ElizabethBuescher1,TilmanAchberger2,IdrisAmusan1,AnthonyGiannini3,CherieOchsenfeld2,AnaRus4,BrettLahner4,OwenHoekenga5,ElenaYakubova4,JeffreyF.Harper6,MaryLouGuerinot7,MinZhang2,DavidE.Salt4,IvanR.Baxter4,8*
1DepartmentofAgronomy,PurdueUniversity,WestLafayette,Indiana,UnitedStatesofAmerica,2DepartmentofStatistics,PurdueUniversity,WestLafayette,Indiana,UnitedStatesofAmerica,3DepartmentofAnimalSciences,PurdueUniversity,WestLafayette,Indiana,UnitedStatesofAmerica,4CenterforPlantEnvironmentalStressPhysiology,PurdueUniversity,WestLafayette,Indiana,UnitedStatesofAmerica,5RobertW.HolleyCenterforAgricultureandHealth,UnitedStatesDepartmentofAgriculture-AgriculturalResearchService,CornellUniversity,Ithaca,NewYork,UnitedStatesofAmerica,6UniversityofNevada,Reno,Reno,Nevada,UnitedStatesofAmerica,7DepartmentofBiologicalSciences,DartmouthCollege,Hanover,NewHampshire,UnitedStatesofAmerica,8PlantGeneticsResearchUnit,UnitedStatesDepartmentofAgriculture-AgriculturalResearchService,DonaldDanforthPlantSciencesCenter,St.Louis,Missouri,UnitedStatesofAmerica
Abstract
Controllingelementalcompositioniscriticalforplantgrowthanddevelopmentaswellasthenutritionofhumanswhoutilizeplantsforfood.Uncoveringthegeneticarchitectureunderlyingmineralionhomeostasisinplantsisacriticalfirststeptowardsunderstandingthebiochemicalnetworksthatregulateaplant’selementalcomposition(ionome).NaturalaccessionsofArabidopsisthalianaprovidearichsourceofgeneticdiversitythatleadstophenotypicdifferences.Weanalyzedtheconcentrationsof17differentelementsin12A.thalianaaccessionsandthreerecombinantinbredline(RIL)populationsgrowninseveraldifferentenvironmentsusinghigh-throughputinductivelycoupledplasma-massspectroscopy(ICP-MS).SignificantdifferencesweredetectedbetweentheaccessionsformostelementsandweidentifiedoverahundredQTLsforelementalaccumulationintheRILpopulations.AlteringtheenvironmenttheplantsweregrowninhadastrongeffectonthecorrelationsbetweendifferentelementsandtheQTLscontrollingelementalaccumulation.Allionomicdatapresentedispubliclyavailableatwww.ionomicshub.org.
Citation:BuescherE,AchbergerT,AmusanI,GianniniA,OchsenfeldC,etal.(2010)NaturalGeneticVariationinSelectedPopulationsofArabidopsisthalianaIsAssociatedwithIonomicDifferences.PLoSONE5(6):e11081.doi:10.1371/journal.pone.0011081˚,Sweden¨rK.Ingvarsson,UniversityofUmeaEditor:Pa
ReceivedJanuary27,2010;AcceptedMay7,2010;PublishedJune14,2010
Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsPublicDomaindeclarationwhichstipulatesthat,onceplacedinthepublic
domain,thisworkmaybefreelyreproduced,distributed,transmitted,modified,builtupon,orotherwiseusedbyanyoneforanylawfulpurpose.
Funding:ThisprojectwasfundedbyagrantfromtheNSFPlantGenomeResearchProgram(DBI-0077378)awardedtoMaryLouGuerinot,DavidEide,JeffHarper,DavidE.Salt,JulianSchroederandJohnWard,NSFArabidopsis2010program(IOS-0419695)awardedtoMaryLouGuerinot,JeffHarper,DavidE.Salt,JulianSchroederandJohnWard,NationalInstitutesofHealth,theNationalInstituteofGeneralMedicine(R01GM78536-01A1)awardedtoDavidE.Salt,MaryLouGuerinotandIvanBaxterandtheIndiana21stCenturyResearchandTechnologyFund(912010479)toDavidE.Salt.OwenHoekengawassupportedbytheNSFPlantGenomeResearchProgram(DBI-0419435)awardedtoLeonKochian,EdwardBuckler,OwenHoekengaandJocelynRose.Thefundershadnoroleinstudydesign,datacollectionandanalysis,decisiontopublish,orpreparationofthemanuscript.CompetingInterests:Theauthorshavedeclaredthatnocompetinginterestsexist.*E-mail:ibaxter@danforthcenter.org
Introduction
Geneticvariationoccurringamongandwithinnaturalpopu-lationsofArabidopsisthalianacanbeusedasatoolforgenediscovery[1–3].A.thalianahasawide-geographicdistribution,producingalargeanddiversegroupofnaturalpopulations,manyofwhichhavebeencollectedasaccessionsthatarecuratedbytheArabidopsisBiologicalResourceCenter(ABRC).Considerablevariationforsuchtraitsasresistancetobioticandabioticstress,development,andmetabolismhasbeendescribed(forrecentreviewssee[3,4]).Observedvariationbetweenaccessionscanbequalitative,definedbyphenotypicdistributionsthatfallintodiscreteclasses,andiscausedbyoneortwomajorloci.Variationcanalsobequantitative,definedbyacontinuousphenotypicdistribution,causedbythecombinedeffectofmultipleloci.Experimentalpopulationsizeplaysamajorroleinthethresholdfordetectionofloci.Smallpopulationsareusefulforidentifyinglociifatraitiscontrolledbyafewlociwithlargephenotypiceffect;
PLoSONE|www.plosone.org
1
however,morecomplextraitscontrolledbymultiplelociwithrelativelysmallphenotypiceffectwillrequirelargeexperimentalpopulations.
Byusingimmortalizedmappingpopulationsknownasrecom-binantinbredlines(RIL),derivedfromavarietyofaccessions,quantitativetraitloci(QTL)havebeenidentifiedfornumerousimportanttraitsrelatedtotheionome[5].Theseincludephosphateaccumulationinseedandshoot[6],nitrogen(N)useefficiency[7,8],aluminum(Al)resistance[9–11],shootcesium(Cs)accumulation[12]andshootselenateaccumulation[13,14].OnceQTLsfortraitsofinteresthavebeenidentified,thegenomictoolsavailableforA.thalianacanbeusedtolocatethegenesthatunderlietheseQTLsandthusdescribethetraitsatamolecularlevel(forareviewsee[15]).SuchanapproachwasrecentlytakeninourlaboratoriestoidentifyAtHKT1,whichencodesaNa-transporter,asthegeneresponsibleforaQTLthatcontrolselevatedNaintwodistinctnaturalaccessionsTsu-1andTs-1[16],AtMOT1,aputativeMotransporterasthegeneresponsibleforan
June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
,80%decreaseinMoaccumulationin7accessions[17]andFPN2,anFeandCotransporterasthegeneresponsibleforincreasedCoin6accessions[18].
AlthoughmanystudieshaveexaminedthegeneticbasisthatcontrolstheaccumulationofsingleelementsinaRILpopulation,onlyasmallnumber[19–21]haveexaminedmultipleelementsinmultiplepopulationstoinvestigatethegeneticarchitectureoftheionome.WehavepreviouslyshownthatphysiologicalresponsestolowFeandPgrowthconditionscauserobust5and6elementsignatures,respectively,inA.thaliana[22],demonstratingthatsingleelementexaminationsofapopulationcannotgiveanaccuratepictureoftheplantionome.Geneticvariationintheresponsetotheseandotherenvironmentalchangesarelikelytoaltertheseelement-to-elementrelationships,leadingtogene6environment6ionomeinteractions.Inthisstudy,weanalyzedasmalldiversitypanelofnaturalaccessionsandthreeRILpopulations.Weshowthatsignificantvariationexistsintheshootandtheseedionomesof12A.thalianaaccessionsacrossdifferentenvironments.TheRILpopulations[Bay-06Shadahara(BaySha),Col-46Ler-0(ColLer)andCvi-06Ler-2(CviLer)]weregrownunderavarietyofenvironmentalconditions(highvs.lowFe)andpopulationsizes.Between17and19elementsweremeasuredforeachpopulation,withbothmacro-andmicroele-mentsrepresented.WedemonstratethatthisvariationiscontrolledbybothMendelianandquantitativetraitloci,andthatalteringtheenvironmenthasalargeeffectonwhichlocicontributetotheobserveddifferences.
