DNA co-methylation modules in postmortem prefrontal cortex tissues of European Australians with alcohol use disorders

Abstract

DNA methylome alterations in the prefrontal cortex (PFC) may contribute to risk for alcohol use disorders (AUDs). We examined postmortem PFC DNA methylomes of 16 male and seven female pairs of AUD and control subjects using Illumina's HumanMethylation450 BeadChip assays. In male AUD subjects, 1,812 CpGs (1,099 genes) were differentially methylated (9.5 × 10(-9) ≤ Pnominal ≤ 7.2 × 10(-4), q < 0.05). In females, no CpGs were associated with AUDs after multiple testing correction (q > 0.05). Twenty-one AUD-associated co-methylation modules were identified in males by co-methylation analysis. The 1,812 CpGs were over-presented by two AUD-associated co-methylation modules (Mturquoise: 1,048 CpGs/683 genes; Mblue: 429 CpGs/304 genes) (Phyper ≤ 0.001). Biological processes enriched for genes in these two modules included neural development and transcriptional regulation. Genes mapped by CpGs in these two modules were enriched in genome-wide association study-identified genes with variants associated with four substance dependence phenotypes or five psychiatric disorders. Additionally, 106 of the 1,812 CpGs were mapped to 93 genes (e.g., AUD-associated genes GRIK3, GRIN2C, and GABRA1) with differential expression in postmortem PFC of male AUD subjects. Our study demonstrates that DNA methylation alterations in the PFC are associated with (and might result in) increased risk of AUDs, and there was a complex DNA methylation-gene expression relationship.

Figure 1. Differentially methylated CpGs in male subjects with alcohol use disorders (AUDs).

(a) Volcano plot of effect size [log₂(fold changes)] against −log₁₀(P values) of 434, 015 CpGs. The red dots represent 1,812 CpGs with q ≤ 0.05 (the horizontal green dash line) and the absolute value of log₂(fold change) >0.03 (two vertical green dash lines), and the black dots represent non-significant CpGs. (b) Kernel density plotting of log₂(fold changes) of 1,812 significant CpGs. The asymmetric plot indicates that a greater proportion (66.3%) of CpGs were hypermethylated in AUD subjects. (c). Hierarchical clustering of the 32 male subjects using a heatmap based on methylation levels of the above 1,812 CpGs (adjusted for age and PMI). The 32 male subjects were clustered into two distinct groups that were consistent with their actual AUD status (the red color: 16 male AUD patients; the blue color: 16 male healthy control subjects). The colors in the heatmap indicate CpG methylation levels (green to red: low to high methylation levels).

Figure 2. Alcohol use disorder (AUD)-associated CpG methylation modules.

(a) Enrichment of the top 1,812 CpGs (identified in male AUD subjects) in each of the 22 modules (including the grey module). The ratio was calculated as the observed number of the top CpGs over the expected number of the top CpGs. (b,c). Correlation of the gene significance (GS) and the module membership (MM) in two large modules (the turquoise module and the blue module).

Figure 3. Pairs of hypo- or hypermethylated CpGs and up- or downregulated genes in postmortem PFC of male AUD subjects.

Vertical and horizontal dash lines indicate significantly changed CpG methylation and gene expression, respectively.