1.IntroductionandBackground

ThereisagreatneedforgeneralmethodstocharacterizetheproteinsthatcontemporaryBIOLOGymakesavailable.Thelistofsuchproteinsneedingfurthercharacterizationisgrowingandincludesproteinsalreadyknowntobeimportantforspecificcellularfunctions,mutantproteinsidentifiedinvivoormadeinvitro,andverylargenumbersofproteinbeingidentifiedbygenomeprojects.Herewedescribetheextensionoftwo-hybridapproachessothattheycanbearonthisproblem.

Therecentsuccessoftwo-hybridsystemsisduetothefactthatmanycellularfunctionsarecarriedoutbyproteinsthattouchoneanother.Forexample,thecomplexprocessoftranscriptioninitiationrequirestheorderedassemblyofnumerousinteractingtranscriptionfactorswithRNApolymeraseandancillaryproteins,intoaproteinmachinethatinitiatestranscription(Guarente,1996;TjianandManiatis,1994).Thismachinecanbeviewedasanetworkofinteractingproteins,ascanthemachinesthatcontrolotherprocesses,suchasDNAreplication,proteintranslation,andthecellcycle.Afullunderstandingoftheseprocesseswillrequireknowledgeof,notonlytheproteins(parts)thatmakeupeachmachine,butalsoofthetopologicalrelationships(connections)thatindividualpartsmakewithoneanother.

Likewise,afullunderstandingofthefunctionofanynewproteinwillrequireknowledgeoftheinteractionsitmakeswithpreviouslyidentifiedproteins.Currently,mostnewproteinsarebeingidentifiedbylargescalesequencingprojects.Formanyofthesenewproteinsthesequencealoneshedslittleornolightontheirfunction.

Two-hybridsystemshavebeenusedtoprobethefunctionofnewproteinseversincetheyweredeveloped(Chienetal.,1991;FieldsandSong,1989).Thefirstapplicationoftwo-hybridmethodstoprobeproteinfunctionwastoexaminetheinteractionsbetweenproteinsisolatedbytwohybridmethodsandrelativelysmallnumbersoftestproteins(seeforexample,Durfeeetal.,1993;Gyurisetal.,1993;Harperetal.,1993;Zervosetal.,1993),buttheirusequicklyspreadtotheanalysisofmanyotherproteins(Choietal.,1994;Kranzetal.,1994;Marcusetal.,1994;PrintenandSprague,1994;VanAelstetal.,1993;Yuanetal.,1993).Inanticipationoftheutilityofapplyingthesemethodstolargersets,weandothersbegandevisingwaystodoso.

Largerscaletwohybridapproachestypicallyrelyoninteractionmating.InthismethodtheproteinfusedtotheDNA-bindingdomain(thebait)andtheproteinfusedtotheactivationdomain(herecalledtheprey)areexpressedintwodifferenthaploidyeaststrainsofoppositematingtype(MATaandMATa),andthestrainsarematedtodetermineifthetwoproteinsinteract.Matingoccurswhenhaploidyeaststrainsofoppositematingtypecomeintocontact,andresultsinfusionofthetwohaploidstoformadiploidyeaststrain.Thus,aninteractioncanbedeterminedbymeasuringactivationofatwo-hybridreportergeneinthediploidstrain.

Asdescribedbelow,interactionmatinghasbeenusedtoexamineinteractionsbetweensmallsetsoftensofproteins(FinleyandBrent,1994;FinleyandBrent,1995;ReymondandBrent,1995),largersetsofhundredsofproteins(R.L.F.andR.B.,unpublished),toscreenlibraries(Bendixenetal.,1994),andtoattempttocomprehensivelymapconnectionsbetweenproteinsencodedbyasmallgenome(Barteletal.,1996).Theprimaryadvantageofthistechniqueisthatitreducesthenumberofyeasttransformationsneededtotestindividualinteractions.Forexample,totestforinteractionsbetweenasetof10baitproteinsand5preyproteinswithoutinteractionmatingwouldrequire50transformationstocreate50strainsthatcarrythepair-wisecombinationsofbaitsandpreys.Withmatinghowever,only15transformationswouldbeneeded;10forthedifferentbaitplasmids,and5forthedifferentpreyplasmids;andtheresultingtwosetsoftransformantswouldbematedtocreatethe50combinations.Themicrobiologyofthematingprocedure(whichisextremelysimple)isdetailedinSection2.

Interactionmatingtechniqueshavefacilitatedanumberoftwo-hybridstudiesofproteinproteininteraction.Amongitsfirstuseswastodeterminethespecificityofinteractorsisolatedinlibraryscreensorinteractorhunts(Harperetal.,1993).Asdescribedinthepreviouschapters,inthefirststepsofaninteractorhunt,oneisolatesgenesthatencodeproteinsthatinteractwithaparticularbait.Beforetheinteractingproteinsarefurthercharacterized,itisnecessarytodetermineiftheirinteractionwiththebaitisspecificbyshowingthattheydonotinteractwithotherunrelatedbaitsorwiththeDNA-bindingdomainportionofthebait.Whenmatingisusedtotestspecificity,thestrainthatcontainstheactivationdomainfusedprotein(prey)ismatedwithdifferentyeaststrainswhichexpresseithertheoriginalbaitproteinorother,preferablyunrelatedbaits,andtheinvestigatorverifiesthatthereportersareonlyactiveindiploidsthatcontaintheoriginalbait(FinleyandBrent,1994;FinleyandBrent,1995;Harperetal.,1993).

Forexample,Harper,Elledgeandcolleaguesusedamatingassaytotestthespecificityofnewlyisolatedinteractors(Harperetal.,1993).Themethodsoftheseinvestigatorsalsocircumventedtheneedtoisolatethepreyplasmid.Intheirexperiments,theyperformedtwo-hybridhuntswithabaitplasmidthatcontainsadominantMarker,CYH2,thatcanbeselectedagainstbyplatingtheyeastonmediumcontainingcycloheximide,whichistoxictoyeastthatcarryCYH2.Yeastisolatedinaninteractorhuntwereplatedoncycloheximideplatestoselectthosethathadlosttheoriginalbaitplasmidbutretainedthelibraryplasmid.Theresultingstrainwasthenmatedwithacollectionofbaitstrains,includingonesthatexpressedtheoriginalbait,todeterminethespecificityofthelibrary-encodedprey.AmatingschemehasalsobeenuseddirectlyinaninteractorhuntbymatingastrainexpressingabaitwithastraintransformedwiththelibraryDNA;here,matingpromisestobypasstheneedtoperformseparatetransformationswithlibraryDNAforeachnewhunt(Bendixenetal.,1994).

