Distinct Roles of Polycomb Group Gene Products in Transcriptionally

CORRECTION Distinct roles of Polycomb group gene products in transcriptionally repressed and active domains of Hoxb8 Yu-ichi Fujimura, Kyo-ichi Isono, Miguel Vidal, Mitsuhiro Endoh, Hiroshi Kajita, Yoko Mizutani-Koseki, Yoshihiro Takihara, Maarten van Lohuizen, Arie Otte, Thomas Jenuwein, Jacqueline Deschamps and Haruhiko Koseki

There were errors published in Development 133, 2371-2381.

The authors have become aware of errors with the display of data in Fig. 4E, Fig. 5A and Fig. 6. These errors are detailed below, and the authors have also provided the original data for these figures, which are included as part of the correction.

In Fig. 4E, the Rnf2 and trimH3-K27 lanes were spliced together without marking, removing intervening lanes from the original gel. In preparing the figure, the Hoxb6 Rnf2 lanes were duplicated as Hoxb8 Rnf2. In addition, contrast adjustments were applied to the original image that obscured weak bands in the Rnf2–/– lanes that we assume are the result of non-specific binding of chromatin to the beads during the immunoprecipitation procedure. Finally, the control lanes are wrongly labelled as Hprt; the negative control used for this experiment was in fact Adam34. These errors are corrected in the revised Fig. 4 and legend shown below, and the original data for this experiment are provided as ‘Fig. 4E original data’.

Fig. 4 (revised)

– – Fig. 4. De-repression of Hox genes in Rnf2 / MEFs and ES cells. (A) Conditional depletion of Rnf2 leads to de-repression of Hoxb8 in MEFs derived from the cranial part of Rnf2fl/fl 9.5 dpc embryos. (Top) Infection of Cre-expressing adenovirus vector to MEFs derived from Rnf2fl/fl embryos (fl/fl) depleted the Rnf2 gene products, whereas the wild-type (+/+) MEFs were unaffected. Lamin B was used as a control. (Bottom) The expression of Hoxb8 was induced by infection of – – Cre-expressing adenovirus vector in Rnf2fl/fl MEFs (fl/fl), but not in the wild type (+/+). (B) Rnf2 / (–/–) ES cells were derived from Rnf2fl/fl (fl/fl) ES cells by – – transient overexpression of Cre-recombinase. Rnf2 was re-expressed by transfecting Rnf2 / ES cells with a construct expressing Myc-tagged Rnf2 (tr). The expression of endogenous and transfected Rnf2 was examined by using anti-Rnf2 (left) and -Myc (middle) antibodies. CBB staining was used as a loading control – – (right). (C) The expression of Hox cluster genes in Rnf2fl/fl (fl/fl), Rnf2 / (–/–) and Rnf2 transfected (tr) ES cells was compared by RT-PCR. The quantity of synthesized cDNA from respective cells was equalized by comparing the relative amounts of β-actin transcripts. (D) The expression of Phc1 and Cbx2 gene – – products was reduced in Rnf2 / ES cells (–/–) in comparison with the wild type (fl/fl), whereas the expression of RYBP (another Rnf2-binding protein that is not found in hPRC-H complex or class 1 PcG proteins) was not altered. (E) Rnf2 association and H3-K27 trimethylation at Hox promoter regions were compared – – between Rnf2fl/fl and Rnf2 / ES cells. For the ‘Input’, genomic DNA extracted from the original whole cell lysate equivalent to the 1/40 volume of that used for the

ChIP analysis was subjected to the PCR. Adam34 was used as a negative control. White lines indicate the removal of intervening gel lanes. DEVELOPMENT

3431 CORRECTION Development (2014) 141, 3431-3436 doi:10.1242/dev.115667

In Fig. 5A, for all samples except Hoxa4, the order in which the lanes are shown in the figure does not represent the order in which they were run on the original gel, where samples were run in the order: Suz12, Eed3, Rnf2, trimH3-K27, Input. For Hoxb1 and Hoxb4, the lanes were exchanged in the figure but this splicing was not appropriately marked. For Hoxb6 and Hoxb8, we failed to exchange these lanes and therefore the labelling was incorrect in the published figure. In addition, the control lanes were wrongly labelled as Hprt; the negative control used was Adam34.Theseerrors are corrected in the revised Fig. 5 and legend shown below, and the original data for this experiment are provided as ‘Fig. 5A original data’.

Fig. 5 (revised)

– – Fig. 5. De-repression of Hox genes in Suz12 / ES cells correlates with reduction of Rnf2 association to Hox genomic regions. (A) The association of – – Suz12, Eed, Rnf2 and H3-K27 trimethylation at the Hox promoter regions in the wild-type and Suz12 / ES cells. Whole-cell lysates prepared from approximately – – the same number of wild type (+/+) and Suz12 / (–/–) ES cells were subjected to ChIP analyses using anti-Suz12, -Eed, -trimethylated H3-K27 and -Rnf2 antibodies. For the ‘Input’, genomic DNA extracted from the original whole cell lysate equivalent to the 1/40 volume of that used for the ChIP analysis was – – subjected to the PCR. Adam34 was used as a control. (B) The expression of Hox cluster genes in the wild type (+/+) and Suz12 / (–/–) ES cells was compared by RT-PCR. Hprt was used as a control. White lines indicate the removal of intervening gel lanes or their exchange.

In Fig. 6A, the original experiment included samples taken from Rnf110+/–; Bmi1–/– animals. These were removed from the published version, but this splicing was not marked. In Fig. 6B, the original experiment included samples taken from Phc1–/–; Phc2+/–. We chose to show only data from the double heterozygote and double homozygote mutant, but in the case of Hoxb8 accidentally removed the Phc1+/–; Phc2+/– lanes instead of Phc1–/–; Phc2+/–. As can be seen in the original data, the results are highly similar for both genotypes. These errors are corrected in the revised Fig. 6 and legend shown below, and the original data for these experiments are provided as ‘Fig. 6 original data’.

Fig. 6 (revised)

–/– –/– –/– –/– Fig. 6. Decreased H3-K9 acetylation at the first exonic region of Hoxb8 in the posterior tissues of Rnf110 Bmi1 and Phc1 Phc2 embryos at 9.5 +/– +/– –/– –/– dpc. (A) Degree of H3-K9 acetylation in the anterior (A) and posterior (P) regions were compared in Rnf110 Bmi1 and Rnf110 Bmi1 embryos. The β-actin –/– +/– promoter was used as a positive control. (B) Degree of H3-K9 acetylation in the anterior (A) and posterior (P) regions were compared in Phc1 Phc2 , +/– –/– –/– –/– Phc1 Phc2 and Phc1 Phc2 embryos. The β-actin promoter was used as a positive control. In this study, the negative control ChIPs (A– and P–) were performed with rabbit IgG. White lines indicate the removal of intervening gel lanes. DEVELOPMENT

3432 CORRECTION Development (2014) 141, 3431-3436 doi:10.1242/dev.115667

The authors apologise to readers for these errors. The editors of the journal and the authors’ Research Integrity Office have examined this paper in detail and find that, although there were substantive errors in data presentation, these do not affect the conclusions of the paper.

Fig. 4E original data

Without lane splicing:

From laboratory notebook: DEVELOPMENT

3433 CORRECTION Development (2014) 141, 3431-3436 doi:10.1242/dev.115667

Fig. 5A original data

Without lane splicing:

