Disruption of the pro-inflammatory, anti-inflammatory cytokines and tight junction proteins expression, associated with changes of the composition of the gut microbiota in patients with irritable bowel syndrome

Background Irritable bowel syndrome (IBS) is a pathologic condition characterized by changes in gut microbiome composition, low-grade inflammation, and disruption of intestinal wall permeability. The interaction between the gut microbiome and the disease manifestation remains unclear. The changing of tight junction proteins and cytokines expression throughout the gastrointestinal tract in IBS patients has not been studied yet. Aim of the study To assess the changes of gut microbiome composition, tight junction proteins, and cytokines expression of intestinal mucosa from the duodenum to the distal part of the colon in IBS patients and healthy volunteers. Methods In 31 IBS patients (16 patients with IBS-D; 15 patients with IBS-C) and 10 healthy volunteers the expression of CLD-2, CLD-3, CLD-5, IL-2, IL-10, and TNF-α in mucosal biopsy specimens was determined by morphological and immune-histochemical methods. The qualitative and quantitative composition of the intestinal microbiota was assessed based on 16S rRNA gene sequencing in both groups of patients. Results The expression of IL-2 and TNF-α was significantly increased in IBS patients compared with the controls (p<0.001), with a gradual increase from the duodenum to the sigmoid colon. The expression of IL-10, CLD-3, and CLD-5 in mucosal biopsy specimens of these patients was lower than in the control group (p<0.001). Increased ratios of Bacteroidetes and decreased ratios of Firmicutes were noted in IBS patients compared to healthy volunteers (p<0.05). Conclusion IBS patients have impaired gut permeability and persisting low-grade inflammation throughout the gastrointestinal tract. Changes in the gut microbiota may support or exacerbate these changes.


Patients cohort
Participants were recruited during their medical treatment at the Sechenov University, Moscow, Russia since June 8 to Septembler 11 2016. Initially, we included in the study 77 patients with symptoms of IBS according to the Rome IV criteria. Non-functional causes for the symptoms were excluded after a detailed evaluation of medical history, physical examination, extensive panel of blood tests, stool analysis, and colonoscopy with biopsies. The exclusion criteria were patients younger than 18 years, older than 59 years, patients with organic bowel disease, organic disease of pancreas and gallbladder, renal disease, hepatic insufficiency, mental illness significantly impairing self-report (schizophrenia, bipolar disorder, or epilepsy), and receiving antibiotic or probiotic therapy for the previous 6 months. After analyzing the medical history and conducting laboratory and instrumental studies, 46 patients were excluded. The final size of the cohort was 31 patients. Additionally, ten healthy volunteers were included in the control group. Written informed consent was obtained from all participants.
The study protocol was approved by the ethics committee of Sechenov University, Moscow, Russian Federation.
Sample size calculation. We did not perform a sample size calculation before the study initiation due to its complexity. Based on an analysis of the previous studies, we found a large difference in the levels of pro-inflammatory cytokines in IBS patients compared with controls. Using G � Power software [21,22] we calculated the analysis power, which reached 0.99 in all cases and, therefore, the sample size is sufficient for achieving statistical significance.
Fixation of the biopsy specimen of the mucous layer was carried out in a 10% solution of neutral formalin for 24 hours, followed by embedding in paraffin. Paraffin sections of 3-5 microns thick were prepared using a microtome. The sections were fixed onto glass slides coated with an adhesive (polylysine, APES) and incubated in a thermostat at 37˚C for 12 hours. Then the sections were dewaxed and dehydrated in a battery of 3 xylenes, 2 absolute alcohols, 2 95% alcohols, 80% and 70% alcohol and distilled water.
The results were derived with both standard quantitative and semi-quantitative methods, depending on the percentage of stained cell nuclei. The intensity of staining was determined by a 6-point scale: 2 points-up to 20% of stained cells; 4 points-from 20% to 40% of stained cells; and 6 points-more than 40% of stained cells. The localization and intensity of staining (membrane and diffusive cytoplasmic staining) for CLDs were separately evaluated.

