您的当前位置：首页 Analysis of Factors Affecting the Accuracy, Reproducibility, and Interpretation of Microbial

Analysis of Factors Affecting the Accuracy, Reproducibility, and Interpretation of Microbial

来源：筏尚旅游网

Analysis of factors affecting the accuracy,

reproducibility, and interpretation of microbial community carbon source utilization patterns.

S K Haack, H Garchow, M J Klug and L J ForneyAppl. Environ. Microbiol.

1995, 61(4):1458.

Updated information and services can be found at: http://aem.asm.org/content/61/4/1458These include:

CONTENT ALERTS

Receive: RSS Feeds, eTOCs, free email alerts (when new articlescite this article), more»

Information about commercial reprint orders: To subscribe to to another ASM Journal go to: http://journals.asm.org/site/misc/reprints.xhtmlhttp://journals.asm.org/site/subscriptions/

Downloaded from http://aem.asm.org/ on January 9, 2014 by guestAPPLIEDANDENVIRONMENTALMICROBIOLOGY,Apr.1995,p.1458–1468Vol.61,No.4

0099-2240/95/$04.00ϩ0

AnalysisofFactorsAffectingtheAccuracy,Reproducibility,and

InterpretationofMicrobialCommunityCarbon

SourceUtilizationPatterns†

SHERIDANKIDDHAACK,1*HELENGARCHOW,1,2MICHAELJ.KLUG,1,2,3ANDLARRYJ.FORNEY1,3CenterforMicrobialEcology,1DepartmentofMicrobiology,3andKelloggBiologicalStation,2MichiganStateUniversity,EastLansing,Michigan48824

Received18August1994/Accepted25January1995

Wedeterminedfactorsthataffectresponsesofbacterialisolatesandmodelbacterialcommunitiestothe95carbonsubstratesinBiologmicrotiterplates.Forisolatesandcommunitiesofthreetosixbacterialstrains,substrateoxidationratesweretypicallynonlinearandweredelayedbydilutionoftheinoculum.Wheninoculumdensitywascontrolled,patternsofpositiveandnegativeresponsesexhibitedbymicrobialcommu-nitiestoeachofthecarbonsourceswerereproducible.Ratesandextentsofsubstrateoxidationbythecommunitieswerealsoreproduciblebutwerenotsimplythesumofthoseexhibitedbycommunitymemberswhentestedseparately.Replicatesofthesamemodelcommunityclusteredwhenanalyzedbyprincipal-componentsanalysis(PCA),andmodelcommunitieswithdifferentcompositionswereclearlyseparatedontheﬁrstPCAaxis,whichaccountedfor>60%ofthedatasetvariation.PCAdiscriminationamongdifferentmodelcommunitiesdependedontheextenttowhichspeciﬁcsubstrateswereoxidized.However,thesubstratesinterpretedbyPCAtobemostsigniﬁcantindistinguishingthecommunitieschangedwithreadingtime,reﬂectingthenonlinearityofsubstrateoxidationrates.Althoughwhole-communitysubstrateutilizationpro-ﬁleswerereproduciblesignaturesforagivencommunity,theextentofoxidationofspeciﬁcsubstratesandthenumbersoractivitiesofmicroorganismsusingthosesubstratesinagivencommunitywerenotcorrelated.Replicatesoilsamplesvariedsigniﬁcantlyintherateandextentofoxidationofseventestedsubstrates,suggestingmicroscaleheterogeneityincompositionofthesoilmicrobialcommunity.

