Polycomb group (PcG) proteins play an important role in the control of developmental gene expression in higher organisms. In mammalian systems, PcG proteins participate in the control of pluripotency, cell fate, cell cycle regulation, X chromosome inactivation and parental imprinting. In this study we have analysed the function of the mouse PcG protein polycomblike 2 (Pcl2), one of three homologues of the Drosophila Polycomblike (Pcl) protein. We show that Pcl2 is expressed at high levels during early embryogenesis and in embryonic stem (ES) cells. At the biochemical level, Pcl2 interacts with core components of the histone H3K27 methyltransferase complex Polycomb repressive complex 2 (PRC2), to form a distinct substoichiometric biochemical complex, Pcl2-PRC2. Functional analysis using RNAi knockdown demonstrates that Pcl2-PRC2 facilitates both PRC2 recruitment to the inactive X chromosome in differentiating XX ES cells and PRC2 recruitment to target genes in undifferentiated ES cells. The role of Pcl2 in PRC2 targeting in ES cells is critically dependent on a conserved PHD finger domain, suggesting that Pcl2 might function through the recognition of a specific chromatin configuration.
In higher organisms the execution and maintenance of cell fate decisions in early development depend critically on epigenetic mechanisms that confer heritable on/off states at specific target loci. The Polycomb group (PcG) proteins play a central role in this, establishing and maintaining gene repression at target loci predominantly via post-translational modification of the core histones. Initially identified as mediators of cellular memory at Hox loci in Drosophila (Lewis, 1978), PcG proteins have been found to be highly conserved and to contribute to developmental gene regulation, the cell cycle, the maintenance of pluripotency and self-renewal capability in embryonic and adult stem cells and to epigenetic silencing on the inactive X chromosome (Xi) and at parentally imprinted loci (for reviews, see Sparmann and van Lohuizen, 2006; Schuettengruber and Cavalli, 2009; Simon and Kingston, 2009).
There are two major multimeric PcG protein complexes that have been widely studied: Polycomb repressive complex (PRC) 1 and 2. The PRC2 complex catalyses histone H3K27 methylation (Cao et al., 2002; Czermin et al., 2002; Muller et al., 2002; Kuzmichev et al., 2002) and is generally thought to be early acting, providing a binding site for subsequent recruitment of PRC1. PRC1 functions as an E3 ligase that specifically monoubiquitylates histone H2A (de Napoles et al., 2004; Wang, H. et al., 2004; Cao et al., 2005; Elderkin et al., 2007). H2A ubiquitylation is important for PcG-mediated silencing (Stock et al., 2007; Nakagawa et al., 2008), although there is also evidence that other direct and/or indirect mechanisms contribute to PRC1 function (Francis et al., 2001; Gambetta et al., 2009).
Mechanisms involved in the targeting of PcG complexes to specific loci remain poorly understood. In Drosophila, the recruitment to Hox genes and probably other target loci is mediated by cis-acting regions that are essential for silencing: the Polycomb response elements (PREs) (Simon et al., 1993; Chan et al., 1994). PREs have not been well defined in mammalian cells, although a recent report described a single example associated with the MafB locus (Sing et al., 2009). In the case of X inactivation, recruitment of PcG proteins is dependent upon the expression of non-coding RNAs (Plath et al., 2003; Silva et al., 2003; de Napoles et al., 2004; Kohlmaier et al., 2004; Plath et al., 2004), and this might also be the case at some imprinted loci (Umlauf et al., 2004; Nagano et al., 2008).
The PRC2 complex comprises three unique core protein components – the histone methyltransferase Ezh2, Eed and Suz12 – and the generic histone-binding proteins RbAp46/48 (also known as Rbbp7/4) (Cao et al., 2002; Czermin et al., 2002; Muller et al., 2002; Kuzmichev et al., 2002). The core PRC2 proteins do not bind DNA, suggesting that co-factors might be important in targeting the complex to specific loci. In this regard, candidate proteins associated with PRC2 have been identified in genetic and biochemical screens. The Jarid2 protein was recently shown to interact with PRC2 in mouse embryonic stem (ES) cells and has been suggested to play a role in PRC2 targeting (Peng et al., 2009; Shen et al., 2009; Li et al., 2010; Pasini et al., 2010) and/or in establishing the poised state at PcG target loci (Landeira et al., 2010). AEBP2, a zinc-finger protein, co-purifies with PRC2 in Hela cells (Pasini et al., 2010) and ES cells (Peng et al., 2009; Shen et al., 2009; Li et al., 2010; Landeira et al., 2010) but its function is as yet undetermined. Finally, the Polycomblike (Pcl) protein associates with PRC2 in Drosophila (O'Connell et al., 2001; Tie et al., 2003; Papp and Muller, 2006) and in mammalian cells (Cao et al., 2008; Sarma et al., 2008) and has been proposed to have a role in stimulating H3K27me3 activity and/or targeting of the complex (Nekrasov et al., 2007; Cao et al., 2008; Sarma et al., 2008).
