Functional Landscape of PCGF Proteins Reveals Both RING1A/B-Dependent-and RING1A/B-Independent-Specific Activities

Abstract

Polycomb repressive complexes 1 and 2 (PRC1 and PRC2) control cell identity by establishing facultative heterochromatin repressive domains at common sets of target genes. PRC1, which deposits H2Aub1 through the E3 ligases RING1A/B, forms six biochemically distinct subcomplexes depending on the assembled PCGF protein (PCGF1-PCGF6); however, it is yet unclear whether these subcomplexes have also specific activities. Here we show that PCGF1 and PCGF2 largely compensate for each other, while other PCGF proteins have high levels of specificity for distinct target genes. PCGF2 associates with transcription repression, whereas PCGF3 and PCGF6 associate with actively transcribed genes. Notably, PCGF3 and PCGF6 complexes can assemble and be recruited to several active sites independently of RING1A/B activity (therefore, of PRC1). For chromatin recruitment, the PCGF6 complex requires the combinatorial activities of its MGA-MAX and E2F6-DP1 subunits, while PCGF3 requires an interaction with the USF1 DNA binding transcription factor.

Keywords: EZH2; H2A ubiquitination; H3K27me3; MGA; MYC; PCGF; PRC1; Polycomb; RING1B; USF1.

Graphical abstract

Figure 1

PCGFs Show Specificity in Target Gene Occupancy (A) ChIP-qPCR analysis for the indicated PCGF proteins at selected target regions in wild-type (WT) and in indicated Pcgf KO mouse ESCs. IgG served as control for ChIP assay. ChIP enrichments are normalized to input. Data represent mean ± SEM. (B) Genomic snapshots of the indicated ChIP-seq profiles at selected gene loci performed as in (A). (C) ChIP-seq cumulative enrichment deposition centered at peak summit for the indicated PCGF proteins performed as in (A). (D) Percentage of co-occupancy of the target genes identified for each indicated PCGF protein with respect to the other datasets. For simplicity, just regions that represent 14% or more of the total PCGF targets are shown in the legend. (E) Genome-wide functional annotation of peaks generated from the indicated ChIP-seq analyses. Promoters are defined as the region around ±2.5 kb from mm9-annotated TSS, and the downstream regions as the first 3 kb after the TES. (F) ChIP-qPCR analysis for the indicated PCGF proteins at selected target regions in the indicate mESC lines. See also Figure S1 and Tables S2 and S3.

Figure 2

Specific PCGF Activities Define Activating and Repressive Modules (A) Heatmaps representing the normalized ChIP-seq intensities for the indicated PCGF proteins over ±4 kb around the TSS of the indicated loci stratified for PCGF co-occupancy in wild-type (WT) mESCs. H3K36me3 intensity was analyzed over the entire gene length (from TSS to TES). CBX7 and RYBP datasets from mESCs were obtained from Morey et al. (2013) and KDM2B from Farcas et al. (2012). (B) Pearson correlation of ChIP-seq signal over the promoter regions (±4 kb from TSS) of annotated RefSeq coding genes (mm9). (C) Average deposition profile of H3K36me3 in WT mESCs over the gene body (from TSS to TES) of PCGF unique bound promoters (left panel) or promoters co-occupied by at least two PCGF proteins (right panel), as indicated. (D) Boxplots showing the expression levels obtained from RNA-seq analyses performed in WT mESC for the indicated PCGF target genes. H3K27me3- and H3K4me3-positive loci served as controls for repressed and active promoters, respectively. (E) Upper panel: GAL4-TK-luciferase reporter system of 293TRex clones expressing inducible Gal4 (empty) or the indicated Gal4-PCGF fusion protein. Lower panel: Luciferase activity triggered by Gal4-fusion recruitment at GAL4-TK-Luciferase promoter is shown as the fold difference relative to the empty control. Luciferase activity was normalized to protein content. Data represent mean ± SEM. (F) Upper panel: GAL4-TK-luciferase reporter system of 293TRex clones expressing inducible Gal4 (empty) or the indicated Gal4-PCGF6 fusion protein. Lower panel: Luciferase assay (as in E) with PCGF6 N-terminally or C-terminally fused to the DNA binding domain of Gal4. The activity is shown as the fold difference relative to the empty control and was normalized to protein content. Data represent mean ± SEM. See also Figure S2 and Table S4.

