聪明文档网

    聪明文档网

    最新最全的文档下载
    当前位置： 首页> lincRNAs act in the circuitry controlling stemness and differentiation

    lincRNAs act in the circuitry controlling stemness and differentiation

    时间：    下载该word文档
    ARTICLE
    doi:10.1038/nature10398
    lincRNAsactinthecircuitrycontrollingpluripotencyanddifferentiation
    MitchellGuttman1,2,JulieDonaghey1,BryceW.Carey2,3,ManuelGarber1,JenniferK.Grenier1,GlenMunson1,GenevaYoung1,AnneBergstromLucas4,RobertAch4,LaurakayBruhn4,XiaopingYang1,IdoAmit1,AlexanderMeissner1,5*,AvivRegev1,2*,JohnL.Rinn1,5*,DavidE.Root1*&EricS.Lander1,2,6
    Althoughthousandsoflargeintergenicnon-codingRNAs(lincRNAshavebeenidentifiedinmammals,fewhavebeenfunctionallycharacterized,leadingtodebateabouttheirbiologicalrole.Toaddressthis,weperformedloss-of-functionstudiesonmostlincRNAsexpressedinmouseembryonicstem(EScellsandcharacterizedtheeffectsongeneexpression.HereweshowthatknockdownoflincRNAshasmajorconsequencesongeneexpressionpatterns,comparabletoknockdownofwell-knownEScellregulators.Notably,lincRNAsprimarilyaffectgeneexpressionintrans.KnockdownofdozensoflincRNAscauseseitherexitfromthepluripotentstateorupregulationoflineagecommitmentprograms.WeintegratelincRNAsintothemolecularcircuitryofEScellsandshowthatlincRNAgenesareregulatedbykeytranscriptionfactorsandthatlincRNAtranscriptsbindtomultiplechromatinregulatoryproteinstoaffectsharedgeneexpressionprograms.Together,theresultsdemonstratethatlincRNAshavekeyrolesinthecircuitrycontrollingEScellstate.
    Themammaliangenomeencodesmanythousandsoflargenon-codingtranscripts1includingaclassof,3,500lincRNAsidentifiedusingachromatinsignatureofactivelytranscribedgenes2–4.TheselincRNAgeneshavebeenshowntohaveinterestingproperties,includingclearevolutionaryconservation2–5,expressionpatternscor-relatedwithvariouscellularprocesses2,6andbindingofkeytranscrip-tionfactorstotheirpromoters2,6,andthelincRNAsthemselvesphysicallyassociatewithchromatinregulatoryproteins4,7.Yet,itremainsunclearwhethertheRNAtranscriptsthemselveshavebio-logicalfunctions8–10.Fewhavebeendemonstratedtohavephenotypicconsequencesbyloss-of-functionexperiments6.Asaresult,thefunc-tionalroleoflincRNAgeneshasbeenwidelydebated.Variouspro-posalsincludethatlincRNAgenesactasenhancerregions,withtheRNAtranscriptsimplybeinganincidentalby-product8,9,thatlincRNAtranscriptsactincistoactivatetranscription11,andthatlincRNAtranscriptscanactintranstorepresstranscription12,13.Wethereforesoughttoundertakesystematicloss-of-functionexperimentsonalllincRNAsknowntobeexpressedinmouseembry-onicstem(EScells2,3.EScellsarepluripotentcellsthatcanself-renewincultureandcangiverisetocellsofanyofthethreeprimarygermlayersincludingthegermline14.Thesignalling14,transcriptional15–17andchromatin15,18–21regulatorynetworkscontrollingpluripotencyhavebeenwellcharacterized,providinganidealsystemtodeterminehowlincRNAsmayintegrateintotheseprocesses.
    HereweshowthatknockdownofthevastmajorityofES-cell-expressedlincRNAshasastrongeffectongeneexpressionpatternsinEScells,ofcomparablemagnitudetothatseenforthewell-knownEScellregulatoryproteins.WeidentifydozensoflincRNAsthatuponloss-of-functioncauseanexitfromthepluripotentstateanddozensofadditionallincRNAsthat,althoughnotessentialforthemaintenanceofpluripotency,acttorepresslineage-specificgeneexpressionpro-gramsinEScells.WeintegratethelincRNAsintothemolecularcircuitryofEScellsbydemonstratingthatmostlincRNAsaredirectly
    13
    regulatedbycriticalpluripotency-associatedtranscriptionfactorsand,30%oflincRNAsphysicallyinteractwithspecificchromatinregu-latoryproteinstoaffectgeneexpression.Together,theseresultsdemonstratearegulatorynetworkinEScellswherebytranscriptionfactorsdirectlyregulatetheexpressionoflincRNAgenes,manyofwhichcanphysicallyinteractwithchromatinproteins,affectgeneexpressionprogramsandmaintaintheEScellstate.
    lincRNAsaffectglobalgeneexpression
    Toperformloss-of-functionexperiments,wegeneratedfivelentiviral-basedshorthairpinRNAs(shRNAs22targetingeachofthe226lincRNAspreviouslyidentifiedinEScells2,3(seeMethodsandSup-plementaryTable1.TheseshRNAssuccessfullytargeted147lincRNAsandreducedtheirexpressionbyanaverageof,75%comparedtoendogenouslevelsinEScells(seeMethods,Fig.1a,SupplementaryFig.1andSupplementaryTable2.Aspositivecontrols,wegeneratedshRNAstargeting,50genesencodingregulatoryproteins,includingbothtranscriptionandchromatinfactorsthathavebeenshowntoplaycriticalrolesinEScellregulation17,20,23;validatedhairpinswereobtainedagainst40ofthesegenes(SupplementaryTable2.Asnega-tivecontrols,weperformedindependentinfectionswithlentivirusescontaining27differentshRNAswithnoknowncellulartargetRNA.WeinfectedeachshRNAintoEScells,isolatedRNAafter4days,andprofiledtheireffectsonglobaltranscriptionbyhybridizationtogenome-widemicroarrays(Fig.1a,seeMethods.Weusedastringentproceduretocontrolfornonspecificeffectsduetoviralinfection,genericRNAinterference(RNAiresponses,or‘off-target’effects.Expressionchangesweredeemedsignificantonlyiftheyexceededthemaximumlevelsobservedinanyofthenegativecontrols,showedatwofoldchangeinexpressioncomparedtothenegativecontrols,andhadalowfalsediscoveryrate(FDRassessedacrossallgenesbasedonpermutationtests(Fig.1b,seeMethods.Thisapproachcontrolsfortheoverallrateofnonspecificeffectsbyestimatingthenumberand
    BroadInstituteofMITandHarvard,7CambridgeCenter,Cambridge,Massachusetts02142,USA.2DepartmentofBiology,MassachusettsInstituteofTechnology,Cambridge,Massachusetts02139,USA.WhiteheadInstituteforBiomedicalResearch,9CambridgeCenter,Cambridge,Massachusetts02142,USA.4GenomicsResearchandDevelopment,AgilentTechnologies,SantaClara,California95051,USA.5StemCellandRegenerativeBiology,HarvardUniversity,Cambridge,Massachusetts02138,USA.6DepartmentofSystemsBiology,HarvardMedicalSchool,Boston,Massachusetts02114,USA.*Theseauthorscontributedequallytothiswork.
    15SEPTEMBER2011|VOL477|NATURE|295
    ©2011MacmillanPublishersLimited.Allrightsreserved

    RESEARCHARTICLE
    a
    InfectwithshRNA
    40%
    Embryonicstemcells
    ControllincRNAhairpinshairpinslinc1244
    Moxd1
    27negativecontrols
    Differentiallyexpressedgenes
    CtgfKD
    Density
    Selectbesthairpin
    Knock-Controldown
    Identifyaffectedgenes
    b
    c0.004
    Nanog0.0030.0020.0010.000
    Sox2Klf4
    ProteincodinglincRNA
    Stat3
    Oct4
    95genes
    outof12lincRNAsaffectedexpressionofagenewithin300kb11.Theexplanationseemstobethatthethresholdinthepreviousstudyfailedtoaccountformultiplehypothesistestingwithinthelocalregion.Accountingforthis,theeffectsonneighbouringgenesarenogreaterthanexpectedbychanceandareconsistentwithourobservationshere(seeMethods.
    AlthoughsomelincRNAscanregulategeneexpressionincis11,24,25,determiningthepreciseproportionofcisregulatorsrequiresmoredirectexperimentalapproaches.WenotethatourresultsareconsistentwithobservedcorrelationsbetweenlincRNAsandneighbouringgenes2,26,whichmayrepresentsharedupstreamregulation2,12orlocaltranscriptionaleffects10,27.Inaddition,thelincRNAsstudiedhereshouldbedistinguishedfromtranscriptsthatareproducedatenhancersites8,9,thefunctionofwhichhasyettobedetermined.
    lincRNAsmaintainthepluripotentstate
    Differentiallyexpressedgenes
    Figure1|FunctionalaffectsoflincRNAs.a,AschematicoflincRNA
    perturbationexperiments.EScellsareinfectedwithshRNAs,knockdownleveliscomputed,thebesthairpinisselectedandprofiledonexpressionarrays,anddifferentialgeneexpressioniscomputedrelativetonegativecontrolhairpins.b,ExampleofalincRNAknockdown.Top:genomiclocuscontainingthelincRNA.Bottom:heat-mapofthe95genesaffectedbyknockdownofthelincRNA,expressionforcontrolhairpins(redlineandexpressionforlincRNAhairpins(bluelineareshown.c,Distributionofnumberofaffectedgenesuponknockdownof147lincRNAs(blueand40well-knownEScellregulatoryproteins(red.PointscorrespondingtofivespecificEScellregulatoryproteinsaremarked.
    magnitudeofobservedeffectsinthenegativecontrolhairpins,wherealleffectsarenonspecific.
    For137ofthe147lincRNAs(93%,knockdowncausedasignifi-cantimpactongeneexpression(SupplementaryTable3,withanaverageof175protein-codingtranscriptsaffected(range:20–936(Fig.1c,SupplementaryFig.2andSupplementaryTable4.Theseresultsweresimilartothoseobtaineduponknockdownofthe40well-studiedEScellregulatoryproteins:38(95%showedsignificanteffectsongeneexpression,withanaverageof207genesaffected(range:28(forDnmt3lto1,187(forOct4(Fig.1c,Supplemen-taryFig.2andSupplementaryTable4.AlthoughsomeindividuallincRNAshavebeenfoundtoleadprimarilytogenerepression12,13,wefindthatknockdownofthelincRNAsstudiedherelargelyledtocomparablenumbersofactivatedandrepressedgenes(Supplemen-taryFig.2andSupplementaryTable4.Toassessoff-targeteffectsfurther,wealsoprofiledtheeffectsofthesecond-bestvalidatedshRNAtargeting10randomlyselectedlincRNAgenes.Inallcases,secondshRNAsagainstthesametargetproducedsignificantlysimilarexpressionchanges(seeMethodsandSupplementaryTable5.TheseresultsindicatethatthevastmajorityoflincRNAshavefunctionalconsequencesonoverallgeneexpressionofcomparablemagnitude(intermsofnumberofaffectedgenesandimpactonlevelstotheknowntranscriptionalregulatorsinEScells.
    WenextsoughttoinvestigatewhetherlincRNAshavearoleinregu-latingtheEScellstate.RegulationoftheEScellstateinvolvestwocomponents:maintainingthepluripotencyprogramandrepressingdifferentiationprograms15.TodeterminewhetherlincRNAshavearoleinthemaintenanceofthepluripotencyprogram,westudiedtheireffectsontheexpressionofNanog,akeytranscriptionfactorthatisrequiredtoestablish28anduniquelymarksthepluripotentstate29,30.WeinfectedEScellscarryingaluciferasereportergeneexpressedfromtheendogenousNanogpromoter31withshRNAstargetinglincRNAsorprotein-codinggenes.Wemonitoredlossofreporteractivityafter8daysrelativeto25negativecontrolhairpinsacrossbiologicalreplicates(seeMethods.Toensurethattheobservedeffectswerenotsimplyduetoareductionincellviability,weexcludedshRNAsthatcausedareductionincellnumbers(seeMethods,SupplementaryFig.4andSupplementaryTable7.Altogether,weidentified26lincRNAsthathadmajoreffectsonendogenousNanoglevelswithmanyatcomparablelevelstotheknockdownoftheknownprotein-codingregulatorsofpluripotencysuchasOct4andNanog(Fig.2aandSupplementaryTable7.ThisestablishesthattheselincRNAshavearoleinmaintainingthepluripotentstate.Tovalidatefurthertheroleofthese26lincRNAsinregulatingthepluripotentstate,weknockeddowntheselincRNAsinwild-typeEScellsandmeasuredmRNAlevelsofpluripotencymarkergenesOct4(alsocalledPou5f1,Sox2,Nanog,Klf4andZfp42after8days.Inallcasesweobservedasignificantreductionintheexpressionofmultiplepluripotencymarkerswith.90%showingasignificantdecreaseinbothOct4andNanoglevels(SupplementaryFig.5andSupplementaryTables8and9.Tocontrolforoff-targeteffects,westudiedadditionalhairpinstargetingtheselincRNAs.For15lincRNAswehadaneffec-tivesecondhairpin.Inall15cases,thesecondhairpinproducedcomparablereductionsinOct4expressionlevels,showingthattheobservationswerenotduetooff-targeteffects(Fig.2bandSup-plementaryTable10.Notably,.90%oflincRNAknockdownsaffectingNanogreporterlevelsledtolossofEScellmorphology(Fig.2candSupplementaryFigs6and7.Thus,inhibitionofthese26lincRNAsleadtoanincreasedexitfromthepluripotentstate.
    0401,0201,0001,080060040020
    24,25
    lincRNAsaffectgeneexpressionintrans
    FollowingtheobservationthatafewlincRNAsactincis,somerecentpapershaveclaimedthatmostlincRNAsactprimarilyincis8,11,26.Wefoundnoevidencetosupportthislatternotion:knockdownofonly2lincRNAsshowedeffectsonaneighbouringgene,only13showedeffectswithinawindowof10genesoneitherside,andonly8showedeffectsongeneswithin300kb;theseproportionsarenogreaterthanobservedforprotein-codinggenes(SupplementaryFig.3andSupplementaryTable6.Inshort,lincRNAsseemtoaffectexpressionlargelyintrans.
    OurresultscontrastwitharecentstudythatconcludedthatlincRNAsactincis,basedontheobservationthatknockdownof7
    296|NATURE|VOL477|15SEPTEMBER2011
    0
    lincRNAsrepresslineageprograms
    TodetermineiflincRNAsactinrepressingdifferentiationprogramswecomparedtheoverallgeneexpressionpatternsresultingfromknockdownofthelincRNAstopublishedgeneexpressionpatternsresultingfrominduceddifferentiationofEScells32,33andassessedsignificanceusingapermutation-derivedFDR34(seeMethods.Thesestatesincludedifferentiationintoendoderm,ectoderm,neuroectoderm,mesodermandtrophectodermlineages.Asapositivecontrolforouranalyticalmethod,weconfirmedtheexpectedresultsthattheexpressionpatterncausedbyOct4knockdownwasstronglyassociatedwiththetrophoectodermlineage35andthepatterncausedbyNanogknockdownwasstronglyassociatedwithendodermdiffer-entiation30(Fig.3a.
    ©2011MacmillanPublishersLimited.Allrightsreserved

