Molecular candidates for early-stage flower-to-fruit transition in stenospermocarpic table grape (Vitis vinifera L.) inflorescences ascribed by differential transcriptome and metabolome profiles

ContentslistsavailableatScienceDirect

Plant

Science

jo u r n al h om ep age :w w w . e l s e v i e r . c o m / l o c a t e / p l a n t s c i

Molecular

candidates

for

early-stage

flower-to-fruit

transition

in

stenospermocarpic

table

grape

(Vitis

vinifera

L.)

inflorescences

ascribed

by

differential

transcriptome

and

metabolome

profiles

Sara

Domingos

_,

_Joana

_Fino

_,

_Octávio

_S.

_Paulo

_,

_Cristina

_M.

_Oliveira

_,

_Luis

_F.

_Goulao

a_{LinkingLandscape,Environment,AgricultureandFood(LEAF),InstitutoSuperiordeAgronomia(ISA),UniversidadedeLisboa,Lisbon,Portugal}

b_{BioTrop,InstitutodeInvestigac¸ãoCientíficaTropicalI.P.(IICT),Lisbon,Portugal}

c_{ComputationalBiologyandPopulationGenomicsGroup,cE3c—CentreforEcology,Evolution,andEnvironmentalChanges,FaculdadedeCiências,}

UniversidadedeLisboa,Lisbon,Portugal

a

r

t

i

c

l

e

i

n

f

o

Articlehistory:

Received6September2015

Receivedinrevisedform8December2015

Accepted18December2015

Availableonline6January2016

Keywords: Fruitset Grapevine Metabolicpathways RNA-Seq Seedless

a

b

s

t

r

a

c

t

Flower-to-fruittransitiondependsofnutrientavailabilityandregulationatthemolecularlevelbysugar andhormonesignallingcrosstalk.However,inmostspecies,theidentitiesoffruitinitiationregulators andtheirtargetsarelargelyunknown.Toascertainthemainpathwaysinvolvedinstenospermocarpic tablegrapefruitset,comprehensivetranscriptionalandmetabolomicanalyseswereconducted specifi-callytargetingtheearlyphaseofthisdevelopmentalstagein‘ThompsonSeedless’.Thehigh-throughput analysesperformeddisclosedtheinvolvementof496differentiallyexpressedgenesand28differently accumulatedmetabolitesinthesampledinflorescences.Ourdatashowbroadtranscriptome reprogram-mingofmoleculetransporters,globallydown-regulatinggeneexpression,andsuggestthatregulationof sugar-andhormone-mediatedpathwaysdeterminesthedownstreamactivationofberrydevelopment. ThemostaffectedgenewastheSWEET14sugartransporter.Hormone-relatedtranscriptionchanges wereobservedassociatedwithincreasedindole-3-aceticacid,stimulationofethyleneandgibberellin metabolismsandcytokinindegradation,andregulationofMADS-boxandAP2-likeethylene-responsive transcriptionfactorexpression.Secondarymetabolism,themostrepresentativebiologicalprocessat transcriptomelevel,waspredominantlyrepressed.Theresultsaddtotheknowledgeofmolecularevents occurringingrapevineinflorescencefruitsetandprovidealistofcandidates,pavingthewayforgenetic manipulationaimedatmodelresearchandplantbreeding.

©2015ElsevierIrelandLtd.Allrightsreserved.

1. Introduction

Fruitset,orflower-to-fruittransition,isthestageinwhichthe ovaryprogressestoagrowingyoungfruit.Fruitcellnumberandthe

Abbreviations:ACO,1-aminocyclopropane-1-carboxylateoxidase;AMP,

adeno-sine5_{-monophosphate;bp,basepair;CW,cellwall;cv,cultivar;DEGs,differentially}

expressedgenes;d,daysafter100%capfall;FDR,falsediscoveryrate;IAA,

indole-3-aceticacid;IAMT,indole-3-acetate-o-methyltransferase;FPKM,fragmentsper

kilobaseofexonpermillionfragmentsmapped;FS,fruitsetstage;GA,gibberellin;

GA20ox,gibberellin20oxidase; GO,GeneOntology;KEGG, Kyoto

Encyclopae-diaofGenesandGenomes;KOG,euKaryoticOrthologousGroups;PCoA,principal

coordinateanalysis;PCD,programmedcelldeath;PMN,PlantMetabolicNetwork;

RNA-Seq,ribonucleicacidsequencing;SAM,S-adenosylmethionine;SNPs,single

nucleotidepolymorphisms.

∗ Correspondingauthor.Presentaddress:ColégioFood,FarmingandForestry,

UniversidadedeLisboa,Portugal.

E-mailaddresses:goulao@reitoria.ulisboa.pt,luisfgoulao@gmail.com

(L.F.Goulao).

ultimatenumberoffruitsformedaredeterminedduringthis devel-opmentalstage,withdecisiveimpactonthefinalfruitsizeandplant yieldpotential.Duringthefirst2weeksafterfullbloom,anabrupt increaseinovarysizeoccursduetocellmultiplicationand,toa lesserextent,tocellenlargement[1–4].Duringthisperiod, repro-ductiveorgansaresignificantlymoresensitivetobioticandabiotic stressesthanduringlaterstagesofberrydevelopmentorvegetative growth[5–10],sothesuccessofthisdevelopmentalstageiscritical bothtospeciesbiologicalsurvivalandagronomicsustainability.

Nowadays,anincreaseintablegrapeworldproductionisclearly evidentand,basedonconsumerpreferences,isaccompaniedbya stronginterestinseedlessvarietiesforfreshconsumption. There-fore,seedlessnessisamongthemainfocusofcurrenttablegrape breedingprogrammes[11].Inseededgrapes,flower-to-fruit tran-sition requires pollination and ovary fertilisation toward seed formation. In seedless grapes, two mechanisms can be distin-guished:parthenocarpy,whichoccurswhentheovaryisableto develop without ovulefertilisation(e.g., cvBlackCorinth), and http://dx.doi.org/10.1016/j.plantsci.2015.12.009

stenospermocarpy,whichprogressesafterpollinationand fertilisa-tionbutwiththeembryoaborting2–4weeksafterfertilisation(e.g., cvThompsonSeedless)[2,12].Inthelattercase,partiallydeveloped seedsorseedtracescanbefound.

Naturalflowershatterisaneventthatoccursduringfruitset, andgenerallyonly20–30%oftheflowersdevelopintoberries[2]. Fruitsetisincreasinglysustainedbyphotoassimilatesandother nutrientsexportedfromphotosyntheticallyactiveleavesthrough thephloem[13]andphotosynthesisedbytheinflorescenceitself. Hence,thesuccessofthisdevelopmentalprocessdependson fac-torssuchasphotoassimilateproduction,transport capacity,and phloemunloadingtothefruit.Nonetheless,themaindeterminant isthefruitsinkstrength[14,15].

Basedontranscriptandmetaboliteanalysesduring fertilisation-dependentand fertilisation-independent fruit formation, itwas furtherverifiedinthefruitmodelspeciestomato(Solanum lycoper-sicon)that,inadditiontonutrientavailability,commonpathways offlower-to-fruittransitionincludemodulationbysugarand hor-monesignallingcrosstalkandregulationbyspecifictranscription factorsfromseveralfamilies[16–24].However,onlyfew compo-nentsofthemechanismregulatingfruitsethavebeenidentified sofar.Accordingtothehypotheticalmodelforsugarsignallingin seedandfruitsetregulation[25],glucose,astheproductofsucrose hydrolysis,actsasasignalmoleculetorepresstheexpressionof programmedcelldeath(PCD)genesandtopromotecelldivision that, in turn,leads tofruit set.Under severestress conditions, phloemsucroseimportisblocked,whichtogetherwitha deple-tionofstarchreservesresultsinreducedglucoselevels,triggering thePCDpathwayandinhibitingcelldivision,consequentlyleading tofruitabortion[26].Auxins,ethylene,andgibberellins(GAs)are otherkeyplayerswithaprominentroleintriggeringand coordi-natingthefruitsetdevelopmentalprocess[22].Auxinsignallingis recognisedasoneoftheearliesteventsinthefruitinitiation cas-cade,particularlythefine-tuningofAUXIN/INDOLE-3-ACETICACID, AUXINRESPONSEFACTOR,andMADS-boxgeneexpression[24].In addition,fruitsetismediatedbyactivatingGAbiosynthesismainly throughtheup-regulation ofGIBBERELLIN OXIDASE20(GA20ox) [21].Polyaminemetabolismshoweddifferentdynamicsaccording toindividualspecies.Levelsofputrescinewereshowntodecrease intomatowhileincreasinginplum(Prunusinsititia)duringfruitset [24,27].Alterationsinsecondarymetabolismwerealsofound dur-ingthisstage.Forinstance,ascorbateantioxidantlevelsdeclined during theanthesisto post-anthesis transition,while its direct precursor,galactonate-1,4-lactone,increasedatthepost-anthesis stage[24].

Despitetheimportanceofgrapevine,bothfromaneconomic pointofviewand asa genomicmodelforstudyingwoodyfruit species [28,29], to date, the development-related global tran-scriptomicandmetabolomicanalysesreportedtargetedonlylater stagesofberrydevelopment[30–35].In thiswork,through dif-ferentialgene expressionandmetaboliteanalyses, weaimedto identifytheputativemolecularcuesresponsibleforinitial flower-to-fruittransitioninstenospermocarpictablegrapes,whichhavea profoundimpactonthelaterstagesofberrydevelopment.

2. Methods

2.1. Samplecollection

Grapeinflorescenceswerecollectedfrom7-year-oldcv Thomp-sonSeedlesscommercialvines(VitisviniferaL.)thatweregrafted on a 140 Ruggeri rootstock and spaced 3×3m. The commer-cialvineyardislocatedinFerreiradoAlentejo,southernPortugal andwasmanagedfollowingstandardfertilisation,irrigation,and pest-managementpractices.Vinesweregrownunderanoverhead

trellis systemcoveredwith plastic,providing favourable condi-tionsforacompressedandhomogeneousfloweringperiod.Climate conditions(photosyntheticallyactiveradiation,temperature,and relativehumidity)weremonitoredabovethevinecanopy (Watch-DogMicroStation,SpectrumTechnologies,Aurora,USA)andshown tobemaintainedduringthesamplingperiod(SupplementaryFig. A.1).Toguaranteethatthe2–3-dayperiodrequiredafter pollina-tionfor fertilisationtobecompleted[9]wasfullytargeted,the inflorescencesamplingwasconductedatthreetimepoints(Fig.1): fruitsetstage1(FS1;3daysafter100%capfall(3d)),FS2(5d),and FS3(7d).Onehundredpercentcapfallcorrespondedtostage69of theBBCHscale[36].Ateachtimepoint,threeinflorescenceswere selectedfromafive-vineplotbasedonuniformityofblossom.An inflorescencewasconsideredtobehomogenouswhenmorethan 90%oftheflowersshowedtobeatthesamestage.Eachbiological replicatewasmadeupofaninflorescence,deprivedfromtherachis, immediatelyfrozeninliquidnitrogenandsubsequentlypowdered andstoredat−80◦_Cuntiluse.