Figure 3

PCGFs Are Specific with Little Compensatory Crosstalk (A) Genomic snapshots of the indicated ChIP-seq profiles at selected target gene loci, performed in WT and in the indicated Pcgf KO mESC clones. (B and C) Boxplots of the normalized intensity profiles of ChIP-seq analyses for PCGF1, PCGF2, PCGF3, or PCGF6 (B) and for RING1B, H2AK119ub1 (H2Aub1), SUZ12, or H3K27me3 (C), performed in WT mESCs, Pcgf1, Pcgf2/4, Pcgf3/5, or Pcgf6 KO ESC clones. Signal enrichment was calculated using a region ±4 kb at unique and co-occupied target genes, as indicated. See also Figures S3 and S4.

Figure 4

PCGF1 and PCGF2 Compensate H2Aub1 Deposition at Specific Targets and Are Dispensable for ESC Viability (A) Phalloidin immunofluorescence staining in wild-type, Pcgf1/2/4, and Pcgf3/5/6 triple KO mESC. Scale bars correspond to 30 μm. (B) Principal component analysis of gene expression levels from RNA-seq analysis performed in WT mESCs and in the indicated KOs. Dashed lines enclose experimental replicates. (C) Volcano plots of –log10 (p value) against log2 fold change representing the differences in gene expression between Pcgf1/2/4 and Pcgf3/5/6 KO mESC clones and WT for all protein coding genes (upper panels) and for PCGF1 and PCGF2 targets or PCGF3 and PCGF6 targets, respectively (bottom panels). (D) Boxplots of normalized ChIP-seq intensity profiles of RING1B, H2AK119ub1 (H2Aub1), SUZ12, and H3K27me3 performed for WT or Pcgf1, Pcgf2/4, and Pcgf1/2/4 KO mESC clones, over ±500 bp (or ±4 kb for H2Aub1 and H3K27me3) around the TSS of target genes unique for PCGF3 or PCGF6, or common to PCGF1/2 or PCGF1/2/6. (E) Genomic snapshots of the ChIP-seq profiles quantified in (D) at selected target gene loci (common or unique, as indicated), performed in WT and the indicated Pcgf KO mESC clones. (F) Boxplots of normalized ChIP-seq intensity profiles of RING1B, H2AK119ub1 (H2Aub1), SUZ12, and H3K27me3 performed in WT or Pcgf6, Pcgf3/5, and Pcgf3/5/6 KO mESC clones over ±500 bp (or ±4 kb for H2Aub1 and H3K27me3) around the TSS of unique or common target genes (as indicated). (G) Genomic snapshots of the ChIP-seq profiles quantified in (F) at selected common or unique target gene loci (as indicated), performed in WT and the indicated Pcgf KO mESC clones. See also Figures S5, S6, and S7 and Tables S5 and S6.

Figure 5

PCGF3 and PCGF6 Activities Are Independent of RING1A/B (A) Percentage of occupancy of the different PCGF proteins at RING1B-bound promoters. (B) Percentage of overlap of RING1B, SUZ12, CBX7, and RYBP at the indicated PCGF-bound promoters. (C) Upper panel: Genomic snapshots of PCGF1 ChIP-seq profiles at selected genomic regions performed in Ring1A^–/–;Ring1B^fl/fl (R1A KO-R1B FL) and Ring1A^–/–;Ring1B^–/– (R1A KO-R1B KO) mESCs. Bottom left: Heatmap showing the normalized signal of PCGF1 ChIP-seq in R1A KO-R1B FL and R1A KO-R1B KO mESC over ±2.5 Kb of PCGF1/RING1B and PCGF1/2/RING1B common target loci, as well as PCGF1 and PCGF1/2 unique target loci. Bottom right: Cumulative quantification of the heatmaps and PCGF1 ChIP-qPCR analysis at selected regions performed in the same mESCs. IgG served as ChIP negative control. ChIP enrichments were normalized to input. Data represent mean ± SEM. (D) As in (C), for PCGF2 ChIP-seq profiles analyzed at the indicated PCGF2-specific targets. (E) Upper panel: Genomic snapshots of PCGF3 ChIP-seq profiles at selected genomic regions performed in R1A KO-R1B FL and R1A KO-R1B KO mESC. Bottom left: Heatmap showing the normalized signal of PCGF3 ChIP-seq in R1A KO-R1B FL and R1A KO-R1B KO mESCs over ±2.5 Kb of PCGF3/RING1B common target loci and PCGF3 unique target loci. Bottom right: Cumulative quantification of the heatmaps and PCGF3 ChIP-qPCR analyses at selected regions performed in the same ESCs. IgG served as ChIP negative control. ChIP enrichments were normalized to input. Data represent mean ± SEM. (F) As in (E), for PCGF6 ChIP-seq profiles analyzed at the indicated PCGF6-specific targets. See also Figure S8.

