Abstract :
[en] Inflammatory bowel disease (IBD), which includes Crohn's disease and ulcerative colitis, is a common complex disease (CCD) characterized by chronic inflammation of the gastrointestinal tract. The etiology of IBD is strongly influenced by genetics, with genome-wide association studies (GWAS) identifying 241 risk loci associated with the disease. The typical span of a risk locus is approximately 250 kb, encompassing between 0 and 100 genes. Coding variants are responsible for disease associations at only 14 of these risk loci. It has been shown that the majority of observed associations can be attributed to regulatory variants that perturb the expression of one or more neighboring genes in cis. Expression quantitative trait locus (eQTL) analyses in disease-relevant cell types obtained from healthy individuals are one method for identifying causal genes. Indeed, it seems reasonable to postulate that common regulatory variants causing eQTL effects can be detected in healthy individuals. Colocalization methods have been developed to quantify the similarity between disease association patterns (DAP) and expression association patterns (EAP) derived from GWAS and eQTL analyses, respectively. However, matching cis-QTLs to uncover putative causal genes has been successful for only 31.5% (63/200) of the IBD risk loci analyzed. The large proportion of "orphan" risk loci may be explained by the fact that the matching eQTLs have not yet been identified. Possible reasons include the absence or underrepresentation of disease-relevant cell types in existing datasets, that some eQTLs are only active in specific contexts (such as the disease process), or that the eQTLs driving CCDs are different from those discovered so far, which often have limited sample sizes compared to GWAS. The aim of this thesis is to investigate the hypothesis that disease-relevant cell types may have been absent or underrepresented in existing datasets. To discover novel eQTLs driving inherited predisposition to IBD, we established the CEDAR-2 cohort with healthy Europeans, dividing it into two parts. In the first part, we collected peripheral blood from 200 individuals and isolated peripheral blood mononuclear cells (PBMCs), as well as 26 circulating immune cell types using fluorescence-activated cell sorting (FACS) and magnetic-activated cell sorting (MACS). We then performed ultra-low input RNA sequencing (RNA-Seq) on each of the more than 5,000 samples. In the second part, we collected intestinal biopsies at three locations (ileum, colon, and rectum) from 60 individuals, performed single-cell RNA-Seq (scRNA-Seq), and defined 43 intestinal cell populations. All individuals were genotyped for approximately 700,000 SNPs and imputed to 6.3 million variants. We developed and applied a novel cell type annotation-free eQTL analysis approach for scRNA-Seq data. We identified 60,113 and 35,010 cis-eQTLs in blood and gut tissues, respectively, affecting 12,926 genes. We merged these eQTLs into 24,745 regulatory modules (RMs), most of which are cell type-specific, location-specific (small versus large intestine), or compartment-specific (blood versus gut). Next, we compared our catalogue of EAP with the DAP in 206 risk loci for IBD using published GWAS summary statistics, employing colocalization methods. We identified matching patterns for 140 risk loci, implicating 556 genes, of which 366 have not previously been associated with IBD. We identify 3.5 times more risk loci and 9.4 times more eQTL genes with DAP-matching EAP with our combined blood and gut dataset, than with whole blood eQTL information from 31,684 individuals. We examined the effect of disease status on the expression of our catalogue of 556 DAP-matching candidate genes in blood and gut tissues, showing that the expression of 119 genes is significantly affected in a manner consistent with the sign of the eQTL effect of risk variants. Notably, eight genes in our list of candidates are or have been clinically tested for treating IBD. We also identified ten genes targeted by drugs that are at least in clinical phase III for diseases other than IBD, which act on their targets in a manner consistent with the effects of IBD risk variants, with the expression of four of these genes being affected by disease status. Additionally, we developed a web browser to visualize our data.
