Plant Cell-2006-Ebbs-1166-76 (4)

Locus-Speci?c Control of DNA Methylation by the Arabidopsis SUVH5Histone Methyltransferase W

Michelle L.Ebbs and Judith Bender1

Department of Biochemistry and Molecular Biology,Johns Hopkins University Bloomberg School of Public Health, Baltimore,Maryland21205

In Arabidopsis thaliana,heterochromatin formation is guided by double-stranded RNA(dsRNA),which triggers methylation of histone H3at Lys-9(H3mK9)and CG plus non-CG methylation on identical DNA sequences.At heterochromatin targets including transposons and centromere repeats,H3mK9mediated by the Su(var)3-9homologue4(SUVH4)/KYP histone methyltransferase(MTase)is required for the maintenance of non-CG methylation by the CMT3DNA MTase.Here,we show that although SUVH4is the major H3K9MTase,the SUVH5protein also has histone MTase activity in vitro and contributes to the maintenance of H3mK9and CMT3-mediated non-CG methylation in vivo.Strikingly,the relative contributions of SUVH4,SUVH5,and a third related histone MTase,SUVH6,to non-CG methylation are locus-speci?c.For example,SUVH4 and SUVH5together control transposon sequences with only a minor contribution from SUVH6,whereas SUVH4and SUVH6 together control a transcribed inverted repeat source of dsRNA with only a minor contribution from SUVH5.This locus-speci?c variation suggests different mechanisms for recruiting or activating SUVH enzymes at different heterochromatic sequences.The suvh4suvh5suvh6triple mutant loses both monomethyl and dimethyl H3K9at target loci.The suvh4suvh5 suvh6mutant also displays a loss of non-CG methylation similar to a cmt3mutant,indicating that SUVH4,SUVH5,and SUVH6together control CMT3activity.

INTRODUCTION

Correct maintenance of heterochromatin patterns is essential for gene regulation and genome stability in eukaryotes(reviewed in Lippman and Martienssen,2004;Matzke and Birchler,2005).In many eukaryotes,including?ssion yeast,Drosophila,mammals, and plants,heterochromatin is marked by histone H3at Lys-9 (H3mK9).In some species,including mammals and plants, heterochromatin is also marked by cytosine methylation.A major mechanism for guiding H3K9and cytosine methylation to appropriate regions of the genome involves double-stranded RNA(dsRNA)–derived species,such as small RNA products of dicer ribonuclease cleavage.For example,in?ssion yeast,cen-tromere repeat dsRNA is diced into small RNAs,which assemble into the RNA-induced initiation of transcriptional gene silencing effector complex to guide H3mK9to centromere repeat regions (Volpe et al.,2003;Verdel et al.,2004).In a potentially related mechanism,small RNAs transfected into human cancer cells can trigger H3mK9on identical promoter sequences,resulting in transcriptional silencing of a downstream gene(Ting et al.,2005). Furthermore,in plants,dsRNA-derived species generated from RNA viruses,transcribed inverted repeats,or products of RNA-dependent RNA polymerases can trigger both H3K9and cytosine methylation on identical DNA sequences(reviewed in Mathieu and Bender,2004).However,effector complexes that mediate the connection between RNA and heterochromatin modi?cations have not yet been characterized in mammalian or plant systems. Plant RNA-directed DNA methylation occurs in both CG and non-CG contexts.Genetic studies in Arabidopsis thaliana have implicated three structurally distinct cytosine methyltransferases (MTases)in this process:the DRM1/DRM2cytosine MTases initiate new DNA methylation imprints in response to a dsRNA signal,whereas the MET1and CMT3cytosine MTases maintain methylation in CG and non-CG contexts,respectively(reviewed in Mathieu and Bender,2004).The DRM cytosine MTases also contribute to the maintenance of non-CG methylation at some target regions,including an Arabidopsis SINE transposon,a direct repeat array MEA-ISR,and transgene reporters for RNA-directed DNA methylation(Cao and Jacobsen,2002;Cao et al.,2003).However,CMT3is the major MTase that maintains non-CG methylation at centromere repeats and transposons,in-cluding DNA transposons and long terminal repeat(LTR)retro-transposons(Bartee et al.,2001;Lindroth et al.,2001;Tompa et al.,2002;Kato et al.,2003;Lippman et al.,2003).

CMT3-mediated non-CG methylation depends on the Su-(var)3-9homologue4(SUVH4)/KYP(hereafter referred to as SUVH4)H3K9MTase,suggesting that CMT3is guided to target sequences by the H3mK9modi?cation(Jackson et al.,2002; Malagnac et al.,2002).Similarly,H3mK9guides cytosine meth-ylation in the fungus Neurospora crassa(Tamaru and Selker, 2001;Tamaru et al.,2003)and in mouse(Lehnertz et al.,2003;Xin et al.,2003).However,at target loci including centromere re-peats and transposons,suvh4mutations confer a weaker loss of non-CG methylation than cmt3mutations(Jackson et al.,2002;

1To whom correspondence should be addressed.E-mail jbender@

https://www.360docs.net/doc/e45566043.html,;fax410-955-2926.

The author responsible for distribution of materials integral to the

?ndings presented in this article in accordance with the policy described

in the Instructions for Authors(https://www.360docs.net/doc/e45566043.html,)is:Judith Bender

(jbender@https://www.360docs.net/doc/e45566043.html,).

W Online version contains Web-only data.

Article,publication date,and citation information can be found at

https://www.360docs.net/doc/e45566043.html,/cgi/doi/10.1105/tpc.106.041400.

The Plant Cell,Vol.18,1166–1176,May2006,https://www.360docs.net/doc/e45566043.html,a2006American Society of Plant Biologists

Malagnac et al.,2002;Lippman et al.,2003).The Arabidopsis genome encodes eight other SUVH putative H3K9MTases (Baumbusch et al.,2001),suggesting that some of these other SUVHs might contribute to the CMT3non-CG methylation pathway. SUVH6has similar in vitro H3K9MTase activity to SUVH4 (Jackson et al.,2004).Therefore,we previously characterized the effects of a suvh6mutation on H3K9and non-CG methylation at CMT3targets,including transposons,centromere repeats,and the endogenous phosphoribosylanthranilate isomerase(PAI)Trp biosynthetic genes(Ebbs et al.,2005).In the Wassilewskija(Ws) ecotype of Arabidopsis,the PAI genes are arranged as a tail-to-tail inverted repeat of two genes,PAI1-PAI4,and two unlinked singlet genes,PAI2and PAI3.Transcription through PAI1-PAI4 from a fortuitous unmethylated promoter upstream of PAI1pro-duces normally polyadenylated PAI1transcripts and longer spe-cies that read into palindromic PAI4sequences to form dsRNA signals for H3K9and cytosine methylation of PAI sequences (Melquist and Bender,2003;Ebbs et al.,2005).Mutations in CMT3strongly reduce non-CG methylation at PAI1-PAI4,PAI2, and PAI3(Bartee et al.,2001).By contrast,mutations in SUVH4 reduce non-CG methylation on PAI2and PAI3but have no effect on methylation patterning at PAI1-PAI4(Malagnac et al.,2002). We found that double mutation of suvh4and suvh6strongly reduced H3K9and non-CG methylation on PAI1-PAI4,indicat-ing that SUVH4and SUVH6act together to maintain epigenetic modi?cations at this transcribed inverted repeat locus(Ebbs et al.,2005).However,suvh4suvh6retained residual PAI1-PAI4 non-CG methylation relative to cmt3.Furthermore,the suvh6 mutation did not enhance partial demethylation of the Ta3LTR retrotransposon,the Mu1DNA transposon,or centromere re-peats conferred by the suvh4mutation relative to cmt3.Here,we show that SUVH5also has H3K9MTase activity in vitro and acts in the CMT3non-CG methylation pathway in vivo.The suvh4 suvh5suvh6triple mutant displays similar strong demethylation of target sequences to a cmt3mutant at the PAI genes,Ta3and Mu1transposons,and centromere repeats,indicating that SUVH4,SUVH5,and SUVH6together control CMT3-mediated non-CG methylation at these loci.

