(A-29) Predicted Regulatory Effects of Gut Microbiome on Estrogen Receptor Beta Signaling in Ulcerative Colitis by Supervised LIVE Modeling

Introduction::
Introduction: Ulcerative colitis (UC) is a chronic colonic inflammatory disorder that involves interactions between gut microbiome dysbiosis, epithelial barrier disruption, and immune hyperactivity. A sex-based difference exists in UC with males 20% more likely to develop UC and 60% more likely to progress to colitis-associated cancer than females. A possible explanation for the difference is the anti-inflammatory and epithelial-protective role of estrogen signaling via estrogen receptor beta (ESR2) in the gut. Here, by using our supervised Latent Interacting Variable Effects (LIVE) modeling, we characterized how gut microbes and metabolites regulate ESR2 signaling in UC and experimentally validated transcriptomic effects of top predictive gut microbiome interactions on ESR2.




Materials and Methods::
Materials and Methods: Gut metabolomic, colonic transcriptomic, and stool 16S rRNA-seq data from 35 unique UC or non-IBD controls subjects were obtained from the IBD Multi-Omics Database. Single sample gene set enrichment analysis (ssGSEA) was used to calculated scores for downstream genes up or down-regulated by ESR2 (ESR2UP/ ESR2DN). Our supervised LIVE modeling framework extracted latent variables (LV) scores of metabolites and 16S data predicting ESR2UP or ESR2DN (sPLSR, R package Mixomics V6.16.3), and integrated them with clinical covariates of sex and UC status to build a multi-omic metamodel using ESR2UP or ESR2DN scores as dependent variables. Significance testing on regression coefficients identified LV interactions synergistically predictive of ER2 pathway activity. Predictive features were extracted via examination of LV loading coefficients, pairwise correlation analysis, and Mann-Whitney testing for univariate differences. The effects of top interacting and predictive gut microbiome features were evaluated in HT-29 female colon cancer cell line. Expression of ESR2 genes were quantified and protein activity assays are implemented to study activated ESR2 translation in the nucleus.




Results, Conclusions, and Discussions::
Results and Discussions: The sPLSR models, using intestinal 16S or metabolomics data to predict ESR2UP or ESR2DN scores, associated the metabolomics with patient clustering by disease status, while the 16S models did not stratify patients on disease status or sex. The multi-omic meta-model accounting for sex and disease status was significantly predictive of ESR2UP and ESR2DN pathway activities, implicating main effects of microbiota in LV1 (p=0.005, p=0.002) and LV2 (p=0.002, p=0.005), UC status (p=0.009, p=0.008), and the interaction effect metabolites:microbiota in LV1 (p=0.004, p=0.0028). For predicting exclusively ESR2UP pathway, significant effects were microbiota:sex in LV2 (p=0.009), metabolite:sex in LV2 (p=0.009), and metabolite:UC in LV2 (p=0.008). For predicting ESR2DN, significant effects were microbiota: sex in LV2 (p=0.01), and metabolite:sex in LV2 (p=0.005), and microbiota:UC in LV1 (p=0.009). Utilizing the metamodel and loadings from sPLSR, 41 interacting microbial species and metabolites were identified via Spearman’s correlation of highly-loaded taxa and metabolites, Mann-Whitney tests and single regression models of metabolite abundance by sex or disease status. The predicted regulatory effects of R. gnavus with acesulfame, and E. rectale with 4 –Methylcathecol on ESR2 activity were validated HT-29 female colon cancer at the transcriptomic level on S100A6 (p-value: 0.01) and RPL21 (p-value: 0.02) genes, indicating a novel protective gut microbiome regulatory mechanism for ulcerative colitis.

Conclusions: We demonstrate the effectiveness of supervised LIVE for modeling relationships between host signaling and microbiome multi-omics data via this investigation into ER2 activity in UC patients. We quantified significant multi-omic microbiome interactions associated with disease status and sex that have an impact on ESR2 signaling that can aid in identifying new therapeutics for UC.









Acknowledgements (Optional): :

References (Optional): :