Article30 September 2004free access Dual histone H3 methylation marks at lysines 9 and 27 required for interaction with CHROMOMETHYLASE3 Anders M Lindroth Anders M Lindroth Department of Molecular, Cell & Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USAPresent address: Division of Nutritional Sciences, College of Agriculture and Life Sciences, Cornell University, 108 Savage Hall, Ithaca, NY 14853, USA Search for more papers by this author David Shultis David Shultis Department of Biochemistry and Molecular Genetics, University of Virginia Health System, Charlottesville, VA, USA Search for more papers by this author Zuzana Jasencakova Zuzana Jasencakova The Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, GermanyPresent address: Dulbecco Telethon Institute, Institute of Genetics and Biophysics CNR, 80131, Napoli, Italy Search for more papers by this author Jörg Fuchs Jörg Fuchs The Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany Search for more papers by this author Lianna Johnson Lianna Johnson Department of Molecular, Cell & Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USA Search for more papers by this author Daniel Schubert Daniel Schubert Institute of Plant Molecular Science (IMPS), School of Biology, University of Edinburgh, Edinburgh, UK Search for more papers by this author Debasis Patnaik Debasis Patnaik New England Biolabs Inc., Beverly, MA, USA Search for more papers by this author Sriharsa Pradhan Sriharsa Pradhan New England Biolabs Inc., Beverly, MA, USA Search for more papers by this author Justin Goodrich Justin Goodrich Institute of Plant Molecular Science (IMPS), School of Biology, University of Edinburgh, Edinburgh, UK Search for more papers by this author Ingo Schubert Ingo Schubert The Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany Search for more papers by this author Thomas Jenuwein Thomas Jenuwein Research Institute of Molecular Pathology (IMP), The Vienna Biocenter, Vienna, Austria Search for more papers by this author Sepideh Khorasanizadeh Corresponding Author Sepideh Khorasanizadeh Department of Biochemistry and Molecular Genetics, University of Virginia Health System, Charlottesville, VA, USA Search for more papers by this author Steven E Jacobsen Corresponding Author Steven E Jacobsen Department of Molecular, Cell & Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USA Molecular Biology Institute, University of California, Los Angeles, Los Angeles, CA, USA Search for more papers by this author Anders M Lindroth Anders M Lindroth Department of Molecular, Cell & Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USAPresent address: Division of Nutritional Sciences, College of Agriculture and Life Sciences, Cornell University, 108 Savage Hall, Ithaca, NY 14853, USA Search for more papers by this author David Shultis David Shultis Department of Biochemistry and Molecular Genetics, University of Virginia Health System, Charlottesville, VA, USA Search for more papers by this author Zuzana Jasencakova Zuzana Jasencakova The Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, GermanyPresent address: Dulbecco Telethon Institute, Institute of Genetics and Biophysics CNR, 80131, Napoli, Italy Search for more papers by this author Jörg Fuchs Jörg Fuchs The Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany Search for more papers by this author Lianna Johnson Lianna Johnson Department of Molecular, Cell & Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USA Search for more papers by this author Daniel Schubert Daniel Schubert Institute of Plant Molecular Science (IMPS), School of Biology, University of Edinburgh, Edinburgh, UK Search for more papers by this author Debasis Patnaik Debasis Patnaik New England Biolabs Inc., Beverly, MA, USA Search for more papers by this author Sriharsa Pradhan Sriharsa Pradhan New England Biolabs Inc., Beverly, MA, USA Search for more papers by this author Justin Goodrich Justin Goodrich Institute of Plant Molecular Science (IMPS), School of Biology, University of Edinburgh, Edinburgh, UK Search for more papers by this author Ingo Schubert Ingo Schubert The Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany Search for more