Table1.ShootandseedionomeofA.thalianaCol-0.
2Element1Shoot
Seed3AverageLiBNaMgPKCaMnFeCoNiCuZnAsSeMoCd
12SD2.939.75217.8429009004700290012.947.620.160.260.7214.241.959.030.810.25
Lineeffect4yesyesyesnoyesyesyesyesyesyesyesyesyesnonoyesno
Average0.876.8464.823283993710690564532.4942.520.270.361.6758.651.3411.711.010.36
SD0.512.114.63614157823195515.4918.570.150.141.1414.420.485.740.410.11
Lineeffectnoyesyesyesnonononoyesnoyesnoyesnonoyesyes
11.8343.88860.14129009700461004500063.59100.631.861.411.8261.061.049.35.522.03
Results
VariationintheshootionomeofArabidopsisthalianaaccessions
TheconcentrationofvariouselementsinhealthyA.thalianashoottissuefromCol-0variesover4ordersofmagnitudedependingontheelementanditsbiologicalfunction(Table1).MacronutrientssuchasMg,P,KandCaaccumulatemorethan9,500mgg21oftheshootdryweight,whereasmicronutrientssuchasB,Mn,Fe,Co,Ni,Cu,ZnandMorangeinconcentrationfrom1–100mgg21.Non-essential,potentiallytoxictraceelementssuchasAs,Se,andCdcanaccumulatetobetween1–10mgg21withoutanyvisiblesymptomsoftoxicity.AnANOVAanalysisoftheaccumulationinthe12accessionsrevealsthatvariationin13ofthe17elementsmeasuredareundergeneticcontrolwithinthepopulation(Table1,Table2,TableS1A).Severalelementsshowedlargevariationbetweenthelowestandhighestaccumu-latingaccessions(Table2).Moshowedthemostvariationwith44significantpairwisedifferencesbetweenaccessions(TableS1).SinceCol-0isthereferenceaccessionandaparentofmanyoftheavailableRILpopulations,wehavealsoincludedatableofelementaldifferencescomparedwithCol-0(Table2).
Allelementspresentedasmgg21.
Datarepresentstheaverage(n=60exceptforLin=30),individualplantsharvestedandanalyzedin3–6separateexperiments.3Datarepresentstheaverage(n=12)ofindividualsamplesfromseedpooledfrom4plantssubsampled3timeseachandanalyzedin2separateexperiments.4ColumnindicatesifthelineeffectissignificantintheANOVA.doi:10.1371/journal.pone.0011081.t001
accessionsintheseedmirrorthedifferencesobservedbetweentheaccessionsintheleaves,however,therearemultiplecomparisonsinwhichsignificantdifferencesintheleavesarenotreflectedintheseed(andviceversa).Forexample,theelevatedshootNaobservedinTs-1isalsoreflectedintheseed,withTs-1showinga161%increaseinseedNacomparedtoCol-0(Table2and3),whiletheotherhighshootNaaccession,Tsu-1doesnotaccumulatesignificantlydifferentamountsofNainitsseedscomparedtoCol-0(Table3).
IdentificationofQTLscontrollingionomicdifferencesinthreeRecombinantInbredPopulations
Toexpandonthegeneticcharacterizationoftheionome,weanalyzedtheelementalcompositionof3RILpopulations:Bay-06Shadahara(BaySha),Col-46Ler-0(ColLer)andCvi-06Ler-2(CviLer).BayShaandCviLerweregrownintwodifferentenvironments(twogrowthmediawithdifferingingredients)foratotalof5differentexperiments.TheexperimentsdifferedinthenumberofRILlinesanalyzed(from93forColLerto411forthesecondBayShaexperiment)andthenumberofplantsanalyzedperline(1–3).TheparentsoftheRILpopulationshowedsignificantdifferencesin53outofthepossible87instances(17elements65populations+SandRbmeasuredinthesecondBayShaexperiment).Weidentified218QTLsinthefiveexperimentsalthoughasignificantnumberarelikelythesameQTLfoundintwoexperimentsofthesamepopulationorareduetothesharedLerparentofCviLerandColLerpopulations(Tables4and5,TableS2).158ofthe218QTLs(72%)werefoundforelementsforwhichtheparentshadasignificant
2
June2010|Volume5|Issue6|e11081
VariationintheseedionomeofArabidopsisthalianaaccessions
TheconcentrationofelementsmeasuredinA.thalianaseedvariesover4ordersofmagnitude,asimilarscaletothatobservedinshoots(Table1).However,thereareseveralsignificantdifferencesbetweentheshootandseedionome,withcertainelementsbeingenrichedorreducedrelativetootherelements.Forexample,onamgg21dryweightbasis,Pdoesnotchangeconcentrationsfromseedtoshoot,butKisapproximately4-foldlowerintheseeds(Table1).Furthermore,ANOVAanalysisoftheelementalcompositionofseedsfromdifferentA.thalianaaccessionsshowedsignificantgeneticcontrolfor8ofthe17elements(Table1,Table3,TableS1).Severalsignificantdifferencesbetween
PLoSONE|www.plosone.org
Nat.IonomicVar.inArab.
Table2.ShootionomevariationacrossA.thalianaecotypescomparedtoCol-0.
NameAcc.#
BNaMgPKCaMnFeCoNiCuZnAsSeMoCd
%differencefromcol-0Cvi-0Est-1Kas-1Mrk-0Mt-0Se-0Ts-1Van-0Ws-0Nd-1Tsu-1Ler-2
CapeVerdiIslandsEastland,RussiaKashmir,IndiaMarktBaden,GermanyMartubaCyrenaika,LibyaSanEleno,SpainTossadelMar,SpainVancouver,CanadaWassilewskija,RussiaNiederzenz,GermanyTsu,Japan
109611501264137413801502155215841602163616408581
127
219
23189
2320
21
272326
2116
273
117
25
1653
45
12
234
232
27
3656
282321419
7342
276281226219
222
2317
2737
22415
24
233
259226227228
224
46
225
AllelementvaluesareinpercentdifferencefromCol-0withdatarepresentingthesignificant(studentt-testP,0.01)averagedifferenceacross2independentexperiments(n=10individualplantsperexperiment).doi:10.1371/journal.pone.0011081.t002
difference,foranaverageof3.0QTLs/elementwhileelementsinwhichtheparentswerenotsignificantlydifferentaveraged1.8QTLs(Tables4and5).Allofthe19major(r2.20%)QTLsweidentifiedcamefromelementsinwhichtheparentsweresignificantlydifferent(Tables4and5,TableS2).FrequencydistributionsshowingdifferencebetweenparentallinesareprovidedinsupplementalFilesS1,S2,S3,S4,S5.ToevaluatetheeffectofdifferentexperimentaldesignsontheabilitytoidentifyQTLsforionomictraits,wecreatedsubsetsofthelargeBayShaexperiment(411linesatn=2,,800samples)andthelargeCviLerexperiment(165linesatn=3).Randomlygeneratedn=1andn=2subsetsoftheCviLerdataidentified69%and95%asmanyQTLsrespectivelyasthen=3datafromwhichtheywerederived.Whendatafromthe165linesusedinthesmallBayShaexperimentwasextractedfromthelargeBaySha411line
experiment,20%fewerQTLswereidentified,suggestingthatthedifferentnumbersofQTLsidentifiedbetweenthelargeandsmallBayShaexperimentsisduetothechangeinthenumberoflines.
