A novel approach for the use of normal colon organoids in the identification of true casual variants and mechanistic interrogation of their gene targets in colon cancer risk

University Of Virginia, Charlottesville VA

Investigators

Abstract

PROJECT SUMMARY Genome-wide association studies (GWAS) have identified over 200 inherited genetic loci associated with colorectal cancer (CRC). However, most GWAS variants (index) are not functional but exist in linkage disequilibrium (LD) with causal variant(s). The prevailing hypothesis is that causal variants map to regulatory elements, such as enhancers, where they disrupt transcription factor binding site (TFBS) recruitment, which then perturbs gene expression and disease risk. Identifying these causal variants (and their gene targets) represents a necessary step for accelerating the translational potential of CRC GWAS. Yet, progress is drastically slowed complex LD structures within these loci, whereby index variants are not only in LD with causal variants but with 10’s-100’s of non-functional variants. Improved variant prioritization strategies are urgently needed if the translational potential of CRC GWAS is to be realized. Recently, our group found a significant overlap between differentially expressed genes associated with CRC environmental factors and genes mapping to CRC GWAS loci following several exposure studies in normal colon organoids. Using a machine learning approach, we also found that accounting for exposure-mediated differences in cell composition in regression models reduced the reporting of significant CRC GWAS genes by 79.6%.This led to our central hypothesis that CRC GWAS genes are cell type specific. To test this hypothesis and determine the role of genes within colon epithelial cells, we performed weighted gene co-expression network analysis (WGCNA) on 15 normal colon organoid lines grown in normal media (enriching for undifferentiated cells) or in media enriching for stem cells. We identified 127 genes within 6 cell type specific modules that mapped to 91 distinct CRC GWAS loci. Through the addition of single nuclear ATAC-Seq, we prioritized a subset of variants for functional follow-up. Through luciferase enhancer assays and CRISPR-cas9 editing approaches, we provide strong evidence of a causal role for one of these variants, rs1675247, and its gene target, DLGAP1-AS2, in CRC risk. Notably, DLGAP1-AS2 was a highly connected gene identified using WGCNA. Here, we propose to expand and improve upon our risk variant and gene prioritization strategy through a well- powered (n=87), integrative analysis of transcriptomic, genotypic and chromatin accessibility differences. We will identify functional mechanisms through which quantitative trait loci (QTL)s disrupt TFBS motifs and/or alter expression (eQTL) of genes driving CRC relevant co-expression modules. Through the integration of CRC GWAS summary statistics, we will directly relate QTL effects to causal CRC risk associations through state-of- the-art fine-mapping and genetic colocalization approaches. Finally, we will functionally validate candidate causal risk variant effects through luciferase enhancer assays and through a high-throughput, cutting-edge CRISPR Droplet Sequencing (CROP-Seq) approach that directly quantifies the effects of variants on single cell gene expression. In doing so, this proposal will aid in improving the translational discovery power of CRC GWAS.