    ARTICLERESEARCH
    Oct4Nanoglinc1253linc1331linc1557linc1562linc1602linc1230linc1335linc1463linc1582linc1612linc1242linc1304linc1356linc1400linc1456linc1490linc1517linc1588linc1617linc1623linc1627linc1633linc1388linc1390linc1307linc1434linc1526linc1552linc1470linc1604
    543210–12
    shOct4shNanogshlinc1230shlinc1253shlinc1562shlinc1602
    shOct4shNanogshlinc1230shlinc1335shlinc1463
    shOct4shNanog
    shlinc1388shlinc1390shlinc1456shlinc1588shlinc1623shlinc1627
    –2.5
    shOct4shNanogshlinc1388shlinc1390shlinc1434shlinc1526shlinc1552
    shOct4shNanog
    shlinc1470shlinc160411
    Fgf5(ectoderm
    Pax6
    (neuroectoderm
    Sox17(endoderm
    BrachyuryCdx2
    (mesoderm(trophectoderm
    a160,000
    Nanog-luciferaseintensity140,000120,000100,00080,00060,00040,00020,000
    LacZRFPGFPLucNanogOct4EsrrbStat3linc1454linc1242linc1434linc1349linc1487linc1411linc1473linc1582linc1283linc1552linc1547linc1368linc1577linc1634linc1609linc1572linc1343linc1448linc1543linc1399linc1385linc1379linc1435linc1405linc1610linc1490
    a
    EctodermNeuroectoderm
    EndodermMesodermTrophectodermLog-expressioncomparedwithcontrols
    b
    0
    NegativesProteinslincRNAs
    b
    Oct4relativeexpression
    1.210.80.60.40.2
    ControlsOct4Sox2Stat3Chd1linc1487linc1434linc1634linc1368linc1349linc1582linc1490linc1547linc1572linc1411linc1399linc1435linc1283
    ControlsChd1Sox2Stat3linc1490linc1349linc1368linc1582linc1411linc1572linc1547linc1435linc1487linc1399linc1283linc1634linc1434
    0
    Figure3|lincRNAsrepressspecificdifferentiationlineages.a,ExpressionchangesforeachlincRNAcomparedtogeneexpressionoffivedifferentiationpatterns.Eachboxshowssignificantpositiveassociation(red,FDR,0.01forOct4andNanog(leftandforlincRNAs(right.b,ExpressionchangesuponknockdownofOct4andNanog(blackbarsandrepresentativelincRNAs(greybarsforfivelineagemarkergenes.Theexpressionchanges(FDR,0.05aredisplayedonalogscaleasthet-statisticcomparedtoapanelofnegativecontrolhairpins.
    c
    Bright
    Hairpin1
    shRFP
    shOct4
    shlinc1368
    Hairpin2shlinc1577
    NegativePositivelincRNAs
    Figure2|lincRNAsarecriticalforthemaintenanceofpluripotency.
    a,ActivityfromaNanogpromoterdrivingluciferase,followingtreatmentwithcontrolhairpins(blackorhairpinstargetingluciferase(green,selectedprotein-codingregulators(red,andlincRNAs(blue.b,RelativemRNAexpressionlevelsofOct4afterknockdownofselectedprotein-coding(redandlincRNA(bluegenesaffectingNanog-luciferaselevels.Thebesthairpin
    (Hairpin1andsecondbesthairpin(Hairpin2areshown.AllknockdownsaresignificantwithaP-value,0.01.Errorbarsrepresentstandarderror(n54.c,MorphologyofEScellsandimmunofluorescencestainingofOct4foranegativecontrolhairpin(blacklineandhairpinstargetingOct4(redline,andtwolincRNAs(blueline.Thefirstrowshowsbright-fieldimages,thesecondrowshowsimmunofluorescencestainingoftheOct4protein,andthethirdrowshowsDAPIstainingofthenuclei.
    Thefactthatknockdownofthese30lincRNAsinducesgeneexpressionprogramsassociatedwithspecificearlydifferentiationlineagesindicatesthattheselincRNAsnormallyareabarriertosuchdifferentiation.Interestingly,mostofthelincRNAknockdowns(,85%thatinducegeneexpressionpatternsassociatedwiththeselineagesdidnotcausethecellstodifferentiateasdeterminedbyNanogreporterlevels(SupplementaryTable7andOct4expression(SupplementaryFig.10.ThisisconsistentwithobservationsforseveralcriticalEScellchromatinregulators,suchasthepolycombcomplex;loss-of-functionoftheseregulatorssimilarlyinduceslineage-specificmarkerswithoutcausingdifferentiation18,36,37.
    Together,thesedataindicatethatmanylincRNAshaveimportantrolesinregulatingtheEScellstate,includingmaintainingthepluripo-tentstateandrepressingspecificdifferentiationlineages.
    DAPIOct4
    lincRNAsaretargetsofEScelltranscriptionfactors
    HavingdemonstratedafunctionalroleforlincRNAsinEScells,wesoughttointegratethelincRNAsintothemolecularcircuitrycontrol-lingthepluripotentstate.First,weexploredhowlincRNAexpressionisregulatedinEScells.Towardsthisend,weusedpublishedgenome-widemapsof9pluripotency-associatedtranscriptionfactors16,38anddeterminedwhethertheybindtothepromotersoflincRNAgenes.Ofthe226lincRNApromoters,75%areboundbyatleast1of9pluripotency-associatedtranscriptionfactors(includingOct4,Sox2,Nanog,c-Myc,n-Myc,Klf4,Zfx,SmadandTcf3withamedianof3factorsboundtoeachpromoter(Fig.4a,SupplementaryFig.11andSupplementaryTable14,comparabletotheproportionreportedforprotein-codinggenes16.Interestingly,thethreecorefactors(Oct4,Sox2andNanogbindtothepromotersof,12%ofallEScelllincRNAsand,50%oflincRNAsinvolvedintheregulationofthepluripotentstate.
    TodetermineiflincRNAexpressionisfunctionallyregulatedbythepluripotency-associatedtranscriptionfactors,weusedshRNAstoknockdowntheexpressionof5ofthe9pluripotency-associatedtran-scriptionfactorgenesforwhichwecouldobtainvalidatedhairpinsandprofiledtheresultingchangesinlincRNAexpressionafter4days.Uponknockdownofatranscriptionfactor,,50%oflincRNAgeneswhosepromotersareboundbythetranscriptionfactorexhibitexpressionchanges(Fig.4a;thisproportioniscomparabletothat
    15SEPTEMBER2011|VOL477|NATURE|297
    Usingthisapproach,weidentified30lincRNAsforwhichknock-downproducedexpressionpatternssimilartodifferentiationintospe-cificlineages(SupplementaryTable11.AmongtheselincRNAs,13areassociatedwithendodermdifferentiation,7withectodermdiffer-entiation,5withneuroectodermdifferentiation,7withmesodermdifferentiationand2withthetrophectodermlineage(Fig.3a.Consistentwiththesefunctionalassignments,weobservedthatmost(.85%ofthe30lincRNAsassociatedwithspecificdifferentiationlineagesshowedupregulationofthewell-knownmarkergenesfortheidentifiedstates17,32uponknockdown(suchasSox17(endoderm,Fgf5(ectoderm,Pax6(neuroectoderm,brachyury(mesodermandCdx2(trophectoderm(Fig.3b,SupplementaryFigs8and9andSupplementaryTables12and13.
    ©2011MacmillanPublishersLimited.Allrightsreserved

