Childhood abuse is associated with methylation of multiple loci in adult DNA

Abstract

Background: Childhood abuse is associated with increased adult disease risk, suggesting that processes acting over the long-term, such as epigenetic regulation of gene activity, may be involved. DNA methylation is a critical mechanism in epigenetic regulation. We aimed to establish whether childhood abuse was associated with adult DNA methylation profiles.

Methods: In 40 males from the 1958 British Birth Cohort we compared genome-wide promoter DNA methylation in blood taken at 45y for those with, versus those without, childhood abuse (n = 12 vs 28). We analysed the promoter methylation of over 20,000 genes and 489 microRNAs, using MeDIP (methylated DNA immunoprecipitation) in triplicate.

Results: We found 997 differentially methylated gene promoters (311 hypermethylated and 686 hypomethylated) in association with childhood abuse and these promoters were enriched for genes involved in key cell signaling pathways related to transcriptional regulation and development. Using bisulfite-pyrosequencing, abuse-associated methylation (MeDIP) at the metalloproteinase gene, PM20D1, was validated and then replicated in an additional 27 males. Abuse-associated methylation was observed in 39 microRNAs; in 6 of these, the hypermethylated state was consistent with the hypomethylation of their downstream gene targets. Although distributed across the genome, the differentially methylated promoters associated with child abuse clustered in genome regions of at least one megabase. The observations for child abuse showed little overlap with methylation patterns associated with socioeconomic position.

Conclusions: Our observed genome-wide methylation profiles in adult DNA associated with childhood abuse justify the further exploration of epigenetic regulation as a mediating mechanism for long-term health outcomes.

Figure 1

Promoter methylation associated with childhood abuse. Heatmap showing MeDIP probe values from the 997 differentially methylated promoters (rows) across all 40 participants (columns). Each promoter is represented by the probe most associated with childhood abuse. Blackened squares above the columns denote non-abuse males, white squares denote those with childhood abuse. Other covariates included are childhood and adulthood socio-economic position (white = low, gray = high). Neither appears to explain the main sample clusters.

Figure 2

Validation of MeDIP results. A. Quantification of methylation differences in the abuse and non-abuse groups by bisulfite pyrosequencing analysis of the SLC17A3 promoter and the PM20D1 first exon and intron. DNA methylation at 14 CpG sites in the SLC17A3 promoter and 12 and 1 CpG sites in the PM20D1 first exon and first intron, respectively, among the abuse and non-abuse groups is shown (N = 10 vs. 26 for SLC17A3; N = 9 vs 23 for PM20D1). One-sided t-tests were applied to each CpG site to test for association of methylation levels with childhood abuse, and false discovery rates were calculated for the resulting p-values in order to correct for multiple testing. All false discovery rates (FDR) were less than 0.1, indicating significant association between CpG methylation levels and childhood abuse. **: FDR < 0.025; *: FDR < 0.05; ++: FDR < 0.1; +: FDR < 0.2. The bars represent average methylation for all subjects in a group and error bars indicate the standard error of the mean. Physical maps of the regions analyzed are presented above the charts where CpG positions are indicated by balloons. The transcription start site (TSS) is indicated by a hook arrow. The positions of the primers used for pyrosequencing (Additional file 3: Table S2) are indicated by arrows. B. Replication of the quantification of the differences in methylation at PM20D1 between the abuse and non-abuse groups in an additional 27 males that were not profiled using MeDIP (N = 7 vs. 20). Pyrosequencing was applied to measure the methylation levels of 13 CpG sites in the first exon and intron of PM20D1.

Figure 3

Differential methylation in the WNT signaling pathway. The KEGG (http://www.genome.jp/kegg/mapper.html) depiction of the WNT signaling pathway is shown with hypermethylated gene promoters (more methylated in the group with childhood abuse) colored red and the hypomethylated gene promoters colored blue.

Figure 4

Megabase regions enriched with methylation differences. Promoter MeDIP differences across chromosomes 16 (A), 17 (B) and X (C) and across two smaller genomic regions (D) and (E) are shown using images obtained from the UCSC Genome Browser. The top track depicts average differences of log abuse – log non-abuse. Each bar in the middle track identifies a significant difference. Bars above or below the horizontal line identify sites with higher or lower methylation in the abused group. The bottom track indicates relative gene abundance across the chromosome.