summedfreehost

SimplifiedandrepresentativebacterialcommunityofmaizerootsBenNiua,JosephNathanielPaulsonb,c,XiaoqiZhengb,d,andRobertoKoltera,1aDepartmentofMicrobiologyandImmunobiology,HarvardMedicalSchool,Boston,MA02115;bDepartmentofBiostatisticsandComputationalBiology,Dana-FarberCancerInstitute,Boston,MA02115;cDepartmentofBiostatistics,HarvardT.
H.
ChanSchoolofPublicHealth,Boston,MA02115;anddDepartmentofMathematics,ShanghaiNormalUniversity,Shanghai200234,ChinaEditedbyStevenE.
Lindow,UniversityofCalifornia,Berkeley,CA,andapprovedFebruary9,2017(receivedforreviewSeptember28,2016)Plant-associatedmicrobesareimportantforthegrowthandhealthoftheirhosts.
Asaresultofnumerouspriorstudies,weknowthathostgenotypesandabioticfactorsinfluencethecompositionofplantmicrobiomes.
However,thehighcomplexityofthesecommunitieschallengesdetailedstudiestodefineexperimentallythemechanismsunderlyingthedynamicsofcommunityassemblyandthebeneficialeffectsofsuchmicrobiomesonplanthosts.
Inthiswork,fromthedistinctivemicrobiotaassembledbymaizeroots,throughhost-mediatedselection,weobtainedagreatlysimpli-fiedsyntheticbacterialcommunityconsistingofsevenstrains(Enterobactercloacae,Stenotrophomonasmaltophilia,Ochrobactrumpituitosum,Herbaspirillumfrisingense,Pseudomonasputida,Curtobacteriumpusillum,andChryseobacteriumindologenes)representingthreeofthefourmostdominantphylafoundinmaizeroots.
Byusingaselectiveculture-dependentmethodtotracktheabundanceofeachstrain,weinvestigatedtherolethateachplaysincommunityassemblyonrootsofaxenicmaizeseedlings.
OnlytheremovalofE.
cloacaeledtothecompletelossofthecommunity,andC.
pusillumtookover.
ThisresultsuggeststhatE.
cloacaeplaystheroleofkeystonespeciesinthismodelecosystem.
Inplantaandinvitro,thismodelcommunityinhibitedthephytopathogenicfun-gusFusariumverticillioides,indicatingaclearbenefittothehost.
Thus,combinedwiththeselectiveculture-dependentquantificationmethod,oursyntheticseven-speciescommunityrepresentingtherootmicrobiomehasthepotentialtoserveasausefulsystemtoexplorehowbacterialinterspeciesinteractionsaffectrootmicro-biomeassemblyandtodissectthebeneficialeffectsoftherootmicrobiotaonhostsunderlaboratoryconditionsinthefuture.
maize|syntheticcommunity|communityassembly|biologicalcontrolPlantsandanimalsgrow,die,and,importantly,evolvesur-roundedbymyriadmicrobes.
Itisthusnotsurprisingthatre-searchfromrecentyearshasidentifiedcomplex,yetstableandpredictable,microbialcommunitiesassociatedwithspecificsitesonandwithinnumerousplants(1–13)andanimals(14–18).
RapidimprovementsinDNAsequencingtechnologiesanddataanalyseshaveledtoanexplosionofdatadescribingthemicrobialcom-munitiesassociatedwithagrowingnumberofplantandanimalspecies.
Fromstudiesperformedtodate,muchhasbeenlearnedaboutthemicrobialcommunities'speciescompositionanddy-namics(5,10,12,17,19).
However,ingreatpart,duetothelargenumberofspecies,rangingfromhundredstothousands,usuallyfoundinnaturalmicrobialcommunitiesassociatedwithplantsandanimals,westillknowrelativelylittleregardingthepropertiesofthesehost-associatedcommunities(20–23).
Oneapproachtoovercomethechallengesinanalyzingthefeaturesofcommunitiesistoestablishsimplerhost-associatedcommunitiesunderthecontrolledconditionsofthelaboratory(2,5,24–26).
Asimplerhost-microbialcommunitysystemcanbeobtainedbystartingwithagerm-freehostthatistheninoculatedwithawell-definedmicrobialcommunity.
Severalsuchgnotobioticmodelsystemshavebeenestablishedandhavebeenusedinthestudyofhowrepresentative,yetsimplified,communitiesinteractwithmodelhosts(2,5,24–26).
Forexample,simplifiedbacterialcommunitiesfromthehumangutweresuccessfullyestablishedingerm-freemiceandusedtogaininsightsintohowsuchcommunitiesareaffectedbydiet(24).
TheleavesofthemodelplantArabidopsisthalianawereinoculatedwithasimplifiedbacterialcommunitytocharacterizehowseveralplantgenesshapedthephyllospheremicrobiota(26).
Asyntheticcommunitycontaining38bacterialstrainswasusedtoevaluatetheabilitiesofA.
thalianamutantswithalteredimmunesystemstosculpttherootmicrobiome(2).
Inaddition,colonizationofArabidopsisrootsandleavesusingacollectionofseveralhundredbacterialisolatesshowedthattheisolatesformedassembliesresemblingthenaturalmicrobiotaontheircognateplantlocations(5).
Maize(Zeamays)isarepresentativemonocotyledonandacropplantofgreatsignificanceinfoodproduction.
Severalstudieshaveanalyzedthecompositionofthemaize-associatedmicrobialcommunities.
Resultsfromsuchstudieshaveshownthatthecompositionofthemaizerhizospheremicrobiotaisgreatlyinflu-encedbyhostgenetics(27–29),soilphysicochemicalproperties(28),andadditionofdifferenttypesoffertilizer(28).
Quiteim-portantly,thebacterialcommunitiesoftherhizospheresur-roundingmaizerootswereshowntobesubstantiallydifferentfromtheadjacentbulksoilintermsofbacterialrichness,diversity,andrelativeabundanceoftaxa(13,27,29–32).
Inaddition,onthemaizerhizoplane,asignificantreductionofmicrobialdiversityrelativetothebulksoilwasobserved(33).
Theseresultsclearlyindicatethatmaizeplantsgreatlyshapethebacterialcommunitycompositionintheirimmediatevicinity.
SignificanceManyspeciesofmicrobescolonizeplantsasmembersofcomplexcommunities.
Thehighcomplexityofsuchplantmicrobialcom-munitiesposesgreatdifficultyforanyexperimentalanalysesaimedatunderstandingtheprinciplesunderlyingsuchmicrobe–plantinteractions.
Inthiswork,weassembledagreatlysimpli-fied,yetrepresentative,syntheticbacterialmodelcommunitythatallowedustostudythecommunityassemblydynamicsandfunctiononaxenicmaizeseedlings.
Thismodelcommunityin-terferedwiththegrowthofaplantpathogenicfungus,thusprotectingtheplant.
Thismodelsystemwillprovetobeausefulsystemforfutureresearchonplant–microbeinteractions.
Authorcontributions:B.
N.
andR.
K.
designedresearch;B.
N.
performedresearch;B.
N.
,J.
N.
P.
,andX.
Z.
analyzeddata;andB.
N.
andR.
K.
wrotethepaper.
Theauthorsdeclarenoconflictofinterest.
ThisarticleisaPNASDirectSubmission.
Datadeposition:Thenext-generationsequencingdataofthe16SrRNAgenesurveyreportedinthispaperhavebeendepositedintheGenbankdatabase(BioProjectIDno.
PRJNA343280,accessionnos.
SRR4253152–SRR4253209).
ThecompletegenomesequencesofthesevenstrainsofthesimplifiedsyntheticcommunityreportedinthispaperhavebeendepositedintheGenBankdatabase(BioProjectIDno.
PRJNA357031,accessionnos.
CP018756,CP018779–CP018786,CP018845,andCP018846).
TheSangersequencingdatahavebeendepositedintheGenbankdatabase(accessionnos.
KX817243–KX817271andKX817273–KX817279).
1Towhomcorrespondenceshouldbeaddressed.
Email:roberto_kolter@hms.
harvard.
edu.
Thisarticlecontainssupportinginformationonlineatwww.
pnas.
org/lookup/suppl/doi:10.
1073/pnas.
1616148114/-/DCSupplemental.
E2450–E2459|PNAS|PublishedonlineMarch8,2017www.
pnas.
org/cgi/doi/10.
1073/pnas.
1616148114DownloadedbyguestonApril9,2021Inthiswork,wechosemaizeasamodelplanttoserveasahosttodevelopasimplifiedroot-associatedbacterialcommunity.
Wegrewgerm-freemaizeseedlingsinsoilanddidtwoselectiveiter-ationstoarriveatagreatlysimplifiedmodelbacterialcommunitycontainingsevenbacterialstrains:Enterobactercloacae,Steno-trophomonasmaltophilia,Ochrobactrumpituitosum,Herbaspirillumfrisingense,Pseudomonasputida,Curtobacteriumpusillum,andChryseobacteriumindologenes.
Thissimplifiedcommunityre-produciblyassembledontherootsurfaces.
Wewereabletofollowthedynamicsofrootcolonizationbyaculture-dependentmethodbasedonagarplatesselectiveforeachofthesevenspecies.
Byremovingonestrainatatime,wewereabletoprobetheroleplayedbyeachspeciesincommunityassembly.
WefoundthatonlytheremovalofE.
cloacaecauseddramaticchangesofthecom-munitycompositions.
ThisfindingsuggeststhatE.
cloacaemightplayakeyroleinmaintainingtheotherstrainsontheroots.
Wealsofoundthatthisseven-speciescommunityprotectsmaizefromcolonizationbyFusariumverticillioides(formerlyFusariummon-iliforme),thecausativeagentofseedlingblightdisease(34–36).
Thereisprecedentfortheuseofsyntheticcommunitiestostudyplant–microbeinteractions.
However,thesestudieshavebeenlargelyfocusedonArabidopsisandhaveusedalargenumberofstrains.
