Reversible Regulation of Promoter and Enhancer Histone Landscape by DNA Methylation in Mouse Embryonic Stem Cells

Abstract

Summary DNA methylation is one of a number of modes of epigenetic gene regulation. Here, we profile the DNA methylome, transcriptome, and global occupancy of histone modifications (H3K4me1, H3K4me3, H3K27me3, and H3K27ac) in a series of mouse embryonic stem cells (mESCs) with varying DNA methylation levels to study the effects of DNA methylation on deposition of histone modifications. We find that genome-wide DNA demethylation alters occupancy of histone modifications at both promoters and enhancers. This is reversed upon remethylation by Dnmt expression. DNA methylation promotes H3K27me3 deposition at bivalent promoters, while opposing H3K27me3 at silent promoters. DNA methylation also reversibly regulates H3K27ac and H3K27me3 at previously identified tissue-specific enhancers. These effects require DNMT catalytic activity. Collectively, our data show that DNA methylation is essential and instructive for deposition of specific histone modifications across regulatory regions, which together influences gene expression patterns in mESCs.

Isoform‐specific localization of DNMT3A regulates DNA methylation fidelity at bivalent CpG islands

“…By reinvestigating published datasets generated in Dnmt ‐TKO and Dnmt3a/3b ‐DKO cells (King et al , ), we indeed observe that the CpG island promoters associated with enriched DNMT3A1 binding display the strongest reduction in H3K27me3 (Fig EV3A–C). And, as previously reported, by re‐expressing DNMT3A1 in Dnmt3a/3b ‐DKO cells, H3K27me3 distribution at these promoters is more effectively restored than by DNMT3A2 or DNMT3B (King et al , ) (Fig EV3B and C). In addition to the H3K27me3 dynamics, we further detect increased DNaseI hypersensitivity and H3K27 acetylation in the absence of DNA methylation in Dnmt ‐TKO cells at the same DNMT3A1‐bound and unmethylated CpG island promoters (Fig EV3D and E), coinciding with frequent deregulation of the associated genes (Fig E).…”

“…We suggest that the balance between DNMT3A1 and TET activity is required to fine‐tune the distribution of methylated and unmethylated CpGs and for restricting Polycomb domain boundaries to the promoters of developmentally regulated genes. This is supported by the increased re‐setting of H3K27me3 in Dnmt3a/3b ‐DKO reconstituted with DNMT3A1 (King et al , ).…”

“…In the absence of DNA methylation, H3K27me3 has been reported to spread into neighboring regions, resulting in depletion of H3K27me3 from CpG island centers and influencing activity of Polycomb‐regulated genes (Fouse et al , ; Lynch et al , ; Brinkman et al , ; Marks et al , ; Reddington et al , ; King et al , ). By reinvestigating published datasets generated in Dnmt ‐TKO and Dnmt3a/3b ‐DKO cells (King et al , ), we indeed observe that the CpG island promoters associated with enriched DNMT3A1 binding display the strongest reduction in H3K27me3 (Fig EV3A–C). And, as previously reported, by re‐expressing DNMT3A1 in Dnmt3a/3b ‐DKO cells, H3K27me3 distribution at these promoters is more effectively restored than by DNMT3A2 or DNMT3B (King et al , ) (Fig EV3B and C).…”

“…Notches display the confidence interval around the median. Scatter plots showing correlated increase in DHS or H3K27ac with loss of H3K27me3 at UMR promoters in Dnmt ‐TKO cells, suggesting that reduction in H3K27me3 leads to increased promoter accessibility and activity. Red dots denote the UMR promoters within cluster 1 bound by DNMT3A1. Data information: Datasets used in (A–C) were obtained from King et al (). Datasets used in (D and E) were obtained from Domcke et al ().…”

Shoring up DNA methylation and H3K27me3 domain demarcation at developmental genes

“…The enzymatic activity of DNMT3A1 could be important for regulating H3K27me3 deposition at bivalent CGIs (Williams et al , ). In Dnmt3a/3b KO or Dnmt‐triple‐KO ES cells, H3K27me3 at bivalent CGIs is reduced, but re‐expression of DNMT3A1 in these KO ES cell lines preferentially re‐targets DNA methylation to bivalent CGI shores and can promote re‐acquisition of H3K27me3 (King et al , ). These results strongly suggest that DNMT3A1 is responsible for sustaining methylcytosine at bivalent CGIs shores, which links with H3K27me3 domain demarcation and potentially subsequent regulation of Polycomb target genes (Fig ).…”