Gut dysbiosis and clinical phases of pancolitis in patients with ulcerative colitis

Abstract Ulcerative colitis (UC) is a frequent type of inflammatory bowel disease, characterized by periods of remission and exacerbation. Gut dysbiosis may influence pathophysiology and clinical response in UC. The purpose of this study was to evaluate whether gut microbiota is related to the active and remission phases of pancolitis in patients with UC as well as in healthy participants. Fecal samples were obtained from 18 patients with UC and clinical‐endoscopic evidenced pancolitis (active phase n = 9 and remission phase n = 9), as well as 15 healthy participants. After fecal DNA extraction, the 16S rRNA gene was amplified and sequenced (Illumina MiSeq), operational taxonomic units were analyzed with the QIIME software. Gut microbiota composition revealed a higher abundance of the phyla Proteobacteria and Fusobacteria in active pancolitis, as compared with remission and healthy participants. Likewise, a marked abundance of the genus Bilophila and Fusobacteria were present in active pancolitis, whereas a higher abundance of Faecalibacterium characterized both remission and healthy participants. LEfSe analysis showed that the genus Roseburia and Faecalibacterium were enriched in remission pancolitis, and genera Bilophila and Fusobacterium were enriched in active pancolitis. The relative abundance of Fecalibacterium and Roseburia showed a higher correlation with fecal calprotectin, while Bilophila and Fusobacterium showed AUCs (area under the curve) of 0.917 and 0.988 for active vs. remission pancolitis. The results of our study highlight the relation of gut dysbiosis with clinically relevant phases of pancolitis in patients with UC. Particularly, Fecalibacterium, Roseburia, Bilophila, and Fusobacterium were identified as genera highly related to the different clinical phases of pancolitis.


| INTRODUC TI ON
The intestinal tract houses a large and diverse community of microorganisms collectively referred to as the gut microbiota. These microorganisms contribute to human health by promoting both immune and metabolic functions (Burman et al., 2016;Chassaing et al., 2017). It is widely accepted that the gut microbiota has a crucial role in regulating the function of the intestinal epithelium, the immune system, and its homeostasis within the gut (Imhann et al., 2018). The term "dysbiosis" refers to an imbalance in the composition and function of the microbiota (Danilova et al., 2019;Nishida et al., 2017;Vemuri et al., 2017), whereas gut dysbiosis along with the altered host immune response has been observed in clinically relevant immunological and inflammatory diseases, such as Ulcerative Colitis (UC), which is a frequent type of Inflammatory Bowel Disease, characterized by periods of remission and exacerbation (Nishida et al., 2018). UC has been classified according to its extent and severity in the so-called Montreal classification, defining the extent as E1 indicates ulcerative proctitis; E2 as UC on the left side; and E3 as extensive UC or pancolitis. Likewise, the severity of the disease is classified into clinical remission (S0), mild disease (S1), moderate disease (S2), and severe disease (S3). Pancolitis is considered the most serious clinical phase of UC, and its involvement represents 10%-15% of all UC (Mohammed Vashist et al., 2018).
Although a causal effect has not been evidenced; nowadays, it is widely accepted that altered interactions between gut dysbiosis and the intestinal immune system promote UC (Imhann et al., 2018), while the precise nature of the intestinal microbiota dysfunction in UC remains to be elucidated. In this sense, the gut microbiota has been considered as a "fingerprint" reflecting the natural history of UC, since it associates with the clinical severity, remission, and flare-up responses (Marchesi et al., 2016). Gut microbiota from patients with UC has been characterized by a reduced number of bacteria with anti-inflammatory capacities and a higher proportion of bacteria with pro-inflammatory properties. Microbiota diversity is also reduced; low abundance of microorganisms like Firmicutes and high abundance of Proteobacteria have been found (Manichanh et al., 2012;Yu, 2018). Rapid development and application of cultureindependent, high throughput DNA-based sequencing technologies have elicited the recognition of such dysbiotic signatures, which may play a role during the early identification of clinical-therapeutic phases of UC, and particularly useful in severe clinical manifestations like pancolitis (Peterson et al., 2008;Rintala et al., 2017).
Despite this notion, the relation of gut dysbiosis with pancolitis has been poorly characterized. Given the increasing UC prevalence worldwide, including Latin American countries (Bosques-Padilla et al., 2011;Farrukh & Mayberry, 2014), along with the strong interest to understand the relation of dysbiotic gut microbiota with most serious phases of UC like pancolitis, the present study aimed to characterize gut microbiota from patients with UC and different clinical phases of pancolitis.