Forexample,asyntheticcommunityof38strainswasusefulindemonstratingtheimportanceofsalicylicacidinshapingtheroot-associatedmicrobiota(2).
Inanotherstudy,amixtureofhundredsofisolateswasusedtoshowanoverlapinthemicrobiotaofleavesandroots(5).
Onestudydiduseagreatlysimplifiedcommunityofsevenstrains,rationallyselectedfromknownplant-associatedmicrobes,toanalyzetheeffectsofArabidopsisgenesonthestructureofleaf-associatedmicrobialcommunities(26).
Thepresentstudyisdistinctfrompriorworkinseveralways.
Thisworkwasdoneusingmaizeasthehostplant;themodelcommunityassemblesonroots,anditssevenmemberswerechosen,inpart,basedonthemicrobesthatthehostselectedonitsroot.
Inad-dition,theabsoluteabundanceofeachmembercanbeaccuratelyandrelativelyeasilytrackedbygrowthonselectivemedia.
Finally,whereaspriorstudieshavefocusedontheeffectsstemmingfromtheplant,thisworkfocusesoncharacterizingthepotentialin-terspeciesinteractionsamongthemembersofthismodelcom-munityanditsbeneficialeffectsonhosts.
Thus,thissimplifiedcommunitymayserveasausefulsystemtogaininsightsintotheprinciplesbehindcommunityassemblyandbeneficialeffectsofplantmicrobiotas.
ResultsMaizeRootsAssembleaDistinctiveMicrobiota.
Toguideoureffortstoestablishasimplifiedmaizerootbacterialcommunity,wesoughtguidancefrompriorstudiesthatanalyzedthespeciescom-positionofplant-associatedmicrobialcommunities.
Itiswellestablishedthatplanthostsexertastrongselectionsuchthatroot-associatedcommunitiesareclearlydifferentfromthosecommuni-tiesfoundintherhizosphereandbulksoils(20,37).
ThisconclusionwasshowninseminalearlystudiesusingthemodelplantArabidopsis(1,2,4,38).
Furthermore,itwasshownthatsuchhostselectionwasdriven,inpart,bytheplantimmunesignalinginArabidopsis(2).
Subsequently,cleardifferencesbetweentherootmicrobiomeandthemicrobiomeofsurroundingsoilshavealsobeenreportedinstudiesonwheat,tomato,cucumber,rice,barley,sugarcane,andeasterncottonwoodbyusingculture-independentanalyses(7,9,10,33,39,40).
Membersoffourbacterialphyla,Proteobacteria,Firmicutes,Bacteroidetes,andActinobacteria,makeupmostofthebacterialdiversityoftherootmicrobiota(1,4,5,9,20,38,39,41–43).
However,thestructureoftherootmicrobiotaatlowertaxonomiclevels,(e.
g.
,genus)variesdependingonthespecifichostandenvironmentalfactors(1,2,4,5,10,38).
Asafirststeptowarddefiningasimplebacterialcommunitythatassemblesreproduciblyandstablyonmaizeroots,wesetouttorepeatinourlaboratorythecharacterizationofthemicrobialcommunitiesformedinrootsgrowninnaturalsoil.
WedidsobysequencingtheV4regionofthe16SrRNAgeneusingtheIlluminaMiSeqplatform(44,45).
WedesignedourapproachtobesimilartopreviouslyreportedstudiesonArabidopsis(1,4)andcollectedsamplesfromthreecompartments(roots,rhizosphere,andbulksoil)(Fig.
S1A).
Insummary,weobtainedatotalof778,601high-qualityreads,andthereadcountpersamplerangedfrom32,411–68,917,withamedianof41,781(DatasetS1).
Aftertheremovaloftheplant-derivedandlow-abundanceoperationaltaxonomicunits(OTUs),thehigh-qualityreadswereclusteredinto3,211OTUsusing≥97%sequenceidentityasthecutoff(DatasetS2).
Ourresultswereconsistentwithpriorstudiesdonewithmaize,wherehostselectionwasshowntoresultincleardifferencesbe-tweenthecommunitycompositionsofroots(33)ortherhizo-sphere(13,27,29–32)andthecommunitycompositionsofsoils(Fig.
S2AandB).
Weobserved13phylaofrelativelyhighabundance(definedashavinganaveragerelativeabundancegreaterthan0.
3%inanyonesample)and17low-abundancephyla(allcombinedintoasinglecolor)(Fig.
1).
Importantly,ashasbeenfoundinpreviousstudies(1,4,5,9,20,33,38,39,41–43),fourphyladominatedthemaizerootmicrobiota(accountingfor95.
3%ofthetotal):Proteobacteria(84.
4%),Firmicutes(4.
3%),Bacter-oidetes(3.
5%),andActinobacteria(3.
1%).
Amongthesefourphyla,Proteobacteriaweresignificantlyenrichedinmaizeroots,whichisinaccordancewithearlyresults(33),andBacteroideteswereenrichedintherhizosphere(Fig.
1,Fig.
S3A,andDatasetsFig.
1.
Histogramofphylaabundances.
"AandB"representbulksoilsampleswhereplantsweregrown.
Niuetal.
PNAS|PublishedonlineMarch8,2017|E2451MICROBIOLOGYPNASPLUSDownloadedbyguestonApril9,2021S3andS4).
Atthegenuslevel,Burkholderia,Herbaspirillum,Curvibacter,Acinetobacter,Enterobacteriaceae,Stenotrophomonas,andPseudomonas(allbeingProteobacteria)andCurtobacterium(amemberoftheActinobacteria)weresignificantlyenrichedintheroots(Fig.
S3BandDatasetsS5andS6).
OurresultsclearlyshowthattherootenrichmentofProteobacteriawasbuiltonthegatheringofspecificgenera,whichindicatesthatmaizeseedlingsestablishroot-inhabitingbacterialcommunitiesbyselectingalimitednumberofgenera.
Thisfindingsuggestedthepossibilityofassemblingarepresentative,yetsimplified,bacterialsyntheticcommunityfromthepoolofdominantgenerabyrootselection.
AssemblyandCharacterizationofSimplifiedBacterialCommunitiesonMaizeRoots.
Oursecondsteptowarddefiningasimplifiedmaizerootcommunitywastoobtaininoculathatcontainedfewerbacterialstrainsthanbulksoilbutwasstillrepresentativeoftheoverallrootbacterialmicrobiota.
Tothisend,wesurface-sterilizedseeds,germinatedthem,andgrewthemfor1wkinsoil.
TherootsoftheresultingseedlingswerecrushedinPBSbuffer,andthereleasedbacteriawereusedasourstartingmaterialfortwopar-allelstrategies,IandII(Fig.
2A).
Inthefirststrategy(I),weusedtheroottissuesuspensiontoinoculatesurface-sterilizedseeds,andtheresultingseedlingsweregrowninsterile1/2MurashigeandSkoog(MS)agar[0.
8%(wt/vol)](adjustingthepHto6.
0tosimulatethepHofsoilAshowninDatasetS7)for7d.
Bacteriawerethenreleasedfromtheseedlingrootsbywashingafterson-ication.
Bacterialdilutionswerethenplatedon0.
1*trypticasesoyagar(TSA)platestoobtainsinglecolonies.
Wesequencedthe16SrRNAgenesfrom61independentcolonies,andthesegenesfellinto16differentspecies(DatasetS8A).
FourteenspecieswerefromthephylumProteobacteria,andtheothertwowereC.
indologenesandC.
pusillum,whicharemembersoftheBac-teroidetesandActinobacteriaphyla,respectively.
The14Proteo-bacteriaisolatescontainedtwospeciesfromeachofthegeneraOchrobactrumandHerbaspirillum;weselectedasinglespeciesfromeach:O.
pituitosumandH.
frisingense.
FouroftheProteo-bacteriaisolatesweremembersofthegenusEnterobacter;wechosetwooftheseisolates(E.
cloacaeandEnterobacterasburiae)becauseapriorpublicationdescribedthemasendophytesofsweetcornroots(46).
Wewerethenleftwith12strains(DatasetS8B).
Wemixedequalvolumesof108cellspermillilitersuspensionsofeachspeciesandusedthemixturetoinoculate10seedlings.
Inoursecondstrategy(II),ratherthaninoculatingtheroottissuesus-pensionontoseedlings,weplateddilutionsofitdirectlyonto0.
1*TSAplatestoobtainisolatedcolonies.
Weselectedcoloniesdis-playingdiversemorphologiesanddeterminedtheir16SrRNAgenesequences.
Fromthesecolonies,weselected33isolates.
Allofthe33isolatesbelongedtofourphyla,Proteobacteria,Firmicutes,Bacteroidetes,andActinobacteria.
Theseisolates,togetherwithtwoisolatesfromstrategyI(Dyellasp.
andC.
pusilum)plusthreewell-studiedrhizobacterialstrains[BacillusamyloliquefaciensFZB42(47),PaenibacilluspolymyxaM-1(48),andAzospirillumbrasilensesp7(49)],madeupaconsortiumof38strains(DatasetS8C).
Equalvolumesofan108cellspermillilitersuspensionofeachofthese38strainsweremixed,andthemixturewasusedtoinoculatesixaxenicseedlings.
After5dofgrowth,wedeterminedthecommunitycompositionspresentoneachofthe16seedlings'roots(10fromstrategyIandsixfromstrategyII)bydeterminingthe16SrRNAgenesequencespresentandanalyzingtheresultsusingtheQuantitativeInsightsintoMicrobialEcology(QIIME)(50)andcluster-freefilter(CFF)(51)approaches(Fig.
2AandDatasetsS9andS10).
Despitethelargedifferencesintheinoculumcompositions,bothstrategiesyieldedrathersimilarsimplifiedroot-associatedcommunities.
Frombothstrategies,membersofthegenusEnterobacterdom-inatedthecommunities(Fig.
2AandDatasetS9).
