<%BANNER%>
RNA-Seq reveals genotype-specific molecular responses to water deficit in eucalyptus
CITATION DOWNLOADS PDF VIEWER
Full Citation
STANDARD VIEW MARC VIEW
Permanent Link: http://ufdc.ufl.edu/AA00008950/00001
 Material Information
Title: RNA-Seq reveals genotype-specific molecular responses to water deficit in eucalyptus
Series Title: BMC Genomics
Physical Description: Archival
Language: English
Creator: Villar, Emilie
Klopp, Christophe
Céline Noirot
Novaes, Evandro
Kirst, Matias
Plomion, Christophe
Gion, Jean-Marc
Publisher: BioMed Central
Publication Date: 2011
 Notes
Abstract: Background: In a context of climate change, phenotypic plasticity provides long-lived species, such as trees, with the means to adapt to environmental variations occurring within a single generation. In eucalyptus plantations, water availability is a key factor limiting productivity. However, the molecular mechanisms underlying the adaptation of eucalyptus to water shortage remain unclear. In this study, we compared the molecular responses of two commercial eucalyptus hybrids during the dry season. Both hybrids differ in productivity when grown under water deficit. Results: Pyrosequencing of RNA extracted from shoot apices provided extensive transcriptome coverage - a catalog of 129,993 unigenes (49,748 contigs and 80,245 singletons) was generated from 398 million base pairs, or 1.14 million reads. The pyrosequencing data enriched considerably existing Eucalyptus EST collections, adding 36,985 unigenes not previously represented. Digital analysis of read abundance in 14,460 contigs identified 1,280 that were differentially expressed between the two genotypes, 155 contigs showing differential expression between treatments (irrigated vs. non irrigated conditions during the dry season), and 274 contigs with significant genotype-by-treatment interaction. The more productive genotype displayed a larger set of genes responding to water stress. Moreover, stress signal transduction seemed to involve different pathways in the two genotypes, suggesting that water shortage induces distinct cellular stress cascades. Similarly, the response of functional proteins also varied widely between genotypes: the most productive genotype decreased expression of genes related to photosystem, transport and secondary metabolism, whereas genes related to primary metabolism and cell organisation were over-expressed. Conclusions: For the most productive genotype, the ability to express a broader set of genes in response to water availability appears to be a key characteristic in the maintenance of biomass growth during the dry season. Its strategy may involve a decrease of photosynthetic activity during the dry season associated with resources reallocation through major changes in the expression of primary metabolism associated genes. Further efforts will be needed to assess the adaptive nature of the genes highlighted in this study.
General Note: Publication of this article was funded in part by the University of Florida Open-Access publishing Fund. In addition, requestors receiving funding through the UFOAP project are expected to submit a post-review, final draft of the article to UF's institutional repository, IR@UF, (www.uflib.ufl.edu/ufir) at the time of funding. The Institutional Repository at the University of Florida (IR@UF) is the digital archive for the intellectual output of the University of Florida community, with research, news, outreach and educational materials
 Record Information
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution.
Resource Identifier: doi - 10.1186-1471-2164-12-538
System ID: AA00008950:00001

Downloads

This item is only available as the following downloads:

( PDF )

( XLS )


Full Text

PAGE 1

RESEARCHARTICLE OpenAccessRNA-Seqrevealsgenotype-specificmolecular responsestowaterdeficitineucalyptusEmilieVillar1,2,3,ChristopheKlopp4,ClineNoirot4,EvandroNovaes5,6,MatiasKirst5,ChristophePlomion2,7and Jean-MarcGion1,2*AbstractBackground: Inacontextofclimatechange,phenotypicplasticityprovideslong-livedspecies,suchastrees,with themeanstoadapttoenvironmentalvariationsoccurringwithinasinglegeneration.Ineucalyptusplantations, wateravailabilityisakeyfactorlimitingproductivity.However,themolecularmechanismsunderlyingthe adaptationofeucalyptustowatershortageremainunclear.Inthisstudy,wecomparedthemolecularresponsesof twocommercialeucalyptushybridsduringthedryseason.Bothhybridsdifferinproductivitywhengrownunder waterdeficit. Results: PyrosequencingofRNAextractedfromshootapicesprovidedextensivetranscriptomecoverage-a catalogof129,993unigenes(49,748contigsand80,245singletons)wasgeneratedfrom398millionbasepairs,or 1.14millionreads.Thepyrosequencingdataenrichedconsiderablyexisting Eucalyptus ESTcollections,adding 36,985unigenesnotpreviouslyrepresented.Digitalanalysisofreadabundancein14,460contigsidentified1,280 thatweredifferentiallyexpressedbetweenthetwogenotypes,155contigsshowingdifferentialexpression betweentreatments(irrigatedvs.nonirrigatedconditionsduringthedryseason),and274contigswithsignificant genotype-by-treatmentinteraction.Themoreproductivegenotypedisplayedalargersetofgenesrespondingto waterstress.Moreover,stresssignaltransductionseemedtoinvolvedifferentpathwaysinthetwogenotypes, suggestingthatwatershortageinducesdistinctcellularstresscascades.Similarly,theresponseoffunctional proteinsalsovariedwidelybetweengenotypes:themostproductivegenotypedecreasedexpressionofgenes relatedtophotosystem,transportandsecondarymetabolism,whereasgenesrelatedtoprimarymetabolismand cellorganisationwereover-expressed. Conclusions: Forthemostproductivegenotype,theabilitytoexpressabroadersetofgenesinresponsetowater availabilityappearstobeakeycharacteristicinthemaintenanceofbiomassgrowthduringthedryseason.Its strategymayinvolveadecreaseofphotosyntheticactivityduringthedryseasonassociatedwithresources reallocationthroughmajorchangesintheexpressionofprimarymetabolismassociatedgenes.Furthereffortswill beneededtoassesstheadaptivenatureofthegeneshighlightedinthisstudy.BackgroundPlantedforestsconstituteonly7%oftheglobalforested area,butcontributetoasignificantproportionofoverall forestgoodsandservices(e.g.upto35%ofindustrial roundwoodsupply).Inthecontextofclimatechange, theadaptationofplantedforestsisessentialforasustainableforestrysector.Theadaptationofindustrial plantationstopresentan dfutureenvironmental conditions(includingextremeweatherevents)depends onseveralfactors,includingthegeneticdiversityofthe materialusedforreforestationandthephenotypicplasticityofindividualgenotype s.Geneticdiversityensures thatforesttreescansurvive,adaptandevolveunder changingenvironmentalconditions[1,2],whereasphenotypicplasticityconstitutesashortertermresponseto environmentalchangesattheindividuallevelofparticularimportanceinlong-livedorganisms,suchastrees [2,3]. Eucalyptus isoneofthekeygeneraamongplanted trees.Thegenusisincludesthemostimportant *Correspondence:gion@cirad.fr1CIRAD,UMRAGAP,CampusdeBaillarguetTA10C,F-34398Montpellier Cedex5,France FulllistofauthorinformationisavailableattheendofthearticleVillar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 2011Villaretal;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommons AttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,andreproductionin anymedium,providedtheoriginalworkisproperlycited.

PAGE 2

hardwoodfibrecropsspeciesplantedworldwide(19millionhectaresaccordingto[4]).Several Eucalyptus speciesgrowrapidlyandarehighlyadaptable.These propertiesledtotheirintroductionworldwide,atlatitudesextendingfromsouthernEuropetoSouthAfrica. Initsnaturalrange(Australiaandsomenearbyislands), Eucalyptus arealsofoundinadiversespectrumofecologicalniches.Thegeneticdiversityof Eucalyptus has beenstudiedextensivelyandremarkablelevelsofvariationhavebeendetectedusingneutralmarkers[5-11] andingenespossiblyinvolvedinadaptivetraits[12-14]. Phenotypicplasticityisalsolikelytoensurebetteradaptationofindividualgenotypestochangingenvironmentalconditions[15]andisofparticularimportancein clonalforestry. Ecophysiologicalstudieshaveshownthatwateristhe principalfactorlimitingstemgrowthin Eucalyptus (e.g. [16,17]).Moreover,somestudieshavereportedthat eucalyptusgenotypesdifferintermsoftheircapacityfor phenotypicmodificationinresponsetowaterdeficit [18-20].Severalphysiologicalmechanismsforcoping withdroughthavebeendescribedinthesespecies:i)the regulationoftranspirationtodecreasewaterloss[21], ii)resourcereallocationfr omtheshoottotheroot,to increasewateruptake[17],andiii)adjustmentofosmoticpotential[22]orprotec tionagainstreactiveoxygen species,topreventdamageduetostress[20].Drought tolerancemechanismshavebeendescribedindetailat themolecularlevelforbothannualandperennialmodel plants,suchas Arabidopsis [23-25]and Populus [26-28], butlittleisknownaboutthemolecularbasisofdrought tolerancein Eucalyptus ,particularlyinfieldconditions. Next-generationsequencing(NGS)providesnew opportunitiesforstudiesofthemolecularplasticityin responsetowaterdeficit.ThehighthroughputofNGS isparticularlyusefulinnon-modelorganismsforwhich fewgenomicresourcesareavailable[29].Moreover, NGSissuitablefortranscri ptprofiling,combiningthe highthroughputofserialanalysisofgeneexpression (SAGE)withthefunctionalannotationcapacityofEST sequencing[30].Thesetechniqueshavebeenwidely usedfortranscriptomeprofiling,particularlyforstudies ofbiotic[31]andabiotic[24]stressresponses,andthe characterisationofdevelop mentalprocesses[32].Considerablesequencingdepthcanbeobtained,makingit possibletoidentifytranscriptomeexpressionvariation [29]. Inplants,theshootapicalmeristem(SAM)isakey organinstemdevelopment.TheSAMinitiatesphytomersandregulatesshootgrowthbyintegratingseveral signals,suchashormones(ABA,auxins,cytokinins)and transcription(e.g.homeobox)[33].Whenplantsare subjectedtoenvironmentalstimuli,theleafdevelopmentalnetworkisadjustedbychangesinshootapex activation[34].In Eucalyptus ,ESTresourceshavebeen developedforvarioustissu es,suchasroots,leavesand wood-formingtissues[3538],butalimitednumberof genomicresourcesareavailableforshootapices,despite theimportantroleofthisorganinplantorganogenesis. Inthisstudy,wecomparedtranscriptprofilesinthe shootapicesoftwoeucalyptusgenotypesusedinindustrialplantations,undertwowateringregimes irrigated (IR) versus non-irrigated(NI).Thetwogenotypesdiffer intheirgrowthratesandecophysiologicalcharacteristics atmaturity,withonegenotypebeingmoreproductive andwateruse-efficientthantheother.Weusedpyrosequencing(Roche454)tosequencenon-normalized cDNAlibrariesconstructedfromshoottipmRNA.After verifyingtechnicalreproducibility,weaddressedthefollowingquestions:i)Arethe remoleculardifferences betweengenotypes,reflectedinthecontrastingphenotypes,anddothesedifferencesaffectspecificpathways orhavearandomeffectonthetranscriptome?ii)Can wedetectmolecularplasticityintheresponsetowater shortageduringthedryseason,andwhichpathwaysare affected?iii)Doesthisplasticitydifferbetweengenotypes(i.e.isthereanygenotype-by-environmentinteraction?),andwhichgenesorpathwaysreflectthese differences?MethodsPlantmaterialWecomparedtheresponseoftwoeucalyptusgenotypes, 1-41(NCBITaxonomyID:764271)and18-50(NCBI TaxonomyID:765255),towatershortageduringthe dryseasonof2008.Thesetwogenotypesareusedin industrialplantationsintheRepublicofCongo.Hybrid 1-41(namedG1inthefollowingsections)wasobtained byopenpollinationof E.alba (themaleparentis unknown)andthehybrid18-50(namedG2)was derivedfromacontrolledpollinationof E.urophylla (genotype14-36)by E.grandis (genotype9-10).These twohybridsdifferintheirgrowthratesandwateruse efficiency(WUE,estimatedbyisotopiccarboncomposition)atmaturity,G2beingsuperiorthanG1.FieldexperimentTreeswerevegetativelypropagatedbyrootedcuttings andestablishedinafieldexperimentinYanika,Republic ofCongo(4 S,1138 E,50mabovesealevel),inJune 2007.Treeswereplantedinplotsof64cuttingsper genotypeandpertreatment,i ncludingabufferzoneof 40plants.Twowateringregimeswereusedduringthe dryseason:noirrigation(NI)andirrigation(IR).Trees werewateredwithsprinklers,toreplenishevapotranspirationlosses,estimatedat3mmperday.Inorderto evaluatetheeffectofwaterdeficitonabove-groundbiomassgrowthandmolecularplasticity,plantmaterialwasVillar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page2of18