2.2. RNAwholetranscriptomedeepsequencing

Total RNA was extracted and purified from 100mg frozen material independently for each biological sample using the RNeasy Plant RNA Extraction Kit and RNase-Free DNase Set (Qiagen,Hilden,Germany)followingthemanufacturer’s instruc-tions, modified by replacing the kit’s extraction buffer by 100mM Tris–HCl, 2% (w/v) CTAB, 25mM EDTA, and 2M NaCl buffer [37]. When traces of contaminant genomic DNA were detected after standard PCR amplification of the Actin 1 (ACT1) gene (GenBank Accession: XM002282480.3; forward primer: 5-CTTCCAGCCATCTCTCATTGG-3 and reverse primer: 5- TGTTGCCATAGAGGTCCTTCC-3), RNA samples were further digestedwithRNase-freeDNaseI(Ambion,LifeTechnologies, Carls-bad, USA). RNA integrity and purity were evaluated by visual inspectionofribosomalbandsafter1.5%agarosegel electrophore-sisandthrougha2100Bioanalyzersystem(AgilentTechnologies, SantaClara,USA)readings.Poly(A)mRNAisolation,cDNA synthe-sis,librarygeneration,indexing,clustergeneration,andRNA-Seq analysesbyIlluminaHiSeq2000RNAsequencingof100-bp paired-endreadswerecarriedoutbyLGCGenomics(Berlin,Germany), usingcommercialservices.

2.3. AlignmentandanalysisofIlluminareads

TherawIllumina100-bppaired-endsequencesweredeposited in the NCBI Sequence Read Archive under accession numbers SRS:853491, SRS:879704,SRS:879828, SRS:879847,SRS:879856, SRS:882229,SRS:882241,SRS:882246,andSRS:882247.

Poorqualitybasereadsandadaptersequencesweretrimmed using Trimmomatic version 0.32 software [38]. Trimmedreads weremappedbacktothereferencegenome[28]forquantification ofgeneexpressionlevelsandglobaltranscriptexpression profil-ingwithineachtimepointusingTophat2version2.0.12[39]with theparameters-D15-R2-L22-iS,1,1.15aswellasend-to-end mode.Thenumbersofreadsmappedtoeachgenepersamplewere extractedfromtheoutputbamfilesofTophat2usingthein-house pythonscriptbamzinga(https://github.com/Nymeria8/bamzinga). Quantificationandnormalisationofgeneexpressionvaluesby frag-mentsperkilobaseofexonpermillionfragmentsmapped(FPKM) wascalculatedbytheCufflinkspackagecuffdiffversion2.2.1[40]. TheMfuzzRpackage version2.26.0[41] processedthe expres-siondatatogenerateclusteringplotswithmissingdataandlow standarddeviationfiltering,withnumberofcentressetto9.

TheRNA-Seqdatawerealsousedtocomparethecv Thomp-sonSeedlesssequencewiththeseededvarietyPinotNoirreference genome [28]. Across the expressed genes in ‘Thompson’

inflo-Fig.1.AspectofrepresentativecultivarThompsonSeedlessinflorescencesduringfruitset.Sampleswerecollectedatfruitsetstage1(FS1;3daysafter100%capfall(3d)),

FS2(5d),andFS3(7d).Scalebaris0.6cm.

rescences under the conditions of the assay, single nucleotide polymorphisms(SNPs)andpost-transcriptionalprocessingevents, suchasconstitutiveandalternativesplicing,weresearched.For thedetectionofSNPs,AtlasSNPsoftwarewasused[42].The Cuf-flinksprogramcuffmergewasusedtoidentifynewsplicingevents betweenthereferencegenomeandourdataset.Toincrease accu-racy,weconservativelyrequiredthatatleast10independentreads mappedintheisoformandatleast10readsmappedacrossthe entireexon–exonjunction.

2.4. Differentialgeneexpression,functionalannotation,and pathwayanalysis

Statistical analysis toidentify differentially expressed genes (DEGs)betweentimepointswasconductedusingtheRpackage version3.0.2fromBioconductor,DESeq2version1.4.5[43], con-sideringanestimationofsizefactors,afalsediscoveryrate(FDR) of0.05,anda-1.5≥log2foldchange≥1.5.

ToobtaintheeuKaryoticOrthologousGroups(KOG)annotation ofgrapevinegenes,Rapsearch2[44] wasusedtosearchagainst functional proteins from Arabidopsis thaliana in the database [45],consideringane-valuecutoffof10−5.FromthetotalDEGs identified,thosethat werenotautomaticallyassigned toa spe-cific KOG functional category were either ascribed in theKOG categories according to the gene description or classified as ‘other’ or ‘of unknown function’. For additional gene assign-mentintheKyotoEncyclopaediaofGenesandGenomes (KEGG [46])andGeneOntology (GO)annotation,Rapsearch2similarity searcheswere locally conducted againsta non-redundant (‘nr’) peptide database. The output was submitted to an in-house-developedscript,Rapsearch2XML(https://github.com/Nymeria8/ Rapsearch2Xml)andthentoBlast2GO[47]forenzyme identifi-cation,metabolicpathwayassignment,andfunctionalannotation usingGOterms.GOenrichedcategorieswereassignedusingthe RBioconductorpackagetopGOversion2.18.0[48],usingaFisher’s exacttestandFDRcorrectionof0.01.

Similarityofexpressionprofilesbetweenbiologicalreplicates wasdeterminedbyPearsoncorrelationcoefficientanalyseswith R3.1.2 software usingnatural logarithm (ln)-transformedread countsfortheDEGsasinput.Alltheanalysesconsideredthree inde-pendentbiologicalreplicatespertreatmentsequenced,providing higherstatisticalrobustnessanddatavalidation.

2.5. Globalmetabolomicanalysis

About200mgofpowderedmaterialfromeachofthethree bio-logical replicatescollectedateach time point werelyophilised, extractedwithmethanol,andanalysedusingtheintegrated plat-form developed by Metabolon® (Durham, USA), consisting of a combination of three independent approaches: (1) ultrahigh performanceliquid chromatography/tandemmassspectrometry (UHLC/MS/MS2) optimised for basic species, (2) UHLC/MS/MS2 optimisedfor acidic species, and (3) gaschromatography/mass spectrometry (GC/MS). Methods were performed as previously described[49–51].

2.6. Differentiallyquantifiedmetabolitesandmappingonto metabolicpathways

Rawareacountsforeachbiochemicalcompoundwererescaled bydividingeachsample’svaluebythemedianvalueforthespecific biochemical.Statisticalanalysisofthedatawasperformedusing Welch’stwo-samplet-testswithArrayStudio(Omicsoft)software. Mappingofnamedmetaboliteswasperformedontogeneral bio-chemicalpathways,asprovidedintheKEGGdatabasesandPlant MetabolicNetwork.Boxplotsweregeneratedforthosecompounds thatshowedasignificantchangeusingboththeWelchtwo-sample t-test(p-value≤0.05)and|log2foldchange|≥1.

2.7. Globalanalysisoftranscriptomicandmetabolomicprofiling Datafor transcriptandmetaboliteprofilinganalyseswas ln-transformed,andnormaldistributionwasverifiedbyhistogram plottingofthenumberofreadsandmetabolitespersampleusing theRsoftwarepackage.Forexploratorydataanalyses,principal coordinate analysis (PCoA) was conducted based on the pair-wisecorrelationmatrixusingtheNTsys-PCversion2.20esoftware package [52].The DCENTERmodule wasusedtotransformthe symmetric matrix to scalar product and EIGEN for eigenvalue decomposition toidentify orthogonalcomponents of the origi-nal matrix modules. A minimum-spanningtree was calculated andsuperimposedtofacilitatethevisualisationofthedistances betweenoperationalunits.Rstatisticalsoftwarewasusedforheat mapconstruction.Heatmap-associatedhierarchicalclusteringand approximateunbiasedandbootstrapprobabilityp-valueswere cal-culatedusingpvclustversion1.3.2[53]withtheUPGMAmethod builtfromthecorrelationdistancemeasuresand1000bootstrap replications.

Table1

RNA-Seqdataoverview.Numbersof100-bpreadsobtainedateachfruitsetstage

sequenced,beforeandafterdatatrimming(meansofthreeindependentbiological

replicates±standarderror(SE)).

Fruitsetstage Rawreadpairs (x1000)

Readpairsafter trimming (x1000) %ofremaining reads FS1 28494±1209 26517±1192 93.0±0.4 FS2 36342±5193 33045±4534 91.1±1.5 FS3 24725±603 22765±462 92.1±0.5 3. Results

3.1. TranscriptomeanalysisbyRNA-Seq

NineRNA-Seq100-bppaired-endreadlibrarieswereprepared

from poly(A) RNA extracted from grape inflorescences,

corre-spondingtothethree time pointssampled(FS1,FS2,and FS3),

eachrepresentedbythethreeindependentbiological replicates

collected.EachcDNAlibraryresultedin≈29million100-bp

paired-endreads.AnoverviewoftherawreaddataisgiveninTable1.The

resultsshowedthat8%ofthereadswereremovedsincethey over-passthethresholdcutoffafterbeingtrimmedbasedonthepresence ofIlluminaadaptersorlowqualitybases.Themajority(62%)of thetotalnumberofreadscouldbemappedbacktothereference genome.Anadditional1.7%ofreadsweremappedtomultiple loca-tionswithinthereferencegenomesequence,andthesereadswere discarded.Statisticsofeachsamplemappingareprovidedindetail inTable2andSupplementaryFig.A.2.

Transcriptionof25703grapevinegeneswasdetectedinatleast onetimetransition,basedoncuffdiffaverageFPKMacross repli-cates.Theclassof10–200FPKMvaluesincludedahighernumber ofgenesinthethreetimepointssampled(Table3).

3.2. Genomefunctionalannotation

Functional annotation of the genome was required for GO enrichment analysesand gene categorisation (see sub-sections 3.5and3.7).Theoverallfunctionalannotationandgene assign-mentwasmadeusingsequencesimilaritysearchesbyRapsearch2 against the KOG and NCBI ‘nr’ databases. The Vitis genome is 96.9%annotated,whichcorrespondsto29048ofthetotal29971 identifiedgenesinthegrapevinegenome.KOGannotation assign-mentsclassified17385genes(58.0%ofthegenome)distributed among 27 different categories, of which the most significant andrepresentativeweresignaltransductionmechanisms(7.5%), post-translationalmodification,proteinturnover,andchaperones (5.0%), and transcription (3.1%). Based ontheKEGG annotation of the genome,2281 genes were identified and assigned to at least one of the 130 pathways. Predominant pathways were purinemetabolism,starchandsucrosemetabolism,andthiamine metabolism,with433,274, and 226assignedunigenes, respec-tively.UsingGOterms,14915geneswerecategorised,annotated, anddistributedintothreemainGOdomains(SupplementaryFig. A.3).

Table3

Classesoftranscriptabundanceateachfruitset(FS)stage.Numbersoftranscripts

detectedatvariouslevelsofabundanceateachtimepoint,ascalculatedby

frag-mentsperkilobaseofexonpermillionfragmentsmapped(FPKM),consideringthe

datafromthreebiologicalreplicates.

