北京百迈客苎麻根侵染线虫转录组测序分析
S.Zhuetal./Gene552(2014)67–7469
wasreportedtodisplayawellstabilityofexpressionintheCOandtheRLN-infectedramie(Liuetal.,2014),wasselectedastheendogenouscontrol.TheprimersequenceofDEGsandactingenewerelistedinTableS1.Thethermalcycleusedwasasfollows:95°Cfor30s,followedby40cyclesof95°Cfor10s,and55°Cfor25s.qRT-PCRwasper-formedintriplicateforeachsample.Relativeexpressionlevelswerede-terminedasdescribedpreviously(LivakandSchmittgen,2001).
3.Results
3.1.Illuminapaired-endsequencinganddenovoassembly
RNAwasextractedindividuallyfromfourwholeplantbodies,ex-cludingroots(twocontrolplants,CO1andCO2,andtwoRLNinfectionplants,Inf1andInf2).RNAfromindividualplantswasusedtoconstructlibrariesofcDNAwithafragmentlengthof200bp.FragmentswerethensequencedusingIlluminapaired-endsequencingtechnology.Atotalof56.3,51.7,43.4,and45.0millionrawsequencingreadswithlengthsof90bpweregeneratedfromthelibrariesofCO1,CO2,Inf1,andInf2,respectively(Table1).Trinitysoftwarewasusedtoassemblethereadsequenceofeachsampledenovo.Sequencesnotextendedoneitherendwerede nedasunigene.Finally,36,934,34,505,32,918,and32,079unigeneswerefoundinCO1,CO2,Inf1,andInf2,respective-ly(Table1).
Allunigenesfromthefoursampleswerealignedbasedonnucleo-tidesequence.Theunigenesfromdifferentsamplesthatoverlappedmorethan35bpwereassembledfurther.Thedenovoassemblyyielded50,486non-redundantunigenes,withanaveragelengthof853.3bpandatotallengthof43.08Mb(Table1).Thelengthofassembledunigenesrangedfrom200to9911bp.Therewere15,640unigenes(31.0%)b300bp,13,578unigenes(26.9%)from301to500bp,7819unigenes(15.5%)from501to1000bp,7725unigenes(15.3%)from1001to2000bp,and5724unigenes(11.3%)N2000bp(Fig.1).
3.2.Functionalannotationandclassi cation
Forvalidationandannotationoftheassembledunigenes,sequencesimilaritysearcheswereconductedintheNr,COG,GO,KEGG,andtheSwiss-ProtproteindatabasewithanE-valuethresholdof10 5.There-sultsindicatedthat,outof50,486unigenes,24,463(48.9%),20,198(40.0%),21,302(42.2%),5137(10.2%),and8541(16.9%)unigenesshowedsigni cantsimilaritytoknownproteinsintheNr,SwissProt,GO,KEGG,andCOGdatabases,respectively(Fig.2).Together,24,820(49.2%)unigenesshowedsimilaritytoknownproteinsinthese veda-tabases(Fig.2).
TheGOdatabaseisaninternationalstandardizedgenefunctionalclassi cationsystem,andithasthreeontologies:molecularfunction,cellularcomponent,andbiologicalprocess.GOfunctionalclassi cationfortheseunigenesassembledshowedthat21,302unigenes(40.0%)withBLASTmatchestoknownproteinswereassignedtoGOclasseswith240,800functionalterms.Ofthese,assignmentstothebiologicalprocessontologymadeupthemajority(123,002;51.1%),followedbycellularcomponent(85,686;35.6%)andmolecularfunction(32,112;13.3%)ontologies(Fig.3).