Importantly,SUVH4,SUVH5,and SUVH6make different rel-ative contributions to non-CG methylation at different loci.For example,although SUVH4and SUVH6control the majority of non-CG methylation at the PAI1-PAI4transcribed inverted re-peat,SUVH4and SUVH5control the majority of non-CG meth-ylation at the Ta3and Mu1transposons.These results suggest locus-speci?c variation in the factors that recruit or activate SUVH enzymes at RNA-directed DNA methylation targets.Thus, the Arabidopsis SUVHs can potentially be exploited for genetic and biochemical identi?cation of novel components that connect heterochromatin-modifying enzymes with dsRNA signals. RESULTS

SUVH5Has in Vitro H3K9MTase Activity

In previous work,we determined that a suvh4suvh6double H3 K9MTase mutant retains residual non-CG methylation on the PAI genes,transposon sequences,and centromere repeats relative to a cmt3cytosine MTase mutant(Ebbs et al.,2005).To identify H3K9MTases that might account for this residual non-CG methylation,we tested the other seven SUVH proteins encoded in Arabidopsis(Baumbusch et al.,2001)for H3K9MTase activity in vitro.The SUVH proteins carry a C-terminal catalytic domain (pre-SET,SET,and post-SET motifs),a central conserved YDG domain,and divergent N termini.To assess catalytic activity,we expressed and puri?ed each SUVH protein as a glutathione S-transferase(GST)fusion to the catalytic domain.These recom-binant proteins were incubated with a puri?ed bovine histone mix (H1,H2A,H2B,H3,and H4)and S-adenosyl-[14C-methyl]L-Met as the methyl group donor(see Methods).Only SUVH4,SUVH5, and SUVH6methylated bovine H3(conserved with Arabidopsis H3;Gendrel et al.,2002)under these conditions(Figure1A). SUVH5also methylated bovine H2A(Figure1A)and the Arabi-dopsis H2A variants HTA2,HTA7,and HTA13in vitro(see Supplemental Figure1online).Mutational analysis of an HTA13 substrate showed that either of two N-terminal lysines could be modi?ed by SUVH5(see Supplemental Figure1online).SUVH4 did not methylate bovine H2A either in a mix of histones or with H2A as the sole substrate,and it did not methylate Arabidopsis HTA13in vitro(Figure1A;data not shown).

To determine the speci?city of SUVH5on H3,we tested for activity against a series of recombinant GST-H3peptide sub-strates(residues1to57):wild type(H3N),a mutant with only Lys-4 (N4),a mutant with only Lys-9(N9),a mutant with only Lys-27 (N27),or a mutant with no Lys residues(NT)(Tachibana et al., 2001).SUVH5methylated only the H3N and N9substrates (Figure1B),indicating speci?city for H3K9.

Histone MTases can add one,two,or three methyl groups to a substrate Lys side chain.Structural analysis of a known histone mono-MTase,the mammalian SET7/9enzyme,versus a known histone tri-MTase,the Neurospora DIM-5enzyme,revealed that this difference in activity can be attributed to a conserved active site position that carries a Tyr in the mono-MTase versus a less bulky Phe in the tri-MTase(Zhang et al.,2003;Collins et al., 2005).All of the Arabidopsis SUVH proteins carry a Tyr at this position and thus are predicted to be mono-or di-MTases but not tri-MTases.Consistent with this structural prediction,both SUVH4and SUVH6have been shown to add one or two but not three methyl groups to the target Lys on a peptide substrate in vitro(Jackson et al.,2004).Furthermore,immunoblot,immuno-cytology,and chromatin immunoprecipitation(ChIP)experiments performed with antibodies speci?c for monomethylated,dime-thylated,or trimethylated H3K9suggest that Arabidopsis pref-erentially uses dimethylated H3K9as a mark for heterochromatin formation and is de?cient in trimethylated H3K9(Jackson et al., 2004).

To test the MTase properties of SUVH5relative to SUVH4,we engineered tri-MTase mutant variants of these enzymes with the conserved Tyr mutated to Phe(SUVH4Y591F and SUVH5Y761F) and assayed their ability to add a third methyl group to an H3 dimethyl K9peptide substrate.Both mutant enzymes methyl-ated the dimethylated substrate,demonstrating that they have tri-MTase activity,in contrast with the corresponding wild-type enzymes,which did not methylate this substrate(Figure1C).The tri-MTase mutant enzymes maintained speci?city for H3K9 similar to the wild-type enzymes.These results show that SUVH4

SUVH5Controls DNA Methylation1167

and SUVH5each can be converted to a tri-MTase by a single mutation,without disruption of overall enzyme function.These results also suggest that,like SUVH4and SUVH6,SUVH5is a mono-/di-MTase.Interestingly,besides its primary MTase ac-tivity against H3,the SUVH4Y591F mutant displayed low-level secondary MTase activity against one of the other histones in the

bovine histone substrate mix (Figure 1C).However,we did not further characterize this secondary activity.SUVH5Contributes to PAI Non-CG Methylation and Transcriptional Silencing in the Absence of SUVH4and SUVH6

To understand the role of SUVH5in the plant,we analyzed two suvh5insertion alleles.The suvh5-1allele is a T-DNA insertion into the catalytic domain–encoding region of the gene isolated in the Ws ecotype,and the suvh5-2allele is a T-DNA insertion into the catalytic domain–encoding region of the gene isolated in the Columbia (Col)ecotype.Both alleles were crossed into the Ws pai1reporter background (see below)as single mutations,dou-ble mutations with suvh4,or triple mutations with suvh4and suvh6.Mutant plants were assayed for changes in non-CG methylation and H3mK9at representative heterochromatic loci targeted by CMT3,including the PAI Trp biosynthetic genes,the Ta3LTR retrotransposon,and the Mu1DNA transposon.PAI and transposon sequences were also tested for transcriptional reactivation.

In wild-type Ws,the PAI1gene in the PAI1-PAI4inverted repeat provides the major source of PAI enzyme,as a result of expres-sion from a fortuitous upstream unmethylated promoter.To monitor the transcriptional activity of the functional but silenced PAI2,we isolated a pai1missense mutation that reduces PAI1enzyme activity without affecting the dsRNA signal for PAI DNA methylation (Bartee and Bender,2001).pai1displays a number of PAI-de?cient phenotypes,including blue ?uorescence under UV light.These phenotypes can be alleviated by mutations that release the transcriptional silencing of PAI2,for example suvh4and cmt3(Bartee et al.,2001;Malagnac et al.,2002).The suvh5-1and suvh5-2alleles were crossed into the Ws pai1background so that we could monitor their effects on PAI2transcriptional silenc-ing and PAI DNA methylation.

In the blue ?uorescence assay for PAI2silencing,the suvh5mutation did not affect ?uorescence in either the pai1or pai1suvh4background relative to the parental background (Figure 2),as observed previously for the suvh6mutation (Ebbs et al.,2005).Both pai1and pai1suvh5displayed strong seedling ?uores-cence,and the pai1suvh4,pai1suvh4suvh5,and pai1suvh4suvh6mutants all displayed reduced but still detectable seed-ling ?uorescence.By contrast,pai1suvh4suvh5suvh6seed-lings were non?uorescent,similar to pai1cmt3seedlings.These ?uorescence phenotypes suggest that SUVH5makes a small contribution to PAI2silencing in the absence of SUVH4and SUVH6and that SUVH6makes a small contribution to PAI2silencing in the absence of SUVH4and SUVH5.The pai1suvh5,pai1suvh4suvh5,and pai1suvh4suvh5suvh6mutants did not display any obvious morphological defects beyond those con-ferred by pai1.In addition,the suvh5-1and suvh5-2alleles did not confer obvious morphological defects in the wild-type Ws and Col backgrounds,respectively.

In previous work,we determined that non-CG methylation and H3mK9of the PAI1-PAI4inverted repeat are controlled by the combined action of SUVH4and SUVH6,such that this locus is demethylated in the suvh4suvh6double mutant but not in the suvh4and suvh6single mutants (Ebbs et al.,2005).However,

the

Figure 1.SUVH5Has in Vitro Histone MTase Activity.

Coomassie blue–stained gels (top panels)and their ?uorograms (bottom panels)are shown.GST-SUVH indicates the positions of full-length recombinant SUVH proteins,and asterisks indicate GST-SUVH trunca-tions.Positions of protein molecular mass markers in kilodaltons are shown at left.

(A)In vitro histone MTase assays were performed with a whole histone mix and each of the nine GST-tagged SUVH proteins.

(B)Speci?city of SUVH5for H3K9.GST-SUVH5was incubated with a bovine histone mix or a GST-H3N-terminal peptide as indicated:H3N is the wild type,N4has K9and K27mutated to Arg,N9has K4and K27mutated to Arg,N27has K4and K9mutated to Arg,and NT has K4,K9and K27mutated to Arg (Tachibana et al.,2001).GST alone was incubated with the histone mix as a control.

(C)Tyr-to-Phe mutations convert SUVH4or SUVH5from an H3K9mono-/di-MTase to an H3K9tri-MTase.GST-SUVH4Y591F,GST-SUVH4,GST-SUVH5Y761F,and GST-SUVH5were assayed for activity on bovine histone mix (mix),GST-H3peptides,or a biotinylated H3K9dimethyl peptide (diMe).