| Study population
In this cross-sectional study, groups of 18 patients with UC and clinical-endoscopic-evidenced pancolitis (active phase n = 9 and remission phase n = 9) as well as 15 healthy participants, attended the Department of Gastroenterology, Centro Médico Nacional '20 de Noviembre' ISSSTE, Mexico City, Mexico, between July 2017 and January 2019. Patients with concomitant irritable bowel syndrome, pseudomembranous colitis, and antibiotic treatment during the previous 4 weeks were excluded. Pancolitis was defined according to clinical, radiological, endoscopic, and histological criteria (Van Assche et al., 2010). All the patients had experienced at least one previous episode of pancolitis before their recruitment. The study at remission phase of pancolitis received therapy based on pharmacological treatment, a fiber-rich diet, and the use of probiotics (Owczarek et al., 2016). Some patients with active pancolitis did not receive treatment due to non-medical reasons, like the inability to attend their follow-up appointment. active and remission phase, gut dysbiosis, gut microbiota, pancolitis <2 for at least 3 months (Siegel et al., 2018;Van Assche et al., 2010).
Healthy participants were volunteers without previous history of chronic disease, belonging to a different family than those with UC, but with a similar diet, as assessed by a 24-h recall (R24H) survey (Parks et al., 2018).

| Stool samples
Stool samples were collected either during hospitalization (active pancolitis) or prepared at home and collected during programmed medical consultation (remission phase and healthy participants); samples were stored at home between 4 and 8°C for up to 24 h, before hospital collection. Samples were collected with the help of a stool sampling kit, which consisted of a plastic lining to cover the toilet, two stool sample tubes with spoons, two plastic bags, and a clipping system for safe closure of the outer bag. Samples were labeled upon arrival, and one part was processed for fecal calprotectin assay; while the remaining was aliquoted and frozen directly at −80°C for further microbiota analyses (Tedjo et al., 2015).

| DNA extraction of fecal samples
Frozen stool samples were thawed on ice, and approximately 200 mg were added to dry-bead tubes with lysis buffer (AllPrep PowerFecal DNA, Qiagen). The stool samples were homogenized followed by a combined chemical and mechanical lysis by using prefilled lysis tubes. Inhibitors commonly present in stool samples were then removed before isolation of nucleic acids. DNA isolation was continued by using the AllPrep DNA MiniElute spin column, according to the manufacturer's instructions. DNA was eluted in 30 μl EB-buffer.
Negative control samples (consisted only of PCR grade water) were handled in the same way as the fecal samples, to rule out contamination during the isolation procedure (Tedjo et al., 2015). A Nanodrop ND-1000 (NanoDrop Technologies), was used to estimate DNA concentrations. DNA concentration was adjusted to a final concentration of 10 ng/µl (Tedjo et al., 2016).

| Amplification and sequencing of bacterial 16S rRNA gene
The V3 and V6 hypervariable regions of the 16S rRNA gene were PCR  (Dubinsky & Braun, 2015;Haas et al., 2011). The PCR products were evaluated by 2% agarose gel electrophoresis and purified. After purification, spectrophotometry was used to quantify the PCR products. Samples were normalized to a final concentration of 2 nM.

| Microbial composition and analysis by Illumina
A two-steps PCR methodology was used to prepare 16S rRNA libraries. For the first-step, extracted DNA was quantified and samples were diluted to the amount of the least concentrated sample. Then 2 μL were used for the PCR reaction (quadruplicates) at the follow-

| Bioinformatic analysis
The Illumina Real-Time Analysis software (version 1.17.28) was used for base calling, image analysis, and error estimation. Sequencing provided read lengths of 300 bp, which were demultiplexed, verifying that the paired ends provided a clear overlap. The paired ends were then linked together with the fastq-join program (http://code. google.com/p/ea-utils/). Separate files of each sample (R1 and R2) were entered in fastq format by using the split_libraries_fastq.
py pipelines. Sequences that had quality value (QV) scores of ≥20 (Phred score of 20) for no-less than 99% of the sequence were selected for further study. All sequences with ambiguous base calls were discarded. Subsequently, the sequences were grouped in Operational Taxonomic Units (OTU), where the pick_closed_refer-ence_otus.py pipelines were used. QIIME, which uses the BIOM format, was used to represent OTU tables (Bolyen et al., 2018;Dubinsky & Braun, 2015;Edgar et al., 2011). Analyses of sequence reads were performed by using SILVA multiclassifier tools with a 97% confidence threshold (Navas-Molina et al., 2013). Subsequent analyses of diversity index were all performed based on this output normalized data (Allali et al., 2017;Aßhauer et al., 2015). To perform the diversity analyses, the core_diversity_analyses.py pipelines were executed with the pipeline alpha_diversity.py. Alpha diversity metrics were calculated with QIIME, that is, the observed OTUs (observed species) and the phylogenetic diversity or complete tree PD (PD_whole_tree) (Bolyen et al., 2018); whereas the weighted distances of UniFrac of the beta diversity were determined with beta_diversity.py pipelines, and the R software v.2.15.3 was used to display the results (Barwell et al., 2015;Chao et al., 2006;Hass et al., 2011). The "Linear discriminant analysis (LDA) effect size (LEfSe)" algorithm was performed with the Galaxy online platform to determine the different relative abundances of bacterial communities among the different groups of patients. The significance thresholds used were those recommended in the program. LEfSe considered statistical significance between the different biological classes with a Kruskal-Wallis test and subsequently analyzed the biological significance with a Wilcoxon test (Segata et al., 2011).