FromstrategyI,theinputstrainsS.
maltophilia,O.
pituitosum,H.
frisingense,P.
putida,C.
pusillum,andC.
indologeneswereallpresentatleastata1.
0%averagerelativeabundance.
Incontrast,theinputstrainsKluyverasp.
andPantoeasepticawerepresentataverylowaveragerelativeabundance(<0.
3%).
Raoultellaornithinolyticawasnearlynotdetected,whereasDyellasp.
wasundetectable.
Surprisingly,eventhoughtheinoculumfromstrategyIIwasmuchmorecom-plex,theresultingcommunitieswereverysimilar.
Inallroots,Enterobacterwasthemostdominanttaxon.
S.
maltophilia,O.
pituitosum,H.
frisingense,P.
putida,C.
pusillum,andC.
indologeneswerealsoallpresent.
Inaddition,twospeciesthatwerenotin-volvedintheinoculumofstrategyI,PaenibacilluspolymyxaandAcinetobactercalcoaceticus,weredetectedintheplantsfromstrategyII.
Insummary,thereweresevenstrainscommontobothFig.
2.
Simplifiedbacterialmodelcommunityofmaizerootswasobtainedbyhost-mediatedselection.
(A)Twostrategieswereusedforisolatingbac-teriathatcouldcolonizerootsurfacesfor5d.
A.
calcoaceticusandP.
poly-myxawereincludedinthe38speciesforinoculationinstrategyII,butwerenotinvolvedinstrategyI.
Inbothstrategies,thestructureprofilesofthebacterialcommunityweremeasuredby16SrRNAgenesequencingandus-ingQIIMEcombinedwithCFFforsequencedataanalysis.
TherelativeabundancesofKluyveraandPantoea,whichpossessanidenticalsequencefortheV4regionofthe16SrRNAgenes,aresummedas"KluyveraandPantoea.
"Threespecies,Chryseobacteriumrhizosphaerae,A.
calcoaceticus,andP.
polymyxa,onlyusedinthe38-speciesinoculumforstrategyIIarecoloredpink.
(B)Dynamicsoftheseven-strainbacterialcommunityonthemaizerootsurfaceweretrackedfromday0immediatelyafterinoculationfor15dbyusing16SrRNAgenesequencing.
The"Cellmix"representsthesevenspeciesmixedandusedastheinoculum.
Theyaxisisrelativeabun-dancecalculatedaspercentagesof16SrRNAgenesequences.
E2452|www.
pnas.
org/cgi/doi/10.
1073/pnas.
1616148114Niuetal.
DownloadedbyguestonApril9,2021strategies:E.
cloacae,S.
maltophilia,O.
pituitosum,H.
frisingense,P.
putida,C.
pusillum,andC.
indologenes[Inpreviousexperi-ments,wehadusedfourdifferentspeciesofEnterobacterinthetwostrategies.
BecausepriorpublicationsindicatedthatE.
cloacaewasagoodmaizerootcolonizer(46,52),wechosetoincludeonlythatspeciesinallsubsequentexperiments.
]Fromtheaboveresultswederivedalistofsevenspeciestomixanduseasinoculaforfurthertestingofcommunityassemblage.
TheyareE.
cloacae,S.
maltophilia,O.
pituitosum,H.
frisingense,P.
putida,C.
pusillum,andC.
indologenes.
Usingamixtureofthesesevenspecies,weinoculatedaxenicseedlingsandfollowedthemicrobialcommunitydynamicsontherootsover15d(Fig.
2BandDatasetsS11andS12).
Althoughtherelativeabundanceofeachofthesevenstrainsvariedoverthedurationofourex-periments,allsevenwerestillpresentevenafter15d.
Quiteimportantly,theresultsvariedlittlefromplanttoplant.
Thus,wewereabletoconstructahighlysimplifiedbacterialcommunitythatreproduciblyassemblesontherootsofmaizeseedlings(Fig.
2B).
Clearly,thisseven-straincommunityisbynomeanstheonlycombinationpossible.
Indeed,giventhatwehadinputmultiplestrainsfromseveraldifferentgenerainsomecases,thequestionremainedastowhetherusingstrainsfromdifferentspeciesofthesamegenuswouldhaveanobservableeffectoncommunitydynamics.
Wedidtestaseven-straincombinationwithadiffer-entspeciesofEnterobacter(E.
asburiae)andobtainedverysim-ilarresults(Fig.
S4AandDatasetS13).
Inaddition,wetestedonenine-straincombinationwherewereplacedO.
pituitosumwithOchrobactrumotriciandaddedP.
polymyxaandA.
calcoaceticus.
Inthiscase,therelativeamountsofthesevengenerausedfortheexperimentsshowninFig.
2Bwerestillverysimilar(Fig.
S4AandDatasetS13).
Thus,fortherestoftheexperimentspresentedhere,weusedthesameseven-straincommunityusedintheexperimentsshowninFig.
2B.
Interestingly,thesevenspeciespresentinthesimplifiedcommunityweusedareallmembersofthedominantgeneraofthemaizerootmicrobiota(Fig.
S3BandC).
Thisresultsuggeststhatthissyntheticcommunitymayberepresentativeoftherootmicrobiota.
Totestthisideafurther,weanalyzedourmicrobiomesequencingdatatodetermineifOTUsrepresentingeachofthesevenspecieswereoriginallypresentintheroots.
Indeed,wefoundrepresentativeOTUspresentintherootsofmaizeseedlings(Fig.
S4BandCandDatasetS14).
Anticipatingthatthesesevenstrainsmaybecomewidelyusedbyotherre-searchers,wewantedtoperformsomeinitialcharacterizationofeachone.
Tothisend,wesequencedtheirgenomesanddepos-itedthesequencesintheNationalCenterforBiotechnologyInformationdatabase.
Thisseven-speciessyntheticcommunitythatassemblesonmaizerootscanserveasausefulsystemforinvestigatingthedynamicsofrootcolonization,characterizingtheunderlyinginterspeciesinter-actions,anddefiningtherolethateachmemberplaysinthecom-munity.
Suchstudiesrequirethattheabundanceofeachspeciescanbeeasily,quickly,andaccuratelyquantified.
Althoughnext-gener-ationsequencing(NGS)–based(culture-independent)methodstoobtainspeciesabundanceareuseful(26,44),theygenerallyhavethreelimitations:(i)arelativelylongturnaroundtime(itmaytakeuptoaweekfromDNAextractionuntiltheacquisitionandfinalanalysesofsequencingdata),(ii)theyarenotparticularlywell-suitedtodeterminingtheabsoluteabundanceofeachspeciesinthecommunityaccurately,and(iii)interferenceofhostDNAandbiasintroducedbyPCRamplification.
Recently,somealternativetech-niqueshavebeenappliedindependentlyortogetherwithNGStoreducetheprocessingtime(26,53),determinetheabsoluteabun-danceofbacteriafromagivenphylum(2,54),andeliminatetheinterferenceofhostDNAandPCRbias(55),butitisstillnoteasytouseonesingleexistingapproachtoovercomeorbypassthethreelimitationssimultaneously,whichpromptedustoseekamoreef-ficientsolution.
Giventhesimplicityofourmodelcommunityandtheculturabilityofallofitsmembers,wedecidedtodevelopaculture-dependent,simple,rapid,andeconomicalmethodtoassesstheabundanceofeachmemberofthecommunity.
DevelopmentofaCulture-DependentMethodforQuantifyingCommunityComposition.
Themainchallengetotheestablishmentofaculture-dependentapproachforquantifyingtheabundanceofeachspeciesandthecompositionoftheseven-speciessimplifiedcommunityishowtodistinguisheachspeciesfromtheothersbyculturing.
Asafirststeptoachievethisgoal,weusedthephenotypemicroarray(PM)technology(56).
Weevaluated288uniqueculturecondi-tionsrelatedtothechemicalsensitivityandtheosmoticresponsewitheachofthesevenbacterialstrains(Figs.
S5AandBandDatasetS15).
Fromthesestudies,weobtainedsevendifferentconditionsthatselectivelypermittedonlythegrowthofeachin-dividualstrain(TableS1).
Foreachstrain,therecoveryefficiencyundertheselectiveconditionwasverycloseto1(TableS1).
Thecompositionsofthemixedcellsuspensionsandthestructureofthemodelcommunityonmaizerootswereassessedbybothcol-ony-formingunit(cfu)countingand16SrRNAgenesequencing.
Bothmethodsyieldedverysimilaraveragerelativeabundances,andthecompositionsofthecellmixsuspensionsdeterminedbythetwomethodswerehighlysimilartothe"expected"ratio;thecompositiondeterminedbycfucountingshowedanespeciallysignificantpositivecorrelation(r=0.
8818,P=0.
0091)totheexpectedratio.
Also,thestructureofthemodelcommunityonmaizerootsdeterminedbythetwomethodswasverysimilar.
Thestructuredeterminedby16SrRNAgenesequencingandnor-malizedbytheribosomalRNAoperonnumberofeachspecieswassignificantlypositivelycorrelatedtothestructuredeterminedbycfucounting(r=0.
9643,P=0.
0014)(Fig.
3andDatasetS16).
Similarresultswereobtainedinasecondindependentexperi-ment(Fig.
S5CandDatasetS16).
Theseresultsindicatethattheculture-dependentmethodwedevelopedcanbeusedtodeterminethecompositionoftheseven-speciescommunityonmaizerootsaccurately.
Inpreviousstudies,thequantificationofplantsyn-theticbacterialcommunitieswasmainlyperformedwithculture-independentmethods:automatedribosomalintergenicspaceranalysis(ARISA)(26,53)or16SrRNAgeneampliconse-quencing(2,5).
Bothmethodshavebeenprovedtobewellsuitedforaccuratedeterminationofthecommunitystructure.