1168The Plant Cell

suvh4suvh6mutant retains residual non-CG methylation on PAI1-PAI4relative to cmt3.

To determine the effects of the suvh5mutation on PAI DNA methylation patterning,we performed both DNA gel blot and bisul?te genomic sequencing assays.For DNA gel blot analysis,we used three methylation-sensitive enzymes with cleavage sites within methylated PAI sequences:Hpa II (sensitive to meth-ylation of either cytosine in 59-CCGG-39),Msp I (sensitive to methylation of only the outer CCG cytosine in 59-CCGG-39),and Hin cII (sensitive to methylation of the outermost cytosines in 59-atGTCAACag-39,where the enzyme recognition sequence is shown in uppercase).In these assays,DNA samples isolated from the suvh5single mutant or the suvh4suvh5double mutant displayed cleavage patterns similar to those of the parental wild type or suvh4,respectively (Figures 3A and 3B).However,Msp I and Hin cII assays revealed that the residual non-CG methylation present on PAI1-PAI4in the suvh4suvh6mutant relative to cmt3was lost in suvh4suvh5suvh6.Thus,at PAI1-PAI4,SUVH5makes a small contribution to the maintenance of non-CG methylation in the absence of SUVH4and SUVH6.

Bisul?te sequencing analysis of the proximal promoter regions of PAI1and PAI2showed that Ws and suvh5carried similar patterns of CG plus non-CG methylation at PAI1and PAI2,suvh4and suvh4suvh5carried similar patterns of CG plus non-CG

methylation at PAI1but mainly CG methylation at PAI2,and cmt3,suvh4suvh6,and suvh4suvh5suvh6carried similar patterns of mainly CG methylation at PAI1and PAI2(see Sup-plemental Figure 2online).Overall,these patterns are consistent with the PAI DNA gel blot DNA methylation assay results (Figure 3),although the subtle differences in residual non-CG methyla-tion between suvh4suvh6and suvh4suvh5suvh6were not within the resolution of the bisul?te sequencing assay.

We used ChIP analysis of the PAI genes to monitor dimethyl H3K9(H32mK9)and dimethyl H3K4(H32mK4),a histone mod-i?cation associated with transcriptional activity (Gendrel et al.,2002;Lippman et al.,2004),in suvh mutant backgrounds.Each PAI locus was monitored with a gene-speci?c PCR primer pair:the PAI1-PAI4locus was ampli?ed at the junction between PAI1and PAI4,PAI2was ampli?ed across a central intron–exon boundary,and PAI3was ampli?ed across a central intron–exon boundary (see Supplemental Table 1online)(Ebbs et al.,2005).The suvh5single mutant displayed H3methylation patterns similar to those of wild-type Ws,with all three PAI loci enriched for H32mK9but not H32mK4(Figure 3C;see Supplemental Figure 3online).The suvh4suvh5double mutant displayed H3methylation patterns similar to those of suvh4:H32mK9was maintained at PAI1-PAI4but lost from PAI2and PAI3,and PAI2and PAI3gained H32mK4.The suvh4suvh5suvh6triple mutant displayed H3methylation patterns similar to those of suvh4suvh6:H32mK9was lost from all three PAI loci,and PAI2and PAI3gained H32mK4.Overall,the PAI H32mK9patterns were consistent with the PAI non-CG methylation patterns (Figure 3).Furthermore,the gain of H32mK4on PAI2in suvh4single and multiple mutants (Figure 3C)was consistent with patterns of PAI2transcriptional activation revealed by ?uorescence phenotypes in the pai1reporter background (Figure 2).

Together,the PAI2silencing phenotypes and the PAI non-CG methylation patterns (Figures 2and 3)suggest that SUVH4,SUVH5,and SUVH6all act at the PAI genes,with the hierarchies SUVH4>SUVH5?SUVH6at the silenced PAI2and PAI3target loci and SUVH4?SUVH6>SUVH5at the PAI1-PAI4transcribed inverted repeat.

We also assayed DNA methylation patterning in a suvh5suvh6double mutant at the PAI genes as well as the transposons Ta3and Mu1,using DNA gel blot assays,and found similar patterns to the wild type at all loci tested (see Supplemental Figure 4online).These results support the view that SUVH4is the major H3K9MTase involved in the CMT3DNA methylation pathway.SUVH5Contributes to Ta3and Mu1Transposon Non-CG Methylation and Transcriptional Silencing in the Absence of SUVH4

The Ta3and Mu1transposons are transcriptionally silent elements that carry CG and non-CG methylation and H3mK9in wild-type backgrounds (Johnson et al.,2002;Lippman et al.,2003).Ta3is a single-copy element (Konieczny et al.,1991),whereas Mu1is one of a group of related elements (Singer et al.,2001)with three sequences detected by DNA gel blot analysis in the Ws ecotype (Ebbs et al.,2005).Both Ta3and Mu1display a partial loss of non-CG methylation in suvh4and a stronger loss of non-CG methylation in cmt3(Jackson et al.,2002;Lippman et al.,

2003).

Figure 2.The suvh4suvh5suvh6Mutant Shows Similar PAI2Transcrip-tional Reactivation to a cmt3Mutant in the pai1Reporter Background.Representative 2-week-old seedlings are shown photographed under visible light (right row)or short-wave UV light (left row).All mutations assayed were in the Ws pai1background,with WT indicating wild type,4indicating suvh4R302*,5indicating suvh5-1,6indicating suvh6-1,and cmt3indicating cmt3illa (Bartee et al.,2001).

SUVH5Controls DNA Methylation 1169

DNA gel blot analysis of Ta3or Mu1using an Msp I digest that monitors DNA methylation in the non-CG context 59-CCG-39revealed that the suvh5single mutant showed similar inhibited cleavage patterns to wild-type Ws,diagnostic of dense CCG methylation (Figures 4A and 4B).By contrast,suvh4suvh5dis-played increased transposon Msp I cleavage,diagnostic of re-duced CCG methylation relative to suvh4or suvh4suvh6.Similar enhanced transposon cleavage patterns were conferred by the suvh5-2allele in the suvh4background (see Supplemental Figure 4online).These results indicate that in the absence of SUVH4,SUVH5controls residual CCG methylation at the Ta3element and at Mu1duplications.

The suvh4suvh5suvh6triple mutant displayed a more com-plete Msp I cleavage pattern than suvh4suvh5at both Ta3and Mu1,diagnostic of a loss of residual CCG methylation (Figures 4A and 4B).In both cases,the Msp I digestion pattern was similar to that of the cmt3non-CG MTase.These results suggest that SUVH6makes a small contribution to the maintenance of non-CG methylation at Ta3and Mu1in the absence of SUVH4and SUVH5.

We used ChIP analysis to monitor H32mK9and H32mK4at Ta3and Mu1in suvh mutant backgrounds.The suvh5single mutant maintained H32mK9similar to wild-type Ws at both transposons (Figure 4C;see Supplemental Figure 3online),consistent with the maintenance of full DNA methylation at these sequences (Figures 4A and 4B).The suvh4mutation reduced Ta3and Mu1H32mK9,such that we could not determine within the sensitivity of the ChIP assay whether this H32mK9depletion was enhanced in the suvh4suvh5and the suvh4suvh5suvh6back-grounds (Figure 4C;see Supplemental Figure 3online).However,at Ta3,the suvh4suvh5and suvh4suvh5suvh6mutants dis-played a gain of H32mK4,diagnostic of transcriptional activa-tion.Furthermore,semiquantitative RT-PCR analysis showed that Ta3was transcriptionally reactivated in the suvh4suvh5and suvh4suvh5suvh6mutants (Figure 4D).Thus,the Ta3transcrip-tional activation patterns correlate with non-CG methylation patterns and demonstrate a role for SUVH5in transposon silenc-ing.Together,these patterns suggest that SUVH4,SUVH5,and SUVH6all act at Ta3,with the hierarchy SUVH4>SUVH5>SUVH6.Mu1did not acquire H32mK4or transcriptional activity in any of the suvh mutant backgrounds tested (Figures 4C and 4D),consistent with the previous observation that this element is not transcriptionally activated by defects in the CMT3

non-CG

Figure 3.SUVH4,SUVH5,and SUVH6All Contribute to PAI1-PAI4Non-CG Methylation.

(A)and (B)DNA gel blot assays for PAI DNA methylation patterning.Genomic DNAfrom theindicated mutants wascleaved with Hpa IIor Msp Iisoschizomers (A)or Hin cII (B)and used in DNA gel blot analysis with a PAI1cDNA probe.P1-P4indicates PAI1-PAI4,P2indicates PAI2,and P3indicates PAI3,with bands diagnostic of methylation on PAI -internal sites denoted with asterisks.