Figure 6

PCGF6 Requires Cooperative E2F and E-Box Recognition for Target Recruitment (A) Western blot analyses using the indicated antibodies in PCGF6 immunoprecipitations from nuclear extracts prepared from the indicated mESC lines. Input served as loading control. (B) De novo motif discovery analysis performed underneath the summit of PCGF6 peaks. Sequence weight matrixes of predicted compared to match DNA binding motifs are shown together with p values. (C) PCGF6 ChIP-qPCR analyses on mESC expressing scrambled (sh Ctrl) or Mga-specific shRNAs at PCGF6 target and a negative region (neg). IgG served as ChIP negative control. ChIP enrichments are normalized to input. Data represent mean ± SEM. (D) Normalized intensity profiles and heatmap of PCGF6 binding in WT mESCs or Pcgf6 KO, a shMga, MgaΔHLH mutant, shE2f6, and a shE2f6 in MgaΔHLH mutant around ±2.5 kb of the TSS of common and unique target loci. (E) Boxplots of the normalized intensity profiles of ChIP-seq analyses for PCGF6 in WT or Pcgf6 KO mESCs, a shMga, MgaΔHLH mutant, shE2f6, and a shE2f6 in MgaΔHLH mutant over ±500 bp respective to the TSS of common and unique target loci. (F) Genomic snapshots of the PCGF6 ChIP-seq profiles at selected genomic regions in WT, shMga, and MgaΔHLH mESCs. (G) Western blot analyses using the indicated antibodies in PCGF6 immunoprecipitations from nuclear extracts prepared from WT or shMga, MgaΔHLH mESCs. IgG served as an unrelated control antibody. Input is shown as loading control. (H) PCGF6 ChIP-qPCR analysis on WT mESCs and MgaΔHLH mESCs expressing scrambled (Ctrl), Max, or E2F6-specific shRNAs at PCGF6 target and a negative region (unrel). IgG served as ChIP negative control. ChIP enrichments are normalized to input. Data represent mean ± SEM. See also Figure S9.

Figure 7

Interactions with USF1/2 Mediate PCGF3 Recruitment to Chromatin (A) De novo motif discovery analysis performed underneath the summit of PCGF3 peaks. Sequence weight matrixes of predicted compared to match DNA binding motifs are shown together with relative p values. (B) Normalized ChIP-seq intensity profile of USF1 binding over ±2.5 kb regions of PCGF3 target loci. (C) Genomic snapshots of PCGF3 and USF1 ChIP-seq profiles at PCGF3 targets in WT mESCs. (D) Relative (percentage) of the different PCGF target genes co-occupied by USF1. (E) ChIP-qPCR analysis for USF1 in WT mESCs at representative PCGF1, PCGF2, PCGF3, and PCGF6 targets. IgG served as a negative ChIP control. ChIP enrichments are normalized to input. Data represent mean ± SEM. (F) Wstern blot analyses using the indicated antibodies in PCGF3 immunoprecipitations from nuclear extracts prepared from WT mESCs. Rabbit IgG served as an unrelated control antibody. Inputs are shown as loading control. (G) Western blot analysis with the indicated antibodies of protein fractions from WT mESC nuclear extracts separated by size-exclusion chromatography. Molecular weights are indicated based on the elution profile of markers in the same conditions. (H) PCGF3 ChIP-qPCR analysis on mESCs expressing scrambled (shCtrl) or combined Usf1- and Usf2 (shUsf)-specific shRNAs. IgG served as negative ChIP control. ChIP enrichments are normalized to input. Data represent mean ± SEM. (I) Genomic snapshots of PCGF3 ChIP-seq profiles in scrambled (shCtrl) and combined Usf1- and Usf2 (shUsf)-specific shRNAs expressing mESCs (left panel) and its relative normalized ChIP-seq intensity profiles for two biological replicates (right panels) over a ±2.5 Kb region at PCGF3/RING1B common and PCGF3 unique target loci. See also Figure S10.