Altered interaction between enteric glial cells and mast cells in the colon of women with irritable bowel syndrome

Enteric glial cells (EGC) and mast cells (MC) are intimately associated with gastrointestinal physiological functions. We aimed to investigate EGC‐MC interaction in irritable bowel syndrome (IBS), a gut‐brain disorder linked to increased intestinal permeability, and MC.


| INTRODUC TI ON
Irritable bowel syndrome (IBS) is a gut-brain disorder characterized by recurrent abdominal pain linked to disturbed bowel habits with a global prevalence of about 11% of the population. 1 The pathogenesis of IBS is driven by a tangle of factors including abnormalities in motility, visceral sensation, brain-gut interaction, psychosocial distress, changes in immune responses, imbalance in gut microbiome, and abnormal intestinal permeability. [2][3][4] Gastrointestinal (GI) functions are predominantly controlled by the enteric nervous system (ENS), an intrinsic network of enteric neurons and enteric glial cells (EGC) 5 that communicates bidirectionally with the central nervous system (CNS), mainly through the vagus nerve, to maintain homeostasis. 6 EGC, distributed throughout all colonic layers, are a diverse population of cells distinguished by their morphology and location within the gut wall. 7 EGC variably express glial cell markers, such as glial fibrillary intermediate filament (GFAP), S100 calciumbinding protein β (S100β) and SRY-box transcriptional factor 10 (Sox-10). This fact suggests functional diversity and plasticity among EGC types 8 that is observed, for example, in response to environmental triggers, such as lipopolysaccharide. 9 EGC are strategically positioned close to enteric neurons, smooth muscle, and intestinal epithelium that together influence neural circuits, taking part in the regulation of GI functions. 5 In the GI tract, EGC are known to participate in epithelial barrier function, 10 fluid secretion, 11 and motility. 12,13 Therefore, EGC may play an important role in the pathophysiology of IBS, but there are still few studies present. Recent experimental and clinical findings have shown changes in EGC in IBS animal models, presenting hyperplasia in myenteric plexus 14 and colonic-mucosal alterations in GFAP expression 15 and S100β-positive area. 16 In IBS patients, these correlated with abdominal pain and bloating.
Mast cells (MC) are active players in the gut and are well known for communicating with neurons within the ENS. 17 MC contribute to low-grade inflammation in the intestinal mucosa 18 and to visceral hypersensitivity in IBS patients. 19 Our group previously showed that MC and vasoactive intestinal polypeptide (VIP), a neuroimmune-endocrine mediator majorly produced by enteric neurons and MC in the gut, contribute to control barrier function in the ileum of healthy humans and in rats submitted to chronic stress. 20 More recently, 21 we reported higher levels of tryptase, an increased density of MC, and percentage of VIP + and VIP receptor 1 (VPAC1) + MC, but not VIP receptor 2 (VPAC2) + MC in IBS compared to healthy controls (HC). Further, we found that translocation of live bacteria through colonic mucosa is partially controlled by MC and VIP in both IBS and healthy subjects. 21 Existing evidence supporting a direct glial-MC communication is limited to the CNS, 22 where, when perturbed, it appears to contribute to the etiology of neurodegenerative diseases such as Parkinson's disease, multiple sclerosis, traumatic brain injury depression, and Autism spectrum disorder. 23 Neuron-MC communication has been studied in human submucosal plexus preparations, showing clear bidirectional signaling. 24 Another study showed that functional dyspepsia patients presenting gliosis displayed impaired neuronal function, which was correlated with MC infiltration in duodenal submucosa. 25 Despite all scientific effort, mechanisms by which the MC-EGC communication influences the IBS pathophysiology remain unclear.
Based on previous findings, we hypothesized that EGC are a functional part of the neuroimmune hub formed by enteric neurons and MC. Since human EGC express toll-like receptors and expression levels are modulated by both pathogenic and probiotic bacteria, 26  IBS subjects were classified based on predominant bowel habit into IBS-diarrhea (IBS-D, n = 7), IBS-constipation (IBS-C, n = 7) and IBS-mixed (IBS-M, n = 16). All patients were experiencing symptoms during the study. Twenty-two age-matched females (mean 31.6 years, range 21-56) without a medical history of GI symptoms or complaints were recruited as HC. Exclusion criteria were organic GI disease, metabolic, neurological, or severe psychiatric disorders, self-reported nicotine intake within 2 months before the study, allergy and use of non-steroidal anti-inflammatory drugs or central acting pain medications. The regional ethical review board approved the study, and all participants gave their written informed consent in accordance with the Declaration of Helsinki.