GarlandandMills(2)recentlyintroducedtheuseofcom-Tointerpretthefunctionaldiversityormetabolicpotentialmunity-levelcarbonsourceutilizationpatternsforcomparisonofmicrobialcommunitiesbasedonBiologpatternsofsub-ofmicrobialcommunitiesfromdifferenthabitats.Theyusedstrateutilization,itisimportanttounderstandfactorswhichcommerciallyavailablemicrotiterplatesthatcontain95carbonsubstrates(BiologGN;Biolog,Inc.,Hayward,Calif.),whichaffectsubstrateutilizationwhenBiologmicroplatesareinocu-theydirectlyinoculatedwithenvironmentalsamplesfromlatedwithmixturesofmicroorganisms.Inprinciple,theBiologfreshwater,saltwater,estuarine,andhydroponicsolutions,assayisdonebycolorimetricallymeasuringtetrazoliumdyefromtherhizosphereofhydroponicallygrownwheat,andfromreductionthatiscoupledtosubstrateoxidation.Thedegreetosoils.Similarly,Winding(5)analyzedthewhole-communitywhicheachofthe95substratesisoxidizedisdeterminedafterBiologGNsubstrateutilizationproﬁlesofsamplestakenfromaﬁxedincubationperiod.Apositiveresponseisidentiﬁedasseveraltypesofforestsoils,aswellasfromsizeclassfractionsanabsorbanceoropticaldensityvaluegreaterthanthatoccur-withinonesoiltype,andZaketal.(6)studiedtheproﬁlesofringintheblankwell.Ifthevaluesforallsubstrate-containingsoilsamplesfromsixdesertplantcommunitiesalonganeleva-wellsareanalyzedbymultivariatestatistics,twotypesofdis-tionalandmoisturegradient.Inallthreestudies,multivariatetinctionsbetweensamplescanbemade.First,thepresenceorstatisticalmethodswereusedtoestablishdistinctionsamongabsenceofapositiveresponsetoeachofthe95substratesortheresultingcommunitysubstrateutilizationproﬁlesandtogroupsofsubstrates(e.g.,polymers)canberecorded.Sec-therebyclassifythecommunitysampleswithrespecttotheirond,whenthesamesubstratesareutilizedbydifferentcom-carbonsourcemetabolism.GarlandandMills(2)proposedmunities,sampleswithconsistentlyhighvaluesforcertainsub-thatsinceseparationofcommunityBiologproﬁlesbymulti-stratescanbesegregatedbymultivariatestatisticalproceduresvariateanalysesisbasedondifferencesincarbonsourceutili-fromthosewithconsistentlylowervaluesforthesamesub-zationbetweensamples,itprovided‘‘afunctionalbasistostrates.Forexample,Zaketal.(6)distinguisheddesertsoildistinctionsamongcommunities.’’Allthreegroupssuggestedmicrobialcommunitiesbybothmethods.

thatinadditiontoestablishingecologicallyrelevant(e.g.,soilItistemptingtoassumethatdifferencesinpatternsofBi-versusfreshwater)classiﬁcationsofmicrobialcommunities,ologsubstrateoxidationbydifferentcommunitiesrepresentsubstrateutilizationproﬁlesmightofferinformationwithre-realdifferencesinthenumbersortypesofmicroorganismsingardtocommunityfunction(2),metabolicpotential(5),orthesecommunitiesandthatthesedifferencesreﬂecttheactiv-functionaldiversity(6).

itiesexpressedintheenvironmentsfromwhichthesampleswereobtained.However,GarlandandMills(2)andWinding(5)bothfoundthatvariationinresponsefromonesampleto*Correspondingauthor.Presentaddress:WaterResourcesDivi-anotherwasnotnecessarilyafunctionofstructuraldifferencession,U.S.GeologicalSurvey,6520MercantileWay,Suite5,Lansing,inthecommunities.Instead,bothnotedastrongcorrelationMI411-5991.

betweeninoculumcelldensityandtherateofcolordevelop-†Contribution786oftheKelloggBiologicalStation.

ment,sothatcommunitydifferencesinsubstrateoxidation

1458

Downloaded from http://aem.asm.org/ on January 9, 2014 by guestVOL.61,1995ANALYSISOFCOMMUNITYSUBSTRATEUTILIZATIONPROFILES1459

recordedataﬁxedtimemightsimplyreﬂectdifferencesinthetotalnumberofmicroorganismsinthecommunities.More-over,bothgroupsfoundthatcolordevelopmentappearedtodependoncellgrowthinsubstrate-containingwells,sothatitwasimpossibletodetermineifallmembersofthecommunitycapableofutilizingthecompoundhadcontributedtothepro-ﬁleoriftheresponseinagivenwellresultedfromgrowthandactivityofasubset,orsinglemember,ofthecommunity.Zaketal.(6)acknowledgedtheselimitationsbutarguedthatsub-strateutilizationproﬁlesnonethelessprovidea‘‘richdataset’’forstudiesofthefunctionanddiversityofmicrobialcommu-nities.However,GarlandandMills(2)suggestedthatfurtherresearchwasrequiredtodeterminewhethercommunity-levelcarbonsubstrateutilizationpatternswererepeatableandtodeterminethecausesofdifferentialsubstrateoxidationamongsamples.