TransgressiveSegregationandEpistasis
Wetestedfortransgressivesegregationusingtwoindependentmethods:thenumberofRILswhichweresignificantlyhigherorlowerthantheparentsgrowninthesametraysorhavingtwoQTLswithoppositealleliceffects.Forelementsinwhichtherewasnotasignificantdifferencebetweentheparents,thepercentageofRILsthatfelloutsidetherangeoftheparentsrangedfrom10%–41%(Tables4and5,FilesS1,S2,S3,S4,S5).ThelikelihoodoffindingtransgressivesegregationbytheoppositealleliceffectstestincreasedasthenumberofQTLsincreased:transgressive
Table3.SeedionomevariationacrossA.thalianaecotypescomparedtoCol-0.
Name
Acc.#Li
BNaMgPKCaMnFeCoNiCuZnAsSeMoCd
%differencefromcol-0Cvi-0Est-1Kas-1Mrk-0Mt-0Ee-0Ts-1Van-0Ws-0Nd-1Tsu-1Ler-2
CapeVerdiIslandsEastland,RussiaKashmir,IndiaMarktBaden,GermanyMartubaCyrenaika,LibyaSanEleno,SpainTossadelMar,SpainVancouver,CanadaWassilewskija,RussiaNiederzenz,GermanyTsu,Japan
109611501264137413801502155215841602163616408581
204
76
212
234
212
69
258
161
2482582825392275
309
72
258
254
55
283243
223
27
69
AllelementvaluesareinpercentdifferencefromCol-0withdatarepresentingthesignificant(studentt-testP,0.01)averagedifferenceacross2independentexperiments(n=10individualplantsperexperiment).doi:10.1371/journal.pone.0011081.t003
PLoSONE|www.plosone.org3June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
Table4.AllQTLsforeachelementintheColLerandBayShaRILpopulations.
ColLer(93lines,n=3)SunshineSoil,LowFe
Herit-HiloRILTrans-p-valueabilityspercadiffbQTLsLiB
1.8E-035.7E-09
44%57%51%52%63%NA61%51%53%50%47%54%38%49%98%51%NA66%53%
16%4%24%16%25%NA21%16%4%3%19%15%23%14%24%29%NA4%20%
NNNNNNAYNNNNNNNNANNANN
01001NA20010100NA0NA23
BayShaSmall(165lines,n=3)SunshineSoil,HighFe
Herit-HiloRILTrans-Majorcp-valueabilityspercadiffbQTLs00000NA00000000NA0NA00
NANA5.5E-101.5E-122.9E-14NA1.8E-151.2E-15NANA7.6E-111.9E-04NA1.5E-03NA1.7E-03NA5.0E-363.3E-06
46%40%79%62%66%NA72%59%43%42%47%39%42%47%55%55%NA80%46%
10%18%47%9%10%NA17%2%15%7%7%8%10%10%21%47%NA2%9%
NYYYYNAYYNNNNYYNYNANN
23227NA8321105603NA21
Majorc00110NA1100000002NA10
BayShaFull(411lines,n=2)PromixSoil,HighFe
Hertit-HiloRILTrans-p-valueabilityspercadiffbQTLsNA4.2E-044.9E-113.2E-153.1E-341.8E-151.7E-422.2E-291.3E-102.3E-07NA3.0E-074.4E-065.4E-09NANA1.8E-357.9E-492.7E-08
55%63%79%70%73%86%75%72%60%66%61%60%63%57%54%57%71%80%53%
17%24%42%21%13%38%15%22%23%26%34%13%19%13%18%11%8%3%10%
NYYYYYYYYYYNYYNYYNN
1376987453423515721
Majorc0011021101000000110
NaNAMg6.1E-08PSKCa
2.0E-14NANANA
Mn3.8E-18Fe
9.3E-11
CoNANi
NA
CuNAZnAsSe
2.7E-03NANA
RbNAMo1.5E-16CdNA
PopulationsizeandreplicatenumberareincludedwitheachRILaswellasenvironment.aTransgressivesegregationasmeasuredbythepercentageofRILssignificantlyoutsideoftherangeoftheparents.bTransgressivesegregationdeterminedbythepresenceofQTLswithdifferentdirectionsoftheadditiveeffect.cMajorQTLwithR2value.0.20.
doi:10.1371/journal.pone.0011081.t004
segregationoccurredin14of19elementsinthelarge(411line)BayShaexperiment(83QTLs),noneoftheelementsintheColLerexperiment(11QTLs)andonlytwooftheelementsinthefirstCviLerexperiment(28QTLs)(Tables4and5).EpistaticinteractionsbetweentheidentifiedQTLswereexaminedusingRQTLinthelargeBayShapopulationandtheCviLerpopulationexperiments.Onlyfivesignificant(p,0.01)interactionswerefoundamongthe53(17+17+19)elementsexamined,noneofwhichwerefoundinthe411line(large)BayShapopulationwiththemostpowertodetectepistaticinteractions(TableS3).
EnvironmentalEffectonElementCorrelationsandQTLDiscovery
Alterationsintheenvironmentorphysiologyofaplantcanaffecttheaccumulationofmultipleelementssimultaneously.VariationinmineraluptakeindifferentenvironmentshasbeendescribedinA.thaliana[19,21,23]andSilenevulgaris[24].Theclearestexampleoftheeffectoftheenvironmentwasobservedintherelationshipsbetweenelementswithinagivenexperiment.WemeasuredthecorrelationbetweeneachpairwisecombinationofelementsfromeveryRILineachofthefivepopulations(Figure1).Whilemanyelementsweresignificantlycorrelatedwithineachofthefiveexperiments,onlythreepairsofelementswerecorrelatedineverypopulation6environmentweanalyzed:Li-Na,Mg-Ca,andCu-Zn,althoughLi-As,Li-Cd,Li-K,Li-Zn,P-Fe,Mg-ZnandZn-Cdwerefoundin4ofthe5experiments(Figure1).
PLoSONE|www.plosone.org
4
TheBayShaandCviLerpopulationswereeachgrownintwodifferentgrowthmediaenvironments.Mostofthesignificantelementalcorrelations(68of85totalcorrelationsinBayShaand66of76totalcorrelationsinCviLer)werefoundinonlyoneoftheenvironments(Figure1).TheCa-MocorrelationwasuniquetotheBayShapopulationandwasfoundinbothenvironments.ThereisathreeelementcorrelatednetworkthatonlyappearsinSunshinegrowthmediumwithsufficientFe,Co-CdispositivelycorrelatedwhilebotharenegativelycorrelatedtoCu(Cu-CdisnotsignificantinCviLer)(Figure1).Interestingly,intheBayShaPromixexperiment,Co-Cuisalsonegativelycorrelated,butCu-Cdispositivelycorrelated.EightothercorrelationsweresharedbetweenthethreepopulationsgrowninsufficientFeSunshinegrowthmedium.IntheFesufficientSunshinegrowthmedium,BayShashared18and22correlationswithCviLerandColLer,respectively(Figure1).WhenthePromixgrownBayShapopulationwascomparedwithsufficientFeSunshineCviLerandColLerpopulations,only8and11sharedcorrelationswereidentified(Figure1).
ComparisonofQTLsacrossenvironments
Totestwhetheralteringtheenvironmentalteredwhichgeneticlocicontroltheionome,wecomparedtheQTLsidentifiedforeachelementinthefiveexperiments.TheonlycommonQTLamongthefivepopulations,regardlessofenvironment,popula-tionsizeorgeneticbackground,istheMoQTLonchromosometwocorrespondingtotheMOT1locus[25].Intheanalysisofthe
June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
Table5.AllQTLsforeachelementintheCviLerRILpopulations.