FS1 FS2 FS3 FPKM>400 341 355 321 FPKM200–400 484 473 471 FPKM10–200 11818 11869 11875 FPKM1–10 7098 6941 7082 FPKM<1 4944 5240 4765 Total 24685 24878 24484

The‘ThompsonSeedless’transcriptomewascomparedtothe

referencegenomeforpolymorphisms.Fromthetotal104595SNPs

identified,71258werefoundtobelocatedincodingregions,22

inintrons,14128inexons,and33315inuncharacterisedregions.

Theanalysisof thetranscribedportionof the‘Thompson

Seed-less’genomealsoallowedforidentifying497newsplicejunctions

(SupplementaryTableB.1).

3.3. Distributionofgeneexpressionpatterns

The25703detectedexpressinggenesweregroupedintonine

pre-definedclusters,bringingtogethergeneswithsimilar

expres-sionpatternsduringthethreetimepointssampled(Fig.2).The

clusterrepresentedbythehighestnumberofgeneswasCluster9, composedofatotalof3699members.Ontheoppositeend, Clus-ter3wasrepresentedby2044genes.SelectionofKOGannotations perclusterretrieved24categoriesassociatedwiththegenes repre-sented.WiththeexceptionofCluster4,inwhichpost-translational modification, protein turnover, and chaperones comprised the mostrepresented category, allclusterswere enriched in signal transduction mechanism-related genes. Cluster 4 was also dif-ferentiated bya higherrepresentation ofgenesassociated with translation,ribosomalstructureandbiogenesis.

3.4. Overalltranscriptomeprofile

A total of 496 DEGs during at leastone of the time transi-tionsinvestigatedwasidentified.Thelistofallgenessignificantly affectedduringfruitset,itsannotationregardingfunctional cate-goriesandgenecodeidentification,thepattern-relatedgenecluster andtherespectivefoldchangearegiveninSupplementaryTable B.2.TheexpressionpatternrepresentedbyCluster1showedthe highestnumberofDEGs,with181genes.Duringthe4-dayperiod defined,thehighestdifferenceingeneexpressionoccurredduring thetransitionfromFS1toFS2,with307DEGsobserved,whilein thefollowing2days,fromFS2toFS3,only137genesshowed dif-ferentialexpression.Consideringthewholetimespaninvestigated, 108additionalgenesshowedstatisticalsignificanceinexpression fromFS1toFS3(Fig.3andSupplementaryTableB.2).

PCoAand hierarchicalclusteringbasedonthe496DEGsare showninFig.4.Principalcoordinate(PC)1andPC2incombination explained71.04%ofthetotalvarianceofthedataset(Fig.4A).In general,PC1differentiatedFS1fromtheremainingclusters,while Table2

Percentage(%)ofmappedreadsineachfruitsetstage.Eachvaluecorrespondstothemeanofthreeindependentbiologicalreplicates±standarderror(SE).Readssingly

mappedarethosethatalignedconcordantly,whiletheirmatepairwasnotmappedbythesoftware.

Fruitsetstage Mappedreads Pairedreads

aligned concordantly Multiple alignmentsof readpairs Readssingly mapped Multiple alignmentsof singlereads FS1 61.5±4.8 54.6±4.4 1.63±0.03 6.90±0.44 1.70±0.06 FS2 66.0±3.1 58.7±2.8 1.63±0.03 7.30±0.37 1.73±0.03 FS3 59.8±3.4 53.0±3.1 1.63±0.03 6.77±0.35 1.73±0.03

Fig.2. Clusteringofgenesaccordingtotheexpressionpattern.Clusteringwasperformedfor25703expressedgenesinninepre-definedclusterswithdistinctexpression profiles,usingtheMfuzzRpackage.Profilesarerepresentedbyexpressionchangesandtimepoints(fruitsetstage(FS)1,FS2,andFS3).Associatedwiththeclustersare euKaryoticOrthologousGroups(KOG)annotatedcategorieswiththeirrespectivepercentagerelativetothetotalnumbersofgenesofeachcluster:Cluster1,3175genes; Cluster2,2860genes;Cluster3,2044genes;Cluster4,2382genes;Cluster5,2915genes;Cluster6,2409genes;Cluster7,3199genes;Cluster8,3019;Cluster9,3699.

PC2allowedfordistinguishingFS2fromFS3samples.Ineachcase,

biologicalsampleswereclusteredtogether,andtheyaredirectly

connectedbytheminimum-spanningtree.Theseresultsindicate

thatinflorescencesfromthesampledtimepointswereaffected

significantlyin theoveralltranscriptome dynamics,providing a

suitable experimental dataset. Hierarchical clustering (Fig. 4B)

showedtheassociationamongsamplesaccordingtotheoverall transcriptomeprofile.Biologicalreplicatescollectedateachstage wereclusteredtogetherwithstrongconfidencebasedonbootstrap

analysesandPearsoncorrelations(Fig.4BandSupplementaryFig. A.4).

3.5. GOenrichmentanalysis

EnrichedcategoriesofDEGsareunevenlyrepresentedacross thedistributionofcategoriesofthegrapevinegenome,suggesting thatpredominanceofagivenpathwayisnotobservedbychance. Fromatotalof496DEGsenrichedin188GOterms,ahigh propor-tion(101)ofGOtermscorrespondedtobiologicalprocesses—81

Fig.3. Diagramshowingthenumbersandtrendsofdifferentiallyexpressedgenes

(DEGs)betweeneachofthethreetimepointsinvestigatedduringfruitset.

Val-uesindicategenespassingcutoffvaluesof−1.5≥log2foldchange≥1.5andfalse

discoveryrate(FDR)=0.05.Greenandredvaluesindicateup-regulatedand

down-regulatedgenes,respectively.Greyvaluesrepresentthenumbersofgenesshared

betweenthetwocomparisons.

to molecularfunction and six to cellularcomponents (Supple-mentaryTableB.3).Acyclicgraphsresultingfromtheenrichment analysisand showing thetop five and top five-related biologi-calprocesses,molecularfunctions,andcellularcomponentsmost affectedateachtimeintervalduringearlyfruitsetareprovided inSupplementaryFig.A.5.Amongbiologicalprocesses,termswith referencetosecondarymetabolitebiosynthesisandmetabolism, cellwall(CW)component biosynthesisandorganisation, carbo-hydrate and lipid biosynthesis, and metabolismwere themost enrichedatFS2andFS3,followedbytermsrelatedtogrowth, mor-phogenesis,pollengermination,andthereproductiveprocess.GO termsrelatedtoresponsetostress,abscission,anddehiscencewere enrichedonlyatFS3whencomparedtoFS1.Regardingmolecular functions,termsweremostlyrelatedtoacyltransferaseand sec-ondarymetabolitesynthase, oxidoreductase,lyase,andesterase activities.Regardingcellularcomponents,themostenriched cat-egorywascellmembrane.Amongthiscategory,pollentubeand mitochondrialmembranetermswerefoundtobemorepresentin FS2andFS3samples,respectively.

3.6. Overallmetabolomicprofile

Fromthe215metabolitessearchedbytheglobalmetabolic anal-ysesconducted,atotalof213weredetectedinatleastoneofthe conditions,and40changedintheirabundanceduringthethree fruitsetstages(SupplementaryTableB.4).Therelativecontentsof 19ofthemetaboliteswerealteredbetweenthefirstandsecond samplingpoints,and28werealteredinthetransitionfromFS1to FS3(11and17weremoreandlessabundantatFS3comparedto FS1,respectively)(Fig.5).

PCoA showed that, based on 71.33% of the total variation explained bythecombination ofPC1 andPC2 endorsedbythe metaboliteprofile,FS1samplesaredifferentfromFS2andFS3,but theselater twotimepointsareindistinguishablebasedontheir quantifiedmetabolome(Fig.6A).Theobserveddispersionbetween replicatesof FS2 andFS3 in theplot canbe explainedonly by biologicalvariation.Individual-plotpairwisecomparisonsofFS1 withFS2andFS1withFS3(SupplementaryFig.A.6)highlighted thatFS1samplesaredistinguishablefrombothFS2andFS3and that thetransitionfrom FS1to FS3would generate moreclear informationtounderstandmetabolomicdynamicsresultingfrom development.Hierarchicalclustering(Fig.6B)basedon

metabo-liteprofileshowedthatonlysamplesfromFS1formedacluster thatwassignificantlyseparatedfromtheothersamples. Regard-ing metabolite association, a cluster composed of putrescine, 2-isopropylmalate,tartarate,adenosine5-monophosphate(AMP), galactinol,andadenosine, whichshowedtobeincreasedin the transitionfromFS1toFS2andFS3,wasseparatedfromtheother metabolites.Diaminopropaneand␥-tocotrienolshowedthe oppo-sitepatternandweregroupedintoadifferentcluster.

3.7. Transcriptomicandmetabolomicprofilebyfunctional category

Fig.7showsthefunctionalannotationdistributionofthe269 DEGs, from which 64 and 205 were automatically and manu-allyassignedtoKOGcategories,respectively.Fromtheremaining 227DEGs,13wereclassifiedas‘otherfunction’and214as ‘gen-eralandunknownfunction’(SupplementaryTableB.2).Secondary metabolitebiosynthesis,transport,andmetabolism;carbohydrate transport andmetabolism;and signaltransduction mechanisms werethemostrepresentativefunctionalcategoriesofDEGs dur-ing both FS1–FS2 and FS2–FS3 time intervals. In thetransition from FS1 to FS3, secondary metabolite biosynthesis, transport, and metabolism; carbohydrate transport and metabolism; and post-translationalmodification,proteinturnover,andchaperones categorieswereenriched.Table4showsthesummaryoftheDEGs duringfruitsetaggregatedbygenefamilyandfunctionalcategory andnumbersofgenesup-anddown-regulatedbetweeneachtime point.

Fig.8A showsthelistof28 metaboliteswhoserelative con-tentsweresignificantlyaffectedatFS3 withrespecttoFS1and themetabolites’assignedfunctionalcategories,KEGGcompound number,andrespectivefoldchange.Amongthesemetabolites,the aminoacidclasswasthemostprevalent(28%),followedby carbo-hydrate(25%);secondarymetabolism(14%);lipid(11%);cofactor, prosthetic group, and electron carrier (11%); and nucleotide (11%). The pattern of relative abundance of the eight most affectedmetabolitesduringflower-to-fruittransitionisshownin Fig.8B.Amongthesemetabolites,2-oxodipate,galactinol,andtwo metabolites from nucleotide metabolism (adenosine and AMP) were increased at FS3, while two metabolites from secondary metabolism (ferulate and gamma-tocotrienol), glycerol, and 2-hydroxyplamitatedecreasedinabundanceduringthesameperiod. AsshowninFig.9,whenintegratingenzymaticandmetabolic changesonmetabolicpathways,theflower-to-fruittransitionstage showsitiscomposedofchangesinspecificstepsmappedondiverse metabolicpathways.Thecompleteenzymeidentificationamong DEGsandrespectivefoldchangesaregiveninSupplementaryTable B.3.