Inadditiontoitsuseininteractorhunts,matingcanbeusedtocharacterizesmallsetsofproteinsasdescribedinSection2.1andProtocol1.Inoneexampleofthisapproach,weusedinteractionmatingtocharacterizeasetofsevenDrosophilaCyclin-dependentkinases(Cdk)interactors,orCdis(FinleyandBrent,1994).StrainsexpressingversionsoftheCdisfusedtoanactivationdomainwerematedwith74differentstrainsexpressingdifferentbaitproteins,includingCdksfromotherspeciesandfouroftheCdisthemselves.Theresultsfromthisstudyillustratethetypesofinformationthatcanbederivedfromsuchacharacterization.First,theexperimentsshowedthatsomeoftheCdisinteractedwithdifferentsubgroupsofsevenhighlyrelatedCdkbaits,suggestingthattheCdisrecognizestructuralfeaturessharedbytheseCdksbutabsentinthenon-interactingCdks;inspectionofanalignmentoftheCdkproteinsequencessuggestedresiduesthatmaybeimportantforspecificinteractionswithcertainCdis.Second,Cdi3,DrosophilaCyclinD,interactedmuchmorestronglywithhumanCdk4thanwithanyoftheotherCdksinthepanelincludingtheDrosophilaCdks,suggestingthattheremaybeanasyetunidentifiedDrosophilaCdk4homologwhichisthetruepartnerforCyclinD.Third,twooftheCdisinteractedwithtwootherCdis,indicatingineachinstancethateachCdihassurfacesforbindingtotheCdkandtoanotherCdi,andsuggestingthattheseproteinsformternaryorhigherordercomplexes.Finally,thedemonstrationthattwoCdiswithnosequencesimilaritytopreviouslyidentifiedproteinsinteractwitheachotheraswellaswiththeCdk,butnotwithapanelofover60otherproteins,providedanadditionalcluetotheirfunctions,stronglysupportingtheideathattheyfunctionalongwiththeCdkinthenetworkofproteinsthatregulatesthecellcycle.Theseresultsdemonstratethatexaminationoftheinteractionsbetweenevensmallnumbersofproteinscanprovideanumberoffunctionalinsights.MuchlargersetsofproteinscanbecharacterizedbyscalinguptheseproceduresasdescribedinSection2.2anddiscussedinSections6and7.

2.Interactionmating

Inthissectionwepresentmethodsforperforminginteractionmatingassaysonsmallorlargesetsofproteinsusingtheinteractiontrap,andinSection3wediscussuseofinteractionmatingwithothertwo-hybridsystems.Theinteractiontrap(seeChapter4andreferencestherein)usestheE.coliproteinLexAastheDNA-bindingdomainandaproteinencodedbyrandomE.colisequences,theB42"acidblob",asthetranscriptionactivationdomain.Bothproteinsareexpressedfrommulticopy(2µ)plasmids;theLexAfusion,orbait,isexpressedfromaplasmidcontainingtheHIS3marker,andtheactivationdomainfusedprotein,orprey,isexpressedfromaplasmidcontainingtheTRP1marker.Inthemostcommonlyusedbaitplasmid,pEG202,thebaitisexpressedfromtheconstitutiveyeastADH1promoter.Relatedbaitplasmidsareavailablewhichexpressthebaitfusedtoanuclearlocalizationsignal(pNLex,seeChapter4),orwhichexpressthebaitconditionallyfromtheGAL1promoter(pGILDA,D.ShaywitzandC.Kaiser,personalcommunication).Themostcommonlyusedpreyplasmid,pJG4-5,expressesproteinsfusedtotheB42activationdomain,theSV40nuclearlocalizationsignal,andanepitopetagderivedfromhemagglutinin,alldrivenbytheyeastGAL1promoterwhichisactiveonlyinyeastgrownongalactose(Gyurisetal.,1993).UseoftheGAL1promotertoexpressthepreyallowstoxicproteinstobeexpressedtransientlyandhelpseliminatemanyfalsepositivesininteractorhunts(Chapter4).TheinteractiontrapusestworeportergenesthatcarryupstreamLexAbindingsites(operators):LEU2andlacZ.TheLEU2reportersareintegratedintotheyeastgenomeandthelacZreporterstypicallyresideon2µplasmidsbearingtheURA3marker,thoughintegratedversionsarealsoavailable(R.L.F.,R.B.,S.Hanes,unpublished).SeveralversionsoftheLEU2andlacZreportershavebeenmadethathavearangeofsensitivitiesbasedonthenumberofupstreamLexAoperators.IngeneraltheLEU2reportersaremoresensitivetoagiveninteractingpairofproteinsthanthelacZreporters(Estojaketal.,1995);however,recentlyhighlysensitivelacZreportershavebeenusedthatcontainseveralLexAoperatorsandtranscriptionterminatorsequencesdownstreamofthelacZgene(S.Hanes,personalcommunication).

Severaldifferentcombinationsofstrains,plasmids,andreporterscanbeusedformating(Section3).Inonecommonversion(FinleyandBrent,1994),thestrainexpressingthebait(baitstrain)isRFY206(MATaura3-52his3Æ200leu2-3lys2Æ201trp1::hisG)transformedwiththeHIS3baitplasmidandaURA3lacZreporterplasmidlikepSH18-34.Thestrainexpressingtheactivationdomain-taggedprotein(preystrain)isEGY48(MATaura3his3leu2::3LexAop-LEU2trp1LYS2)transformedwiththeTRP1preyplasmid.Patchesofthesetwostrainsonagarplatesarebroughtintocontactbyreplicaplating(seebelow)andgrownonarichmediumovernight.Duringthistimecellsinthepatchesmateandfusetoformdiploids.Thecellsarethentransferredbyreplicaplatingtoplatesonwhichonlydiploidscangrow:theseplateslackuracil,histidine,andtryptophansothatneitherparentalhaploidcangrowonthem.Toavoidanadditionalstep,thediploidselectionplatesarealsoindicatorplates,whichallowsaninteractiontobescoredbytestingforexpressionofthereportergenes.IntheprotocolspresentedherethelacZreporterismeasured,usingdiploidselectionindicatorplatescontainingX-Gal,achromogenicsubstrateforthelacZgeneproduct.However,itisworthmentioningthatexpressionoftheLEU2reportercanalsobeeasilyscoredbyputtingthediploidsonplatesthatlackleucine,andthatthefuturewilllikelybringotherreporters.FurThermore,becausebothreportergenesexhibitareducedsensitivityindiploidstrainscomparedtohaploidstrains,themostsensitiveversionsofthelacZorLEU2reportersarerecommendedforinteractionmatingassays.

Variantsofthissimpleprocedurearesometimesuseful.Inparticular,becausesomebaitsactivatetranscriptionbythemselves,itisoftenusefultoconditionallyexpressthepreyproteinsothatonescorespatchesthatshowanincreaseinreportergeneexpressioninthepresenceoftheprey.Todothis,thediploidsareplacedontwodifferentX-Galplates,onethatcontainsgalactose,whichresultsinexpressionoftheprey,andonethatcontainsglucosewhichrepressesexpressionoftheprey.Here,aninteractionbetweenthebaitandpreyisdetectedwhenthediploidyeastcontainingthemturnmoreblueonthegalactoseX-GalplatethanontheglucoseX-Galplate.

