Introduction

In the gut, intestinal epithelial cells form a structural barrier, delineating the intestinal lumen from the lamina propria¹. Within the lumen, microbes exert essential functions in maintaining gastrointestinal homeostasis and functionality². Regardless of commensal or pathogenic nature of the bacteria, their interaction with gut epithelium is inevitable. During ulcerative colitis (UC), both mucosal barrier and microbial composition are disturbed³. However, it is still unclear whether naturally occurring gut bacteria might influence the control or management of the protective lining of cells in the colon.

Aims & Methods

The aim of the study was to identify which faecal microbiota components remain unaltered during UC and assess the interplay of these bacteria with colonic epithelial cells.
The study was approved by the Kaunas Regional Biomedical Research Ethics Committee (protocol no. BE-2-31). Faecal microbiota was analysed by employing 16S rRNA-gene v1-v2 region sequencing on stool samples of active and quiescent UC patients (n=47) and non-IBD controls (n=25). Functional analysis was performed using colonic epithelial organoid-derived monolayers of UC patients (n=9) and non-IBD controls (n=8). Monolayers were co-cultured with Escherichia coli (ATCC 25922) and Phocaeicola vulgatus (ATCC 8482) bacteria type strains. The interplay between bacteria and colonic epithelial cells was assessed by analysing the expression of gene markers for epithelial barrier integrity (ZO1), pathogen recognition (TLR4) and stress induction (HSPA1A, HSPB1).

Results

First, 16S rRNA-gene sequencing revealed a persistent bacterial imbalance in both active and quiescent UC patients compared to non-IBD controls. Specifically, Chao1, Shannon, and Simpson α-diversity indices were significantly lower (p_Wilcoxon < 0.05) in both active and quiescent UC patient groups when compared to non-IBD controls. Microbial community clusters, assessed by Bray-Curtis dissimilarity index, also differed significantly between UC patients and non-IBD controls (p_PERMANOVA < 0.05). Subsequent co-occurrence and differential abundance analyses identified that in contrast to healthy individuals, UC patients exhibited statistical significance (p_{adj. Wilcoxon} < 0.05) in the relative abundance of 5 genera, including Alistipes, Mediterraneibacter, Paraprevotella, etc., while 35 genera were found to be present at comparable levels (p_{adj. Wilcoxon} ≥ 0.05). Of them, the most abundant ones were Phocaeicola, Collinsella, Roseburia, Holdemanella, and Bacteroides. Functional analysis, utilizing co-cultures of commensal bacteria and colonic epithelial cells of active UC patients and non-IBD controls, showed that both E. coli and P. vulgatus did not induce pathogen-pattern recognition and stress responses. Interestingly, both species demonstrated potential to provoke diverse tight junction-related cellular reactivity (however, due to high variation in patient-to-patient responses, results did not reach statistical significance (p_{adj. Wilcoxon} ≥ 0.05)). More precisely, a contact with E. coli and P. vulgatus tended to increase the expression of ZO1 in non-IBD epithelial cells (FC = 3.59 and FC = 2.27, respectively) compared to untreated cells, while an opposite trend was observed in epithelial cells of UC patients (FC = 0.67 and FC = 0.75, respectively).

Conclusion

Diverse colonic epithelial cell responses to distinct commensal bacteria species imply the contribution of stably abundant gut microbiota to the impairment of mucosal barrier integrity during UC.

References

¹Horowitz, A., Chanez-Paredes, S.D., Haest, X. et al. Paracellular permeability and tight junction regulation in gut health and disease. Nat Rev Gastroenterol Hepatol 20, 417–432 (2023)
²Fan, Y., Pedersen, O. Gut microbiota in human metabolic health and disease. Nat Rev Microbiol 19, 55–71 (2021)
³Kobayashi, T., Siegmund, B., Le Berre, C. et al. Ulcerative colitis. Nat Rev Dis Primers 6, 74 (2020)

Introduction

Patients with ulcerative colitis (UC) often have unpredictable response over time. Using clinical parameters has proven useful, however, deeper understanding of the molecular and cellular mechanisms that influence therapeutic outcomes in head-to-head trials is lacking. Here we performed a large-scale transcriptomics analysis of tissue biopsies derived from moderate-to-severe UC patients treated with vedolizumab (VDZ) or adalimumab (ADA) in the VARSITY trial (Sands, 2019).

Aims & Methods

The aim of this study was to identify biomarkers of response to VDZ and ADA treatment. We performed bulk RNA sequencing of 954 biopsies collected using the flexible sigmoidoscopy from pre-treatment/baseline, week (w)14 and w52. Our dataset comprised complete longitudinal tissue sampling of n=151 non-responders, n=117 responders based on clinical remission at w52. We employed comparative analysis to pinpoint differentially expressed genes (DEG) and differential pathways for ADA and VDZ treated patients. We then employed systems network strategies to identify gene modules (Choobdar, 2019) associated with clinical parameters. Finally, we utilized multivariate trajectory model (Nagin, 2018) to characterise longitudinal patient trajectories based on study endpoints and integrated transcriptomics signatures to patient variable response (Schreiber, 2023).

Results

Variable baseline gene expression profile was associated with prior exposure to anti-TNFα, corticosteroid or immunomodulator treatment. DEG and gene-modules linked to VDZ response in tissue biopsies revealed inflammatory, fibrotic, epithelial, and neuronal pathways, including but not limited to JAK/STAT, G-Protein Coupled Receptor signalling, ECM interactions, oxidative phosphorylation, and neuroglial signatures. Importantly, these consistent gene and gene-module predictors of clinical remission were detected at pre- and post-treatment (w14 and w52) to VDZ. For ADA treated patients, no DEG and gene-modules were significantly associated with treatment-response at baseline after FDR correction. Upon relaxing stringency, we were able to find pathways such as neutrophil degranulation, MAPK signalling, and Fc-mediated immune regulation linked to ADA response.
Cell type contributions were in silico inferred with deconvolution analysis of bulk gene expression data. In agreement with the pathway analysis, we identified epithelial, plasma, stromal populations as baseline biomarkers of response to VDZ, while specific CD4+, CD8+ T cell subsets, follicular B cells and inflammation-associated fibroblasts were significantly elevated in non-responders compared to the responders to VDZ from baseline to post-treatment. In ADA non-responding patients, CD4+ T cells were significantly increased compared to responders at baseline and post-treatment. Gene-module predictors were linked to distinct patients' trajectories based on symptoms; the greatest correlations with transcriptomics findings were observed between our extreme responding groups from trajectory analyses.

Conclusion

We present first-of-a-kind analysis using large-scale clinical trial integrating longitudinal bulk RNA-seq analysis with clinical outcomes in response to two different therapies. For VDZ, as opposed to ADA, we were able to find strong baseline predictors also preserved after treatment. We hope that similar approaches could support clinical precision medicine applications to help patients achieve faster and lasting disease control.

References

Sands, BE, et al. N Engl J Med 2019;381(13):1215-1226.
Choobdar, S, et al. Nat Methods 2019;16(9):843-852.
Nagin DS, et al. Statistical Methods in Medical Research 2018;27(7):2015-2023
Schreiber, et al., United European Gastroenterology Journal 2023; 11 (Supplement 8)

Disclosure

All authors are employees and shareholders of Takeda.