BRAHMA ATPase of the SWISNF Chromatin Remodeling Complex Acts as a Positive Regulator of Gibberellin-Mediated Responses in Arabidopsis

BRAHMA ATPase of the SWISNF Chromatin Remodeling Complex Acts as a Positive Regulator of Gibberellin-Mediated Responses in Arabidopsis BRAHMAATPaseoftheSWI/SNFChromatinRemodeling ComplexActsasaPositiveRegulatorofGibberellin- MediatedResponsesinArabidopsis RafalArchacki1,DanielBuszewicz2,TomaszJ.Sarnowski2,ElzbietaSarnowska3?a,AnnaT.Rolicka1, TakayukiTohge4,AlisdairR.Fernie4,YusukeJikumaru5?b,MaciejKotlinski2,RoksanaIwanicka- Nowicka1,2,KatarzynaKalisiak2,JacekPatryn1,JoannaHalibart-Puzio2,YujiKamiya5,SethJ.Davis3?c, MartaK.Koblowska1,2,AndrzejJerzmanowski1,2 * 1Department of Plant Molecular Biology, Faculty of Biology, University of Warsaw, Warsaw, Poland, 2Institute of Biochemistry and Biophysics, Polish Academy of Sciences,Warsaw,Poland,3Max-PlanckInstituteforPlantBreeding,Cologne,Germany,4MaxPlanckInstituteofMolecularPlantPhysiology,Potsdam-Golm,Germany, 5RIKENPlantScienceCenter,Tsurumi-ku,Yokohama,Kanagawa,Japan Abstract SWI/SNF chromatin remodeling complexes perform a pivotal function in the regulation of eukaryotic gene expression. Arabidopsis (Arabidopsis thaliana) mutants in major SWI/SNF subunits display embryo-lethal or dwarf phenotypes, indicating their critical role in molecular pathways controlling development and growth. As gibberellins (GA) are major positiveregulatorsofplantgrowth,wewantedtoestablishwhetherthereisalinkbetweenSWI/SNFandGAsignalingin Arabidopsis.Thisstudyrevealedthatinbrm-1plants,depletedinSWI/SNFBRAHMA(BRM)ATPase,anumberofGA-related phenotypictraitsareGA-sensitiveandthatthelossofBRMresultsinmarkedlydecreasedlevelofendogenousbioactiveGA. Transcriptional profiling of brm-1 and the GA biosynthesis mutant ga1-3, as well as the ga1-3/brm-1 double mutant demonstratedthatBRMaffectstheexpressionofalargesetofGA-responsivegenesincludinggenesresponsibleforGA biosynthesisandsignaling.Furthermore,wefoundthatBRMactsasanactivatoranddirectlyassociateswithpromotersof GA3ox1,aGAbiosyntheticgene,andSCL3,implicatedinpositiveregulationoftheGApathway.ManyGA-responsivegene expressionalterationsinthebrm-1mutantarelikelyduetodepletedlevelsofactiveGAs.However,theanalysisofgenetic interactions between BRM and the DELLA GA pathway repressors, revealed that BRM also acts on GA-responsive genes independently of its effect on GA level. Given the central position occupied by SWI/SNF complexes within regulatory networks controlling fundamental biological processes, the identification of diverse functional intersections of BRM with GA-dependentprocessesinthisstudysuggestsaroleforSWI/SNFinfacilitatingcrosstalkbetweenGA-mediatedregulation andothercellularpathways. Citation:ArchackiR,BuszewiczD,SarnowskiTJ,SarnowskaE,RolickaAT,etal.(2013)BRAHMAATPaseoftheSWI/SNFChromatinRemodelingComplexActsas aPositiveRegulatorofGibberellin-MediatedResponsesinArabidopsis.PLoSONE8(3):e58588.doi:10.1371/journal.pone.0058588 Editor:MiguelA.Blazquez,InstitutodeBiolog??aMolecularyCelulardePlantas,Spain ReceivedDecember8,2012;AcceptedFebruary5,2013;PublishedMarch11,2013 Copyright: ß 2013 Archacki etal. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited. Funding:ThisworkwassupportedbyMinisterstwoNaukiiSzkolnictwaWyzszego(MNiSW)andNationalScienceCentregrants:NN301158735and2011/01/D/ NZ1/01614 for AJ and RA, Marie Curie Reintegration Grant FP7-PEOPLE-2010-RG 268313 and W29/7. PR/2011 (MNiSW) for TJS, and Deutsche Forschunggemeinschaft–DFG-DAgrant1061/2-1forSJD.URLs:()()(. eu/mariecurie-actions/home.html)().Thefundershadnoroleinstudydesign,datacollectionandanalysis,decisiontopublish,or preparationofthemanuscript. CompetingInterests:Theauthorshavedeclaredthatnocompetinginterestsexist. *E-mail:andyj@ibb.waw.pl ?aCurrentaddress:MarieCurieMemorialCancerCenter,Warsaw,Poland ?bCurrentaddress:DepartmentofBiosciences,TeikyoUniversity,Utsunomiya,Japan ?cCurrentaddress:DepartmentofBiology,UniversityofYork,York,UnitedKingdom forassembly,overallstoichiometryandrecruitmentofSWI/SNF Introduction to target loci [3,4]. Arabidopsis has two major orthologs of the The SWI/SNF chromatin remodeling complexes are evolu- ATPase (BRM and SYD) and four orthologs of SWI3 (SWI3A, tionarilyconservedmultimericassemblagesofproteinsthatusethe SWI3B, SWI3C, SWI3D), which gives the potential to assemble energy of ATP hydrolysis to disrupt DNA-histone interactions. complexeswithdifferentcombinationsofsubunits[5,6].Asglobal ThroughtheirabilitytoregulateaccesstonucleosomalDNAthey regulatorstheArabidopsisSWI/SNFcomplexesareessential.This exert profound effects on transcriptional activity [1]. SWI/SNF- is reflected by the embryo-lethal phenotypes of single swi3a and mediatedchromatinremodelinghasbeenshowntoplayacentral swi3bmutantsandofdoublebrm/sydmutants[7,8].Duetopartial roleincellproliferation,differentiationanddevelopment[2].All redundancybetweentheBRMandSYDATPases,singlemutants SWI/SNF complexes possess a catalytic subunit (ATPase) in their respective genes are viable. The phenotypes of these associated with a set of accessory core subunits, including mutants,andofmutantsintheSWI3CandSWI3Dsubunits,are homologs of yeast SNF5 and SWI3 proteins which are essential PLOSONE | www.plosone.org 1 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling dwarf or semi-dwarf with numerous aberrations in organ de- participationofSWI/SNFchromatinremodelinginthemediation velopment[9,10,7]. ofGAresponses. Whilesomeoftheprocessesdisruptedinswi/snfmutantshave been revealed [11,12,13], the global pattern of changes in the Results regulatory networks that could lead to their strong and complex PlantsDepletedofBRMShowGA-relatedPhenotypic developmentalphenotypesislargelyunknown.Recently,SYDand TraitsandIncreasedSensitivitytoGABiosynthesis BRM ATPases were shown to interact with LEAFY and SEPALLATA3 proteins in order to control floral organ identity, Inhibitor actingantagonisticallytoPolycombrepressors[14].Thereisalso Arabidopsisbrmnullmutants(suchasbrm-1anditsphenocopy evidence linking SWI/SNF complexes with hormonal pathways. brm-6, studied here) [9,44], are depleted in the SWI/SNF-type SYDisinvolvedintheregulationofjasmonicacid-andethylene- ATPaseBRManddisplayadwarfphenotypewithcharacteristic dependent genes [15], and SWI3B is an interaction partner of short and branched roots, dark green coloration, closed flowers, HAB1, a key element in ABA signaling [16]. A transcriptional underdeveloped stamens, male sterility and delayed flowering profiling studyofbrmandsyd nullmutantsidentifiedthatseveral under non-inductive short-day conditions. Inthese respects, they genesinvolvedinauxinandGAsignalingwereaffected[8].These resembletosomeextentthephenotypeofmutantswithsuppressed dataandthepropertiesofArabidopsisswi/snfmutantsprompted GA signaling or biosynthesis, such as ga1-3 [45] or ga3ox [20], us to examine whether there is a functional link between GA whichhavereducedlevelsofendogenousgibberellins(Figure1A, signaling and SWI/SNF complex-mediated chromatin remodel- FigureS1A–E).However,thebrmnullplantsalsodisplayanumber ing. of unique features, like curled leaves and homeotic changes in GAsaremajorpromotersofplantgrowthanddevelopmentthat flowers[9,44],whicharenotsharedbyGAbiosynthesismutants. are involved in various processes including seed germination, TheGA-relatedphenotypictraitsofbrmnullmutantareconsistent vegetative growth, flowering and stress responses [17,18,19]. withthemicroarraytranscriptprofilingofBezhanietal.