papers by this author Thomas Jenuwein Thomas Jenuwein Research Institute of Molecular Pathology (IMP), The Vienna Biocenter, Vienna, Austria Search for more papers by this author Sepideh Khorasanizadeh Corresponding Author Sepideh Khorasanizadeh Department of Biochemistry and Molecular Genetics, University of Virginia Health System, Charlottesville, VA, USA Search for more papers by this author Steven E Jacobsen Corresponding Author Steven E Jacobsen Department of Molecular, Cell & Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USA Molecular Biology Institute, University of California, Los Angeles, Los Angeles, CA, USA Search for more papers by this author Author Information Anders M Lindroth1, David Shultis2, Zuzana Jasencakova3, Jörg Fuchs3, Lianna Johnson1, Daniel Schubert4, Debasis Patnaik5, Sriharsa Pradhan5, Justin Goodrich4, Ingo Schubert3, Thomas Jenuwein6, Sepideh Khorasanizadeh 2 and Steven E Jacobsen 1,7 1Department of Molecular, Cell & Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USA 2Department of Biochemistry and Molecular Genetics, University of Virginia Health System, Charlottesville, VA, USA 3The Institute of Plant Genetics and Crop Plant Research (IPK), Gatersleben, Germany 4Institute of Plant Molecular Science (IMPS), School of Biology, University of Edinburgh, Edinburgh, UK 5New England Biolabs Inc., Beverly, MA, USA 6Research Institute of Molecular Pathology (IMP), The Vienna Biocenter, Vienna, Austria 7Molecular Biology Institute, University of California, Los Angeles, Los Angeles, CA, USA *Corresponding authors. Department of Biochemistry and Molecular Genetics, University of Virginia Health System, 1300 Jefferson Park Avenue, Charlottesville, VA 22908, USA. Tel.: +1 434 243 4817; Fax: +1 434 924 5069; E-mail: [email protected] of MCD Biology, University of California, Los Angeles, PO Box 951606, Los Angeles, CA 90095-1606, USA. Tel.: +1 310 825 0182; Fax: +1 310 206 3987; E-mail: [email protected] The EMBO Journal (2004)23:4286-4296https://doi.org/10.1038/sj.emboj.7600430 Correction(s) for this article Dual histone H3 methylation marks at lysines 9 and 27 required for interaction with CHROMOMETHYLASE304 May 2011 PDFDownload PDF of article text and main figures. ToolsAdd to favoritesDownload CitationsTrack CitationsPermissions ShareFacebookTwitterLinked InMendeleyWechatReddit Figures & Info Both DNA methylation and post-translational histone modifications contribute to gene silencing, but the mechanistic relationship between these epigenetic marks is unclear. Mutations in two Arabidopsis genes, the KRYPTONITE (KYP) histone H3 lysine 9 (H3K9) methyltransferase and the CHROMOMETHYLASE3 (CMT3) DNA methyltransferase, cause a reduction of CNG DNA methylation, suggesting that H3K9 methylation controls CNG DNA methylation. Here we show that the chromodomain of CMT3 can directly interact with the N-terminal tail of histone H3, but only when it is simultaneously methylated at both the H3K9 and H3K27 positions. Furthermore, using chromatin immunoprecipitation analysis and immunohistolocalization experiments, we found that H3K27 methylation colocalizes with H3K9 methylation at CMT3-controlled loci. The H3K27 methylation present at heterochromatin was not affected by mutations in KYP or in several Arabidopsis PcG related genes including the Enhancer of Zeste homologs, suggesting that a novel pathway controls heterochromatic H3K27 methylation. Our results suggest a model in which H3K9 methylation by KYP, and H3K27 methylation by an unknown enzyme provide a combinatorial histone code for the recruitment of CMT3 to silent loci. Introduction Gene silencing in eukaryotes is associated with several epigenetic mechanisms including post-translational histone modifications, DNA methylation, and small RNAs, and accumulating evidence suggests extensive crosstalk between these mechanisms (Elgin and Grewal, 2003; Fischle et al, 2003a). Transcriptionally active or silenced genes are associated with distinct combinations of post-translational histone modifications important for regulating gene expression (Strahl and Allis, 2000; Turner, 2000). Methylation of the histone H3 tail at lysine 9 (H3K9) is associated with