Understanding the interplay between CpG island-associated gene promoters and H3K4 methylation

Abstract

The precise regulation of gene transcription is required to establish and maintain cell type-specific gene expression programs during multicellular development. In addition to transcription factors, chromatin, and its chemical modification, play a central role in regulating gene expression. In vertebrates, DNA is pervasively methylated at CG dinucleotides, a modification that is repressive to transcription. However, approximately 70% of vertebrate gene promoters are associated with DNA elements called CpG islands (CGIs) that are refractory to DNA methylation. CGIs integrate the activity of a range of chromatin-regulating factors that can post-translationally modify histones and modulate gene expression. This is exemplified by the trimethylation of histone H3 at lysine 4 (H3K4me3), which is enriched at CGI-associated gene promoters and correlates with transcriptional activity. Through studying H3K4me3 at CGIs it has become clear that CGIs shape the distribution of H3K4me3 and, in turn, H3K4me3 influences the chromatin landscape at CGIs. Here we will discuss our understanding of the emerging relationship between CGIs, H3K4me3, and gene expression.

Keywords: Chromatin; CpG islands; DNA methylation; H3K4me3; Transcription.

Fig. 1

Histone H3 lysine 4 methylation and the H3K4 methyltransferase complexes. (A) and (B) Schematic depictions of the distribution of H3K4 methylation across (A) yeast and (B) vertebrate genes. Arrows indicate transcription start sites. Gene bodies are shown with black boxes representing exons. (C) The subunit composition of mammalian H3K4 histone methyltransferases. The H3K4 HMTs are split into three groups based on the homology of their catalytic subunit: SET1A/B, MLL1/2 and MLL3/4. Shared subunits are depicted in the same colour and position.

Fig. 2

CpG islands and H3K4me3. Heatmaps illustrating non-methylated DNA (BioCAP) [47], RNAPII (transcription) [267], H3K4me3, CFP1, SET1A [61] and MLL2 [25] ChIP-seq signal across all CpG islands in mouse embryonic stem cells. The heatmaps are ranked based on RNAPII signal. H3K4me3 occurs broadly across CpG islands genome-wide. MLL2 associates with the majority of CGIs, whereas CFP1 and SET1A are enriched at CpG islands that are bound by RNAPII.

Fig. 3

Mechanisms that regulate H3K4me3 at CpG islands. (A) H3K4me3 HMT complexes can deposit H3K4me3 across all CpG islands by binding directly to non-methylated CpG dinucleotides via their ZF-CxxC domains. The position of the transcription start site is illustrated as an arrow and a legend for the chromatin modifications is shown at the bottom of this figure. (B) Mechanisms to enrich H3K4me3 at actively transcribed gene promoters. These include interaction of HMT complexes with transcription factors (TFs) or interactions with the transcription machinery during initiation and elongation. (C) A mechanism to amplify and maintain H3K4me3 once initiated. CFP1 stabilises SET1A/B complexes at actively transcribed CGI-associated gene promoters through multivalent interactions with non-methylated CpG dinucleotides and pre-existing H3K4me3. (D) A metaplot illustrating the distribution of H3K4me3 and H2BK120ub1 ChIP-seq signal at CGI associated gene promoters in mouse embryonic fibroblasts [105]. H3K4me3 is distributed around the TSS and over the CGI, whereas H2BK120ub1 peaks downstream of the TSS and is also enriched throughout the gene body. The schematic below indicates the position of the TSS as an arrow, CGI promoter as a green box and gene body as a black box. (E) Mechanisms through which demethylation by KDM5 proteins could shape H3K4me3. These include dynamic turnover of H3K4me3, focussing of H3K4me3 at gene promoters, and removal of spuriously deposited H3K4me3 in gene bodies.

Fig. 4

The relationship between H3K4me3 and transcriptionally permissive chromatin. (A) H3K4me3 has been proposed to counteract methylation of CpG islands by preventing both binding and activation of the de novo DNA methyltransferases (DNMTs). The position of the transcription start site is illustrated as an arrow in each panel and a legend for the chromatin modifications is shown below the panels. (B) H3K4me3 counteracts acquisition of repressive chromatin modifications by preventing binding of histone methyltransferases (HMTs) while promoting removal of repressive methylation by stabilising binding of demethylases. In addition to inhibiting its catalytic activity, H3K4me3 may also inhibit PRC2 by preventing binding of NuRD and hence indirectly reduce PRC2 association with chromatin. (C) Histone acetyltransferases (HATs) and chromatin remodellers bind to H3K4me3 and contribute to an accessible and transcriptionally permissive chromatin architecture at CGI-associated gene promoters. (D) The TAF3 subunit of TFIID binds H3K4me3 and promotes pre-initiation complex formation, providing a direct link to transcription. This interaction is enhanced by histone acetylation.

Fig. 5

Do CpG island-associated gene promoters and chromatin bistability shape transitions in gene expression? (A) Both the Polycomb system (PRC1 and PRC2) and H3K4 histone methyltransferases (HMTs) can engage with or ‘sample’ CGIs via ZF-CxxC domain-containing proteins. Self-reinforcing positive feedback loops and mutual antagonism inherent to these systems could create bistable chromatin states at CGIs. (B) A schematic illustrating the proposed bistable chromatin state at CpG island-associated gene promoters. The transcriptionally repressive Polycomb state is sustained by feedback mechanisms that also antagonise the permissive state (A). We envisage that this Polycomb dependent repressive state constrains transcription until gene activation signals reach a threshold where inhibition is overcome and transcription initiates effectively. This would then result in a transcription-dependent switch to a permissive chromatin state, characterised by H3K4me3 and transcriptional activity (green arrow). This transcriptionally permissive state is maintained through self-reinforcing feedback mechanisms that also antagonise the Polycomb state (A). The permissive state is thereby maintained unless the activatory signal drops below a certain level where feedback can no longer be sustained and transcription does not persist. This would then cause a switch back to the repressive Polycomb chromatin state (red arrow). (C) The interaction between regulatory signals and bistable chromatin states (A,B) at CGI-associated gene promoters could shape gene expression transitions such that graded analogue gene regulatory inputs are translated into switch-like digital gene expression outputs.