CviLerLow(151lines,n=1)SunshineSoil,LowFep-value
LiBNaMgPSKCaMnFeCoNiCuZnAsSeRbMoCd
5.7E-05NA1.4E-083.3E-041.5E-05NANANA2.4E-07NANANANANANANANA8.2E-087.3E-05
Herit-abilityNANANANANANANANANANANANANANANANANANANA
HiloRILsperca21%27%13%24%39%NA25%34%11%28%19%23%21%15%26%38%NA9%13%
Trans-diffbNNYNNNAYNNNNNNNNNNANN
QTLs13113NA2132101302NA31
Majorc00100NA0000000000NA10
CviLerHigh(161lines,n=3)SunshineSoil,HighFep-value2.2E-045.1E-162.7E-111.9E-311.4E-46NA7.0E-191.0E-29NA7.2E-05NA2.2E-03NANA6.1E-11NANA1.0E-291.2E-08
HiloRIL
Herit-abilitysperca52%70%48%59%78%NA57%56%68%51%47%45%37%63%54%53%NA81%56%
34%25%7%6%2%NA12%6%40%13%12%24%15%41%9%13%NA4%11%
Trans-diffbQTLsYYYYYNANNYNNNNYNNNANN
27455NA3162103500NA22
Majorc11001NA0000000000NA10
PopulationsizeandreplicatenumberareincludedwitheachRILaswellasenvironment.aTransgressivesegregationasmeasuredbythepercentageofRILssignificantlyoutsideoftherangeoftheparents.bTransgressivesegregationdeterminedbythepresenceofQTLswithdifferentdirectionsoftheadditiveeffect.cMajorQTLwithR2value.0.20.
doi:10.1371/journal.pone.0011081.t005
parentsgrownwiththeCviLerpopulation,sevenoftheelementsmeasured,Li,Na,Mg,P,Ca,MoandCd,weresignificantlydifferentinbothgrowthconditions,threewerenotsignificantlydifferentineithercondition,andsevenelementswereonlysignificantinonecondition(Tables4and5,TableS2).Fortheelementsthatweresignificantlydifferentintheparentsinbothconditions,4ofthe7QTLswerefoundinbothconditions(Tables4and5,Figure2,TableS2).IntheanalysisoftheparentsgrownwiththeBayShapopulations,nineoftheelementsmeasured,Na,Mg,P,K,Ca,Ni,Zn,MoandCd,weresignificantlydifferentinbothgrowingconditions,twowerenotsignificantlydifferentineithercondition,andeightelementsweresignificantlydifferentinonlyonecondition(Tables4and5,TableS2).OfthenineelementsinbothBayShapopulationsthatweresignificantlydifferentintheparents,20of54QTLswerefoundinbothgrowingconditions(Tables4and5,Figure2,TableS2).
ComparingQTLsfromtheColLerandCviLerpopulationsgrowninsufficientFeSunshinegrowthmedium,weidentifiedfourcommonQTLs(Figure2,TableS2).However,onlyoneofthoseQTL(Mo)wassharedwiththeBayShapopulationgrowninthesamemedium.WhencomparingtheBayShapopulationswithColLer,threeofthe11ColLerQTLsappeartobeinsimilarlocations(Figure2,TableS2).TheCviLerpopulationsgrownindifferentenvironmentssharefourcommonQTLs(Figure2)andtheBayShapopulationsshare25QTLs(Figure2,TableS2).
PLoSONE|www.plosone.org
5
Discussion
NaturalvariationintheA.thalianashootandseedionome
IdentificationofvariationinelementalaccumulationamongA.thalianaaccessionsprovidesanexcellentstartingpointforidentifyinggenesimportantforregulationoftheplantionome,andforunderstandinghowtheionomerespondstodifferentgrowthconditionsandstresses.AlloftheA.thalianaaccessionsprofiledinthisstudyhadatleastoneelementthatwassignificantlydifferent(p,0.01)fromtheCol-0referenceaccession.Severalaccessions,forexampleEst-1andNd-1,hadnosignificantdifferencesinelementalaccumulationbetweenthem,eventhoughtheywerecollectedfromgeographicallydistantsites[26].Oneexplanationisthattheyareadaptedtosoilswithsimilarmineralcontents.Unfortunately,welackinformationonthetypeofsoilfromwhichmostoftheseaccessionswerecollected.TheionomicsignatureofCviismarkedlydifferentfromthatofalltheotheraccessionsinthisstudy.ThelargedifferenceinionomicprofilesbetweenCviandtheotheraccessionsismirroredinthegeneticdistanceofCvifromotheraccessionsasshownbygenome-scaleanalysisofsequencepolymorphisms[26].Thevariationobservedintheaccumulationofmostelementsbetweenshootsandseedsmaybeattributedtothelargedifferencesinthebiochemicalandphysiologicalfunctionsofthesetissues.Forthisreason,itismoreappropriatetofocusonelement-to-elementandaccession-to-accessioncomparisonswhencontrast-ingtheseedandshootionomes.Phosphorusisthefourthmost
June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
Figure1.Elementalcorrelationsforthe5RILpopulations.Solidlinesrepresentapositivecorrelationvalue.Dashedlinesrepresentnegativecorrelationvalues.Thickersolidanddashedlinesindicatecorrelations.0.5or,20.5,respectively.ColLer,CviLerhighFe,andBayShasmallwereallgrowninFe-sufficientSunshinegrowthmedium.TheotherCviLerpopulationwasgrowninSunshinegrowthmediumwithlowFewatering,whiletheLargeBayShapopulationwasgrowninPromixgrowthmedium.doi:10.1371/journal.pone.0011081.g001
abundantelementintheshoots,butthesecondhighest(andwithin10%ofK)intheseeds,likelyreflectingtheroleofthePcontainingmoleculephytateastheanioninstoragecrystalsforcationslikeCa,K,Mg,MnandZn[27].ComparisonoftheseedNaaccumulationofthethreehighshootNaaccessionssuggestsanactivemechanismtoeithertransportNatotheseedinTs-1orexcludeNafromtheseedinWs-0andTsu-1.Ofthe8accessionsthathaveshootMolevelsthataresignificantlydifferentfromCol-0,5showsimilardifferencesintheseedionome,whileNd-1haslowerMointheshoots,buthigherMointheseedsandtheremainingtwolinesarenotsignificantlydifferentintheseeds.Thewidedisparityinionomicsignaturesofeachaccessioninseedsandshootssuggeststhatthemechanismsgoverningelementalaccu-mulationaredistinctinthesetissues.WatersandGrusak[28]demonstratedthatbothremobilizationfromtheleavesandcontinuedsupplyfromtherootscouldcontributetoseedmineralloadinginCol-0,Cvi-0andLer-0,suggestingmultiplecontrolpointsinwhichnaturalvariationcouldhavedifferenteffectsontheseedandleafionomes.Sinceconductingthisscreen,wehaveidentifiedseveralofthegenesunderlyingthevariationdetectedintheleavesofthe12accessions.WeidentifiedHKT1,encodingaNa-transporter,asthegeneunderlyingtheQTLresponsibleforelevatedshootNainbothTsu-1andTs-1[16],MOT1,aputativeMotransporter,asthegeneunderlyingthelowMoinLer-2,Ws-0andVan-0[17]andFPN2,anFeandCotransporterasthegeneresponsibleforincreasedCoinTs-1andSe-0[18].