| Questionnaires and GI symptom diaries
Questionnaire data from all participants were collected at inclusion for sample characterization and for the assessment of IBS-related variables. Symptom diaries were obtained from IBS patients, but not HC. Patients recorded their GI symptoms for 14 consecutive days using validated diary cards. 27 The symptoms (abdominal pain, nausea, bloating) and every single bowel movement and stool consistency (defined by the Bristol Chart, BSC) were reported along a 24-h time axis. The values were manually scored, and the mean frequency of symptom episodes/week and symptom duration/day was extracted from the diary data.

| IBS severity scoring system (IBS-SSS)
IBS-SSS is a five-item questionnaire evaluating overall IBS symptom severity by assessing the frequency and the intensity of abdominal pain and distension, the satisfaction with bowel habits, and interference with daily life. 28 Each item generates a score between 0 and 100 with a maximal sum score of 500. Sum scores indicate mild (75-175), moderate (175-300), or severe (>300) disease.

| Hospital Anxiety and Depression Score (HADS)
The HADS was used to measure symptoms of anxiety and depression. 29 The scale consists of seven items of anxiety (HADS-A) and depression subscales (HADS-D), with scores on each subscale ranging from 0 to 21.

| Sigmoidoscopy
Flexible sigmoidoscopies were performed according to a standardized protocol regarding stretching and thickness of the biopsies.
Participants were fasting for 8 h and completed a colon preparation with an enema (klyx®) early the same morning. The procedure was performed without sedation with scope insertion approximately 30-40 cm ascending from linea dentata. Colonic biopsies were taken with biopsy forceps without a central lance and placed directly in ice-cold-oxygenated Krebs buffer. 30

| Fecal samples
Fecal samples from 33 patients with IBS and 20 HC were collected in feces containers (Sarstedt, Helsingborg, Sweden) within 2 weeks prior to the sigmoidoscopy and within 24 h sent to the laboratory in room temperature (RT). Immediately after arrival, samples were frozen at −80°C.

| Microbiota analysis
Fecal microbiota was analyzed by the Genetic Analysis GA-map® Dysbiosis Test (Genetic Analysis AS, Oslo, Norway), which is a gut microbiota DNA analysis tool to identify dysbiosis. Briefly, the test involves sample homogenization, mechanical bacterial cell disruption, and total bacterial gDNA extraction as previously described. 31 The test utilizes 54 predetermined bacterial DNA markers targeting the 16S rRNA sequence in seven variable regions (V3-V9) that measure the abundance of bacteria according to the intensity of the fluorescent signal detection. The 54 DNA probes on the GA-map targeted ≥300 bacteria on different taxonomic levels and were selected based on the ability to distinguish between HC, IBS, and IBD patients. 31 Analyses of bacterial abundances were performed using the fluorescent signal.

| Ussing chambers experiments
Colonic biopsies from IBS patients and HC were mounted in Ussing chambers as previously described. 30 After 40 min of equilibration, To investigate the influence of glial cell mediators on permeability, biopsies from 3 HC were mounted in Ussing chambers and added 100 µM S-nitrosoglutathione (GSNO) (Sigma-Aldrich), 3 nM GDNF (Thermo Fisher), or 100 nM S100β, (Sigma-Aldrich), or Krebs buffer as control, on the mucosal side. Concentrations were based on previous publications. 10,32,33 The paracellular marker fluorescein isothiocyanate (FITC)-dextran 4000 (Sigma-Aldrich) was added on the mucosal sides at 2.5 nM, and serosal samples were collected over time and analyzed for FITC-dextran passage at 488 nm in a VICTOR™ X3 multileader plate reader.

| Immunofluorescence
Biopsies from IBS and HC were fixed, embedded in paraffin, sectioned at 5 µm, and individually stained for 1:500 rabbit-anti-GFAP (DakoCytomation) and 1:250 mouse-anti-S100β (Invitrogen) overnight at 4°C. Antibody specificity was verified by Western blotting on human intestinal tissues and cell lines. Slides were rinsed and in- Staining for MC-tryptase, VIP, VPAC1, and VPAC2 is described elsewhere. 21

| Determination of GFAP, GDNF and S100β levels in biopsy lysates by ELISA
To measure GFAP, GDNF, and S100β levels, we employed sandwich ELISA kits. Biopsy lysates diluted 1:20, cell culture medium, and standard samples were added to plates pre-coated with primary antibody. Plates were incubated at 37°C, and after 90 min, a secondarybiotinylated antibody was added. After 60 min at 37°C, plates were further treated following manufacturer's instructions (Nordic Biosite).
Absorbance was measured at 450 nm in VersaMax Tunable Microplate Reader (Molecular Devices) using SoftMax pro 5 (Molecular Devices).
The software generated a standard curve from which the concentrations of the samples were calculated. Biopsy lysates values were corrected to dilution and normalized to protein concentration.