heterochromatic gene silencing (Lachner et al, 2001; Nakayama et al, 2001; Schotta et al, 2002). H3K9 methylation creates a binding site for the chromodomain of Heterochromatin-associated Protein 1 (HP1), an important silencing protein found in many different organisms (Bannister et al, 2001; Jacobs et al, 2001; Lachner et al, 2001; Nakayama et al, 2001). Lysines can accept three methyl groups, and can therefore be monomethylated, dimethylated, or trimethylated (hereafter denoted m, m2, and m3), and recent evidence suggests that there may be functional differences between these methylation states (Dutnall, 2003). For instance, in mammalian cells, H3K9m3 is preferentially localized to pericentromeric heterochromatin, while H3K9m and H3K9m2 are localized to euchromatin (Peters et al, 2003; Rice et al, 2003). In Neurospora, silent loci are associated with H3K9m3 (Tamaru et al, 2003). However, in Arabidopsis, both H3K9m and H3K9m2 are found at silent loci, with H3K9m3 virtually absent (Jackson et al, 2004). Methylation of histone H3 lysine 27 (H3K27) is also associated with gene silencing in several systems. For instance, mammals show an enrichment of H3K27m at heterochromatin and H3K27m2 in euchromatin (Peters et al, 2003), and methyl H3K27 also marks the inactive X chromosome (Plath et al, 2003; Okamoto et al, 2004; Rougeulle et al, 2004). Furthermore, homeotic gene silencing in animals is associated with H3K27 methylation controlled by polycomb group (PcG) proteins (reviewed in Cao and Zhang, 2004). In Arabidopsis, silencing of the FLC locus is associated with H3K27 methylation, which depends on VRN2, a homolog of the PcG member Su(z)12 (Bastow et al, 2004; Sung and Amasino, 2004). The eukaryotic enzymes shown thus far to methylate H3K27 are the mammalian SET domain-containing protein G9a and the PcG Enhancer of zeste (E(Z)) proteins. G9a can methylate both H3K9 and K27 in vitro (Tachibana et al, 2001). However, while its role in methylating H3K9 in vivo has been characterized, its ability to methylate H3K27 in vivo has yet to be demonstrated (Tachibana et al, 2002; Rice et al, 2003). E(Z) proteins exist in conserved protein complexes responsible for H3K27 methylation of homeotic genes and of the inactive X chromosome (Cao et al, 2002; Czermin et al, 2002; Kuzmichev et al, 2002; Muller et al, 2002). One component of these complexes is the Extra Sex Combs (ESC) WD-40 protein, which is required for H3K27 methylation activity in vivo, and which is found in both animal and plant PcG complexes (Czermin et al, 2002; Kuzmichev et al, 2002; Muller et al, 2002; Hsieh et al, 2003). The polycomb protein (Pc) binds to methyl H3K27 via its chromodomain suggesting a possible mechanism for its action in the silencing of homeotic genes (reviewed in Cao and Zhang, 2004). The Arabidopsis genome encodes three E(Z) homologs, CURLY LEAF (CLF), SWINGER (SWN) (formerly named EZA1), and MEDEA (MEA) (reviewed in Hsieh et al, 2003). CLF encodes a repressor of the floral homeotic gene AGAMOUS (Goodrich et al, 1997). SWN is a very close homolog of CLF and appears to act in a partially redundant fashion with CLF (Chanvivattana et al, 2004). MEA is an imprinted gene encoding a repressor of gene expression and endosperm development, and is expressed in the developing female gametophyte and endosperm (reviewed in Hsieh et al, 2003). The FERTILIZATION INDEPENDENT ENDOSPERM (FIE) gene encodes the sole Arabidopsis ESC homolog. FIE mutations were initially identified by their endosperm development phenotypes, which are similar to those of mea mutants. Subsequently, using a fie partial loss-of-function allele, fie TK114, FIE was shown to be involved in floral homeotic gene repression (Kinoshita et al, 2001). Thus, the plant polycomb complexes likely act in the same fashion as in animals in the long-term repression of developmental regulatory genes (Hsieh et al, 2003). Recent reports suggest a close relationship between histone H3K9 methylation and DNA methylation in several organisms (Tamaru and Selker, 2001; Jackson et al, 2002; Lehnertz et al, 2003). For instance, Neurospora mutants lacking histone H3K9 methylation show a complete loss of DNA methylation in all sequence