In this study we have analysed the function of the mouse Pcl2 (Mtf2 – Mouse Genome Informatics) protein, one of three homologues of Pcl found in mammalian cells. We find that Pcl2 is expressed at high levels during early embryogenesis and in ES cells. Pcl2 interacts with the core PRC2 complex to form a stable and distinct biochemical complex, Pcl2-PRC2. Functional analysis using RNAi knockdown demonstrates that Pcl2-PRC2 is important in PRC2 recruitment to the Xi in differentiating XX ES cells and also for PRC2 recruitment to target genes in undifferentiated ES cells. A conserved PHD finger domain in Pcl2 is required for PRC2 targeting in ES cells.
MATERIALS AND METHODS
Cell lines and embryos
Growth and maintenance of trophoblast stem (TS) and ES cell lines were as described previously (Penny et al., 1996; Mak et al., 2002). ES cell differentiation was achieved by embryoid body formation via removal of LIF from the medium. The MP fibroblast cell line was derived in house from a male PGK mouse.
Preimplantation and postimplantation mouse embryos were obtained from timed mating of C57B16×CBA F1 animals as described previously (Sheardown et al., 1997).
Immunofluorescence (IF) and RNA fluorescent in situ hybridisation (FISH) analyses
Preparation of cells and of embryos for IF, RNA FISH and Immuno-RNA FISH were as previously described (Sheardown et al., 1997; Mak et al., 2002; Silva et al., 2003; de Napoles et al., 2004). IF, RNA FISH and Immuno-RNA FISH were carried out as described previously (Mak et al., 2002; Silva et al., 2003; de Napoles et al., 2004). When using Pcl2 antibody for IF, cells were pre-extracted prior to fixation by treatment with 0.4% Triton X-100 (1 minute, 4°C). Primary antibodies were diluted in 5% normal goat serum in 0.2% fish gelatin: H3K27me3, Ezh2 and Suz12 (all Millipore) 1:500; Eed (Sewalt et al., 1998) 1:100. Pcl2 mouse monoclonal antibody was from H.K. and is described in a separate publication (Li et al., 2011). Pcl2 antibody incubation was carried out in CanGetSignal solution (Toyobo, Japan; 1:10). Images were acquired on a Leica DMRB fluorescence microscope using a CCD camera or, alternatively, on a Leica TCS SP5 confocal microscope using LAS AF software. For Immuno-RNA-FISH of ES cell colonies, images were acquired on a Deltavision real-time (RT) microscope (Applied Precision; 63× objective with a numerical aperture of 1.4). Images were deconvolved with Softworx, and ImageJ was used for further image processing. Maximum intensity projections are shown.
Generation of stable Pcl2-FLAG cell lines
Pcl2 full-length cDNA was amplified from PGK12.1 ES cell mRNA using oligonucleotides 5′-GGATCCACCATGAGAGACTCTACAGGAGCA-3′ and 5′-GGTACCACCTCCGGATGCAGTCGCTCCTTCCCA-3′. cDNA was cloned N-terminally to the FLAG tag into the 5′ BamHI and 3′ KpnI sites of the pCBA-2×FLAG vector (van den Berg et al., 2008). pCBA-Pcl2-2×FLAG construct was transfected into PGK12.1 ES cells using Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions and stable transformants were selected with geneticin (400 μg/ml, Gibco) for 10 days. Individual colonies were picked, expanded and analysed by western blotting for stable expression of Pcl2-2×FLAG.
Pcl2 deletion constructs were generated by ‘PCR sewing’ using the above construct as a template and the oligonucleotides listed in Table S2 in the supplementary material to induce deletion of specific domains by fusion of two regions of the cDNA. Cloning of PCR products in pCBA-2×FLAG, transfection and clone selection were as described above.
Generation of stable Pcl2 knockdown cell lines
For each short hairpin RNA (shRNA) construct, four complementary oligonucleotides with overhangs and loop sequence were synthesised, annealed and cloned into the BglII/HindIII sites of the pSuper.neo/gfp vector (OligoEngine). Hairpin constructs were transfected into PGK12.1 ES cells using Lipofectamine 2000. Stable transformants were selected with geneticin (400 μg/ml) for 4 days and by FACS sorting for GFP expression. Sorted cells were plated at clonal density and individual colonies were picked, expanded and Pcl2 levels were analysed by western blotting. shRNA target sequences are listed in Table S3 in the supplementary material.