| In vitro intracellular calcium [Ca 2+ ] i responses live cell imaging
To study the ability of MC in evoking responses in EGC and vice-versa, we employed live cell recordings using [Ca 2+ ] i mobilization assay with Data were presented as ΔF/F0 (difference between rise in signal corrected to background and expressed as % relative to baseline).

| GSNO, GDNF, and S100β effects on VIPinduced degranulation in HMC-1.1
To investigate the effect of the EGC products GDNF and S100β on MC, a MC-degranulation assay was set up 39   were based on previous publications. 10,32,33 Cells were collected and lysed to quantify the degranulation rate of HMC-1.1 with a βhexosaminidase assay, as previously described. 39 Results were normalized to maximum degranulation and expressed as the percentage of β-hexosaminidase release relative to control samples.

| EGC cell line activation by tryptase, serotonin (5-HT), histamine, VIP, and live bacteria
To evaluate EGC activation in vitro, EGC cell line CRL-2690 were grown until confluence (~10 × 10 6 cells) in DMEM supplemented with 10% fetal bovine serum +1% penicillin/streptomycin. Cells were washed in DPBS and kept in non-supplemented DMEM without phenol red. Cells were then treated with 1 nM tryptase, 100 µM 5-HT, 10 µM histamine, or 3 μM VIP. In addition, EGC were exposed to S. typhimurium, E. coli HM427, or Yersinia enterocolitica (cell:bacteria, 1:100) After 24 h, medium was collected to determine the levels of S100β and GDNF by ELISA, and cell protein lysates were subjected to Western blotting, as described above, to measure the increase in GFAP expression as an glial activation signature (gliosis).

| Statistics
Correlational data were calculated in Mplus, formal statistical inference employed a nonparametric bootstrap. Data were processed using GraphPad Prism 8 (GraphPad Software Inc); all correlational data were treated as nonparametric. For correlations, Spearman's correlation settings were employed; for comparison between group's correlations, linear regression was performed to test differences between slopes (intercorrelation). Comparison between 2 groups was done with Mann-Whitney test, while comparison between >2 groups was done by one-way analysis of variance (ANOVA), followed by Tuckey's multiple comparison test. Statistical significance was set as *p < 0.05, **p < 0.01, ***p < 0.001 and ****p < 0.0001. For microbiota, a principal component analysis of similarities and correlation analysis were calculated using R version 4.0.2 and packages vegan and labdsv and visualized with ggplot2.  (Table 1). Prospectively recorded GI symptom diary confirmed that IBS patients had a moderate-to-severe symptom burden ( Table 2).

| Reactive EGC in the colon of IBS patients
Immunofluorescence showed increased levels of mucosal GFAP in biopsies of IBS compared to HC, p < 0.001 ( Figure 1A) while there was no difference in S100β expression ( Figure 1B). Immunofluorescence finding of GFAP was confirmed by Western blotting ( Figure 1C Figure 1E). Measurements of GDNF, measured by ELISA in biopsy lysates, revealed decreased levels in IBS compared to HC, p < 0.05 ( Figure 1F). Spearman analysis revealed no significant correlations between mucosal parameters and symptoms (data not shown).

| Decreased levels of Actinomycetales in IBS but no relation between microbiota and EGC
Out of the 54 probes investigated, all identifying a phylogenetic group of bacteria, only the group Actinomycetales showed out to differ between HC and IBS. There were significantly lower levels in IBS patients, p < 0.01 (Figure 2A), with a majority of the lower data points belonging to the IBS-D subgroup (red dots in Figure 2A Figure 3B), GFAP versus VPAC1 + MC ( Figure 3C), but not between GFAP and VPAC2 + MC ( Figure 3D).

| EGC phenotype correlates differently to live bacterial passage through colonic biopsies of patients with IBS and HC
To evaluate the involvement of EGC in barrier function, we first analyzed intercorrelations from IBS and HC on the passage of live commensal or pathogenic bacteria (an event partly controlled by MC and VIP signaling) and GFAP expression. In HC, higher expression of GFAP correlated with less E. coli and Salmonella passing through colonic mucosa. In IBS, however, there was a slightly positive correlation between GFAP and the number of live bacteria translocating through the mucosa. Intercorrelation analysis between IBS and HC showed that GFAP levels had a significant interaction with the passage of Salmonella ( Figure 3E), but not to E. coli, p = 0.12 Figure S2).