PAGE 3

sampledinSeptember2008,16monthsafterthetrees wereplanted.Thedryseasonbeganapproximatelyon May15th2008-therefore,treesunderNItreatment weresubjectedtofourmonthswithoutrainfallbythe timesampleswerecollected.SoilwatercontentVolumetricwatercontent(VWC)wasmeasuredbytime domainreflectometry(TDR;T rasesystem,Soilmoisture,SantaBarbara,CA).FourseriesofTDRprobesper genotypeandpertreatmentwereinstalledhorizontally, atsixdepths(0.15,0.5,1,2,3and4m).Meanvalues werecalculatedfromthefourreplicatedmeasurements ateachdepth.BiomassproductionWeharvested11treespertreatmentanddissectedthem intothefollowingcompartments:stem,deadbranches, livingbranchesandleaves.Eachcompartmentwas weighedinthefield.Representativesubsamplesofeach compartmentwerethenharvested,andweighedbefore andafterdryingat65Ctoconstantweight.Watercontentwascalculatedforeachofthesesubsamplesand usedtoestimatetotaldrybiomassforeach compartment.Totalabovegroundbiomass(thesumof allthecompartment)wasanalysedbytwo-way ANOVA,withR(RDevelopmentCoreTeam),accordingtothefollowingmodel: Xi j k= + a Gi+ b T j + c ( G T )i j + i jk whereXijkistheabove-groundbiomassingenotype i (G1orG2),treatment j (NIorIR)andreplicate k.a,b and c aretheregressioncoefficientsof G ,thegenotypic effect, T ,thetreatmenteffectand G T theinteraction betweengenotypeandtreatment,and ijkistheresidual.cDNAsynthesisTheexperimentaldesign,fromtissuesamplingtolibrary constructionandsequencing,isdescribedinfigure1. Shootapiceswerecollectedfromthreetreesfromeach genotypeandtreatment,andimmediatelyfrozenin liquidnitrogen.TwoRNAextractionsofthreeapices fromeachtreewereperformed,aspreviouslydescribed byReid etal .[39].RNAsamplesweretreatedwith TurboRNase-FreeDNase(A mbion,Austin,TX,USA) andpurifiedwiththeRNeasyPlantMiniKit(Qiagen, Valencia,CA,USA).RNAconcentrationandquality wereanalysedwithanAgilentTechnologies2100 Irrigated (IR) Treatment 1 4118 3 trees3 t r 3 apices / tree 3 apices / tree 3 apices / tree Sample Sample Sample Genotype Field sampling RNA extraction cDNA library Sample 1 Sample 2 Sample 3constructionMultiplex 1 Ru n Run A 454 se q uencin g Tagging and multiplexing qg SS1: 1 41IR_A SS2: 18 50IR_A SS3: 1 41NI_A SS4: 18 50NI_A SS5: 1 4 1 SS6: 18 5 SS7: 1 4 1 SS8: 18 5 Sequencing sets Non Irrigated (NI) 50 1 41 18 50 r ees 3 trees 3 trees3 apices / tree 3 apices / tree 3 apices / tree 3 apices / tree 3 apices / tree Sample Sample Sample Sample Sample Sample 4 Sample 5 Sample 6 Sample 7 Sample 8 n B Run C Multiplex 2 1 IR_B 5 0IR_B 1 NI_B 5 0NI_B SS9: 1 41IR_C SS10: 18 50IR_C SS11: 1 41NI_C SS12: 18 50NI_C Figure1 Procedureused,fromtissuesamplingtosequencing .Twogenotypes(G1andG2)weresubjectedtotwowateringregimes(IR andNI).Shootapicesfromthreetreespergenotypeandpertreatmentwerecollectedinthefield.TotalRNAwasextractedfromthreeapices pertree.TworeplicateRNAextractionswerecarriedoutfortheconstructionoftwoindependentreplicatecDNAlibrariespergenotypeandper treatment(resultingin8templatesforcDNAlibraryconstruction).Forsequencing,eachcDNAlibrarywastagged,andtwoindependent multiplexeswerecreatedbypoolingonesampleforeachgenotypeandtreatmentcombination.Multiplex#1wassequencedwithtwo454RocheFLXTitaniumhalf-runs,resultingineightsequencingsets,whereasmultiplex#2wassequencedwithonehalf-runandresultedinfour sequencingsets. Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page3of18

PAGE 4

Bioanalyser(AgilentTechnologies,Mississauga,ON, USA)andaND-1000Spectrophotometer(NanoDrop, Wilmington,DE,USA).The threeRNApreparations perreplicateandpercondition(correspondingtothree trees),werepooledinequalproportions,providingtemplatesforcDNAlibrariesS1-S8(figure1).Full-length cDNAwasobtainedfrom1 gofRNA,withtheSmart cDNALibraryConstructionKit(Clontech,Mountain View,CA,USA),accordi ngtothemanufacturer s instructions.WeamplifiedthecDNAwithPCRAdvantageIIPolymerase(Clontech,MountainView,CA, USA),over16cycles(7sat95C,20sat66C,and4 minat72C).ThiscDNAamplificationprocedurewas repeatedeighttimesinseparatetubesforeachsample, withpoolingtogiveatotalof6 gofcDNAfragments longerthan1,000bpquantif iedwithanAgilentTechnologies2100Bioanalyzer.EightcDNAlibraries(S1-S8) wereconstructed,givingtwobiologicalreplicatesforthe twogenotypes(G1,G2)subjectedtothetwowatering regimes(IR,NI).Libraryconstruction,454-sequencing,sequencequality controlandassemblyWenebulised5 gofeachcDNAsampletoamean fragmentsizeof650bpandligatedittoanadaptor, accordingtostandardprocedures[40].EachcDNA librarywastaggedwithMultiplexIdentifiers(MID)barcodeadaptors,andtwoindependentmultiplexeswere createdbypoolingonesamplefromeachgenotypeand treatment.Multiplex#1comprisedsamplesS1,S3,S5 andS7,whereasmuliplex#2comprisedsamplesS2,S4, S6andS8.Onehalf-run(runA)ofsequencingwas initiallycarriedoutformultiplex#1onaGS-FLXTitaniumplatform(454LifeScience,Brandford,CT,USA) atCogenics(Meylan,France).Twohalf-runsofsequencingformultiplex#1(runB)andmultiplex#2(runC) werethenperformedbyAgencourt(Beverly,MA,USA) onaGS-FLXTitaniumsequencer.BasecallingwithGSFLXSystemsoftwaregenerated353,344high-quality readsforthefirsthalf-runandin785,322readsforthe secondcompleterun.Sequencesweredepositedatthe NCBIshort-readarchive(SRA)underaccessionnumber SRA012867.2(Figure2).The454-sequencingreads (1,138,666fromthisstudyand1,041,876fromNovaes et al .[13])werescreenedbycross_match(http://bozeman. mbt.washington.edu/phrap.docs/phrap.html)forprimers andadaptorsandthenmasked.Foreach454-sequencing read,thelongestnon-maskedregionwasextractedand furthertrimmedwithSeqClean(http://compbio.dfci.harvard.edu/tgi/).Theshorterregionswerediscardedto eliminatepotentialchimeras.Sequenceswereassembled aspreviouslydescribed[41],withTGICL[42],usingthe 12setsofsequencingdatafromthisstudyandthefour setsofsequencingdataobtainedfor E.grandis [accessionnumberSRA001122]byNovaes etal .[13].In parallel,allreadswerestoredintheNG6system(http:// vm-bioinfo.toulouse.inra.fr/ng6/,project:BIOGECO eucalyptus)andthreekindsofanalysiswereperformed foreachofthe16sequencingsets,aspreviously described[41]:i)BLASTsearchfor E.coli ,phageand yeastcontaminants,ii)readqualityanalysisandiii) removalofsequencesthatweretoolongortooshort, sequenceswithanexcessoferrors(morethan4%ofN), low-complexitysequencesandduplicatedreads,using Pyrocleanerprogram.Onlyunigeneelements(UE) resultingfromsequencesge neratedinthisexperiment (the E.spp sequencingset)wereusedfordigitalgene expressionanalysis.DigitalgeneexpressionanalysisContigswithlessthan10readsforthe12sequencing setsgeneratedinthisstudywereeliminatedfrom furtherstatisticalanalyses.Forthe14,460remaining contigs,thenumbersofreadspersequencingsetand percontigwereusedtoassessgeneabundance.Two typesofstatisticalanalysiswereperformed.First,pairwisecomparisonswerecarriedoutbetweengenotypes (G2vs.G1sequencingsets,irrespectiveoftreatment) andbetweentreatments(IRvs.NIsequencingsets,irrespectiveofgenotype).Fouradditionalcomparisonswere carriedoutforeachgenotypeandeachtreatment,as follows:IRvs.NIforgenotypesG1andG2,G1vs.G2 fortreatmentsIRandNI.Statisticaltests,basedonthe useoftheMARSmethodintheDEGseqpackage[43] wereperformedtoassessdifferentialexpression[43]. Second,two-wayanalysisofvariance(ANOVA)was performedoncontigs,makinguseofthethreereplicates (runA,BandC)pertreatmenttoestimaterandomvariationandtestthegenotype(G),treatment(T)andgenotype-treatmentinteraction(GxT)effects.Transcript abundancewasnormalizedbydividingthenumberof readsbythesequencelengthofthecontigsandthe totalnumberofsequencesineachsequencingset.Contigswithaq-value<0.05intheDEGseqtest(i.e.after falsediscoveryratecorrections)[44]andwithp-values <0.05afterANOVAwereconsideredtobedifferentially expressedandwereextracte dforfurtheranalysis.The 14,460genesanalysedwereclassifiedintofourclasses: notsignificantlydifferentiallyexpressed(NS),andshowinggenotype("G contigs),treatment("T contigs)or genotypetreatment("GxT contigs)effects.Forthe comparisonofexpressionlevels,weusedlog2-transformedfold-changesbetweencontigabundancesinthe variouscontrastsobtainedintheDEGseqanalysis.FunctionalannotationofdifferentiallyexpressedgenesContigswereassignedaputativefunctionbyBLASTX [45],usingvariouspublicdatabases:UniProtKB/Swiss-Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page4of18

PAGE 5

Prot(release57.1),RefSeqProtein(release34),Pfam (release23.0),withane-valuecut-offvalueof10e-5. SequenceswerealsocomparedtoTIGR sassembliesof Arabidopsis_thaliana(rel ease14),Helianthus_annuus (release6),Populus(release4),Picea(release3)and Vitisvinifera(release6),withane-valuecut-offof10e-2. GeneOntologytermswerea ssignedviatheUniprotKB accessionandclusteredwithBlast2GO[46].Thedifferentialdistributionsofeachclassofeffect(T,G,GxT andNS)betweenBiologicalProcesses,MolecularFunctionsandCellularComponentswereassessedusing Fisher sexacttests,withasignificancethresholdof0.05. PathwayanalysiswascarriedoutwithMapman[47]. Differentiallyexpressedcontigswereassignedtofunctionalcategories(orbins)byMercator(http://mapman. gabipd.org/web/guest/mer cator).Adedicatedpathway mapwascreatedtorepresentmostofthesecontigs. TheWilcoxonranksumtestwasusedtoidentifydifferentiallyregulatedbins.ResultsMonitoringofsoilwatercontentAfactorialdesignincludingtwogenotypes(G1andG2) andtwowaterregimes(irriga tedIRvs.non-irrigated NI)wasestablishedinafieldtrialinoneofthemain areasofeucalyptusplantationintheRepublicofCongo. Theexperimentwasevaluatedoveraperiodoftwo years.Soilwatercontent(SWC)wasmonitored throughouttheexperimentatsixdepths(0.15-4m),to assesswateravailabilityindifferentexperimentalconditions.Inthisstudy,wefocusontheeffectofwater availabilityintheseconddryseason(afterfourmonths withoutrainfall)onbiomassproductionandthe transcriptome. IntheNItreatment,SWCvariedfrom4.5%to8%, andnosignificantdifferencewasfoundbetweenthetwo genotypes(Figure3).SWCva lueswereclosetowilting point(pF4.2),i.e.,whenplantsceasedtobeableto absorbsoilwater.IntheIRtreatment,SWCranged E. spp (454 sff files)SRA0128672 Sequence extraction (sfffile) (454 sff files) SRA012867 2 1,138,666 reads Adaptor / primer detection (cross_matc Extraction of the longest sequence witho u Cleaning (Seqclean) TGICL Clustering (mgblast) 1,994,741 reads Cluster splitting (sclust) Assembly (CAP3) 231715lt UE 86 full length Eucalyptus cDNA GenBank Homology search (Blast) 231 715 euca l yp t us UE UniProtKB, RefSeq_rna, RefSeq_Protein, TIGR, GO Eucalyptus Contig Browser E. grandis* SRX000427(454 sff files) SRX000427 (454 sff files) 1,041,876 reads Contaminant detection ( Blast ) NCBI NG6 h) u t X () Read quality analysis Cleaning (Pyrocleaner) NG6 202,279 UE UE specific to E. spp library 49,748 contigs and 80,245 singletons 14,460 contigs Abundance analysis: # reads 10 DEG Seq R ANOVA Pairwisetests 1,651 differentiallyexpressedcontigs 1,445 differentially expressed contigs: 1280 G; 155 T; 274 GxT Functional analysis Visualisationof pathways Mapman BLAST2GO GO Comparison Figure2 Pipelineusedforsequenceanalysis .Whiteboxescorrespondtosequencegeneration,redboxescorrespondtolibraryquality control,assemblyandannotationandgreenboxescorrespondtoabundanceandfunctionalanalysis.Thesoftwaresuiteorprogramusedis indicatedintheupperpartoftheboxandtheresultsofeachstepoftheanalysisareshownunderthebox.UE:unigeneelement; G":genetic effect, T":treatmenteffectand GxT":genotypebytreatmentinteractioneffect.* E.grandis .sfffileswereobtainedfromNovaes etal.(2008). a : availableathttp://genotoul-contigbrowser.toulouse.inra.fr:9092/Eucalyptus/index.html b:availableathttp://vm-bioinfo.toulouse.inra.fr/ng6/,user= euca;password=euca33bio! Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page5of18

