Dissecting the role of H3K27 acetylation and methylation in PRC2 mediated control of cellular identity

Abstract

The Polycomb repressive complexes PRC1 and PRC2 act non-redundantly at target genes to maintain transcriptional programs and ensure cellular identity. PRC2 methylates lysine 27 on histone H3 (H3K27me), while PRC1 mono-ubiquitinates histone H2A at lysine 119 (H2Aub1). Here we present engineered mouse embryonic stem cells (ESCs) targeting the PRC2 subunits EZH1 and EZH2 to discriminate between contributions of distinct H3K27 methylation states and the presence of PRC2/1 at chromatin. We generate catalytically inactive EZH2 mutant ESCs, demonstrating that H3K27 methylation, but not recruitment to the chromatin, is essential for proper ESC differentiation. We further show that EZH1 activity is sufficient to maintain repression of Polycomb targets by depositing H3K27me2/3 and preserving PRC1 recruitment. This occurs in the presence of altered H3K27me1 deposition at actively transcribed genes and by a diffused hyperacetylation of chromatin that compromises ESC developmental potential. Overall, this work provides insights for the contribution of diffuse chromatin invasion by acetyltransferases in PRC2-dependent loss of developmental control.

Fig. 1

PRC2 enzymatic activity is not required for its association to chromatin. a Schematic representation of the CRISPR/Cas9 strategies used to generate knock-out (KO) and knock-in (KI) mESCs. Scissors indicate the position of the sgRNAs used for Cas9 targeting. b Western blot analysis using the indicated antibodies with protein extracts obtained from WT, Ezh1 KO, Ezh2 KO, Ezh1/2 dKO, and Ezh1 KO-Ezh2 Y726D mouse ESC lines. Vinculin and histone H3 were used as loading controls. c Immunoprecipitations using EZH2 antibody with total protein extracts from the indicated cell lines. Western blots were performed with the indicated antibodies. d Heatmaps representing spike-in normalized H3K27me3 and normalized SUZ12 ChIP-seq intensities ± 5 kb around the transcription start sites (TSS) of the Polycomb-bound promoters in the indicated cell lines. Promoters were ranked according to their ChIP-seq intensities in WT mESCs. Enrichment plots representing the average distribution of H3K27me3 and SUZ12 ± 5 kb around TSS are shown in the upper panels. Target regions (N = 3968) were selected considering SUZ12 and RING1B peaks with P ≥ 1 × 10^–1 in wild-type mESCs. e Overlap between the H3K27me3- and SUZ12-bound promoters in the indicated cell lines. f Representative genomic snapshots for H3K27me3 and SUZ12 ChIP-seq analyses performed in the indicated cell lines. Regions of H3K27me3 spreading outside the boundaries of SUZ12 chromatin association are highlighted

Fig. 2

PRC2-EZH1 is unable to spread H3K27me3 deposition but preserves target genes repression. a Illustration describing the criteria used to select SUZ12 inside peak regions, its boundaries, and the regions selected for the analysis at the 5′ and 3′ outside SUZ12 peak boundaries. b Boxplots representing H3K27me3 density distribution in WT and Ezh2 KO cells within PRC2 peaks, as well as 1.2 kb outside the 5′ and 3′ ends. P-values were determined using a Student’s t-test. c Left panels, boxplots representing the distribution of the H3K27me3 density ratio between the H3K27me3 density inside PRC2 peaks (PEAK in a) and at 5′ (left panel) or 3′ (right panels) spreading regions (5′ or 3′ spread in a). P-values were determined using a Student’s t-test. Right panels, the correlation for the IN/OUT H3K27me3 density ratio between WT and Ezh2 KO cells. The linear correlation coefficient (r) and coefficient of determination (r²) are shown in the graphs. d Left panels, boxplots representing the distribution of the H3K27me3 density ratio between the H3K27me3 density inside PRC2 peaks (PEAK in a) and at 5′ (left panel) or 3′ (right panels) spreading regions (5′ or 3′ spread in a) considering the top 20% H3K27me3 enriched promoters in Ezh2 KO cells based on P-values. P-values were determined using a Student’s t-test. Right panels, the correlation for the IN/OUT H3K27me3 density ratio between WT and Ezh2 KO cells. The linear correlation coefficient (r) and coefficient of determination (r²) are shown in the graphs. e Volcano plots of differentially expressed genes in the indicated cell lines compared to WT. Log2FC ≥ 1

Fig. 3

De novo PRC2 recruitment is independent of H3K27 methylation. a The experimental strategy used for de novo recruitment experiments. b Western blot analyses with the indicated antibodies using protein extracts obtained from WT, Ezh1/2 dKO, and Ezh1/2 dKOs cells transiently transfected with vectors expressing human EZH2 or EZH2Y731D (Y726D in mouse). Empty vector was used as negative control of transfection. Vinculin, histone H3, and histone H2A served as loading controls. c Heatmaps representing the spike-in normalized H3K27me3 ChIP-seq intensities and the normalized SUZ12 ChIP-seq signals ± 5 kb from TSS of SUZ12-bound promoters in the cell lines presented in a. Promoters were ranked according to SUZ12 intensities in WT mESCs. Upper panels, enrichment plots representing the average distribution of H3K27me3 and SUZ12 ± 5 kb around TSS. d Genomic snapshots for H3K27me3 and SUZ12 ChIP-seq analyses performed in the indicated cell lines at the Wnt5a locus