(2007)[8] LevelsofactiveGAsaretightlycontrolledthroughtranscriptional who reported that several hormone pathway genes, including regulationofgenesencodingGA20-oxidases(GA20ox)andGA3- those of the GA-pathway, are mis-expressed in brm-101 (another oxidases(GA3ox),responsibleforthelatestepsofGAbiosynthesis, brmnullallele)andsyd-2mutants. aswellasGA2-oxidases(GA2ox),responsibleforGAdegradation The above data suggest that BRM plays a role in GA [20,21,22].GAsignalinginitiateswiththebindingofGAtooneof biosynthesis and/or signaling. To test this hypothesis, we its receptors (GID1a, b, and c in Arabidopsis), triggering examinedbrm-1phenotypesinthepresenceoftheGAbiosynthesis proteasomal degradation of the master growth repressors: the inhibitor, paclobutrazol (PAC). The brm-1 plants were more DELLA proteins [18,19,23,24,25]. Arabidopsis has five DELLA sensitive to PAC than wild type plants, since brm-1 homozygotes proteins: RGA, GAIandRGL1-3. Genetic analyses have shown couldnotberecoveredaftergerminatingbrm-1/BRMprogenyon thatthedifferentDELLAsperformbothspecificandoverlapping mediumcontaining10mMPAC,theconcentrationatwhichwild functionsduringdevelopment[26,27].AtlowGAconcentrations, typeplantsdisplayedagerminationrateofabout70%(Figure1B). DELLAproteinsaccumulateandactasrepressorsofgrowthand Wealsotestedgrowthresponsesofbrm-1plantsgrownfromseeds otherGA-regulateddevelopmentalprocesses[28,29,30].Gain-of- incubated with exogenous GA to ensure germination. The function DELLA mutants, or mutants with decreased levels of presence of 10mM PAC severely affected the development of active GA, like ga1-3, which is defective in an early-step of GA brm-1mutants,whichfailedtosurvivebeyond25days,whilewild biosynthesis,arecharacterizedbyadwarfphenotypeandstrongly typeplantscontinuedtogrowundertheseconditions(Figure1C). impairedgermination,flowering,andfertility.Conversely,loss-of- As PAC can possibly interfere with the biosynthesis of other function DELLA mutations lead to suppression of the ga1-3 hormones, we also examined growth on medium containing phenotype[27,31].Morerecently,DELLAswerefoundtointeract 10mMPACsupplementedwithexogenousGAs.Inthiscase,the withthelight-responsivetranscriptionfactorsPIF3andPIF4inthe brm-1 plants germinated and were viable, suggesting that the nucleus, prompting a model in which they act primarily by enhanced reaction of these mutants to PAC is linked with its inhibitingtranscriptionalregulators[32,33,34,35,36,37,38,39].In inhibitoryeffectsontheGApathway(Figure1BandFigureS1F). additiontoproteasome-dependentregulationofDELLAlevelsby Since brm-1 is a highly pleiotropic mutant, defects in many active GA, their activities are also controlled by other proteins differentfunctionscouldhavepotentiallyinfluencedtheoutcome such as the N-acetylglucosamine transferase SPINDLY [40,41], of the PAC assays and the PAC hypersensitivity of brm-1 plants andSCARECROW-LIKE3(SCL3)thatwasrecentlyproposedto might be due to the additive effect of GA deficiency and earlier actasanattenuatorofDELLAs[42,43]. defects in growth and/or development resulting in the dwarf Inthisstudy,wedemonstratethatBRM,acatalyticsubunitof statureofadultbrm-1plants.Toexaminethispossibility,weused SWI/SNF complexes, affects the expression of a significant aweakbrm-3mutantinwhichaT-DNAinsertioninthe39portion number of GA-responsive genes in an opposite manner to of the BRM gene gives rise to a truncated protein lacking a C- DELLAs. This is consistent with our finding that the level of terminalfragmentof454aminoacids(approximately1/5thofthe active GA is markedly decreased in the brm null mutant. molecule). Although the brm-3 mutant exhibits only mild de- Moreover, we show that BRM activates GA3ox1 and binds to velopmentalandgrowthdefects(FigureS2)[46],PACtreatment chromatin in the vicinityofits promoter, suggesting thatit plays hadamuchgreaterinhibitoryeffectonthegerminationofbrm-3 a direct role in the positive regulation of GA biosynthesis. than on wild type seeds, while germination on medium without However, we also show by genetic analyses, that BRM controls PAC was normal for both the brm-3 and the wild type. As a number of GA-responsive genes independently of its effect on expected,atripledellamutant(rga/rgl1/rgl2)wasinsensitivetoPAC GA biosynthesis. We also reveal that in addition to targeting treatment (Figure 1D).Moreover, inthe presenceofPAC, brm-3 GA3ox1,BRMpositivelyregulatesanddirectlyassociateswiththe had significantly shorter roots than wild type plants (Figure 1E). promoterofSCL3,ageneperformingregulatoryfunctionsinthe Thus, both null and weak brm mutations confer increased GApathway.OurdemonstrationthatBRMinteractswiththeGA sensitivity to a GA biosynthesis inhibitor, indicating that this signaling pathway at different levels, is the first evidence for PLOSONE | www.plosone.org 2 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling Figure1.brmmutantsshowGA-relatedphenotypictraitsandincreasedsensitivitytopaclobutrazol.(A),Comparisonofbrm-1andga1- 3mutantsgrownonKMSmediumfor18daysunderlong-dayconditions.(B),Germinationofthebrm-1mutantisabolishedinthepresenceof 10mM PAC and rescued upon addition of exogenous gibberellin. The progeny of brm-1/BRM plants were analyzed 14 days after sowing. (C), Phenotypeofbrm-1plantsgrownfor25dayson10mMPACafterincubationofseedswithexogenousGA.(D),Germinationassayofwildtype,brm-3 and3xdella(rga/rgl1/rgl2)lines.Seedcoatruptureafter14dayswasscoredasgermination.(E),Rootelongationassayofwildtypeandbrm-3plants grownfor12daysonPAC-containingmedium.BarsinA,CandE=5mm. doi:10.1371/journal.pone.0058588.g001 phenotypic trait is not a secondary effect of earlier GA- metabolites in the GA biosynthesis and degradation pathways, independentgrowthdefectscausedbythebrmmutation. andofGA4,apredominantbioactiveformofGAinArabidopsis [48,22], in 4-week old wild type, brm-6 (brm null allele) [44] and ga1-3(inCol-0ecotype)[27]plantsgrowninsoil.LevelsofGA4in brmNullMutantsShowBothGA-sensitiveand- brm-6andga1-3plantswerearound50%and15%ofthewildtype insensitiveTraitsandhaveReducedLevelsofBioactive value, respectively (Table 1). Analysis of GA34, an inactive GA catabolite derived from active GA4, showed that its level was TolearnmoreabouttherelationshipbetweenGAsignalingand strongly reduced in brm-6 plants, indicating that the decrease in BRM-mediatedprocesses,weexaminedtheresponsivenessofthe activeGA4inthismutantresultedfromdefectiveGAbiosynthesis brm-1 mutant to exogenous GAs. The ability to germinate on rather than enhanced GAdegradation. Interestingly, the level of PAC+GAmedium(Figure1B)indicatedthatthebrmnullmutation GA12,butnotGA9(adirectprecursorofGA4),wasalsoreduced doesnotabolishGAperception.Moreover,inthepresenceofGA, inbrm-6comparedwiththewildtype(Table1).Therelativelyhigh brm-1 plants displayed significantly increased hypocotyl growth levelofGA9isprobablyduetothedecreasedrateofitsconversion (Figure2A,B)andagreatlyacceleratedonsetoffloweringunder to GA4, while the reduced level of GA12 might be caused by short-dayconditions(i.e.daystoflowering),suchthatthemutants increased20-oxidationofthisform(seeDiscussion). floweredsimilarlytowildtypeplants(Figure2C,TableS1).This Theaboveanalysisindicatedthatthebrmnullmutantispartly suggestedthatatleastsometraitsofbrmmutantsmaybecausedby GAdeficient,suggesting adirectorindirect roleofBRMin GA GA deficiency, which can thus be overcome by application of biosynthesis. However, unlike in typical GA-deficient mutants, exogenousGAs,asisthecaseforGAbiosynthesismutants. treatment with GA did not reverse some of the other ga1-3-like Therefore, we next compared, through combined liquid traits of brm-1 plants. Although the cotyledon size, rosette radius chromatography-mass spectrometry [47], the levels of key PLOSONE | www.plosone.org 3 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling Table1.Concentrationofgibberellinsinwildtypeand mutantlines. GA12 GA9 GA4 GA34 wildtype 5.206(0.327) 1.730(0.105) 4.269(0.315) 6.153(0.159) 2.063(0.384) 1.911(0.186) 2.697(0.125) brm-6 2.782(0.258) ga1-3 n.d. n.d. 0.583(0.072) 0.095(0.095) Thevaluesareng/gdryweight(s.e.).Theyarethemeansofthreebiological replicatesexceptfortheGA12 andGA9measurementsinga1-3,forwhich2 replicateswereused.n.d.