
ARTICLES https://doi.org/10.1038/s41594-018-0023-y Cryo-EM structures of PRC2 simultaneously engaged with two functionally distinct nucleosomes Simon Poepsel 1,2, Vignesh Kasinath1,2 and Eva Nogales 1,2,3,4* Epigenetic regulation is mediated by protein complexes that couple recognition of chromatin marks to activity or recruitment of chromatin-modifying enzymes. Polycomb repressive complex 2 (PRC2), a gene silencer that methylates lysine 27 of histone H3, is stimulated upon recognition of its own catalytic product and has been shown to be more active on dinucleosomes than H3 tails or single nucleosomes. These properties probably facilitate local H3K27me2/3 spreading, causing heterochromatin formation and gene repression. Here, cryo-EM reconstructions of human PRC2 bound to bifunctional dinucleosomes show how a single PRC2, via interactions with nucleosomal DNA, positions the H3 tails of the activating and substrate nucleosome to interact with the EED subunit and the SET domain of EZH2, respectively. We show how the geometry of the PRC2–DNA interac- tions allows PRC2 to accommodate varying lengths of the linker DNA between nucleosomes. Our structures illustrate how an epigenetic regulator engages with a complex chromatin substrate. ovalent modification of the N-terminal tails of histone pro- such as H3K4me3 (ref. 11) or H3K36me3 (ref. 12) or the interaction teins forming the protein core of the nucleosomes that pack- with noncoding RNAs13,14 and auxiliary subunits15–17. Cage DNA in eukaryotes is a fundamental mechanism of Biochemical studies have shown that the activity of PRC2 is sig- epigenetic gene regulation. Histone-modifying enzymes catalyze nificantly higher on dinucleosomes and higher-order chromatin the deposition or removal of histone marks, which can in turn be structures than on histone tails or mononucleosomes, a property bound by specific recognition modules within larger protein assem- that is not mechanistically understood but may also contribute to blies that serve gene regulatory functions1. The faithful orchestra- the spreading of the PRC2 silencing mark4,18. Indeed, many ques- tion of gene regulatory patterns, for example, during development, tions have remained unanswered. How does PRC2 engage with critically relies on the interplay of sensing and altering the chroma- its natural chromatin substrates, nucleosomal arrays? Does PRC2 tin state. Consequently, both of these activities are typically found interact with core histones via the acidic patch used by many in gene regulatory complexes. The dynamic nature of chromatin nucleosome-binding proteins? Does the interaction instead involve poses a challenge to studies aiming to elucidate these processes, nucleosomal DNA? Can more than one nucleosome be engaged by both in cellular context and in reconstituted systems, and detailed a single PRC2? If so, can such engagement occur in the context of structural studies of epigenetic regulators have so far been limited neighboring nucleosomes, and how does nucleosome spacing and to histone peptide-bound complexes or single functional modules geometry affect PRC2 engagement? bound to nucleosomes2,3. Here, we provide direct visualization of PRC2–chromatin inter- Trimethylation of lysine 27 on histone H3 (H3K27me3), cata- actions through cryo-EM structures of PRC2 in the specific context lyzed by PRC2, leads to gene silencing of developmental and cell- of dinucleosomes containing one unmodified substrate nucleosome fate-determining genes within multicellular organisms4. All four and one activating, H3K27me3-containing nucleosome. This com- core PRC2 subunits, enhancer of zeste homolog 2 (EZH2), embry- bination is particularly relevant for our understanding of H3K27me3 onic ectoderm development (EED), suppressor of zeste 12 (SUZ12) spreading, as it functionally corresponds to a boundary condition in and RBAP46 or RBAP48, have been proposed to contribute to which the state of one nucleosome can directly affect the activity of histone-tail binding4–9. Engagement of H3K27me3 by the recog- the complex on the neighboring one. We find that PRC2 interacts nition module EED characteristically leads to allosteric activation with the histone H3 tails and with the nucleosomal DNA, but not of the catalytic su(var)3–9, enhancer-of-zeste and trithorax (SET) with other histones or the histone core. Of special functional rel- domain within EZH2, a mechanism that has been characterized in evance, our reconstructions show how the specific geometry of the detail only using peptide ligands. Crystal structures of the catalytic assembly allows the simultaneous engagement of both nucleosomes lobes of a fungal9 and human10 PRC2, comprised of EZH2, EED, by the same complex, even in the context of various linker lengths. and the C-terminal VEFS domain of SUZ12, bound to stimulatory