contexts (Tamaru and Selker, 2001). In addition, in a screen for mutations that derepress the silencing of the heavily CNG (where N=A, T, C, or G) methylated and silenced Arabidopsis SUPERMAN locus, we previously isolated two genes with similar loss-of-function phenotypes, encoding the DNA methyltransferase CMT3 (Lindroth et al, 2001) and the H3K9-specific histone methyltransferase KYP (Jackson et al, 2002). These same two genes were cloned from an independent mutant screen for suppressors of silencing at the PAI loci (Bartee et al, 2001; Malagnac et al, 2002). The cmt3 and kyp mutants show a loss of methylation, primarily at CNG sites, and cause reactivation of the expression of a subset of retrotransposons. We found that kyp mutants, but not cmt3 mutants, show major losses of H3K9 methylation at affected loci, suggesting that H3K9 methylation acts upstream to control CNG DNA methylation (Johnson et al, 2002). Here we address possible mechanisms by which H3K9 methylation targets CMT3 to methylate CNG sites. We previously found that, unlike the chromodomain of HP1, the chromodomain of CMT3 did not bind to K9 methylated histone H3 peptides. We therefore proposed a model in which CMT3 was indirectly tethered to methylated histones by the plant HP1 homolog, LHP1 (Gaudin et al, 2001; Jackson et al, 2002). However, our genetic analyses of lhp1 mutants, as well as that previously published (Malagnac et al, 2002), do not support a role for LHP1 in the control of DNA methylation. Instead, we find that the CMT3 chromodomain can indeed interact with methylated histone H3 tails, but only if they are simultaneously methylated at H3K9 and H3K27. Furthermore, we show that H3K9 and H3K27 methylations are both enriched at silent loci where CMT3 acts. This suggests that a dual methylation mark is required for CMT3-dependent CNG methylation. Interestingly, the H3K27 methylation present at heterochromatin is not affected by mutations in KYP or in several Arabidopsis PcG related genes including the E(Z) homologs, suggesting that a novel pathway controls heterochromatic H3K27 methylation. Results LHP1 is not required for CMT3 activity in vivo We previously showed that the Arabidopsis HP1 homolog, LHP1 (Gaudin et al, 2001), binds methylated H3K9 (Jackson et al, 2002). Furthermore, we showed that LHP1 binds to CMT3 in vitro. We therefore proposed a model in which KYP methylates H3K9, which results in binding of LHP1 and the subsequent recruitment of CMT3 to methylated chromatin (Jackson et al, 2002). This model predicts that loss-of-function alleles of LHP1 should resemble loss-of-function alleles of kyp and cmt3. To test this, we obtained two recessive alleles of lhp1 including a complete deletion (LHP1 is the only HP1 homolog in Arabidopsis). These alleles were previously described as the terminal flower2 mutants, and show a number of interesting developmental abnormalities including early flowering and inflorescences that prematurely terminate in floral structures (Larsson et al, 1998; Kotake et al, 2003). Both tfl2 alleles have the same phenotype, and are here referred to as lhp1-3 and lhp1-4. Using Southern blot analysis with methylation-sensitive endonucleases, we tested several methylated DNA sequences known to be affected by kyp and cmt3 for a loss of methylation in the lhp1 mutants. We have previously shown that CMT3 methylates heterochromatic regions such as the pericentromeric Ta3 retrotransposon and centromeric 180 bp (CEN) repeats, as well as the silent FWA and SUP loci, which are present in otherwise euchromatic regions. The lhp1 alleles did not affect DNA methylation at either CG or CNG at any of the loci tested (Figure 1A–C). We also found that two retrotransposon related sequences (Ta3 and TSI) that show reactivation of expression in the cmt3 and kyp mutants are not reactivated in the lhp1 mutants (Figure 1D) (Jackson et al, 2002). Figure 1.lhp1 mutants do not affect DNA methylation or gene silencing in heterochromatin. (A–C) Southern blot analysis with the DNA methylation-sensitive endonuclease CfoI and HpaII for CG sites, and BglII and MspI for CNG sites, probed with centromeric 180 bp repeat (CEN), Ta3, and FWA. fwa-1 is a mutant