–notdetermined. doi:10.1371/journal.pone.0058588.t001 BRMAffectstheExpressionofBothGABiosynthesisand SignalingGenes SincetheaboveresultsimplicatedBRMinGAbiosynthesisand also suggested a role in GA signaling, we used qRT-PCR to examinetheexpressionofgenesencodingenzymesinvolvedinthe latestepsofGAbiosynthesis(GA20ox1,GA20ox2,GA3ox1,GA3ox2) andGAinactivation(GA2ox1,GA2ox2),aswellasgenescodingfor GAreceptors(GID1a,b,c)andSCARECROW-LIKE3(SCL3 ),which isbelievedtoactintheGApathwaybyattenuatingtheDELLA repressors[42,43].ThelevelsoftheGA3ox1,GA3ox2,GA2ox1 ,and SCL3 transcripts were significantly decreased in brm-1 compared with wild type plants. The abundance of the GA20ox1 and 2, GID1a and b transcripts was increased, while GA2ox2 and GID1c were not significantly changed compared with the wild type (Figure3A).Toconfirmthesedata,wethenexaminedlevelsofthe same transcripts in the weak brm-3 mutant. Consistent with its increasedsensitivitytoPAC(Fig.1DandE),theexpressionofthe majority of GA pathway genes (except GA3ox2 and GA2ox1 ) was also changed, albeit slightly, in this mutant (Figure 3A). The relatively small changes in expression in brm-3 correspond tothe mild phenotypic effects seen in this mutant under normal conditions, and are similar to the findings of a previous study Figure2.GAresponsesofthebrm-1mutant.(A,B),Elongationof showinglessseverechangesintheexpressionofhomeoticgenesin brm-1 hypocotyls and roots in response to 1mM GA4. Plants were brm-3thaninbrm-1[46].Theresultsoftheseanalysesconfirmed grownonKMSmediumfor8daysunderlong-daysconditionsinthe that GA pathway genes are mis-regulated in null and weak brm presenceorabsenceof1mMGA4.GAapplicationcausedconsiderable mutants. elongationofthehypocotyls,buthadlittleeffectonbrm-1rootgrowth. Bar=5mm. (B), Hypocotyl length of plants grown as in A. Presented dataarethemeansof12measurements6s.d.(C),Floweringofbrm-1 BRMOccupiesthePromotersoftheGA3ox1andSCL3 plantsinresponsetoexogenousgibberellins.Plantsweregrowninsoil Genes under short-day conditions and treated with 10mM GA3. At least 15 ThedecreasedexpressionofGA3ox1islikelytobetheprimary plants of each line/condition were scored. Data are the means 6 s.d. cause ofthe reduced GA4 content in brm nullmutants, sincethe Asterisks indicate significant differences from the wild type plants (p,0.01). GA3ox1enzymecatalyzesconversionofprecursorGA9toGA4in doi:10.1371/journal.pone.0058588.g002 the final step of GA biosynthesis in Arabidopsis [22]. GA biosynthesisandGID1genesareknowntobefeedbackregulated by GAs. TheGA20ox andGID1 genes are up-regulated, whereas and stem height of brm-1 mutant plants increased upon GA GA2ox genes are downregulated under low GA conditions treatment, they remained much smaller than in wild type plants [23,49,50]. We thus hypothesized that increased expression of (Figure2A,TableS1).Inaddition,whengrowninthepresenceof GA20ox and GID1 genes, as well as decreased expression of the GA,therootsofbrm-1remainedsignificantlyshortenedcompared GA2ox1 gene in brm mutants is a secondary effect caused by with those of wild type plants (Figure 2A). There was also no a feedback mechanism compensating downregulation of GA3ox1 reversion by GA treatment of brm-1 flower phenotypes, and the and depletion of active GA, and possibly also downregulation of mutantplantsremainedcompletelymalesterile(notshown).Thus, SCL3, which encodes an important factor involved in the anumberoftheGAdeficiency-likeaberranttraitscausedbythe maintenance of GA pathway homeostasis (Figure 3B) [25]. brm null mutation (short stature, short roots, flower defects and Therefore, we examined whether BRM regulates GA3ox1 and male sterility) did not change or changed only slightly upon GA SCL3 expression by directly interacting with their regulatory treatment. Some of these brm mutant traits might be strongly sequences. A chromatin immunoprecipitation (ChIP) assay was influenced by defects in processes that involve BRM, but are performed using wild type and brm-1 plants and anti-BRM independentoftheGApathway,whileotherscouldbecausedby antibody [44]. Enrichment of GA3ox1 and SCL3 promoter GA-insensitivedefectsinGAsignaling. sequenceswasdetected,whiletherewasnosignificantenrichment of promoter sequences of GA20ox2 or GID1b (Figure 3C, Figure PLOSONE | www.plosone.org 4 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling seeds.Thiscouldbeduetotheseverelyunderdevelopedrootsof ga1-3/brm-1(Figure4A,C).AstheGAlevelinga1-3isatleast3- foldlower thanin brm-1, theadditivetraits ofthedouble brm-1/ ga1-3 mutant could be the result of GA pathway-independent functionsofBRM.However,enhancementofthega1-3phenotype has also been reported for mutants in genes acting as positive regulators of downstream GA responses, such as scl3 [42,43]. Therefore,theobservedadditiveeffectcouldalsobedue,atleast in part, to down-regulation of SCL3. Nonetheless, in spite of the enhanced phenotype of the ga1-3/brm-1 double mutant, the growth of brm-1 plants on 10mM PAC (Figure 1B) resulted in astrongerphenotype.Thiscouldbeexplainedbythelowamount of GAs present in the ga1-3 mutant ([51], this work) and the possibleblockingeffectofPAConothersignalingpathways.The ga1-3 phenotype was shown to be further strengthened by PAC treatment[52],whichisconsistentwithbothoftheseexplanations. WenextexaminedthelevelsofGA3ox1andSCL3transcriptsin thega1-3/brm-1doublemutantincomparisonwiththega1-3single mutant.ThetranscriptsofGA3ox1andSCL3areknowntobeup- regulated under low GA conditions as part of the feedback regulation of the GA pathway [23,25]. RT-qPCR analysis demonstratedthatthebrmnullmutationcausesamarkeddecrease in GA3ox1 and SCL3 transcript levels in the ga1-3 background (Figure4F),indicatingthatBRMalsocontributestotheregulation ofthesegenesunderlowGAconditions.Theoveralleffectsofthe double ga1-3/brm-1 mutation suggest that BRM, in addition to positivelyregulatingtheGAlevel,probablyfunctionsasaregulator ofGAresponses,atleastinpartbypromotingSCL3expression. TheTranscriptionalProfileofbrm-1Overlapswiththatof thega1-3Mutant To investigate how BRM contributes to global GA-dependent Figure3.BRMdirectlyregulatestheexpressionoftheGA3ox1 transcriptional regulation, we compared the transcript profiles of and SCL3 genes. (A), RT-qPCR analysis of relative transcript levels of GAbiosynthesisandsignalinggenesin18-d-oldwildtype,brm-1and 18-d-oldseedlingsofga1-3,brm-1anddoublega1-3/brm-1mutant brm-3 lines. The housekeeping genes PP2A and GAPC were used as linesandwildtypeplants,grownonMSmedium(seeFigure4A; normalization controls. RT-qPCR data are the means 6 s.d. of 3 Tables S2, S3, S4). Microarray transcriptome analysis revealed biological replicates. Transcript levels in the wild type were set to 1. a considerable overlap between the single brm-1 and ga1-3 Asterisksindicatesignificantdifferencesfromthewildtypeplantswith mutants: over 40% of genes that were mis-expressed in ga1-3 p,0.05 (*) or p,0.01 (**). (B), Simplified model of the GA signaling (comparedwiththewildtype),alsoshowedmis-expressioninbrm-1 pathway.(C),BRMrecruitmenttothepromotersofGA3ox1andSCL3in wildtypeandbrm-1plants,analyzedbyChIP-qPCR.Thesignalobtained (Figure 5A, B; Table S5). This showed that the expression of forthePP2ApromoterregionwasusedtonormalizetheqPCRresultsin asignificantnumberofGA-responsivegenesisalsodependenton eachsample.Distal(d)andproximal(p)promotersequencesrelativeto theBRM-containingSWI/SNFcomplex,whichisconsistentwith thestartcodonofeachgenewereanalyzed.Foldenrichmentofeach a positive role for BRM in GA biosynthesis and signaling. In regioninthewildtypewascalculatedrelativetothebrm-1sample.The agreement with our qRT-PCR data, the two direct targets of valueofChIPenrichmentinbrm-1wassetto1.Dataarethemeans6 s.e. from 3 reactions in one ChIP experiment. Similar results were BRM, GA3ox1 and SCL3, were present among the genes down- obtainedinseparateexperiments. regulatedinbrm-1comparedwiththewildtype(amongrecognized doi:10.1371/journal.pone.0058588.g003 GA biosynthesis and signaling genes, the microarray and qRT- PCRdatawereinconsistentonlyforGA20ox2;TablesS2–S3). FunctionalclassificationbasedonGeneOntologyshowedthat S3).Thus,theGA3ox1andSCL3genesrepresentdirecttargetsof overall, the brm-1 mutation had a much broader effect on gene BRM,whichisconsistentwiththenotionthatBRMisinvolvedin theregulationofbothGAbiosynthesisandGAsignaling. expression than ga1-3 (Figure S5A, B), confirming that BRM regulates many processes independent of gibberellins. However, gene clusters commonly regulated by both BRM and GAs were ga1-3/brm-1DoubleMutantShowsAdditiveand also identified (Table 2; Figure S5). In both the ga1-3 and brm-1 SynergisticTraits gene sets, genes involved in