Binding of the substrate nucleosome by a rigid DNA-binding inter- methylated peptides have offered clues about the structural rear- face on the CXC domain of EZH2 positions the H3 tail optimally rangements within PRC2 that lead to activation, which is thought to for modification by EZH2. On the other side of the complex, a more contribute to local spreading of H3K27me3 and the establishment flexible binding surface involving the WD40 domain of EED allows of heterochromatin domains. Regulatory mechanisms controlling for the recognition of an activating H3K27me3 in the context of PRC2 function also include inhibition by active chromatin marks geometrically diverse chromatin substrates. Our structures support 1California Institute for Quantitative Biology (QB3), University of California, Berkeley, Berkeley, CA, USA. 2Molecular Biophysics and Integrative Bio- Imaging Division, Lawrence Berkeley National Laboratory, Berkeley, CA, USA. 3Howard Hughes Medical Institute, University of California, Berkeley, Berkeley, CA, USA. 4Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, CA, USA. *e-mail: [email protected] NATURE STRUCTURAL & MOLecULAR BIOLOGY | www.nature.com/nsmb © 2018 Nature America Inc., part of Springer Nature. All rights reserved. ARTICLES NATURE STRUCTURAL & MOLECULAR BIOLOGY an H3K27me3-based PRC2 activation and spreading mechanism, It was possible to unambiguously assign the densities of the top which had been proposed previously on the basis of biochemical catalytic lobe, comprising EZH2, EED, and the VEFS domain of data, linking activation of the SET domain with the right engage- SUZ12, and a bottom lobe density for RBAP48 to published crys- ment of a new PRC2 substrate. Our work also provides a framework tal structures (Fig. 1b). AEBP2 and the N-terminal part of SUZ12 to probe the mechanism of how differences in nucleosome spacing probably correspond to the remaining unassigned densities local- might lead to changes in methyltransferase activity. ized to the bottom lobe, in agreement with our previous study using genetic labels19 and with our more recent high-resolution structures Results of a six-component complex (PRC2–AEBP2–JARID2)20. The cryo- Structure of a human PRC2 and its interaction with dinucleo- EM structure of human PRC2 was obtained following mild cross- somes. For our structural studies of PRC2 interactions with chro- linking of the complex, which is absolutely necessary to maintain matin, we decided to explore a specific functional configuration, the integrity of the complex during the harsh process of sample with a minimal, structurally tractable substrate: a dinucleosome that blotting and vitrification that is used to generate a frozen hydrated includes an unmodified and an H3K27me3-modified nucleosome. sample for cryo-EM visualization (Methods). Cross-linking, how- We generated recombinant heterodinucleosomes with 35 base pairs ever, proved incompatible with nucleosome binding, as assessed (bp) of linker DNA (DiNcl35) by ligating mononucleosomes harbor- by both EMSA and cryo-EM visualization (data not shown), most ing pseudotrimethylated H3K27 (Nclmod) and unmodified H3 (Nclsub) likely caused by the titration of key functional lysines needed for (Supplementary Fig. 1) and purified recombinant human PRC2 nucleosome engagement. Disruption of dinucleosome binding to composed of the core subunits EZH2, EED, SUZ12, and RBAP48 PRC2 was observed even when cross-linker was added after forma- and the cofactor AEBP2 (Fig. 1a and Supplementary Fig. 1c). AEBP2 tion of the PRC2–dinucleosome complex, thus suggesting that the has been shown to have an overall stabilizing effect on the complex off rate of the dinucleosome was fast enough for the cross-linker to through extensive interactions with other subunits19 and may play compete for the lysines. Therefore, both the negative-stain analysis a role in chromatin binding based on its proposed DNA-binding and the following cryo-EM study of PRC2–dinucleosome interac- properties15. We hereafter refer to this five-subunit PRC2 assembly tions had to be carried out in the absence of cross-linker. simply as PRC2. Binding of our dinucleosomes to PRC2 was tested Analysis of our negative-stain 2D class averages showed several using electrophoretic mobility shift assay (EMSA) (Supplementary distinct populations of PRC2–dinucleosome complexes (Fig. 1c and Fig. 1c). Reference-free 2D classification of an initial negative-stain Supplementary Fig. 3a). In all of the class averages, we observed one EM dataset showed typical views of PRC2 with two nucleosomes of the nucleosomes positioned by the top catalytic lobe of PRC2, in bound (Supplementary Fig. 1d). In order to be able to
Details
-
File Typepdf
-
Upload Time-
-
Content LanguagesEnglish
-
Upload UserAnonymous/Not logged-in
-
File Pages14 Page
-
File Size-