Gene6Environment6IonomeInteractions
Wedetectedastronginteractionbetweentheenvironmentandthegeneticcontroloftheionomeinouranalysisofelement-to-PLoSONE|www.plosone.org
6
elementgeneticcorrelationsandcomparisonsofQTLsdetectedinthesamepopulationindifferentenvironments.Twooftherobustgeneticcorrelationsidentifiedinall5experiments(Mg-CaandCu-Zn)havebeenidentifiedbyotherresearchersinavarietyofspeciesandenvironments[28–34].Thethreerobustlycorrelatedpairsofelements(Li-Na,Mg-CaandCu-Zn)havesimilarchemicalproperties,suggestingthatsharedbiochemicalaccumu-lationpathwaysaccountsfortheobservedcorrelation.Manyoftheothersignificantcorrelationsthatwedetectedwerespecifictoagivengrowthmedium,RILpopulationorcombinationofboth.Thissuggeststhatmanyoftheserelationshipsarenottheresultofaspecificpathwayforaccumulatingtheseelements,butareindirecteffectsofchangesinthebiochemistryorphysiologyoftheplantsinresponsetodifferentenvironmentalconditionssuchastheresponsestolowFeandPidentifiedbyBaxteretal.[22].Inagreementwiththishypothesis,wealsoobservedalargenumberofenvironment-specificQTLs.ForexampleintheCviLerpopulationgrowninhighandlowFeconditions,thelargesteffectQTLweobservedwasforPintheFesufficientconditions,whichexplained33%ofthevariance(TableS2).SmallereffectQTLsforLi,KandFewereobservedatthesamelocation.Interestingly,noneoftheQTLswereobservedwhentheRILpopulationwasgrownunderlowFeconditions,whichcorrespondswiththelossofanydifferencebetweenCvi-1andLer-2parentinP,KandFe,andthecorrelationsbetweenLi-Fe,Li-KandP-KunderlowFe.IthaspreviouslybeenobservedthatPstatusregulatesFestatus,withhighPreducingFeaccumulationandincreasingexpressionofIRT1encodingtheprimaryrootFe-transporter[35,36].ItispossiblethattheseresponsesareattributedtoreducedFebioavailabilityinthegrowthmediumcausedbyelevatedPdrivingtheprecipitationofFeasFe-phosphate[37].Wealsoobserveda
June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
Figure2.ChromosomemapswithQTLnotedforeachelementinwhichQTLwereidentified.ThewhitecirclewithinthecolorfulboxesrepresentstheestimatedlocationoftheQTL.A.QTLidentifiedinRILColLer.B.QTLidentifiedinRILBaySha,Sunshinegrowthmedium.C.QTLidentifiedinRILBaySha,Promixgrowthmedium.D.QTLidentifiedinRILCviLer,highFeenvironment.E.QTLidentifiedinRILCviLer,lowFeenvironment.
doi:10.1371/journal.pone.0011081.g002
setofco-localizedQTLsonchromosomeoneandcorrespondingsignificantelementcorrelationsintheBayShapopulation:Mg,KandCahadQTLsinbothenvironments,whiletheFeQTLandMg-FeandCa-Fecorrelations,wereonlyfoundinthelargeBaySha,Promixgrowthmediumexperiment.Ghandilyanetal.[19]reportedsimilarphenomena,observingthatQTLsidentifiedformultipleelementsacrosspopulationswerecontingentonenvironmentandtypeoforgan(ieseedortissue)thetraitwasmeasured,whileWatersandGrusak[20]observedthatalargenumberofQTLsforseedelementalaccumulationintheColLerpopulationwerenotdetectedinallexperimentsconductedoveraperiodofyears.
ImplicationsofExperimentdesignandParentDifferencesforQTLDiscovery
AstheQTLstudiesreportedhererequiredaconsiderableamountofeffortandresources,weinvestigatedseveraldifferentexperimentaldesignstooptimizeQTLdiscoverywhilelimitingthenumberofsamplesanalyzed.Whilewediddetectsometransgressivesegregation,findingRILpopulationsinwhichtheparentsaredifferentfortheelementofinterestisclearlythebestwaytoidentifyQTLscontrollingaspecificelement.Acrossallexperiments,majorQTL(s),whicharemucheasiertofinemap,werefarmorelikelytobefoundwhentheparentsweresignificantlydifferentforthatelement.TheIonomicshubdatabase(www.ionomicshub.org)nowhasdatafor.350accessions,
PLoSONE|www.plosone.org
7
includingmostavailableRILpopulationparentsandisanexcellentresourceforfindingaccessionpairswhichdifferforagivenelement.Reducingthenumberofreplicatesfrom3to2intheCviLerpopulation(138lineswhichhaddatafor3samples)onlyreducedthenumberofQTLsidentifiedby5%whiledecreasingthenumberoflinesinthelargeBayShapopulationfrom411to165reducedthenumberofQTLsidentifiedby22%.Thissuggeststhatforionomics,likeothertraits[38],morelinesaremorebeneficialthanmorereplicates.Withthepossibilityofanalyticalorbiologicaloutliers,webelievethatn=2shouldbetheminimumnumberofreplicatesperline.However,theredoesnotappeartobeaneedtoincreasethenumberofplantsanalyzedforeachlinebeyondtwoifitwouldreducethenumberoflinesanalyzedormaketheexperimentscostorscopeunfeasible.
Severalstudiesexaminingelementalaccumulation[19,21,23]inA.thalianaRILpopulationshaveidentifiedmultipleepistaticinteractions.Incontrast,wefoundnosignificant(p-value,0.01)epistaticeffectsinthelargeBayShapopulationandonlyfivesignificantepistaticeffectsinthetwoCviLerexperiments,afewmorethanwouldbeexpectedbychancealone.NoepistaticinteractionswerefoundbetweenQTLsidentifiedusingcompositeintervalmapping(CIM).With411linesintheBayShapopulationwehadmorethansufficientstatisticalpowertodetect2wayepistaticinteractions.ThedifferenceamongpreviouslypublishedstudiesandoursmaybeduetoamoreconservativepermutationbasedsignificancecutoffinourRqtlanalysis.Ultimately,resolutionofthisquestionwillrequirethecloningofgenes
June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
underlyingtheseQTLsandidentifyingepistaticinteractionsbetweenthegenes.
Correlationanalysis
Foreachpairwisecombinationofelementsintheexperiment,PearsoncorrelationcoefficientswerefoundusingthetraycenteredsampledataforpairwisecompleteobservationsutilizingthecorrfunctioninR.Statisticallysignificantcorrelationswereidentifiedusingthet-distributionwithn-2degreesoffreedom(wheren=411intheBayShadata)wheret=(corr*sqrt(n22))/(sqrt(12corr2)),orequivalentlyusingtheF-distributionwith1andn-2degreesoffreedomwhereF=(corr2*(n22))/(12corr2).AconservativeBonferronicorrectionwasappliedtothealphalevelof0.05toadjustforthe19elements.Atotalof171pairwiseelementalcombinationsexist(19choose2=(19)*(1921)/2=171).Thus,onlycorrelationshavingap-valuebelow0.05/171(,2.92461024)wereidentifiedasbeingsignificant.
Conclusion
WehavedemonstratedthatnaturalaccessionsofA.thalianaprovideanexcellentresourceforionomicgenediscovery.Thereisastrongeffectofthegrowthenvironmentonboththeelement-to-elementcorrelationsandtheQTLsunderlyingelementalaccu-mulation.Alltheionomicdataforshoottissuediscussedispubliclyaccessibleforviewing,downloadandre-analysisattheonlinePurdueIonomicsInformationManagementSystem(PiiMS;accessedatwww.ionomicshub.org).