stress responses, the circadian clock, In order to genetically test the contribution of BRM to GA- flowering, and responses to light and hormones were highly relatedresponses,wegeneratedaga1-3/brm-1doublemutantline. enriched(Table2).Theseprocessesareknowntobeinfluencedby When compared with either of the single mutants, homozygous gibberellins[53,54,55,56,32]andwerealsoenrichedintranscrip- ga1-3/brm-1 plants showed a number of additive as well as tional analyses of DELLA-responsive genes [57,58,59]. Interest- synergistic traits including increased dwarfism, very short roots ingly,inouranalysis,thegreatestenrichmentinbothmutantgene andtheinabilitytoflowerunderlong-dayconditions(Figure4A– setswasingenesinvolvedinresponsestoauxinstimuluswhichis E; Figure S4). Double mutant plants were also less viable when consistentwiththeextensivecross-talkbetweentheGAandauxin growninsoil:onlyafewga1-3/brm-1homozygoteswererecovered pathways [59,60,61] (Table 2, Figure S5A). Moreover, examina- after germinating about 1000 heterozygous ga1-3/brm-1/BRM tion of the molecular function categories identified a highly PLOSONE | www.plosone.org 5 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling Figure4.ga1-3/brm-1mutantphenotypes.(A–B),Phenotypesofthega1-3,brm-1andga1-3/brm-1mutantsgrownonMSmedium(18-d-old seedlings,A)orinsoil(22-d-oldplants,B).Bars=10mm.(C–F),Quantitativecharacterizationofbrm-1,ga1-3andga1-3/brm-1mutants:rootlengthof 18-d-oldseedlings(C),rosettediameteratmaturity(D)andfloweringtimeunderLDconditions(E).Dataarethemeans6s.d.,10plantsofeachline werescored,exceptforga1-3/brm-1(7plants).*Allga1-3/brm-1plantsexceptonefailedtoflowerbytheendoftheexperiment(80days).(F),RT- qPCRanalysisofrelativetranscriptlevelsofGA3ox1andSCL3in20-d-oldwildtype,brm-1,ga1-3,andga1-3/brm-1lines.RT-qPCRdataarethemeans 6s.d.of3biologicalreplicates.Transcriptlevelsinthewildtypeweresetto1. doi:10.1371/journal.pone.0058588.g004 enriched cluster of genes encoding carboxylesterases and pectin- were mostly additive, although for some of the analyzed genes, related enzymes, that was similarly affected in ga1-3 and brm-1they could be classified as synergistic, as the ga1-3 and brm-1 mutants(FigureS5B,C),indicatingthatBRMcanregulatelarge mutations enhanced the action of one another (Table S7). The GA-dependentgenefamilies. additive changes, as hypothesized above with respect to pheno- The inclusion of data for the double ga1-3/brm-1 mutant typicchanges,couldbeduetoGA-independenteffectsofthebrm-1 revealed a common set of 446 genes mis-expressed in all three mutation on GA-responsive genes. One possible explanation for mutants(Figure5A;TableS6).Itisnoteworthythatover90%of thegenesaffectedsynergistically,isthatGAsignalingandBRM- genesinthisoverlappinggroupdisplayedasimilarpatternofmis- mediated chromatin remodeling converge on the same targets withsomefunctionalinteractions. expression(eitherup-ordown-regulation)inthebrm-1andga1-3 mutant backgrounds (Figure 5C). This showed that BRM depletion (brm-1 mutant) or a significant decrease in GA content DELLAMutationsPartiallySuppressthebrmPhenotype (ga1-3 mutant) have a similar effect on gene expression, further Wereasonedthatsomeofthetraitsofbrmmutantsareprobably supportingapositiveroleforBRMinGAbiosynthesis.Consistent due to reduction in the levels of active GAs. Traits such as withthisfinding,themajorityofthesegenes(about60%)showed decreased germination and viability in the presence of PAC and non-additive expression levels in the double mutant compared delayed flowering under short-days conditions (Figure 1B, withthesinglemutants.Itislikelythatthesegenesreactmainlyto Figure2C)canbereversedbyexogenousGAs,whichareknown decreased GA levels caused by both brm-1 and ga1-3 mutations. to act predominantly through the destruction of DELLA repressors. On the other hand, and as stated above, some GA- Interestingly, the remaining genes from the overlapping group (about 40%) exhibited enhanced mis-expression in the double related traits of brm-1 plants, like short roots and reduced plant mutantcomparedwithbothsinglemutants(Figure5B;TableS7), size, showed additive changes in the double ga1-3/brm-1 mutant consistentwiththemoreseverephenotypiceffectsobservedinga1- andwereonlymarginallyamelioratedbyGAtreatmentofbrm-1 3/brm-1plants.Thechangesingeneexpressioninthissub-group (Figure2A,TableS1).TofurtherinvestigatetheroleofBRMin PLOSONE | www.plosone.org 6 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling Table2.GeneOntology(GO)categoriesstatisticallyover- representedamonggenesdifferentiallyexpressedinboth ga1-3andbrm-1mutants. GOcategory(biologicalprocess) p-value Numberofgenes responsetoauxinstimulus 3.89E210 29 1.90E24 circadianrhythm 8 6.55E24 responsetoredorfarredlight 13 cellularcarbohydratemetabolicprocess 0.0169 21 photoperiodism,flowering 5 0.0360 responsetogibberellinstimulus 0.0445 8 responsetosaltstress 0.0461 15 doi:10.1371/journal.pone.0058588.t002 possible that BRM exerts some additional DELLA-independent regulatoryeffectonGA-responsivegenes. BRMActsThroughDistinctMechanismstoRegulateGA- responsiveGenes To further investigate the mechanisms by which BRM can regulate GA-responsive genes, we focused on putative GA- and BRM-dependent genes showing enhanced expression in the ga1- 3/brm-1 double mutant. As genes that are responsive to GA are oppositely regulated by gibberellins and DELLAs [58], we examined the effect of the della mutations on the expression of selected genes by using brm-1/3xdella mutants in the ga1-3 background. In agreement with the phenotype of ga1-3/brm-1 and the microarray data, the effects of the ga1-3 and brm-1 Figure5.Transcriptionalprofileofbrm-1overlapswiththatof ga1-3. (A), Overlap betweendifferentially regulated genes in mutants mutations were apparently additive for the genes EXP5 brm-1, ga1-3 and ga1-3/brm-1 identifiedinmicroarray data,shownby (AT3G29030) and OFP16 (AT2G32100), and synergistic for a Venn diagram. (B), Genes up- and down-regulated in all three CYS2 (AT2G31980) and LTP2 (AT2G38530) (Figure 6C). All of mutants,shownbyaheatmap.Thecolorscalerepresentsnormalized these genes were DELLA-responsive, since removal of RGA, expressionlevels.(C),94%ofthegenescommonlymis-expressedinall RGL1,andRGL2inthega1-3backgroundchangedtheirtranscript threemutantsshowexpressionchangesinasimilardirection.Green– genesmis-regulatedonlyinbrm-1;blue–genesmis-regulatedinbrm-1, levelsinthewildtypedirection,althoughinthecaseofEXP5and ga1-3 and ga1-3/brm-1; orange – genes mis-regulated in a similar LTP2 they did not reach the wild type level of gene expression directioninallthreemutants;gray–genesmis-regulatedinanopposite (possiblybecausethesegenesarealsounderthecontroloftheGAI directioninoneofthemutants. and/or RGL3 DELLA proteins). In agreement with the micro- doi:10.1371/journal.pone.0058588.g005 arraydata(Figure5C),thisanalysisdemonstratedthatBRMand DELLAs have opposing effects on the expression of GA- GA-mediatedresponses,weconstructedabrm-1/3xdellaline(brm- dependent genes. Notably, in the ga1-3/brm-1/3xdella line the 1/rga-28/rgl1-2/rgl2-13), in which three of the five Arabidopsis expression levels of EXP5, OFP16, CYS2 and LTP2 were DELLAgenesweremutated.Consistentwiththeobservedeffects intermediate between those in the ga1-3/brm-1 and ga1-3/3xdella of GAs (Figure 1B), these mutations fully restored the ability of mutants,indicatinganadditiveeffectofBRMandDELLAs,and brm-1/3xdella mutant plants to germinate on PAC-containing suggesting that DELLAs and SWI/SNF can independently medium(Figure6A).However,thetripledellamutationhadaless regulatetheexpressionofthesegenetargets. prominent effectontherootlengthofthe brm-1plants grownin thepresenceofPAC(Figure6B),whichisprobablyduetobrm-1- Discussion specificdevelopmentaldefects.Thisinterpretationissupportedby theobservationthattherootphenotypeisstrongerinbrm-1thanin BRMPositivelyRegulatesGABiosynthesis ga1-3, and that this phenotype is additive in the ga1-3/brm-1 Inthis study, weexamined the functionallinks between BRM doublemutant(Figure4AandC).Similarlysmalleffectsonroot ATPase, a catalytic subunit required for SWI/SNF-dependent