TheresearchdescribedherewasconductedtodeterminefactorswhichinﬂuencesubstrateoxidationwhenBiologmicro-platesareinoculatedwithsamplesfrommicrobialcommuni-ties.Weusedsoilbacterialisolates,modelbacterialcommu-nitiescomposedofcombinationsofthesestrains,andactualsoilmicrobialcommunities.Wewishedtodetermine,forax-enicculturesandmodelcommunities,(i)whethertherewasarelationshipbetweengrowthinwellsduringmicroplateincu-bationandsubstrateoxidation,(ii)theinﬂuenceofinoculumdensityontherateofsubstrateoxidation,(iii)whethertherateandextentofoxidationofagivensubstratewerereproduciblewheninoculumdensitywascontrolled,and(iv)whetherthepatternofpositiveandnegativeresponsestoall95substratesinBiologmicroplateswasreproducible.Usingtwomodelsoilbacterialcommunitiesforwhichthetaxonomicstructureandcarbonsubstrateutilizationpatternsoftheindividualmemberswereknownapriori,weaddressedwhetherBiologproﬁlesofwholecommunitiesreﬂectedthetaxonomicstructureofthosecommunitiesandwhetherwecoulddetectcontributionsofindividualpopulationstothewhole-communityBiologproﬁle.Finally,weexaminedcommunity-levelproﬁlesofactualsoilmicrobialcommunitiesandinterpretedtheresultsofmulti-variatestatisticalanalysesofmodelandactualsoilmicrobialcommunityresponsesinlightofknowledgeobtainedfromtheworkwithisolates.Ourresultsverifytheutilityofthisap-proachfortheclassiﬁcationofmicrobialcommunitiesandprovideinformationonfactorswhichinﬂuencesubstrateoxi-dationandmustbetakenintoaccountwheninterpretingthefunctionalsigniﬁcanceofwhole-communitysubstrateutiliza-tionproﬁles.

MATERIALSANDMETHODS

Bacterialisolates.TenbacterialisolateswereobtainedfromsoilsoftheLongTermEcologicalResearchsiteattheKelloggBiologicalStation,MichiganStateUniversity.Theseisolateswereidentiﬁed,byphenotypiccharacteristicsandfattyacidproﬁles(3)astwoBacillusspp.,twoArthrobacterspp.,twoPseudomonasspp.,threeCytophagaspp.,andoneCorynebacteriumsp.AllcultureshadbeenisolatedonR2Aagar(Difco,Detroit,Mich.)andmaintainedonR2Aagarslantsintherefrigeratorwithregularsubculturing.These10strainswereroutinelyculturedat28ЊConR2Aagarorinthecorrespondingbrothmediumthatlackedagar(designatedR2Abroth).Brothcultureswereaeratedbyshaking.Standardcurvescorrelatingcellnumber(obtainedbyacridineorangedirectcounts)andbiomass(micrograms[dryweight]permilliliter)withopticaldensityat590nm(OD590)wereobtainedforeachstrainduringgrowthinR2Abrothat28ЊCandwereusedforestimationofcellnumberorbiomassinallexperimentswhereindicated.ViablecellcountsweredeterminedbyplatingonR2Aagar.

Biologsubstrates.ForallstudiesreportedhereinweusedtheBiolog,Inc.,gram-negative(GN),gram-positive(GP),ornosubstrate(MT)96-wellmicro-plates.Eachofthe95substratewellsinstandardBiologGNandGPmicroplatescontains(indriedform)acomplex,low-concentration,bufferednutrientmedium(1),atetrazoliumredoxdye,andacarbonsubstrate.Thesubstrateblankwellcontainsthenutrientmediumanddyebutnosubstrate.TheGNsubstratesweredividedbyGarlandandMills(2)into11categories:polymers(substrates1to5),carbohydrates(substrates6to33),esters(substrates34and35),carboxylicacids(substrates36to59),brominatedchemicals(substrate60),amides(substrates61to63),aminoacids(substratesto83),aromaticchemicals(substrates84to87),amines(substrates88to90),alcohols(substrates91and92),andphosphor-ylatedchemicals(substrates93to95).GPmicroplatescontain62ofthesamesubstratesasGNmicroplatesplus33differentsubstratesinsimilarcategories.MTmicroplatescontainthedriednutrientmediumandredoxdyebutnosubstrateinanywell.Stocksolutions(0.5%,wt/vol)ofglucose,galactose,L-asparagine,aspartate,cellobiose,mannitol,andL-serine(allfromSigmaChem-icalCo.,St.Louis,Mo.)werepreparedin50mMphosphatebuffer(pH7)andsterilizedbypassagethroughasterile0.22-␮m-pore-sizeﬁlter(MilliporeCorp.,Bedford,Mass.).StocksolutionsweredispensedtoMTmicroplatewellsin75-␮laliquots.Blankwellsreceived75␮lofbuffer.

Microplateprotocols.Pure-cultureinoculaformicroplatesweretypicallypre-paredinoneoftwoways.Culturesgrownonsolidmediumfor24hwereswabbedfromtheagarsurfaceandresuspendedin50mMphosphatebuffer(pH7).Culturesgrowninbrothmediumwerecentrifuged(10,000ϫg,for10to15min),thesupernatantwasremoved,andthecellswereresuspendedin50mMphos-phatebuffer(pH7)andheldfor1hbeforeinoculationofmicroplates.BiologGNorGPmicroplatewellswereinoculatedwith150␮lofthesecellsuspensionsthathadbeenadjustedtotheappropriatedensity(GN,ϳ3ϫ108cellsperml,GP,ϳ4.5ϫ108cellsperml)bycomparisonwiththeturbiditystandardssuppliedbythemanufacturer.MTplateswereinoculatedwithcellsuspensionshavinganOD590ofϳ0.45.Isolates,combinationsoftwoorthreeisolates,andbufferwereaddedtoMTmicroplatewellssuchthatthetotalvolumeofinoculumequalled75␮lbutthenumberofcellsofeachisolateineachwellwasequivalent.AllMTplatewellsalsocontained75␮lofsubstratesolutionorbuffer(blank).