MaterialsandMethodsPlantGrowth
AlloftheseedsfortheA.thalianaaccessionsusedinthisstudywereobtainedfromtheABRC.Theaccessionswereplantedinseven(5.25065.250)potsorintworowsofa20-row(10.506210)tray.Theplantingpatternwasvariedacrosstraystoreducepositionaleffects.Plantsweregrowninaclimate-controlledroomat19–24uCwith10hoursoflightat80to100mEfor36to40days.ThegrowthmediumwereSunshinemixLB2(SunGroHorticulture)(screenedthrougha1/4inchmesh)andPromix(PremierHorticulture).Bothmixturesarepeatbased,buttheydifferintheidentityandgradeoftheothercomponents.Notably,Sunshinehasgypsum,whilePromixhasvermiculiteaswellasaddedmacro-andmicronutrients.BothgrowthmediawereamendedwithLi,Na,Co,Ni,As,Se,Rb,SrandCdatsub-toxicconcentrations[39].TheCviLerlowFe(n=1,151individuals)populationwasgrowninSunshinemixLB2andwateredwith0.25XHoaglandssolution+2.5ml/LFetartrate.TheCviLer(n=3,161individuals)andBayShasmall(n=3,165individuals)populationwasgrowninSunshinemixLB2andwateredwith0.25XHoaglandssolution(FileS7)withadditionalFe(1mlFeHBED/L).ThelargeBayShapopulations(n=2,411individuals)wasgrowninPromix(PremierHorticulture)andalsowateredwith0.25XHoaglandssolution+1mlFeHBED/L.ColLerpopulation(n=3,93individuals)wasgrowninSunshinemixLB2andwateredwith0.25XHoaglandssolution+1mlFeHBED/L.Allplantswerewateredat3to4dayintervals.
QTLAnalysis
Eachtrayforeachofthefivepopulationsexaminedwasnormalizedasfollows.ThedataforeachelementwasdividedintoquartilesandtheInterQuartileRange(IQR),theupperandlowerboundsofthemiddletwoquartiles,wasdetermined.Elementconcentration,whichwasoutsidetherangeoflowerboundminus3timesIQRtoupperboundplus3timesIQR,wasremoved.Eachtraywascenteredsothattheaverageofthetwoparentlinesgrowninthetraywasthesameacrossalltrays.ThemeanvalueacrossalltraysforeachlinewasthenusedforQTLanalysis.ThemarkersetswereobtainedfromtheNaturalwebsite(CviLer,www.dpw.wau.nl/natural/),theBayShawebsite(http://dbsgap.versailles.inra.fr/vnat/Documentation/33/DOC.html)andSing-eretal.[41]forColLer.WeusedreducedmarkersetsfortheCviLerandColLermapping.ThemarkermapsforallQTLmappingexperimentsareincludedassupplementalFileS6.NotethatthechromosomenumberingandorientationdoesnotmatchthefinalQTLresults,aswechangedtheoutputvaluestomatchthepublishedmaps.Weperformedcompositeintervalmapping(CIM)usingQTLCartographerversion1.17f[42],withCIM[43,44]model6,awalkspeedof2cM,awindowof5cM,usingtheforwardandbackwardregressionmethod.Todeterminethresholdvalues,thepermutationmethodwasused[45]with1000permutationsperelementperpopulation(TableS3).
Afterlocatingallmaineffect(single)QTL,epistaticinteractionsbetweentwolociwereinvestigatedusingthescantwofunctioninthesoftwareR/qtl[46].Testswereconductedbetweenallpairwiseloci(bothwithinandbetweenchromosomes)usingtheHaley-Knottregressionwitha2cMwalkingspeed.Onethousandpermutationswereperformedforeachoftheionomictraitstodeterminethegenome-widesignificancethreshold.
IonomicAnalysis
Plantsweresampledbyremoving2–3leaves(0.001–0.005gfreshweight)andwashingwith18MVwaterbeforebeingplacedinPyrexdigestiontubes.Sampledplantmaterialwasdriedfor24hrat88uC,andweighedbeforeopen-airdigestioninPyrextubesusing0.7mLconcentratedHNO3(MallinckrodtARselectgrade)at110uCfor5hours.Eachsamplewasdilutedto6.0mLwith18MVwaterandanalyzedforLi,B,Na,Mg,P,K,Ca,Mn,Fe,Co,Ni,Cu,Zn,As,Se,MoandCd(andRbandSrinsomeexperiments)onanElanDRCeICP-MS(PerkinElmerSciex).NISTtraceablecalibrationstandards(ULTRAScientific,NorthKingstownRI)wereusedforthecalibration.Seedfromtwoplantsofeachaccessionwasobtainedbyincreasingtheirdaylengthto24hours.Threeaccessions(1372,1602and1264)werekeptat4uCfor1monthtoinduceflowering.Theseedwasanalyzedsimilarlytotheplanttissue.Theentiregrowthandanalysisprocedurewasrepeatedtomeasurereproducibility.AllexperimentsweremanagedusingthePurdueIonomicsInforma-tionManagementSystem(PiiMS)[40],andallionomicdataispubliclyavailableforviewing,downloadandreanalysisatwww.ionomicshub.org.
PLoSONE|www.plosone.org
8
ExperimentalDesign
Totesttheeffectsofchangingthenumberofreplicatesorthenumberoflines,weperformedtwoinsilicoexperimentsusingtheQTLdata.1.FromtheCviLerhighFepopulation,wetookthe138linesinwhich3sampleswereanalyzedperlineandmade20subsetsoftheoutlierremoved,traycentered,data:10subsetsinwhich2sampleswererandomlyselectedfromeachlineand10subsetsinwhich1samplewasselectedfromeachline.WethenperformedQTLanalysisasdescribedaboveonthe20subsetsaswellasthefulln=3dataforthe138lines.TableS4acontainsthemeannumberofQTLsidentifiedwithineachsetof10experiments.2.WetookthedatafromthelargeBayShapopulationandmadeasubsetofthe165linesthatwereanalyzedinthesmallBayShapopulationperformedQTLanalysisasdescribedabove.AcomparisonofthenumberofQTLsidentifiedinthisexperimentwiththatofthefull411linesisshowninTableS4b.
June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
SupportingInformation
TableS1Significantpairwisecomparisonsbetween12acces-
Foundat:doi:10.1371/journal.pone.0011081.s006(0.06MBPDF)
FileS3
sionsofA.thaliana.Lineeffectindicatessignificantdifferences(p,0.05)betweenaccessionsforthatelement.A.A.thalianaleafdata.B.A.thalianaseeddata.
Foundat:doi:10.1371/journal.pone.0011081.s001(0.03MBDOC)
TableS2AllQTLforeachelementacrossall5RIL
populations.LOD(logarithmoftheodds)scoreabovetheLODthresholdisindicatedforeachQTL.QTLregionareindicatedbyMIstartandMIendandtheprojectedQTLlocationisgivenincM.CofactorsindicatesthenumberofcofactorsusedintheCIMmodel.Thresholdvaluesindicatethe99%confidenceintervalderivedfrom1000permutations.
Foundat:doi:10.1371/journal.pone.0011081.s002(0.37MBDOC)
Two-wayepistaticinteractionsforeachRILpopula-tionacrossall5chromosomes.Lod.fullisthelog-oddsratioofthefullmodelwithtwolociandtheirinteractioncomparedtothenullmodelwithnoQTL.Lod.fv1isthelog-oddsratioofthefullmodelcomparedtothebestsingleQTLmodelwithonelocusoneitherchromosomeAorB(notnecessarilyatthesamelocationasthefullmodelloci).Lod.intisthelog-oddsratiooftheinteractiontermwhichisfoundbycomparingthefullmodelwithaninteractionterm,tothetwoQTLmodelwithnointeractionterm.Lod.addisthelog-oddsratiooftheadditiveeffects,foundbycomparingthetwoQTLmodel(nointeractionterm)tothenullmodelwithnoQTL.Lod.av1isthelog-oddsratiocomparingthetwoQTLmodelwithnointeractionterm,tothebestsingleQTLmodelwithonelocusoneitherchromosomeAorB(notnecessarilyatthesamelocationasinthetwoQTLmodel).