However,regardingthemeasurementoftheabsoluteabundanceofeachspeciesinthecommunities,ARISAonlyworksinasemi-quantitativeway(26,53),whereas16SrRNAgeneampliconse-quencingneedstobecombinedwithadditionaltechnique(s)(e.
g.
,catalyzedreporterdeposition-fluorescenceinsituhybridization)(54).
Inthiswork,ourculture-dependentmethodislow-costandveryfast,anditpermitsaccuratedeterminationoftheabsolutecfunumbersofeachspeciespresentontheroots.
ContributionofEachSpeciestotheAbilityoftheOtherSpeciestoRemainPresentintheCommunity.
Recently,muchhasbeenlearnedregardingthestructureofthemicrobialcommunitiespre-sentontherootsofdifferentplantspecies(1,3–7,9,10).
However,thesecommunitiesaresocomplexthatourknowledgeofhowtheyassembleandwhatroleeachspeciesplaysindeterminingcommu-nitystructureisrudimentaryatbest.
Theseven-speciescommunitythatwehavedefinedhererepresentsamuchsimplersystemtocarryoutanalysesoncommunityassemblyandtherolethatindividualspeciesplayintheprocess.
Notonlydowehaveasimplesystembutwehavealsodevelopedconditionsthatallowustodeterminetheabsoluteabundanceofeachspeciesontherootveryquickly.
Asafirststeptowardunderstandingcommunityassembly,wedeterminedtheeffectofremovingindividualspeciesoneatatime.
Weusedthesevenmixesofsixspecieseach(withoneofthespeciesfromtheoriginalseven-membercommunityremoved),plustheseven-speciesmix,toinoculateaxenicseedlingsandthenfollowedthedynamicsofcommunitycompositionbydeterminingtheabundanceofeachspecies(bycfucounting).
TheaverageNiuetal.
PNAS|PublishedonlineMarch8,2017|E2453MICROBIOLOGYPNASPLUSDownloadedbyguestonApril9,2021relativeabundancesobtainedfromtwoindependentexperimentsaredisplayedasbarchartsinFig.
4andFig.
S6A(DatasetS17).
ThedifferentcommunitycompositionsobtainedwereclusteredbasedontheBray–Curtis(BC)dissimilarityindex.
Wealsode-terminedtheα-diversityofeachcommunitybycalculatingitsShannonindex(H)scores.
Theseven-speciescommunity(C7)dynamicsparalleltheresultsofasimilarexperimentalreadyshowninFig.
2B,reinforcingtheconclusionthatthiscommunity'sassemblyanddynamicsarereproducible.
Forsixofthesevensingle-speciesremovalexperiments,theoveralldynamicsarebroadlysimilartotheoveralldynamicsobservedwiththeseven-speciescommunities(i.
e.
,theirHscoresremainrelativelyhighalongthethreesamplingtimes),whichsuggeststhesecom-munitieskeeprelativelyhighevenness.
Instarkcontrast,re-movalofE.
cloacae(Ecl)leadstodramaticchangesincommunitycompositionandα-diversity.
WithoutE.
cloacae,thecommunitiesquicklybecamedominatedbyC.
pusillum,andbyday15,theotherfivespecieswereatornearextinction,drivingtheHscorestonear0(0.
005±0.
004).
TheremovalofP.
putida(Ppu)alsoledtoanincreaseintherelativeabundanceofC.
pusillum,butitstakeoverwasneverasdramatic,andtheresultingHscoresremainedrelativelyhigh.
AkeycontrolwastodetermineifthemaizeseedlingswereindeednecessarytoyieldthedynamicsofthecommunitystructureweobservedandshowinFig.
4andFig.
S6A;thepossibilityremainedthattheresultsweresimplyduetobacterialinteractionsinMSagar,irrespectiveoftheplant.
Totestthispossibility,weperformedcontrolexperimentsinthesamegnotobioticsystembutwithoutmaizeseedlings.
Theseven-speciesmixwasinoculatedinMSagar;wethenfollowedthedynamicsofbacterialcommunitycompositionfor15dandfoundthatthecompositionsofthecommunitiesonmaizerootandinMSagarareindeedsignifi-cantlydifferent(Fig.
S7A–CandDatasetsS18AandS19A).
Inaddition,whenweremovedE.
cloacaeandfollowedthedynamicsinMSagarwithoutseedlings,C.
pusillumdidnottakeover(Fig.
S7D–FandDatasetsS18BandS19B).
Thus,ourkeyobservationsindeeddependonthepresenceofthemaizeseedlings.
Itwasnotclearwhetherthechangesintherelativeabundanceofeachspecies,asdisplayedinFig.
4,wereduetoovergrowthofC.
pusillum,disappearanceoftheotherspecies,orboth.
Ourculture-dependentassay,wherewedetermineabsoluteabundances,Fig.
3.
Comparisonoftheculture-dependentselectiveplatemethodandtheculture-independent16SrRNAgenesequencingmethodforde-terminingthecompositionoftheseven-speciessyntheticcommunity.
Thetwomethodswereappliedinparalleltosamplesofmixedcellsuspensions(Cellmix)andto15-d-oldmaizerootscolonizedwiththeseven-straincommunitylabeledas"Inplanta.
"Theoutputsfromthe16SrRNAgenesequencingandtheculture-dependentmethodaremarkedas"Reads"and"CFUs,"respectively.
"Normalized"indicatesthattheresultsobtainedwithsequencingwerenormalizedwiththeribosomalRNAoperonnumberofeachofthesevenspecies,and"Expected"meansthesupposedcompositionsofthemixedcellsuspensions.
TheSpearman'srankcorrelationcoefficientswerecalculatedforcellmixcompositionsvs.
expectedratio,andcommunitystructurewasdeterminedby16SrRNAgenesequencingvs.
communitystructuredeterminedbycfucounting.
Fig.
4.
Clusteringanalysisofthecompositionsoftheseven-speciessyntheticcommunityandsevensix-speciescommunitiesalongthe15-dgrowthofmaizeseedlings.
Thesampleswerecollectedatday5,day10,andday15.
Sma,Opi,Cpu,Ecl,Cin,Hfr,andPpurepresentthesix-speciescommunitiesresultingfromtheremovalofS.
maltophilia,O.
pituitosum,C.
pusillum,E.
cloacae,C.
indologenes,H.
frisingense,andP.
putida,respectively.
ThecommunitiesareclusteredbasedontheBCdissimilarityindex.
E2454|www.
pnas.
org/cgi/doi/10.
1073/pnas.
1616148114Niuetal.
DownloadedbyguestonApril9,2021allowedustoresolvethismatter.
InFig.
5andFig.
S6B(DatasetS20),weplottedtheabsoluteabundancesofeachspeciesinC7andthecommunitylackingE.
cloacae(Ecl)alongdifferentsamplingtimes.
InC7,theabsoluteabundancesoffivespeciesincreasedafterinoculation,whereasP.
putidaandH.
frisingensedecreasedinnumber(Fig.
5A).
ThegrowthofE.
cloacaeandS.
maltophiliareachedsaturationatday5,andtheabundanceofC.
indologenesreacheditsmaximumatday10.
Afterward,thesethreespeciesmaintainedrelativelystablelevels.
Incontrast,O.
pituitosumandC.
pusillumkeptgrowingduringthe15doftheexperiment.
Themaximumabundancesofthefivespecieswere7.
9*104to1.
3*106cfu/mg(offreshweightofmaizeroot),upfrom2.
1*103to6.
1*104cfu/mgonday0,determinedimmediatelyafterinoculation.
P.
putidaandH.
frisingensedisplayedacontinuousdecline.
Atday15,theabundancesofthesetwospeciesdroppedto4.
7*102and6.
3*102cfu/mgfrom9.
9*104cfu/mgand5.
3*104cfu/mg,respectively(Fig.
5A).
Incontrast,intheEclexperiment(Fig.
5B),onlytheabundancesofC.
pusillumandH.
frisingenseroseafterinoculation,whereastheabundancesoftheotherfourspeciesdroppedsharplyfrom5.
6*103to1.
5*105cfu/mgtolessthan1.
0*102cfu/mg.
H.
frisingensegrewuntilday5,whenitsabundancereachedthemaximum,andthenquicklydecreasedfrom1.
5*105cfu/mgtolessthan1.
0*102cfu/mg.
C.
pusillumgrewuntilday10,anditsabundanceincreasedfrom4.
9*103cfu/mgto1.
8*106cfu/mgandremainedatarelativestablelevel(Fig.
5B).
OurresultsshowthattheremovalofE.
cloacaecausedthedecreaseoftheabsoluteabundancesofS.
maltophilia,O.
pituitosum,andC.
indologenes.
Moreover,theabundancesofP.
putidaandH.
frisingensedroppedevenfasterinEclthaninC7,whereasthegrowthofC.
pusillumwasdramaticallystimulated.
Thus,theresultsprovedthatthecommunitydynamicsweobservedfromFig.
4andFig.
S6AresultedfrombothovergrowthofC.
pusillumanddisappear-anceoftheotherspecies.
Togainabettersenseofthereproducibilityofthecommunitydynamics,weanalyzedtheresultsobtainedfromeachindividualplant.
Tothisend,wecomparedtheBCdistancebetweeneachsix-speciescommunityandtheseven-speciesmodelcommunityplantbyplant(Fig.
6,Fig.
S6C,andDatasetS21).
ThevaluesoftheBCdistancebetweenthebacterialcommunityoneachseed-linginoculatedwiththemixesofsixspeciesandtheseven-speciesmodelcommunitycolonizedoneachplantat5,10,and15dpostinoculationareplottedinFig.
6.
Thedatapointsforeachcomparisonwithineachplantare,ingeneral,narrowlydistributed,andtheresultsarealsoverysimilarfromplanttoplantforthesameremovalandtimepoint.
Thus,ingeneral,theseresultscorroboratethereproducibilityofthismodelsystem.
Themajorityofdatapointsfromthesix-speciescommunities(excludingEcl)representBCdistancevaluesoflessthan0.