| EGC mediators control paracellular permeability
To test whether GSNO, GDNF, and S100β control barrier function, colonic biopsies were mounted in Ussing chambers. All treatments significantly decreased the passage of FITC-Dextran 4000 ( Figure 3F).

| In vitro [Ca 2+ ] i responses in EGC and MC suggest their ability to communicate
The

GDNF was significantly increased by tryptase and Y. enterocolitica
( Figure 5E-F).

| DISCUSS ION
In this study, we explored EGC phenotypical changes in IBS and how the activation of EGC is associated with MC-VIP-increased translocation of live bacteria through the colonic mucosa. 21 We found an increased expression of GFAP in the colonic mucosa of IBS, in which EGC acquire a reactive phenotype. 40 We did not count the number of GFAP + EGC nor S100β + EGC, instead, we measured the intensity of GFAP expression, which in the literature is considered as a sign of EGC activation (gliosis). 40  decreased S100β-stained area in the colon of IBS patients that was negatively correlated with both frequency and intensity of pain and bloating. However, no differences in S100β and GFAP protein expression were found. S100β is constitutively expressed by a subset of EGC, 8 presenting neuroprotective properties at low concentrations and pro-inflammatory properties when upregulated. In the present study, no changes in S100β could be identified, and the EGC phenotype in IBS did not correlate with abdominal pain nor bloating.
GFAP is expressed in CNS astrocytes, non-myelinating Schwann cells in the peripheral nervous systems, and in EGC. 41 GFAP expression is regulated by several compounds such as lipopolysaccharide. 42 Increased S100β and GFAP expression were observed in human small bowel-derived EGC exposed to lipopolysaccharide and IFNγ. 43   and S100 calcium-binding protein β (S100β) on permeability to fluorescein isothiocyanate (FITC) dextran 4000 across colonic biopsies mounted on Ussing chambers for 120 min. Data are expressed as median (IQR), two-way ANOVA, followed by Bonferroni's test, *p < 0.05, **p < 0.01 and ***p < 0.001; each symbol represents the pool of three independent experiments. Data in A-E were processed with Spearman correlation followed by linear regression to test differences between slopes cells (oligodendrocytes, non-myelinating Schwann cells and astrocytes) and EGC, the latter represents a distinct glial lineage. 48 Understanding the role EGC in the gut is imperative to comprehend the impact that EGC-MC interaction may have in health and during intestinal disease such as IBS.
MC are often found in close proximity to nerves. 49 Barbara et al.