(C)ChIP analysis of H32mK4and H32mK9patterning on the PAI genes.Primer sets speci?c for PAI1-PAI4(indicated as PAI1),PAI2,PAI3,or ACTIN were used to amplify PCR products from total input chromatin (I),no-antibody mock precipitation control (M),chromatin immunoprecipi-tated with H3anti-dimethyl K4antibodies (2mK4),or chromatin immu-noprecipitated with H3anti-dimethyl K9antibodies (2mK9)from the indicated mutants.GelStar-stained PCR products are shown.These results were reproduced in three independent experiments (see Supple-mental Figure 3online),with a representative data set shown.ACTIN is a control unmethylated transcribed gene previously determined to be en-riched for H32mK4but not H32mK9in wild-type and DNA methylation–de?cient mutant backgrounds (Johnson et al.,2002;Ebbs et al.,2005).Genotypes are as described for Figure 2.

1170The Plant Cell

methylation pathway (Lippman et al.,2003).Transcriptional activation of Mu1was observed only when CG methylation was lost,as in a met1CG MTase mutant (Figure 4D)(Lippman et al.,2003).

SUVH4,SUVH5,and SUVH6Together Control the Majority of CMT3-Mediated Repetitive Sequence DNA Methylation We also examined the effects of suvh mutations on the mainte-nance of non-CG methylation at the 180-bp centromere (CEN )repeats and at 5S rDNA pericentromeric repeats,using Msp I DNA gel blot analysis.At the CEN repeats,the suvh5and suvh6single mutants displayed similar cleavage patterns to the wild

type (Figure 5A).The ladder of cleaved products was partially shifted downward to a similar extent in the suvh4,suvh4suvh5,and suvh4suvh6mutants,indicative of a partial loss of CCG methylation.The ladder of cleaved products was more strongly shifted downward in the suvh4suvh5suvh6mutant to a similar extent as in cmt3,demonstrating a stronger loss of CCG meth-ylation.Thus,SUVH4,SUVH5,and SUVH6together control the majority of CMT3-mediated DNA methylation at the CEN repeats,with the hierarchy SUVH4>SUVH5?SUVH6.

At the 5S rDNA repeats,the suvh5and suvh6mutants had similar Msp I cleavage pro?les to wild-type Ws,whereas the suvh4,suvh4suvh5,and suvh4suvh6mutants had a similar partial downward shift in the ladder of cleaved products,and the suvh4suvh5suvh6triple mutant had a stronger downward shift (Figure 5B).These results indicate that at the 5S rDNA repeats,as at the CEN repeats,SUVH4,SUVH5,and SUVH6together control CMT3-mediated DNA methylation,with the hierarchy SUVH4>SUVH5?SUVH6.However,in contrast with other loci tested for DNA methylation patterning,the 5S rDNA repeats displayed Msp I cleavage patterns diagnostic of residual CCG methylation in suvh4suvh5suvh6relative to cmt3.

SUVH4,SUVH5,and SUVH6Together Control H3K9Monomethylation

A previous study found that suvh4maintains similar levels of H3monomethyl K9(H31mK9)to the wild type at heterochromatin targets,including the Ta3retrotransposon (Jackson et al.,2004).To determine whether SUVH5and SUVH6contribute to H31mK9,we performed ChIP analysis for this modi?cation

on

Figure 4.SUVH5Contributes to Ta3and Mu1Transposon Non-CG Methylation and Ta3Transcriptional Silencing in the Absence of SUVH4.(A)and (B)DNA gel blot assays for transposon DNA methylation patterning.DNA from the indicated mutants was cleaved with Msp I (A)or Hin dIII plus Msp I (B)and used in DNA gel blot analysis with a Ta3probe (A)or an Mu1probe (B).Arrowheads at left indicate the positions of fully cleaved bands.

(C)ChIP analysis of H32mK4and H32mK9patterning on transposons.Primer sets speci?c for Ta3,Mu1,or ACTIN were used to amplify PCR products from total input chromatin (I),no-antibody mock precipitation control (M),chromatin immunoprecipitated with H3anti-dimethyl K4antibodies (2mK4),or chromatin immunoprecipitated with H3anti-dimethyl K9antibodies (2mK9)from the indicated mutants.GelStar-stained PCR products are shown.These results were reproduced in three independent experiments (see Supplemental Figure 3online),with a representative data set shown.

(D)Semiquantitative RT-PCR analysis of transposon transcription.Primer sets speci?c for Ta3,Mu1,or ACTIN were used for RT-PCR analysis of total RNA prepared from 3-week-old plants of the indicated genotypes.The no-RT control is shown for the ACTIN primer set.

Genotypes are as described for Figure 2.The cmt3met1mutant (Ebbs et al.,2005)was used as a control for RT-PCR

analysis.

Figure 5.SUVH4,SUVH5,and SUVH6Together Control the Majority of CMT3-Mediated DNA Methylation at Repetitive Sequences.

DNA gel blot assays for repetitive sequences.DNA from the indicated mutants was cleaved with Msp I and used in DNA gel blot analysis with a 180-bp centromere repeat probe (A)or a 5S rDNA probe (B).Genotypes are as described for Figure 2.

SUVH5Controls DNA Methylation 1171

transposon and PAI gene sequences in suvh mutants (Figure 6;see Supplemental Figure 6online).We found that SUVH4,SUVH5,and SUVH6together control H31mK9,with different relative contributions at different loci.

At Ta3and Mu1transposon sequences,the suvh4,suvh5,and suvh6single mutants maintained similar levels of H31mK9to the wild type (Figure 6A).However,H31mK9was partially reduced in the suvh4suvh5and suvh4suvh6double mutants and strongly reduced in the suvh4suvh5suvh6triple mutant,indicating that SUVH4,SUVH5,and SUVH6together control H31mK9at these loci.At Mu1,the loss of H31mK9in suvh4suvh5,suvh4suvh6,and suvh4suvh5suvh6cannot be attributed to transcriptional reactivation induced by the loss of non-CG methylation,because Mu1is not transcriptionally active in these backgrounds (Figure 4D).At the PAI1-PAI4transcribed inverted repeat,the suvh single mutants and the suvh4suvh5double mutant maintained similar levels of H31mK9to the wild type (Figure 6B).However,H31mK9was reduced in suvh4suvh6and lost in suvh4suvh5suvh6.This loss of H31mK9cannot be attributed to transcriptional hyperactivation induced by DNA demethylation,because PAI1-PAI4maintains a constant level of expression from its upstream unmethylated promoter even when internal non-CG methylation is lost in the cmt3background (Ebbs et al.,2005).Instead,the loss of H31mK9in suvh4suvh6and suvh4suvh5suvh6suggests that SUVH4and SUVH6together control the majority of H31mK9at PAI1-PAI4.

At the PAI2and PAI3target loci,the suvh4single mutant displayed reduced H31mK9(Figure 6C),in parallel with the loss of H32mK9,non-CG methylation,and transcriptional reactiva-tion (Figures 2and 3)(Ebbs et al.,2005).At PAI2,residual H31mK9was maintained in suvh4suvh5and suvh4suvh6but lost in suvh4suvh5suvh6,in parallel with patterns of residual transcrip-tional activation (Figure 2).Therefore,SUVH4controls the ma-jority of H31mK9at PAI2and PAI3,but this control could occur indirectly through the maintenance of H32mK9,non-CG meth-ylation,and transcriptional silencing.The stronger loss of H31mK9from PAI3than from PAI2in suvh4probably re?ects stronger transcriptional reactivation of PAI3as a result of less extensive promoter PAI sequence identity and DNA methylation (Ebbs et al.,2005).DISCUSSION

RNA-directed heterochromatin formation is associated with H3mK9in ?ssion yeast,Drosophila ,mammals,and plants,indicat-ing a fundamental mechanistic connection between RNA signals and H3K9MTases (Volpe et al.,2003;Lippman et al.,2004;Pal-Bhadra et al.,2004;Ting et al.,2005).In the plant Arabidopsis ,H3mK9further guides non-CG methylation mediated by the CMT3cytosine MTase (Jackson et al.,2002;Malagnac et al.,2002).Here,we show that three Arabidopsis proteins with H3K9MTase activity in vitro—SUVH4,SUVH5,and SUVH6—act together to guide H3K9monomethylation and dimethylation and CMT3-mediated non-CG methylation.The three SUVH proteins make different relative contributions to the maintenance of H3K9and DNA methylation at different loci,suggesting locus-speci?c mechanisms for their recruitment or activation.The suvh4suvh5suvh6triple mutant displays similar strong non-CG de-methylation patterns to a cmt3mutant at the PAI genes (Figure 3),the Ta3and Mu1transposons (Figure 4),and centromere repeats (Figure 5),indicating that the three SUVH proteins control the majority of CMT3-mediated DNA methylation.