lacking methylation of the FWA gene. cmt3-7 (included as a control) reduces CNG methylation, and therefore shows a downshifted banding pattern. (D) Semiquantitative RT–PCR showing reactivation of retroelement sequences in cmt3 but not in lhp1 mutants. (E) Immunohistochemical staining of wild-type and lhp1 interphase nuclei with dimethyl H3K9 antibodies preferentially labeling heterochromatic chromocenters. The lhp1-4 mutant does not affect distribution of methyl H3K9. (F) Floral phenotype of the clark kent-st (clk-st) lhp1-4 double mutant. The left flower, a clk-st mutant, shows supernumerary stamens and a defective gynoecium. The middle flower, an lhp1-4 mutant, shows a terminal flower with three normal gynoecia. The right flower, a clk-st lhp1-4 double mutant, shows a compound terminal flower containing two clk-st-like defective gynoecia. Download figure Download PowerPoint In fission yeast, it has been shown that spreading and distribution of methyl H3K9 are dependent on Swi6 (the yeast HP1 homolog) (Hall et al, 2002). We therefore tested if lhp1 mutants altered dimethyl H3K9 distribution in Arabidopsis. The lhp1 mutant did not show a change in the distribution of dimethyl H3K9, since staining was still clearly localized to heterochromatic chromocenters (Figure 1E). CMT3 and KYP are important in maintaining non-CG methylation and gene silencing of the SUP locus in a stably hypermethylated superman allele, clk-st. To analyze if LHP1 is involved in this process, we introduced the lhp1-4 mutant into the clk-st strain and looked for suppression of the superman floral phenotype. We found that, unlike the cmt3 and kyp mutants, the lhp1 mutants do not suppress silencing of clk-st (Figure 1F). Thus our genetic analysis does not support a role for LHP1 in recruitment of CMT3. These results also agree with data showing that a weak lhp1 allele (Gaudin et al, 2001) does not reduce DNA methylation of the PAI loci or of the CEN repeats (Malagnac et al, 2002). We therefore conclude that LHP1 is unlikely to play an important role in the recruitment of CMT3 to methylated chromatin. Binding of CMT3 to the doubly methylated N-terminus of histone H3 We revisited the hypothesis that CMT3 might bind directly to methylated histones through its chromodomain. The CMT3 chromodomain sequence is closely related to those of HP1 and Pc that bind to methylated H3K9 and H3K27, respectively (Figure 2A and B). Analyses of the structures and thermodynamics of the HP1 (Jacobs and Khorasanizadeh, 2002; Nielsen et al, 2002) and Pc chromodomains (Fischle et al, 2003b; Min et al, 2003) in complex with methylated H3 tails have shown that each of these chromodomains makes a 1:1 complex with its target histone H3 tail. Further, both HP1 and Pc use an aromatic cage involving three conserved aromatic residues to recognize the methyllysine side chain in the H3 tail. The CMT3 chromodomain also contains aromatic residues located at the same sequence positions (F382, W409, and Y412; Figure 2B). Figure 2.Interaction of the CMT3 chromodomain with the N-terminal tail of histone H3. (A) Domain composition of CMT3. Roman numerals denote the conserved motifs within the catalytic DNA methyltransferase region. (B) Alignment of the chromodomains of CMT3 and ZMET2 (a functional CMT gene from Zea mays; Papa et al, 2001) proteins with those of the HP1, LHP1, and Pc proteins. Secondary structure elements drawn above the sequence correspond to the structure of the HP1 chromodomain. Positions F382, W409, and Y412, highlighted in red, show the sites of conserved aromatic residues that form an aromatic cage. (C) Fluorescence polarization-based binding results for the interaction of the CMT3 chromodomain with H3 peptides. The CMT3 chromodomain binds to a doubly methylated H3K9/K27 peptide, but not to either the unmodified or singly methylated peptides. Binding was eliminated by mutation of the conserved aromatic residue F382 to alanine. (D) Binding of CMT3 chromodomain to methyl H3K9/K27 as measured by ITC. The solid line corresponds to the fit of the integrated heats of injections to an interaction stoichiometry of 0.5; see the inset for the thermodynamic parameters of the best fit of