InoculaformodelcommunitieswereobtainedbygrowingculturesoftheindividualbacterialstrainsinR2Abrothat28ЊCforapproximately16h.TheOD590after16hwascorrelatedwithcellcount(cellspermilliliter)orbiomass(micrograms[dryweight]permilliliter)fromstandardcurvespreviouslyestab-lishedforeachbacterialstrain.Culturevolumesrequiredtoachievespeciﬁcproportionsofbacterialstrainswerecombinedandcentrifugedtoobtainthe‘‘community’’cellpellet,whichwasresuspendedin50mMphosphatebuffer(pH7)for1hpriortoinoculationofGNmicroplates.Communitiesweredesignedsuchthattheﬁnalcelldensity(totalofallbacterialstrainsinthecommunity)wouldfallwithintherecommendedrangeforinoculationofBiologGNmicro-plates.

Threebacterialstrainswereusedtodetermineifsubstrateoxidationwasrelatedtocellgrowthfollowinginoculation.EightreplicatewellsofBiologMTplatesthatcontained0.25%(wt/vol)galactoseassubstratewereinoculatedwitheachbacterialstrain.Theinitialcelldensitywaseitherϳ108or107cellsperml.Colordevelopment(OD590)wasrecordedat1-hintervalsfor24handthenonceagainat48h.At1,3,6,9,19,25,and48h,oneoftheeightreplicatewellsforeachbacterialstrainorcombinationwassampledtodetermineviablecellcount.Microplateswereincubatedat28ЊCandwerereadat590nmineithertheBiologmicroplatereaderwithMicrolog2NsoftwareortheEL312eKineticsreader(Bio-TekInstruments,Inc.,Winooski,Vt.)withKineticalcIIsoftware.InthestandardMicrolog2Nsoftwaresuppliedbythemanufacturer,apositiveresultmustbeofhigherOD590thanthesubstrateblankreadingandmustbeatleast40%ofthehighestpositivesubstrateresponseforthatmicroplate.ThisthresholdwasusedthroughoutourstudyforassignmentofapositiveornegativeresponseforallGNorGPsubstrates.TheBio-TekEL312ereaderandKineti-calcIIsoftwarewereusedtoobtainkineticreadingsofcolordevelopmentinMTplates.Blankwellswereassignedtosectorsofeachmicroplateasappropriate,andthereadingat590nmofeachblankwellwassubtractedfromthereadingofeachsubstrate-containingwellinthatsector,ateachreadingtime.

EnvironmentaleffectsonisolateBiologGNproﬁles.Atleastsevenindepen-dentBiologGNproﬁleswereobtainedat24and72hforeachbacterialstrainfollowinggrowthinR2Abrothat28ЊC.Theresponsetoeachsubstrate(i.e.,alwayspositive,alwaysnegative,orvariableoverthesevenreplicates)wasde-termined.Thesereferenceproﬁleswereusedasthestandardagainstwhicheffectsofgrowthmedium,growthtemperature,andinoculumdensity,aswellastheresultsofourmodelcommunityexperiments,wouldbeassessed.BiologGNproﬁleswerethenobtainedforeachofthe10bacterialstrainsgrownunderthefollowingconditions:(i)onnutrientagar,trypticsoybrothagar(trypticsoybrothsolidiﬁedwith3%[wt/vol]agar)(bothfromSigmaChemicalCo.),orR2Aagarat28ЊCand(ii)inR2Abrothat17and32ЊC.Inthelattercase,Biologmicro-plateswereincubatedatthegrowthtemperature.Eachproﬁlewasdeterminedintriplicateandreadingswereobtainedat24and72h.Proﬁleswerealsoobtained(threereplicates;0-,4-,7-,11-,24-,and96-hreadings)foreachbacterialstrainafterinoculationtoGNmicroplatesat0.1timestherecommendedcelldensity.Forgram-positivespecies,BiologGPproﬁlesaftergrowthinR2AbrothandonR2Aagarwerealsoobtained.

Modelcommunities.Themodelsoilbacterialcommunitiesusedwereidenticaltothosepreviouslyreported(3).Modelbacterialcommunitieswereconstructedbycombiningknownproportionsof6(communityI-AandI-B)or4(communityII-AandII-B)ofthe10isolates.AandBdesignatereplicateexperimentsconductedinNovember1992(A)andApril1993(B).CommunityIconsistedofequalproportionsofthefollowingsixsoilbacteria:Corynebacteriumsp.strain5,Pseudomonassp.strain14,Cytophagasp.strains19and72,Arthrobactersp.strain42,andBacillussp.strain81.CommunityIIconsistedofBacillussp.strain94

Downloaded from http://aem.asm.org/ on January 9, 2014 by guest1460HAACKETAL.APPL.ENVIRON.MICROBIOL.