Foundat:doi:10.1371/journal.pone.0011081.s003(0.04MBDOC)
TableS3
FrequencyplotsofparentallinesandRILsforeach
elementacrossthe5RILpopulations.X-axisrepresentsthecenteredPPM(SeeMethods)ofindicatedelement.Y-axisindicatesfrequencyofoccurrence.Blackverticallinesindicatethe95%confidenceintervaloftheparentsdistribution(i.e.lowerparent21.96SD(pooled)tohigherparent+1.96SD(pooled)).ComparisonBaySha,growninPromixsoil.
Foundat:doi:10.1371/journal.pone.0011081.s007(0.06MBPDF)
FrequencyplotsofparentallinesandRILsforeach
elementacrossthe5RILpopulations.X-axisrepresentsthecenteredPPM(SeeMethods)ofindicatedelement.Y-axisindicatesfrequencyofoccurrence.Blackverticallinesindicatethe95%confidenceintervaloftheparentsdistribution(i.e.lowerparent21.96SD(pooled)tohigherparent+1.96SD(pooled)).ComparisonofCviLer,highFeenvironment.
Foundat:doi:10.1371/journal.pone.0011081.s008(0.06MBPDF)
FrequencyplotsofparentallinesandRILsforeach
elementacrossthe5RILpopulations.X-axisrepresentsthecenteredPPM(SeeMethods)ofindicatedelement.Y-axisindicatesfrequencyofoccurrence.Blackverticallinesindicatethe95%confidenceintervaloftheparentsdistribution(i.e.lowerparent21.96SD(pooled)tohigherparent+1.96SD(pooled)).ComparisonofCviLer,lowFeenvironment.
Foundat:doi:10.1371/journal.pone.0011081.s009(0.06MBPDF)
FileS4
FileS5
FileS6Estimatedgeneticmaps.Estimatedgeneticmapsusing
Resultsofinsilicoexperimentaldesignsimulations.A.
NumberofQTLsdetectedinthesubsetofCviLerlineswith3samplesanalyzed.Firsttworowsindicatetheaverageof10randomlygeneratedsubsetswithn=1or2,thirdrowindicatesthenumberofQTLsidentifiedwiththefulln=3dataset.B.QTLsidentifiedfromthelargeBayShaexperimentwheneitherthefull411linesorthesubsetof165linescorrespondingtothesmallerBayShasetwasused.
Foundat:doi:10.1371/journal.pone.0011081.s004(0.05MBDOC)
TableS4
FileS1FrequencyplotsofparentallinesandRILsforeach
markerdataforourRILpopulationsinQTLCartographerdisplay.Mapfunctionandunitofmeasurementareincludedatthetopofeachmapestimatefollowedbynumberofchromosomes,totalnumberofmarkersmapped,meanandstandarddeviationformarkersandinter-markerdistance.Thetableisarepresen-tationofmarkerdistancebetweenmarkers,acrossall5Arabidopsisthalianachromosomes.Finally,alistofmarkernameandorderacrosschromosomesisincludedforeachpopulation.1.MapdatafortheBayShapopulations.2.MapdatafortheColLerpopulation.3.MapdatafortheCviLerLowFepopulation.4.MapdatafortheCviLerHighFepopulation.
Foundat:doi:10.1371/journal.pone.0011081.s010(0.03MBTXT)
HoaglandsMediaRecipe.ModifiedHoaglandsmedia
usedinthisstudy.
Foundat:doi:10.1371/journal.pone.0011081.s011(0.04MBDOC)
FileS7
elementacrossthe5RILpopulations.X-axisrepresentsthecenteredPPM(SeeMethods)ofindicatedelement.Y-axisindicatesfrequencyofoccurrence.Blackverticallinesindicatethe95%confidenceintervaloftheparentsdistribution(i.e.lowerparent21.96SD(pooled)tohigherparent+1.96SD(pooled))1.1.ComparisonofColLer.
Foundat:doi:10.1371/journal.pone.0011081.s005(0.06MBPDF)
FileS2FrequencyplotsofparentallinesandRILsforeach
Acknowledgments
WewouldliketothanktheArabidopsisBiologicalResourceCenter(ABRC)andthePurdueIonomicsFacility.EB,TA,IA,CO,AG,andIBanalyzedtheSunshinehighFeBayShapopulationaspartofDr.RebeccaDoerges’QTLclass(Stats549)atPurdueUniversityandgratefullyacknowledgeherassistanceandfeedback.
elementacrossthe5RILpopulations.X-axisrepresentsthecenteredPPM(SeeMethods)ofindicatedelement.Y-axisindicatesfrequencyofoccurrence.Blackverticallinesindicatethe95%confidenceintervaloftheparentsdistribution(i.e.lowerparent21.96SD(pooled)tohigherparent+1.96SD(pooled))2.ComparisonofBaySha,growninSunshineSoil.
AuthorContributions
Conceivedanddesignedtheexperiments:JFHMLGDESIB.Performedtheexperiments:ARBLEY.Analyzedthedata:EBTAIAAGCOBLOHMZIB.Wrotethepaper:EBTAMLGDESIB.
PLoSONE|www.plosone.org9June2010|Volume5|Issue6|e11081
Nat.IonomicVar.inArab.
References
1.Alonso-BlancoC,KoornneefM(2000)NaturallyoccurringvariationinArabidopsis:anunderexploitedresourceforplantgenetics.TrendsPlantSci5:22–29.
2.ParanI,ZamirD(2003)Quantitativetraitsinplants:beyondtheQTL.TrendsGenet19:303–306.
3.KoornneefM,Alonso-BlancoC,VreugdenhilD(2004)NaturallyoccurringgeneticvariationinArabidopsisthaliana.AnnuRevPlantBiol55:141–172.4.KeurentjesJ,KoorneefM,VreugdenhilD(2008)Quantitativegeneticsintheageofomics.CurrOpinPlantBiol11:123–128.
5.SaltD,BaxterI,LahnerB(2008)Ionomicsandthestudyoftheplantionome.AnnRevPlantBio59:709–733.
6.BentsinkL,YuanK,KoorneefM,VreugdenhilD(2003)ThegeneticsofphytateandphosphateaccumuulationinseedsandleavesofArabidopsisthaliana,usingnaturalvariation.TheorApplGenet106:1234–1243.
7.RauhL,BastenC,BucklerSt(2002)QuantitativetraitlocianalysisofgrowthresponsetovaryingnitrogensourcesinArabidopsisthaliana.TheorApplGenet104:743–750.
8.LoudetO,ChaillouS,MerigoutP,TalbotecJ,Daniel-VedeleF(2003)QuantitativetraitlocianalysisofnitrogenuseefficiencyinArabidopsis.PlantPhysiol131:345–358.
9.HoekengaOA,MaronLG,PinerosMA,CancadoGM,ShaffJ,etal.(2006)AtALMT1,whichencodesamalatetransporter,isidentifiedasoneofseveralgenescriticalforaluminumtoleranceinArabidopsis.ProcNatlAcadSciUSA103:9738–9743.
10.KobayashiY,KoyamaH(2002)QTLanalysisofAltoleranceinrecombinant
inbredlinesofArabidopsisthaliana.PlantCellPhysiol43:1526–1533.
11.HoekengaOA,VisionTJ,ShaffJE,MonforteAJ,LeeGP,etal.(2003)
IdentificationandcharacterizationofaluminumtolerancelociinArabidopsis(Landsbergerecta6Columbia)byquantitativetraitlocusmapping.Aphysio-logicallysimplebutgeneticallycomplextrait.PlantPhysiol132:936–948.
12.PayneK,BowenH,HammondJ,HamptonC,LynnJ,etal.(2004)Natural
geneticvariationincaesium(Cs)accumulationbyArabidopsisthaliana.NewPhytol162:535–548.
13.ZhangL,ByrnePF,Pilon-SmitsEA(2006)Mappingquantitativetraitloci
associatedwithselenatetoleranceinArabidopsisthaliana.NewPhytol170:33–42.