the data using the MicroCal software. Download figure Download PowerPoint Using fluorescence anisotropy binding assays, we measured the affinity of the CMT3 chromodomain for various histone H3 tail peptides methylated at K9, K27, or both. We found no appreciable binding to peptides with either methylation mark alone. However, when both methylations were present in the same histone H3 peptide, we found a strong interaction (KD=0.9±0.1 μM) (Figure 2C). This mode of binding suggests that the CMT3 chromodomain may form two independent binding sites for each methyllysine, or alternatively form a homodimer for binding to the doubly methylated H3 tail in a 2:1 complex. To establish more thoroughly the thermodynamics and stoichiometry of this interaction, we carried out isothermal titration calorimetry (ITC) and found that the molar ratio of the CMT3 chromodomain bound to the doubly methylated peptide is 0.5±0.1, with the remaining parameters measured as follows: KD=9±1 μM, ΔH=−8.8 kcal/mol, ΔG=−6.5 kcal/mol, and T ΔS=−2.3 kcal/mol. This indicates that the association of two CMT3 chromodomains as a homodimer allows the high-affinity interaction with the doubly methylated histone H3 tail (Figure 2D). The KD values measured by fluorescence and ITC for CMT3 binding to the doubly methylated H3 tail are in the range of 0.9–9 μM. Importantly, these KD values are similar to those measured previously for HP1 and Pc chromodomains binding to their respective singly methylated H3 tails (4–5 μM) (Fischle et al, 2003b). To test the role of the putative aromatic cage of CMT3 in binding methylated lysines, we mutated residue F382 to alanine and then tested this mutant chromodomain using fluorescence anisotropy binding studies (Figure 2C). This mutation resulted in a complete defect in binding to the doubly methylated H3 tail, confirming the importance of the predicted aromatic cage residues of the CMT3 chromodomain in target methyllysine binding. Together, these data suggest that while the same aromatic cage in HP1, Pc, and CMT3 chromodomains may be required for target recognition, the CMT3 chromodomain is unique in its ability to cooperate via two subunits to bind to a doubly methylated H3 tail. As such, the chromodomain of CMT3 is reminiscent of the chromo shadow domain of HP1, which also forms a homodimer (Thiru et al, 2004). Since homodimerization does not occur with HP1 or Pc chromodomains despite their overall high sequence similarity with CMT3, we suggest that the six-residue insertions within the β-strand 1 segment of the chromodomain (Figure 2B), or nonconserved residues within the CMT3 chromodomain, contribute to a dimerization interface of this protein. Interestingly, the maize homolog, ZMET2 (Papa et al, 2001), also contains a similar insertion suggesting a related function for another plant chromodomain that resides within a DNA methyltransferase fold. Colocalization of methyl H3K9 and K27 at heterochromatin The in vitro CMT3 binding data suggest a model in which CMT3 targeting to chromatin involves a combinatorial double methylation signal. This model predicts that H3K27 methylation should be localized to loci controlled by CMT3. To test this, we performed chromatin immunoprecipitation (ChIP) experiments using antibodies to mono-, di-, and trimethyl H3K27 (Peters et al, 2003; Perez-Burgos et al, 2004), looking for enrichment at silent CMT3-controlled loci. We tested two loci found in pericentromeric heterochromatin that we previously found were affected in cmt3 mutants, the Ta3 retrotransposon and the 180 bp CEN repeat (Lindroth et al, 2001; Johnson et al, 2002). Both loci showed enrichment of mono- and dimethyl H3K27 over ACTIN, a euchromatic gene used for normalization (Figure 3A). The monomethyl H3K27 signal was stronger than the dimethyl signal, and we did not observe trimethyl signal over the no antibody controls. These results suggest that both H3K9 and H3K27 methylations are enriched at DNA methylated and silent loci in pericentromeric heterochromatin. Interestingly, while the kyp mutation reduced the H3K9 methylation at these loci, it did not affect the H3K27 methylation (Figure 3A). Figure 3.Colocalization of methyl H3K9 and