Cell fractionation and western blot analysis
The preparation of nuclear extracts and cell fractionation were essentially as described previously (Elderkin et al., 2007; van den Berg et al., 2010). For western blotting, proteins were separated on SDS-PAGE gels and transferred (45 minutes, 15V) to PVDF membranes in 1× Transfer buffer (48 mM Tris, 39 mM glycine, 0.037% SDS, 20% methanol) using a Bio-Rad semi-dry blotting system. Enhanced chemiluminescence detection was performed as recommended by the manufacturer (GE Healthcare). The following antibodies and dilutions were used for western blotting: Eed mouse monoclonal (Sewalt et al., 1998), 1:200; Ezh2 rabbit polyclonal (Millipore), 1:1000; Suz12 rabbit polyclonal (Millipore), 1:1000; M2 FLAG (Sigma), 1:2000; Ring1B (Rnf2 – Mouse Genome Informatics) mouse monoclonal (de Napoles et al., 2004), 1:100; YY1 rabbit polyclonal (Santa Cruz), 1:200; HP1γ (Millipore), 1:1000; Rbpj (Millipore), 1:1000; lamin B goat polyclonal (Santa Cruz), 1:2000; H3K27me3 (Millipore), 1:1000; H2A, H4, H3K4me1, H3K4me3 (Millipore), 1:2000; and H3, H3K4me2, H3K27me1, H3K27me2 rabbit polyclonal (Millipore), 1:5000. To improve detection, Pcl2 antibody was diluted (1:100) in CanGetSignal solution according to the manufacturer's instructions. Acid-extracted histones were prepared as described previously (de Napoles et al., 2004).
Immunoprecipitation (IP) and mass spectrometry analysis
IP of Pcl2-FLAG from ES cell nuclear extracts using the mouse M2 FLAG antibody (Sigma) and mass spectrometry analysis were carried out as previously described (van den Berg et al., 2010).
Chromatin Immunoprecipitation (ChIP) and gene expression analysis
Cells were fixed in suspension in 1% formaldehyde for 10 minutes, quenched with 125 mM glycine for 5 minutes and lysed in 1% SDS, 10 mM EDTA pH 8.0, 50 mM Tris-HCl pH 8.1 and protease inhibitors for 10 minutes at 4°C. Chromatin was fragmented to 0.3-0.5 kb by sonication. One hundred and fifty micrograms of chromatin was used per IP. Chromatin was diluted 1:10 in 1% Triton X-100, 2 mM EDTA pH 8.0, 150 mM NaCl, 20 mM Tris-HCl pH 8.1, protease inhibitors and pre-cleared with 30 μl of protein A or G agarose beads (Millipore) for 90 minutes at 4°C. Chromatin was subjected to IP overnight at 4°C using 5 μl of the following antibodies: anti-Ezh2 (Diagenode), anti-Suz12 (Diagenode), anti-H3K27me3 (Millipore) and anti-H3K4me3 (Millipore). For all experiments, 2 μl of anti-mouse IgG (Sigma) was used to measure background levels and 2 μl of anti-histone H3 (Abcam) served as input control. Thirty microlitres of protein G or protein A agarose beads were used to pull down immunocomplexes for 2 hours at 4°C. DNA was reverse crosslinked, eluted in 1% SDS, 0.1 M NaHCO3 and precipitated with isopropanol. DNA was resuspended in 100 μl H2O and analysed by real-time PCR using SYBR Green PCR Master Mix (Bio-Rad) following the manufacturer's instructions, on a Chromo4 real-time PCR system (Bio-Rad). Enrichment was normalised to 10% of input DNA.
For FLAG ChIP, following fixation as above cells were washed once in 10 mM Hepes pH 7.5, 200 mM NaCl, 1 mM EDTA, 0.5 mM EGTA, protease inhibitors and lysed in 25 mM Tris pH 7.5, 150 mM NaCl, 5 mM EDTA, 1% Triton X-100, 0.1% SDS, 0.5% sodium deoxycholate, protease inhibitors. One hundred and fifty micrograms of chromatin was used for IP with 5 μl M2 FLAG antibody (Sigma) overnight at 4°C, and subsequently with 30 μl of protein A or G agarose beads for 2 hours at 4°C.
Each ChIP experiment was performed independently three times. Primers used are shown in Table S4 in the supplementary material. RT-PCR for PcG target loci was carried out using standard methods and primers are listed in Table S5 in the supplementary material.
Gel filtration analysis
For Pcl2-PRC2 complex analysis (Fig. 2D), a Superose 6 SMART gel filtration column (GE Healthcare) was pre-calibrated with the gel filtration standards thyroglobin (669 kDa), γ-globulin (158 kDa), ovalbumin (44 kDa), myoglobulin (17 kDa) and vitamin B12 (1.35 kDa) (Bio-Rad). Purified Pcl2-FLAG complex obtained by IP was loaded onto the column pre-equilibrated with the buffer used for IP. Fractions were collected at a flow rate of 40 μl/minute and analysed by western blotting.
For gel filtration of PRC2 complexes in nuclear extracts, 4 μg of high-salt nuclear extracts from ES cells were loaded on a Superose 6 SMART gel filtration column pre-equilibrated with low-salt buffer (10% glycerol, 40 mM Hepes pH 7.6, 150 mM KCl, 0.5 mM EDTA, 1 mM DTT). Fifty-two fractions of 250 μl were collected and analysed by western blotting. Prior to the assay, the column was pre-calibrated using the gel filtration standards Blue Dextran 2000 (2 MDa), thyroglobin, apoferritin (440 kDa), γ-globulin, conalbumin (75 kDa) and ovalbumin (GE Healthcare).