Fig. 4

PRC2 recruitment is not affected by acute loss of H2Aub1 deposition. a Western blot analysis with the indicated antibodies of total protein extracts obtained from Ezh1 KO-Ezh2 Y726D cells upon treatment with DMSO (vehicle) or 10 µM MG132 for 6 h. TP53 served as positive control for MG132 treatment. Vinculin and histone H2A were used as loading controls. b Heatmaps representing the spike-in normalized H2AK119ub ChIP-seq intensities and the normalized SUZ12 ChIP-seq signals ± 5 kb around TSS of SUZ12-bound promoters/targets in Ezh1 KO-Ezh2 Y726D cells upon treatment with DMSO or MG132. Promoters were ranked according to their intensities in WT mESCs. Enrichment plots representing the average distribution of H2AK119ub and SUZ12 ± 5 kb around TSS are shown in the upper panels. c Genomic snapshot of H2AK119ub1 and SUZ12 of the analysis performed in b at the Wnt5a locus. d Experimental strategy used for de novo recruitment performed in Rosa26:Cre-ERT2, Ring1a KO, Ring1b fl/fl, Ezh1 KO, Ezh2 KO ESC transduced with a doxycycline inducible lentiviral vector expressing EZH2. e Western blot analysis with the indicated antibodies of total protein extracts obtained from cells described in d, treated with EtOH (vehicle) or OHT (0.5 μM) alone or in combination with doxycycline (1 μg/ml). Vinculin was used as loading control. f qPCR of SUZ12 ChIP in the same cells used in e, upon treatment with EtOH or OHT (0.5 μM) alone or in combination with doxycycline (1 μg/ml). Rabbit IgG served as negative control. Enrichments are normalized to % INPUT. Data are represented as mean ± SEM

Fig. 5

EZH1-mediated accumulation of H3K27me2 at promoters sustains maintenance of PRC1 repressive domains. a Heatmaps representing the spike-in normalized H3K36me3, H3K27me1, H3K27me2, and H3K27me3 ChIP-seq intensities within the entire normalized gene length plus a ± 1 kb region around all annotated RefSeq genes (N = 31027). Regions were clustered into Polycomb targets, H3K36me3 positive and negative regions. PcG targets were ranked for the intensity of H3K27me3 deposition while the rest of RefSeq genes for ascending intensity of H3K36me3 in WT cells. Upper boxplots represent the average normalized intensities of H3K36me3, H3K27me1, H3K27me2, and H3K27me3 in the indicated cell lines in the different gene clusters. b Representative genomic snapshots of RING1B ChIP-seq analyses performed in the indicated cell lines. c Heatmaps representing the normalized RING1B and the spike-in normalized H3K27me1, H3K27me2, and H3K27me3 ChIP-seq intensities ± 5 kb around TSS of Polycomb-bound promoters in the indicated cell lines. Promoters were ranked according to their intensities in WT mESCs. Enrichment plots representing the average distribution of RING1B, H3K27me1, H3K27me2, and H3K27me3 ± 5 kb around TSS are shown in the upper panels

Fig. 6

H3K27me2/3 deposition, but not PRC2 occupancy, prevents H3K27ac invasion at PcG repressed promoters. a Heatmaps representing the normalized SUZ12 and the spike-in normalized H3K27ac ChIP-seq signals ± 5 kb around all TSS annotated in RefSeq. Promoters were clustered into Polycomb targets, H3K36me3 positive and H3K36me3 negative. PcG targets were ranked for the intensity of H3K27me3 deposition while the rest of RefSeq genes for ascending intensity of H3K36me3. b Boxplots representing the normalized intensities of SUZ12 and the spike-in normalized intensities of H3K27me3 and H3K27ac in the indicated cell lines at Polycomb-bound promoters (upper panels) and at Polycomb not-bound promoters (bottom panels). P-values were determined by Wilcoxon test

Fig. 7

Diffused accumulation of H3K27ac directly contributes to PRC2-mediated differentiation defects. a Pictures of embryoid bodies at day 7 (d7) of differentiation, from the indicated ESC lines (shown at ×4 magnification). b Western blot analysis using the indicated antibodies with protein extracts obtained from WT, Ezh2 KO, Ezh1/2 dKO, Ezh1 KO-Ezh2 Y726D, and Ezh1 KO-Ezh2 R685C mouse ESC lines. Histone H3 was used as loading control. c ChIP-qPCR of H3K27ac in the indicated mESC lines. IgG rabbit served as negative control. Enrichments are normalized to % INPUT. Data are represented as mean ± SEM. d Heatmaps of qPCR relative expression analyses for the indicated genes at d0, d5, d7, and d9 during ESC differentiation of the cells indicated shown in a. Gapdh expression served as a normalization control