1 Institut Curie, 26 Rue d���Ulm, 75005 Paris, France. 2 CNRS UMR3215, 26 Rue d���Ulm, 75005 Paris, France. 3 INSERM U934, 26 Rue d���Ulm, 75005 Paris, France. 4 Howard Hughes Medical Institute, Department of Biochemistry, New York University School of Medicine, 522 First Ave, New York, New York 10016, USA. Polycomb group proteins maintain the gene-expression pattern of different cells that is set during early development by regulating chromatin structure. In mammals, two main Polycomb group complexes exist ��� Polycomb repressive complex 1 (PRC1) and 2 (PRC2). PRC1 compacts chromatin and catalyses the monoubiquitylation of histone H2A. PRC2 also contributes to chromatin compaction, and catalyses the methylation of histone H3 at lysine 27. PRC2 is involved in various biological processes, including differentiation, maintaining cell identity and proliferation, and stem-cell plasticity. Recent studies of PRC2 have expanded our perspectives on its function and regulation, and uncovered a role for non-coding RNA in the recruitment of PRC2 to target genes. Tthat he term Polycomb (Pc) initially referred to a Drosophila mutant that displayed improper body segmentation1. It was suggested Polycomb encodes a negative regulator of the homeotic genes that are required for segmentation2. The Polycomb group (PcG) now defines a set of genes characterized by mutations that result in similar phenotypes to those of Polycomb. The crucial role of PcG proteins dur- ing development is highlighted by early embryonic lethality in mice after the deletion of genes encoding some of these proteins (Eed, Ezh2 (also known as Enx-1), Suz12 and Ring1B (Rnf2)). The antagonistic activities of the PcG and the trithorax families of proteins culminate in the maintenance, throughout development and adulthood, of the appropriate patterns of homeotic gene expression in a spatially defined manner3. PcG proteins are found in several families of multiprotein complexes, including the Polycomb repressive complexes PRC1 and PRC2 (Fig. 1). Two other PcG complexes were characterized in Drosophila, PHO-repressive complex (PhoRC) and Polycomb repres- sive deubiquitinase (PR-DUB), and their components have orthologues in mammals however, the conservation of their functions has not yet been addressed4���6. Polycomb-mediated gene silencing is thought to rely mostly on the regulation of chromatin structure, in part through post-transla- tional modification (PTM) of histones. Hence, the PRC2 complex is responsible for the methylation (di- and tri-) of Lys 27 of histone H3 (H3K27me2/3)3,6 through its enzymatic subunits EZH1 and EZH2, whereas the PRC1 complex monoubiquitylates Lys 119 of histone H2A (H2AK119ub) via the ubiquitin ligases RING1A and RING1B (Fig. 1). In addition, some PRC1 complexes can regulate gene expression by compacting chromatin in a manner independent of enzymatic activ- ity7. The PRC1 component Pc (known as CBX in mammals) binds specifically to the product of PRC2 catalysis, H3K27me3, leading to the hypothesis that PRC1 functions downstream of PRC2. Although this premise is still cited in the literature, its operational status is equivocal as there are genes targeted by PRC2 that lack H2AK119ub8 and genes targeted by PRC1 in the absence of PRC2 (refs 9, 10). Nonetheless, PRC2 and PRC1 are often both required to maintain gene repression. Owing to the pivotal role of PRC2 in the coordination of PcG protein function, the still partial characterization of PRC1 and PRC1-like com- plexes in mammals, and the existence of up-to-date reviews on PRC1 (refs 3, 6), this Review focuses primarily on mammalian PRC2. After considering PRC2 in terms of evolution, we evaluate the newly appre- ciated, variable composition of PRC2 and describe the function of its catalytic product and its localization. Finally, we