3.7.1. Aminoacidtransportandmetabolism

Genesencodingadenylylsulphatereductases(EC1.8.4.8)that participateinsulphurmetabolism,ahypotheticalproteininvolved in glutathione metabolism (EC 2.3.2.2), and amino acid and oligopeptide transporter-related genes were down-regulated. Genesencodinggelatinasesandperoxidases(EC1.11.1.7)involved inphenylalaninemetabolismandencodinganendopeptidasewere alsodifferentiallyexpressed.Ontheotherhand,aglobalincrease ofsomemetabolitesfromaminoacidmetabolismwasobserved, composedofshikimateandtryptophanfromtheshikimate path-way;2-oxodipate and2-isopropylmalate,involvedinlysineand leucinebiosynthesis,respectively;andputrescinefrompolyamine metabolism. Aspartate accumulation showed the opposite pat-tern.Anincreaseofmethioninesulphoxide(theoxidisedformof methionine)andadecreaseofS-adenosylhomocysteine,withboth metabolites putatively involved in ethylene biosynthesis, were verified.

Table4

DEGscombinedbyKOGfunctionalannotation.Thenumbersofup-anddown-regulatedgenesareindicatedinthegreenandredsquares,respectively,ateachtimeinterval.

TheidentityofthosegenesisprovidedinSupplementaryTableB.2.(Forinterpretationofthereferencestocolourinthistablelegend,thereaderisreferredtothewebversion

Fig.4. Principalcoordinateanalysis(PCoA)(A)andhierarchicalclustering(B)ofexpressionvaluesatdifferentsampledperiods.(A)Green,grey,andredindicatethethree differenttimepoints–FS1,FS2,andFS3–andtheirrespectivereplicates.Samplesareconnectedbyaminimum-spanningtree.Principalcoordinate(PC)1explains45.32% andPC225.72%ofthetotalvariation.(B)Differentcolumnsrepresentindependentbiologicalreplicates.Yellowtonesrepresentgeneswithhigherexpression,whileblue tonesrepresentgeneswithlowerexpression.Thestrengthofeachdendrogramnodewasestimatedwithabootstrapanalysisusing1000permutations.Valuesrepresented ontheleftsideofinternalnodesaretheapproximateunbiasedp-values(AU);boldanditalicvaluesontherightsiderepresentthebootstrapprobabilityvalue(BP).(For interpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

3.7.2. Carbohydratetransportandmetabolism

DEGs involved in carbohydrate metabolism were

down-regulatedduringthefirsttimeintervalinvestigated,while,inthe

secondinterval,themajorityofDEGswereup-regulated.Cluster

5(Fig.2)washighlyrepresentedwithinthisfunctionalcategory, includinggenesencodingcellulosesynthases(EC2.4.1.12),

beta-Fig.5. Diagramshowingthemetaboliteswithsignificantlydifferentabundances

ateachofthethreetimepointssampledduringearlyfruitset.Valuesindicate

genespassingthep-value≤0.05cutoff.Greenandredvaluesindicatenumbersof

metaboliteswithincreasedanddecreasedaccumulation,respectively.Greyvalues

representthenumberofmetabolitessharedbetweenthetwocomparisons.

galactosidases(EC3.2.1.23),andatonoplast-intrinsicaquaporin. However,genes encoding eugenolsynthase, sugar transporters, expansin,amylase (EC3.2.1.2), and UDP-glucuronicacid decar-boxylase(EC4.1.1.35)wereexclusivelydown-regulatedduringthe fruitsetstagesstudied.Genesencoding proteinsofglycosyl(EC 2.4.1.14),galactosyl,andgalacturonosyl(E.C2.4.1.134)transferases and CW-relatedenzymes, suchasmembers frompectate lyase (EC4.2.2.2),endoglucanase,polygalacturonase(EC3.2.1.15), glu-cosidase,andpectinesterase/pectinesteraseinhibitor(EC3.1.1.11) families,werealsodifferentiallyexpressedduringtheperiod inves-tigated. Regarding metabolites, pyruvate, citrate, fumarate, and chiro-inositolwereshowntobelessabundantatFS3thaninFS1. Conversely,increasedgalactinol,ribitol,andtartaratewere mea-sured.

3.7.3. Coenzymetransportandmetabolism

Genes encoding thiazole, ketoacyl-CoA, and chalcone syn-thase enzymes were down-regulated at FS2 compared to FS1, whilehydroxycinnamoyl-CoAshikimate/quinatewasup-regulated at the same period. S-adenosylmethionine (SAM)synthase (EC 2.5.1.6),involved inethylenebiosynthesis,andbluecopper pro-teinswerealsodifferentiallyexpressed.Nicotinateribonucleoside andmethylphosphaterelativecontentsdecreasedduringthestages investigated.

3.7.4. Inorganiciontransportandmetabolismandintracellular trafficking

The majority of genes encoding inorganic ion transporter proteins showed a down-regulation pattern. Genes encoding K+_/H+_{-antiportersandplasmamembraneH}+_{-transportingATPase} weredown-regulated duringthefirst time interval butshiftto an up-regulation in the second interval, while carbonic anhy-drasesshowedanup-regulatedpatternduringthewholeperiod sampled.Adown-regulationofmembersencodingexocyst compo-nentsandsecretorycarriermembraneproteinswasalsoperceived. GenesencodingSNAREandclathrinassemblyproteinswere down-regulatedatthefirsttimeintervalandup-regulatedatthesecond one.

3.7.5. Lipidtransportandmetabolism

The majority of genes involved in lipid transport and metabolismpathwaysweredown-regulatedatFS2incomparison with FS1, including, among others, genes encoding

medium-chain-fatty-acid-CoA ligase (EC 6.2.1.25), acyl-desaturases (EC 1.14.19.2), and a WAX protein. Two of these genes, encod-ing patellin and monoacylglycerol lipase proteins, inverted the expressiontrendatFS3comparedtoFS2.Non-specificlipid trans-ferproteinswereup-regulatedduringboth time intervals.Two genesthatencodeO-acyltransferases,O-ACYLTRANSFERASEWSD1 andO-ACYLTRANSFERASEWSD1-LIKE,assignedunderthedifferent enzymaticclassificationsEC2.3.1.20andEC2.3.1.75,were down-andup-regulated,respectively.Fattyacids,glycerolipids,andsterol metabolitesshoweddecreasedamountsinthesampled inflores-cencesduringthestagesinvestigated.

3.7.6. Nucleotidetransportandmetabolism

Onegeneclassifiedtoencodeadihydropyrimidinase(EC3.5.2.2) wasdown-regulated, and other encoding a putative RNA poly-merase(EC2.7.7.6)wasup-regulatedduringthetransitionfrom FS2toFS3.AdenosineandAMPhadaccumulatedatFS3compared to FS1, while N6-carbamoylthreonyladenosine relative content decreased.

3.7.7. Secondarymetabolitebiosynthesis,transport,and catabolism

Mostofthegenesinvolvedinsecondarymetabolite biosynthe-sis,transport,andcatabolismshowedadown-regulationpattern and cluster 2 (Fig. 2)washighly represented withinthis func-tionalcategory, including22 genesencoding stilbenesynthases (EC2.3.1.95),whichareinvolvedinphenylpropanoidmetabolism, andmyrcene(EC4.2.3.15)and limonenesynthases(EC4.2.3.16, EC 4.2.3.20), known to be involved in terpenoid metabolism. On the other hand,genes encoding linalooland nerolidol syn-thases(EC4.2.3.25),alsoinvolvedinterpenoidmetabolism,were up-regulatedinthetransitionfromFS1 toFS2.Genesencoding l-ascorbateoxidases,ABCtransporters,cytochromeP450-related proteins, carotenoidcleavagedioxygenase, andflavonol sulpho-transferasewereamongtheotherDEGs.

Allmetabolites assigned tothe secondary metabolism func-tional category significantly decreased theirabundance at FS3, namely gamma-tocotrienol from tocopherol metabolism, feru-late, involved in phenylpropanoid metabolism, procyanidin B1 andtrimerfromflavonoidmetabolism,andhydroquinone beta-D-glucopyranosidefromthebenzenoidfamily.

3.7.8. Cellcyclecontrolandcytoskeleton

Genesfromthecellcyclecontrol,celldivision, and chromo-some partitioning category were up-regulated.Genes encoding actin-related proteins, assigned to cytoskeleton function, were down-regulatedatFS2comparedtoFS1andup-regulatedduring thesecondtimeinterval,whilekinesinprotein-relatedgeneswere down-regulatedatbothtimeintervals.

3.7.9. Energyproductionandconversion

Genesencoding sarcosine oxidase(EC 1.5.3.1) and cytokinin dehydrogenase (EC 1.5.99.12) were up-regulated, while those encoding a voltage-gated shaker and vacuolar H+_{-ATPase were} down-regulated.Onegeneencodingaglycerophosphoryldiester phosphodiesterasewasdown-regulatedatFS2compared toFS1 butup-regulatedduringthelasttimeinterval.

3.7.10. Transcriptionfactors,translation,andpost-translational modification

Genes from the MYB transcription factor superfamily were mostly up-regulated during the timespan investigated, while MADS-box, MEIS and therelated HOX domain, elongation fac-tor SPT6, negative regulator of transcription and GATA-4/5/6 transcription factors weredown-regulated. Genes encoding the

Fig.6.Principalcoordinateanalysis(PCoA)(A)andhierarchicalclustering(B)ofmetaboliterelativecontentatthedifferenttimes.(A)Green,grey,andredindicatethethree

differenttimepoints–fruitsetstage(FS)1,FS2,andFS3–andtheirrespectivereplicates.Samplesareconnectedbyaminimum-spanningtree.Principalcoordinate(PC)

1explains55.74%andPC215.59%ofthetotalvariationexplainedbytheplot.(B)Theindependentbiologicalreplicatesareillustratedindifferentcolumns.Yellowtones

representmoreabundantmetabolites,whilebluetonesrepresentlessabundantmetabolites.Thestrengthofdendrogramnodeswasestimatedwithabootstrapanalysis

using1000permutations.Valuesrepresentedontheleftsideofinternalnodesaretheapproximateunbiasedp-values(AU);boldanditalicvaluesontherightsiderepresented

thebootstrapprobabilityvalue(BP).(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

AP2-likeethylene-responsivefactor, GT-2,and heat-shock tran-scriptionfactorswereup-regulatedinthetransitionfromFS1to FS2,FS2toFS3,and FS1toFS3,respectively. Genesencoding a tRNAmethyltransferaseandribosomalproteins, fromthe trans-lation, ribosomal structure, and biogenesis functional category, weredown-regulatedinoursampleset.Conversely,the

major-ityof genesinvolved inpost-translational modification,protein turnover,andchaperonesshowedanup-regulatedpattern. 3.7.11. Signaltransduction,defencemechanisms,andother functions

Regardingthe signaltransductionmechanisms, themajority ofDEGsweredown-regulatedduringthetransitionfromFS1to

Fig.7.Functionalannotationdistributionforthe269differentiallyexpressedgenes(DEGs)assignedtoaspecificeuKaryoticOrthologousGroups(KOG)functionalcategory.

