The Mll2 branch of the COMPASS family regulates bivalent promoters in mouse embryonic stem cells

Abstract

Promoters of many developmentally regulated genes, in the embryonic stem cell state, have a bivalent mark of H3K27me3 and H3K4me3, proposed to confer precise temporal activation upon differentiation. Although Polycomb repressive complex 2 is known to implement H3K27 trimethylation, the COMPASS family member responsible for H3K4me3 at bivalently marked promoters was previously unknown. Here, we identify Mll2 (KMT2b) as the enzyme catalyzing H3K4 trimethylation at bivalentlymarked promoters in embryonic stem cells. Although H3K4me3 at bivalent genes is proposed to prime future activation, we detected no substantial defect in rapid transcriptional induction after retinoic acid treatment in Mll2-depleted cells. Our identification of the Mll2 complex as the COMPASS family member responsible for H3K4me3 marking bivalent promoters provides an opportunity to reevaluate and experimentally test models for the function of bivalency in the embryonic stem cell state and in differentiation.

Figure 1

Mll2 is required for the trimethylation of Histone H3 lysine 4 at bivalent homeotic genes. (a) The COMPASS family of H3K4 methylases in Drosophila (3 members on the left) and mammals (6 members, on the right). Subunits common to all COMPASS family members from yeast to human are shown in green; complex-specific subunits are shown in blue, orange and purple. (b) Mll2 and Mll3 mRNA levels after RNAi-mediated knockdown in mouse embryonic stem cells with shRNAs targeting GFP (shGFP), Mll3 (shMll3) or Mll2 (shMll2). Expression was determined by qRT-PCR and is shown relative to Actin (Actb). Results are shown as means and s.d. (n=2 technical replicates, representative of three biological replicate experiments). (c) ChIP-seq track file examples of H3K4me3 at mouse Homeobox (Hox) gene clusters. Red and black bars above the tracks indicate bivalent and non-bivalent regions, respectively. H3K27me3 data from Mikkelsen et al. is shown for comparison (purple). (d) ChIP-seq track examples of bivalent and non-bivalent chromatin in control and Mll2 shRNA-treated cells. Bivalently marked genes such as Prr18, Brachyury (T), Vgll4 and Syn2 are shown with red bars above the tracks as in (c).

Figure 2

Mll2 is required for the implementation of bivalency genome-wide. (a) The H3K4me3 occupancy change in the Mll2- and Mll3-depleted mouse embryonic stem cells for all H3K4me3 enriched genes. Left: ChIP-seq enrichment profiles for +/− 5kb around the TSS of all H3K4me3 enriched genes. Bivalent genes are shown as a separate group from the H3K4me3-only modified genes. Right: H3K4me3 occupancy log2 fold-change after depletion of Mll2 (shMLL2/shGFP) or Mll3 (shMLL3/shGFP) measured +/− 5kb around TSS. (b) Percentage of bivalent and non-bivalent genes with H3K4me3 occupancy loss. The percentages are shown at four levels: total H3K4me3-enriched genes, genes with more than 50% decrease of H3K4me3, genes with more than 75% decrease, and genes with more than 87.5% decrease. Numbers shown are the total number of genes for each level. H3K27me3 data in a,b are from Wamstad et al. (c) Average-gene occupancy plots of H3K4me3 in wild-type (shGFP) (blue line), shMll2 (red line) and shMll3 (green line) knockdown embryonic stem cells. Top and bottom panels show non-bivalent and bivalent genes, respectively. Plots are averaged from both H3K4me3 ChIP-seq biological replicates. (d) Gene Ontology analysis of genes with H3K4me3 loss after Mll2 knockdown. Benjamini-corrected p-values are shown.

Figure 3

Mll2 depletion has little effect on transcriptional induction kinetics in response to retinoic acid. (a) Expression analysis of mouse V6.5 embryonic stem cells infected with GFP shRNA (shGFP) or Mll2 shRNA (shMll2) lentivirus and induced to differentiate with 2µM retinoic acid for 6 (RA 6) or 12 hours (RA 12). Heatmaps of expression and H3K4me3 occupancy are shown. Genes with expression increased by 6 and 12 hours of RA treatment are sorted by the wild-type occupancy of H3K4me3 and separated as bivalent or not bivalent. Left: Expression heatmap in control and shMLL2 treated cells. Middle: log2 fold changes for expression after Mll2 RNAi. Induction defects would appear as an increase in green in the 6 and 12 hour timepoints. Right: log2 fold change of H3K4me3 occupancy. (b) Scatter plots of log2 fold changes in expression after RA induction at 6 hours (top panel) or 12 hours (bottom panel). Each dot represents a gene that was induced by 6 hours. Correlation coefficients are calculated in the log2 scale. (c) Gene expression track examples before and after RA induction. (d) Induction kinetics at one and three hours RA treatment. Results are shown as means and s.d., (n=2 technical replicates, representative of 3 biological replicate experiments).