reference annotation for the human and mouse genomes, Frankish [/bib_ref] (build 37) annotations after removing ribosomal protein genes. Fine-mapped variants were also overlapped with ATAC-seq peaks across 16 hematopoietic progenitor cell populations and ATAC-RNA count correlations calculated using Pearson coefficients for hematopoietic progenitor cell RNA counts of genes within 1 Mb of the ATAC peaks and these were used to identify putative target genes of fine-mapped variants that overlapped ATAC-seq peaks [bib_ref] Lineage-specific and single-cell chromatin accessibility charts human hematopoiesis and leukemia evolution, Corces [/bib_ref] [bib_ref] Inherited myeloproliferative neoplasm risk affects haematopoietic stem cells, Bao [/bib_ref] [bib_ref] Integrated single-cell analysis maps the continuous regulatory landscape of human hematopoietic differentiation, Buenrostro [/bib_ref] [bib_ref] Interrogation of human hematopoiesis at single-cell and single-variant resolution, Ulirsch [/bib_ref]. We also looked up the SIFT [bib_ref] SIFT missense predictions for genomes, Vaser [/bib_ref] and PolyPhen 52 scores for these fine-mapped variants using the SNPnexus v.4 annotation tool [bib_ref] SNPnexus: a web server for functional annotation of human genome sequence variation..., Oscanoa [/bib_ref] to identify coding variants with predicted functional consequences. Finally, we used the Open Targets Genetics resource [bib_ref] Open Targets Genetics: systematic identification of trait-associated genes using large-scale genetics and..., Ghoussaini [/bib_ref] to identify the most likely target gene of the lead variant at each locus as per Open Targets and used this in our omnibus target gene prioritization scheme described below.
To prioritize putative target genes at the P lead-variant < 5 × 10 −8 loci identified by our GWAS of overall CH, DNTM3A-CH, TET2-CH and large/small clone size CH, we combined gene-level genome-wide significant results from (1) MAGMA and (2) SMR with (3) PPI network hub status of the gene, (4) variant-to-gene searches of the Open Targets database for lead variants, and overlap between fine-mapped variants and (5) gene bodies, (6) regions with accessible chromatin (ATAC-seq peaks) across 16 hematopoietic progenitor cell populations that were also correlated with nearby gene expression (RNA sequencing) in the same cell populations and (7) missense variant annotations from SIFT and PolyPhen. Genes nominated by at least two of the seven approaches were listed (except where only one of the seven methods nominated a single gene in a region in which case that gene was listed) and the genes nominated by the largest number of approaches represented the most likely targets at each locus. We also evaluated the 'druggability' of the prioritized functional target genes in the context of known therapeutics and ongoing drug development using the Open Targets Platform 56 and canSAR 57 v.1.5.0 databases. The database canSAR provides chemistry-based (assesses the likely 'ligandability' of a protein based on the chemical properties of compounds tested against the protein itself and/or its homologs) and antibody-based (assesses if a target is potentially suitable for antibody therapy) predictions.
Phenome-wide association scan for lead variants. We used PhenoScanner V2 (refs. [bib_ref] PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations, Kamat [/bib_ref] [bib_ref] PhenoScanner: a database of human genotype-phenotype associations, Staley [/bib_ref] with catalog set to 'diseases & traits' , P value set to '5E-8' , proxies set to 'EUR' and r 2 set to '0.8' to search for published phenome-wide associations between our lead variants or variants in strong LD (r 2 > 0.8) with the lead variants and other diseases and traits.
MR analyses. MR 111,112 uses germline variants as instrumental variables to proxy an exposure or potential risk factor and evaluate evidence for a causal effect of the exposure or potential risk factor on an outcome. Due to the random segregation and independent assortment of alleles at meiosis, MR estimates are less susceptible to bias from confounding factors as compared with conventional observational epidemiological studies. As the germline genome cannot be influenced by the environment after conception or by preclinical disease, MR estimates are also less susceptible to bias due to reverse causation. MR estimates represent the association between genetically predicted levels of exposures or risk factors and outcomes, as compared with conventional observational epidemiological estimates, which represent direct associations of the exposure or risk factor levels with outcomes. Effect allele harmonization across GWAS summary statistics datasets followed by MR analyses were performed using the TwoSampleMR v.0.5.6 R package. The CH phenotypes were considered as both exposures (to identify consequences of genetic liability to CH) and outcomes (to identify risk factors for CH). When considering CH phenotypes as outcomes, germline variants associated with putative risk factors or exposures at P < 5 × 10 −8 were used as genetic instruments for the risk factors/exposures, except for the appraisal of circulating cytokines and growth factors [bib_ref] Genome-wide association study identifies 27 loci influencing concentrations of circulating cytokines and..., Ahola-Olli [/bib_ref] wherein variants associated with cytokines/growth factors at P < 10 −5 were used as instruments. IVW analysis 113 was the primary analytic approach with pleiotropy-robust sensitivity analyses carried out using the MR-Egger 78 and weighted median 77 methods. A full list of external GWAS data sources used for MR analyses is provided in . We also conducted an MR-PheWAS evaluating overall CH and DNMT3A-CH as exposures (using variants associated with these at P < 10 −5 ) and 1,434 disease and trait outcomes in the UKB data using summary genetic association statistics for the outcomes that were generated by the Neale lab (http://www.nealelab.is/uk-biobank) and accessed via the TwoSampleMR v.0.5.6 R package and the Integrative Epidemiology Unit (IEU) OpenGWAS project portal. FDR control was applied to the MR-PheWAS IVW analysis P values.
## Statistics and reproducibility.
No statistical method was used to predetermine sample size. The experiments were not randomized and investigators were not blinded during the experiments and outcome assessment. Participants were excluded from the GWAS due to genetic sex mismatch, excess heterozygosity (>3 s.d. from the mean) and relatedness (only one of each set of participants who were third-degree relatives or closer were retained). To summarize, our study design included observational genomic analyses of CH in 200,453 individuals across ancestries, genome-wide association and post-GWAS analyses for five CH traits (overall, DNMT3A, TET2, large clone and small clone CH) in 184,121 individuals of European ancestry, followed by trans-ancestry genetic association analyses in 12,398 individuals, and replication genetic association analyses in an additional 221,285 individuals of European ancestry-all from the UKB. [fig_ref] Figure 1 |: Characterization of CH in the uKB [/fig_ref] | Characterization of CH in the uK Biobank. a, Histogram stratified by sex showing the age distribution of individuals in the UKB cohort (n=200,453). b, Overall percentage of females and males in the UKB cohort. c, Percentage of the most common self-reported ancestry groups in the UKB cohort. Ancestry groups with a frequency lower than 1% were grouped under the 'Other ancestry group' category. d, Number of individuals with 1, 2, 3, and 4 somatic mutations. More than 90% of individuals with CH had only one driver mutation identified. e, Percentages of different CH mutation types identified. f, Relative prevalence of each of the six base substitution types amongst the identified CH mutations. g, Density plot showing the variant allele fraction (VAF) distribution of all CH somatic mutations. h, Density plot showing similar VAF distribution for different mutation types. Mean and median are indicated for g and h. | Age distribution of CH by mutant gene, clone size, and sex. a, Prevalence of CH in the cohort with advancing age. The blue line represents the smoothed model fitted to a generalized additive model with 95% confidence interval (CI; gray shadow). b, Prevalence of CH by age stratified by the top eight most frequently mutated genes. Colored lines represent the smoothed model fitted to a generalized additive model with 95% CI (colored shadows). Y-axis is log-scaled. c, Clone size, estimated by the variant allele fraction (VAF), increases with age. The blue line represents the smoothed model fitted to a generalized additive model and the shadow represents the 95% CI. d, Empirical cumulative distribution (ECD) of the age of individuals with CH stratified by sex. CH was observed one year earlier in females than in males (median 61 versus 62 years; P=1.6x10 −4 , two-sided pairwise Wilcoxon rank sum test).
## Extended data
## Extended data
[fig] Figure 1 |: Characterization of CH in the uKB. a, Composite plot summarizing mutations in the 10 most common driver genes in 10,924 individuals with CH. Each column in the waterfall plot represents a single individual, with mutation types color-coded. Bars on the left quantify mutations per gene as a percentage of all CH mutations identified. Violin plots on the right show the distribution of VAFs, with vertical lines representing the median and dots with horizontal lines the mean ± s.d. b, Empirical cumulative distribution (ECD) of the age of individuals with CH overall (black) and stratified by the eight most common driver genes. Compared with DNMT3A, mutations in ATM were observed 3 yr earlier (P = 7.2 × 10 -4 ), while mutations in ASXL1, PPM1D, SRSF2 and SF3B1 were observed 1 (P = 2.7 × 10 −8 ), 1 (P = 8.5 × 10 −6 ), 2 (P = 5.7 × 10 −10 ) and 3 (P = 6.5 × 10 −6 ) years later, respectively. Differences were calculated using two-sided pairwise Wilcoxon rank sum tests. c, Bar plot showing the female to male (F:M) ratio of CH carriers with mutations in the ten most common driver genes. GNB1 (P = 2.3 × 10 −3 ) and DNMT3A (P = 3.2 × 10 −11 ) show a higher F:M ratio, while PPM1D (P = 6.4 × 10 −3 ), TP53 (P = 1 × 10 −3 ), JAK2 (P = 5.7 × 10 −3 ), SF3B1 (P = 3.5 × 10 −4 ), ASXL1 (P = 5.7 × 10 −28 ) and SRSF2 (P = 3.8 × 10 −14 ) show lower F:M ratio. 'Other' represents the remaining driver genes grouped together and 'Ctrl' the ratio for individuals without CH. Dotted vertical line shows the F:M ratio observed in the full cohort (F:M = 1.2). P values are from a chi-squared test comparing the distribution for each gene with 'Ctrl'. [/fig]
[fig] Figure 3 |: Cell-type-specific enrichment of the CH polygenic signal. a, Heritability enrichment of CH across histone marks profiled in ten cell-type groups. b, Heritability enrichment of CH across open chromatin regions identified by ATAC-seq in hematopoietic progenitor cells/lineages at different stages of differentiation. Partitioned heritability cell-type group analysis in the LDSC software was used to compute these enrichments and corresponding P values. The data underlying the figures are available inSupplementary Tables 14 and 15. CNS, central nervous system; GI, gastrointestinal; CLP, common lymphoid progenitor; CMP, common myeloid progenitor; MPP, multipotent progenitor; GMP, granulocyte/ macrophage progenitor; LMPP, lymphoid-primed multipotent progenitor; NK, natural killer cell; Mono, monocyte; Erythro, erythroid progenitor. [/fig]
[fig] Figure 4 |: Manhattan plots displaying genome-wide associations between common germline genetic variants and each of five CH traits. The y axes depict P values (−log 10 ) for associations derived from the noninfinitesimal mixed model association test implemented in BOLT-LMM. The x axes depict chromosomal position on build 37 of the human genome (GRCh37). The dotted lines indicate the genome-wide significance threshold of P = 5 × 10 −8 . [/fig]
[fig] Figure 5 |: Gene-level association and PPI network analyses. a, Gene-level associations in the 6q21 region within 25 kb of CRIP3, that is, between GRCh37 positions 43,017,448 and 43,526,535 on chromosome 6. The x axis lists all the genes in this region that were tested by both MAGMA and SMR. MAGMA uses a multiple linear principal components regression model while SMR is based on the Wald test. CRIP3 was the only gene located more than 1 Mb away from a GWAS-identified lead variant that was found to be associated with CH at gene-level genome-wide significance by both MAGMA and SMR. The y axis depicts the P value (−log 10 ) for association in the MAGMA and SMR analyses. The gene-level genome-wide significance threshold in MAGMA (P = 2.6 × 10 −6 after accounting for 19,064 genes tested) is indicated by the blue dashed line and in SMR (P = 3.2 × 10 −6 after accounting for 15,672 genes tested) by the orange dotted line. Both CRIP3 and SRF had SMR HEIDI P > 0.05 indicating colocalization of the GWAS and eQTL associations. [/fig]
[fig] Figure 6 |: IVW MR forest plots with CH traits as outcomes. a-c, ORs for CH risk are represented as per [/fig]
[table] Table 34: Symbols represent OR markers, and OR marker symbols with corresponding P < 0.05 are represented by filled circles. Error bars represent 95% CIs. Sample sizes for the smoking, alcohol, BMI, WHRadjBMI, T2D, apolipoproteins B and A-I, LDL, HDL and triglycerides analyses are provided in Supplementary Table 34. Sample sizes for the LTL, IEAA, Hannum and three blood cell count analyses are provided in Supplementary Table 35. Full results, including from sensitivity analyses, are presented in Supplementary Tables 36-38. WHRadjBMI, waist-to-hip ratio adjusted for BMI; LDL, low-density lipoprotein cholesterol; IEAA, intrinsic epigenetic age acceleration. [/table]
|
10.22038/ijbms.2021.52890.11923
|
CCBY
|
8087848
|
33995949
|
s2orc_pubmed_articles
|
Immunogenicity evaluation of the HIV-1 Tat containing polyepitope DNA vaccine adjuvanted with CpG-ODNs in mice
Objective(s):HIV-1 is still considered a serious threat to human health, and accessibility of a suitable and efficient vaccine is urgently needed to address the disease burden. DNA vaccines employ the cells of the vaccinated hosts for in situ production of the vaccines. This strategy is an alternative and effective approach for traditional vaccination against high-risk pathogens, e.g., HIV-1. On the other hand, polyepitope vaccines, containing several immunogenic and conserved epitopes from virus vital regulatory and structural proteins, could more efficiently induce cellular and humoral immune responses against different clades of the virus. Materials and Methods: Herein, we comparatively investigated the immunogenic potency of the HIV-1 polytope DNA vaccine containing CpG oligodeoxynucleotides (CpG-ODNs) in BALB/c mice. To this end, after verifying the expression of the recombinant sequence in the eukaryotic HEK 293 cell line, it was amplified and extracted in the prokaryotic host cells (E. coli DH5α)) and then formulated and administered intramuscularly (IM) to the experimental mice (on days 0, 14, and 28) with and without CpG-ODNs adjuvant. Results: Taken together, the results demonstrated that CpG-ODNs adjuvanted DNA vaccine could significantly elicit cellular and humoral immune responses in the immunized animals in comparison with the control ones (P<0.05). Conclusion: Regarding the obtained results and also considering the advantages of polytopic and DNA vaccines, this approach might be considered a new regimen in HIV-1/AIDS vaccination.There are no conflicts of interest in this research.
# Introduction
Since identification of HIV-1/AIDS, many people have been affected by the virus worldwide and almost 50% of them have died. Thus, accessibility to a suitable efficient vaccine is in urgent need to address the disease burden [bib_ref] HIV vaccine development at the turn of the 21st century, Girard [/bib_ref]. Although different types of vaccines have been developed [bib_ref] How can improve DNA vaccine modalities as a therapeutic approach against HIV..., Habibzadeh [/bib_ref] , they failed to provide the desired protection due to some stumbling blocks, including the high genetic diversity of the virus, its capacity to evade the adaptive immune response, induction of weak antibody responses, etc. [bib_ref] Challenges in the development of an HIV-1 vaccine, Barouch [/bib_ref].
Discovery of DNA vaccines (1980s) circumvented the need for production of protein purification vaccines in laboratories. These vaccines employ the cells of the vaccinated hosts for in situ production of antigens and keep long-lived effector activity [bib_ref] DNA Vaccines-A Review, Lavanya [/bib_ref] [bib_ref] DNA vaccines: A review, Liu [/bib_ref]. In this vaccination approach, the DNA is finally converted to the antigen of interest following the plasmid delivery and can then induce immune responses and mimics the natural infection. This in situ intracellular production of antigens not only elicits humoral immunity but also leads to induction of cellular immune responses [bib_ref] DNA Vaccines-A Review, Lavanya [/bib_ref] [bib_ref] DNA vaccines: A review, Liu [/bib_ref]. It is established that an efficient HIV-1 vaccine has to induce durable cross-clade humoral and cellular immune responses [bib_ref] HIV vaccine development at the turn of the 21st century, Girard [/bib_ref]. To meet these critical requirements, various attempts have been made to assemble cocktails of conserved and protective immunogenic epitopes derived from different HIV-1 proteins. This approach has some valuable advantages including no risk for integration of gene into the host cell, no infection triggering during immunization, flexibility in polyepitope DNA vaccine designing, etc. The latter can elicit several types of immune responses that broaden the spectrum of induced immune responses and induce robust immunogenicity of vaccines [bib_ref] Infectious bronchitis virus poly-epitope-based vaccine protects chickens from acute infection, Tan [/bib_ref]. Designing an effective polyepitope vaccine needs some important considerations such as the immunogenicity of the desired epitope, conservancy of the desired epitope in different HIV-1 clades, and similar affinity of the selected epitopes for binding to MHC molecules [bib_ref] Formulation of chitosan with the polyepitope HIV-1 protein candidate vaccine efficiently boosts..., Jazaeri [/bib_ref].
Several clinical trials have shown that DNA vaccines induce relatively modest immunogenicity in large animals, especially humans [bib_ref] Enhancement of HIV-1 DNA vaccine immunogenicity by BCG-PSN, a novel adjuvant, Sun [/bib_ref]. Therefore, the immunogenicity of DNA vaccines could be reinforced by using different vectors, adjuvants, or delivery systems [bib_ref] DNA Vaccines-A Review, Lavanya [/bib_ref].
In this research, the immunogenicity of the HIV-1 ployepitopic DNA vaccine in mice was evaluated in the presence and absence of CpG-ODNs as an adjuvant. In this regard, after verification of the recombinant vaccine expression in the eukaryotic host cells (HEK 293 cell line), its amplification (in E. coli DH5α) and extraction was carried out and it was then formulated either alone or adjuvanted with CpG-ODNs and finally administered IM to the experimental mice on days 0, 14, and 28. The immunogenicity of the candidate vaccines was then comparatively assessed and analyzed.
# Materials and methods
# Materials
Goat anti-mouse (IgG 1 and IgG 2a ) secondary antibodies and CpG oligodeoxynucleotides (CpG-ODNs) adjuvant were purchased from Sigma (St. Louis, MO, USA). Ampicillin antibiotic was provided from Invitrogen (Carlsbad, CA, USA). The kits for the 5-Bromo-2-deoxyuridine (BrdU) test and lactate dehydrogenase (LDH) assay were provided by Takara. The restriction endonucleases were purchased from Fermentas. Cell culture medium RPMI-1640 was supplied by Gibco. Cytokine assay kit was obtained from Due Set, and Merck provided all of the other chemicals. Data reproducibility was confirmed by at least three independent experimental repeats and the presented results are typical experimental data.
## Polyepitope vaccine characteristics
The pcDNA polyepitope vector was obtained from Iran's Pasteur Institute (IPI, Tehran, Iran). Some important criteria have been carefully met in designing the recombinant vector. It includes six amino acid fragments from the immunogenic and conserved epitopes of the HIV-1 Tat, Pol, Gag, and Env antigens that have been tandemly conjoined as a full-length polytope using the LosAlamos HIV immunology database (http:// www.hiv.lanl.gov/content/immunology/index.html). 3 spacer sequences (AAA and AAY) have been inserted between the epitopes for optimization of cleavage by the proteasome and also reducing the possibility of formation of junctional epitopes. CLC Main Workbench 5.5 software (CLC bio, USA) was applied to assess and reduce the hydrophobic residue accumulation (9) (supplementary table).
# Western blotting analysis
After ensuring the presence of the recombinant sequence in the pcDNA3.1 vector (using double digestion with BamHI and XhoI restriction endonucleases) and its transfection into the eukaryotic host cell (HEK 293 cell line, as one of the potent eukaryotic expression systems), the expression of the candidate vaccine was verified by using Western blot analysis. Briefly, HEK 293 cell line was seeded in a 75 cm 2 Cell Culture Flask and once the confluency of the cells reached 70-90 percent, they were transfected by the recombinant pcDNA3.1 expression vector, using lipofectamine 2000 reagent.
48 and 72 hr post transfection, cell harvesting was carried out by centrifugation and the cells were then lysed in a lysis buffer (40 mM Tris HCl (pH 7.4), 2 M BioLite, 7 M urea, 50 mM dithiothreitol (DTT), 0.2% 3-[(3-Cholamidopropyl)dimethylammonio]-1-propanesulfonate (CHAPS), 4% (w/v) thiourea, protease inhibitor cocktail (Roche) and 200 µM phenylmethylsulfonyl fluoride (PMSF). We did not determine the cell lysate concentration (total protein), and to qualitatively confirm the protein expression, we loaded and ran the maximum possible amount of the extract on 12% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) well. Blotting was then carried out according to the procedure (10), and a chemiluminescence image station (Kodak) was used for visualization of the protein band.
## Amplification and expression of the recombinant vector
The recombinant pCDNA3.1 vector was extracted and transformed into the competent cells (E. coli (DH5α)) by a chemical method. The transformed cells were cultivated for 12-16 hr in 250 ml of Luria-Bertani broth (LB) medium containing ampicillin (50 μg ml −1 ) in 1 L Erlenmeyer and the temperature and vigorous shaking were adjusted to 37 °C and 250 rpm, respectively. Harvesting of the cells was carried out by centrifugation at 5000 g, 4 °C, 20 min. After removing the supernatant, the pellet was used for plasmid extraction using a commercial endotoxin-free Qiagen plasmid extraction kit. The content of endotoxin was < 0.1 EU/ µg DNA (endotoxin units per µg plasmid DNA), and the amount of the supercoiled plasmid was > 80%.
## Mice and vaccination protocol
The immunogenicity of the candidate DNA vaccines (with and without CpG-ODNs adjuvant) was tested in 6-8 week old BALB/c mice. National Institutes of Health Guidelines for the Care and Use of Laboratory Animals (NIH publications No. 8023, revised 1978) were carefully considered during the work. The animals were divided randomly into 4 groups and 6 mice per group. Immunization was carried out intramuscularly (IM) 3 times at 2-week intervals (on days 0, 14, and 28) with the DNA vaccine alone or adjuvanted with CpG-ODNs [fig_ref] Table 1: The immunization protocol for the HIV-1 DNA vaccine candidates [/fig_ref].
## Measurement of antibody levels
Bleeding was done seven days after the final vaccination. Sera samples from the immunized mice were separated and our previously optimized indirect ELISA method was used to measure the levels of IgG subclasses (IgG 1 and IgG 2a ) [bib_ref] Formulation of chitosan with the polyepitope HIV-1 protein candidate vaccine efficiently boosts..., Jazaeri [/bib_ref].
In brief, 100 μl of 10 μg/ml of the protein vaccine buffered with PBS was coated into 96-well ELISA Maxisorp plates overnight and the temperature was adjusted to 4 °C. The protein was recombinantly expressed in a prokaryotic system (E. coli BL21 (DE3)) and purification was performed by Ni-NTA affinity resin. The levels of the elicited antibodies were finally determined according to our previous method using horseradish peroxidase (HRP) conjugated goat antimouse IgGs (7).
## Cellular immunity assays
For assessment, the cellular immune responses induced by the HIV-1 polyepitope DNA vaccines, proliferation of lymphocytes, production of interferongamma (IFN-γ), and cytotoxic activity of T lymphocytes (CTL) were measured as the hallmarks of cellular immunity. For this purpose, 2 weeks after the third and last vaccination, isolation of the vaccinated mice splenocytes, preparation of single-cell suspensions, and development of ELISA assays were performed as mentioned previously (7). To assay the lymphocyte proliferation which was performed using 5-Bromo-2-deoxyuridine (BrdU) kit, concanavalin-A (Gibco, with a final concentration 5 μg ml -1 ) was utilized as the positive control while complete culture medium and unstimulated wells were respectively used as the blank and negative control, Stimulation of the cells' suspensions was carried out using the HIV-1 corresponding polyepitope protein as antigen recall. The optical density (OD) values were recorded at 450 nm with a Microplate Reader (Biohit 8000, USA), and the stimulation index (SI) calculated as OD of stimulated wells/OD of unstimulated wells, were used to report the cell proliferation.
Measurement of IFN-γ produced by the immunized mice was also carried out using R&D Systems commercial kit and IFN-γ cytokine quantity (pg ml -1 ) was calculated by plotting the standard curve. To assay the CTL activity, the suspensions obtained from splenocytes of the vaccinated animals were used as effector cells and the mouse myeloma Sp2-0 tumor cell line was applied as the target cells. The standard lactate dehydrogenase (LDH) assay was performed for in vitro detection of the cell-mediated killing of target cells at various ratios of effector/target cells. The percent of cytotoxicity was finally accounted by the corresponding formula (7).
The complete cell lysate was obtained using Triton X-100 (1%) and the spontaneous release was examined with target cells prepared in RPMI-1640 containing BSA (2%).
# Statistical analysis
All experiments presented in this paper were performed in triplicates and repeated 3 times. Procession and analysis of data were carries out using Graphpad Prism software (version 6.01) and presented as means±standard deviation (SD) using the one-way ANOVA method.
# Results
## The polyepitope dna vaccine production and preparation
The existance of the HIV-1 polyepitope DNA fragment in the recombinant pcDNA3.1 plasmid was verified using double enzymatic digestion and agarose gel analysis (1% agarose) . The expression of the HIV-1 polyepitope protein in the HEK 293 cell line was also verified by Western blot analysis . The vector amplification and extraction were then carried out and DNA concentration was determined using a spectrophotometer (Picodrop-Pico200, UK)
## Evaluation of humoral immunity
The results of the assessment of the specific IgG 1 levels in the serum samples of vaccinated animals suggested that immunization of the mice with the candidate DNA vaccines could significantly increase the antibody levels in the vaccinated animals in comparison with the control groups . Similar results were also obtained for the specific IgG 2a amounts in the sera of the vaccinated mice (P<0.05). The animals vaccinated with the CpG-ODNs containing DNA vaccine produced greater amounts of both antibodies (IgG 1 and IgG 2a ) in comparison with those who received the candidate DNA vaccine alone. Nevertheless, no significant difference was observed (P>0.05).
## Examination of cellular immunity
BrdU method was used for evaluation of the For further investigation of the cellular immune responses induced by the DNA vaccines, the amount of IFN-γ secreted by Th1 cells was evaluated using an ELISA method. IFN-γ secretion occurred upon specific recognition of the HIV-1 polyepitope antigens that had been expressed in the eukaryotic host cells and used as antigen recall. According to the results , increased secretion of IFN-γ was measured in response to the DNA candidate vaccines (P<0.05) and its quantity was remarkably higher in the animals receiving DNA vaccine adjuvanted with CpG-ODNs than in the group immunized with non-adjuvanted DNA vaccine, indicating elicitation of potent Th1-type responses in this group.
With regard to the determinant role of CTLs in the clearance of several viral infections, LDH was evaluated as one of the most important indicators of CTLs activity upon particular recognition of the antigens presented by APC cells. According to the results, the HIV-1 polyepitope DNA vaccine adjuvanted with CpG-ODNs could considerably increase the cytotoxicity responses in comparison with other formulations in the immunized animals .
# Discussion
HIV-1 is a significant challenge in terms of vaccine development. Traditional vaccines like live virus or even attenuated virus are very effective, although they are too risky against HIV-1. Therefore, other alternative vaccination modalities such as recombinant DNA or protein vaccines should be developed [bib_ref] DNA vaccines: A review, Liu [/bib_ref]. In recent years, DNA based vaccine modality has been largely considered due to some brilliant privileges including safety, relative simplicity for construction, production and administration, scalability with high purity and stability [bib_ref] A DNA-based candidate HIV vaccine delivered via in vivo electroporation induces CD4..., Kopycinski [/bib_ref] , and up to now, some researchers have applied naked DNA as a single vaccine modality. Several pieces of evidence show DNA vaccination has successfully protected macaques from chronic viremia (4). It is indicated that an efficacious protective vaccine against HIV-1 must provoke both protective antibodies and strong T cell responses. To this goal, neutralizing epitopes or conserved T cell epitopes have been utilized for HIV-1 polyepitope vaccine design. Taken together, this strategy has been promising in recent years for designing and producing new HIV-1 vaccines [bib_ref] Development of a novel AIDS vaccine: The HIV-1 transactivator of transcription protein..., Cafaro [/bib_ref] [bib_ref] HIV-1 gp120: A target for therapeutics and vaccine design, Cicala [/bib_ref].
Yang et al. employed some conserved epitopes as an HIV-1 DNA vaccine and their results suggested that the candidate vaccine could induce vigorous cellular immune responses and yielded partial protection [bib_ref] A recombinant multi-epitope protein MEP1 elicits efficient long-term immune responses against HIV-1..., Yang [/bib_ref]. Reguzova and colleagues also reported expression of the HIV-1 polyepitope T cell immunogens, which placed together as a DNA vaccine construct, could cause CD8 + and CD4 + T cell responses in the immunized animals [bib_ref] Design and evaluation of optimized artificial HIV-1 poly-T cell-epitope immunogens, Reguzova [/bib_ref]. The results of previous studies also revealed that the combination of genes encoding the important HIV-1 epitopes (from both structural and regulatory viral proteins) induced much more immunogenicity in comparison with the very same structural or regulatory genes when used alone [bib_ref] HIV-1 Gag p24-Nef fusion peptide induces cellular and humoral immune response in..., Mahdavi [/bib_ref]. So far, various regulatory and structural proteins of HIV-1, e.g., Pol, Gag, Vif, Nef, Tat, and Env have been considered to create polyepitope vaccines [bib_ref] Control of HIV-1 replication in vitro by vaccineinduced human CD8+ T cells..., Ahmed [/bib_ref]. These proteins are heavily involved in the life cycle and pathogenesis of the virus viral and are considered the best candidates to produce efficient and safe HIV/ADIS vaccines [bib_ref] Boosting Tat DNA vaccine with Tat protein stimulates strong cellular and humoral..., Alipour [/bib_ref]. According to the literature, the inhibition of these viral vital proteins could noticeably cause the decline of the virus infectivity at different steps of disease [bib_ref] Clustered epitopes within a new poly-epitopic HIV-1 DNA vaccine shows immunogenicity in..., Jafarpour [/bib_ref]. Very effective control of the progression of AIDS is attainable using the HIV-1 Gag protein [bib_ref] A covalent HIV vaccine: is there hope for the future?, Paul [/bib_ref] , and targeting the viral Pol and Tat proteins could greatly decrease the pathogenicity of the virus [bib_ref] Physicochemical studies on the structural stability of the HIV-1 vaccine candidate recombinant..., Falahati [/bib_ref]. Considering the critical role of the HIV-1 Env protein in the virus-lymphocytes (T cells) interactions, the induction of immune responses against this viral protein could therefore affect the HIV-1 neutralization [bib_ref] Clustered epitopes within a new poly-epitopic HIV-1 DNA vaccine shows immunogenicity in..., Jafarpour [/bib_ref].
Different researchers have also considered using effective adjuvants and different vaccine delivery systems in vaccine design. These approaches help significantly improve the efficiency of vaccines via modulation and enhancement of the immune responses and reduction of antigen doses [bib_ref] Combination of cell penetrating peptides and heterologous DNA prime/protein boost strategy enhances..., Kadkhodayan [/bib_ref]. It is indicated that using efficient adjuvants in the HIV-1 candidate vaccine formulation would be a suitable approach to increase the efficacy of vaccines, and it elicited robust immune responses in vaccinated models [bib_ref] Cloning, expression and purification of a novel multi-epitopic HIV-1 vaccine candidate: A..., Arabi [/bib_ref].
CpG-ODNs exist at high repetition in prokaryotic DNAs but rare in eukaryotic ones that are able to bind to their cognate TLR-9 receptors on surfaces of cells and cause the activation of the innate immune system. The innate immune activation leads to production of Th1 and pro-inflammatory cytokines and chemokines including IL-12, IL-6, IL-1, and TNF-α [bib_ref] A DNA-based candidate HIV vaccine delivered via in vivo electroporation induces CD4..., Kopycinski [/bib_ref] [bib_ref] CpG DNA as a vaccine adjuvant, Bode [/bib_ref] [bib_ref] Vaccine adjuvants: Role and mechanisms of action in vaccine immunogenicity, Marciani [/bib_ref] [bib_ref] Approaches to enhancing immune responses stimulated by CpG oligodeoxynucleotides, Mutwiri [/bib_ref] [bib_ref] The ability of CpG oligonucleotides to protect mice against Francisella tularensis live..., Rees [/bib_ref]. By using CpG-ODNs as adjuvant in a vaccine formulation we could imitate the immune stimulatory activity of bacterial DNA when it infects eukaryotic cells. After internalization by target cells, CpG-ODNs reach the late endosomal/ lysosomal compartment and are released to trigger immune stimulation by signaling cascades involving the up-regulation of NF-ĸB. This human-compatible adjuvant also improves the antigen-presenting function of macrophages, dendritic cells, and monocytes, induces the proliferation of B cells and indirectly stimulates the immunoprotective activity of natural killer cells when it recruits T cells to the site of CpG-ODNs administration [bib_ref] A DNA-based candidate HIV vaccine delivered via in vivo electroporation induces CD4..., Kopycinski [/bib_ref] [bib_ref] CpG DNA as a vaccine adjuvant, Bode [/bib_ref] [bib_ref] Vaccine adjuvants: Role and mechanisms of action in vaccine immunogenicity, Marciani [/bib_ref] [bib_ref] Approaches to enhancing immune responses stimulated by CpG oligodeoxynucleotides, Mutwiri [/bib_ref] [bib_ref] The ability of CpG oligonucleotides to protect mice against Francisella tularensis live..., Rees [/bib_ref]. As a result, CpG-ODNs will be able to act as a potent vaccine adjuvant. Besides that, it could induce faster, longer-lasting, and stronger cellular and humoral immune responses which can be seen following either systemic or mucosal candidate vaccine administration [bib_ref] A DNA-based candidate HIV vaccine delivered via in vivo electroporation induces CD4..., Kopycinski [/bib_ref] [bib_ref] CpG DNA as a vaccine adjuvant, Bode [/bib_ref] [bib_ref] Vaccine adjuvants: Role and mechanisms of action in vaccine immunogenicity, Marciani [/bib_ref] [bib_ref] Approaches to enhancing immune responses stimulated by CpG oligodeoxynucleotides, Mutwiri [/bib_ref] [bib_ref] The ability of CpG oligonucleotides to protect mice against Francisella tularensis live..., Rees [/bib_ref].
A study documented the ability of CpG-ODNs adjuvant in eliciting cellular immune responses and activation adaptive and innate immune responses through induction of plasmacytoid DC and B cells [bib_ref] HIV-1 Tat-based vaccines: an overview and perspectives in the field of HIV/AIDS..., Caputo [/bib_ref]. It could also significantly enhance the number and survival of CD8 + T cells [bib_ref] Induction of NK activity in murine and human cells by CpG motifs..., Ballas [/bib_ref]. CpG-ODNs could reinforce humoral and cellular immunity. A study showed that addition of CpG-ODNs as an adjuvant to the formulation of the HIV-1 Tat-based protein vaccines could enhance humoral and cellular immunity. Another study also indicated that a pcDNA3.1-tat DNA vaccine containing CpG-ODNs caused increased humoral immunity as well as lymphocyte proliferation and IFN-γ cytokine release in the mice model [bib_ref] Subcutaneous administration CpG-ODNs acts as a potent adjuvant for an HIV-1-tat-based vaccine..., Panahi [/bib_ref]. Results of another study suggest that the DNA vaccine adjuvanted with CpG-ODNs induced specific IgG 1 , IgG 2a , IFN-γ, and CTL responses remarkably. By contrast, the very same DNA vaccine without adjuvant was unable to elicit significant immune responses [bib_ref] Adjuvant effect of multi-CpG motifs on an HIV-1 DNA vaccine, Kojima [/bib_ref]. Given all studies mentioned above, we sought to examine if CpG-ODNs adjuvant could improve the immunogenicity of the HIV-1 polyepitope DNA vaccine candidate.
As mentioned before, induction of humoral responses and production of neutralizing antibodies, in particular in portal routes of HIV-1 (mucosal and systemic routes), is very important and has to be considered in designing prophylactic vaccines [bib_ref] DNA Vaccines-A Review, Lavanya [/bib_ref] [bib_ref] DNA vaccines: A review, Liu [/bib_ref]. Specific sera antibodies (particularly against gp41 and gp120) have been very efficient in prevention of HIV-1 infection [bib_ref] DNA Vaccines-A Review, Lavanya [/bib_ref] [bib_ref] DNA vaccines: A review, Liu [/bib_ref]. Our results showed that pcDNA3.1-HIV-1 polyepitope adjuvanted with CpG-ODNs could significantly increase IgG 1 and IgG 2a antibodies in the immunized animals compared to the controls .
Cellular immune responses are the next important factor in controlling viral load and prevention of disease progression to AIDS, which should also be considered in a therapeutic vaccine's design against HIV-1 [bib_ref] DNA Vaccines-A Review, Lavanya [/bib_ref] [bib_ref] DNA vaccines: A review, Liu [/bib_ref]. In this regard, we evaluated the induction of lymphocyte proliferative responses, cytokine IFN-γ production, and cytotoxicity activity of CTL as criteria for assessment of the candidate vaccines' potencies in the elicitation of cellular immune responses [bib_ref] HIV-1 Gag p24-Nef fusion peptide induces cellular and humoral immune response in..., Mahdavi [/bib_ref]. This finding revealed that the HIV-1 polyepitope DNA based candidate vaccine could significantly strengthen lymphocyte proliferative responses, cytokine secretion, and cytotoxicity activity of CTLs in comparison with the control groups [fig_ref] Figure 3: Proliferation of lymphocytes in the vaccinated mice [/fig_ref].
# Conclusion
Taken together, the results of our study revealed that the HIV-1 polyepitope DNA candidate vaccine formulated with CpG-ODNs could significantly elicit immune responses in the vaccinated animals. It seems that the combination of some effective approaches in one immunization regime (using polytopic DNA vaccine along with applying human-compatible CpG-ODNs adjuvant and selection of a suitable route for vaccine administration) could boost the vaccine efficacy and activate simultaneously both arms of the immune system in immunized mice. Although this candidate vaccine still needs more investigation for its immunogenicity, efficacy profile, and safety, and further studies should be held to evaluate its potency, the findings of the present study show the efficacy of this HIV-1 polyepitopic candidate vaccine adjuvanted with CpG-ODNs and this approach could be considered a new approach in HIV-1 vaccine design.
[fig] Figure 1, Figure 2: (a) Analysis of the extracted pcDNA3.1-HIV-1-polyepitope vector on 1% agarose gel. Lane 1, the pcDNA3.1-HIV-1-polyepitope vector that was digested by BamHI and XhoI. The 600 bp bond shows the HIV-1-polyepitope fragment; lane M, molecular size marker. (b) Western blotting of the pure recombinant protein; lane M, molecular size marker; lane 1, the pure HIV-1-polyepitope recombinant protein expressed in the eukaryotic host cell (HEK 293) Examination of the humoral immunity of HIV-1 polytopic adjuvanted and non-adjuvanted DNA vaccines in mice. The test was carried out 1 week after the final administration and the titers of (a) IgG1 and (b) IgG2a were measured by an indirect ELISA method N.S.: Not Significant a induction of lymphocyte proliferative responses and the amounts of SI values were calculated. Data showed that vaccination of animals with HIV-1 polyepitope DNA vaccines (with and without the CpG-ODNs adjuvant) could elicit proliferation of lymphocytes in the immunized mice in comparison with the control ones (Figure 3) and the difference was remarkable (P<0.05). [/fig]
[fig] Figure 3: Proliferation of lymphocytes in the vaccinated mice. Administration of the candidate DNA vaccines was intramuscularly performed 3 times at 2-week intervals. Spleen lymphocytes were extracted from individual mice 14 days after the third and final immunization, cultured and stimulated with the HIV-1-polyepitope protein. The subsequent proliferation responses were analyzed and represented as the stimulation index (SI) values. The data are from triplicate experiments and expressed as the mean of SI values±SD of (P<0.05) [/fig]
[fig] Figure 4, Figure 5: Assessment of cellular responses to the candidate DNA vaccines. Production of IFN-γ in the splenocytes isolated from the animals vaccinated with the polyepitopic DNA vaccines (in the presence and absence of CpG-ODNs) was evaluated using the ELISA method (P<0.05) Induction of cytotoxicity responses against the HIV-1polyepitope DNA vaccine was evaluated by lactate dehydrogenase assay. Data was indicated by the measurement of the amount of the LDH release as a result of the specific cytolysis of the cells that present the desired epitopes. Triton X-100 cell lysate was used as a positive control (P<0.05) [/fig]
[table] Table 1: The immunization protocol for the HIV-1 DNA vaccine candidates [/table]
|
10.3389/fphar.2021.699193
|
CCBY
|
8296636
|
34305607
|
s2orc_pubmed_articles
|
Recent Progress in Environmental Toxins-Induced Cardiotoxicity and Protective Potential of Natural Products
Humans are unconsciously exposed to environmental toxins including heavy metals as well as various pesticides, which have deleterious effects on human health. Accumulating studies pointed out that exposure to environmental toxins was associated with various cardiopathologic effects. This review summarizes the main mechanisms of cardiotoxicity induced by environmental toxins (cadmium, arsenic and pesticides) and discusses the potential preventive effects of natural products. These findings will provide a theoretical basis and novel agents for the prevention and treatment of environmental toxins-induced cardiotoxicity. Furthermore, the limitations of current studies, future needs and priorities are discussed.
# Introduction
The term "toxicity" refers to the relative ability of exogenous chemical substances to cause direct or indirect damage after contacting with living organisms or entering living organisms. Toxicity is closely related to dose, route of exposure, and duration of exposure. Accumulating evidence demonstrated the toxicity of drugs or other environmental factors to organs and systems. In mammals, aminoglycoside compounds are selectively toxic to renal proximal tubule cells and sensory hair cells of inner ear, resulting in nephrotoxicity and ototoxicity. In a recent prospective study, it was found that approximately one out of 2,300 amoxicillin-clavulanic acid users experienced liver damage. The heavy metal lead, as a neurotoxin, has an adverse effect on human cognition and behavior. Although the whole organs in the human body are vulnerable, heart as a vital organ is noteworthy.
Heart disorders such as arrhythmias and ischemic heart disease, are one of the leading causes of morbidity and mortality in the world. In addition to pathological factors, drug factors and environmental factors can also lead to heart disease. Cardiomyocytes are considered to be the main target of doxorubicin cardiotoxicity, and the clinical application of doxorubicin is limited by cumulative and dose-related cardiotoxicity, which may lead to congestive heart failure. In the past few decades, evidence on the role of exposure to environmental toxins in cardiac damage has increased rapidly.
Environmental toxins, including heavy metals, pesticides, petroleum by-products and other toxic chemicals, are not easily degraded and have a long residual period. Moreover, they can be enriched in the human body through bioconcentration and food chain, thereby causing deleterious health effects. Heavy metals refer to inorganic elements with a density greater than 5 g/cm 3 which can be classified into two groups based on their toxicity: essential and non-essential heavy metal. 1) Essential heavy metals are harmless or relatively less harmless at low concentration . 2) Non-essential metals are highly toxic even at low concentration (Cd, Hg, As, and Cr). Metalloids such as arsenic usually fall into the category of heavy metals due to its similar chemical properties and environmental behavior. Heavy metal contamination in water and soil has rapidly increased during the last few decades due to fossil fuel combustion, urban waste disposal, mining and smelting, and the application of fertilizers and pesticides. Recent studies have provided convincing evidence that exposure to heavy metals is associated with increased risks of diabetes and hypertension, which are strong risk factors for cardiovascular diseases. Exposure to arsenic is associated with various cardiopathologic effects. For example, arsenic exposure may cause arrhythmia by prolonging the QT interval and accelerating intracellular calcium overload. As a toxic heavy metal, cadmium can damage the cardiovascular system, leading to heart disorders such as myocardial infarction, cardiomyopathy, and heart failure.
The post World War II era witnessed a great change in agriculture practices, which led to the emergence of mechanization and the development of various chemical pesticides. Pesticides are often used to keep crops healthy and free from disease and pests. Moreover, pesticides have a wider range of uses, such as acaricides and rodenticides for non-plant use to control disease vectors. The main chemical pesticides include organochlorines, organophosphates and carbamates. With the expansion of agricultural production scale, the application of pesticides has shown a sharp upward trend. The use of pesticides has indubitably increased food production and eased the problem of famine to a certain extent, but the threat of pesticides to life and health has gradually emerged and cannot be ignored. Rare arrhythmias can appear later after organophosphorus poisoning, even if the state of illness improves markedly, which may be caused by cardiotoxic effects, metabolic acidosis and electrolyte disturbances. The cardiotoxic effect of aluminum phosphide leads to oxidative stress, deplete myocardial energy ATP and induce apoptosis in animals. Paraquat poisoning induces alterations in myocardial function and left ventricular geometry, such as enlarged left ventricular end-systolic diameter.
Natural products with multiple biological activities are becoming a significant beginning of novel agents and own various pharmaceutical development potentials. This review summarizes current understanding of environmental toxinsinduced cardiotoxicity and discusses the potential preventive effects of natural plants.
## Mechanisms of environmental toxins-induced cardiotoxicity
## Oxidative stress
Oxidative stress is a typical phenomenon of many cardiovascular abnormalities, as well as the most widely studied and recognized mechanism of iatrogenic cardiotoxicity of environmental toxins. The expression and coordinated induction of antioxidant enzymes are mediated by the antioxidant response element, which is a key mechanism to resist chemically induced oxidative/electrophilic stress. Nuclear factor erythroid-2 related factor 2 (Nrf2) is an important transcription factor that regulates cellular oxidative stress, which binds to antioxidant response element sites and regulates antioxidant-mediated gene expression and induction. Endogenous antioxidants including superoxide oxide dismutase (SOD), catalase (CAT) and glutathi-one peroxidase (GPx) have been identified as Nrf2-regulated antioxidant enzymes and constitute a line of defense against reactive oxygen species (ROS), which play a vital role in maintaining redox balance and protecting cells from oxidative damage. Once the production of ROS and the internal antioxidant defense system is out of balance, cells will experience oxidative stress. A study conducted bydemonstrated that the heart, compared to liver, has a much lower concentration of endogenous antioxidants, which indicates that the heart may be more sensitive to peroxidation damage due to its limited antioxidant capacity in scavenging oxygen free radicals.
Early studies have shown that heavy metal Cd promotes the release of iron in biomembrane and this free iron participates in Fenton-type reactions, thus generating ROS. Another study reported that Cd may competitively bind to the q 0 site of cytochrome b of mitochondrial complex III, leading to the accumulation of semiubiquinone at the q 0 site, generating ROS and causing oxidative stress . Moreover, as a thiol affinity metal, free Cd mainly targets intracellular glutathione (GSH)-a reactive oxygen scavenger. The depletion of the GSH pool will lead to insufficient clearance of Cd, which triggers the disorder of the cell redox balance . The same conclusion can also be obtained in the study of the cardiotoxicity mechanism of arsenic and other heavy metals. It is speculated that the overproduction of mitochondrial reactive oxygen species (mtROS) may lead to the occurrence of various cardiovascular diseases. Arsenic-induced overproduction of mtROS destroys myocardial mitochondrial complex I-IV, thereby down-regulating the expression of mitochondrial respiratory chain complex. Earlier, it was also believed that the increased production of mtROS caused mitochondrial swelling and abnormal mitochondrial membrane potential.
Just like many drugs, lithium may directly or indirectly induce oxidative stress through mitochondria via the redox cycling or by promoting iron accumulation and oxidation/nitrification modification of essential mitochondrial proteins; in this way, it plays a key role in the development of myocardial dysfunction.
Fine particulate matter (PM2.5) can produce a large number of ROS and reactive nitrogen species (RNS), caused the unbalance of cell homeostasis and then excessive ROS may result in damage of nuclear DNA and mitochondria including mitochondrial permeabilization, membrane potential decreasing and mitochondrial swelling, followed by further apoptosis in cardiomyocytes.
It is reported that the redox state of experimental animals exposed to organophosphate is generally abnormal. Georgiadis and his colleagues pointed out that oxidative stress is the main mode of action of organophosphates, leading to side effects on myocardial tissue. The fact that longterm exposure to organophosphorus pesticides inhibit the activity of cholinesterase, increased the production of ROS, which disrupts the balance of antioxidant enzymes, thereby leading to oxidative stress. As for other pesticides, most of them will experience the same process. Phosphorus aluminum can inhibit antioxidant enzymes and iron release from transferrin, causing Fenton's and Haber-Weiss reactions to produce iron-catalyzed ROS, thereby inducing oxidative stress. Paraquat can trigger a continuous redox cycle reaction, producing a large amount of ROS, leading to oxidative stress and impairing myocardial contractility, ultimately causing heart failure. Chlorpyrifos can cause myocardial tissue disorders, myocardial fiber degeneration and connective tissue edema. It is speculated that these changes may be due to increased secretion of mtROS and oxidative stress in heart tissue, which lead to decreased mitochondrial energy, decreased secretion of proteolytic enzymes, and DNA disintegration to induce apoptosis.
## Changes in cardiac ion channels and ion disorders
Cardiomyocytes are rich in sodium channels, and pyrethroids are believed to be able to convert the voltage-dependent activation and inactivation of sodium channels into hyperpolarization potentials. Type I pyrethroid, tefluthrin, and type II pyrethroid, fenpropathrin and α-cypermethrin, alter the time course of sodium channel current by changing the relative ratio of fast and slow inactivation current, modify the voltage dependence of I (Na), prolong the ventricular action potential and cause postdepolarization, indicating an arrhythmogenic activity. Ion content plays a role in many physiological processes. Proper ion concentration is essential to ensure the normal function of the entire body, especially the heart. Atrazine induced cardiotoxicity via the modulation of cardiac ATPase including Na + -K + -ATPase, Ca 2+ -ATPase, Mg 2+ -ATPase and changes in the transcription of pump subunits, leading to ionic disorder. PM2.5 significantly inhibited mitochondrial Na + -K + -ATPase and Ca 2+ -ATPase activities, implying that PM2.5 causes dysfunction of sodium pump and calcium pump. One of the typical features of NaAsO 2 -induced cardiotoxicity is the accumulation of intracellular calcium. Myocardial L-type calcium channel plays a vital role in maintaining calcium ion balance. The accumulation of intracellular calcium in myocardial tissue may cause various abnormalities, including ventricular arrhythmia and systolic dysfunction.
## Inflammation
Excessive toxic metals impair immune function and the accumulation of immune complexes, and lead to cardiovascular diseases through a series of interrelated processes, such as the uncontrolled release of inflammatory cytokines. Cd poisoning increases the production of TNF-α, IL-2 and IL-6 by stimulating circulating monocytes and tissue macrophages, leading to the synthesis and release of proinflammatory cytokines, which may be attributed to the increase in oxidized low-density lipoprotein (ox-LDL) caused by cadmium. Arsenic may directly induce atherosclerosis by increasing the mRNA transcription of growth factors such as granulocyte-macrophage colony stimulating factor (GMCSF) and TGFα, and inflammatory cytokines such as TNF-α and IL-6and multiple evidence indicate atherosclerosis is closely related to ischemic heart disease.