demonstrated that MC and afferent nerves proximity may contribute
to the abdominal pain in IBS. 50 Nevertheless, interaction between MC and enteric nerves has also been described. 51  Medium was collected and used for determination with ELISA of (C) S100 calcium-binding protein β (S100β) release induced by neuronal and MC mediators, (D) S100β release induced by live bacteria, (E) glial-derived neurotrophic factor (GDNF) release induced by neuronal and MC mediators, (F) GDNF release induced by live bacteria. Data are expressed as median (IQR), one-way ANOVA, followed by Dunns' test or Kruskal-Wallis nonparametric t test, *p < 0.05, **p < 0.01 and ****p < 0.0001; each symbol represents independent experiments signaling in the gut implies the activation of adrenergic and serotonergic receptors whose expression and function on VIP-positive neurons has been demonstrated. 9 The VIP-dependent translocation of bacteria 21 correlated differently to EGC phenotype in IBS and HC. GFAP expression in the colonic mucosa of HC negatively correlated with the bacterial passage, suggesting protective involvement of EGC. Whereas in IBS, higher GFAP levels were linked to the loss of beneficial EGC activity.
GFAP physiological expression may reflect protective mechanisms promoted by EGC, whilst IBS-reactive EGC might influence its microenvironment negatively, due to altered release of glial mediators.
In fact, we showed that EGC activation induced by tryptase, 5-HT, histamine and VIP, displayed a variable pattern of S100β and GDNF release compared with that induced by live bacteria, even though all the experimental conditions had led to increased GFAP expression, suggesting that EGC-reactive phenotype can result in a much more complex response than simply GFAP upregulation. Interestingly, IBS biopsies presented decreased GDNF levels in relation to HC.
In contrast to our finding, Lin and colleagues 58 showed increased GDNF protein levels in colonic biopsies from patients with IBS-D.
Another recent study from Lee et al. 59  As referred above, tryptase induced gliosis followed by increased release of GDNF by EGC in our in vitro assay. On the other hand, IBS biopsies presented lower levels of GDNF and more activated MC (thus releasing more tryptase) and EGC, which would point to a higher release of GDNF by EGC. However, other mediators, such as histamine and VIP, also induced gliosis, yet kept GDNF levels to that of untreated EGC, suggesting that other mediators can block GDNF release by activated EGC. In addition to that, GDNF is highly expressed by epithelial cells, 33 hence biopsy GDNF levels more likely reflect overall mucosal levels rather than EGC-GDNF only.
We used GSNO and GDNF to confirm their ability to tighten the colonic epithelia of HC biopsies in Ussing chambers. In addition, we tested whether S100β could also influence passage through the biopsies. All treatments resulted in diminished passage of FITC-dextran, indicating that all these EGC mediators (at physiological concentrations) influence the paracellular route through the colonic mucosa.
Both MC and EGC respond to environmental triggers, that is, bacteria and/or their products in the gut. 62 Physiological loads of pathogens and/or antigens breaking into the epithelial layer are consequentially sampled by resident immune cells in the intestines, which could explain the positive association between GFAP and MC numbers in HC. This load of pathogens and antigens within the mucosa may be slightly higher, or altered in its composition, in IBS patients, 21 which could potentially be a disturbing factor to EGC-MC interaction.
Fettucciari and colleagues recently suggested a role of EGC in the pathogenesis of Clostridium (C.) difficile infection. 63 Further, authors hypothesized that EGC surviving C. difficile toxins acquire a senescence state which could be associated with an increased risk of IBS. 64 In the present study, we could not find any correlation between microbiota composition and GFAP or S100β expressions, although we did find decreased levels of Actinomycetales. Actinomycetales is an order of bacteria where one member is the Actinomycetes, known to have antibacterial activity in the GI tract 65  to EGC mediators S100β and GDNF, indicating their ability in communicating with each other. Furthermore, in vitro MC-degranulation assay showed that HMC-1.1 exposed to either S100β or GDNF kept normal response. Both GDNF and S100β prevented VIP-induced degranulation of HMC-1.1. Even though, in vitro models are limited by the nature of the cells, our findings, provide substantial evidence of regulatory roles between EGC and MC.
Our data showed that VIP + MC correlated positively with GFAP expression in IBS patients. Even though MC are not the main source of VIP, 67 MC contain 68 and release 69 VIP and may thereby contribute to EGC activation. VIPergic enteric neurons most likely influence both MC and EGC activities in the GI. Altogether this indicates that this neuroimmune hub in the gut appears to ground significant events that ultimately contribute to developing IBS.
In summary, the study suggests that EGC-MC interaction in the human colon presents distinct profiles in health and in IBS, as overviewed in Figure 6. Despite limitations, we provided compelling evidence suggesting that, under homeostasis, EGC-increasing GFAP level is followed by increased MC numbers. The MC are VIP-tuned to regulate bacterial translocation in the colon, contributing to the barrier function. In IBS, however, dysregulated VIP circuitry together with reactive EGC destabilize MC. This might contribute to weakening the barrier and, in turn, may result in increased passage of antigens, creating a feedback loop. Altogether, these findings point to a VIP-tuned EGC-MC activity in the human colon with a potential relevance to IBS.

ACK N OWLED G M ENTS
The authors thank all healthy volunteers and patients for their par-

D I SCLOS U R E S
The authors have no conflicting interests to declare.

F I G U R E 6 Schematic overview of enteric glial cells (EGC) and mast cells (MC) interactions in the colon as suggested by our findings.
In healthy controls (HC), EGC and MC likely contribute to homeostatic conditions by inhibiting the passage of bacteria. On the contrary, in irritable bowel syndrome (IBS), reactive EGC expressing increased glial fibrillary acidic protein (GFAP) are related to more vasoactive intestinal polypeptide (VIP) + MC less VIP receptor 1 (VPAC1) + MC, and increased translocation of pathogenic bacteria. Furthermore, EGC mediators such as S100 calcium-binding protein β (S100β) and glial-derived neurotrophic factor (GDNF) are known to be released in higher amounts under glial activation (gliosis). By in vitro experiments, we showed that GDNF induces degranulation of MC in the presence of bacterial products, and that MC mediators such as VIP (also and majorly produced by enteric neurons), histamine, and tryptase, known to regulate intestinal permeability, elicit EGC responses

AUTH O R CO NTR I B UTI O N S
FMF designed the study, performed experiments, analyzed data, and wrote the manuscript; MCB performed experiments, analyzed data, co-wrote the manuscript, and made the graphical artwork; CML contributed to microbiota analysis, visualization, and interpretations; MPJ was involved in statistical analysis and interpretation; SAW designed the study, provided financial support, interpreted data and co-wrote the manuscript; ÅVK designed the study, provided financial support, interpreted data, co-wrote the manuscript, and was responsible for the final version. All authors read and approved the final manuscript.