5.
Thisfindingindi-catesthatthecompositionsofthecommunitiesinhabitingeachindividualplantweresimilartothecompositionofC7.
Notably,theBCdistancevaluesrosestrikinglyandmaintainedrelativelyhighvalues(0.
81–0.
96)whenE.
cloacaewasremoved.
Also,inPpu,threeplantsatday15showedrelativelyhighBCdistancevalues(0.
76–0.
80).
Theresultsareconsistentwiththedendro-gramsshowninFig.
4,wherePpuandEclclustertogetheratday15.
Takentogether,ourresultspointatE.
cloacaeasthemostimportantspeciesintheseven-speciescommunityintermsofthecommunity'sabilitytomaintainallsevenspeciespresentatalltimes.
Earlierstudieswithsyntheticbacterialcommunitiesshowedthatplantimmunesignalingisinvolvedinsculptingtherootmicrobiome(2)andthatplanthostgeneticfactorsaffectthephyllospheremicrobiota(26)andthemicrobiotaspecializationtotheirrespectivecognatehostorgans(5).
Thus,priorstudiesweremainlyfocusedontheeffectsofhostchangesonthecommunitydynamicsorassembly.
Incontrast,inourworkwiththeseven-speciescommunity,effortsareprimarilyplacedonrevealingtheFig.
5.
Growthofeachstrainintheseven-speciessyntheticcommunity(A)andthesix-speciescommunitywithoutE.
cloacae(B)onmaizerootsde-terminedbycfucounting.
Fig.
6.
ComparisonofBCdistancesbetweeneachsix-speciescommunityandtheseven-speciesmodelcommunityplantbyplant.
EachdotrepresentsaBCdistancevaluebetweenthebacterialcommunityononeplantcolo-nizedbyasix-speciescommunityandthebacterialcommunityononeplantcolonizedbytheseven-speciesmodelcommunity.
Theasteriskindicatesastatisticallysignificantdifference(*P<0.
05).
Sma,Opi,Cpu,Ecl,Cin,Hfr,andPpudesignatethesix-speciescommunitiesasinFig.
4.
Thetri-angle,square,andcircularsymbolsrepresentthesamplescollectedat5d,10d,and15dafterinoculation,respectively.
Fisher'sleastsignificantdif-ference(LSD)testwasusedfortheanalysis.
Niuetal.
PNAS|PublishedonlineMarch8,2017|E2455MICROBIOLOGYPNASPLUSDownloadedbyguestonApril9,2021roleofpotentialinteractionsamongbacteriainassemblingthecommunityonmaizeroots.
InhibitionEffectsoftheModelCommunityAgainstthePathogenF.
verticillioides.
Inadditiontostudyingthemodelcommunityassemblyonmaizeroots,weevaluateditspotentialbeneficialeffectsonthehostplants.
Therootmicrobiotahasbeendemon-stratedtocontributetothegrowthandhealthofhostplants(21).
First,wecomparedtheseedgerminationrate,rootfreshweight,androotlengthofthecommunity-treatedplantswiththosepa-rametersinsterileplants.
Nosignificantdifferencesinthesepa-rametersweredetectedbetweenthecommunity-treatedandsterileplants(Fig.
S8andDatasetS22).
However,wedidfindeffectsofthemodelcommunitywhenweinvestigateditsinter-actionswithaplantpathogenicfungus.
WeexaminedtheinhibitoryeffectofthecommunityagainstthefungalpathogenF.
verticillioides,thecauseofmaizeseedlingblight(34–36).
Throughaninvivofungalcolonizationassay,wefoundthatthegrowthofF.
verticillioidesmyceliaonthesurfacesoftheseedsinoculatedwiththecommunitywassignificantlydelayedcomparedwiththegrowthonthebacteria-freeseeds(MovieS1)andtheseedsinoculatedwithEscherichiacoliDH5α(Fig.
7andDatasetS23A).
Twodaysafterinoculation,thefungalmyceliastartedappearingonthesurfacesofseedstreatedwithF.
verticillioidesandF.
verticillioidesjointlywithE.
coli.
Theaveragecolonizationratesofthetwotreatmentswerethesame,13%(Fig.
7A).
Onthethirddayafterinoculation,myceliawerepresentonthesurfacesofnearlyalloftheseedstreatedwithF.
verticillioides(100%)andF.
verticillioidesjointlywithE.
coli(97%).
However,therewasnofungalgrowthvisibleonthesurfacesofseedsinoculatedwithF.
verticillioidesjointlywiththemodelcommunityuntilthefourthdayafterinoculation.
Atthatpoint,thepercentageofseedscolonizedbyfungiwasonly3%,significantlylowerthanthepercentageoftheseedsinoculatedwithF.
verticillioides(100%)andF.
verticillioidesjointlywithE.
coli(100%)(Fig.
7AandB).
SuchdelayedgrowthofF.
verticillioideslasteduntiltheninthdaypostinoculation.
Atthatpoint,thefungalcolonizationpercentageoftheseedsinoculatedwithF.
verticillioidesjointlywiththemodelcommunityreached82%.
However,thiscolonizationpercentagewasstillsignificantlylowerthanthecolonizationpercentageofseedstreatedwithF.
verticillioidesandF.
verticillioidesjointlywithE.
coli(P=0.
0303)(DatasetS23A).
Onthetenthdaypost-inoculation,thedifferencesbetweenthecolonizationpercentagesoftheF.
verticillioidesjointlywiththemodelcommunityandtheothertwobecamenonsignificant(Fig.
7A).
Wealsocomparedtheseverityofthemaizeseedlingblightofthethreetreatments.
OurresultsrevealedthattheseedlingstreatedwithF.
verticillioidesjointlywiththemodelcommunitydisplayedthelowestdiseaseseverityindex(Fig.
7CandDatasetS24A;diseaseranksareprovidedinFig.
S9A).
Also,muchlessfungalgrowthwasobservedonfungi-colonizedseedsoftheF.
verticillioidesjointlywithcommunitythanthefungalgrowthoftheothertwotreatments(Fig.
7B).
Inshort,thisseven-speciescommunityiscapableofreducingmaizeseedlingblightbydelayingthecolonizationofF.
verticillioides.
Thisfindingsuggeststhepotentialofusingthiscommunityforbiologicalcontrolofthemaizeseedlingblight.
WealsotestedtheinhibitorypropertyofthebacterialcommunityagainstF.
verticillioidesinvitroonpotatodextroseagarandnutrientagarplates.
OurresultsshowedthatthecommunitysignificantlysuppressedthegrowthofF.
verticillioidesonbothtypesofagarplates(Fig.
7DandDatasetS25).
Next,wetestedtheinhibitoryeffectsofeachofthesevenspe-ciesindividuallyagainstF.
verticillioides.
WefoundallofthesevenspeciesdelayedthecolonizationofF.
verticillioidesandreducedtheseverityoftheseedlingblighttovaryingdegrees(Fig.
S9BandCandDatasetsS23BandS24B).
Amongthesevenspecies,E.
cloacaeshowedthehighestefficiencyindelayingF.
verticillioidesFig.
7.
Antagonisticeffectoftheseven-speciesmodelcommunityagainstF.
verticillioides.
(A)Delayingeffectoftheseven-speciesmodelcommunityonF.
verticillioidescolonization.
(B)Inhibitoryeffectoftheseven-speciescommunityonF.
verticillioidesmycelialgrowthonthesurfacesofmaizeseeds.
Photographsweretakenonday4andday10afterinoculation.
(Scalebars:2mm.
)(C)Biocontroleffectoftheseven-speciescommunityagainstmaizeseedlingblightcausedbyF.
verticillioides.
(D)Inhibitoryeffectsoftheseven-speciescommunityonthegrowthofF.
verticillioidesonpotatodextroseagar(PDA)andnutrientagar(NA)platesinvitro.
Fve,treatmentwithF.
verticillioidesalone;Fve+C7,jointtreatmentwithF.
verticillioidesandtheseven-speciesmodelcommunity;Fve+H5α,jointtreatmentwithF.
verticillioidesandE.
coliDH5α.
Asterisksindicatethatdifferencesamongthemeansrepresentedbythecolumnsarestatisticallysignificant(*P<0.
05).
ns,nonsignificantdifference.
Fisher'sLSDtestwasusedfortheanalysis.
E2456|www.
pnas.
org/cgi/doi/10.
1073/pnas.
1616148114Niuetal.
DownloadedbyguestonApril9,2021colonization.
ThecolonizationpercentageofseedlingstreatedwithE.
cloacaewassignificantlylowerthanthecolonizationpercentageofseedlingstreatedwithF.
verticillioidesandF.
verticillioidesplusE.
coliuntilthesixthdaypostinoculation,2dearlierthantheseven-speciescommunitytreatment(Fig.
S9B).
Similarly,E.
cloacaealsoexhibitedthebestbiocontroleffectagainstmaizeseedlingblightamongthesevenmembersofthecommunity.
TheaveragediseaseseverityindexoftheseedlingstreatedwithE.
cloacaewasaround60,whichissignificantlylowerthantheaveragediseaseseverityindexofseedlingstreatedwithF.
verticillioidesandF.
verticillioidesjointlywithE.
coli,butstillhigherthantheaveragediseaseseverityindexoftheseven-speciescommunitytreatment(Fig.
S9CandDataset24B).
Thus,ourresultsindicatethatalthougheachoneofthesevenmembersofthecommunityshowedsomebiocontrolef-fectsagainstF.
verticillioidesindividually,sucheffectswerenotasstrongasthebiocontroleffectsexhibitedbytheentirecommunity.
Thisfindingcorroboratesthediversityresistancehypothesis(57,58).
TheincreasedresistancetoF.
verticillioidesinvasioninthecommunity-treatedmaizeseedlingsmaybeduetoahighnumberofspeciesinteractions(57)andintensifiedcompetitionfornichespace(59,60).