Secondarymetabolitebiosynthesis,transport,andmetabolismandcarbohydratetransportandmetabolismwerethemostrepresentativefunctionalcategories.Inthe

transitionfromFS1toFS3,thepost-translationalmodification,proteinturnoverandchaperonescategoryalsowasparticularlyenriched.

FS2andincludedgenesencodingserine/threonineproteinkinases, mitogen-activated protein kinase kinase kinases, and calcium-bindingmessenger proteins. Among thesegenes, five encoding serine/threonineproteinskinases,twocalcium-bindingmessenger proteins,andonemembraneproteinwereup-regulatedfromFS2 toFS3.

AllDEGsidentifiedinvolvedindefencemechanismswere down-regulatedinthetransitionfromFS1toFS2excepttheCHITINASE gene,whichwasup-regulatedunderapatternrepresentedby Clus-ter8(Fig.2andTable4).Genesinvolvedinhormonemetabolism, namely1-AMINOCYCLOPROPANE-1-CARBOXYLATEOXIDASE(ACO), GA20ox,andINDOLE-3-ACETATE-O-METHYLTRANSFERASE1(IAMT1) weredifferentiallyexpressed.DEGsencodingasenescence-related protein,MATEeffluxfamilyproteins,majorfacilitatorsuperfamily membranetransportproteins,andmajorallergenproteinswere down-regulatedduringthetimeperiodtargeted.Genesencodinga pollen-specificproteinweredown-regulatedinthetransitionfrom FS1toFS2andup-regulatedduringthefollowingtimeinterval.

4. Discussion

4.1. RNA-Seqcomputationaldatavalidationbyanalysisof variabilityofbiologicalreplicates

Usingnext-generationsequencingtechnologyasaconvenient tool for comparativetranscriptome analyses of species withor withoutgenomesequences[54,55],thetechnicalvariabilityofthe dataisincreasinglynegligible,whiletheestablishmentof biolog-icalreplicatesallows fordecidingwhetherobserveddifferences betweengroupsoforganismsexposedtodifferenttreatmentsare simplyrandomorrepresenta‘true’biologicaldifferenceinduced byagiventreatment.Followingtheapproachpreviouslyproposed [56,57], the significant correlation verified between biological

replicatesusedaspartoftheexperimentaldesign (Supplemen-taryFig.A.4)confirmsthatourRNA-Seqdataisreproducibleand precise.Previousstudiesaddressingdichotomybetween biolog-icaland technical replicatesshowedthat biological variation is oftenlargerthantechnicalvariation,underscoringtheimportance ofincludingbiologicalreplicatesinthestudydesign[58].Taking advantageoftheuseofbiologicalreplicates,thePCoAand hierar-chicalclusteringconfirmedthesampleseparationintothreeclasses sharingsimilarexpressionsignaturesaccordingtothespecifictime pointduringfruitset(Fig.4).

4.2. Geneexpressionandmetabolicpatternsduringfruitset Thegeneralmoleculardynamicsthatoccurredduringfruitset highlightedagenedown-regulationpatternobservedfor5days after100%capfallinmostfunctionalcategories(Table4)thatwas accompaniedbydecreasedmetaboliteabundance,mainly concern-ing carbohydrates,cofactors, lipids, and secondary metabolism, duringthelastinvestigatedtimepoint(Fig.8).Inaddition,atotal of12.5% ofDEGs showedaninversion trendin expression pat-ternsatFS2,indicatingthatatthistimepointadevelopmentalor stress-resultingsignaloccurs.Thesameexpressiondynamicswas observedintomatoforashortperiod,rangingfrom2daysbefore bloomto4daysafterfullbloom,demonstratingalackofcontinuum inovarydevelopmentfromflowerbudtopost-anthesis[24]. 4.3. Secondarymetabolismregulation

Functionalannotation(Fig.5)andthetopfiveGOenrichedterms (SupplementaryTableB.2)highlightedsecondarymetabolismas themostrepresentativebiologicalprocess.

ThelargemultigenicfamilyofMYBtranscriptionfactors,with membersknowntoberegulatedbysugarsandtocontrolflavonoid

Fig.8.Summaryofthemetabolitesthatsignificantlychangedinabundanceduringtheearlyfruitsetpointsinvestigated.(A)Listofmetabolitessignificantlyaffectedatfruit

setstage(FS)3comparedtoFS1(p-value≤0.05),assignedfunctionalcategories,KyotoEncyclopaediaofGenesandGenomes(KEGG)compoundnumber,andrespective

accumulationfoldchange.(B)Relativecontentevolutionofallmetaboliteswithhighlysignificantdifferences(|foldchange|≥1)betweenFS1andFS3.

biosynthesis[59],showedup-regulationduringfruitset(Table4). Basedonourresults,wecanhypothesisethatthesignallingsugar galactinoland hexokinaseenzymecouldbeinvolved in regula-tionofMYBexpression.SpecificMYBtranscriptionfactorswere previouslydescribedtobepositive[60,61]ornegative[62] regula-torsofflavonoidbiosynthesisinseveralspecies.Ourdatarevealed repressionofflavonoid-and phenylpropanoid-relatedpathways (Figs.8and9)andrepressionofthephenylalanineprecursor, ␤-alanine,biosynthesisviaspermidinebyprimaryamineoxidase(EC 1.4.3.21)andviauracilbydihydropyrimidinase(EC3.5.2.2)(Fig.9). Theobserveddown-regulationofgenesencodingABCtransporters (Table 4)indicated alsoreduced transport andaccumulation of flavonoidsininflorescencesduringfruitset[63].

Theregulationof cytochrome P450-relatedgenes; terpenoid metabolism (EC 4.2.3.15, EC 4.2.3.25); protective compound biosynthesis,suchasEUGENOLSYNTHASE1-LIKE,LACCASE-14-LIKE, andCHITINASE;genesencodingketoacyl-CoAsynthases;andWAX2 protein,whichcontributestocuticularwaxandsuberin biosynthe-sis,indicatedthataresponsetoenvironmentalstimuliisactivated duringfruitset,agreeingwithGOenrichedtermsassociatedwith environmentalstressresponses and biotic stimuli (Table 4 and SupplementaryTableB.3).Theregulationofl-ASCORBATEOXIDASE HOMOLOGduringthethreefruitsetstagesinvestigated,withthe

changeofexpressionpatternatFS2,alsoindicatesaresponseto oxidativestressconditions.

4.4. Regulationofnutrienttransportinthedevelopingfruit

Thepassivesymplasticphloemunloadingpathwayinthefruit walland seedcoat,throughplasmodesmata,predominates dur-ingtheearlystageofberrydevelopment[64].However,ourdata, which indicate that extensive transcriptome reprogrammingof nutrienttransporters(Table4)isinduced,agreewiththeevidence thatnutrienttransportacrossmembranesrequiring membrane-embeddedamino acids,peptides, sugar, secondary metabolites, andinorganiciontransportersisstronglyregulatedduringfruit set[25,65,66].

Theobserveddown-regulationofmembranesugartransporter genes,namelythoseencodingsucrosetransportproteinSUC2and bidirectional sugar transporter SWEET1,4, 5, and 14, indicates a down-regulation of phloem long-distancetransport (sucrose) andphotosynthateshort-termtransportoraccumulation(glucose) levels[67–70].TheexpressionofSWEET14wasfoundinthe flower-to-fruittransitionstagewitha4-foldrepressionatFS3compared toFS1(SupplementaryTableB.2),suggestingasinklimitationfor sugarspreviously observed in fruit set(reviewed by [25]).The

Fig.9.Changesinenzymegeneexpressionandmetabolitequantificationmappedontosimplifiedmetabolicpathways.Redandgreenarrowsrepresentchangesin

metabo-liteaccumulationatfruitsetstage(FS)3comparedtoFS1.Redandgreensquaresrepresentdown-andup-regulationofthetranscriptsattheFS2stage(corresponding

totransitionfromFS1toFS2)andFS3(comprisetransitionsfromFS1toFS3andfromFS2toFS3).Descriptionofenzymecodes:1.11.1.7,peroxidase;1.14.19.2,acyl

desaturase;1.4.3.21,primaryamineoxidase;1.5.3.1,sarcosineoxidase;1.5.99.12,cytokinindehydrogenase;1.8.4.8,adenylylsulphatereductase;2.1.1.37,cytosine

5-methyltransferase;2.3.1.20,acyltransferase;2.3.1.75,wax-estersynthase;2.3.1.92,sinapoyltransferase;2.3.1.95,stilbenesynthase;2.3.2.2,␥-glutamyltransferase;2.4.1.12,

cellulosesynthase;2.4.1.134,␤-galactosyltransferase;2.4.1.14,sucrose-phosphatesynthase;2.5.1.6,S-adenosylmethioninesynthase;2.7.1.1,hexokinase;2.7.7.6,RNA

poly-merase;3.1.1.11,pectinesterase;3.2.1.15,polygalacturonase;3.2.1.2,␤-amylase;3.2.1.23,␤-1,3-galactosidase;3.5.2.2,dihydropyrimidinase;3.5.4.10,IMPcyclohydrolase;

3.6.1.3,ATPmonophosphatase;4.1.1.35,UDP-glucuronatedecarboxylase;4.2.2.2,pectatelyase;4.2.3.15,myrcenesynthase;4.2.3.16,(4S)-limonenesynthase;4.2.3.20,

(R)-limonenesynthase;4.2.3.25,linaloolsynthase.(Forinterpretationofthereferencestocolourinthisfigurelegend,thereaderisreferredtothewebversionofthisarticle.)

geneVvSUC2/SUT2,whichparticipatesinphloemorpost-phloem unloading,firstdescribedasweaklyexpressedinberries[71],was showntobeinvolvedinonsetofberrydevelopment.Inadditionto theknownimpairedflowerformationandberrysetresultingfrom carbonlimitationduetoinadequateratesofsourceleaf photosyn-thesis[9,72],theseresultsindicatethesinkstrengthlimitationis animportantstepduringtheflower-to-fruittransitionstage. 4.5. Carbohydratemetabolismandotherenergysources

Therepressionofcarbohydrate-relatedpathwaysatthe tran-scriptomelevelduringthefirsttimeintervalinvestigated,which canbecorrelatedwithdecreased metabolitecontentassociated withglycolysisand thetricarboxylic acidcycle,wasinvertedat FS2(Fig.8andTable4)and couldindicatethestartingpointof cellmetabolismstimulationthatwouldleadtoberryformation. The perceivedup-regulation of two genes encoding glycogenin enzymes (VIT07s0005g01970, VIT07s0005g01980), among the glycosyltransferases, suggeststhat conversion ofglucose tothe energystoragepolymerglycogenisfavoured(SupplementaryTable

B.2).Thedemandforenergycanalsobefulfilledbyincreased glyco-genbiosynthesis via sarcosine oxidase(EC 1.5.3.1). In addition, adeninesynthesisedviacytokinindehydrogenase (EC1.5.99.12) canbeconvertedtoadenosinewiththeincreasedlevelsobserved andcanbeinvolvedinenergytransferprocesses(ATPandADP) (Figs.8and9).