## Protective effects of natural products against environmental toxins-induced cardiotoxicity
## Cadmium-induced cardiotoxicity
Heavy mental Cadmium (Cd), a ubiquitous environmental pollutant, was listed as the first category human carcinogen by the International Agency for Research on Cancer (IARC) in 1993. Human exposure to Cd can be through many sources, including eating food contaminated by pesticides and fertilizers, smoking cigarettes, and working in cadmium-contaminated workplace, with smoking being a major risk factor. Like other toxic heavy metals, Cd is greatly concentrated in the upper levels of food chain . Its extremely long biological half-life (about 20-30 years) and low rate of excretion from the body lead to Cd continuous bioaccumulation in soft tissues, which inevitably induces acute and chronic tissue damage and does great harm to important organs, such as liverand kidney, even our heart, depending on the time and dose. It has been long realized that the toxicity induced by Cd triggers biochemical and physiological changes in the heart. A growing number of data has shown that excessive exposure to Cd is closely related to cardiovascular disease, including cardiac death and myocardial infarction, according to the epidemiological studies.
It is universally acknowledged that oxidative stress is the major event induced by Cd poisoning. Cd exerts toxic effects via the generation of free radicals and subsequent ROS accumulation, causing tissue damage to a great extent. Given the correlation between Cd exposure and oxidative stress, attention turned to focus on natural products rich in antioxidants in order to inhibit the cardiotoxicity mediated by Cd. Documented probable cadmium exposure routes and mechanisms of natural products against Cdinduced cardiotoxicity are presented in .
## Natural products
Accumulating studies has proved the beneficial effects of natural products for its protection against Cd induced toxicity. Quercetin is known as robust antioxidant and free radical scavenger against oxidative stress. A study conducted by Milton Prabu et al. evaluated the effects of quercetin on cardiac marker enzymes, lipid peroxidation products, lipid profile, membrane bound ATPases and antioxidant status in Cd-intoxicated rats, which demonstrated that quercetin pretreatment at a dose of 50 mg/kg body mass (bm) per day could prevent oxidative damage to the rat hearts when supplemented with Cd (5 mg/kg bm/day) for 28 days (Milton.
Grape seed proanthocyanidins, effective bioactive components extracted from natural grape seeds, have drawn a great slew of attention due to their extensive biological and pharmacological properties. In Cdadministered (CdCl 2 , 5 mg/kg bm, 4 weeks) male Wistar rats, grape seed proanthocyanidins (100 mg/kg bm, p. o.) was found to own cardioprotective effects by functioning as in-vivo antioxidant and was capable of inhibiting the membrane disturbances in cardiomyocytes, apoptotic pathway and inflammation. Besides, the signals from phenolic antioxidants present in grape seed proanthocyanidins lead to phosphorylation of Nrf2 and/or redox modulation of Nrf2/Keap1, leading to separation of Nrf2 from Nrf2/Keap1, thereby restoring Nrf2 expression in the heart of rats. Therefore, grape seed proanthocyanidins recuperated the Cd-induced oxidative stress mediated cardiac dysfunction.
## Plant extracts
Tinospora cordifolia (Willd.) Miers ex Hook. f. & Thomson, belonging to the family Menispermaceae Juss., is a medicinal herb used in Ayurveda for treating varied metabolic disorders and toxic conditions. Antioxidant and cardioprotective activity of Tinospora cordifolia (Willd.) Miers ex Hook. f. & Thomson against streptozotocin induced diabetic rats has been reported. Moreover, its stem methanolic extract for male albino Wistar rats can significantly attenuate Cdinduced lipid peroxidation and protein carbonylation, reduced heart histological changes and decreased the activities of membrane bound ATPases at dose of 5 mg Cd/kg bm.
Allium cepa L. (Onion), one of the most widely and commonly consumed vegetables in the genus Allium L., shows considerable antioxidant value. Onion consumption correlates with low rates of coronary heart disease. Several studies have demonstrated cardioprotective effects of A. cepa extract by lowering serum cholesterol and blood pressure. In Cdadministered (1 mg/kg bm, s. c.) male Sprague-Dawley rats, A. cepa extract effectively attenuated Cd-induced histological alterations, apoptosis in the cardiomyocytes and decreased levels of the enzymatic antioxidants FIGURE 1 | The cadmium exposure routes and protective mechanisms of natural products against cadmium-induced cardiotoxicity. CK-MB, creatine kinase-MB; LDH, Lactic dehydrogenase; SOD, Superoxide dismutase; CAT, catalase; GPx, Glutathione peroxidase; GST, Glutathione S-Transferase; GSH, glutathione; NF-lB, nuclear factor-lB; NO, nitric oxide; TNF-α, tumor necrosis factor-α; Nrf2, NF-E2-related factor 2; HO-1, heme oxygenase-1; MDA, Malondialdehyde.
Frontiers in Pharmacology | www.frontiersin.org July 2021 | Volume 12 | Article 699193
probably via its antioxidant and anti-apoptotic activity. Summarized above the mentioned, although the natural products supplementation has a protective effect on Cdinduced oxidative heart damage in rats, the exact molecular mechanism has not been elucidated. Moreover, with the exception of in-vivo studies, the protective activity of these natural products in Cd-induced cardiotoxicity should be verified through in-vitro studies.
## Arsenic-induced cardiotoxicity
Arsenic is naturally occurring metalloid element that abundant in Earth's crust and biosphere. Due to its high water-solubility, the main source of human exposure to arsenic is drinking water contaminated with arsenic. In addition, the use of arsenic-containing herbicides, pesticides, and burning fossil fuels can cause a high risk of arsenic poisoning. Once absorbed, arsenic is redistributed to almost the entire body organ system, including the heart. Arsenic exposure gives rise to myocardial damage, arrhythmiaand left ventricular hypertrophy, which correlates with pathogenesis of myocardial tissue. The possible mechanisms of arsenic-induced cardiotoxicity mainly include oxidative stress, DNA fragmentation, changes in cardiac ion channels. At present, there are various drugs for the treatment of the above-mentioned arsenic-induced heart diseases, such as commonly used angiotensin-converting enzyme inhibitors and β-blockers. However, the non-negligible side effects limit their clinical application. In recent years, more and more attention has been focused on the chemical components extracted from natural products due to their broad pharmacological activity and low toxicity. Scientists hope to find inspiration from it and explore new methods for the treatment of arsenic-induced cardiotoxicity. As is depicted in , the possible arsenic exposure routes and mechanism of natural products against cadmium-induced cardiotoxicity have been summarized.
## Natural products
Arjunolic acid, one of the constituents existed in the bark of Terminalia arjuna (Roxb. ex DC.) Wight & Arn., has been shown to provide significant cardioprotection in isoproterenol-induced myocardial necrosis. mentioned that arsenic poisoning (10 mg/kg bm, p. o., 2 days) can cause severe oxidative damage to the heart tissue in mice, which can be prevented by treating with arjunolic acid (20 mg/kg bm, 4 days). The overall preventive effect of arjunolic acid may be due to its scavenging free radicals and metal chelating properties, FIGURE 2 | The arsenic exposure routes and protective mechanisms of natural products against arsenic-induced cardiotoxicity. AST, Aspartate transaminase; ALT, alanine aminotransferase; ROS, reactive oxygen species; IL-β, interleukin-β; eNOS, endothelial NOS; NOX2, NADPH oxidase two; NOX4, NADPH oxidase four; TSH, total sulfhydryl; NPSH, non-protein sulfhydryl; AMPK, adenosine monophosphate-activated protein kinase; GR, glutathione reductase.
Frontiers in Pharmacology | www.frontiersin.org July 2021 | Volume 12 | Article 699193 5 thereby reducing the arsenic load in the cells. Another research revealed that arjunolic acid could regulate oxidative phosphorylation in the mitochondria and subsequently inhibit ROS generation through abrogated p47phox-Ser345 phos phorylation in stimulated and myocardial infarction neutrophils, which is the first report showing the protective effect of arjunolic acid on p47phox phosphorylation and mitochondrial bioenergetics. Moreover, arjunolic acid as a Peroxisome Proliferator-Activated Receptorα (PPARα) agonist up-regulates PPARα, leading to repression of TGF-β signaling, especially by inhibiting TGF-β activated kinase1phosphorylation, which reduces the activity of p38MAPK and NFκBp65, ultimately reversing hypertrophy-related myocardial fibrosis. These results provide new insights for in-depth exploration of the protective mechanism of arjunolic acid against arsenic-induced cardiotoxicity.
(-)-Epigallocatechin-3-gallate, the most abundant and active compound in green tea, possesses a potent antioxidant capacity and exhibits versatile pharmacological activities. Treatment with (-)-epigallocatechin-3-gallate (50 mg/kg bm, i. g., 30 days) significantly attenuated myocardial injury, suppressed oxidative damage and myocardial apoptosis induced by arsenic in rats (50 mg/kg bm, 30 days) and H9c2 cells (1 µM). This study proved that (-)-epigallocatechin-3-gallate plays a protective role by reducing the production of ROS, maintaining the intracellular calcium ion concentration, and reducing activation of caspase-3. The protective effect of (-)-epigallocatechin-3gallate on arsenic-induced cardiotoxicity was evaluated in vivo and in vitro, and the mechanism may be attributable to its potent antioxidant capacity.
The plant of Madhuca indica J. F. Gmel., belonging to the family Sapotaceae Juss., commonly known as Mahua, possesses a robust inhibitory effect on oxidative stress and inflammation. Its bioactive component extraction-quercetin has been proven to own the potential to fight against cardiotoxicity induced by isoproterenol and ischemia/reperfusion injury. found that administration of quercetin (10 and 20 mg/kg bm, p. o.) showed a significant protective effect against arsenic-induced (5 mg/kg bm, 28 days) oxido-nitrosative stress and myocardial injury by regulating Nrf2, PPAR-γ, and apoptosis (c-fos and c-jun).
Silibinin is a polyphenolic flavonoid extracted from the seeds of milk thistle, possessing strong antioxidant and cardioprotective activities. It has been successfully employed as a protective agent against arsenic-induced in vivo model of hepatotoxicity. Pre-administration of silibinin (75 mg/kg bm, 28 days) in arsenic-intoxicated (5 mg/kg bm, p. o., 28 days) rats shows potent protective efficacy against oxidative stress and cardiac injury. The treatment of silibinin facilitates the restoration of antioxidant status and normal histological architecture of cardiac tissue by inhibiting the induction of prooxidants (e.g., NOX2 and NOX4) and enhancing antioxidants. p-Coumaric acid is a naturally occurring hydroxycinnamic acid derivative which is widely found in fruits and vegetables as a dietary polyphenol. According to, p-coumaric acid has a protective effect similar to the standard antioxidant vitamin C on the rat heart damage caused by the pretreatment of sodium arsenite (5 mg/kg bm, 30 days) with respect to oxidative stress and cardiac tissue markers due to its antioxidant properties. Subsequent experiments detected changes in the mRNA expression profiles of inflammatory cytokines, transcription factors, MAP kinases and apoptotic proteins in myocardial tissue. The results show that daily oral p-coumaric acid (75 and 100 mg/kg bm, 30 days) before sodium arsenate exposure can regulate the changes in the above-mentioned mRNA expression profile. However, further studies should be performed to identify the exact mechanism on the protective effects of p-coumaric acid against arsenic toxicity.
Oleic acid, a natural triterpene compound mainly found in olive oil, has been shown to have therapeutic potential for improving pathological conditions including CVDs (Kris-Etherton, 1999). Samanta et al. pointed out that arsenic can cause cardiac hypertrophy in both mice and rat H9c2 cardiomyocytes. Interestingly, they observed that oleic acid (100 µM) reduced the expression of NFATc3 by activating AMPK and increasing the localization of FoxO, thereby helping to relieve cardiac hypertrophy induced by arsenic (1 µM) in H9c2 cells. Oleic acid (12.5 mg/kg bm, 14 days) has also been observed to help improve cardiac hypertrophy in arsenic-exposed mice (arsenic trioxide, 4 mg/kg bm, i. p., at every alternate 2 days for 6 weeks). AMPK-FoxO1-NFATc3 pathway acting in arsenic-mediated cardiac hypertrophy is a novel finding, which contributes to find out novel avenues to treat the condition.
Ellagic acid, a dimeric derivative of gallic acid, is a natural polyphenol component widely found in various fruits and nuts. Ellagic acid treatment (30 mg/kg bm, 14 days) could reduce oxidative stress by reducing the levels of MDA and NO and enhancing the activity of endogenous antioxidant enzymes such as CAT, SOD, GPx, thereby effectively alleviating the cardiotoxicity induced by arsenic (10 mg/kg bm, p. o., 21 days). However, further research is needed to accurately understand the underlying cellular mechanisms.
Naringin is a flavanone glycoside with antioxidant, antiinflammatory and anti-apoptotic efficacy, mainly found in citrus fruits, especially in the peels of tangerines, sweet oranges and lemons. Naringin (40 and 80 mg/kg, p. o., 28 days) can reverse the significant changes in electrocardiogram, hemodynamics and left ventricular contractile function caused by chronic administration of sodium arsenite (5 mg/kg bm, p. o., 28 days). It was testified that naringin ameliorates arseniteinduced cardiotoxicity and hyperlipidemia via regulation of TGF-b/Smad-3 and Nrf-2/HO-1 pathways along with a reduction in myocardial apoptosis. Hesperidin, an isomer of naringin, similarly shows a robust protective potential against sodium arsenite-induced (10 mg/kg bm, 15 days) cardiotoxicity by reducing oxidative stress, apoptosis and preventing inflammation.
Frontiers in Pharmacology | www.frontiersin.org July 2021 | Volume 12 | Article 699193 6
## Plant extracts
Trichosanthes dioica Roxb., commonly known as parwal, is mainly cultivated as a vegetable distributed in India. T. dioica root has been reported to possess some pharmacological properties, including anti-inflammatory and antioxidant activity. Bhattacharya et al. indicated that T. dioica root treatment prior to arsenic intoxication (10 mg/kg bm, p. o., 8 days) possessed remarkable alleviative effect against arsenicinduced cardiotoxicity in Wistar albino rats, as evidenced by significant prevention of alterations in body weight, heart weight, and hematological and serum biochemical parameters. In addition, T. dioica root treatment significantly modulated all of the myocardial antioxidative parameters and reduced DNA fragmentation. The experiment also proved that at higher doses, the effect of T. dioica root is comparable to that of ascorbic acid and quercetin. It is speculated that the active ingredient may be cucurbit type triterpene aglycone. The alleviative effect of T. dioica root against arsenic-induced (10 mg/kg, p. o., 8 days) cardiotoxicity is also reflected in aqueous extract of its fruit.
Corchorus olitorius L. belongs to the family Malvaceae Juss. Its leaves and tender stems are rich in several phenolic antioxidative compounds. According to Das et al., oral administration of aqueous extract of C. olitorius leaves prior to NaAsO 2 -intoxication significantly protected cardiac tissue against arsenic-induced (10 mg/kg bm, p. o., 10 days) oxidative impairment and prevented hyperlipidemia and DNA fragmentation. The cardioprotective effect may be due to the presence of substantial quantity of phytophenolics and flavonoids in the extract.
In addition to looking for new agents with the potential to combat environmental toxin-induced cardiotoxicity, the current focus has shifted to determine crucial signaling molecules/ pathways in environmental toxin-induced cardiotoxicity. The up-regulation of pro-apoptotic proteins Bax, caspase-3, c-fos and c-jun and the down-regulation of anti-apoptotic protein TGF-β and Bcl-2 are involved in arsenic-induced cardiomyocyte apoptosis. Studies have demonstrated that arsenic-induced elevated oxidative stress is associated with alteration in the Nrf-2/HO-1 pathway. Besides, Arsenicinduced cardiac hypertrophy is mediated by reducing the expression of AMPK and FoxO1, thereby increasing the expression of NFATc3. Collectively, studying the signaling pathways acting in arsenic-induced cardiotoxicity will help to find novel therapeutic avenues. However, the exact molecular target of natural products remained unclear. Trace its root, lack of knockdown or overexpression methods to verify the aforementioned signaling molecules, and even no gene knockout animal models are used to evaluate the cardioprotective activity of natural products.
## Pesticide-induced cardiotoxicity
Pesticides is one of the agricultural chemicals with the largest consumption and the most abundant categories, which play a vital role in increasing agricultural production and solving human food problems. However, the use of pesticides inevitably becomes an important factor in environmental pollution. Pesticides are inherently toxic to living organisms, so its adverse effects on humans and other organisms are inevitable. A large amount of epidemiological and experimental evidence shows that there is a connection between exposure to pesticides and the incidence of multiple human diseases. Exposure to pesticides has been associated with several cardiovascular complications including electrocardiogram abnormalities, myocardial infarction, functional remodeling, histopathological insults, such as hemorrhage, vacuolization, signs of apoptosis and degeneration. In addition, it also includes increased systemic and cardiac-tissue-specific oxidative stress and DNA alterations in cardiac cells that could lead to functional impairment. Early studies have shown that pesticide chemicals may induce oxidative stress, leading to the production of free radicals and alterations in the antioxidant or oxygen free radical scavenging enzyme system. Growing evidence show that oxidative stress is a main apoptosis stimulating factor in different diseases including cardiovascular diseases and ROS induces apoptosis. Hence, this process may be inhibited by abundant natural antioxidants. A summary of possible pesticides exposure routes and mechanisms of natural products against pesticidesinduced cardiotoxicity is presented in.
## Herbicides-induced cardiotoxicity
Atrazine is an extensively used herbicide of the triazine class. Due to its soil leaching properties, atrazine is easy to be leached by rainwater, irrigation water to deeper soil, or enters rivers and lakes with surface runoff. In addition, atrazine remains in surface water and groundwater for a long time and is difficult to degrade, thus causing serious pollution to the ecological environment. Exposure to atrazine can cause health risks to humans and other animal species. As early as the 1990s, atrazine and its anabolic metabolites (for example, atrazine-decrystallized-desubstituent) have been detected in heart tissue (du Preez and van Vuren, 1992) and cause cardiovascular disease during poisoning intoxication. Curcumin is a polyphenol compound isolated from Curcuma longa L. (turmeric), which has been widely used in the food industry as a common natural pigment for a long time. Curcumin's powerful anti-inflammatory and antioxidant effects, coupled with its ability to improve endothelial function, can reduce the risk of heart disease. concluded that the administration of curcumin (400 mg/kg bm, 21 days) can improve the cardiotoxicity induced by ATR via regulating redox state, mitochondrial function and caspase-3 expression. Lycopene, a carotenoid existed in plant foods, has a higher content in ripe red plant fruits, especially in tomatoes, carrots, watermelons. With robust free radical scavenging and antioxidant ability, lycopene has attracted considerable attention as a potential chemopreventive agent against diseases such as CVD. Atrazine induced cardiotoxicity by regulating the activity of cardiac ATPase and the transcription of its subunits, thereby triggering ion disorders, while supplemented lycopene (5 mg/kg Frontiers in Pharmacology | www.frontiersin.org July 2021 | Volume 12 | Article 699193 bm, 21 days) in mice significantly counteracted atrazine-induced cardiotoxicity by regulating ATPase activity and subunit transcription. Therefore, lycopene shows a significant chemopreventive potential for atrazine-induced cardiotoxicity. Paraquat is a quick-acting contact-killing herbicide which has been widely used due to its good weeding effect and low environmental pollution. However, in the past few decades, there have been many deaths of paraquat, mainly due to accidental or voluntary ingestion. Once ingested orally, paraquat has extremely strong toxic effects on multiple organs, causing lung, heart, liver and kidney failure and even death. Because no effective treatment has been developed so far, the paraquat poisoning incident is extremely distressing but powerless. Medicinal plants provide abundant resources for screening new therapeutic agents. Tanshinone IIA, one of the active components of Danshen (Salvia miltiorrhiza Bunge), has a wide range of powerful pharmacological effects, including antioxidant and anti-inflammatory effects. As a clinical formulation of Tanshinone IIA, sodium tanshinone IIA sulfate is commonly used to improve myocardial blood and oxygen supply in clinical practice. According to Zhang et al., sodium tanshinone IIA sulfate therapeutically inhibits paraquat-induced myocardial cell apoptosis in rats via the enhancement of Bcl-2, the inhibition of Bax and modulating the Nrf2 pathway .
## Insecticide-induced cardiotoxicity
## Organophosphates-induced cardiotoxicity
Approximately 40% of all pesticides commercially produced and used are organophosphates, which are widely used in the agricultural field. Organophosphates induced toxicity is mainly via inhibition of carboxyl ester hydrolases, particularly acetylcholinesterase, and it is reported that the most significant side effect of this mode of action is the oxidative stress induced in the myocardial tissue.
Malathion is a low-toxic organophosphorus pesticide.conducted cell proliferation studies in malathiontreated cells, using various natural and synthetic antioxidants, including butylated hydroxyltoluene, trolox, quercetin, (−)-epicatechin, ascorbic acid, curcumin, and gallic acid. Among all the antioxidants mentioned above, gallic acid showed the most significant protection against oxidative stress. Syzygium cumini (L.) Skeels is a traditional medicinal plant with various bioactive compounds distributed in all parts of the plant. S. cumini methanolic pulp extract, rich in gallic acid, was taken to study its effect on malathion-induced toxicity. The result showed that with the treatment with S. cumini Frontiers in Pharmacology | www.frontiersin.org July 2021 | Volume 12 | Article 699193 8 methanolic pulp extract, malathion-induced morphological changes, ROS levels and nuclear deformities have decreased. Increment in collagen content could also be clearly observed. Hence, S. cumini methanolic pulp extract could ameliorate the oxidative stress in H9c2 cells caused by malathion, which indicates that it has antioxidant properties and protective effects on malathion-induced cardiotoxicity.
Chlorpyrifos is a moderately toxic organophosphate insecticide, which was once banned from being used on vegetables due to excessive residues. The acute toxicity of chlorpyrifos mostly affects the nervous system and cardiovascular system. It is said that chlorpyrifos poisoning induces heart tissue to produce abundant free radicals, making it particularly vulnerable to oxidative stress and peroxidative damage. As a traditional Chinese medicine, saffron (Crocus sativus L.) has been used for centuries around the world, known for its extensive pharmacological activities including antioxidant. The efficacy of saffron is mainly attributed to the presence of crocin, which can effectively prevent cardiotoxicity, hepatotoxicity and DNA damage. Studies have shown that crocin can improve the histopathological changes caused by doxorubicin, and attenuated the cardiotoxicity induced by diazinon. proved that the cardiotoxicity of chlorpyrifos is mainly due to oxidative stress and crocin can alleviate this toxic effect by its antioxidant property. Propolis, a colloidal natural product with a variety of pharmacological activities, has been widely used in medicine, health food, cosmetics and other fields. Propolis contains a lot of flavonoids and polyphenols which correlates with its physiological activity, such as antiinflammatory, antioxidant and immunomodulatory activities. A recent study has shown that propolis intake may reduce the risk of cardiovascular diseases caused by chlorpyrifos exposure by the improvement of PON1 and XO mRNA genes regulation leading to the accumulation of the cellular enzymatic and/or non-enzymatic antioxidants. Pomegranate (Punica granatum L.) is a common fruit rich in polyphenols such as ellagic acid and ellagitannin. Administration of either its juice or peel methanol extract has shown protective potential against chlorpyrifos-induced cardiotoxicity potentially via its antioxidant, antiapoptotic and membrane stabilizing properties.
Diazinon is a moderately toxic, novel, broad-spectrum organophosphorus pesticide. Subchronic exposure to diazinon can induce mitochondrial-mediated apoptosis in rat cardiac tissue. The above-mentioned crocin can also play a protective role in diazinon-induced cardiotoxicity by reducing lipid peroxidation and alleviating apoptosis. Nigella sativa L. is a member of family Ranunculaceae Juss. Its seeds, the main source of the active ingredients of the plant, have long been used as a treatment for various pathological diseases in the Middle East and Far East. As a strong antioxidant and anti-inflammatory agent, thymoquinone has proved to be the most bioactive ingredient in N. sativa seeds.
Sub-acute exposure to diazinon significantly induces oxidative damage and inhibits the antioxidant defense system. Thymoquinone supplement could inhibit diazinon-induced cardiotoxicity and improved cholinesterase activity in rats via the mechanism of free radical scavenging.
## Other insecticide-induced cardiotoxicity
Aluminum phosphide (ALP) is one of the most widely used metal phosphides, often used as a solid fumigant to exterminate pests. Once ALP is exposed to water or acid, it will release highly toxic phosphine gas. Hence, whether it is taken orally or inhaled phosphine gas, it may cause severe poisoning or even death. Due to the extreme toxicity of ALP, more than 70% of people exposed die from its detrimental effects on various organs of the body and their main cause is cardiotoxicity. ALP-induced cardiovascular disturbances include refractory hypotension, dysrhythmia, and congestive heart failure. The primary mechanism of aluminum phosphide poisoning is the inhibitory effect of phosphine on cytochrome C oxidase. Other mechanisms refer to oxidative stress, calcium and magnesium complexation, and damage to the electron transport chain. Increasing studies have shown that the active compounds found in plants are effective against chemical and druginduced cardiotoxicity. Chrysin is a flavonoid compound with a wide range of pharmacological activities. Recently, a study in animal model showed the protective effect of chrysin against doxorubicin-induced cardiomyopathy. The protective effect of chrysin on doxorubicin-induced cardiotoxicity is exerted via the mitochondrial apoptosis pathway and the inhibition of oxidative stress. Similarly, Khezri et al. found that treatment with chrysin significantly reduced the formation of ROS, lysosomal damage, mitochondrial damage and lipid peroxidation induced by ALP. A recent study conducted by Jahedsani et al. showed that apigenin also has the same protective effect as chrysin against ALP-induced myocardial damage. Since these studies were performed in primary rat cardiomyocytes or isolated mitochondria, further studies are needed to verify the cardioprotective effects by animal trials and clinical trials on humans. Echinophora cinerea (Boiss.) Hedge & Lamond is a native Iranian plant, whose erial parts are often used as food seasoning in cheese and yogurt. Research suggests that a flavonoid glucoside extracted from E. cinerea showed cytoprotective effect against oxidative stress induced by hydrogen peroxide in PC12 cells. Haydari et al. demonstrated that administration of the E. cinerea extract can improve bradycardia, hypotension, and conduction disturbances of the rats heart caused by ALP poisoning. It can also increase the level of antioxidant enzymes and protect human body from ALPinduced oxidative damage which shows a dose-dependent characteristic to a certain extent.
Lindane, an artificial chlorinated hydrocarbon pesticide which was first introduced as a scabicide for human use in the 1950s, was once widely used for agricultural and public health pest control. Its persistence in the environment, high toxicity to mammals, and resistance to biodegradation have led many developed and developing countries to ban or restrict its use. As early as 2005, it has reported that lindane could cause oxidative stress, lipid peroxidation and changes in the levels of enzymatic antioxidants in the rat heart. found that oral administration of gallic acid or quercetin protects against lindane-induced myocardial damage, possibly via maintaining the level of endogenous antioxidant enzymes and membrane-bound ATPase activity, and inhibiting lipid peroxidation (Vijaya.
Thiamethoxam, the second-generation nicotinic insecticide, is widely used to protect cotton, sorghum and other crops from being attacked by insects and pests. Due to its slow metabolism in plants and soil and high solubility in water, thiamethoxam poses a major risk to aquatic ecosystems and human health. H9c2 cardiomyoblastes and in vivo using Wistar rat model to confirm the toxic effect of thiamethoxam on the heart for the first time. Fenugreek (Trigonella foenum-graecum L.), an annual herb of the family Fabaceae Lindl., is considered to be one of the oldest medicinal plants. a polysaccharide from fenugreek seeds, called fenugreek seed water polysaccharide. Administration of this polysaccharide to thiamethoxam-treated rats shows a strong protective effect on the oxidative stress of the heart, which is manifested in the significant improvement of enzymatic and non-enzymatic antioxidants and the restoration of histopathological changes in the heart tissue.
Fenugreek seed water polysaccharide owns a great cardioprotective potential, but its specific mechanism needs to be further elucidated.
Throughout these studies mentioned above, we found that most studies on the protective effects of natural products in pesticides-induced cardiotoxicity were carried out in animals. However, in-vitro studies are indispensable for in-depth study of the action mechanism and complex signal networks of natural products. Due to differences in the metabolism and physiology of animals and humans, the animal models used cannot fully simulate the development of human diseases. Therefore, further clinical trials are needed to evaluate effectiveness and safety of natural products.
## Future needs and priorities
People are exposed to environmental toxins in various ways unconsciously. Once the environmental toxins enter the human body, they are difficult to eliminate and the content in the body slowly accumulates over time, eventually reaching the cumulative dose, damaging tissues and even organs. Although few specific environmental toxins directly target the heart, their more or less toxic effects on the heart cannot be ignored. Regrettably, despite the existence of common cardioprotective drugs in the pharmaceutical market, there is a lack of specific targeted treatment. Cardiotoxicity is still a major medical problem. Therefore, there is an urgent need to study and determine the potential molecular mechanisms and signal transduction pathways that antagonize environmental toxins-induced cardiotoxicity. Early studies mostly revealed the important role of oxidative stress, and the corresponding antagonistic method is to remove ROS. As a recent research report, the strong antioxidant resveratrol significantly inhibits the production of ROS, thereby inhibiting oxidative stress, and thus plays a protective effect on PM2.5-induced heart defects in zebrafish embryos. Although this method has been successful in cells and experimental animal models to some extent, the use of ROS scavengers (such as N-Acetyl-Lcysteine and L-carnitine) does not seem to be popular in the clinical practice, which indicated that oxidative stress is not the only cause of cardiotoxicity. Growing studies provided potent evidence that besides oxidative stress, environmental toxinsinduced cardiotoxicity can also be affected by inflammation and changes in cardiac ion channels and ion disorders. Most natural products have multiple "targets" and may affect more than one signaling pathway. The development of multi-site targeted natural products and multi-component traditional Chinese herbal medicine can be used for the treatment of cardiotoxicity induced by environmental toxins.
The occurrence and development of cardiotoxicity are accompanied by the damage and even the death of cardiomyocytes, indicating that cardiomyocytes death may be the main cause of cardiotoxicity. The mechanism of environmental toxins-induced cardiotoxicity is sophisticated, but existing research point out that its ultimate result is cardiomyocyte apoptosis. In addition to apoptosis, a common form of cell death, drug-induced cardiomyocytes death forms also include autophagy, necrosis and other regulated forms, such as necroptosis, pryoptosis, and ferroptosis. Hence, whether there are other forms of cell death in cardiotoxicity induced by environmental toxins remains to be further studied.
As mentioned above, it can be found that the active ingredients in daily food, such as lycopene in tomatoes and ellagic acid in nuts and fruits, have a beneficial effect on the cardiotoxicity induced by environmental toxins, which suggests that people, especially workers who are occupationally exposed to these environmental toxins may be able to reduce the risk of poisoning by consciously increasing their intake of these foods. Epidemiological studies are needed for further confirmed.
Although in pre-clinical studies, natural products have made significant progress in preventing cardiotoxicity caused by environmental toxins, they have not yet been converted into clinical use. The main hindrance to the development of natural products-based cardioprotective adjuvants attributes to the low bioavailability of natural products. Actually, natural products must be supplied in sufficient doses to enter the systemic circulation in the form of their natural structures or metabolites and reach target tissues to exert biological activity. As typical natural products with strong antioxidant capacity, polyphenols have effective protective effects on cardiotoxicity induced by environmental toxins. The antioxidant activity of polyphenols depends on their ability to scavenge free radicals, donate hydrogen atoms or electrons, or chelate metal cations, and the structure is a key determinant of their free radical scavenging and metal chelating activity. However, due to Frontiers in Pharmacology | www.frontiersin.org July 2021 | Volume 12 | Article 699193 low bioavailability and kinetic limitations, the direct antioxidant activity of polyphenols seems to be insufficient in the body. As , the concentrations of polyphenols used in in-vitro studies usually range from μmol/L to mmol/L, while plasma metabolite concentrations rarely exceed nmol/L after normal dietary intake. Although polyphenols have been shown to have indirect antioxidant capacity by regulating gene expression and endogenous antioxidant enzyme defense systems as described in our review, it is limited. Therefore, strategies such as structural modification of natural products and changing dosage forms are needed to improve the bioavailability of natural products. Environmental toxins-induced cardiotoxicity models are mostly established on animals, primary cultured cells or cell lines. A suitable experimental model that simulates the physiology of the human heart is needed. Several natural products have been shown to have cardioprotective effects on environmental toxins-induced cardiotoxicity both in vitro and in vivo, but they are not thorough enough, and mechanisms of action have not been fully elucidated. Moreover, most of the existing research focuses on oxidation, inflammation and apoptosis, and more attention should be paid to the complex signal network, which is the core of cardiomyocyte survival and dysfunction. However, we have to admit that in-vivo and in-vitro studies have certain limitations. One of the disadvantages of in-vitro studies is that the effective dose used is much higher than concentration that can be achieved in humans. In addition, in-vitro studies use prototype compounds of natural products, rarely involving their active metabolites. As for in-vivo studies, due to differences in metabolism, genomics and physiology, the animal models used do not fully mimic the development of human diseases. The results obtained in in-vivo studies need to be transformed into humans. Collectively, a large number of compounds are proven to be effective in vivo and in vitro studies, but most of them cannot be converted into evidence for therapeutic benefits. From the discovery of new agents to the conversion into clinical use, there is still a long way to go to develop natural products to combat cardiotoxicity induced by environmental toxins.
# Conclusion
When evaluating the toxicity of environmental toxins, cardiotoxicity is an important consideration because myocardial damage can be irreversible and fatal. Mechanism of environmental toxins-induced cardiotoxicity is sophisticated and involves multiple targets and pathways, making it difficult to take corresponding intervention measures. This article reviews the mechanism of environmental toxins-induced cardiotoxicity and summarizes the protective effects of natural products on environmental toxins-induced cardiotoxicity. A list of valuable natural products has been mentioned in.
Medicinal plants can be used as a source of useful novel compounds to develop effective treatments to combat cardiotoxicity in patients exposed to environmental toxins.
# Author contributions
YY and SW wrote the manuscript. BZ and WL revised the manuscript. All authors read and approved the final version of the manuscript for publication.
|
10.1016/j.jdcr.2023.09.007
|
CCBY
|
10618619
|
37920705
|
s2orc_pubmed_articles
|
Diffuse purpuric eruption in a patient with acute myeloid leukemia
## Case presentation
A 67-year-old Caucasian female newly diagnosed with acute myeloid leukemia (AML) was initiated on chemotherapy with daunorubicin 60 mg/m 2 (days 1-3) and cytarabine 100 mg/m 2 (days 1-7).The patient developed a progressive, purpuric eruption on day 20 post-treatment initiation.She had diffuse, purpuric macules involving her trunk, chest, extremities, face, scalp, and palate (Fig .Her eruption became more confluent in the days following her initial presentation (Fig .Bloodwork revealed severe pancytopenia with negative autoimmune markers, infectious work-up, and normal complements.Punch biopsy revealed extravasated red blood cells in the superficial dermis without any vasculitis or inflammation (Fig 3
## D.
Progressive papular purpuric eruption e Correct.A frequently used antineoplastic for hematologic malignancies, cytarabine can cause a progressive papular purpuric eruption that develops during or days to weeks after treatment.Unlike other more serious dermatoses induced by cytarabine use, this purpuric eruption shows evidence of hemorrhage and purpura with several possible concomitant features such as spongiosis, acantholysis and dyskeratosis, and/or lymphocytic infiltrate on biopsy.Patients often also have evolution of their eruption with involvement of the skin folds. [bib_ref] Generalized benign cutaneous reaction to cytarabine, Ruben [/bib_ref] [bib_ref] High-dose cytosine arabinoside-induced cutaneous reactions, Cetkovsk A [/bib_ref] Neutrophilic eccrine hidradenitis e Incorrect.While most associated with cytarabine, neutrophilic eccrine hidradenitis presents with sterile neutrophilic infiltrate in the dermis. [bib_ref] Mucocutaneous reactions to chemotherapy, Susser [/bib_ref] [bib_ref] High-dose cytosine arabinoside-induced cutaneous reactions, Cetkovsk A [/bib_ref] [bib_ref] Florid skin rash in acute myeloid leukaemia, Ammannagari [/bib_ref] Question 2: Which of the following risk factors is associated with increased incidence of the above described and other cytarabineassociated cutaneous reactions?Compared to other hematologic malignancies including non-Hodgkin's lymphoma, myelodysplastic syndrome, and acute lymphoblastic leukemia, patients with AML have been noted to have significantly higher rates of cutaneous reactions to cytarabine. [bib_ref] Risk factors for cytarabineinduced cutaneous toxicity in patients with haematological malignancies, Morio [/bib_ref] C. Concurrent steroid use e Incorrect.Patients that receive steroids as part of their chemotherapy regimen or infusion pretreatment have significantly fewer adverse cutaneous reactions including cytarabine-related papular purpuric eruptions. [bib_ref] Risk factors for cytarabineinduced cutaneous toxicity in patients with haematological malignancies, Morio [/bib_ref] D. Lower cytarabine doses e Incorrect. Altho cytarabine associated cutaneous reactions have been observed in patients that are cytarabine-na€ ıve or on low-dose therapy, cytarabine-related papular purpuric eruptions have been most often noted in patients on high dose therapy as indicated for hematologic malignancies like AML. [bib_ref] High-dose cytosine arabinoside-induced cutaneous reactions, Cetkovsk A [/bib_ref] [bib_ref] Risk factors for cytarabineinduced cutaneous toxicity in patients with haematological malignancies, Morio [/bib_ref] E. Female sex e Incorrect.Similar rates of cutaneous toxicity to cytarabine have been observed when comparing male and female patients. [bib_ref] Risk factors for cytarabineinduced cutaneous toxicity in patients with haematological malignancies, Morio [/bib_ref] Question 3: Which of the following treatment strategies is most suitable for this diagnosis?A. Systemic steroids e Incorrect.Unlike cytarabine syndrome and more serious drug reactions like Stevens-Johnson syndrome, cytarabine-associated papular purpuric eruptions do not require systemic steroid therapy if symptoms are well-controlled with topicals agents alone. [bib_ref] Generalized benign cutaneous reaction to cytarabine, Ruben [/bib_ref] B. Symptomatic treatment e Correct.Cytarabineinduced papular purpuric eruptions are self-limited and resolve within 3-20 days after onset with no further sequalae.Treatment is centered around symptomatic management of pruritus and burning with topical corticosteroids, emollients, and antihistamines as needed.An additional short oral steroid taper may be prescribed in cases where patients are severely symptomatic. [bib_ref] Generalized benign cutaneous reaction to cytarabine, Ruben [/bib_ref] [bib_ref] High-dose cytosine arabinoside-induced cutaneous reactions, Cetkovsk A [/bib_ref] C. Chemotherapy cessation e Incorrect. We fueled by an interplay of hypersensitivity, immune-mediated effects, and direct cell toxicity, cytarabine-associated papular purpuric eruptions do not preclude patients from future cytarabine therapy. [bib_ref] Generalized benign cutaneous reaction to cytarabine, Ruben [/bib_ref] Patients can often continue or successfully rechallenge cytarabine without rash reappearance. [bib_ref] High-dose cytosine arabinoside-induced cutaneous reactions, Cetkovsk A [/bib_ref] [bib_ref] Florid skin rash in acute myeloid leukaemia, Ammannagari [/bib_ref] D. Photopheresis e Incorrect.Photopheresis targets aberrant white blood cells which do not directly underlay the pathogenesis of cytarabineassociated papular purpuric eruptions. [bib_ref] Generalized benign cutaneous reaction to cytarabine, Ruben [/bib_ref] [bib_ref] High-dose cytosine arabinoside-induced cutaneous reactions, Cetkovsk A [/bib_ref] E. Doxycycline e Incorrect.With both antiinflammatory and antimicrobial properties, doxycycline has been used to prophylactically prevent skin toxicity across a variety of cancer therapies; however, it has not been indicated for use in cutaneous reactions associated with cytarabine chemotherapy. [bib_ref] Generalized benign cutaneous reaction to cytarabine, Ruben [/bib_ref] [bib_ref] High-dose cytosine arabinoside-induced cutaneous reactions, Cetkovsk A [/bib_ref] [bib_ref] Florid skin rash in acute myeloid leukaemia, Ammannagari [/bib_ref]
[fig] A. 5 B.: Patient age [50 years of age B. Underlying diagnosis of AML C. Concurrent steroid use D. Lower cytarabine doses E. Female sex Answers: A. Patient age [50 years of age e Incorrect.While cytarabine-related papular purpuric eruptions and other toxicities have been observed across a wide range of patient ages, they occur more commonly in patients \50 years of age independent of other characteristics.Underlying diagnosis of AML e Correct. [/fig]
|
10.1186/1479-5868-8-90
|
CCBY
|
3176145
|
21854622
|
s2orc_pubmed_articles
|
Physical activity and sedentary behaviours in Greek-Cypriot children and adolescents: a cross-sectional study
Background: There are no data on physical activity and sedentary behaviours of Greek-Cypriot children and adolescents, and no study to date examined the association between these two behaviours in this population. The purpose of this study was to document the prevalence of physical activity and sedentary behaviours among Greek-Cypriot adolescents and examine the association between physical activity and a range of sedentary behaviours. Logistic regression analyses were performed to examine the association between physical activity and sedentary behaviours. Methods: A cross-sectional study among 1,966 Greek-Cypriot children and adolescents was conducted in 2008/ 2009. Data were collected by means of a questionnaire across primary, middle, high and technical/vocational schools.Results: Overall 52.3% and 52.4% of the participants met physical activity and television viewing guidelines respectively. Boys and younger children were more likely to meet guidelines. Boys who attended sports clubs for two or more times per week were more likely to be physically active (OR = 3.4), and those who listened to music for one or less than one hour per day were less likely to be physically active (OR = 0.6). Girls who attended sports clubs for two or more times per week and who watched television for two or less than two hours per day were more likely to be physically active, (OR = 3.0 and OR = 1.5 respectively). Girls who reported travelling by car/bus/ motorbike for one or less than one hour per day were more likely to actively travel to school (OR = 1.8).Conclusions: Findings from this study provide limited support for the displacement hypothesis whereby sedentary behaviours displace physically active time. About 50.0% of Greek children and adolescents in Cyprus meet existing physical activity and television viewing guidelines. Encouraging children to attend sports clubs for at least two times per week may markedly improve their physical activity levels.
# Background
Participation in physical activity has been found to result in health benefits including improved bone mineral density and improved indices of cardiovascular health such as blood pressure and overweight and obesity among children [bib_ref] Systematic review of the health benefits of physical activity and fitness in..., Janssen [/bib_ref] [bib_ref] Evidence based physical activity for school-age youth, Strong [/bib_ref]. Excessive television watching results in greater levels of overweight and obesity across children in many countries [bib_ref] Relationship of physical activity and television watching with body weight and level..., Andersen [/bib_ref] [bib_ref] Television viewing and its association with overweight in Colombian children: results from..., Gomez [/bib_ref] [bib_ref] Child overweight in France and its relationship with physical activity, sedentary behaviour..., Lioret [/bib_ref]. A recent study also suggested that television watching and computer use are positively associated with aggression and alcohol use [bib_ref] HBSC Physical Activity Focus Group: Interrelationships of adolescent physical activity, screen-based sedentary..., Iannotti [/bib_ref].
Studies that examined the combined effects of television and physical activity on overweight and obesity also indicate that low levels of physical activity and high levels of television watching among younger [bib_ref] BMI from 3-6 y of age is predicted by TV viewing and..., Jago [/bib_ref] and older children [bib_ref] Combined influence of physical activity and television viewing on the risk of..., Eisenmann [/bib_ref] [bib_ref] Combined influence of physical activity and screen time recommendations on childhood overweight, Laurson [/bib_ref] are associated with increased levels of overweight and obesity.
Current guidelines recommend that young people should engage in physical activity for at least 60 minutes of moderate to vigorous intensity per day [bib_ref] Systematic review of the health benefits of physical activity and fitness in..., Janssen [/bib_ref] [bib_ref] Health enhancing physical activity for young people: Statement of the United Kingdom..., Cavill [/bib_ref] and should watch television for no more than 2 hours per day [bib_ref] Sedentariness, small-screen recreation, and fitness in youth, Hardy [/bib_ref]. Nevertheless, findings from a number of studies from different countries suggest that young people do not meet these guidelines. Data from self-reports of children's physical activity suggest that about 42% of [bib_ref] Sedentariness, small-screen recreation, and fitness in youth, Hardy [/bib_ref] [bib_ref] Physical activity levels of Canadian children and youth: current issues and recommendations, Katzmarzyk [/bib_ref] [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] [bib_ref] Physical activity and sedentary behaviors among Finnish youth, Tammelin [/bib_ref] [bib_ref] The prevalence of sedentary behaviour and physical activity in leisure time: A..., Biddle [/bib_ref] [bib_ref] Is spending time in screenbased sedentary behaviors associated with less physical activity:..., Melkevik [/bib_ref] year-olds from Canada [bib_ref] Physical activity levels of Canadian children and youth: current issues and recommendations, Katzmarzyk [/bib_ref] , 35% of 15-18 yearsolds from the U.S.and about one-third of young people from Europe [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] meet the recommendation of at least 60 minutes of moderate to vigorous intensity activity per day. Likewise, 46% of [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] year-olds from Finland [bib_ref] Physical activity and sedentary behaviors among Finnish youth, Tammelin [/bib_ref] and 45% of Scottish adolescents watch more than 2 hours of television per day [bib_ref] The prevalence of sedentary behaviour and physical activity in leisure time: A..., Biddle [/bib_ref]. A recent cross national investigation indicated that the percentage of adolescents from North America and Europe exceeding the recommended amount of television watching per day is 77% [bib_ref] Is spending time in screenbased sedentary behaviors associated with less physical activity:..., Melkevik [/bib_ref].
A number of studies have examined the association between physical activity and television watching. This investigation has been based on the displacement hypothesis that states that engaging in sedentary activities displaces or reduces physical activity [bib_ref] Physical activity and sedentary behaviors among Finnish youth, Tammelin [/bib_ref] [bib_ref] Physical activity and sedentary behaviours in youth: issues and controversies, Biddle [/bib_ref]. A longitudinal study failed to reveal a relationship between year-to-year changes in television viewing and changes in moderate to vigorous physical activity among [bib_ref] Combined influence of physical activity and screen time recommendations on childhood overweight, Laurson [/bib_ref] [bib_ref] Health enhancing physical activity for young people: Statement of the United Kingdom..., Cavill [/bib_ref] [bib_ref] Sedentariness, small-screen recreation, and fitness in youth, Hardy [/bib_ref] [bib_ref] Physical activity levels of Canadian children and youth: current issues and recommendations, Katzmarzyk [/bib_ref] year-olds suggesting that these two behaviours are two separate constructs [bib_ref] Longitudinal relationship between television viewing and leisure-time physical activity during adolescence, Taveras [/bib_ref]. Lack of association between television watching and other sedentary behaviours was also observed in cross-sectional studies assessing weekly frequency of physical activity participation [bib_ref] Children's physical activity, TV watching and obesity in Cyprus: the CYKIDS study, Lazarou [/bib_ref] , and active commuting to school [bib_ref] Social-Ecological correlates of active commuting to school among high school students, Robertson-Wilson [/bib_ref].
On the contrary, Tammelin et al. [bib_ref] Physical activity and sedentary behaviors among Finnish youth, Tammelin [/bib_ref] have found a negative association between television watching/computer use and self-reported physical activity in a sample of 6928, [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] year-old Finnish youth and Hager [bib_ref] Television viewing and physical activity in children, Hager [/bib_ref] observed in a sample of 40 boys aged 9-12 that those who watched television after school were less likely to be active as assessed by accelerometer in comparison to those who did not watch television. Reviewing the evidence on the association between sedentary behaviours and obesity development, Rey-Lopez et al. [bib_ref] Sedentary behaviour and obesity development in children and adolescents, Rey-Lopez [/bib_ref] concluded that it is not known whether sedentary behaviour displaces physical activity, and findings from a metaanalysis indicate that the relationship between physical activity and television watching, playing video games or using computers receives very little empirical support [bib_ref] Relationships between media use, body fatness and physical activity in children and..., Marshall [/bib_ref].
Findings from the above studies suggest that the evidence supporting the association between physical activity and sedentary behaviours are contradictory. It is therefore important to assess both of these behaviours as they may both need to be targeted for physical activity promotion. This is especially important as young peoples' physical activity levels tend to decline as they move through adolescence [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] [bib_ref] Physical activity levels and patterns of 9-and 15-yr-old European children, Riddoch [/bib_ref]. Furthermore, while findings regarding the association between age and screen time behaviours are mixed [bib_ref] Child overweight in France and its relationship with physical activity, sedentary behaviour..., Lioret [/bib_ref] [bib_ref] How do school-day activity patterns differ with age and gender across adolescence?, Olds [/bib_ref] a recent cross national investigation suggests that older adolescents are more likely to be spending more than two hours daily in cumulative screen time [bib_ref] Is spending time in screenbased sedentary behaviors associated with less physical activity:..., Melkevik [/bib_ref].
While data on physical activity among Greek-Cypriot elementary school children exist [bib_ref] Children's physical activity, TV watching and obesity in Cyprus: the CYKIDS study, Lazarou [/bib_ref] [bib_ref] Pedometer-assessed physical (ambulatory) activity in Cypriot children, Loucaides [/bib_ref] , to our knowledge there are no data that examine the physical activity levels and sedentary behaviours of Greek-Cypriot adolescents with reference to physical activity and television watching guidelines across different levels of education. This is especially important as Greek-Cypriot children's levels of overweight and obesity are increasing [bib_ref] Prevalence of overweight and obesity among 11-year-old children in Cyprus, Savva [/bib_ref]. As obesity prevention interventions need to be tailored to the needs of local participants, an understanding of the physical activity and screen-viewing behaviours of Cypriot youth is urgently needed. Further, examining the association between physical activity and sedentary behaviours in a unique population may help enrich existing evidence. A recent study also indicated that television watching has dominated the assessment of sedentary behaviours [bib_ref] The prevalence of sedentary behaviour and physical activity in leisure time: A..., Biddle [/bib_ref] , and there is a need to consider a wide range of sedentary activities when examining the association with physical activity [bib_ref] The prevalence of sedentary behaviour and physical activity in leisure time: A..., Biddle [/bib_ref] [bib_ref] Relationships between media use, body fatness and physical activity in children and..., Marshall [/bib_ref] [bib_ref] Association between sedentary behaviour, physical activity, and obesity: inactivity among active kids, Wong [/bib_ref]. Lastly, as only one study was located that examined the association between active commuting to school and sedentary behaviours [bib_ref] Social-Ecological correlates of active commuting to school among high school students, Robertson-Wilson [/bib_ref] , more data are needed that examine this association. To address these issues, this study examined the association between physical activity (moderate to vigorous and active travelling) and multiple sedentary behaviours in a sample of Greek-Cypriot youth.