DNA isolation and 16S library preparation
Fecal samples from 21 patients with IBS and 10 healthy volunteers were collected. Total DNA was isolated using AmpliPrimeDNA-sorb-AM kit (NextBio, Russia) for clinical specimens, according to the manufacturer's protocol. The isolated DNA was stored at -20˚C. For qualitive and quantitative assessment of the isolated DNA we used NanoDrop 1000 equipment (Thermo Fisher Scientific, USA). 16S library preparation was carried out according to the protocol of 16S Metagenomic Sequencing Library Preparation (Illumina, USA), which is recommended for Illumina MiSeq sample prep. The first round of amplification of V3-V4 16S rDNA variable regions was performed using the following primers: forward (TCGTCGGCAGCGTCAGATGTG TATAAGAGACAG-CCTACGGGNGGCWGCAG) and reverse (GTCTCGTGGGCTCGGAGATGTG TATAAGAGACAG-GACTACHVGGGTATCTAATCC) [23]. These primers are aimed at the amplification of bacterial (more than 90%) but not archaeal (less than 5%) rRNA genes. The approximate size of target amplicons was 440-460 bp. The amplification program (Applied Biosystems 2720 Thermal Cycler, USA) was the following: 1) 95˚C-3m; 2) 30 cycles: 95C -30s; 55˚C-30s; 72˚C-30s; 3) 72˚C-5m; 4) 4˚C. The derived amplicons were purified with AgencourtAMPure XP (Beckman Coulter, USA) beads according to the manufacturer's protocol. The second amplification round was used for double-indexing samples with a combination of specific primers. The amplification program was the following: 1) 95˚C-3m; 2) 8 cycles: 95˚C-30s; 55˚C-30s; 72˚C-30s; 3) 72˚C-5m; 4) 4˚C.
The purification of products of the second PCR round was also carried out using Agencour-tAMPure XP. Concentration of the derived 16S rDNA libraries was measured using Qubit 2.0 fluorimeter (Invitrogen, USA) using Quant-iT dsDNA High-Sensitivity Assay Kit. The purified amplicons were mixed equimolarly according to the derived concentration values. Quality of the libraries was evaluated using Agilent 2100 Bioanalyzer (Agilent Technologies, USA) and Agilent DNA 1000 Kit. Sequencing was carried out on MiSeq machine (Illumina) and MiSeq Reagent Kit v2 (paired-end reads, 2x250 nt).

Bioinformatics analysis of 16S sequencing data. Statistical analysis
The obtained sequences of forward and reverse reads were merged using MeFiT and CASPER, before quality-trimming and filtering with DADA2. For the most samples more than 99.5% reads were successfully merged. Further, the merged reads were analyzed by the DADA2 package (a part of the Bioconductor project) for R [24]. The analysis included the following steps: 1) the primer sequences were removed using cutadapt; 2) the reads were filtered by quality; 3) error distribution models were derived based on read quality profiles; 4) the sequencing errors were corrected; 5) RSV (ribosomal sequence variant; an analog of OTU) sequences were obtained; 6) chimeric RSVs were eliminated. Next, taxonomic annotation of the derived RSVs was carried out using DADA2 and the GreenGenes 13.8 16S reference sequence database [25].
Statistical analysis was carried out using Statistica for Windows 10.0 (StatSoft Inc.). Qualitative characteristics were described using absolute and relative (%) indicators, and quantitative characteristics using the median and 95% confidence interval (median [95% CI]). For the evaluation of statistical significance of differences between two and more independent groups, we used Mann-Whitney and Kruskal-Wallis non-parametric tests, respectively. For qualitative indicators, the Chi-squared test and, when necessary, exact Fisher test were used. For the correlation analysis, we used Kendall's tau approach. Analysis of variance (ANOVA) was used when comparing more than two groups with quantitative indicators. In some cases, p<0.05 was considered as the significant level. In multiple comparisons, the Bonferroni correction was used (p-value threshold is defined as p = 0.05/n, where n is the number of comparisons of the same data).

Results
Main characteristics of the observed groups are summarized in Table 1. There were no significant differences in gender distribution and average age.

The expression of pro-inflammatory cytokines in the mucosal biopsy specimens of IBS patients was significantly higher than in the control group
The biopsies of the duodenum, ileum, cecum and sigmoid colon were performed in IBS patients and healthy volunteers. A statistically significant increase in the expression of the proinflammatory cytokine TNF-α was found in the mucosal biopsy of IBS patients from duodenum compared to healthy individuals (3.5 ± 0.5 versus 1.1 ± 0.3 points, respectively), ileum (3.6 ± 0.5 and 1.1 ± 0.3, respectively), cecum (5.4 ± 0.5 and 1.0 ± 0.2) and sigmoid colon (5.6 ± 0.5 and 1.2 ± 0.3; p < 0.0001 for all localizations; Fig 1).

The gradient of cytokines and tight junction proteins expression in the mucosal biopsy IBS patients and in control group
The expression level of pro-inflammatory cytokines increased gradually from duodenum to sigmoid colon in IBS group, while in control it remained the same (Fig 6). Statistically significant changes in TNFα expression were observed between ileum and duodenum and ileum and cecum (p<0.05), for IL-2 between ileum and duodenum (p<0.05).