growthwereobservedoncrossingthebrm-1/ga1-3mutantwiththe chromatin remodeling, and GA-signaling in Arabidopsis. Our 3xdellamutantline(FigureS6A),andupontreatmentofbrm-1or interestinthecross-talkbetweenthesetwopathwaysfollowedthe ga1-3/brm-1 plants with exogenous GAs (Figure 2A and Figure observationthatplantswithmutationsinBRMresembleinseveral S6B). The strongly repressed growth of brm-1 on PAC was also respects mutants with suppressed GA signaling or biosynthesis. attenuated by the triple della mutation, and the brm-1/3xdella Moreover, brm mutants showed increased sensitivity to GA mutantlineshowedanintermediatephenotypecomparedwiththe biosynthesisinhibitionwhichcouldbereversedbytreatmentwith parental lines, especially later in development (Figure 6B and exogenous GAs or by mutation of genes encoding the DELLA Figure S6C). While the partial rescue in brm-1/3xdella comparedrepressors. Similarly, the delayed flowering of the brm-1 mutant with 3xdella may be caused by higher levels or activity of the undershort-dayconditionsrevertedtothewildtypepatternupon treatmentwithexogenousGAs.Wealsofoundahighlysignificant remaining two DELLA proteins (GAI and RGL3), it is also PLOSONE | www.plosone.org 7 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling Figure6.BRMactsthroughdistinctmechanismstoregulateGA-mediatedresponses.(A),Germinationofthebrm-1mutanton10mMPAC isrescuedbythetripledellamutation.Theprogenyofbrm-1/BRMplantswereanalyzed10daysaftersowing.(B),Phenotypesof3-week-oldplants grownon2.5mMPAC.Thebrm-1/3xdellalineshowsanintermediategrowthphenotype.Bar=5mm.(C),RT-qPCRanalysisofrelativetranscriptlevels oftheOFP16,EXP5,CYS2andLTP2genesin18-d-oldwildtype,brm-1,ga1-3,ga1-3/brm-1,ga1-3/3xdellaandga1-3/brm-1/3xdellalines.Transcript levelsinthewildtypeweresetto1.Dataarethemeans6s.d.of3biologicalreplicates.(D),ModeloftheroleofBRMinregulatingtheexpressionof GA-responsivegenes.BRMpositivelyregulatestheGA3ox1andSCL3genesinvolvedinGAbiosynthesisandsignaling,andprobablythroughthis influencestheexpressionofmanyGA-responsivegenesintheoppositemannertoDELLArepressors.Inaddition,BRMseemstoactonasubsetof GA-responsivegenesindependentlyofDELLArepressors.Alsointhiscase,theeffectexertedbyBRMistypicallyintheoppositedirectiontothatof DELLAsandisobservedbothforgenesup-anddown-regulatedbytheSWI/SNFcomplex(blueandredlines,respectively). doi:10.1371/journal.pone.0058588.g006 overlap between the transcriptional profiles of brm-1 and GA phenotype of brm plants is more similar to that ofthe semidwarf biosynthesismutantga1-3.Inagreementwiththeseobservations, ga3ox1 mutant than the severe ga1-3 mutant [20]. We also thelevelofbioactiveGA4wasconsiderablyreducedinabrmnull hypothesizedthatthealteredexpressionofothergenesinvolvedin mutant compared with the wild type, although it was still more GA biosynthesis and signaling in brm plants is a consequence of than 3 times higher than in the ga1-3 mutant. GA4 depletion in feedbackcontrolinresponsetodecreasedGA3ox1expressionand BRM-deficientplantswasconsistentwiththeresultsofqRT-PCR GAcontent.Thiswassupportedbythefactthatwewereunableto analysisshowingthatexpressionoftheGA3ox1gene,encodingGA detect BRM on promoters of GA20ox2 and GID1b by ChIP 3-oxidase, responsible for the last step of synthesis of bioactive analysis. Moreover, there was a decrease of about 2-fold in the GA4, was down-regulated by brm mutations (Figure 3A). More- levelsofthemetabolitesGA12andGA34inbrm(Table1),similar over, ChIP experiments demonstrated that the promoter of to that previously described in the ga3ox1 mutant [20,62]. The GA3ox1isboundbyBRM,suggestingthatSWI/SNFremodeling reduction in GA12 observed in these studies was explained by is directly involved in transcription of this gene. Part of the increasedactivityofGA20oxenzymesduetofeedbackregulation complexphenotypeofbrm-1andthelargenumberofGA-related [20,62]. On the other hand, the increase in the level of the genes found to be misregulated in this mutant can therefore be metaboliteGA9inbrmwasconsiderablylowerthanthatreported accounted for by mild GA-deficiency caused mostly by GA3ox1 for ga3ox1 [20,62], raising the possibility that BRM can also downregulation. In agreement with this interpretation, the influenceGAbiosynthesisbydifferentmeans. PLOSONE | www.plosone.org 8 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling BRMAffectsGASignaling BRMAffectsGA-mediatedResponsesinDiverseWays Taken together, the findings of this study implicate BRM as A ga1-3/brm-1 mutant showed additive or synergistic pheno- apositiveregulatorofGA-mediatedresponsesandrevealdiverse types (Figure 4A–E). In agreement with these effects, transcrip- tional analysis of this double mutant revealed additive or (both direct and indirect) interactions between SWI/SNF BRM synergistic changes in the expression of many genes that are ATPase and the GA pathway. This resembles the complex involvementofSWI/SNFchromatinremodelingintheregulation affected in both ga1-3 and brm-1. Thus, BRM, in addition to promotingGAsynthesis,mayalsocounteractthenegativeeffects of flowering [13], where BRM appears to repress important flowering regulators that act in different genetic pathways. By ofDELLAsinadifferentway.Indeed,weshowedthataswellas highlightingtheinteractionsbetweenBRMandtheGApathway, promoting GA3ox1 expression, BRM also positively and directly ourresultsdiscloseanotherlayerofcomplexityandsuggestarole regulates the SCL3 gene, encoding a positive regulator of GA for BRM-dependent SWI/SNF chromatin remodeling in the responses that was proposed to act by attenuating DELLA integration of GA-controlled responses with other signaling repressors [42,43]. This suggests that BRM not only regulates pathways.AcandidategeneforsuchregulationisGA3ox1 ,shown GA biosynthesis, but also affects GA signaling. The down- inthisstudytobeadirecttargetofBRM,andwhosetranscription regulation of SCL3 (Figure 4F) could, at least in part, be is tightly regulated by both developmental and environmental responsible for the observed additive effects in the ga1-3/brm-1 stimuli[17,19,22]. doublemutant(atboththephenotypicandtranscriptionallevels), Interestingly, mutants in another chromatin remodeling factor sincethescl3nullmutationwaspreviouslyshowntoenhancethe PICKLE (PKL), a chromodomain-containing Snf2-type ATPase, ga1-3phenotype[42,43]. are also semi-dwarf, resembling GA-response mutants, and their ThepresenceofBRMonthepromotersofSCL3andGA3ox1, characteristic pickle-root phenotype is greatly enhanced by twogeneshomeostaticallyregulatedbyGAsignalingcomponents treatment with GA-biosynthesis inhibitors and decreased by GA [17,22,25], suggests that BRM-containing SWI/SNF complexes treatment[63].PKLhasbeenshowntocontrolalargenumberof might be recruited specifically to these target sequences by GA-responsive genes by acting in parallel to GA signaling [64]. transcriptional regulators acting in the GA signaling pathway. However,incontrasttothesituationinthebrmmutant,pklplants Interestingly,afewexamplesofDNA-bindingregulatorsthatare were found to have increased levels of active gibberellin. Thus, likely to act in GA homeostasis and could potentially serve to both of these chromatin remodeling complexes seem to act as recruit chromatin remodeling complexes have been reported positive regulators of the GA pathway, although probably by [17,22]. differentmechanisms. BRMcanRegulateGA-responsiveGenesinaDELLA- MaterialsandMethods independentManner OurphysiologicalanalysesrevealedthatsomeGA-relatedtraits PlantLinesandGrowthConditions of the complex brm-1 phenotype were not fully reversed by GA Arabidopsis thaliana wild type and all mutant lines were of the treatmentnorbyatripledellamutation(Figure2A,6B,andFigure Columbia-0(Col-0)ecotype.Thebrm-1,brm-3,andbrm-6mutant S6). We therefore investigated whether BRM could act on GA- alleles were characterized previously [9,46,44]. The ga1-3 line responsive genes independently of DELLAs. By comparing the introgressed into Col-0 and the ga1-3/rga-28/rgl1-2/rgl2-13 line transcriptional response of putative GA- and BRM-dependent [27]werekindlyprovidedbyDr.Tai-pingSun.Toobtainga1-3/ genes in ga1-3/brm-1 and