FIG.1.Numberofpositiveresponsesat24and72h,asmeasuredbyMicrolog2Nsoftware,foreachofthe95BiologGNsubstratesafterinoculationofsevenreplicatemicroplateswithCytophagasp.strain72orPseudomonassp.strain31.

(whichmadeuphalfofthecommunityonacellnumberorbiomassbasis)andCorynebacteriumsp.strain5,Pseudomonassp.strain31,andCytophagasp.strainstrainsandwassimilarfororganismsofthesamegenus.Biolog47,eachmakingupone-sixthofthecommunity.

GNproﬁlesforsevenreplicatemicroplatesofCytophagasp.Soilcommunities.Soilwasobtainedfromagreenhouseplantingsofcorn,atstrain72andPseudomonassp.strain31at24and72haretheterminationofanexperimentconductedbyHalCollinsattheKelloggdepictedinFig.1.Forbothstrains,somesubstrateswereBiologicalStation,MichiganStateUniversity.SoilwasoriginallyobtainedfromasinglelocationattheKelloggBiologicalStationLongTermEcologicalRe-alwaysutilizedandwerescoredpositiveinsevenofsevensearchSite.Itwashomogenized,dried,andplacedin10-gal(38-liter)plasticreplicatesat24h,whileothersubstrateswerescorednegativepots.Waterwasaddedtoﬁeldcapacity,andcornwasplantedineachpot2weeksineveryreplicate,evenafter72hofincubation.

later.Ineachpot,asinglecornplantwasallowedtoreachphysiologicalmaturity,ThethreeCytophagaspp.exhibitedproﬁlessimilartothatofatwhichtimesoilsamples(100g)weretakenfromthreelocationswithintherootzoneandthreelocationswithinthebulksoilfromeachofsixpots(36samples).strain72(Fig.1);eachconsistentlyusedrelativelyfew(18toEachofthe361-gsamplesofsoilwassuspendedin5mlof50mMphosphate25)ofthe95substratesonGNmicroplatesandmetabolizedbuffer(pH7)andagitatedfor15min.ApproximatelyonegranuleofCaClprimarilycarbohydrates(substrates6to33),aminoacids(sub-equivalentamountofMgCO2andan3wereaddedtoeachsuspension,andthere-stratesto83),andpolymers(substrates1to5).Thetwosultingﬂocwasallowedtosettle.Thisleftaslightlyturbidsupernatant,whichwasdirectlyinoculatedontoBiologMTmicroplates.TheODPseudomonasstrainsmetabolizedprimarilycarboxylicacidswereallowed590valuesoftheinoc-ulatedmicroplateswereread,andthentheplatestositatroom(substrates36to59),aminoacids,andvariouscarbohydrates.temperatureuntilϳ20h,whencolordevelopmentbegantobevisible.AtthisNeitherusedthecarbohydratepolymers(␣-cyclodextrin,dex-time,themicroplateswereplacedintheBio-Tekreader,andtheODtrin,andglycogen)usedbyallthreeCytophagaspp.Coryne-2-hintervalsforthenext24h.Theabilityofeachcellsuspension590wasdeterminedattoutilizeasparagine,aspartate,glucose,cellobiose,carboxymethylcellulose,N-bacteriumstrain5used21ofthesubstratesonGNmicroplatesacetylglucosamine,andferulicacid(SigmaChemicalCo.)wasdeterminedwith(Tween40and80,sevencarbohydrates,fourcarboxylicacids,BiologMTplates.

bromosuccinate,succinamicacid,twoaminoacids,twoesters,PCA.Principal-componentsanalysis(PCA)wasperformedwithSAS/STAT2,3-butanediol,andglycerol).

(StatisticalAnalysisSystem;SASInstituteInc.,Cary,N.C.)version6.03underthePRINCOMPprocedure.DataforeachreplicateofeachcommunityintheTheBacillusandArthrobacterspp.failedtoconsistentlyyieldmodelcommunityexperimentswereenteredasthepercentchangefromthepositiveresponsesonBiologGN(orGP)microplates.After24substrateblankvalueprovidedbytheBiologsoftware.Dataforexperimentswithor72h,microplatesinoculatedwiththeBacillusstrainswouldMTplateswereenteredastheblank-adjustedODtypicallyexhibitonlyonetothreeveryweakpositiveresponsessoftware.