Transient enrichment of Pcl2 on the Xi at the onset of X inactivation
We set out to investigate whether mammalian Pcl proteins play a role in X chromosome inactivation, building on previous observations demonstrating that the Xi is an important PcG target (Mak et al., 2002; Silva et al., 2003; de Napoles et al., 2004). We first carried out indirect immunofluorescence (IF) experiments in XX (female) trophoblast stem (TS) cells using an antibody specific for mouse Pcl2 (Li et al., 2011). Pcl2 was found to be diffusely localised throughout much of the nucleoplasm and, in addition, was highly enriched in a single domain that colocalised with a concentrated signal for the core PRC2 protein Suz12, a defined marker of the Xi chromosome (Fig. 1A). No focal Pcl2 staining was observed in XY TS cells (data not shown). Thus, Pcl2 is concentrated on the Xi chromosome in a similar manner to core components of both PRC1 and PRC2 complexes.
TS cells provide an in vitro model for the imprinted form of X inactivation that occurs in extraembryonic lineages of mouse embryos. To test for an involvement of Pcl2 in the random form of X inactivation we analysed differentiating XX mouse ES cells. Previous studies have demonstrated that modification of Xi chromatin in this model system occurs in a progressive, stepwise process that is initiated by an accumulation of Xist RNA (reviewed by Heard, 2005). Recruitment of PRC2 and PRC1 complexes and accumulation of the associated histone modifications H3K27me3 and H2AK119ub1 occur early, immediately following the onset of Xist expression (Mak et al., 2002; Silva et al., 2003; de Napoles et al., 2004). The accumulation of PRC1 and PRC2 proteins on the Xi diminishes as differentiation proceeds, but low or moderate levels are presumably maintained as both H3K27me3 and H2AK119ub1 persist (Silva et al., 2003; de Napoles et al., 2004). To study nuclear localisation of Pcl2 in random X inactivation, XX ES cells were differentiated in vitro for 7 days, and samples corresponding to each day of differentiation were analysed by indirect IF using Pcl2- and Suz12-specific antibodies (Fig. 1B). Pcl2 staining closely mirrored that seen for Suz12. Xi foci were not detected in undifferentiated cells, consistent with both X chromosomes being active. However, strong Xi foci were seen in early-stage differentiated cells and the localised signal diminished progressively as differentiation proceeded (Fig. 1C). Pcl2 foci were undetectable in XX mouse embryo fibroblast (MEF) cell lines known to have an Xi (not shown), similar to results previously reported for core PRC2 proteins (Plath et al., 2003; Silva et al., 2003). Consistent with these observations, western blot analysis (Fig. 1D) and quantitative RT-PCR (see Fig. S1A in the supplementary material) demonstrated that Pcl2 is highly expressed in undifferentiated ES cells and during early stages of differentiation and that levels drop rapidly at later differentiation stages (Fig. 1D and see Fig. S1B in the supplementary material).
In a proportion of mitotic cells, Pcl2 staining was observed on a single condensed chromosome, presumably the Xi (Fig. 1E). This signal was localised in a characteristic banded pattern observed previously for Xist RNA (Duthie et al., 1999) and PcG core components or associated histone modifications (Mak et al., 2002; de Napoles et al., 2004). Together, these results demonstrate that Pcl2 is enriched on Xi and that Pcl2 recruitment coincides with early stages of X inactivation in vitro.
To assess the accumulation of Pcl2 on Xi in vivo during normal development, IF analysis was carried out on mouse embryos using antibodies to H3K27me3 to counterstain the Xi. In blastocysts (E3.5), in which X inactivation is paternally imprinted (Mak et al., 2004), Pcl2 Xi foci were detected in the majority of cells in approximately half of all embryos examined (presumptive XX versus XY embryos) (Fig. 1F). At earlier preimplantation stages we failed to detect Pcl2 Xi foci, similar to results obtained previously for PRC2 core proteins (Mak et al., 2004; Okamoto et al., 2004). To examine Pcl2 association with Xi in random X inactivation we carried out IF on dissociated embryonic cells from individual XX postimplantation embryos. Pcl2 Xi foci were detected in the majority of cells at E6.5 and E7.5 (Fig. 1G), but not in later-stage embryos. This again mirrors the results obtained for core PRC2 proteins, in which enrichment on Xi is only observed during a developmental window when levels of the proteins are high (Mak et al., 2002; Silva et al., 2003; de Napoles et al., 2004). In summary, Pcl2 is enriched on the Xi chromosome specifically during those developmental stages at which PRC2 proteins were shown to be recruited to the Xi.
Pcl2 in mouse ES cells forms a stable complex with core PRC2 proteins
Biochemical purification of the Drosophila Polycomblike protein and of the human Polycomblike homologue PHF1 (PCL1) demonstrated an interaction with the PRC2 complex (O'Connell et al., 2001; Nekrasov et al., 2007; Cao et al., 2008; Sarma et al., 2008). To determine whether Pcl2 associates with PRC2 components, we purified FLAG-tagged Pcl2 (Pcl2-FLAG) from ES cells and analysed the purified fractions by western blot and mass spectrometry. PGK12.1 ES cells stably expressing Pcl2-FLAG were characterised and we selected two clones, B2 and E1, that expressed full-length Pcl2-FLAG at levels similar to endogenous Pcl2 (Fig. 2A). Analysis of the distribution of Pcl2-FLAG in nuclear fractions revealed an identical distribution to that of endogenous Pcl2, being found predominantly in S3, the high-salt fraction (Fig. 2B).
