Whole genome analysis of the methylome and hydroxymethylome in normal and malignant lung and liver

Abstract

DNA methylation at the 5-postion of cytosine (5mC) is a well-established epigenetic modification which regulates gene expression and cellular plasticity in development and disease. The ten-eleven translocation (TET) gene family is able to oxidize 5mC to 5-hydroxymethyl-cytosine (5hmC), providing an active mechanism for DNA demethylation, and may also provide its own regulatory function. Here we applied oxidative bisulfite sequencing to generate whole-genome DNA methylation and hydroxymethylation maps at single-base resolution in paired human liver and lung normal and cancer. We found that 5hmC is significantly enriched in CpG island (CGI) shores while depleted in CGIs themselves, in particular at active genes, resulting in a 5hmC but not 5mC bimodal distribution around CGI corresponding to H3K4me1 marks. Hydroxymethylation on promoters, gene bodies, and transcription termination regions showed strong positive correlation with gene expression within and across tissues, suggesting that 5hmC is a mark of active genes and could play a role gene expression mediated by DNA demethylation. Comparative analysis of methylomes and hydroxymethylomes revealed that 5hmC is significantly enriched in both tissue specific DMRs (t-DMRs) and cancer specific DMRs (c-DMRs), and 5hmC is negatively correlated with methylation changes, particularly in non-CGI associated DMRs. Together these findings indicate that changes in 5mC as well as in 5hmC and coupled to H3K4me1 correspond to differential gene expression in tissues and matching tumors, revealing an intricate gene expression regulation through interplay of methylome, hydroxyl-methylome, and histone modifications.