From laboratory notebook: DEVELOPMENT

3434 CORRECTION Development (2014) 141, 3431-3436 doi:10.1242/dev.115667

Fig. 5A original data (cont.) DEVELOPMENT

3435 CORRECTION Development (2014) 141, 3431-3436 doi:10.1242/dev.115667

Fig. 6 original data

Without lane splicing: DEVELOPMENT

3436 complex, known asclass1,includesExtra sex combs(Esc)and et al.,2002;MullerCao2002).Thefirsttype of types ofmultimericproteincomplexes (Shaoet al.,1999;Czermin al., 1992).PcGgeneproductshave beenshown toformatleasttwo histone tails(Milneetal.,2002;NakamuraTamkun et complex andassociatewithcatalyticactivities thatmodifycore genes encodesubunits oftheSWI/SNFchromatinremodeling concern theremodelingofchromatinstructure,asseveral ofthese embryogenesis. AmajorfunctionoftrxGgeneproductsappearsto the chromatin,freezetranscriptionalstatesdeterminedearlyin nuclear factors that,byformingmultimericproteincomplexes on (Kennison andTamkun, 1988).The genes have beenidentifiedassuppressors oftrxGmutations repression (Paro, 1995;Pirrotta,1997).Indeed,some ofthePcG appropriate segments, whereasthePcGareinvolved intheir products arerequiredtomaintaintheactivity ofHoxgenesinthe trithorax group(trxG)andPolycomb(PcG).ThetrxGgene genes (Hoxgenes)ismaintainedbytwo groupsofproteins:the Amsterdam, TheNetherlands. Vienna, Austria. Vienna, Institute ofMolecularPathology, Biocenter, TheVienna DrBohrgasse7,A-1030 of Amsterdam, Kruislaan406,1098SMAmsterdam, TheNetherlands. *These authorscontributedequallytothis work Netherlands. Developmental andCellBiology, Ramiro deMaeztu9,28040Madrid,Spain. In INTRODUCTION KEY WORDS:Polycomb,Hox,Mouse,Chromatin, Immunoprecipitation transcriptionally repressedandactivedomainsof active statusviaprofoundregulationofH3-K9acetylation.Thepresentstudyindicatesdistinctrolesforclass2PcGcomplexe outside theirexpressiondomain.Withinthedomains,class2PcGcomplexesappearedtomaintaintranscriptionall and class2PcGcomplexes,mediatedbytrimethylatedH3-K27,playdecisiverolesinthemaintenanceofHoxgenerepression the PcGclass2componentRnf2or marks temporallycoincidewiththemisexpressionofHoxgenesinPcGmutants.Functionalanalyses,usingmutantallelesimpairi Hoxb8 immunoprecipitation revealedthattheassociationofPcGproteins,andH3-K9acetylationH3-K27trimethylationaround methylated onlysine4(H3-K4),andoftranscriptionallyinactivechromatintrimethylated27(H3-K27).Chromatin tissues. Inparallel,wefollowedthedistributionofhistonemarkstranscriptionallyactiveH3acetylatedonlysine9(H3-K focused onthebindingpatternsofPcGgeneproductstoflankingregions To addressthemolecularmechanismsunderlyingPolycombgroup(PcG)-mediatedrepressionofHoxgeneexpression,wehave Yu-ichi Fujimura transcriptionally repressedandactivedomainsof Distinct rolesofPolycombgroupgeneproductsin Development 133,2371-2381(2006)doi:10.1242/dev.02405 Accepted 12April 2006 † 3 1 Japan. Medicine, Hiroshima University, 1-2-3Kasumi,Minami-ku,Hiroshima 734-8553, Yokohama 230-0045,Japan. Haruhiko Koseki Yoshihiro Takihara Author forcorrespondence (e-mail:[email protected]) Department ofStemCellBiology, Research InstituteforRadiationBiologyand RIKEN Research CenterforAllergyandImmunology, 1-7-22Suehiro, Tsurumi-ku, Drosophila 4 Division ofMolecularGenetics,TheNetherlandsCancerInstitute,1066CX were distinctintissuesexpressingandnotthegene.Weshowthatdevelopmentalchangesoftheseepigenetic 7 Hubrecht Laboratory, Uppsalalaan83584CTUtrecht, The , theregionally restrictedexpression ofhomeotic 1, 1,† *, Kyo-ichiIsono 3 2 Centro deInvestigacionesBiologicas,Departmentof , MaartenvanLohuizen 5 Swammerdam InstituteforLifeSciences,University trxG Suz12 1, and *, MiguelVidal mutation decreasingH3-K27trimethylation,revealedthatinteractionsbetweenclass1 PcG Hoxb8 genes encode 4 , ArieOtte 6 Research gene. 2 , Mitsuhiro Endoh 5 , ThomasJenuwein maintenance has beenrevealed previously bythe effect ofmutations Nakamura etal.,2002).Theinvolvement ofclass2PcG inthis E(z), mediatesH3-K27trimethylation (Milneetal.,2002; H3 (H3-K4)atseveral Hoxgenes,whereasEzh2,ahomologueof that theMllproteinregulates theacetylationoflysine4histone restricted expression of Hox clustergenes.Indeed,ithasbeenshown 1999), suchmodificationscould beinvolved inmaintaining spatially with activities thatmodifyhistonetails(van derVlagandOtte, 1996). AsMllandclass1PcGcomplex areknown tobeassociated involvement inrepressingHoxgeneexpression (Schumacheretal., mutation of post coitus(dpc)(Yu etal.,1998).Conversely, ahypomorphic Hox geneexpressions hasbeenobserved inmice lacking trxG andbothclassesofPcGgeneproducts.Silencingseveral cluster genesdependsonthecontrolofmammalianhomologues of Levine etal.,2002;Shao1999). arrays andsubsequenttranscriptioninvitro(Kingetal.,2002; efficiently blockSWI/SNF-dependentremodelingofnucleosomal Paro, 1998;Wang etal.,2004b).Purifiedclass2corecomplexes which inturnmaysilencetarget genesflanked byPREs(Cavalli and serves asasignalfortherecruitmentofclass2PcGcomplexes, PcG-mediated trimethylationofhistoneH3lysine27(H3-K27) silence oftarget genes.Recently, ithasbeenreportedthatclass1 elements (PREs),thiscomplex canmaintainthetranscriptional interacting withDNA regulatory sequencestermedPcGresponse in (Ph), Dring(Sex combextra; Sce)andSex combonmidleg (Scm) contains Polycomb(Pc),Posteriorsex combs(Psc),Polyhomeotic 2002; Caoetal.,2002).Thesecondtypeofcomplex (class2) modification ofhistonetails(Czerminetal.,2002;Muller and methyltransferaseactivities suggestsafunctioninthe Enhancer ofzeste[E(z)],anditsassociationwithhistonedeacetylase mammalian homologueof In mammals,ithasbeendemonstratedthattheexpression ofHox Drosophila Eed 1 (Shao etal.,1999).Ithasbeenshown that,by , Hiroshi Kajita , ahomologueof Hoxb8 6 , JacquelineDeschamps gene inexpressingandnon-expressing rspiatrx Drosophila 1 , Yoko Mizutani-Koseki RESEARCH ARTICLE Drosophila Esc Hoxb8 , between8.5and9.5days 7 , revealed its and 9) and s in Mll 2371 ng y , a 1 ,

DEVELOPMENT ( Pcgf2 MATERIALS ANDMETHODS shown toinvolve theregulation ofH3-K9acetylation. PcG complex proteinsinthetranscriptionallyactive domainwas domain, asitisthecasein in maintainingtherepressionofHoxgenesoutsidetheirexpression complex, mediatedbytrimethylatedH3-K27,playsadecisive role methylation, weshow thattherecruitmentofclass2PcG 2 PcGcomplexes, and By usingmutantallelesfor differs inembryonictissuesexpressing andnotexpressing thegene. H3-K4 methylationandH3-K27trimethylationaround known torequiretheclass2PcGproteinsforitsposterior restriction. chromatin) tothegenomicregion flanking and H3-K27trimethylation(amarkoftranscriptionallyinactive H3-K4 methylation(marksoftranscriptionallyactive chromatin) documenting theassociationofPcGproteins,H3-K9acetylation, Hox lociindeveloping embryos.We have addressedthisissueby tail modificationshasnotyetbeenwidelydocumentedaroundthe because therelationshipbetweenPcGproteinbindingandhistone products toregulate Hoxgenesisnotwellunderstood,inpart the molecularmechanismsusedbymammaliantrxGandPcGgene of Hoxgenesoutsidetheirexpression domains.However, todate, proteins mayparticipateinmaintainingtranscriptionallysilentstates cranially thereafter(Akasakaetal.,2001).Thus,class2PcG tissues of8.5dpcembryos,but was progressively de-repressed Drosophila Psc in micedoublydeficientfor Lugt etal.,1996;Suzuki2002;Takihara etal.,1997).Notably, in 2372 rmtyae 3K abtAtsrmUpstateBiotechnology(LakePlacid,NY) UpstateBiotechnology(LakePlacid,NY) UpstateBiotechnology(LakePlacid,NY) Hameretal.(2002) Petersetal.(2003) Hameretal.(2002) Petersetal.(2003) Antiserum Petersetal.(2003) Abcam(Cambridge,UK) UpstateBiotechnology(Lake Placid,NY) Antiserum Petersetal.(2003) Schoorlemmeretal.(1997) Antiserum Suzukietal.(2002) Monoclonal Antiserum Monoclonal Miyagishimaetal.(2003) Antiserum Schoorlemmeretal.(1997) Rabbit Antiserum Antiserum Antiserum Rabbit Antiserum Antiserum Companyorreference Rabbit Atsutaetal.(2001) Antiserum Mouse Mouse Rabbit Monoclonal Rabbit Antiserum Rabbit Trimethylated H3-K4 Rabbit Rabbit Rabbit Rabbit Monoclonal Dimethylated H3-K4 Rabbit Trimethylated H3-K27 Mouse Dimethylated H3-K27 orantiserum? Rabbit Trimethylated H3-K9 Mouse Dimethylated H3-K9 Acetylated H3-K9 Suz12 Species Ezh2 Eed Rnf110 Cbx2 Phc1 Phc1 Ring1 Rnf2 Specificity Table 1.Alistofantibodiesusedinthisstudy Antibodies usedarelistedintheTable 1. Antibodies AdCre virusatMOI5. i.f.u./ml andmouseembryonicfibroblast(MEFs)wereinfectedwith The generationofRnf2 Mice andcelllines described previously (deNapoles etal.,2004).Thegenerationof (Kanegae etal.,1995).AdCreviruswas concentratedtobe2.0 expressing theCRErecombinase(AdCre)was carriedoutasdescribed CRE-mediated deletionofRnf2inMEFcelllinesusingadenovirus deficient miceandEScellswillbedescribedinaseparatepublication. Ring1B We show thattheassociationofPcGproteins,H3-K9acetylation, Bmi1 – MouseGenomeInformatics), RESEARCH ARTICLE ) onHoxgeneexpression (Akasakaetal.,1996;van der ( Pcgf4 , Hoxb6 – MouseGenomeInformatics), Suz12 fl expression was normallylocalizedincaudal allele andRnf2mutantEScellshasbeen Rnf2 Drosophila causing astrongdecreaseinH3-K27 Rnf110 , whichencodeconstituentsofclass . Thepositive roleofclass2 and Phc1 Bmi1 Hoxb8 Monoclonal ( rae28 , homologuesof Rnf110 , aHoxgene ) and ( Mel18 Hoxb8 Suz12 ϫ Rnf2 10 –9 - ; 2005a). ChIP analysisbyusingEScellswas performed asdescribed(Isonoetal., supplementary material).PrimersusedinthisstudyarelistedTable 2. of experiments wereperformedatleastthreetimes(seeFig.S3inthe the initiallysatesandanenrichmentvalue was determined.Every series estimated byreferringtotheserialdilutionsof‘Input’DNA isolatedfrom of eachgenomicregion inimmunoprecipitatedgenomicDNA was immunoprecipitated DNA wereusedastemplates.Therelative quantity out semi-quantitative PCR,seriallydiluted‘Input’DNA and carry for anti-Rnf2immunoprecipitationweresubjectedtothePCR.To and P–)derived from thesamevolume ofthechromatinfractionasused 20 ngofgenomicDNA was used.Mock-immunoprecipitatedDNA (A– DNA loadedinlane‘1’weresubjectedtoPCRreactions.Usually, 10to (see Fig.S2Ainthesupplementarymaterial). DNA andband-intensitieswerecomparedafterethidiumbromidestaining electrophoresed for5minutesinanagarosegel,next toseriallydilutedinput immunoprecipitation, thealiquotsofimmunoprecipitatedDNA were ‘Input’DNA (seeFigs1-3).To measuretheDNA yieldafter as WCE throughthesameprocedureasdescribedabove anddesignated Genomic DNA was alsoisolatedfromtheoriginalchromatinfractionor extracted byphenol-chloroformandconcentratedethanolprecipitation. heating at65°Cforreverse crosslinking,supernatantswerecollected, incubation with500 with 50 after 3hoursincubationwithProteinASepharosebeads. with rocking,for2hourstoovernight. Immunecomplexes werecaptured protein extracts wereincubatedwiththecorrectamountofantibodies,at4°C to performoptimalimmunoprecipitationforeachantibody. Precleared 40 wereaddedtothechromatinfractionorwholecelllysates(WCE)inorder This processwas bypassedinsomecases.CorrectamountsofNaClandNP- described inFig.S1thesupplementarymaterial(Orlandoetal.,1997). the embryonictissueswas purifiedbyCsClisopycnic centrifugationas developmental stages,underastereomicroscope.Chromatinfractionfrom Embryonic tissuesofinterestweredissectedfromembryos,atthedifferent Chromatin immunoprecipitation (ChIP)analysis 2005a). western blotanalysiswereperformedasdescribedpreviously (Isonoetal., Preparation ofwhole-cellextracts fromembryos,EScellsandMEFs using Mx3005Pmultiplex quantitative PCRsystems(Stratagene). was quantifiedbyreferringto adjusted bytheexpression of cellular RNA subjectedtoreverse transcriptase(RT)-PCR forHoxgeneswas performed asdescribedpreviously (Isonoetal.,2005a).Quantityoftotal RNA extraction fromEScells,reverse transcriptionandPCRreaction was Expression analysesforRNAandprotein Equivalent amountsofimmunoprecipitatedDNA tothatof‘Input’ To isolategenomicDNA fromimmunecomplexes, beadsweretreated ␮ g/ml ofRNaseAat37°Cfor30minutesfollowed byovernight ␮ g/ml proteinaseK/0.5%SDSat37°C.After3hours Gapdh ␤ -actin. expression byreal-timePCRanalyses Hoxb8 expression shown inFig.3A Development 133(12)