2.1Interactionmating-smallscale

Itisofteninformativetolookforinteractionsbetweensmallsetsofproteins,orbetweenagivenproteinandatestsetoftentoahundredproteins.Thetestset,forexample,mightcontaindifferentallelicformsoftheoriginalbait,setsofstructurallyrelatedproteins,setsofproteinsknownorsUSPectedtobeinvolvedinsomeprocess,andunrelatedproteinsusedtodemonstratethespecificityofaninteraction.Protocol1describesaconvenientmethodtotestsmallsetsofproteinsforinteractions.

Thecollectionsofbaitandpreystrainsusedherecanbemaintainedonyeastplatesstoredat4oCfortwotothreemonths,orstoredfrozenforseveralyears(seeProtocol2).Formating,thetwostrainsarefirststreakedtotheappropriateselectionplates:thebaitstrains(RFY206containingtheURA3lacZreporterplasmidandHIS3baitplasmid)arestreakedtoplateslackinguracilandhistidine-u-hGlu)tomaintainselectionforthetwoplasmids;thepreystrains(EGY48containingtheTRP1preyplasmid)arestreakedtoplateslackingtryptophan(-wGlu)tomaintainselectionforthepreyplasmid.ThehaploidstrainsarethenbroughtintocontactbyplacingbothplatessequentiallyonthesamereplicavelvetandliftingthedoubleimprintwithaYPDplate(seeProtocol1).Ifthebaitstrainsarestreakedinparallelhorizontalstripesandthepreystrainsarestreakedinverticalstripes,physicalcontactbetweenthestrainswilloccurattheintersectionsofthestripesontheYPDplate.Afterabriefperiodofgrowthtoallowdiploidstoform,theyeastaretransferredtodiploidselectionindicatorplatesbyreplicaplating.DiploidcoloniesthatcontainapairofinteractingbaitandpreyproteinsaremoreblueonthegalactoseX-GalplatethantheglucoseX-Galplate.

________________________________________________________________________

Protocol1.Matingassay-smallscalefortensofdifferentbaitorpreyproteins.

Materials

_Baitstrains:S.cerevisiaestrainRFY206(MATaura3-52his3Æ200leu2-3lys2Æ201trp1::hisG)transformedwithaURA3plasmidcontainingalacZreporter,suchaspSH18-34,andvariousHIS3baitplasmids,suchasderivativesofpEG202thatproducedifferentLexAfusions.Eachbaitstrainwillcontainadifferentbaitplasmid.

Preystrains:S.cerevisiaestrainEGY48(MATaura3his3leu2::3LexAop-LEU2trp1LYS2)transformedwithTRP1preyplasmids,suchasderivativesofpJG4-5thatproducedifferentactivationdomain-taggedproteinsorpreys

Sterilewoodenapplicatorsticks(e.g.FisherBrand01-340)

Minimalglucoseyeastplateslackinguracilandhistidine(-u-hGlu),seeChapter4

Minimalglucoseplateslackingtryptophan(-wGlu),seeChapter4

YPDplates,seeChapter4

MinimalX-Galglucoseplateslackinguracil,histidine,andtryptophan(-u-h-wGluX-Gal),seeChapter4.

MinimalX-Galgalactose/raffinoseplateslackinguracil,histidine,andtryptophan(-u-h-wGal/RafX-Gal),seeChapter4

Replicaplaterandsterilereplicavelvets

Optional

Minimalglucoseplateslackinguracil,histidine,tryptophan,andleucine(-u-h-w-lGlu),seeChapter4

Minimalgalactose/raffinoseplateslackinguracil,histidine,tryptophan,andleucine(-u-h-w-Gal/Raf),seeChapter4

Method

1.Streakdifferentbaitstrainsinhorizontalparallelstripesona-u-hGluplate.Streaksshouldbeatleast3mmwideandatleast5mmapart,withthefirststreakstartingabout15mmfromtheedgeoftheplate.A100mmplate(whichforsomereasonistypically90mmindiameter)willhold8differentbaitstrains.Createaduplicateplateofbaitstrainsforeachdifferentplateofpreystrainstobeused.

2.Likewise,streakdifferentpreystrainsinverticalparallelstripesona-wGluplate.Asacontrolforbaitsthatmayactivatetranscription,includeapreystrainthatcontainsthepreyvectorpJG4-5notencodingafusionprotein(i.e.encodingonlytheactivationdomain).Createaduplicateplateofpreystrainsforeachplateofbaitstrainstobeused.

3.Incubateplatesat30oCuntilthereisheavygrowthonthestreaks.Whentakenfromreasonablyfreshcultures,forexampleplatesthathavebeenstoredat4oCforlessthanamonth,streakedRFY206-derivedbaitstrainstakeabout48hourstogrowandEGY48-derivedpreystrainstakeabout24hours.

4.Pressaplateofpreystrainstoareplicavelvet,evenlyandfirmlysothatyeastfromallalongeachstreakareleftonthevelvet.Thisplatemaybereusedifnecessary.Pressaplateofbaitstrainstothesamereplicavelvet.Thisplateofbaitstrainscannotbereusedasitisnowcontaminatedwithpreystrains.

5.LifttheimpressionofthebaitandpreystrainsfromthevelvetbypressingaYPDplateonit.IncubatetheYPDplatefor24hoursat30oC.

6.ReplicaYPDplatestothefollowingdiploidselection,indicatorplates:-u-h-wGluX-Gal,-u-h-wGal/Raf,and(optional:-u-h-w-lGlu,and-u-h-w-lGal/Raf).TheYPDplateshouldcontainsufficientgrowthtoenableasingleimpressiononthevelvettobeliftedbyatleastfourindicatorplates.

7.Patchcontrolstrains(seetext)ontotheindicatorplatesandincubateat30oC.Examineresultsdaily.Diploidswillgrowandbluecolorwilldevelopwithin2days.

________________________________________________________________________

2.2Interactionmating-largescale

Withafewmodifications,theproceduredescribedabovecanbeusedtotestforinteractionsbetweenasinglepreyproteinandhundredsofbaits(Protocol3,seeFigure1below).Largepanelsofbaitstrainscanbecollectedandstoredfrozenindefinitely(Protocol2)andthenscreenedagainstanynumberofpreys.Onesuchsetofbaitstrainscontainsover700differentLexAfusionproteinsfromourownworkandfromnumerousotherlabsthatusetheinteractiontrap(R.L.F.,R.B.,A.Reymond,unpublished).ScreeningaproteinagainstsuchapanelenablesonetoquicklytestitsABIlitytointeractwithalargenumberofknownproteins,mostofwhichhavebeencharacterizedtosomeextent,andhavebeenchosenforstudybecauseoftheirknownorsuspectedinvolvementinsomebiologicalprocess.Thus,thefindingofaninteractionbetweenatestedproteinandamemberofthepanelcanoftenleadtoimmediatecluesaboutthebiologicalfunctionofbothproteins(seeSection5).Whilethenumberofproteinsintheexistingpanelisfarlessthanthenumberofproteinsinagoodlibrary,thisapproachdoesoffertheadvantageofscreeningthetestproteinagainstasetofproteinsenrichedforthoseofcurrentinteresttothebiologicalcommunity.Itisworthnotingthattheseproteinscomefrommanydifferentorganismsinwhichtheyareexpressedindifferenttissuesandatdifferentdevelopmentalstages.ThusitbecomespossIBLetoidentifyinteractingpartnersthathavenotyetbeenisolatedfromthesamespecies,orthatarenotexpressedintissuesfromwhichinteractionlibrarieshavebeenmade.