Pcl2-FLAG protein was immunoprecipitated using anti-FLAG agarose beads from nuclear extracts of PGK12.1 Pcl2-FLAG cell lines and eluted with FLAG tri-peptide. Non-transfected PGK12.1 ES cells were used as a control. Western blot analysis revealed that Pcl2-FLAG co-immunoprecipitates with the core PRC2 components Ezh2, Suz12 and Eed isoform 3, the most abundant isoform in mouse ES cells (Fig. 2C). The chromatin protein YY1, which was used as a negative control, was not detected. Pcl2-FLAG did not significantly deplete the pool of PRC2 proteins, as indicated by retention of signal in the flow-through fractions (Fig. 2C, Ft), suggesting that Pcl2-PRC2 complexes represent a relatively small proportion of the cellular PRC2 complement.
Mass spectrometry analysis was carried out on Pcl2-FLAG immunoprecipitation (IP) fractions to identify proteins that co-purify. Fig. 2E and Table S1 in the supplementary material summarise the data obtained from three independent IP experiments. The PRC2 proteins Ezh2, Suz12 and Eed, which constitute the catalytic core of the complex, were detected in all three experiments, as was Pcl2. RbAp46 and RbAp48, which have previously been shown to co-purify with PRC2, were not detected consistently, although RbAp48 was found in one experiment. No other proteins gave a high mascot score (>500) in more than one experiment (see Table S1 in the supplementary material), suggesting that the Pcl2-PRC2 complex does not have additional components.
To further analyse Pcl2-PRC2 subunit composition, Pcl2-FLAG IP fractions were analysed by size exclusion chromatography on a Superose 6 column. Previous studies have estimated the size of the core PRC2 complex purified from Drosophila and mammalian cells to be ~550-600 kDa (Kuzmichev et al., 2002; Nekrasov et al., 2005; Sarma et al., 2008) and that Pcl-PRC2 and PHF1-PRC2 purified from Drosophila embryos and Hela cells, respectively, sediment at ~600 kDa (Nekrasov et al., 2007; Cao et al., 2008; Sarma et al., 2008). Consistent with these findings, we detected Pcl2-FLAG and Suz12 proteins in fractions corresponding to ~600 kDa (Fig. 2D).
Pcl2 modulates the biochemical properties of PRC2 complexes and facilitates PRC2 recruitment to the Xi
To analyse the function of Pcl2 we established ES cell lines in which Pcl2 expression was stably knocked down by shRNA. A number of constructs expressing shRNAs that target different parts of the Pcl2 transcript were transfected into PGK12.1 XX cells. Two shRNAs, CL3.5 and CL4, that resulted in a clear reduction in Pcl2 protein levels in a number of independent clones (see Fig. S2 in the supplementary material) were selected for further analysis.
We initially assessed PRC2 complexes in nuclear extracts from wild-type and knockdown cell lines using size exclusion chromatography. As shown in independent experiments (Fig. 3A and see Fig. S2C in the supplementary material), Pcl2 and the core PRC2 components Ezh2 and Suz12 co-elute in a major peak of 600-700 kDa (centred on fraction 21) and as high molecular weight complexes of 1-2 MDa (fractions 3-15). In Pcl2-depleted nuclear extracts we observed a shift in the size distribution of the major peak to ~500-550 kDa and a reduction in the abundance of the high molecular weight PRC2 complexes. Rbpj, a chromatin factor that is not associated with PRC2, eluted at equivalent fractions in both control and Pcl2-depleted nuclear extracts (Fig. 3A). The size shift seen for the major PRC2 peak might be attributable to the absence of Pcl2, although this seems unlikely given that Pcl2-PRC2 represents only a relatively small fraction of total PRC2 (Fig. 2C).
We went on to test the role of Pcl2 in PRC2 function in X chromosome inactivation. Indirect IF analysis was carried out using antibodies against Ezh2, Eed, Suz12 and H3K27me3 on two independent PGK12.1 Pcl2 shRNA ES cell lines and a control cell line differentiated in vitro for 3 days, a timepoint at which Xist upregulation and PRC2 recruitment to the Xi have occurred in a significant proportion of cells (Sheardown et al., 1997; Silva et al., 2003). We also analysed Xist RNA expression by RNA fluorescent in situ hybridisation (FISH) to exclude indirect effects on Xist expression or cell differentiation. For each experiment, the number of foci corresponding to the Xi was counted and divided by the total number of cells analysed. We observed a strong reduction in the number of cells with Xi domains detected using antibodies to Suz12, Ezh2 and Eed in both PGK12.1 Pcl2 shRNA clones relative to the shRNA control cell line (Fig. 3B). H3K27me3 Xi domains were also reduced in Pcl2 knockdown cells, although to a lesser extent. Robust Xist RNA domains were observed in ~35% of cells in all cases, which is similar to the frequency observed for PGK12.1 cells in previous studies (Sheardown et al., 1997), indicating that reduced PRC2 recruitment is not due to indirect effects.