Therefore, the purpose of this study was twofold: 1) to document the prevalence of physical activity and sedentary behaviours across different levels of education in Cyprus and 2) to examine the association between physical activity (moderate to vigorous and active travelling) and a range of sedentary behaviours.
# Methods
## Participants
Students from 25 schools from all districts under the control of the Republic of Cyprus were invited to participate in this study including grade six students from nine elementary schools (n = 448), grade 7-9 students from six middle schools (n = 656), grade 10-12 students from five high schools (n = 479) and from five technical schools (n = 383). Technical schools offer vocational rather than academic training. Letters were sent to the head-teachers of each school informing them of the procedures involved. All head-teachers gave their consent, and students from all grade six classes from the elementary schools and randomly selected classes of students from middle, high and technical schools were invited to complete questionnaires while at school. Parental informed consent was obtained by all students who completed questionnaires. The protocol for this study was approved by the Cyprus Pedagogical Institute and by the Cyprus Ministry of Education and Culture.
## Measures
## Physical activity
Physical activity was assessed with four items modified from the Youth Risk Behavior Survey [bib_ref] Statewide prevalence and correlates of walking and bicycling to school, Evenson [/bib_ref]. Two of these items assessed weekly frequency students participated in moderate ('Physical activity that does not make you sweat or breathe hard such as walking, slow bicycling and volleyball') and vigorous ('Physical activity that makes you sweat and breathe hard such as running, playing basketball, playing football and swimming') physical activity respectively. Responses for these items were on an eight-point scale ranging from 'not at all' to 'seven days'. Two further items assessed the usual duration that students participated in moderate and vigorous activities with four response options including 'up to 30 minutes', 'up to one hour', 'up to one and a half hour' and 'more than one and a half hour'. A recent review concluded that the Youth Risk Behavior Survey has good validity including convergent validity with accelerometry [bib_ref] An assessment of self-reported physical activity instruments in young people for population..., Biddle [/bib_ref].
Two other items also assessed physical activity related behaviours. The first asked students to indicate their usual mode of travel to school with four possible responses including bus or car, motorcycle, bicycle and walk. The second item asked students to indicate the weekly frequency they attended a sports club. Responses for this item were on a six-point scale ranging from 'not at all' to 'more than four times'.
## Sedentary behaviours
Eight different sedentary behaviours were assessed including television watching, video/dvd watching, playing video games (e.g. X-Box), in front of the computer, studying or doing homework, talking on the phone, listening to music, and traveling in the car/bus/motorcycle. Students were asked to indicate the usual time (hours per day) that they spent on each of the above activities. Responses were on a six-point scale and ranged from 'zero hours' to 'more than four hours'.
# Data analysis
A principal components analysis with varimax rotation was conducted on the eight items assessing sedentary activities to examine whether these sedentary behaviours could be grouped in different factors. The initial principal components analysis resulted in the extraction of two factors. However, because the internal consistency reliability of the second factor was markedly improved (from α = .56 to α = .67) after deleting the item 'hours per day studying' the factor analysis was conducted for a second time without including this item. The results of the factor analysis with the seven items included are presented in [fig_ref] Table 1: Factor analysis of the items assessing sedentary behaviours [/fig_ref]. Two factors were extracted explaining 57.70% of the variance, KMO = 0.807, Bartlett's Test of Sphericity χ 2 (21) = 2848, p < 0.001. Four items relating to screen-based activities loaded on factor one and was therefore named 'Screen-based sedentary activities' and three items loaded on factor two and was named 'Non-screen based sedentary activities'. Scores of the items that loaded on each factor were summed up and a composite score was obtained for each student on the two types of activities.
Students were classified as physically active if they participated in moderate to vigorous physical activity for at least 60 minutes per day for seven days [bib_ref] Systematic review of the health benefits of physical activity and fitness in..., Janssen [/bib_ref] [bib_ref] Health enhancing physical activity for young people: Statement of the United Kingdom..., Cavill [/bib_ref] and were considered to satisfy the recommendation for daily time watching television if they watched two or less than two hours of television per day [bib_ref] Sedentariness, small-screen recreation, and fitness in youth, Hardy [/bib_ref]. Independent samples t-tests were employed to examine potential differences between boys and girls and physically active and inactive students across the eight sedentary behaviours and the weekly frequency of sports club attendance. Effect sizes (Cohen's d) were also calculated to examine the practical significance of the differences between group means. Chi-square tests were used to examine potential differences in the percentages of adolescents across gender and level of education that satisfied the physical activity and television viewing recommendations. Three series of adjusted logistic regression analyses (one for each of the genders and one for the whole sample) were performed with physical activity (60 or more minutes of moderate to vigorous physical activity per day versus less activity) as the dependent variable and each of the sedentary behaviours and the variable assessing weekly frequency of sports club attendance as the independent variables. The same sets of analyses were repeated with travel mode status (active versus nonactive traveling to school) as the dependent variable. Students were classified as active travelers if they reported as usual mode of travel to school walk or bicycle. Reponses on time spent in sedentary behaviours and weekly frequency of sports clubs attendance were dichotomized based on median values. Independent variables with a significant association with the dependent variable at the bivariate level were entered in a logistic regression model. Level of entry at the model was set at p = 0.01. Because students were nested in schools, we used robust (Huber-White sandwich estimates) standard errors to take account of clustering (non-independence between pupils from the same school) in the computation of 95% confidence intervals and p-values. Analyses were performed using the Complex Samples procedure in the Statistical Package for the Social Science (PASW Statistics 18.0, Chicago, IL, USA) and alpha was set at 0.05.
# Results
Out of the 1966 students who completed questionnaires, 52.4% were boys. Mean age of participants was 14.7 ± 2.2. The majority of participants (84.2%) lived in the four towns of Cyprus (Nicosia, Lemesos, Larnaca, Paphos) and the rest lived in rural areas. [fig_ref] Table 2: Descriptive statistics and t-tests of gender differences in sport club attendance and... [/fig_ref] present the results of the independent samples t-tests across gender (boys and girls), and physical activity (active and inactive) respectively, on each of the sedentary activity items, the two composite sedentary activity variables and the item assessing times per week attending sports clubs. The only large effect size difference between boys and girls was observed on the item assessing hours per day playing video games, whereas the only large effect size difference between active and inactive children was observed on the item assessing times per week attending sports clubs.
## Physical activity and television viewing prevalence
Overall 52.3% of the participants were classified as physically active, with boys more likely to be physically active than girls, (χ 2 (1) = 36.19, p < 0.001) (59.0% versus 45.2%). Statistically significant gender differences were observed across all levels of education. A statistically significant difference was also observed across levels of education, (χ 2 (3) = 83.33, p < 0.001) with a higher percentage of students from primary and middle schools meeting physical activity recommendations in comparison to students from technical and secondary schools. presents the percentages of physically active students across gender and level of education.
On the whole, 52.4% of the participants met the recommendation of watching ≤ 2 hours of television per day with boys more likely to meet the recommendation, (χ 2 (1) = 6.87, p < 0.01) (55.3% versus 49.3%). Statistically significant gender differences were observed across
## Figure 1
Percentages of adolescents across gender and level of education that meet physical activity recommendations. Statistically significant gender differences in primary (p < 0.05), middle (p < 0.001), technical (p < 0.01) and secondary schools (p < 0.05). Statistically significant differences between primary and technical (p < 0.001) and between primary and secondary schools (p < 0.001) and statistically significant differences between middle and technical (p < 0.001) and between middle and secondary schools (p < 0.001). primary and middle schools. A statistically significant difference was also observed across levels of education, (χ 2 (3) = 23.31, p < 0.001) with a higher percentage of students from primary schools meeting recommendations in comparison to students from middle and technical schools and a higher percentage of students from high schools meeting recommendations in comparison to students from technical schools. presents the percentages of students meeting the recommendation across gender and level of education. [fig_ref] Table 4: Logistic regression models predicting activity status [/fig_ref] presents the results of the logistic regression analyses with physical activity level (inactive versus active) as the dependent variable. Boys from high schools and technical education schools were less likely to be physically active (OR = 0.4, 95%CI: 0.3-0.6 and OR = 0.5, 95%CI: 0.3-0.8 respectively) than boys from primary schools. Boys who attended sports clubs for two or more times per week were more likely to be physically active, (OR = 3.4, 95% CI: 2.7-4.2), and boys who listened to music for one or less than one hour per day were less likely to be physically active, (OR = 0.6, 95% CI: 0.5-0.8). Girls from high schools and technical education schools were less likely to be physically active (OR = 0.5, 95%CI: 0.3-0.8 and OR = 0.4, 95%CI: 0.2-0.8 respectively) than girls from primary schools. Girls who attended sports clubs for two or more times per week were more likely to be physically active, (OR = 3.0, 95% CI: 2.2-4.2), and girls who watched television for two or less than two hours per day were more likely to be physically active, (OR = 1.5, 95% CI: 1.1-2.1).
## Associations between physical activity and sedentary behaviours
## Figure 2
Percentages of adolescents across gender and level of education that meet television viewing recommendation. Statistically significant gender differences in primary (p < 0.05) and middle (p < 0.05) schools. Statistically significant differences between primary and middle (p < 0.001) and between primary and technical/vocational schools (p < 0.001) and statistically significant differences between high and technical schools (p < 0.05).
Additional File 1 presents the results of the logistic regression analyses with travel mode status to school (inactive versus active travel) as the dependent variable. Boys from technical education schools were less likely to travel by active mode to school (OR = 0.3, 95% CI: 0.2-0.5) than boys from primary schools. Girls who reported travelling by car/bus/motorbike for one or less than one hour per day were more likely to actively travel to school, (OR = 1.8, 95% CI: 1.1-2.9).
# Discussion
The purpose of this study was to examine the prevalence of physical activity and sedentary behaviours in a sample of Greek children and adolescents in Cyprus and present evidence on the association between these two behaviours. On the whole, 52.3% of the participants met current guidelines recommending that young people should engage in moderate to vigorous physical activity for at least 60 minutes per day. These prevalence estimates are slightly higher than self-reported estimates reported in the US and Canada [bib_ref] Physical activity levels of Canadian children and youth: current issues and recommendations, Katzmarzyk [/bib_ref]. While, in general, about one third of young people from European countries meet these recommendations, according to Armstrong and Welsman [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] comparison between countries should be made with caution as wide variations are observed across countries. Further, comparison is even more complicated because of the different measures adapted in each study to measure physical activity. For example, studies using accelerometers in national and international studies, indicate that the proportion of adolescents meeting these recommendations vary between 2.0 to 61.0% [bib_ref] Disparities in physical activity and sedentary behaviors among US children and adolescents:..., Whitt-Glover [/bib_ref] and 62.0% to 97.6% [bib_ref] Physical activity levels and patterns of 9-and 15-yr-old European children, Riddoch [/bib_ref]. As this is the first study that presents data on physical activity prevalence based on international guidelines among Cypriot youth from different ages, it may be used for comparison purposes until more data from a more representative sample using objective measures of physical activity is obtained.
Our results indicate that boys are more active than girls across all levels of education with the highest prevalence estimates observed among boys from middle and primary schools (69.4% and 68.2% respectively) and the lowest among girls from technical and high schools (26.4% and 34.4% respectively). Further, their appears to be a marked decline in children's physical activity levels after middle school (i.e. after 14-15 years of age) whereby the overall percentages of physical active adolescents in primary and middle schools were 63.2% and 59.8% respectively while the respective percentages for high and technical schools were 37.8% and 44.3%. These findings are in agreement with studies from European countries [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] [bib_ref] Physical activity levels and patterns of 9-and 15-yr-old European children, Riddoch [/bib_ref] and from North America [bib_ref] Physical activity levels of Canadian children and youth: current issues and recommendations, Katzmarzyk [/bib_ref] [bib_ref] Disparities in physical activity and sedentary behaviors among US children and adolescents:..., Whitt-Glover [/bib_ref] indicating gender and age related differences in physical activity levels. Interestingly, in the study by Whitt-Glover et al. [bib_ref] Disparities in physical activity and sedentary behaviors among US children and adolescents:..., Whitt-Glover [/bib_ref] age-related differences were observed from the age of 12, while in the current study the marked decrease was observed in the age of 15. This may be partly explained by the increased homework obligations among older students. These findings suggest that girls may be especially targeted for physical activity interventions as well as children older than 15 in order to reduce the marked decline of physical activity observed.
Sedentary activities that children devoted most of their time to included television watching (2.7 hours per day), listening to music (2.5 hours per day), in front of the computer (2.1 hours per day) and doing homework (1.7 hours per day). These findings are similar to studies from Scotland [bib_ref] The prevalence of sedentary behaviour and physical activity in leisure time: A..., Biddle [/bib_ref] and Hungary [bib_ref] The prevalence of sedentary behaviours and physical activity in Hungarian youth, Hamar [/bib_ref] where television watching, doing homework, and playing computer/video games were among the top five most time consuming sedentary activities. The only large effect size difference observed between boys and girls was in hours per day playing video games (means were 2.1 and 0.8 respectively), a finding that confirms findings from other countries [bib_ref] A descriptive epidemiology of screenbased media use in youth: A review and..., Marshall [/bib_ref] [bib_ref] Use of television, videogames, and computer among children and adolescents in Italy, Patriarca [/bib_ref]. Further, mean hours per day spent watching television in the current study are within the range (1.8 to 2.8 hours per day) reported in a review study by Marshall et al. [bib_ref] A descriptive epidemiology of screenbased media use in youth: A review and..., Marshall [/bib_ref]. Total daily hours spent on screen-based activities (television, video games, DVDs, computer) and non screen-based sedentary activities (talking on the phone, listening to music and motorized transport) were 7.7 and 5.4 respectively. While time spent in both of these types of sedentary behaviours appears to be extensive, it should be noted that screen based activities may be done concurrently with non-screen based activities such as watching television and listening to music or talking on the phone. Interestingly, both of these values observed in the current study are in the range of 5.5 to 8.5 accelerometer derived mean hours per day spent in sedentary activities reported by Whitt-Glover et al. [bib_ref] Disparities in physical activity and sedentary behaviors among US children and adolescents:..., Whitt-Glover [/bib_ref] in a large sample of adolescents from the US.
The finding that about half (52.4%) of the adolescents met the recommendation of watching television for less than two hours per day indicates that there is a need to reduce the time spent in front of the television. These estimates are comparable to data from [bib_ref] Health enhancing physical activity for young people: Statement of the United Kingdom..., Cavill [/bib_ref] [bib_ref] Sedentariness, small-screen recreation, and fitness in youth, Hardy [/bib_ref] [bib_ref] Physical activity levels of Canadian children and youth: current issues and recommendations, Katzmarzyk [/bib_ref] year-old children from mainland Greece [bib_ref] Is spending time in screenbased sedentary behaviors associated with less physical activity:..., Melkevik [/bib_ref] and [bib_ref] The physical activity patterns of European youth with reference to methods of..., Armstrong [/bib_ref] year-old Finnish adolescents [bib_ref] Physical activity and sedentary behaviors among Finnish youth, Tammelin [/bib_ref] but are more favorable than estimates from children in Italy [bib_ref] Use of television, videogames, and computer among children and adolescents in Italy, Patriarca [/bib_ref] and Canada [bib_ref] Television viewing, computer use and total screen time in Canadian youth, Mark [/bib_ref] where 38.0% and 25.0% of children respectively met the recommendation of watching television for less than two hours per day. In our study, boys were more likely to meet the recommendation in comparison to girls (55.3% versus 49.3% respectively), a finding that contradicts findings from previous studies [bib_ref] Physical activity and sedentary behaviors among Finnish youth, Tammelin [/bib_ref] [bib_ref] Is spending time in screenbased sedentary behaviors associated with less physical activity:..., Melkevik [/bib_ref] [bib_ref] Television viewing, computer use and total screen time in Canadian youth, Mark [/bib_ref]. This finding may be partly explained by the large amount of time that boys spent in other competing sedentary activities such as video game playing, or other pursuits such as sports clubs attendance as observed in this study. Boys from primary and high schools were most likely to meet recommendations (65.7% and 58.6% respectively) and girls from technical and middle schools were the least likely to meet recommendations (37.0% and 44.7% respectively). In general, our findings support previous work that suggests that the percentage of children that meets the recommendations decreases as they grow older [bib_ref] Use of television, videogames, and computer among children and adolescents in Italy, Patriarca [/bib_ref] [bib_ref] Television viewing, computer use and total screen time in Canadian youth, Mark [/bib_ref]. Of interest is the low percentage of girls from technical and middle schools that meet recommendations, a finding that suggests that these groups should be especially targeted for intervention programmes.
In general, our findings provide limited support for the displacement hypothesis, as only two significant associations in the subgroup analyses were observed between physical activity and sedentary behaviours. Interestingly, boys who listened to music for less than one hour per day were less likely to be active in comparison to those who listened to music for more than one hour per day. To our knowledge, this is a novel finding and more research is needed to confirm the present association. A possible explanation may be that during physical activity young people may find music both enjoyable and motivating [bib_ref] Young people's exposure to loud music. A summary of the literature, Vogel [/bib_ref] and therefore, those who listen to music for more than one hour per day may simultaneously be more likely to engage in physical activity. The only significant association observed between physical activity and screen based activities was in the girls' analyses where those girls who watched television for less than two hours per day were more likely to be physically active. Previous research has produced contrasting results with some studies failing to show any associations [bib_ref] Longitudinal relationship between television viewing and leisure-time physical activity during adolescence, Taveras [/bib_ref] [bib_ref] Children's physical activity, TV watching and obesity in Cyprus: the CYKIDS study, Lazarou [/bib_ref] , other studies showing associations only with boys [bib_ref] Television viewing and physical activity in children, Hager [/bib_ref] and other studies showing small associations with the whole sample [bib_ref] Physical activity and sedentary behaviors among Finnish youth, Tammelin [/bib_ref].
While our study assessed the association between physical activity and a number of sedentary behaviours as well as between physical activity and composite variables of screen-based and non-screen based sedentary activities, the fact that only two significant associations were observed supports previous research that physical and sedentary behaviours are two separate constructs [bib_ref] Longitudinal relationship between television viewing and leisure-time physical activity during adolescence, Taveras [/bib_ref] and that both need to be targeted in potential intervention programmes to promote physical activity. This is also enhanced by the lack of a significant association between active commuting to school and screen-based sedentary activities, a finding that supports a previous study conducted in Canada [bib_ref] Social-Ecological correlates of active commuting to school among high school students, Robertson-Wilson [/bib_ref].
Another important finding of the present study is the strong association between physical activity and weekly times of sports clubs attendance whereby children who attended sports clubs for two or more times per week were at least three times more likely to meet physical activity recommendations. This finding supports previous results with Greek-Cypriot children using a fourday physical activity recall [bib_ref] Correlates of physical activity in a Cypriot sample of sixth grade children, Loucaides [/bib_ref] and pedometers [bib_ref] Correlates of pedometer-assessed physical activity in Cypriot elementary school children, Loucaides [/bib_ref] as well as findings from the United States using accelerometers, where children accumulated additional 20-minutes of moderate-to-vigorous-activity while attending after school programmes [bib_ref] Physical activity levels among children attending after-school programs, Trost [/bib_ref]. A higher percentage of boys than girls (61.6% and 44.6% respectively) reported attending sports clubs for two or more times per week. Furthermore, there was a graded decrease in the percentages of adolescents attending sports clubs for two or more times per week from primary (68.8%), middle (58.6%), technical and high schools (45.7% and 38.5% respectively). These differences in sports clubs attendance may partly explain gender and age related differences in the percentages of adolescents meeting physical activity recommendations.
While, to our knowledge, this is the first study to examine physical activity and sedentary behaviours in relation to appropriate guidelines in a large sample of Greek children and adolescents in Cyprus from different levels of education, a number of limitations are also worth addressing. First, the cross-sectional design of the present study precludes the inference of cause and effect relationships between physical activity and sedentary behaviours. Second, physical activity was assessed via self-report and future studies within the Cypriot context that examine the relationship between physical activity and sedentary behaviours should adopt objective measures of physical activity including accelerometers or pedometers. Incorporating an objective measure of physical activity behaviour, at least from a subsample, would strengthen the results of this study. Third, while a number of sedentary activities were assessed, students were asked to indicate the usual time (hours per day) that they spent on each of the above activities. Reporting sedentary activities using one-day recalls or diaries, rather than using 'the usual time' might have improved the validity of these measures. While assessing sedentary behaviours such as television viewing with single items is subject to measurement error, this approach has been used in a large number of studies and is appropriate for surveillance studies [bib_ref] Measurement of television viewing in children and adolescents: a systematic review, Bryant [/bib_ref]. Fourth, socioeconomic status data were not collected in this study and possible differences in SES between levels of education might have biased the results. Furthermore, assessment of physical activity and sedentary behaviours did not differentiate between weekdays and weekends.
# Conclusions
Our results indicate that about 50.0% of Greek children and adolescents in Cyprus meet existing physical activity and television viewing guidelines with marked gender and educational level differences. This study provided limited support for the displacement hypothesis indicating that both physical activity and sedentary behaviours need to be targeted when implementing intervention programmes for promoting physical activity and decreasing sedentary behaviours. Encouraging children to enroll and attend sports clubs for at least two times per week may markedly improve their physical activity levels.
# Additional material
Additional file 1: Logistic regression models predicting travel mode to school (non-active versus active traveling) from sports club attendance and sedentary activities. This table presents odds ratios and confidence intervals from the analyses examining the association between travel mode to school (non-active versus active traveling), sports club attendance and sedentary activities.
[table] Table 1: Factor analysis of the items assessing sedentary behaviours [/table]
[table] Table 2: Descriptive statistics and t-tests of gender differences in sport club attendance and sedentary activities [/table]
[table] Table 3: Descriptive statistics and t-tests of differences between inactive and active students on sport club attendance and sedentary activities [/table]
[table] Table 4: Logistic regression models predicting activity status (non-active versus active) from sports club attendance and sedentary activities [/table]
|
10.1089/nat.2022.0060
|
CCBY
|
9940807
|
36516128
|
s2orc_pubmed_articles
|
Consensus Guidelines for the Design and In Vitro Preclinical Efficacy Testing N-of-1 Exon Skipping Antisense Oligonucleotides
on behalf of the N = 1 CollaborativeAntisense oligonucleotides (ASOs) can modulate pre-mRNA splicing. This offers therapeutic opportunities for numerous genetic diseases, often in a mutation-specific and sometimes even individual-specific manner. Developing therapeutic ASOs for as few as even a single patient has been shown feasible with the development of Milasen for an individual with Batten disease. Efforts to develop individualized ASOs for patients with different genetic diseases are ongoing globally. The N = 1 Collaborative (N1C) is an umbrella organization dedicated to supporting the nascent field of individualized medicine. N1C recently organized a workshop to discuss and advance standards for the rigorous design and testing of splice-switching ASOs. In this study, we present guidelines resulting from that meeting and the key recommendations: (1) dissemination of standardized experimental designs, (2) use of standardized reference ASOs, and (3) a commitment to data sharing and exchange.ASOs with the same chemistry often behave similarly with regard to biodistribution and safety. For this reason, we 20 AARTSMA-RUS ET AL.
# Introduction
A ntisense oligonucleotides (ASOs) offer the unique opportunity for sequence-specific targeting of gene transcripts. This can be exploited to reduce expression of transcripts that give rise to toxic gain-of-function proteins via activation of RNase H, or to modulate RNA splicing to promote or restore expression of partially or fully functional proteins for diseases caused by haploinsufficiency or complete loss of proteins [bib_ref] Antisense technology: an overview and prospectus, Crooke [/bib_ref]. As such, ASOs offer potential treatment avenues for a plethora of common and rare genetic diseases, in particular those impacting the liver, central nervous system, and retina, tissues to which ASOs can be delivered efficiently. GalNac conjugates allow very robust ASO uptake by hepatocytes after systemic delivery, whereas local delivery is sufficient for broad uptake within the central nervous system and retina (via intrathecal or intravitreal injection, respectively) [bib_ref] Delivery of oligonucleotide-based therapeutics: challenges and opportunities, Hammond [/bib_ref]. Although local treatment is invasive, a low rate of ASO turnover allows for relatively infrequent dosing: every 3-4 months for the central nervous system and 6 months for the eye.
Several oligonucleotides that require intravitreal or intrathecal delivery have received marketing authorization by the Food and Drug Administration (FDA) to treat eye and motor neuron diseases (fomivirsen, Macugen, and nusinersen) [bib_ref] Antisense technology: an overview and prospectus, Crooke [/bib_ref]. However, it has also demonstrated that it is possible to use ASOs to treat genetic mutations found in as few as an individual patient carrying a unique mutation within an academic setting [bib_ref] Patient-customized oligonucleotide therapy for a rare genetic disease, Kim [/bib_ref]. Milasen, developed at Boston Children's Hospital, was a mutation-specific ASO for a child with Batten disease who carried an intronic mutation in the MFSD8 gene that resulted in cryptic splicing.
ASOs targeting the cryptic exon were designed and resulted in restoration of normal splicing and protein production when tested in patient-derived cell lines. After rat safety studies, an investigational new drug application (IND) was filed with the FDA, and investigational treatment with Milasen was initiated less than a year after discovering the pathogenic variant. Milasen administration was associated with a clear drop in the frequency and duration of epileptic seizures and a slower functional decline [bib_ref] Patient-customized oligonucleotide therapy for a rare genetic disease, Kim [/bib_ref]. However, due to the late stage of her disease, Milasen could not reverse accumulated neuronal damage, and ultimately, the patient passed away 3 years later, although with improved quality of life.
Inspired by this example, initiatives such as n-Lorem were established [bib_ref] Addressing the needs of patients with ultra-rare mutations one patient at a..., Crooke [/bib_ref] and also academic groups set out to develop N = 1 ASOs for individual patients such as the Dutch Center for RNA Therapeutics and the ''1 Mutation 1 Medicine'' (1M1M) initiative [bib_ref] N of 1' therapies need a better model, Aartsma-Rus [/bib_ref] , and results of academic initiatives to develop individualized ASOs for FUS and C9orf72 amyotrophic lateral sclerosis have been published [bib_ref] Antisense oligonucleotide silencing of FUS expression as a therapeutic approach in amyotrophic..., Korobeynikov [/bib_ref] [bib_ref] Suppression of mutant C9orf72 expression by a potent mixed backbone antisense oligonucleotide, Tran [/bib_ref]. The N = 1 Collaborative (N1C) was set up as an umbrella organization to align and facilitate individualized treatment development efforts by sharing best practices and learning from successes and failures. N1C aims to provide tools for the different steps involved in N = 1 ASO development such as guidelines for patient/mutation selection, preclinical ASO design and testing, safety and toxicity tests, and regulatory aspects, as well as a toolkit to measure treatment effects in individual patients.
To develop consensus, interactive workshops on specific topics are organized by the N1C. Each workshop is attended by stakeholders involved in N = 1 ASO development: researchers in academia and industry, clinicians, foundations, and patients and their family members. While topics of specific workshops may appeal more to some stakeholders, N1C aims to solicit input from all stakeholder groups to inform and advance best practices for N = 1 ASO development.
This article is the result of an N1C workshop on ASO design and preclinical testing that was held online on July 25, 2022. The workshop covered both exon skipping and RNase H ASOs with the goal of advancing standards for development that are simultaneously rigorous and efficient (reflecting the clinical needs of patients involved in these early efforts, who often have rapidly progressive diseases for which ''time is neurons''). This article will focus on the design of exon skipping ASOs, while an accompanying forthcoming article will focus on RNase H ASOs.
## Why guidelines and data-sharing are important
While any given mutation-specific ASO may apply to as few as a single patient, this approach could collectively provide benefit to vast numbers of individuals. Successful expansion of this interventional approach will, however, re-quire establishing community standards to ensure rigor in design and development.
Based on the workshop about design and preclinical testing for exon skipping ASOs, we here provide guidelines and considerations. These can benefit preclinical researchers to plan and conduct ASO experiments at a high scientific standard. They also serve as a checkbox to clinicians and patients/families as clinical treatment with an ASO should only be initiated following experiments that were performed properly. This is the first version of these guidelines, based on the current best practices and available knowledge in 2022. We anticipate that with time and increased insight and knowledge, updates will be produced.
Notably, these guidelines focus on proper experimental designs for preclinical ASO development and assume as a prerequisite that due clinical diligence has been performed to assess if exon skipping is expected to be therapeutic in the first place (for more details, we refer the reader to Synofzik et al. [bib_ref] Preparing n-of-1 antisense oligonucleotide treatments for rare neurological diseases in Europe: genetic,..., Synofzik [/bib_ref]. Furthermore, the important topic of safety evaluations (to assess if lead compounds are safe before initiating treatment of patients) is beyond the scope of these guidelines. We refer the reader also to the N-of-1+ Oligonucleotide Therapeutics Society briefing document for further information on the full ASO development process: https://www.oligotherapeutics.org/ rare-disease-task-force/rare-disease-briefing-document/ and to the N = 1 Collaborative website for up-to-date considerations with regard to safety testing, preparing for clinical trials, and other relevant topics.
## Exon skipping
There are three general ways in which ASO-mediated exon skipping can be used to treat genetic diseases [fig_ref] FIG. 1: Ways to use ASOmediated exon skipping to restore protein production [/fig_ref]. The first is skipping a cryptic or pseudoexon (from here on referred to as cryptic exon) [bib_ref] Antisense oligonucleotidebased rescue of aberrant splicing defects caused by 15 pathogenic variants..., Tomkiewicz [/bib_ref]. Variants within the intron have been for long time considered innocuous. However, an increasing number of intronic variants are being described as the cause of rare diseases. These variants can cause part of an intron to be included into the final mRNA, leading to a disruption of the reading frame and therefore reducing protein levels. ASOs targeting these cryptic exons can restore normal splicing and thus normal protein production. Milasen is an example of an ASO targeting a cryptic exon [bib_ref] Patient-customized oligonucleotide therapy for a rare genetic disease, Kim [/bib_ref]. The advantage of this type of exon skipping is that it restores functional mRNA and protein production.
The second type of exon skipping aims to skip in-frame constitutive exons that harbor a pathogenic variant-most typically, one that causes loss of functional protein, such as a stop gain or frameshift variant [bib_ref] Ataxin-3 protein modification as a treatment strategy for spinocerebellar ataxia type 3:..., Evers [/bib_ref]. Exon skipping strategies can be designed to bypass the pathogenic variant to allow production of an internally deleted protein. Whether this protein will be (partially) functional will depend on if the skipped exon encodes crucial domains for protein structure or function. For this type of exon skipping, studies to confirm the protein is functional and stable are crucial. One can consider to design ASOs targeting the mutation to achieve exon skipping in an allele-specific manner [bib_ref] Update of genetic variants in CEP120 and CC2D2A-With an emphasis on genotypephenotype..., Barroso-Gil [/bib_ref]. However, this will not always be possible, either because the ASOs are not selective enough or because the mutated region is a suboptimal target site for ASO-mediated exon skipping.
Finally, exon skipping can be used for haploinsufficiency diseases by skipping ''poison exons'' [bib_ref] Antisense oligonucleotides increase Scn1a expression and reduce seizures and SUDEP incidence in..., Han [/bib_ref]. Poison exons are This prevents protein production. ASOs targeting the cryptic exon can prevent inclusion into the mRNA allowing production of a normal mRNA and normal protein. (C-E) ASO-mediated skipping of constitutively spliced exons can restore the production of partially functional proteins in multiple ways: (C) by restoring the reading frame, to allow the production of an internally deleted, but partially functional protein; (D) by skipping an in-frame exon containing a nonsense or frameshifting variant, which will bypass the variant, while maintaining the reading frame, to allow the production of an internally deleted, but partially functional protein; (E) by skipping an inframe exon containing a toxic gain of function variant, which will allow the production of an internally deleted protein that is partially functional, rather than a toxic protein. (F) For some genes, transcripts containing poison exons, short naturally occurring exons containing a stop or a frameshift, are produced. These transcripts are subjected to nonsense-mediated mRNA decay. ASO-mediated skipping of poison exons can increase the amount of functional transcripts produced and thus increase the amount of protein, which can be therapeutic for haploinsufficiency diseases. ASO, antisense oligonucleotide.
short, naturally occurring, highly conserved alternative exons that contain a premature termination codon, and when spliced into the transcript targets the transcript for nonsensemediated decay, leading to reduced protein levels. A current example of using ASOs to increase protein levels by skipping a poison exon is for Dravet syndrome, an autosomal dominant disease, which is caused by variants in the SCN1A gene causing reduced expression of voltage-gated sodium channel alpha subunit Na v 1.1. Interestingly, transcript isoforms containing an exon that disrupts protein production are expressed at relatively high levels in brain. Skipping this ''poison exon'' increases the amount of functional protein, thus compensating for the haploinsufficiency. This approach is called targeted augmentation of nuclear gene output and restores normal transcripts and proteins. This approach can only be exploited when transcripts with poison exons are produced in the target tissue.
Notably, ASO splice modulation can also result in exon inclusion, for example, nusinersen induces the inclusion of exon 7 in SMN2 transcripts [bib_ref] Antisense technology: an overview and prospectus, Crooke [/bib_ref]. However, ASOs that induce exon inclusion are harder to develop, typically being dependent on a detailed understanding of the splicing regulatory signals near the exon to be included [bib_ref] How the discovery of ISS-N1 led to the first medical therapy for..., Singh [/bib_ref]. In these guidelines, we will only focus on the use of ASOs to induce skipping of cryptic, constitutive, or poison exons.
## Designing exon skipping asos
Guidelines for the design of exon skipping ASOs have been published based on comparison of effective and ineffective ASOs for dystrophin exon skipping [bib_ref] Overview on AON design, Aartsma-Rus [/bib_ref]. Furthermore, a software tool to help predict effective exon skipping ASOs has been developed based on published information on exon skipping ASOs [bib_ref] eSkip-Finder: a machine learning-based web application and database to identify the optimal..., Chiba [/bib_ref]. These tools are not yet perfected and are based on limited information. Nevertheless, there are some general rules that can enhance the chance of identifying an effective ASO:
(1) A GC percentage of 40%-60% appears most optimal.
(2) ASOs of 18-22 nucleotides appear most optimal.
(3) Avoid stretches of three or more cytosines in the target sequence, as stretches of three or more Gs in the ASO are very prone to self-structure. (4) Targeting splicing enhancer sites within an exon increases the chance of efficient exon skipping. (5) For constitutive exons, targeting the first 30% of the exon and targeting exon-internal sequences rather than splice sites increase the chance of finding an effective exon skipper. (6) In some cases, effective ASOs do overlap splice sites.
We suggest that in these cases the majority of the ASO be designed to target the exon, but with 1-5 nucleotides covering the donor or acceptor splice site. This helps avoid hybridization with unintended targets and helps enable the discovery of ASOs with appropriate GC percentages (although both of those parameters should be checked carefully for splicesite-targeted ASOs). (7) For pathogenic variants inducing cryptic exons, targeting splicing enhancer sites increases the chances of cryptic exon skipping; however, blocking weak splice sites has also proven effective in some cases
We would like to stress that the guidelines and the eSkip-Finder software will need further optimization and are currently based primarily on skipping constitutive exons. Several studies have confirmed that similar rules apply to cryptic and poison exons [bib_ref] Antisense oligonucleotide screening to optimize the rescue of the splicing defect caused..., Garanto [/bib_ref]. However, the information on these cases is even more limited than for skipping constitutively spliced exons. Larger data sets of effective and ineffective ASOs are required to improve ASO design tools.
Candidate ASOs should be checked for uniqueness in the transcriptome using in silico NCBI BLAST analysis. In addition, one should check the genome as well, to rule out overlap with intronic regions and exon-intron junctions. To avoid potential off-target exon skipping, ASOs having homology with other transcripts of 17 or more consecutive nucleotides should be avoided. Assuming the 2¢-O-methoxyethyl phosphorothioate (MOE PS) chemistry of nusinersen is used, ASOs with partial homology of 15 or more consecutive nucleotides to a large number of transcripts should be avoided. While they are unlikely to result in exon skipping, these transcripts may compete with binding to the target transcript resulting in lower efficiency. Note that some third-generation chemical modifications such as locked nucleic acids (LNA) have very high affinity for RNA and likely will already hybridize to shorter homology stretches.
When partial homology with non-target transcripts cannot be avoided, one should consider when and where the target transcript is expressed in relation to the time and tissue that will be treated. For example, if a transcript is only expressed embryonically, it will not be present in patients. Similarly, if the off-target transcript is present only in the central nervous system, while the liver is targeted, this likely will not present a problem.
In addition, the region where the ASOs bind is also important. If an ASO hybridizes directly on an exon or the boundaries, it more likely can affect splicing. In contrast, in the intronic region, chances are lower; yet this should still be verified.
Finally, when one is aware of a potential nonspecific target of an ASO and it is expressed also in the target tissue, one can evaluate in relevant cell types whether the ASO indeed causes exon skipping. Given how challenging it is to develop an effective ASO, it is not likely that ASOs with partial sequence overlap result in efficient exon skipping. However, it can happen and exon skipping events have been reported in cultured cells despite two or three mismatches between the ASO and the skipped exon target sites [bib_ref] Hybridization-mediated off-target effects of splice-switching antisense oligonucleotides, Scharner [/bib_ref].
If possible, we recommend designing and testing up to 10 ASOs per target exon in initial preclinical studies. This may not be possible if the exon is too short, contains CCC motifs at inconvenient locations, there is homology with other transcripts, or there are repetitive sequences or motifs in the region that are recurrently found in our genes (eg, Alu repeats, microsatellites).
If needed, a second round of ASO design can be performed to optimize ASO efficacy (eg, making the ASOs slightly longer or shorter or moving them 1-2 nucleotides to the left or the right). recommend using MOE PS and 5-methylated cytosine and uridine residues for the development of mutation-specific exon skipping ASOs, that is, the chemical modifications used for nusinersen. For this chemistry, pharmacokinetic and pharmacodynamic data after intrathecal delivery are publicly available for multiple species including humans. While the sequence can have profound effects on pharmacokinetic properties and each sequence needs to be tested for safety, the use of MOE PS enables a shorter in vivo safety package in a single species (https://www.fda.gov/regulatory-information/ search-fda-guidance-documents/ind-submissions-individuali zed-antisense-oligonucleotide-drug-products-administrativeand-procedural).
However, the 2¢-O-methyl phosphorothioate (2OMePS) chemistry without methylated cytosine and uridine residues is a cheaper alternative that generally shows similar efficacy in in vitro experiments. As such, it is possible to perform initial target optimization studies with 2OMePS ASOs and to switch to the MOE PS chemistry with methylated cytosines and uridines for lead compounds. In that case, it has to be confirmed that MOE PS ASOs are similarly efficacious in vitro before moving to safety studies.
Other chemistries can be considered as well. Whether this is opportune depends on the regulatory jurisdiction. In the United States, N = 1 treatment can only be initiated after an IND filing with the FDA, which involves rigorous safety studies in rodents. However, in Europe, N = 1 treatment can be performed in a ''named patient'' setting, which does not involve regulatory approval, provided there is a clinical track record for chemical modification used [bib_ref] N of 1' therapies need a better model, Aartsma-Rus [/bib_ref]. In that case, not using the nusinersen chemistry (MOE PS) will unduly delay treatment initiation.
## Cell line considerations
As the ASOs target human exons, human cell models are needed to establish ASO efficiency, as well as potential offtarget effects if identified by BLAST analysis. ASO optimization is ideally carried out in cell cultures that are easy to work with (eg, fibroblasts or neuroblastoma cells rather than neuronally differentiated induced pluripotent stem cells [iPSCs]), but ideally also relevant to the tissue type impacted in disease. When the target exon is not mutated (constitutive or poison exons), wild-type cells can be used. In cases of cryptic splicing mutations or other mutated target exons, patient-derived or gene-edited cells are required.
Obviously, the target transcript has to be expressed in the selected cultured cells. In case of cryptic or poison exons, inclusion of these exons in the transcript has to be confirmed in the cell system used, as these processes can be tissuespecific. In cases where the gene is expressed, but the mutation causes a premature termination codon or an out-offrame transcript, nonsense-mediated mRNA decay (NMD) can potentially reduce the levels of mutated transcripts rendering it undetectable. In those cases, blocking the NMD process with cycloheximide or a similar reagent will allow the detection of the mutated transcript and therefore allow assessing the efficacy of the ASO in reducing the mutant transcript and increasing the functional one. This is usually the case of cryptic exons, and several studies have shown that the use of cycloheximide revealed the splicing defect, which was not detectable in untreated conditions [bib_ref] Deep-intronic ABCA4 variants explain missing heritability in Stargardt disease and allow correction..., Sangermano [/bib_ref] [bib_ref] Identification and rescue of splice defects caused by two neighboring deep-intronic ABCA4..., Albert [/bib_ref] [bib_ref] A novel nonsense variant in ARID1B causing simultaneous RNA decay and exon..., Sofronova [/bib_ref].
While iPSCs differentiated to neuronal lineages may not be the first choice due to cost, labor, and time considerations, for some transcripts this will be the only option. Note that transdifferentiation using NGN2 can be an alternative for neuronal differentiation. For other cell types, this might be more difficult. For example, in the retina, although there are protocols for transdifferentiation of fibroblasts to photoreceptor cells using four transcription factors [bib_ref] The manner of decay of genetically defective EYS gene transcripts in photoreceptordirected..., Seko [/bib_ref] , the results are not yet optimal and not all genes of interest are detected. In those cases, alternative models in midigenes for initial screening and patient-derived iPSC cellular models for final validation are required. Notably, minigenes generally are too small to provide sufficient intronic context.
Generally, the use of minigenes or midigenes is not recommended for discovery of exon skipping ASOs because these systems take part of the gene out of its genetic context. Furthermore, the minigenes or midigenes are often expressed in a cell type that is easy to transfect, but where the expression of splicing factors may differ from the cell type where the transcripts are normally expressed. These differences will influence the inclusion of poison and cryptic exons (eg, the cryptic exon is not included in the midigene) as well as ASO efficacy (eg, ASOs are efficacious in the midigene system but not in patient-derived cells of the relevant tissue).
While the use of midigenes and minigenes is not recommended for the reasons outlined above, there will be cases where this is the only option to optimize ASOs in a timely manner. In that case, it is crucial to validate that transcription is similar to the target tissue and to confirm the optimal ASOs in a patient-derived cell setting. Of note, it is important that for cryptic exons caused by deep-intronic variants creating a new splice site, this system has been extremely useful to identify splicing defects in several retinal genes [bib_ref] ABCA4 midigenes reveal the full splice spectrum of all reported noncanonical splice..., Sangermano [/bib_ref]. However, when the insertion of such a cryptic exon was caused by the generation of a splicing enhancer, both midigenes and patient-derived fibroblasts do not always recapitulate the cryptic exon inclusion identified in retinal cells [bib_ref] Identification and rescue of splice defects caused by two neighboring deep-intronic ABCA4..., Albert [/bib_ref]. This highlights again the importance of the molecular and genetic context, suggesting that validation in relevant models is required.
## Controls to be included
## Control considerations
It is crucial to take along control ASOs for various aspects [bib_ref] Guidelines for experiments using antisense oligonucleotides and double-stranded RNAs, Gagnon [/bib_ref]. Controls should have the same chemistry composition as the targeting ASOs. We recommend the following controls in all experiments:
Transfection controls. The purpose of these controls is to confirm if the transfection worked properly and may also confirm localization of the ASO (nuclear ideally). Especially if no effective ASO has yet been identified for the target and/or a new cell type is being used, confirming uptake is crucial to avoid concluding that ASOs are ineffective if indeed transfection or delivery failed.
Researchers can use a fluorescently labeled ASO for this purpose, testing localization by microscopy. However, when possible, an ASO known to work efficiently provides the ideal transfection control since it measures functional delivery. This can also target an exon in another transcript and can be the same as the fluorescently labeled ASO mentioned above. If this ASO gives a poor skip, you know that transfection process is not optimal. Even an RNase-H-recruiting ASO targeting a ubiquitously expressed and easy to silence transcript, such as the noncoding RNA MALAT1, can serve as a useful control for efficient transfection.
Nontargeting ASO. An ASO that does not target the exon is required to rule out exon skipping due to transfection or ASO treatment per se (some alternative splicing may be influenced by this). For functional assays, treatment effects should always be compared with a nontargeting control ASO, as transfection or ASO treatment can influence multiple cellular physiological processes.
It is known that certain ASO motifs can trigger immune responses or change physiological processes in the cells [bib_ref] Mechanisms and applications of immune stimulatory CpG oligodeoxynucleotides, Krieg [/bib_ref]. We therefore recommend using nontargeting ASOs for which it is known that they are well tolerated and do not cause obvious phenotype or gene expression changes in the cells of interest. Nontargeting controls should ideally have a similar GC percentage and length to the targeting ASO, and they should have the same chemical modification.
Different options of nontargeting ASOs exist:
A previously identified control ASO known to be well tolerated in previous contexts generally does not contain motifs that can trigger immune responses or change physiological processes (may have different GC percentage and length as the targeting ASO) Scrambled ASO (will have the same GC percentage and length as the targeting ASO, but may contain undesirable motifs and may target other exons in the transcriptome) Sense ASO (ie, complementary to the active antisense sequence) will by definition have the same GC percentage and length as the targeting ASO but may contain undesirable motifs and may target other exons in the transcriptome
We recommend using a previously identified, generic nontargeting control ASO with no known side effects for initial screening studies. However, when functional studies are performed with lead compounds, scrambled and/or sense ASO controls should also be included.
The N1C is currently collecting information to produce an online database of control ASOs with confirmed ''good behavior'' to facilitate selecting a control ASOs for the cell type of choice.
We recommend using untreated cells as an additional negative control. All the types of nontargeting ASO controls above should show similar readouts relative to the untreated cells: if this is not the case, investigators needs to test additional controls and carry out additional experiments to try to understand what is happening. It may be that the act of transfection or delivery is perturbing the process of interest. The ''untreated cells'' control is therefore a sort of reality check on the specificity of the effect. However, the nontargeting ASOs are considered the key controls, and when measuring functional effects, the reference sample should be the ASO control rather than untreated cells.
## Transfection versus gymnosis
Transfection reagents are very useful to deliver ASOs to cultured cells. We recommend using transfection reagents during the first screening round to assess whether ASOs are effective or not. Generally concentrations of 100 nM or less are sufficient, and efficacy can generally be assessed after 24 h. If you need higher concentrations to induce exon skipping, likely the ASO will not be efficient in vivo because the local concentrations in tissues will be lower for most cells. It should be appreciated that transfection of ASOs is extremely efficient and will negate any ASO-specific aspects that can influence efficiency of delivery (eg, length, protein binding).
We therefore recommend to also assess ASO efficacy after gymnotic uptake (naked uptake) for effective ASOs [bib_ref] Silencing of gene expression by gymnotic delivery of antisense oligonucleotides, Soifer [/bib_ref]. Here, higher doses of ASOs are needed (0.5-10 mM), and efficacy should be assessed after 72 h or longer. For some cell types, performing gymnotic uptake experiments in 9 mM CaCl 2 (drastically) improves efficiency, although some cell types do not tolerate this [bib_ref] Calcium-mediated in vitro transfection technique of oligonucleotides with broad chemical modification compatibility, Wada [/bib_ref].
As gymnotic uptake experiments are generally more predictive of in vivo uptake, we recommend basing candidate selection on gymnotic uptake results in relevant models. If gymnosis works very efficiently for the cell model you are using, one could forego the transfection step and only use gymnosis.
## Measuring efficacy on rna level
To measure efficacy and efficiency of exon skipping levels on RNA, RT-PCR analysis can be performed. Several aspects have to be considered here:
## Primer selection
It is possible that, in addition to the target exon, one or more additional exons are also skipped. Therefore, on top of primers targeting flanking exons, primer pairs further away from the target exon should also be included to study this. For larger exons, PCR amplification of the wild-type product might be problematic. The design of primers recognizing either the wild-type exon-exon boundaries versus primers recognizing the novel skipped exon-exon boundaries can be useful in this case. Furthermore, if additional exons are skipped, this can result in out-of-frame transcripts with premature stop codons, which may be quickly degraded and thus go undetected. We suggest, as described above, to use cycloheximide or similar reagents to block NMD and detect these unexpected/undesired transcripts.
## Amplification bias
Smaller fragments will often be amplified more efficiently during PCR than longer fragments. This means that exon skipping levels will generally be overestimated with RT-PCR analysis when based on densitometry analysis of agarose gels. Even a 1% more efficient amplification will result in a 40% increase after 35 cycles (1.4 times overestimation), whereas a 10% more efficient amplification results in a 2,500% increase (25 times overestimation). At the same time, larger fragments will be better labeled and visualized in a gel as they will bind more intercalating dies, creating a bias in the opposite direction. When using Bioanalyzer capillary electrophoresis technology, fragment concentrations will be provided and length will be taken into account [bib_ref] A multicenter comparison of quantification methods for antisense oligonucleotide-induced DMD exon 51..., Hiller [/bib_ref].
In any case, when different ASOs are compared within the same PCR, the amplification bias will apply to all samples so selecting the most optimal ASOs will still work. However, [bib_ref] A novel nonsense variant in ARID1B causing simultaneous RNA decay and exon..., Sofronova [/bib_ref] AARTSMA-RUS ET AL.
one should refrain from drawing conclusions on efficiency and efficacy, as this will be a semiquantitative value that can lead to under-or overestimation.
Notably, ASOs will be present in the RNA as well and they will be able to bind to the target exon in transcripts and amplified fragments. This can also interfere with amplification efficiency, especially for ASOs with chemical modifications that result in very high affinity binding to the target (eg, LNA). This will also result in a preferential amplification of the skip fragment as the ASO will only bind to the fulllength fragment and not the skip fragments.
Quantitative polymerase chain reaction or digital droplet PCR Quantitative polymerase chain reaction (qPCR) analysis can be used to quantify amounts of ''non-skipped'' and ''skipped'' fragments. Here, one can normalize for amplification efficiency. Primers have to be carefully designed here to avoid amplification of both the ''skipped'' and the ''nonskipped'' fragment for the ''non-skipped'' primers. One can design primers on the exon-exon junctions to avoid this, but sometimes exon-exon junctions are similar. Therefore, it has to be validated that that the primers are specific only for the intended target (''skipped'' or ''non-skipped'') with similar efficiencies. As each primer pair can have different efficiencies, this can lead to relative over-or underestimation of the skip, and therefore, efficacy should not be seen as absolute amounts. Also, for qPCR, there can be interference of ASOs binding to transcripts and amplification fragments containing the targeted exon.
The golden standard using absolute quantification is the digital droplet PCR (ddPCR) [bib_ref] A multicenter comparison of quantification methods for antisense oligonucleotide-induced DMD exon 51..., Hiller [/bib_ref] or RNA-sequencing. However, these are expensive techniques and may not be available to everyone. There are cases, however, where ddPCR or RNA-sequencing may be the only option (eg, to discriminate between two transcripts of similar length). For ddPCR, one will also have to confirm specificity of the ddPCR probes for skip and non-skip products. Furthermore, while this provides an absolute quantification, one should realize that in vitro studies do not directly translate to the in vivo situation, where the efficacy likely will be lower.