Theseinplantainteractionsbetweentheseven-speciessyntheticcommunityandF.
verticillioidesmayprovetobeapromisingsystemtoinvestigatethebiologicalcontroleffectsofplantmicrobiotas.
DiscussionThatplantsselectivelyrecruitmicrobesfromthesoiltoestablishacharacteristicyetverycomplexmicrobiotaontheirrootshasbeendemonstratedbynumerousstudies(1–9,61,62).
Thiscom-plexitymakesdetailedmechanisticstudiesonhowrootmicrobialcommunitiesassembleandfunctionchallenging,giventhetoolsavailabletoday.
Onepossiblewaytogaininsightsintothework-ingsofrootmicrobialcommunitiesistoestablishgreatlysimplifiedmultispeciescommunitiesabletocolonizeplantrootsstably.
Theresultspresentedinthisworkprovidearobustandrepresentativeseven-speciescommunityofthemaizerootmicrobiota,whichreproduciblyassemblesandpersistsontherootsofaxenicmaizeseedlings.
Beforethedevelopmentoftheseven-speciescommunity,weanalyzedthemicrobiotathatassembledontherootsofmaizeseedlingsgrowninsoil.
Clearly,thecommunitiesthatassembledonthoserootswereverydifferentfromthecommunitiesfoundintherhizosphereandtheoriginalsoil(DatasetS26).
Thoseresultsareconsistentwithearlierfindingsonotherplantspecies(1,2,4,7,9,10,20,33,37–40).
Inpreviousstudiesonthemaizemicro-biome,theremarkabledifferencebetweenthebacterialcommu-nityoftherhizosphere(13,27,29–32)orrhizoplane(33)andthebacterialcommunityofbulksoilwasalsodemonstrated.
Thus,ourfindingscorroboratetheknowledgeoftheniche-mediatedplantmicrobiotastructure.
Basedontheseresults,whichindicatedthatthemaizeseedlingsselectasubsetofspeciespresentinbulksoil,weusedhost-mediatedmicrobiotaselectionasanaidinobtainingagreatlysim-plifiedsyntheticcommunitythatwasrepresentativeofthemorecomplexinitialcommunitiesandabletocolonizeroots.
Suchap-proacheshavebeenrecentlyproposedandappliedtoimproveplantandanimalhealth(63–65).
Thesestrategiesemphasizetheselectionofmicrobialcommunitiesindirectlythroughthehost(63).
Aftertworoundsofhostselection(strategyIinFig.
2A),weobtainedasim-plifiedbacterialsyntheticcommunityconsistingofsevenspeciesthatweremembersofthemaizerootmicrobiota(Fig.
S4BandC).
Amongthesevenstrains,P.
putidaisthemostabundantintherootmicrobiome,whichisinconsistentwiththepreviousfindingsthatPseudomonasdominatedinmaizerhizosphere(29,30,66)andrhi-zoplane(33)bacterialcommunities,wheretherelativeabundanceofPseudomonaswasupto67%(33).
Inaddition,Herbaspirillum,Enterobacter,Stenotrophomonas,Chryseobacterium,andOchro-bactrumwereallreportedasdominantmembersofthemaizeroot-associatedmicrobiota(13,66),whereasCurtobacteriumwascharacterizedasanendophyteinhabitingtherootofsweetcorn(46).
Therefore,ourhost-mediatedmicrobiotaselectionworkedefficientlytocapturethedominantmembersofthemaizerootmicrobiome.
Byusingthisapproachoflettingthehostselectbeforeisolatingindividualbacterialspecies,mostnoncandidatespecieswereex-cluded.
Anotheradvantageofthisselectionmodelisthatthereisnoneedtoconsiderlargenumbersofcombinationsofcandidatebacterialstrains,becausetheplanthostselectsthecommunityasawhole.
Consideringthatdifferentplantspeciesmayshareafairportionoftheirrootmicrobiota(1,4,5,9,10,20,38,39,41–43),ahostselectionapproachmightleadtocommunitiesthatassembleandaremaintainedindifferentplantspecies.
Theseven-speciesmodelbacterialcommunitywehavedevelopedismuchsimplerthanthesyntheticcommunitiesdevelopedusingArabidopsisroots(2,5).
However,thesevenspeciesarestilltaxonomicallyrepre-sentativeofthemaizerootmicrobiota(Figs.
S3BandCandFig.
S4BandC).
Thus,thissimplifiedcommunityshouldserveasausefullaboratorysystemtostudyrootassemblagesandmaizebacterialmicrobiotainteractionsindetail.
Nonetheless,weac-knowledgethatitisimpossibleforthisseven-speciesmodelcommunitytopossessallofthebacterialinteractionsandfunc-tionswithintherootmicrobiota.
Wealsonoticethattheratioofthesevenstrainschangedsignificantlyinthesimplifiedcommunitycomparedwiththeirratiosintherootmicrobiota(Fig.
S4C).
Similarchangeswerealsoobservedinasyntheticbacterialcom-munityrepresentingthemostabundantphylaintheArabidopsisphyllosphere,wheretherelativeabundanceofRhodococcussp.
belongingtoActinobacteriawas1%(67,68)andincreaseddra-maticallyto40%inthesyntheticcommunity(26),whichledtothefindingthatthedistributionsoftheRhodococcussp.
andothermemberswerelesseveninthesyntheticcommunitythaninthephyllospheremicrobiome,andthattheratioofProteobacteriaandActinobacteriadecreasedtoaround1:1inthedefinedcommunity(26)fromabout15:1inthephyllospheremicrobiota(41).
Thisoutcomemaybeduetothevariationinthenumberofmicrobialinteractions.
Inthesyntheticcommunities,theabsenceofsomanyothermicrobesmayleadtodifferentinteractionnetworksamongthesyntheticcommunitymembers,whichprobablygivesrisetothedifferentrelativeabundances.
Webelievethatourdevelop-mentofaculture-dependentassaytoobtaintheabsoluteabun-dancesofindividualsfromeachspeciesquicklyandeasilymakesthismodelsystemveryattractive.
Theturnaroundtime,accuracy,cost,andeaseofmanipulationmakethisassayvery"user-friendly"comparedwithnucleicacidsequencing.
Oneofthekeyfundamentalquestionsinstudiesofmicrobialcommunitiesishowmultiplespeciesofbacteriacoexistandas-sembleintoacommunity.
Innaturalsettings,theassemblageofabacterialcommunitycanbeinfluencedbydiversefactors,in-cludingtheinteractionswithinthecommunity,thehosts,andthespatialdistributionsofthemembers.
Bacterialinteractions(69),hostgenotypes(2,26),nichespecificities(5),andspatialdistri-butions(70)arekeyfactorsfortheassemblageofsimplesyntheticcommunities.
Duetothehighcomplexityofthemicrobiotainenvironmentalsettingssuchasplantroots,itisdifficulttoprobethesignificanceofbacterialinteractionsexperimentally.
Thesimplifiedseven-speciesbacterialcommunitythatestablishesonmaizerootshasallowedustoinvestigateexperimentallytheim-portanceofeachmembertotheassemblageofthecommunity.
Specifically,weperformedcolonizationassayswithallcommunitycombinationslackingonespecies.
Amongthesevenuniquemixeslackingonespecies,onlyoneledtodramaticchangesincom-munitycomposition:WhenweremovedE.
cloacae,C.
pusillumtookoverthepopulation.
ThisresultsuggeststhatE.
cloacaeisakeymemberforassemblageofthecommunity,andisreminiscentoftheecologicalconceptofkeystonespecies(71,72).
Fromtheabsoluteabundancedataweobtained(Fig.
5),wefoundthattheremovalofE.
cloacaeledtoasharpdecreaseintheabundancesofNiuetal.
PNAS|PublishedonlineMarch8,2017|E2457MICROBIOLOGYPNASPLUSDownloadedbyguestonApril9,2021S.
maltophilia,O.
pituitosum,andC.
indologenes.
Incontrast,C.
pusillumgrewmuchmorerapidlyintheabsenceofE.
cloacae.
ThesepopulationdynamicssuggestthatE.
cloacaemayinteractpositivelywithS.
maltophilia,O.
pituitosum,andC.
indologenesandnegativelywithC.
pusillum.
However,themoleculardetailsofthebacterialinteractionswithintheseven-speciescommunityarelikelyverycomplexandwillbethesubjectoffuturestudies.
Theplant-associatedmicrobiotacanserveasaprotectivebarrieragainstinvadingpathogens(21,58,63).
Suchbeneficialcommunitiesarecapableofcontrollingpathogensinbothdirectandindirectways(21).
Recently,alteringthemicrobiotatoprotecthostsfrompathogensinwaysthatdonotnecessarilykillthepathogenshavebeenproposedandapplied(73–75).
How-ever,becauseofthehighcomplexity,knowledgeofthemecha-nismsthroughwhichthemicrobiotaprotectsplantsfrompathogensisverylimited.
Ourgreatlysimplifiedsyntheticcom-munitycontrolledmaizeseedlingblightcausedbyF.
verticillioides(34)throughinhibitingfungalcolonization(Fig.
7A)andarrestinghyphalexpansiongrowth(Fig.
7D)undergnotobioticconditionsinthelaboratory.
Althoughwedonotyetunderstandtheunderlyingmolecularmechanismsforthisbiologicalcontrol,thereishopethatthesimplicityofthecommunitywillmakesuchstudiespos-sibleinthenearfuture.
MaterialsandMethodsThebacterialmicrobiotaofthemaizeroots,rhizosphere,andbulksoilwasanalyzedbysequencingthe16SrRNAgeneamplifiedbyPCRfromthegenomicDNAextractedfromthethreesampletypesdiscussedaboveundergreenhouseconditions.
ThesequencingwasperformedontheIlluminaMiSeqplatform(44,45)(DatasetS27).
TheanalysisofsequencingdatawascarriedoutwiththeQIIME(50)pipelineandCFF(51)approaches.