Forsomeproteins,thisapproachoffersadditionaladvantagesoverscreeningalibraryusingatrADItionaltwo-hybridscheme.ProteinsthatactivatetranscriptionwhenfusedtoLexAoranotherDNA-bindingdomaincanbedifficulttouseinconventionalinteractorhunts.Thoughmethodsareavailabletoreducethesensitivityofthereportergenes(Durfeeetal.,1993;Estojaketal.,1995;Chapter2,3,4)itisnotalwayspossibletoreducethereportersensitivitybelowthethresholdofactivationforsomebaits.Moreover,reductioninreportersensitivitycarrieswithittheriskthatthereporterswillnotdetectweaklyinteractingproteins.Furthermore,spontaneouslyoccurringyeastmutations,forexamplethosethatincreasethecopynumberofthebaitplasmid,canincreasetheactivatingpotentialofweaklyactivatingbaits(R.L.F.,R.B.,A.Mendelsohn,unpublisheddata);suchmutationsaretypicallyscoredaspositiveintheearlystagesofaninteractorhunt,andtheyarenotreadilydetectedinschemeswherethespecificitytestisperformedbyremovingthebaitplasmidfromthestraincontainingthepreyandmatingthestrainwithotherbaitstrains.Thus,analternativeforproteinsthatactivatetranscriptionasbaits,istousethemaspreystoscreenexistingpanelsofbaits,orevenlibrariesofbaits.Interactionmatingapproachesalsohaveclearadvantagesforproteinsthataresomewhattoxictoyeast;thepreyvectorallowsconditionalexpressionoftoxicproteinsinthepresenceofabait,andoftentheinteractioncanbeobservedasthereportersareactivatedevenifthecellsareinviable.Anexampleoftheuseofinteractionmatingtogetherwithalargepanelofbaitstrainstocharacterizeaproteinthatbothactivatestranscriptionandistoxictoyeast,DrosophilaCyclinE(Finley,Zavitz,Thomas,Richardson,Zipursky,andBrent,inprep),isdiscussedinSection7.

Figure1.Matingassayforinteractionsbetweenapreyand96baits

Figure1.

Top.Theplateontheleftholds96differentyeaststrainsinpatches(orcolonies)thateachexpressadifferentbaitprotein.Theplateontherightholds96patches,eachofthesameyeaststrain(preystrain)thatexpressesaproteinfusedtoanactivationdomain(prey).TheplateofbaitstrainsandtheplateofpreystrainsareeachpressedtothesamereplicavelvetandtheimpressionisliftedwithaplatecontainingYPDmedium.AfteronedayofgrowthontheYPDplate,duringwhichtimethetwostrainsmatetoformdiploids,theYPDplateispressedtoanewreplicavelvetandtheimpressionisliftedwithaplatecontainingdiploidselectionmediumandanindicatorlikeX-Gal.Bluepatches(darkspots)ontheX-GalplateindicatethatthelacZreporteristranscribed,suggestingthatthepreyinteractswiththebaitatthatlocation.

________________________________________________________________________

Protocol2.Collectingbait(andprey)strains

Materials

Freezingmedia:1:1solutionofminimalglucosemedialackingappropriateaminoacids(e.g.-u-hGluforbaitstrains):sterileglycerolsolution(65%(v/v)glycerol,0.1MMgSO4,25mMTris-HClpH7.4)

1.0to1.5mlcryotubes

Yeaststrainsfreshlystreakedtominimalglucoseplates

Sterilewoodenapplicatorstrips

Methods

1.Streakbaitstrainsto-u-hGluplates,orpreystrainsto-wGluplates,andincubateat30oCfor24to48hours.Yeastshouldbetakenfromtheplatesandfrozennomorethan4daysafterbeingstreaked.

2.Withasterilewoodenapplicatorstick,grabadollopofyeastfromtheplatesandinoculate0.5mloffreezingsolutioninacryotube.Vortexlightly.ThissolutionshouldhaveanOD600over3.0.

3.Alternatively,inoculate0.5mlof-u-hGluliquidmediatoanOD600lessthan0.2,incubateat30oCwithshakinguntilOD600=1.5to2.0(logphase),andadd0.25mlofthiscultureto0.25mlofsterileglycerolsolutioninacryotube.

4.Freezebyplacingcryotubesin-80oCfreezer.Moststrainscanberecoveredafteruptoatleasttwoyearsbyscrapingthesurfaceoftheiceandstreakingtominimalglucoseplates.Avoidallowingentirecontentsofcryotubetothaw.

________________________________________________________________________

________________________________________________________________________

Protocol3.Matingassay-largescaleforhundredsofdifferentbaitorpreystrains.

Materials

Freshlystreakedbaitandpreystrains(seeProtocol1)

Onesetofthefollowing150x15mmplatesforeachtestofinteractionsbetweenanactivationdomain-taggedprotein(inapreystrain)and96baits(baitstrains):-u-h-Glu;-wGlu;YPD;-u-h-wGluX-Gal;-u-h-wGal/RafX-Gal

Replicaplaterandsterilevelvetsfor150mmdiameterplates.(Areplicadevisecanbefashionedfromaboxof200µlpipettipsbystretchingavelvetoverthetopofthebox)

96-prongdevice(e.g.DanKarMC-96)with3mmdiameterflatendedmetalprongsina96-wellconfiguration.Similardevicescanbeusedin48-wellconfigurationsforusewith100mmplates.

0.5to4.0mlsterilizedtubesarrangedina96-wellconfigurations(e.g.clustertubessuchasCostar#4411).Ideallythesetubecanbecappedandfrozenat-80oC.

-u-hGluliquidmedia,seeChapter4

-wGluliquidmedia,seeChapter4

Sterileglycerolsolution(65%(v/v)glycerol,0.1MMgSO4,25mMTris-HClpH7.4)

Methods

1.Itismostconvenienttoplacelargenumbersofbaitstrainsina96-wellconfiguration(Figure1).Thiscanbedonebyinoculating2mlof-u-hGlumediainclustertubesandgrowingtoOD600=1.5to2.0.Aftermakingplatesfromthesecultures(seestep2below)addanequalvolumeofsterileglycerolsolution,capandfreezeat-80oC.

