Dissecting the Role of Unfolded Protein Response Pathway in Inflammatory Bowel Disease: In Silico mRNA-miRNA Co-Expression Analysis

Genetic models, specifically in the intestinal epithelium, have provided avenues to understandthe diverse pathways whereby intestinal epithelial cells direct the mucosal immune state inIBD. Among these pathways of note is the unfolded protein response (UPR) induced bystress in the endoplasmic reticulum (ER). Remarkably, the genetic risk factors alone do notexplain the paradigm of ER stress and inflammation. Moreover, inflammation modulatesthe expressions of microRNAs (miRNAs) that influence the production of mRNAs or proteins.We performed a gene expression, miRNA profile, pathways and in silico mRNA-miRNA co-expression analyses in biopsies of IBD patients. Methods: Paired expression levels of mRNAsand miRNAs were measured on 15 Crohn's Disease (CD)(11 male) and 15 ulcerative colitis(UC) (11 male) inflamed and in the corresponding non-inflamed tissues, using Human Gene1.0 ST Array and miRNA 2.0 Array, on Affymetrix platform. Differential expression (DE),for both mRNA and miRNA profiles, was evaluated by performing paired t-tests, and theFalse Discovery Rate (FDR) by the Benjamini and Hochberg procedure. The gene set ReactomeUnfolded Protein Response was analyzed in terms of enrichment with respect to differentialexpression by using Random Set Method. Results: Comparing inflamed vs non inflamedtissues by pathway analysis, the whole UPR pathway, that comprises 20 genes, was associatedto both UC (P<0.00001) and CD (P<0.04). In inflamed UC and CD biopsies, 14 (ATF4,ATF6, CEBPZ, DNAJB11, DNAJB9, DNAJC3, EDEM1, EIF2AK3, EIF2S1, ERN1, HERPUD1,HSPA5, SERP1 and XBP1) (P<0.016, FDR<0.09) and 7 (ATF6, DNAJB11, DNAJB9,DNAJC3, EDEM1, SERP1 and XBP1) (P<0.027, FDR<0.16) genes, respectively, wereup regulated. A bipartite network was constructed linking one gene with a miRNA whenthey result differentially correlated (DC) with a P<0.01. In UC patients 23 miRNAs wereDE (P<0.05) and also DC with at least one gene in UPR pathway. Of these, 5 were DCwith at least 3 genes: miR-194*, miR-200C, miR-595, miR-627, miR-548I. In CD patients32 miRNAs DE (P<0.05) were also DC with at least one gene in the pathway. Of these,6 were DC with at least 3 genes: miR-369-5P, miR-380, miR-493*, miR-503, miR-675 andmiR-2355. By analyzing in UC specimens the three major arms of UPR pathway, correspond-ing to ATF4, ATF6 and XBP1 genes, we found that these genes were DC with 2, 3 and 6 miRNAs, respectively; in CD , ATF6 was DC with 7, and XBP1 with 14 miRNAs. Among hubgenes identified for specific miRNAs, the XPB1 gene was the most enriched of connections.Conclusions: Our analysis revealed that the UPR pathway is involved in the pathogenicmechanisms, principally in UC patients, and a number of miRNAs are both differentiallyexpressed and correlated with the UPR genes, shedding a new light on the paradigm of theER stress and inflammation.