    RESEARCHARTICLE
    a
    PromoterbindingOct4Sox2NanogKlf4Smad1Tcf3ZfxcMycnMyc
    Cmc1
    SOX2Oct4
    linc1405
    Chr9(427kbEomes
    177lincRNApromoters
    Core
    regulatedMyc
    regulated
    linc1405
    NenfChr1
    (155kb
    linc1428
    Klf4c-Mycn-Myc
    Atf3
    61372849NANANANA45
    %Downregulated
    linc1428
    Oct4Sox2NanogKlf4Klf5Klf2Zfp42EsrrbThap11Stat3n-Myc
    b
    Transcriptionfactor
    knockdown
    Transcriptionfactor
    knockdown
    Nanog
    Sox2
    Oct4
    Transcriptionfactor
    knockdown
    Klf4Klf2n-MycEsrrb
    0–1–2–3–4–5–6–7–8–9
    Oct4linc1405
    0–1–2–3–4–5–6–7–8–9
    linc1428
    Figure4|lincRNAsaredirectregulatorytargetsoftheEScell
    transcriptionalcircuitry.a,Aheat-maprepresentingChIP-Seqenrichmentsforninetranscriptionfactors(columnsatlincRNApromoters(rows.ThepercentageofboundlincRNAsdownregulateduponknockdownofthe
    transcriptionfactorisindicatedinboxes.NA,notmeasured.Right:examplesoflincRNAsfromtwoclusters(‘coreregulated’and‘Mycregulated’showingtheirgenomicneighbourhoodandtranscriptionfactorbinding.b,Left:aheat-maprepresentingchangesinlincRNAexpression(rowsafterknockdownof11transcriptionfactors(columns.Middle:effectofknockdownofSox2,Oct4andNanogonexpressionlevelsoflinc1405(greyandOct4(black.Right:effectofknockdownofKlf2,Klf4,n-MycandEsrrbonexpressionlevelsoflinc1428.
    profiledtheco-precipitatedlincRNAsusingadirectRNAquantifica-tionmethod41(seeMethods.Weperformedimmunoprecipitationofthepolycombcomplexacrossfivebiologicalreplicatesandeightmock-IgGcontrols,andweassessedsignificanceusingapermutationtest(seeMethodsandSupplementaryFig.16.Altogether,weidentified24lincRNAs(,10%oftheEScelllincRNAsthatwerestronglyenrichedforbothpolycombcomponents(Fig.5bandSupplementaryTable17.TodetermineiflincRNAsinteractwithadditionalchromatinproteins,wesystematicallyanalysedchromatin-modifyingproteinsthathavebeenshowntohavecriticalrolesinEScells18–21,42.Specifically,wescreenedantibodiesagainst28chromatincomplexes(seeMethods,SupplementaryFig.14andSupplementaryTable18andidentified11additionalchromatincomplexesthatarestronglyandreproduciblyassociatedwithlincRNAs(seeMethodsandSup-plementaryFigs15and16.Thesechromatincomplexesareinvolvedin‘reading’(Prc1,Cbx1andCbx3,‘writing’(Tip60/P400,Prc2,Setd8,EsetandSuv39h1and‘erasing’(Jarid1b,Jarid1c,andHdac1histonemodifications,aswellasachromatin-associatedDNAbindingprotein(Yy1(Fig.5a.Altogether,wefoundthat74(,30%oftheEScelllincRNAsareassociatedwithatleast1ofthese12chromatincom-plexes(Fig.5bandSupplementaryTable17.Althoughmostoftheidentifiedinteractionsarewithrepressivechromatinregulators,thisisprobablyduetolimitationsofourselectioncriteriaandavailableantibodies.
    ManylincRNAsarestronglyassociatedwithmultiplechromatincomplexes(Fig.5b.Forexample,weidentified8lincRNAsthatbindtothePrc2H3K27andEsetH3K9methyltransferasecomplexes(writersofrepressivemarksandtheJarid1cH3K4demethylasecomplex(aneraserofactivatingmarks.Consistentwiththis,thePrc2andEsetcomplexeshavebeenreportedtobindatmanyofthesame‘bivalent’domains21andtoassociatefunctionallywiththeJarid1ccomplex43.Similarly,weidentifiedadistinctsetof17lincRNAsthatbindtothePrc2complex(‘writer’ofK27repressivemarks,Prc1complex(‘reader’ofK27repressivemarksandJarid1bcomplex(‘eraser’ofK4activating
    a
    Mark
    Mark
    T-statisticcomparedwithcontrols
    seenforprotein-codinggeneswhosepromotersareboundbythetranscriptionfactor(SupplementaryFig.12.Thestrongbutimper-fectcorrelationbetweentranscription-factor-bindingandeffectoftranscription-factorknockdownisconsistentwithpreviousobserva-tions39andmayreflectregulatoryredundancyinthepluripotencynetwork40.Inaddition,weprofiledtheknockdownofanadditional7pluripotency-associatedtranscriptionfactors(includingEsrrb,Zfp42andStat3.Altogether,for,60%oftheEScelllincRNAs,weidentifiedasignificantdownregulationuponknockdownof1ofthese11transcriptionfactors(Fig.4bandSupplementaryTable15.
    Afterretinoic-acid-induceddifferentiationofEScells,theEScelllincRNAsshowtemporalchangesacrossthetimecoursewith,75%showingadecreaseinexpressioncomparedtountreatedEScells(SupplementaryFig.13andSupplementaryTable16.Notably,allofthelincRNAsshowntoregulatepluripotencyaredownregulateduponretinoicacidtreatment(SupplementaryFig.13.OurresultsestablishthatlincRNAsaredirecttranscriptionaltargetsofpluripotency-associatedtranscriptionfactorsandaredynamicallyexpressedacrossdifferenti-ation.Collectively,theseresultsdemonstratethatlincRNAsareanimportantregulatorycomponentwithintheEScellcircuitry.
    T-statisticcomparedwithcontrols
    147lincRNAs
    MarkMarkMark
    MarkMark
    Chromatinreaders
    Jarid1b
    Jarid1c
    Ring1b
    Chromatinwriters
    Tip60/P400
    Suv39h1
    Chromatinerasers
    Hdac1
    Setd8
    Cbx1
    Cbx3
    b
    Prc2
    Eset
    Yy1
    Cluster1:
    K27reader/writer,K4eraser
    Cluster2:
    K27/K9writer,K4eraserCluster3:K9writersCluster4:
    K9writer/reader
    lincRNAsbinddiversechromatinproteins
    ToexplorehowlincRNAscarryouttheirregulatoryroles,westudiedwhetherlincRNAsphysicallyassociatewithchromatinregulatoryproteinsinEScells.WepreviouslyshowedthatmanyhumanlincRNAscaninteractwiththepolycombrepressivecomplex4,acom-plexthathasacriticalfunctionalroleintheregulationofEScells18,19.TodeterminewhethertheEScelllincRNAsphysicallyassociatewiththepolycombcomplex,wecrosslinkedRNA–proteincomplexesusingformaldehyde,immunoprecipitatedthecomplexusingantibodiesspecifictoboththeSuz12andEzh2componentsofpolycomb,and
    298|NATURE|VOL477|15SEPTEMBER2011
    Figure5|lincRNAsphysicallyinteractwithchromatinregulatoryproteins.a,Aschematicoftheclassesofchromatinregulatorsprofiled:readers(blue,writers(orangeanderasers(green.b,Aheat-mapshowingtheenrichmentof74lincRNAs(rowsfor1of12chromatinregulatorycomplexes(columns.Thenamesarecolour-codedbychromatin-regulatorymechanism.Majorclustersareindicatedbyverticallineswithadescriptionofthechromatincomponents.
    ©2011MacmillanPublishersLimited.Allrightsreserved
    74lincRNAs

    ARTICLERESEARCH
    marks(Fig.5b,aswellasotherfunctionallyconsistentreader,writeranderasercombinations(SupplementaryFig.17.Oneofseveralpoten-tialmodelsconsistentwiththesedataarethatlincRNAsmaybindtomultipledistinctproteincomplexes,perhapsservingas‘flexiblescaf-folds’tobridgefunctionallyrelatedcomplexesaspreviouslydescribedfortelomeraseRNA44.
    TodetermineiftheidentifiedlincRNA–proteininteractionshaveafunctionalrole,weexaminedtheeffectsongeneexpressionresultingfromknockdownofindividuallincRNAsthatarephysicallyasso-ciatedwithparticularchromatincomplexesandfromknockdownofgenesencodingtheassociatedcomplexitself(seeMethods.For.40%oftheselincRNA–proteininteractions,weidentifiedahighlysignificantoverlapinaffectedgeneexpressionprogramscomparedtojust,6%forrandomlincRNA–proteinpairs(seeMethodsandSupplementaryTable19.Othercasesmayreflectthelimitedpowertodetecttheoverlaps,becausespecificlincRNA–proteincomplexesmayberelatedtoonlyafractionoftheoverallexpressionpatternmediatedbythechromatincomplex.
    Together,thesedatademonstratethatmanyEScelllincRNAsphys-icallyassociatewithmultipledifferentchromatinregulatoryproteinsandthattheseinteractionsareprobablyimportantfortheregulationofgeneexpressionprograms.
    Sox2Oct4
    Nanog
    TF-YTF-X
    Sox2Oct4
    Nanog
    PolymeraseTF-YTF-XPolymerase
    PromoterPromoter
    Promoter
    Discussion
    AlthoughthemammaliangenomeencodesthousandsoflincRNA
    genes,fewhavebeenfunctionallycharacterized.Weperformedanunbiasedloss-of-functionanalysisoflincRNAsexpressedinEScellsandshowthatlincRNAsareclearlyfunctionalandprimarilyactintranstoaffectglobalgeneexpression.WeestablishthatlincRNAsarekeycomponentsoftheEScelltranscriptionalnetworkthatarefunctionallyimportantformaintainingthepluripotentstate,andthatmanyaredownregulatedupondifferentiation.TheEScelllincRNAsphysicallyinteractwithchromatinproteins,manyofwhichhavebeenpreviouslyimplicatedinthemaintenanceofthepluripotentstate18,20,21.Inadditiontochromatinproteins,lincRNAsinteractwithotherproteincomplexesincludingmanyRNA-bindingproteins(datanotshown.
    OurdatasuggestamodelwherebyadistinctsetoflincRNAsistranscribedinagivencelltypeandinteractswithubiquitousregula-toryproteincomplexestoformcell-type-specificRNA–proteincom-plexesthatcoordinatecell-type-specificgeneexpressionprograms(Fig.6.BecausemanyofthelincRNAsstudiedhereinteractwithmultipledifferentproteincomplexes,theymayactascell-type-specific‘flexiblescaffolds’44tobringtogetherproteincomplexesintolargerfunctionalunits(Fig.6.ThismodelhasbeenpreviouslydemonstratedfortheyeasttelomeraseRNA44andsuggestedfortheXIST45andHOTAIR46lincRNAs.ThehypothesisthatlincRNAsserveasflexiblescaffoldscouldexplaintheunevenpatternsofevolutionaryconservationseenacrossthelengthoflincRNAgenes3:themorehighlyconservedpatchescouldcorrespondtoregionsofinteractionwithproteincomplexes.
    AlthoughamodeloflincRNAsactingas‘flexiblescaffolds’isattrac-tive,itisfarfromproven.TestingthehypothesisforlincRNAswillrequiresystematicstudies,includingdefiningallproteincomplexeswithwhichlincRNAsinteract,determiningwheretheseproteininteractionsassembleonRNA,andascertainingwhethertheybindsimultaneouslyoralternatively.Moreover,understandinghowlincRNA–proteininteractionsgiverisetospecificpatternsofgeneexpressionwillrequiredeterminationofthefunctionalcontributionofeachinter-actionandpossiblelocalizationofthecomplextoitsgenomictargets.
    PluripotencyprogramLineageprogram
    Figure6|AmodelforlincRNAintegrationintothemolecularcircuitryofthecell.ES-cell-specifictranscriptionfactors(suchasOct4,Sox2andNanogbindtothepromoterofalincRNAgeneanddriveitstranscription.ThelincRNAbindstoubiquitousregulatoryproteins,givingrisetocell-type-specificRNA–proteincomplexes.Throughdifferentcombinationsofproteininteractions,thelincRNA–proteincomplexcangiverisetounique
    transcriptionalprograms.Right:asimilarprocessmayalsoworkinothercelltypeswithspecifictranscriptionfactorsregulatinglincRNAs,creatingcell-type-specificRNA–proteincomplexesandregulatingcell-type-specificexpressionprograms.
    lincRNAs,40protein-codinggenesand27negativecontrolswerehybridizedtoAgilentmicroarrays.DifferentiallyexpressedgenesweredefinedashavinganFDR,5%andfold-change.2-foldcomparedtocontrols.
    Screeningforpluripotencyeffects.Nanog-luciferaseEScells31wereinfectedandmeasuredafter8days.Hitswereidentifiediftheyreducedluciferaselevels(z,26acrossallreplicatesanddidnotreduceAlamarBluelevels.Hitswerevalidatedinwild-typeEScellsbymeasuringmRNAlevelsofOct4,Nanog,Sox2,Klf4andZfp42.Oct4expressionwasassessedusingimmunofluorescencestainingandmorphologywasvisuallyassessed.
    Lineageexpressioneffects.Lineageexpressionprogramsweredefinedusingpublisheddatasets(GeneExpressionOmnibusGSE12982,GSE11523,andGSE4082andcuratedgeneexpressionsignatures32,33.Overlapsingeneexpres-sioneffectswereassessedusingamodifiedGSEA34.ExpressionchangesinlineagemarkersweredeterminedusingqPCR.
    Transcriptionfactorbindingandregulation.ChIP-Seqdatawasdownloaded(GSE11724andGSE11431,alignedandanalysed.lincRNApromoterswereprevi-ouslydefinedusingH3K4me3peaks3.ChangesinexpressionofthelincRNAsuponknockdownofthetranscriptionfactorswereanalysedusingAgilentmicroarrays.Chromatinbindingandoverlapinexpression.EScellswerecrosslinkedwithformaldehyde,lysed,immunoprecipitated,washedandreversecrosslinked.RNAwashybridizedtotheNanostringcodeset.Wetestedantibodiesfor28chromatincomplexesandselectedsuccessfulantibodiesthathad.10lincRNAsexceedingafivefoldchangeandhadsignificantenrichmentsacross3replicates.Wecom-paredtheoverlapingeneexpressionusingamodifiedGSEA34.
    FullMethodsandanyassociatedreferencesareavailableintheonlineversionofthepaperatwww.nature.com/nature.Received28March;accepted26July2011.Publishedonline28August2011.1.2.
    TheFANTOMConsortium.Thetranscriptionallandscapeofthemammaliangenome.Science309,1559–1563(2005.
    Guttman,M.etal.Chromatinsignaturerevealsoverathousandhighlyconservedlargenon-codingRNAsinmammals.Nature458,223–227(2009.
    15SEPTEMBER2011|VOL477|NATURE|299
    METHODSSUMMARY
    RNAiexpressioneffects.WeclonedfiveshRNAstargetingeachlincRNAintoapuromycin-resistantlentiviralvector22.EScellswereplatedonpre-gelatinized96-wellplatesandinfectedwithlentivirusbeforeadditionofirradiatedDR4mouseembryonicfibroblasts(MEFs.Mediacontaining1mgml21puromycinwasadded24hafterinfection.On-targetknockdownwasassessedafter4daysandthebesthairpinshowingaknockdown.60%wasselected.RNAfrom147
    ©2011MacmillanPublishersLimited.Allrightsreserved

