Integrative analysis of 5-methyl- and 5-hydroxymethylcytosine indicates a role for 5-hydroxymethylcytosine as a repressive epigenetic mark

Since the discovery of 5-hydroxymethylcytosine (5hmC) as a prominent DNA base modification found in mammalian genomes, an emergent question has been what role this mark plays in gene regulation. 5hmC is hypothesized to function as an intermediate in the demethylation of 5-methylcytosine (5mC) and also in reactivation of silenced regulatory elements, including promoters and enhancers. Further, weak positive correlations have been observed between gene body 5hmC and gene expression. We previously demonstrated that ME-Class, which uses a high-resolution model of whole-genome bisulfite sequencing data, is an effective tool to understand relationships between 5mC and expression. In this work, we present ME-Class2, a machine-learning based tool to perform integrative 5mCG, 5hmCG and expression analysis. Using ME-Class2 we analyze whole-genome single-base resolution 5mC and 5hmC datasets from 20 primary tissue and cell samples to uncover relationships between 5hmC and expression. The addition of 5hmC improves model performance for tissues with high-levels of 5hmC such as the brain. Our analysis further indicates that conversion of 5mC to 5hmC within 2kb of the transcription start site associates with distinct functions depending on the summed level of 5mC + 5hmC. Unchanged levels of 5mC + 5hmC (i.e. conversion from 5mC to stable 5hmC) associate with repression. Meanwhile, decreases in 5mC + 5hmC (i.e. 5hmC-mediated demethylation) associate with gene activation. As more large-scale, genome-wide, differential DNA methylation studies become available, tools such as ME-Class2 will prove invaluable to interpret epigenomic data and guide mechanistic studies into the function of 5hmC.