To confirm these observations we carried out dual IF/RNA FISH analysis to detect both Ezh2 and Xist RNA in control and Pcl2 knockdown cells. Xist domains depleted for Ezh2 staining were detected in a significant proportion of Pcl2 knockdown cells when compared with wild-type controls (Fig. 3C,D). These observations demonstrate that Pcl2 has a role in the recruitment and/or maintenance of PRC2 on the Xi.
Pcl2 is important for the recruitment of PRC2 to Polycomb targets in ES cells
Recruitment of PRC2 complexes to Xi is dependent on Xist RNA (Mak et al., 2002; Silva et al., 2003) and the importance of Pcl2 in this process might constitute a specialised function. To explore this further, we analysed the role of Pcl2 at other known PcG target loci. Recent studies have demonstrated that, in undifferentiated ES cells, PcG complexes collaborate to repress a large number of genes that encode master regulators of embryonic and extraembryonic lineages (Boyer et al., 2006; Bracken et al., 2006; Lee et al., 2006). The promoters of many of these target genes exhibit an unusual chromatin configuration, termed bivalent chromatin, showing an enrichment for PcG-mediated repressive histone modifications, together with histone modifications that mediate transcriptional activity, such as H3K4me3 and H3/H4 acetylation (Azuara et al., 2006; Bernstein et al., 2006). Moreover, bivalent domains are associated with the presence of poised RNA polymerase II (Stock et al., 2007).
To determine whether Pcl2 localises to the promoters of known PRC2 target genes in ES cells, we carried out ChIP analysis. The Pcl2 antibody described above is not suitable for ChIP (data not shown), and we therefore carried out experiments using anti-FLAG antibody in PGK12.1 ES cell lines stably expressing Pcl2-FLAG. ChIP analysis for the histone modification H3K27me3 was carried out in parallel (Fig. 4A). We tested a number of known PcG target loci that encode regulators of extraembryonic and embryonic lineages. As a negative control we analysed Oct4 (Pou5f1 – Mouse Genome Informatics), which encodes a transcription factor that is highly expressed in ES cells, and the housekeeping gene beta-2 microglobulin (B2m). As expected, H3K27me3 levels were enriched at the promoters of PcG target loci but not at promoters of genes expressed in ES cells. Pcl2-FLAG was also enriched at these loci, with the relative level of enrichment at individual promoters closely mirroring that seen for H3K27me3. This result suggests that Pcl2-PRC2 localises to PRC2 target genes.
To investigate the function of Pcl2 at target loci in ES cells, we assessed levels of H3K27me3, H3K4me3 and of PRC2 core components in Pcl2 knockdown cell lines (Fig. 4B). We observed slightly reduced levels of both H3K27me3 and H3K4me3, although this varied between cell lines. Similarly, global levels of H3K27 and H3K4 methylation, as assessed by western blot analysis, were not significantly affected in the Pcl2 shRNA cell lines (Fig. 4C). In marked contrast, target occupancy of the PRC2 core components Ezh2 and Suz12 was strongly reduced in Pcl2 knockdown cell lines (Fig. 4B). We presume that the reduced PRC2 occupancy at target loci in Pcl2 knockdown cells is nevertheless sufficient to attain near normal levels of H3K27me3. Taken together, these results suggest that Pcl2 facilitates PRC2 recruitment to target loci.
We went on to determine whether Pcl2 knockdown results in the derepression of PRC2 target genes, as occurs in cells lacking core PRC2 proteins (Azuara et al., 2006; Boyer et al., 2006; Lee et al., 2006; Pasini et al., 2007). Pcl2 transcript levels were clearly reduced in the knockdown cell lines, as expected, but we did not detect elevated expression of the PcG target genes analysed (see Fig. S3 in the supplementary material). This is consistent with the relatively small decrease in H3K27me3 observed at target loci (Fig. 4B). In summary, Pcl2 plays a role in the recruitment of PRC2 complexes to target loci in ES cells. Whereas Pcl2 depletion significantly reduces PRC2 occupancy, H3K27me3 levels are only marginally affected and target genes remain fully repressed.
PRC2 targeting in ES cells requires the PHD2 domain of Pcl2
Pcl protein and the three mammalian homologues are characterised by the presence of a Tudor domain in the N-terminal region and two centrally located Plant Homeodomain (PHD) fingers (Lonie et al., 1994; O'Connell et al., 2001; Wang, S. et al., 2004). PHD and Tudor domains are found in many chromatin proteins, where they mediate interactions with histones (and other proteins), often through binding specific methylated residues (Bienz, 2006; Mellor, 2006). Previous studies have suggested that the PHD domains of Pcl are required for the interaction with E(z) in D. melanogaster (O'Connell et al., 2001). To determine the domain requirements for Pcl2 interactions with PRC2 and for the PRC2 targeting function we derived ES cell lines that express Pcl2-FLAG with or without specific domains (Fig. 5A and see Fig. S4 in the supplementary material) and carried out co-IP and ChIP analyses.
