2018-DNA甲基化阅读器复合物可增强基因转录.pdf

MOLECULAR BIOLOGYA DNA methylation reader complexthat enhances gene transcriptionC.Jake Harris1*,Marion Scheibe2*,Somsakul Pop Wongpalee1,Wanlu Liu1,Evan M.Cornett3,Robert M.Vaughan3,Xueqin Li4,Wei Chen4,Yan Xue1,Zhenhui Zhong1,5,Linda Yen1,William D.Barshop6,Shima Rayatpisheh6,Javier Gallego-Bartolome1,Martin Groth1,Zonghua Wang5,7,James A.Wohlschlegel6,Jiamu Du4,Scott B.Rothbart3,Falk Butter2|,Steven E.Jacobsen1,8|DNA methylation generally functions as a repressive transcriptional signal,but it is alsoknown to activate gene expression.In either case,the downstream factors remainlargely unknown.By using comparative interactomics,we isolated proteins in Arabidopsisthaliana that associate with methylated DNA.Two SU(VAR)3-9 homologs,the transcriptionalantisilencing factor SUVH1,and SUVH3,were among the methyl reader candidates.SUVH1and SUVH3 bound methylated DNA in vitro,were associated with euchromatic methylation invivo,and formed a complex with two DNAJ domain-containing homologs,DNAJ1 and DNAJ2.Ectopic recruitment of DNAJ1 enhanced gene transcription in plants,yeast,and mammals.Thus,the SUVH proteins bind to methylated DNA and recruit the DNAJ proteins to enhanceproximal gene expression,thereby counteracting the repressive effects of transposoninsertion near genes.DNA methylation frequently marks trans-posable elements(TEs)in eukaryotic ge-nomes(13).In plants,the RNA-directedDNA methylation(RdDM)pathway is re-sponsible for the initial establishment ofmethylation in CG,CHG,and CHH contexts(4).TE insertions can exert a transcriptional effectonneighboringgenes(58),andpromotermeth-ylation is typically associated with gene repres-sion(9).However,exceptionsexistwherepromotermethylationisrequiredforgeneexpression(1014).The downstream factors that perceive meth-ylation to mediate these divergent transcrip-tional effects are still poorly characterized,andlittle is known of how methylation can stimu-late gene transcription.To identify proteins in Arabidopsis thalianathat recognize methylated DNA,we incubatednuclear extract from floral bud tissue with eithermethylated or unmethylated biotinylated double-stranded DNA oligonucleotides,affinity purifiedthe DNA,and subjected the associated proteinsto high-resolution mass spectrometry followedby label-free comparative analysis(15)(fig.S1).WeusedDNAsequencesthatarenaturallymethyl-ated in vivo and two distinct DNA sequences foreach of the CG,CHG,and CHH methylation con-texts(fig.S2).A total of 41 proteins were signifi-cantly methyl enriched in at least one pull-downassay,including many candidates with knownor predicted methyl-binding activity involvedin gene silencing and methylation control(fig.S3).By requiring that candidates be significantlyenriched in both DNA sequences for each ofCG,CHG,and CHH,we obtained a stringentlist of 10 candidates(Fig.1A).Of these,relativelylittle is known aboutthe role of the highly relatedSUVH1 and SUVH3 proteins(16)or the DNAJproteins.Recently,SUVH1 was isolated from an anti-silencing screen and was shown to promote theexpressionofpromotermethylatedgenes(17).AsSUVH1 and SUVH3 contain a SET-and RING-associated(SRA)domain(18),they are predictedto bind methylated DNA directly.Using fluores-cence polarization(FP)and microscale thermo-phoresis(MST),we confirmed an SRA-dependentmethyl-bindingpreferenceforrecombinantSUVH1and SUVH3 proteins,respectively,in CG,CHG,and CHH contexts(Fig.1B and fig.S4).Chroma-tin immunoprecipitation sequencing(ChIP-seq)oftransgeniclinesexpressingFLAG-taggedSUVH1or SUVH3 showed that their localization wasessentially identical(fig.S5A)and that they co-localized with CHH methylation deposited by theRdDM pathway(Fig.2A and fig.S5B).SUVH1andSUVH3displayedenrichment directly