ga1-3/brm-1/3xdella mutants, we brm-1andga1-3/brm-1/rga-28/rgl1-2/rgl2-13(ga1-3/brm-1/3xdella) identified genes that were affected in an additive manner by lines, heterozygous brm-1 mutant plants were crossed with ga1-3 and ga1-3/rga-28/rgl1-2/rgl2-13 homozygous lines, respectively, BRM and DELLAs, suggesting that BRM can also control GA- responsive genes by acting in parallel to DELLA repressors followed by PCR screening of mutant alleles in the segregating (Figure 6D). Alternatively, the partial rescue in brm-1/3xdella populations. The triple della (rga-28/rgl1-2/rgl2-13) and brm-1/ 3xdella lines were obtained by screening the same population. comparedwith3xdellamaybecausedbyhigherlevelsoractivityof the remaining two DELLA proteins, GAI and RGL3. It should Primers used for genotyping are listed in Table S8. Due to the sterility of all lines containing the homozygous brm-1 mutation, also be noted that it is not yet known whether SCL3 down- segregating progeny weresownfor each analysis andgenotyped. regulation could also influence brm-1/3xdella phenotypes. While For analysis of plants with the ga1-3 background, seeds were thergamutationwasshowntobeepistatictothescl3nullmutation imbibed in 100mM GA for 3d at 4uC and then washed in root length assays in the presence of PAC, this epistasis was 3 found to be only partial at later developmental stages [42,43]. thoroughlyinwaterbeforesowing.Plantsweregrownunderlong- day (LD; 16h light/8h dark) or short-day (SD; 8h light/16h Clearly,furtherstudies–likeectopicexpressionofSCL3inthebrm dark) conditions at 18–23uC, with 70% humidity and 200mM mutant background – are required to determine the extent to 22 21 m s light intensity. Seedlings were cultivated in medium which the decreased SCL3 level accounts for the GA-insensitive containing K Murashige and Skoog (MS) salts (Sigma-Aldrich), part of the brm phenotype. Finally, while the positive regulatory 0.5%(w/v)sucroseand0.8%(w/v)agar,pH5.8,orinsoil. functionofBRMinrespecttoGA3ox1andSCL3isfullyconsistent withtheobservedoppositeeffectsofBRMandDELLAsonGA- PACandGATreatment responsivegenes,itisstillsurprising,giventheoccurrenceofGA- Forgerminationassays,seedsofthewildtypeandbrm-3,brm-1/ independentregulationbyBRM,thattheexpressionofover90% BRM,brm-1/BRM/3xdellaand3xdellagenotypesweresownonMS oftheoverlappinggenesbehavessimilarly(up-ordown-regulation platescontainingdifferentconcentrationsofPACor10mMPAC withrespecttothewildtype)ineachofthesinglemutants(ga1-3 +10mM GA3. Segregating progeny of brm-1/BRM plants were andbrm-1)andinthedoublega1-3/brm-1mutant(Figure5C).This genotyped using primers specific for the wild type and mutant may indicate that there are other, as yet unrevealed, levels of alleles in order to confirm or exclude the brm-1 genetic functionalinteractionbetweenBRMandGAsignaling. background. To analyze growth responses of brm-1 or brm-1/ 3xdella mutants to PAC, seeds of wild type, brm-1/BRM, brm-1/ PLOSONE | www.plosone.org 9 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling BRM/3xdella and 3xdella genotypes were sown on MS plates Real-TimeRT-qPCRAnalyses and containing 10mM PAC or 10mM PAC +10mM GA3 Aerialpartsof18-d-oldseedlingsofwildtype,brm-1,ga1-3,ga1- cultivatedfor25days.Inordertopromoteequalgerminationof 3/brm-1, ga1-3/rga/rgl1/rgl2 and ga1-3/brm-1/rga/rgl1/rgl2 lines allseeds,theywerepre-incubatedwith100mMGA3,thenrinsed were used for analyses. RNA was extracted from plant material thoroughly and sown on PAC-containing media. To analyze the usingtheRNeasyplantminikit(Qiagen)andDNAwasremoved GA response, plants were grown in soil and treated with 10mM by TURBO DNase-treatment (Ambion). A first-strand cDNA GA by spraying twice a week, or they were grown on plates synthesis kit (Roche) was used to prepare cDNA from 1m g of 3 containing K MS medium supplemented with 1mM or 2mM RNA.Aliquots(2ml)of5-folddilutedcDNAsampleswereusedas GA4,placedverticallyunderlong-dayconditionsat22uC.Forthe templates in 20ml reactions containing LightCycler 480 SYBR Green I Master mix (Roche) and specific primers for PCR setofdataforwhichdifferencesweresmall,statisticalsignificance wasestimatedbydeterminingPvalueusingStudent’st-test. amplification in a LightCycler 480 System (Roche), as recom- mended by the manufacturer. The final primer concentrations were0.5mMandtheannealingtemperaturewassetat58–60uC. QuantificationofPlantHormones TheRT-qPCRdatawereanalyzedwithLightCycler480Software Aerialpartsofsoil-grownwildtype,brm-6andga1-3plantswere version 1.3. PP2A and GAPc housekeeping genes were used as harvestedattheendofthedayandimmediatelyfrozeninliquid normalizationcontrolsandgavesimilarresults.Eachexperiment nitrogen. Plant hormones were quantified according to Plackett wasperformedusingatleasttwoindependentbiologicalreplicates, etal., (2012) [51] by using a 6410 Triple Quad LCMS (Agilent and the specificity of real-time PCR products was confirmed by Technologies,SantaClara,CA,USA)withanAgilent1200series meltingcurveanalysisandagarosegelelectrophoresis.Fortheset rapid resolution liquid chromatography system fitted with of data for which differences were small (Figure 3A), statistical aZORBAXEclipseXDB-C18column(1.8mm,2.1650mm). significancewasestimatedbydeterminingPvalueusingStudent’s t-test.SpecificprimersusedinqPCRreactionsarelistedinTable MicroarrayTranscriptomeAnalysis S8. Total RNA was extracted from shoots of 18-d-old wild type, brm-1, ga1-3, and ga1-3/brm-1 seedlings using the RNeasy plant ChromatinImmunoprecipitation mini kit (Qiagen) according to the manufacturer’s protocol, ChIP experiments were performed as described by Gendrel followed by TURBO DNase treatment (Ambion). The quantity etal.(2005)[67]withsomemodifications.Aerialpartsof15-d-old andqualityoftheisolatedRNAwasdeterminedusingaNanoDrop seedlingsofthewildtypeandbrm-1mutant(negativecontrol)were ND1000 spectrophotometer (Nanodrop technologies) and RNA used as the source of chromatin. Anti-BRM antibody [44] was integrity wasassessedwithaBioanalyzer2100(AgilentTechnol- bound to Dynabeads Protein A (Invitrogen) and incubated with ogies). 100ng of RNA were used for aRNA synthesis with aliquots of 10-fold diluted chromatin (,100mg of DNA). The a GeneChip 39 IVT - Express Kit (Affymetrix), and 15mg of isolated DNA was resuspended in 100ml of water. ChIP labeledandfragmentedaRNAwerehybridizedwithArabidopsis enrichment of putative BRM targets was determined by qPCR ATH1 genome arrays, according to the manufacturer’s recom- using LightCycler 480 SYBR Green I Master mix (Roche). mendations (Affymetrix). Three biological replicates were exam- Reactionswereperformedwith2mlofimmunoprecipitatedDNA inedforeachgenotype. as template. A standard curve was established for each pair of primers.TheamountofChIPDNAwascalculatedbasedonthe MicroarrayDataAnalysis standard curve and then normalized to the PP2A promoter Microarray hybridization signals were recorded and processed sequencesignal(controllocus)foreachsample.Foldenrichmentof using AffymetrixH GeneChipH Command ConsoleH Software each region in the wild type was adjusted relative to the brm-1 (AGCC). sample. In independent BRM-ChIP experiments, enrichments All processed samples passed the quality control tests. The were also determined by subjecting the input and immunopreci- resulting CEL files were further analyzed with the Partek pitatedDNAtoPCRandvisualizingamplifiedbandsbyethidium Genomics Suite (Partek). A GC-RMA normalization was con- bromide staining after separation on agarose gels (Fig. S3). The 2S2-usequencewasusedasapositivecontrolasithasbeenshown ducted. Principle Components Analysis (PCA) for all genes revealed significant separation based on genotype. To exclude tobindBRMprotein[12].18SrDNAservedasnegativecontrol genesthatwerenotexpressedintheplantmaterial,anon-specific forBRMbinding.PrimersusedinChIPexperimentsarelistedin TableS8. filterwasappliedusingtheMAS5.0algorithm.Onlythosegenes identified as ‘‘present’’ in at least one of the three replicates of agenotypewereincludedinfurtheranalysis,and16,824of22,810 SupportingInformation passed the filtering criteria. A two-way ANOVA was performed Figure S1 Examples of phenotypic traits of the brm-1 and genes with a false discovery rate (FDR) of ,0.05 were null mutant resembling those of mutants with sup- consideredsignificantlyalteredintheirexpressioninthemutants pressedGAbiosynthesisorsignaling.