    RESEARCHARTICLE
    3.4.5.6.7.8.9.10.11.12.13.14.15.16.17.18.19.20.21.22.23.24.25.26.27.28.29.30.31.
    Guttman,M.etal.Abinitioreconstructionofcelltype-specifictranscriptomesinmouserevealstheconservedmulti-exonicstructureoflincRNAs.NatureBiotechnol.28,503–510(2010.
    Khalil,A.M.etal.ManyhumanlargeintergenicnoncodingRNAsassociatewithchromatin-modifyingcomplexesandaffectgeneexpression.Proc.NatlAcad.Sci.USA106,11667–11672(2009.
    Ponjavic,J.,Ponting,C.P.&Lunter,G.Functionalityortranscriptionalnoise?EvidenceforselectionwithinlongnoncodingRNAs.GenomeRes.17,556–565(2007.
    Mattick,J.S.ThegeneticsignaturesofnoncodingRNAs.PLoSGenet.5,e1000459(2009.
    Koziol,M.J.&Rinn,J.L.RNAtrafficcontrolofchromatincomplexes.Curr.Opin.Genet.Dev.20,142–148(2010.
    DeSanta,F.etal.AlargefractionofextragenicRNApolIItranscriptionsitesoverlapenhancers.PLoSBiol.8,e1000384(2010.
    Kim,T.K.etal.Widespreadtranscriptionatneuronalactivity-regulatedenhancers.Nature465,182–187(2010.
    Ebisuya,M.,Yamamoto,T.,Nakajima,M.&Nishida,E.Ripplesfromneighbouringtranscription.NatureCellBiol.10,1106–1113(2008.
    Ørom,U.A.etal.LongnoncodingRNAswithenhancer-likefunctioninhumancells.Cell143,46–58(2010.
    Huarte,M.etal.AlargeintergenicnoncodingRNAinducedbyp53mediatesglobalgenerepressioninthep53response.Cell142,409–419(2010.
    Rinn,J.L.etal.FunctionaldemarcationofactiveandsilentchromatindomainsinhumanHOXlocibynoncodingRNAs.Cell129,1311–1323(2007.
    Smith,A.G.Embryo-derivedstemcells:ofmiceandmen.Annu.Rev.CellDev.Biol.17,435–462(2001.
    Jaenisch,R.&Young,R.Stemcells,themolecularcircuitryofpluripotencyandnuclearreprogramming.Cell132,567–582(2008.
    Chen,X.etal.Integrationofexternalsignalingpathwayswiththecore
    transcriptionalnetworkinembryonicstemcells.Cell133,1106–1117(2008.Ivanova,N.etal.Dissectingself-renewalinstemcellswithRNAinterference.Nature442,533–538(2006.
    Boyer,L.A.etal.Polycombcomplexesrepressdevelopmentalregulatorsinmurineembryonicstemcells.Nature441,349–353(2006.
    Bernstein,B.E.etal.Abivalentchromatinstructuremarkskeydevelopmentalgenesinembryonicstemcells.Cell125,315–326(2006.
    Fazzio,T.G.,Huff,J.T.&Panning,B.AnRNAiscreenofchromatinproteinsidentifiesTip60-p400asaregulatorofembryonicstemcellidentity.Cell134,162–174(2008.
    Bilodeau,S.,Kagey,M.H.,Frampton,G.M.,Rahl,P.B.&Young,R.A.SetDB1contributestorepressionofgenesencodingdevelopmentalregulatorsandmaintenanceofEScellstate.GenesDev.23,2484–2489(2009.
    Moffat,J.etal.AlentiviralRNAilibraryforhumanandmousegenesappliedtoanarrayedviralhigh-contentscreen.Cell124,1283–1298(2006.
    Hu,G.etal.Agenome-wideRNAiscreenidentifiesanewtranscriptionalmodulerequiredforself-renewal.GenesDev.23,837–848(2009.
    Plath,K.,Mlynarczyk-Evans,S.,Nusinow,D.A.&Panning,B.XistRNAandthemechanismofXchromosomeinactivation.Annu.Rev.Genet.36,233–278(2002.Koerner,M.V.,Pauler,F.M.,Huang,R.&Barlow,D.P.Thefunctionofnon-codingRNAsingenomicimprinting.Development136,1771–1783(2009.
    Ponjavic,J.,Oliver,P.L.,Lunter,G.&Ponting,C.P.Genomicandtranscriptionalco-localizationofprotein-codingandlongnon-codingRNApairsinthedevelopingbrain.PLoSGenet.5,e1000617(2009.
    Sproul,D.,Gilbert,N.&Bickmore,W.A.Theroleofchromatinstructureinregulatingtheexpressionofclusteredgenes.NatureRev.Genet.6,775–781(2005.
    Silva,J.etal.Nanogisthegatewaytothepluripotentgroundstate.Cell138,722–737(2009.
    Chambers,I.etal.FunctionalexpressioncloningofNanog,apluripotencysustainingfactorinembryonicstemcells.Cell113,643–655(2003.Mitsui,K.etal.ThehomeoproteinNanogisrequiredformaintenanceofpluripotencyinmouseepiblastandEScells.Cell113,631–642(2003.
    Brambrink,T.etal.Sequentialexpressionofpluripotencymarkersduringdirectreprogrammingofmousesomaticcells.CellStemCell2,151–159(2008.
    32.Sherwood,R.I.etal.Prospectiveisolationandglobalgeneexpressionanalysisof
    definitiveandvisceralendoderm.Dev.Biol.304,541–555(2007.
    33.Aiba,K.etal.Definingdevelopmentalpotencyandcelllineagetrajectoriesby
    expressionprofilingofdifferentiatingmouseembryonicstemcells.DNARes.16,73–80(2009.
    34.Subramanian,A.etal.Genesetenrichmentanalysis:aknowledge-based
    approachforinterpretinggenome-wideexpressionprofiles.Proc.NatlAcad.Sci.USA102,15545–15550(2005.
    35.Niwa,H.,Miyazaki,J.&Smith,A.G.QuantitativeexpressionofOct-3/4defines
    differentiation,dedifferentiationorself-renewalofEScells.NatureGenet.24,372–376(2000.
    36.Pasini,D.,Bracken,A.P.,Hansen,J.B.,Capillo,M.&Helin,K.Thepolycombgroup
    proteinSuz12isrequiredforembryonicstemcelldifferentiation.Mol.Cell.Biol.27,3769–3779(2007.
    37.Jiang,H.etal.RoleforDpy-30inEScell-fatespecificationbyregulationofH3K4
    methylationwithinbivalentdomains.Cell144,513–525(2011.
    38.Marson,A.etal.ConnectingmicroRNAgenestothecoretranscriptionalregulatory
    circuitryofembryonicstemcells.Cell134,521–533(2008.
    39.Kunarso,G.etal.Transposableelementshaverewiredthecoreregulatorynetwork
    ofhumanembryonicstemcells.NatureGenet.42,631–634(2010.
    40.Jiang,J.etal.AcoreKlfcircuitryregulatesself-renewalofembryonicstemcells.
    NatureCellBiol.10,353–360(2008.
    41.Geiss,G.K.etal.Directmultiplexedmeasurementofgeneexpressionwithcolor-codedprobepairs.NatureBiotechnol.26,317–325(2008.
    42.Dey,B.K.etal.ThehistonedemethylaseKDM5b/JARID1bplaysaroleincellfate
    decisionsbyblockingterminaldifferentiation.Mol.Cell.Biol.28,5312–5327(2008.
    43.Cloos,P.A.,Christensen,J.,Agger,K.&Helin,K.Erasingthemethylmark:histone
    demethylasesatthecenterofcellulardifferentiationanddisease.GenesDev.22,1115–1140(2008.
    44.Zappulla,D.C.&Cech,T.R.YeasttelomeraseRNA:aflexiblescaffoldforprotein
    subunits.Proc.NatlAcad.Sci.USA101,10024–10029(2004.
    45.Wutz,A.,Rasmussen,T.P.&Jaenisch,R.Chromosomalsilencingandlocalization
    aremediatedbydifferentdomainsofXistRNA.NatureGenet.30,167–174(2002.46.Tsai,M.C.etal.LongnoncodingRNAasmodularscaffoldofhistonemodification
    complexes.Science329,689–693(2010.SupplementaryInformationislinkedtotheonlineversionofthepaperatwww.nature.com/nature.
    AcknowledgementsWethankD.Rivera,T.Green,T.Bhimdi,G.Verstappen,C.Surka,S.Silver,A.Brown,D.LamandO.Ramfortechnicalhelp;C.Gifford,S.MarkoulakiandR.Jaenischforprovidingcelllinesusedinthisstudy;P.Tsang,B.Curry,A.TsalenkoandAgilentTechnologiesformicroarrayandtechnicalhelp;B.ChallisandActiveMotifforantibodies;G.Geiss,R.BoykinandNanostringtechnologiesfortechnicalhelp;E.WangandC.BurgeforhelpwithRNAimmunoprecipitationexperimentsandhelpfuldiscussions;P.Gupta,A.Gnirke,J.Cassady,E.Lieberman-Aiden,M.CabiliandM.Thompsonfordiscussionsandideas;andL.Gaffneyforassistancewithfigures.M.GuttmanisaVertexscholar.ThisworkwasfundedbyNHGRI,aCenterforExcellenceforGenomicScience,theMerkinFoundationforStemCellResearch,andfundsfromtheBroadInstituteofMITandHarvard.
    AuthorContributionsM.GuttmanandE.S.L.conceivedanddesignedtheoverallprojectwithhelpfromA.M.,A.R.,J.L.R.andD.E.R.;M.GuttmanandJ.D.designedexperimentswithhelpfromJ.K.G.,X.Y.(RNAi,B.W.C.(pluripotencyassaysandI.A.(RNAIP;M.Guttman,J.D.,G.M.,A.B.L.,R.A.andG.Y.performedexperiments;M.
    Guttman,J.D.andM.Garberanalyseddata;L.B.,A.M.andD.E.R.providedreagents;andM.GuttmanandE.S.L.wrotethemanuscript.
    AuthorInformationMicroarraydatahavebeendepositedintheGeneExpressionOmnibus(GEOunderaccessionnumberGSE30245.Reprintsandpermissionsinformationisavailableatwww.nature.com/reprints.Theauthorsdeclarecompetingfinancialinterests:detailsaccompanythefull-textHTMLversionofthepaperatwww.nature.com/nature.Readersarewelcometocommentontheonlineversionofthisarticleatwww.nature.com/nature.CorrespondenceandrequestsformaterialsshouldbeaddressedtoM.Guttman(mguttman@mit.eduorE.S.L.(lander@broadinstitute.org.
    300|NATURE|VOL477|15SEPTEMBER2011
    ©2011MacmillanPublishersLimited.Allrightsreserved