At all loci we examined except the PAI1-PAI4transcribed inverted repeat,the suvh4single mutation reduced H32mK9and/or non-CG methylation,whereas the suvh5and suvh6single mutations had no effect (Figures 3to 5)(Ebbs et al.,2005).Thus,SUVH4plays a dominant role relative to SUVH5and SUVH6in maintaining heterochromatin-associated modi?cations at most heterochromatin targets.The dominant role of SUVH4could be attributable to higher levels of protein expression,intrinsically better MTase activity,or more ef?cient recruitment to target sequences than SUVH5or SUVH6.

At PAI1-PAI4,suvh4suvh6but not the suvh5suvh6or suvh4suvh5mutant had reduced H3mK9and non-CG methylation (Figures 3and 6B;see Supplemental Figure 4online).These results show that either SUVH4alone or SUVH6alone can fully maintain heterochromatin-associated modi?cations at PAI1-PAI4,with only a minor role for SUVH5.By contrast,suvh4suvh5displayed a stronger loss of residual non-CG methylation from Ta3and Mu1transposons than suvh4suvh6,and suvh4suvh5but not suvh4suvh6showed transcriptional reactivation of Ta3(Figure 4).These data indicate that SUVH5makes a greater contribution than SUVH6to heterochromatin modi?cations at these transposons.The different relative contributions of

SUVH5

Figure 6.SUVH4,SUVH5,and SUVH6Together Control H31mK9.ChIP analysis of H31mK9patterning on Ta3and Mu1transposons (A),the PAI1-PAI4transcribed inverted repeat (indicated as PAI1)(B),the silenced PAI2and PAI3genes (C),and the unmethylated transcribed control gene ACTIN (D).Primer sets speci?c for each gene were used to amplify PCR products from total input chromatin (I),no-antibody mock precipitation control (M),or chromatin immunoprecipitated with H3anti-monomethyl K9antibodies (1mK9)from the indicated mutants.GelStar-stained PCR products are shown.These results were reproduced in three independent experiments (see Supplemental Figure 6online),with a rep-resentative data set shown.Genotypes are as described for Figure 2.

1172The Plant Cell

and SUVH6at PAI1-PAI4versus Ta3and Mu1transposons suggest that there are locus-speci?c differences in recruitment or activation of these proteins.

The heterochromatic loci where we observed different relative SUVH activities have different transcriptional activities and de-rive the dsRNA signal for heterochromatin formation from differ-ent sources.The PAI1-PAI4inverted repeat(SUVH4?SUVH6> SUVH5)directly produces dsRNA by constitutive transcription from an upstream promoter,PAI2and PAI3(SUVH4>SUVH5?SUVH6)are targeted for DNA methylation in trans by PAI1-PAI4 RNA but are not themselves transcribed,and the Ta3and Mu1 transposons(SUVH4>SUVH5>SUVH6)probably generate dsRNA through RNA-dependent RNA polymerase action on transposon-derived transcripts(Melquist and Bender,2003; Xie et al.,2004).These differences in RNA production and pro-cessing could underlie the different SUVH activity patterns.For example,the different sources of dsRNA could feed into different RNA effector complex variants that differentially recruit the SUVH proteins.Alternatively,different RNA signals could promote dif-ferent patterns of histone modi?cations on target loci that would lead to differential recruitment of SUVH proteins.It is unlikely that the primary DNA sequence or cytosine methylation pattern on a locus determines the relative activity of the SUVH proteins, because PAI1and PAI2are almost identical in sequence and DNA methylation(Luff et al.,1999)and yet have different patterns of dependence on the SUVH proteins for the maintenance of H3 K9and non-CG methylation.

At the Ta3and Mu1transposons,the suvh4single mutation was suf?cient to reduce H32mK9and non-CG methylation,but it did not affect H31mK9levels or confer transcriptional activation (Figures4and6A).These patterns suggest that H31mK9is more easily maintained than H32mK9by the remaining active SUVH enzymes in suvh4.Consistent with this view,a previous in vitro analysis showed that SUVH4and SUVH6are better mono-MTases than di-MTases:both enzymes catalyze monomethylation of a peptide substrate within minutes but catalyze dimethylation only after several hours(Jackson et al.,2004).The slow kinetics of SUVH-catalyzed dimethylation in vitro suggests that the relative contribution of a SUVH enzyme to dimethylation of a particular target site in vivo might re?ect how stably it is associated with that target.Our?nding that SUVH4,SUVH5,and SUVH6to-gether maintain H31mK9as well as H32mK9and non-CG methylation suggests that in plants the H31mK9modi?cation is part of the CMT3-mediated DNA methylation pathway,rather than serving a novel signaling role.The simplest explanation of our results is that SUVH4,SUVH5,and SUVH6catalyze both H3 1mK9and H32mK9in vivo,but we cannot exclude the possibility that these enzymes control H3mK9patterns through an indirect mechanism,such as alterations in other histone modi?cation patterns.

The six SUVH proteins that lack in vitro histone MTase activity under our assay conditions(Figure1A)are all structurally diver-gent in their catalytic domains from the three active SUVH proteins(Baumbusch et al.,2001).However,in an in vivo setting, with a full complement of interacting factors and substrate histones in a nucleosomal/chromatin context,some of these proteins might also have H3K9MTase activity.For example, SUVH2has H3K9MTase activity against nucleosomal sub-strates in vitro and reduced H3mK9in vivo(Naumann et al., 2005).Therefore,the residual DNA methylation detected at the 5S rDNA repeats in the suvh4suvh5suvh6triple mutant relative to cmt3(Figure5B)might re?ect residual H3mK9at these sequences mediated by SUVH2or other SUVH proteins.It is also possible that some of the other SUVH proteins act as MTases on plant-speci?c histone variants or nonhistone substrates.

Our?nding that SUVH5can methylate both H3K9and Arabidopsis histone H2A variants in vitro(Figure1;see Supple-mental Figure1online)suggests that this enzyme might catalyze both H3K9and H2A methylation in vivo.However,the lack of DNA methylation phenotypes in the suvh5single mutant(Figures 3to5;see Supplemental Figure5online)suggests that SUVH5 H2A MTase activity does not play a role in DNA methylation patterning unless this activity is masked by redundancy with other H2A histone MTases.In addition,a more extensive analysis of the phenotypes conferred by suvh5,especially under non-standard growth conditions,might reveal a unique pathway controlled by SUVH5versus SUVH4or SUVH6.Although the functions of the different Arabidopsis H2A variants remain to be elucidated,H2A variants in fungal and mammalian systems have been shown to be important for speci?c processes,including telomeric silencing(Wyatt et al.,2003),proper segregation of telomeres during meiotic prophase1(Fernandez-Capetillo et al., 2003),protecting euchromatin from heterochromatin spreading (Meneghini et al.,2003),and the nonhomologous end-joining DNA repair pathway(Celeste et al.,2002).

Similar to the cmt3mutant(Bartee et al.,2001;Lindroth et al., 2001),the suvh4suvh5suvh6triple H3K9MTase mutant dis-plays no developmental defects.This observation suggests that the primary function of the three SUVH enzymes is to make histone modi?cations that guide CMT3-mediated non-CG meth-ylation to transposons and repeated sequences.However,other SUVH proteins might have locus-speci?c histone methylation functions connected to developmental regulation rather than genome defense.Other plant species,such as maize(Zea mays), also contain multiple SUVH-related genes(Springer et al.,2003), suggesting diversi?cation of SUVH function as a common theme in plant evolution.As we have shown here for SUVH4,SUVH5, and SUVH6,maintaining multiple H3K9MTases provides par-tially redundant modes of protection from different classes of aberrant or invasive DNA sequences.

METHODS

Expression of Recombinant Proteins

The catalytic segment extending from just downstream of the YDG domain through the C terminus of each Arabidopsis thaliana Ws SUVH gene(Baumbusch et al.,2001)was subcloned into either pGEX4T-1or pGEX4T-2(Pharmacia)to make an N-terminal GST fusion protein ex-pression construct.Tyr-to-Phe mutations in GST-SUVH4(residue591) and GST-SUVH5(residue761)were created using oligonucleotide-directed mutagenesis(Kunkel et al.,1987).The GST-H3peptide plasmids were a gift of Yoichi Shinkai(Tachibana et al.,2001).