PAGE 6

from11%to17%,andexceededfieldcapacity(pF2.0), indicatingthatwaterwasnotalimitingfactorfortree growth.SWCwasalsohigherintheareasurrounding genotypeG2thaningenotypeG1,exceptattwodepths (1mand4m).Thisresultsuggeststhatthetwogenotypesabsorbwaterpreferentiallyfromdifferentdepths, possiblybecausetheirrootsystemdevelopsdifferently.EffectofwaterdeficitonbiomassproductionGenotype(p<0.0001)andtreatment(p<0.001)had significanteffectsonabove-groundbiomassproduction (stem,deadbranches,livingbranchesandleaves)(Figure 4).MeanbiomasswasmuchhigherforgenotypeG2 (12.8kg)thanforgenotypeG1(8.6kg)confirmingearlierfindings[48].Notsurprisingly,meanbiomasswas higherfortheIRtreatment(11.8kg)thanfortheNI treatment(9.7kg).Atwo-wayANOVAshowedthatthe GxTinteractioneffectwasn otsignificant(Additional file1),butrelativebiomasslosswasnonethelesslower forgenotypeG2(12%)thanforgenotypeG1(24%). TheseresultssuggestthatthegrowthofgenotypeG2is lessaffectedbywatershortagecomparedtoG1.Sequencingofthe Eucalyptus shootapextranscriptomeWechosetosequencetheshootapextranscriptometo studythemolecularresponseofeucalyptustowaterdeficit,duetoitsroleinshootorganogenesis[49].Shoot apiceswerepooledfromthreetreesinfoursetsofconditions(2genotypes2treatments).Threehalf-runs (A,B,C)of454-RocheFLXTitaniumsequencingprovided353,344(runA),405,223(runB)and380,099 (runC)reads,foratotalof1,138,666sequences(398 Mb).Themeanreadlengthwas334bpforrunA,369 bpforrunBand344bpforrunC(Table1).Reads wereslightlyshorterforrunA,withahigherabundance ofreadscomprisedbetween400-420bpinlength, whereasrunsBandCwerecharacterisedbyreadsof 460-480bpinlength(Addit ionalfile2).Toincrease contiglengthoftheassembly,wecombinedallthe reads(1,138,666)generatedinthisstudy( E.spp sequencingset)withotherGS-20andGS-FLX454reads (1,041,876reads)fromvariousorgansof E.grandis ( E. grandis sequencingset;[13]). Figure2summarisesthe variousstagesinsequenceanalysis,correspondingto dataqualitycontrol,readassembly,annotationand abundanceanalysis.Afterremovalofvectorandadaptor sequences,1,994,741readswereavailableforassembly. AssemblywithTGICLgenerated231,715unigeneelements(UE)comprising80,854contigsand150,861singletons.Removaloflow-qualitysequencesand duplicatedreadswiththe NG6platformresultedina totalof202,279UE(69,584contigsand132,690singletons),fromwhich129,993UE(49,748contigsand 80,245singletons)wereidentifiedinthe E.spp sequencingset(thisstudy)andusedfordigitalgeneexpression analysis(Table2).Contributionofthethreehalf-runreplicatesIntotal,90,579UE(70%of E.spp UE)didnothave sequencesimilaritytothe E.grandis sequencereads fromNovaes etal .[13].Mostofthesesequenceswere singletons(71,761),althoughsomewerecontigs (18,818),correspondingto27%ofthe E.spp contigs.A BLASThomologysearch(cut-off:10-10)ofpublished eucalyptusdatabases(ESTsfromGenBankavailableon April2010;454-ESTsgeneratedbyNovaes etal .,[13]; 454-ESTsfromJGIfrom E.globulus xylemandleaftissues;IlluminacontigsgeneratedbyMizrachi etal .,[38]) showedthat21,401UE(comprising3,066contigsand 18,335singletons)didnotmatchanyknownsequence. Thus,theresourcedescribedheregreatlyextendsthe 1 41 IR 18 50 IR 1 41NI 18 50NI 1 41 NI 18 50 NI p F2 p F4.2 Figure3 Profileofsoilwatercontent(SWC) .Meansandstandard errorsofSWCatsixdepths,forbothgenotypes(G1andG2), subjectedtotwowaterregimes(irrigated:IRandnotirrigated:NI). SWCatwiltingpoint(pF4.2)andatfieldcapacity(pF2)were calculatedfromdataobtainedatasiteclosetotheexperimental field(Laclau,personalcommunication). Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page6of18

PAGE 7

listofgenesknowntobeexpressedin Eucalyptus whichwillbecriticalfortheannotationofthegenome sequence.DuetothesmallernumberofreadsinrunA, thetotalnumberofUEincludingreadsgeneratedfrom runAwasalsosmaller(58,763)thanthatgenerated fromtheothertworuns(67,467inrunBand67,756in runC).Eachsupplementaryhalf-runproducedbetween 16and21%newcontigsforasecondhalf-run,and between5and7%forathirdhalf-run(Additionalfile 3).Vega-Arreguin etal .[50]reportedsimilartrendsin maize,withaplateauofgenerepresentationreached afterthethirdsuccessiveGS-20454-sequencingrun. Thenumberofreadsgeneratedwasthereforeconsideredsufficienttosamplemostexpressedgenes. ThePearson scorrelationcoefficients(figure5) obtainedforreadfrequenciesinthe12sequencingsets from E.spp (87%onaverage)andthefoursequencing setsfrom E.grandis (86%onaverage)weresimilar. 10 12 14 16biomass (kg)A A B C 0 2 4 6 8 G 1 IR G 1 NI G 2 IR G 2 NIAbove ground Figure4 Above-groundbiomass .Boxplotofabove-groundbiomasscalculatedfor11treespertreatment,forbothgenotypes(G1andG2), subjectedtotwowaterregimes(irrigated:IRandnotirrigated:NI).Centrelineandoutsideedgeofeachboxindicatethemedianandrangeof innerquartilearoundthemedian,respectively;verticallinesonthetwosidesoftheboxrepresentthefirstandtheninthdecile,respectively. LettersindicatethegroupsobtainedinBonferronitestsformultiplepairwisecomparisons. Table1Summarystatisticsforthethree454-sequencinghalf-runsSampleGenotypeTreatmentSequencing set Runs#of Reads #of Reads #ofbpAveragelengthof reads(bp) Averagelengthof reads(bp) 1G1IR58,92119,313,318328 3G2IRSS2:G2IR_A92,16530,365,092329 5G1NISS3:G1-NI_ArunA353,34495,50031,350,808328334 7G2NISS4:G2NI_A106,75837,126,644348 1G1IRSS5:G1IR_B139,13751,085,203367 3G2IRSS6:G2IR_BrunB405,223112,05141,110,040367 5G1NISS7:G1-NI_B59,90721,833,839364369 7G2NISS8:G2NI_B94,12835,374,789376 2G1IRSS9:G1IR_C93,65132,338,261345 4G2IRSS10:G2IR_C90,51131,595,445349 6G1NISS11:G1NI_CrunC380,09983,53027,929,481334344 8G2NISS12:G2NI_C112,40738,802,862345 TOTAL1.5runGS-FLX Titanium 1,138,6661,138,666398,225,782350350 Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page7of18

PAGE 8

However,themeancorrelationwasmuchweaker(52%) betweenthe E.spp and E.grandis sequencingsets,suggestingthatdifferentfractionsofthetranscriptomehad beensampledfromthesetwostudiesand/orthatgene expressiondifferedbetweenbothsequencingsets.As expected,correlationsweres trongerbetweenreplicates (92%;illustratedbysquaresinfigure5)thanbetween differentsamples(86% ) in E.spp sequencingsets,suggestingahighleveloftechnicalrepeatability. Withinconditions,corre lationsbetweentheG1and G2sequencingsetswererobustandsimilarbetween treatments:90%fortheIRtreatmentand88%forthe NItreatment,suggestingthatthesetwogenotypesdisplayedsimilarpatternsofgeneexpressionwhenplaced inthesameenvironmentalconditions.Correlations betweentheIRandNIsequencingsetswereslightly weakerandastrongercontrastwasobservedwithin genotypes:onaverage,86%forgenotypeG1and82% forgenotypeG2.Theweakercorrelationsobtainedfor G2suggestthatthisgenotypehadamorepronounced responsetowaterdeficitthangenotypeG1.HomologysearchBLASTsearch(e-valuecut-offof10e-5)resultsaresummarisedinTable3.Afunctionalannotationwasobtained morefrequentlyforcontigs-70%ofthecontigsharbouredsimilaritytosequencesinproteindatabases,and 75%tosequencesinnucleicaciddatabases.Incontrast, forsingletonsonly39%and9 %ofthecontigshadsimilaritytoproteinandnucleicacidsequences,respectively. Thesedifferenceswereexpected,giventhelongermean lengthofthecontigs(734bp)comparedtosingletons (319bp),theirlowerabundanceandthefactthatsingletonsaremorelikelytobesequencingartefacts.Alarger numberofnucleicacidsequenceshadsimilarityto Arabidopsis (55%ofannotatedsequences),followedby Vitis (25%),and Populus (6%).Thegreatersimilarityto Arabidopsis genesmaybeduetothecloserphylogeneticrelationshipof Eucalyptus to Arabidopsis (bothbelongtothe eurosidIIphylogeneticclade)thanto Populus (eurosidI, [51]).The Arabidopsis genomehasalsobeenannotated ingreaterdetailedthanthe Populus genome.Interestingly,thesimilarcharacteristicsofeucalyptusandpoplar intermsofgrowthhabitsdonottranslatedintohigher similarityofthesequencestranscribed. AccordingtoGeneOntology(GO)classification,38,190 UE(25%ofthe E.spp sequencingsetUE)wereassociated withatleastonebiologicalprocess(BP),molecularfunction (MF)orcellularcomponent(CC).TheproportionsofUE annotatedineachcategoryweregenerallysimilartothose obtainedin Arabidopsis (Additionalfile4),suggestingthat the E.spp sequencingsetsareappropriatefortheanalysisof geneexpressiononabroadrangeoffunctionalcategories.TranscriptabundanceanalysisAfterremovingcontigsrepresentedbyfewerthan10 readsinallthe E.spp sequencingsets,14,460contigs Table2AssemblystatisticsfromTGICL(I),andfigures obtainedafterPyrocleaneranalysis(II),forthesetof sequencesreportedhere(III)TGICL(I)Pyrocleaner (II) E.spp (III) #Contigs80,85469,58449,748 #readsincontigs1,843,8061,386,859851,751 Averagelengthofallcontigs (bp) 552608734 #LargeContigs>500bp34,07633,96232,694 Averagelengthoflargecontigs900901912 #Singletons150,861132,69580,245 #UE231,715202,279129,993 SS5:G1IR_B0.92SS9:G1IR_C0.880.93SS3:G1 NI_A0.870.850.85SS7:G1 NI_B0.880.890.900.94SS11:G1NI_C0.780.800.840.870.91SS2:G2IR_A0.890.890.890.900.910.82SS6:G2IR_B0.880.900.910.890.910.830.92SS10:G2IR_C0.870.910.930.870.910.830.920.93SS4:G2NI_A0.810.810.810.860.880.870.820.820.82SS8:G2NI_B0.810.840.830.860.890.890.820.830.830.94SS12:G2NI_C0.790.840.840.840.880.920.800.820.830.910.94SRR0016580.440.420.440.480.480.430.480.470.460.440.420.41SRR0016590.530.500.500.560.560.500.570.550.540.510.490.480.77SRR0016600.540.510.530.570.580.520.580.570.560.520.510.500.790.95SRR0016610.560.520.530.610.600.540.600.580.580.550.530.520.760.940.94SS1: G 1IRA SS5: G 1IRB SS9: G 1IR C SS3: G 1NIA SS7: G 1NIB SS11: G 1NI C SS2: G 2IRA SS6: G 2IRB SS10: G 2IR C SS4: G 2NIA SS8: G 2NIB SS12: G 2NI C SRR001658 SRR001659 SRR001660 Figure5 Pearsoncorrelationcoefficientforreadfrequencies(withincontigs)betweensequencingsets .SS1-SS12aresequencingsets from E.spp ;SRR001658-SRR001661aresequencingsetsfrom E.grandis (Novaes etal. ,2008).Thecolourscaleindicatesthestrengthofthe correlation,fromred(strongestcorrelations)togreen(weakestcorrelations).Allthecorrelationcoefficientsweresignificant,withap-value <0.001. Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page8of18

PAGE 9

remainedforabundanceanalysis.Twostatisticaltests wereperformedinseriestodetectdifferencesinthe expressionlevelsofthese14,460contigs,amongthe fourexperimentalconditions(G1IR,G1NI,G2IRand G2NI).First,aDEG-seqtest[43]identified1,651differentiallyexpressedcontigs(FDR 5%).Atwo-way ANOVAwasthenperformedtoassesstheeffectsofthe twomainfactors(GandT)andtheirinteraction(GxT) onthenumberofreadspercontig.Thisanalysisidentified1,445contigswithatleastonesignificanteffect(pvalue 5%;figure6,additionalfile5).Withanerror rateof5%,only83falsepositivesareexpectedamong the1,445contigs. Mostofthedifferentiallyexpressedcontigs(1,280) showedagenotypeeffect("Gcontigs ),with624 G contigs overexpressedingenotypeG1(positivelog2transformedfold-changebe tweencontigabundancein G1vs.G2)vs.656 Gcontigs underexpressedinG1( negativelog2-transformedfold-change;figure7A).Of the656contigsoverexpressedingenotypeG2,289 (44%)wereexpressedonlyinthatgenotype(withno correspondingreadsintheG1sequencingsets)whereas only55contigs(9%)showedthereversetrend,suggestingthatG2mayexpressofalargersetofgenesordifferentsplicingvariants. Atotalof155contigsshowingatreatmenteffect("T contigs";figure7B)wereidentifiedwithsimilarnumbers overexpressedinthetwotreatments(81inIRand74in NI).Thirteen Tcontigs wereexpressedonlyinNI conditions,whereasallthe Tcontigs overexpressedin IRconditionswerealsofoundinNIsequencingsets, suggestingthatfew specificgenes areupregulatedin responsetowaterdeficitbutthatthesetofgenes expressedinfavourableconditionsisalsoexpressedata lowerlevelinstressedplants. Finally,274contigscorrespondedto GxT contigs. Thelargernumberof GxT contigsthanof Tcontigs suggeststhatsomeoftheobservedmolecularplasticity isundergeneticcontrol.Only11 GxTcontigs displayedsignificantdifferentialexpressionbetweentheIR andNIconditionsingenotypeG1(figure7C),whereas 112 GxTcontigs displayedsuchbehaviouringenotype G2(figure7D),suggestingamorepronouncedresponse inG2.Similarly,48 GxTcontigs weredifferentially expressedbetweenthetwogenotypesinIRconditions (figure7E),whereas228 GxTcontigs weredifferentiallyexpressedbetweenthetwogenotypesfortheNI treatment(figure7F).Theseresultssuggestthat,despite therathersimilarexpressionpatternsforthetwogenotypesinIRconditions,waterdeficitinducedamolecular responsespecifictoeachgenotype,reflectingdifferent strategiestorespondtowatershortageduringthedry season.Blast2GOfunctionalanalysisGeneOntology(GO)analysiswasperformedon9,058 contigs(of14,460contigscontainingmorethan10 reads).Thesecontigswereassignedtobiologicalprocesses(BP),with11mainsubcategories(7,593contigs), cellularcompartments(CC),withfivemainsubcategories(7,347contigs),andmolecularfunctions(MF), witheightmainsubcategories(7,638contigs). Table3Annotationresultsforproteinhits,nucleicacidhits,andGeneOntologies(GO):BiologicalProcess(BP), CellularComponent(CC)andMolecularFunction(MF)E.spp ProteinhitsSSequencingsetNucleicacidhitsGO-BPGO-CCGO-MF #UE129,99366,135(51%)44,652(34%)32,835(25%)30,836(24%)33,222(26%) #Contigs 49,748 34,951(70%)37,383(75%)18,091(36%)17,242(35%)18,355(37%) #Singletons31,184(39%) 80,245 7,269(9%)14,744(18%)13,594(17%)1,4867(19%)%ofsequencesrelatedtoE.spp.librariesisshowninbrackets. 1066 100 41 9 49 124 56 9 Genotype*Treatment Genotype Treatmen t Figure6 Differentiallyexpressedcontigs .VenndiagramindicatingthenumberofdifferentiallyexpressedcontigsshowingaG,Tand/orGxT effect,forthe14,460contigsanalysed. Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page9of18