DEVELOPMENT Dual functionsofPcGcomplexesinHoxtranscription Hoxb8 genes was addressed.PcGassociationsatgenomicregions flanking embryos involves thedirect bindingofPcGgeneproductstoHox First, theissueofwhetherPcG-silencingHoxgenesindeveloping Hoxb8 Association ofmammalianPcGproteins tothe RESULTS material). Amounts ofDNA fromrespective tissueswerefurther absence ofprimary antibody(seeFig.S2Ain thesupplementary antibodies was four- toeightfoldhigherthanthatbroughtdown inthe fraction, theamountofDNA immunoprecipitatedbyanti-Rnf2 supplementary material).Usually, foragiven volume ofchromatin DNA isolated fromtheinitiallysates(seeFig.S2Ain genomic DNA was measuredbyreferringtoseriallydilutedgenomic material) (Suzukietal.,2002).Quantityofimmunoprecipitated Phc1, Ring1,Cbx2andRnf110(seeFig.S1inthesupplementary and subjectedtoimmunoprecipitationusingantibodiesagainstRnf2, Chromatin fractionsoftherespective embryonic tissueswerepurified and caudaltissuesweresubjectedtotheanalysis(Fig. 1A). after theremoval ofinternalvisceraandhindlimb buds. Onlycranial of thepinnaprimordiumandposterioredgesforelimbbuds, in theneuraltube,embryosweredissectedtransversely atthelevels 7th prevertebrae intheparaxialmesodermand atthecaudalhindbrain dpc. Astherostralboundaryof at12.5 and active caudaltissuesbyChIPusingembryonic tissues Adam34 Hoxa4 Hoxb9 Hoxb8 Positions* Hoxb6 Hoxb3 Hoxb8 Reverse Forward Hoxb8 Primer Table 2.PCRprimersusedinthisstudy *Position of5 itlD gatgtttagtaggatgcaaactccattata ggtagtagctttctgatggt D1 distal 1-516 actgtttgctttgctgctgtttag aacaggagacagagaaactggtac 1 proximal were comparedbetweentranscriptionallyrepressedcranial genomic region at12.5dpc Ј -most nucleotideofregion 1wasarbitrarilydesignatedas‘1’. 4cgacatggttgttggctgtt11581-11787 11220-11408 12309-12829 10964-11218 aacgaatttattgagaattc 10501-10792 tgtaggtctggcggcctcgcttt 10774-10985 10304-10527 taacataaccctcctggcaggccg 9908-10134 10115-10323 agccatcctcctgattcag accatcagaaagctactacc ctgacagaacttggctctgatg ctgtcatcagtcactctct 9685-9927 tctgagaactcccagcata D2 ttgtaagccctctttgaagct 8978-9505 atgagtataggagtctct 9484-9706 ctgcacctagtagtg 25 gtcatctttctggagtgata agagagtgactgatgacag accactctttgactctgtgt 24 8561-9085 catcagagccaagttctgtcag 23 acacccagaactgagctctc tatgctgggagttctcaga 7994-8584 22 7560-8032 7061-7529 ctgagcttcgcatccaggggta agagactcctatactcat ctagggcctgagagcactgagc 21 5746-6083 tatcactccagaaagatgac 20 6060-6315 acacagagtcaaagagtggt 6592-7041 19 agtcggggacgttttagtgtc gagagctcagttctgggtgt 5593-5764 18 cccaagaggaggcgagcctgg gctcagtgctctcaggccctag 17 cactagccaccagcctgggta tacccctggatgcgaagctcag caatgctcacagcgcgcatgc 16 4292-4565 tcacgtggtcagaagagg tctccacagcccccataaaac cagaagctattacgagatactacc 15 5045-5568 aggccaggctcgcctcctcttggg 4531-5091 14 tctctggtaactagaaccag cgctcttgggaagagatctaccca 13 ttagcatgcgcgctgtgagcattg 446-1006 cttgagggcgcatccaggg 12 ctccttcccttctcttgggggtcc caaagactgatgtgggggagt tggagctggagaaggagttccta 11 cacggcgcacgggttctgctggta tagaacagcgaagcctgcaaaagt actgttttatgggggctgtggaga 10 ccctggatgcgccctcaag 9 ttctacggctacgaccctctgcag 8 agggatgagaagggccgaggg gccgcggctgccatgcaggcttag 7 ggtgtttttccgtgactcccccac 6 tatgactacctcgttgtttg 5 gggtataaatttctgaaggttaag 4 3 2 tatgatagct gcggttatgatctattactac cgtgatggatgcygctagcc tggtagacagacggcaggct gcttgcagtctgcgtactgc accagccggcggcggctacg atgagtgggactaaggccctg agctccagccctggcttcgc ttggacgtttgcctcgactc atgtccatttctgggacgct atgagctcttatttcgtcaa atgagttcctatttcgtgaa atgcagaaagccacctacta Hoxb8 expression isatthelevel ofthe adjusted byPCRforthe Rnf2 tothedistal element(DE)locatedbetween embryonic tissues. Likewise, asignificantdifferential association of regulatory elementsof 5 1993). Therefore,Rnf2differentially bindstotheproximalcis- Hoxb8 transgenic approaches;regions 3and13includethepromotersof regions BH1100andKA thathave previously beenidentified by Regions 1and2,1011,correspondtothe region 17and18,nosignificantassociationofRnf2was observed. 19, 20,21,22,23,24,25orD2(Fig.1B,F, SupplementalFig.3).In tissues, whereasthiswas notthecaseforregions 4,5,6,7,8,12,14, and D1was morethanfourtimesdifferent incranialandcaudal a negative control(Brachvogel etal.,2002)(Fig.1B-E). Cbx2 orRnf110inembryonictissuesandthusturnedouttoserve as exclusively inadulttestes,isnotboundbyany ofRnf2,Ring1,Phc1, materials (Fig.1B-E).The enrichment value foreachregion of initial lysateswas alsosubjectedtothePCRevaluate the top; seeTable 2).SeriallydilutedgenomicDNA isolatedfromthe pairs ofprimersdefiningdiscreteregions inthe and posteriortissuesweresubjectedtosemi-quantitative PCRusing Equivalent amountsofDNA immunoprecipitatedfromtheanterior embryonic tissues(seeFig.S2Binthesupplementarymaterial). The associationofRnf2withregions 1,2,3,9,10,11,13,15,16 and Hoxb7 , respectively (Charitéetal.,1995;Vogels etal., tbx2 Ј Hoxb gene, whichisalsoboundbyRnf2in adam34 RESEARCH ARTICLE ee incranialandcaudal genes Hoxb8 gene, whichisexpressed in immunoprecipitated Hoxb Hoxb4 Hoxb8 locus (Fig.1F, and regulatory Hoxb5 2373