    ARTICLERESEARCH
    METHODS
    EScellculture.V6.5(genotype129SvJae3C57BL/6andNanog-luciferase31EScellswereco-culturedwithirradiatedC57BL/6MEFs(GlobalStem;GSC-6002Conpre-gelatinizedplatesaspreviouslydescribed47.Briefly,cellswereculturedinmESmediaconsistingofknockoutDMEM(Invitrogen;10829018supplementedwith10%FBS(GlobalStem;GSM-6002,1%penicillin-streptomycin(Invitrogen;15140-163,1%L-glutamine(Invitrogen;25030-164,0.001%b-mercaptoethanol(Sigma;M3148-100MLand0.01%ESGRO(Millipore;ESG1106.
    PickinglincRNAgenecandidates.UsingourpreviouscatalogueofK4-K36definedlincRNAs2alongwiththereconstructedfull-lengthsequenceswedeter-minedusingRNA-Seq3,wedesignedshRNAhairpinstargetingeachlincRNAidentifiedinbothsets.Specifically,weusedtheconservativeK4-K36definitionsfromourpreviouswork2thatwereexpressedinmouseEScells.Wefurtherfilteredthelisttoincludeonlymulti-exoniclincRNAsthatwerereconstructedinmouseEScells3.Together,thisyielded226lincRNAgenes.
    Pickingprotein-codinggenecandidates.Weselectedproteincodinggenecon-trolsconsistingofbothtranscriptionfactorsandchromatinproteins.Theseproteinswereselectedbasedontheirwell-characterizedroleinregulatingmouseEScellsandincludeOct4(Pou5f135,48,Sox2(refs17,49Nanog(refs29,30,Stat3(ref.50,Klf4(ref.51andZfp42(Rex152.Inaddition,weselectedadditionaltranscriptionalandchromatinregulatorsthatwereidentifiedbyRNAiscreensasregulatorsofpluripo-tency17,20,23and/orwerefoundinsmallerfocusedstudiestohavecriticalrolesinthemaintenanceofthepluripotentstate(suchasCarm1(ref.53,Chd1(ref.54,Thap11(ref.55,Suz12(refs18,19,36andSetdb1(refs21,56.AfulllistisprovidedinSupplementaryTable2.
    shRNAdesignrules.ForeachlincRNAwedesignedfivehairpinsbyextendingthepreviouslydescribeddesignrules22accountingforthesequencecontentofthehairpin,miRNAseedmatches,uniquenesstothetargetcomparedtothetran-scriptomeandthegenome,andnumberoflincRNAisoformscovered.
    ForeachlincRNAweenumeratedall21-mersub-sequencesandscoredthemasfollows:(1a‘clampscore’wascomputedbylookingatthenucleotidesatpositions18,19and20.IfallthreepositionscontainedanA/Titwasassignedascoreof4,iftwopositionswereA/Titwasassignedascoreof1.5andifonewasA/Titwasassignedascoreof0.8.Wethenlookedatpositions16,17,and21;ifallthreewereA/Titwasassignedascoreof1.25,iftwowereA/Titwasassignedascoreof1.1,andifonewasA/Tiswasassignedascoreof0.8.Theclampscorewascomputedastheproductofthesetwoscores.(2A‘GCscore’wascomputedbylookingatthetotalGCpercentageofthe21-mersequence.Ifthesequencewas,25%GCitwasassignedascoreof0.01,ifitwas,55%itwasassignedascoreof3,ifitwas,60%itwasassignedascoreof1,andif.60%itwasassignedascoreof0.01.(3A‘4-merpenalty’of0.01wasassignedforanyhairpincontainingthesamenucleotidein4subsequentnucleotides.(4A‘7GCpenalty’of0.01wasassignedtoanyhairpincontainingany7consecutiveG/Cnucleotides.(5WeremovedallhairpinscontaininganAineitherposition1orposition2ofthehairpin.(6Weremovedallhairpinscontainingarepeatmaskednucleotide.(7Finally,wecomputeda‘miRNA-seedpenalty’bylookingattheforwardpositions11–17,12–20and13–19ofthehairpinaswellasthereversecomplementofpositions14–20,15–21,or16–21plusa39C.WethenlookedupwhetherthesepositionsmatchedknownmiRNAseedsandwithwhatfrequency.Wecomputedthescoresfortheforwardandreversepositionsanddefinedthescoreastheproductoftheforwardandreversescores.Thefinalscoreforeachhairpinsequenceisdefinedastheproductofallsevenscores.
    Wethensortedthecandidatehairpinsequencesbyscore,breakinghigh-scoringtiesbythetotalnumberoflincRNAisoformsthatarecoveredbythehairpin.WethenalignedeachhairpinsequenceagainstboththegenomeandtheRefSeq-definedtranscriptome(NCBIRelease39,andfilteredanyhairpinwithfewerthanthreemismatchestoanyothergeneorpositioninthegenome.CandidatesequenceswerechosenforshRNAproductionbyfirstpickingthehighestscoringcandidateandthenproceedingtosuccessivelylowerscores.Aseachhairpinwasselected,allotherhairpinsoverlappingthishairpinwereremoved.WerepeatedthisprocessuntilweidentifiedfivehairpinsthatcoveredeachlincRNA.
    shRNAcloningandvirusprep.Wedesigned1,143hairpinstargeting226lincRNAgenes.Ofthese,wesuccessfullycloned1,010hairpinstargeting214lincRNAs.Thesehairpinswereclonedintoavectorcontainingapuromycinresistancegeneandincorporatedintoalentiviralvectoraspreviouslydescribed22.Briefly,syntheticdouble-strandedoligosthatrepresentastem-loophairpinstruc-turewereclonedintothesecond-generationTRC(theRNAiConsortiumlentiviralvector,pLKO.5;theexpressionofagivenhairpinproducesashRNAthattargetsthegeneofinterest.Lentiviruswaspreparedaspreviouslydescribed22.Briefly,100ngofshRNAplasmid,100ngofpackagingplasmid(psPAX2and10ngofenvelopeplasmid(VSV-Gwereusedtotransfectpackagingcells(293TwithTransIT-LT1(MirusBio.Viruswasharvested48and70hafter
    transfection.Twoharvestswerecombined.Virustitresweremeasuredasprevi-ouslydescribed22.Briefly,wemeasuredvirustitresbyinfectingA549cellswithappropriatelydilutedviruses.Twenty-fourhoursafterinfection,puromycinwasaddedtoafinalconcentrationof5mgml21andtheselectionproceededfor48h.Thenumberofsurvivingcells,whichiscorrelatedtovirustitre,wasmeasuredbyAlamarBlue(BioSourcestainingusingtheEnvision2103Multilabelplatereader(PerkinElmer.
    Infectionandselectionprotocol.V6.5EScellsorNanog-luciferaseEScellswereplatedatadensityof5,000cellsperwell(8-daytimepointor25,000cellsperwell(4-daytimepointin100mlmESmediaontopre-gelatinized96-welldishes(VWR;BD356689.Cellswereinfectedwith5mlofalentiviralshRNAstockandincubatedat37uCfor30min.Puromycin-resistantDR4MEFs(GlobalStem;GSC-6004Gwerethenaddedtotheplatesatadensityof,6,000cellsperwellandincubatedovernightat37uC,5%CO2.After24h,allmediawasremovedfromthecellsandreplacedwithmediacontaining1mgml21puromycin.Mediawasthenchangedeveryotherdaywithfreshmediacontaining1mgml21puromycin.Theend-pointdependedontheassayandwaseither4daysafterinfection(knockdownvalidationandmicroarraysor8days(reportersandqPCRofmarkergenes.
    RNAextraction.EScellswereinfectedandlysedatday4with150mlofQiagen’sRLTbufferandthreereplicatesofeachvirusplatewerepooledforRNAextrac-tionusingQiagen’sRNeasy96-wellcolumns(74181.RNAextractionwascom-pletedfollowingQiagen’sRNeasy96-wellprotocolwiththefollowingmodifications:450mlof70%ethanolwasaddedto450mltotallysatebeforethefirstspin.AnadditionalRPEwashwasaddedtotheprotocol,foratotalofthreeRPEwashes.
    lincRNAprimerdesignandpre-screen.lincRNAprimersweredesignedusingprimer3(http://frodo.wi.mit.edu/primer3/.Specifically,wedesignedprimersspanningexon–exonjunctionsbyspecifyingeachoftheregionsaspreferredinclusionregionsintheprimer3program.Whenalow-scoringprimerpair(primerpenalty,1wasavailableitwasused.Ifnonewasavailable,wethenidentifiedallprimersthatcontainedampliconsthatspannedanexon–exonjunc-tion.Inafewcases,whenwecouldnotidentifyaprimerpairspanninganexon–exonjunction,wedesignedprimerswithinanexonofthelincRNA.Foreachprimerpair,wetestedthespecificityagainstthetranscriptome57(RefSeqNCBIRelease39andthegenome(MouseMM9usingtheisPCR(http://genome.ucsc.edu/cgi-bin/hgPcrprogram.Specifically,werequiredthattheprimerpairamp-lifythelincRNAgeneandnoothergenomicofgeneamplicon.
    Foreachprimerpair,wevalidatedthequantificationandspecificitybeforeuse.Specifically,wetestedprimersinqPCRreactionsusingadilutionseriesofmouseEScDNAincludinganoreversetranscriptase(RTsample.Weexcludedanyprimerthatdidnothaverobustquantificationacrossa64-folddilutioncurve,hadhighsignalinthenoRTsample,orhadlowdetectableexpressionintheundilutedsample(cyclenumber.34.Forprimersthatfailedthisvalidationweredesignedandtestednewprimers.
    KnockdownvalidationusingqPCR.TodetermineiflincRNAhairpinswereeffectiveatknockingdownthelincRNAofinterest,weinfectedeachhairpinintomouseembryonicstemcells,selectedforlentiviralintegration,andmeasuredchangesinthetargetedlincRNAexpressionlevel.WeisolatedtotalcellularRNAafter4days;thistimepointwaschosentoallowforidentificationofrobustchangeswhileminimizingsecondaryeffectsduetodifferentiationoftheEScells.WereasonedthatthiswouldallowustodeterminemoredirecteffectsduetoRNAiratherthantodifferentiation.
    GenepanelswereconstructedthatcontainedallfivehairpinstargetingagenealongwithanemptyvectorcontrolpLKO.5-nullTandtheGFP-targetinghairpinclonetechGfp_437s1c1.cDNAwasgeneratedusing10mlofRNAand10mlof23cDNAmastermixcontaining53TranscriptorRTReactionBuffer(Roche,DTT,MMLV-RT(Roche,dNTPs(Agilent;200415-51,Random9-meroligos(IDT,Oligo-dT(IDTandwater.cDNAwasdiluted1:9andquantitativePCRwasperformedusing250nMofeachprimerin23Sybrgreenmastermix(RocheandrunonaRocheLight-Cycler480.TargetlincRNAexpressionandGapdhlevelswerecomputedforeachpanel.lincRNAexpressionlevelswerenormalizedbyGapdhlevelsandthisnormalizedvaluewascomparedtothereferencecontrolhairpinswithinthepanel.Knockdownlevelswerecomputedastheaverageofthefolddecreasecomparedtothetwocontrolhairpins.Hairpinsshowingaknockdowngreaterthan60%oftheendogenouslevelwereconsideredvalidatedandthebestvalidatedhairpinfromalincRNApanelwasselectedformicroarraystudies.
    Pickingcandidatesformicroarrayanalysis.ToassesstheeffectsofalincRNAongeneexpression,weprofiledthechangesingeneexpressionafterknockingdowneachlincRNAgene.Specifically,foreachlincRNAwithatleastonevali-datedhairpinweprofiledthegenome-wideexpressionlevelchangesafterknock-downacrosstwoindependentinfections(seeabove.Tocontrolforexpression
    ©2011MacmillanPublishersLimited.Allrightsreserved

