Mendelian inheritance of trimodal CpG methylation sites suggests distal cis-acting genetic effects

Environmentally influenced phenotypes, such as obesity and insulin resistance, can be transmitted over multiple generations. Epigenetic modifications, such as methylation of DNA cytosine-guanine (CpG) pairs, may be carriers of inherited information. At the population level, the methylation state of such "heritable" CpG sites is expected to follow a trimodal distribution, and their mode of inheritance should be Mendelian.

Our findings suggest that CpG methylation is maintained over larger genomic distances. Furthermore, nearly half of the SNPs associated with these trimodal sites were also associated with the expression of nearby genes (P = 4.08 × 10-6), implying a regulatory effect of these trimodal CpG sites.

Using the Illumina Infinium 450 K DNA methylation array, we determined DNA CpG-methylation in blood cells from a family cohort 123 individuals of Arab ethnicity, including 18 elementary father-mother-child trios, we asked whether Mendelian inheritance of CpG methylation is observed, and most importantly, whether it is independent of any genetic signals. Using 40× whole genome sequencing, we therefore excluded all CpG sites with possibly confounding genetic variants (SNP) within the binding regions of the Illumina probes.

We identified a total of 955 CpG sites that displayed a trimodal distribution and confirmed trimodality in a study of 1805 unrelated Caucasians. Of 955 CpG sites, 99.9% observed a strict Mendelian pattern of inheritance and had no SNP within +/-110 nucleotides of the CpG site by design. However, in 97% of these cases a distal cis-acting SNP within a +/-1 Mbp window was found that explained the observed CpG distribution, excluding the hypothesis of epigenetic inheritance for these clear-cut trimodal sites. Using power analysis, we showed that in 46% of all cases, the closest CpG-associated SNP was located more than 1000 bp from the CpG site. Conclusions: Our findings suggest that CpG methylation is maintained over larger genomic distances. Furthermore, nearly half of the SNPs associated with these trimodal sites were also associated with the expression of nearby genes (P = 4.08 × 10-6), implying a regulatory effect of these trimodal CpG sites.