Pragmatically, we recommend using standard RT-PCR to compare ASO efficiencies and select lead candidates if possible. We do urge users to be careful when mentioning exon skipping percentages as a quantitative measure, and to clearly state this is only semiquantitative and likely an overestimation of the real situation.
When cycloheximide treatment is needed to observe nonproductive transcripts, each condition should be performed with cycloheximide treatment. Only the nontreated condition needs to be performed both with and without cycloheximide. This will also serve as validation of those transcripts appearing upon inhibiting NMD. For cryptic exon containing transcripts subjected to NMD, blocking NMD will allow assessment of a decrease of the aberrant transcript in parallel with assessing an increase of the wild-type transcript.
## Assessing effects on protein and/or functional level
The studies to assess protein restoration or reduction of functional deficits will depend on the protein involved and the cellular function of this protein. As such, we cannot make detailed recommendations. However, a common procedure we suggest to perform for lead candidates is first confirming protein restoration. In cases where protein restoration is detected, additional studies to assess protein stability and normalization of functional deficits can be performed as well. Protein analysis relies on the availability of antibodies and the most straightforward test is Western blot analysis. For approaches aiming to skip an in-frame exon, Western blot analysis can also confirm the production of a shorter protein. We realize that, in some cases, performing these studies will not be possible (eg, because no antibody is available, the protein is difficult to blot, or because no deficits are seen in cultured cells).
When these experiments are possible, inclusion of the proper controls (see above) is crucial to avoid drawing falsepositive or false-negative conclusions. In these cases, the functional effects have to be offset to the control ASO reference. However, as mentioned previously, taking along an untreated sample as well is recommended to assess whether the treatment procedure has an effect on protein or function per se. Many of the confounding effects described above (such as PCR artifacts and primer or ASO interference) do not apply to a study of protein-level expression, so positive protein-level data are of significant help in building confidence that the ASO is performing as expected.
## Takeaway messages for clinicians and families
These guidelines aim to facilitate exon skipping ASO development in a standardized and scientifically rigorous manner to allow obtaining high-quality efficiency and efficacy data, which will be useful to optimize software for exon skipping ASO design. However, there are also takeaway messages for clinicians who will be provided with an ASO to treat a patient and patients and families who are given the possibility to be treated by such an ASO.
It is clear that developing and optimizing ASOs preclinically is complicated. However, toward N = 1 treatment, a collaborative effort is required, involving different perspectives that are not all covered by preclinical researchers' expertise. Clinicians need to ask critical questions about using exon skipping as a therapeutic option for a specific variant, patient, and disease: Does the rationale make sense? Will exon skipping indeed lead to a functional protein? Will this result in clinical benefit for the specific patient? What will this benefit look like and it is possible to measure it? Is it possible to treat this patient with intrathecal injection or does the pathology prevent this (eg, scoliosis)? What would that benefit look like? Patients and families should also question whether the expected benefit would outweigh the burden of a repetitive invasive treatment.
While it may not be possible for clinicians and patients and families to understand the nuances of the preclinical studies, checking if quality controls were included should be possible. Also the level of detail can be easily assessed: does the preclinical work involve only RNA analysis or demonstrates protein-level rescue and functional effects of treatment? Demonstrating functional effects, when possible, is obviously of higher value in terms of showing promise for translation to further development toward patient use. When a researcher provides an ASO for clinical use in patients, both patients and clinicians should question if relevant safety studies were conducted.
## Recommendations for the future
There are three takeaways from our meeting. First, it is crucial to disseminate best practices, which we aspire to do with this publication. Second, it is important to have standardized reference ASOs, which is a resource the N1C will compile. Finally, data sharing will be crucial: N1C will only be able to achieve its goals if we implement data sharing standards to allow each effort to inform the next, supporting a continual improvement process that maximizes efficiency, efficacy, and safety. Sharing these data will allow critically important retrospective analyses of ASO efficacy and safety and enable the derivation of open-source computational design algorithms to further support the community. The quality of the design software will depend on the quality of the data submitted. Ideally, studies are performed with scientific rigor, and all required positive and negative controls were included, which brings us back to the importance of best practices.
Developing N = 1 ASOs takes the combined and active participation of a group of experts, including preclinical researchers, clinicians, genetic counselors, patients, pharmacological and toxicology experts, hospital pharmacists, regulators, and ethicists. While a team of people is needed for each specific N = 1 ASO and each individual patient, processes and procedures can be streamlined across efforts. The N1C aims to provide best practices and guidelines to give each of the stakeholders the tools to be actively involved and to facilitate development of N = 1 ASOs for the patients who need them. We also welcome you to submit your effective and ineffective ASO sequences to the N = 1 database once it comes online. The preclinical design and development of exon skipping ASOs is a small part of this process, but we hope that the guidelines will be useful to those developing N = 1 ASOs.
# Author disclosure statement
Authors disclose being members of the N = 1 collaborative. Authors do not disclose anything else related to this work.
[fig] FIG. 1: Ways to use ASOmediated exon skipping to restore protein production. (A) Normally, the protein coding information is dispersed over exons in a gene. The gene transcript will contain both exons and introns. During splicing, the introns are removed, resulting in the messenger RNA (mRNA), which is translated into a protein. (B) Cryptic splicing variants cause part of an intron to be recognized as an exon and aberrantly included in the mRNA. [/fig]
|
10.1002/per.2088
|
CCBY
|
5324657
|
28303077
|
s2orc_pubmed_articles
|
Using Response Surface Analysis to Interpret the Impact of Parent–Offspring Personality Similarity on Adolescent Externalizing Problems
Personality similarity between parent and offspring has been suggested to play an important role in offspring's development of externalizing problems. Nonetheless, much remains unknown regarding the nature of this association. This study aimed to investigate the effects of parent-offspring similarity at different levels of personality traits, comparing expectations based on evolutionary and goodness-of-fit perspectives. Two waves of data from the TRAILS study (N = 1587, 53% girls) were used to study parent-offspring similarity at different levels of personality traits at age 16 predicting externalizing problems at age 19. Polynomial regression analyses and Response Surface Analyses were used to disentangle effects of different levels and combinations of parents and offspring personality similarity. Although several facets of the offspring's personality had an impact on offspring's externalizing problems, few similarity effects were found. Therefore, there is little support for assumptions based on either an evolutionary or a goodness-of-fit perspective. Instead, our findings point in the direction that offspring personality, and at similar levels also parent personality might impact the development of externalizing problems during late adolescence.
Personality has been associated with the development of externalizing problems (e.g. [bib_ref] Do substance use risk personality dimensions predict the onset of substance use..., Malmberg [/bib_ref] [bib_ref] Adolescent personality, problem behaviour and the quality of the parent-adolescent relationship, Manders [/bib_ref] [bib_ref] Internalizing and externalizing problems in adolescence: General and dimension-specific effects of familial..., Ormel [/bib_ref] [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref]. Recently, studies of the association between personality and externalizing problems have started to examine personality within the broader social context (see also [bib_ref] The social consequences of personality: Six suggestions for future research, Back [/bib_ref]. That is, individuals might be affected by their own personality, but also by the personality of important others and by the match between both personalities. Whereas various studies have examined personality similarity between peers and romantic partners (e.g. [bib_ref] Two personalities, one relationship: Both partners' personality traits shape the quality of..., Robins [/bib_ref] [bib_ref] Emerging late adolescent friendship networks and big five personality traits: A social..., Selfhout [/bib_ref] , little is known regarding personality similarity between parents and offspring. Only two previous studies examined parent-offspring personality similarity [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref] , suggesting that personality similarity might play an important role in offspring's development. However, it is not known whether the level of personality traits affects the impact of personality similarity. It is possible that similarity has a different impact at low or high levels of personality traits, or might differ depending whether the parent or the offspring has a higher level of certain personality traits.
Personality has been characterized as 'relatively stable individual differences in affect, behaviour, and cognition' . The Big Five model [bib_ref] Personality development: Stability and change, Caspi [/bib_ref] [bib_ref] Validation of the five-factor model of personality across instruments and observers, Mccrae [/bib_ref] captures such individual differences in five traits: Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness to experience. Of these traits, Extraversion, Neuroticism, and Openness have most consistently been associated with negative outcomes (e.g. [bib_ref] Linking "big" personality traits to anxiety, depressive, and substance use disorders: A..., Kotov [/bib_ref]. Studies have indicated that Extraversion was not or only moderately positively associated with externalizing problems (such as aggression, antisocial behavior, or delinquency), Openness was not or negatively associated with externalizing problems, and Neuroticism was positively associated with externalizing problems [bib_ref] The "little five": Exploring the nomological network of the five-factor model of..., John [/bib_ref] [bib_ref] Personality, antisocial behavior, and aggression: A meta-analytic review, Jones [/bib_ref] [bib_ref] Longitudinal associations between personality traits and problem behavior symptoms in adolescence, Klimstra [/bib_ref] [bib_ref] Structural models of personality and their relation to antisocial behavior: A meta-analytic..., Miller [/bib_ref] [bib_ref] Examining antisocial behavior through the lens of the Five Factor Model of..., Miller [/bib_ref].
The two studies investigating parent-offspring similarity in personality indicated that, when similarity mattered, it was associated with fewer externalizing problems [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref]. From an evolutionary perspective, such similarity might be beneficial. As personality is heritable, genetic factors account for approximately 40%-60% of individual differences in personality (e.g. [bib_ref] A behavioral genetic study of the overlap between personality and parenting, Spinath [/bib_ref] , similarity in personality might indicate genetic similarity. From an evolutionary perspective, fathers would have thus more proof that they are the genetic father of their offspring and might therefore be more inclined to help kin that have similar personality characteristics (see [bib_ref] Father-offspring resemblance predicts paternal investment in humans, Alvergne [/bib_ref] [bib_ref] Evolution and proximate expression of human paternal investment, Geary [/bib_ref].
The two previous studies investigating similarity used Qcorrelations [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref] and difference scores. Such analyses do not differentiate between pairs who have similarly low or high personality traits. For example, offspring and parents who are both low on Neuroticism or both high on Neuroticism would receive the same score-indicating a high similarity. Moreover, such studies did not differentiate between the offspring or parent scoring higher or lower on a certain trait. Thus, a parent with higher Neuroticism than the offspring would have received the same difference score compared to offspring having higher Neuroticism than the parent does. Little is known, however, about whether the level of personality traits affects the impact of personality similarity; thus, whether effects differ for pairs who score similarly low or high on personality traits. From a goodness-of-fit perspective, an individual's temperament should match the demands and expectations of the social environment [bib_ref] The import of temperament for psychosocial functioning: Tests of a goodness of..., Lerner [/bib_ref] [bib_ref] Temperament and goodness of fit: Implications for developmental psychopathology, Seifer [/bib_ref]. Similarity in positive traits, such as Extraversion or Openness, might lead to a better mutual understanding. Such similarity might thus be associated with better outcomes for the offspring, such as less externalizing problems. However, from an interpersonal circumplex perspective (see, [bib_ref] When do opposites attract? Interpersonal complementarity versus similarity, Dryer [/bib_ref] [bib_ref] An informal history of the interpersonal circumplex tradition, Wiggins [/bib_ref] dissimilarity, or complementarity, might be preferable for some negative personality characteristics. In line with a goodness-of-fit perspective [bib_ref] The import of temperament for psychosocial functioning: Tests of a goodness of..., Lerner [/bib_ref] [bib_ref] Temperament and goodness of fit: Implications for developmental psychopathology, Seifer [/bib_ref] , similarity at high levels of negative traits might lead to a bad fit between parent and offspring as the demands and expectations between the offspring's temperament and social environment (i.e. the parent) are suboptimal. Therefore, similarity at high levels of negative traits, such as Neuroticism, might be associated with a worse fit and thus might be associated with more externalizing problems.
A promising way to overcome the methodological limitations of earlier studies investigating similarity in parent offspring personality is using polynomial regression analyses (see [bib_ref] The social consequences and mechanisms of personality: How to analyse longitudinal data..., Nestler [/bib_ref] [bib_ref] Polynomial regression with response surface analysis: A powerful approach for examining moderation..., Shanock [/bib_ref]. Such analyses allow differentiating between effects based on similarity at lower or higher levels of certain personality traits. Moreover, such analyses allow differentiating between parents or offspring scoring higher on certain traits. However, polynomial regression analyses have not yet been used to examine the effects of parent-offspring similarity in personality.
This study aimed to investigate the effects of parentoffspring similarity at different levels of personality traits, comparing expectations based on evolutionary and goodness-of-fit perspectives. This study combined a confirmatory and exploratory approach. The hypotheses were theory driven and confirmatory, while the comparison of different models was exploratory as we did not have a priori expectations which specific models might best fit our data. It was expected that from an evolutionary perspective, (1a) similarity in personality characteristics was beneficial regardless of the type of personality trait or the level of the personality trait. However, from a circumplex or goodnessof-fit perspective, (1b) similarity in negative traits (i.e. Neuroticism) was expected to have a negative impact on externalizing problems. Last, from an evolutionary perspective, (2) similarity effects were expected to be stronger for fathers than for mothers.
# Methods
## Participants and procedure
This study is part of the TRacking Adolescents' Individual Lives Survey (TRAILS), an ongoing prospective cohort study based on a sample representative of the Dutch population, investigating the emotional, social, and mental development from preadolescence into adulthood. Parental informed consent was obtained after the procedures had been fully explained. Detailed information about sample selection and analysis of non-response bias has been reported elsewhere [bib_ref] Cohort profile: The dutch 'TRacking adolescents' individual lives, Huisman [/bib_ref] Participants included 1587 adolescents who had filled out the personality questionnaire at age 16.2 (SD = 0.7, 51.7% girls, Time 3 of the TRAILS study) along with both their biological mother and (self-reported) biological father, from here on referred to as mother and father. This selection creates a relatively homogeneous group with respect to parental influence during childhood. At age 19.0 (SD = 0.5, 53.0% girls, Time 4 of the TRAILS study), 1488 participants (93.8%) filled out the externalizing problems questionnaire. Analyses were thus based on 1587 participants starting at age 16, when personality was first assessed.
## Measures
Personality (Age 16). To assess adolescent and parent personality, we used six facets of the Revised Neuroticism-Extroversion-Openness Personality-Inventory (NEO-PI-R), measured at age 16. The NEO-PI-R is a personality questionnaire consisting of 30 facet scales covering the Five-Factor Model of personality. Due to time constraints during data collection, facets were a priori selected based on their relevance to behavioral problems (e.g. [bib_ref] Personality, antisocial behavior, and aggression: A meta-analytic review, Jones [/bib_ref]. From the broad domain of Neuroticism, we used the facets anger hostility (Cronbach's α = .71), impulsivity (α = .51), and vulnerability (α = .77). Anger hostility relates to the tendency to experience anger and frustration, impulsiveness to low inhibition control and a strong activating response to cravings and urges, and vulnerability to susceptibility to stress. From the broad domain of Extraversion, we used the facets assertiveness (α = .75) and excitement-seeking (α = .58). Assertiveness reflects social dominance; excitement-seeking the need for high-intensity stimulation. From the broad domain of Conscientiousness, we used the facet self-discipline (α = .76), which measures the capacity to begin and complete tasks despite distractions. Available answers ranged from 1 (fully disagree) to 5 (fully agree). After recoding reversed items, facet scores were each based on the mean of eight questions.
Externalizing problems (Age 19). The Adult Self Report (ASR) has been widely used to assess self-report symptom dimensions. Symptom dimensions in the externalizing domain (α = .89) that are covered by the ASR are aggression and delinquent behavior. The mean score of 29 items was used, based on a three-point Likert scale as 0 (not true) to 2 (very or often true).
# Analysis strategy
In order to assess the joint impact of parent and offspring personality on externalizing problems, it is important to take levels of personality into account. Similarity patterns, also called fit patterns, have been developed to assess different types of similarity between two predictor variables. Such patterns are based on two main assumptions. First, there is an optimal match between two variables such as parent and offspring personality traits. Second, deviation from this optimal match leads to less optimal outcomes and bigger deviations will have more impact on the outcomes. Therefore, using similarity patterns, it can be estimated whether there is an optimal level of similarity in personality when predicting offspring's externalizing problems. Polynomial regression analysis can be used to compare several types of similarity patterns.
Polynomial regression analyses investigate linear effects of predictor variables, quadratic effects of predictor variables, and effects of the interaction between the predictor variables. Specifically, an intercept (b0), a linear (b1), and quadratic (b3) effect of the offspring, a linear (b2) and quadratic effect of the parent (b5), and an interaction between the linear effects of parent and offspring (b4) are estimated. Due to the combination of quadratic terms and an interaction term, interpretations of polynomial regressions are notoriously difficult. To facilitate interpretation, Response Surface Analyses have been developed (see [bib_ref] On the use of Polynomial Regression Equations as an alternative to difference..., Edwards [/bib_ref] [bib_ref] Polynomial regression with response surface analysis: A powerful approach for examining moderation..., Shanock [/bib_ref].
Response surface analyses provide a visual representation of the outcomes of polynomial regressions (see [fig_ref] Figure 1: Full polynomial regression analysis [/fig_ref] , based on similarity (or congruence) and dissimilarity (or incongruence) between two variables. The x-axis indicates the level of offspring's angry hostility, the y-axis indicates the level of parent angry hostility, and the z-axis indicates the level of offspring's externalizing problems. The dots in the figure represent participants, half of the participants are within the black line (bag plot) on the surface of [fig_ref] Figure 1: Full polynomial regression analysis [/fig_ref] and half of the participants are outside of this line. Two parameters (a1 and a2) assess effects among a Line of Congruence, or the line of similarity. The Line of Congruence is an imaginary line where parent and offspring have similar scores. For example, in [fig_ref] Figure 1: Full polynomial regression analysis [/fig_ref] , this line would run from the near corner where both parent and offspring have low scores to the end at the far corner where both parent and offspring have high scores. These effects assess how externalizing behavior is associated with personality when parent and offspring have similar scores. They indicate a linear slope (a1) and quadratic slope (a2) of similarity of parent and offspring personality on externalizing problems. Thus, significant effects indicate that similarity of parent and offspring personality traits is associated with externalizing problems.
Other linear (a3) and quadratic (a4) terms indicate whether there is a dissimilarity effect of personality on externalizing problems, along a Line of Incongruence. This Line of Incongruence runs from the left corner where parents score high and offspring cores low, to the right corner where offspring scores high and parents score low. The linear slope effect (a3) indicates the likelihood for higher externalizing problems when the offspring scores higher than the parent on a personality trait. The quadratic effect (a4) indicates whether externalizing problems are especially likely at high or low levels of dissimilarity. Thus, significant effects indicate that dissimilarity in personality impacts externalizing problems.
One potential problem of polynomial regressions, however, is overfitting the data. Therefore,suggested five simpler fit models, which are nested under the full polynomial model and use fewer degrees of freedom. Two of these fit models are mainly targeted at incommensurable measures (i.e. not measured on a similar scale) or used when there are theoretical expectations that the variables have a dissimilar impact on the outcome variable. As that is not the case for this study, these two models were disregarded. The other three fit models were compared with the full polynomial regression model and regular regression models.
The first types of models assume that there is no main effect of parent or offspring personality on the outcome variable, but allow for (dis)similarity effects. Thus, the level of the personality trait does not affect externalizing problems, but it does matter how (dis)similar parent and offspring are in personality characteristics. These models are thus in line with assumptions based on the evolutionary model that similarity in personality affects externalizing problems regardless of the level of personality traits. The sub model shifted squared difference model (SSDQ) models an effect of (dis) similarity, but optimal levels of (dis)similarity do not have to be at numerical equality. Thus, this model takes into account that the optimal match might not be when both parent and offspring have exactly the same score but allows the optimal match to be off from the numerical equality (for example if the optimal match is when offspring scores higher than parents).
The second types of fit models also assume (dis)similarity effects, but they also take the impact of the level of parent and offspring personality on externalizing problems into account. Thus, it also models how at similar levels of personality these traits are associated with externalizing problems. First, the sub model basic rising ridge model (RR) assumes that there is a main effect of (dis)similarity but also an effect of personality at similar levels of parents and offspring personality when predicting externalizing problems. Again, the shifted version of the rising ridge model (SRR) takes into account that the optimal match might not be when both parent and offspring have the exact same score.
These effects were estimated using the RSA package in R, guidelines fromwere used for model selection. The main determinant for model selection was the corrected Akaike Information Criterion (AICc). Models with smaller AICc better fit the data. These weights can be compared using model weights called 'Akaike weights', which give the probability that a model is the best model of the candidate models, difference scores higher than two indicate significantly worse model fits. As AICc indices only indicate whether models are better compared to other models, rather than the absolute plausibility of models, R 2 adj should be used to assess the explained variance. If the explained variance (R 2 adj ) is significant, results can be interpreted. All variables were centered to facilitate interpretation. A score of zero thus means that participants had an average score, within their role (i.e. father, mother, or offspring). Positive scores indicate scoring higher than average on a personality trait while negative scores indicate scoring lower than average. [fig_ref] Table 1: Correlations between the main study variables [/fig_ref] shows the correlations between the main study variables, indicating that there was a significant, albeit modest, correlation between offspring's personality facets and the same father's or mother's personality facets; correlations ranged from 0.10 to 0.23. Offspring's, mother's, and father's anger hostility, impulsiveness, and excitement-seeking were positively correlated to offspring's externalizing problems, and self-discipline was negatively correlated to offspring's externalizing problems. Vulnerability was significantly positively correlated to offspring's externalizing problems for offspring and mother, but not for father, and assertiveness was not correlated to offspring's externalizing problems for offspring, mother, or father. [fig_ref] Table 2: Outcomes of the fit-analyses of offspring and parent personality predicting externalizing problems [/fig_ref] indicates that offspring externalizing problems was explained by all personality facets, as indicated by significant adjusted R 2 adj effects. Effect sizes range from 0.006 for the model including father's assertiveness to 0.185 for the model including father's anger hostility.
# Results
## Externalizing problems
Effects of anger hostility. Mother-offspring and fatheroffspring similarity on externalizing problems was best modeled (see [fig_ref] Table 3a: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref] by full polynomial regression models (mother effects: a1 = 0.023, SE = 0.002, p < .001; a2 = 0.001, SE = 0.001, p = .018; a3 = 0.013, SE = 0.002, p < .001; a4 = 0.000, SE = 0.001, p = .489 (n.s.); father effects: a1 = 0.022, SE = 0.002, p < .001; a2 = 0.001, SE = 0.001, p = .014; a3 = 0.013, SE = 0.001, p < .001; a4 = À0.001, SE = 0.001, p = .083 (n.s.)). [fig_ref] Figure 1: Full polynomial regression analysis [/fig_ref] shows these outcomes for the father-offspring effects; motheroffspring effects were similar (see . The x-axis indicates the level of offspring's angry hostility, the y-axis indicates the level of parent angry hostility, and the z-axis indicates the level of offspring's externalizing problems. The significant a1 and a2 effects indicate effects along the line of similarity; there is a linear and quadratic prediction from similarity in anger hostility on externalizing problems. An increase in anger hostility, when both parent and offspring have similar scores, of both parent and offspring is associated with an increase in externalizing problems, and this increase in externalizing problems tends to escalate at higher levels of anger hostility. Along the line of dissimilarity, the a4 effect was non-significant; the degree of dissimilarity did not impact externalizing behavior. However, as indicated by a positive a3 effect, the direction of dissimilarity did impact externalizing behavior. Effects were stronger when the offspring has higher anger hostility than the parent has rather than vice versa.
Effects of impulsivity for the mother-offspring similarity hypothesis were best modeled by offspring effects only (see [fig_ref] Table 3b: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref]. There was a significant linear (b1 = 0.019, SE = 0.002, p < .001) and quadratic effect (b3 = 0.001, SE = 0.000, p = .004) of offspring's impulsivity on offspring's externalizing problems. Thus, independent of mother's impulsivity, offspring's impulsivity was positively associated with externalizing problems and the association increased at higher levels of impulsivity. Father-offspring similarity was best modelled by a full polynomial regression model (a1 = 0.026, SE = 0.002, p < .001; a2 = 0.003, SE = 0.001, p < .001; a3 = 0.012, SE = 0.003, p < .001; a4 = 0.000, SE = 0.000, p = .621 (n. s.)). The significant a1 and a2 effects indicate effects along the line of similarity; there was a linear and quadratic prediction from similarity in impulsivity on externalizing problems. An increase in impulsivity at similar levels of impulsivity of both father and offspring was associated with an increase in externalizing problems. Moreover, this effect tends to escalate at higher levels of impulsivity. Along the line of dissimilarity, the a4 effect was non-significant. Therefore, the degree of dissimilarity did not impact externalizing behavior. However, as indicated by a positive a3 effect, the direction of dissimilarity did matter. Effects were stronger when the offspring has higher impulsivity than the parent has rather than vice versa. Effects of vulnerability. Mother-offspring and fatheroffspring similarity on externalizing problems were best modelled by effects of the offspring's vulnerability only (see [fig_ref] Table 3c: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref]. Both for the mother-offspring and the fatheroffspring model, there was a linear effect (b1 = 0.043, SE = 0.006, p < .001) and a quadratic effect (b3 = 0.013, SE = 0.00, p = .001) for offspring's vulnerability predicting externalizing problems. In sum, independent on the vulnerability of mother or father, offspring's vulnerability is positively associated with externalizing problems, and this tends to escalate at higher levels of vulnerability.
Effects of assertiveness mother-offspring similarity was best modeled (see [fig_ref] Table 3d: Notes [/fig_ref] by a Rising Ridge model. Although the Rising Ridge model had the lowest AICc, other models such as the full polynomial model or offspring only effects were equally good candidate models; as the Delta AICc was less than two. This Rising Ridge model indicates that more similarity is associated with less externalizing problems , regardless of the level of assertiveness at which mother and offspring were similar. There was no significant linear (a1 = 0.003, SE = 0.002, p = .092 (n.s.)) or quadratic (a2 = 0.000, SE = 0.000, p = 1.000 (n.s.)) effect of assertiveness on externalizing problems along the line of similarity. Only the a4 effect was significant (a4 = 0.001, SE = 0.001, p = .023), indicating more similarity in assertiveness of mother and Anger hostility (see [fig_ref] Table 3a: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref] for more details) Mother offspring Full polynomial model 0.179*** Father offspring Full polynomial model 0.185*** Impulsivity (see [fig_ref] Table 3b: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref] for more details) Mother offspring
Only offspring effects 0.123*** Father offspring Full polynomial model 0.156*** Vulnerability (see [fig_ref] Table 3c: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref] for more details)
Mother offspring Only offspring effects 0.066*** Father offspring
Only offspring effects 0.066*** Assertiveness (see [fig_ref] Table 3d: Notes [/fig_ref] for more details)
Mother offspring Rising Ridge model 0.007*** Father offspring
Only offspring effects 0.006** Excitement seeking (see [fig_ref] Table 3e: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref] for more details)
Mother offspring Only offspring effects 0.031*** Father offspring Full polynomial model 0.043*** Self-discipline (seefor more details)
Mother offspring Full polynomial model 0.107*** Father offspring
Only offspring effects 0.096*** Note: **p < .01. ***p < .001. offspring is associated with less externalizing problems. Father-offspring effects of assertiveness were best modeled by offspring effects only. The linear effect was nonsignificant (b1 = 0.03, SE = 0.054, p = .054 (n.s.)), but there was a small quadratic effect of offspring assertiveness on externalizing problems (b3 = 0.000, SE = 0.000, p = .026). This indicates that especially offspring with lower or higher levels of assertiveness experienced more externalizing problems compared to their peers who had average assertiveness (see . Effects of excitement seeking. Mother-offspring similarity on externalizing problems were best modeled (see [fig_ref] Table 3e: Model comparison for the prediction of externalizing problems by mother, father, and... [/fig_ref] by only offspring's effects. In this model, externalizing problems were significantly predicted by offspring's excitement seeking (b1 = 0.009, SE = 0.001, p < .001), and the quadratic effect of offspring's excitement seeking (b3 = 0.001, SE = 0.000, p = .002). Thus, independent on mother's excitement seeking, the offspring excitement seeking is positively associated with externalizing problems and has a tendency to escalate at higher levels of offspring excitement seeking. Effects of father-offspring similarity in excitement seeking on externalizing problems was best modeled by the full polynomial regression model (a1 = 0.013, SE = 0.002, p < .001; a2 = 0.000 SE = 0.000, p = .431 (n.s.); a3 = 0.002, SE = 0.003, p = .461 (n.s.); a4 = 0.001, SE = 0.001, p = .129 (n.s.)). The significant a1 effect indicates that along the line of similarity, there is a linear prediction from similarity in excitement seeking on externalizing problems. This effect was linear rather than quadratic, as indicated by the non-significant a2 effect. Thus, at similar levels of excitement seeking, there was a positive association linear association between excitement seeking and externalizing problems. Along the line of dissimilarity, the a4 effect was non-significant, thus the degree of dissimilarity did not impact externalizing behavior. Furthermore, as indicated by a non-significant a3 effect, effects did not depend on the direction of dissimilarity. It did not matter whether the father or offspring had higher excitement seeking.
Effects of self-discipline mother-offspring similarity on externalizing problems were best modeled (seeby a full polynomial regression model (a1 = À0.018, SE = 0.002, p < .001; a2 = 0.000, SE = 0.000, p = .304 (n.s).; a3 = À0.008, SE = 0.002, p < .001; a4 = À0.001, SE = 0.00, p = .018). shows that the significant negative a1 and the non-significant a2 effects indicate that there is a negative linear effect along the line of similarity of self-discipline on externalizing problems. An increase in self-discipline, while mother and offspring have similar self-discipline, of mother and offspring is associated with a decrease in externalizing problems. The significant negative a3 effects indicate that the direction of difference in self-discipline matters. Externalizing problems are more likely when the mother has higher self-discipline compared to the offspring, rather than vice versa. The negative a4 effect, indicating the curvature along the line of dissimilarity, was also significant. This indicates that externalizing problems are especially likely when mother and offspring have a similar level of self-discipline. For father-offspring similarity, only the offspring's self-discipline predicted offspring's externalizing problems. There was a linear effect of offspring's self-discipline on externalizing problems (b1 = À0.013, SE = 0.001, p < .001), but the quadratic effect was non-significant (b3 = 0.000, SE = 0.000, p = .088 (n. s.)). Thus, independent of father's self-discipline, offspring's self-discipline is negatively associated with externalizing problems.
# Discussion
This study set out to explore the effects of parent-offspring personality similarity on externalizing problems. Hypotheses based on an evolutionary and on a goodness-of-fit perspective were compared, using sophisticated analyses and response surface plots. Findings indicated that, in contrast to earlier studies (e.g. [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref] , similarity was mostly unrelated to offspring externalizing problems. One exception was offspring-mother similarity in assertiveness, a facet of Extraversion. Similarity between mother and offspring was associated with fewer externalizing problems, independent of the level of assertiveness. Notably, similarity in mother-offspring self-discipline was negatively rather than positively associated with externalizing problems. With an increased similarity in self-discipline for mother and offspring, the chance of later externalizing problems for the offspring increased. Hypotheses based on an evolutionary perspective therefore received no support: Similarity was not beneficial regardless of the trait or the level of the trait nor did similarity matter more for fathers than for mothers. There was limited support for the hypothesis based on a goodness-of-fit or interpersonal circumplex perspective. Although similarity was beneficial for a facet of Extraversion, a facet which is associated with lower externalizing problems, it was detrimental for a facet of Conscientiousness which is also associated with fewer externalizing problems. However, other findings indicated that both effects of parent's and offspring's personality matter, and at similar levels of personality these personality facets were associated with externalizing problems.
## Personality similarity and externalizing problems
Three facets of Neuroticism were investigated: anger hostility, impulsivity, and vulnerability. Offspring's Neuroticism predicted offspring's externalizing problems, in line with previous findings (e.g. [bib_ref] Longitudinal associations between personality traits and problem behavior symptoms in adolescence, Klimstra [/bib_ref] [bib_ref] Structural models of personality and their relation to antisocial behavior: A meta-analytic..., Miller [/bib_ref]. Based on the goodness-of-fit perspective, it was expected that similarity at higher levels of parent and offspring anger hostility was associated with more externalizing problems. However, rather than an effect of similarity, at similar levels of angry hostility of both parent and offspring predicted externalizing problems. Furthermore, externalizing problems were more likely when the offspring had higher anger hostility than the parent did rather than vice versa. Moreover, for mother-offspring impulsivity and both mother and father-offspring vulnerability, only the offspring's characteristics affected offspring's externalizing problems. Higher levels of impulsivity and vulnerability were associated with more externalizing problems. For father-offspring impulsivity, both father and offspring personality were associated with externalizing problems at similar levels of this facet. Some previous studies did not find a significant association between children's Neuroticism and externalizing problems (e.g. [bib_ref] The "little five": Exploring the nomological network of the five-factor model of..., John [/bib_ref]. Possibly, especially parent's angry hostility is important in explaining the association between Neuroticism and offspring's externalizing problems. Broader indicators of Neuroticism might fail to detect effects based on more specific facets of personality. Angry hostility, impulsivity, and vulnerability have been associated with externalizing problems, while other facets of Neuroticism such as anxiety, or self-consciousness have not always been associated with externalizing problems (e.g. [bib_ref] Personality, antisocial behavior, and aggression: A meta-analytic review, Jones [/bib_ref] [bib_ref] Examining antisocial behavior through the lens of the Five Factor Model of..., Miller [/bib_ref]. In sum, for the facets anger hostility, impulsivity, and vulnerability of Neuroticism there were almost no differences between the findings for father-offspring and mother-offspring similarity, and higher scores of the offspring on all investigated Neuroticism facets were related to more externalizing problems.
Two facets of Extraversion were investigated: assertiveness and excitement seeking. In line with previous studies (e.g. [bib_ref] The "little five": Exploring the nomological network of the five-factor model of..., John [/bib_ref] [bib_ref] Personality, antisocial behavior, and aggression: A meta-analytic review, Jones [/bib_ref] [bib_ref] Structural models of personality and their relation to antisocial behavior: A meta-analytic..., Miller [/bib_ref] , it was expected that Extraversion would be weakly or not associated with externalizing problems. In line with these expectations, assertiveness and excitement seeking only explained a small, although statistically significant, portion of externalizing problems. The explained variances were mainly based on offspring-only effects, which means that the offspring's (and not the parent's) Extraversion predicted the offspring's future externalizing problems. Only for excitement seeking both father's and offspring's excitement seeking mattered, excitement seeking at similar levels for both father's and offspring's excitement seeking was positively associated with offspring's externalizing problems. One exception was mother-offspring similarity in assertiveness. Offspring who differed from their mother in their level of assertiveness were more likely to experience externalizing problems, compared to offspring who were more similar to their mother. There was, however, no direct effect of mother's or offspring's level of assertiveness in predicting externalizing problems. Therefore, the association between facets of Extraversion and offspring's externalizing problems depends both on the facet of Extraversion and the parent being studied.
One facet of Conscientiousness was studied: selfdiscipline. Conscientiousness has been associated with less externalizing problems (e.g. [bib_ref] The "little five": Exploring the nomological network of the five-factor model of..., John [/bib_ref] [bib_ref] Personality, antisocial behavior, and aggression: A meta-analytic review, Jones [/bib_ref] [bib_ref] Longitudinal associations between personality traits and problem behavior symptoms in adolescence, Klimstra [/bib_ref] [bib_ref] Structural models of personality and their relation to antisocial behavior: A meta-analytic..., Miller [/bib_ref] [bib_ref] Examining antisocial behavior through the lens of the Five Factor Model of..., Miller [/bib_ref]. An increase in self-discipline at similar levels of this facet of Conscientiousness for both mother and offspring was associated with less externalizing problems. Furthermore, externalizing problems were associated with similarity for conscientiousness. More rather than less similarity of self-discipline was associated with externalizing problems. Moreover, the effects of offspring's self-discipline were larger than the effects of mother's self-discipline in explaining externalizing problems. Father's self-discipline was not associated with offspring's externalizing problems. Thus, self-discipline, a facet of Conscientiousness, was associated with less externalizing problems, and this was based on mother's and offspring's level of Conscientiousness.
In sum, rather than parent-offspring similarity, offspring's personality at the age of 16 seems to be most important in explaining offspring's externalizing problems at the age of 19. With an exception of the effects of mother and offspring in assertiveness, all facets of offspring's personality were associated with future externalizing problems. Similarity only predicted externalizing behavior for mother's and offspring's assertiveness and self-discipline. For the other facets, both parents and offspring or only the offspring's personality was associated with offspring's externalizing problems. Therefore, there is little support for the two hypotheses based on an evolutionary perspective: Similarity in parent-offspring personality was mostly not beneficial, and for self-discipline even detrimental, and motheroffspring rather than father-offspring similarity in personality was associated with externalizing problems. Possibly mothers' personality, rather than fathers' personality, has a higher impact on offspring's future externalizing problems as mothers, in general, spend more time with their offspring. In light of a goodness-of-fit or interpersonal circumplex perspective, only for a facet of Extraversion similarity was associated with less externalizing problems and for a facet of Conscientiousness similarity was even associated with more externalizing problems. Thus, only for one facet negatively associated with externalizing problems similarity was associated with less externalizing problems, while for another facet negatively associated with externalizing problems similarity was even detrimal. For other facets, similarity was not associated with externalizing problems. Thus, there was little support for the hypotheses based on a goodness-of-fit or interpersonal circumplex perspective. However, for quite some facets of personality traits, the offspring's personality mattered most in predicting externalizing problems. Therefore, regardless of parents' personality, the offspring personality affected future externalizing problems.
# Strengths and limitations
The current study was the first to use both polynomial regression analyses and similarity fit indices to identify how parent-offspring personality similarity best predicted offspring's externalizing problems. Polynomial regression analyses overcome two major shortcomings of earlier studies. First, similarity was not expected to be equal at different levels of personality traits; the impact of similarity at higher and lower levels of personality traits was allowed to be different. Second, different scores were not expected to be symmetrical, effects of parents having a higher score on a personality trait than their offspring was not expected to have the same impact as offspring having a higher score on a personality trait than the parent. Third, this study combined a confirmatory and exploratory approach. The hypotheses were theory driven and confirmatory, while the comparison of different models was exploratory as we did not a priori have expectations which specific models might best fit our data. Future studies might be able to also use confirmatory approaches to a priori identify the best way to model effects of parents offspring similarity. Furthermore, our findings were based on a large representative dataset of Dutch adolescents.
There were also some limitations to this study. First, as [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref] investigated early adolescents (around 13 years old), age differences between the samples might help explain why we did not find many effects of parentoffspring similarity. TRAILS data only assessed Big Five facets of parent and offspring personality when participants were 16 years old. Parents might have more influence on offspring during early rather than late adolescence. Second, this study did not control for externalizing behavior at age 16, while predicting externalizing behavior at age 19 as this is not possible in the current fSRM package. Moreover, the measures for externalizing problems differed between the assessments in the TRAILS study. As analyses were already highly complex, taking into account externalizing problems at age 16 may further complicate the model beyond the current aims of the study. Longitudinal studies investigating the interplay between parent and offspring personality and externalizing problems might shed light on more complex longitudinal developments. For example, parenting styles might have a different impact depending on offspring's personality [bib_ref] The additive and interactive effects of parenting and children's personality on externalizing..., Prinzie [/bib_ref]. Moreover, offspring's problems might interact with the temperament of mothers (Atzaba-Poria, [bib_ref] It takes more than one for parenting: How do maternal temperament and..., Atzaba-Poria [/bib_ref]. Also, testing multiple analyses in such a large sample increases the possibility of chance findings. However, we did take effect size into account to assess the magnitude of our findings. Last, underlying processes through which similarity in personality might impact externalizing behavior were not investigated. Future studies might aim to further disentangle this process, where it might be possible that dissimilarity in personality leads to a worse relationship quality, which in turn affects externalizing behavior development (see [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref]. Although [bib_ref] Parent-offspring similarity in personality and adolescents' problem behaviour, Van Tuijl [/bib_ref] did not find relationship quality to impact effects based on personality similarity, possibly the use of complex analyses such as polynomial regression analyses might help uncover such processes.
# Conclusion
This study investigated different levels of parent-offspring personality similarity, and how these were associated with offspring's externalizing problems. From an evolutionary and a goodness-of-fit perspective similarity in personality was expected to be beneficial, at least at certain levels of traits. Therefore, they were expected to be associated with fewer externalizing problems for the offspring. However, hardly any similarity effects were found. Therefore, there is little support for assumptions based on either model. Although it might be argued that similarity leads to a more optimal match between parents and offspring, our findings point in the direction of additive effects of parent and offspring personality on the development of externalizing problems. Adolescents were especially likely to experience externalizing problems when both they and their parents had similar high levels of negative personality traits. Moreover, for several facets of personality, only the offspring's personality had an impact on offspring's externalizing problems. Therefore, interventions might identify adolescents based on their own personality as for almost all traits this had an impact on externalizing problems; including the personality of parents might only be beneficial when studying certain facets of personality traits.
[fig] Figure 1: Full polynomial regression analysis: Father-offspring similarity in anger hostility is predicting externalizing problems. [Colour figure can be viewed at wileyonlinelibrary.com] [/fig]
[fig] Figure 2, Figure 3: Full polynomial regression analysis: Mother-offspring similarity in anger hostility is predicting externalizing problems. [Colour figure can be viewed at wileyonlinelibrary.com] Rising Ridge model: Mother-offspring similarity in assertiveness is associated with less externalizing problems. [Colour figure can be viewed at wileyonlinelibrary.com] [/fig]
[fig] Figure 4, Figure 5: Full polynomial regression analysis: Father-offspring similarity in assertiveness predicting externalizing problems. [Colour figure can be viewed at wileyonlinelibrary.com] Full polynomial regression analysis: Mother-offspring similarity in self-discipline predicting externalizing problems. [Colour figure can be viewed at wileyonlinelibrary.com] [/fig]
[table] Table 1: Correlations between the main study variables [/table]
[table] Table 2: Outcomes of the fit-analyses of offspring and parent personality predicting externalizing problems [/table]
[table] Table 3a: Model comparison for the prediction of externalizing problems by mother, father, and offspring anger hostility. Ordered by delta AICc [/table]
[table] Table 3b: Model comparison for the prediction of externalizing problems by mother, father, and offspring impulsivity. Ordered by delta AICc Notes: K, number of parameters; AICc, corrected Akaike Information Criterion; CFI, Comparative fit index; R 2 , variance explained of the model; pmodel, p value [/table]
[table] Table 3c: Model comparison for the prediction of externalizing problems by mother, father, and offspring vulnerability. Ordered by delta AICc [/table]
[table] Table 3d: Notes: k, number of parameters; AICc, corrected Akaike Information Criterion; CFI, Comparative fit index; R 2 , variance explained of the model; pmodel, p value [/table]
[table] Table 3e: Model comparison for the prediction of externalizing problems by mother, father, and offspring excitement seeking. [/table]
|
10.1038/s41597-022-01396-1
|
CCBY
|
9184483
|
35680901
|
s2orc_pubmed_articles
|
A global database of woody tissue carbon concentrations
Woody tissue carbon (C) concentration is a key wood trait necessary for accurately estimating forest C stocks and fluxes, which also varies widely across species and biomes. However, coarse approximations of woody tissue C (e.g., 50%) remain commonplace in forest C estimation and reporting protocols, despite leading to substantial errors in forest C estimates. Here, we describe the Global Woody Tissue Carbon Concentration Database (GLOWCAD): a database containing 3,676 individual records of woody tissue C concentrations from 864 tree species. Woody tissue C concentration data-i.e., the mass of C per unit dry mass-were obtained from live and dead woody tissues from 130 peer-reviewed sources published between 1980-2020. Auxiliary data for each observation include tissue type, as well as decay class and size characteristics for dead wood. In GLOWCAD, 1,242 data points are associated with geographic coordinates, and are therefore presented alongside 46 standardized bioclimatic variables extracted from climate databases. GLOWCAD represents the largest available woody tissue C concentration database, and informs studies on forest C estimation, as well as analyses evaluating the extent, causes, and consequences of inter-and intraspecific variation in wood chemical traits.
# Background & summary
Forests play a critical role in the global carbon (C) cycle, with the world's forests storing an estimated 861 ± 66 Pg C across tropical (~471 Pg C), boreal (~272 Pg C), and temperate forest ecosystems (~119 Pg C) [bib_ref] A large and persistent carbon sink in the world's forests, Pan [/bib_ref]. At the same time, C cycling in forested biomes is highly dynamic and transient, with estimates indicating that forests sequester between ~2.15 to 2.4 Pg C y −1 globally on average [bib_ref] A large and persistent carbon sink in the world's forests, Pan [/bib_ref] [bib_ref] Role of forest regrowth in global carbon sink dynamics, Pugh [/bib_ref]. Throughout the 2000s, structurally intact old-growth forests accounted for ~0.85 Pg C y −1 , while C sequestration was ~1.30 Pg C y −1 in secondary forests 2 . Tropical regions are particularly important in sequestering atmospheric carbon dioxide (CO 2 ) in both regenerating [bib_ref] Carbon sequestration potential of second-growth forest regeneration in the Latin American tropics, Chazdon [/bib_ref] [bib_ref] Biomass resilience of Neotropical secondary forests, Poorter [/bib_ref] [bib_ref] Mapping carbon accumulation potential from global natural forest regrowth, Cook-Patton [/bib_ref] and intact forests [bib_ref] A large and persistent carbon sink in the world's forests, Pan [/bib_ref] [bib_ref] Increasing carbon storage in intact African tropical forests, Lewis [/bib_ref] [bib_ref] Asynchronous carbon sink saturation in African and Amazonian tropical forests, Hubau [/bib_ref]. Nevertheless, recent analyses from both temperate [bib_ref] First signs of carbon sink saturation in European forest biomass, Nabuurs [/bib_ref] and tropical regions [bib_ref] Asynchronous carbon sink saturation in African and Amazonian tropical forests, Hubau [/bib_ref] have indicated that the magnitude of C sinks in old-growth forests are declining.
The amount of C stored within, and transferred to and from, trees and forests have been estimated from fieldor remote-sensing-based observations of tree attributes, which are used to obtain estimates of tree-or forest aboveground biomass (AGB) [bib_ref] A large and persistent carbon sink in the world's forests, Pan [/bib_ref] [bib_ref] Changes in forest production, biomass and carbon: results from the 2015 UN..., Köhl [/bib_ref] [bib_ref] High-resolution forest carbon stocks and emissions in the Amazon, Asner [/bib_ref]. Estimates of AGB are then converted into C estimates by multiplying these values by a woody tissue C concentration, commonly referred to in the literature as a C fraction 13-16 (i.e., the mass of C per unit dry mass). Accurate woody tissue C concentration data are therefore critical in (1) accurately estimating terrestrial forest C budgets and sequestration rates 17 , (2) estimating the C emissions associated with land-use change 18 , and ultimately (3) informing decision-making related to the identification of forests with high C storage capacity 11 . Indeed, the Intergovernmental Panel on Climate Change's (IPCC) Tier 3 C accounting protocols suggests that a "specific carbon fraction…should also be incorporated" when estimating C stocks and fluxes in AGB 13 . Moreover, woody tissue C concentration data can be employed in studies on the abiotic or biotic predictors of variation in -and possible adaptive significance of -wood chemical traits across tree species 19,20 , as well as evaluating the role that different sample extraction, preparation, and analytical methods have on wood C fractions 17 . Owing at least in part to a lack of large woody tissue C datasets, these research areas have received relatively little attention in comparison to other suites of plant traits 21 .
To date, most C estimation and reporting protocols use generic approximations of woody tissue C concentrations (namely, an assumption that 50% of AGB is comprised of C 13 ), which has led to substantial systematic errors in forest C estimates. For example, our recent analyses indicated that generic woody tissue C fractions overestimate C stocks by approximately 8.9% in tropical forests 19 . Similar issues exist for the accounting of C stocks and fluxes in dead wood, with recent analyses indicating that generic dead wood C fractions may result in dead wood C pools being overestimated by ~3.0 Pg C globally 22 . Although multiple studies evaluating woody tissue C concentrations in trees globally through field-or meta-analyses now exist 19,23-25 , there is no single woody tissue C data repository to aid researchers in accessing and using these data.
To address these issues, we created and describe here the "Global Woody Tissue Carbon Concentration Database" (hereafter GLOWCAD 26 ), which contains woody tissue C concentrations measured on live and dead tree tissues, spanning all forested biomes. By organizing and standardizing data from a range of taxonomic groups and woody tissue-types (described below), GLOWCAD represents a resource that helps improve our understanding of both global forest C dynamics and inter-and intraspecific variability in wood chemical traits. GLOWCAD only includes data from peer-reviewed sources. In addition to associated information on the taxonomic identities and woody tissue types for each woody tissue C data point, GLOWCAD includes geographical and associated bioclimatic data obtained from climate databases 27 .
Data records in GLOWCAD are stored in 3 easy-to-use Comma Separated Values (.csv) spreadsheets [fig_ref] Figure 1: Structure of Global Woody Tissue Carbon Concentration Database [/fig_ref]. All spreadsheets comprise plain text, with the first spreadsheet (titled "Wood Carbon Database") containing the core data (i.e., woody tissue C concentrations and related information), while the other spreadsheets provide descriptive supporting information including references (titled "References") and column descriptions (titled "Column Descriptions"). GLOWCAD has been made publicly available through the Dryad Digital Repository, with existing applications including studies on: (1) woody tissue C concentrations variation across live trees 19,23,25 ; (2) variation in dead woody tissue C concentrations 22 ; (3) relationships between woody tissue C concentrations and tree life-history strategies 19,22 ; and (4) climate correlates of woody tissue C concentrations in trees 28 .
# Methods
Literature review. Data compilation expanded earlier versions of the GLOWCAD first initiated in 2012 25 , and more recently published in 2018 19 and 2021 29 . GLOWCAD is therefore based on a systematic search on primary literature of all peer-reviewed papers that cited previously published studies on woody tissue C concentrations 19,23-25 . We searched key terms "carbon", "tree", "wood carbon", "coarse woody debris", "dead wood", and "wood nutrient", as well as "carbon" alongside major tree tissue types (including "wood", "bark", "root" and "stem"), within four web-based platforms (Google Scholar, Web of Science, Web of Knowledge, and Scopus), in order to identify peer-reviewed publications that present species-specific woody tissue C concentration data.
In addition to peer-reviewed papers, other sources of data included in GLOWCAD include the TRY Plant Trait Database (v. 5.0) Wood C data attributes. To be included in GLOWCAD, the species-specific binomial nomenclature and tissue-specific information for each woody tissue C sample was required. A detailed field and lab methodology was also necessary, in order to maximize our sample size while permitting reliable species-and tissue-specific analysis. Where a single paper contained multiple tissue-and species-specific woody tissue C records, all the published values were recorded. In the majority of cases, woody tissue C data were extracted directly from published tables or from supplementary data of the articles. In instances where woody tissue C data were published as figures, the data was extracted using the WebPlotDigitizer v4.2 software 41 . If species-by-tissue-specific woody tissue C data were not published, the corresponding authors were contacted to provide data.