    RESEARCHARTICLE
    changesduetoviralinfection,weperformedfiveindependentinfectionscontain-ingnoRNAihairpin(pLKO.5-nullT.ThiscontrolhairpinwasembeddedineachRNApreparationplate.Tocontrolforeffectsduetoanoff-targetRNAieffect,weprofiled27distinctnegativecontrolhairpinswhichdonothaveaknowntargetinthecell.Thesehairpinsincluded6RFPhairpins,10GFPhairpins,6luciferasehairpinsand5LacZhairpins.ThesehairpinsprovideameasurementofthevariabilityoftheRNAiresponsetriggeredduetononspecificeffects.Furthermore,weprofiledhairpinstargeting147lincRNAs,including10withasecondbesthairpin,and40protein-codinggenesinbiologicalreplicate.Thehairpinsandtheirreplicateswererandomlydistributedacross796wellplatesandpreparedinbatches.EachRNApreparationbatchcontainedonepLKOhairpinandoneclonetechGfp_437s1c1hairpininarandomlocationontheplate.Tominimizebatcheffects,theplatelocationsofthebiologicalreplicateswerescrambledandthepositionswithintheplateswerescrambledforallhairpinsandreplicates.
    Agilentmicroarrayhybridization.UsingAgilent’sOne-ColourQuickAmpLabellingkit(5190-0442,weamplifiedandlabelledtotalRNAforhybridizationtoprototypemouselincRNAarrays(G4140-90040accordingtomanufacturer’sinstructionswithafewvariations.ThecustomAgilentSurePrintG38x60Kmousearraydesignusedforthisstudy(G4102A,AMADID025725G4852Ahasprobesto21,503Entrezgenesand2,230lincRNAgenes.Anewupdatedversionofthismousedesigniscommerciallyavailablethatcontainsprobesto34,017Entrezgenetargetsaswellas2,230lincRNAgenes(G4825A.ThecRNAsampleswerepreparedbydiluting200ngofRNAin8.3mlwaterandaddingpositivecontrolone-colourRNAspike-inmix(Agilent,5188-5282thatwasdilutedserially1:20,then1:25andfinally1:10.WeannealedtheT7promoterprimerfromthekitbyincubatingat65uCfor10min.WepreparedthecDNAmastermixandaddedittotheannealedRNAandincubatedat40uCfor2h,followedby65uCfor15min.WepreparedthecRNAtranscriptionmastermixandaddedittothecDNAandincubatedat40uCfor2hprotectedfromlight.WepurifiedthelabelledcRNAusingQiagen’sRNeasy96-wellcolumns(Qiagen,74181byadding350mlofQiagenRLT(withoutBMEtothecRNAfollowedbytheadditionof250mlof95%ethanolbeforeapplyingtotheplatecolumn.Aftera4minspinat6,000r.p.m.,wewashedthecolumnsthreetimeswith800mlbufferRPE.Wedriedthecolumnsbyspinningfor10minandelutedthecRNAwith50mlofwater.WemeasuredthecRNAyieldanddyeincorporationusingtheNanodrop8000Microarraymeasurementsetting.Wemixed600ngofcRNAwithblockingagentandfragmentationbuffer(Agilent,5190-0404andfragmentedfor30mininthedarkat60uC.Weadded23hybridizationbuffertoeachsampleandloaded40mlontoan8-packHybridizationgasket.Weplacedthemicroarrayslidesontop,sealedinthehybridizationchamber,andincubatedfor18hat65uC.Wewashedtheslidesfor1mininroomtemperatureGEWashBuffer1andthenfor1minin37uCGEWashBuffer2(Agilent5188-5327,notritonaddition.WescannedthemicroarraysusinganAgilentScannerC(G2565CAusingthefollow-ingsettings:dyechannel5red&green,scanregion5scanarea(61321.6mm,scanresolution53mm.Wepreparedallofthesamplessimultaneouslyusinghomogenousmastermixestolimitvariability.Fragmentationandhybridizationwasstaggeredovertimeinbatchesof3to4slides(24to32samples.
    Arrayfiltering,normalizationandprobefiltering.EacharraywasprocessedanddataextractedusingtheAgilentfeatureextractionsoftware(G4462AA,Version10.7.3.Sampleswereretainediftheypassedallthefollowingqualitycontrolstatistics:AnyColourPrcntFeatNonUnifOL,1;eQCOneColourSpikeDetectionLimit.0.01and,2.0;Metric_absGE1E1aSlopebetween0.9and1.2;Metric_gE1aMedCVProcSignal,8;gNegCtrlAveBGSubSig.210and,5;Metric_gNegCtrlAveNetSig,40;gNegCtrlSDevBGSubSig,10;Metric_gNonCntrlMedCVProcSignal,8;Metric_gSpatialDetrendRMSFilterMinusFit,15;SpotAnalysis_PixelSkewCookiePct.0.8and,1.2.
    GeneexpressionvaluesweredeterminedusingthegProcessedSignalintensityvalues.Probeswereflaggediftheywerenotdetectablewellabovebackgroundorhadanexpressionlevellowerthanthelowestdetectablespike-incontrolvalue.Thevalueswereflooredacrossallsamplesbytakingthemaximumoftheminimumnon-flaggedvaluesacrossallexperiments.Anyvaluelessthanthismaximumvaluewassettothemaximum.Thisconservativelyeliminatesanydetectionvariabilityacrossthesamplesduetostringencyorotherarrayvariables.
    Theresultofthisisasinglevalueforeachprobeperarray.Tonormalizeexpressionvaluesacrossarrays,weperformedquantilenormalizationasprevi-ouslydescribed58.Briefly,werankedeacharrayfromlowesttohighestexpression.Foreachrank,wecomputedtheaverageexpressionandeachexperimentwiththisvalueattheassociatedrank.Foreachprobe,wecomputedthedifferencebetweenthesecondsmallestexpressionvalueandthesecondlargestexpressionvalue.Ifthisdifferencewaslessthan2,wefilteredtheprobe.Thismetricwaschosentoeliminatebiasduetosinglesampleoutliers.
    IdentifyingsignificantgeneexpressionhitsfromRNAiknockdowns.Tocon-trolforeffectsduetononspecificeffectsofshRNAs,weprofiled27distinct
    negativecontrolhairpinswhichdonothaveaknowntargetinthecell.Thesehairpinsprovideameasurementofthevariabilityoftheexpressionprofilesduetorandomvariabilityortriggeredby‘off-target’effectsoftheshRNAlentiviruses.Assumingthatanyobservedeffectsinthenegativecontrolhairpinsareduetooff-targeteffectsandobservedeffectsinthetargetinghairpinsincludeamixofbothoff-targeteffectsandon-targeteffects,weusepermutationsofthenegativecontrolstoassignaFDRconfidencelevelforbeinganon-targethittoeachgene.Assuch,agenewouldonlyreachgenome-widesignificanceifthenumberofgenesandscaleoftheeffectwasmuchlargerthanwouldbeobservedrandomlyamongalloftheexpressionchangesfoundforthenegativecontrolhairpin.Specifically,foreachgenewecomputedat-statisticbetweenshRNAstargetingthelincRNAandcontrolshRNAsamples.Toassessthesignificanceofeachgenewepermutedthesampleandcontrolgroupsretainingtherelativesizesofthegroupsandcomputingthesamet-statistic.WethenassignedanFDRvaluetoeachgenebycomputingtheaveragenumberofvaluesinthepermutedt-statisticsthatweregreaterthantheobservedvalueofinterestanddividedthisbythenumberofallobservedt-statisticsthatweregreaterthantheobservedvalue.WedefinedgenesassignificantlydifferentiallyexpressediftheFDRwas,5%andthefold-changecomparedtothenegativecontrolswas.2-fold.Usingthisapproach,aneffectwouldonlyreachasignificantFDRifthescaleissignificantlylargerthanwouldbeobservedinthenegativecontrols.KnockdownofalincRNAwasconsideredtohaveasignificanteffectongeneexpressionifweidentifiedatleast10genesthathadaneffectthatpassedallofthecriteria.
    Gene-neighbouranalysis.WeidentifiedneighbouringgenesbasedontheRefSeqgenomeannotation57(NCBIRelease39.WeexcludedfromanalysisallRefSeqgenesthatcorrespondedtoourlincRNAofinterestbutincludedallothercodingandnon-codingtranscripts.WeidentifiedasignificanthitasanylincRNAaffect-inganeighbourwithin10genesoneithersidewithanFDR,0.05andtwofoldexpressionchange.Tocomputetheclosestaffectedneighbour,weclassifiedallgenesaffecteduponknockdownofthelincRNAsusingthesamecriteriaabove.WecomputedthedistancebetweeneachaffectedgeneandthelocusofthelincRNAgene(andprotein-codinggenethatwasperturbedandtookthemin-imumabsolutedistanceacrossallaffectedgenes.
    Analysisofexpectednumberofneighbouringgenesthatwillchangebychance.Todeterminetheexpectednumberofdifferentiallyexpressed‘neighbouring’genesoccurringbychanceassumingthattheknockdownhasnoeffectongeneexpression,wecalculatedtheaveragenumberofgenesina300-kbwindowaroundarandomlyselectedgeneinthehumanandmousegenome.Wecalculatedthistobe11.2(humanand11.8(mouse.Forsimplicity,wewillconservativelyroundthisdownto11.Assumingthatnogenesarechangingbetweentheknockdownandcontrol,usinganominalP-value,whichhasauniformdistributionunderthenullhypothesis(nothingeffected,wewouldexpecttoseeadifferencecalledin5%ofcasesataP-valueof0.05.Ifwetestonelocus,whichhasonaverage11neighbours,wewouldexpecttoidentify0.55hitsbychance(1130.0550.55.However,ifwenowtest12lociwewouldexpecttosee6.6(1230.55knockdownsthatappeartohaveaneffectunderthenullhypothesis.
    LuciferaseanalysisofNanogESlines.EScellscontainingaNanog-luciferaseconstruct31wereinfectedinbiologicalduplicateandmonitoredafter7days.LuciferaseactivitywasmeasuredusingBright-Glo(Promega.Allreagentsandcellswereequilibratedtoroomtemperature.100mlBright-Glosolutionwasaddedtoeachplatewell.Plateswereincubatedinthedarkatroomtemperaturefor10minandluciferasewasmeasuredonaplatereader.TheluciferaseunitswerenormalizedtothecontrolhairpinsandaZ-scorecomparedtothenegativecontrols(excludingluciferasehairpinswascomputed.Foreachhairpin,wecomputedaZ-scorerelativetothenegativecontrolhairpinsandidentifiedhitsreducingluciferaselevelsmorethan6standarddeviations(Z,26forbothindependentreplicates.Inallcaseswewereabletoidentifyasignificantreductioninluciferaselevelswhenusingdistincthairpinstargetingluciferase.Toexcludehitsthatwereduetoanoverallreductioninproliferation(whichwouldalsocauseareductionofNanogpositivecellsinthisread-outweexcludedallhairpinsthatcausedareductioninproliferationasmeasuredbyAlamarBlueincorporation(describedbelow.AlamarBlueincorporationwasmeasuredinthesamecellsimmediatelybeforereadingoutNanog-luciferaselevels.
    AlamarBlueanalysisofESlines.Aftera7-dayinfection,Nanog-luciferasecellviabilitywasmeasuredusingAlamarBlue(Invitrogen;DAL1025.AlamarBluewasmixedwithmESmediaina1:10ratio,addedtothecellsandincubatedat37uCfor1h.Absorbancereadingsat570nmweretaken.Tocontrolforpossibleeffectsduetovirustitre,wemeasuredAlamarBlueincorporationonbothpuromycintreatedandnon-puromycintreatedsamplesforeachinfection.
    mRNAanalysisofpluripotencymarkers.V6.5EScellswereinfectedwithshRNAstargetinglincRNAs,protein-codinggenes,and21negativecontrols.After8days,RNAwasextractedandmRNAlevelsoftheOct4,Nanog,Sox2,
    ©2011MacmillanPublishersLimited.Allrightsreserved