| Fecal calprotectin test
Fecal calprotectin (FC) was measured as a marker of intestinal inflammation by using a commercial ELISA (MyBioSource), following the manufacturer's instructions. Optical densities were read at 405 nm with a microplate ELISA reader. Samples were tested in duplicate, and results were calculated from a standard curve and expressed as μg/g stool (Chang & Cheon, 2018).

| Statistical analysis
Data normality was evaluated with the Shapiro-Wilk test.
Quantitative data were compared by non-paired, two-tail, t test, or U-Mann Whitney, as appropriate. Analyses of the sequences were carried out in the QIIME and R software. Multivariate nonparametric ANOVA was used to determine the differences in the abundance of the microbial community between groups, whereas Unifrac was used to compare the abundance of the specific microbiota and the concentration of fecal calprotectin, and it was visualized by principal coordinate analysis. To test whether the clusters of microbiota from the study conditions were different between them, Unifrac p-values, based on principal coordinate analysis applied to the matrix distance, were performed to allow pairwise comparison of microbiota from clinical phases of pancolitis and healthy controls (Caporaso et al., 2010;Lawley & Tannock, 2017). Finally, the Area Under the Curve (AUC) was calculated to explore whether the relative abundance of the bacterial genus most frequently observed (cutoff value according to ROC analysis) may predict UC severity. The Statistical Package for Social Sciences SPSS v.18.0. was used, and p-values of ≤0.05 (2-tailed) were considered to be statistically significant.

| Microbial composition and diversity
The analysis of microbiome from fecal samples showed the relative abundance of OTUs at different taxonomic levels (Figure 1a, b, Table 2). OTUs were created out of the filtered tags and were grouped at a similarity of 97%. This gave a total of 1533 OTUs for the 33 samples used in this study. Taxonomic composition at the level of phyla is summarized in Figure 1a : Table A1, A2, and A3). Regarding bacterial alpha diversity comparison, pancolitis activity was related to the lowest community richness (Chao index) and diversity (Shannon index) (Figure 1c, d), whereas community richness and diversity were similar between remission pancolitis and healthy participants.
Likewise, significant differences in species dominance of microbiota (Simpson index) (Figure 1e) were found between active vs. remission pancolitis and healthy participants (Appendix : Table A4).
Interestingly, the relative abundance of the most frequent bacterial genus observed in active pancolitis was significantly different from those corresponding to remission pancolitis and healthy partic-  Quantitative data were resumed as mean ± SD and qualitative data as n (%). Statistical analysis was performed with a two-way U-Mann Whitney and Fisher's test, as appropriate.