Each published woody tissue C record was then classified according to the forest biome in which it was sampled. A small number of studies (e.g. 42 ) presented both boreal and temperate data, which were differentiated in our database based on the sampling location coupled with a consultation of species distribution maps. Species taxonomy was first recorded as presented as in published articles. A final list of taxa was then compared with, and resolved according to, the Taxonomic Name Resolution Service v. 4.0 43 . Both original and resolved taxonomy is maintained in GLOWCAD. Inclusion of new published data was halted as of Dec. 31, 2020.
Dead wood C data attributes. When classifying dead wood data, we considered three primary factors associated with woody tissue decomposition and related chemical change: A) decay class (DC), B) position, and C) size (diameter and length). In the majority of publications, dead woody tissue C values were reported along a conventional 1-5 DC scale. These values were included in GLOWCAD as published, while noting the DC scale employed. In cases where DC was reported as a two-category range (e.g. DC 1-2), the higher DC was included in GLOWCAD. In cases where a multiple category DC was presented (e.g. DC 3-5), the middle DC value was used in GLOWCAD. In the few instances DC was reported along a 0-5 point scale (where DC of 0 was defined as dead and not live wood), dead wood reported with a DC of 0 was classified as DC 1. Lastly, in a subset of papers the number of years since tree death (instead of DC) was reported. In these cases, years since death were converted to DC based on published decay class transition metrics (e.g. 44 ). When classifying position of dead wood, "standing" referred to snags and suspended woody debris, and "downed" referred to anything sampled from the forest floor. The default position was "downed" for the few publications that did not specify position.
GLoWCaD structure. The structure of GLOWCAD is simple to navigate [fig_ref] Figure 1: Structure of Global Woody Tissue Carbon Concentration Database [/fig_ref]. Within GLOWCAD, all the woody tissue C data is present under the "Wood Carbon Database" spreadsheet. In this spreadsheet, a unique number (i.e., 'unique.id') of all woody tissue C data is specified beside the reference from which it was obtained. The value of the 'reference.number' corresponds to the detailed citation presented in the "References" spreadsheet, which links the 'reference.number' with the author(s)' name and publication year, title, journal, volume, issue, and pages.
When inputting woody tissue C data from publications into GLOWCAD, the latitude and longitude were also recorded in the database when explicitly stated in the original publication. General climate information such as mean annual temperature (MAT) and mean annual precipitation (MAP) of the study region were recorded as an average. The study regions' latitude and longitude were also used to further describe its climate with WorldClim (v.2) data 27 . However, when a range of geographic coordinates or a map was provided, climate data were not generated from these since averages MAT and MAP may be imprecise. We used MAT and MAP obtained from WorldClim (v.2) to label the study region's dominant Whittaker biome 45 , and therefore categorize the region as one of Boreal forest, Subtropical desert, Temperate grassland/desert, Temperate rain forest, Temperate seasonal forest, Tropical rain forest, Tropical seasonal forest/savanna, or Woodland/shrubland. A list containing the details collected from each publication is presented (Supplementary . Bioclimatic variables and other climate data associated with each study location were retrieved from WorldClim (v.2) 27 and added alongside woody tissue C data (Supplementary .
Previous versions of GLOWCAD. GLOWCAD is the fourth iteration of the woody tissue C dataset, though these earlier versions did not use the same acronym, and contained differing sets/ subsets of data based on different research questions. Three earlier versions are publicly hosted in the TRY Plant Trait Database, such that: 1) the first version contained n = 973 observations of dead wood C only, from 121 species; 2) the second version contained n = 1,145 observations of live woody tissue C only, from 415 species paired with geographic coordinates www.nature.com/scientificdata www.nature.com/scientificdata/ and climate data; and 3) the third version contained n = 2,432 observations of live woody tissue C only, from 636 species including all of the observations of the previous version .
GLOWCAD is a single data product which consolidates the dead and live woody tissue C observations of all prior iterations (where n = 3,405), and includes 271 new woody tissue C observations from 10 additional publications 31-40,46 . In sum, n = 3,676 data points in the GLOWCAD version described here. Unlike previous versions, GLOWCAD also includes information on growth habit or 'woodiness' (described below) and the original binomial nomenclatures as listed in their publications.
## Data records
GLOWCAD is stored in .csv format at the Dryad Digital Repository (https://doi.org/10.5061/dryad.18931zcxk). Data outputs consist of a single database of 3,676 woody tissue C observations from 130 sources 17,23,24,30-40,42,44,46-159 published between 1980 and 2020 , which includes C concentrations of woody tissues from 864 tree/ shrub species sampled across all continents except Antarctica . While data exists from papers published since 1980, the large majority (86%) of the data in GLOWCAD (n = 3,154 data points) is derived from sources published in or after 2010 [fig_ref] Figure 1: Structure of Global Woody Tissue Carbon Concentration Database [/fig_ref].
GLOWCAD includes woody tissue C values from 414 genera and 107 families, with the Pinaceae (n = 927 data points), Fabaceae (n = 383 data points), Fagaceae (n = 335 data points), Cupressaceae (n = 159 data points) and Betulaceae (n = 146 data points) being most well represented . Across biomes, most woody tissue C data in GLOWCAD are derived from tropical forests (n = 1,513 data points), followed by www.nature.com/scientificdata www.nature.com/scientificdata/ temperate (n = 1,202 data points), subtropical/ Mediterranean (n = 518 data points) and boreal (n = 301 data points) forests. Across the entire database, woody tissue C ranged from 18.4-75.1% [fig_ref] Figure 4: Variation in woody tissue C concentrations [/fig_ref].
In GLOWCAD, 73% of data points were obtained from woody tissue measurements of live plants (n = 2,671 data points), while the remaining 27% (n = 1,005 data points) came from dead plant measurements. In regard to tissue types, stems (inclusive of heartwood and sapwood; n = 1,523 data points), roots (inclusive of fine-root and coarse-root; n = 986 data points) and branches (inclusive of both large and small branches/twigs; n = 619 data points) were most well represented [fig_ref] Figure 1: Structure of Global Woody Tissue Carbon Concentration Database [/fig_ref]. Additionally, woody tissue C data were retrieved from publications spanning a wide climatic range, with a MAT ranging from −5.4-29 °C (across n = 1,326 data points), and MAP ranging from 160-5,130 mm (n = 1,455 data points).
The foremost drying method employed by publications incorporated into GLOWCAD was conventional oven-drying (n = 1,941 data points), while the least common was the Minimizing the Loss of Carbon (MLC) method described by Jones and O'Hara (2016 96 ; n = 9 data points). Drying temperatures ranged widely from www.nature.com/scientificdata www.nature.com/scientificdata/ 18-110 °C, with drying durations spanning 5-360 hours. The majority of publications made use of Elemental Analyzers (corresponding to n = 2,760 data points) when estimating woody tissue C concentrations. In sum, 34% of observations in GLOWCAD (n = 1,241 data points) were associated with exact geographic coordinates of their sampling locations (i.e, not a range of latitude and longitude), and only these observations were assigned climate information from WorldClim (v.2) 27 .
## Technical validation
Trait data validation. records included in GLOWCAD were obtained from peer-reviewed scientific journals, or indirectly, through the TRY Functional Trait Database or Global Root Traits Database. Each specific record is linked to its original reference, allowing users to verify and validate the accuracy of tissue C data and data source. All data in GLOWCAD was thoroughly screened to ensure accuracy, and appropriate methods of data acquisition. Specifically, woody tissue C values had to be measured directly, and not approximated based on secondary sources. Data that did not meet these criteria were excluded from GLOWCAD.
Taxonomic validation. Across the 40-year period during which data was collected [fig_ref] Figure 1: Structure of Global Woody Tissue Carbon Concentration Database [/fig_ref] , tree species may have been misidentified or had their taxonomic information updated. To address these discrepancies and ensure that the most up-to-date taxonomic information is included in GLOWCAD, taxonomic information was directly recorded from original papers, and then verified and adjusted accordingly to reflect the appropriate name listed in the Taxonomic Name Resolution Service v. 4.0 33 . All woody tissue C records included binomial nomenclature, and records without this degree of specificity were omitted from GLOWCAD. Phylogenetic coverage associated with the resolved taxonomy within GLOWCAD are presented in [fig_ref] Figure 5: Phylogenetic coverage of species represented within the Global Woody Tissue Carbon Concentration... [/fig_ref].
## Growth habit validation. growth habit or 'woodiness' was evaluated for all species included in glowcad
to ensure that woody tissue C data corresponded only to woody plant species, based on a functional definition of "woody": i.e., having a persistent aboveground stem [bib_ref] Three keys to the radiation of angiosperms into freezing environments, Zanne [/bib_ref]. Therefore, all species were cross-referenced with those included in a global growth habit dataset [bib_ref] Three keys to the radiation of angiosperms into freezing environments, Zanne [/bib_ref] and growth habits -defined here as trees, shrubs, or shrub/treewere assigned. Species of the Arecaceae (palm) family (n = 8 species) were also included in GLOWCAD (n = 32 data points) since these 1) met the functional definition of "wood" and are 2) are important contributors to aboveground biomass C in Neotropical forests, relative to other biogeographic locations [bib_ref] The global abundance of tree palms, Muscarella [/bib_ref] , monocot species [bib_ref] Mapping global bamboo forest distribution using multisource remote sensing data, Du [/bib_ref] and non-conventional woody species (e.g. tree ferns) [bib_ref] Coarse woody debris carbon storage across a mean annual temperature gradient in..., Iwashita [/bib_ref].
Climate data validation. Bioclimatic variables were assigned to observations which were accompanied by the specific geographic coordinates (excluding ranges) of their sample location, using WorldClim (v.2). In GLOWCAD 34% of woody tissue C observations include a WorldClim-derived estimate of MAT and MAP (n = 1,241 data points). Linear regression models indicated that statistically significant positive relationships existed between (1) publication-vs. WorldClim-derived estimates of MAT (p<0.001, r 2 = 0.95, model slope = 0.97 ± 0.01 (s.e.)) and (2) publication-vs. WorldClim-derived estimates of MAP (p<0.001, r 2 = 0.82, model slope = 1.23 ± 0.02 (s.e.)).
## Usage notes
GLOWCAD is openly available for use in any application. It can be accessed via (1) the DRYAD Digital Repository (https://doi.org/10.5061/dryad.18931zcxk), (2) a GitHub repository, (3) the TRY Plant Trait Database, and (4) upon request to the corresponding author. GLOWCAD is licensed under CC-BY 4.0.
## Code availability
All analyses used to generate figures and summary statistics were performed in R (v.4.1.2). No custom computer code or algorithms were used to generate the data presented in the manuscript.
www.nature.com/scientificdata www.nature.com/scientificdata/ 11. Asner, G. P. Tropical forest carbon assessment: integrating satellite and airborne mapping approaches.
[fig] Figure 1: Structure of Global Woody Tissue Carbon Concentration Database (GLOWCAD). Teal boxes represent the three spreadsheets contained in GLOWCAD and include the column names of each record. Details for all measurements in the "Wood Carbon Database" worksheet are described in Supplementary Tables 1 and 2. Thick gray lines indicate links between worksheets, while gray dashed line indicates a sub-table containing sub-units of a primary variable. [/fig]
[fig] Figure 2, Figure 3: Number of peer-reviewed publications (Panel (a)) and observations (Panel (b)) included in Global Woody Tissue Carbon Concentration Database (GLOWCAD) across publication year. Gray bars indicate the number of publications or woody tissue carbon values included in GLOWCAD (corresponding to the y-axis), while black circles and dotted lines correspond to a cumulative probability density function (corresponding to the z-axis). Woody tissue carbon concentration sampling sites for data sources included in the Global Woody Tissue Carbon Concentration Database (GLOWCAD). Data point colours correspond to tree status, where dead woody tissue is represented in green and live woody tissue is represented in purple. Point sizes are proportional to the number of woody tissue C observations recorded at a site on a continuous scale, ranging from 1-85 observations. [/fig]
[fig] Figure 4: Variation in woody tissue C concentrations (% dry mass) in GLOWCAD. Panel (a) compares the distribution of C concentrations between live woody tissues (light grey bars) and dead woody tissues (dark grey bars). Panel (b) compares the distribution of C concentrations among woody tissue types: stem, heartwood and sapwood (purple bars); root, coarse root and fine root (blue bars); branch and twig (green bars); and bark (yellow bars). [/fig]
[fig] Figure 5: Phylogenetic coverage of species represented within the Global Woody Tissue Carbon Concentration Database (GLOWCAD). Colours mapped onto the phylogenetic tree correspond to 1) two major plant clades including gymnosperms (in blue, n = 100 species), angiosperms (in purple, n = 772 species), and 2) angiosperm 'palm' family, Arecaceae (in yellow, n = 8 species). Each branch represents a species in GLOWCAD (n = 864 total). Bars on the outer ring depict the sample sizes for each species (number of observations proportional to the logarithm of base 10), which are presented in full in Supplementary Tables 3-5. Phylogeny here is based on the Angiosperm Phylogeny Group megatree (R2012089.new) with branch lengths corresponding to clade ages based on fossil records165,166 . (2022) 9:284 | https://doi.org/10.1038/s41597-022-01396-1 [/fig]
|
10.1186/s13613-023-01206-w
|
CCBY
|
10587047
|
37858003
|
s2orc_pubmed_articles
|
Incidence of asymptomatic catheter-related thrombosis in intensive care unit patients: a prospective cohort study
Background Catheter-related thrombosis (CRT) incidence, rate, and risk factors vary in literature due to differences in populations, catheters, diagnostic methods, and statistical approaches.The aim of this single-center, prospective, observational study was to assess incidence, incidence rate (IR), cumulative incidence, and risk factors by means of IR ratio (IRR) of asymptomatic CRT in a non-oncologic Intensive Care Unit (ICU) population.CRT development was assessed daily by means of ultrasound screening.The proportions of patients and catheters developing CRT and CRT incidence rates, expressed as the number of events per catheter-days (cd), were calculated.Kalbfleisch and Prentice's method was used to estimate the cumulative incidence of CRTs.Univariate and multivariable Poisson regression models were fitted to calculate IRR in risk factors analysis.Results Fifty (25%, 95% CI 19-31) out of 203 included patients, and 52 (14%, 95% CI 11-18) out of 375 catheters inserted developed CRT 2) CRTs/1000*cd], after 5 [3-10] days from insertion.Forty-six CRTs (88%) were partial thrombosis.All CRTs remained asymptomatic.Obesity and ECMO support were patient-related protective factors [IRR 0.24 (0.10-0.60), p = 0.002 and 0.05 (0.01-0.50), p = 0.011, respectively].The internal jugular vein had higher CRT IR than other sites [20.1 vs. 5.9 CRTs/1000*cd, IRR 4.22 (1.22-14.63),p = 0.023].Pulmonary artery catheter and leftside cannulation were catheter-related risk factors [IRR 4.24 (2.00-9.00),p < 0.001 vs. central venous catheters; IRR 2.69 (1.45-4.98),p = 0.002 vs. right cannulation, respectively].No statistically significant effect of the number of simultaneously inserted catheters , p = 0.708] and of the catheterization length ), p = 0.155] was detected.The ICU length of stay was longer in CRT patients (20 [15-31] vs. 6 [4-14] days, p < 0.001), while no difference in mortality was observed.Conclusions CRTs are frequent but rarely symptomatic.This study suggests that obesity and ECMO are protective factors, while pulmonary artery catheter, internal jugular vein and left-side positioning are risk factors for CRT.
# Background
Insertion of a catheter in a central vein for fluid infusion and/or monitoring is required in a large proportion of patients admitted to the intensive care unit (ICU).Complications of catheter placement include catheter-related thrombosis (CRT), pulmonary embolism, catheterrelated bloodstream infection, catheter malfunction, and the development of post-thrombotic syndrome [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref].
The endothelial lesion, produced by the catheter insertion, and the consequently reduced blood flow around it lead to fibrin deposition, proliferation, and adherence of endothelial and smooth muscle cells to the catheter and vein walls, thus forming a thrombus, which can invade the lumen until the occlusion of the vein occurs [bib_ref] Central venous catheter-related thrombosis, Geerts [/bib_ref].
Most CRTs remain subclinical [bib_ref] Central venous catheter-related thrombosis, Geerts [/bib_ref] [bib_ref] Deep vein thrombosis associated with central venous catheters-a review, Van Rooden [/bib_ref] , and the literature lacks studies in which a daily screening is performed to evaluate the formation of the thrombosis, although asymptomatic.In addition, most studies focus on cancer or pediatric patients or on peripherally inserted central catheters (PICCs) [bib_ref] Peripherally inserted central venous catheters and central venous catheters related thrombosis in..., Bonizzoli [/bib_ref] [bib_ref] Risk factors for catheter-related thrombosis (CRT) in cancer patients: a patientlevel data..., Saber [/bib_ref] [bib_ref] Epidemiology, diagnosis, prevention and treatment of catheter-related thrombosis in children and adults, Baumann Kreuziger [/bib_ref].As a result, there are still areas of uncertainty regarding the incidence, timing, and risk factors associated with CRT, which may differ from those of deep vein thrombosis.Indeed, different catheters' positioning and characteristics could play an essential role in CRT pathogenesis [bib_ref] Femoral vs jugular venous catheterization and risk of nosocomial events in adults..., Parienti [/bib_ref].Therefore, we designed a prospective cohort study to determine the incidence of asymptomatic CRTs through a daily ultrasound (US) screening and to identify the possible patient and catheter-related risk factors contributing to their development.
# Methods
This is a prospective, single-center, observational cohort study on adult patients admitted to ICU from September 14th, 2020, to October 6th, 2022.Written informed consent was collected according to Italian and European regulations for clinical studies performed on critically ill patients.If patients could not provide their consent at enrollment, delayed consent was obtained.Data collected from patients who later refused consent after regaining consciousness was permanently deleted and excluded from the statistical analysis.This study was conducted following the amended Declaration of Helsinki.This study was approved by the Institutional Ethical Committee (Comitato Etico Milano Area 2 approved the study entitled "Trombosi Asintomatica Catetere Correlata: studio prospettico di coorte", protocol n° 0010404 on March 17th, 2020) and was pre-registered at clinicaltrials.gov(NCT04503135).This study is reported according to STROBE guidelines.
Patients entered the study only when catheterized in the enrolling ICU, where the following types of catheters were used:central venous catheters (CVCs, Teleflex Medical Europe, Ireland, 2-5 lumens, 20 cm, 7-8.5 Ch); [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] hemodialysis venous catheters (HDCs, Teleflex Medical Europe, Ireland, 2-3 lumens, 16-25 cm, 12-14 Ch); [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] pulmonary artery catheters (PACs, Edwards Lifesciences, California, 3 lumens, 110 cm, 7.5 Ch), positioned through an introducer (Teleflex Medical Europe, Ireland, 10 cm, 8 Ch).Patients catheterized outside the ICU environment (i.e., emergency department, general ward, operating theatre) and then admitted to ICU did not fulfill the inclusion criteria.Once a patient was enrolled in the study, any new catheter inserted was considered a new observation.The exclusion criteria for this study were:age < 18 years; [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] congenital thrombophilia; [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] ongoing neoplastic disease; [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] refusal to consent.Before the insertion of any catheter, a thorough assessment of internal jugular veins (IJVs) and subclavian veins was conducted using the Rapid Central Vein Assessment (RaCeVA) protocol [bib_ref] Rapid central vein assessment (RaCeVA): a systematic, standardized approach for ultrasound assessment..., Spencer [/bib_ref] and the compression ultrasound (CUS) of the femoral veins to evaluate all potential cannulation sites with a Philips Affiniti 70 L12-4 Transducer ultrasound (Philips, Amsterdam, Netherlands).
Every catheter was positioned by a trained secondyear Intensive Care resident using real-time ultrasound guidance and under the supervision of an experienced intensivist.After the catheter placement, the correct positioning was verified via a chest X-ray, except for femoral-inserted catheters.
Ultrasound monitoring of both catheterized and not catheterized veins to check the occurrence of any thrombosis was performed daily.The study ended when one of the following conditions was met:removal of the catheter (or, in case of multiple catheterizations, removal of the last remaining catheter); [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] diagnosis of catheterrelated thrombosis; [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] 28 days following the last catheter insertion or patient discharge from the ICU, whichever came first; [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] patient's death.If multiple catheterizations were performed, follow-up for all catheters ended after the first CRT was diagnosed.The follow-up of the last patient ended on October 21st, 2022.
## Crt diagnosis
Diagnosis of CRT was performed through ultrasounds, which included direct visualization of the intraluminal thrombus in combination with either one of the following criteria:partial vein compressibility or partial absence of blood flow in difficult-to-compress veins (i.e., subclavian), and that was considered partial thrombosis, or [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] complete vein incompressibility or complete absence of blood flow, and that was considered complete thrombosis.Blood flow was studied using Color Doppler Flow Imaging [bib_ref] Diagnosis and management of upper extremity deep-vein thrombosis in adults, Grant [/bib_ref] [bib_ref] Catheter-related thrombosis: risks, diagnosis, and management, Linenberger [/bib_ref] [bib_ref] Management of occlusion and thrombosis associated with long-term indwelling central venous catheters, Baskin [/bib_ref].Ultrasound findings were, thus, classified as:no thrombosis; [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] partial thrombosis; [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] complete thrombosis.Whenever a thrombosis was suspected during the daily US assessment performed by the intensivist, a specialistic US study was asked to a radiologist to confirm or rule out the diagnosis.Asymptomatic thrombosis was defined whenever a thrombosis was imaging-diagnosed without developing any clinical objective alteration on physical examination (i.e., tenderness, warmth, erythema or cyanosis, edema, superficial venous dilation) or without raising in the physician in charge the suspicion of pulmonary embolism.
## Catheter maintenance
The nursing staff in the ICU conducted three daily checks to ensure adherence of the medications to the vascular access site [bib_ref] Vascular access devices: securement and dressings, Gabriel [/bib_ref].To ensure proper care of the vascular access, each lumen was kept patent by infusing maintenance fluids or using a neutral displacement clave after 10 ml of saline solution intermittent flush [bib_ref] Best practice for delivering small-volume intermittent intravenous infusions, Harding [/bib_ref].In addition, 4% sodium citrate was utilized as a lock solution for hemodialysis catheters when patients were not undergoing Renal Replacement Therapy [bib_ref] Sodium citrate 4% locking solution for central venous dialysis catheters-an effective, more..., Grudzinski [/bib_ref].
## Data collection
Data were collected using REDCap electronic data capture tools hosted at Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico [bib_ref] Research electronic data capture (REDCap)-a metadata-driven methodology and workflow process for providing..., Harris [/bib_ref] [bib_ref] The REDCap consortium: Building an international community of software platform partners, Harris [/bib_ref].During the enrollment phase, the following patient information was recorded: age, sex, Body Mass Index (BMI), admission diagnosis, Charlson Comorbidity Index, Sequential Organ Failure Assessment (SOFA) Score, diagnosis, date of ICU admission, laboratory data (hematocrit, platelets, International Normalized Ratio and activated partial thromboplastin time ratio, fibrinogen, D-dimers).Whenever, for clinical purposes, a catheter was placed, the following catheter-related data were recorded: kind of catheter inserted, catheter's outer diameter, number of lumens, antimicrobic treatment, and catheter tip position at chest X-ray, catheterized vein and its diameter, and number of cannulation attempts.The following patient-related data were recorded daily: anticoagulant and antiplatelet therapies, need for extracorporeal membrane oxygenation (ECMO) support or surgical interventions, and results of the RaCeVa protocol and femoral veins' CUS.Moreover, patients' ICU outcomes and lengths of stay (LOS) were collected at the end of hospitalization.
The following cutoffs were used to differentiate prophylactic from therapeutic anticoagulation: 80 mg/day for enoxaparin, 2.5 mg/day for fondaparinux, and 200 Units/ kg/day for Unfractionated Heparins.Argatroban administration was always therapeutic.
## Outcomes
The primary study outcomes were the incidence, the incidence rate, and the cumulative incidence of CRT in an ICU population.Secondary outcomes included:the potential risk factors associated with CRTs; [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] the proportion of partial or complete central venous thrombosis; [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] the percentage of patients with central venous catheters who developed non-catheter-related thrombosis.We, therefore, described the clinical measures that have been taken following the diagnosis of CRT.
# Statistical analysis
Categorical variables are reported as absolute numbers and percentages.Continuous variables are presented as mean (standard deviation, SD) or median [interquartile range, IQR] for normal and non-normal distributions.Patient and catheter characteristics were described according to the CRT occurrence or not, without statistical comparisons.The proportions of patients and catheters developing CRT and 95% confidence intervals (95% CI) were calculated.Moreover, CRT incidence rates (IR) and 95% CI, expressed as events/1000 catheter-days, were calculated, considering multiple catheterizations on the same patient as different observations.Catheter cumulative incidence of CRT was estimated using the Kalbfleisch and Prentice method [bib_ref] The analysis of failure times in the presence of competing risks, Prentice [/bib_ref] , considering death during ICU and thrombosis of another catheter as competing events.
Univariate and multivariable random-intercept Poisson regression models, with patients as random-effect, were fitted to calculate incidence rate ratios (IRR) and 95% CI of CRT.Multivariable models included admission disease and SOFA as fixed adjustment covariates; ECMO support, surgical interventions, the number of simultaneous catheters inserted on the same patient, and days of catheterization as time-dependent adjustment covariates.Analyses were performed with Stata 17 (StataCorp.2017, Texas, USA) and JMP Pro 16 (SAS Institute Inc., Cary, NC, USA).
# Results
During the study period, 216 out of 1615 patients needed catheterization during ICU stay and were eligible to be enrolled in the study.Three patients refused to participate, and eight patients had oncologic comorbidities.Two hundred and five patients were enrolled.Data collection was wrongly performed on two patients, and thus, they were excluded from the analysis.The remaining 203 patients entered the analysis.Enrolment occurred 0 [0-1] days after ICU admission, see Fig. [fig_ref] Figure 1: Study flowchart [/fig_ref] , flowchart, for details.
One-hundred-thirty-two (65%) were males, and the median age was 60 [49-70] years.At enrollment, SOFA Score and Charlson's Comorbidity Index were 4 [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] [bib_ref] Central venous catheter-related thrombosis, Geerts [/bib_ref] [bib_ref] Deep vein thrombosis associated with central venous catheters-a review, Van Rooden [/bib_ref] and 2 [bib_ref] Symptomatic pulmonary embolus after catheter removal in children with catheter related thrombosis:..., Jaffray [/bib_ref] [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] , respectively.Non-surgical diseases were the most important causes of ICU admission (178/203, 88%), and Acute Respiratory Distress Syndrome (ARDS) was the most represented diagnosis (73/203, 36%).COVID was the most represented cause of ARDS (56/73, 77%).One-hundred-five (105/203, 52%) patients were on antithrombotic prophylaxis at enrollment, whereas 80 (39%) were fully anticoagulated.One-hundred-eighteen (58%) patients needed more than one catheter during ICU stay.Twenty-one (10%) patients were treated with ECMO, and 53 (26%) underwent surgical intervention before or during ICU stay.Patients' characteristics are presented in Table [fig_ref] Table 1: Characteristics of enrolled patients CRT Catheter-related thrombosis, BMI, Body Mass Index, SOFA,... [/fig_ref].
Three-hundred-seventy-five catheters were placed and followed for 2941 catheter-days (cd).Table [fig_ref] Table 2: Characteristics of catheters inserted CVC central venous catheter, RA right atrium, SVC... [/fig_ref] describes the catheters' characteristics in detail.
The median duration of catheter follow-up was 5 [3-10] days.CVCs (266, 71%) were the most frequently positioned, followed by PACs (63, 17%) and HDCs (46, 12%).The most common site of cannulation was the IJV (302, 81%).In total, 255 (68%) catheters were positioned on the right side of the body, of which 206 (81%) were in the right IJV.Additional file 1: Table reports details of the type, site, and side of cannulation divided by catheters with and without CRT.More than one cannulation attempt was performed 25 times (7%).Catheter-vein diameter ratio was 0.21 [0.16-0.25].
Fifty out of 203 (25%, 95% CI 19-31) patients developed a CRT.Two patients had CRTs on two distinct catheters on the same day.The proportion of CRTs was 52 out of 375 (14%, 95% CI 11-18), and the incidence rate (95% CI) was 17.7 (13.5-23.2) CRTs/1000*cd.Figure [fig_ref] Figure 2: Cumulative incidence of catheter-related thrombosis [/fig_ref] shows the catheters' cumulative incidence of CRT.The median time interval between catheter placement and CRT was 5 [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] [bib_ref] Central venous catheter-related thrombosis, Geerts [/bib_ref] [bib_ref] Deep vein thrombosis associated with central venous catheters-a review, Van Rooden [/bib_ref] [bib_ref] Peripherally inserted central venous catheters and central venous catheters related thrombosis in..., Bonizzoli [/bib_ref] [bib_ref] Risk factors for catheter-related thrombosis (CRT) in cancer patients: a patientlevel data..., Saber [/bib_ref] [bib_ref] Epidemiology, diagnosis, prevention and treatment of catheter-related thrombosis in children and adults, Baumann Kreuziger [/bib_ref] [bib_ref] Femoral vs jugular venous catheterization and risk of nosocomial events in adults..., Parienti [/bib_ref] shows catheter-related risk factors analysis.
Five adverse events were reported during catheter positioning: four hematomas and one accidental carotid artery puncture.One catheter needed immediate replacement after X-rays showing contralateral anonymous vein tip placement.No iatrogenic pneumothorax was registered.
Eighteen (36%) patients were on therapeutic anticoagulation at CRT diagnosis.Seventeen (34%) patients started anticoagulation after CRT diagnosis, whereas in 15 patients (30%), no clinical changes in therapy were decided by the ICU doctor in charge.Where in doubt, the decision to anticoagulate the patient was taken jointly by the intensivist and the hematologist.In Additional file 4: Table , the characteristics of these two sub-cohorts of patients are described.No statistical differences between other variables were found except for non-COVID ARDS as the admission reason.None of the catheters malfunctioned due to CRT; thus, none was immediately removed.Neither bleeding related to the start of anticoagulation therapy nor thrombotic complications in not treated patients were observed.
During the study period, nine patients (4%) developed a deep venous thrombosis in a non-catheterized vein.
Mortality in ICU did not differ between patients who suffered vs. those who did not suffer from CRT (23% vs. 30%, p = 0.323), while ICU LOS was longer in CRT patients vs. 6 [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] [bib_ref] Central venous catheter-related thrombosis, Geerts [/bib_ref] [bib_ref] Deep vein thrombosis associated with central venous catheters-a review, Van Rooden [/bib_ref] [bib_ref] Peripherally inserted central venous catheters and central venous catheters related thrombosis in..., Bonizzoli [/bib_ref] [bib_ref] Risk factors for catheter-related thrombosis (CRT) in cancer patients: a patientlevel data..., Saber [/bib_ref] [bib_ref] Epidemiology, diagnosis, prevention and treatment of catheter-related thrombosis in children and adults, Baumann Kreuziger [/bib_ref] [bib_ref] Femoral vs jugular venous catheterization and risk of nosocomial events in adults..., Parienti [/bib_ref] [bib_ref] Rapid central vein assessment (RaCeVA): a systematic, standardized approach for ultrasound assessment..., Spencer [/bib_ref] [bib_ref] Diagnosis and management of upper extremity deep-vein thrombosis in adults, Grant [/bib_ref] [bib_ref] Catheter-related thrombosis: risks, diagnosis, and management, Linenberger [/bib_ref] , p < 0.001).The ICU LOS in CRT patients is primarily due to time spent in the ICU after CRT development rather than before (days between ICU admission and CRT 6 [bib_ref] Infectious complications of central venous catheters increase the risk of catheter-related thrombosis..., Van Rooden [/bib_ref] [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] [bib_ref] Central venous catheter-related thrombosis, Geerts [/bib_ref] [bib_ref] Deep vein thrombosis associated with central venous catheters-a review, Van Rooden [/bib_ref] [bib_ref] Peripherally inserted central venous catheters and central venous catheters related thrombosis in..., Bonizzoli [/bib_ref] [bib_ref] Risk factors for catheter-related thrombosis (CRT) in cancer patients: a patientlevel data..., Saber [/bib_ref] [bib_ref] Epidemiology, diagnosis, prevention and treatment of catheter-related thrombosis in children and adults, Baumann Kreuziger [/bib_ref] [bib_ref] Femoral vs jugular venous catheterization and risk of nosocomial events in adults..., Parienti [/bib_ref] vs. days between CRT and ICU outcome 12 [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] [bib_ref] Central venous catheter-related thrombosis, Geerts [/bib_ref] [bib_ref] Deep vein thrombosis associated with central venous catheters-a review, Van Rooden [/bib_ref] [bib_ref] Peripherally inserted central venous catheters and central venous catheters related thrombosis in..., Bonizzoli [/bib_ref] [bib_ref] Risk factors for catheter-related thrombosis (CRT) in cancer patients: a patientlevel data..., Saber [/bib_ref] [bib_ref] Epidemiology, diagnosis, prevention and treatment of catheter-related thrombosis in children and adults, Baumann Kreuziger [/bib_ref] [bib_ref] Femoral vs jugular venous catheterization and risk of nosocomial events in adults..., Parienti [/bib_ref] [bib_ref] Rapid central vein assessment (RaCeVA): a systematic, standardized approach for ultrasound assessment..., Spencer [/bib_ref] [bib_ref] Diagnosis and management of upper extremity deep-vein thrombosis in adults, Grant [/bib_ref] [bib_ref] Catheter-related thrombosis: risks, diagnosis, and management, Linenberger [/bib_ref] [bib_ref] Management of occlusion and thrombosis associated with long-term indwelling central venous catheters, Baskin [/bib_ref] [bib_ref] Vascular access devices: securement and dressings, Gabriel [/bib_ref] [bib_ref] Best practice for delivering small-volume intermittent intravenous infusions, Harding [/bib_ref] [bib_ref] Sodium citrate 4% locking solution for central venous dialysis catheters-an effective, more..., Grudzinski [/bib_ref] [bib_ref] Research electronic data capture (REDCap)-a metadata-driven methodology and workflow process for providing..., Harris [/bib_ref] [bib_ref] The REDCap consortium: Building an international community of software platform partners, Harris [/bib_ref] [bib_ref] The analysis of failure times in the presence of competing risks, Prentice [/bib_ref].
# Discussion
The main finding of the study was that a daily ultrasound screening revealed a high proportion of asymptomatic CRTs: 25% of enrolled patients and 17% of the catheters developed CRT, corresponding to an incidence rate of 17.7 CRTs/1000*cd.Of note, none of the 52 CRTs detected when asymptomatic became later symptomatic.We also identified some catheter-related characteristics as risk factors for CRT (catheter type and catheter site and side) and some patient-related ones as protective factors (BMI and ECMO treatment).Furthermore, we highlighted that neither the number of simultaneous inserted catheters nor the duration of catheterization itself are risk factors for CRT.Finally, we described the anticoagulation management post-CRT diagnosis decided by the physicians in charge.Previous literature describing CRT incidence is available, but results vary significantly due to different study designs (retrospective vs. prospective), catheter selection (PICCs [bib_ref] Peripherally inserted central catheter (PICC)-related thrombosis in critically Ill patients, Zochios [/bib_ref] vs. CVCs [bib_ref] Complications of intravascular catheters in ICU: definitions, incidence and severity a randomized..., Günther [/bib_ref] , population selection (pediatric [bib_ref] Central venous catheter-associated deep vein thrombosis in critically ill pediatric patients: risk..., Johnson [/bib_ref] vs. adult patients or only-cancer [bib_ref] International clinical practice guidelines for the treatment and prophylaxis of thrombosis associated..., Debourdeau [/bib_ref] vs. non-cancer patients), and CRT diagnostic method (venography [bib_ref] van den Abbeele AD. The treatment and outcome of cancer patients with..., Frank [/bib_ref] vs. US.).Timsit et al. [bib_ref] Central vein catheter-related thrombosis in intensive care patients: Incidence, risks factors, and..., Timsit [/bib_ref] , in a prospective multicentric study conducted in three medicalsurgical ICUs, observed a higher proportion of CRT (33% of 208 patients) than what we found.Some important differences in the study design can explain this discrepancy.First, their population included cancer patients.Second, improvement in catheter materials technology (i.e., less thrombogenic) could have lowered the risk of CRTs in our population [bib_ref] Epidemiology, diagnosis, prevention and treatment of catheter-related thrombosis in children and adults, Baumann Kreuziger [/bib_ref].Third, we performed US-guided catheterization, and this could have allowed us to reduce the number of catheter positioning attempts.Gunther et al. [bib_ref] Complications of intravascular catheters in ICU: definitions, incidence and severity a randomized..., Günther [/bib_ref] reported a higher incidence rate of catheter thrombosis (33 CRTs/1000*cd).Still, again, they included cancer patients and studied both arterial and central venous catheters without reporting disaggregated data from the various catheter types.
Recently, Wu et al. [bib_ref] Daily point-of-care ultrasound-assessment of central venous catheter -related thrombosis in critically ill..., Wu [/bib_ref] , in a prospective, multicenter study, including 1262 patients, reported a CRT in 16.9% of patients.Similar to us, they performed a daily US assessment to detect CRT formation before any symptoms developed.Some relevant differences between the studies should be noted.In particular, they did not include HDCs and PACs, and the sites of catheterization were different, with a lower proportion of catheters placed in the IJV (54% vs. 81%) and a higher rate of subclavian catheterization (30% vs. 2%).Interestingly, none of their CRT was symptomatic.
Female sex, older age, and higher BMI are commonly considered risk factors for venous thrombosis [bib_ref] Risk factors associated with catheter-related venous thrombosis: a meta-analysis, Liu [/bib_ref] [bib_ref] Common risk factors for both arterial and venous thrombosis, Lowe [/bib_ref].However, all previous studies evaluated risk factors using odds (and odds ratios) of developing CRTs, thus neglecting time at risk.We did consider time by calculating incidence rates and found no association between CRT occurrence and sex; age ≥ 65 years had a 30% higher CRT rate (although with wide CIs), and, surprisingly, obesity was associated with markedly lower CRT incidence rate.All these patient's characteristics could affect the duration of ICU length-of-stay [bib_ref] Obesity and mortality, length of stay and hospital cost among patients with..., Nguyen [/bib_ref] [bib_ref] Effect of obesity on intensive care morbidity and mortality: a meta-analysis*, Akinnusi [/bib_ref].Therefore, we think that IRR rather than the odds ratio may be more correct in describing CRT's risk and protective factors.This could be why our risk factors analysis results are slightly different from previous literature.Indeed, despite the median BMI of the two cohorts of patients being similar ]kg/m 2 in CRT patients vs. 26.1 ]kg/m 2 in no CRT patients), our signal in the analysis comparing IR is very strong: the higher the BMI, the slower the thrombus formation.However, the novelty of this result surely deserves further studies to be confirmed.
In our study, the most frequent diagnosis of admission was COVID-19 ARDS.Although the demonstrated hypercoagulability of COVID patients [bib_ref] Hypercoagulability of COVID-19 patients in intensive care unit: a report of thromboelastography..., Panigada [/bib_ref] , the disease was not associated with a higher risk of CRT in our patient.This could be due to the enhanced anticoagulation regimen adopted in our institution for this disease.Anticoagulation was not highlighted as a protective factor.Nevertheless, we think the lower incidence rate of CRT in ECMO patients could be traced back to the strict anticoagulation monitoring of these patients, which, in our center, is performed at least three times a day.
PACs had a higher rate of CRTs rather than CVCs and HDCs.We did not find any study evaluating this kind of catheter specifically.Still, we can hypothesize that the need for a large introducer [bib_ref] Reducing catheter-related thrombosis using a risk reduction tool centered on catheter to..., Spencer [/bib_ref] , our preference for IJV positioning, and the higher length of the catheter are possible explanations for this result.Moreover, we observed lower CRTs in the femoral rather than the IJV approach and left-side positioning as a risk factor for CRT.These results are consistent with previously published studies [bib_ref] Intravascular complications of central venous catheterization by insertion site, Parienti [/bib_ref] [bib_ref] Which central venous catheters have the highest rate of catheterassociated deep venous..., Malinoski [/bib_ref] [bib_ref] Complication and failures of central vascular access device in adult critical care..., Takashima [/bib_ref].
The employment of US guidance during each catheter positioning maneuver may have contributed to the observed low incidence of adverse events and their relatively mild severity.
We described heterogeneous therapeutic management of patients after CRT diagnosis, and we failed to identify any patient characteristic that could have guided the physician's decision to start full anticoagulation, except for ARDS as the admission diagnosis.Deciding to start anticoagulation, for how long, and with which drug are still topics of research and discussion in symptomatic CRT, especially those of upper extremities [bib_ref] Central venous catheter-related thrombosis in children and adults, Sridhar [/bib_ref].CHEST guidelines [bib_ref] Antithrombotic therapy for VTE disease: antithrombotic therapy and prevention of thrombosis, Kearon [/bib_ref] suggested not removing the catheter if functional, starting anticoagulation, and continuing it at least 3 months after its removal.The grading of this recommendation is a 2C.Of note, we did not observe any adverse events due to the choice of not starting anticoagulation, and, according to Wu et al. [bib_ref] Daily point-of-care ultrasound-assessment of central venous catheter -related thrombosis in critically ill..., Wu [/bib_ref] , most of their CRTs regress spontaneously without it.Our study and Wu's [bib_ref] Daily point-of-care ultrasound-assessment of central venous catheter -related thrombosis in critically ill..., Wu [/bib_ref] , taken together, account for a total of 263 patients who developed asymptomatic CRTs.Even without a specific daily screening, in routine clinical settings, some of them could have been diagnosed with CRT incidentally and thus, according to guidelines, anticoagulated for months.Future research should investigate if, in these cases, applying the same guidelines of symptomatic thrombosis is suitable.
Similar to the results of Wu [bib_ref] Daily point-of-care ultrasound-assessment of central venous catheter -related thrombosis in critically ill..., Wu [/bib_ref] , our patients who suffered from CRT had longer ICU stays but no mortality difference.At first glance, explaining this with a higher risk of CRT in long-lasting hospitalized patients may be tempting.Actually, CRT patients spent much time hospitalized after CRT diagnosis, not before.Since asymptomatic CRTs are unlikely to cause longer hospitalization, we can speculate that they are, at least, an early sign of a possible long ICU stay.
Our study has several strengths.The most important is the daily assessment of CRT, which allowed us to promptly detect thrombosis, even if asymptomatic.We were able to report the estimates of a time-dependent event as incidence rate rather than proportion.Second, we considered a heterogeneous population with different kinds of catheters inserted, which may provide us with the information currently lacking in the scientific literature.Third, we asked a radiology physician to confirm or rule out every doubt or suspected CRT identified during the ICU physician's daily assessment, to increase our accuracy in CRT diagnosis [bib_ref] Diagnostic accuracy of point-of-care ultrasound for catheter-related thrombosis in children, Li [/bib_ref].
Our study has several limitations, too.First, we censored follow-up at patient ICU discharge.Thus, we cannot account for CRTs that could have developed later.Second, contrary to other studies [bib_ref] Daily point-of-care ultrasound-assessment of central venous catheter -related thrombosis in critically ill..., Wu [/bib_ref] [bib_ref] Intravascular complications of central venous catheterization by insertion site, Parienti [/bib_ref] , in which US screening was continued for 2 days after catheter removal or after CRT, we did not plan to continue with a US follow-up, and we were not able to describe the evolution of CRTs in anticoagulated vs. not anticoagulated patients.Third, we evaluated asymptomatic thrombosis on behalf of clinical objective alteration at physical examination.Thus, we cannot exclude that some unconscious patients were subjectively symptomatic.Fourth, the habits of our ICU physicians (i.e., use of PAC as gold-standard hemodynamic monitoring, low rate of subclavian CVCs positioning), the choice to include only patients catheterized in the ICU and not in other hospital settings, and the monocentric nature of our study may limit the external validity of our results.Finally, we still have many unanswered questions regarding other possible risk factors for CRT, such as whether different US puncture techniques (i.e., long axis vs. short axis or in-plane vs. out-of-plane approach), catheter-related infections, or even catheter's depth of insertion could affect the development of CRTs [bib_ref] The relationship between the thrombotic and infectious complications of central venous catheters, Raad [/bib_ref].
# Conclusions
In our patient population, CRTs were frequent, rarely symptomatic, and most were partial thrombosis.The kind and the site of the catheter could play a role in favoring or protecting from CRT. Obesity and ECMO treatment were protective factors.Further studies are required to define the clinical impact of CRTs and should investigate the best coagulation management after CRT diagnosis.
## Abbreviations
[fig] Figure 1: Study flowchart [/fig]
[fig] Figure 2: Cumulative incidence of catheter-related thrombosis [/fig]
[table] Table 1: Characteristics of enrolled patients CRT Catheter-related thrombosis, BMI, Body Mass Index, SOFA, Sequential Organ Failure Assessment; ICU, Intensive Care Unit; ARDS, Acute Respiratory Distress Syndrome; COVID, Coronavirus Disease; Hct, Hematocrit; INR, International Normalized Ratio; aPTT, activated partial thromboplastin time; ECMO, Extracorporeal Membrane Oxygenation a Including asthma and chronic obstructive pulmonary disease exacerbation, cardiogenic shock, diabetic ketoacidosis, hypoglycemic coma, intoxication, heat stroke, botulinum poisoning, and status epilepticus b Including lung and liver transplantation, bowel perforation and osteomyelitis c Including patients admitted to ICU for surgical disease and patients who needed surgery during ICU stay for any reason [/table]
[table] Table 2: Characteristics of catheters inserted CVC central venous catheter, RA right atrium, SVC superior vena cava, PA pulmonary artery a 315 catheters included [Excluding catheters inserted in the femoral vein (n = 60)] [/table]
|
10.3390/foods12112134
|
CCBY
|
10253141
|
37297382
|
s2orc_pubmed_articles
|
Botanical Origin Influence on Some Honey Physicochemical Characteristics and Antioxidant Properties
Five types of honey (multifloral, sunflower, linden, rapeseed, and acacia), from Southern Romania, were classified using chemometrics methods coupled with IR spectroscopy. The botanical origin's effect on the physicochemical characteristics of honey was studied to highlight the most valuable plant source of honey. Except for antioxidant activity, the moisture, ash, electrical conductivity (EC), pH, free acidity (FA), total sugar content (TSC), hydroxymethylfurfural (HMF), total phenolic (TPC), tannin (TTC), and flavonoid content (TFC) were significantly influenced by the botanical origin of the honey. The results showed that sunflower honey had the highest moisture (15.53%), free acidity (16.67 mEq kg −1 ), electrical conductivity (483.92 µS cm −1 ), phenolics (167.59 mg GAE 100 g −1 ), and flavonoids (19.00 mg CE 100 g −1 ), whereas multifloral honey presented the highest total sugar content (69.64 g Glu 100 g −1 ). The highest HMF content was found in linden honey (33.94 mg kg −1 ). The HMF contents of all tested honey were within the standard recommended limit, and they confirmed that the tested honey was free of any heat treatment. All five types of tested honey presented a safe moisture content for storage and consumption (12.21-18.74%). The honey s free acidity was in the range of 4.00 to 25.00 mEq kg −1 ; this indicated the freshness of the samples and the absence of any fermentation processes in the tested honey. Honey with a total sugar content over 60% (except for linden honey, with 58.05 g glucose 100 g −1 ) showed the characteristic of nectar-derived honey. The elevated antioxidant activity of honey was correlated with its high moisture, flavonoids, and HMF, whereas the tannins and HMF were positively correlated with ash and electrical conductivity. The higher content of phenolics, flavonoids, and tannins was correlated with higher free acidity. The chemometric method, coupled with ATR-FTIR spectra, revealed a clear separation between linden honey from acacia, multifloral, and sunflower honey.
# Introduction
Honey is a natural food produced by honeybees, Apis melifera, from the nectar of blossoms or exudates of trees and plants giving nectar honey or honeydews [bib_ref] Contribution of honey in nutrition and human health: A review, Alvarez-Suarez [/bib_ref].
Honey production is an enzymatic process, completed with dehydration. Honeybees consume nectar and pollen as carbon and nitrogen sources, respectively, and both of these foods are subjected to gut processing, an enzymatic process of breaking down the nectar's sugar into simple sugars (mainly glucose and fructose). The second step is to pass this nectar/sugar mixture to the younger honeybees, who convert it to honey via another enzymatic step involving three enzymes secreted by the hypopharyngeal glands of workers: alpha-glucosidase (breaks sucrose, the majority nectar component, into glucose and fructose), amylase (hydrolyses the starches that contaminate the nectar), and the glucose oxidase (converts glucose into gluconic acid and peroxide, both of which are responsible
# Materials and methods
## Chemicals and reagents
All chemicals and reagents were purchased from Merck, Darmstadt, Germany.
## Honey samples
During 2021-2022, twenty-four samples of Romanian honey were analyzed, with the following geographical origins (GO): sunflower (S) from Arges-Costesti and Arges-Gliganu (AG-C, and AG-G), linden (L) from Giurgiu-Bolintin and Tulcea-Topolog (GR-B, and TL-T), rapeseed (R) from Teleorman-Branceni, Arges-Costesti, and Arges-Gliganu (TR-B, AG-C, and AG-G), multifloral (M) from Tulcea-Casimcea and Arges-Mozaceni (TL-C, and AG-MZ), and acacia (A) from Arges-Costesti, Arges-Mosoaia, and Arges-Vedea (AG-C, AG-MO, and AG-V). The twenty-four honey samples were produced by bee populations of the Apis mellifera Carpatica race in the small or medium-sized apiaries of Southern Romania. All samples were collected directly from the primary producers, without any thermal treatment (based on the fact that raw honey is allowed for marketing in Romania). Immediately after harvesting, the samples were subjected to physicochemical, biochemical, and spectral analysis.
## Botanical origin identification
The botanical origin (BO) of the Romanian honey samples was confirmed using melissopalynological analysis, in accordance with the methodology suggested by [bib_ref] Methods of melissopalynology, Louveaux [/bib_ref] [bib_ref] Methods of melissopalynology, Louveaux [/bib_ref]. Both the qualitative results (pollen spectrum of the honey sample) and quantitative results (number of pollen grains per gram of honey) were registered. The pollen spectrum of each honey sample was determined by counting at least 800 pollen grains. Only the pollen grain types with frequencies higher than 1% were considered. For each honey sample, the relative frequency classes were determined in accordance with the international melissopalynological nomenclature using the terms: 'dominant pollen' (more than 45% of pollen grains counted), 'accompanying pollen' (representing 15-45%), 'important minor pollen' (3-15%), and 'minor pollen' (less than 3%) [bib_ref] Methods of melissopalynology, Louveaux [/bib_ref].
## Physicochemical determinations
The physicochemical parameters were determined in accordance with the Harmonised Methods of the International Honey Commission (2009).
