Interactive effects of genotype with prenatal stress on DNA methylation at birth

Intrauterine stress exposure is associated with offspring health. DNA methylation (DNAm) is a putative underlying mechanism, but large population-based studies reported limited associations between prenatal stress and DNAm. Recent research has shown that environmental factors in interaction with genetic variants are better predictors of DNAm than environment or genotype alone. We investigated whether interactions of maternal prenatal stress with genetic variants are associated with DNAm at birth. We examined 2963 mother-child pairs from the population-based Generation R Study and Avon Longitudinal Study of Parents and Children, using a harmonized, comprehensive cumulative prenatal stress measure. We tested genome-wide genotype-by-prenatal stress interactions on epigenome-wide DNAm (GxEmodel), and models including only genetic variants (Gmodel) or prenatal stress (Emodel) as predictors. Follow-up analyses included Gene Ontology analyses and mediation analyses of prenatal alcohol intake, smoking, gestational age, and birth weight. We report two independent gene-by-prenatal-stress interactions on DNAm after multiple testing correction, including five genetic variants in CHD2 and ORC5, and two DNAm sites in EPPK1. By comparison, the Gmodel showed 691,202 associations and the Emodel showed three associations in genes AHRR, GFI1, and MYO1G, which could largely be explained by prenatal smoking. Genes linked to suggestive GxEmodel results were often involved in neuronal development. Our results provide some support of interaction of prenatal stress with the child's genome on DNAm of genes related to neuronal development. Based on these models, genetic main effects on DNA methylation at birth were much more abundant than gene-by-prenatal stress interactions were.