| DISCUSS ION
Our main finding was the significant differences of fecal microbiota composition from patients with active vs. remission pancolitis, with potential clinical application. Our study population was constituted of young aged patients with UC and severe stage of pancolitis.
Scarce studies have explored gut microbiota in such population, probably due to the low prevalence of pancolitis between cases with UC. However, gut dysbiosis observed in patients with active vs. remission pancolitis in the present study is comparable with other reports (Alam et al., 2020;Danilova et al., 2019;Franzosa et al., 2019;Halfvarson et al., 2017;Imhann et al., 2018;Kumari et al., 2013;Sha et al., 2013). Our results were further validated by comparison with gut microbiota from healthy participant controls from a family who shares a similar diet and they are expected to exert a lower influence on the gut microbiota composition. These results are similar to those obtained by Franzosa et al., 2019, Kumari et al., 2013Sha et al., 2013. Particularly, the findings of a reduced proportion of the genera Faecalibacterium and Roseburia in active pancolitis, and their restoration in remission pancolitis, has also been observed in previous reports (Khan et al., 2019;Man et al., 2011;Palmela et al., 2018;Vigsnaes et al., 2012). Such characterization is relevant due to scanty information regarding microbiota abundance in the remission phase of pancolitis, whereas consistent identification of specific genus in the remission phase may be useful to design more efficient therapeutic strategies, prompted to reduce UC severity. Interestingly, a particular bacterial composition Other findings were the higher abundance of the phylum Proteobacteria, and particularly the expansion of the genus Bilophila in active pancolitis. It is known that the relative abundance of Bilophila is promoted by diets enriched in saturated fats, which increase bacterial resistance to bile elimination. Furthermore, a change in the type of fat consumed affects the composition of gut microbiota, which may modify the onset and severity of UC (Devkota & Chang, 2015;Pittayanon et al., 2020;Torres et al., 2018). Dietary modifications involving excessive consumption of fried food, dairy products, and wheat flour are associated with the development of severe diarrhea in patients with active pancolitis (Keshteli et al., 2019). In the present study, we consider that there is no significant effect derived from the modification of the diet, since the population consumed a soft diet with abundant hydration; without a specific recommendation for dietary restrictions, even during active pancolitis.
Certain species of Fusobacterium show pro-inflammatory, invasive, and adherent capacity to the intestinal mucosa, while the increased proportion of Bilophila in the gut promotes an immune response mediated by Th1, resulting in the development of colitis in experimental mice models (Bashir et al., 2016;Chen et al., 2020;Hirano et al., 2018;Liu et al., 2019;Ohkusa et al., 2002;Tahara et al., 2015;Wright et al., 2015). Although a direct pathophysiological mechanism is not possible to elucidate from the present study, we can propose that the relative abundance of some species is associated with the degree of inflammation and pancolitis, derived from the inverse relationship observed between the abundance of Fecalibacterium and Roseburia with calprotectin, a biomarker of severity of UC, which was consistent with a recent report (Björkqvist et al., 2019;Yu et al., 2019). Likewise, differences in bacterial richness, diversity, and dominance were highly related to the clinical scenarios studied. Remarkably, remission pancolitis and healthy participants showed the highest relative abundance of the phylum Firmicutes, which contributed to most of the bacterial diversity and richness (Björkqvist et al., 2019;Ganji-Arjenaki & Rafieian-Kopaei, 2018;Jandhyala et al., 2015). Further analyses of cluster distribution of bacterial communities showed differences in active pancolitis, as compared to remission pancolitis and healthy participants, which was consistent with previous studies showing a difference in the structure of microbiota between active pancolitis and healthy participants (Forbes et al., 2016;Havenaar, 2011;Louis & Flint, 2017).
Furthermore, studies characterizing gut microbiota composition and its modification during pancolitis are relevant, since (a) pancolitis provides a higher risk for colorectal cancer, whereas gut dysbiosis is thought to facilitate colorectal cancer development; (b) the study of gut microbial communities during clinical phases of pancolitis contributes to a better understanding of potential interactions with the host immune response; (c) characterization of a specific genus of gut microbial communities may own potential clinical application derived from their association with pancolitis or remission phases; and (d) specific microbial manipulation, concomitant to antibiotic use, is currently used as a therapeutic approach for UC (Alard et al., 2018;Devkota & Chang, 2015;Galazzo et al., 2019).
Finally, gut dysbiosis has been proposed as an important contributing factor to the increasing prevalence of pancolitis, with a potential role for the related clinical-therapeutic phases (Halfvarson et al., 2017;Miyoshi et al., 2018;Petrof et al., 2013). Consistently, we found a significant ability of the genus Bilophila and Fusobacterium to selectively associate with cases of activity/remission pancolitis (Fukuda & Fujita, 2014;Guo et al., 2019).
To our knowledge, this is the first study that investigated the Here, we provide a broad investigation of the fecal microbial community in Mexican patients presenting pancolitis. We demonstrate differences in the microbiota communities in patients with active pancolitis, remission pancolitis, and healthy participants. Selective association of gut dysbiosis with active/remission pancolitis may set the basis for further applications of non-invasive methods, clinically useful for early identification of disease severity.

ACK N OWLED G M ENTS
This study was funded by the E-015 institutional program and

CO N FLI C T S O F I NTE R E S T
None declared.

E TH I C S S TATEM ENT
The study was carried out according to the 1975 ethical guidelines of the Declaration of Helsinki. All participants provided written informed consent. The study was approved by the Local Committees