An atlas of genetic effects on the monocyte methylome across European and African populations
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Genetic regulation of DNA methylation in immune cells may mediate complex disease risk. However, current epigenomic studies are constrained by microarray CpG coverage, mixed-cell tissues, and limited representation of diverse ancestries. Thus, we generated a whole-genome, multi-ancestry atlas of genetic effects on the purified monocyte methylome.
Methods
We first performed whole-genome bisulfite sequencing (WGBS) of purified peripheral blood monocytes and whole-genome sequencing (WGS) from 160 African American (AA) and 298 European American (EA) participants, profiling around 25 million CpG sites. Next, we identified cis -methylation quantitative trait loci (meQTLs), estimated cis heritability, and evaluated replication against large external meQTL resources. We further trained population-specific DNAm imputation models and applied them to methylome-wide association studies (MWAS) of 41 traits using genome-wide association study summary statistics from the Million Veteran Program. Type 2 diabetes signals were further evaluated using Mendelian randomization and Bayesian colocalization. We also conducted exploratory trans-meQTL mapping.
Results
We identified 1,480,064 and 1,527,480 CpG sites with at least one cis-meQTL in AA and EA populations, respectively, including 543,869 shared sites and extensive population-specific regulation attributable to both allele-frequency differences and effect-size heterogeneity. Cis-meQTL effects replicated robustly in external datasets: effect sizes correlated strongly with prior studies (EA Pearson’s r = 0.76; 90.8% concordant directions; AA Pearson’s r = 0.71; 86.6% concordant directions). We built DNAm prediction models with cis -h 2 > 0.01 for 2,677,714 CpG sites in AA and 1,976,046 CpG sites in EA, achieving mean cross-validated prediction R 2 of 0.20 and 0.18. Across 41 traits, MWAS 23,650 significant methylation-phenotype associations (2,116 in AA and 21,534 in EA), of which ∼98% were not interrogated by Illumina 450K/EPIC arrays. For type 2 diabetes, MWAS identified 20 CpG sites in AA and 4,023 CpG sites in EA, with substantial support from Mendelian randomization and colocalization. Exploratory trans-meQTL mapping detected widespread long-range associations, with limited cross-study overlap but high directional concordance among shared signals.
Conclusions
This whole-genome, monocyte-resolved, multi-ancestry methylome atlas and accompanying imputation resource expand interpretable methylation variation beyond array-based studies and enable multi-ancestry integration of genetic, epigenetic, and genome-wide association study data to prioritize immune-cell regulatory mechanisms for complex disease.