590readingsobtainedfromtheBio-Tek(i.e.,OD590onlyslightlyabovethatofthesubstrateblank).ThesubstratewellsidentiﬁedaspositivewerenotconsistentRESULTS

fromassaytoassay,andthispatternwasoftenaccompaniedbycolorinthesubstrateblankwell.SlightlydifferentresultswereBiologGNproﬁlesforisolates.ThepatternofpositiveandobtainedforthetwoArthrobacterspecies.Thesestrainsoftennegativeresponsestothe95substratesinBiologGNmicro-producedcolorinthesubstrateblankwellandoccasionallyplateswasshowntobecharacteristicofandgenerallyrepro-exhibitedaproﬁlewithclearpositiveandnegativeresponses.ducibleforCytophaga,Pseudomonas,andCorynebacterium

Unfortunately,theseproﬁleswereinconsistentfromassayto

Downloaded from http://aem.asm.org/ on January 9, 2014 by guestVOL.61,1995ANALYSISOFCOMMUNITYSUBSTRATEUTILIZATIONPROFILES1461

FIG.2.Colordevelopmentproﬁlesafterinoculationattwodifferentinitialcelldensities(3ϫ108cellsperml[AandC]and3ϫ107cellsperml[BandD])forCytophagasp.strain72andPseudomonassp.strain31,displayingpositive,negative,andvariablereadingsonselectedBiologGNsubstrates.

assayandevenwithinreplicatemicroplatesinoculatedwiththenegative,andvariableresponsescanbeseeninFig.2,whichsameculture,sothatareferenceproﬁleforthesetwostrainsshowsthetimecourseofcolordevelopment(ODcouldnotbeobtained.

590)onse-lectedsubstratesforeitherCytophagasp.strain72orPseudo-ThelackofsubstrateutilizationbyBacillusstrainswasnotmonassp.strain31followinginoculationattwodifferentinitialduetotetrazoliumdyetoxicity(bothspecieswouldgrowindensities.SincetherateofsubstrateoxidationwasnotalwaysR2Abrothwithtetrazoliumdyeadded)ortosporeformationlinearthroughoutthetestperiod,differentscorescouldbepriortoinoculationofmicroplates.Nogrowthmediumorobtaineddependingonwhenthemicroplatewasread.FortemperaturetestedresultedinaconsistentBiologresponseforexample,wellscouldbescorednegativeatearlyreadingsandeitherBacillusstrain.Themanufacturersuggeststhatoligotro-positivelater(e.g.,fructoseforPseudomonassp.strain31)orphicmicroorganismsmaygiveanall-negativeresponseinBi-positiveatearlyreadingsbutnegativelater(e.g.,compare10-ologmicroplatesandthatcolorinthesubstrateblankwelland96-hvaluesforrhamnoseforCytophagasp.strain72).Incouldbeduetosporeformationinthewells,celllysis,oraddition,therateandextentofcolordevelopmentdependedutilizationofendogenousorextracellularstoragepolymers.Tooninoculumdensity(Fig.2BandD).Therefore,assignmentofcheckourtechnique,wegrewBacilluscereusATCC14579byapositiveornegativeresponseat24hforweaklyoxidizedrecommendedmethodsandinoculatedittoBiologGP(andsubstrateswasinconsistentfromonereplicatemicroplatetoGN)microplates.Incontrasttooursoilstrains,thisstrainanother.Incontrasttovariablesubstrates,substrateswhichproducedclearlypositiveresultsafter24handwasidentiﬁedwerealwaysscoredpositiveforaparticularstrain(forexample,bytheBiologsoftwareasBacillusthuringiensis/cereus(similar-dextrinforCytophagasp.strain72andasparagineandorni-ityindex,0.51).ThemanufactureralsosuggeststhatparticularthineforPseudomonassp.strain31[Fig.2])wereoxidizedgrowthmediamayberequiredtoelictaBiologresponsefromrapidlyandtoagreaterextent.

bacterialgenera.Ourexperimentalobjectivesrequiredidenti-Additionalexperimentsweredonetodeterminetheeffectofcalgrowthconditionsforeachstrainmakingupourmodelinoculumdensityandcellgrowthinmicroplatewellsonsub-communities;therefore,wemadenoefforttoﬁndalternativestrateoxidationkinetics.PreviousresultsshowedthatCyto-growthconditionsthatwouldresultinconsistentBiologre-phagasp.strain72alwaysexhibitedapositiveresponseonsponsesfortheBacillusandArthrobacterisolates.Webelievegalactose,thatPseudomonassp.strain31alwaysexhibitedathatourapproachreﬂectswhatoccurswhennaturalmicrobialvariableresponse,andCorynebacteriumsp.strain5wouldnotcommunitiesareextractedfromsoilandinoculatedtoBiologoxidizegalactose.Weexamineddetailsofgalactoseoxidationplates,sinceinsituconditionsareunlikelytobeoptimalforkineticsbythesethreestrainsandmeasuredcellgrowthofstimulatingaBiologresponsebyeverycommunitymember.eachstrainfollowinginoculationtoBiologmicroplatewellsatKineticsofsubstrateoxidationbyisolates.Thevariabletwodifferentinoculumdensities(Fig.3).Cytophagasp.strainresponsesofCytophaga,Pseudomonas,andCorynebacterium72exhibitedarapidandextensiveoxidationofgalactose(Fig.strainswereshowntoresultfromweak,nonlinearratesof3AandB)andeithermaintainedviablecellcountat108cellssubstrateoxidationcombinedwithadependenceofsubstrateperml(Fig.3A)orgrewto108cellsperml(Fig.3B)andthenoxidationrateoninoculumcelldensity.Thebasisforpositive,maintainedtheircellcountatthisdensity.Pseudomonassp.