CWbiosynthesisaccompaniesthemandatoryremodellingthat provides the flexibility required for cell expansion during fer-tilisedflowerandpediceldevelopment[22,24]andisfine-tuned tocopewiththedoublesigmoidalgrowthpatternofberry devel-opment[73].EvidenceofCWremodellingprocesseswasgainedby changesinexpressionofgenesencodingglycosyltransferaseand cellulosesynthase(EC2.4.1.12)enzymes,whicharerequiredfor pectinandcellulosebiosynthesis;andexpansins,endoglucanases, pectinesterases(EC3.1.1.11),polygalacturonases(EC3.2.1.15),and ␤-1,3-galactosidases (EC 3.2.1.23),which facilitate CW polymer modification(Fig.9andTable4).GenesencodingCW disassembly-associated enzymescan alsobeimplicated in abscission of the non-fertilisedflowerorgans[74],whicharepresentinourwhole inflorescencesamples.ThiscorroborateswiththeGOenrichment

analysis(SupplementaryTableB.3),whichrevealedtermsrelated toCWorganisationandmodificationandflowerorganabscission biologicalprocesses.

4.6. Sugarsignallingpathways

Sugarsmay control a wide range of activities in plantcells through transcriptional and post-translational regulation. The increasedgalactinolquantifiedduringfruitset(Fig.8)mayaffect expressionofdifferentgenes,suchasencodingsugartransporters, components of the photosystems, enzymes of carbohydrate metabolism,enzymesofsecondarymetabolism,andtranscription factors[59].TheregulationofHEXOKINASE-1(VIT09s0002g03390) (Table4)wasreportedtobeinvolved insugar-sensingand sig-nalling through CW invertase and sucrose synthase regulation [75]andsugarcontroloftransporters[59]ingrapeberries.Here, wecanassociatethisregulationwithearlystagesoffruitsetin grapevine,duringwhichhexosesugarsignalsthatregulatethecell cycleandcelldivisionprogrammesseemedtobegenerated[76]. Theenhancedexpressionofgenesencodingheat-shockproteins (HSP90,HSP70,andsHSP26/42)andoneheatstresstranscription factor(Table4andSupplementaryTableB.2)canbehypothesised toprotectCWinvertasefrommisfoldingforcorrecttargetingand function[77].Theseproteinsareknowntobeexpressedinresponse tostressconditionsandduringparticularstagesofthecellcycle [78],performchaperonefunctionbystabilisingnewproteins,and playimportantrolesinsignalling,trafficking,andpathogen resis-tance[79].

4.7. Hormonemetabolismandsignallingpathways

Thediscernedregulationofauxin-,GA-,andethylene-related genesduringfruitset(Table4)confirmstheimportanceofplant hormones as mediators of thefruit developmentalsignal after pollination [21–24]. Our data also revealed the mediation of auxin biosynthesis and signalling [80] via down-regulation of IAMT,an enzymeresponsiblefor inactivationofindole-3-acetic acid(IAA), and down-regulationof auxin-associated MADS-box genes [24]. It can be speculated that these two MADS-box genes(VIT14s0066g01640,VIT18s0001g01760)(Supplementary TableB.2)areinstrumental intriggeringthegrapevinefruitset programme. The up-regulation of GIBBERELLIN 20 OXIDASE 2 (GA20ox2)(Table4)suggeststhatGAbiosynthesisoccursduring initialsteps of fruit set, as previously proposed [21,81]. Addi-tionally, the observed regulation of genes encoding predicted ACO(Table4)andSAMsynthase1-like(EC2.5.1.6)(Fig.8)and theincreased methionine sulphoxide contentderived fromthe intermediate methionine indicated changes in ethylene-related pathways,aspreviouslyreported[24].Likewise,theup-regulation ofageneencodinganAP2-likeethylene-responsivetranscription factorsuggestedtheactivationoftheethylenesignallingpathway toactasafruitsettrigger[24].Sincepolyaminesandethyleneshare SAMasacommonintermediate,SAMmaybealternatively chan-nelledtowardthepolyaminepathway,whichisalsocloselyrelated tofruitset[82].Theobservedincreaseinputrescineindicatedan inductionofpolyaminebiosynthesisinthegrapevinepost-anthesis stage,whichcontrasts withthepreviousobservationin tomato [24].

Therefore,itcanbediscussedthatthedown-regulationofthis IAAinhibitor probably resultsin increasedlevelsof active IAA, which,togetherwithinductionofethylenesignalling,would trig-gerfruitsetandwouldstimulatethesynthesisofbioactiveGAs responsibleforfruitgrowththroughpromptingcelldivisionand expansion.On theother hand,theobserved up-regulation of a cytokinindehydrogenase(VIT11s0016g02110)mightbe associ-atedwithadecreaseincytokininlevels.Previousworkingrapevine

demonstratedthataninductionofcytokininbiosynthesisoccurs about13daysafterpollination[83].

4.8. Pollenviability,fertilisation,andcelldivisionphase

In‘ThompsonSeedless’,fruitsetisdependentonsuccessful pol-linationandfertilisation;however,seeddevelopmentabortsatan earlystageafterfertilisation[12].Theobservedregulationofgenes relatedtothepollenviability(Table4),includinggenesencoding enzymesresponsibleforcellulosesynthesis(EC2.4.1.12),pectate lyase(EC4.2.2.2),amylase(EC3.2.1.2),andacyl-CoAsynthase, iden-tifiedspecificmembersforsynthesisofenergy,cellulose,callose, andsporopolleninforpollendevelopment[84–86].PollentubeCW developmentalsorequireschangesinlipidmetabolism,suchas palmitatethatcanbeconvertedinto2-hydroxypalmitate,which showeddecreasedrelative abundance3–7d(Fig.8).Changesin gene expressioninvolved in activevesicle traffickingtodeliver secretoryvesiclessuggestpolarisedrapidgrowthpollentube reg-ulationviaexo-andendocytosis[87](Table4).Thisevidenceand theregulationofgenesencodingpollen-specificproteinsfromthe majorlatexproteinregulatoryfamily[88]agreeswiththeenriched GOtermscorrespondingtoantherdehiscencebiologicalprocesses andpollentubecellularcomponents.

Simultaneously,duringfruitsetstagesinvestigated,aperiodof rapidcelldivisiontakesplace,asverifiedbyup-regulationofgenes assignedtothecellcyclecontrol,celldivision,andchromosome partitioningfunctionalcategory(Table4),correspondingtoovary development[3].

5. Conclusions

During the decisive fruit set developmental stage, carbohy-dratepathway,secondarymetabolism,and proteinmodification and turnover mechanisms were the most affected functional categories. An increased cell division-related gene expression, requiredforfruitgrowthinitiation,wasverified.Genesencoding signallingproteins,suchasserine/threoninekinasesand calcium-bindingmessengerproteins, l-ascorbate oxidase,chitinase,and CW-modifyingenzymes,showedadiscontinuousparticipationin thisdevelopmentalstagewithachangeinexpressionpatternat FS2.Itwasalsoshowntobecharacterisedbypathwaysthat medi-atesugar-andhormone-controlledresponses.Theparticipationof auxin,GA,ethylene,andcytokinin-relatedgenesfromtheearlier fruitsetstageinvestigated (FS1),maintainingthesame expres-sionpatternuntilFS3,suggeststhathormoneactionisoneofthe constanttriggersignalsneededfortheflower-to-fruittransitionto proceed.Therefore,weproposethatrepressionofnutrient trans-porttogetherwithincreasedlevelsofactiveIAAandinductionin ethylenesignallingwilltriggerfruitsetandstimulatethesynthesis ofbioactiveGAsresponsibleforfruitgrowththroughstimulation of celldivision and expansion.Therefore, this first global tran-scriptomicandmetabolomicanalyseshelpelucidatethemolecular mechanismsoccurringininflorescencesduringtheflower-to-fruit transitionstage.

Acknowledgements

ThisworkwassupportedbythePortugueseFundac¸ãoparaa CiênciaeaTecnologia(FCT)throughproject‘VitiShade: PTDC/AGR-GPL/116923/2010’,thePh.D.grantSFRH/BD/69076/2010toSD,and byProDeRAction4.1:PROdUVA23921/2/3/4.WeacknowledgeDr. DannyAlexanderandhisteamatMetabolon(Durham,USA)for metabolomicanalysesandDr.MartinHeineandhisteamatLGC Genomics(Berlin,Germany)fortranscriptomesequencing.

AppendixA. Supplementarydata

Supplementarydataassociatedwiththisarticlecanbefound,in theonlineversion,athttp://dx.doi.org/10.1016/j.plantsci.2015.12. 009.

References

[1]B.G.Coombe,Relationshipofgrowthanddevelopmenttochangesinsugars, auxins,andgibberellinsinfruitofseededandseedlessvarietiesofVitis vinifera,PlantPhysiol.35(1960)241–250.

[2]N.K.Dokoozlian,Grapeberrygrowthanddevelopment,in:L.P.Christensen (Ed.),RaisinProductionManual,UniversityofCalifornia,Agriculturaland NaturalResources,Oakland,2000,pp.30–37.

[3]H.Ojeda,A.Deloire,A.Carbonneau,A.Ageorges,C.Romieu,Berry

developmentofgrapevines:relationsbetweenthegrowthofberriesandtheir DNAcontentindicatecellmultiplicationandenlargement,Vitis38(1999) 145–150.

[4]C.Pratt,Reproductiveanatomyincultivatedgrapes:areview,Am.J.Enol. Vitic.22(1971)92–109.

[5]J.S.Boyer,J.E.McLaughlin,Functionalreversiontoidentifycontrollinggenes inmultigenicresponses:analysisoffloralabortion,J.Exp.Bot.58(2007) 267–277.

[6]Y.Jin,D.A.Ni,Y.L.Ruan,Post-translationalelevationofcellwallinvertase activitybysilencingitsinhibitorintomatodelaysleafsenescenceand increasesseedweightandfruithexoselevel,PlantCell21(2009)2072–2089.

[7]W.M.Kliewer,Effectofhightemperaturesduringthebloom-setperiodon fruit-setovulefertility,andberrygrowthofseveralgrapecultivars,Am.J. Enol.Vitic.28(1977)215–222.

[8]H.Suwa,H.Hakata,H.A.El-Shemy,N.T.Adu-Gyamfi,S.Nguyen,P.K.Lightfoot, K.Mohapatra,Hightemperatureeffectsonphotosynthatepartitioningand sugarmetabolismduringearexpansioninmaize(ZeamaysL.)genotypes, PlantPhysiol.Biochem.48(2010)124–130.

[9]M.C.Vasconcelos,M.Greven,C.S.Winefield,M.C.T.Trought,V.Raw,The floweringprocessofVitisvinifera:areview,Am.J.Enol.Vitic.60(2009) 411–434.

[10]K.E.Zinn,M.Tunc-Ozdemir,J.J.Harper,Temperaturestressandplantsexual reproduction:uncoveringtheweakestlinks,J.Exp.Bot.61(2010)1959–1968.