Pcl2-FLAG constructs lacking the Tudor (ΔT), PHD1 (ΔPHD1), PHD2 (ΔPHD2) or both PHD domains (ΔPHD1+2) were generated and transfected into PGK12.1 ES cells and stable cell lines that expressed the mutant proteins were selected (Fig. 5A and see Fig. S4 in the supplementary material). The Pcl2ΔPHD2 construct was poorly expressed in all cell lines tested (see Fig. S4 in the supplementary material) and was therefore excluded from subsequent analyses. To determine which domains of Pcl2 are important for interaction with PRC2 core components, the mutant Pcl2-FLAG proteins were immunoprecipitated using an anti-FLAG antibody bound to agarose beads and eluted using FLAG peptides. Western blot analysis was performed using antibodies to FLAG, Ezh2 and HP1γ (Cbx3 – Mouse Genome Informatics), the latter providing a negative control. The results demonstrated that all Pcl2 deletion mutants efficiently co-immunoprecipitate Ezh2, indicating that other regions of Pcl2 must be important for the interaction with PRC2 core components (Fig. 5B).
We then examined whether the Tudor and/or PHD domains of Pcl2 play a role in its localisation to the promoters of PRC2 target genes by performing ChIP on the cell lines described above. Consistent with the observations made for the full-length Pcl2-FLAG cell line, H3K27me3 levels were enriched at the promoters of target loci in all cell lines analysed. ΔT Pcl2-FLAG and ΔPHD1 Pcl2-FLAG localised at target loci similar to wild-type Pcl2-FLAG (Fig. 5C). By contrast, ΔPHD1+2 Pcl2-FLAG localisation was strongly impaired. These observations suggest that the PHD2 domain of Pcl2 is crucial for mediating PRC2 recruitment/stabilisation at target loci.
In this study we have shown that Pcl2, one of three mammalian homologues of Drosophila Pcl, is expressed predominantly in early development and that it forms a stable complex with the core PRC2 polycomb proteins Ezh2, Suz12 and Eed. We demonstrate that this complex, Pcl2-PRC2, plays an important role in the recruitment and/or stabilisation of PRC2 both on the Xi and at PcG target genes in ES cells. The PRC2 targeting activity of Pcl2-PRC2 is critically dependent on PHD finger domain 2, suggesting that Pcl2 might function through the recognition of a specific chromatin configuration.
Pcl2 is highly expressed in early development and in ES cells, suggesting that it contributes to PcG function at these stages. In the case of ES cells there is evidence that Pcl2 is a direct target of the pluripotency factors Oct4 and Nanog (Loh et al., 2006; Walker et al., 2007), and this is likely to be important in determining the Pcl2 expression pattern. The PRC2 protein Eed has similarly been shown to be a direct target of the ES cell transcription factor circuitry (Ura et al., 2008).
Our data suggest that high-level expression of Pcl2 in ES cells and in early embryogenesis is necessary to support the function of PRC2 core complexes in repressing PcG target loci, notably the key lineage determinants defined as PcG targets in ES cells (Boyer et al., 2006; Mikkelsen et al., 2007). A recent report investigating Pcl2 function in ES cells also demonstrated a role for Pcl2 in PRC2 targeting (Walker et al., 2010) and, in addition, found that depletion of Pcl2 enhances the self-renewal characteristics of ES cells and inhibits their differentiation. Our study did not directly address the latter, although we note that in our experiments Pcl2 knockdown did not overtly inhibit differentiation as assessed by embryoid body formation and induction of Xist RNA expression, the latter being directly linked to the suppression of the pluripotency transcription factor network (Navarro et al., 2008). These differences might be attributable to shRNA-mediated knockdown efficiency or the different ES cell lines used in the respective studies.
PRC2 levels and, more notably, H3K27me3 were only partly reduced following Pcl2 depletion in ES cells and there was no derepression of PcG target loci as has been reported following knockout of PRC2 core components (Azuara et al., 2006; Boyer et al., 2006; Pasini et al., 2007). Moreover, a previous study demonstrated that Pcl2-deficient mice show only mild phenotypes – specifically, minor posterior skeletal transformations (Wang et al., 2007). By contrast, Drosophila Pcl mutant embryos exhibit a strong phenotype and early embryo lethality (Duncan, 1982; Breen and Duncan, 1986). A likely explanation for the mild phenotype of mouse Pcl2 mutants is that the other Pcl homologues, Pcl1 and Pcl3 (Phf1 and Phf19 – Mouse Genome Informatics), functionally compensate for Pcl2 depletion. It will be important in future to address this point, for example by analysing combined knockout animals and/or ES cells.