Downloaded from http://aem.asm.org/ on January 9, 2014 by guest1462HAACKETAL.APPL.ENVIRON.MICROBIOL.

FIG.3.Colordevelopmentproﬁles(OD590;opensymbols)andcelldensity(CFUpermilliliteronR2Aagar;solidsymbols)inwellsofBiologMTmicroplatescontaining0.25%(wt/vol)galactose,followinginoculationwiththreedifferentbacterialstrainseachattwodifferentinitialconcentrations.(A,C,andE)Initialinoculumdensity,ϳ108CFU/ml;(B,D,andF)initialinoculumdensity,ϳ107CFU/ml.

strain31exhibitedamoremoderaterateofgalactoseoxidationEnvironmentaleffectsonBiologGNproﬁlesofisolates.yetgaveasimilarpatternofviablecellcountstothatofCyto-Environmentalconditions(growthtemperature,growthme-phagasp.strain72.ForbothCytophagasp.strain72anddium,solidversusliquidculture)hadverylittleaffectonsub-Pseudomonassp.strain31,cellcountsintheblankwellwerestrateutilizationproﬁlesbyisolates.Wenotedonlyafewmaintainedatorincreasedto108CFU/mlwithoutsigniﬁcantexceptions.Pseudomonassp.strain31didnotuseformicacidcolordevelopment(datanotshown).ThepredominanteffectaftergrowthonR2Aagarornutrientagar,eventhoughthisofinoculumdilutionforCytophagasp.strain72andPseudo-substratewasalwaysusedaftergrowthinR2Abroth.Cyto-monassp.strain31wasalagintheinitiationofcolordevel-phagasp.strain72usedﬁveadditionalsubstratesaftergrowthopment.Corynebacteriumsp.strain5didnotoxidizegalactose,onnutrientagarortrypicsoybrothagarandusedtwoaddi-andtheviablecellcountdecreasedthroughoutthecourseoftionalsubstratesaftergrowthonR2Aagar,ascomparedwiththeexperiment(Fig.3EandF).Therefore,whilecellgrowthwhenitwasgrowninR2Abroth.Theonlyobservedtemper-doesoccurinBiologmicroplatewells,theamountofgrowthisatureeffectwasthatPseudomonassp.strain31occasionallynotcorrelatedtotherateorextentofsubstrateoxidation.Theusedformicacidaftergrowthat17ЊCbutneveratothertem-rateandextentofoxidationofparticularsubstratesforisolatesperatures.

inoculatedatsimilardensitieswerereproducible,however.GNBiologproﬁlesofmodelcommunities.Substrateutiliza-

Downloaded from http://aem.asm.org/ on January 9, 2014 by guestVOL.61,1995ANALYSISOFCOMMUNITYSUBSTRATEUTILIZATIONPROFILES1463

FIG.4.Numberofpositiveresponsesat72h,asmeasuredbyMicrolog2Nsoftware,foreachofthe95BiologGNsubstratesafterinoculationofreplicatemicroplateswithmodelbacterialcommunitiescomposedofsixsoilbacterialisolates(communityI)orfoursoilbacterialisolates(communityII).AandBdesignatetwoindependentconstructionsoftheidenticalmodelcommunity.

tionproﬁlesofwholemicrobialcommunitieswererepeatablethreePCaxesaccountedfor97%ofthevariationinthedata.anduniqueforeachcommunity.Forthesetests,weusedTheﬁrstPCaxiswasmosthighlycorrelatedwithoxidationofmodelbacterialcommunitiescomposedofsix(communityI)m-inositol,cis-aconiticacid,citricacid,D-gluconicacid,␣-ke-orfour(communityII)strains.WespeciﬁcallyincludedBacil-toglutaricacid,D-saccharicacid,L-asparagine,andhydroxylusandArthrobacterstrainsineachcommunity,toseeiftheseL-proline(allϳ0.26).Therewerenoequivalentnegativeload-isolateswouldaffectcommunityproﬁlesinanyway,eveningsforPCaxis1.Again,theseweresubstrateswiththehighestthoughwhentestedasaxenicculturestheyfailedtooxidizeanypercentchangevaluesincommunityII-AandII-Bandrela-