PAGE 10

Consistentwiththenumberofannotationspercategory (BP,CC,MF),twosubcategorieswerefoundtobe stronglyrepresented:>65%ofthecontigswereassigned tocellularandmetabolicprocessesforBP,morethan 40%wereassignedtocellsand85%toorganellesfor CC,and>40%wereassignedtobindingandcatalytic activityforMF.Figure8showsthedistributionofcontigsbetweenthesesubcategoriesasafunctionofthe fourclassesofeffects("NS G T ,and GxT contigs).Thehomogeneityoftherelativeabundanceofcontigsbetweenthe significant classes("G T ,or GxT )andthe notsignificant class("NS )ineachGO categorywasassessedwithFisher sexacttests.InBP, G contigswereoverrepresentedinfoursubcategories (responsetostimulus,developmentalprocess,deathand multiorganismprocess). T contigswereoverrepresentedinonlyonesubcategory(responsetostimulus). Finally, GxT contigswereoverrepresentedinthree subcategories(responsetostimulus,deathanddevelopmentalprocess).Thesedifferencesinrelativeabundance suggestthatgenesrelatedtodefencereactionsarethe maincontributorstodifferencesbetweensignificantand NS contigs.ForCC,onlyonesubcategory(extracellularregion)presentedahigherrelativeabundanceforall threesignificanteffects.ForMF, G contigswereoverrepresentedintwosubcategories(structuralmolecule activityandmoleculartransduceractivity),whereas T contigswereoverrepresentedinonlyonesubcategory contigsG effect : [1-41/18-50] G1 ( 624 conti g s ) A G effect : [G1/G2] -15 -10 -5 0 5 10FCDE GxT effect : 1-41[IR/NI] (g) G2 (656 contigs)C GxTeffect : G1[IR/NI] -4-3-2-10123DE-contigs IR (7 contigs) NI (4 contigs) FC DE-contigsGxT effect : IR[1-41/18-50] G1 (31 contigs) G2 (17contigs) E GxTeffect : IR[G1/G2] -8-6-4-20246FC (17 contigs) contigsT effect : [IR/NI] IR ( 81 conti g s ) B Teffect : [IR/NI] -8-6-4-20246FCDE GxT effect : 18-50[IR/NI] (g) NI (74 contigs)D GxTeffect : G2[IR/NI] -10 -5 0 5 10DE-contigs IR (57 contigs) NI (55 contigs) FC DE-contigsGxT effect : NI[1-41/18-50] G1 (103 contigs) G2F GxTeffect : NI[G1/G2] -10 -5 0 5 10FC (125 contigs) Figure7 Fold-change(FC)distribution .Log2-transformedFCwerecalculatedwithDEGseqandplottedforeachclassofeffect.(A)Genotype effect:contigsoverexpressedingenotypeG1arepresentedwithpositivevalues(+),whereascontigsunderexpressedingenotypeG1have negativevalues(-).Similarly,(B)treatmenteffects:overexpressedinIR(+)andunderexpressedinIR(-);(C)genotypetreatmenteffectfor genotypeG1:IR(+)andNI(-);(D)genotypetreatmenteffectforgenotypeG2:overexpressedinIR(+)andunderexpressedinIR(-);(E) genotypetreatmenteffectfortreatmentIR:overexpressedforG1(+)andunderexpressedforG1(-);(F)genotypetreatmenteffectfor treatmentNI:overexpressedforG1(+)andunderexpressedforG1(-).Thenumberofcontigsoverexpressedisshowninbrackets. Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page10of18

PAGE 11

(moleculartransduceractivity);for GxT contigs,two othersubcategories(catalyticactivityandantioxidant activity)presentedhigherrelativeabundancesthan NS .AnalysisofmetabolicpathwayswithMapManOfthe1,445contigsdisplayingsignificantdifferential expression,the1,280 Gcontigs didnotenablecharacterisationofspecificmol ecularprocesses(i.e.didnot showanyclearco-regulationwithgenesofthesame biosynthesispathway).For 95%ofthesecontigs,many differentgenesfromdifferentmolecularprocesses wereactivated,dependingontheexperimentalcondition(Additionalfile6).Insomeinstances(5%of G contigs),somebinspresentedspecificoverexpression inonegenotype(Additionalfile7).Forexample,contigsrelatedtoethylenebiosynthesisandcellorganisationwereoverexpressedmainlyingenotypeG2, whereascontigsrelatedtop hotosynthesis,nitrilases, calciumsignallingandpathogenesis-relatedprotein binswereoverexpressedingenotypeG1.Serineproteases(9contigs)wereexpressedmorestronglyingenotypeG1(p-value=0.038),whereasubiquitinE3encodingproteins(23contigs)wereexpressedmore stronglyingenotypeG2(p-value=0.096),suggesting thatproteolysisoccurredviadifferentpathwaysinthe twogenotypes. Analysisofthemetabolicpathwaysfor T contigswas limitedbecauseofthesmallnumberofcontigs(155), whichweredistributedinseveralbins(Additionalfile 8).However,theexpressionofgenesrelatedtocarbohydratedegradationandethylenebiosynthesiswerefound tobestrongerfortheNItreatment,whereastheexpressionofgenesrelatedtoribosomalproteinsynthesisand celldevelopmentappeared tobestrongerfortheIR treatment(Additionalfile7). Differentpatternswereobservedfor GxT contigs (figure9):i)Somepathways(25%of GxT contigs)displayedsimilarpatternsinthetwogenotypes,butwith responsesofdifferentmagnitudes(i.e.scaleplasticity,as definedbyLynchandWalsh[52]).PhotosystemcomponentstendedtobeoverrepresentedingenotypeG2in theIRcondition,whereasfewdifferenceswereobserved betweenconditionsforgenotypeG1.Conversely,genes relatedtocellorganisationandPR-proteinsweremore likelytobeoverexpressedintheNIconditioningenotypeG2comparedtogenotypeG1,ii)Interestingly, mostofthepathways(75%of GxT"contigs)displayed oppositetrendsinthetwogenotypes(i.e.are-ranking 80% 90% 100 % ntigs * * 30% 40% 50% 60% 70% v e number of co* * * ** 0% 10% 20% 30% c ess c ess t ion u lus tion t ion c ess esis c ess e ath c ess cell n elle p lex m en g ion d ing i vity vity vity i vity i vity vity i vity Relati v * * * * * * * * * * cellular pro c metabolic pro c localiza t response to stim u biological regula m ponent organiza t velopmental pro c o mponent biogen r organismal pro c d e u lti organism pro c orga n r omolecular com p r ane enclosed lu m extracellular re g bin d catalytic act i transporter acti u ral molecule acti ion regulator act i a r transducer act i antioxidant acti e ctron carrier act i GO u b categories cellular co m de cellular c o multicellula m u mac r memb r struct u transcript molecul a el e BiologicalProcess CellularCompartment MolecularFunction s u GO c ategories Biological Process Cellular Compartment Molecular Function NS G DE T DE(6850) (636) (93) c NS G DE T DEClasses of effect NS G DE T DE(6930) (598) (86) (6679) (575) (75) GxT DE(149) GxT DE GxT DE(142) (130) Figure8 DistributionofGeneOntologycategoriesbetweeneffectclasses .PercentageofcontigsannotatedforBiologicalProcess,Cellular ComponentandMolecularFunctionGOcategory(level2).Non-significantcontigs(NS)andcontigsdisplayinggenotype(G),treatment(T),or genotypetreatment(G*T)effectsareshown.ThenumbersinbracketsarethenumbersofannotatedcontigsperclassofcontigsandGO category.*IndicatesdistributionssignificantlydifferentbetweenNSandotherclassesofeffect,inFisher sexacttestswithathresholdof0.05. Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page11of18

PAGE 12

interactioneffect[52]).Forexample,intheNItreatment,genesrelatedtopheny lpropanoidbiosynthesis, auxinbiosynthesis,heatstress,andlightsignallingwere overexpressedingenotypeG1,whiletheywhereunderexpressedingenotypeG2.Conversely,genesrelatedto globalprimarymetabolism(particularlystarchdegradationandribosomalproteinsynthesis)andreceptor kinaseswereoverexpressedingenotypeG2,butunderexpressedingenotypeG1intheNIcondition(additional file7).Thisresultstronglys uggeststhatdifferentmetabolicpathwaysandgeneswereactivatedinthesetwo genotypesinresponsetowatershortage.Thus,thesetwo genotypesexhibitdifferentmolecularstrategiestocope withwaterdeficitduringthedryseason.DiscussionGenediscoveryandexpressionanalysisbyNGSNext-generationsequencing(NGS)technologiesare becomingthemethodofchoiceforlarge-scaletranscriptomeanalysis,evenfornon-mo delspecies(e.g.[29,41], reviewedin[53]).Severaltechnologieshavebeendeveloped,differingessentiallyinthenumberofreadsgeneratedandreadlength(reviewedin[54]),makingit possibletocataloguethegenesexpressedandtomonitorgeneexpression. In Eucalyptus ,Mizrachi etal .[38]generated3.93Gbp ofshortreads(36-60bp)withsequencing-by-synthesis technologyfromIllumina,andassembledthis information denovo into18,894contigs(Illumina-contigs)longerthan200bp(22.1Mbpintotal).Inthis study,weobtained0.398Gbpofsequenceswithlonger reads(meanof350bp),whichwereassembledinto 48,950contigs(454-sequencingcontigs)withmorethan 200bpeach(36.5Mbpintotal).Wewerethusableto assemblemorereads,i.e.9.2%ofthesequencingset, thatisamuchhigherratethanthe0.56%reportedby Mizrachi etal .[38].BLASTsearchesforsequencesimilaritiesbetweenthetwodatasetsshowedthat86%ofthe contigswerecommontobothstudies(42,550454sequencingcontigsmatched16,278Illuminacontigs). However,eachIlluminacontigmatchedameanoffive 454-sequencingcontigs,indicatingthatUEdetected withourapproachwereprobablyconfoundedinthe short-readassembly.Inaddition,theshortIlluminacontigsmayrepresentdomainssharedbymultipleproteins, confirmingthedifficultyinvolvedinassemblingshort readsintotranscriptionalunits[55].Alternativelywe cannotruleoutthefactthatgenesweresplitinmultiplecontigsinthe454assemblybecauseofthelackof coveragecomparedtoIllumina sshortreads.Finally,we foundthat36,985454-UEdidnotmatchanypreviously describedeucalyptusESTs,andthat43%oftheseUE (18%ofthe454-sequencingcontigs)displayednomatch withanynucleicacidorproteinsequencepublishedfor anyotherspecies.Therefore,our454-sequencingdata considerablyenrichedthe Eucalyptus ESTcollection. Figure9 Distributionof GxT contigsbetweenmetabolicpathways .Eachsquarerepresentsthelog2-transformedfold-changeof abundancebetweenirrigated(IR)andnon-irrigated(NI)treatmentsforonecontig:forgenotypeG1(A)andG2(B).Contigsinbluewere overexpressedfortheIRtreatmentandcontigsinredwereoverexpressedfortheNItreatment. Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page12of18