    ARTICLERESEARCH
    Klf4andZfp42pluripotencymarkerswereanalysedusingqPCR.PrimersequencesarelistedinSupplementaryTable9.EachsamplewasnormalizedtoGapdhlevels.Significancewasassessedcomparedtothenegativecontrolhairpinsusingaone-tailedt-test.
    Tocontrolforoff-targeteffects,weanalysedadditionalhairpinsagainstthe26lincRNAsaffectingNanog-luciferaselevels.Ofthe26lincRNAs,weidentified15lincRNAsthatcontainedanadditionalhairpinthatreducedlincRNAexpressionby.50%.V6.5EScellswereinfectedwiththebestandadditionalhairpinacrossbiologicalreplicatesforthese15lincRNAsand21negativecontrolhairpins.RNAwasextractedafter8daysandOct4expressionlevelsweredeterminedusingqPCR.Significancewasassessedrelativetothenegativecontrolsusingaone-tailedt-test.
    Immunofluorescence.Wecrosslinkedcellsin4%paraformaldehydefor15min,andwashedin13PBSthreetimes.Topermeabilizethecells,wewashedwith13PBS10.1%Tritonandthenblockedin13PBS10.1%Triton11%BSAfor45minatroomtemperature.Weincubatedcellswithanti-Pou5f1antibody(SantaCruz:SC-9081at1:100dilutioninblockingsolutionfor1.5hatroomtemperatureandthenwashedinblockingsolutionthreetimes.Next,weincubatedcellsinanti-rabbitsecondaryantibodycoupledtoGFP(JacksonImmunoResearch:111-486-152atadilutionof1:1,000inblockingsolutionfor45min.Finally,wethoroughlywashedcellsinblockingsolutionthreetimes,andaddedvectashieldcontainingDAPI(VWR:101098-044toeachwell.
    Publicdatasetcuration.Traditionally,lineagemarkersareusedtoidentifychangesinphenotypicstates.Althoughthesemarkerscanbegoodindicatorsofdifferentiationpotential,therearetwomajorlimitationswiththisapproach.First,therearemultiplegenesthatareassociatedwitheachlineagesosimplylookingatonecanoftenbemisleading.Second,thisapproachonlyworksforclassifyingstateswithwell-characterizedmarkergenesbutwouldnotworkforacomprehensivecharacterizationofthefunctioninthecell.Therefore,wedecidedtotakeadifferentapproachandlookattheentiregeneexpressionprofileofeachlincRNAknockdowntodeterminewhatcellstateeachlincRNAresembles.WecuratedasetofESperturbationsanddifferentiationstatesfrompubliclyavailablesources.Specifically,weusedtheNCBIe-utils(http://eutils.ncbi.nlm.nih.gov/toprogrammaticallyidentifyallpublisheddatasetscontainingkeywordsassociatedwithembryonicstemcells.Wefilteredthelisttoonlyincludemousedatasetsthatweregeneratedacrossoneofthreecommercialarrayplatforms(Affymetrix,AgilentandIllumina.Followingthisapproach,wemanuallycuratedthelisttoincludedatasetsassociatedwithEScellperturbations(geneticdeletions,RNAi,orchemicalperturbationsanddifferentiationorinduceddifferentiationprofiles.Thiscurationyielded41GEOdatasetscorrespondingto.150samples.Specifically,wedefineddifferentiationlineagestatesusingthefollowingdatasets.(1Neuroectoderm:wedownloadedadataset(GSE12982correspondingtomouseEScellscontainingaSox1–GFPreporterconstruct.UpondifferentiationofSox1–GFPEScellsintoembryoidbodies(EBs,Sox1–GFP-positivecellswerecollectedandtheirglobalexpressionwasprofiled59.Inaddition,wedownloadedadataset(GSE408260correspondingtodirectneuroectodermdifferentiation61.(2Mesoderm:wedownloadedthesamedataset(GSE12982asabove,wheretheauthorsdifferentiatedbrachyury–GFPreporterEScellsintoEBsandsortedandprofiledbrachyury–GFP-positivecells59.
    (3Endoderm:wedownloadedadataset(GSE11523correspondingtomouseEScellswhichwereengineeredtooverexpressGATA633.GATA6overexpressionhasbeenshowntodriveEScellsintoaprimitiveendoderm-likestate62.
    (4Ectoderm:wedownloadedadataset(GSE408260correspondingtomouseEScellsdifferentiatedintoprimitiveectoderm-likecellswithdefinedmedia61.(5Trophectoderm:wedownloadedadataset(GSE1152333correspondingtomouseEScellswhichwereengineeredtodepleteOct435.Thesecellshavebeenshowntoenteratrophectoderm-likestate35.Toensurespecificitytothetrophectodermstate,wealsocomparedtheexpressioneffectstotrophoblaststemcells33.ForalllincRNAsidentified,werequiredasignificantenrichmentforbothinducedOct4knockoutandtrophoblaststemcellprograms.
    Inaddition,foralllineagestatesweusedacurateddiscretegeneexpressionsignatureofdifferentiationwhichwaspreviouslyfunctionallytestedandshowntocorrespondspecificallytodifferentiationintotheassociatedstates63.
    Continuousenrichmentanalysisandphenotype-projectionanalysis.Todeter-minerelationshipsbetweenlincRNAknockdownsandfunctionalstates,weusedamodifiedGeneSetEnrichmentAnalysis34approachthataccountsforthecon-tinuousnatureofthetwodatasets,similartopreviouslydescribedexten-sions34,64,65.ForeachlincRNAknockdownbyfunctionalpairwecomputeacontinuousenrichmentscore.Specifically,(1foreachlincRNAknockdownwecomputeanormalizedscorematrixcomparedtoapanelofnegativecontrolhairpinsbycomputingat-statisticforeachgenebetweenthereplicatelincRNAknockdownexpressionvaluesandthecontrolknockdownvalues.(2Foreachexperiment,wesortthematrixbythenormalizedscoresuchthatthemost
    differentiallyexpressedupregulatedgeneisfirstandthemostdifferentiallyexpresseddownregulatedgeneislast.UsingthisorderingwesortthefunctionaldatasetsuchthattheorderingcorrespondstothedifferentialrankofthelincRNAknockdownset.(3WecomputeascoreSiastherunningaverageofvaluesfromthefirstpositiontopositioni.WethendefinetheenrichmentscoreEasthemaximumoftheabsolutevalueofSiforallvaluesofi.10.Werequirei.10toavoidsmallfluctuationsinthebeginningoftherankedlistcausingfluctuationsintheenrichmentscore.ThisscoreiscomputedforeachlincRNAknockdownbyfunctionalset.BecausewehavemanylincRNAknockdownsandfunctionalsets,inrealitywehaveamatrixofscoresandwewillrefertotheenrichmentscoreoftheithknockdownandjthfunctionalsetasEij.
    Toassessthesignificanceofthesescores,wecomputeapermutation-derivedFDRandassignaconfidencevalueforeachprojection.Specifically,toassessthesignificanceofEij,wepermutethelincRNAknockdownsamplesandcontrolsamplesandcomputetheenrichmentscoreforeachpairacrossallpermutations.ToaccountfortheFDRassociatedwithmanylincRNAsandfunctionalsets,weusethevaluesofallpermutationsdirectlytoassesstheFDRlevelofEij.Specifically,toassesstheFDRforeachenrichmentvalueEij,weaccumulateallthepermutationvaluesforalllincRNAknockdownsandfunctionalsetsandcomputethenumberofvaluesgreaterthanEijaswellasavectorofvaluesgreaterthanEijcorrespondingtoeachpermutation.TheFDRiscomputedastheaveragenumberofpermutedvaluesgreaterthanEijdividedbytheobservednumbergreaterthanEij.Usingthisapproach,weassignanFDRvaluetoeachlincRNAknockdownbyfunctionalsetandidentifysignificanthitsasthosewithanFDR,0.01.
    Tohighlighttheaccuracyofthisapproach,weobservedthatforpubliclyavail-ablegeneperturbationsforwhichwealsoperturbedthegenewewereabletoidentifyasignificantassociationoftargetgenesin,75%ofcases.Althoughtheremainingfewdidnotpassourconservativesignificancecriteria,theyalsoshowedincreasedenrichmentsconsistentwiththeircommoneffects.Inaddition,theprojectedeffectsarehighlyreproducibleacrossdistinctexperimentsoriginatingfrommanygroupsandacrossmultipleexpressionplatforms.Highlightingthespecificityofthisapproach,wenotethattherearemanyprofilesforwhichnolincRNAhadasimilareffect.
    Analysisofgene-expressionoverlapsbetweenindependenthairpinknock-downs.TodeterminewhetherindependenthairpinstargetingthesamelincRNAgenesharecommongenetargets,wecomputedacontinuousenrichmentscoredescribedabove.Briefly,wecomputedat-statisticforbothhairpinsagainstthenegativecontrols.Wethentookthesecondbesthairpinandsortedthegenes.Wescoredthebesthairpinaffectedgenesbasedonthisrankedorder.WeassessedthesignificanceofthisenrichmentbypermutingthesamplesandcontrolsandassignedanFDRoftheoverlapoftheexpressioneffect(asdescribedabove.Discretegenesetanalysis.DiscretegenesetswereanalysedusingtheGeneSetEnrichmentAnalysiswithaslightmodificationtothescoringproceduretobemoreanalogoustoourcontinuousscoringprocedure(describedabove.Specifically,wecomputedtheaverageoftheexpressionchanges(definedbythet-statisticforallgeneswithinthediscretegenesetuponknockdown63.Significancewasassessedbypermutingthecontrolandsamplelabelsandre-computingtheaveragestatisticforeachpermutation.TheFDRwasassessedoffofthesevaluesasdescribedabove.
    Lineagemarkergeneanalysis.Wecuratedlineagemarkergenesetsfrompub-lishedworkandpubliclyavailablesources17,32,63.Weidentifiedlineagemarkergenesassignificantlyupregulatedusingthedifferentialexpressioncriteriaoutlinedabove.WevalidatedtheexpressionoftheselineagemarkergenesforaselectedsetoflineagemarkergenesusingqPCR(asdescribedaboveaftera4-dayinfection.Specifically,welookedattheexpressionofFgf5(ectoderm,Sox1(neuroecto-derm,Sox17(endoderm,brachyury(mesodermandCdx2(trophectoderm.PrimersequencesarelistedinSupplementaryTable9.ExpressionestimateswerenormalizedtoGapdhandcomparedtoapanelof25negativecontrolhairpins.IdentifyingboundlincRNApromoters.Weobtainedgenome-widetranscrip-tionfactorbindingdatainmouseEScellsfromtwosources.ThetranscriptionfactorsOct4,Sox2,NanogandTcf3weredownloadedfromtheGeneExpressionOmnibus(GSE11724andc-Myc,n-Myc,Zfx,Stat3,Smad1,Klf4andEsrrbfromGEO(GSE11431.ForeachChIP-Seqdataset,therawreadswereobtainedfromtheSRA(http://www.ncbi.nlm.nih.gov/sraandprocessedasfollows.(1Thereadswereallalignedtothemousegenomeassembly(buildMM9usingtheBowtiealigner66,requiringasinglebestplacementofeachread.Allreadswithmultipleacceptableplacementswereremovedfromtheanalysis.(2BindingsitesweredeterminedfromthealignedreadsusingtheMACS67(http://liulab.dfci.harvard.edu/MACS/algorithmusingthedefaultparameterswith–mfold8toaccountforvaryingreadcountsinthelibraries.(3lincRNApromoterregionsweredefinedaspreviouslydescribed2,3usingthelocationoftheK4me3peaksoverlap-pingorwithin5kbofthetranscriptionalstartsitedeterminedbyRNA-Seq
    ©2011MacmillanPublishersLimited.Allrightsreserved

