Identification of methylation and hydroxymethylation haplotype blocks as distal regulatory elements aids in deconvolution of heterogeneous brain tissues

Abstract

5-Hydroxymethylcytosine (5hmC) is an oxidation product of 5-methylcytosine (5mC), and adjacent CpG sites in the mammalian genome can be co-methylated and co-hydroxymethylated due to the processivity of DNMT and TET enzymes. We applied TAB-seq and oxBS-seq to selectively detect 5hmC and 5mC, respectively, at base resolution in the mouse cortex, olfactory bulb and cerebellum tissues. We found that majority of the called 5hmC CpG sites frequently had 5mC modification simultaneously and enriched in gene body regions of neuron development related genes in brain tissues. These results supported a prominent role of oxidizing 5mC to 5hmC as new epigenetic mark. Strikingly, by a systematic search of regions that show highly coordinated methylation and hydroxymethylation (MHBs and hMHBs), we found that MHBs significantly overlapped with hMHBs in gene body regions which further supported that oxidized 5mC to 5hmC co-ordinately in a subset of cells within heterogeneous brain tissues. Consistently, using a metric called methylation haplotype load, we defined a subset of 1,361 tissue-specific MHBs and 3,818 shared MHBs which were predominantly regulatory elements, and aids in deconvolution of heterogeneous brain tissues. Our results provide new insights into the role of co-ordinately oxidized 5mC to 5hmC as distal regulatory elements may involve in regulating tissue identity.