The initial observation that Pcl2 localises to the Xi chromosome led us to consider that it might be important for Xist RNA-dependent recruitment of PRC2 during X inactivation. The Pcl proteins have two PHD finger domains and a Tudor domain, the latter being found in a number of proteins involved in RNA biogenesis and processing. We observed that the second PHD finger domain is important for PRC2 occupancy at target loci in ES cells. Based on the known function of PHD finger domains (Bienz, 2006; Mellor, 2006), this suggests that Pcl2 might recognise a specific chromatin configuration or histone modification landscape. The function of the other conserved domains of Pcl2 (Tudor and PHD1) is not apparent from our studies, although neither appears to be essential for complex formation or PRC2 occupancy at the target loci we analysed. We cannot rule out the possibility that these domains facilitate the recruitment of PRC2 to other PcG targets including the Xi chromosome.
The role of Pcl2 in facilitating PRC2 recruitment to the Xi chromosome and at PcG target loci in ES cells could be attributable to a function in the recognition of the primary targeting cues, such as transcription factors, the chromatin landscape or specific RNAs/RNA-binding proteins, or, alternatively, to a role in retaining or stabilising PRC2 occupancy following initial recruitment. The importance of the latter possibility has been highlighted by recent studies demonstrating that PRC2 occupancy is maintained following removal of an initial recruitment signal (Hansen et al., 2008), and that this might be mediated by a direct interaction of the Eed/Esc protein with H3K27me3 (Margueron et al., 2009). It has been suggested that this feedback mechanism is important for maintaining high levels of H3K27me3 at target loci through S phase and the cell cycle and a similar model could be suggested for Pcl2 function mediated through the recognition of PcG target chromatin configuration via PHD domain 2. This explanation would more readily account for the role of Pcl2 in the recruitment of PRC2 both to Xi, which is thought to involve direct targeting by Xist RNA (Plath et al., 2003; Silva et al., 2003; Kohlmaier et al., 2004), and to target loci in ES cells, at which PRC2 recruitment is more likely to be determined by transcription factor binding and/or chromatin features (Endoh et al., 2008; Margueron et al., 2009).
Affinity purification demonstrated that Pcl2 interacts with the Ezh2, Eed and Suz12 core components of the PRC2 complex in a stable biochemical complex with a molecular weight of ~600 kDa. This is in close agreement with data for Pcl complexes purified from Drosophila embryos (Nekrasov et al., 2007) and PHF1 (PCL1) complexes from Hela cells (Cao et al., 2008; Sarma et al., 2008). Similar to these studies, our analysis indicates that the Pcl2-PRC2 complex does not include stoichiometric levels of any other proteins. Jarid2, recently described as a major PRC2 interactor in ES cells, and AEBP2, a zinc-finger protein observed to co-purify with PRC2 in Hela cells and ES cells, were not detected, suggesting that the interaction of these components with PRC2 might be mutually exclusive with that of Pcl2. Consistent with this, a recent study found that Pcl2, Aebp2 and Jarid2 co-purify with the PRC2 core protein Eed from ES cells, but that Pcl2 levels are much lower in complexes that co-purify with Jarid2 (Landeira et al., 2010).
Although purified Pcl2-PRC2 complexes migrate on gel filtration columns with a molecular weight of ~600 kDa, which is similar to that of core PRC2 complexes purified in a number of independent studies, direct gel filtration of ES cell nuclear extracts revealed a size distribution ranging from 600-700 kDa up to 2 MDa. High molecular weight PRC2 species have also been described previously, both in Drosophila (Tie et al., 2003) and Hela cells (Kuzmichev et al., 2005). Following depletion of Pcl2 we observed a shift in the size distribution of the major peak to ~500-550 kDa and a reduction in the abundance of the higher molecular weight (1-2 MDa) PRC2 species. One interpretation of this result is that the larger complexes include additional subunits for which recruitment is Pcl2 dependent. Alternatively, high molecular weight species could arise through multimerisation of core PRC2 complexes, an idea favoured by recent studies revealing that the N-terminus of the core PRC2 protein Eed/Esc interacts with histone H3 and dimerises in a phosphorylation-dependent manner (Tie et al., 2005; Margueron et al., 2009). Although we cannot differentiate between these possibilities, the absence of additional stoichiometric components in purified Pcl2-PRC2 and the fact that Pcl2 depletion affects the biochemical properties of total PRC2, only a small proportion of which is associated with Pcl2 at any given time, lead us to favour the idea that Pcl2 mediates the formation of higher-order PRC2 complexes.
In summary, this study highlights Pcl2 as an important PRC2 co-factor that functions in early development and in ES cells to facilitate the recruitment/retention of the complex at target loci.
We thank Naveenan Navaratnam and Rob Klose for advice on biochemical studies; Amanda Fisher, Stephan Sauer, David Landeira and members of the N.B. lab for valuable feedback and discussion; and Anne-Valérie Gendrel for critical reading of the manuscript. This work was supported by the MRC UK and the Wellcome Trust. M.C. was supported by a D. Phil. GABBA Studentship from the Fundação para a Ciência e Tecnologia, Portugal. Deposited in PMC for release after 6 months.
Competing interests statement
The authors declare no competing financial interests.
Supplementary material for this article is available at http://dev.biologists.org/lookup/suppl/doi:10.1242/dev.053652/-/DC1
- Accepted January 17, 2011.
- © 2011.