discuss the biological roles of PRC2, and propose a model for its recruitment to target genes that involves non-coding RNA (ncRNA). Evolution of PRC2 The core PRC2 complex, which is conserved from Drosophila to mammals, comprises four components: EZH1/2, SUZ12, EED and RbAp46/48 (also known as RBBP7/4). The composition of PRC1 complexes is more variable, with only two core common components ��� RING1A/B together with BMI1, MEL18 (PCGF2) or NSPC1 (PCGF1)6,11 (Fig. 1). The presence of PRC2 in various unicellular eukaryotes led to the suggestion that it existed in the last common unicellular ancestor, but it was lost at times during evolution as exemplified by the cases of Schizosaccharomyces pombe and Saccharomyces cerevisiae, in which PRC2 is absent12. Notably, the PRC2 components, in contrast to those of PRC1, underwent little duplication in mammals, with vertebrates containing two copies of enhancer of zeste homologue, EZH1 and EZH2 (ref. 12). Drosophila has two copies of the EED homologue, ESC and ESCL. Although ESC and ESCL are interchangeable13, the same might not be true for EZH1 and EZH2, which have different expression pat- terns. EZH1 is present in both dividing and differentiated cells, whereas EZH2 is found only in actively dividing cells. Also, PRC2 complexes containing EZH1 (PRC2���EZH1) in lieu of EZH2 have low methyl- transferase activity compared with PRC2���EZH2 (ref. 14). This indi- cates that PRC2���EZH2 establishes cellular H3K27me2/3 levels through its EZH2-mediated methyltransferase activity, and that PRC2���EZH1 restores H3K27me2/3 that could have been lost after histone exchange or through demethylase activity. Moreover, PRC2���EZH1 and ���EZH2 have distinct chromatin-binding properties, as illustrated by the specific chromatin-compaction property of PRC2���EZH1 (ref. 14). In contrast to mammals, PRC2 evolved towards a greater complex- ity in plants, with species such as Arabidopsis thaliana having up to 12 homologues of PRC2 components15. A homologue of the mam- malian and S. pombe heterochromatin protein 1 (HP1) that binds to H3K9me3 also exists in plants and is denoted LHP1. LHP1 binds to H3K27me3 and interacts with the RING1 homologues AtRING1A and AtRING1B, suggesting the existence of a PRC1-like complex The Polycomb complex PRC2 and its mark in life Rapha��l Margueron1,2,3 & Danny Reinberg4 2 0 J a n u a R y 2 0 1 1 | V O L 4 6 9 | n a T u R E | 3 4 3 REVIEW doi:10.1038/nature09784 �� 2011 Macmillan Publishers Limited. All rights reserved

in plants15. Whereas EZH1 and EZH2 target the same genes and are expected to contribute to the same silencing pathway16, the plant PRC2 complexes were reported to have distinct functions15. On the basis of these criteria, we speculate that PRC2 evolved from a function that was partially redundant with gene silencing through the H3K9me3 pathway, gaining a more specific role as multicellular organ- isms acquired specific cell lineages. PRC2 comprises more than four components The first PRC2 purifications led to the identification of the four com- ponents that are required for its enzymatic activity in vitro. It was recently shown that PRC2 contains several other polypeptides (Fig. 1) ��� AEBP2, PCLs and JARID2 ��� the functions of which are described below. Of note, other proteins transiently interact with PRC2 (for exam- ple, DNMTs, HDAC1 and SIRT1), but their effect on PRC2 function is unclear, and as such, they are not discussed further here. AEBP2 is a zinc-finger protein that was identified as part of the PRC2 complex. It interacts with several PRC2 components to enhance its enzymatic activity17, and co-localizes with PRC2 at some target genes18. AEBP2 was postulated to bind DNA with an apparently relaxed specificity18. PCL1, PCL2 and PCL3 (also known as PHF1, MTF2 and PHF19, respectively) are the