All samples were prepared in triplicate.
Moisture and ash (mineral content) were determined gravimetrically via oven drying at 105-110 - C, and the calcination of dry residue was determined at 550-600 - C, until the samples were brought to a constant mass. The results were expressed as a percentage (%) of the moisture and ash content.
Electrical conductivity (EC) was determined by measuring the electrical resistance of aqueous solutions of honey consisting of 20% dry matter, with a multimeter C-561, at 20 - C. The results were expressed as micro Siemens per centimeters (µS cm −1 ). pH was measured using aqueous honey solutions, consisting of 10% dry matter, using a multimeter Consort C-561.
Free acidity (FA) was determined by titrating aqueous solutions of honey consisting of 10% dry matter, using 0.1 M sodium hydroxide solution, until it reached pH 8. [bib_ref] A variety of volatile compounds as markers in Palestinian honey from Thymus..., Odeh [/bib_ref]. The results were expressed in milliequivalents of acids per kg of honey (mEq kg −1 ).
Total sugars content (TSC) expressed as g glucose (Glu) 100 g −1 was determined colorimetrically by following the methodology suggested by [bib_ref] Colorimetric method for determination of sugars and related substances, Dubois [/bib_ref] [bib_ref] Colorimetric method for determination of sugars and related substances, Dubois [/bib_ref].
5-hydroxymethylfurfural (HMF) was determined in aqueous solutions of honey consisting of 20% dry matter using the measurement obtained from the absorbances of the filtered solutions, after clarification with Carrez I and II solutions, at 284 and 336 nm (White method [bib_ref] Spectrophotometric method for hydroxymethylfurfural in honey, White [/bib_ref]. The results were expressed in milligrams HMF per kg of honey (mg HMF kg −1 ).
## Bioactive compound determinations
Total polyphenols content (TPC) was determined using the methodology suggested by Ciucu-Paraschiv and Hoza (2021) [bib_ref] The effect of foliar application with organic and inorganic products on the..., Ciucu-Paraschiv [/bib_ref]. Honey solutions consisting of 40% dry matter in absolute ethanol were used. The results were expressed as a mg gallic acid equivalent (GAE) 100 g −1 .
Total tannin content (TTC) was determined using the methodology suggested by [bib_ref] Evolution of the polyphenols, flavonoids, and tannins content in walnut leaves and..., Giura [/bib_ref] [bib_ref] Evolution of the polyphenols, flavonoids, and tannins content in walnut leaves and..., Giura [/bib_ref] , using aqueous solutions of honey consisting of 40% dry matter. The results were expressed as a mg gallic acid equivalent (GAE) 100 g −1 .
Total flavonoid content (TFC) was determined using the methodology suggested by Tudor-Radu et al. (2016) [bib_ref] Assessment of ascorbic acid, polyphenols, flavonoids, anthocyanins and carotenoids content in tomato..., Tudor-Radu [/bib_ref] , using honey solutions consisting of 40% dry matter in absolute ethanol and the results were expressed as a mg catechin equivalent (CE) 100 g −1 .
Total antioxidant activity was determined using the methodology suggested by [bib_ref] Valorisation of aronia melanocarpa pomace for development of functional ingredients with high..., Lazar [/bib_ref] [bib_ref] Valorisation of aronia melanocarpa pomace for development of functional ingredients with high..., Lazar [/bib_ref] , using ethanolic solutions of honey consisting of 40% dry matter. The results were expressed as a percentage of the inhibition of 2,2-diphenyl-1-picrylhydrazyl (DPPH I%).
## Uv-vis and atr-ftir analysis
The spectral measurements were made with a UV-Vis Perkin-Elmer Lambda25 and an FTIR Jasco 6300 spectrometer.
An ATR accessory equipped with a diamond crystal (Pike Technologies, Madison, Wisconsin, USA) allows the collection of FTIR spectra directly on a sample without any special preparation. The FTIR spectra were recorded in the region of 4000-400 cm −1 , with a TGS detector, and apodization Cosine. The spectral data were processed with JASCO Spectra Manager software, version 2. Samples were scanned at a 4 cm −1 resolution, accumulation: 100 scans. Background reference spectra were recorded using air after every sample to minimize the interference due to carbon dioxide and water vapor in the atmosphere. Between measurements, the ATR crystal was carefully cleaned using pure acetone (Sigma-Aldrich Co., Saint Louis, MO, USA), then, it was dried with soft tissue [bib_ref] Rapid method for the discrimination of Romanian wines based on mid-infrared spectroscopy..., Topală [/bib_ref] [bib_ref] ATR-FTIR spectroscopy coupled with chemical and chemometric analysis to distinguish between some..., Topală [/bib_ref] [bib_ref] Attenuated total reflection Fourier transform infrared (ATR-FTIR): A method for the biochemical..., Topală [/bib_ref].
All measurements were taken at room temperature (T = 23 - C). For each sample, three replicate spectra were recorded to ensure spectral reproducibility and to assess analytical precision; then, the average spectrum was complete.
# Chemometric analysis
Infrared Spectra were exported from Spectra Manager, in an ASCII (dx) format, into the Unscrambler Software (Edition X 10.4, Camo Oslo, Norway) for chemometric analysis. Spectra were preprocessed using the second-derivative transformation, the Savitzky-Golay derivation. The use of spectra derivatives with the Savitzky-Golay algorithm as a chemometric pre-processing technique has been widely reported in most classifications that are based on FTIR spectroscopy [bib_ref] Rapid method for the discrimination of Romanian wines based on mid-infrared spectroscopy..., Topală [/bib_ref] [bib_ref] ATR-FTIR spectroscopy coupled with chemical and chemometric analysis to distinguish between some..., Topală [/bib_ref] [bib_ref] Attenuated total reflection Fourier transform infrared (ATR-FTIR): A method for the biochemical..., Topală [/bib_ref]. Multivariate analysis (e.g., principal component analysis, PCA; hierarchical cluster analysis, HCA; linear discriminate analysis, LDA) was previously often used to evaluate and/or classify honey depending on its chemical composition, physicochemical, or biological properties [bib_ref] Physico-chemical, enzymatic, mineral and colour characterization of three different varieties of honeys..., Nayik [/bib_ref]. The principal component analysis (PCA) model was developed using cross-validation. PCA was performed on both the entire spectral range (4000 to 400 cm −1 ) and on the MIR 'fingerprint' (1700-750 cm −1 and 1200-950 cm −1 ). Validation: Cross Validation. Algorithm: Singular Value Decomposition (SDV).
# Statistical analysis
Two-way ANOVA, followed by the Duncan Multiple Range Test at a significance level of α = 0.05 (IBM SPSS 20), were used to study the influence of the botanical origin (BO), year, botanical origin × year interaction, and geographical origin (GO), year, and geographical origin × year interaction on honey quality indicators.
Data were reported as the mean ± standard deviation of at least three replications. Pearson correlation coefficients, r, (at a significance level of 95%) were calculated using IBM SPSS 20 software to measure the strength of the linear relationships between honey quality indicators. Only statistically significant correlations were discussed.
# Results and discussion
# Melissopalynological analysis
The pollen analysis showed that sunflower honey had the principal pollen Helianthus annuus (48.5-74.1%), linden honey had the principal pollen Tilia tomentosa (31.8-61.4%), rapeseed honey had the principal pollen Brassica napus (46.1-81.7%), and acacia honey had the principal pollen Robinia pseudoacacia (24.3-53.1%).
According to [bib_ref] Quality evaluation of light-and dark-colored Hungarian honeys, focusing on botanical origin, antioxidant..., Bodó [/bib_ref] [bib_ref] Quality evaluation of light-and dark-colored Hungarian honeys, focusing on botanical origin, antioxidant..., Bodó [/bib_ref] , the minimum percentage of pollen required to classify honey as monofloral is 45%. The authors also noted some exceptions: for Lamiaceae origin and thyme honey, the honey requires at least 18% pollen, sage honey requires 20% pollen, as does acacia honey, according to [bib_ref] Pollen, physicochemical, and mineral analysis of Croatian acacia honey samples: Applicability for..., Uršulin-Trstenjak [/bib_ref] [bib_ref] Pollen, physicochemical, and mineral analysis of Croatian acacia honey samples: Applicability for..., Uršulin-Trstenjak [/bib_ref].
In our study, the minimum percentage of pollen was above the mentioned limit, which certifies the monofloral origin of the samples. Based on pollen analysis, the Romanian honey samples were classified in accordance with the botanical origin. The categories of botanical origin were as follows: acacia (six samples), linden (four samples), rapeseed (six samples), sunflower (four samples), and multifloral honey (four samples).
As secondary pollen contributors, Malus (apple), Pyrus (pear), Cerasus (cherry), Taraxacum (dandelion), Antirrhinum (snapdragon), Corylus (hazelnut), and Salix (willow) were found for rapeseed honey, and Brassica, Malus (apple), Pyrus (pear), Cerasus (cherry), Taraxacum (dandelion), Fragaria (strawberry), and Prunus persica (peach) were found for acacia honey. Linden honey also contains pollen from Robinia, Hypericum (St. John's wort), Rubus (blackberry, raspberry), Achilea (yarrows), and Sinapis alba (white mustard), whereas sunflower honey contains pollens from Tilia (linden), Matricaria (chamomile), Achilea (yarrow), and Silybum (milk thistle). In multifloral samples, the main pollen was Brassica (12.3-37.5%), followed by Helianthus (sunflower, 8.0-14%), Phacelia (phacelia heliotrope), Trifolium (white/sweet clover), and Mentha (mint).found that in multifloral honey, the main species contributing to pollen were Brassica napus (dominant) followed by Bifora radians (wild bishop), Tilia (linden), Prunus (plum), Plantago (fleaworts), and Echium (blueweed). For acacia honey, the authors noted that there was 5-58% Robinia pseudoacacia pollen, in rapeseed honey there was 52-93% Brassica pollen; however, in multifloral honey, the Brassica pollen content was less than 40%. In linden honey, 28.3-88.3% the Tilia pollen was reported as being lower than 45%; this was similar to our study. In sunflower honey, the Helianthus annuus pollen content was 57.7-65.5% [bib_ref] Main European unifloral honeys: Descriptive sheets, Persano Oddo [/bib_ref]. Uršulin-Trstenjak et al. (2017) [bib_ref] Pollen, physicochemical, and mineral analysis of Croatian acacia honey samples: Applicability for..., Uršulin-Trstenjak [/bib_ref] found very large oscillations with regard to pollen percentages in acacia honey: between 22% and 71% (average of 43.55%) depending on the harvest area. [bib_ref] Antioxidant activity and phenolic profile of selected organic and conventional honeys from..., Halagarda [/bib_ref] found that the accompanying pollens (pollen presented in a proportion of 15-45%) in multifloral honey are as follows: rapeseed (38.9-44.6%), white/sweet clover (17.1-39.7%), raspberry (21.8-38.9%), linden (22.7%), phacelia (19.6%), and buckwheat (17.2%). Oroian and Ropciuc (2017)found smaller variation amplitudes and lower maximums for acacia and sunflower honey compared with the present study (45.1-49.6% and 60.1-68.7%, respectively). All these differences (some of them quite large) are mainly due to the effect of climatic factors. The late spring frosts, increased nebulosity, the rains, the strong wind, and also the very high temperatures during the flowering period, cause damage to the floral structures, and thus, they prevent the bees from flying or they simply reduce the accumulation of carbohydrates available to the pollinators. There is no doubt that the spontaneous flora in the area where the beehives were located, which were different from one region to another, left their mark on the results of the melissopalynological analysis.
## Botanical origin effect on honey quality indicators
The values of some physicochemical parameters, presented in [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] (i.e., moisture, ash, electrical conductivity (EC), pH, free acidity (FA), and total sugar content (TSC)) and [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref] (i.e., hydroxymethylfurfural (HMF), total phenolic (TPC), tannin (TTC), flavonoid content (TFC), and antioxidant activity (DPPH I%)), for the five analyzed types of honey, indicate the influence of the botanical origin (BO), the study year (Year), and the BO × Year interaction.
All five types of tested honey presented a safe moisture content [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] for storage and consumption (12.21-18.74%, data not presented), falling below the maximum value (20%) established by the Codex Alimentarius standard. Moisture can negatively influence the quality of honey when stored as it creates favorable conditions for microbial activity (fermentation, for example). Hence, the presence of a large amount of water must be removed from the honey (to ensure its stability during the storage period) as it affects other quality parameters of the honey. Therefore, high humidity is related to the reduction of thermosensitive compound content (vitamin C, for example); this is because, in order to remove the moisture, the duration of the heat treatment needs to be extended, or higher temperatures are required for dehydration. In addition, as [bib_ref] Honey moisture reduction and its quality, Singh [/bib_ref] [bib_ref] Honey moisture reduction and its quality, Singh [/bib_ref] mentioned, the heat treatment, as well as the extended storage time, increased the HMF content of honey, especially for honey with a low pH. For this reason, HMF is considered to be a marker of excessive heat treatment, but it is also used to check the adulteration of honey with glucose syrup. On the other hand, [bib_ref] Alteration of antioxidative properties of longan flower-honey after high pressure, ultra-sonic and..., Chaikham [/bib_ref] [bib_ref] Alteration of antioxidative properties of longan flower-honey after high pressure, ultra-sonic and..., Chaikham [/bib_ref] reported an increase in TPC and TFC levels in thermally treated honey (at 50 - C and 70 - C), but the treatment did not have a significant effect on antioxidant activity. On the contrary, some studies [bib_ref] Effects of prolonged heating on antioxidant activity and colour of honey, Turkmen [/bib_ref] [bib_ref] Changes of antioxidant activity and formation of 5-hydroxymethylfurfural in honey during thermal..., Kowalski [/bib_ref] show a significant increase in the antioxidant activity of honey following heat treatments. Even in conditions where honey does not suffer deterioration under dehydration temperatures, the presence of high humidity is correlated with higher costs, as thermal treatment is necessary to bring the honey to a water content that does not affect its stability during storage. It is appreciated that ash content is an important quality indicator of honey; it reflects the content of mineral elements, and it is also dependent on the botanical origin as it is accepted that an ash content of less than 0.6% indicates the floral origin of the honey [bib_ref] Physical, biochemical and antioxidant properties of some Indian honeys, Saxena [/bib_ref]. All honey samples analyzed in our study had an ash content [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] (higher than that reported in the present study), an average level was reported by the authors for multifloral honey, and the minimum level was found in acacia honey. Electrical conductivity (EC) is another parameter used for honey quality control and to certify its botanical origin and purity [bib_ref] Electrical conductivity and acidity of honey, Baloš [/bib_ref]. EC is correlated with organic acid content, mineral salts, proteins, and the honey's color; a lighter color indicates lower conductivity and a darker color indicates higher conductivity [bib_ref] Electrical conductivity and acidity of honey, Baloš [/bib_ref] [bib_ref] Detection of the electrical conductivity and acidity of honey from different areas..., Yadata [/bib_ref]. In this study, EC [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] ranged from 169 to 615 µS cm −1 (data not presented), and according to Directive 2014/63/EU, the analyzed samples come from nectar honey (origin certified with EC values lower than 0.8 mS cm −1 and ash values below 0.6%). Moreover, except for sunflower honey, in 2021, the average EC was lower than 500 µS cm −1 , a value which, according to, is considered the maximum limit that is reached, with some exceptions, such as EC in the case of pure floral honey. The authors also mention the fact that EC values between 500 and 800 µS cm −1 are attributed to mixed honey. Interestingly, the multifloral honey presented EC values lower than 500 µS cm −1 . Higher values, compared with the upper limit of this study, were 0.637 mS cm −1 (for Romanian honey) and 0.689 mS cm −1 (honey from Bulgaria), whereas the minimums were set at 0.097 and 0.083 mS cm −1 [bib_ref] Phenolic and Total Flavonoid Contents and Physicochemical Traits of Romanian Monofloral Honeys, Albu [/bib_ref]. In general, the EC values reported for Romanian linden and sunflower honey were the highest, multifloral honey had average EC, and rapeseed and acacia honey had the lowest values [bib_ref] Phenolic and Total Flavonoid Contents and Physicochemical Traits of Romanian Monofloral Honeys, Albu [/bib_ref] [bib_ref] Advanced Characterization of Monofloral Honeys from Romania, Pauliuc [/bib_ref]. Nevertheless, it must be noted that there were some exceptions with regard to monofloral honey, that had even higher EC values than the 0.8 mS cm −1 limit mentioned by some authors. These include: P. aviculare (Knot weed), Gossypium sp. (cotton honey), Paliurus spina-christi (Jerusalem thorn), and Persea americana (avocado honey) [bib_ref] Legislation of honey criteria and standards, Thrasyvoulou [/bib_ref].
The pH values recorded for the analyzed honey samples [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] were within the standard limits (pH 3.40-6.10) that ensure the honey's freshness, and they were in accordance with the Codex Alimentations (2001). Unlike pH, free acidity (FA) showed very high variations [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref]. Rapeseed honey had the lowest FA (5.25 mEq kg −1 ), followed by the group consisting of linden and acacia honey (7.13 and 7.50 mEq kg −1 , respectively), whereas sunflower honey showed high FA (16.67 mEq kg −1 ). The level of free acidity in the range of 4.00 to 25.00 mEq kg −1 (below the maximum allowed of 50 mEq kg −1indicated the freshness of all samples and the absence of any fermentation processes in the tested honey. Honey generally has a slightly acidic character due to the content of organic acids (predominantly gluconic acid); this is also an indicator of the botanical and geographical origin of the honey and it contributes to the appearance (color) and taste of the honey [bib_ref] Honey: Chemical composition, stability and authenticity, Da Silva [/bib_ref]. Highly acidic honey is the result of sugar fermentation, which is responsible for both the honey's taste and its microbiological stability. Moreover, it is also positively correlated with the honey's mineral content [bib_ref] Physicochemical characteristics of honey from different origins, El Sohaimy [/bib_ref]. High pH values (which differ depending on the studied year) for linden honey, and medium values for acacia honey, were determined by Albu et al. [bib_ref] Phenolic and Total Flavonoid Contents and Physicochemical Traits of Romanian Monofloral Honeys, Albu [/bib_ref] , whereas for rapeseed, acacia, and multifloral honey, lower pH values were found in the Czech Republic and Poland [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref]. Free acidity is considered to be an indicator that the honey can be subjected to long storage times or ineffective heat treatments; this is because fermentation is a process that increases the honey's natural FA. Moreover, according to some authors, FA increased slightly in honey that was stored, especially after the first 20 months [bib_ref] Honey: Chemical composition, stability and authenticity, Da Silva [/bib_ref]. FA is related to honey content in tartaric, citric, oxalic, acetic, and other organic acids, but it also depends on nectar or bee secretions [bib_ref] Detection of the electrical conductivity and acidity of honey from different areas..., Yadata [/bib_ref].
Multifloral honey showed the highest total sugar content (TSC), 69.64 g Glu 100 g −1 , followed by acacia (63.79 g Glu 100 g −1 ), whereas the lowest level of sugars was recorded for linden honey (58.05 g Glu 100 g −1 ). Rapeseed and sunflower honey had a similar sugar content to both acacia and linden honey (62.00 and 60.82 g Glu 100 g −1 , respectively). TSC over 60% (except for linden honey, with 58.05 gGlu 100 g −1 ) showed characteristics of nectar-derived honey. The TSC content determined for multifloral honey in this study fell below the values reported by Abdulkhaliq and Swaileh (2017) [bib_ref] Physico-chemical properties of multi-floral honey from the West Bank, Abdulkhaliq [/bib_ref]. Moreover, higher sugar contents were also determined for acacia honey [bib_ref] The development of a biochemical profile of acacia honey by identifying biochemical..., Mărghitas [/bib_ref] [bib_ref] Quality assessment of honey in three different geographical areas from Republic of..., Chirsanova [/bib_ref] , which were similar to that of rapeseed honey [bib_ref] Quality assessment of honey in three different geographical areas from Republic of..., Chirsanova [/bib_ref]. However, TSC in multifloral honey fell within the limits found by . It is generally considered that nectar-derived honey has a minimum level of 60% sugar (as a sum of fructose and glucose), whereas honeydewderived honey has a smaller minimum sugar content of 45%. Regarding the Tilia species, Jacquemart et al. (2018) [bib_ref] Tilia trees: Toxic or valuable resources for pollinators?, Jacquemart [/bib_ref] mentioned that nectar is the main source of carbohydrates for flower-visiting bees, and they correlated observations of mortality among bees visiting Tilia flowers with data from the literature regarding the presence of toxic carbohydrates (mannose), of some alkaloids (mainly nicotine), and the low carbohydrate content of these flowers (the authors state that the bees die of hunger). This latter observation could justify the low level of total sugar in the analyzed linden honey samples, which could remove the uncertainty regarding its floral origin. [bib_ref] Tilia trees: Toxic or valuable resources for pollinators?, Jacquemart [/bib_ref] [bib_ref] Tilia trees: Toxic or valuable resources for pollinators?, Jacquemart [/bib_ref] analyzed the sugar content of the nectar of some linden species and found that TSC decreased in the order T. tomentosa, T. platyphillos, T × europaea, T. cordata. In Romania, the first species to bloom is T. platyphyllos; after 10-15 days, the downy lime T. cordata blooms, and after 21-22 days, T. tomentosa blooms. In total, this equates to a period of approximately 30 days (from June to July), though it varies from one locality to another and from one year to another, depending on pedoclimatic conditions [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref]. The nectar secretion of linden flowers begins at temperatures of at least 16 - C, they increase visibly after 20 - C, and they stop completely over 33 - C. Therefore, during the flowering period, drought, strong and cold winds, heavy rains that produce large amounts of water and have long duration periods (June is often a rainy month in Romania), low temperatures, and fogs cause damage to flowers and reduce or stop their nectar secretions [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref]. Low TSC in linden honey could be explained as a consequence of both the lack of sugars in linden nectar and the rainy season in June. Juan-Borrás et al. (2014) [bib_ref] Effect of country origin on physicochemical, sugar and volatile composition of acacia,..., Juan-Borrás [/bib_ref] compared linden honey from Romania with linden honey from the Czech Republic and found that the highest amount of glucose plus fructose was in honey from the Czech Republic (75% sum of glucose plus fructose). Linden honey from Romania contained a higher total sugar content (71% sum of glucose plus fructose) than was found in the present study. In the same study, the sum of the two sugars decreased in sunflower honey (76.6%), linden honey (73.9%), and acacia honey (73.7%), and was substantially higher than those found in our study. This aspect could be justified by the different climatic conditions during the years in which the honey was produced and the honey's geographical origins in the two studies.
5-hydroxymethylfurfural (HMF) is an intermediate that is formed during the Maillard reaction, wherein heat treatment is applied to acidic honey, and the honey is subjected to conditions of prolonged storage. In general, the presence of HMF is associated with a drop in honey quality. As previously discussed, treating honey with heat does not always negatively influence its quality, and some products with antioxidant activity are also formed in the Maillard reaction. However, in the case of the Maillard reaction, we can also speak of a reduction in the nutritional quality of honey, as some of the essential amino acids are destroyed [bib_ref] Honey: Chemical composition, stability and authenticity, Da Silva [/bib_ref]. Regarding the limits of HMF concentrations in honey, some authors cite the absence of HMF in fresh honey, though a maximum level of 40 mg HMF kg −1 is allowed according to European legislation, and an even higher level (by 60 mg kg −1 ) is allowed according to Brazilian legislation. Moreover, as Codex Alimentarius (2001)claims, for some tropical-origin honey, HMF should not exceed 80 mg kg −1 . The HMF content of all tested honey [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref] fell within the International Honey Commission Standards'standard recommended limit (the maximum concentration of 40 mg kg −1 ), and it was confirmed that the tested honey was not subjected to any heat treatment. The highest HMF content was found in linden honey (33.94 mg kg −1 ). Compared with our study, most of the data reported in the literature indicated lower levels (sometimes even tenfold) of HMF [bib_ref] Advanced Characterization of Monofloral Honeys from Romania, Pauliuc [/bib_ref] [bib_ref] Physicochemical parameters prediction and authentication of different monofloral honeys based on FTIR..., Pauliuc [/bib_ref] [bib_ref] Characterization and classification of Romanian acacia honey based on its physicochemical parameters..., Crăciun [/bib_ref] [bib_ref] Physicochemical parameters and microbiological status of honey produced in an urban environment..., Matović [/bib_ref] , whereas linden honey was also mentioned by [bib_ref] Physicochemical parameters and microbiological status of honey produced in an urban environment..., Matović [/bib_ref] [bib_ref] Physicochemical parameters and microbiological status of honey produced in an urban environment..., Matović [/bib_ref] due to its high HMF level (17.86 mg kg −1 ). Nevertheless, found higher HMF content levels, at 90.7 and 117.7 mg kg −1 , and they cited authors that explained that the high HMF content was as a result of the heat treatment (responsible for an increase of up to 145.5 mg kg −1 ) and inadequate storage conditions. The antioxidant activity of honey is mainly due to phenolic compounds, the main sources of which are pollen and nectar. The secondary metabolism products of plants are synthesized under abiotic and biotic stress conditions. Another role that these phenolic compounds have is that of attractants for pollinators, which are also consumed by the bees, along with the nectar, and they are later transferred into the honey. Among the compounds with antioxidant activity analyzed in the study, the total phenolic content (TPC) showed a variation that was lower compared with those reported by mention that the main phenolic compounds present in honey are flavonoids and some phenolic acids, which are compounds that influence its taste and appearance (color in particular). In a previous study, [bib_ref] Determination of flavonoid and phenolic acid contents of clover, cotton and citrus..., Hamdy [/bib_ref] [bib_ref] Determination of flavonoid and phenolic acid contents of clover, cotton and citrus..., Hamdy [/bib_ref] stated that the flavonoids in honey mainly come from nectar and pollen, but also propolis. Another, even earlier study by Tomás-Barberán et al. (1993) [bib_ref] Flavonoids in honey of different geographical origin, Tomás-Barberán [/bib_ref] showed that the ratio between propolis-derived flavonoids and pollen-nectar-derived flavonoids could be correlated with the geographical origin of the honey. The authors hypothesized that flavonoids derived from propolis are found in higher proportions in European honey (Spain and Italy), and in the temperate zones of the northern hemisphere (where poplar predominates), than in honey from other regions. Unlike our results,determined that linden honey has a flavonoid content of 32.0 µg rutin equivalents (RE) g −1 , and almost twice the lower level of flavonoids in rapeseed honey (13.5 µg RE g −1 ). Moreover, unlike the present study (in which TFC = 12.81% of TPC, only slightly higher compared with TTC), [bib_ref] Identification of flavanoids in sunflower honey, Sabatier [/bib_ref] [bib_ref] Identification of flavanoids in sunflower honey, Sabatier [/bib_ref] stated that in monofloral honey, flavonoids comprise the majority of phenolic compounds (up to 42%). Data presented in the literature regarding TFC vary greatly. Thus, Al-Farsi et al. (2018)found that in an acacia species of honey (Acacia tortilis) levels of flavonoids were 2143 mg kg −1 (1613-2890 mg kg −1 ), and TPC varied around the value of 2236 mg kg −1 (1624-2898 mg kg −1 ). In the same study, the TFC and TPC levels found in multifloral honey were 925 mg kg −1 (521-1354 mg kg −1 ) and 1066 mg kg −1 (842-1384 mg kg −1 ), respectively.
Compared with this study, the analysis of four types of honey from the Czech Republic [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref] indicated a reduction of DPPH I% in the following order: linden, multifloral, acacia, rapeseed honey, and higher values of this indicator. A similar ranking, but lower values compared with the discussed study, was obtained by the same authors for the same types of honey that were collected from Poland [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref].
Except for antioxidant activity (DPPH), all analyzed parameters were significantly influenced by the botanical origin of honey (p < 0.001).
The results presented in [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] showed that sunflower honey had the highest moisture (15.53%), ash (0.21%), electrical conductivity (483.92 µS cm −1 ), free acidity (16.67 mEq kg −1 ), phenolics (167.59 mg GAE 100 g −1 ), tannins (69.05 mg GAE 100 g −1 ), flavonoids (19.00 mg CE 100 g −1 ), and antioxidant activity (28.16%), whereas multifloral honey presented the highest total sugar content (69.64 g Glu 100 g −1 ). A significant effect of the year (p < 0.001) on honey moisture content was noted in 2022, where higher moisture was found in the honey samples (15.16%) compared with 2021 (13.65%) [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref]. A significant effect of the botanical origin × year interaction (BO × year) (p < 0.001) was highlighted by the differences in humidity recorded for sunflower and linden honey in 2021 and 2022 [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] , but this was not observed in rapeseed, multifloral, and acacia honey (p > 0.05). In addition, although in the first year of the study, there were no significant oscillations in terms of humidity among the five types of honey, they were noted in 2022. [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] showed no differences between the ash contents of the average samples collected over the two years, and no significant effect of the botanical origin × year interaction (p > 0.05) was observed. However, it was noted that there was no common trend in terms of increasing or decreasing ash content between 2021 and 2022, and the only significant oscillations in ash content were determined for linden and multifloral honey [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref]. A similar effect concerning the botanical origin (BO) in 2021 and 2022 was observed [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] ; rapeseed honey had the lowest ash content and sunflower honey had the highest ash content.
Although EC did not vary significantly over the studied years (p = 0.075), a significant effect of the botanical origin × year interaction (p = 0.007) was highlighted; the EC significantly decreased for acacia and rapeseed honey between 2021 and 2022 [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref]. No differences regarding the effect of the botanical origin, during either of the studied years, were observed.
Significant differences (p = 0.015) were recorded between the average pH levels recorded in 2021 (4.38) and 2022 . In addition, a significant effect of the botanical origin × year of study interaction was also highlighted. Related to this effect, it was observed that the strong differences between the honey pH levels recorded in 2021 (the values were arranged in four groups in accordance with homogeneity) were mitigated the following year (when only two homogeneous subsets were displayed) [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref]. Except for linden honey (pH decreased significantly between 2021 and 2022), for all other botanical origins, the honey pH levels increased between 2021 and 2022, with a significant increase for sunflower and acacia honey only [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref].
On average, the most acidic honey was collected in 2021 (9.82 mEq kg −1 ), compared with 7.50 mEq kg −1 in 2022 (p < 0.001, [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref]. The differences between the years, depending on the honey's botanical origin, were particularly apparent upon observation of the increase of FA in linden honey (from low, in 2021, to medium acidity, in 2022), and the reduction of FA in acacia honey (from medium, in 2021, to low acidity, in 2022 [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref].
Furthermore, on average, the sugar content (TSC) of honey collected in the second year of the study was greater than in 2021 [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] , and except for acacia honey, this tendency was observed across all other four types of honey [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref]. In addition, among the five types of honey, a significant variation in TSC between 2021 and 2022 was found only for linden honey. Similar variations in both FA and TSC were noted in 2021 and 2022; FA had the highest level in sunflower honey and the lowest level in rapeseed honey, whereas the highest TSC level was found in multifloral honey, and the minimum level was found in linden honey [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref].
Regarding the effect of the year, a higher HMF level was observed in 2021 (23.36 mg kg −1 ) compared with 2022 (20.07 mg kg −1 ) [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref]. As shown in [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref] , except for sunflower honey, significant differences in HMF content were registered between 2021 and 2022.
During the two study years, TPC decreased from 129.48 mg GAE 100 g −1 in 2021 to 98.65 mg GAE 100 g −1 in 2022 [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref]. Along with this reduction, the differences between honey samples were accentuated; in 2022, the honey types, based on TPC content, were rigidly segregated into four groups, starting with sunflower honey, followed by multifloral honey, rapeseed honey, linden honey, and finally, acacia honey. The most stable TPC content between 2021 and 2022 was found in rapeseed, linden, and multifloral honey.
Honey tannins (TTC) represented about 43.18% of the total phenolic content (data not presented). A more intense variation in TTC compared with TPC [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref] was found, mainly due to the effect of the year (p < 0.001), but also the honey's botanical origin (p < 0.001) and botanical origin × year interaction (p = 0.001). The highest TTC was observed in sunflower and acacia honey (69.05 and 62.73 mg GAE 100 g −1 , respectively). The average tannin content was found in multifloral honey (45.15 mg GAE 100 g −1 ), and the lowest TTC in linden honey (28.72 mg GAE 100 g −1 ). No significant differences were recorded between the tannin contents of rapeseed honey (39.00 mg GAE 100 g −1 ) and multifloral or linden honey. On average, TTC decreased in 2022 compared with 2021, similarly to TPC. The most intense reductions were recorded for sunflower honey (by 60.60 mg GAE 100 g −1 ) and multifloral honey (by 42.85 mg GAE 100 g −1 ), whereas the smallest difference was observed for rapeseed honey (12.49 mg GAE 100 g −1 ), although all were statistically assured (p < 0.05, [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref].
The total flavonoid content (TFC) represented approximately 12.81% of the TPC of the honey analyzed in the study (data not presented). Given the significant effect of the botanical origin (p < 0.001, [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref] , TFC showed a maximum level in sunflower honey (19.00 mg CE 100 g −1 ), followed by multifloral honey (17.39 mg CE 100 g −1 ), rapeseed honey (13.74 mg CE 100 g −1 ), linden honey (12.93 mg CE 100 g −1 ), and acacia honey (11.83 mg CE 100 g −1 ). Similarly to TPC, linden honey did not differ significantly compared with rapeseed or acacia honey in terms of TFC. The level of flavonoids decreased nonsignificantly from 2021 to 2022 (15.02 and 14.20 mg CE 100 g −1 , respectively). Nevertheless, the most intense reduction was observed for sunflower honey (5.81 mg CE 100 g −1 ). In addition, an exception was registered in the case of multifloral honey, where the TFC increased in 2022 by 3.08 mg CE 100 g −1 . Nevertheless, linden honey, followed by rapeseed, and acacia honey, appeared to have the most stable flavonoid content.
Unlike other honey quality indicators, antioxidant activity (DPPH I%) was influenced only by the botanical origin of honey (p = 0.001, [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref]. Two groups of values were observed: a group with high DPPH I% (24.95-28.16%), in which sunflower, linden, rapeseed, and multifloral honey were found, and a class with low antioxidant activity, in which only acacia honey (19.57%) was included. Similarly to TFC, rapeseed honey presented with an unchanged level of antioxidant activity, as did the linden and acacia samples. Botanical origin had a slightly stronger effect on DPPH I% in 2022 (25.82%) compared with 2021 (24.10%), but it was non-significant.
## Geographical origin effect on honey quality indicators
Significant variations in honey quality indicators were also observed between samples collected from different regions, and these differences varied depending on the year in which the honey was produced . Some exceptions were still observed. For example, the moisture content of the rapeseed, multifloral, and acacia honey samples varied depending on the collection area, as well as the year of the study (as previously noted). The lack of GO × year interaction was due to the similar evolution of moisture in all samples , regardless of geographical origin (in all cases the moisture contents determined in 2022 were higher than in 2021 for rapeseed and acacia, whereas for multifloral honey, the differences between 2021 and 2020 were insignificant both in TL-C and AG-AZ). In addition, for sunflower honey, the variations between the collection areas and between years in terms of ash content were insignificant . However, the evolution of the ash content over the years was different in AG-C (ash increased in 2022 by 0.10%) compared with AG-G (ash decreased in 2022 by 0.9%). The ash content of linden honey did not significantly depend on the collection area; a decrease in ash content in 2022 was evident both in TL-T and in GR-B. The ash content of rapeseed honey varied, but insignificantly, between 2021 and 2022, and for multifloral honey, no combined GO × year effect was registered, neither in the case of ash nor in the case of FA.
Among the compounds with antioxidant activity , the only deviation observed was that of flavonoids in linden honey, with insignificant variations found across the years of the study.
The indicators that showed constant evolution, with higher levels in 2022 in all regions, were moisture, HMF (with the small exception of sunflower honey from AG-G), TPC (with the two exceptions of rapeseed honey from TR-B and multifloral honey from TL-C), and TTC.
Differences between honey collected from different regions have been frequently reported in the literature [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref] [bib_ref] Physicochemical parameters as a tool for the assessment of origin of honey, Lazarević [/bib_ref] [bib_ref] Quality authentication and geographical origin classification of honey of Amhara region, Ethiopia..., Yayinie [/bib_ref] [bib_ref] Labeling regulations and quality control of honey origin: A review, Mădaş [/bib_ref] , and the differences observed in cases where the honey had the same botanical origin were attributed to the climatic conditions of the collection areas [bib_ref] Physicochemical parameters as a tool for the assessment of origin of honey, Lazarević [/bib_ref]. [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref] [bib_ref] The influence of geographical origin on honey composition studied by Polish and..., Tomczyk [/bib_ref] found that linden honey from the two regions (Poland and Slovakia) differed the most in terms of antioxidant and physicochemical parameters, and in both countries, rapeseed honey exhibited the most similar properties. Some authors even managed to classify the types of honey according to their geographical origin, with the help of a chemometric model [bib_ref] Initial study of honey adulteration by sugar solutions using midinfrared (MIR) spectroscopy..., Kelly [/bib_ref] [bib_ref] The development of a biochemical profile of acacia honey by identifying biochemical..., Mărghitas [/bib_ref] [bib_ref] Characterization and classification of Romanian acacia honey based on its physicochemical parameters..., Crăciun [/bib_ref]. [bib_ref] Labeling regulations and quality control of honey origin: A review, Mădaş [/bib_ref] [bib_ref] Labeling regulations and quality control of honey origin: A review, Mădaş [/bib_ref] stated that it is difficult to find adequate markers with which to establish the origin of honey, especially because these markers vary greatly with the botanical origin of the honey. For this reason, more sensitive methods are recommended for the identification of honey, such as chromatographic techniques; for instance, HPLC or GC-MS, but also Infrared and Raman spectroscopy. In any case, for the development of an accurate method with which to identify geographical origin, it is necessary to utilize very large databases.
## Correlation matrix between quality indicators of honey
In the analysis of the correlations between the quality indicators, it is necessary to take into account the fact that the analyzed honey samples were neither subjected to thermal treatment nor were they stored for a long time. The correlation matrix indicates that moisture correlated negatively with TTC (r = −0.352 **) and positively with DPPH I% (r = 0.273 *). Likewise, other correlations established by DPPH I% were positive ones, such as ash (r = 0.280 *), HMF (r = 0.237 *), and TFC (r = 0.292 *), though it correlated negatively with TTC (r = −0.286 *). A high level of ash was correlated with high EC (r = 0.689 **), FA (r = 0.470 **), HMF (r = 0.285 *), TPC (r = 0.464 **), TTC (r = 0.265 *), TFC (r = 0.405 **), and DPPH I% (r = 0.280 *). Except for DPPH I%, the same types of correlations were established in the case of EC. Honey with a low pH showed a high level of sugar (r = −0.307 **), TPC (r = −0.536 **), TTC (r = −0.511 **), and TFC (r = −0.593 **), and sugar was positively correlated with TFC (r = 0.289 *). Finally, positive correlations were observed between TPC, TTC, and TFC (r = 0.738 **, for TPC and TTC, r = 0.806 ** for TPC and TFC, r = 0.419 ** for TTC and TFC).
The analysis of the correlations between TTC and moisture content, in accordance with the botanical origin of the honey (data not-presented), indicated that, in all cases, the correlations were negative; this was very significant for linden (r = −0.879 ***), distinctly significant for acacia (r = −0.626 **) and sunflower (r = −0.709 **), and significant for rapeseed (r = −0.534 *). The only case in which the correlation between TTC and moisture was insignificant (although still negative) was multifloral honey (r = −0.188). To an extent, the correlations between TPC and moisture (data not presented) were similar: significantly negative for sunflower (r = −0.601 *) and rapeseed (r = −0.485 *), and distinctly significant for acacia (r = −0.714 **). In this case, for multifloral honey, the TPC correlation with moisture was insignificantly positive (r = 0.246), and for linden, although negative, it was not statistically significant (r = −0.496). Regarding the correlation between TTC, TPC, and moisture, over the two years of study (data not presented), it could be observed that in 2021, the correlations were negative, and in 2022, they were positive, but both were statistically insignificant.
Regarding the correlation between the biochemical quality parameters of honey, similarly significant positive correlations between TFC and DPPH I% in Algerian honey, as well as some Malaysian samples, were also reported by [bib_ref] Physicochemical and antioxidant properties of Algerian honey, Khalil [/bib_ref] [bib_ref] Physicochemical and antioxidant properties of Algerian honey, Khalil [/bib_ref]. The authors [bib_ref] Physicochemical and antioxidant properties of Algerian honey, Khalil [/bib_ref] also reported a significant positive correlation between TPC and DPPH I%, but with a lower correlation coefficient compared with TFC (r = 0.615 * for TPC and r = 0.888 ** for TFC). They also found strong correlations between DPPH I% and proline (r = 0.956 **) and ascorbic acid (r = 0.785 **) [bib_ref] Physicochemical and antioxidant properties of Algerian honey, Khalil [/bib_ref]. In our study, the correlation between antioxidant activity (DPPH I%) and TFC was positive and significant, whereas the correlation between antioxidant activity and TPC, although positive, was not statistically significant. These results indicate that, although found in low concentrations compared with other classes of phenolic compounds, honey s flavonoids are among the main contributors to its antioxidant activity.
The antioxidant activity of honey is due to components such as flavonoids, phenolic acids, enzymes, and vitamins, but also minerals, such as copper and iron [bib_ref] Antioxidant activity of three honey samples in relation with their biochemical components, Chua [/bib_ref]. Some authors identified 54 mineral elements in honey, classified into major, minor, and heavy metals [bib_ref] Physicochemical properties, minerals, trace elements, and heavy metals in honey of different..., Solayman [/bib_ref]. In our study, honey with high moisture contents also presented a high ash content and high antioxidant activity. This relationship between humidity and antioxidant activity could be justified by the ash content. It is known that honey contains several mineral elements, but the most important, from the point of view of antioxidant activity, are manganese (cofactor of enzymes with an antioxidant role), copper (involved in the synthesis of superoxide dismutase), zinc (involved in the production of antioxidants and synthesis of superoxide dismutase), selenium (involved in the synthesis of glutathione peroxidase) and iron (with a role in the neutralization of active radicals) [bib_ref] Vitamins and minerals functioning as antioxidants with supplementation considerations, Mcdowell [/bib_ref]. . Correlation matrix between honey moisture, ash, electrical conductivity (EC), pH, free acidity (FA), total sugar content (TSC), HMF, total phenolic (TPC), tannin (TTC), flavonoid (TFC) content, and antioxidant activity, expressed as DPPH radical inhibition activity of honey (r values are presented).
## Ash
## Atr-ftir and chemometric analysis
To compare the honey samples, FTIR spectroscopy was used as an efficient method. [fig_ref] Figure 1: Cont. [/fig_ref] shows the ATR-FTIR spectra of the tested honey with major high bands. The characteristic differences between the FTIR spectral analyses for honey samples were observed. Five major areas were identified in the MIR domain, and the fingerprint region was localized between 3600-900 cm −1 . Area 1 (3350-3600 cm −1 ) was assigned to the stretching vibrations of OH (from water, alcohols, phenols, and carbohydrates). Area 2 (2800-2900 cm −1 ) corresponds with the C-H stretching vibrations of CH 3 and CH 2 from lipids and lipid derivatives. Area 3 is complex (1500-1760 cm −1 ), and corresponds with bending vibrations C=O, C-N stretching (acids, amide I), and amide II absorption (primarily N-H bending coupled with a C-N stretching vibrational mode). Area 4 (1500-1230 cm −1 ) corresponds with stretching C-O, deformation C-H, and deformation N-H, whereas Area 5 (1230-915 cm −1 ) is assigned to C-O stretching in carbohydrates and the phosphate band (and 965 cm −1 to fructose [bib_ref] Application of FTIR-ATR spectroscopy to the quantification of sugar in honey, Anjos [/bib_ref]. The representative ATR-FTIR spectrum of sunflower honey with the described regions is presented in [fig_ref] Figure 1: Cont. [/fig_ref]. shows the exact position of the bands, together with the assignment of relevant vibrations, in specific functional groups. In carboxylic acids and alcohols, the O-H stretching vibration band is quite wide, measuring in the range of 3300-2500 cm −1 , with the largest band measuring at 3000 cm −1 [bib_ref] Analytical method development using FTIR-ATR and FT-Raman spectroscopy to assay fructose, sucrose,..., Nickless [/bib_ref] ; this is the same area as the stretching vibration region for carbon and aromatic C-H groups. The peaks around 2930 cm −1 are characteristic of C-H stretching in carboxylic acids and NH 3 stretching in free amino acids [bib_ref] Analytical method development using FTIR-ATR and FT-Raman spectroscopy to assay fructose, sucrose,..., Nickless [/bib_ref] [bib_ref] Differentiation of Anatolian honey samples from different botanical origins by ATR-FTIR spectroscopy..., Gok [/bib_ref]. The absorption band measuring at around 1640 cm −1 is due to both water and a small amount of protein molecules [bib_ref] Honey discrimination using Fourier transform-infrared spectroscopy, Bunaciu [/bib_ref]. The peaks measuring from 1175 to 940 cm −1 corresponded with C-O stretching in carbohydrates, as follows: 1148 cm −1 was specific to sucrose; 1087 cm −1 and 1043 cm −1 indicated the presence of glucose and fructose; and 983 cm −1 and 965 cm −1 indicated the presence of fructose [bib_ref] Honey discrimination using Fourier transform-infrared spectroscopy, Bunaciu [/bib_ref] [bib_ref] Quantification of saccharides in multiple floral honeys using Fourier transform infrared microattenuated..., Tewari [/bib_ref] [bib_ref] Application of FTIR-ATR spectroscopy to the quantification of sugar in honey, Anjos [/bib_ref]. In the analysed Romanian honey samples, the stretching vibration band of the C=O carboxylic acids group measured between 1760-1690 cm −1 , although the exact position of the band depended on whether the acid was saturated or unsaturated, dimerized or associated, and so on [bib_ref] Antioxidant activity of three honey samples in relation with their biochemical components, Chua [/bib_ref]. shows the exact position of the bands, together with the assignment of relevant vibrations, in specific functional groups. In carboxylic acids and alcohols, the O-H stretching vibration band is quite wide, measuring in the range of 3300-2500 cm −1 , with the largest band measuring at 3000 cm −1 [bib_ref] Analytical method development using FTIR-ATR and FT-Raman spectroscopy to assay fructose, sucrose,..., Nickless [/bib_ref] ; this is the same area as the stretching vibration region for carbon and aromatic C-H groups. The peaks around 2930 cm −1 are characteristic of C-H stretching in carboxylic acids and NH3 stretching in free amino acids [bib_ref] Analytical method development using FTIR-ATR and FT-Raman spectroscopy to assay fructose, sucrose,..., Nickless [/bib_ref] [bib_ref] Differentiation of Anatolian honey samples from different botanical origins by ATR-FTIR spectroscopy..., Gok [/bib_ref]. The absorption band measuring at around 1640 cm −1 is due to both water and a small amount of protein molecules [bib_ref] Honey discrimination using Fourier transform-infrared spectroscopy, Bunaciu [/bib_ref]. The peaks measuring from 1175 to 940 cm −1 corresponded with C-O stretching in carbohydrates, as follows: 1148 cm −1 was specific to sucrose; 1087 cm −1 and 1043 cm −1 indicated the presence of glucose and fructose; and 983 cm −1 and 965 cm −1 indicated the presence of fructose [bib_ref] Honey discrimination using Fourier transform-infrared spectroscopy, Bunaciu [/bib_ref] [bib_ref] Quantification of saccharides in multiple floral honeys using Fourier transform infrared microattenuated..., Tewari [/bib_ref] [bib_ref] Application of FTIR-ATR spectroscopy to the quantification of sugar in honey, Anjos [/bib_ref]. In the analysed Romanian honey samples, the stretching vibration band of the C=O carboxylic acids group measured between 1760-1690 cm −1 , although the exact position of the band depended on whether the acid was saturated or unsaturated, dimerized or associated, and so on [bib_ref] Antioxidant activity of three honey samples in relation with their biochemical components, Chua [/bib_ref].
There were no significant differences observed between the MIR spectra of the analyzed honey samples. Nevertheless, using chemometric analysis, honey sample discrimination was possible. For the selected regions, 1700-750 cm −1 and 1200-950 cm −1 , a good result when discriminating between multifloral honey and acacia honey was found. For the considered honey samples, the first three principal components (PCs) represented 99% of the total variance (PC1 = 92%, PC2 = 5%, and PC3 = 2%). This indicates that these three components were sufficient to provide a good separation between the groups [fig_ref] Figure 2: Two-dimensional scores obtained from the PCA of FTIR spectra of honey for... [/fig_ref]. This region includes the region at 1150-1000 cm −1 , which was frequently preferred for the spectral analysis of carbohydrates during IR spectroscopy. Linden honey was separated from acacia honey, multifloral honey, and sunflower honey, respectively. There were no significant differences observed between the MIR spectra of the analyzed honey samples. Nevertheless, using chemometric analysis, honey sample discrimination was possible. For the selected regions, 1700-750 cm −1 and 1200-950 cm −1 , a good result when discriminating between multifloral honey and acacia honey was found. For the considered honey samples, the first three principal components (PCs) represented 99% of the total variance (PC1 = 92%, PC2 = 5%, and PC3 = 2%). This indicates that these three components were sufficient to provide a good separation between the groups [fig_ref] Figure 2: Two-dimensional scores obtained from the PCA of FTIR spectra of honey for... [/fig_ref]. This region includes the region at 1150-1000 cm −1 , which was frequently preferred for the spectral analysis of carbohydrates during IR spectroscopy. Linden honey was separated from acacia honey, multifloral honey, and sunflower honey, respectively. . The location of the maxima of absorption bands FTIR in the tested honey samples. [fig_ref] Figure 1: Cont. [/fig_ref]. ATR-FTIR spectra of honey samples with different botanical origins: sunflower (a), rapeseed (b), acacia (c), linden (d), and multifloral (e) (line continuous-2021, do ed line-2022).
# Conclusions
The best-represented pollen was from Brassica napus in rapeseed honey (46.1-81.7%), followed by Helianthus annus in sunflower honey .1%), Tillia tomentosa in linden honey (31.8-61.4%), and Brassica pollen was best represented in multifloral honey (12.3-37.5%).
The honey's botanical origin represented the main variability source. Sunflower honey stood out given its multiple qualities (ash, free acidity, antioxidant compounds, and antioxidant activity). It was followed by multifloral honey, with its maximum sugar level, an appreciable content in terms of antioxidant compounds, and antioxidant activity, and linden honey, with its lower of sugar and organic acid content, but high antioxidant activity. Based on the quality parameter variations between 2021 and 2022, the least stable honey was linden honey, followed by sunflower and acacia honey, whereas rapeseed honey was the opposite. Among the phenolic compounds, flavonoids are the most strongly positively correlated with the antioxidant activity of honey, and they are found in higher concentrations in honey with lower pH levels, but high sugar content.
The chemometric method, coupled with ATR-FTIR spectra, revealed a clear separation between linden honey acacia, multifloral, and sunflower honey.