overNRPE1 sites(19)the largest subunit of theRdDM component RNA polymerase V(Pol V)(Fig.2B and fig.S5C)and showed preferentiallocalization over short TEs and at the edges oflong TEs(Fig.2C and fig.S5D),which are hall-marksofRdDMlocalization(20,21).Therewasapositive correlation between SUVH1 and SUVH3enrichment and RdDM-deposited CHH methyl-ation(mCHH)at both local and genome-widescales(fig.S5,E to H).Using random forestregression,we observed that mCHH was thestrongest predictor for SUVH1 binding in vivo(Fig.2,D and E).The nearly perfect colocalization of SUVH1with RdDM sites predicts that RdDM pathwaymutants might reduce SUVH1 occupancy.ChIP-seq of SUVH1 in nrpd1,nrpe1,or drm1/2 RdDMmutant backgrounds(4)showed that SUVH1enrichment was essentially eliminated(Fig.2Fand fig.S6).To exclude the possibility that inter-action with RdDM proteins,rather than DNAmethylation itself,was responsible for SUVH1recruitment,we compared ChIP-seq results foranSRAdomainaminoacidchangemutantwithtyrosine-277 mutated to alanine(Y277A)thatabrogated methyl binding,SUVH1Y277A(Fig.1B).Indeed,SUVH1Y277Ashowed highly reduced re-cruitment and association with CHH methyla-tion(Fig.2G and fig.S7).Whole-genome bisulfite sequencing(WGBS)revealed that SUVH1 ChIP-seq peaks were char-acterized by local CHH methylation maxima andthat in suvh1,suvh3,and double mutant suvh1suvh3plants,methylationlevelswereunperturbed(17)(fig.S8A).This indicated that SUVH1 andSUVH3 are not required for methylation main-tenance and act strictly as methyl readers.RNAsequencing(RNA-seq)of suvh1,suvh3,and suvh1suvh3 confirmed many ofthe previously identified(17)promoter methylated genes that requireSUVH1 for expression(fig.S8B)and showed re-duced expression at genes proximal to RdDMsites(22)(fig.S8C).SUVH1 and SUVH3 might enhance transcrip-tion by directly impacting chromatin(18),as bothencode SET domains of the SU(VAR)3-9 familythat typically methylate histone H3s lysine-9(23).However,we were unable to detect his-tone methyltransferase(HMT)activity in vitro(fig.S9)or changesindimethylationofhistone3lysine-9(H3K9me2)levelsinsuvh1suvh3mutantsin vivo(17)(fig.S10).Furthermore,SUVH1Y524FandSUVH1Y638FpredictedHMTcatalyticmutants(18),but not the SUVH1Y277Amethyl-binding mu-tant,were able to complement suvh1,indicatingthat HMT activity is nonessential for function invivo(fig.S11).Chromatinaccessibility,asprofiledby ATAC-seq(a sequencing technique based onan assay for transposase-accessible chromatin),was also unchanged in suvh1 suvh3 mutants(fig.S12).Next,weassessedwhetherSUVH1andSUVH3might enhance transcription by acting as arecruitment platform(24).Immunoprecipitationfollowed by massspectrometry(IP-MS)of SUVH1andSUVH3identifiedthat eachpulleddowntheother and also DNAJ1 and DNAJ2(Figs.1A and3A and fig.S13).IP-MS of DNAJ1 and DNAJ2RESEARCHHarris et al.,Science 362,11821186(2018)7 December 20181 of 41Department of Molecular,Cell and Developmental Biology,University of California at Los Angeles,Los Angeles,CA90095,USA.2Quantitative Proteomics,Institute of MolecularBiology,55128 Mainz,Germany.3Center for Epigenetics,VanAndel Research Institute,Grand Rapids,MI 49503,USA.4National Key Laboratory of Plant Molecular Genetics,CASCenter for Excellence in Molecular Plant Sciences,ShanghaiCenter for Plant Stress Biology,Shanghai Institutes forBiological Sciences,Chinese Academy of Sciences,Shanghai201602,China.5State Key Laboratory of Ecological PestControl for Fujian and Taiwan Crops,Fujian Agriculture andForestry University,350002 Fuzhou,China.6Department ofBiological Chemistry,David Geffen School of Medicine,University of California,Los Angeles,Los Angeles,CA 90095,USA.7Institute of Oceanography,Minjiang University,350108Fuzhou,China.8Howard Hughes Medical Institute,Universityof California at Los Angeles,Los Angeles,CA,USA.