three mammalian orthologues of Drosophila Polycomblike (PCL). They share the same protein motifs: a tudor domain, two plant homeodomain (PHD) finger proteins, a PCL extended domain and a carboxy-terminal domain tail19 (Fig. 1). PCL proteins interact with PRC2 through EZH2, and to some extent through SUZ12 and the his- tone chaperones RbAp46 and RbAp48 (ref. 20). Genome-wide studies showed that PCL2 co-occupied PRC2 target genes21,22. Various functions have been attributed to PCLs, from the regulation of PRC2 enzymatic activity20,23 to the gene recruitment of PRC2 (refs 21, 24). Mammalian PCLs are expressed in a tissue-specific manner21, and this redundancy could explain apparent discrepancies between studies. The phenotypes associated with PCL mutation in Drosophila and Xenopus, and the co- localization and interaction of PCLs and PRC2, point to PCL proteins having a crucial role in PRC2 function. Understanding the underlying molecular mechanisms will probably require a detailed understanding of how PCLs interact with chromatin. JARID2 is the founding member of the Jumonji family of proteins that catalyses the demethylation of histone proteins, yet it lacks the key residues necessary for cofactor binding and is devoid of enzymatic activity. Its deletion in mice results in severe defects in cardiovascular and liver development25. The C-terminal half of JARID2 contains some conserved regions such as the ARID domain (a potential DNA-binding domain), the JmjC and JmjN domains, and a zinc finger (Fig. 1). JARID2 was identified as a PRC2 component, and biochemical studies have EED RNA SUZ12 Nucleic acid? SANT ncRBD SANT CXC SET Box 1 Box 2 66% 56% 76% 84% EZH2 SUZ12 1 746 1 741 Zn VEFS EED Histone H3 1 Binding to repressive trimethylated lysines 441 WD40 WD40WD40WD40 WD40WD40 WD40 1 425 RbAp48 JARID2 1 1234 DNA binding DNA binding PRC2 interaction & regulation repression domain Zn JmjN JmjC ARID Tudor PHD PHD 1 559 1 504 Zn Zn Zn Modified histone recognition domain? PCL homology Nucleic-acid-binding domain? PCL1 AEBP2 PRC2 PRC1 Di-/trimethylation of H3K27 Chromatin compaction Monoubiquitylation of H2AK119 Chromatin compaction Activity? BMI1 HPH1 RYBP X Z MEL18 NSPC1 HPH2 HPH3 RING1A/B RING1A/B RING1A/B CBX2 CBX4 CBX6 CBX7 CBX8 BMI1... RING1A/B ? SUZ12 EZH1/2 SET EED AEBP2 RbAp46/48 JARID2 PCL Y a b WD40 WD40 WD40 WD40 WD40 WD40 Figure 1 | The Polycomb complexes PRC1 and PRC2. a, Diagrams representing the composition of PRC2 and PRC1 are shown. In PRC1, the diagrams shown on the left correspond to the classical PRC1 complexes, whereas those on the right correspond to the so-called PRC1-like complexes. Owing to their homology with the Drosophila PSC protein, we assumed that the BMI1-, MEL18- and NSPC1-containing PRC1 complexes could compact chromatin. The ���pocket��� shape of the CBX proteins represents the chromodomain that specifically recognized H3K9/27me3. HPH1, 2 and 3 denote human polyhomeotic homologue 1, 2 and 3. X, Y and Z denote various proteins such as SCMH1/2, FBXL10, E2F6 and JARID1D that could contribute to the formation of PRC1-like complexes, whose exact composition is still enigmatic. b, Characterized domains with potential functions are indicated for each PRC2 component. In EZH2, box 1 and 2 refer to domains based on sequence homology, and the numbers below the scheme indicate the percentage similarity between mouse and Drosophila homologues for the corresponding domain. CXC, cysteine-rich domain ncRBD, non-coding- RNA-binding domain SANT, SWI3, ADA2, N-CoR and TFIIIB DNA- binding domain SET, Su(var)3-9, enhancer of zeste, trithorax domain VEFS, conserved among VRN2���EMF2���FIS2���SU(Z)12 WD40, short ~40 amino acid motifs. 3 4 4 | n a T u R E | V O L 4 6 9 | 2 0 J a n u a R y 2 0 1 1 Review insight �� 2011 Macmillan Publishers Limited. All rights reserved