DEVELOPMENT diagonal line. the ‘Input’,andare represented bytheblackandgraybars,respectively. Genomic regions leftunexaminedare covered byboxes DNA from anteriorandposterior tissueswasestimatedbyreferring to‘Input’DNAisolated from theinitiallysatesandenrichm pairs listedinTable 1are shownbybars andnumericallyindicated.Therelative quantityofeachgenomic region inimmunoprec Oosterveen etal.,2003).Putative promotor regions are indicatedbyfoldedarrows. Thegenomicregions examinedbyPCRusings acting regulatory elements are represented byoverlyingboldbarsindicated asBH1100,KAandDE(Charitéetal.,1995;Vogels shown atthetop.Exonicregions are indicatedbyblackboxesandtheexonnumbersare numericallyshownintheboxes.Position and Rnf110associationtothe did appeartobesignificantlystrongerinthecaudaltissues. caudal tissues(Fig.1C).Nevertheless, inregion 3,theassociation for region 8,withoutany obvious differences betweencranialand association ofPhc1extended throughallregions examined except for regions 3and7(Fig.1D).IncontrasttoRnf2binding,the Similar toRnf2,Ring1associationwas seenincranialtissues,except (Oosterveen etal.,2003)was alsoseenincranialand caudaltissues. 2374 ( Fig. 1.ComparisonofPcGassociationswith ( chromatin fractionasusedforanti-Rnf2immunoprecipitation were subjectedtothePCR.The used. Immunoprecipitated DNAwasalsoseriallydiluted.Mock-immunoprecipitated DNA(A–andP–)derivedfrom thesamevolumeof immunoprecipitated DNAtothatof‘Input’loadedintolane1were subjectedtoPCRreactions. Usually10 to 20ngofgenomi isolated from theoriginalchromatin fractionsdesignated as‘Input’(seeFig.S2Ainthesupplementarymaterial)andequivalen antibody. AmountsofgenomicDNAimmunoprecipitated byanti-Rnf2 (A+orP+)were quantifiedbycomparingwithserially dilutedg between anteriorandposteriortissues.Thechromatin fractionpurifiedfrom AorPtissuewassubjectedtotheimmunoprecipitat (paraxial mesodermplusneuroectoderm) were subjected totheChIPanalyses.( regions 2and14.Allexperimentswere conductedas describedusinganti-Rnf110antibody. ( volume ofthechromatin fractionasusedforanti-Cbx2and-Ring1immunoprecipitation were subjectedtothePCR. ( subjected toPCRwere equivalenttothatof‘Input’DNAloadedintolane1.Mock-immunoprecipitated DNA(A–andP–)derivedfro of Cbx2andRing1toregions 2and14.AmountsofgenomicDNA(A+P+)immunoprecipitated byanti-Cbx2andanti-Ring1antibo A C ComparativeanalysisofPhc1associationtoregions 2and14betweenanteriorposteriortissues.( ) Embryonictissuesusedinthisstudy. embryosat12.5dpcwere dissectedasillustratedandanterior(A)posteri Wild-type ) RESEARCH ARTICLE Hoxb8 genomic surrounding between anteriorandposteriortissues.Genomicorganizationaround Hoxb8 genomic surrounding at12.5dpcbetweenanteriorandposteriortissues. bound to (Fig. 1D).Therefore,theseresultsindicatethatPhc1andCbx2are caudal tissueswithrespecttothechromatinassociationofCbx2 Furthermore, therewas nosignificantdifference betweencranialand both cranialandcaudaltissues(Fig.1E).Theseexperiments were in thecranialtissuesonly, but toregions 1,2,7and14itwas seenin state ofthegene.Rnf110associationtoregions 3and10was seen B ) ComparativeanalysisofRnf2associationtoregions 2,14andD1 Hoxb8 F ) SchematiccomparisonsofRnf2,Ring1,Phc1,Cbx2 adam34 genomic region irrespective ofthetranscriptional locus wasusedasanegativecontrol. D Comparativeanalysesforassociation ) E ) Rnf110associationto Development 133(12) t amountsof ent valuesagainst ipitated genomic crossed witha et al.,1993; or (P)tissues ion withanti-Rnf2 s ofknowncis- pecific primer Hoxb8 c DNAwas m thesame enomic DNA the dies gene is

DEVELOPMENT ( ‘Input’ DNAisolatedfrom the initiallysatesandenrichmentvaluesagainstthelysateare represented bytheblackand relative quantityofeachgenomic region inimmunoprecipitated genomicDNA from anteriorandposteriortissueswasestimatedby immunoprecipitation were subjectedtothePCR.(Right)SchematiccomparisonofH3-K9 acetylationbetweenanteriorandposterior Mock-immunoprecipitated DNA(A–andP–)derivedfrom thesamevolumeofchromatin fractionasusedforanti-acetylatedH3-K immunoprecipitated genomic DNA(A+andP+)byanti-acetylatedH3-K9subjectedtoPCRwere equivalenttothatof‘Input’DNAloa prepared from anteriorandposteriorpartsof12.5dpcembryoswere subjectedtoChIPanalysesbyusinganti-acetylatedH3-K9. are shown.( (right) are shown. K27 trimethylationat Differences betweenH3-K9acetylationandH3- gene. Rnf2 componentalsobindintissuesactively expressing theHox where thegeneisrepressed,whereasform(s)lackingatleast 2 PcGcomplexes predominantlyassociatewith supplementary material).Insummary, thecompleteformofclass performed threetimeswithsimilarresults(seeFig.S3inthe Dual functionsofPcGcomplexesinHoxtranscription between anteriorandposteriortissues. Fig. 2.ComparisonofH3-K9acetylation,H3-K4methylationand H3-K27trimethylationin locus andMEFsthatarederived from state ofHoxgeneshasbeenseenindeveloping embryosat correlation betweenhistonemodificationsandthetranscriptional and K14acetylatedresidues(Agaliotietal.,2002).Alinear recruitment ofTBPthroughtheinteractionTAF1 withtheH3-K9 marks have beenshown tobeoneoftheprerequisitesforefficient expressing embryonictissues,was addressed.These epigenetic K4 methylationat Following this,theissueofdegree ofH3-K9acetylationandH3- embryonic tissues acetylation andH3-K4methylationaroundthe between anteriorandposteriorembryonictissues.H3-K9 active transcriptionofHoxgenes,theseextents werecompared K9 acetylationandH3-K4methylationmaybeprerequisitesfor controls (Rastegar etal.,2004;Milne2002).Therefore,asH3- B ) Comparativeanalysisofdi-andtrimethylation ofH3-K4attheregion 2,7,8and9.Representative results (left)andschema C ) Comparativeanalysisoftrimethylation ofH3-K27attheregion 2,3and8.Representative results (left)andschematicsummar Hoxb8 genomic regions inexpressing andnon- Hoxb8 in cranialandcaudal Mll ( A ) ComparativeanalysisofH3-K9acetylationatregions 2,7,8and9.(Left)Whole-celllysates(WCE) mutants andwild-type Hoxb8 Hoxb8 was more in tissues Hoxd4 dpc, hasbeenreportedasresultingfromthe Loss ofspatialrestriction histone H3modificationsat Developmental kineticsofRnf2-associationand any oftheregions examined (Y.F. andH.K.,unpublished). trimethylation ofH3-K9ordimethylationH3-K27wereseenat expressing caudalpart(Fig.2B).Nosignificantlevels ofdi-or transcriptionally silentcranialembryonicpartthanintheHox- regions examined, H3-K27trimethylationwas moreabundant inthe compared betweencranialandcaudalembryonicregions. Atallthe degree ofH3-K9andH3-K27methylationatregions 2,3and8was to bemediatedbyclass1PcGcomplexes (Caoetal.,2002).The al., 2003).Inparticular, trimethylationofH3-K27hasbeenshown repression ateuchromaticpositions(Fischleetal.,2003;Lachner inactivation, PcG-mediatedgenesilencingandtranscriptional including theformationofcentromericheterochromatin,X transcriptionally silentchromatinandisinvolved inmany functions, (Rastegar etal.,2004)(Fig.2A,B). in thenon-expressing anteriorpartasreportedbyRastegar etal. abundant inthetranscriptionallyactive caudalembryonicpartthan progressively silencesHoxgeneexpression withinthistimewindow mutation (Akasakaetal.,2001).Similarly, The methylationofH3-K9andH3-K27constitutesimprintsfor Hoxb8 Hoxb6 genomic surrounding at12.5dpc expression, between8.5and9.5 RESEARCH ARTICLE Hoxb8 graybars,respectively. Rnf110 tic summary(right) Amounts of Mll referring to tissues.The / ded inlane1. 9 Bmi1 deficiency double y 2375

