The early-life exposome modulates the effect of polymorphic inversions on DNA methylation

Abstract

Polymorphic genomic inversions are chromosomal variants with intrinsic variability that play important roles in evolution, environmental adaptation, and complex traits. We investigated the DNA methylation patterns of three common human inversions, at 8p23.1, 16p11.2, and 17q21.31 in 1,009 blood samples from children from the Human Early Life Exposome (HELIX) project and in 39 prenatal heart tissue samples. We found inversion-state specific methylation patterns within and nearby flanking each inversion region in both datasets. Additionally, numerous inversion-exposure interactions on methylation levels were identified from early-life exposome data comprising 64 exposures. For instance, children homozygous at inv-8p23.1 and higher meat intake were more susceptible to TDH hypermethylation (P = 3.8 × 10^-22); being the inversion, exposure, and gene known risk factors for adult obesity. Inv-8p23.1 associated hypermethylation of GATA4 was also detected across numerous exposures. Our data suggests that the pleiotropic influence of inversions during development and lifetime could be substantially mediated by allele-specific methylation patterns which can be modulated by the exposome.

Fig. 1. Inversion status as methylation quantitative trait loci (mQTL) of multiple CpG sites within and surrounding three common human inversions.

The first column in the plot panel corresponds to inv-8p23.1, the second to inv-16p11.2, and the third to inv-17q21.31. a–c Manhattan plots for the significance of the associations between the differential methylation of the CpG sites and the inversion genotypes in child blood cells (N = 1009). The x axes show the chromosome position (±1 Mb between the inversions’ breakpoints). The y axes show the –log₁₀ (P-value). The dashed red line indicates Bonferroni’s threshold of significance. Green points are CpG sites with significant associations and those in gray are nonsignificant. The orange block illustrates the inversions’ region. d–f Principal component (PC) analysis for methylation levels of CpG sites within and surrounding the inversions, revealing remarkably distinctive methylation patterns among the different inversion statuses. Blue points illustrate noninverted homozygous (N/N), yellow illustrates heterozygous (N/I), and orange illustrates inverted homozygous (I/I) individuals. In parenthesis, the methylation variance explained by each PC. g–i Manhattan plots of differentially methylated CpG sites, depending on the inversion genotypes in fetal heart DNA (N = 40).

Fig. 2. Inversion-exposure interactions as methylation quantitative trait loci (mQTL) of multiple CpG sites within and surrounding three common human inversions.

a Number of exposures per family in the early-life exposome from the HELIX project. b–d Manhattan plots showing the significance of the associations (N = 1009) between the differential methylation of the CpG sites and the inversion-exposure interactions across all 64 exposures in (a) and the genotypes of three human inversions at 8p.23.1 (b); 16p11.2 (c); and 17q21.31 (d), illustrated by the orange block. The x axes show the chromosome position (±1 Mb between the inversions’ breakpoints). The y axes show the –log₁₀ (P-value) of the associations. The dashed red line indicates Bonferroni’s threshold of significance. Significant results are colored according to the family exposure (a) and labeled according to the closest gene to the CpG (Illumina annotation). Gray points are not significant.

Fig. 3. Interaction and forest plots for TDH, GATA4, and TRMT9B genes.

a Interaction plot illustrating differences across inv-8p23.1 genotypes in the association between cg01489256 (TDH) methylation and meat intake (expressed in servings per week). Methylation means the given meat-intake status and inversion genotype are represented with their 95% confidence intervals (N = 1009). b Forest plot showing the meta-analysis effect estimates of inv-8p23.1–meat-intake interaction on cg01489256 methylation across HELIX cohorts. c Interaction plot illustrating differences across inv-8p23.1 genotypes in the association between cg26020513 (GATA4) methylation and manganese (N = 1009). d Forest plot showing the meta-analysis effect estimates of inv-8p23.1–manganese interaction on cg26020513 methylation across HELIX cohorts. e Interaction plot illustrating differences across inv-8p23.1 genotypes in the association between cg08196601 (TRMT9B) methylation and parental smoking (N = 1009). f Forest plot showing the meta-analysis effect estimates of the inv-8p23.1–parental smoking interaction on cg08196601 methylation across HELIX cohorts. Blue points and lines illustrate noninverted homozygous (N/N), yellow illustrates heterozygous (N/I), and orange illustrates inverted homozygous (I/I) individuals. The error bar represents one standard deviation.