2.Usethe96-prongdevice,sterilizedinethanolandflame,totransferbaitstrainsfromtheculturetothecenterofa150mm-u-hGluplate.Eachplatecancontain96differentbaitstrains.Tensofidenticalplatescanbemadefromoneculture.Incubatetheplatesat30oCfor48hoursoruntilallbaitstrainshavegrowntocolonies5mmindiameter.Theseplatescanbestoredat4oCforupto2monthsandusedtoinoculateanotherliquidculturewhenmoreplatesareneeded.Severalpositionsoneachplateshouldcontaincontrolstrainswithbaitsthatactivatevariouslevelsoftranscription(seeSection4andTable1).

3.Inoculate50mlof-wGluliquidmediawithapreystrainandgrowat30oCwithshakingtoOD600=1.5to3.0.Pourthecultureintoasterile150mmplate,orintothesteriletopfromaboxof200µlpipets,andusethe96-prongdevice,sterilizedinethanolandflame,totransferthecultureto-wGluplates.Ontheseplates,all96positionswillcontainthesamepreystrain.

4.FollowthereplicaplatingprocedurefromProtocol1tocombinethebaitandpreystrainstoaYPDplate,andthenaftergrowthontheYPDplateat30oCfor24hours,replicatoX-Galindicator,diploidselectionplates(-u-h-wGluX-Galand-u-h-wGal/RafX-Gal)(seeFigure1above).

5.Examineresultsaftertwodays.

________________________________________________________________________

3.Interactionmatingassaywithotheryeasttwo-hybridsystems

Inadditiontotheinteractiontrap,manyotheryeasttwo-hybridsystemshavebeendeveloped(seeChapter1andAllenetal.,1995;FieldsandSternglanz,1994;MendelsohnandBrent,1994,forreviews).Alloftheseallowtheanalysisofindividualprotein-proteininteractions,andpermitinteractorhuntstoisolatenewproteinsthatinteractwithabait.Insomeinstancesplasmidsorstrainsfromonesystemcanbeusedinanother,butoftenthecomponentsareincompatible.Mostoften,theyeastselectablemarkersonthedifferentcomponentsdiffer.Inaddition,systemsthatuseGal4astheDNAbindingdomaincannotbeusedwithyeaststrainsthathaveawild-typeGAL4gene,andtherefore,sincetheGal4proteinisrequiredtoactivatetheGAL1promoter,cannotbeusedwithsystemsthatusetheGAL1promotertodriveexpressionofthepreyprotein.Finally,useofinteractionmatingrequirescarefulattentiontothematingtypesofthestrainsandtheselectablemarkersusedtoselectthediploids.

4.Recordingtheresults

Interactionbetweenbaitandpreyresultsintheinteractionphenotypes:growthofthestrainonmediumlackingleucine,andtranscriptionalactivationofthelacZreporterandproductionofactiveß-galactosidase.OnX-Galplatestheß-galactosidasecleavestheX-Galsubstrate,producingaproductwhichturnstheyeastcolonyblue.TheamountofcolorprovidesafastandsimplemethodtoapproximatetheleveloflacZexpressioninastrain.AninteractionisscoredwhenathediploidcolonyismoreblueontheX-GalplatecontaininggalactosethantheX-Galplatecontainingglucose.

Scoringtheseinteractionsbenefitsfrominclusionofanumberofcontrols.TocontrolforcommonvariationsbetweentheX-Galplates,itisusefultoincludecontrolstrainsthatcontainbaitswhichactivatetranscriptiontovaryingextents.Table1showssomebaitswithknownactivatingabilities.InclusionofsuchstrainsoneveryX-Galplateenablesonetonormalizetheamountofblueproducedbyaninteraction.Itisalsousefultoincludeacontrolstraintocheckthattheplatescontainthecorrectcarbonsources,andensurethattheGAL1promoterwhichdrivestheexpressionofthepreyproteinisactivatedontheGal/RafplatesandnottheGluplates.Anidealcontrolofthisnatureconsistsofadiploidstrainderivedfromamatingassay,whichexpressesaninteractingpairofbaitandpreyproteins,suchasanyoneofanumberofwell-characterizedinteractingpairs(FinleyandBrent,1994;Gyurisetal.,1993;Zervosetal.,1993).AnalternativetousingX-Galplatesistoperformafilterliftassayforß-galactosidaseactivityingrowndiploidcolonies(Chapter).Finally,everybaitshouldbetestedtoseeif,andhowmuch,itactivatestranscriptionintheabsenceofaprey,whichcanbesimplyaccomplishedbymatingthebaitstrainstoastraincontainingtheemptypreyvector.Thus,atrueinteractionwithapreyproteinisscoredwhentheamountofgalactose-dependentactivationofthelacZreporter(e.g.amountofblue)exceedstheamountproducedintheabsenceofaprey.

Table1.Activatingandnon-activatingbaits

5.Interpretinginteractiondata

5.1Qualitativeinterpretation

Forlargeamountsofinformationflowingfrominteractionmatingexperiments,theproblemofdeterminingwhetherindividualinteractionsaremeaningfulismultiplied.Weconsideranumberoftheseseparately.

Trueandfalsepositives.Anygiveninteractionwithaffinitytighterthan10-6willgetdetected.Althoughtheremayexistaweakpositivecorrelationbetweenapparenttightnessandbiologicalsignificance,manyapparentlyweakinteractionsarerealwhilesomestrongonesarenot.Theproblemofdeterminingwhichinteractionshavebiologicalsignificanceisthereforenottrivial.Atthemoment,themostsatisfyingwaytoshowbiologicalsignificanceistoverifytheinteractionbyadifferent,biochemicaltechnique,preferablyco-precipitationfromacellinwhichbothproteinsareexpressed.However,theinteractiondataalonecanoftenpointoutprobabletrueandfalsepositives.Forexample,ourexperienceindicatesthathighlyspecificinteractions,suchasbetweenaproteinthatbindstooneorasmallsetofhighlyrelatedproteinsandnottohundredsofunrelatedproteins,aregoodcandidatestopursueasbiologicallyrelevant.Conversely,wetendtogivelessweighttointeractionsbetweenproteinsthataresticky,orinvolvingthoseproteinssoubiquitousinthelifeofthecell(e.g.,membersoftheubiquitinsystemorheatshockproteins)thattheinteractionsmightbemeaningfulbutrelativelyuninformative.

Trueandfalsenegatives.Aproblemlessfrequentlyconsideredisthatofinteractionsthatarenotobserved.Twoobservationssuggestthatmanyinteractionsthatshouldbeobservedarenot.Oneisthatinlibraryscreensproteinsthatshouldbefoundoccasionallyarenot.Althoughfailuretorecoverexpectedproteinsinthisinstancemightbeduetotrivialconsiderations,suchastheabsenceoftheproteinfromthelibraryused,anotherfactsuggeststherecouldbeotherreasons.Therearenowanumberofexamplesinwhichknowninteractionsareeithernotobserved,oraresubjecttodirectionality,beingobservedonlywhenoneofthetwoproteinsisabaitandtheotheraprey(seeforexample,Estojaketal.,1995).Ourcurrentdoctrinefordeterminingthatindividualinteractionsdonotoccuristhatfulllengthandtruncatedputativepartnersmustbetestedinallcombinationsofbaitsandpreys,withthemostsensitivereporters,beforetheinvestigatorcantentativelyconcludethatthetwoproteinsdonottouch.Sincethisisimpracticalformatingexperimentsthatinvolvealargenumberofbaitsandpreys,suchasthegenomewideapproachesdiscussedbelow,weareresignedthatfalsenegativeswillarise,andwedonotgivetheabsenceofinteractionanyweightinourdataanalysis.Thisdoctrinemaychangeasmoresensitivedetectionmethodsaredesigned.