*These authors contributed equally to this work.Present address:Department of Microbiology,Faculty of Medicine,Chiang MaiUniversity,Chiang Mai,Thailand 50200.Present address:Genomic Institute of the Novartis Research Foundation,10675John Jay Hopkins Dr.,San Diego,CA 92121,USA.Present address:Helmholtz Zentrum Mnchen,Institute of Biochemical PlantPathology,85764 Neuherberg,Germany.|Corresponding author.Email:f.butterimb-mainz.de(F.B.);jacobsenucla.edu(S.E.J.)on December 27,2018 http:/science.sciencemag.org/Downloaded from showed that each of these pulled down theother and also SUVH1 and SUVH3(Fig.3A andfig.S13),indicating that SUVH1,SUVH3,DNAJ1,and DNAJ2 interact in vivo.We confirmed theinteractions between SUVH1 and SUVH3 withDNAJ and DNAJ2 by coimmunoprecipitation inNicotiana benthamiana and in yeast two-hybridassays(figs.S14 and S15).To assess the strengthof the interaction,we expressed all four proteinsin the same bacterial cell and performed affinitypurification of either SUVH1 or SUVH3,findingthatbothDNAJ1andDNAJ2remainedassociatedeven under 500mM NaCl conditions(fig.S16).DNAJ1 and DNAJ2 lack any discernible methyl-bindingdomain,butthey are robustly associatedwithSUVH1andSUVH3,suggestingthatSUVH1and SUVH3 may be responsible for recruitingDNAJ1 and DNAJ2 to methylated DNA(Fig.1A).WerepeatedaCHHcontextpulldownexperimentwithsuvh1 suvh3anddnaj1dnaj2doublemutantplants.DNAJ1 and DNAJ2 were no longer asso-ciated with methyl-DNA in suvh1 suvh3,whileSUVH1 and SUVH3 methyl-DNA binding wasunaffected in dnaj1 dnaj2(fig.S17).Thus,SUVH1and SUVH3 are required to recruit DNAJ1 andDNAJ2 to methylated DNA.We performedChIP-seq of DNAJ1 and DNAJ2 and found atight genome-wide correlation with SUVH1 andSUVH3(Fig.3,BandC,andfig.S18,AandB).Aswith suvh1 suvh3,there was no effect on DNAmethylation levels in dnaj1 dnaj2 mutants,con-sistent with a downstream reader function(fig.S18C).To assess whetherDNAJ1 and DNAJ2arerequired for the transcriptional enhancementactivity of SUVH1 and SUVH3,we performedRNA-seq on dnaj1,dnaj2,and double mutantdnaj1 dnaj2 plants.The dnaj1 dnaj2 transcrip-tome wasstrongly positively correlated with thatofsuvh1 suvh3(Fig.3Dandfig.S19),and RdDMproximal genes showed reduced expressionin both suvh1 suvh3 and dnaj1 dnaj2 doublemutants(fig.S20).ROS1 is one of the fewloci known to require methylation forexpres-sion(11,12),and indeed we observed reducedexpression of ROS1 in both the suvh1 suvh3 anddnaj1 dnaj2 backgrounds,despite methylationlevels being maintained(fig.S21).Furthermore,genes withpromoters proximaltoSUVH1peaksgenerally showed reduced expression in boththe suvh1 suvh3 and dnaj1 dnaj2 double mu-tants(Fig.3E).Together,these data indicatethat DNAJ1 and DNAJ2 interact with SUVH1and SUVH3,are recruited to sites of RdDM,and promote the expression of proximal genes.The yeast two-hybrid experiments revealedthat binding domain(BD)-fused DNAJ1 inducedexpression of the reporter even when cotrans-formed with an unfused activation domain con-struct(fig.S15).Thissuggested that DNAJ1alonemay be sufficient to stimulate expression of thereporter,which we confirmed in a yeast one-hybrid assay(fig.S22A).We fused DNAJ1 to azinc finger protein(ZF108)(24)behind theUBQ10 promoter and cotransformed it intoN.benthamiana with a reporter construct con-tainingeithertheZF108 targetsiteorascrambledtarget site in the promoter region.Expression ofthe ZF108 target reporter was increased by ap-proximately threefold above that of the scrambledpromoter(fig.S22B).ToassesswhetherDNAJ1canfunction in a mammalian context(25),we trans-fected N2a cells andfound that Gal4 DNA-bindingdomain(Gal4BD)fused DNAJ1 was able to