Maternal One‐Carbon Biomarkers and the Infant Methylome: The Role of Seasonality Using a Case Study from The Gambia

Abstract

IntroductionThe periconceptional period represents a window in which environmentally‐induced epigenetic modifications could have significant consequences for offspring health. Metastable Epialleles (MEs) are genomic loci whose methylation state varies between individuals, but where variation is correlated across tissues originating from all three germ layers in a single individual, and is independent of genotype. They provide a powerful device for studying the influence of the periconceptional environment on the offspring epigenome. Here we assess whether the profile of periconceptional maternal nutritional biomarkers predicting infant ME methylation differs according to the contrasting nutritional backgrounds afforded by Gambian rainy and dry seasons.MethodsWe measured 1‐carbon biomarkers in maternal plasma at mean 8.7 weeks gestation, back‐extrapolated to conception, and CpG methylation at 50 ME loci in their infants at mean age 3.3 months (N=120 mother‐child pairs). We tested for interactions between seasonality and biomarker concentrations on mean ME methylation z‐score. We also analysed biomarker principal components (PCs) in an attempt to capture co‐variation.ResultsPlasma folate was not associated with methylation in the dry season, but in the rainy season a one SD increase was associated with a decrease in mean methylation of 0.14 SD (95% CI: −0.26, −0.01; p=0.023; p‐value for interaction = 0.019). Plasma choline was also not associated with methylation in the dry season, however, in the rainy season a one SD increase was associated with an increase in mean methylation of 0.16 SD (95% CI: 0.02, 0.3; p=0.023; p‐value for interaction = 0.030). Conversely, homocysteine decreased total mean methylation in the dry season by 0.13 SD (95% CI: −0.25, −0.01; p=0.039) but showed no association in the rainy season (p‐value for interaction=0.030). PC1 captured co‐variation in the folate metabolism cycle and predicted dry season methylation. PC2 represented the betaine metabolism pathway and predicted rainy season methylation. Overall we found individual biomarkers showed little individual effect on methylation, but joint effects captured by the PC analysis generally resulted in stronger associations. When compared to predictive models combining the seasons the models stratified by season fitted better and the retained coefficients were stronger, with preliminary evidence suggesting an interaction between season of conception and 1‐carbon biomarkers with methylation (fig. 1).ConclusionFuture 1‐carbon related nutritional interventions may have different impacts on DNA methylation according to the nutritional background against which they are implemented.Support or Funding InformationMRC Population & Systems Medicine Board Grant