Multimericcomplexes.Finally,itisworthnotingthatonecanbuildupchainsofindividualbinaryinteractionstosuggesthigherordercomplexes.Thishasworkedwell,forexamplewithproteinsinsignaltransduction(Choietal.,1994;Marcusetal.,1994;PrintenandSprague,1994),andtheadventofmatingtechniqueshasmadeiteveneasiertobuildupsuchpatterns(FinleyandBrent,1994;C.KaiserandD.Shaywitz,personalcommunication).

5.2Quantitativeinterpretation

Notwo-hybridtechnique-particularlythematingtechniquesdescribedinProtocols1and3-allowsprecisequantitation,andanyinteractionsidentifiedmustbestudiedfurthertodeterminebiologicalsignificanceandbiophysicalcharacteristics.However,somequantitativeinformationdoesinhereinthedata.Theamountofß-galactosidaseactivityinthecellisproportionaltotheleveloflacZtranscriptionsothatsomeinformationaboutthestrengthofinteractionoftwoproteinsmightbederivedfrommeasuringß-galactosidaseactivity.Thoughmeasurementofß-galactosidaseactivitywithaliquidassay(GuarenteandPtashne,1981;RoseandBotstein,1983)isnotpracticalforlargenumbersofstrains,alesspreciseindicationofenzymeactivitycanbederivedfromthecoloroftheyeastcolonyonanX-Galindicatorplate;forexample,darkblue,lightblue,orwhitecoloniescorrespondtohigh,moderate,orlowtonolacZtranscription.Despitethiscorrelationbetweentranscriptionlevelsandß-galactosidaseactivity,onemustusecautioninusingß-galactosidaseactivitytocomparerelativeaffinitiesofdifferentbaitandpreypairs.Manyvariablescouldaffecttheinteractionphenotypes,includingthestabilityofthetwofusionproteins,transportofthefusionsintothenucleus,andtheabilityofthebaittobindDNA.Theseconsiderationsmakeitimprudenttousetwo-hybriddatatocompareaffinitiesbetweensetsofunrelatedproteins.

Itis,however,oftenpossibletomakemeaningfulcomparisonsoftheaffinityofasinglepreyproteinforseveralrelatedbaits.Suchacomparisonreliesontwoassumptionsthataregenerallycorrectandcanbeexperimentallyverified:thattheprey,whichcanbedetectedwithantibodiestoitsepitopetag,isexpressedatthesamelevelineachdiploid,andthatthebaits,whichcanbedetectedwithanti-LexAantibodyandwhoseDNAbindingcanbequantitatedbyarepressionassay(BrentandPtashne,1984),occupytheoperatorstosimilarextents.

5.3Inferenceoffunctionfrompatternofinteractions

Onereasonfordevelopinginteractionmatingtechniqueswasthehopethatitwouldrevealcontactsbetweentestproteinsandknownproteinsthatwouldprovidecluestothefunctionofthetestproteins.Thisturnedoutbetrue(seeforexample,Section7).However,ourfirstexperimentsrevealedthatcluestofunctionmightalsobederivedfromthepatternofinteractionsaproteinmakes,withoutreferencetothebiochemicalidentityoftheinteractingproteins.Asimpleexample,takenfromourfirstexperiments,illustratesthispoint.Cdi4andCdi11bothinteractwithDrosophilaCdc2candinteractionmatingexperimentsalsorevealedthatCdi4interactswithCdi11(FinleyandBrent,1994).Fromthepatternofinteractionsalone,thesedataareconsistentwiththeideathatCdi4,Cdi11andCdc2ccouldformathreeproteincomplex.Itispossiblethatothersuchpatternsofinteractions,particularlyconjoinedwiththecrudeaffinitydata,mightsignalothersortsofregulators.Thealgorithmicanalysisofconnectivitydataforpatternsofthistypeisanimportantareaoffutureresearch.

6.Libraryscaleandgenome-widecharacterizationofproteinnetworks

Interactionmatingschemescanalsobeusedonalargerscale,forscreeninglibraries,and,eventuallytocharacterizecomplexgenomes.Onesuchschemeistomateapoolofcellscontainingdifferentactivationdomain-taggedproteinsagainstabaitprotein.Anotheristheconverseoftheoriginaltwo-hybridsystem.Inthisapproach,alibraryofdifferentproteinsfusedtoaDNA-bindingdomainisusedinaninteractorhunttofindproteinsthatinteractwithaspecificactivation-taggedprotein.Historically,thedrawbacktosuchapproacheshasbeenthatlibrariesthatexpressproteinsfusedtoDNA-bindingdomainswillcontainalargenumberproteinsthatactivatetranscriptionwhenbroughttoDNA(MaandPtashne,1987),complicatingthetaskofidentifyingyeastinwhichthereportersareactiveduetothepresenceofaninteractingprotein.Onewaytocircumventthisdifficultywouldbetointroducethelibraryintoayeaststrainthatcontainedacounter-selectablereportergene(e.g.LexAop-LYS2andLexAop-URA3),selectagainstthoseyeastthatcontainedactivators,andthenmatethe"depleted"librarywithyeastoftheoppositematingtypethatcontainthetestprotein.Yetanotherwayistoexpresstheactivationdomain-taggedproteinsfromaconditionalpromoterlikeGAL1andcomparereporteractivationbetweenreplicaplatesonwhichtheyareandarenotexpressed,asdescriedinProtocol1and3,andinChapter4).

Recently,Barteletalappliedtwo-hybridtechnologytocharacterizeasmallgenome(Barteletal.,1996).TheysetouttoidentifyalldetectablebinaryinteractionsbetweenproteinsencodedbythebacteriophageT7genome.Theydidthisbymakingtwolibraries,oneofDNA-bindingdomainhybridsandoneofactivationdomainhybrids,expressedinyeaststrainsofoppositematingtype.Theythenmatedapoolofyeastthatcontainedtheentirelibraryofactivationdomainhybridswith30,000ofthestrainsexpressingDNA-bindingdomainfusions,ingroupsoftensotheycouldreadilysingleoutthosethatactivatedtranscription.TheyselecteddiploidsinwhichtheHIS3reporterwasactivatedandscreenedforactivationofasecondlacZreporterusingafilterassay.Inthiswaytheyidentified19binaryinteractionsbetweenT7encodedproteins.Theyfurtherperformedindividualinteractorhuntstesting34specificDNA-bindinghybridsagainsttheentireactivationdomainlibrary,and11specificactivationdomainhybridsagainsttheentireDNA-bindingdomainhybridlibrary,againbyinteractionmating,andidentified3additionalinteractions.Finally,theymadeamatrixofalloftheyeastexpressingDNA-bindingdomainhybridsinvolvedinaninteractionmatedwithyeastexpressingalloftheactivationdomainhybridsinvolvedinaninteractiontoidentifythreemoreinteractions.