PAGE 13

RNA-Seqisalsoaninterestingapproachtoobtaina comprehensivedigitalgeneexp ressionprofileforspecifictissues,celltypesordevelopmentalprocesses.Inthis study,thehighdegreeofrep eatabilityobservedforthe threereplicatesmadeitpossibletotestG,TandGxT effectswithastatisticalsupport.Wewereabletomonitortheexpressionof14,460UEandtoidentify1,445 UEdisplayingatleastonesignificanteffect.Sometechnicalbiases,suchasnon-linearamplificationandalack ofsequencingdepthmayhaveresultedinalackofprecisioninthepredictionofgeneexpressionby454sequencing(Additionalfile9).Shortread-basedsequencingapproachesprovideamplereadcoverageandgenerallygivebetterpredictionsofgeneexpression[56,57]. Thus,acombinationoflongandshortreadsmaybe seenasareasonablestrategyfortheanalysisofgene expression[55,58,59].Wit hthiscombinedstrategy, long-readsequencingcanbeusedtoestablishacomprehensivecatalogueoftranscriptionalunits,whileshort readsmappedontothisassemblyprovidegreater sequencingdepth,improvingpredictionsofabundance.Behaviourofvariancecomponentsatthephenotypicand molecularlevelsWeobservedthatphenotypicandmolecularvariation areaccountedforprincipallybygenotypicdifferences. Indeed,above-groundbiomassandcontigabundance wereinfluencedprincipallybygenotype.Above-ground biomasswas,onaverage,49%higherforG2thanfor G1,andmostofthedifferentiallyexpressedcontigs (1,280of1,455)inthisstudypresentedagenotypic effect.Thenumberofcontigsoverexpressedinoneor theotherofthegenotypeswassimilar-624and656 contigswereoverexpressed inG1andG2,respectively. Noparticulargenotypicsignatureintermsoffunctional categoriesorpathwayswasdetected.Thetwogenotypes differedstronglyinphenotype(notonlyittermsof growthpotential,butalsointermsofleafmorphology, stomataldistributionandwateruseefficiency),butit remainsunclearwhetherthesedifferencesintranscript abundancewereresponsiblefortraitvariation,neutral, orsimplyinvolvedinreprod uctiveisolationbetween parentalspecies.Indeed,differencesingeneexpression betweenspecieshavebeenreportedinthefieldofecologicalgenomics[60,61]andinterpretedasamechanism ofspeciation.Inourstudy,bothgenotypeswerehybrid combinationsbetweendifferentspecies.Thismayhave increasedthenumberofdifferencesbetweentheirtranscriptomes.Furtherinvestigationsabouttheroleofgene expressioninecologicalspeciationisanimportantquestion,particularlyforeucalyptus,inwhichspeciescomplexesarecommon[62]. Thevarianceaccountedforbygenotype-by-environmentinteraction(GEI)atthephenotypicandmolecular levelwasalsosignificantinthisstudy.TheNItreatment resultedinasignificantlylowerabove-groundbiomass, andthisdifferencewasgreaterfortheleastproductive ofthetwogenotypes,G1(24%),thanforG2(12%).The GandTeffectsaccountedfor56%and13%ofthe above-groundbiomassvariation,respectively,whereas GxTeffectsaccountedforonly0.2%ofthevariance.At themolecularlevel,wealsofoundahigherproportion ofgenesdisplayingGeffects(1,280contigs,8.8%ofthe contigsscreened)but,surprisingly,wefoundfewer genesdisplayingTeffects(155contigs,1.1%ofthecontigsscreened)thanGxTeffects(274contigs,1.9%ofthe contigsscreened). Whileonly11contigsweredifferentiallyexpressed betweenthetwotreatmentsingenotypeG1,112contigs showeddifferentialexpressioningenotypeG2(4contigs displayeddifferentialexp ressionforbothgenotypes). Moreover,whenthewhole E.spp sequencingsetwas screened,genotypeG2presentedalargernumberof specificcontigs(10.5%ofthe E.spp sequencingset)than G1(5.5%ofthe E.spp sequencingset).Theseresults suggestthatalargersetofgenesisactivatedingenotype G2,leadingtothetriggeringofspecificresponsesto waterdeficit.ThishighermolecularsensitivityofgenotypeG2mayconferadvantagesultimatelyresultingina greatercapacitytocopewithwaterdeficitduringthe dryseasonand,therefore,instrongergrowthcapacity (table4).GenesdisplayingGEIeffectsreflectdifferencesinsignal perceptionandresponsestrategyContigsdisplayingGEIeffectscouldbeclassifiedinto twogroupsaccordingtothefunctionoftheproteins encoded[63]:i)regulatoryproteinsresponsiblefor droughtsignaltransductionandresponsetriggering,and ii)functionalproteinsinvolvedincellprotection, damagerepairandthemaintenanceofcellactivity. Regardingregulatoryfunctions,thegenesinvolvedin thebiosynthesisofhormones,suchasethyleneandauxinsinparticular(aldo/ketoreductase,proteinsofthe ethylene-responsivefamily,2-oxoglutarate-dependent dioxygenase)weremostlyoverexpressedingenotypeG1 andunderexpressedinG2,intheNIcondition.Genes actingassecondmessengersinthetransductionofhormonalsignaltostomatalguardcells[64,65]alsodisplayedGEIeffects:intheNIcondition,genesinvolved incalciumsignallingwerepredominantlyoverexpressed inG1,whereastheresponseofG2todroughtpreferentiallyinvolvedreceptorkinases.Wealsoidentifiedother signaltransducers,suchas light-inducedproteinsand heat-shockproteins,whichmayberelatedtoother typesofstressinducedbywaterdeficit,includingosmoticstressduetopHvariationsandoxidativestressdue totheaccumulationofreactiveoxygenspecies(ROS)Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page13of18

PAGE 14

[66,67].Thesepathwaysweremostlyoverexpressedin G1andunderexpressedinG2duringthedryseason. Pathogenesis-related(PR) proteingeneswereoverexpressedinG2,butdisplayedalessclear-cutpatternof expressioninG1.PRproteinswereinitiallyreportedto beinducedbyhormonesorROSinresponsetobiotic stress[68,69],buttheyhavealsobeenshowntobe involvedinotherabioticstresses[70].Lee etal .[69] alsosuggestedthatPRproteinsmaybeusedasstorage proteinswhengrowthislimitedbyenvironmentalfactors.Sometranscriptionfactorsrespondedstronglyto watershortageingenotypeG2.Two,inparticular, encodedfactorshomologousto AtMYB12 and AtMYB85 ,whichhavebeenshowntoregulatesecondarymetabolism(flavonoidandligninbiosynthesis, respectively)in Arabidopsis [71].Theseresultssuggest thatwatershortageinducesdifferentcellularstresscascades,perceiveddifferentlybythetwogenotypes. Stresssignaltransducersinteracttotriggertheregulationofgeneexpressionforthemaintenanceofthree mainfunctions:cellprotection,damagerepairandthe maintenanceofcellactivity.Ourresultssuggestthat moregenesrelatedtocellprotectionwereinvolvedin theresponsetowatershortageingenotypeG1thanin genotypeG2.Protectionagainstdroughtstressseemsto involvemostlycarbohydrates,with11contigsdisplaying GEIeffects,and,toalesserextent,polyamines,which maymodulatesomeionchannels[72]. Bycontrasttothetrendsobservedforgenesrelated tocellprotection,moregenesrelatedtodamagerepair seemtobeexpressedduringthedryseasoningenotypeG2thaningenotypeG1,particularlythoserelated tocellorganisation.Theoverexpression,duringNI treatment,ofgenesrelat edtoprimarymetabolism, includingcarbohydrate,lipidandproteinsynthesisand degradation,suggeststhatresourcesarereallocatedfor therepairofcellstructuresortheformationofnew structuresindroughtstressedplants.Thepatternsof expressionofsecondarymetabolismgenesdiffered betweenthetwogenotypesaswell.Asanexample,G1 displayedahighernumberofgenesrelatedtoterpenoidsandflavonoidssynthesis(thatmayprotect againstoxidativestress)overexpressedintheNItreatmentcomparedtogenotypeG2.However,thecontrastsbetweenNIRandIRtreatmentsweremuch higherforG2.Conversely,genesrelatedtoligninbiosynthesis(e.g. CCoAOMT )wereoverexpressedinthe IRconditiononlyforG2. Generelatedtophotosynthesiswerefoundtobe under-expressedinG2subjectedtoNItreatment, whereasnovariationwasfoundinG1.Othermetabolic processes,suchcelldevelopmentandtransport,controlledbygenesencodingwaterorsugarchannels, decreasedinNItreatment,particularlyinG2.These resultsconfirmthetrendsobservedbyAlexandersson et al .[73]in Arabidopsis.Theseauthorsstudiedtheexpressionof18genesencodingaquaporinsandshowedthat mostofthesegenesweredownregulatedinleavessubjectedtoagradualwaterdeficit.Interestingly,cellactivity seemedtobemorereducedatthetranscriptionallevel forG2,althoughthisgenotypegrewmorestrongly.Itis possiblethat,duringwaterdeficit,genotypeG2reduces itsrateofphotosynthesisandreallocatesresources(as suggestedbychangesinprimarymetabolism)topreserve itscellstructuresandabilitytoresumegrowthwhenconditionsbecomemorefavourable.EvolutionaryimplicationbehindGEIWefoundthat31ofthe274contigsdisplayingaGxT effectwereabsentfromtheG1sequencingset(11.3%), Table4Summaryofphenotypicandmolecularplasticityevidencedforthetwostudiedgenotypes(G1andG2), betweenirrigated(IR)andnonirrigated(NI)treatmentsG1 G2 IR NI IR NI Above-ground biomass decrease stable Transcript abundance Hormones + Secondarymessengers +(lightsignalling) (lightsignalling)++(receptorkinases) Otherstressrelated genes + (heat-shockproteins)++(PRproteins) Transcriptionfactors stable ++ Cellularprotection + Damagesrepair +(carbohydratessecondary metabolism), ++(cellorganisation,lipids,proteins) (secondary metabolism) Cellularactivity maintenance stable (photosynthesis,transport)Abbreviations:+/-meansthatinagivencategorymorethan50%ofthegenesareover/underexpressedinNI.++/ meansthatinagivencategorymorethan 50%ofthegenesareover/underexpressedinNIwithalog2-tranformedfoldchange<-0.8or>0.8.Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page14of18

PAGE 15

whereasonlytwosuchcontigswereabsentinG2(0.7%). Unfortunately,of31contigsabsentinG1,16couldnot beenassignedtoahomologgenein Arabidopsis .The otherscorrespondedtogenesrelatedtocellorganisation (ankyrins),ethylenesynthesis,proteinmetabolism,PR proteinsandreceptorkinases.Thesegenesmaybeconsiderednon-essentialfortreedevelopment,andare thereforeunlikelytobesubjecttoselectionconstraints. Landry etal .[74]foundanoverrepresentationofnonessentialgenes(thedeletionofwhichisnotlethal) amonggenesdisplayingGEIin Saccharomycescerevisiae .Theyproposedtwohypothesestoaccountforthe activityofthesegenesbeingcompensatedincells:i) metabolicbuffering:non-codedmetabolitesmaybe reroutedthroughthemetabolicnetwork,andii)genetic buffering:paralogousgenesmaysupplythemissing function.Weshowedintheresultssection(Additional file5)thatdifferencesbetweengenotypesmaybe accountedforbythepreferentialexpressionofdifferent membersorsplicingformsofgenesfromthesame family.Thisobservationmayconfirmthehypothesisof geneticbuffering. Scaleplasticitywasobservedfor146ofthe274differentiallyexpressedcontigs:genotyperankswereconservedbetweentreatmentsbutonegenotypereacted moreorlessstronglytotheenvironmentalvariation. Conversely,90contigsshowedachangeinranking betweengenotypes(rankplasticity).Landry etal .[74] hypothesisedthattheset wotypesofGEIwouldhave differenteffectsontheevolutionofplastictraits.Inthe caseofscaleplasticity,whatevertheenvironment,selectionwouldresultinthesamefavouredgenotype, whereasinthecaseofrankplasticity,differentgenotypeswouldbeselectedindifferentenvironments.In thepresentstudy,wefounddifferentiallyexpressed genespresentingbothscaleplasticity(62%)andrank plasticity(38%),indicating differenttypesofreaction normsonwhichnaturalselectionwouldacton.ConclusionsWeshowedthatnext-generationsequencingisapowerfultoolfortranscriptomescreening:with398Mbof sequence,wewereabletoassembleESTsinto69,584 contigs,withremaining80,245singletons,andtodeterminetherelativeabundanceof14,460contigseach comprisingmorethan10reads.Largedifferences betweengenotypes,intermsofphenotypicbehaviour andtranscriptomeregulation,wereobservable.Differencesingeneexpressionbetweenthetwogenotypes appeartoaffectthewholetranscriptome,ratherthan specificpathways.Thegenotype-specificresponseto watershortage(i.e.GxTeffect)wasmorepronounced thantheresponsecommontobothgenotypes(i.e.T effect).Thegenesdisplayinggeneticallycontrolled plasticitywerefoundtobelongtoanumberofdifferent pathwaysessentiallyrelatedtosignaltransductionand primarymetabolism.Themoreproductivegenotype, G2,expressalargersetofgenes,leadingtothetriggeringofspecificmolecularresponses.Moreover,GxT interactionresultsprincipallyfromalackofmolecular responseingenotypeG1,togetherwithastrong responseofgenotypeG2(table4).Theabilitytoregulatemoreactivelyitstranscriptomemightbeakeycomponentinthemaintenanceofbiomassinwaterdeficit conditions. Finally,althoughthisstudyprovidescluestotheway inwhichdifferentgenotypesactivatetheirtranscriptomeswhensubjectedtowaterdeficit,moreresearchis requiredtounderstandthemolecularmechanisms involvedduringthedryseason.First,thereisaneedto characterizereactionnorminabroadergeneticbackground[75].Second,epigenet icsorpost-transcriptional regulationmechanismsthatarewellknowntointerfere withabioticstressresponses[76,77]deservespecific investigations.AdditionalmaterialAdditionalfile1:ResultsoftheANOVAforabove-groundbiomass Additionalfile2:Distributionofreadlengthforthethreehalf-runs Additionalfile3:Increasingcoveragewithsuccessiveruns .Number ofcontigsrepresentedineachhalf-runorcombinationofseveralhalfruns.Performingasecondhalf-runincreasedcontigcoveragebyan averageof18%,andathirdhalfrunincreasedcoveragebyanaverage of6%. Additionalfile4:ComparisonofthedistributionofGeneOntology (GO)categoriesbetween Eucalyptusspp unigeneelements(UE)and Arabidopsis annotatedunigenes .ProportionofeachGOcategory (BiologicalProcess,CellularComponentandMolecularFunction)found inthe E.spp sequencingsetandintheannotated Arabidopsis genome. Additionalfile5:Differentiallyexpressedcontigsdisplaying significantTandGxTeffects .EachcontigshowingTorGxTis associatedwithafunctionalcategory(asdefinedbyMercator:http:// mapman.gabipd.org/web/guest).Log2-transformedfoldchangebetween abundanceinirrigatedandnonirrigatedlibrariesareindicatedforeach genotype,aswellasp-valueofTandGxTeffectsanalyzedbythe ANOVA. Additionalfile6:Distributionof G contigsbetweenfunctional pathways .Eachsquarerepresentsthelog2-transformedfold-changeof abundancebetweengenotypes1-41and18-50foronecontig.Contigs ingreenwereoverexpressedingenotype1-41andcontigsinbluewere overexpressedingenotype18-50. Additionalfile7:SignificantcategoriesfortheWilcoxonranksum test,accordingtoMapmananalysisforthepairwisecomparisonof differentiallyexpressedcontigsdisplayinggenotype(G),treatment (T)andgenotypetreatment(GxT)effects .**Categoriesdifferentially expressedatanerrorratethresholdof0.05*Categoriesdifferentially expressedatanerrorratethresholdof0.1 Additionalfile8:Distributionof T contigsbetweenseveral functionalpathways .Eachsquarerepresentsthelog2-transformedfoldchangeofabundancebetweenirrigated(IR)andnon-irrigated(NI) treatmentsforonecontig.ContigsinbluewereoverexpressedfortheIR treatmentandcontigsinredwereoverexpressedfortheNItreatment.Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page15of18