[11]C.C.Nwafor,I.Gribaudo,A.Schneider,R.Wehrens,M.S.Grando,L.Costantini, Transcriptomeanalysisduringberrydevelopmentprovidesinsightsinto co-regulatedandalteredgeneexpressionbetweenaseededwinegrape varietyanditsseedlesssomaticvariant,BMCGenomics15(2014)1030–1051.

[12]F.Varoquaux,R.Blanvillain,M.Delseny,P.Gallois,Lessisbetter:new approachesforseedlessfruitproduction,TrendsBiotechnol.18(2000) 233–242.

[13]G.Lebon,B.Wojnarowiez,F.Holzapfel,N.Vaillant-Gaveau,C.Clément,Sugars andfloweringinthegrapevine(VitisviniferaL.),J.Exp.Bot.59(2008) 2565–2578.

[14]D.Werner,N.Gerlitz,R.Stadler,Adualswitchinphloemunloadingduring ovuledevelopmentinArabidopsis,Protoplasma248(2011)225–235.

[15]C.Zhang,K.Tanabe,F.Tamura,K.Matsumoto,A.Yoshida,13_{C-photosynthate}

accumulationinJapanesepearfruitduringtheperiodofrapidfruitgrowthis limitedbythesinkstrengthoffruitratherthanbythetransportcapacityof thepedicel,J.Exp.Bot.56(2005)2713–2719.

[16]M.DeJong,M.Wolters-Arts,R.Feron,C.Mariani,W.H.Vriezen,TheSolanum lycopersicumauxinresponsefactor7(SlARF7)regulatesauxinsignaling duringtomatofruitsetanddevelopment,PlantJ.57(2009)160–170.

[17]M.DeJong,M.Wolters-Arts,C.García-Martínez,W.H.Vriezen,TheSolanum lycopersicumauxinresponsefactor7(SlARF7)mediatescross-talkbetween auxinandgibberellinsignallingduringtomatofruitsetanddevelopment,J. Exp.Bot.62(2011)617–626.

[18]I.Olimpieri,F.Siligato,R.Caccia,L.Mariotti,N.Ceccarelli,G.P.Soressi,A. Mazzucato,Tomatofruitsetdrivenbypollinationorbytheparthenocarpic fruitallelearemediatedbytranscriptionallyregulatedgibberellin biosynthesis,Planta226(2007)877–888.

[19]L.Pascual,J.M.Blanca,J.Ca ˜nizares,F.Nuez,Analysisofgeneexpressionduring thefruitsetoftomato:acomparativeapproach,PlantSci.173(2007) 609–620.

[20]Z.Ren,Z.Li,Q.Miao,Y.Yang,W.Deng,Y.Hao,Theauxinreceptorhomologue inSolanumlycopersicumstimulatestomatofruitsetandleafmorphogenesis,J. Exp.Bot.62(2011)2815–2826.

[21]J.C.Serrani,R.Sanjuán,O.Ruiz-Rivero,M.Fos,J.L.García-Martínez,Gibberellin regulationoffruitsetandgrowthintomato,PlantPhysiol.145(2007) 246–257.

[22]W.H.Vriezen,R.Feron,F.Maretto,J.Keijman,C.Mariani,Changesintomato ovarytranscriptomedemonstratecomplexhormonalregulationoffruitset, NewPhytol.177(2008)60–76.

[23]H.Wang,B.Jones,Z.Li,P.Frasse,C.Delalande,F.Regad,S.Chaabouni,A. Latché,J.C.Pech,M.Bouzayen,ThetomatoAux/IAAtranscriptionfactorIAA9 isinvolvedinfruitdevelopmentandleafmorphogenesis,PlantCell17(2005) 2676–2692.

[24]H.Wang,N.Schauer,B.Usadel,P.Frasse,M.Zouine,M.Hernould,A.Latché, J.C.Pecha,A.R.Fernie,M.Bouzayen,Regulatoryfeaturesunderlying

pollination-dependentand-independenttomatofruitsetrevealedby transcriptandprimarymetaboliteprofiling,PlantCell21(2009)1428–1452.

[25]Y.L.Ruan,J.W.Patrick,M.Bouzayen,S.Osorio,A.R.Fernie,Molecular regulationofseedandfruitset,TrendsPlantSci.17(2012)656–665.

[26]J.E.McLaughlin,J.S.Boyer,Sugar-responsivegeneexpression,invertase activity,andsenescenceinabortingmaizeovariesatlowwaterpotentials, Ann.Bot.94(2004)675–689.

[27]P.DeDios,A.J.Matilla,M.Gallardo,Flowerfertilizationandfruitdevelopment promptchangesinfreepolyaminesandethyleneindamsonplum(Prunus insititiaL.),J.PlantPhysiol.163(2006)86–97.

[28]O.Jaillon,J.M.Aury,B.Noel,A.Policriti,C.Clepet,A.Casagrande,N.Choisne,S. Aubourg,N.Vitulo,C.Jubin,etal.,Thegrapevinegenomesequencesuggests ancestralhexaploidizationinmajorangiospermphyla,Nature449(2007) 463–467.

[29]R.Velasco,A.Zharkikh,M.Troggio,D.A.Cartwright,A.Cestaro,D.Pruss,M. Pindo,L.M.FitzGerald,S.Vezzulli,J.Reid,etal.,Ahighqualitydraftconsensus sequenceofthegenomeofaheterozygousgrapevinevariety,PLoSOne2 (2007)1326–1343.

[30]R.Cramer,K.A.Ghan,R.L.Schlauch,H.Tillett,A.Heymann,M.Ferrarini,S. Delledonne,M.Zenoni,Transcriptomicanalysisofthelatestagesofgrapevine (Vitisviniferacv.CabernetSauvignon)berryripeningrevealssignificant inductionofethylenesignalingandflavorpathwaysintheskin,BMCPlant Biol.14(2014)370–390.

[31]L.G.Deluc,J.Grimplet,M.D.Wheatley,R.L.Tillett,D.R.Quilici,C.Osborne,D.A. Schooley,K.A.Schlauch,J.C.Cushman,G.R.Cramer,Transcriptomicand metaboliteanalysesofCabernetSauvignongrapeberrydevelopment,BMC Genomics8(2007)429–470.

[32]J.Grimplet,L.G.Deluc,R.L.Tillett,M.D.Wheatley,K.A.Schlauch,G.R.Cramer, J.C.Cushman,Tissue-specificmRNAexpressionprofilingingrapeberry tissues,BMCGenomics8(2007)187–209.

[33]C.Sweetman,D.C.Wong,C.M.Ford,D.P.Drew,Transcriptomeanalysisatfour developmentalstagesofgrapeberry(Vitisviniferacv.Shiraz)provides insightsintoregulatedandcoordinatedgeneexpression,BMCGenomics13 (2012)691–715.

[34]L.Venturini,A.Ferrarini,S.Zenoni,G.B.Tornielli,M.Fasoli,S.DalSanto,A. Minio,G.Buson,P.Tononi,E.D.Zago,G.Zamperin,D.Bellin,M.Pezzotti,M. Delledonne,DenovotranscriptomecharacterizationofVitisviniferacv. Corvinaunveilsvarietaldiversity,BMCGenomics14(2013)41–53.

[35]S.Zenoni,A.Ferrarini,E.Giacomelli,L.Xumerle,M.Fasoli,G.Malerba,D. Bellin,M.Pezzotti,M.Delledonne,Characterizationoftranscriptional complexityduringberrydevelopmentinVitisviniferausingRNA-Seq,Plant Physiol.152(2010)1787–1795.

[36]D.Lorenz,K.Eichhorn,H.Bleiholder,R.Klose,U.Meier,E.Weber, Phänologischeentwicklungsstadienderrebe(VitisviniferaL.ssp.vinifera): codierungundbeschreibungnachdererweitertenBBCH-Skala,Vitic.Enol. Sci.49(1994)66–70.

[37]S.Chang,J.Cairney,AsimpleandefficientmethodforisolatingRNAfrompine trees,PlantMol.Biol.Rep.11(1993)113–116.

[38]A.M.Bolger,M.Lohse,B.Usadel,Trimmomatic:aflexibletrimmerforIllumina sequencedata,Bioinformatics30(2014)2114–2120.

[39]D.Kim,G.Pertea,C.Trapnell,H.Pimentel,R.Kelley,S.L.Salzberg,TopHat2: accuratealignmentoftranscriptomesinthepresenceofinsertions,deletions, andgenefusions,GenomeBiol14(2013)36–48.

[40]C.Trapnell,B.A.Williams,G.Pertea,A.Mortazavi,G.Kwan,M.J.vanBaren,S.L. Salzberg,B.J.Wold,L.Pachter,Transcriptassemblyandquantificationby RNA-Seqrevealsunannotatedtranscriptsandisoformswitchingduringcell differentiation,Nat.Biotechnol.28(2010)511–515.

[41]L.Kumar,M.E.Futschik,Mfuzz:asoftwarepackageforsoftclusteringof microarraydata,Bioinformation2(2007)5–7.

[42]Y.Shen,Z.Wan,C.Coarfa,R.Drabek,L.Chen,E.A.Ostrowski,Y.Liu,G.M. Weinstock,D.A.Wheeler,R.A.Gibbs,F.Yu,ASNPdiscoverymethodtoassess variantalleleprobabilityfromnext-generationresequencingdata,Genome Res.20(2010)273–280.

[43]M.I.Love,W.Huber,S.Anders,Moderatedestimationoffoldchangeand dispersionforRNA-SeqdatawithDESeq2,GenomeBiol.15(2014)550–570.

[44]Y.Zhao,H.Tang,Y.Ye,RAPSearch2:afastandmemory-efficientprotein similaritysearchtoolfornext-generationsequencingdata,Bioinformatics28 (2012)125–126.

[45]R.L.Tatusov,N.D.Fedorova,J.D.Jackson,A.R.Jacobs,B.Kiryutin,E.V.Koonin, D.M.Krylov,R.Mazumder,S.L.Mekhedov,A.N.Nikolskaya,B.S.Rao,S. Smirnov,A.V.Sverdlov,S.Vasudevan,Y.I.Wolf,J.J.Yin,D.A.Natale,TheCOG database:anupdatedversionincludeseukaryotes,BMCBioinformatics4 (2003)41–54.

[46]M.Kanehisa,M.Araki,S.Goto,M.Hattori,M.Hirakawa,M.Itoh,T.Katayama, S.Kawashima,S.Okuda,T.Tokimatsu,Y.Yamanishi,KEGGforlinking genomestolifeandtheenvironment,NucleicAcidsRes.36(2008)480–484.

[47]A.Conesa,S.Götz,J.M.García-Gómez,J.Terol,M.Talón,M.Robles,Blast2GO: auniversaltoolforannotation,visualization,andanalysisinfunctional genomicsresearch,Bioinformatics21(2005)3674–3676.

[48]A.Alexa,J.Rahnenfuhrer,topGO:enrichmentanalysisforgeneontology.R

Packageversion2.18.0,2010.

