The Cauliflower mosaic virus (CaMV) open reading frame VI product (P6) is essential for the viral infection cycle. It controls translation reinitiation of the viral polycistronic RNAs and forms cytoplasmic inclusion bodies (viroplasms) where virus replicat

930ThePlantCell

withP6whenplacedinanunrelatedsequencecontext.P6boundtoA:P42butnottoP42alone(Figure1D)andtoGST:AbutnottoGST(seebelow),thusfurtherdemonstratingthatdomainAcanactindependentlyoftherestoftheaminoacidsequenceinmediatingP6self-associationinvitro.Takentogether,theresultsoftheabovefarproteingelblotexperimentsprovideevidencethattheN-terminalregionencompassesthedomainrequiredforP6self-interactioninvitro.

TheNTerminusofP6IsanEssentialDeterminantforBoththeFormationofViroplasmsandTheirLocalizationintheCytoplasm

ToobtainfurtherinformationabouttheimportanceoftheN-terminalregionofP6intheformationofinclusionbodies,tobacco(Nicotianatabacum)BY-2cellsweretransfectedwithrecombinantpCK-EGFPplasmidscodingforfull-lengthP6andtwodeletedversions(AandP6DA)fusedtotheCterminusoftheenhancedgreen uorescentprotein(EGFP).Theresultsde-scribedbelowarerepresentativeofatleastfourindependenttransfectionexperimentsandobservationbyconfocallaserscanningmicroscopy(CLSM)at20hpost-transfection.

AfterbombardmentofBY-2cellswithplasmidsexpressingtheEGFP:P6fusionprotein,;80%oftransfectedcellscontainedlargecytoplasmicinclusionbodies(3to5mmindiameter)withpittedsurfaces,generallyintheproximityofthenucleus(Figure2A,panels1and2).Theinclusionbodieswereformedbynumeroussmalleraggregates,mostofwhichappearedashollowdonut-likestructures(Figure2A,panel3).Toexcludethepossibilitythattheirformationwasartifactual(i.e.,asaresultoftheEGFPmoietyfusedtotheP6),protoplastswerepreparedfromCaMV-infectedturnipplantsasdescribedbyKobayashietal.(1998), xedandimmunolabeledwithanti-P6andsecond-aryantibodiescoupledtothe uorochromeAlexa568.Obser-vationsbyCLSMrevealedthattheviroplasmsthusproducedinthecontextofanauthenticviralinfectionhadasimilarstructure(Figure2B),demonstratingthattheEGFPmoietyhasnopro-nouncedeffectontheself-assemblyofEGFP:P6intobaccocells.MoreovertheseresultsalsoillustratethatP6moleculesassembleproperlyandindependentlyofthecellularcontextbecausesimilarlyshapedaggregateswereformedincellsfromhost(turnip)andnon-host(tobacco)plants.

Approximately20%ofthetobaccocellsexpressingtheEGFP:P6fusionproteincontainedaggregatesofvariablesizesbut<2mmindiameter(Figure2A,panels4and5),whichprobablycorrespondtoearlystagesofviroplasmformation.Thesmalleraggregatesgenerallywerescatteredinthecyto-plasm,althoughtheywerealsosometimesfoundwithinthenucleuswhenthecellswereanalyzedbyCLSM.ThepresenceofsuchaggregateswithinthenucleusmayindicatethatEGFP:P6moleculesweretransportedtothenucleus(seebelow)andwerethenunabletoexitaftertheirself-assemblybecauseofthelargesizeoftheresultingaggregates.

IncontrastwithEGFP:P6,EGFP:P6DA(Figure3A)didnotformaggregatesintobaccocells(Figure3B,panels1and2)butwasmainlyfoundinthenucleusandinparticularwithinthenucleolus.ThisresultstronglysuggeststhattheN-terminalregionofP6isadeterminantnecessaryfortheformationofaggregatesandthat

italsocontainssignal(s)involvedinthetargetingand/orretentionofP6withinthecytoplasm.SimilarlytoEGFP:P6DA,expressionofEGFP:Anevergaverisetoaggregatesofanysize,buttheproteinwasinsteaddistributeduniformlyinthecytoplasm(Figure3B,panels3and4).ThefailureofEGFP:Atoaccumulateinthenucleustoasigni cantextentwassomewhatsurprisingbecauseitissuf cientlysmallthatitwouldbeexpectedtobeabletotraversenuclearporesbypassivediffusion.

Takentogether,thesedatashowthattheN-terminalregionofP6isnecessarybutapparentlynotsuf cientfortheformationofviroplasmsand,thus,thatotherregion(s)ofP6contributetothisprocess.OurfailuretodetectEGFP:AinthenucleusstronglysupportstheideathatthecytosoliclocalizationofP6isgovernedbydomainA.

TheN-TerminalP6–P6InteractionDomainIsConservedamongCaMVStrains

ComparisonoftheN-terminalsequenceofP6fromCaMVstrainCabb-BJIwithitscounterpartsfromotherCaMVstrains(CabbS,CabbS-Japan,NY8153,CM1841,W260,D/H,D4,B29,Xin/Jin,andBari)showedidentityrangingfrom83to97%.TheregioncanbedividedintotwosubdomainsthatwehavedesignatedA1(aminoacids1to83)andA2(aminoacids84to112),respectively(Figure4A).Theformeristhemostconserved(87to99%ofidentity)withtwonotableinvariantsequences:I1(aminoacids11to20)andI2(aminoacids63to83).NotethatI1alsocontainsapentapeptidemotifEKILM(residues11to15)thatisidenticalatfourof vepositionstotheupsteamsequenceEKLLM(motifi1;Figure4B,top).ThesequenceI1formspartofapredicteda-helixlocatedneartheNterminus(positions4to31)thatcontainsaseriesoffoursuccessiveheptadsequences(abcdefg),whereresiduesinpositionsaanddarehydrophobicasinLeuzippermotifs(Figure4B).Thissequenceispredictedbycomputeranalysis(Berger,1995)todimerizeviaacoiled-coilstructurewith>98%probability.SubdomainA2ismorevariable(62to93%identity)anddoesnotpossessaconservedmotifamongCaMVstrains;ithasapredictedbsheetcon guration.

Asa rststeptowardfurtherde ningtheP6–P6interactiondomain,thesequencescodingforsubdomainsA1andA2wereampli edbyPCRandclonedintothepGEX-2TKvectortoproduceGST-taggedproteins.Thecapacityofthesefusionproteinstoself-interactandtointeractwithP6orGST:Aweretestedbyfarproteingelblotexperiments.RadiolabeledP6interactedwithGST:AandGST:A1butnotwithGST:A2(Figure4C,leftpanel),showingthatsubdomainA1isresponsiblefortheinteractionbetweenP6molecules.Thisresultwascon rmedbythe ndingthat,wheneitherGST:AorGST:A1wasusedasoverlay,theyinteractedwithP6andwiththemselvesbutnotwithGST:A2,whereasthelatterwasunabletobindeithertoP6ortoanyofthefusionproteins(Figure4C,rightpanels).NotethatnoneofthefusionproteinsinteractedwithGST,excludingthepossi-bilitythattheobservedinteractionswereduetodimerizationofthetag.Takentogether,theseresultsprovideevidencethatthe83N-terminalaminoacidsencompassthedomainrequiredforP6self-interactioninvitroand,hence,probablyfortheformationofviroplasms.