Fine-mapping and molecular characterisation of primary sclerosing cholangitis genetic risk loci

Abstract

Genome-wide association studies of primary sclerosing cholangitis have identified 23 susceptibility loci. The majority of these loci reside in non-coding regions of the genome and are thought to exert their effect by perturbing the regulation of nearby genes. Here, we aim to identify these genes to improve the biological understanding of primary sclerosing cholangitis, and nominate potential drug targets. We first build an eQTL map for six primary sclerosing cholangitis-relevant T-cell subsets obtained from the peripheral blood of primary sclerosing cholangitis and ulcerative colitis patients. These maps identify 10,459 unique eGenes, 87% of which are shared across all six primary sclerosing cholangitis T-cell types. We then search for colocalisations between primary sclerosing cholangitis loci and eQTLs and undertake Bayesian fine-mapping to identify disease-causing variants. In this work, colocalisation analyses nominate likely primary sclerosing cholangitis effector genes and biological mechanisms at five non-coding (UBASH3A, PRKD2, ETS2 and AP003774.1/CCDC88B) and one coding (SH2B3) primary sclerosing cholangitis loci. Through fine-mapping we identify likely causal variants for a third of all primary sclerosing cholangitis-associated loci, including two to single variant resolution.

Fig. 1. Summary of fine-mapping.

Summary of the fine-mapping results across 15 PSC risk loci.

Fig. 2. Colocalisation results for ETS2 Chr21:40466744.

ETS2 Chr21:40466744 regional association plot for (a) PSC GWAS data (N_Cases= 4,796; N_Ctr = 19,955), (b) colocalising (PP4$\ge 0.8$) eQTL data for ETS2 in monocytes (N = 194) and (c) colocalising (PP4$\ge 0.8$) H3K27ac histQTL data in monocytes (N = 174). The most-likely causal variant (rs4817987 PP = 0.87) is shown in purple. LD information is calculated from PSC GWAS data.

Fig. 3. Colocalisation results for PRKD2 Chr19:47205707.

PRKD2 Chr19:47205707 regional association plots for (a) PSC GWAS data (N_Cases = 4796; N_Ctr = 19,955) and (b) colocalising (PP4$\ge 0.8$)) eQTL data for PRKD2 in monocytes (N = 194). The most-likely causal variant (rs313829, PP = 0.94) is shown in purple. LD information is calculated from PSC GWAS data.

Fig. 4. Colocalisation results for UBASH3A Chr21:43855067.

UBASH3A Chr21:43855067 regional association plots for (a) PSC GWAS data (N_Cases = 4796; N_Ctr = 19,955) and (b) colocalising (PP4$\ge 0.8$) eQTL data for UBASH3A in CD4 + T-cells (N = 171) and (c) colocalising (PP4$\ge 0.8$) spliceQTL data for UBASH3A in CD4 + T-cells (N = 171). The most likely causal variant (rs1893592 PP4 = 0.99), is shown in purple. LD information is calculated from PSC GWAS data.

Fig. 5. Study flow diagram.

Study flow diagram for generation and collection of eQTL datasets.