stim-ulatetranscriptionofthereporterby5-to10-fold(fig.S22C).Next,wegeneratedstabletransgenicA.thalianalines using the UBQ10:ZF108-DNAJ1 construct.The first-generation independent transgenic linesdisplayed severe morphological defects(fig.S23).RNA-seq and ChIP-seq(Fig.4A)on these UBQ10:ZF108-DNAJ1 lines showed that up-but notdown-regulatedgenesweresignificantly enrichedfor overlap with ZF108-DNAJ1 ChIP-seq peaks(observed over expected=2.26,hypergeometrictest P=7.7e71)(fig.S24).As controls,we gener-ated UBQ10:ZF108-YPET and UBQ10:DNAJ1(without ZF108)transgenic plants and found noHarris et al.,Science 362,11821186(2018)7 December 20182 of 4Fig.1.Comparative interactomics identifies methyl reader proteins.(A)Heatmap of methyl-binding preferences for proteins identifiedas significantly enriched in two different underlying DNA sequencesper methyl-cytosine(mC)context(mCG,mCHG,mCHH).NA,the proteinwas not detected.FWA,MEA,SDC,and SUP represent four in vivo-methylated loci.Probes are listed in fig.S2.(B)FP binding assays toquantify the interaction of SUVH1 with methylated or unmethylated probesin CG,CHG,and CHH contexts(left)or an amino acid change version,SUVH1Y277A,predicted to abrogate methyl binding(18)(right).Bindingaffinities are indicated by dissociation constants(Kd)values.Errorbars represent SEM of technical replicates.The data are representativeof two independent experiments.RESEARCH|REPORTon December 27,2018 http:/science.sciencemag.org/Downloaded from Harris et al.,Science 362,11821186(2018)7 December 20183 of 4Fig.2.SUVH1 isrecruited by RdDM-associated mCHH.(A)SUVH1 enrichment atloci defined by loss ofmethylation(hypomethyla-tion).Differentially methy-lated regions(DMRs)inmutant genotypes are indi-cated.The DRM1 andDRM2 methyltransferasesare responsible for mCHHat RdDM target sites,whilemCG,mCHG,and hetero-chromatic mCHH aremaintained by MET1,CMT3,and CMT2,respec-tively(1).*,met1 hypo CGDMRs that overlap withdrm1/2 hypo CHH DMRswere removed.(B)SUVH1enrichment at NRPE1peaks.(C)SUVH1 enrich-ment at NRPE1-associatedshort(5-kb)TEs.(D)Relativeimportance of genomicfeatures in predicting SUVH1 binding,based on the random forest regressoralgorithm.Error bars represent SEM from five random permutations of thetraining set.(E)Area under receiveroperating characteristic curves(AUC)modelaccuracyusingallfeatures(left)vs.accuracyusingmCHHalone(right).(F)Boxplot of SUVH1 enrichment in suvh1,nrpe1,nrpd1,and drm1 drm2mutantbackgroundsatSUVH1peaks.(G)ScatterplotofSUVH1overSUVH1Y277Aenrichment vs.mCHH methylation percentage at SUVH1 peaks.Line of best fitis shown in blue,with adjusted R2and P values indicated.Data in the lowerpanel indicate kernel density for mCHH.Average methylation levels andenrichmentarecalculatedfromthe200-bpregionssurroundingthepeaksummits.Fig.3.SUVH1,SUVH3,DNAJ1,andDNAJ2 interact,colocalize,and arerequired for theexpression ofproximal genes.(A)IP-MS results fortagged lines.Onlyproteins present ineach of the fourtransgenic but notwild-type(WT)pull-downs are presented.NSAF,normalizedspectral abundancefactor,averaged fromtwo biological repli-cates.(B)Represent-ative browser trackshowing ChIP-seq ofSUVH1,SUVH3,DNAJ1,and DNAJ2(normalized reads,FLAG-tagged versionsminus WT)(top fourlines)and methylationfraction(bottomthree lines)at amethylated locus.(C)Pearsons correlation of genome-wide ChIP-seqprofiles at 1-kb resolution.H3K23ac from(20)was used as an outgroupcontrol.(D)Scatterplot of FPKM fold change over WTof dnaj1 dnaj2double vs.suvh1 suvh3 double at genes that were differentially expressedin suvh1 suvh3.Line of best fit is shown in red,with adjusted R2and P values indicated.(E)Boxplot of expression change for genesproximal to SUVH1 binding sites.n,number of genes.*P 0.05(Mann-Whitney test).RESEARCH|REPORTon December 27,2018 h