Maternal and neonatal one-carbon metabolites and the epigenome-wide infant response

Abstract

Maternal prenatal status, as encapsulated by that to which a mother is exposed through diet and environment, is a key determinant of offspring health and disease. Alterations in DNA methylation (DNAm) may be a mechanism through which suboptimal prenatal conditions confer disease risk later in life. One-carbon metabolism (OCM) is critical to both fetal development and in supplying methyl donors needed for DNAm. Plasma concentrations of one-carbon metabolites across maternal first trimester (M1), maternal term (M3), and infant cord blood (CB) at birth were tested for association with DNAm patterns in CB from the Michigan Mother and Infant Pairs (MMIP) pregnancy cohort. The Illumina Infinium MethylationEPIC BeadChip was used to quantitatively evaluate DNAm across the epigenome. Global and single-site DNAm and metabolite models were adjusted for infant sex, estimated cell type proportions, and batch as covariates. Change in mean metabolite concentration across pregnancy (M1 to M3) was significantly different for S-adenosylhomocysteine (SAH), S-adenosylmethionine (SAM), betaine, and choline. Both M1 SAH and CB SAH were significantly associated with the global distribution of DNAm in CB, with indications of a shift toward less methylation. M3 SAH and CB SAH also displayed significant associations with locus-specific DNAm in infant CB (FDR<0.05). Our findings underscore the role of maternal one-carbon metabolites in shifting the global DNAm pattern in CB and emphasizes the need to closely evaluate how dietary status influences cellular methylation potential and ultimately offspring health.

Keywords: DNA methylation; Metabolomics; One-carbon metabolism; Prenatal exposure.

Figure 1.

One-carbon metabolite levels across time points. Peak intensities averaged within M1, M3, and CB are displayed (circles) and lines represent the 95% confidence interval for each mean.

Figure 2.

Network representing Pearson correlations between metabolites across all three sample types. Correlations between two metabolites that are significant at p<0.05 are displayed. Individual metabolites in a given sample type are represented by each node. Edges connecting two nodes indicate a significant correlation between the two (at p-value <0.05). Thicker edges within the network correspond to more highly significant correlations.