Protein expression plasmids were transformed into Escherichia coli strain BL21Codon Plus RIL(Stratagene).Bacterial cultures for GST-SUVH proteins were grown at258C in1liter of2XYT medium supple-mented with100mg/L ampicillin and30mg/L chloramphenicol.Cultures

SUVH5Controls DNA Methylation1173

were induced when they reached midlog phase by adding isopropylthio-b-galactoside to a0.1mM?nal concentration and then grown for an additional4h.GST-H3(Figure1)and GST-HTA(see Supplemental Figure1online)bacterial expression was performed in a similar manner, except that cultures were grown and induced at378C.

Induced cells were lysed in50mM Tris,pH7.5,0.1mM EDTA,0.1% Triton X-100,1mg/mL lysozyme(Sigma-Aldrich),15units/mL DNase, and a full complement of protease inhibitors(1mL/4g wet cell weight; Sigma-Aldrich P-8340)by freeze/thaw.GST fusion proteins were puri?ed from cell lysates using glutathione-coupled Sepharose beads according to the manufacturer’s instructions(Amersham Biosciences).Proteins were concentrated using Ultra-free0.5concentrators(Millipore),and protein concentration was determined by SDS-PAGE and Coomassie Brilliant Blue R250staining.

In Vitro Histone MTase Assays

Calf thymus histone mix(H1,H2A,H2B,H3,and H4)was purchased from Roche Molecular Biochemicals.The biotinylated dimethyl H3K9peptide was purchased from Upstate Biotechnology.MTase assays were per-formed based on a previous protocol(Rea et al.,2000).Brie?y,10m g of recombinant GST-SUVH protein was incubated with10m g of a his-tone substrate and300nCi of S-adenosyl-[methyl-14C]L-Met(100m M; Amersham)in methylase activity buffer(50mM Tris,pH8.5,20mM KCl, 10mM MgCl2,10mM b-mercaptoethanol,and250mM sucrose)at378C for60min in a total volume of50m L.Reactions were separated on an18% SDS-PAGE gel.Proteins were visualized by Coomassie Brilliant Blue R250staining,and the14C signal was visualized by EN3HANCE (Amersham)?uorography.

In Vitro Analysis of SUVH5MTase Activity on Arabidopsis H2A Variant Substrates

The Arabidopsis genome encodes13histone H2A(HTA)variants,all of which except HTA4are detectably expressed(Callard and Mazzolini, 1997;Mysore et al.,2000).Although the central sequences of these variants are highly conserved with each other and the mammalian H2A sequences,the N-terminal sequences are divergent(see Supplemental Figure1A online).To determine whether any of the Arabidopsis HTA proteins could serve as in vitro substrates for methylation by SUVH5,we expressed,puri?ed,and assayed11variants as full-length N-terminal GST fusion proteins.Full-length cDNA clones for HTA1(Nam et al.,1999), HTA2,HTA3,HTA5,HTA7,HTA9(Callard and Mazzolini,1997),HTA10, HTA11(Callard and Mazzolini,1997),and HTA13from the Col ecotype were obtained from the ABRC at Ohio State University.HTA6and HTA8 full-length cDNAs were ampli?ed by RT-PCR from Col RNA.Each of these full-length cDNAs was subcloned into the pGEX*vector(Haldeman et al., 1997)to make an N-terminal GST fusion protein construct.

We found that variants HTA2,HTA7,and HTA13were used as SUVH5 substrates,with HTA13being the most robustly methylated(see Sup-plemental Figure1B online),whereas HTA1,HTA3,HTA5,HTA6,HTA8, HTA9,HTA10,and HTA11were not detectably methylated(data not shown).For GST-HTA2,GST-HTA7,and GST-HTA13,both the full-length proteins and truncations of between31and35kD were observed to be used as SUVH5substrates(see Supplemental Figure1B online).Because these truncations are most likely missing C-terminal residues,this?nding suggested that SUVH5recognizes residues in HTA N-terminal tails.

To determine more precisely which HTA residues are methylated,we focused on HTA13as a model substrate.We made single Lys-to-Arg mutations at residues13and14of HTA13(K13R,K14R),counting from the?rst amino acid after the predicted Met translational start codon as residue1,and a double mutation of Lys-13and Lys-14(KDM).All mutant variants were expressed,puri?ed,and assayed as full-length N-terminal GST fusion proteins.This analysis showed that only the KDM mutation abolished SUVH5-catalyzed methylation,suggesting that SUVH5can methylate either K13or K14in HTA13(see Supplemental Figure1C online).We obtained similar results for mutations of the analogous Lys residues in HTA2(data not shown).

Given the pro?le of Arabidopsis HTA variants that serve as SUVH5 methylation substrates(see Supplemental Figures1A and1B online),the speci?city of SUVH5for Lys-13and Lys-14in HTA13(see Supplemental Figure1C online),and the speci?city of SUVH5for Lys-9in histone H3 (Figure1B),we predicted a recognition sequence context of(V/A)A(R/ K*)(R/K*)S for this enzyme,with K*indicating potential substrate Lys residues.To test this prediction,we made a series of context mutations in the HTA13substrate,including V11A,A12K,K13T,S15A,and S15P. These mutations were chosen based on sequences of HTAs that were not substrates for SUVH5.For example,HTA3,HTA5,and HTA6all have a T before the K14-analogous residue,whereas HTA9and HTA11have a P after the K14-analogous residue(see Supplemental Figure1A online).Of the context mutations tested,V11A,A12K,K13T,and S15A reduced methylation,whereas only the S15P mutation abolished SUV5-catalyzed methylation(see Supplemental Figure1C online).Thus,it is likely that particular combinations of amino acid differences in the N termini of nonsubstrate HTAs act together to block SUVH5activity.

Plant Materials

The Ws suvh5-1T-DNA insertion mutation was obtained from the FLAGdb/FST mutant collection at the Institute of Agronomic Research in Versailles,France(Samson et al.,2002).The left border insertion junction lies2071bp downstream from the ATG translational start codon of SUVH5in the exonic region that encodes the SET portion of the catalytic sequences.The Col suvh5-2T-DNA insertion mutation was obtained from the SALK Institute Genomic Analysis Laboratory mutant collection via the ABRC(Alonso et al.,2003).The left border insertion junction lies2258bp downstream from the ATG translational start codon of SUVH5in the exonic region that encodes the SET portion of the catalytic sequences.Each suvh5allele was crossed with Ws pai1(Bartee and Bender,2001),Ws pai1suvh4R302*(Malagnac et al.,2002),Ws pai1 suvh6-1(Ebbs et al.,2005),or Ws pai1suvh4R302*suvh6-1(Ebbs et al., 2005),and PCR-based genotype markers were used to identify progeny that were homozygous for the three PAI loci from Ws(Luff et al.,1999)and homozygous for each suvh mutation.At least three independent lines were isolated for each suvh mutant,and DNA gel blot assays for PAI and transposon DNA methylation were used to determine that there was no variation in DNA methylation phenotypes among different lines with the same suvh genotype.A representative line of each suvh genotype was used for ChIP analysis(Figures3C,4C,and6)and PAI bisul?te sequenc-ing analysis(see Supplemental Figure2online).PCR primers to amplify the suvh5-1insertion are FSTLB4(59-CGTGTGCCAGGTGCCCACGGA-ATAGT-39)and SUVH5R2(59-GCTTATTCAGACAGAACTGAAC-39),which yield a550-bp product.PCR primers to amplify the suvh5-2insertion are SALKLB(59-GCGTGGACCGCTTGCTGCAACT-39)and SUVH5AKFR (59-ATCACTAAGACAACATCAGTATGATCAAG-39),which yield a650-bp product.PCR primers to amplify the intact SUVH5gene from either suvh5-1or suvh5-2are SUVH5R2and SUVH5GAR2(59-GATGGTCTTTG-CAATGTTG-39),which amplify an844-bp product.PCR-based assays to score the suvh4R302*and suvh6-1mutations were described previously (Malagnac et al.,2002;Ebbs et al.,2005).

Plant genomic DNA preparations and DNA gel blots were performed as described previously(Melquist et al.,1999).

ChIP Analysis

ChIP assays were performed using a previously described method (Gendrel et al.,2002)starting with0.7g of leaf tissue from3-week-old plants grown in soilless potting mix(Fafard mix2)under continuous

1174The Plant Cell

illumination.Chromatin was immunoprecipitated with anti-H3dimethyl K4antibodies(Upstate Biotechnology),with anti-H3dimethyl K9anti-bodies(a gift of T.Jenuwein),with anti-H3monomethyl K9antibodies (Upstate Biotechnology),or carried through the protocol with no antibody added as a control(mock precipitation).PCR ampli?cation of immuno-precipitated DNA was performed as described previously(Ebbs et al., 2005).PCR-ampli?ed products from ChIP template DNA were visualized on a2.5%agarose gel stained with GelStar(Cambrex).Each ChIP assay was performed in three independent experiments(see Supplemental Figures3and6online),with results from representative experiments shown in Figures3C,4C,and6.ChIP primer sequences used in this study are listed in Supplemental Table1online.