Thesimplifiedsyntheticcom-munitieswereassembledontherootsofaxenicmaizeseedlingsgrowninanMSagar-basedgnotobioticsystem.
TheselectivegrowthconditionforeachofthesevenstrainsofthemodelcommunitywasdeterminedwithPMtechnol-ogy(56).
Thedynamicsofthecompositionsoftheseven-speciesmodelcom-munityandthesix-speciescommunities(resultingfromtheremovalofeachoneofthesevenspecies)werefollowedbycfucountingonselectiveplatesand16SrRNAgenesequencing.
Thebiologicalcontroleffectsoftheseven-speciessyntheticcommunityagainstthemaizeseedlingblightdiseasewereevaluatedunderlaboratoryconditions.
TheinhibitoryeffectsofthemodelcommunityagainstF.
verticillioideswerealsotested.
Furtherdetailsofmate-rialsandmethodologyareprovidedinSIMaterialsandMethods.
ACKNOWLEDGMENTS.
WethankKojiYasudafordataanalysis;Dr.
CharlesBaconforhisgenerousgiftofF.
verticillioidesstrains;Dr.
ScottChimileskifortheimaginganalysesoffungalgrowthinhibition;Dr.
AlejandroReyesandDr.
MengWuforvaluableadviceondataanalysisandcommentsonthemanuscript;andDr.
NicholasGriffin,Dr.
KorneliaSmalla,Dr.
Ching-HongYang,andmembersoftheR.
K.
laboratoryforvaluableadvice.
B.
N.
wassupported,inpart,byaChinaGreenHealthAgriculturePostdoctoralFellowship.
J.
N.
P.
wassupportedbyNIHGrants5R01-HL111759and1U01-CA190234(toJohnQuackenbush).
ThisworkwassupportedbyNIHGrantGM58218(toR.
K.
).
1.
LundbergDS,etal.
(2012)DefiningthecoreArabidopsisthalianarootmicrobiome.
Nature488(7409):86–90.
2.
LebeisSL,etal.
(2015)PLANTMICROBIOME.
Salicylicacidmodulatescolonizationoftherootmicrobiomebyspecificbacterialtaxa.
Science349(6250):860–864.
3.
WagnerMR,etal.
(2016)Hostgenotypeandageshapetheleafandrootmicrobiomesofawildperennialplant.
NatCommun7:12151.
4.
BulgarelliD,etal.
(2012)RevealingstructureandassemblycuesforArabidopsisroot-inhabitingbacterialmicrobiota.
Nature488(7409):91–95.
5.
BaiY,etal.
(2015)FunctionaloverlapoftheArabidopsisleafandrootmicrobiota.
Nature528(7582):364–369.
6.
CardinaleM,GrubeM,ErlacherA,QuehenbergerJ,BergG(2015)Bacterialnetworksandco-occurrencerelationshipsinthelettucerootmicrobiota.
EnvironMicrobiol17(1):239–252.
7.
deSouzaRSC,etal.
(2016)Unlockingthebacterialandfungalcommunitiesassem-blagesofsugarcanemicrobiome.
SciRep6:28774.
8.
BergG,SmallaK(2009)Plantspeciesandsoiltypecooperativelyshapethestructureandfunctionofmicrobialcommunitiesintherhizosphere.
FEMSMicrobiolEcol68(1):1–13.
9.
EdwardsJ,etal.
(2015)Structure,variation,andassemblyoftheroot-associatedmicrobiomesofrice.
ProcNatlAcadSciUSA112(8):E911–E920.
10.
Ofek-LalzarM,etal.
(2014)Nicheandhost-associatedfunctionalsignaturesoftherootsurfacemicrobiome.
NatCommun5:4950.
11.
BeckersB,etal.
(2016)Ligninengineeringinfield-grownpoplartreesaffectstheendospherebacterialmicrobiome.
ProcNatlAcadSciUSA113(8):2312–2317.
12.
RitpitakphongU,etal.
(2016)ThemicrobiomeoftheleafsurfaceofArabidopsisprotectsagainstafungalpathogen.
NewPhytol210(3):1033–1043.
13.
HaicharFZ,etal.
(2008)Planthosthabitatandrootexudatesshapesoilbacterialcommunitystructure.
ISMEJ2(12):1221–1230.
14.
OkumuraR,etal.
(2016)Lypd8promotesthesegregationofflagellatedmicrobiotaandcolonicepithelia.
Nature532(7597):117–121.
15.
MarkWelchJL,RossettiBJ,RiekenCW,DewhirstFE,BorisyGG(2016)Biogeographyofahumanoralmicrobiomeatthemicronscale.
ProcNatlAcadSciUSA113(6):E791–E800.
16.
ForslundK,etal.
;MetaHITconsortium(2015)Disentanglingtype2diabetesandmetformintreatmentsignaturesinthehumangutmicrobiota.
Nature528(7581):262–266.
17.
OhJ,etal.
;NISCComparativeSequencingProgram(2014)Biogeographyandin-dividualityshapefunctioninthehumanskinmetagenome.
Nature514(7520):59–64.
18.
YasudaK,etal.
(2015)Biogeographyoftheintestinalmucosalandlumenalmicro-biomeintherhesusmacaque.
CellHostMicrobe17(3):385–391.
19.
SunagawaS,etal.
;TaraOceanscoordinators(2015)Oceanplankton.
Structureandfunctionoftheglobaloceanmicrobiome.
Science348(6237):1261359.
20.
BulgarelliD,SchlaeppiK,SpaepenS,VerLorenvanThemaatE,Schulze-LefertP(2013)Structureandfunctionsofthebacterialmicrobiotaofplants.
AnnuRevPlantBiol64:807–838.
21.
BerendsenRL,PieterseCM,BakkerPA(2012)Therhizospheremicrobiomeandplanthealth.
TrendsPlantSci17(8):478–486.
22.
KubinakJL,etal.
(2015)MyD88signalinginTcellsdirectsIgA-mediatedcontrolofthemicrobiotatopromotehealth.
CellHostMicrobe17(2):153–163.
23.
O'ConnorRM,etal.
(2014)Gillbacteriaenableanoveldigestivestrategyinawood-feedingmollusk.
ProcNatlAcadSciUSA111(47):E5096–E5104.
24.
FaithJJ,McNultyNP,ReyFE,GordonJI(2011)Predictingahumangutmicrobiota'sresponsetodietingnotobioticmice.
Science333(6038):101–104.
25.
McNultyNP,etal.
(2011)Theimpactofaconsortiumoffermentedmilkstrainsonthegutmicrobiomeofgnotobioticmiceandmonozygotictwins.
SciTranslMed3(106):106ra106.
26.
BodenhausenN,Bortfeld-MillerM,AckermannM,VorholtJA(2014)Asyntheticcommunityapproachrevealsplantgenotypesaffectingthephyllospheremicrobiota.
PLoSGenet10(4):e1004283.
27.
PeifferJA,etal.
(2013)Diversityandheritabilityofthemaizerhizospheremicrobiomeunderfieldconditions.
ProcNatlAcadSciUSA110(16):6548–6553.
28.
AiraM,Gómez-BrandónM,LazcanoC,BthE,DomínguezJ(2010)Plantgenotypestronglymodifiesthestructureandgrowthofmaizerhizospheremicrobialcommu-nities.
SoilBiolBiochem42(12):2276–2281.
29.
BouffaudML,etal.
(2012)IsdiversificationhistoryofmaizeinfluencingselectionofsoilbacteriabyrootsMolEcol21(1):195–206.
30.
García-SalamancaA,etal.
(2013)Bacterialdiversityintherhizosphereofmaizeandthesurroundingcarbonate-richbulksoil.
MicrobBiotechnol6(1):36–44.
31.
GomesN,etal.
(2001)Bacterialdiversityoftherhizosphereofmaize(Zeamays)grownintropicalsoilstudiedbytemperaturegradientgelelectrophoresis.
PlantSoil232(1-2):167–180.
32.
SanguinH,etal.
(2006)Potentialofa16SrRNA-basedtaxonomicmicroarrayforanalyzingtherhizosphereeffectsofmaizeonAgrobacteriumspp.
andbacterialcommunities.
ApplEnvironMicrobiol72(6):4302–4312.
33.
OfekM,Voronov-GoldmanM,HadarY,MinzD(2014)Hostsignatureeffectonplantroot-associatedmicrobiomesrevealedthroughanalysesofresidentvs.
activecom-munities.
EnvironMicrobiol16(7):2157–2167.
34.
BaldwinTT,etal.
(2014)MaizeseedlingblightinducedbyFusariumverticillioides:AccumulationoffumonisinB1inleaveswithoutcolonizationoftheleaves.
JAgricFoodChem62(9):2118–2125.
35.
GlennAE,GoldSE,BaconCW(2002)Fdb1andFdb2,Fusariumverticillioideslocinecessaryfordetoxificationofpreformedantimicrobialsfromcorn.
MolPlantMicrobeInteract15(2):91–101.
36.
KommedahlT,WindelsCE(1981)Root-,Stalk-,andEar-InfectingFusariumSpeciesonCornintheUSA(ThePennsylvaniaStateUnivPress,UniversityPark,PA).
37.
Reinhold-HurekB,BüngerW,BurbanoCS,SabaleM,HurekT(2015)Rootsshapingtheirmicrobiome:Globalhotspotsformicrobialactivity.
AnnuRevPhytopathol53:403–424.
38.
SchlaeppiK,DombrowskiN,OterRG,VerLorenvanThemaatE,Schulze-LefertP(2014)QuantitativedivergenceofthebacterialrootmicrobiotainArabidopsisthalianarelatives.
ProcNatlAcadSciUSA111(2):585–592.
39.
BulgarelliD,etal.
(2015)Structureandfunctionofthebacterialrootmicrobiotainwildanddomesticatedbarley.
CellHostMicrobe17(3):392–403.
40.
GottelNR,etal.
