Prioritization and functional analysis of GWAS risk loci for Barrett's esophagus and esophageal adenocarcinoma

Abstract

Genome-wide association studies (GWAS) have identified ~20 genetic susceptibility loci for esophageal adenocarcinoma (EAC), and its precursor, Barrett's esophagus (BE). Despite such advances, functional/causal variants and gene targets at these loci remain undefined, hindering clinical translation. A key challenge is that most causal variants map to non-coding regulatory regions such as enhancers, and typically, numerous potential candidate variants at GWAS loci require testing. We developed a systematic informatics pipeline for prioritizing candidate functional variants via integrative functional potential scores (FPS) consolidated from multi-omics annotations, and used this pipeline to identify two high-scoring variants for experimental interrogation: chr9q22.32/rs11789015 and chr19p13.11/rs10423674. Minimal candidate enhancer regions spanning these variants were evaluated using luciferase reporter assays in two EAC cell lines. One of the two variants tested (rs10423674) exhibited allele-specific enhancer activity. CRISPR-mediated deletion of the putative enhancer region in EAC cell lines correlated with reduced expression of two genes-CREB-regulated transcription coactivator 1 (CRTC1) and Cartilage oligomeric matrix protein (COMP); expression of five other genes remained unchanged (CRLF1, KLHL26, TMEM59L, UBA52, RFXANK). Expression quantitative trait locus mapping indicated that rs10423674 genotype correlated with CRTC1 and COMP expression in normal esophagus. This study represents the first experimental effort to bridge GWAS associations to biology in BE/EAC and supports the utility of FPS to guide variant prioritization. Our findings reveal a functional variant and candidate risk enhancer at chr19p13.11 and implicate CRTC1 and COMP as putative gene targets, suggesting that altered expression of these genes may underlie the BE/EAC risk association.

Figure 1

Functional analytic pipeline schematic summary.

Figure 2

Functional potential scores (FPS). (A) Histogram distribution of FPS for 1565 candidate risk SNPs in 19 GWAS susceptibility regions. (B) Scatter plot of FPS for individual candidate risk variants in each of these 19 regions. Red, GWAS index SNP.

Figure 3

Genome Browser plots for selected GWAS loci. (A) Region 13 (9q22.32/BARX1), (B) Region 19 (19p13.11/CRTC1). red: index variants, blue r² > 0.80; SE/Enh, predicted super-enhancer or enhancer regions based on H3-K27ac ChIP-seq profiles in specific primary tissues: gastric (dark green), esophagus (brown) or stomach smooth muscle (light green) (Hsniz 2013). CER1: 9:96715651-96 716 450 (rs11789015/rs7872123); CER2: 19:18817324-18 818 396 (rs10423674) [http://genome.ucsc.edu].

Figure 4

Luciferase reporter enhancer activity assays. The CER at 19p13.11 (CER2) was assessed for allele-specific activity, alongside positive and negative control fragments. Three independent experiments were performed in triplicate in two EAC cell lines (OE19 and OE33).

Figure 5

CRISPR-Cas9 genome editing of putative enhancer on 19p13.11. DNA gel electrophoresis showing genome editing of CER2, containing SNP rs10423674, in OE19 and OE33 lines. Region targeted by CRISPR gRNAs is ~0.45 kb. PCR amplified region is ~2 kb. gRNA (−): cells transfected with cas9 vector and gRNA empty vector (−); gRNA (+): cells transfected with cas9 vector and guide RNA target vectors.

Figure 6

Gene expression changes following CRISPR-Cas9 deletion of putative enhancer on 19p13.11. CER2 was targeted for deletion in OE19 (left) and OE33 (right) cells using CRISPR-Cas9 technology. Pools of transfected cells were analyzed using TaqMan gene expression assays for CRTC1, COMP, CRLF1, TMEM59L, KLHL26 in triplicate, in three independent experiments. NC: mock transfected parental cells; Cas9 + gRNA: cells transfected with Cas9 vector and guide RNA target vectors. ^*P < 0.05; ^**P < 0.01 and ^***P < 0.001.

Figure 7

Regional 3D chromatin interactions with putative enhancer containing rs10423674. Promoter capture-Hi-C profiles from multiple primary tissues were queried for fragments spanning the candidate risk SNP. UCSC Genome Browser plot of significant interactions between such fragments and promoters of neighboring genes (±150 kb) [http://genome.ucsc.edu].