Semiquantitative RT-PCR

Total RNA was isolated from4-week-old plants and treated with RNase-free DNase(Promega).RT-PCR was performed as described previously for Mu1(Singer et al.,2001),Ta3(Johnson et al.,2002),or ACTIN(Ito et al., 2004).Control reactions without reverse transcriptase were performed to ensure that no background signal was generated from contaminating DNA. Accession Numbers

Arabidopsis Genome Initiative locus identi?ers are as follows:SUVH1 (At5g04940),SUVH2(At2g33290),SUVH3(At1g73100),SUVH4(At5g13960), SUVH5(At2g35160),SUVH6(At2g22740),SUVH7(At1g17770),SUVH8 (At2g24740),SUVH9(At4g13460),ACTIN(At5g09810),HTA1 (At5g54640),HTA2(At4g27230),HTA3(At1g54690),HTA5(At1g08880), HTA6(At5g59870),HTA7(At5g27670),HTA8(At2g38810),HTA9 (At1g52740),HTA10(At1g51060),HTA11(At3g54560),HTA13 (At3g20670),PAI1(At1g07780),PAI2(At5g05590),and PAI3(At1g29410). Supplemental Data

The following materials are available in the online version of this article.

Supplemental Figure1.SUVH5Has Histone H2A MTase Activity in Vitro.

Supplemental Figure 2.Bisul?te Genomic Sequencing of DNA Methylation Patterning on the PAI1and PAI2Proximal Promoters in suvh Mutants.

Supplemental Figure3.Replicates of ChIP Analysis of H32mK4and H32mK9Patterning.

Supplemental Figure4.The suvh5suvh6Mutant Does Not Display DNA Methylation Defects.

Supplemental Figure5.Two suvh5Alleles Confer Similar Loss of Transposon Non-CG Methylation in the Absence of SUVH4.

Supplemental Figure6.Replicates of ChIP Analysis of H31mK9 Patterning.

Supplemental Table1.ChIP Primer Sequences.

ACKNOWLEDGMENTS

We thank Yoichi Shinkai for GST-H3plasmids;the ABRC for HTA plasmids,the5S rDNA plasmid pCT4.2,and the Salk collection suvh5-2 T-DNA insertion mutant;the Institute of Agronomic Research for the suvh5-1T-DNA insertion mutant;and Thomas Jenuwein for H3anti-dimethyl K9antibodies.We also thank Cecile Pickart for technical advice and Brandi Rocci for assistance with HTA expression.This work was supported by National Institutes of Health Grant GM-61148to J.B.and by training grants T32ES-07141(National Institute of Environ-mental Health Sciences)and T32CA-09110(National Cancer Institute) to M.L.E.

Received January24,2006;revised February16,2006;accepted March1, 2006;published March31,2006.

REFERENCES

Alonso,J.M.,et al.(2003).Genome-wide insertional mutagenesis of Arabidopsis thaliana.Science301,653–657.

Bartee,L.,and Bender,J.(2001).Two Arabidopsis methylation-de?ciency mutations confer only partial effects on a methylated endogenous gene family.Nucleic Acids Res.29,2127–2134. Bartee,L.,Malagnac,F.,and Bender,J.(2001).Arabidopsis cmt3 chromomethylase mutations block non-CG methylation and silencing of an endogenous gene.Genes Dev.15,1753–1758. Baumbusch,L.O.,Thorstensen,T.,Krauss,V.,Fischer, A., Naumann,K.,Assalkhou,R.,Schulz,I.,Reuter,G.,and Aalen, R.B.(2001).The Arabidopsis thaliana genome contains at least29 active genes encoding SET domain proteins that can be assigned to four evolutionarily conserved classes.Nucleic Acids Res.29,4319–4333.

Callard,D.,and Mazzolini,L.(1997).Identi?cation of proliferation-induced genes in Arabidopsis thaliana.Characterization of a new member of the highly evolutionarily conserved histone H2A.F/Z variant subfamily.Plant Physiol.115,1385–1395.

Cao,X.,Aufsatz,W.,Zilberman, D.,Mette,M.F.,Huang,M.S., Matzke,M.,and Jacobsen,S.E.(2003).Role of the DRM and CMT3methyltransferases in RNA-directed DNA methylation.Curr. Biol.13,2212–2217.

Cao,X.,and Jacobsen,S.E.(2002).Locus-speci?c control of asym-metric and CpNpG methylation by the DRM and CMT3methyltrans-ferase https://www.360docs.net/doc/e45566043.html,A99(suppl.4),16491–16498. Celeste,A.,et al.(2002).Genomic instability in mice lacking histone H2AX.Science296,922–927.

Collins,R.E.,Tachibana,M.,Tamaru,H.,Smith,K.M.,Jia,D.,Zhang, X.,Selker,E.U.,Shinkai,Y.,and Cheng,X.(2005).In vitro and in vivo analyses of a Phe/Tyr switch controlling product speci?city of histone lysine methyltransferases.J.Biol.Chem.280,5563–5570.

Ebbs,M.L.,Bartee,L.,and Bender,J.(2005).H3lysine9methylation is maintained on a transcribed inverted repeat by combined action of SUVH6and SUVH4methyltransferases.Mol.Cell.Biol.25,10507–10515.

Fernandez-Capetillo,O.,Liebe,B.,Scherthan,H.,and Nussenzweig, A.(2003).H2AX regulates meiotic telomere clustering.J.Cell Biol. 163,15–20.

Gendrel,A.V.,Lippman,Z.,Yordan,C.,Colot,V.,and Martienssen, R.A.(2002).Dependence of heterochromatic histone H3methylation patterns on the Arabidopsis gene DDM1.Science297,1871–1873. Haldeman,M.T.,Xia,G.,Kasperek,E.M.,and Pickart,C.M.(1997). Structure and function of ubiquitin conjugating enzyme E2-25K:The tail is a core-dependent activity element.Biochemistry36,10526–10537. Ito,T.,Wellmer, F.,Yu,H.,Das,P.,Ito,N.,Alves-Ferreira,M., Riechmann,J.L.,and Meyerowitz, E.M.(2004).The homeotic protein AGAMOUS controls microsporogenesis by regulation of SPOROCYTELESS.Nature430,356–360.

Jackson,J.P.,Johnson,L.,Jasencakova,Z.,Zhang,X.,PerezBurgos, L.,Singh,P.B.,Cheng,X.,Schubert,I.,Jenuwein,T.,and Jacobsen, S.E.(2004).Dimethylation of histone H3lysine9is a critical mark for DNA methylation and gene silencing in Arabidopsis thaliana. Chromosoma112,308–315.

SUVH5Controls DNA Methylation1175

Jackson,J.P.,Lindroth,A.M.,Cao,X.,and Jacobsen,S.E.(2002). Control of CpNpG DNA methylation by the KRYPTONITE histone H3 methyltransferase.Nature416,556–560.

Johnson,L.,Cao,X.,and Jacobsen,S.(2002).Interplay between two epigenetic marks.DNA methylation and histone H3lysine9methyl-ation.Curr.Biol.12,1360–1367.

Kato,M.,Miura,A.,Bender,J.,Jacobsen,S.E.,and Kakutani,T. (2003).Role of CG and non-CG methylation in immobilization of transposons in Arabidopsis.Curr.Biol.13,421–426.

Konieczny,A.,Voytas,D.F.,Cummings,M.P.,and Ausubel,F.M. (1991).A superfamily of Arabidopsis thaliana retrotransposons.Ge-netics127,801–809.

Kunkel,T.A.,Roberts,J.D.,and Zakour,R.A.(1987).Rapid and ef?cient site-speci?c mutagenesis without phenotypic selection. Methods Enzymol.154,367–382.

Lehnertz, B.,Ueda,Y.,Derijck, A.A.,Braunschweig,U.,Perez-Burgos,L.,Kubicek,S.,Chen,T.,Li,E.,Jenuwein,T.,and Peters, A.H.(2003).Suv39h-mediated histone H3lysine9methylation directs DNA methylation to major satellite repeats at pericentric heterochro-matin.Curr.Biol.13,1192–1200.

Lindroth,A.M.,Cao,X.,Jackson,J.P.,Zilberman,D.,McCallum, C.M.,Henikoff,S.,and Jacobsen,S.E.(2001).Requirement of CHROMOMETHYLASE3for maintenance of CpXpG methylation. Science292,2077–2080.