Table1
SummaryoftranscriptomesassembledbyIlluminasequencing.SampleReadnumberUnigenenumberTotallengthMeanlength(million)assembled(Mb)(bp)CO156.336,93434.90944.8CO251.734,50533.76978.4Inf143.432,91832.34982.6Inf245.032,07931.58984.3Total
196.4
50,486
43.08
853.3
parisonofgeneexpressionlevelsbetweenCOramieandInframieTodeterminetheexpressionlevelsofthe50,486unigenesassem-bleddenovo,allsequencingreadswererealignedtotheunigenes.Thenumberofreadsalignedtoaunigeneappearinginthelibrarywasusedtoestimateitsexpressionlevelusingthenormalizedvalueofthegene'suniquelyalignedreadcountsperkilobaseofexonmodelpermil-lionreads(RPKM).DifferencesinreadfrequenciesbetweentheCOandInflibrarieswereusedtodeterminechangesingeneexpressioninre-sponsetoroot-lesionnematodeinfection.TheRPKMvaluesofallunigeneswerecomparedbetweenthetwoCOreplicatesandbetweenthetwoInfreplicates.Thereweresigni cantcorrelationsbetweenthetwoCOreplicatesandbetweenthetwoInfreplicates,withthecorrela-tioncoef cientsof0.95and0.99,respectively.Scatterdiagrams,inwhichthelogarithmicRPKMvaluesofeachgeneinthetworeplicatesofeachtreatmentwereassignedascoordinatevaluesoftwoaxes,showedthatalldatapointsweredistributedintheregionofthediago-nal(Fig.4).AlltheseresultssuggestedthattheabundancesoftheseunigenetranscriptsinthetwoCOreplicatelibrariesandinthetwoInfreplicatelibrariesweresimilar.
ComparisonsoftheaverageRPKMvaluesofthetworeplicatesforeachgenebetweenCOandInframiewereperformedtoscreenthedif-ferentiallyexpressedgenes(DEGs)involvedintheRLN-defensere-sponse.Atotalof777DEGs(592genesup-regulatedand185genesdown-regulated)withatleastatwo-folddifferencebetweenCOramieandInframiewereidenti ed(FDRb0.01)(Fig.5,TableS1).AmongtheseDEGs,16(12genesup-regulatedand4genesdown-regulated)genesshowedmorethanatwenty-folddifferencebetweenCOramieandInframie(Fig.5,TableS2).Therewere40transcriptionfactors,whichweredistributedinto21families,regulatedbyRLNinfection(Table2,TableS3).
3.4.ValidationoftheexpressiondifferenceofDEGs
Inordertovalidatetheexpressionpro lingbyIlluminasequencing,theexpressionlevelsoftwelvegenes,includingtwocatalasegenes,twoNACtranscriptionfactors,threeperoxidasegenesand vedisease-resistancerelatedgenes,werefurtheranalyzedbyqRT-PCR.TheqRT-PCRresultshowedthatsevengeneswereup-regulatedexpressionand vegeneswasdown-regulatedexpression(Table3)(Pb0.05).ThetrendofexpressionchangesoftheseselectedgenesbasedonqRT-PCRwassimilarwiththosedetectedbyIllumina-sequencingmethod.However,thechangefoldsofthesegeneexpressionlevelsrespondingtoRLN-infectiondetectedbyqRT-PCRhadsomedifferencewiththosedetectedbyIlluminasequencing(Table3).
3.5.Functionalclassi cationofDEGsandthepathwaysenrichedbyDEGsTheCOGdatabasewasusedtoclassifytheorthologousgeneprod-ucts.EveryproteinintheCOGdatabaseisassumedtohaveevolvedfromanancestorprotein.Thewholedatabaseisbuiltoncodingproteinswithcompletegenomes,aswellasonsystemevolutionrelationshipsofbacteria,algae,andeukaryotes.AllDEGswerealignedusingtheCOGda-tabasetoclassifypotentialfunctions.Intotal,214DEGswereassignedtothe23COGclassi cationswith335functionalterms(Fig.6).TheGOfunctionalclassi cationofDEGswasperformedand694DEGswereclassi edinto51GOcategorieswith7641functionalterms(Fig.3).InmostGOcategories,theratioofDEGsagainstbackgroundunigenesrangedfrom2%to4%.Threefunctionalcategories,metallochaperoneac-tivity,antioxidantactivityandcellproliferation,contain6.67%,7.87%,and8.08%oftheDEGsintheirtotalunigenes,respectively,whichsug-geststhatthesethreefunctionalcategorieswereenrichedbyDEGs.
Thein uenceofRLNinfectiononbiologicalpathwayswasevaluatedbyenrichmentanalysisofDEGs.Atotalof61pathwaysinramiewerepossiblyaffectedbyRLNinfection(TableS4).Threepathways,phenylal-aninemetabolism,carotenoidbiosynthesis,andphenylpropanoid