The presented results revealed a series of correlations between compounds (such as the relationship between phenolic compounds and moisture content), which, to be correctly understood and explained, require further study.
# Supplementary materials:
The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/foods12112134/s1, [fig_ref] Table 1: Honey moisture, ash, electrical conductivity [/fig_ref] : Honey moisture, ash, electrical conductivity (EC), pH, free acidity (FA), and total sugar content (TSC) influenced by the study year depending on honey botanical origin (BO); [fig_ref] Table 2: Honey hydroxymethylfurfural [/fig_ref] : Honey hydroxymethylfurfural (HMF), total phenolic (TPC), tannin (TTC), flavonoid content (TFC), and antioxidant activity (DPPH I%) influenced by the study year depending on the honey's botanical origin (BO); : Honey moisture, ash, electrical conductivity (EC), pH, free acidity (FA), and total sugar content (TSC) influenced by geographical origin (GO), year, and GO × year interaction; : Honey hydroxymethylfurfural (HMF), total phenolic (TPC), tannin (TTC), flavonoid content (TFC), and antioxidant activity (DPPH I%) influenced by geographical origin (GO), year and GO × year interaction for each botanical origin (BO); : Honey moisture, ash, electrical conductivity (EC), pH, free acidity (FA), and total sugar content (TSC) influenced by the study year depending on the geographical origin (GO) for each botanical origin (BO) of honey; : Honey hydroxymethylfurfural (HMF), total phenolic (TPC), tannin (TTC), flavonoid content (TFC), and antioxidant activity (DPPH I%) influenced by the study year depending on geographical origin (GO) for each BO of honey; [fig_ref] Figure 1: Cont. [/fig_ref] : Representative ATR-FTIR spectrum of sunflower honey (1-stretching vibrations of OH, 2-C-H stretching vibrations of CH3 and CH2 from lipids, 3 and 4-C=O, amide I, amide I, C-N from proteins, 5-C-O stretching in carbohydrates, phosphate band).
# Data availability statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy.
[fig] Foods 2023 ,: 12, x FOR PEER REVIEW 16 of 24 [/fig]
[fig] Figure 1: Cont. [/fig]
[fig] Figure 2: Two-dimensional scores obtained from the PCA of FTIR spectra of honey for the first two PCs (a), and PC3 versus PC1 (b). [/fig]
[fig] Author: Contributions: Conceptualization, I.C.M., C.M.T. and L.E.V.; methodology, C.M.T. and L.E.V.; software, I.C.M., C.M.T. and L.E.V.; validation, I.C.M., C.M.T. and L.E.V.; investigation, C.M.T., L.E.V., C.E. and S.E.; resources, C.M.T. and L.E.V.; writing-original draft preparation, I.C.M., C.M.T., L.E.V., C.E. and S.E.; writing-review and editing, I.C.M., C.M.T., C.E., S.E. and L.E.V.; supervision, I.C.M. and L.E.V. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. [/fig]
[table] Table 1: Honey moisture, ash, electrical conductivity (EC), pH, free acidity (FA), and total sugar content (TSC) influenced by botanical origin (BO), study year (Year), and BO × Year interaction (means ± SD are presented). [/table]
[table] Table 2: Honey hydroxymethylfurfural (HMF), total phenolic (TPC), tannin (TTC), flavonoid content (TFC), and antioxidant activity (DPPH I%) influenced by botanical origin (BO), study year (Year), and BO × year interaction (means ± SD are presented). BO = sunflower, linden, rapeseed, multifloral, and acacia; year = 2021 and 2022; BO-year = each group of honey from five botanical origins (sunflower, linden, rapeseed, multifloral, and acacia), which were collected in 2021 and 2022, were analyzed separately. For each BO mean, at least 12 determinations are presented (12 determinations for S, L, and M, and 18 determinations for R and A). For each year, means of at least 12 determinations were presented (2 for S, L, and M, and 3 for R and A). For each BO-Year mean, at least 6 determinations were presented (6 for S, L, and M, and 9 for R and A). Means with the same letter in each column are not significantly different at a 5% level, in accordance with Duncan s Multiple Range Test. p values for the significance of the BO and Year influence were presented and calculated in accordance with the One-Way Analysis of Variance (at a significance level of α = 0.05). p values for the significance of the BO × Year influence were calculated in accordance with the Two-Way Analysis of Variance (at a significance level of α = 0.05). S = sunflower honey, L = linden honey, R = rapeseed honey, M = multifloral honey, A = acacia honey. [/table]
|
10.1186/s12912-021-00614-2
|
CCBY
|
8186200
|
34103020
|
s2orc_pubmed_articles
|
Comparing formative and summative simulation-based assessment in undergraduate nursing students: nursing competency acquisition and clinical simulation satisfaction
Background: Formative and summative evaluation are widely employed in simulated-based assessment. The aims of our study were to evaluate the acquisition of nursing competencies through clinical simulation in undergraduate nursing students and to compare their satisfaction with this methodology using these two evaluation strategies. Methods: Two hundred eighteen undergraduate nursing students participated in a cross-sectional study, using a mixed-method. MAES© (self-learning methodology in simulated environments) sessions were developed to assess students by formative evaluation. Objective Structured Clinical Examination sessions were conducted to assess students by summative evaluation. Simulated scenarios recreated clinical cases of critical patients. Studentsṕ erformance in all simulated scenarios were assessed using checklists. A validated questionnaire was used to evaluate satisfaction with clinical simulation. Quantitative data were analysed using the IBM SPSS Statistics version 24.0 software, whereas qualitative data were analysed using the ATLAS-ti version 8.0 software.Results:Most nursing students showed adequate clinical competence. Satisfaction with clinical simulation was higher when students were assessed using formative evaluation. The main students' complaints with summative evaluation were related to reduced time for performing simulated scenarios and increased anxiety during their clinical performance.Conclusion:The best solution to reduce students' complaints with summative evaluation is to orient them to the simulated environment. It should be recommended to combine both evaluation strategies in simulated-based assessment, providing students feedback in summative evaluation, as well as evaluating their achievement of learning outcomes in formative evaluation.
# Background
Clinical simulation methodology has increased exponentially over the last few years and has gained acceptance in nursing education. Simulation-based education (SBE) is considered an effective educational methodology for nursing students to achieve the competencies needed for their professional future [bib_ref] Simulation-based learning in nurse education: systematic review, Cant [/bib_ref] [bib_ref] Simulation-based learning in higher education: a meta-analysis, Chernikova [/bib_ref] [bib_ref] Effectiveness of simulation-based nursing education depending on fidelity: a meta-analysis, Kim [/bib_ref] [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref]. In addition, simulationbased educational programs have demonstrated to be more useful than traditional teaching methodologies [bib_ref] Effectiveness of simulation-based nursing education depending on fidelity: a meta-analysis, Kim [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref]. As a result, most nursing faculties are integrating this methodology into their study plans [bib_ref] The role of simulation in developing communication and gestural skills in medical..., Bagnasco [/bib_ref]. SBE has the potential to shorten the learning curve for students, increase the fusion between theoretical knowledge and clinical practice, establish deficient areas in students, develop communication and technical skills acquisition, improve patient safety, standardise the curriculum and teaching contents, and offer observations of real-time clinical decision making [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref] [bib_ref] The effects of simulation-based learning using standardized patients in nursing students: a..., Oh [/bib_ref] [bib_ref] Recognizing and managing a deteriorating patient: a randomized controlled trial investigating the..., Stayt [/bib_ref].
SBE offers an excellent opportunity to perform not only observed competency-based teaching, but also the assessment of these competencies. Simulated-based assessment (SBA) is aimed at evaluating various professional skills, including knowledge, technical and clinical skills, communication, and decision-making; as well as higher-order competencies such as patient safety and teamwork skills [bib_ref] Simulation-based learning in nurse education: systematic review, Cant [/bib_ref] [bib_ref] Simulation-based learning in higher education: a meta-analysis, Chernikova [/bib_ref] [bib_ref] Effectiveness of simulation-based nursing education depending on fidelity: a meta-analysis, Kim [/bib_ref] [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref]. Compared with other traditional assessment methods (i.e. written or oral test), SBA offers the opportunity to evaluate the actual performance in an environment similar to the 'real' clinical practice, assess multidimensional professional competencies, and present standard clinical scenarios to all students [bib_ref] Simulation-based learning in nurse education: systematic review, Cant [/bib_ref] [bib_ref] Simulation-based learning in higher education: a meta-analysis, Chernikova [/bib_ref] [bib_ref] Effectiveness of simulation-based nursing education depending on fidelity: a meta-analysis, Kim [/bib_ref] [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref].
The main SBA strategies are formative and summative evaluation. Formative evaluation is conducted to establish students' progression during the course. This evaluation strategy is helpful to educators in improving students' deficient areas and testing their knowledge [bib_ref] A framework for evaluating and planning assessments intended to improve student achievement, Nichols [/bib_ref]. Employing this evaluation strategy, educators give students feedback about their performance. Subsequently, students self-reflect to evaluate their learning and determine their deficient areas. In this sense, formative evaluation includes an ideal phase to achieve the purposes of strategy: the debriefing [bib_ref] The benefits of debriefing as formative feedback in nurse education, Cant [/bib_ref]. International Nursing Association for Clinical Simulation and Learning (INACSL) defines debriefing as a reflective process immediately following the simulation-based experience where 'participants explore their emotions and question, reflect, and provide feedback to one another'. Its aim is 'to move toward assimilation and accommodation to transfer learning to future situations' [bib_ref] INACSL Standards of Best Practice: Simulation SM Simulation Glossary, Inacsl Standards Committee [/bib_ref]. Therefore, debriefing is a basic component for learning to be effective after the simulation [bib_ref] Successful debriefing-best methods to achieve positive learning outcomes: a literature review, Dufrene [/bib_ref] [bib_ref] A systematic review of the effectiveness of simulation debriefing in health professional..., Levett-Jones [/bib_ref]. Furthermore, MAES© (according to its Spanish initials of self-learning methodology in simulated environments) is a clinical simulation methodology created to perform formative evaluations [bib_ref] Self-learning methodology in simulated environments (MAES©): elements and characteristics, Díaz [/bib_ref]. MAES© allows evaluating specifically nursing competencies acquired by several nursing students at the same time. MAES© is structured through the union of other active learning methodologies such as self-directed learning, problem-based learning, peer education and simulation-based learning. Specifically, students acquire and develop competencies through self-directed learning, as they voluntarily choose competencies to learn. Furthermore, this methodology encourages students to be the protagonists of their learning process, since they can choose the case they want to study, design the clinical simulation scenario and, finally, actively participate during the debriefing phase [bib_ref] Self-learning methodology in simulated environments (MAES©): elements and characteristics, Díaz [/bib_ref]. This methodology meets all the requirements defined by the INACSL Standards of Best Practice [bib_ref] INACSL Standards of Best Practice: Simulation SM : Participant Evaluation, Inacsl Standards Committee [/bib_ref]. Compared to traditional simulation-based learning (where simulated clinical scenarios are designed by the teaching team and led by facilitators), the MAES© methodology (where simulated clinical scenarios are designed and led by students) provides students nursing a better learning process and clinical performance [bib_ref] Improving simulation performance through self-learning methodology in simulated environments (MAES©), Díaz Agea [/bib_ref]. Currently, the MAES© methodology is used in clinical simulation sessions with nursing students in some universities, not only in Spain but also in Norway, Portugal and Brazil [bib_ref] Perceptions about the self-learning methodology in simulated environments in nursing students: a..., Díaz Agea [/bib_ref].
In contrast, summative evaluation is used to establish the learning outcomes achieved by students at the end of the course. This evaluation strategy is helpful to educators in evaluating students' learning, the competencies acquired by them and their academic achievement [bib_ref] A framework for evaluating and planning assessments intended to improve student achievement, Nichols [/bib_ref]. This assessment is essential in the education process to determine readiness and competence for certification and accreditation [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref] [bib_ref] Summative simulated-based assessment in nursing programs, Oermann [/bib_ref]. Accordingly, Objective Structured Clinical Examination (OSCE) is commonly conducted in SBA as a summative evaluation to evaluate students' clinical competence [bib_ref] Assessment of clinical competence using an objective structured clinical examination (OSCE), Harden [/bib_ref]. Consequently, OSCE has been used by educational institutions as a valid and reliable method of assessment. OSCE most commonly consists of a 'round-robin' of multiple short testing stations, in each of which students must demonstrate defined clinical competencies, while educators evaluate their performance according to predetermined criteria using a standardized marking scheme, such as checklists. Students must rotate through these stations where educators assess students' performance in clinical examination, technical skills, clinical judgment and decision-making skill during the nursing process [bib_ref] Assessment of clinical competence using an objective structured clinical examination (OSCE), Harden [/bib_ref] [bib_ref] The objective structured clinical examination (OSCE): optimising its value in the undergraduate..., Mitchell [/bib_ref]. This strategy of summative evaluation incorporates actors performing as simulated patients. Therefore, OSCE allows assessing students' clinical competence in a real-life simulated clinical environment. After simulated scenarios, this evaluation strategy provides educators with an opportunity to give students constructive feedback according to their achieved results in the checklist [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref] [bib_ref] Summative simulated-based assessment in nursing programs, Oermann [/bib_ref] [bib_ref] Assessment of clinical competence using an objective structured clinical examination (OSCE), Harden [/bib_ref] [bib_ref] The objective structured clinical examination (OSCE): optimising its value in the undergraduate..., Mitchell [/bib_ref].
Despite both evaluation strategies are widely employed in SBA, there is scarce evidence about the possible differences in satisfaction with clinical simulation when nursing students are assessed using formative and summative evaluation. Considering the high satisfaction with the formative evaluation perceived by our students during the implementation of the MAES© methodology, we were concerned if this satisfaction would be similar using the same simulated clinical scenarios through a summative evaluation. Additionally, we were concerned about the reasons why this satisfaction would be different using both strategies of SBA. Therefore, the aims of our study were to evaluate the acquisition of nursing competencies through clinical simulation methodology in undergraduate nursing students, as well as to compare their satisfaction with this methodology using two strategies of SBA, such as formative and summative evaluation. In this sense, our research hypothesis is that both strategies of SBA are effective in acquiring nursing competencies, but student satisfaction with the formative evaluation is higher than with the summative evaluation.
# Methods
## Study design and setting
A descriptive cross-sectional study using a mixedmethod and analysing both quantitative and qualitative data. The study was conducted from September 2018 to May 2019 in a University Centre of Health Sciences in Madrid (Spain). This centre offers Physiotherapy and Nursing Degrees.
## Participants
The study included 3rd-year undergraduate students (106 students participated in MAES© sessions within the subject 'Nursing care for critical patients') and 4th-year undergraduate students (112 students participated in OSCE sessions within the subject 'Supervised clinical placements -Advanced level') in Nursing Degree. It should be noted, 4th-year undergraduate students had completed all their clinical placements and they had to approve OSCE sessions to achieve their certification.
## Clinical simulation sessions
To assess the clinical performance of 3rd-year undergraduate students using formative evaluation, MAES© sessions were conducted. This methodology consists of 6 elements in a minimum of two sessions [bib_ref] Self-learning methodology in simulated environments (MAES©): elements and characteristics, Díaz [/bib_ref] : Team selection and creation of group identity (students are grouped into teams and they create their own identity), voluntary choice of subject of study (each team will freely choose a topic that will serve as inspiration for the design of a simulation scenario), establishment of baseline and programming skills to be acquired through brainstorming (the students, by teams, decide what they know about the subject and then what they want to learn from it, as well as the clinical and non-technical skills they would like to acquire with the case they have chosen), design of a clinical simulation scenario in which the students practice the skills to be acquired (each team commits to designing a scenario in the simulation room), execution of the simulated clinical experience (another team, different from the one that has designed the case, will enter the high-fidelity simulation room and will have a simulation experience), and finally debriefing and presentation of the acquired skills (in addition to analysing the performance of the participants in the scenario, the students explain what they learned during the design of the case and look for evidence of the learning objectives).
Alternatively, OSCE sessions were developed to assess the clinical performance of 4th-year undergraduate students using summative evaluation. Both MAES© and OSCE sessions recreated critically ill patients with diagnoses of Exacerbation of Chronic Obstructive Pulmonary Disease (COPD), acute coronary syndrome haemorrhage in a postsurgical, and severe traumatic brain injury.
It should be noted that the implementation of all MAES© and OSCEs sessions followed the Standards of Best Practice recommended by the INACSL [bib_ref] INACSL Standards of Best Practice: Simulation SM Simulation Glossary, Inacsl Standards Committee [/bib_ref] [bib_ref] INACSL Standards of Best Practice: Simulation SM Simulation Design, Inacsl Standards Committee [/bib_ref] [bib_ref] INACSL Standards of Best Practice: Simulation SM Facilitation, Inacsl Standards Committee [/bib_ref] [bib_ref] INACSL Standards of Best Practice: Simulation SM Debriefing, Inacsl Standards Committee [/bib_ref]. In this way, all the stages included in a high-fidelity session were accomplished: pre-briefing, briefing, simulated scenario, and debriefing. Specifically, a session with all nursing students was carried out 1 week before the performance of OSCE stations to establish a safe psychological learning environment and familiarize students with this summative evaluation. In this pre-briefing phase, we implemented several activities based on practices recommended by the INACSL Standards Committee [bib_ref] INACSL Standards of Best Practice: Simulation SM Simulation Design, Inacsl Standards Committee [/bib_ref] [bib_ref] INACSL Standards of Best Practice: Simulation SM Facilitation, Inacsl Standards Committee [/bib_ref] and Rudolph, Raemer, and Simon [bib_ref] Establishing a safe container for learning in simulation: the role of the..., Rudolph [/bib_ref] for establishing a psychologically safe context. Although traditional OSCEs do not usually include the debriefing phase, we decided to include this phase in all OSCEs carried out in our university centre, since we consider this phase is quite relevant to nursing students' learning process and their imminent professional career.
Critically ill patient's role was performed by an advanced simulator mannequin (NursingAnne® by Laerdal Medical AS) in all simulated scenarios. A confederate (a health professional who acts in a simulated scenario) performed the role of a registered nurse or a physician who could help students as required. Occasionally, this confederate could perform the role of a relative of a critically ill patient. Nursing students formed work teams of 2-3 students in all MAES© and OSCE sessions. Specifically, each work team formed in MAES© sessions received a brief description of simulated scenario 2 months before and students had to propose 3 NIC (Nursing Interventions Classification) interventions, and 5 related nursing activities with each of them, to resolve the critical situation. In contrast, the critical situation was presented to each work team formed in OSCE sessions for 2 min before entering the simulated scenario. During all simulated experiences, professors were monitoring and controlling the simulation with a sophisticated computer program in a dedicated control room. All simulated scenarios lasted 10 min.
After each clinical simulated scenario was concluded, a debriefing was carried out to give students feedback about their performance. Debriefings in MAES© sessions were conducted according to the Gather, Analyse, and Summarise (GAS) method, a structured debriefing model developed by Phrampus and O'Donnell. According to this method, the debriefing questions used were: What went well during your performance?; What did not go so well during your performance?; How can you do better next time?. Additionally, MAES© includes an expository phase in debriefings, where the students who performed the simulated scenario establish the contributions of scientific evidence about its resolution [bib_ref] Self-learning methodology in simulated environments (MAES©): elements and characteristics, Díaz [/bib_ref]. Each debriefing lasted 20 min in MAES© sessions. In contrast, debriefings in OSCE sessions lasted 10 min and they were carried out according to the Plus-Delta debriefing tool [bib_ref] Standards of best practice: simulation standard VI: the debriefing process, Decker [/bib_ref] , a technique recommended when time is limited. Consequently, the debriefing questions were reduced to two questions: What went well during your performance?; What did not go so well during your performance?. Within these debriefings, professors communicated to students the total score obtained in the appropriate checklist. Each debriefing lasted 10 min in OSCE sessions. After all debriefings, students completed the questionnaires to evaluate their satisfaction with clinical simulation. In OSCE sessions, students had to report their satisfaction only with the scenario performed, which took part in a series of clinical stations.
In summary, [fig_ref] Table 1: Required elements for formative and summative evaluation according to the Standards of... [/fig_ref] shows the required elements for formative and summative evaluation according to the Standards of Best Practice for participant evaluation recommended by the INACSL [bib_ref] INACSL Standards of Best Practice: Simulation SM : Participant Evaluation, Inacsl Standards Committee [/bib_ref]. It should be noted that our MAES© and OSCE sessions accomplished these required elements.
## Instruments
## Clinical performance
Professors assessed students' clinical performance using checklists ('Yes'/'No'). In MAES© sessions, checklists were based on the 5 most important nursing activities included in the NICselected by nursing students. [fig_ref] Table 2: Formative evaluation [/fig_ref] shows the checklist of the most important NIC interventions and its related nursing activities selected by nursing students in the Exacerbation of Chronic Obstructive Pulmonary Disease (COPD) simulated scenario. In contrast, checklists for evaluating OSCE sessions were based on nursing activities selected by consensus among professors, registered nurses, and clinical placement mentors. Nursing activities were divided into 5 categories: nursing assessment, clinical judgment/decision-making, clinical management/nursing care, communication/ interpersonal relationships, and teamwork. [fig_ref] Table 3: Summative evaluation [/fig_ref] shows the checklist of nursing activities that nursing students had to perform in COPD simulated scenario. During the execution of all simulated scenarios, professors checked if the participants perform or not the nursing activities selected.
## Formative evaluation summative evaluation
Formative evaluation is conducted to:
- Monitor progress toward achieving outcomes.
- Provide ongoing formative feedback.
- Support participant's clinical competencies.
- Identify and close gaps in knowledge and skills.
- Assess readiness for real-world experiences.
- Facilitate teaching and learning.
Summative evaluation is conducted:
- At a discrete point in time (i.e., at the end of a course or certain time period).
- In a safe learning environment.
- After orientation to the environment and equipment.
- Appropriate level of fidelity necessary to achieve the participant outcomes.
- Utilizing a standardized format and scoring methods (i.e., utilizing a standardized scenario that includes information on when to cue, scenario length of time, and other scenario details). Use a theoretically based method to determine passing or cut scores where appropriate.
Use small group ratio, ideally a minimum ratio of one facilitator per three to five students.
## Select a valid and reliable instrument.
Provide rater training for observation-based evaluation.
Establish interrater reliability when more than one rater required.
Inform participants in advance of the evaluation.
Provide summative feedback to participant about achievement of outcomes.
## Clinical simulation satisfaction
To determine satisfaction with clinical simulation perceived by nursing students, the Satisfaction Scale Questionnaire with High-Fidelity Clinical Simulation [bib_ref] Clinical simulation as a learning tool in undergraduate nursing: validation of a..., Alconero-Camarero [/bib_ref] was used after each clinical simulation session. This questionnaire consists of 33 items with a 5-point Likert scale ranging from 'strongly disagree' to 'totally agree'. These items are divided into 8 scales: simulation utility, characteristics of cases and applications, communication, selfreflection on performance, increased self-confidence, relation between theory and practice, facilities and equipment and negative aspects of simulation. Cronbach's α values for each scale ranged from .914 to .918 and total scale presents satisfactory internal consistency (Cronbach's α value = .920). This questionnaire includes a final question about any opinion or suggestion that participating students wish to reflect after the simulation experience.
# Data analysis
Quantitative data were analysed using IBM SPSS Statistics version 24.0 software for Windows (IBM Corp., Armonk, NY, USA). Descriptive statistics were calculated to interpret the results obtained in demographic data, clinical performance, and satisfaction with clinical simulation. The dependent variables after the program in the two groups were analyzed using independent ttests. The differences in the mean changes between the two groups were analyzed using an independent t-test. Cohen's d was calculated to analyse the effect size for ttests. Statistical tests were two-sided (α = 0.05), so the statistical significance was set at 0.05. Subsequently, all students' opinions and comments were analysed using the ATLAS-ti version 8.0 software (Scientific Software Development GmbH, Berlin, Germany). All the information contained in these qualitative data were stored, managed, classified and organized through this software. All the reiterated words, sentences or ideas were grouped into themes using a thematic analysis. It should be noted that the students' opinions and comments were preceded by the letter 'S' (student) and numerically labelled.
# Results
A total of 218 nursing students participated in the study (106 students were trained through MAES© sessions, whereas 112 students were assessed through OSCE sessions). The age of students ranged from 20 to 43 years (mean = 23.28; SD = 4.376). Most students were women (n = 184; 84.4%).
In formative evaluation, professors checked 93.2% of students selected adequately both NIC interventions and its related nursing activities for the resolution of the clinical simulated scenario. Subsequently, these professors checked 85.6% of students, who participated in each simulated scenario, performed the nursing activities previously selected by them. In summative evaluation, students obtained total scores ranged from 65 to 95 points (mean = 7.43; SD = .408). Descriptive data for each scale of satisfaction with clinical simulation questionnaire, t-test, and effect sizes (d) of differences between two evaluation strategies are shown in [fig_ref] Table 4: Descriptive data, t-test and effect sizes [/fig_ref]. Statistically significant differences were found between two evaluation strategies for all scales of the satisfaction with clinical simulation questionnaire. Students´satisfaction with clinical simulation was higher for all scales of the questionnaire when they were assessed using formative evaluation, including the 'negative aspects of simulation' scale, where the students perceived fewer negative aspects. The effect size of these differences was large (including the total score of the questionnaire) (Cohen's d values > .8), except for the 'facilities and equipment' scale, which effect size was medium (Cohen's d value > .5).shows specifically descriptive data, t-test, and effect sizes (d) of differences between both evaluation strategies for each item of the clinical simulation satisfaction questionnaire. Statistically significant differences were found between two evaluation strategies for all items of the questionnaire, except for items 'I have They perform a focused respiratory exploration through appropriate pulmonary auscultation (5 points)
They recognise correctly signs and symptoms of respiratory distress, including SaO 2 (5 points)
They assess correctly haemodynamic signs and symptoms (5 points where students informed being more aware and worried in summative evaluation sessions. Most effect sizes of these differences were small or medium (Cohen's d values ranged from .238 to .709). The largest effect sizes of these differences were obtained for items 'timing for each simulation case has been adequate' (d = 1.107), 'overall satisfaction of sessions' (d = .953), and 'simulation has made me more aware/worried about clinical practice' (d = -.947). In contrast, the smallest effect sizes of these differences were obtained for items 'simulation allows us to plan the patient care effectively' (d = .238) and 'the degree of cases difficulty was appropriate to my knowledge' (d = .257).
In addition, participating students provided 74 opinions or suggestions expressed through short comments. Most students' comments were related to 3 main themes after the thematic analysis: utility of clinical simulation methodology (S45: 'it has been a useful activity and it helped us to recognize our mistakes and fixing knowledge', S94: 'to link theory to practice is essential'), to spend more time on this methodology (S113: 'I would ask for more practices of this type', S178: 'I feel very happy, but it should be done more frequently'), and its integration into other subjects (S21: 'I consider this activity should be implemented in more subjects', S64: 'I wish there were more simulations in more subjects'). Finally, students´comments about summative evaluation sessions included other 2 main themes related to: limited time of simulation experience (S134: 'time is short', S197: 'there is no time to perform activities and assess properly') and students´anxiety (S123: 'I was very nervous because people were evaluating me around', S187: 'I was more nervous than in a real situation').
# Discussion
The most significant results obtained in our study are the nursing competency acquisition through clinical simulation by nursing students and the different level of their satisfaction with this methodology depending on the evaluation strategy employed.
Firstly, professors in this study verified most students acquired the nursing competencies to resolve each clinical situation. In our study, professors verified that most nursing students performed the majority of the nursing activities required for the resolution of each MAES© session and OSCE station. This result confirms the findings in other studies that have demonstrated nursing competency acquisition by nursing students through clinical simulation [bib_ref] Effectiveness of patient simulation manikins in teaching clinical reasoning skills to undergraduate..., Lapkin [/bib_ref] [bib_ref] Revisiting "a critical review of simulation-based medical education research, Mcgaghie [/bib_ref] , and specifically nursing competencies related to critical patient management [bib_ref] Recognizing and managing a deteriorating patient: a randomized controlled trial investigating the..., Stayt [/bib_ref] [bib_ref] Nurse students learning acute care by simulationfocus on observation and debriefing, Abelsson [/bib_ref].
Secondly, students' satisfaction assessed using both evaluation strategies could be considered high in most items of the questionnaire, regarding their mean scores (quite close to the maximum score in the response scale of the satisfaction questionnaire). The high level of satisfaction expressed by nursing students with clinical simulation obtained in this study is also congruent with empirical evidence, which confirms that this methodology is a useful tool for their learning process [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref] [bib_ref] Clinical simulation as a learning tool in undergraduate nursing: validation of a..., Alconero-Camarero [/bib_ref] [bib_ref] Concept analysis of simulation as a learning strategy in the education of..., Bland [/bib_ref] [bib_ref] Psychometric testing on the NLN student satisfaction and self-confidence in learning, design..., Franklin [/bib_ref] [bib_ref] The development and psychometric testing of the satisfaction with simulation experience scale, Levett-Jones [/bib_ref] [bib_ref] Evaluating best educational practices, student satisfaction, and self-confidence in simulation: a descriptive..., Zapko [/bib_ref].
However, satisfaction with clinical simulation was higher when students were assessed using formative evaluation. The main students' complaints with summative evaluation were related to reduced time for performing simulated scenarios and increased anxiety during their clinical performance. Reduced time is a frequent complaint of students in OSCE [bib_ref] The objective structured clinical examination (OSCE): optimising its value in the undergraduate..., Mitchell [/bib_ref] [bib_ref] OSCE best practice guidelines-applicability for nursing simulations, Kelly [/bib_ref] and clinical simulation methodology [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref] [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref]. Professors, registered nurses, and clinical placement mentors tested all simulated scenarios and their checklist in this study. They checked the time was enough for its resolution. Another criticism of summative evaluation is increased anxiety. However, several studies have demonstrated during clinical simulation students' anxiety increase [bib_ref] Effects of simulation on nursing student stress: an integrative review, Cantrell [/bib_ref] [bib_ref] Causes of student anxiety during simulation: what the literature says, Nielsen [/bib_ref] and it is considered as the most disadvantage of clinical simulation [bib_ref] Simulation-based learning in nurse education: systematic review, Cant [/bib_ref] [bib_ref] Simulation-based learning in higher education: a meta-analysis, Chernikova [/bib_ref] [bib_ref] Effectiveness of simulation-based nursing education depending on fidelity: a meta-analysis, Kim [/bib_ref] [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref] [bib_ref] The role of simulation in developing communication and gestural skills in medical..., Bagnasco [/bib_ref] [bib_ref] The effects of simulation-based learning using standardized patients in nursing students: a..., Oh [/bib_ref] [bib_ref] Recognizing and managing a deteriorating patient: a randomized controlled trial investigating the..., Stayt [/bib_ref] [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref]. In this sense, anxiety may influence negatively students' learning process [bib_ref] Effects of simulation on nursing student stress: an integrative review, Cantrell [/bib_ref] [bib_ref] Causes of student anxiety during simulation: what the literature says, Nielsen [/bib_ref]. Although the current simulation methodology can mimic the real medical environment to a great degree, it might still be questionable whether students´performance in the testing environment really represents their true ability. Test anxiety might increase in an unfamiliar testing environment; difficulty to handle unfamiliar technology (i.e., monitor, defibrillator, or other devices that may be different from the ones used in the examinee's specific clinical environment) or even the need to 'act as if' in an artificial scenario (i.e., talking to a simulator, examining a 'patient' knowing he/she is an actor or a mannequin) might all compromise examinees' performance. The best solution to reduce these complaints is the orientation of students to the simulated environment [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref] [bib_ref] Summative simulated-based assessment in nursing programs, Oermann [/bib_ref] [bib_ref] Assessment of clinical competence using an objective structured clinical examination (OSCE), Harden [/bib_ref] [bib_ref] The objective structured clinical examination (OSCE): optimising its value in the undergraduate..., Mitchell [/bib_ref]. Nevertheless, it should be noted that the diversity in the satisfaction scores obtained in our study could be supported not by the choice of the assessment strategy, but precisely by the different purposes of formative and summative assessment. In this sense, there is a component of anxiety that is intrinsic in summative assessment, which must certify the acquisition of competencies [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref] [bib_ref] A framework for evaluating and planning assessments intended to improve student achievement, Nichols [/bib_ref] [bib_ref] Summative simulated-based assessment in nursing programs, Oermann [/bib_ref]. In contrast, this aspect is not present in formative assessment, which is intended to help the student understand the distance to reach the expected level of competence, without penalty effects [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref] [bib_ref] A framework for evaluating and planning assessments intended to improve student achievement, Nichols [/bib_ref].
Both SBA strategies allow educators to evaluate students' knowledge and apply it in a clinical setting. However, formative evaluation is identified as 'assessment for learning' and summative evaluation as 'assessment of learning' [bib_ref] Assessment for learning: a wider (classroom-researched) perspective is important for formative assessment..., Gavriel [/bib_ref]. Using formative evaluation, educators' responsibility is to ensure not only what students are learning in the classroom, but also the outcomes of their learning process [bib_ref] Summative and formative assessment, Taras [/bib_ref]. In this sense, formative assessment by itself is not enough to determine educational outcomes [bib_ref] Objective structured clinical examination as an educational initiative for summative simulation competency..., Wunder [/bib_ref]. Consequently, a checklist for evaluating students' clinical performance was included in MAES© sessions. Alternatively, educators cannot make any corrections in students' performance using summative evaluation [bib_ref] Summative and formative assessment, Taras [/bib_ref]. Gavriel [bib_ref] Assessment for learning: a wider (classroom-researched) perspective is important for formative assessment..., Gavriel [/bib_ref] suggests providing students feedback in this SBA strategy. Therefore, a debriefing phase was included after each OSCE session in our study. The significance of debriefing recognised by nursing students in our study is also congruent with the most evidence found [bib_ref] The benefits of debriefing as formative feedback in nurse education, Cant [/bib_ref] [bib_ref] Successful debriefing-best methods to achieve positive learning outcomes: a literature review, Dufrene [/bib_ref] [bib_ref] A systematic review of the effectiveness of simulation debriefing in health professional..., Levett-Jones [/bib_ref] [bib_ref] High-fidelity simulation debriefing in nursing education: a literature review, Neill [/bib_ref]. Nursing students appreciate feedback about their performance during simulation experience and, consequently, debriefing is considered as the most rewarding phase in clinical simulation by them [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref] [bib_ref] Systematic review of the literature on simulation in nursing education, Norman [/bib_ref]. In addition, nursing students in our study expressed they could learn from their mistakes in debriefing. Learn from error is one of the most advantages of clinical simulation shown in several studies [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref] [bib_ref] Error as allies: error management training in health professions education, King [/bib_ref] and mistakes should be considered learning opportunities rather than there being embarrassment or punitive consequences [bib_ref] Examining organizational learning in schools: the role of psychological safety, experimentation, and..., Higgins [/bib_ref].
Furthermore, nursing students who participated in our study considered the practical utility of clinical simulation as another advantage of this teaching methodology. This result is congruent with previous studies [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref]. Specifically, our students indicated this methodology is useful to bridge the gap between theory and practice [bib_ref] Rethinking theory and practice: Preregistration student nurses experiences of simulation teaching and..., Hope [/bib_ref] [bib_ref] Integration of theory and practice: experiential learning theory and nursing education, Lisko [/bib_ref]. In this sense, clinical simulation has proven to reduce this gap and, consequently, it has demonstrated to shorten the gap between classrooms and clinical practices [bib_ref] The role of simulation for learning within pre-registration nursing education-a literature review, Ricketts [/bib_ref] [bib_ref] Effectiveness of patient simulation in nursing education: meta-analysis, Shin [/bib_ref] [bib_ref] Rethinking theory and practice: Preregistration student nurses experiences of simulation teaching and..., Hope [/bib_ref] [bib_ref] Integration of theory and practice: experiential learning theory and nursing education, Lisko [/bib_ref]. Therefore, as this teaching methodology relates theory and practice, it helps nursing students to be prepared for their clinical practices and future careers. According to Benner's model of skill acquisition in nursing, nursing students become competent nurses through this learning process, acquiring a degree of safety and clinical experience before their professional careers [bib_ref] The use of simulation to address the acute care skills deficit in..., Nickless [/bib_ref]. Although our research indicates clinical simulation is a useful methodology for the acquisition and learning process of competencies mainly related to adequate management and nursing care of critically ill patients, this acquisition and learning process could be extended to most nursing care settings and its required nursing competencies.
## Limitations and future research
Although checklists employed in OSCE have been criticized for their subjective construction [bib_ref] Simulation-based assessments in health professional education: a systematic review, Ryall [/bib_ref] [bib_ref] Summative simulated-based assessment in nursing programs, Oermann [/bib_ref] [bib_ref] Assessment of clinical competence using an objective structured clinical examination (OSCE), Harden [/bib_ref] [bib_ref] The objective structured clinical examination (OSCE): optimising its value in the undergraduate..., Mitchell [/bib_ref] , they were constructed with the expert consensus of nursing professors, registered nurses and clinical placement mentors. Alternatively, the self-reported questionnaire used to evaluate clinical simulation satisfaction has strong validity. All simulated scenarios were similar in MAES© and OSCE sessions (same clinical situations, patients, actors and number of participating students), although the debriefing method employed after them was different. This difference was due to reduced time in OSCE sessions. Furthermore, it should be pointed out that the two groups of students involved in our study were from different course years and they were exposed to different strategies of SBA. In this sense, future studies should compare nursing students' satisfaction with both strategies of SBA in the same group of students and using the same debriefing method. Finally, future research should combine formative and summative evaluation for assessing the clinical performance of undergraduate nursing students in simulated scenarios.
# Conclusion
It is needed to provide students feedback about their clinical performance when they are assessed using summative evaluation. Furthermore, it is needed to evaluate whether they achieve learning outcomes when they are assessed using formative evaluation. Consequently, it should be recommended to combine both evaluation strategies in SBA. Although students expressed high satisfaction with clinical simulation methodology, they perceived a reduced time and increased anxiety when they are assessed by summative evaluation. The best solution is the orientation of students to the simulated environment.
[fig] •: With a video recording of the evaluation to allow review by multiple trained evaluators Requires formally trained facilitators (see INACSL Standard: Facilitation). [/fig]
[table] Table 1: Required elements for formative and summative evaluation according to the Standards of Best Practice for participant evaluation recommended by the International Nursing Association for Clinical Simulation and Learning (INACSL, 2016) [/table]
[table] Table 2: Formative evaluation: Checklist of the most important NIC interventions and its related nursing activities[28] selected by nursing students in Exacerbation of Chronic Obstructive Pulmonary Disease (COPD) simulated scenario Explain all procedures, including sensations likely to be experienced during the procedure Provide factual information concerning diagnosis, treatment, and prognosis Stay with the patient to promote safety and reduce fear [/table]
[table] Table 3: Summative evaluation: Checklist of nursing activities performed by nursing students in Exacerbation of Chronic Obstructive Pulmonary Disease (COPD) simulated scenario [/table]
[table] Table 4: Descriptive data, t-test and effect sizes (d) of differences between two evaluation strategies for scales of clinical simulation satisfaction (n = 218) [/table]
|
10.1371/journal.pone.0185297
|
CCBY
|
5624581
|
28968413
|
s2orc_pubmed_articles
|
UV imaging reveals facial areas that are prone to skin cancer are disproportionately missed during sunscreen application
Application of sunscreen is a widely used mechanism for protecting skin from the harmful effects of UV light.However, protection can only be achieved through effective application, and areas that are routinely missed are likely at increased risk of UV damage.Here we sought to determine if specific areas of the face are missed during routine sunscreen application, and whether provision of public health information is sufficient to improve coverage.To investigate this, 57 participants were imaged with a UV sensitive camera before and after sunscreen application: first visit; minimal pre-instruction, second visit; provided with a public health information statement.Images were scored using a custom automated image analysis process designed to identify areas of high UV reflectance, i.e. missed during sunscreen application, and analysed for 5% significance.Analyses revealed eyelid and periorbital regions to be disproportionately missed during routine sunscreen application (median 14% missed in eyelid region vs 7% in rest of face, p<0.01).Provision of health information caused a significant improvement in coverage to eyelid areas in general however, the medial canthal area was still frequently missed.These data reveal that a public health announcement-type intervention could be effective at improving coverage of high risk areas of the face, however high risk areas are likely to remain unprotected therefore other mechanisms of sun protection should be widely promoted such as UV blocking sunglasses.
# Introduction
Despite increasing sun awareness and sun protection usage, between 70-90 percent of basal cell carcinomas (BCCs) develop in sun-exposed head and neck regions, and 5 to 10 percent of all skin cancers occur on the eyelids alone [bib_ref] Incidence of eyelid basal cell carcinoma in England: 2000-2010, Saleh [/bib_ref].Specifically within England, 33610 eyelid BCCs were recorded in the 11 years between 2000 and 2010 [bib_ref] Incidence of eyelid basal cell carcinoma in England: 2000-2010, Saleh [/bib_ref].Within the eyelid area, the medial canthus, a region where the medial corner of the upper and lower eyelids meet, has been shown to be not only a particularly common site for BCC, but is also associated with poor prognosis [bib_ref] Basal cell carcinoma of the medial canthal region, Abraham [/bib_ref] [bib_ref] Treatment options and future prospects for the management of eyelid malignancies: an..., Cook [/bib_ref] [bib_ref] Risk factors for orbital exenteration in periocular Basal cell carcinoma, Iuliano [/bib_ref] [bib_ref] Basal cell carcinoma of the eyelid and periocular skin, Margo [/bib_ref].It has been postulated that the high prevalence of non-melanoma skin cancer on the eyelids is due to the skin being the thinnest on the body and hence specifically vulnerable to damage from prolonged ultraviolet (UV) light exposure, a well-established risk factor for BCCs and for squamous cell carcinomas [bib_ref] Skin cancer risk in outdoor workers: a European multicenter case-control study, Trakatelli [/bib_ref] [bib_ref] Is occupational solar ultraviolet irradiation a relevant risk factor for basal cell..., Bauer [/bib_ref] [bib_ref] Risk and protective factors for sporadic basal cell carcinoma: results of a..., Walther [/bib_ref] [bib_ref] Impact of residential UV exposure in childhood versus adulthood on skin cancer..., Ransohoff [/bib_ref] [bib_ref] Non-melanoma skin cancer: occupational risk from UV light and arsenic exposure, Surdu [/bib_ref].Therefore, the importance of adequately protecting this vulnerable area is clear, and the use of sunscreen formulations has been widely promoted.
Use of sunscreens for sun protection requires two conditions to be met: i) adequate quantities of the substance to be applied with appropriate frequency of reapplication, and ii) effective coverage of all sun exposed areas.Importantly, it has been demonstrated that even when the frequency of application and quantity applied are appropriate, the application technique in terms of coverage, is often inadequate [bib_ref] Consumer acceptability and compliance: the next frontier in sunscreen innovation, Wang [/bib_ref] [bib_ref] Teaspoon rule revisited: proper amount of sunscreen application, Isedeh [/bib_ref].Studies investigating sunscreen application to the face with emphasis on identifying commonly missed areas have very rarely been performed, however, one study in 1994 suggested inferior application to the medial canthal area in a study of 50 participants [bib_ref] Pitfalls in sunscreen application, Loesch [/bib_ref].In the 23 years since this publication, through numerous public health information streams, awareness of the risks associated with UV exposure has increased dramatically.However, the positive health benefits of UV exposure in terms of supporting the complex sensory functions of the skin, including effects on brain, neuroendocrine, and immune function, as well as the deleterious effects of vitamin D insufficiency have been widely demonstrated leading to guidelines designed to balance the benefits of UV exposure against the potential DNA damage [bib_ref] Sensing the environment: regulation of local and global homeostasis by the skin's..., Slominski [/bib_ref] [bib_ref] Ultraviolet B stimulates proopiomelanocortin signalling in the arcuate nucleus of the hypothalamus..., Skobowiat [/bib_ref] [bib_ref] UVB Activates Hypothalamic-Pituitary-Adrenal Axis in C57BL/6 Mice, Skobowiat [/bib_ref] [bib_ref] Skin Exposure to Ultraviolet B Rapidly Activates Systemic Neuroendocrine and Immunosuppressive Responses, Skobowiat [/bib_ref] [bib_ref] Cutaneous hypothalamic-pituitaryadrenal axis homolog: regulation by ultraviolet radiation, Skobowiat [/bib_ref] [bib_ref] Key role of CRF in the skin stress response system, Slominski [/bib_ref] [bib_ref] The role of vitamin D in cancer prevention, Garland [/bib_ref] [bib_ref] Do sunscreens increase risk of melanoma in populations residing at higher latitudes?, Gorham [/bib_ref]. Therefore an updated investigation into application habits is warranted.Recent data suggest that sunscreens are increasingly becoming the method of choice for sun protection meaning that it is more important than ever that sunscreen is applied effectively [bib_ref] Patient awareness and sun protection behaviour following excision of basal cell carcinoma, De Blacam [/bib_ref] [bib_ref] Sunscreen Increasingly Overshadows Alternative Sun-Protection Strategies, Koch [/bib_ref].
To date, the majority of sunscreen application publications use surrogates in place of real sun creams to determine coverage.Often these surrogates are of different texture or visibility which may influence application [bib_ref] Sunscreen application at the beach, Lademann [/bib_ref] [bib_ref] Ultraviolet radiation and the skin: Photobiology and sunscreen photoprotection, Young [/bib_ref].Recent improvements in the availability of UV sensitive cameras have opened up the possibility of investigating sunscreen application directly using actual formulations available to the public and obtaining superior image quality compared with the use of a Wood's lampor by imaging fluorescent creams [bib_ref] An evaluation of photographic methods to demonstrate the uniformity of sunscreen applied..., Grencis [/bib_ref].UV photographic imaging has thus been demonstrated as being useful as a method not only to assess sunscreen application but also to assess skin damage and drive behavioral change in sun bed users [bib_ref] Using UV photography to reduce use of tanning booths: a test of..., Gibbons [/bib_ref] [bib_ref] Sun damage in ultraviolet photographs correlates with phenotypic melanoma risk factors in..., Gamble [/bib_ref] [bib_ref] Effects of appearance-based interventions on sun protection intentions and self-reported behaviors, Mahler [/bib_ref].In this study, we have adopted the UV imaging approach to determine if skin cancer prone facial regions are ineffectively covered, and if an information based intervention could be used to improve sunscreen application.
# Materials and methods
# Ethical approval
The study was approved by University of Liverpool Ethics Review Board with approval number 201606181.Written consent was obtained from all participants prior to both phases of the trial.The individuals who's images have been used in this manuscript have given written informed consent (as outlined in PLOS consent form) to publish their case details here.
## Study design
Sample sizes were determined based on preliminary data where 4 participants were analysed from two identical non-intervention visits and one intervention visit (SD = 8).Power analyses were performed based on 95% power and 5% type I error rate; to detect an increase of at least 5% after intervention requires 57 participants assuming paired t-test.57 people (27 male and 30 female) were recruited in October to December through poster advertising and an email cascade to all staff and students of the Institute of Ageing and Chronic Disease, University of Liverpool.There were no exclusion criteria based on demographics, ethnicity or other personal criteria, however, excluded from the study were volunteers who self-identified as having allergies to sun lotion.Volunteers were required to fill in a pre-questionnaire before participating in the study (S1 Fig), requiring them to confirm whether they had used any form of sun protection previously, and whether they had any known allergies.Participants were also requested to self-identify their skin-type based on provided Fitzpatrick scale diagram.
Participants were then allowed to self-select from either SPF50 spray or SPF50 cream sunscreen formulations (both Nivea, Birmingham, UK) and instructed to apply sunscreen in their usual manner.No instructions were provided regarding volume of solution to use.UV images were acquired before and after sunscreen application.The same participants were invited to return for a second visit two weeks later, where they received an information sheet stating "Skin cancers most commonly occur on sun exposed areas of the body.Most skin cancers occur in the face with 10% of all skin cancers occurring on the sensitive eyelid area which is thinner and therefore more sensitive to sunlight.Using sunblock reduces the risk of getting skin cancer."They were then given the same sunscreen formulation as used initially and imaged before and after application.Participants were not shown the images from their first visit prior to sunscreen application.Following the second phase of the study, participants were requested to complete a second questionnaire asking about their experience and to identify behavioural trends within the population (S2 Fig).
## Image acquisition
Participants were sat in front of a plain white background and images captured using a tripod mounted DSLR (Canon EOS Rebel XTi 400D) with 60mm EF-S macro lens (both; Canon, Surrey, UK).The camera was modified to be sensitive only to UV light by replacing the internal hot mirror with a UV band pass filter (Lifepixel, Mukilteo, WA, USA) and were photgraphed with the use of an electronic flash (Vivitar Auto Thyristor Model 285 with fresnel lens removed, Vivitar, Edison, NJ, USA).Camera settings (F 2.8, ISO 1800, shutter speed 1.2s,), lighting and distances from the camera were kept constant throughout.The original images were greyscale ten million pixels images in JPG format.
# Image analysis
Manual segmentation was performed using image J (NIH, Bethesda, MA) by an observer manually drawing around areas deemed to be not covered using the freehand selection tool.Due to observers reporting difficulty in identifying glare/reflection and resultant intraobserver variability, this approach was deemed ineffective.Therefore, an automated image analysis method was developed to objectively detect, segment and quantify the areas of the face within the UV images that were not covered by sunscreen.To counteract differences in skin tones and reflection, contrast limited adaptive histogram equalisation was first applied to the images using openCV (http://opencv.org/) [bib_ref] Color Image-Enhancement through 3-D Histogram Equalization, Trahanias [/bib_ref].This divided the image into small blocks and each of these blocks were then histogram equalized using the following algorithm:
Let f be the source image and m x n matrix from the small block The pixel intensity values in a greyscale image vary from 0-255 The probability of an image having intensity x is given by: p x ¼ pixels with intensity n m x n ; n ¼ 0; 1; . . .; 255:
[formula] g m;n ¼ 255 X f m;n x¼0 p x [/formula]
Next, the dlib package (http://dlib.net/)was used to detect facial landmarks within the image.These landmarks were used as markers for determining the facial region and theletterbox region surrounding the eyes of the participant.The landmark located at the inner eye was used to define the medial canthus.The face was segmented from the original image and the relative letterbox points collated for later use.Gaussian blur (openCV code library) was then applied to the normalised image to smooth the image and reduce unwanted noise [bib_ref] A Class of Fast Gaussian Binomial Filters for Speech and Image-Processing, Haddad [/bib_ref].The Gaussian equation for a 2D image is defined as:
[formula] G x; y ð Þ ¼ Ae À ðxÀ u x Þ 2 2y 2 x À ðyÀ m y Þ 2 2y 2 y [/formula]
Where μ is the mean pixel value and θ represents the variance.