(A,B),Semi-dwarfism comparedwiththewildtype.Thegenelistswerethenfilteredto and dark green coloration. (C), Short and branched roots. (D), select those with a fold change of .1.5. Gene list comparisons Closedflowers.(E),Underdevelopedstamens.(F),brm-1homozy- wereperformedusingthePartekGenomicsSuite.Geneexpression gous mutants germinate and are viable when grown on 10mM datashownasheat-mapswasstandardizedbythedefaultmethod PAC-containingmediumsupplementedwith10mMGA3.Atthis inthesoftware(z-scoreconversion)toreceivevaluesbetween22 concentration of GA3, the growth phenotype of wild type plants and +2. Overlap analysis of genes differentially expressed in the didnotfullyrecover.20-d-oldplantsareshown.Bar=5mm. brm-1andga1-3mutantswasperformedusingFisher’sexacttest. (TIF) Gene ontology analyses were performed with the FatiGO [65] FigureS2 Comparisonofbrm-3andbrm-1mutants.14- (Table2)andGOrilla[66](FigureS5)tools.Themicroarraydata and20-d-oldplantsgrownonMSmedium(A)orinsoil(B)under have been deposited in Gene Expression Omnibus and are LDconditionsareshown,respectively.Bars=5mm. accessiblethroughGEOSeriesaccessionnumberGSE26848. PLOSONE | www.plosone.org 10 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling (TIF) (XLSX) Figure S3 ChIP analysis of potential BRM targets and Table S3 Genes that exhibit significantly different controlgenes.The2S2-upromoter[12]and18SrDNAserved expressioninga1-3mutantcomparingtowildtype. as positive and negative controls for BRM binding, respectively. (XLSX) PrimersequencesusedinChIPanalysisarelistedinTableS8. Table S4 Genes that exhibit significantly different (TIF) expression in ga1-3/brm-1 mutant comparing to wild FigureS4 Floweringofthega1-3/brm-1doublemutant. type. The ga1-3/brm-1 mutant is usually unable to flower under long- (XLSX) day conditions (A, B). Treatment with 10mM GA3 restores its TableS5 Genesshowingdifferentialexpressionbothin ability to flower (C). 36-d-old (A) and 54-d-old (B, C) plants are brm-1andga1-3mutants. shown. (XLSX) (TIF) TableS6 Differentiallyexpressedgenesinga1-3/brm-1 Functional analysis of genes commonly S5 Figure mutant that are also changed in the same direction in regulated by BRM and GAs. (A, B), Genes misregulated in brm-1andga1-3. ga1-3andbrm-1mutants,classifiedbasedonGeneOntology(GO) (XLSX) categories of biological processes (A) and molecular function (B). Charts were generated using the Gene Ontology Enrichment Table S7 Genes from overlapping list showing en- AnalysisandVisualizationtool() hanced misexpression in ga1-3/brm-1 comparing to *65+.Therewereonly18geneswiththe‘‘gibberellin-responsive’’ brm-1andga1-3. GO term in the ga1-3 microarray dataset; 8 of them were also (XLSX) presentintheoverlappinggene-set.(C),Expressionlevelsofgenes Table S8 Oligonucleotides used in genotyping, RT- encoding carboxylesterases and pectin-related enzymes in micro- qPCRandChIP.[68–70]. arraydataforga1-3,brm-1andga1-3/brm-1lines. (TIF) (DOCX) Figure S6 Effect of DELLA mutations or GA treatment Acknowledgments onbrm-1andga1-3/brm-1mutantphenotypes.(A),Root lengthofga1-3/brm-1/3xdellacomparedwithga1-3/3xdellaplants. Wethankmembersofourlaboratoryforcritiqueandhelpfuladvice,Tai- 12-d-oldplantsareshown.(B),Rootlengthof8-d-oldga1-3/brm-1 pingSunforprovidingmutantlines,BartoszLangeforhelpwithstatistical plantscomparedwithga1-3plantsgrowninthepresenceof2mM analysis, and Tomoe Nose for help with hormone measurements. E.S. thanksJanuszSiedleckiforencouragement. GA . (C), Growth phenotype of brm-1/3xdella plants grown on 4 10mMPAC.40-d-oldplantsareshown.Bars=5mm. AuthorContributions (TIF) Conceived and designed the experiments: RA DB TJS MKK AJ. TableS1 EffectofGAapplicationonsizeandflowering Performedtheexperiments:RADBTJSYJRI-NKKJPJH-P.Analyzed ofbrm-1plants. thedata:TTARFESATRMK.Contributedreagents/materials/analysis (DOCX) tools:TTARFSJDYK.Wrotethepaper:RAAJ. Table S2 Genes that exhibit significantly different expressioninbrm-1mutantcomparingtowildtype. References 1. Narlikar GJ, Fan H-Y, Kingston RE (2002) Cooperation between complexes 11. Kwon CS, Hibara K, Pfluger J, Bezhani S, Metha H, et al. (2006) A role for thatregulatechromatinstructureandtranscription.Cell108:475–487. chromatinremodelinginregulationofCUCgeneexpressionintheArabidopsis 2. cotyledonboundary.Development133:3223–3230. ClapierCR,CairnsBR(2009)Thebiologyofchromatinremodelingcomplexes. AnnuRevBiochem78:273–304. 12. TangX,HouA,BabuM,NguyenV,HurtadoL,etal.(2008)TheArabidopsis 3. BRAHMA chromatin-remodeling ATPase is involved in repression of seed Martens JA, Winston F (2003) Recent advances in understanding chromatin maturationgenesinleaves.PlantPhysiol147:1143–1157. remodelingbySwi/Snfcomplexes.CurrOpinGenDev13:136–142. 4. Yang X, Zaurin R, Beato M, Peterson CL (2007) Swi3p controls SWI/SNF 13. Farrona S,Hurtado L,March-D??azR, Schmitz RJ,FlorencioFJ,etal.(2011) assemblyandATP-dependentH2A-H2Bdisplacement.NatStructMolBiol14: Brahma is required for proper expression of the floral repressor FLC in 540–547. Arabidopsis.PLoSONE6:e17997. 14. 5. Jerzmanowski A (2007) SWI/SNF remodeling and linker histones in plants. WuMF,SangY,BezhaniS,YamaguchiN,HanSK,etal.(2012)SWI2/SNF2 chromatin remodeling ATPases overcome polycomb repression and control BiochimBiophysActa1769:330–345. 6. floralorganidentitywiththeLEAFYandSEPALLATA3transcriptionfactors. Kwon SB, Wagner D (2007) Unwinding chromatin for development and ProcNatlAcadSciUSA109:3576–3581. growth:afewgenesatatime.TrendsGenet23:403–412. 7. SarnowskiTJ,RiosG,JasikJ,SwiezewskiS,KaczanowskiS,etal.(2005)SWI3 15. WalleyJW,RoweHC,XiaoY,ChehabEW,KliebensteinDJ,etal.(2008)The chromatin remodeler SPLAYED regulates specific stress signaling pathways. subunitsofputativeSWI/SNFchromatinremodelingcomplexplaydistinctroles PLoSPathog4:e1000237. duringArabidopsisdevelopment.PlantCell17:2454–2472. 16. SaezA,RodriguesA,SantiagoJ,RubioS,RodriguezPL(2008)HAB-1-SWI3B 8. Bezhani S, Winter C, Hershman S, Wagner JD, Kennedy JF, et al. (2007) interactionrevealsalinkbetweenabscisicacidsignalingandputativeSWI/SNF Unique, Shared, and Redundant Roles for the Arabidopsis SWI/SNF chromatin-remodelingcomplexesinArabidopsis.PlantCell20:2972–2988. Chromatin Remodeling ATPases BRAHMA and SPLAYED. Plant Cell 19: 403–416. 17. Sun TP (2008) Gibberellin metabolism, perception and signaling pathways in Arabidopsis.ArabidopsisBook6:e0103. 9. Hurtado L, Farrona S, Reyes JC (2006) The putative SWI/SNF complex 18. subunit BRAHMA activates flower homeotic genes in Arabidopsis thaliana. HauvermaleAL,AriizumiT,SteberCM(2012)Gibberellinsignaling:atheme PlantMolBiol62:291–304. andvariationsonDELLArepression.PlantPhysiol160:83–92. 19. Achard P, Genschik P (2009) Releasing the brakes of plant growth: how GAs 10. Wagner D, Meyerowitz EM (2002) SPLAYED, a novel SWI/SNF ATPase homolog,controlsreproductivedevelopmentinArabidopsis.CurrBiol12:85– shutdownDELLAproteins.JExpBot60:1085–1092. 20. 94. MitchumMG,YamaguchiS,HanadaA,KuwaharaA,YoshiokaY,etal.(2006) Distinct and overlapping roles of two gibberellin 3-oxidases in Arabidopsis development.PlantJ45:804–818. PLOSONE | www.plosone.org 11 March2013 | Volume 8 | Issue 3 | e58588 BRMChromatinRemodelerandGASignaling 21. RieuI,ErikssonS,PowersSJ,GongF,GriffithsJ,etal.(2008)Geneticanalysis tionofsiteofABAbiosynthesis,ABAtransportandhormoneinteractionsPlant revealsthatC19-GA2-oxidationisamajorgibberellininactivationpathwayin CellPhysiol51:1988–2001. Arabidopsis.PlantCell20:2420–2436. 48. Talon M, Koornneef M, Zeevaart JAD (1990) Endogenous gibberellins in YamaguchiS(2008).Gibberellinmetabolismanditsregulation.AnnuRevPlant 22. Arabidopsisthalianaandpossiblestepsblockedinthebiosyntheticpathwaysof Biol59:225–251. thesemidwarfga4andga5mutants.ProcNatlAcadSciUSA87:7983–7987. 23. SchwechheimerC(2008)Understandinggibberellicacidsignaling–arewethere 49. IuchiS,SuzukiH,KimYC,IuchiA,KuromoriT,etal.