Background: With the comprehensive characterization of 14,287 primary genetic association with 2,923 plasma proteins in 54,219 UKBB participants, the recently published UKBB pQTL dataset provides a rich resource to elucidate the biological mechanisms underlying proteo-genomic discoveries and accelerate the development of biomarkers for complex diseases such as IBD. Herein we utilized Mendelian Randomization (MR), a powerful tool for causal inference based on GWAS summary statistics, to identify the proteins contribute to IBD.
Methods: European pQTL GWAS summary statistics of 2,923 proteins were downloaded from the UKBB website and fine-mapping was performed using SusieR. For each protein, index SNPs from fine-mapping were extracted for each credible set and Instrumental Variables (IVs) for MR were selected based on a threshold of F-statistics>10. Phenoscanner was further utilized to exclude the pleiotropic variants in the IVs and only proteins with >2 IVs were included in the MR analysis. IBD GWAS summary statistics from European population were obtained from de Lange et al (NG 2017). Data curation and MR analysis was performed using the package TwoSampleMR and penalized MR-Egger approach was used for MR inference.
Results: 1,237 proteins with >2 IVs were included in current analysis, with a significance threshold of 4.04E-5 after Bonferroni’s Correction. 28 proteins in CD, 16 in UC and 18 in IBD achieved statistical significance.
Among the top signals are well known IBD related proteins such as MST1 (In CD: beta = 0.19, P=1.89E-23; in UC: beta=0.15, P= 5.82E-11; in IBD: beta= 0.16, P= 7.10E-18) , IL18R1 (in CD: beta = -0.15, P = 2.46E-11), and IL12B (in UC: beta = -0.29, P=9.70E-14).
In addition, MR analysis also unveiled novel targets such as ITGB6 (Integrin alpha V beta 6, in CD, beta = -5.33, P = 2.32E-6), FGF19 (Fibroblast growth factor 19, in CD, beta = 0.78, P = 2.72E-10), MUC2 (Mucin 2, in CD, beta = -0.18, P = 1.71E-7), AGER (advanced glycosylation end product receptor, in UC, beta =-0.87 , P=8.06E-12), CD274 (an immune inhibitory receptor ligand, in UC, beta = 0.32,P = 1.41E-6) , HYAL1 (Hyaluronidase 1 which degrade hyaluronan, one of the major glycosaminoglycans of the extracellular matrix, in IBD, beta = 0.30, P=1.74E-10) and CSNK1D (Casein Kinase I Isoform Delta, in IBD, beta = -2.28, P = 1.61E-8).
Conclusion: Utilizing the UKBB pQTL dataset, we validated causal effect of known proteins with IBD. Novel causal proteins were also indicated which might lead to new biomarkers and potential therapeutic targets. Further analyses, including colocalization analysis, network analysis and integration with eQTL datasets and additional IBD associated loci are in progress.