DEVELOPMENT black andgraybars, respectively. Stagesleft unexamined are indicated by boxescrossed witha diagonal line. estimated byreferring tothose of‘Input’DNAisolatedfrom theinitiallysatesandenrichment valuesagainstthe‘Input’are posterior tissues.Therelative quantityofeachgenomicregion inimmunoprecipitated genomicDNAfrom anteriorandposteriort antibodies. ( prepared from theanterior(A)andposterior(P)tissueswere used,whereas WCEwere usedforanti-acetylatedH3-K9and-trimet using anti-Rnf2,chromatin fractions bud afterremoving theviscera. ForChIP hindbrain andcaudallytotheforelimb level oftheposteriorend (~35 somites)were dissectedatthe ChIP analyses.( and respective WCEwere subjectedto at thelevelofsomite9/10boundary into anterior(A)andposterior(P)pieces dpc (about25somites)were bisected to ChIPanalyses.( unsegmented region (P)were subjected somite 9(A)andposterior WCE prepared from cranialregion upto newly generatedsomiteboundary. between somites9and10,atthe into three piecesattheboundary at the~13-somitestagewere dissected subjected toChIPanalyses.( prepared from respective tissueswere unsegmented (P)regions, andWCE segmented (A)andposterior boundary andseparatedintoanterior level ofthenewlygeneratedsomite somite embryoswere bisectedatthe described below. ( anterior andposteriortissuesas 1. Embryoswere bisectedintothe tissue of8.0dpcembryowasarbitrarily The between anteriorandposteriortissues. Relative quantitywascompared PCR againsttheexpression of stages, asquantiﬁedbyusingreal-time caudal tissuesatvariousdevelopmental expression of putative promotor region. H3-K27 trimethylationatthe association, H3-K9acetylationand Fig. 3.Temporal changesinRnf2 to laterdevelopmental stagesacrossthe association, H3-K9acetylationandH3-K27trimethylationinearly and Wijgerde, 1993).We carriedoutthekineticanalysesofRnf2 relative quantityprogressively increaseduntil12.5dpc(Deschamps expression of extensive insituhybridizationanalyses,caudallyrestricted by usingreal-timePCR(Fig.3A).Asreportedpreviously by Hoxb8 at be thecrucialperiodwhenPcGassociationandH3-K9acetylation (Yu etal.,1998).Thissuggeststhatthisdevelopmental stage might 2376 at thesix-somitestageandinbothcranialcaudalregions atthe dpc (Fig.3E).H3-K9acetylationwas observed in the cranialregion association graduallyincreasedandreachedamaximallevel at10.5 later stagesupto12.5dpc(Fig.3C-E).Therelative quantityofRnf2 tissue, attheeight-somitestage(Fig.3B).Thiswas alsothecaseat seen atthesix-somitestage.Itwas ﬁrstseen,exclusively incranial (region 3inFig.1F).In8.0dpcembryos,noRnf2associationwas Hoxb8 Prior totheChIPanalyses,were-examined theexpression of Hoxb8 gene inthecranialandcaudaltissuesquantitative manner RESEARCH ARTICLE play theirroleinmodulatinggeneexpression. / Gapdh F ) SchematiccomparisonsofRnf2association, H3-K9acetylationandH3-K27trimethylationattheregion 3betweenanteriorand Hoxb8 E Hoxb8 ) Embryosat10.5dpc D B ratio intheanterior ) Six-andeight- ) Embryosat9.5 in thecranialand was alreadyestablishedat8.0dpcandits C ( ) Embryos A Gapdh ) The Hoxb8 . Hoxb8 promoter region association ofRnf2withthe in thecaudal,region from8.5to12.5dpc.Insummary, adifferential trimethylation atregion 3continuedtobeseeninthecranial,but not region in9.5and10.5dpcembryosthan8.0dpc.H3-K27 eight-somite stage.Itwas presentathigherlevels inthecaudal transcription. Thishypothesisissupportedbyourprevious Differential associationofRnf2to genes repression ofHox Evidence fortherole ofRnf2intranscriptional spatially restrictedexpression ofthe and H3-K27trimethylationmaybeinvolved tomaintainthe observations indicatethattheRnf2association,H3-K9 acetylation established at9.5dpcandmaintainedto12.5dpc.These dpc. Likewise, differential trimethylationofH3-K27was already amounts ofacetylatedH3-K9inthisregion alsoincreasedupto12.5 10.5 dpc(Fig.3G)(DeschampsandWijgerde, 1993).Relative established from8.0dpconwards, andreachedcompletionaround tissues notexpressing Hoxb8 Hoxb8 suggests itsrepressive rolein Hoxb8 promoter region, region 3,was Hoxb8 genomic surroundingin . Development 133(12) represented bythe issues was hylated H3-K27 Hoxb8

DEVELOPMENT to thePCR. ‘Input’, genomicDNAextractedfrom theoriginalwholecelllysateequivalentto1/40volume ofthatusedfortheChIPanal were derivedfrom relative amountsof Rnf2 transfected(tr)EScellswascompared byRT-PCR. The quantityofsynthesizedcDNAfrom respective cells wasequalizedby derived from thecranialpartof transcriptional repressionof Rnf2 associationisessentialforthemaintenanceof infected MEFscomparedwiththeuninfectedcontrol.Therefore demonstrated morethana16-foldincreaseof days afterAd-Creinfection.Two independentexperiments for completion.Theanalysisof in westernanalysisofinfectedcellculture(Fig.4A)andtook2days as described(deNapolesetal.,2004).Rnf2depletionwas evident cultures withadenovirus expressing theCRErecombinase(Ad-Cre) Subsequent deletionof 9.5 dpcembryosinwhich Rnf2 Rnf2 of conditionally depletedfunctional establishment ofthisrepressedstatus.To addressthisquestion,we was functionallyinvolved inthemaintenanceratherthanearly al., 2002).However, itwas notdeterminedwhetherRnf2association de-repression ofseveral Hoxgeneswas seenat11.5dpc(Suzukiet experiments thatusedanhypomorphicalleleof Dual functionsofPcGcomplexesinHoxtranscription 2004). PrimaryMEFswerederived fromthecranialpartof flanked withloxPsites(Voncken etal., 2003;deNapolesetal., not altered. ( the wildtype(fl/fl),whereas theexpression ofRYBP (another Rnf2-bindingprotein thatisnotfoundinhPRC-Hcomplex orclass Fig. 4.De-repression ofHoxgenesin of Hoxb8wasinducedbyinfectionCre-expressing adenovirusvectorin (fl/fl) depletedtheRnf2geneproducts, whereas thewild-type (+/+)MEFswere unaffected. LaminBwasusedasacontrol. (Bottom (middle) antibodies.CBBstaining was usedasaloadingcontrol (right).( a constructexpressing Myc-tagged Rnf2(tr).Theexpression ofendogenousandtransfected Rnf2wasexaminedbyusinganti-Rnf2 Hoxb8 fl -null mice,wegeneratedaconditionalalleledesignatedas in whichexon 2,containingtheATG initiationcodon,was was established.Becauseofearlyembryoniclethality Hprt E ) Rnf2associationandH3-K27trimethylation atHoxpromoter regions were compared between was usedasanegative control. Rnf2 ␤ -actin transcripts.( fl/fl (fl/fl) EScellsbytransientoverexpression ofCre-recombinase. Rnf2wasre-expressed bytransfecting Rnf2 Hoxb8 Rnf2 was achieved byinfectionofMEF Hoxb8 Hoxb8 Rnf2 fl/fl . 9.5 dpcembryos.(Top) InfectionofCre-expressing adenovirusvector toMEFsderivedfrom D Rnf2 ) Theexpression ofPhc1and Cbx2geneproducts wasreduced in expression was repressed. after theexpression domain expression continuedfor4 –/– MEFs andEScells. Hoxb8 Rnf2 , inwhichthe expression in Rnf2 fl/fl ( A C ) ConditionaldepletionofRnf2leadtode-repression of ) Theexpression ofHoxcluster genesin Rnf2 transcriptional repressionofHoxgenesinEScells. results confirmthatRnf2associationisrequiredtomediatethe (Chambeyron andBickmore,2004;Chambeyron etal.,2005). geneexpression bothinEScellsandthedeveloping neuraltube Hox territory, whereasdecondensation andreorganization accompanied geneswithinthechromosome cluster andthelocationofHox was accompaniedbybothchromatincondensationoftheHox repression embryonic stem(ES)cellswas investigated becauseHox Rnf2 Hoxb3 Loss offunctionalRnf2resultedinde-repression 13-3, weregeneratedbyCRE-mediatedexcision ofexon 2(Fig.4B). of of endogenousRnf2.RT-PCR analysisrevealed that the expression expressed transgene-encodedRnf2atalevel equivalent tohalfthat Myc-tagged Rnf2.Asshown inFig.4B,transfected mutantEScells carried out,transfectingmutantEScellswithaconstructexpressing consequence ofmutatingRnf2,complementationexperiments were (Fig. 4C).To testifde-repressionofHoxgenesisadirect same degree asintheparental Seven After this,therepressive roleofRnf2inundifferentiated Hoxa1 fl/fl –/– MEFs (fl/fl),butnotinthewildtype(+/+).( , ES cellderivatives ofonetheselines,themaleEScellline, Hoxb8 Rnf2 , Hoxb3 fl/fl , Hoxb13 ES cellshave beenderived (deNapolesetal.,2004). , Hoxb8 , Hoxc9 and Rnf2 Hoxd8 , Hoxd8 Rnf2 Rnf2 –/– Rnf2 RESEARCH ARTICLE ES cells(–/–)incomparisonwith fl/fl was repressedalbeitnottothe fl/fl and fl/fl and ES cells(Fig.4C).These (fl/fl), Hoxd10 Rnf2 B ) Rnf2 Rnf2 ysis wassubjected Rnf2 –/– Hoxb8 1 PcGproteins) was comparing the ES cells.Forthe but notof Rnf2 ) Theexpression –/– –/– (left) and-Myc Hoxa1 –/– ES cellswith (–/–) and (–/–) EScells in MEFs fl/fl embryos , Hoxa7 Hoxc8 2377 ,