(2007)Multipleloss-of- yet?CurrOpinPlantBiol11:9–15. function of Arabidopsis gibberellin receptor AtGID1s completely shuts down 24. HarberdNP,BelfieldE,YasumuraY(2009)Theangiospermgibberellin-GID1- agibberellinsignal.PlantJ50:958–966. DELLA growth regulatory mechanism: how an ‘‘inhibitor of an inhibitor’’ 50. MiddletonAM,Ubeda-TomasS,GriffithsJ,HolmanT,HeddenP,etal.(2012) enablesflexibleresponsetofluctuatingenvironments.PlantCell21:1328–1339. Mathematical modeling elucidates the role of transcriptional feedback in Sun T-P (2011) The molecular mechanism and evolution of the GA-GID1- 25. gibberellinsignaling.ProcNatlAcadSciUSA109:7571–7576. DELLAsignalingmoduleinplants.CurrBiol21:338–345. 51. PlackettAR,PowersSJ,Fernandez-GarciaN,UrbanovaT,TakebayashiY,et 26. LeeS,ChengH,KingKE,WangW,HeY,etal.(2002)Gibberellinregulates al.(2012).AnalysisofthedevelopmentalrolesoftheArabidopsisgibberellin20- Arabidopsis seed germination via RGL2, a GAI/RGA-like gene whose oxidases demonstrates that GA20ox1, -2, and -3 are the dominant paralogs. expressionisup-regulatedfollowingimbibition.GenesDev16:646–658. PlantCell24:941–960. Tyler L,Thomas SG,HuJ, DillA,Alonso JM, etal.(2004)DELLA proteins 27. 52. Arizumi T, Murase K, Sun TP, Steber CM (2008) Proteolysis-independent and gibberellin-regulated seed germination and floral development in Arabi- downregulation of DELLA repression in Arabidopsis by gibberellin receptor dopsis.PlantPhysiol135:1008–1019. GIBBERELLININSENSITIVEDWARF1.PlantCell20:2447–2459. 28. KoorneefM,ElgersmaA,HanhartCJ,vanLoenen-MartinetEP,vanRijnL,et 53. Achard P, Cheng H, De Grauwe L, Decat J, Schoutteten H, et al. (2006) al.(1985)AgibberellininsensitivemutantofArabidopsis.PhysiolPlant65:33– Integration of plant responses to environmentally activated phytohormonal 39. signals.Science311:91–94. Peng J, Carol P, Richards DE, King KE, Cowling RJ, et al. (1997) The 29. 54. AranaMV,Mar??n-delaRosaN,MaloofJN,Bla?zquez MA,Alabad??D(2011) Arabidopsis GAI gene defines a signaling pathway that negatively regulates CircadianoscillationofgibberellinsignalinginArabidopsis.ProcNatlAcadSci gibberellinresponses.GenesDev11:3194–3205. USA108:9292–9297. Silverstone AL, Ciampaglio CN, Sun T-P (1998) The Arabidopsis RGA gene 30. 55. Hanano S., Domagalska M., Nagy F., and Davis, S J. (2006) Multiple encodesatranscriptionalregulatorrepressingthegibberellinsignaltransduction phytohormonesinfluencedistinctparametersoftheplantcircadianclock.Genes pathway.PlantCell10:155–169. Cells.11:1381–1392. ChengH,QinL,LeeS,FuX,RichardsDE,etal.(2004)Gibberellinregulates 31. 56. Domagalska MA, Sarnowska E, Nagy F, Davis SJ (2010) Genetic analyses of Arabidopsis floral development via suppression of DELLA protein function. interactionsamonggibberellin,abscisicacid,andbrassinosteroidsinthecontrol Development131:1055–1064. offloweringtimeinArabidopsisthaliana.PLoSOne,5:e14012. 32. FengS,MartinezC,GusmaroliG,WangY,ZhouJ,etal.(2008)Coordinated Achard P, Renou JP, Berthome? R, Harberd NP, Genschik P (2008) Plant 57. regulationofArabidopsisthalianadevelopmentbylightandgibberellins.Nature DELLAs restrain growth and promote survival of adversity by reducing the 451:475–480. levelsofreactiveoxygenspecies.CurrBiol18:656–660. 33. deLucasM,DaviereJM,Rodriguez-FalconM,PontinM,Iglesias-PedrazJM, 58. et al. (2008) A molecular framework for light and gibberellin control of cell Zentella R, Zhang ZL, Park M, Thomas SG, Endo A, et al. (2007) Global analysis of DELLA direct targets in early gibberellin signaling in Arabidopsis. elongation.Nature451:480–484. PlantCell19:3037–3057. ArnaudN,GirinT,SorefanK,FuentesS,WoodTA,etal.(2010)Gibberellins 34. controlfruitpatterninginArabidopsisthaliana.GenesDev24:2127–2132. 59. Gallego-Bartolome? J, Alabad??D, Bla?zquez MA (2011). DELLA-induced early 35. Gallego-Bartolome?J,MinguetEG,Mar??nJA,PratS,Bla?zquezMA,etal.(2010) transcriptional changes during etiolated development in Arabidopsis thaliana. Transcriptional diversification and functional conservation between DELLA PLoSONE6:e23918. proteinsinArabidopsis.MolBiolEvol27:1247–1256. 60. Willige BC, Isono E, Richter R, Zourelidou M, Schwechheimer C (2011) 36. Hou X,LeeLYC,Xia K,Yen Y,Yu H(2010) DELLAs modulate jasmonate Gibberellin regulates PIN-FORMED abundance and is required for auxin signalingviacompetitivebindingtoJAZs.DevCell19:884–894. transport-dependent growth and development in Arabidopsis thaliana. Plant 37. Josse EM, Gan Y, Bou-Torrent J, Stewart KL, Gilday AD, et al. (2011) A Cell23:2184–2195. DELLA in disguise: SPATULA restrains the growth of the developing 61. O’NeillDP,DavidsonSE,ClarkeVC,YamauchiY,YamaguchiS,etal.(2010) Arabidopsisseedling.PlantCell23:1337–1351. Regulation of the gibberellin pathway by auxin and DELLA proteins. Planta 38. Bai MY, Shang JX, Oh E, Fan M, Bai Y, et al. (2012) Brassinosteroid, 232:1141–1149. gibberellin and phytochrome impinge on a common transcription module in 62. HuJ,Mitchum MG,Barnaby N,AyeleBT,OgawaM,etal.(2008)Potential Arabidopsis.NatCellBiol14:810–817. sites of bioactive gibberellin production during reproductive growth in 39. Gallego-Bartolome?J,MinguetEG,Grau-EnguixF,AbbasM,LocascioA,etal. Arabidopsis.PlantCell20:320–336. (2012) Molecular mechanism for the interaction between gibberellin and 63. Henderson JT, Li HC, Rider SD, Mordhorst AP, Romero-Severson J, et al. brassinosteroid signaling pathways in Arabidopsis. Proc Natl Acad Sci U S A (2004)PICKLEactsthroughouttheplanttorepresstheexpressionofembryonic 109:13446–13451. traitsandmayplayaroleingibberellin-dependentresponses.PlantPhysiol134: Jacobsen SE, Olszewski NE (1993) Mutations at the SPINDLY locus of 40. 995–1005. Arabidopsisaltergibberellinsignaltransduction.PlantCell5887–896. 64. ZhangH,RiderSDJr,HendersonJT,FountainM,ChuangK,etal.(2008)The 41. SilverstoneAL,TsengTS,SwainSM,DillA,JeongSY,etal.(2007).Functional CHD3 remodeler PICKLE promotes trimethylation of histone H3 lysine 27. analysisofSPINDLYingibberellinsignalinginArabidopsis.PlantPhysiol.143: JBiolChem283:22637–22648. 987–1000. 65. Al-ShahrourF,D??az-UriarteR,DopazoJ(2004)FatiGO:awebtoolforfinding Heo JO, Chang KS, Kim IA, Lee MH, Lee SA, et al. (2011). Funneling of 42. significant associations of Gene Ontology terms with groups of genes. gibberellinsignalingbytheGRAStranscriptionregulatorscarecrow-like3inthe Bioinformatics20:578–580. Arabidopsisroot.ProcNatlAcadSciUSA108:2166–2171. 66. EdenE,NavonR,SteinfeldI,LipsonD,YakhiniZ(2009)GOrilla:AToolFor 43. ZhangZL,OgawaM,FleetCM,ZentellaR,HuJ,etal.(2011)Scarecrow-like3 Discovery And Visualization of Enriched GO Terms in Ranked Gene Lists. promotesgibberellinsignalingbyantagonizingmastergrowthrepressorDELLA BMCBioinformatics10:48. inArabidopsis.ProcNatlAcadSciUSA108:2160–2165. 67. Gendrel AV, Lippman Z, Martienssen R, Colot V (2005) Profiling histone Archacki R, Sarnowski TJ, Halibart-Puzio J, Brzeska K, Buszewicz D, et al. 44. modificationpatternsinplantsusinggenomictilingmicroarrays.NatMethods2: (2009) Genetic analysis of functional redundancy of BRM ATPase and 213–218. ATSWI3C subunits of Arabidopsis SWI/SNF chromatin remodelling com- 68. Silverstone AL, Mak PY, Mart??nez EC, Sun TP (1997) The new RGA locus plexes.Planta229:1281–1292. encodes a negative regulator of gibberellin response in Arabidopsis thaliana. Koorneef M, van der Veen JH (1980) Induction and analysis of gibberellin 45. Genetics146:1087–1099. sensitive mutants in Arabidopsis thaliana (L.) Heynh. Theor Appl Genet 58: 257–263. 69. AchardP,LiaoL,JiangC,DesnosT,BartlettJ,etal.(2007)DELLAscontribute toplantphotomorphogenesis.PlantPhysiol143:1163–1172. 46. Farrona S, Hurtado L, Reyes JC. (2007) A nucleosome interaction module is 70. required for normal function of Arabidopsis thaliana BRAHMA. J Mol Biol. OhE,YamaguchiS,HuJ,YusukeJ,JungB,etal.(2007)PIL5,aphytochrome- interacting bHLH protein, regulates gibberellin responsiveness by binding 373:240–250. directly to the GAI and RGA promoters in Arabidopsis seeds. Plant Cell 19: 47. Kanno Y, Jikumaru Y, Hanada A, Nambara E, Abrams SR, et al. (2010) Comprehensive hormone profiling in developing Arabidopsis seeds: Examina- 1192–1208. PLOSONE | www.plosone.org 12 March2013 | Volume 8 | Issue 3 | e58588