PAGE 16

Additionalfile9:Supportinginformation:validationofdigital profilesbyanalyzingexpressionbyRT-qPCRon36genes AcknowledgementsandFunding WethankAndrasNdeko,AndrMabiala,JolPolidoriandDr.AubainR. Saya(CRDPI,Rep.ofCongo)forsettinguptheexperimentaltrial.Thisarticle ispartofthePhDthesisofEmilieVillar,supervisedbyJean-MarcGionand ChristophePlomion.EVwassupportedbyCIRAD.Thisworkwassupported bygrantsfromCIRAD,ATPproject Plasticitphnotypiquedesprennes souscontraintehydriqueauchamp ,andthe ConseilRgionald Aquitaine ("ABIOGEN projectno.Presage32973).Someoftheexperiments(RT-qPCR) presentedherewereperformedattheGenomeTranscriptomefacilityof Bordeaux(grantsfromthe ConseilRgionald Aquitain nos.20030304002FA and20040305003FAandfromtheEuropeanUnion,FEDERno.2003227). Authordetails1CIRAD,UMRAGAP,CampusdeBaillarguetTA10C,F-34398Montpellier Cedex5,France.2INRA,UMR1202BIOGECO,F-33610Cestas,France.3CRDPI, BP1291,PointeNoire,RpubliqueduCongo.4Plateformebioinformatique Genotoul,UR875BiomtrieetIntelligenceArtificielle,INRA,31326CastanetTolosan,France.5SchoolofForestResourcesandConservation,Universityof Florida,POBox110410,Gainesville,USA.6UniversidadeFederaldeGois, CaixaPostal131,CEP74690-900,Goinia,Brazil.7UniversitdeBordeaux, UMR1202BIOGECO,F-33610Cestas,France. Authors contributions EV:participatedinthefieldwork,performedthemolecularwork,carriedout thestatisticalanalysisanddraftedthemanuscript.CKandCNperformedthe bioinformaticwork(assemblyandannotation).ENandMKprovided E. grandis sequencingsetsandparticipatedinthestatisticalanalysis.CP participatedinco-ordinationofthemolecular,bioinformaticandstatistical workandindraftingthemanuscript.JMGparticipatedinthedesignofthe study,co-ordinationofthefield,molecular,bioinformaticandstatisticalwork anddraftingthemanuscript.Allauthorsreadandapprovedthefinalversion ofthemanuscript. Received:19July2011Accepted:2November2011 Published:2November2011 References1.KrutovskiiKV,NealeDB: Forestgenomicsforconservingadaptivegenetic diversity. ForestGeneticResources 2001,6-8. 2.HamrickJL: Responseofforesttreestoglobalenvironmentalchanges. ForestEcologyandManagement 2004, 197 :323-335. 3.JumpAS,PenuelasJ: Runningtostandstill:adaptationandtheresponse ofplantstorapidclimatechange. EcologyLetters 2005, 8 :1010-1020. 4. Eucalyptusuniversalis.Globalcultivatedforestsmap. 2008[http://www. git-forestry.com/]. 5.McGowenMH,WiltshireRJE,PottsBM,VaillancourtRE: Theoriginof Eucalyptusvernicosa,auniqueshrubeucalypt. BiologicalJournalofthe LinneanSociety 2001, 74 :397-405. 6.JonesRC,SteaneDA,PottsBM,VaillancourtRE: Microsatelliteand morphologicalanalysisofEucalyptusglobuluspopulations. Canadian JournalofForestResearch-RevueCanadienneDeRechercheForestiere 2002, 32 :59-66. 7.HolmanJE,HughesJM,FenshamRJ: AmorphologicalclineinEucalyptus: ageneticperspective. MolecularEcology 2003, 12 :3013-3025. 8.TripianaV,BourgeoisM,VerhaegenD,VigneronP,BouvetJM: Combining microsatellites,growth,andadaptivetraitsformanaginginsitugenetic resourcesofEucalyptusurophylla. CanadianJournalofForestResearchRevueCanadienneDeRechercheForestiere 2007, 37 :773-785. 9.PaynKG,DvorakWS,JanseBJH,MyburgAA: Microsatellitediversityand geneticstructureofthecommerciallyimportanttropicaltreespecies Eucalyptusurophylla,endemictosevenislandsineasternIndonesia. TreeGenetics&Genomes 2008, 4 :519-530. 10.ButcherPA,WilliamsER: Variationinoutcrossingratesandgrowthin EucalyptuscamaldulensisfromthePetfordRegion,Queensland; Evidenceofoutbreedingdepression. SilvaeGenetica 2002, 51 :6-12. 11.ShepherdM,SextonTR,ThomasD,HensonM,HenryRJ: Geographicaland historicaldeterminantsofmicrosatellitevariationinEucalyptuspilularis. CanadianJournalofForestResearch-RevueCanadienneDeRecherche Forestiere 2010, 40 :1051-1063. 12.McKinnonGE,PottsBM,SteaneDA,VaillancourtRE: Populationand phylogeneticanalysisofthecinnamoylcoAreductasegenein Eucalyptusglobulus(Myrtaceae). AustralianJournalofBotany 2005, 53 :827-838. 13.NovaesE,DrostDR,FarmerieWG,PappasGJ,GrattapagliaD,SederoffRR, KirstM: High-throughputgeneandSNPdiscoveryinEucalyptusgrandis, anuncharacterizedgenome. BmcGenomics 2008, 9 :312. 14.KulheimC,YeohSH,MaintzJ,FoleyWJ,MoranGF: ComparativeSNP diversityamongfourEucalyptusspeciesforgenesfromsecondary metabolitebiosyntheticpathways. BmcGenomics 2009, 10 :452. 15.SultanSE: Phenotypicplasticityforplantdevelopment,functionandlife history. TrendsinPlantScience 2000, 5 :537-542. 16. StapeJL,BinkleyD,RyanMG: Productionandcarbonallocationina clonalEucalyptusplantationwithwaterandnutrientmanipulations. ForestEcologyandManagement 2008, 255 :920-930. 17.CampionJM,NkosanaM,ScholesMC: BiomassandNandPpoolsin above-andbelow-groundcomponentsofanirrigatedandfertilised EucalyptusgrandisstandinSouthAfrica. AustralianForestry 2006, 69 :48-57. 18.EylesA,PinkardEA,MohammedC: Shiftsinbiomassandresource allocationpatternsfollowingdefoliationinEucalyptusglobulusgrowing withvaryingwaterandnutrientsupplies. TreePhysiology 2009, 29 :753-764. 19.TatagibaSD,PezzopaneJEM,ReisEFd,PenchelRM: Performanceofsix clonesofeucalyptusinresponsetosubstratewateravailability. EngenharianaAgricultura 2009, 17 :179-189. 20.ShvalevaAL,SilvaFCE,BreiaE,JouveL,HausmanJF,AlmeidaMH, MarocoJP,RodriguesML,PereiraJS,ChavesMM: Metabolicresponsesto waterdeficitintwoEucalyptusglobuluscloneswithcontrasting droughtsensitivity. TreePhysiology 2006, 26 :239-248. 21.PitaP,PardosJA: Growth,leafmorphology,wateruseandtissuewater relationsofEucalyptusglobulusclonesinresponsetowaterdeficit. Tree Physiology 2001, 21 :599-607. 22.MerchantA,PeukeAD,KeitelC,MacfarlaneC,WarrenCR,AdamsMA: PhloemsapandleafdeltaC-13,carbohydrates,andaminoacid concentrationsinEucalyptusglobuluschangesystematicallyaccording tofloodingandwaterdeficittreatment. JournalofExperimentalBotany 2010, 61 :1785-1793. 23.HarbA,PereiraA: ScreeningArabidopsisGenotypesforDroughtStress Resistance. PlantReverseGenetics:MethodsandProtocols 2011,191-198. 24.DeyholosMK: Makingthemostofdroughtandsalinitytranscriptomics. PlantCellandEnvironment 2010, 33 :648-654. 25.LefebvreV,KianiSP,Durand-TardifM: AFocusonNaturalVariationfor AbioticConstraintsResponseintheModelSpeciesArabidopsisthaliana. InternationalJournalofMolecularSciences 2009, 10 :3547-3582. 26.CohenD,Bogeat-TriboulotMB,TisserantE,BalzergueS,MartinMagnietteML,LelandaisG,NingreN,RenouJP,TambyJP,ThiecDl, HummelI: Comparativetranscriptomicsofdroughtresponsesin Populus:ameta-analysisofgenome-wideexpressionprofilinginmature leavesandrootapicesacrosstwogenotypes. BMCGenomics 2010, 11 :630,(12November2010). 27.BertaM,GiovannelliA,SebastianiF,CamussiA,RacchiML: Transcriptome changesinthecambialregionofpoplar(PopulusalbaL.)inresponseto waterdeficit. PlantBiology 2010, 12 :341-354. 28.HamanishiET,RajS,WilkinsO,ThomasBR,MansfieldSD,PlantAL, CampbellMM: IntraspecificvariationinthePopulusbalsamiferadrought transcriptome. PlantCellandEnvironment 2010, 33 :1742-1755. 29.VeraJC,WheatCW,FescemyerHW,FrilanderMJ,CrawfordDL,HanskiI, MardenJH: Rapidtranscriptomecharacterizationforanonmodel organismusing454pyrosequencing. MolecularEcology 2008, 17 :1636-1647. 30. TorresTT,MettaM,OttenwalderB,SchlottererC: Geneexpressionprofiling bymassivelyparallelsequencing. GenomeResearch 2008, 18 :172-177. 31.BarakatA,DiLoretoDS,ZhangY,SmithC,BaierK,PowellWA,WheelerN, SederoffR,CarlsonJE: ComparisonofthetranscriptomesofAmerican chestnut(Castaneadentata)andChinesechestnut(Castaneamollissima)Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page16of18