H3K27 at heterochromatin. (A) ChIP analysis showing methyl H3K9 and H3K27 enrichment at the Ta3 retrotransposon and CEN repeats, relative to ACTIN. kyp affects levels of methyl H3K9 but not mono-, di-, or trimethyl H3K27. ACTIN is used for normalization, and the amount of chromatin used in each sample (for both the Ta3 and CEN experiments) was adjusted so that an equal amount of ACTIN was amplified. WCE, whole-cell extract. (B) Distribution of mono-, di-, and trimethyl H3K27 in interphase leaf nuclei. Mono- and dimethyl H3K27 are enriched at DAPI-stained chromocenters, while trimethyl H3K27 signals are more evenly distributed. kyp reduces dimethyl H3K9 at heterochromatin, but not H3K27 methylation. (C) ChIP analysis showing enrichment of both H3K9 and H3K27 methylation at SUPERMAN and FWA relative to ACTIN. Download figure Download PowerPoint To assess whether H3K27 methylation is generally associated with heterochromatin, we assayed the distribution of methyl H3K27 in different nuclear compartments using immunohistochemical staining of interphase nuclei (Figure 3B). We utilized the same mono-, di-, and trimethyl H3K27 antibodies to look for colocalization with DAPI staining chromocenters where most of the heterochromatin resides. H3K27m preferentially localized to chromocenters with very little staining in euchromatin. H3K27m2 localized strongly to chromocenters and to a lesser extent to euchromatin. Trimethyl staining was observed in small speckles, not coinciding with DAPI-stained chromocenters. While the level of H3K9m2 at chromocenters was reduced in the kyp mutant, all staining patterns with the H3K27 methyl antibodies were unaffected (Figure 3B). These results agree with the ChIP analyses and show that KYP is not responsible for the H3K27 methylation at silent heterochromatin. Finally, we used ChIP analysis to access the levels of H3K27 methylation at two CMT3-controlled loci that are silent but reside in otherwise euchromatic regions of the genome, SUPERMAN and FWA (Figure 3C). We found that both H3K9 and H3K27 methylations were enriched at these loci. And again, while the kyp mutation reduced the H3K9 methylation, it did not reduce the levels of H3K27 methylation. Histone methyltransferases affecting H3K27 methylation It is currently unknown which genes in Arabidopsis are responsible for methylating H3K27 sites in heterochromatin. Our findings predict that mutations in these genes should reduce CMT3 recruitment to chromatin, and thus should mimic the kyp and cmt3 mutants (reducing CNG methylation). In Arabidopsis, the most obvious candidate genes are the PcG related E(Z) homologs, since animal E(Z) proteins are known to methylate H3K27 (Cao et al, 2002; Czermin et al, 2002; Kuzmichev et al, 2002; Muller et al, 2002). The three plant E(Z) genes are MEA, CLF, and SWN. MEA is paternally imprinted and the maternal copy is only expressed in the endosperm and not in the rest of the plant (Reyes and Grossniklaus, 2003). It is therefore very unlikely that MEA is involved in global H3K27 methylation in heterochromatin. The other two genes, CLF and SWN, are widely expressed, and are potentially good candidates for global H3K27 methyltransferases (Chanvivattana et al, 2004). We therefore generated a double loss-of-function clf and swn mutant. The clf-50 allele is a complete deletion of the CLF locus, and the swn-3 allele is a T-DNA insertion upstream of the catalytic SET domain (Chanvivattana et al, 2004). The clf-50 swn-3 double mutant is viable but shows severe developmental abnormalities (Chanvivattana et al, 2004). Another Arabidopsis PcG member is FIE, a homolog of the Drosophila WD-40 protein Extra sex combs (ESC). Since FIE is a single-copy gene in Arabidopsis, fie mutants are likely to affect all H3K27 methylation targeted by E(Z) protein-containing complexes (Reyes and Grossniklaus, 2003). Although fie null mutants are embryonic lethal, likely due to the role of FIE in imprinting and early endosperm development, a transgenic allele was isolated, fie TK114, in which FIE is expressed only during very early embryo development, but not during later seedling development (Kinoshita et al, 2001). This allele causes ectopic expre