Bythismeanstheydetectedatotalof25interactions.Someoftheinteractionswerepreviouslyknown,whileothersconfirmedinteractionsthathadbeensuspectedbasedongeneticorbiochemicalstudies.Mostimportantly,10oftheinteractionsdetectedinthistwo-hybridtourdeforceidentifiedconnectionsbetweenproteinsnotpreviouslyknowntointeract.Thisnewinformationcontainsbothcluestothefunctionofindividualproteinsandcluesastohowsomemayfunctiontogether.Anadditionalwindfallfromthisapproach,madepossiblebythefactthatthetwolibrariesweremadefromrandomfragmentsoftheT7genome,wastheidentificationofanumberofpreviouslyunsuspectedintramolecularinteractions.Thedetectionoftheseintramolecularinteractionssuggestedpossiblehomo-oligomericproteincontactsaswellasinterdomaincontactsthatmightpromotetheformationoftertiarystructure.Thesuccessofthisgenome-wideapproachdemonstratesthatinteractionmatingtechniquescanbeusedtoidentifythenetworksofinteractingproteinsencodedbymorecomplexgenomes.Thechartingofsuchconnectionsbetweenproteinswillprovideinsightsintothefunctionsofindividualproteinsandleadtoabetterunderstandingofhowgroupsofproteinscontrolbiologicalprocesses.

7.Conclusions

Thefewyearssincetheadventoftwo-hybridsystemshasproventheirutilityinthestudyofdefinedproteininteractions,inidentificationofnewinteractingproteins,andinthechartingofgeneticnetworksofproteinsinvolvedinprocessesfromsignaltransductiontotranscriptionregulation.Thesetremendoussuccessessuggestthattwo-hybridapproacheslikethosediscussedinthischaptermayeventuallybeusedtoidentifyalloftheproteinproteincontactsmadeinacelloranorganism.

Beforethistime,anotherneedisclear.Sequencingprojectslikethehumangenomeinitiativewillsoonprovideuswiththesequencesofalloftheexpressedproteins.Agooddealofinsightintothefunctionoftheseproteinscanbederivedfromtheirsequencesalone,butultimatelymustbecombinedwithotherformsofinformationtounderstandthebiologyindetail.Informationaboutcontactsmadebytheproteinsofagenomewillcomplementandaugmentthesequenceinformation.Suchinformationwilllikelycomefromincrementalscalingupofthemethodsdescribedhere,aswellasfromscaledupversionsofideassuchasthosedevelopedbyBarteletal(Barteletal.,1996).Connectiondatawillalsocomefromthethousandsoflabsusingtwo-hybridsystemstoidentifyandcharacterizespecificproteins.Finally,itmayalsocomefromrecenteffortstoidentifyalloftheproteinsinthenetworksofinteractingproteinsinacellusingrapidsequentialtwo-hybridinteractorhuntsthatusetheproteinsisolatedinonehuntasstartingpointsforfurtherhunts,inasortof"proteininteractionwalk"(R.L.F.,unpublished).

AsdiscussedinSection5,alltwo-hybridapproachesinevitablyproducefalsepositives,interactionsthatdonotoccurinanybiologicalsetting.Thus,althoughitwillberichininformation,connectivitymapsderivedfromtwo-hybriddatawillnecessarilybeimprecise.Thisneednotbethoughtofasasignificantdrawbackofgenome-widetwo-hybridapproaches,provideditisborneinmindthattheinformationinaproteinlinkagemapderivesitsutilityinprovidingcluestoimportantinteractionswhichmustbeexploredwithfurtherstudyusingothermethods.

Oneexampleofaninsightintoproteinfunctionfromalargescaletwo-hybridapproachistheidentificationoftheDrosophilaproteinRoughex,Rux,asaproteinthatinteractsstronglyandspecificallywithDrosophilaCyclinE(Finley,Zavitz,Thomas,Richardson,Zipursky,andBrent,inprep).Rux,a335aminoacidproteinwhosesequencegivesnocluestoitsfunction(Thomasetal.,1994),wasinapanelof600baitproteinsthatwetestedforinteractionwithaCyclinEprey.Itwasknownthatruxisrequiredfornormaleyedevelopment;lossoffunctionruxmutantshaverougheyesandaberrantcellcycleregulationintheeyeimaginaldiscfromwhichtheeyedevelops(Thomasetal.,1994).ThomasetalshowedthatastripeofcellsinthemorphogeneticfurrowofthedevelopingeyediscmustarresttransientlyintheG1phaseofthecellcycleforproperdevelopmentandthisG1arrestfailsinruxmutanteyediscs.Combinedwiththisinformation,thefindingthatRuxinteractsdirectlywithCyclinE,aproteinknowntoberequiredforprogressionthroughG1,immediatelysuggestedthatRuxmodulatedcyclinactivity,andinspiredustoundertakespecificgeneticandbiochemicalexperimentstotestthehypothesis.

Scaledupinteractionmatingassaysarelikelytobeusefulintheanalysisofgeneticdiseasesandothercomplexgenetictraits.Thefirstversionofthisidea,whichhasalonghistory,isthatgenesthatmodifythefunctionofothergenesmayparticipateinthesameprocess.Alessobviouscorollaryofthisideabecameapparentseveralyearsago:that,amongtheproteinsthatinteractwithaproteininvolvedinadisease,thosethatinteractdifferentlywithwild-typeanddiseasestateallelicformsoftheproteinarelikelytobeinvolvedinthedisease.Recently,ReymondandBrentundertookatestofthisidea(ReymondandBrent,1995).TheystudiedtheproteinencodedbytheINK4humantumorsuppressorgene,p16.Wildtypep16interactswithtwohumanCyclin-dependentkinases,Cdk4andCdk6toinhibittheiractivity.Asexpected,interactionmatingshowedthatallelesofp16foundincancer-pronefamiliesaredeficientintheirinteractionwiththekinases.Twounexpectedconclusionsarosefromtheseexperiments.Oneallele,p16-G101W,showeddecreasedinteractionwithCdk4butnotwithCdk6,suggestingthatitsroleindiseaseisunrelatedtoitsactiononCdk6.Furthermore,anotherallele,p16-I49T,whichisalsofoundinthecontrolpopulation,isdeficientininteractionwithCdk4,suggestingthatthisallelemayalsocontributetoatumor-pronephenotype.Thesefindingsunderscorethefactthatinteractionmatingwithdifferentallelesinapopulationwillcontributetotheanalysisofcomplexpolygenictraits.