PAGE 17

inresponsetothechestnutblightinfection. BMCPlantBiology 2009, 9 :51, (9May2009). 32.AlagnaF,D AgostinoN,TorchiaL,ServiliM,RaoR,PietrellaM,GiulianoG, ChiusanoML,BaldoniL,PerrottaG: Comparative454pyrosequencingof transcriptsfromtwoolivegenotypesduringfruitdevelopment. BMC Genomics 2009, 10 :399,(26August2009). 33.TraasJ,VernouxT: Theshootapicalmeristem:thedynamicsofastable structure. PhilosophicalTransactionsoftheRoyalSocietyofLondonSeriesBBiologicalSciences 2002, 357 :737-747. 34.Lopez-JuezE,DillonE,MagyarZ,KhanS,HazeldineS,deJagerSM, MurrayJAH,BeemsterGTS,BogreL,ShanahanH: Distinctlight-initiated geneexpressionandcellcycleprogramsintheshootapexand cotyledonsofArabidopsis. PlantCell 2008, 20 :947-968. 35.KellerG,MarchalT,SanClementeH,NavarroM,LadouceN,WinckerP, CoulouxA,TeulieresC,MarqueC: Developmentandfunctional annotationofan11,303-ESTcollectionfromEucalyptusforstudiesof coldtolerance. TreeGenetics&Genomes 2009, 5 :317-327. 36.RengelD,SanClementeH,ServantF,LadouceN,PauxE,WinckerP, CoulouxA,SivadonP,Grima-PettenatiJ: Anewgenomicresource dedicatedtowoodformationinEucalyptus. BMCPlantBiology 2009, 9 :36, (27March2009). 37.FariaDA,MamaniEMC,PappasMR,PappasGJ,GrattapagliaD: ASelected SetofEST-DerivedMicrosatellites,PolymorphicandTransferableacross 6SpeciesofEucalyptus. JournalofHeredity 2010, 101 :512-520. 38.MizrachiE,HeferCA,RanikM,JoubertF,MyburgAA: Denovoassembled expressedgenecatalogofafast-growingEucalyptustreeproducedby IlluminamRNA-Seq. BMCGenomics 2010, 11 :681,(1December2010). 39.ReidKE,OlssonN,SchlosserJ,PengF,LundST: Anoptimizedgrapevine RNAisolationprocedureandstatisticaldeterminationofreference genesforreal-timeRT-PCRduringberrydevelopment. BMCPlantBiology 2006, 6 :11,6(Nov.). 40.MarguliesM,EgholmM,AltmanWE,AttiyaS,BaderJS,BembenLA,BerkaJ, BravermanMS,ChenYJ,ChenZT, etal : Genomesequencingin microfabricatedhigh-densitypicolitrereactors. Nature 2005, 437 :376-380. 41.UenoS,ProvostGl,LegerV,KloppC,NoirotC,FrigerioJM,SalinF,SalseJ, AbroukM,MuratF, etal : BioinformaticanalysisofESTscollectedby Sangerandpyrosequencingmethodsforakeystoneforesttreespecies: oak. BMCGenomics 2010, 11 :650,(23November2010). 42.PerteaG,HuangXQ,LiangF,AntonescuV,SultanaR,KaramychevaS,LeeY, WhiteJ,CheungF,ParviziB, etal : TIGRGeneIndicesclusteringtools (TGICL):asoftwaresystemforfastclusteringoflargeESTdatasets. Bioinformatics 2003, 19 :651-652. 43. WangLK,FengZX,WangX,WangXW,ZhangXG: DEGseq:anRpackage foridentifyingdifferentiallyexpressedgenesfromRNA-seqdata. Bioinformatics 2010, 26 :136-138. 44.BenjaminiY,HochbergY: ControllingtheFalseDiscoveryRate-a PracticalandPowerfulApproachtoMultipleTesting. JournaloftheRoyal StatisticalSocietySeriesB-Methodological 1995, 57 :289-300. 45.AltschulSF,GishW,MillerW,MyersEW,LipmanDJ: BasicLocalAlignment SearchTool. JournalofMolecularBiology 1990, 215 :403-410. 46.GotzS,Garcia-GomezJM,TerolJ,WilliamsTD,NagarajSH,NuedaMJ, RoblesM,TalonM,DopazoJ,ConesaA: High-throughputfunctional annotationanddataminingwiththeBlast2GOsuite. NucleicAcids Research 2008, 36 :3420-3435. 47.UsadelB,PoreeF,NagelA,LohseM,Czedik-EysenbergA,StittM: Aguide tousingMapMantovisualizeandcompareOmicsdatainplants:acase studyinthecropspecies,Maize. PlantCellandEnvironment 2009, 32 :1211-1229. 48.Safou-MatondoR,DeleporteP,LaclauJP,BouilletJP: Hybridandclonal variabilityofnutrientcontentandnutrientuseefficiencyinEucalyptus standsinCongo. ForestEcologyandManagement 2005, 210 :193-204. 49.HalperinW: OrganogenesisatShootApex. AnnualReviewofPlant PhysiologyandPlantMolecularBiology 1978, 29 :239-262. 50.Vega-ArreguinJC,Ibarra-LacletteE,Jimenez-MorailaB,MartinezO,VielleCalzadaJP,Herrera-EstrellaL,Herrera-EstrellaA: Deepsamplingofthe Palomeromaizetranscriptomebyahighthroughputstrategyof pyrosequencing. BMCGenomics 2009, 10 :299,(6July2009). 51.SoltisDE,SoltisPS,ChaseMW,MortME,AlbachDC,ZanisM,SavolainenV, HahnWH,HootSB,FayMF, etal : Angiospermphylogenyinferredfrom 18SrDNA,rbcL,andatpBsequences. BotanicalJournaloftheLinnean Society 2000, 133 :381-461. 52.LynchM,WalshJB: GeneticsandAnalysisofQuantitativeTraits Sunderland, MA;1998. 53.EkblomR,GalindoJ: Applicationsofnextgenerationsequencingin molecularecologyofnon-modelorganisms. Heredity 2010,aop. 54.MetzkerML: APPLICATIONSOFNEXT-GENERATIONSEQUENCING Sequencingtechnologies-thenextgeneration. NatureReviewsGenetics 2010, 11 :31-46. 55.ZerbinoDR,BirneyE: Velvet:Algorithmsfordenovoshortreadassembly usingdeBruijngraphs. GenomeResearch 2008, 18 :821-829. 56.MorozovaO,MarraMA: Applicationsofnext-generationsequencing technologiesinfunctionalgenomics. Genomics 2008, 92 :255-264. 57. ZhouX,SuZ,SammonsRD,PengYH,TranelPJ,StewartCN,YuanJS: Novel softwarepackageforcross-platformtranscriptomeanalysis(CPTRA). Bmc Bioinformatics 2009, 10(Suppl11) :S16. 58.WallPK,Leebens-MackJ,ChanderbaliAS,BarakatA,WolcottE,LiangHY, LandherrL,TomshoLP,HuY,CarlsonJE, etal : Comparisonofnext generationsequencingtechnologiesfortranscriptomecharacterization. BMCGenomics 2009, 10 :347,(1August2009). 59.NowrousianM: Next-GenerationSequencingTechniquesforEukaryotic Microorganisms:Sequencing-BasedSolutionstoBiologicalProblems. EukaryoticCell 2010, 9 :1300-1310. 60.HegartyMJ,BarkerGL,BrennanAC,EdwardsKJ,AbbottRJ,HiscockSJ: Changestogeneexpressionassociatedwithhybridspeciationinplants: furtherinsightsfromtranscriptomicstudiesinSenecio. Philosophical TransactionsoftheRoyalSocietyB-BiologicalSciences 2008, 363 :3055-3069. 61.PaveySA,CollinH,NosilP,RogersSM: Theroleofgeneexpressionin ecologicalspeciation. YearinEvolutionaryBiology 2010, 1206 :110-129. 62.PottsBM,DungeyHS: Interspecifichybridizationof<i> Eucalyptus</i>:keyissuesforbreedersandgeneticists. NewForests 2004, 27 :115-138. 63.ShinozakiK,Yamaguchi-ShinozakiK: Genenetworksinvolvedindrought stressresponseandtolerance. JournalofExperimentalBotany 2007, 58 :221-227. 64.Hong-BoS,Li-YeC,Ming-AnS: Calciumasaversatileplantsignal transducerundersoilwaterstress. Bioessays 2008, 30 :634-641. 65.SchroederJI,KwakJM,AllenGJ: Guardcellabscisicacidsignallingand engineeringdroughthardinessinplants. Nature 2001, 410 :327-330. 66.HutinC,NussaumeL,MoiseN,MoyaI,KloppstechK,HavauxM: Earlylightinducedproteinsprotectarabidopsisfromphotooxidativestress. ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesof America 2003, 100 :4921-4926. 67.MartindaleJL,HolbrookNJ: Cellularresponsetooxidativestress: Signalingforsuicideandsurvival. JournalofCellularPhysiology 2002, 192 :1-15. 68.KitajimaS,SatoF: Plantpathogenesis-relatedproteins:Molecular mechanismsofgeneexpressionandproteinfunction. Journalof Biochemistry 1999, 125 :1-8. 69.LeeBR,JungWJ,LeeBH,AviceJC,OurryA,KimTH: Kineticsofdroughtinducedpathogenesis-relatedproteinsanditsphysiologicalsignificance inwhitecloverleaves. PhysiologiaPlantarum 2008, 132 :329-337. 70.PrzymusinskiR,RucinskaR,GwozdzEA: Increasedaccumulationof pathogenesis-relatedproteinsinresponseoflupinerootstovarious abioticstresses. EnvironmentalandExperimentalBotany 2004, 52 :53-61. 71. DubosC,StrackeR,GrotewoldE,WeisshaarB,MartinC,LepiniecL: MYB transcriptionfactorsinArabidopsis. TrendsinPlantScience 2010, 15 :573-581. 72.YamaguchiK,TakahashiY,BerberichT,ImaiA,TakahashiT,MichaelAJ, KusanoT: Aprotectiveroleforthepolyaminespermineagainstdrought stressinArabidopsis. BiochemicalandBiophysicalResearchCommunications 2007, 352 :486-490. 73.AlexanderssonE,FraysseL,Sjovall-LarsenS,GustavssonS,FellertM, KarlssonM,JohansonU,KjellbomP: Wholegenefamilyexpressionand droughtstressregulationofaquaporins. PlantMolecularBiology 2005, 59 :469-484. 74.LandryCR,OhJ,HartlDL,CavalieriD: Genome-widescanrevealsthat geneticvariationfortranscriptionalplasticityinyeastisbiasedtowards multi-copyanddispensablegenes. Gene 2006, 366 :343-351. 75.Hodgins-DavisA,TownsendJP: Evolvinggeneexpression:fromGtoEto GE. TrendsinEcology&Evolution 2009, 24 :649-658. 76.ChinnusamyV,ZhuJK: Epigeneticregulationofstressresponsesin plants. CurrentOpinioninPlantBiology 2009, 12 :133-139.Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page17of18

PAGE 18

77.MazzucotelliE,MastrangeloAM,CrosattiC,GuerraD,StancaAM, CattivelliL: Abioticstressresponseinplants:Whenpost-transcriptional andpost-translationalregulationscontroltranscription. PlantScience 2008, 174 :420-431.doi:10.1186/1471-2164-12-538 Citethisarticleas: Villar etal .: RNA-Seqrevealsgenotype-specific molecularresponsestowaterdeficitineucalyptus. BMCGenomics 2011 12 :538. Submit your next manuscript to BioMed Central and take full advantage of: Convenient online submission Thorough peer review No space constraints or color gure charges Immediate publication on acceptance Inclusion in PubMed, CAS, Scopus and Google Scholar Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Villar etal BMCGenomics 2011, 12 :538 http://www.biomedcentral.com/1471-2164/12/538 Page18of18

PAGE 19

SourceDDLSum of squaresMean of squaresFPr > F Genotype1183.059183.05975.868< 0.0001 Treatment143.19243.19217.9010.0001 Genotype x Treatment10.9340.9340.3870.537

PAGE 20

20000 25000 30000 35000 40000 45000 50000 m ber of readsRUN A 15000 20000 25000 30000 35000 40000 45000 50000 m ber of readsRUN B 15000 20000 25000 30000 35000 40000 45000 50000 m ber of readsRUN C 0 5000 10000 150000 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 Nu m Read length 0 5000 10000 15000 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 Nu m Read length 0 5000 10000 15000 0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420 440 460 480 500 520 540 560 580 600 620 640 660 680 700 Nu m Read length

PAGE 21

ABC One run One run A+BA+CB+CA+B+C Two runs Three runs Two runs Three runs

PAGE 23

FigureS3–RepartitionofG-DEcontigsinsev e Eachsquarerepresentlog2transformedratio b et w contig.Contigsingreenareover-expressedinG1 a G2 Log2 ratio G1 e ralfunctionalpathways w eenabundanceingenotypeG1andG2forone a n d contigsinblueareover-expressedinG2.

PAGE 24

Category# contigsmean FC# contigsmean FC# contigsmean FC# contigsmean FC# contigsmean FC# contigsmean FChormone metabolism44 2**5 0.777 0.1490.259 1.42*10 0.51 hormone metabolism : auxin8 1.221 1.532 0.2122.67*2 1.4421.45 hormone metabolism : ethylene signal transductio n 21 3.22**1 6.36*0NA3 2.372 5.24**3 5.39** signalling : calcium141.89**0NA6 0.4840.5760.8762.56* signalling : ligh t 100.1550.113 0.4722.88*30.2733.52* signalling : receptor kinase s 16 1.0711.0911.253 2.74*1 0.893 4* stress73 0.0513 0.6817 0.27190.2220 0.81**21 0.21 abiotic stress : heat10 0.856 0.269 0.7*101.48*10 1.34**100.68 biotic stress : PR proteins290.51**3 1.633 0.23 2.76*4 0.45 0.88 regulation of transcription : MYB related transcription factor famil y 30.5221.5*0NA0NA0NA0NA regulation of transcription : unclassified6 2.152 0.0410.1617.42*11.2518.51* nitrilases72.12*10.970NA0NA0NA0NA photosystem380.94*100.73130.1130.99*130.38131.28 cell37 2.31**5 1.6230.035 2.7**5 1.376 2.91** major CHO metabolism : starch degradation2 3.673 2.76**10.562 2.16*1 0.792 2.67* lipid metabolism14 0.591 5.0750.256 1.4751.19*6 0.49 protein degradation : serine proteas e 91.75**0NA0NA1 2.6110.051 4.43 protein degradation : ubiquitin E 3 23 2.04*3 1.995 0.396 0.450.026 0.52 protein synthesis : ribosomal protein47 0.4731.26*21.93**3 1.63 0.533 3.72* secondary metabolism : phenylpropanoid s 240.1120.844 0.2432.97**41.7**44.16** development14 0.6221.58*6 0.6360.6460.4271.32 **Categories differentially expressed ay an error threshold of 0.0 5 *Categories differentially expressed ay an error threshold of 0. 1 "GxT"contigs: "G" contigs:"T"contigs:"GxT"contigs:"GxT"contigs:"GxT"contigs: IR (1 41 vs 18 50)NI (1 41 vs 18 50) (1 41 vs 18 50)(IR vs NI)1 41 (IR vs NI)18 50 (IR vs NI)

PAGE 25

FigureS4–Repartitionof“T”contigsinsever a Eachsquarerepresentlog2transformedratio b et w (NI)treatmentsforonecontig.Contigsinbluear e expressedinNI. NI Log2 ratio IR a lfunctionalpathways w eenabundanceinirrigated(IR)andnon-irrigated e over-expressedinIRandcontigsinredareover-

PAGE 26

Validation by RT-qPCR Methods The transcript abundance of 36 ge nes was estimated by RT-qPCR fo r three types of sample: i) cDNA samples (obtained with the Smart c DNA Library Construction kit) used for sequencing, ii) RNAs used for library constr uction and sequencing, and iii) RNAs extracted from independent samples. The extracted RNAs obtained in (ii) and (iii) were reversetranscribed with the ImpromII Reverse Transc ription System (Promega, Madison, WI, USA) according to the manufacturer's instructions. Primer pairs were designed from the 454-reads with Primer3 software (Rozen and Skaletsky, 2000) (see supporting f ile 1). Real-time PCR was performed on 384-well plates, and was m onitored with a Light Cycler 480 (Roche Applied Science, Indianapolis, IN, USA) at the Genome and Transcriptome Facility of Bordeaux (France, https://www4.bordeaux-aquitaine.inra.fr/pgtb), France. Each reaction mixture contained 3 l of cDNA preparation (diluted 1/50 for Improm reverse-transcribed cDNA and 1/1000 for SMART reverse-transcri bed cDNA), 5 l of 2 x ABsolute QPCR SYBR Green Mix (ABgene, Epsom, UK) and 70 nM gene-specific primer in a total volume of 10 l. The cDNA samples and master mix were pipetted with a Hamilton workstation (Hamilton Robotics, Reno, NV, USA). Each r eaction was performed in duplicate. The fluorescent signal was evaluated and starting concentrations we re calculated with LinRegPCR (Ramakers et al. 2003; Ruijter et al. 2009). The relative abundance of transcripts was normalised against the mRNA levels for four constitutively expressed housekeeping genes: elongation factor 1EF1elongation factor 2 (eEF2), gl yceraldehyde 3P dehydrogenase (G3PDH) and cyclophilin (Cyp), by calculating a norma lisation factor with GeNorm (Pattyn et al. 2003). Log2-transformed ratios between treatments were generated for each genotype, to facilitate comparison of the results w ith those of digital expression profiling.