    RESEARCHARTICLE
    reconstruction.(4ThetranscriptionfactorbindinglocationsandlincRNApro-moterlocationswereintersectedandtheenrichmentlevelofthepeakoverlappingalincRNApromoterwasassignedtranscriptionfactorbindingenrichmentforeachlincRNA.Wedefinedtranscriptionfactorbindinglocationsforprotein-codinggenesinacomparableway.(5Toexcludethepossibilitythatsomeofthisbindingmightbeduetotranscriptionfactorbindingatdistalenhancers,weexcludedallbindingeventsthatshowedevidenceofP300—aproteinassociatedwithactiveenhancers68—localization.Altogether,weonlyidentified,5%ofpromotersoverlappingwithanyP300enrichmentsignal,aslightlylowerpercent-agethanidentifiedforprotein-codinggenepromoterswithdetectableP300signal.Identifyingtranscription-factor-regulatedlincRNAgenes.lincRNAprobesontheAgilentmicroarraywereanalysedusingthedifferentialexpressionmethodologydescribedaboveafterknockdownofthetranscriptionfactorandcomparisontothenegativecontrolhairpins.ToconfirmtheexpressionchangesoftheselincRNAs,wehybridized12transcriptionfactorknockdownsonacustomlincRNAcodesetusingtheNanostringnCounterassay41(LIN-MES1-96.Theknockdownswereprofiledinbiologicalduplicatealongwith15negativecontrols.RegulatedlincRNAswereidentifiedusingthedifferentialexpressionapproachdescribedabove.
    Nanostringprobe-setdesign.NanostringprobesagainstlincRNAgenesweredesignedfollowingthestandardnanostringdesignprincipleswiththefollowingmodificationsspecificallyforthelincRNAprobes.(1Toexcludepossiblecross-hybridization,probeswerescreenedforcross-hybridizationagainstboththestandardmousetranscriptomeaswellasabackgrounddatabaseconstructedfromallthelincRNAsequences.(2Toaccountforisoformcoverage,afirstpassdesignattemptedtoselectaprobethatwouldtargetasmanyisoformsaspossibleforeachlincRNA.IncaseswhereitwasnotpossibletotargetallisoformsforagivenlincRNA,theprobethattargetedthelargestnumberwasselected,andadditionalprobeswerechosenwhenpossibletotargettheremainingisoforms.(3ThestandardrestrictionsonmeltingtemperatureandsequencecompositionwererelaxedtoincludeprobesforasmanylincRNAsaspossible.
    Retinoicaciddifferentiation.V6.5cellswereculturedongelatin-coateddishesinmESmediaintheabsenceofLIF.5mMofretinoicacidwasaddeddailyandcellsamplesweretakendailyfor6days.RNAwasextractedusingQiagen’sRNeasyspincolumnsfollowingthemanufacturer’sprotocol.
    Westernblots.30mgofmESCnuclearproteinextractswererunon10%Bis-Trisgels(InvitrogenNP0316BOXinMOPSbuffer(InvitrogenNP0001at75Vfor20minfollowedby120Vfor1h.Gelswereincubatedfor30minin20%methanoltransferbuffer(InvitrogenNP0006-1andtransferredontoPVDFmembranes(Invitrogen831605at20Vfor1husingtheBio-Radsemi-drytransfersystem(170-3940.MembraneswereblockedinBlotto(Pierce,37530atroomtemper-aturefor1h.AntibodiesweredilutedinBlottoandmembraneswereincubatedovernightat4uC.Antibodiesweredilutedinthefollowingconcentrations.Ezh21:2,000,Suz121:5,000,hnRNPH1:1,000,Ruvbl21:1,000,Jarid1b1:500,Hdac11:250,Cbx61:500,Yy11:500.Allantibodiestestedwereraisedinrabbit.Thenextday,membraneswerewashed33in0.1%TBSTfor5mineach.Themembraneswereprobedwithanti-rabbit-horseradishperoxidase(GEHealthcare;NA9340Vata1:10,000dilution,washed33in0.1%TBST,incubatedinECLreagent(GEHealthcareRPN2132andexposed.
    CrosslinkedRNAimmunoprecipitation.V6.5mEScellswerefixedwith1%formaldehydefor10minatroomtemperature,quenchedwith2.5Mglycine,washedwith13PBS(33harvestedbyscraping,pelleting,andre-suspendedinmodifiedRIPAlysisbuffer(150mMNaCl,50mMTris,0.5%sodiumdeoxycho-late,0.2%SDS,1%NP-40supplementedwithRNaseinhibitors(Ambion,AM2694andproteaseinhibitors.ForUVcrosslinkingexperiments,cellswereirradiatedwith254nmUVlight.Cellswerekeptoniceandcrosslinkedin13PBSusing400,000mjoulescm22.
    CellsuspensionwassonicatedusingaBranson250Sonifierfor3320scyclesat20%amplitude.10mlofTurboDNase(Ambion,AM2238wasaddedtosonicatedmaterial,incubatedat37uCfor10min,andspundownatmaxspeedfor10minat4uC.Protein-Gbeadswerewashedandpre-incubatedwithantibodiesfor30minatroomtemperature.Lysateandbeadswereincubatedat4uCfor2h.Beadswerewashed33usingthefollowingwashbuffer(13PBS,0.1%SDS,0.5%NP-40followedby23usingahighsaltwashbuffer(53PBS,0.1%SDS,0.5%NP-40andcrosslinkswerereversedandproteinsweredigestedwith5mlproteinase-K(NEB,P8102Sat65uCfor2–4h.RNAwaspurifiedusingphenol/chloroform/isoamylalcoholandRNAwasprecipitatedinisopropanol.
    Nanostringhybridization.500ngoftotalRNAwashybridizedfor17husingthelincRNAcodeset.ThehybridizedmaterialwasloadedintothenCounterprepstationfollowedbyquantificationonthenCounterDigitalAnalyserfollowingthemanufacturer’sprotocol.ForRNAimmunoprecipitationexperiments,weusedamodifiedprotocol.Afterreversecrosslinking,RNAwasextractedusingphenol/chloroformandethanolprecipitationmethodsandre-suspendedin10mlofH2O.5mloftheelutedmaterialwashybridizedfor17husingthelincRNAcodeset.
    Nanostringanalysis.Probevalueswerenormalizedtonegativecontrolprobesbydividingthevalueoftheprobebythemaximumnegativecontrolprobe.Probevalueswereflooredtoanormalizedvalueof3(threefoldhigherthanmaximumnegativecontrol.Probeswithnovaluegreaterthanthisflooracrossallsampleswereremovedfromtheanalysis.Thevalueswerelogtransformed.Tocontrolforvariabilitybetweenrunsanddifferentinputmaterialamounts,wenormalizedallsamplessimultaneouslyusingthequantilenormalizationapproachdescribedabove.Theresultisasetofnormalizedlog-expressionvaluesforeachprobenormalizedacrossallexperiments.
    ValidationofRNAimmunoprecipitationmethods.Tovalidateourformaldehyde-basedRNAimmunoprecipitationmethodweimmunoprecipitatedtheRNAbind-ingproteinhnRNPH,whichhasaroleinmRNAsplicing69andidentifiedtheassociatedRNAs.Consistentwithknowninteractions,weidentifiedastrongenrichmentforitsbindingtointronicregionsofmRNAgenes.WevalidatedtheseobservedresultsinmouseEScellsbyperformingUV-crosslinkingexperiments70–72andidentifiednearlyidenticalresults.WeidentifiedasimilarcorrelationbetweentheUVandformaldehydecrosslinkedsamplesasforbiologicalreplicatesofUVcrosslinkedsamplesandformaldehydecrosslinkedsamplesandhighlycomparableenrichments(datanotshown.
    Antibodyselection.Weselectedchromatinproteinsthathavebeenimplicatedinregulationofthepluripotentstatealongwiththeirknownassociated‘reader’,‘writer’and‘eraser’complexes.Specifically,wetestedantibodiesagainst40chro-matinproteins,correspondingto28chromatincomplexes.Inmanycases,wetestedmultipleantibodiesagainstthesametargetproteintotrytoidentifyanantibodythatworkedwellforimmunoprecipitation.Afulllistoftestedcom-plexesandtheirassociatedantibodiesislistedinSupplementaryTable18.
    Determiningsignificantchromatin–lincRNAenrichments.Wetestedeachantibodyusingformaldehydecrosslinkedcellsandhadatwo-stepprocedureforconsideringanantibodysuccessful.(1Wetestedallselectedantibodiesinbatches,witheachbatchcontainingamock-IgG(SantaCruznegativecontrolandhnRNPH(Bethylpositivecontrol.Batcheswithvariabilityineitherthemock-IgGorhnRNPHcontrolswereexcludedandretested.Foreachsuccessfulbatch,wecomputedenrichmentforeachlincRNAbetweenthetestedantibodyandmock-IgG.Weconsideredanantibodysuccessfulinthefirststepifthehighestenrichmentlevelexceededafivefoldchangecomparedtothemock-IgGcontrolandmorethan10lincRNAsexceededthisthreshold.Althoughthisapproachcanyieldfalsepositives(antibodiesthatpassbutarenotefficientitsignificantlyreducedthenumberofantibodiestobetestedinthenextstep.(2Forallantibodiesthatsuccessfullypassedthefirstcriterion,weperformedimmu-noprecipitationontwoadditionalbiologicalreplicatesalongwith4mock-IgGcontrols.Wecomputedat-statisticforeachlincRNAcomparedtothecontrolsandassessedthesignificanceusingapermutationtest,bypermutingthesamplesandIgGsamples(asabove.Hitswereconsideredsignificantiftheyexceedat-statisticcutoffof2(logscalecomparedtothecontrolsandhadanFDR,0.2.WeallowedaslightlyhigherFDRcutoffbecausethenumberofpermutationswasfarsmalleryieldinglowerpowertoestimatetheFDR.OnlyantibodiesyieldingsignificantlincRNAswereconsideredsuccessful.Intotal,weidentified12ofthe28complexes(55antibodieswithatleastonesuccessfulantibody.
    DeterminingsignificantoverlapsbetweenlincRNAandchromatinproteinknockdowneffects.TodeterminethefunctionaloverlapbetweenthelincRNAandthechromatincomplexesitphysicallyinteractswith,wecomparedtheeffectsongeneexpressionuponknockdownofthelincRNAandtheassociatedproteincomplex.Todothis,weusedthegeneexpressionprofilesdeterminedforeachlincRNAknockdownandknockdownsof9ofthe12identifiedchromatincom-plexesforwhichwehadgoodhairpins.WedefinedeachinteractionbetweenalincRNAandprotein,andcomputedacontinuousenrichmentscore,generatedallpermutationsofthecontrolhairpinsandsamplehairpinsandassignedanFDRtothescores(asdescribedabove.AtanFDR,0.05weidentified43%oftheinteractionstobesignificant.For69%oftheinteractions,wewereabletoidentifyanoverlapatanFDR,0.1.
    47.Meissner,A.,Eminli,S.&Jaenisch,R.Derivationandmanipulationofmurine
    embryonicstemcells.MethodsMol.Biol.482,3–19(2009.
    48.Nichols,J.etal.Formationofpluripotentstemcellsinthemammalianembryo
    dependsonthePOUtranscriptionfactorOct4.Cell95,379–391(1998.
    49.Avilion,A.A.etal.Multipotentcelllineagesinearlymousedevelopmentdependon
    SOX2function.GenesDev.17,126–140(2003.
    50.Niwa,H.,Burdon,T.,Chambers,I.&Smith,A.Self-renewalofpluripotent
    embryonicstemcellsismediatedviaactivationofSTAT3.GenesDev.12,2048–2060(1998.
    51.Nakatake,Y.etal.Klf4cooperateswithOct3/4andSox2toactivatetheLefty1core
    promoterinembryonicstemcells.Mol.Cell.Biol.26,7772–7782(2006.
    52.Brons,I.G.etal.Derivationofpluripotentepiblaststemcellsfrommammalian
    embryos.Nature448,191–195(2007.
    ©2011MacmillanPublishersLimited.Allrightsreserved

    ARTICLERESEARCH
    53.Torres-Padilla,M.E.,Parfitt,D.E.,Kouzarides,T.&Zernicka-Goetz,M.Histone
    argininemethylationregulatespluripotencyintheearlymouseembryo.Nature445,214–218(2007.
    54.Gaspar-Maia,A.etal.Chd1regulatesopenchromatinandpluripotencyof
    embryonicstemcells.Nature460,863–868(2009.
    55.Dejosez,M.etal.Roninisessentialforembryogenesisandthepluripotencyof
    mouseembryonicstemcells.Cell133,1162–1174(2008.
    56.Yuan,P.etal.EsetpartnerswithOct4torestrictextraembryonictrophoblast
    lineagepotentialinembryonicstemcells.GenesDev.23,2507–2520(2009.57.Pruitt,K.D.,Tatusova,T.,Klimke,W.&Maglott,D.R.NCBIReference
    Sequences:currentstatus,policyandnewinitiatives.NucleicAcidsRes.37,D32–D36(2009.
    58.Yang,Y.H.etal.NormalizationforcDNAmicroarraydata:arobustcomposite
    methodaddressingsingleandmultipleslidesystematicvariation.NucleicAcidsRes.30,e15(2002.
    59.Shen,X.etal.EZH1mediatesmethylationonhistoneH3lysine27and
    complementsEZH2inmaintainingstemcellidentityandexecutingpluripotency.Mol.Cell32,491–502(2008.
    60.Aiba,K.etal.Definingadevelopmentalpathtoneuralfatebyglobalexpression
    profilingofmouseembryonicstemcellsandadultneuralstem/progenitorcells.StemCells24,889–895(2006.
    61.Ying,Q.L.,Stavridis,M.,Griffiths,D.,Li,M.&Smith,A.Conversionofembryonic
    stemcellsintoneuroectodermalprecursorsinadherentmonoculture.NatureBiotechnol.21,183–186(2003.
    62.Morrisey,E.E.etal.GATA6regulatesHNF4andisrequiredfordifferentiationof
    visceralendoderminthemouseembryo.GenesDev.12,3579–3590(1998.63.Bock,C.etal.ReferencemapsofhumanESandiPScellvariationenablehigh-throughputcharacterizationofpluripotentcelllines.Cell144,439–452(2011.64.Barbie,D.A.etal.SystematicRNAinterferencerevealsthatoncogenicKRAS-drivencancersrequireTBK1.Nature462,108–112(2009.
    65.Lamb,J.etal.Theconnectivitymap:usinggene-expressionsignaturestoconnect
    smallmolecules,genes,anddisease.Science313,1929–1935(2006.
    66.Langmead,B.,Hansen,K.D.&Leek,J.T.Cloud-scaleRNA-sequencingdifferential
    expressionanalysiswithMyrna.GenomeBiol.11,R83(2010.
    67.Zhang,Y.etal.Model-basedanalysisofChIP-Seq(MACS.GenomeBiol.9,R137
    (2008.
    68.Visel,A.etal.ChIP-seqaccuratelypredictstissue-specificactivityofenhancers.
    Nature457,854–858(2009.
    69.Katz,Y.,Wang,E.T.,Airoldi,E.M.&Burge,C.B.AnalysisanddesignofRNA
    sequencingexperimentsforidentifyingisoformregulation.NatureMethods7,1009–1015(2010.
    70.Licatalosi,D.D.etal.HITS-CLIPyieldsgenome-wideinsightsintobrainalternative
    RNAprocessing.Nature456,464–469(2008.
    71.Ule,J.etal.CLIPidentifiesNova-regulatedRNAnetworksinthebrain.Science302,
    1212–1215(2003.
    72.Wang,Z.,Tollervey,J.,Briese,M.,Turner,D.&Ule,J.CLIP:constructionofcDNA
    librariesforhigh-throughputsequencingfromRNAscross-linkedtoproteinsinvivo.Methods48,287–293(2009.
    ©2011MacmillanPublishersLimited.Allrightsreserved

    免费下载 Word文档免费下载： lincRNAs act in the circuitry controlling stemness and differentiation

    TOP热门搜索

    相关推荐：

    公文写作材料

    • ￥29.8

      ￥45 每天只需1.0元
      1个月 推荐

    • ￥9.9

      ￥15
      1天

    • ￥59.8

      ￥90
      3个月

    选择支付方式

    • 微信付款
    会员须知 联系在线客服 登录账号
    郑重提醒：支付后，系统自动为您完成注册

    请使用微信扫码支付(元)

    订单号：
    支付后，系统自动为您完成注册
    遇到问题请联系 在线客服

    常用手机号：
    用于找回密码
    图片验证码：
    看不清？点击更换
    短信验证码：
    新密码：
     
    绑定后可用手机号登录
    请不要关闭本页面，支付完成后请点击【支付完成】按钮
    遇到问题请联系 在线客服

    侵权投诉  © 2013-2019 http://www.tcm999.cn   站点地图 |  手机版

    本站文档均来自互联网及网友上传分享，本站只负责收集和整理，有任何问题可通过上访投诉通道进行反馈