(2011)DistinctmicrobialcommunitieswithintheendosphereandrhizosphereofPopulusdeltoidesrootsacrosscontrastingsoiltypes.
ApplEnvironMicrobiol77(17):5934–5944.
41.
BodenhausenN,HortonMW,BergelsonJ(2013)BacterialcommunitiesassociatedwiththeleavesandtherootsofArabidopsisthaliana.
PLoSOne8(2):e56329.
42.
GuttmanDS,McHardyAC,Schulze-LefertP(2014)Microbialgenome-enabledinsightsintoplant-microorganisminteractions.
NatRevGenet15(12):797–813.
43.
HacquardS,etal.
(2015)Microbiotaandhostnutritionacrossplantandanimalkingdoms.
CellHostMicrobe17(5):603–616.
E2458|www.
pnas.
org/cgi/doi/10.
1073/pnas.
1616148114Niuetal.
DownloadedbyguestonApril9,202144.
CaporasoJG,etal.
(2011)Globalpatternsof16SrRNAdiversityatadepthofmillionsofsequencespersample.
ProcNatlAcadSciUSA108(Suppl1):4516–4522.
45.
CaporasoJG,etal.
(2012)Ultra-high-throughputmicrobialcommunityanalysisontheIlluminaHiSeqandMiSeqplatforms.
ISMEJ6(8):1621–1624.
46.
McinroyJA,KloepperJW(1995)Surveyofindigenousbacterialendophytesfromcottonandsweetcorn.
PlantSoil173(2):337–342.
47.
ChenXH,etal.
(2007)Comparativeanalysisofthecompletegenomesequenceoftheplantgrowth-promotingbacteriumBacillusamyloliquefaciensFZB42.
NatBiotechnol25(9):1007–1014.
48.
NiuB,etal.
(2013)PolymyxinPistheactiveprincipleinsuppressingphytopathogenicErwiniaspp.
bythebiocontrolrhizobacteriumPaenibacilluspolymyxaM-1.
BMCMicrobiol13(1):137.
49.
KumarS,etal.
(2012)Bacteriophytochromecontrolscarotenoid-independentre-sponsetophotodynamicstressinanon-photosyntheticrhizobacterium,AzospirillumbrasilenseSp7.
SciRep2:872.
50.
CaporasoJG,etal.
(2010)QIIMEallowsanalysisofhigh-throughputcommunityse-quencingdata.
NatMethods7(5):335–336.
51.
TikhonovM,LeachRW,WingreenNS(2015)Interpreting16Smetagenomicdatawithoutclusteringtoachievesub-OTUresolution.
ISMEJ9(1):68–80.
52.
HintonDM,BaconCW(1995)Enterobactercloacaeisanendophyticsymbiontofcorn.
Mycopathologia129(2):117–125.
53.
RametteA(2009)Quantitativecommunityfingerprintingmethodsforestimatingtheabundanceofoperationaltaxonomicunitsinnaturalmicrobialcommunities.
ApplEnvironMicrobiol75(8):2495–2505.
54.
EickhorstT,TippktterR(2008)Improveddetectionofsoilmicroorganismsusingfluorescenceinsituhybridization(FISH)andcatalyzedreporterdeposition(CARD-FISH).
SoilBiolBiochem40(7):1883–1891.
55.
LundbergDS,YourstoneS,MieczkowskiP,JonesCD,DanglJL(2013)Practicalinno-vationsforhigh-throughputampliconsequencing.
NatMethods10(10):999–1002.
56.
BochnerBR(2009)Globalphenotypiccharacterizationofbacteria.
FEMSMicrobiolRev33(1):191–205.
57.
CaseTJ(1990)Invasionresistancearisesinstronglyinteractingspecies-richmodelcompetitioncommunities.
ProcNatlAcadSciUSA87(24):9610–9614.
58.
WeiZ,etal.
(2015)Trophicnetworkarchitectureofroot-associatedbacterialcom-munitiesdeterminespathogeninvasionandplanthealth.
NatCommun6:8413.
59.
vanElsasJD,etal.
(2012)Microbialdiversitydeterminestheinvasionofsoilbyabacterialpathogen.
ProcNatlAcadSciUSA109(4):1159–1164.
60.
KennedyTA,etal.
(2002)Biodiversityasabarriertoecologicalinvasion.
Nature417(6889):636–638.
61.
Inceoglu,Al-SoudWA,SallesJF,SemenovAV,vanElsasJD(2011)Comparativeanalysisofbacterialcommunitiesinapotatofieldasdeterminedbypyrosequencing.
PLoSOne6(8):e23321.
62.
HardoimPR,etal.
(2011)Riceroot-associatedbacteria:Insightsintocommunitystructuresacross10cultivars.
FEMSMicrobiolEcol77(1):154–164.
63.
MuellerUG,SachsJL(2015)Engineeringmicrobiomestoimproveplantandanimalhealth.
TrendsMicrobiol23(10):606–617.
64.
SwensonW,WilsonDS,EliasR(2000)Artificialecosystemselection.
ProcNatlAcadSciUSA97(16):9110–9114.
65.
Panke-BuisseK,PooleAC,GoodrichJK,LeyRE,Kao-KniffinJ(2015)Selectiononsoilmicrobiomesrevealsreproducibleimpactsonplantfunction.
ISMEJ9(4):980–989.
66.
ChauhanPS,ChaudhryV,MishraS,NautiyalCS(2011)Unculturedbacterialdiversityintropicalmaize(ZeamaysL.
)rhizosphere.
JBasicMicrobiol51(1):15–32.
67.
VorholtJA(2012)Microbiallifeinthephyllosphere.
NatRevMicrobiol10(12):828–840.
68.
DelmotteN,etal.
(2009)Communityproteogenomicsrevealsinsightsintothephysiologyofphyllospherebacteria.
ProcNatlAcadSciUSA106(38):16428–16433.
69.
MeeMT,CollinsJJ,ChurchGM,WangHH(2014)Syntrophicexchangeinsyntheticmicrobialcommunities.
ProcNatlAcadSciUSA111(20):E2149–E2156.
70.
KimHJ,BoedickerJQ,ChoiJW,IsmagilovRF(2008)Definedspatialstructurestabilizesasyntheticmultispeciesbacterialcommunity.
ProcNatlAcadSciUSA105(47):18188–18193.
71.
Cottee-JonesHEW,WhittakerRJ(2012)Perspective:Thekeystonespeciesconcept:Acriticalappraisal.
FrontBiogeogr4(3):117–127.
72.
SteeleJA,etal.
(2011)Marinebacterial,archaealandprotistanassociationnetworksrevealecologicallinkages.
ISMEJ5(9):1414–1425.
73.
ValePF,etal.
(2016)Beyondkilling:CanwefindnewwaystomanageinfectionEvolMedPublicHealth2016(1):148–157.
74.
XueC,etal.
(2015)ManipulatingthebananarhizospheremicrobiomeforbiologicalcontrolofPanamadisease.
SciRep5:11124.
75.
VentorinoV,etal.
(2016)Chestnutgreenwastecompostingforsustainableforestmanagement:Microbiotadynamicsandimpactonplantdiseasecontrol.
JEnvironManage166:168–177.
76.
LaneDJ(1991)16S/23SrRNAsequencing.
NucleicAcidTechniquesinBacterialSystematics,edsStackebrandtE,GoodfellowM(Wiley,NewYork),pp115–175.
77.
EdgarRC(2010)SearchandclusteringordersofmagnitudefasterthanBLAST.
Bioinformatics26(19):2460–2461.
78.
EdgarRC,HaasBJ,ClementeJC,QuinceC,KnightR(2011)UCHIMEimprovessensi-tivityandspeedofchimeradetection.
Bioinformatics27(16):2194–2200.
79.
CaporasoJG,etal.
(2010)PyNAST:Aflexibletoolforaligningsequencestoatemplatealignment.
Bioinformatics26(2):266–267.
80.
PriceMN,DehalPS,ArkinAP(2009)FastTree:Computinglargeminimumevolutiontreeswithprofilesinsteadofadistancematrix.
MolBiolEvol26(7):1641–1650.
81.
PaulsonJN,StineOC,BravoHC,PopM(2013)Differentialabundanceanalysisformicrobialmarker-genesurveys.
NatMethods10(12):1200–1202.
82.
PaulsonJN,PopM,BravoHC(2015)metagenomeSeq:Statisticalanalysisforsparsehigh-throughputsequencing.
Bioconductorpackage1.
10.
0.
Availableatcbcb.
umd.
edu/software/metagenomeSeq.
AccessedMay15,2016.
83.
vanDiepeningenAD,deVosOJ,ZelenevVV,SemenovAM,vanBruggenAH(2005)DGGEfragmentsoscillatewithorcountertofluctuationsincultivablebacteriaalongwheatroots.
MicrobEcol50(4):506–517.
84.
Buddrus-SchiemannK,SchmidM,SchreinerK,WelzlG,HartmannA(2010)RootcolonizationbyPseudomonassp.
DSMZ13134andimpactontheindigenousrhizo-spherebacterialcommunityofbarley.
MicrobEcol60(2):381–393.
85.
WilsonK(2001)PreparationofgenomicDNAfrombacteria.
CurrProtocMolBiol,10.
1002/0471142727.
mb0204s56.
86.
ChinC-S,etal.
(2013)Nonhybrid,finishedmicrobialgenomeassembliesfromlong-readSMRTsequencingdata.
NatMethods10(6):563–569.
87.
SherwoodR,HagedornDJ(1958)DeterminingCommonRootRotPotentialofPeaFields(AgriculturalExperimentStation,UniversityofWisconsin,Madison,WI).
88.
JovelJ,etal.
(2016)Characterizationofthegutmicrobiomeusing16Sorshotgunmetagenomics.
FrontMicrobiol7:459.
Niuetal.
PNAS|PublishedonlineMarch8,2017|E2459MICROBIOLOGYPNASPLUSDownloadedbyguestonApril9,2021