DEVELOPMENT Suz12 locus, intheabsenceoftrimethylatedH3-K27,was examined using Therefore, Hoxgeneexpression andRnf2association at the of Pc(Caoetal.,2002;CzerminMuller2002). interaction betweentrimethylatedH3-K27andthechromodomain facilitate therecruitmentofclass2PcGcomplexes viadirect In expression gene on Rnf2associationatHoxlociand Functional involvementoftrimethylatedH3-K27 K27 trimethylation. deficiency affects Hoxgeneexpressions withoutchanginglocalH3- promoter regions werealmostunchanged(Fig.4E).Therefore, Rnf2 analyses revealed thatlocallevel ofH3-K27trimethylationatHox in The expression ofSuz12andEzh2werenotsignificantlychanged change ofH3-K27trimethylationacrosstheHoxgenomicregions. Rnf2 expression levels ofothercomponentsclass2PcGcomplexes in of thePolycombcorecomplex inSf9cells(Francisetal.,2001),the suggested thatRing1isanimportantcomponentinthestabilization functions ofclass2PcGcomplexes was investigated. Asithasbeen 2378 whether de-repressionofHoxgenesin in transcriptionallyrepressedregion furtherpromptedustoask stability oftheclass2PcGcomplexes. thus likely thatRnf2impactstheHoxexpression by regulating the Rnf2 lossspecificallyaffects theexpression ofPhc1andCbx2.Itis transcription of H complex, was notaltered(Garciaetal.,1999)(Fig. 4D).Asthe RYBP, anotherRnf2-bindingprotein,whichisnotfoundinhPRC- 4D; M.E.andH.K.,unpublished).Bycontrast,theexpression of function alleleof investigated. Anti-Suz12,-Eedand-trimethylatedH3-K27 and H3-K27trimethylationattheHoxpromotorregions was methylation (Pasini etal.,2004).First,theassociationof class1PcG than 10%ofthewildtypebut didnotsignificantlychangeH3-K9 Suz12 reducedthelevels ofH3-K27tri-anddimethylationtoless derived fromcrossesofheterozygousmutants.Theabsence Cbx2 geneproductswas obviously reducedin al., 2004)(K.I.andH.K.,unpublished). phenotype almostidenticaltothatreportedbyPasini et al. (Pasini et Next, themethodbywhichRnf2deficiency impactsonthe The coincidenceofRnf2associationandH3-K27trimethylation Rnf2 Drosophila –/– –/– RESEARCH ARTICLE –/– ES cellswas examined. Theexpression ofbothPhc1and ES cells.We have independentlygeneratedaloss-of- ES cells(Fig.4D).Concordantwiththisresult,ChIP , trimethylationonH3-K27hasbeenshown to Phc1 Suz12 and and thehomozygousmutantsexhibited a Cbx2 was notalteredin Rnf2 Suz12 –/– Rnf2 ES cellsinvolves a –/– Rnf2 –/– ES cellswere ES cells(Fig. –/– ES cells, Hoxb Rnf2 associationin in ordertomediaterepressionofHoxgenes.We wentontoexamine complexes, orassociationofthecomplexes, arelikely toberequired Therefore, localH3-K27trimethylationmediatedbyclass1PcG express moreHoxgenetranscriptsthanthewildtype(Fig.5B). to Results fromthisstudyalsoshow thatPcGgeneproductsassociate acetylation transcription iscoupledtoincreased H3-K9 The positiverole ofPcGcomplexeson interactions isnotnecessarilyexcluded. the recruitmentofclass2complexes viaprotein-protein demonstrated in regulatory regions, asalready association ofRnf2withHox Therefore H3-K27trimethylationmaybeaprerequisiteforthe Suz12 Hox promoterfragmentsfromthewild-typeEScells(Fig.5A).In antibodies specificallyimmunoprecipitatedsignificantamountsof the expression of local H3-K27trimethylationinundifferentiated ES cells.Second, class 1PcGcomplexes associatelocallytoHoxgenesandmediate Eed associationandH3-K27trimethylationatHoxgenes.Therefore, Suz12 immunoprecipitation.Thelossofsignificantlyreduced promoters was significantlyreducedin the of PcGcomplexes explains thedropingeneexpression levels within knock-in loci(deGraaff etal.,2003).Inaddition,apositive action severely impairthetranscriptionfromHox/ decrease thetranscriptionlevel ofendogenous Null mutationsinthe positive functionofclass2PcGcomplexes (deGraaff etal.,2003). embryonic tissues.Thisisinlinewiththerecentlyuncovered expression level in colocalization, weinvestigated whether thedropin controls, andMllBmi1proteinsdisplaydiscretesubnuclear acetylation correlateswithHoxgeneexpression in de Graaff etal.,2003;Isono2005b).Asthelevel ofhistoneH3 Bmi1/Rnf110 in thecaudalregion of et al.,2002;deGraaff etal.,2003).Thedegree ofH3-K9acetylation involved achangeinH3-K9 acetylation(Hansonetal.,1999;Milne Hoxb8 Suz12 Hoxb8 Hoxb8 –/– –/– , Hoxb9 ES cells,noHoxDNA couldbedetectedatalluponanti- ES cellsbyRT-PCR. genomic regions inthetranscriptionallyactive caudal expression domainsknown tooccuraround9.5dpcin and and Drosophila Phc1 Hoxa1 Rnf110 Suz12 Hoxc6 in thewild-typeand K27 trimethylationattheHoxpromoter regions ( Rnf2 associationtoHoxgenomicregions. number ofwildtype(+/+)and lysates prepared from approximately the same (–/–) EScellswascompared byRT-PCR. Suz12 Fig. 5.De-repression ofHoxgenesin cluster genesinthewildtype(+/+) and was usedasacontrol. ( the ChIPanalysiswassubjectedto thePCR. equivalent tothe1/40volumeof that usedfor extracted from theoriginal wholecelllysate antibodies. Forthe‘Input’,genomicDNA Suz12, -Eed,-trimethylatedH3-K27and-Rnf2 cells were subjectedtoChIPanalysesusinganti- A TheassociationofSuz12,Eed,Rnf2andH3- ) Rnf110 / Rnf110 Phc2 , –/– Hoxa4 –/– / was comparedbetweenwild-typeand . Theroleofclass1PcGcomplexes in Bmi1 ES cells.Rnf2associationtotheHox double mutants(Akasakaetal.,2001; ES cellscorrelates withreduction of –/– and Suz12 Bmi1 , and Hoxb1 Phc1 –/– –/– Phc1 Suz12 suz12 embryos was muchlower , loci have beenshown to ES cellswereshown to B Hoxb3 Development 133(12) Theexpression ofHox ) / Phc2 –/– –/– lacZ ES cells.Whole-cell ES cells(Fig.5A). suz12 , Mll double mutants Hoxb4 Hoxb1 reporters and mutants and –/– (–/–) ES suz12 , , andto Hoxb6 Hoxb8 Hprt –/– ,

