Introduction
DNA methylation (DNAm), an epigenetic modification to cytosine, is involved in mediating the developmental regulation of gene expression and function, as well as transcriptional processes such as genomic imprinting and X chromosome inactivation.1, 2 Although often regarded as a mechanism of transcriptional repression, the relationship between DNAm and gene expression is highly complex and not fully understood.3 Gene-body DNAm, for example, is often associated with active expression4 and also influences other transcriptional processes, including alternative splicing and promoter usage.5 This dynamic property of DNAm means it can vary across samples and might underlie phenotypic differences. There is growing interest in characterizing the variation of DNAm across populations6, 7 and in the role of DNAm in disease, and recent epigenome-wide association studies (EWASs) have identified robust associations between variable DNAm and cancer,8 as well as a diverse range of other complex phenotypes, including rheumatoid arthritis [MIM: 180300],9 body-mass index,10 schizophrenia [MIM: 181500],11 and Alzheimer disease [MIM: 104300].12 Characterizing the complex relationship between genetic, epigenetic, and transcriptomic variation will increase understanding about the mechanisms underpinning health and disease phenotypes. Twin and family studies have demonstrated that population-level variation in DNAm is under considerable genetic control, although these effects vary across genomic loci, developmental stages, and different cell and tissue types.13, 14, 15, 16, 17 Studies in a variety of tissues, including brain, whole blood, pancreatic islet cells, and adipose tissue, have identified widespread associations between common DNA sequence variants and DNAm.17, 18, 19, 20, 21, 22 These DNAm quantitative trait loci (mQTLs) are primarily cis-acting, are enriched in regulatory chromatin domains and transcription-factor binding sites, and have been shown to colocalize with gene expression quantitative trait loci (eQTLs).3, 17, 23
There is considerable interest in using mQTLs, along with other types of molecular QTLs, to interpret the functional consequences of common genetic variation associated with human traits, especially because the actual gene(s) involved in mediating phenotypic variation are not necessarily the most proximal to the lead SNPs identified in genome-wide association studies (GWASs). Of note, GWAS variants are enriched in enhancers and regions of open chromatin,24, 25 reinforcing the hypothesis that most common genetic risk factors influence gene regulation rather than directly affecting the coding sequences of transcribed proteins.26 Importantly, evidence for the co-localization of genetic variants associated with both phenotypic and regulatory variation is not sufficient to show that the overlapping association signals are causally related; additional analytical steps are needed to distinguish pleiotropic effects—i.e., where the same variant is influencing both outcomes, although not necessarily dependently—from those that are an artifact of linkage disequilibrium (LD). We recently extended the use of one approach—summary-data-based Mendelian randomization (SMR), which was initially used in conjunction with expression quantitative trait loci (eQTL) data27—to prioritize genes for GWAS-nominated loci using mQTL data.28
Building on our previous work, we used the Illumina EPIC array and imputed SNP data to identify mQTLs associated with variable DNAm at ∼850,000 sites across the genome in samples from the Understanding Society UK Household Longitudinal Study (UKHLS) (n = 1,111). We then used these mQTLs within the SMR framework to refine genetic association data from publicly available GWAS datasets in order to prioritize genes involved in 63 complex traits and diseases. We subsequently used the SMR approach to identify pleiotropic relationships between DNAm and variable gene expression by using publicly available whole-blood gene eQTL data. Our mQTL database and SMR results are available via a searchable online database as a resource to the research community (see Web Resources).