Theabilitytoconductscaled-uptwohybridanalysishascomeatagoodtime.Thetrickleofnewgenesandalleleshasbecomeatorrent.Robustandgeneralapproachestotheunderstandingofgeneandpathwayfunctionwillhelpustothenextstepofbiologicalunderstanding.

BacktoFinleyLabHomePage

7.Acknowledgments

WethankL.LokandmembersoftheBrentlaboratory,pastandpresent,forhelpfuldiscussions,A.Mendelsohnforassistanceinworkingouttheinteractionmatingassay,andA.Reymondforhelpincollectingandmaintainingthebaitpanel.WealsothankP.Colas,E.Golemis,andC.Girouxforhelpfulcommentsonthemanuscript.R.B.wassupportedbyHoeschtAGandanAmericanCancerSocietyFacultyResearchAward.

--------------------------------------------------------------------------------

8.References

Allen,J.B.,Walberg,M.W.,Edwards,M.C.,andElledge,S.J.(1995).FindingProspectivepartnersinthelibrary:thetwo-hybridsystemandphagedisplayfindamatch.TrendsinBiochem.20,511-516.

Bartel,P.L.,Roecklein,J.A.,SenGupta,D.,andFields,S.(1996).AproteinlinkagemapofEscherichiacolibacteriophageT7.NatureGenetics12,72-77.

Bendixen,C.,Gangloff,S.,andRothstein,R.(1994).Ayeastmating-selectionschemefordetectionofprotein-proteininteractions.NucleicAcidsResearch22,1778-1779.

Brent,R.,andPtashne,M.(1984).Abacterialrepressorproteinorayeasttranscriptionalterminatorcanblockupstreamactivationofayeastgene.Nature312,612-615.

Chien,C.-T.,Bartel,P.L.,Sternglanz,R.,andFields,S.(1991).Thetwo-hybridsystem:Amethodtoidentifyandclonegenesforproteinsthatinteractwithaproteinofinterest.Proc.Natl.Acad.Sci.USA88,9578-9582.

Choi,K.Y.,Satterberg,B.,Lyons,D.M.,andElion,E.A.(1994).Ste5tethersmultipleproteinkinasesintheMAPkinasecascaderequiredformatinginS.cerevisiae.Cell78,499-512.

Durfee,T.,Becherer,K.,Chen,P.-L.,Yeh,S.-H.,Yang,Y.,Kilburn,A.E.,Lee,W.-H.,andElledge,S.J.(1993).Theretinoblastomaproteinassociateswiththeproteinphophatasetype1catalyticsubunit.GenesandDev.7,555-569.

Estojak,J.,Brent,R.,andGolemis,E.A.(1995).Correlationoftwo-hybridaffinitydatawithinvitromeasurements.MolCellBiol15,5820-5829.

Fields,S.,andSong,O.(1989).Anovelgeneticsystemtodetectprotein-proteininteractions.Nature340,245-246.

Fields,S.,andSternglanz,R.(1994).Thetwo-hybridsystem:anassayforprotein-proteininteractions.TrendsGenet.10,286-292.

Finley,R.L.,Jr.,andBrent,R.(1994).InteractionmatingrevealsbinaryandternaryconnectionsbetweenDrosophilacellcycleregulators.ProcNatlAcadSciUSA91,12980-12984.

Finley,R.L.,Jr.,andBrent,R.(1995).Interactiontrapcloningwithyeast.InDNACloning2,ExpressionSystems:APracticalApproach,B.D.HamesandD.M.Glover,eds.(Oxford:OxfordUniversityPress),pp.169-203.

Golemis,E.A.,andBrent,R.(1992).FusedproteindomainsinhibitDNAbindingbyLexA.Mol.Cell.Biol.12,3006-3014.

Guarente,L.(1996).Transcriptionalcoactivatorsinyeastandbeyond.TrendsinBiochem20,517-521.

Guarente,L.,andPtashne,M.(1981).FusionofEschericiacolilacZtothecytochromecgeneofSaccharomycescerevisiae.Proc.Natl.Acad.Sci.USA78,2199-2203.

Gyuris,J.,Golemis,E.,Chertkov,H.,andBrent,R.(1993).Cdi1,ahumanG1andSphaseproteinphosphatasethatassociateswithCdk2.Cell75,791-803.

Harper,J.W.,Adami,G.R.,Wei,N.,Keyomarsi,K.,andElledge,S.J.(1993).Thep21Cdk-interactingproteinCip1isapotentinhibitorofg1cyclin-dependentkinases.Cell75,805-816.

Kranz,J.E.,Satterberg,B.,andElion,E.A.(1994).TheMAPkinaseFus3associateswithandphosphorylatestheupstreamsignalingcomponentSte5.GenesDev8,313-27.

Lech,K.,Anderson,K.,andBrent,R.(1988).DNA-boundFosproteinsactivatetranscriptioninyeast.Cell52,179-184.

Ma,J.,andPtashne,M.(1987).Anewclassoftranscriptionalactivators.Cell51,113-119.

Marcus,S.,Polverino,A.,Barr,M.,andWigler,M.(1994).ComplexesbetweenSTE5andcomponentsofthepheromone-responsivemitogen-activatedproteinkinasemodule.Proc.Natl.Acad.Sci.USA91,7762-7766.

Mendelsohn,A.R.,andBrent,R.(1994).Applicationsofinteractiontraps/two-hybridsystemstobiotechnologyresearch.Curr.Op.Biotechn.5,482-486.

Printen,J.A.,andSprague,G.F.,Jr.(1994).Protein-proteininteractionsintheyeastpheromoneresponsepathway:Ste5pinteractswithallmembersoftheMAPkinasecascade.Genetics138,609-619.

Reymond,A.,andBrent,R.(1995).p16proteinsfrommelanoma-pronefamiliesaredeficientinbindingtoCdk4.Oncogene11,1173-1178.

Rose,M.,andBotstein,D.(1983).ConstructionanduseofgenefusionstolacZ(beta-galactosidase)thatareexpressedinyeast.MethodsEnzymol101,167-80.

Thomas,B.J.,Gunning,D.A.,Cho,J.,andZipursky,L.(1994).CellcycleprogressioninthedevelopingDrosophilaeye:roughexencodesanovelproteinrequiredfortheestablishmentofG1.Cell77,1003-1014.

Tjian,R.,andManiatis,T.(1994).Transcriptionalactivation:Acomplexpuzzlewithfeweasypieces.Cell77,5-8.

VanAelst,L.,Barr,M.,Marcus,S.,Polverino,A.,andWigler,M.(1993).ComplexformationbetweenRASandRAFandotherproteinkinases.Proc.Natl.Acad.Sci.,U.S.A.90,6213-6217.

Yuan,Y.O.,Stroke,I.L.,andFields,S.(1993).Couplingofcellidentitytosignalresponseinyeast:interactionbetweenthealpha1andSTE12proteins.GenesDev7,1584-97.

Zervos,A.S.,Gyuris,J.,andBrent,R.(1993).Mxi1,aproteinthatspecificallyinteractswithMaxtobindMyc-Maxrecognitionsites.Cell72,223-232.