PAGE 27

Results We selected a subset of 36 contigs for valid ation of their expression levels by RT-qPCR. Sixteen displayed stable levels of expressi on (0.52) in the digita l expression analysis, between genotypes or treatments, or displaying GxT effects. Four st ably expressed genes were chosen with GeNorm for data normalisation: these genes encoded elongation factor 1-alpha ( EF1a ), eukaryotic translation elongation factor 2 ( eEF2 ), cyclophilin ( CYP ), and glycerol-3-phosphate dehydrogenase ( G3PDH ). RT-qPCR was carried out on three sets of RNA/cDNA samples: i) 8 cDNA samples (used for 454-sequencing) obta ined by SMART amplification of eight RNA samples corresponding to the four conditions with two replicates (set #1), ii) eight new cDNA samples obtained by linear reverse transcription (RT) of the same RNA samples (set #2), iii) eight new cDNA samples obtained by RT of eight new RNA samples (set #3). Gene expression, as analysed by RT-qPCR, wa s strongly correlated in sets #2 and #3 (significant Spearman correlation of 93.4%), sugge sting a high degree of homogeneity in the pattern of expression be tween biological replicates. Correl ations between gene expression in these two sets and in set #1 were weaker, with a mean correlati on of 68.3%. The non-linear SMART amplification may account for these diffe rences in expression patterns and suggests that caution is required when using this tech nique for expression va lidation. We therefore chose not to use this method in subsequent an alyses. Spearman’s correlations between digital expression profiles (454-results) and the two sets of RT-qPCR results were significant, but relatively low: 35.3% (set #2), 28.7% (set #3). The weakness of these co rrelations may be accounted for by the successive biases of th e method: non-linear amplification by SMART and low sequencing depth, generating only a limite d number of reads fo r digital expression analysis. Nevertheless, DEGseq analysis of the 11 contigs with significant fold-changes

PAGE 28

between IR and NI treatments for the two ge notypes in 454-sequencing data, confirmed the expression patterns of seven genes in set #1, te n genes in set #2, and seven genes in set #3 (supporting file 2). The reaction norms of molecular variations cannot be fully described by 454 quantification (Spearman’s correlations be tween the results obtained with the two methods vary from 28.7% to 35.3%), the pattern of molecular plasticity of significant contigs was similar for the different type s of expression quantification. Supporting file 1 – Primers used for validation by RT-qPCR Gene CodeGene NameContig NamePrimer Forward (5' 3')Primer Reverse (5' 3') SHINE1DNA binding / transcription factorF0F0A3T01A00IG.l.eu.2TCAAGAA ATGCAGCAAGACGGACCCAGTTGGAGTCTGAGC LP1nonspecific lipid transfer protein 1F0F0A3T01A00NX.l.eu.2AGGGGATCGACTTCAACCTCGAGCTCGTGGACCTCACTTC SIP2hydrolase, hydrolyzing O glycosyl compoundsF0F0A3T01A032B.l.eu.2CGATTTG GCGCTTACTCTTCACCGGCCATCTGTACATCTC LIM1transcription factorF0F0A3T01A03RQ.l.eu.2AGAGGGA AACAATGGCTCCTGCTTACAACCAACCCGAGAA RD26NAC domain containing protein 2; transcription factorF0F0A3T01A0DJA.l.eu.2CAACA CACCTCCGCAATATGTCCCAACCTTCCAACTTCAC DCAM adenosylmethionine decarboxylase family proteinF0F0A3T01A0ESR.l.eu.2ATCCAGAGGGTGTTGGTCTGCTCCAGCACTGTACCCCTTC CHIABbasic chitinaseF0F0A3T01A0IP5.l.eu.2; F0F0A3T01A3DVE.l.eu.2CAAGGGCTTCTACACTTACGAGCAGTATCCCCACGCATAG ELIP1(EARLY LIGHT INDUCABLE PROTEIN); chlorophyll bindingF0F0A3T01A0S92.l.eu.2TCCCT CTTCATCGCCTACACGTCTGCCATTGATCCTCTCC PHS2ALPHA GLUCAN PHOSPHORYLASE 2F0F0A3T01A0SF1.l.eu.2TTCTAGCTTCCCTCCCTTCCGTTAGGCGGAGTGCATTTGT CCoAOMT2caffeoyl CoA 3 O methyltransferase, putativeF0F0A3T01A0Z5S.l.eu.2; F0F0A3T01BLVGJ.l.eu.2GCAAGATGAGAAGAACCATGAGCACGAAGTCCCGGTAGTA UCRubiquinol cytochrome C reductase complex, putativeF0F0A3T01A1SQD.l.eu.2GCTGTCTCCTGACGTTTGTGCCGGCTATTTCGACTCCTTC DELTA TIPdelta tonoplast integral proteinF0F0A3T01A2Q8I.l.eu.2CCGTGGTCAGTGGAGACTTCCATTGACAAGAGGTGCATGG CesA1CELLULOSE SYNTHASE 1F0F0A3T01A31VD.l.eu.2TAAAGTGCTTGCTGGCATTGTGAGAAGCGATGTCCATTTG PIP1Bplasma membrane intrinsic protein 1;2F0F0A3T01A344C.l.eu.2TTCTGCAACATGGCTACACCATTCCTCTTGGCATCAGTGG CCoAOMT1caffeoyl CoA 3 O methyltransferase, putativeF0F0A3T01A4G7A.l.eu.2GTATGTGAGGTACTACAGGGCTCTCTTTTTCGTGGGTTCT LHY1 2LATE ELONGATED HYPOCOTYLF0F0A3T01A4GAO.l.eu.2GAAAGAGGATAATGGGCCAAGCCTCGGGAAATTATGAGCAG RBCLlarge subunit of RUBISCOF0F0A3T01A5ANA.l.eu.2GCTAAGAACTACGGTAGAGCTTCACCTGTTTCAGCCTGTG ERD15EARLY RESPONSIVE TO DEHYDRATION 15F0F0A3T01A5C8L.l.eu.2ATATTTCGCCGTCGTTCAAGATAGAGGGGCATCTGGGTTC PGR5PROTON GRADIENT REGULATION 5F0F0A3T01A5OGX.l.eu.2 ; F0F0A3T01A1FPP.l.eu.2TGGCTTGTTGCTAGGGTTTCATTGATCGCGAGCACTTTTC GDR1GLUTATHIONE DISULFIDE REDUCTASEF0F0A3T01A5Q34.l.eu.2TAAGTGTGGTGCAACGAAGGAACCCGTCTTGTGACTGACC 2 ODD2 oxoglutarate dependent dioxygenaseF0F0A3T01A8608.l.eu.2CTCCAGATCATGTCCAACGACAAACAGGCTTTCACAATCG HYP1HYPOTHETICAL PROTEIN 1F0F0A3T01A8IGK.l.eu.2CCGGTGATGTGAGAGAAACCAGCGGTATACGCGACTTGAG RPP1A 60S acidic ribosomal protein P1F0F0A3T01AFPEY.l.eu.2CACCAATGTTGGTTCTGGTGAGACCAAAGCCCATGTCATC GHP3glycosyl hydrolase family 3 proteinF0F0A3T01AJF18.l.eu.2GGGTTG CAGACGTTCTGTTTATGAGGATCCCCAACATTCA DXR1 DEOXY D XYLULOSE 5 PHOSPHATE REDUCTOISOMERASEF0F0A3T01AJOYE.l.eu.2TGAGATCATTCCAGGGGAACGCTATGTCCTTGCCTGCTTC HEN4HUA ENHANCER 4; nucleic acid bindingF0F0A3T01AYA64.l.eu.2GAGAGGAAGGTGGATGCTTGTGAGCTGCATGAGTTTGCTC HVA22abscisic acid responsive HVA22 family proteinF0F0A3T01B0RI8.l.eu.2TCGAGAGCAAATCACCAATGCTCTCACCGTCTGCACTCTG LHY1 1LATE ELONGATED HYPOCOTYLF0F0A3T01B9JR1.l.eu.2CGTTGAAAGAGGGAGTTCAGGGAGCAGTTTGCATCTCATCTG POLGpolygalacturonase, putativeF0F0A3T01BFXHG.l.eu.2CGAAACCAAGATGTGTGTGGCGATCAAGACCTCCTGCTTC CER6CUTICULAR 1; catalyticF0F0A3T01EOHZ5.l.eu.2TCCATATTCCCGAAGTGGTCCGCAATCCACCTATCAAACC

PAGE 29

Supporting file 2 – Comparison of fold-change (FC) ratio between RT-qPCR and digital analysis for 30 genes. Expression assessed by determining the amount of cDNA obtained with the SMART construction kit (Set #1) and in two biological replicates (Set #2 and S et #3) by linear reverse transcription (RT) was compared with the results for 454-sequencing. The mean log2-transformed FC values for the two biological replicates are indicated Significant fold change according to DEGseq analysis, at a q value threshold of 0.05. a: fold change could not be calculated because no reads were found for the genotype considered in either the IR or the NI sequencing set. b: fold change could not be calculated because no expression was reported for the NI treatment, for the genotype considered. c: fold change could not be calculated because no expression was reported for the IR treatment, for the genotype considered. Gene CodeGene NameContig Namenb reads 1 4118 50 1 4118 50 1 4118 50 1 4118 50Set #1/ 454Set #2/ 454Set #3/ 454 LP1nonspecific lipid transfer protein 1F0F0A3T01A00NX.l.eu.227511.42* 1.6*1.36 0.63.12 n.a. ( )cn.a. (+) 2.58YYY ELIP1(EARLY LIGHT INDUCABLE PROTEIN); chlorophyll bindingF0F0A3T01A0S92.l.eu.2317 0.073.32* 0.77 0.34 n.a. (+)b0.024.320.51NYY SIP2hydrolase, hydrolyzing O glycosyl compoundsF0F0A3T01A032B.l.eu.246301.98* 1.13 0.812.892.354.02 0.64NYN LHY1 2LATE ELONGATED HYPOCOTYLF0F0A3T01A4GAO.l.eu.22750.624*2.193.134.460.624.172.15YYY DCAM adenosylmethionine decarboxylase family proteinF0F0A3T01A0ESR.l.eu.23920.341.88*0.851.593.60.263.221.3YYY DELTA TIPdelta tonoplast integral proteinF0F0A3T01A2Q8I.l.eu.26370.021.24* 0.022.214.122.282.392.26YYY LHY1 1LATE ELONGATED HYPOCOTYLF0F0A3T01B9JR1.l.eu.2750.774.93*1.890.090.380.520.490.21YYY PIP1Bplasma membrane intrinsic protein 1;2F0F0A3T01A344C.l.eu.21770.064.43*1.191.12.13 2.05n.a. (+)n.a. ( )YNN PHS2ALPHA GLUCAN PHOSPHORYLASE 2F0F0A3T01A0SF1.l.eu.2430.87 2.53* 0.35 0.40.35 0.68 0.22 0.61NYN PGR5PROTON GRADIENT REGULATION 5F0F0A3T01A5OGX.l.eu.22090.492.14*1.253.452.262.592.143.7YYY CHIABbasic chitinaseF0F0A3T01A0IP5.l.eu.2290 1.11* 1.07* 1.41 1.12 7.5 5.71 6.97 10.93YYY UCRubiquinol cytochrome C reductase complex, putativeF0F0A3T01A1SQD.l.eu.23030.451.040.230.21n.a. (+)0.832.88 0.08YYN HEN4HUA ENHANCER 4; nucleic acid bindingF0F0A3T01AYA64.l.eu.222 0.13 3.480.65 0.822.24 0.348.99 0.17YNN POLGpolygalacturonase, putativeF0F0A3T01BFXHG.l.eu.221 n.a.a 2.11 0.51 1.8n.a. 2.02n.a. 1.89YYY SHINE1DNA binding / transcription factorF0F0A3T01A00IG.l.eu.2205 0.07 0.65 0.140.55n.a. 3.69n.a. 2.12NYY 2 ODD2 oxoglutarate dependent dioxygenaseF0F0A3T01A8608.l.eu.2461.99 1.480.750.682.99 2.053.34 2.88NYY HYP1HYPOTHETICAL PROTEIN 1F0F0A3T01A8IGK.l.eu.223 0.54 2.140.780.595.260.334.72.11NNN HVA22abscisic acid responsive HVA22 family proteinF0F0A3T01B0RI8.l.eu.2183.193.78 0.71 0.914.35 6.264.54 1.97NNN DXR1 DEOXY D XYLULOSE 5 PHOSPHATE REDUCTOISOMERASEF0F0A3T01AJOYE.l.eu.2681.311.260.944.212.451.142.953.62YYY GHP3glycosyl hydrolase family 3 proteinF0F0A3T01AJF18.l.eu.2550.46 1.151.45 1.481.65 0.590.93 1.42YYY ERD15EARLY RESPONSIVE TO DEHYDRATION 15F0F0A3T01A5C8L.l.eu.2162 0.130.59 0.23 0.12.840.190.660.36NNN RPP1A 60S acidic ribosomal protein P1F0F0A3T01AFPEY.l.eu.215 3.3 2.03 0.590.081.731.94 0.550.26NNN LIM1transcription factorF0F0A3T01A03RQ.l.eu.2156 0.99 0.4 4.921.9n.a. 2.24n.a. 3.06NYY RD26NAC domain containing protein 2; transcription factorF0F0A3T01A0DJA.l.eu.251 0.130.880.943.983.01 1.011.79n.a.NNY CCoAOMT2caffeoyl CoA 3 O methyltransferase, putativeF0F0A3T01A0Z5S.l.eu.2114 0.570.68 0.120.52 0.040.35 0.04 0.27YYN GDR1GLUTATHIONE DISULFIDE REDUCTASEF0F0A3T01A5Q34.l.eu.2114 0.580.59 0.31.144.311.855.08 0.48YNN CesA1CELLULOSE SYNTHASE 1F0F0A3T01A31VD.l.eu.2169 0.940.21 0.41 1.52.65 0.43n.a. (+) 3.6NNN CCoAOMT1caffeoyl CoA 3 O methyltransferase, putativeF0F0A3T01A4G7A.l.eu.2550.41 0.44 0.070.190.140.070.270.09YYY RBCLlarge subunit of RUBISCOF0F0A3T01A5ANA.l.eu.2233 0.10.060.89 0.210.97 0.061.17 0.26NNN CER6CUTICULAR 1; catalyticF0F0A3T01EOHZ5.l.eu.216 0.390.56 0.63 3.032.11 0.324.23 3.13NNN Fold Change Similarity Log2 FC : [IR/NI] 454Set 1 (SMART)Set 2 (RT)Set 3 (RT)