Lippman,Z.,et al.(2004).Role of transposable elements in hetero-chromatin and epigenetic control.Nature430,471–476.

Lippman,Z.,and Martienssen,R.(2004).The role of RNA interference in heterochromatic silencing.Nature431,364–370.

Lippman,Z.,May,B.,Yordan,C.,Singer,T.,and Martienssen,R. (2003).Distinct mechanisms determine transposon inheritance and methylation via small interfering RNA and histone modi?cation.PLoS Biol.1,E67.

Luff,B.,Pawlowski,L.,and Bender,J.(1999).An inverted repeat triggers cytosine methylation of identical sequences in Arabidopsis. Mol.Cell3,505–511.

Malagnac,F.,Bartee,L.,and Bender,J.(2002).An Arabidopsis SET domain protein required for maintenance but not establishment of DNA methylation.EMBO J.21,6842–6852.

Mathieu,O.,and Bender,J.(2004).RNA-directed DNA methylation. J.Cell Sci.117,4881–4888.

Matzke,M.A.,and Birchler,J.A.(2005).RNAi-mediated pathways in the nucleus.Nat.Rev.Genet.6,24–35.

Melquist,S.,and Bender,J.(2003).Transcription from an upstream promoter controls methylation signaling from an inverted repeat of endogenous genes in Arabidopsis.Genes Dev.17,2036–2047. Melquist,S.,Luff,B.,and Bender,J.(1999).Arabidopsis PAI gene arrangements,cytosine methylation and expression.Genetics153, 401–413.

Meneghini,M.D.,Wu,M.,and Madhani,H.D.(2003).Conserved histone variant H2A.Z protects euchromatin from the ectopic spread of silent heterochromatin.Cell112,725–736.

Mysore,K.S.,Nam,J.,and Gelvin,S.B.(2000).An Arabidopsis histone H2A mutant is de?cient in Agrobacterium T-DNA integration.Proc. https://www.360docs.net/doc/e45566043.html,A97,948–953.

Nam,J.,Mysore,K.S.,Zheng,C.,Knue,M.K.,Matthysse,A.G.,and Gelvin,S.B.(1999).Identi?cation of T-DNA tagged Arabidopsis mu-tants that are resistant to transformation by Agrobacterium.Mol.Gen. Genet.261,429–438.

Naumann,K.,Fischer,A.,Hofmann,I.,Krauss,V.,Phalke,S.,Irmler, K.,Hause,G.,Aurich,A.C.,Dorn,R.,Jenuwein,T.,and Reuter,G. (2005).Pivotal role of AtSUVH2in heterochromatic histone methyla-tion and gene silencing in Arabidopsis.EMBO J.24,1418–1429.Pal-Bhadra,M.,Leibovitch,B.A.,Gandhi,S.G.,Rao,M.,Bhadra,U., Birchler,J.A.,and Elgin,S.C.(2004).Heterochromatic silencing and HP1localization in Drosophila are dependent on the RNAi machinery. Science303,669–672.

Rea,S.,Eisenhaber, F.,O’Carroll, D.,Strahl, B.D.,Sun,Z.W., Schmid,M.,Opravil,S.,Mechtler,K.,Ponting,C.P.,Allis,C.D., and Jenuwein,T.(2000).Regulation of chromatin structure by site-speci?c histone H3methyltransferases.Nature406,593–599. Samson,F.,Brunaud,V.,Balzergue,S.,Dubreucq,B.,Lepiniec,L., Pelletier,G.,Caboche,M.,and Lecharny,A.(2002).FLAGdb/FST:A database of mapped?anking insertion sites(FSTs)of Arabidopsis thaliana T-DNA transformants.Nucleic Acids Res.30,94–97. Singer,T.,Yordan,C.,and Martienssen,R.A.(2001).Robertson’s Mutator transposons in A.thaliana are regulated by the chromatin-remodeling gene Decrease in DNA Methylation(DDM1).Genes Dev. 15,591–602.

Springer,N.M.,Napoli,C.A.,Selinger,D.A.,Pandey,R.,Cone,K.C., Chandler,V.L.,Kaeppler,H.F.,and Kaeppler,S.M.(2003).Com-parative analysis of SET domain proteins in maize and Arabidopsis reveals multiple duplications preceding the divergence of monocots and dicots.Plant Physiol.132,907–925.

Tachibana,M.,Sugimoto,K.,Fukushima,T.,and Shinkai,Y.(2001). Set domain-containing protein,G9a,is a novel lysine-preferring mammalian histone methyltransferase with hyperactivity and speci?c selectivity to lysines9and27of histone H3.J.Biol.Chem.276, 25309–25317.

Tamaru,H.,and Selker,E.U.(2001).A histone H3methyltransferase controls DNA methylation in Neurospora crassa.Nature414,277–283. Tamaru,H.,Zhang,X.,McMillen,D.,Singh,P.B.,Nakayama,J., Grewal,S.I.,Allis, C.D.,Cheng,X.,and Selker, E.U.(2003). Trimethylated lysine9of histone H3is a mark for DNA methylation in Neurospora crassa.Nat.Genet.34,75–79.

Ting,A.H.,Schuebel,K.E.,Herman,J.G.,and Baylin,S.B.(2005). Short double-stranded RNA induces transcriptional gene silencing in human cancer cells in the absence of DNA methylation.Nat.Genet. 37,906–910.

Tompa,R.,McCallum,C.M.,Delrow,J.,Henikoff,J.G.,van Steensel, B.,and Henikoff,S.(2002).Genome-wide pro?ling of DNA methyl-ation reveals transposon targets of CHROMOMETHYLASE3.Curr. Biol.12,65–68.

Verdel,A.,Jia,S.,Gerber,S.,Sugiyama,T.,Gygi,S.,Grewal,S.I., and Moazed,D.(2004).RNAi-mediated targeting of heterochromatin by the RITS complex.Science303,672–676.

Volpe,T.,Schramke,V.,Hamilton,G.L.,White,S.A.,Teng,G., Martienssen,R.A.,and Allshire,R.C.(2003).RNA interference is required for normal centromere function in?ssion yeast.Chromosome Res.11,137–146.

Wyatt,H.R.,Liaw,H.,Green,G.R.,and Lustig,A.J.(2003).Multiple roles for Saccharomyces cerevisiae histone H2A in telomere position effect,Spt phenotypes and double-strand-break repair.Genetics164, 47–64.

Xie,Z.,Johansen,L.K.,Gustafson, A.M.,Kasschau,K.D.,Lellis, A.D.,Zilberman,D.,Jacobsen,S.E.,and Carrington,J.C.(2004). Genetic and functional diversi?cation of small RNA pathways in plants.PLoS Biol.2,E104.

Xin,Z.,Tachibana,M.,Guggiari,M.,Heard, E.,Shinkai,Y.,and Wagstaff,J.(2003).Role of histone methyltransferase G9a in CpG methylation of the Prader-Willi syndrome imprinting center.J.Biol. Chem.278,14996–15000.

Zhang,X.,Yang,Z.,Khan,S.I.,Horton,J.R.,Tamaru,H.,Selker, E.U.,and Cheng,X.(2003).Structural basis for the product speci-?city of histone lysine methyltransferases.Mol.Cell12,177–185.

1176The Plant Cell

DOI 10.1105/tpc.106.041400

; originally published online March 31, 2006;

2006;18;1166-1176Plant Cell Michelle L. Ebbs and Judith Bender

SUVH5 Histone Methyltransferase

Arabidopsis Locus-Specific Control of DNA Methylation by the

This information is current as of November 11, 2014

Supplemental Data https://www.360docs.net/doc/e45566043.html,/content/suppl/2006/03/31/tpc.106.041400.DC1.html References

https://www.360docs.net/doc/e45566043.html,/content/18/5/1166.full.html#ref-list-1This article cites 51 articles, 27 of which can be accessed free at:Permissions

https://https://www.360docs.net/doc/e45566043.html,/ccc/openurl.do?sid=pd_hw1532298X&issn=1532298X&WT.mc_id=pd_hw1532298X eTOCs

https://www.360docs.net/doc/e45566043.html,/cgi/alerts/ctmain Sign up for eTOCs at:

CiteTrack Alerts

https://www.360docs.net/doc/e45566043.html,/cgi/alerts/ctmain Sign up for CiteTrack Alerts at:

Subscription Information

https://www.360docs.net/doc/e45566043.html,/publications/subscriptions.cfm is available at:Plant Physiology and The Plant Cell Subscription Information for ADVANCING THE SCIENCE OF PLANT BIOLOGY

? American Society of Plant Biologists