Qualitative and quantitative analysis of intestinal microbiota composition based on 16S rRNA gene sequencing revealed significant changes between IBS patients and healthy individuals
As a result of sequencing of 16S rRNA genes of fecal microbiome taken from 31 IBS and 10 healthy volunteers, a total of 12267 ribosomal sequence variants (RSV) were inferred. Among them, 4716 (62%) RSVs were treated as potentially chimeric (they correspond to 4.6% reads)   Fig 7).
Considering fecal microbial alpha diversity, ideally, the Chao1 index should reflect the species diversity of the microbiome (and Shannon index-both its diversity and balance), i.e. it should be calculated at the species level. Nevertheless, in most cases, we can really operate only at the RSV or genus/family level because of very poor annotation of reads at the species level. However, each RSV does not correspond to each species of bacteria. The number of RSVs may be either less than the real number of species (when sequencing too small fragments of the rRNA genes) or, as a rule, much greater than the actual number of species (because of presence of non-corrected sequencing errors, non-eliminated chimeras) [26]. Therefore, since we are mainly interested in comparing diversity between patient groups, we based our conclusions on Chao1 and Shannon indices at the genus level. The values of Shannon index for IBS patients (1.86 [1.17; 2.89]) tended to be lower than for healthy individuals (2.19 [1.38; 2.64]), but the difference was not statistically significant (p = 0.2) mainly because of the limited cohort size. We noticed the same tendency at the family and RSV levels. As it was expected, the values of the indices calculated at the family level were lower (and at RSV level-higher) than at the genus level (Figs 8 and 9). We noticed that the values of Chao1 indices were strongly correlated with the reads counts (Spearman's rank correlation coefficient r = 0.67, p < 0.001), whereas Shannon index values demonstrated only slight correlations (r = 0.15, p = 0.02). The strong correlations between Choa1 index and reads counts may indicate: a) the excessive number of RSVs (because of sequencing errors/chimeras) or b) presence of lowly abundant taxons, which were not completely covered with the current sequencing depth. Since, in these cases it is better to consider only Shannon index values calculated at the genus level.

Discussion
Despite the extensive research, etiology and pathophysiology of IBS remain incompletely understood. Proposed mechanisms involved in the pathogenesis include increased intestinal permeability [27], changes in the immune system reactivity [28], and gut microbiota composition [29,30], visceral hypersensitivity, and impaired gut motility [31,32].
Disruption of the intestinal barrier permeability due to a decrease of membrane protein expression leads to the inflammatory response of the intestinal wall, which results in changes in gut sensitivity and motility [2,5,33,34]. Inflammatory changes in the intestinal wall may also contribute to an increase in the number of Bacteroidetes in the intestinal microbiota composition [20]. Bacteroides, in turn, can dissolve mucosal glycoproteins [35], aggravate the intestinal barrier permeability due to the splitting of dense contact proteins [36]. On the other hand, previously it was shown that SCFAs modulate the expression of claudines 3, 4 and occludin, which are capable of increasing the barrier properties of the epithelium [37,38]. The predominant bacteria that produce SCFAs in human microbiota are classified as family Ruminococcaceae (cluster IV) and genus Eubacterium (cluster XIVa), both are included in the order Clostridiales, class Clostridia, phylum Firmicutes [39]. Our study supports this hypothesis, according to the obtained data the abundance of butyrate producing bacteria is lower in IBS patients than in healthy individuals (Fig 10).
The low abundance of SCFA-producing bacteria in intestinal microbiota may exacerbate existing impaired permeability and promote immune system activation [38]. The mast cells degranulation causes release of inflammatory mediators (histamine, serotonin and proteases) resulting in lymphocyte activation and cytokine imbalance [40]. Patients with IBS were found to have higher levels of proinflammatory IL-6, IL-8, IL-1β, TNF-α and lower levels of antiinflammatory IL-10 in both serum and intestinal mucosa [41,42]. In addition, mast cells degranulation was shown to reduce the expression of tight junction proteins, probably through tryptase release [32].
Thus, the combination of disturbed intestinal permeability and altered composition of the intestinal microbiota lead to the developing of IBS symptoms. The expression gradient of proinflammatory cytokines (TNFα and IL-2) may be associated with an increase in the number of microbial cells in the distal part of colon [43].
We admit that this study has some limitations such as the lack of direct gut permeability measurement, precise analysis of gut microbiota for patients with different variants and duration of IBS. A broader cytokine panel would be useful for a more detailed analysis of gut inflammation. Also, up to date, no data exists on how disease severity or duration can influence the intestinal microbiome, permeability, and degree of intestinal wall inflammation. Larger cohorts of patients to examine would be beneficial for future studies. The study of the tight junction proteins expression, the composition of the intestinal microbiota, the degree of inflammation in the intestinal wall, and their correlation with different duration of IBS history and symptoms severity may allow us to use them as biomarkers, and create a new approach for the therapy.

Conclusion
Our study confirms the results of previous articles related to the role of impaired gut permeability and changes in the composition of the intestinal microbiota in patients with IBS. However, the relationship between these two pathogenetic mechanisms has not been studied before. A better understanding of the pathogenesis of IBS will improve the diagnosis and treatment of these patients, but further research including larger patient's cohorts and a more detailed analysis of the gut microbiota is needed.