[49]A.Evans,C.DeHaven,T.Barrett,Integrated,nontargetedultrahigh performanceliquidchromatography/electrosprayionizationtandemmass spectrometryplatformfortheidentificationandrelativequantificationofthe small-moleculecomplementofbiologicalsystems,Anal.Chem.81(2009) 6656–6667.

[50]A.M.Evans,M.W.Mitchell,H.Dai,C.D.DeHaven,Categorizingion:featuresin liquidchromatography/massspectrometrymetobolomicsdata,

Metabolomics2(2012)110.

[51]T.Ohta,N.Masutomi,N.Tsutsui,T.Sakairi,M.Mitchell,M.V.Milburn,J.A. Ryals,K.D.Beebe,L.Guo,Untargetedmetabolomicprofilingasanevaluative tooloffenofibrate-inducedtoxicologyinFischer344malerats,Toxicol. Pathol.37(2009)521–535.

[52]F.J.Rohlf,NTsys-PC:NumericalTaxonomyandMultivariateAnalysisSystem, ExeterSoftware,Setauket,NewYork,2005.

[53]R.Suzuki,H.Shimodaira,Pvclust:anRpackageforassessingtheuncertainty inhierarchicalclustering,Bioinformatics22(2006)1540–1542.

[54]L.Gao,Z.J.Tu,B.P.Millett,J.M.Bradeen,Insightsintoorgan-specificpathogen defenseresponsesinplants:RNA-Seqanalysisofpotatotuber–Phytophthora infestansinteractions,BMCGenomics14(2013)340–351.

[55]J.J.Liu,R.N.Sturrock,R.Benton,TranscriptomeanalysisofPinusmonticola primaryneedlesbyRNA-Seqprovidesnovelinsightintohostresistanceto Cronartiumribicola,BMCGenomics14(2013)884–899.

[56]M.Fasoli,S.DalSanto,S.Zenoni,G.B.Tornielli,L.Farina,A.Zamboni,A. Porceddu,L.Venturini,M.Bicego,V.Murino,A.Ferrarini,M.Delledonne,M. Pezzotti,Thegrapevineexpressionatlasrevealsadeeptranscriptomeshift drivingtheentireplantintoamaturationprogram,PlantCell24(2012) 3489–3505.

[57]D.Thakare,R.Yang,J.G.Steffen,J.Zhan,D.Wang,R.M.Clark,X.Wang,R.Clark, X.Xiangfeng,R.Yadegari,RNA-Seqanalysisoflaser-capturemicrodissected cellsofthedevelopingcentralstarchyendospermofmaize,GenomicsData2 (2014)242–245.

[58]J.H.Bullard,E.Purdom,K.D.Hansen,D.Dudoit,Evaluationofstatistical methodsfornormalizationanddifferentialexpressioninmRNA-Seq experiments,BMCBioinformatics11(2010)94–106.

[59]F.Lecourieux,C.Kappel,D.Lecourieux,A.Serrano,E.Torres,P.Arce-Johnson, S.Delrot,Anupdateonsugartransportandsignallingingrapevine,J.Exp.Bot. 65(2013)821–832.

[60]J.Bogs,F.W.Jaffé,A.M.Takos,A.R.Walker,S.P.Robinson,Thegrapevine transcriptionfactorVvMYBPA1regulatesproanthocyanidinsynthesisduring fruitdevelopment,PlantPhysiol.143(2007)1347–1361.

[61]A.Gonzalez,M.Zhao,J.M.Leavitt,A.M.Lloyd,Regulationoftheanthocyanin biosyntheticpathwaybytheTTG1/bHLH/Mybtranscriptionalcomplexin Arabidopsisseedlings,PlantJ.53(2008)814–827.

[62]J.S.Park,J.B.Kim,K.J.Cho,C.I.Cheon,M.K.Sung,M.G.Choung,K.H.Roh, ArabidopsisR2R3-MYBtranscriptionfactorAtMYB60functionsasa transcriptionalrepressorofanthocyaninbiosynthesisinlettuce(Lactuca sativa),PlantCellRep.27(2008)985–994.

[63]E.Braidot,M.Zancani,E.Petrussa,C.Peresson,A.Bertolini,S.Patui,F.Macrì, A.Vianello,Transportandaccumulationofflavonoidsingrapevine(Vitis viniferaL.),PlantSignal.Behav.3(2008)626–632.

[64]X.Y.Zhang,X.L.Wang,X.F.Wang,G.H.Xia,Q.H.Pan,R.C.Fan,F.Q.Wu,X.C.Yu, D.P.Zhang,Ashiftofphloemunloadingfromsymplasmictoapoplasmic pathwayisinvolvedindevelopmentalonsetofripeningingrapeberry,Plant Physiol.142(2006)220–232.

[65]A.Sanders,R.Collier,A.Trethewy,G.Gould,R.Sieker,M.Tegeder,AAP1 regulatesimportofaminoacidsintodevelopingArabidopsisembryos,PlantJ. 59(2009)540–552.

[66]M.C.Shelden,S.M.Howitt,B.N.Kaiser,S.D.Tyerman,Identificationand functionalcharacterisationofaquaporinsinthegrapevine,Vitisvinifera, Funct.PlantBiol.36(2009)1065–1078.

[67]L.Q.Chen,B.H.Hou,S.Lalonde,H.Takanaga,M.L.Hartung,X.Q.Qu,W.J.Guo, J.J.Kim,W.Underwood,B.Chaudhuri,D.Chermak,G.Antony,F.F.White,S.C. Somerville,M.B.Mudgett,W.B.Frommer,Sugartransportersforintercellular exchangeandnutritionofpathogens,Nature468(2010)527–532.

[68]J.Chong,M.C.Piron,S.Meyer,D.Merdinoglu,C.Bertsch,P.Mestre,TheSWEET familyofsugartransportersingrapevine:VvSWEET4isinvolvedinthe interactionwithBotrytiscinerea,J.Exp.Bot.65(2014)6589–6601.

[69]K.Shiratake,Geneticsofsucrosetransporterinplants,GenesGenom. Genomics1(2007)74–80.

[70]T.L.Slewinski,Diversefunctionalrolesofmonosaccharidetransportersand theirhomologsinvascularplants:aphysiologicalperspective,Mol.Plant4 (2011)641–662.

[71]D.Afoufa-Bastien,A.Medici,J.Jeauffre,P.Coutos-Thévenot,R.Lemoine,R. Atanassova,M.Laloi,TheVitisviniferasugartransportergenefamily: phylogeneticoverviewandmacroarrayexpressionprofiling,BMCPlantBiol. 10(2010)245–266.

[72]S.Domingos,P.Scafidi,V.Cardoso,A.E.Leitao,R.DiLorenzo,C.M.Oliveira,L.F. Goulao,FlowerabscissioninVitisviniferaL.triggeredbygibberellicacidand shadedisclosesdifferencesintheunderlyingmetabolicpathways,Front. PlantSci.6(2015)457–474.

[73]J.C.Goulão,P.Fernandes,Tacklingthecellwallofthegrapeberry,in:M.M. Gerós(Ed.),TheBiochemistryoftheGrapeBerry,BenthamSciencePublishers, Sharjah,2012,pp.172–193.

[74]C.Ruperti,F.Bonghi,S.Ziliotto,A.Pagni,S.Rasori,P.Varotto,J.J.Tonutti, Characterizationofamajorlatexprotein(MLP)genedown-regulatedby ethyleneduringpeachfruitletabscission,PlantSci.163(2002)265–272.

[75]X.Q.Wang,L.M.Li,P.P.Yang,C.L.Gong,Theroleofhexokinasesfromgrape berries(VitisviniferaL.)inregulatingtheexpressionofcellwallinvertaseand sucrosesynthasegenes,PlantCellRep.33(2014)337–347.

[76]S.Bihmidine,C.T.Hunter,C.E.Johns,K.E.Koch,D.M.Braun,Regulationof assimilateimportintosinkorgans:updateonmoleculardriversofsink strength,Front.PlantSci.4(2013)177–191.

[77]Z.Li,W.P.Palmer,A.P.Martin,R.Wang,F.Rainsford,Y.Jin,J.W.Patrick,Y. Yang,Y.L.Ruan,Highinvertaseactivityintomatoreproductiveorgans correlateswithenhancedsucroseimportinto,andheattoleranceof,young fruit,J.Exp.Bot.63(2012)1155–1166.

[78]E.Vierling,Therolesofheat-shockproteinsinplants,Annu.RevPlantPhysiol. PlantMol.Biol.42(1991)579–620.

[79]M.H.Al-Whaibi,Plantheat-shockproteins:aminireview,J.KingSaudUniv. Sci.23(2011)139–150.

[80]J.Li,R.Yuan,NAAandethyleneregulateexpressionofgenesrelatedto ethylenebiosynthesis,perception,andcellwalldegradationduringfruit abscissionandripeningin‘Delicious’apples,J.PlantGrowthRegul.27(2008) 283–295.

[81]L.Giacomelli,O.Rota-Stabelli,D.Masuero,A.K.Acheampong,M.Moretto,L. Caputi,U.Vrhovsek,C.Moser,GibberellinmetabolisminVitisviniferaL. duringbloomandfruit-set:functionalcharacterizationandevolutionof grapevinegibberellinoxidases,J.Exp.Bot.64(2013)4403–4419.

[82]A.Aziz,Spermidineandrelatedmetabolicinhibitorsmodulatesugarand aminoacidlevelsinVitisviniferaL.:possiblerelationshipswithinitialfruitlet abscission,J.Exp.Bot.54(2003)355–363.

[83]P.Dauelsberg,J.T.Matus,M.J.Poupin,A.Leiva-Ampuero,F.Godoy,A.Vega,P. Arce-Johnson,Effectofpollinationandfertilizationontheexpressionofgenes relatedtofloraltransition,hormonesynthesis,andberrydevelopmentin grapevine,J.PlantPhysiol.168(2011)1667–1674.

[84]C.DeAzevedoSouza,S.S.Kim,S.Koch,L.Kienow,K.Schneider,S.M.McKim, G.W.Haughn,E.Kombrink,C.J.Douglas,Anovelfattyacyl-CoAsynthetaseis requiredforpollendevelopmentandsporopolleninbiosynthesisin Arabidopsis,PlantCell21(2009)507–525.

[85]A.Krichevsky,S.V.Kozlovsky,G.W.Tian,M.H.Chen,A.Zaltsman,V.Citovsky, Howpollentubesgrow,Dev.Biol.303(2007)405–420.

[86]S.G.Palusa,M.Golovkin,S.B.Shin,D.N.Richardson,A.S.N.Reddy,

Organ-specific,developmental,hormonal,andstressregulationofexpression ofputativepectatelyasegenesinArabidopsis,NewPhytol.174(2007) 537–550.

[87]L.Zonia,T.Munnik,Uncoveringhiddentreasuresinpollentubegrowth mechanics,TrendsPlantSci.14(2009)318–327.

[88]B.L.Lytle,J.Song,N.B.delaCruz,F.C.Peterson,K.A.Johnson,C.A.Bingman, G.N.Phillips,B.F.Volkman,StructuresoftwoArabidopsisthalianamajorlatex proteinsrepresentnovelhelix-gripfolds,Proteins76(2009)237–243.