Characterizing the genetic architecture of DNA methylation: using molecular QTLs to refine genetic association signals for complex traits

Eilis Hannon (1), Tyler Gorrie-Stone(2), Melissa Smart(3), Joe Burrage(1), Amanda Hughes(3), Yanchun Bao(3), Meena Kumari (3), Leonard Schalkwyk (2), Jonathan Mill (1,*)

Multi-omics studies that explore the interplay of genetics with gene regulation and expression seek to enhance existing studies interesting in identifying the specific disease-associated genes. There has been much interest in the role of one particular regulatory mark DNA methylation (DNAm) in health and disease, however, most studies to date have ignored the role of genetic, or even transcriptomic, variation limiting the potential interpretation. In this study, we describe the most comprehensive analysis of common genetic variation on DNAm to date, using Illumina EPIC array data profiled in the UK Household Longitudinal study to identify DNAm quantitative trait loci (mQTLs). We demonstrate the utility of these data to refine GWAS signals for 63 complex traits and investigate the relationship between DNAm and gene expression. Applying Summary data-based Mendelian Randomization (SMR) we report 1,662 pleiotropic associations between 36 complex traits and 1,246 DNAm sites. DNAm sites associated with complex traits overlapped both regulatory features such as DNAse I hypersensitivity sites, and genes more than expected by chance, with supporting evidence from gene expression quantitative trait loci for a number of the gene-trait associations. Finally, we used SMR to characterize the relationship between DNAm and gene expression identifying 6,798 pleiotropic associations between 5,420 DNAm sites and 1,702 genes. While most genes were associated with one or more DNAm methylation site, not all DNAm sites were associated with gene expression, supporting the notion that the relationship between these domains is not as simple as often claimed.