DEVELOPMENT of H3-K9acetylationintranscriptionallyactive regions. class 2PcGgeneproductsto GeneVI previous findingsinthe This isreminiscentof domains wherethegeneistranscriptionallyactive andrepressed. genomic region inembryonic PcG complexes bindtotheHoxb of transcriptionalstatus.Thisimpliesthatdifferent formsofclass2 members, Phc1andCbx2,isobserved atallAPlevels, irrespective embryonic tissues,whereasthebindingofotherPcGclass2 gene isseenpredominantlyintranscriptionallysilentanterior PcG developmental timewindow aswhenHoxgenesarederegulated in expressing andnon-expressing embryonictissuesoccurinthesame establishment ofdifferential PcGbindingandhistonemarksin requiredforPcGbinding.Inaddition,weshowed thatthe is genederepression,andthatH3-K27trimethylation leads toHox genetic impairmentofbothPcGbinding,andH3-K27trimethylation developing embryos.Moreover, theresultsdemonstratedthat with themaintenanceoftranscriptionalsilencingaHoxgenein as H3-K27trimethylationmarkinginactive chromatin,correlates specific, Rnf2-containingformoftheclass2PcGcomplex, aswell The mainoutcomeofthisstudywas toshow thatbindingofa DISCUSSION than in Dual functionsofPcGcomplexesinHoxtranscription seen in to thecranialpartwhereascraniocaudallygradedacetylationwas acetylation level inthecaudalpartof actin promoterwas almostequivalent (Fig.6A).Similarly, H3-K9 complexes atHoxgenesmayfacilitate subsequentbinding of Rnf2- trimethylation, H3-K27trimethylation mediatedbyclass1PcG alters Hoxexpression without changinglocallevels ofH3-K27 Hox genesisreducedin mediation ofthistranscriptional repression.AsRnf2associationto of bothRnf2associationand H3-K27trimethylationinthe of Hoxgenesin class 2PcGcomplexes andH3-K27trimethylation. De-repression has beenshown tocorrelatewiththeassociationofRnf2-containing 2004). directed tohistoneH2A(Wang etal.,2004a;deNapoles role ofRnf2ismediatedthroughitsE3ubiquitinligaseactivity regulate itsstability(Francisetal.,2001).Itisalsopossiblethatthe be alimitingprocesstomediatetranscriptionalrepression and case, incorporationoftheRnf2componentintocomplex might efficiently thanform(s)lackingtheRnf2component.Ifthisis 2 PcGcorecomplex maymediatetranscriptionalrepressionmore and Paro, 1997).Thereforethecomplete,‘perfect’formofclass and Pscarepresentirrespective ofthetranscriptionalstatus(Strutt exclusively associatedwithtranscriptionally silentgenes,whilePh Rnf2 associationtoknown regulatory elementsofthe rncitoa ersino o genesinthedeveloping embryo Transcriptional repressionofHox mutants. Phc1 Rnf110 complex in +/– +/+ Phc2 Rnf2 Bmi1 Drosophila +/– and +/– embryos (Fig.6B).Thereforeassociationof Suz12 , whereastheH3-K9acetylationat Suz12 Hox mutant EScellsand mutant cellsreveal therequirement SL-2 cells,wherethePcproteinis b8 isrequiredforthemaintenance Phc1 –/– Phc2 –/– Engrailed/Inv/ was equivalent Rnf2 mutation Hoxb8 ␤ - Interaction between class1and2PcGcomplexes mediated by transcription, byformingcomplexes ofdifferent composition. embryonic territories,which are silentoractive for action remainunaddressed. et al.,2003).Themolecularmechanisms underlyingthispositive compartment wheremostpre-mRNA synthesistakes place(Cmarco localization ofseveral PcGproteins in theperichromatin genes areexpressed. Thisisconsistentwithpredominantsubnuclear maintenance ofH3-K9acetylationinembryonictissueswhere Hox complexes, whichlacktheRnf2component,are also involved inthe 2001; Yu etal.,1998; Isonoetal.,2005b).Intriguingly, class2PcG of Drosophila genes seemstohave beenevolutionarily conserved between Therefore, molecularcircuitryunderlyingPcGsilencingofHox repressed chromatinstatus(Shaoetal.,1999;Levine etal.,2002). factors, suchasSWI/SNFcomplex, leadingtothe formationofa ofnucleosomeremodeling complexes mayinturnprevent theaccess RecruitmentofRnf2-containingPcG containing PcGcomplexes. and lossofitstranscriptionin that Hoxgenein during embryogenesistemporallycoincideswithde-repression of differential bindingoftheRnf2andH3-K9acetylationat mutations (Hansonetal.,1999).Moreover, theestablishmentof consistent withtheantagonisticpropertiesof Hoxb8 each otherandcooperatetomaintaintheanteriorboundariesof containing PcGcomplexes andacetylatedH3-K9maycounteract repression. Therefore,ourdatasuggestthattheassociationsofRnf2- containing class2PcGcomplexes tomediatethistranscriptional cranial partof9.5dpcembryossuggeststheinvolvement ofRnf2- Hoxb8 repressed atearlyandlaterdevelopmental stages.De-repressionof at the anterior, non-Hoxexpressing tissues, someacetylationofH3-K9 dpc andolderembryosaremoreheavily acetylatedatH3-K9than axis. Althoughthetranscriptionallyactive posteriortissuesof9.5 modifications ofthehistonetailincreasescraniocaudallyalong (Milne etal.,2002;Rastegar etal.,2004).We have shown thatthese is likely toinvolve H3-K9acetylationandH3-K4methylation al., 2002;CzerminetMuller2002). H3-K27, mediatestherepressionatHoxgenesinvivo (Caoet chromatin, andthattheassociation,whichinvolves trimethylated different meansbywhichclass2PcGcomplexes bindtothe 1999). Theimplicationofthesefindingsisthatthereareatleasttwo with thenucleosomaltemplateslackinghistonetails(Shaoetal., shown theassociationofpurifiedorreconstitutedPcGcomplexes K27. Thisisconsistentwithprevious biochemicaldatathathave embryonic tissues,despitethelackofhistoneH3trimethylatedat In conclusion,class2PcGgene productsplaydistinctrolesin h aneac frgoal etitdepeso fHxgenes The maintenanceofregionally restrictedexpression ofHox Hoxb8 Drosophila compared in acetylation intheanterior(A)andposterior(P)regions were embryos. The and region of Fig. 6.Decreased H3-K9acetylationatthefirstexonic compared in K9 acetylationintheanterior(A)andposterior(P)regions were ␤ with rabbitIgG. this study, thenegativecontrol ChIPs(A–andP–)were performed expression upondepletionofRnf2inMEFsderived fromthe -actin promoter wasusedasapositivecontrol. ( expression atmid-gestationalstagesandlater. Thisis is seeninanteriorregions where Phc1 and mammals.ItisalsonotablethatCbx2,ahomologue –/– Pc, bindsto Hoxb8 Bmi1 Phc2 Rnf110 Phc1 ␤ -actin promoter wasusedasapositivecontrol. In / –/– Rnf110 –/– in theposteriortissuesof Phc2 embryos at9.5dpc. +/– Bmi1 +/– and Hoxb8 , Mll +/– Phc1 Phc1 and RESEARCH ARTICLE homozygotes (Akasakaetal., +/– Rnf110 Phc2 in transcriptionallyactive / Phc2 –/– ( –/– Hoxb8 double homozygotes, A and Bmi1 ) Degree ofH3-K9 Rnf110 Phc1 B Mll –/– ) Degree ofH3- expression is –/– embryos. The Phc2 and –/– Bmi1 Hoxb8 Hoxb8 –/– Bmi1 2379 –/–