The image was then mapped to Hue Saturation Value and thresholding performed to produce a binary segmented mask of the image [bib_ref] HSV Color Space Based Segmentation of Region of Interest in Satellite Images, Ganesan [/bib_ref].Results for applying the substance were determined from this binary mask and reported as a percentage of the pixels in each image/letter box region.A binary, yes/no classification was used for the medial canthal region, where images were scored as not covered if the segmentation detected any missed skin within the defined region.
## Statistical approaches
Data were tested for normality with Shapiro Wilk test.Mann-Whitney tests were performed to compare coverage between eyelid regions and non eyelid regions relative to gender, skin type, and sun cream versus sun spray.Spearman's correlation was used to assess correlation between percentages of eyelid regions missed with percentage of rest of face.Wilcoxon tests were used to assess the improvement in coverage.Chi square test was used to assess change in medial canthal region coverage.Differences were deemed statistically significant where p<0.05.
# Results
## Eyelid regions and medial canthal areas are disproportionately missed during routine sunscreen application
In order to study how people normally apply sunscreen to their face and thereby identify problem areas, we recruited 57 participants (27 male, 30 female) in October to December 2016 and photographed them using a UV sensitive camera before (Fig 1A or after (Fig 1B sunscreen application.UV light is absorbed by melanin and sunscreen, so areas of high pigment or sunscreen coverage appear darker in these photographs whereas non-pigmented skin areas or without sunscreen appear lighter [bib_ref] Using UV photography to reduce use of tanning booths: a test of..., Gibbons [/bib_ref].Analysis of a pre-study questionnaire (S1 Fig) showed all users self-identified as previously having applied sunscreen and nine had applied moisturizer or makeup containing SPF on the day of imaging.No exclusions were required based prior application of SPF nor on skin tone/ethnicity, as fresh application of SPF50 could be detected by the UV camera in all cases (Fig 1A [fig_ref] Fig 1: Fig 1 [/fig_ref].
In order to quantify the areas that were/were not covered by sunscreen, we initially attempted to manually segment the images by drawing around the regions using image J software.However, this proved problematic with observers reporting difficulty in differentiating between reflection or glare and true failure in coverage, giving rise to high intraobserver variability in obtained data (S1A Fig) Therefore, in order to remove this subjectivity and account for glare, an automated script was developed.Initially, the images were histogram normalized to counteract differences in skin tone and flash reflection.Previously it has been demonstrated that manual segmentation is not only time-consuming, but also produces observer-dependent results, especially for a non-medical study without a clinical definition, which are more reliant on images of uniform light and tone [bib_ref] In vivo hippocampal measurement and memory: a comparison of manual tracing and..., Cherbuin [/bib_ref] [bib_ref] A comparison of automated segmentation and manual tracing for quantifying hippocampal and..., Morey [/bib_ref].We therefore elected to continue to use the automated system with the understanding that although this may increase risk of underestimation of area missed, it would reduce the risk of obtaining type I errors.Preliminary visual analyses suggested that the eyelid regions were missed with higher frequency compared with the rest of the face.As the eyelid area is particularly prone to skin cancer development [bib_ref] Epidemiologic characteristics and clinical course of patients with malignant eyelid tumors in..., Cook [/bib_ref] [bib_ref] Clinical factors influencing periocular surgical defects after Mohs micrographic surgery, Carter [/bib_ref] , a letter box region encompassing both eyelids, periorbital regions and the bridge of the nose was isolated from the other facial regions to specifically assess whether this observation reflected a true trend in application behaviour (yellow box in Fig 1C .Analysis of these data revealed the median percentage of the whole face missed to be 10% (range 0-22%, Fig [fig_ref] Fig 2: Fig 2 [/fig_ref].Interestingly, a significantly higher percentage of the eyelid region was missed compared with the rest of the face not including the eyelid regions (eyelid median 14%, median non-eyelid 7%, p<0.001Mann-Whitney test, Fig [fig_ref] Fig 2: Fig 2 [/fig_ref].Within the cohort, there was a weak positive correlation between the amount of the face not including eyelid missed and the percentage of eyelid region missed (r 2 = 0.19, Spearman correlation 0.84, p<0.01,Fig 2B indicating the level of eyelid coverage is more generally related to the overall sunscreen application ability.Comparison between males and females revealed no significant difference between genders (Fig [fig_ref] Fig 2: Fig 2 [/fig_ref].However, analysis on the basis of self-reported skin type indicated that those with skin types 1 and 2 performed slightly worse than those with skin types 3 or higher, these As the periorbital/medial canthal regions (magenta box in Fig [fig_ref] Fig 1: Fig 1 [/fig_ref] are particularly at risk for more aggressive BCC [bib_ref] Basal cell carcinoma of the medial canthal region, Abraham [/bib_ref] [bib_ref] Epidemiologic characteristics and clinical course of patients with malignant eyelid tumors in..., Cook [/bib_ref] [bib_ref] Clinical factors influencing periocular surgical defects after Mohs micrographic surgery, Carter [/bib_ref] , a secondary binary analysis (covered/not covered) was performed on this region, revealing that 44 of the 57 failed to cover this region (Fig [fig_ref] Fig 2: Fig 2 [/fig_ref].
In addition to its use for identifying sunscreen application, UV photography also enables the visualization of areas of existing sun damage in lighter complexion individuals.Careful examination of our before sunscreen application images revealed numerous examples of sun damage spots in eyelid regions supporting the concept that this area is at risk for sun damage (Fig 1F .
## Provision of simple risk indication information improved the sunscreen coverage of eyelid regions
Next we sought to determine if increased awareness of eyelid cancer risk would be sufficient to drive an improved coverage of the at risk areas.Various methods of behavioural intervention to promote sun-protection have been previously reported, with varying levels of success.Methods have included written information [bib_ref] Effects of appearance-based interventions on sun protection intentions and self-reported behaviors, Mahler [/bib_ref] [bib_ref] Beyond the usual suspects: target group-and behavior-specific factors add to a theory-based..., Schuz [/bib_ref] , photo aging imaging interactive presentation [bib_ref] Evaluation of a multicomponent appearance-based sun-protective intervention for young women: Uncovering the..., Jackson [/bib_ref] and psychosocial modelling to assess sun protective behaviour [bib_ref] A psychosocial model of sun protection and sunbathing in young women: The..., Jackson [/bib_ref].In this study we chose to use written information as this approach could easily be adopted in the labelling of bottles, moreover studies have shown that awareness of risk is sufficient to drive behavioural changes specifically related to sunscreen use [bib_ref] Worry about skin cancer mediates the relation of perceived cancer risk and..., Kiviniemi [/bib_ref].
The same 57 participants were invited to return for a second visit, the study was carried out as previously however prior to sunscreen application participants were provided an information sheet stating; "Skin cancers most commonly occur on sun exposed areas of the body.Most skin cancers occur in the face with 10% of all skin cancers occurring on the sensitive eyelid area which is thinner and therefore more sensitive to sunlight.Using sunblock reduces the risk of getting skin cancer."
Image analysis revealed at the second visit participants missed a median of 8% of the whole face compared with 10% in the initial visit (Fig 3A [fig_ref] Fig 3: Fig 3 [/fig_ref].Range 0-20%, Wilcoxon Signed Ranks test Z -3.63, p<0.001).The percentage of the eyelid region missed in visit 2 showed a statistically significant improvement to a median 10% compared with 14% missed in visit 1 (Fig 3B, eyelid regions from all participants in S4 Fig, range 0-24%, Z -4.66 p<0.001), the noneyelid regions showed a below significant improvement to 6% missed from 7% (Fig 3B, range 0-19%, Z-1.82, p = 0.069).
Encouragingly, analysis on a per person basis revealed the greatest improvement in eyelid coverage to be to be observed in those who had initially achieved the lowest coverage (Fig 3C, r 2 = 0.34 Spearman correlation 0.66, p<0.01,Fig 3D, 11/57 missed >20% in visit 1 compared with 5/57 in visit 2).Although statistically significant, the overall eyelid region improvement was relatively small and coverage of the medial canthal area remained poor with 37 of 57 participants still failing to cover this region (Fig 3E [fig_ref] Fig 3: Fig 3 [/fig_ref].
During the study, participants could choose between use of sun cream or sun spray and were asked to use the same method on both visits.No significant difference was observed in regions missed between users of sun cream and sun spray (Fig 4A, Mann Whitney test).This was unsurprising as all sun spray users first sprayed the solution into their hands and then applied it to their face, as per the manufacturer suggested application method, rather than spraying directly onto the face.
In order to gain an insight into potential reasons why the eyelid regions were missed more frequently, participants were invited to complete a short online questionnaire 1-2 weeks after completing both parts of the study (S2 Fig) Response rate was 78% (44/57).Participants reported approximately equal ease of application (Fig 4B, median score of 80/100 for cream, 73/100 for spray) however, there was a slight perception by the sun cream users of more
# Discussion
In this study, we have demonstrated that although overall sunscreen application to the face regularly achieves high levels of coverage, problem areas still exist in the eyelid regions, particularly the medial canthus.However, we demonstrate that provision of a short information sheet is sufficient to drive a small but statistically significantly improved coverage to eyelids regions, which was particularly effective in those individuals that initially performed poorly.These data highlight the need for greater public awareness of eyelid cancer risk and suggest that simple pubic information intervention could be effective.However, our data also demonstrate that, despite improved awareness, the medial canthal areas were still frequently ineffectively covered and therefore use of alternate strategies for protection, such as UV blocking sunglasses, should be promoted wherever possible.This will have the dual effect of protecting cancer at risk areas and also protecting eyes from UV damage, thereby reducing incidence of corneal damage, macular degeneration and cataract formation [bib_ref] Ultraviolet radiation as a risk factor for cataract and macular degeneration, Roberts [/bib_ref] [bib_ref] UV-blocking spectacle lens protects against UVinduced decline of visual performance, Liou [/bib_ref].The participants in our study were drawn from University students and staff members and therefore there is a selection biased toward well-educated participants.Moreover, our participants reflected the local University population and were skewed toward lighter skin types.This has two important implications for the generalizability of our findings.First; one would predict that our cohort would have a base line cancer-risk awareness levels and as such are likely to be vigilant when applying sun protection [bib_ref] Worry about skin cancer mediates the relation of perceived cancer risk and..., Kiviniemi [/bib_ref] [bib_ref] Factors associated with sun protection compliance: results from a nationwide cross-sectional evaluation..., Sattler [/bib_ref].This may suggest that the disproportionately poor coverage of the eyelid region could be even more striking in the general population.It should be noted, that in our post questionnaire just under half of the respondents did not cite a specific reason for missing the eyelids so, despite a perceived general awareness, their application was still relatively poor.Second; our cohort may respond more to our information based intervention strategy than a broader cross section of the public and therefore we may be over estimating the magnitude of the improvement.However, information based intervention is a standard and proven efficacious approach in wide variety of contexts in diverse groups of patients/participants and as such we do not believe this to be a major limitation here [bib_ref] Information sheets for patients with acute chest pain: randomised controlled trial, Arnold [/bib_ref] [bib_ref] Effect of an intervention on observed sun protection by vacationers in a..., Buller [/bib_ref] [bib_ref] Effects of tailored risk communications for skin cancer prevention and detection: the..., Glanz [/bib_ref].
When considering the data within this study it is important to consider the behaviour modifying effect wearing sun protection has.If sunscreen application is only performed when preparing for an extended period in the sun, ineffective application will lead to a false sense of security in terms of perceived protection.Our data strongly indicate that in routine application, the eyelid region and particularly the medial canthal areas are relatively poorly protected hence at increased risk.The belief that the whole face is protected may repeatedly increase UV exposure to vulnerable areas that have been missed as people spend longer exposed to the sun.
An important additional consideration is that any public health message must weigh the benefits of reduced UV induced DNA damage against the positive health benefits of UV-B induced vitamin D production.Specifically, insufficient levels of vitamin D have been shown to occur in 50% of the UK adult population with seasonal variations showing severe deficiencies during winter and spring [bib_ref] Vitamin D deficiency: Diagnosis and patient centred management, Kalra [/bib_ref] [bib_ref] Hypovitaminosis D in British adults at age 45 y: nationwide cohort study..., Hypponen [/bib_ref].Low vitamin D levels are associated with increased risk of certain cancer subtypes, and other health risks including bone disease, muscle weakness and diabetes mellitus [bib_ref] The role of vitamin D in cancer prevention, Garland [/bib_ref] [bib_ref] Do sunscreens increase risk of melanoma in populations residing at higher latitudes?, Gorham [/bib_ref] [bib_ref] The case for a comprehensive national campaign to prevent melanoma and associated..., Manson [/bib_ref] [bib_ref] Diet, environmental factors, and lifestyle underlie the high prevalence of vitamin D..., Zgaga [/bib_ref] [bib_ref] In defense of the sun: An estimate of changes in mortality rates..., Grant [/bib_ref].Dietary vitamin D supplementation can reduce some of these health risks [bib_ref] The role of vitamin D in cancer prevention, Garland [/bib_ref] [bib_ref] Diet, environmental factors, and lifestyle underlie the high prevalence of vitamin D..., Zgaga [/bib_ref] , however, it has also been demonstrated that relatively short UV exposure is sufficient to produce the daily recommended serum vitamin D levels [bib_ref] Estimates of beneficial and harmful sun exposure times during the year for..., Samanek [/bib_ref].This is reflected in National Institute for Clinical Excellence (NICE) guidelines which recommend short periods of non-protected exposure to ensure adequate vitamin D production.The findings presented here do not countermand these recommendations, but rather that emphasis be made that, when protection is required, extra attention be paid that sunscreens are applied in a way that does not repeatedly leave the same areas unprotected each time.
Taken together, our findings strongly suggest that a public information campaign is warranted to stress the importance of eye protection from the sun.The ongoing problem of irritation caused by the sun cream/spray needs to be addressed with an appropriate education system put in place to educate the public in newer tear-free formulations [bib_ref] A tear-free, SPF50 sunscreen product, Yan [/bib_ref] , whilst the importance of seeking alternate protection mechanisms should be further emphasized.
[fig] Fig 1: Fig 1. UV imaging as a mechanism to identify regions of incomplete sunscreen application.a) UV images before and after sunscreen application.b) UV images of eyelid region showing the impact of SPF15 makeup compared with SPF50 sun cream.c) UV images analysis steps; top left and middle panels, before and after images from sunscreen application.Top right panel, facial landmarks identified by dlib package: green box; cropped facial region, yellow box; eyelid region, magenta boxes; medial canthal areas.Bottom left, cropped facial region.Bottom middle, hue saturation value (HSV) heat map produced from grayscale image, bottom right binary mask generated from thresholding the HSV heat map.https://doi.org/10.1371/journal.pone.0185297.g001 [/fig]
[fig] Fig 2: Fig 2. Eyelid regions and medial canthal areas are disproportionately missed during routine sunscreen application.a) Box and whisker plot of percentage of indicated region missed as detected by automated image analysis software.Line represents median, boxes represent 25 to 75 th percentile, whiskers 5 th and 95 th percentile, outliers denoted by black dots, n = 57.* denote significant difference between bracketed groups p<0.01 Mann Whitney test b) Dot plot of percentage area of rest of face missed versus percentage missed in eyelid region.Each dot represents one individual from the trial, n = 57.Spearman correlation coefficient 0.84, p<0.01.c) Box and whisker plots comparing male and females sunscreen application effectiveness plotted as percentage missed of the indicated regions, plotted as in b), d) box and whisker plot comparing application with self-assessed skin type, grouped as types 1 or 2 (n = 42) compared with types 3 or higher (n = 15), plotted as for b), e) Bar chart of percentage of population that either completely covered or failed to cover medial canthal regions.f) Representative UV images of six participants eyelid regions without sunscreen application.Note the dark spots indicating presence of UV damaged skin i.e, areas of pigmentation deep in the dermis that are invisible to the naked eye but visible to UV photography.https://doi.org/10.1371/journal.pone.0185297.g002 [/fig]
[fig] Fig 3: Fig 3. Provision of a simple information sheet improves eyelid coverage.a) Representative images without SPF (left), after SPF50 application during routine application (visit 1), or after receiving cancer risk information sheet (visit 2).b) Box and whisker plot of percentage of region missed as detected by automated image analysis software.Line represents median, boxes represent 25 to 75 th percentile, whiskers 5 th and 95 th percentile, outliers denoted by black dots, n = 57.* denote significant difference between bracketed groups, p<0.01 Wilcoxon Signed Ranks test.c) Dot plot showing percentage coverage change against initial percentage eyelid area missed for all participants (n = 57).Values above 0 on this plot indicate improved coverage.Pearson correlation coefficient 0.66, p<0.01.d) Bar chart showing percentage of study population who failed to cover either >20% of their eyelid regions (white), 10-20% (grey) or 0-10%.e) Bar chart of percentage of population that either completely covered or failed to cover medial canthal regions.f) Bar chart of medial canthal area coverage comparing on an individual basis coverage in visit 1 and visit 2, x 2 p>0.05.https://doi.org/10.1371/journal.pone.0185297.g003 [/fig]
[fig] Fig 4: Fig 4.There is no difference between application of sun cream or sun spray.a) Box and whisker plot of percentage of region missed as detected by automated image analysis software.Line represents median, boxes represent 25 to 75 th percentile, whiskers 5 th and 95 th percentile, outliers denoted by black dots, n = 57.p>0.05 Mann Whitney test b) Box and whisker plot of participants Likert scale responses regarding ease of sun cream/spray application.Line represents median, boxes represent 25 to 75 th percentile, whiskers 5 th and 95 th percentile, n = 40.c, d and e) pie charts representing participants' responses to indicated questions.https://doi.org/10.1371/journal.pone.0185297.g004 [/fig]
|
10.3390/diagnostics12081935
|
CCBY
|
9406318
|
36010285
|
s2orc_pubmed_articles
|
Identification of EGFR as a Biomarker in Saliva and Buccal Cells from Oral Submucous Fibrosis Patients—A Baseline Study
Citation: Moorthy, A.; Venugopal, D.C.; Shyamsundar, V.; Madhavan, Y.; Ravindran, S.; Kuppuloganathan, M.; Krishnamurthy, A.; Sankarapandian, S.; Ganapathy, V.; Ramshankar, V. Identification of EGFR as a Biomarker in Saliva and Buccal Cells from Oral Submucous Fibrosis Patients-A Baseline Study.
# Introduction
Oral cancer accounts for more than 30% of all cancers and poses a significant burden in the Indian subcontinent [bib_ref] Challenges of the Oral Cancer Burden in India, Coelho [/bib_ref]. Tobacco and betel nut chewing habits are the main contributing factors to its increased incidence. Leukoplakia, erythroplakia, and palatal changes due to reverse smoking, which are tobacco-associated, and Oral Submucous Fibrosis (OSMF), associated with arecanut use, have been grouped within oral potentially malignant disorders (OPMD). Among the various OPMD, OSMF is highly prevalent in South-East Asia, with a high malignant transformation rate of 7-13% [bib_ref] Imtiyaz Oral Submucous Fibrosis -Current Concepts in Etiopathogenesis. People's, Gupta [/bib_ref]. Early identification of high-risk OPMD using potential biomarkers can reduce the carcinomatous transformation, thereby reducing the morbidity and mortality associated with the disease [bib_ref] Epidermal Growth Factor Receptor Detection in Serum and Saliva as a Diagnostic..., Zanotti [/bib_ref] [bib_ref] Use of Saliva Protein Biomarkers for Diagnosis of Oral Cavity Cancer, Chu [/bib_ref]. Over recent years, biomarkers in various samples, such as serum, saliva, fresh tissue, exfoliated buccal cells, and formalin-fixed paraffin blocks (FFPE), have been studied [bib_ref] Epidermal Growth Factor Receptor Copy Number in Potentially Malignant Oral Disorders and..., Bagan [/bib_ref] [bib_ref] Epidermal Growth Factor Receptor (EGFR) as a Target in Cancer Therapy: Understanding..., Ciardiello [/bib_ref] [bib_ref] Clinical Significance of EGFR, Her-2 and EGF in Oral Squamous Cell Carcinoma:..., Bernardes [/bib_ref]. Among the various samples used for diagnostic purposes, saliva is found to be a protein-rich fluid that is directly in contact with the oral lesions and represents an ideal source for the development of biomarkers for early detection, monitoring the progression of OPMD, and also in assessing the treatment response [bib_ref] Use of Saliva Protein Biomarkers for Diagnosis of Oral Cavity Cancer, Chu [/bib_ref]. Numerous potential biomarkers have been evaluated in saliva and are overexpressed in OPMD and OSCC [bib_ref] Epidermal Growth Factor Receptor Copy Number in Potentially Malignant Oral Disorders and..., Bagan [/bib_ref].
Epidermal Growth Factor Receptor (EGFR) is a tyrosine kinase receptor involved in various cellular activities [bib_ref] Elevated Levels of Transforming Growth Factor Alpha and Epidermal Growth Factor Receptor..., Grandis [/bib_ref]. EGFR (also known as ErbB-1/HER1) is a 170 KDs transmembrane glycoprotein belonging to the ErbB family of receptor tyrosine kinases (RTK) [bib_ref] EGF Activates Its Receptoeractions That Autoinhibit Ectodomain Dimerization, Ferguson [/bib_ref] [bib_ref] Signalling Mechanisms and Therapeutic Opportunities, Yarden [/bib_ref] , involved in signal pathways associated with cancer development and progression and associated with several gene mutations [bib_ref] Epidermal Growth Factor Receptor in Relation to Tumor Development: EGFR-Targeted Anticancer Therapy, Okamoto [/bib_ref] [bib_ref] Epidermal Growth Factor Receptor in Relation to Tumor Development: EGFR Gene and..., Mitsudomi [/bib_ref]. Overexpression of EGFR has an unfavorable clinical outcome, poor prognosis, and low survival rates in Oral Squamous Cell Carcinoma (OSCC) [bib_ref] Levels of TGF-Alpha and EGFR Protein in Head and Neck Squamous Cell..., Grandis [/bib_ref] [bib_ref] Reduction in and Preventive Effects for Oral-Cancer Risk with Antidepressant Treatment, Chung [/bib_ref] [bib_ref] Epidermal Growth Factor Receptor Copy Number Alterations Correlate with Poor Clinical Outcome..., Temam [/bib_ref]. The reliability of EGFR is proven because the molecular targets inhibiting EGFR receptors are currently under clinical trial [bib_ref] Epidermal Growth Factor Receptor (EGFR) as a Target in Cancer Therapy: Understanding..., Ciardiello [/bib_ref]. EGFR has been found to play a role in cell invasion by induction into the Epithelial to Mesenchymal Transition [bib_ref] Activation of EGFR Promotes Squamous Carcinoma SCC10A Cell Migration and Invasion via..., Zuo [/bib_ref] [bib_ref] Epithelial-Mesenchymal-Transition Induced by EGFR Activation Interferes with Cell Migration and Response to..., Holz [/bib_ref]. Hence, EGFR is overexpressed in OSCC and OPMD. EGFR was found to be overexpressed in leukoplakia and oral submucous fibrosis, thus suggesting a valuable diagnostic marker for early malignancy [bib_ref] Quantitative Immunoexpression of EGFR in Oral Potentially Malignant Disorders: Oral Leukoplakia and..., Meka [/bib_ref]. Similar studies in Oral Submucous Fibrosis showed increased immunohistochemical expression of EGFR [bib_ref] Evaluation of TGF-Alpha and EGFR Expression in Oral Leukoplakia and Oral Submucous..., Srinivasan [/bib_ref] [bib_ref] Quantitative Estimation of PCNA, c-Myc, EGFR and TGF-Alpha in Oral Submucous Fibrosis-an..., Srinivasan [/bib_ref] and Oral Squamous Cell Carcinoma [bib_ref] Evaluation of Epidermal Growth Factor Receptor Expression by a New Scoring System..., Verma [/bib_ref]. In a study done by Zanotti et al., comparing serum and salivary EGFR in oral cancer, it was found that saliva was a more reliable diagnostic and prognostic marker compared to serum [bib_ref] Epidermal Growth Factor Receptor Detection in Serum and Saliva as a Diagnostic..., Zanotti [/bib_ref]. However, there are contradictory results reported with salivary EGFR regarding its reliability as a diagnostic and prognostic marker in saliva [bib_ref] Clinical Significance of EGFR, Her-2 and EGF in Oral Squamous Cell Carcinoma:..., Bernardes [/bib_ref].
To the best of our knowledge, the existing literature on EGFR expression in saliva and exfoliated buccal cells of OSMF and OSCC is limited. Immunoexpression of EGFR in OPMD with dysplasia compared to expression levels in different grades of OSCC will further aid in understanding the role of this biomarker in monitoring the prognosis of OPMD. Hence, the study aims to assess the expression of EGFR in saliva and exfoliated buccal cells of OSMF and OSCC patients. The current study also attempts to validate the expression of EGFR in tissue samples of OSMF and OSCC using real-time polymerase chain reaction (RT -PCR) and the immunoexpression of EGFR in OPMD with dysplasia and OSCC using IHC.
# Materials and methods
## Patient samples
The study was approved by the Institutional Ethical Committee (IEC No. CSP/19/MAY/ 77/169) and was conducted at the Department of Oral Medicine and Radiology, Sri Ramachandra Institute of Higher Education and Research, from June until August 2019. Written informed consent was obtained from all the study participants. Patient demographic details, medical history, habits, and details of clinical examination were recorded in the proforma.
A total of n = 49 subjects were enrolled for the study and divided into Group A (n = 24), comprising Oral Submucous Fibrosis, Group B (n = 10), comprising OSCC, and Group C (n = 15), comprising healthy controls. Subjects who were less than 18 years of age, patients who were currently under treatment or previously treated for OSCC or OSMF, and those who had not given informed consent for participation were excluded from the study. Biopsy was performed in Group A and Group B patients as a routine diagnostic workup for histopathological confirmation and further treatment planning.
## Sample collection
Unstimulated whole saliva was collected from Group A, B, and C by passive expectoration and spit into a 50-mL sterile tube containing proteinase inhibitor (Proteinase inhibitor cocktail (P2714, Sigma Aldrich). Patients were asked to refrain from drinking, eating, chewing tobacco, or smoking 1 h before collecting saliva. Whole saliva samples were transferred to 1.5-mL sterile microtubes and centrifuged for 3 min at 13,000 rpm. The supernatants were immediately aliquoted and stored at −80 - C within 60 min after saliva collection. The buccal exfoliated cells from the same patients were collected using a sterile buccal swab, followed by 2 min rinse with 10 mL distilled water. The samples were transferred to 1.5-mL sterile microtubes and centrifuged for 3 min at 13,000 rpm, immediately aliquoted, and stored at −80 - C [bib_ref] 2D-DIGE-Based Proteomic Profiling with Validations Identifies Vimentin as a Secretory Biomarker Useful..., Sivagnanam [/bib_ref].
In addition, formalin-fixed paraffin-embedded (FFPE) sections fom healthy controls (n = 11) and retrospective samples of OPMD with dysplasia (n = 56) and OSCC (n = 106) were obtained.
## Measurement of egfr levels
Saliva and buccal samples were thawed on ice and centrifuged at 3000 rpm at 48 - C before analysis. EGFR concentrations were determined using a commercial sandwich enzyme-linked immunosorbent assay (ELISA), according to the manufacturer's instructions (Human EGFR ELISA kit; Wuhan Fine Biotech Co., Ltd., Hubei, China). The plate was washed twice before adding standard, sample, and control (zero) wells. 100 µL standard and sample was added to each well and incubated for 90 min at 37 - C, followed by aspiration, and further the plates were washed twice. 100 µL of Biotin-labeled antibody working solution was added to each well and incubated for 60 min at 37 - C, aspirated, and the ELISA plate was washed thrice. Additionally, 100 µL of HRP-Streptavidin Conjugate (SABC) working solution was added to each well, incubated for 30 min at 37 - C, then aspirated and the plate was washed five times. The next step was followed by the addition of 90 µL TMB substrate and incubated for 15-30 min at 37 - C. Subsequently, 50 µL stop solution was added and the absorbance was read at 450 nm immediately. Experiments were performed in duplicates and the average value was taken for the analysis. EGFR levels were determined using standard curves, reading the optical density at 450 nm on an automatic plate reader (ROBONiK; readwell TOUCH, ELISA Plate Analyser). Results were tabulated and subjected to analysis by SPSS 15 software [bib_ref] Enzyme-Linked Immunosorbent Assay of Epidermal Growth Factor Receptor in Lung Cancer: Comparisons..., Pfeiffer [/bib_ref].
## Immunohistochemistry
Immunohistochemistry was performed on 4 µm sections obtained from formalinfixed paraffin-embedded tissue (FFPE) samples. Sections were taken on slides coated with 3-aminopropyl-triethoxysilane (APES). The sections were deparaffinized in xylene and rehydrated using absolute alcohol. Endogenous peroxidase activity was quenched by immersing the sections for 10 min in 0.03% hydrogen peroxide in distilled water, followed by a distilled water wash. Antigen retrieval was done with 0.05 M Tris EDTA Buffer (pH-9) in a pressure cooker for 20 min. Sections were pre-incubated with 2% bovine serum albumin (BSA) for 40 min. The sections were incubated with primary antibody EGFR (PU335: polyclonal rabbit anti-EGFR, Biogenex, CA, USA) overnight at 4 - C in 100% moisture. The BioGenex Super Sensitive TM Detection System (Biogenex, CA, USA) was used to detect expression. Hematoxylin-counterstained sections were dehydrated using ascending grades of isopropyl alcohol and xylene and mounted in DPX. Known positive controls and negative controls were used. The expression of EGFR was graded and compared to the absolute normal oral mucosa. In 10 high power fields (40×), several positive cells were counted in the epithelium and connective tissue (connective tissue cells such as fibroblasts and inflammatory cells), and the % positivity was computed. Counting was done on a computer display using the software ProgRes CapturePro v2.8.8. Briefly, EGFR expression was assessed semi-quantitatively by evaluating the percentage of epithelial CT cells and, based on the expression levels of the respective proteins, they were classified as follows: (i) Mild Positive-10% to 50% EGFR expression, (ii) Intermediate Positive-60% to 90% EGFR expression, and (iii) Strong Positive-100% EGFR expression [bib_ref] Proteomics Based on 2D Gel Electrophoresis and Mass Spectrometry with Validations: Identification..., Venugopal [/bib_ref]. The staining intensity was measured at several levels of the epithelium (basal, stratum spinosum, and superficial). Similarly, expression in the connective tissue was also counted. Scoring for Diagnostics 2022, 12, 1935 4 of 12 IHC was done by an oral pathologist who was blinded to the clinical details of the included patient samples.
## Tissue homogenization
The collected patient tissue samples (Healthy controls = 9; OSMF = 9, OSCC = 25) were sliced into small pieces using a sterile surgical blade. The tissue pieces were transferred into individual 2-mL centrifuge tubes each containing 300 µL of TRIzol reagent (Thermo Fischer Scientific Inc., Waltham, MA, USA). Two sterile beads were added to all the tubes and they were placed in a Tissue LyserLT (Qiagen Inc., Venlo, The Netherlands) for 15 min. After the lysis of tissues, 700 µL of TRIzol reagent was added to the tubes, resuspended, and transferred to fresh 1.5 mL centrifuge tubes [bib_ref] Optimising Bacterial DNA Extraction from Faecal Samples: Comparison of Three Methods, Smith [/bib_ref].
## Rna isolation
First, 0.2 mL of chloroform per 1 mL of TRIzol reagent was added and the tubes were vortexed for 10 to 15 s, followed by incubation at room temperature for 3 min. The samples were centrifuged at 12,000 rpm, for 15 min at 4 - C. (The mixture was separated into a lower phenol-chloroform phase, an interphase, and a colorless upper aqueous phase. RNA remained in the aqueous phase). The aqueous phase was transferred to a fresh tube. The RNA was precipitated by mixing with isopropyl alcohol (0.5 mL of isopropyl alcohol per 1 mL of TRIzol reagent). The samples were incubated at room temperature for 10 min and centrifuged at 12,000 rpm for 10 min at 4 - C. The RNA precipitate formed a white pellet on the side and bottom of the tube. The supernatant was removed and 1 mL of 75% ethanol per 1 mL of TRIzol was added. The sample was mixed by vortexing and centrifuged at 7500 rpm for 5 min at 4 - C.
The supernatant was removed and the RNA pellet was dried briefly. The RNA pellet was not allowed to dry completely as this may greatly decrease its solubility. The RNA was dissolved in RNase-free water by flipping the tube a few times, and it was incubated for 10 min at 55 to 60 - C.
## Cdna conversion
The Quantitect Reverse Transcription Kit (Qiagen Inc.) was used for the reverse transcription of the RNA samples. A total of 2 µg of RNA was required for the cDNA conversion, which consisted of 2 major steps: elimination of genomic DNA and reverse transcription. The purified RNA sample was incubated with 2 µL of gDNA wipe-out buffer and briefly incubated at 42 - C for 2 min. The reaction mix was immediately transferred to ice. After gDNA elimination, the RNA sample was ready for reverse transcription using a master mix prepared with the Quantiscript Reverse Transcriptase, Quantiscript RT Buffer, and RT Primer mix. The entire reaction took place at 42 - C and was inactivated at 95 - C. The prepared cDNA was stored at −40 - C [28].
## Real-time pcr
SYBR Green-based real-time amplification was performed using the QuantiNova SYBR Green RT-PCR Kit (Qiagen Inc., Venlo, The Netherlands). A 20-µL reaction was set up, which contained 10 µL of 2× QuantiNova SYBR Green RT-PCR Master Mix, 1 µL of each forward and reverse primer [fig_ref] Table 1: EGFR primer sequences [/fig_ref] , 6 µL of nuclease-free water, and 2 µL of cDNA. The thermal profile was 30 min at 50 - C, 15 min at 95 - C, 45 cycles of 15 s at 94 - C, 30 s at Tm, 30 s at 72 - C, ending with a melting curve from 60 - C to 90 - C. All real-time amplifications were run on a Rotor Gene Q Real-Time PCR system (Qiagen). Triplicate reactions were performed for gene expression studies using EGFR, and the mean expression value was computed for the subsequent analysis. The relative expression level of the genes was calculated using the (2-ddct) method [bib_ref] Development of a Novel, Genome Subtraction-Derived, SARS-CoV-2-Specific COVID-19-Nsp2 Real-Time RT-PCR Assay and..., Yip [/bib_ref].
# Statistical analysis
Categorical variables were presented as percentages and continuous variables as mean ± standard deviation (SD). Clinical characteristics and outcomes were compared between the groups (OSMF vs. OSCC vs. healthy controls) using Dunnett's T3 post-hoc test for categorical variables or continuous variables. Student's t-test was used to calculate statistically significant differences between the groups in EGFR expression during real-time PCR. Statistical analyses were performed using SPSS package v 23 SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).
# Results
## Measurement of egfr levels
The mean value of salivary EGFR in Group A (OSMF) was found to be 60.32 + 10.2 ng/mL, in Group B (OSCC) was found to be 71.63 + 10.09 ng/mL, and in Group C (healthy controls) was found to be 65.1 + 9.08 ng/mL [fig_ref] Figure 1: Salivary EGFR levels of healthy controls vs [/fig_ref]. The difference was statistically significant for OSCC in comparison with OSMF and healthy controls, with an F value of 4.760 and with p value 0.01 (<0.05), using Dunnett's T3 post-hoc test. were performed for gene expression studies using EGFR, and the mean expression value was computed for the subsequent analysis. The relative expression level of the genes was calculated using the (2-ddct) method [bib_ref] Development of a Novel, Genome Subtraction-Derived, SARS-CoV-2-Specific COVID-19-Nsp2 Real-Time RT-PCR Assay and..., Yip [/bib_ref].
# Statistical analysis
Categorical variables were presented as percentages and continuous variables as mean ± standard deviation (SD). Clinical characteristics and outcomes were compared between the groups (OSMF vs. OSCC vs. healthy controls) using Dunnett's T3 post-hoc test for categorical variables or continuous variables. Student's t-test was used to calculate statistically significant differences between the groups in EGFR expression during realtime PCR. Statistical analyses were performed using SPSS package v 23 SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA).
# Results
## Measurement of egfr levels
The mean value of salivary EGFR in Group A (OSMF) was found to be 60.32 + 10.2 ng/mL, in Group B (OSCC) was found to be 71.63 + 10.09 ng/mL, and in Group C (healthy controls) was found to be 65.1 + 9.08 ng/mL [fig_ref] Figure 1: Salivary EGFR levels of healthy controls vs [/fig_ref]. The difference was statistically significant for OSCC in comparison with OSMF and healthy controls, with an F value of 4.760 and with p value 0.01 (<0.05), using Dunnett's T3 post-hoc test.. The mean value of exfoliated buccal cells EGFR concentration in Group A (OSMF) was found to be 454.54 + 89.50 ng/mL, in Group B (OSCC) was found to be 528.79 + 62.25 ng/mL, and in Group C (healthy controls) was found to be 545.47 + 109.72 ng/mL [fig_ref] Figure 2: EGFR expression in exfoliated buccal cells in healthy controls vs [/fig_ref]. The difference was found to be statistically significant for OSCC in comparison with OSMF and healthy controls, with an F value of 5.257 and p value 0.009 (<0.01), using Dunnett's T3 post-hoc test. The mean value of exfoliated buccal cells EGFR concentration in Group A (OSMF) was found to be 454.54 + 89.50 ng/mL, in Group B (OSCC) was found to be 528.79 + 62.25 ng/mL, and in Group C (healthy controls) was found to be 545.47 + 109.72 ng/mL [fig_ref] Figure 2: EGFR expression in exfoliated buccal cells in healthy controls vs [/fig_ref]. The difference was found to be statistically significant for OSCC in comparison with OSMF and healthy controls, with an F value of 5.257 and p value 0.009 (<0.01), using Dunnett's T3 post-hoc test.
## Immunohistochemistry
Expression of EGFR was studied in 173 patients in different age groups, in which mild positive immunoexpression was found in 23 patients (13.3%), intermediate positive immunoexpression was found in 51 patients (29.5%), strong positive immunoexpression was found in 87 patients (50.3%), and 12 patients (6.9%) tested negative [fig_ref] Table 2: Clinical histopathological features in EGFR-positive groups in healthy controls and OSMF and... [/fig_ref]. EGFR overexpression was increased in oral cancer and OSMF compared to healthy controls (p = 0.000; χ 2 = 68.620). EGFR overexpression was significantly correlated (p = 0.000; χ 2 = 85.409) with various grades of dysplasia, and well-differentiated (WDSCC) and moderately differentiated squamous cell carcinoma (MDSCC) [fig_ref] Table 3: Clinical histopathological features in EGFR-positive groups in dysplasia and cancer patients [/fig_ref]. [fig_ref] Figure 3: IHC for EGFR in healthy controls [/fig_ref] shows the expression of EGFR in healthy controls showing negative expression. [fig_ref] Figure 4: IHC of EGFR in OSMF samples [/fig_ref] shows the expression of EGFR in OSMF, showing intense cytoplasmic and nuclear positivity. [fig_ref] Figure 5: IHC for EGFR in dysplasia [/fig_ref] show the expression of EGFR in dysplasia and OSCC samples, respectively, showing intense cytoplasmic and nuclear positivity.
## Immunohistochemistry
Expression of EGFR was studied in 173 patients in different age groups, in which mild positive immunoexpression was found in 23 patients (13.3%), intermediate positive immunoexpression was found in 51 patients (29.5%), strong positive immunoexpression was found in 87 patients (50.3%), and 12 patients (6.9%) tested negative [fig_ref] Table 2: Clinical histopathological features in EGFR-positive groups in healthy controls and OSMF and... [/fig_ref]. EGFR overexpression was increased in oral cancer and OSMF compared to healthy controls (p = 0.000; χ 2 = 68.620). EGFR overexpression was significantly correlated (p = 0.000; χ 2 = 85.409) with various grades of dysplasia, and well-differentiated (WDSCC) and moderately differentiated squamous cell carcinoma (MDSCC) [fig_ref] Table 3: Clinical histopathological features in EGFR-positive groups in dysplasia and cancer patients [/fig_ref]. [fig_ref] Figure 3: IHC for EGFR in healthy controls [/fig_ref] shows the expression of EGFR in healthy controls showing negative expression. [fig_ref] Figure 4: IHC of EGFR in OSMF samples [/fig_ref] shows the expression of EGFR in OSMF, showing intense cytoplasmic and nuclear positivity. [fig_ref] Figure 5: IHC for EGFR in dysplasia [/fig_ref] show the expression of EGFR in dysplasia and OSCC samples, respectively, showing intense cytoplasmic and nuclear positivity.
## Real-time pcr
The expression of EGFR was studied in tissue samples using real-time PCR for different patient categories [fig_ref] Table 1: EGFR primer sequences [/fig_ref]. Based on the obtained results, EGFR exhibited maximum expression in OSCC tissue samples in comparison to OSMF and healthy controls. We found a significant 18-fold upregulation of the EGFR gene in OSCC samples and three-fold upregulation in OSMF samples compared to healthy controls [fig_ref] Figure 7: Differential expression of EGFR in healthy controls vs [/fig_ref]. The EGFR expression was found to be statistically significant using Student's t-test for the studied samples (OSCC vs. OSMF vs. healthy controls), with p value 0.037 (<0.05).
## Real-time pcr
The expression of EGFR was studied in tissue samples using real-time PCR for different patient categories [fig_ref] Table 1: EGFR primer sequences [/fig_ref]. Based on the obtained results, EGFR exhibited maximum expression in OSCC tissue samples in comparison to OSMF and healthy controls. We found a significant 18-fold upregulation of the EGFR gene in OSCC samples and threefold upregulation in OSMF samples compared to healthy controls [fig_ref] Figure 7: Differential expression of EGFR in healthy controls vs [/fig_ref]. The EGFR expression was found to be statistically significant using Student's t-test for the studied samples (OSCC vs. OSMF vs. healthy controls), with p value 0.037 (<0.05).
The expression of EGFR was studied in tissue samples using real-time PCR for different patient categories (Tables S1 and S2). Based on the obtained results, EGFR exhibited maximum expression in OSCC tissue samples in comparison to OSMF and healthy controls. We found a significant 18-fold upregulation of the EGFR gene in OSCC samples and three-fold upregulation in OSMF samples compared to healthy controls [fig_ref] Figure 7: Differential expression of EGFR in healthy controls vs [/fig_ref]. The EGFR expression was found to be statistically significant using Student's t-test for the studied samples (OSCC vs. OSMF vs. healthy controls), with p value 0.037 (<0.05).
# Discussion
The primary function of epidermal growth factor receptor (EGFR) is to maintain homeostasis and epithelial tissue development. Epithelial cancers, including OSCC, have been found to overexpress EGFR and hence it is worth exploring the expression in OPMD [bib_ref] Epidermal Growth Factor and Epidermal Growth Factor Receptor: The Yin and Yang..., Bodnar [/bib_ref]. Therefore, early detection of the malignant potential of high-risk OPMD could be
# Discussion
The primary function of epidermal growth factor receptor (EGFR) is to maintain homeostasis and epithelial tissue development. Epithelial cancers, including OSCC, have been found to overexpress EGFR and hence it is worth exploring the expression in OPMD [bib_ref] Epidermal Growth Factor and Epidermal Growth Factor Receptor: The Yin and Yang..., Bodnar [/bib_ref]. Therefore, early detection of the malignant potential of high-risk OPMD could be evaluated using EGFR, especially using saliva and exfoliated buccal cells as a diagnostic sample, which can prove to be a viable screening tool.
Over the years, numerous techniques, such as IHC, fluorescence in situ hybridization (FISH), and PCR, have been used to identify the overexpression of EGFR, EGFR copy number gains (CNG), and EGFR mutations. However, numerous studies have employed IHC because of its ease and simplicity [bib_ref] Dysregulation and Detection Methods of EGFR in Oral Cancer, Somarriva [/bib_ref]. Studies evaluating the expression of EGFR in OSCC and leukoplakia have shown a significant correlation in OSCC, when compared to leukoplakia [bib_ref] Asynchronous Modulation of Transforming Growth Factor Alpha and Epidermal Growth Factor Receptor..., Grandis [/bib_ref] [bib_ref] Dysregulation of Epidermal Growth Factor Receptor Expression in Premalignant Lesions during Head..., Shin [/bib_ref] [bib_ref] Expression of P53, Epidermal Growth Factor Receptor, c-ErbB2 in Oral Leukoplakias and..., Singla [/bib_ref]. The increased expression of EGFR in OPMD such as leukoplakia and OSMF favors EGFR as a valuable diagnostic marker [bib_ref] Quantitative Immunoexpression of EGFR in Oral Potentially Malignant Disorders: Oral Leukoplakia and..., Meka [/bib_ref]. However, the study mentioned above evaluated the EGFR expression in dysplastic cases of oral leukoplakia and OSMF, with no attempt to compare the grades of dysplasia. In the present study, we have shown significant differences between different grades of dysplasia in OPMD and different histological grades of OSCC. The results of the immunoexpression of EGFR in our study are in accordance with a study performed by Mahendra et al., comparing leukoplakia and OSCC, which showed increased EGFR expression with increasing grades of dysplasia and the highest expression in OSCC [bib_ref] Epidermal Growth Factor Receptor Protein: A Biological Marker for Oral Precancer and..., Mahendra [/bib_ref]. Immunoexpression of EGFR was correlated with the survival rate of oral cancer patients, and the authors have concluded EGFR to be an independent prognostic factor for the assessment of survival rates [bib_ref] Correlation of EGFR Expression with Survival Rate in Patients with Oral Squamous..., Baghai Naini [/bib_ref]. With extensive literature support for EGFR using IHC, limited studies are available on saliva and exfoliated buccal cells as a diagnostic sample. Hence, the current study evaluated the EGFR expression in saliva and buccal cells, considering the advantage of patient compliance during screening procedures. The expression of EGFR in saliva was supported valuably by evaluating the immunoexpression in OSCC using IHC and quantification of gene expression using qPCR. In a previous study performed by , evaluating the EGFR copy number using RT-PCR showed significant expression in the advanced stages of oral malignancy compared to OPMD, with higher expression in non-homogenous leukoplakia compared to homogenous leukoplakia [bib_ref] Epidermal Growth Factor Receptor Copy Number in Potentially Malignant Oral Disorders and..., Bagan [/bib_ref]. Similar observations were obtained in our study, where EGFR expression was highest in the OSCC group, followed by the OSMF group, compared to healthy controls. The results further validate the role of EGFR in progression to malignancy.
EGFR is found to be a transmembrane receptor. However, the extracellular domain of EGFR could be released by proteolytic cleavage and exfoliated from the surfaces of cells; hence, it is found to be detected in other body fluids, mainly saliva [bib_ref] EGF-Receptor Regulation of Matrix Metalloproteinases in Epithelial Ovarian Carcinoma, Hudson [/bib_ref]. In our study, the salivary EGFR levels in OSCC were high, attributed to high EGFR levels in actively dividing tumor cells. This is in accordance with the study performed by where the salivary EGFR levels were higher than levels in serum; there was a significant correlation between tumor stage and survival [bib_ref] Epidermal Growth Factor Receptor Detection in Serum and Saliva as a Diagnostic..., Zanotti [/bib_ref]. However, in the study performed by Bernandes et al., salivary EGFR levels were not elevated in OSCC and did not correlate significantly with the clinical pathological parameters [bib_ref] Clinical Significance of EGFR, Her-2 and EGF in Oral Squamous Cell Carcinoma:..., Bernardes [/bib_ref]. Similarly, in a study comparing salivary EGFR levels in premalignant conditions, OSCC and normal patients did not show a significant association, despite OSCC showing higher salivary EGFR [bib_ref] Salivary Egfr as Biomarker for Oral Cancer & Pre-Cancer: A Case Control..., Irfan [/bib_ref].
The present study showed a decrease in salivary EGFR in OSMF compared to OSCC, in accordance with previous studies with OPMD compared to OSCC [bib_ref] Expression of P53, Epidermal Growth Factor Receptor, c-ErbB2 in Oral Leukoplakias and..., Singla [/bib_ref] [bib_ref] Salivary Egfr as Biomarker for Oral Cancer & Pre-Cancer: A Case Control..., Irfan [/bib_ref]. The reduction in EGFR levels in the saliva of OSMF patients could be attributed to the epithelial atrophy observed in OSMF and the possible pathway of TGFβ-mediated Epithelial to Mesenchymal Transition (EMT) [bib_ref] Loss of Epidermal Growth Factor Receptor Expression in Oral Squamous Cell Carcinoma..., Kimura [/bib_ref].
A potential limitation of our study was the smaller sample size and unequal number of samples in different groups for ELISA. However, the attempt to validate the results of saliva using a substantial number of samples in IHC and qPCR sought to overcome this limitation. Thus, the current research could aid in conducting studies with larger samples, with a standardized technique for saliva collection and estimation to obtain reproducible results in confirming the diagnostic utility of salivary EGFR as a biomarker.
# Conclusions
EGFR expression could be considered a promising diagnostic marker and its evaluation in saliva and exfoliated buccal cells using ELISA could serve as a valuable screening diagnostic tool for OPMD. The validation studies using IHC have shown increased immunoexpression of EGFR in OSCC compared to OPMD. Furthermore, gene expression studies performed using RT-PCR validate the upregulation of EGFR in OSCC in comparison to OSMF and healthy controls. Further studies with larger samples of OSMF and OSCC patients will shed light on the reliability of salivary EGFR as a diagnostic marker in identifying the possible malignant risk.
Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/diagnostics12081935/s1. [fig_ref] Table 1: EGFR primer sequences [/fig_ref] : Expression of EGFR in OSCC samples; [fig_ref] Table 2: Clinical histopathological features in EGFR-positive groups in healthy controls and OSMF and... [/fig_ref] : Expression of EGFR in OSMF samples.
[fig] Figure 1: Salivary EGFR levels of healthy controls vs. OSMF vs. OSCC. * Values are expressed as mean ± SD. Statistical significance analyzed by Dunnett's T3 post-hoc test at p < 0.05. [/fig]
[fig] Figure 2: EGFR expression in exfoliated buccal cells in healthy controls vs. OSMF vs. OSCC. * Values are expressed as mean ± SD. Statistical significance analyzed by Dunnett's T3 post-hoc test at p < 0.01. [/fig]
[fig] Figure 3: IHC for EGFR in healthy controls (under 20× magnification). [/fig]
[fig] Figure 4, Figure 4: IHC of EGFR in OSMF samples (under 20× magnification). IHC of EGFR in OSMF samples (under 20× magnification). [/fig]
[fig] Figure 4: IHC of EGFR in OSMF samples (under 20× magnification). [/fig]
[fig] Figure 5: IHC for EGFR in dysplasia (under 20× magnification). [/fig]
[fig] Figure 6: IHC for EGFR in OSCC (under 20× magnification). [/fig]
[fig] Figure 7: Differential expression of EGFR in healthy controls vs. OSMF vs. OSCC. * Statistical analysis performed using Student's t-test. [/fig]
[table] Table 1: EGFR primer sequences. [/table]
[table] Table 2: Clinical histopathological features in EGFR-positive groups in healthy controls and OSMF and OSCC patients. [/table]
[table] Table 3: Clinical histopathological features in EGFR-positive groups in dysplasia and cancer patients. [/table]
|