14.ZhangL,Abdel-GhanySE,FreemanJL,AckleyAR,SchavonM,etal.(2006)
InvestigationofselenatetolerancemechanismsinArabidopsisthaliana.PhysiolPlant.PhysiolPlant128:212–223.
15.BorevitzJO,NordborgM(2003)Theimpactofgenomicsonthestudyofnatural
variationinArabidopsis.PlantPhysiol132:718–725.
16.RusA,BaxterI,MuthukumarB,GustinJ,LahnerB,etal.(2006)Natural
VariantsofAtHKT1EnhanceNa(+)AccumulationinTwoWildPopulationsofArabidopsis.PLoSGenet2:e210.
17.BaxterI,HosmaniP,RusA,LahnerB,BorevitzJ,etal.(2009)Rootsuberin
formsanextracellularbarrierthataffectswaterrelationsandmineralnutritioninArabidopsis.PLoSBiol5:e1000492.
18.MorrisseyJ,BaxterIR,LeeJ,LiL,LahnerB,etal.(2009)Therferroportin
metaleffluxproteinsfucntioninironandcolbalthomeostasisinArabidopsis.ThePlantCell.
19.GhandilyanA,IlkN,HanhartC,MbengueM,BarbozaL,etal.(2009)Astrong
effectofgrowthmediumandorgantypeontheidentificationofQTLsforphytateandmineralconcentrationsinthreeArabidopsisthalianaRILpopulations.JournalofExperimentalBiology60:1409–1425.
20.WatersB,GrusakMA(2008)Quantitativetraitlocusmappingforseedmineral
concentrationsintwoArabidopsisthalianarecombinantinbredpopulations.NewPhytol179:1033–1047.
21.GhandilyanA,BarbozaL,Tisne´S,GrainerC,ReymondM,etal.(2009)
GeneticsanalysisidentifiesquantitativetraitlocicontrollingrosettemineralconcentrationsinArabidopsisthalianaunderdrought.NewPhytol184:180–192.22.BaxterIR,VitekO,LahnerB,MuthukumarB,BorghiM,etal.(2008)Theleaf
ionomeasamultivariablesystemtodetectaplant’sphysiologicalstatus.PNAS105:12081–12086.
23.LoudetO,Saliba-ColombaniV,CamilleriC,CalengeF,GaudonV,etal.
(2007)NaturalvariationforsulfatecontentinArabidopsisthalianaishighlycontrolledbyAPR2.NatureGenetics39:896–900.
24.ErnstWHO,NelissenHJM,TenBookumWM(2000)Combinationtoxicology
ofmetal-enrichedsoils:PhysiologicalresponsesofaZn-andCd-resistantecotypesofSilenevulgarisonpolymetallicsoils.EnvironmentalandExperimentalBotany43:55–71.
25.BaxterI,MuthukumarB,ParkH,BuchnerP,LahnerB,etal.(2008)Variation
inmolybdenumcontentacrossbroadlydistributedpopulationsofArabidopsisthalianaiscontrolledbyamitochondrialmolybdenumtransporter(MOT1).PLoSBiol4:e1000004.
26.NordborgM,HuTT,IshimoY,ToomajianC,ZhengHG,etal.(2005)The
patternofpolymorphisminArabidopsisthaliana.PLOSBiol3:e196.
27.OteguiMS,CappR,StaehelinLA(2002)DevelopingseedsofArabidopsisstore
differentmineralsintwotypesofvacuolesandintheendoplasmicreticulum.PlantCell14:1311–1327.
28.WatersB,GrusakMA(2008)Whole-plantmineralpartitioningthroughoutthe
lifecycleinArabidopsisthalianaecotypesColumbia,Landsbergerecta,CapeVerdeIslands,andthemutantlineysl1ysl3.NewPhytol177:389–405.
29.LiuJ,YangJ,LiR,ShiL,ZhangC,etal.(2009)Analysisofgeneticfactorsthat
controlshootmineralconcentrationsinrapeseed(Brassicanapus)indifferentboronenvironments.PlantandSoil320:255–266.
30.WuJ,YuanY,ZhangX,ZhaoJ,SongX,etal.(2008)MappingQTLsfor
mineralaccumulationandshootdrybiomassunderdifferentZnnutritionalconditionsinChinesecabbage(BrassicarapaL.ssp.pekinensis).PlantandSoil310:25–40.
31.RaboyV,DickinsonDB,BelowFE(1984)Variationinseedtotalphosphorus,
phyticacid,zinc,calcium,magnesium,andproteinamonglinesofGlycinemaxandG.soja.CropScience24:431–434.
32.VreugdenhilD,AartsMGM,KoornneefM,NelissenH,ErnstWHO(2004)
NaturalvariationandQTLanalysisforcationicmineralcontentinseedsofArabidopsisthaliana.PlantCellEnviron27:828–839.
33.BroadleyMR,J.P.H,KingG,AstleyD,H.C.B,etal.(2008)Shootcalciumand
magnesiumconcentrationsdifferbetweensubtaxa,arehighlyheritable,andassociatewithpotentiallypleiotropiclociinBrassicaoleracea.PlantPhysiol146:1707–1720.
34.Garcia-OliveiraA,TanL,FuY,SunC(2009)Geneticidentificationof
QuantitativeTraitLociforcontentsofmineralnutrientsinricegrains.JournalofIntegrativePlantBiology51:84–92.
35.MissonJ,RaghothamaKG,JainA,JouhetJ,BlockMA,etal.(2005)Agenome-widetranscriptionalanalysisusingArabidopsisthalianaAffymetrixgenechipsdeterminedplantresponsestophosphatedeprivation.ProcNatlAcadSciUSA102:11934–11939.
36.HirschJ,MarinE,FlorianiM,ChiarenzaS,RichaudP,etal.(2006)Phosphate
deficiencypromotesmodificationofirondistributioninArabidopsisplants.Biochimie88:1767–1771.
37.WardJT,LahnerB,YakubovaE,SaltD,RaghothamaKG(2008)Theeffectof
ironontheprimaryrootelongationofArabidopsisduringphosphatedeficiency.PlantPhysiol147:1181–1191.
38.Flint-GarciaSA,ThuilletA,YuJ,PressoirG,RomeroSM,etal.(2005)Maize
associationpopulation:ahigh-resolutionplatformforquantitativetraitlocusdissection.ThePlantJournal44:1054–1064.
39.LahnerB,GongJ,MahmoudianM,SmithEL,AbidKB,etal.(2003)Genomic
scaleprofilingofnutrientandtraceelementsinArabidopsisthaliana.NatBiotechnol21:1215–1221.
40.BaxterI(2007)PurdueIonomicsInformationManagementSystem(PIIMS).An
IntegratedFunctionalGenomicsPlatform.PlantPhysiol;10.1104/pp.106.092528.41.SingerT,FanY,ChangHS,ZhuT,HazenSP,etal.(2006)Ahigh-resolution
mapofArabidopsisrecombinantinbredlinesbywhole-genomeexonarrayhybridization.PLoSGenet2:e144.
42.BastenC,WeirBS,ZengZB(2000)QTLCartographerversion1.17f.http://
statgen.ncsu.edu/qtlcart/.
43.ZengZB(1994)Precisionmappingofquantitativetraitloci.Genetics136:
1457–1468.
44.JansenRC(1994)Highresolutionofquantitativetraitsintomultiplelocivia
intervalmapping.GeneticsGenetics136:1447–1455.
45.ChurchhillGA,DoergeRW(1994)Empiricalthresholdvaluesforquantitative
traitmapping.Genetics138:963–971.
46.BromanKW,WuH,SenŒ,ChurchhillGA(2003)R/qtl:QTLmappingin
experimentalcrosses.Bioinformatics19:889–890.
PLoSONE|www.plosone.org10June2010|Volume5|Issue6|e11081
因篇幅问题不能全部显示，请点此查看更多更全内容