DEVELOPMENT Deschamps, J.andWijgerde, M. de Napoles,M.,Mermoud,J.E.,Wakao, R.,Tang, Y. A.,Endoh,M., de Graaff, W., Tomotsune, D.,Oosterveen, T., Takihara, Y., Koseki,H.and Garcia, E.,Marcos-Gutierrez, C.,delMarLorente, M.,Moreno, J.C.and Czermin, B.,Melfi,R.,McCabe,D.,Seitz,V., Imhof,A.andPirrotta, V. Cao, R.,Wang, L., Wang, H.,Xia,L.,Erdjument-Bromage, H.,Tempst, P., Brachvogel, B.,Reichenberg,D.,Beyer, S.,Jehn,B.,vonderMark,K.and Atsuta, T., M.,Akasaka,T. Fujimura,S.,Moriya,H.,Vidal, andKoseki,H. Akasaka, T., vanLohuizen,M.,derLugt,N.,Mizutani-Koseki,Y., Kanno, Akasaka, T., Kanno,M.,Mieza,M.A.,Balling,R.,Taniguchi, M.andKoseki, Hanson, R.D.,Hess,J.L.,Yu, P., B.D.,Ernst, A., vanLohuizen,M.,Berns, Hamer, K.M.,Sewalt,R. G.,denBlaauwen,J.L.,Hendrix,T., Satijn,D.P. and Francis, N.J.,Saurin,A.Shao,Z.andKingston,R.E. Cavalli, G.andParo, R. Cmarko, D.,Verschure, P. J.,Otte,A.P., vanDriel,R. andFakan,S. Agalioti, T., Chen,G.andThanos,D. References http://dev.biologists.org/cgi/content/full/133/12/2371/DC1 Supplementary materialforthisarticleisavailableat Supplementary material of H3-K27trimethylation. involved inmaintainingatranscriptionallyactive status,independent expression domain,class2PcGcomplexes are within theHox outside theirexpression domains,asseenin trimethylated H3-K27playdecisive rolesinHoxgenerepression 2380 Isono, K.,Mizutani-Koseki, Y., Komori, T., Schmidt-Zachmann, M.S.and Fischle, W., Wang, Y. andAllis, C. D. Charité, J.,deGraaff, W., Vogels, R.,Meijlink,F. andDeschamps,J. Chambeyron, S.andBickmore, W. A. Chambeyron, S.,DaSilva,N.R.,Lawson,K.A.andBickmore, W. A. expression domains. methyltransferase activitythatmarkschromosomal Polycombsites. Drosophila enhancerofZeste/ESCcomplexeshaveahistoneH3 heritable genesilencingandXinactivation. Polycomb group proteins Ring1A/BlinkubiquitylationofhistoneH2Ato Appanah, R.,Nesterova, T. B.,Silva,J.,Otte,A.P., M.etal. Vidal, 13362-13367. transcriptionally silencedinPolycombmutants. Deschamps, J. 185-196. EMBO J. of themammalianPolycombcomplex,andwithtranscriptionfactorYY1. M. Vidal, 545-556. Polycomb-group silencing. Jones, R.S.andZhang,Y. 210. of theDisintegrinandMetalloprotease-Domain (ADAM)family. Bielke, W. proteins. (2001). Production ofmonoclonalantibodiesagainstmammalianRing1B compartment ofthemammaliannucleus. Polycomb group genesilencingproteins are concentratedintheperichromatin expression. synergy andrequirement formaintenancebutnotinitiationofHoxgene Mice doublydeficientforthePolycombGroup genesMel18andBmi1reveal M., Taniguchi, A.andKoseki,H. M.,Alkema,Berns, M.,Vidal, 1522. the anteroposterior specificationoftheaxialskeleton. H. histone acetylationcodeforahumangene. regulators ofhomeoticdevelopment. Mammalian Trithorax andpolycomb-group homologues are antagonistic van derLugt,N.M.,Shashikant,C.S., Ruddle,F. H.,Seto,M.etal. group proteins. Otte, A.P. Reconstitution ofafunctionalcore polycombrepressive complex. cassettes inhistonebiologyandbeyond. 305. elements increases responsiveness topositionalinformation. Regulation oftheHoxb-8gene:synergismbetweenmultimerizedcis-acting conveys epigeneticinheritanceduringmitosisandmeiosis. 14377. development. Nuclear reorganization oftheHoxBcomplexduringmouseembryonic Dev. nuclear reorganization oftheHoxBlocusuponinductiontranscription. (1996). Arole formel-18,aPolycombgroup-related vertebrategene,during 18 RESEARCH ARTICLE , 1119-1130. 18 Hybridoma (1999). RYBP, anewrepressor protein thatinteractswithcomponents (2002). Molecularcloningandexpression analysisofanovelmember (2002). Apanelofmonoclonalantibodies againsthumanpolycomb , 3404-3418. Development Development (2003). RandomlyinsertedandtargetedHox/reporter fusions Hybrid. Hybridomics Dev. Biol. 20 (1998). TheDrosophila Fab-7chromosomal element , 43-46. 128 Science 132 (2002). RoleofhistoneH3lysine27methylationin , 1587-1597. 152 , 2215-2223. (1993). Two phasesintheestablishmentof , 473-480. 298 21 (2003). Binaryswitchesandmodification (2002). Decipheringthetranscriptional Proc. Natl.Acad.Sci.USA (2004). Chromatin decondensationand , 1039-1043. , 245-252. Nature J. CellSci Dev. Cell Cell Proc. Natl.Acad.Sci.USA 111 425 . Drosophila , 381-392. , 475-479. 116 7 , 663-676. Development , 335-343. (2001). Cell Dev. Biol. 93 Gene 96 . Inaddition, , 505-518. , 14372- Mol. Cell Cell 122 (2003). 288 171 (1995). (2004). (2005). 111 (2001). , 1513- (1999). (2002). , 203- 100 Genes , 294- , Hox 8 , , Shao, Z.,Raible,F., Mollaaghababa,R.,Guyon,J.Wu, C.T., Bender, W. Schoorlemmer, J.,Marcos-Gutierrez, C.,Were, F., Martinez,R.,Garcia, E., sn,K.,Fujimura,Y., Shinga,J.,Yamaki,Isono, M.,O-Wang, J.,Takihara, Y., Suzuki, M.,Mizutani-Koseki,Y., Fujimura,Y., Miyagishima,H.,Kaneko,T., Strutt, H.andParo, R. Shumacher, A.,Faust,C.andMagnuson,T. Rastegar, M.,Kobrossy, L.,Kovacs,E.N.,Rambaldi,I.andFeatherstone,M. Levine, S.S.,Weiss, A.,Erdjument-Bromage, H.,Shao,Z.,Tempst, P. and Peters, A.H.,Kubicek,S.,Mechtler, K.,O’Sullivan,R.J.,Derijck,A.A.,Perez- Kanegae, Y., Lee,G.,Sato,Y., Tanaka, M.,Nakai,Sakaki,T., Sugano,S. Takihara, Y., Tomotsune, D.,Shirai,M.,Katoh-Fukui, Y., Nishii,K.,Motaleb, Tamkun, J.W., Deuring,R.,Scott, M.P., Kissinger, M.,Pattatucci,A. Milne, T. A.,Briggs,S.D.,Brock, H.W., Martin,M.E.,Gibbs,D.,Allis,C.D. Lachner, M.,O’Sullivan,R.J.andJenuwein,T. King, I.F., Francis,N.J.andKingston,R.E. Kennison, J.A.andTamkun, J.W. Pirrotta, V. Muller, J.,Hart,C.M.,Francis,N.Vargas, M.L.,Sengupta,A.,Wild,B., Miyagishima, H.,Isono,K.,Fujimura,Y., Iyo,M.,Takihara, Y., Masumoto,H., Oosterveen, T., Niederreither, K.,Dolle,P., Chambon,P., Meijlink,F. and Nakamura, T., Mori,T., Tada, S.,Krajewski,W., Rozovskaia,T., Wassell, R., Pasini, D.,Bracken,A.P., Jensen,M.R.,Denchi,E.L.andHelin, K. Paro, R. Orlando, V., Strutt,H.andParo, R. requires theessentialspliceosomalprotein Sf3b1. Mol. Cell.Biol. Drosophila melanogasterhas different compositionsatdifferent targetgenes. Polycomb complex. and Kingston,R.E. rhombomere boundariesin the mousehindbrain. repressor thatinteractswiththePolycomb-M33protein andisexpressed at Satijn, D.P., Otte,A.P. M. andVidal, 8103. Koseki, H. methyltransferase activity. Polycomb-repression of Mammalian polyhomeotichomologuesPhc2andPhc1actinsynergytomediate Murahashi, Y., Takada, Y., Mizutani-Koseki,Y. andKoseki,H. the anterior-posterior axisinmice. (2002). InvolvementofthePolycomb-group geneRing1Binthespecification of Takada, Y., Akasaka,T., Tanzawa, H.,Takihara, Y., Nakano,M.etal. global regulator ofanterior-posterior inmice. patterning anterior from posteriorembryonic compartments. (2004). SequentialhistonemodificationsatHoxd4regulatory regions distinguish Dev. 22 compositionally andfunctionallyconservedinflieshumans. Kingston, R.E. for histonelysinemethylation. repress transcriptioninvitro. Polycomb group complexesestablishaselectiveblocktotemplateaccessibility 85 Polycomb andantennapediamutationsinDrosophila. mammalian chromatin. (2003). DecPartitioningandplasticityofrepressive histone methylationstatesin Burgos, L.,Kohlmaier, A.,Opravil,S.,Tachibana, M.,Shinkai,Y. etal. Res. recombinant adenovirusexpressing site-specificCre recombinase. and Saito,I. Development gene leadstoaltered anteroposteriorandneuralcrest patterning defects. Targeted disruption ofthemousehomologueDrosophila polyhomeotic M. A.,Nomura,M.,Tsuchiya, R.,Fujita,Y., Shibata,Y. etal. Hox genepromoters. and Hess,J.L. complex. Histone methyltransferaseactivityofaDrosophila Polycombgroup repressor Miller, E.L.,O’Connor, M.B.,Kingston,R.E.andSimon,J.A. Histochem. CellBiol. configuration changesofthechromatin inmitoticandmeioticprophase. proteins, Ring1BandRae28/Ph1,from thechromatin correlates with M.andKoseki,H. Vidal, Suz12 isessentialformousedevelopmentandEZH2histone vivo formaldehydecross-linking. 5 Deschamps, J. transcriptional regulation. methyltransferase thatassemblesasupercomplex ofproteins involvedin Dubois, G.,Mazo,A.,Croce, C.M.andCanaani,E. 295-297. Ј , 6070-6078. , 8136-8140. Hoxb genesinposteriorhindbrain. 23 7 , 249-258. (1995). Propagating memoryoftranscriptionalstates. , 3816-3821. Cell (1997). PcGcomplexesandchromatin silencing. (2005a). MammalianPolycomb-mediatedrepression of (1995). Efficient geneactivationinmammaliancellsbyusing 111 124 17 (2002). MLLtargetsSETdomainmethyltransferaseactivityto (2003). Retinoidsregulate theanteriorexpression boundaries of (2002). Thecore ofthePolycombrepressive complexis , 6773-6783. , 197-208. , 3673-3682. Cell 120 (1999). Stabilizationofchromatin structure byPRC1,a Mol. Cell (1997). Thepolycombgroup protein complexof Mol. Cell Hox 98 , 111-119. Mol. Cell EMBO J. , 37-46. (2003). DissociationofmammalianPolycomb-group genes. Mol. Cell.Biol. J. CellSci. 10 Methods 12 , 1107-1117. Development (1988). Dosage-dependentmodifiersof 23 (1997). Analysisofchromatin structure byin Mol. Cell.Biol. 10 , 1577-1589. EMBO J. , 4061-4071. , 1119-1128. (1997). Ring1Aisatranscriptional 116 11 22 (2002). Nativeandrecombinant (1996). Positionalcloningofa , 205-214. , 2117-2124. , 7919-7928. 22 (2003). Anepigeneticroad map 129 , 262-269. Genes Dev. EMBO J. Mol. Cell.Biol 25 Development 133(12) , 4171-4183. , 6694-6706. Proc. Natl.Acad.Sci.USA (2002). ALL-1isahistone Nature Curr. Opin.Genet. 16 rnsGenet. Trends , 5930-5942. 19 . , 536-541. 383 Mol. Cell.Biol. (1997). 24 Nucleic Acids (2002). (2005b). Hox , 8090- , 250-253. (2004). genes 11 ,

DEVELOPMENT Dual functionsofPcGcomplexesinHoxtranscription Voncken, J.W., Roelen,B.A.,Roefs,M.,deVries, S.,Verhoeven, E.,Marino, van der Lugt, N. M., Alkema, M., Berns, A.andDeschamps,J. van derLugt,N.M.,Alkema,Berns, Vogels, R.,Charite,J.,deGraaff, W. andDeschamps,J. van derVlag,J.andOtte,A.P. SNF2/SWI2. homeotic genesstructurallyrelated totheyeasttranscriptionalactivator Kaufman, T. C.andKennison,J.A. transgenic mice. acting elementscooperatetosetHoxb-7(Hox-2.3)expression boundariesin Genet. the humanpolycomb-group protein EEDinvolveshistonedeacetylation. Mech. Dev. Polycomb-group homologBmi-1isaregulator ofmurineHoxgeneexpression. 23 , 474-478. 58 Cell , 153-164. 68 Development , 561-572. 118 (1999). Transcriptional repression mediatedby , 71-82. (1992). brahma:aregulator ofDrosophila (1993). Proximal cis- (1996). The Nat. Wang, H.,Wang, L.,Erdjument-Bromage, M.,Tempst, H.,Vidal, P., Jones,R. Wang, L.,Brown, J.L.,Cao,R.,Zhang,Y., Kassis,J.A.andJones,R.S. Yu, S.E.andKorsmeyer, B.D.,Hanson,R.Hess,J.L., Horning, S.J. silencing. S. andZhang,Y. 100 causes gastrulationarrest andcellcycleinhibition. S., Deschamps,J.andvanLohuizen,M. Cell (2004b). Hierarchical recruitment ofpolycombgroup silencingcomplexes. 10636. maintenance factorinmorphogenesis. (1998). MLL,amammaliantrithorax-group gene,functionsasatranscriptional , 2468-2473. 14 , 637-646. Nature 431 (2004a). RoleofhistoneH2AubiquitinationinPolycomb , 873-878. Proc. Natl.Acad.Sci.USA RESEARCH ARTICLE (2003). Rnf2(Ring1b)deﬁciency Proc. Natl.Acad.Sci.USA 95 , 10632- 2381 Mol.

DEVELOPMENT