substrates.Figure4depictsthe72-hBiologGNproﬁleforeachcommunityfromtwoindependentexperiments.PCAclearlyindicatedthedegreetowhichproﬁlesforeachcommunitywereuniqueandreproducible.PCAscoresobtainedafteranalysisofthecovariancematrixof72-hOD590valuesaredepictedinFig.5.PCAbasedonthecovariancematrixdistin-guishedthetwomodelcommunities,andclusteredreplicatesofeachcommunityfrombothexperiments,ontheﬁrstPCaxis(Fig.5).TheﬁrstcovariancePCaxisaccountedfor65.4%,thesecondaccountedfor27.3%andthethirdaccountedfor3.9%ofthevariationinthedata,foracummulativetotalof96.6%.PCaxis1hadhighestloadingsonm-inositol,D-rafﬁnose,␣-ke-toglutaricacid,hydroxyL-proline,L-ornithine,andurocanicacid(allϳ0.20)andonlactulose(Ϫ0.21).Theseweresub-stratesforwhichtherewereconsistentdifferencesinthede-greeofsubstrateoxidation(i.e.,ODtwocommunities.CommunityIIreplicates590values)betweentheexhibitedmuchhigher72-hOD590valuesonm-inositol,D-rafﬁnose,␣-ketogl-utaricacid,hydroxyL-proline,L-ornithine,andurocanicacidandmuchlowervaluesonlactulosethandidcommunityIFIG.5.ResultsofPCAof72-hODinoculationwithmodel590readingstakenfromBiologGNmicroplatesfollowingbacterialcommunitiescomposedreplicatesineitherexperiment.

ofsixsoilbacterialisolates(communityI)orfoursoilbacterialisolates(com-Proﬁlesobtainedforeachcommunityatearliertimes(24h)munityII).AandBdesignatetwoindependentconstructionsoftheidenticalwerealsouniqueandreproducibleforeachcommunity.Inthismodelcommunity.PC1,PC2,PC3,ﬁrstthreeprincipal-componentaxes.Sym-case,PCaxis1accountedfor%ofthevariationandtheﬁrst

bols:U,communityI-A;E,communityI-B;P,communityII-A;F,communityII-B.

Downloaded from http://aem.asm.org/ on January 9, 2014 by guest14HAACKETAL.tivelylowvaluesinCommunityI-AandI-Bat24h,andthesubstratesincludesomeofthoseimportantintheanalysisofthe72-hdata.

EffectofcommunitycompositiononBiologGNproﬁles.Positiveresponsestosubstratesincommunityproﬁlesgener-allyreﬂectedthepresenceofspeciﬁcstrainscapableofusingthosesubstrates.BiologGNproﬁlesofaxenicallygrownstrainswereusedtoidentify‘‘signaturecompounds’’thatindicatedthepresenceofparticularstrainsinourmodelcommunityproﬁles.Forexample,allthe‘‘signature’’substratesforPseudomonassp.strain14or31(D-alanine,alaninamide,D-glucosaminicacid,L-histidine,D-galactonicacidlactone,mal-onicacid,orhydroxyL-proline)werepositiveafter72hofincubation.Similarly,thecarbohydratepolymersandglycyl-L-asparticacidwerepositiveinbothcommunitiesat72h,indi-catingthepresenceofCytophagasp.strains19and72incommunityI-AandI-BanduniquelyindicatingthepresenceofCytophagasp.strain47incommunityII-AandII-B.Inaddition,␤-methylglucoside(Cytophagasp.strain72)andD-rafﬁnose(Cytophagasp.strain47)uniquelyindicatedthepresenceofthesestrainsintheirrespectivecommunities.Onlyoneunexplainedpositivereadingwasfound.D-Rafﬁnose(sub-strate27)wasscoredasapositiveresponseforeveryreplicateofcommunityIat72h,eventhoughnoneofthecommunitymembersusedthissubstratewhentestedalone.

Negativeresponsesatlaterreadingtimesgenerallyindicatedtheabsenceofstrainscapableofusingagivensubstrate.TheonlyexceptionwasthatofthethreesubstratesindicativeofthepresenceofCorynebacteriumsp.strain5ineachcommunity,xylitol(substrate33,Fig.4)wasnegativeineverycasefollow-ing72hofincubation.Atearlierreadingtimes,false-negativeresultsweremorefrequentlynoted,sincesignaturesubstrateswerenotnecessarilypositiveafter24hofincubation.Forexample,2,3-butanediol(substrate91,Fig.4)andthymidine(substrate87),indicativeofCorynebacteriumsp.strain5,werenegativeat24hforeachcommunity.ThepresenceofAr-throbactersp.strain42orBacillussp.strain81incommunityIorofBacillussp.strain94,whichmadeuphalfthetotalcellpopulationofcommunityII,appearedtohavenoeffectoncommunityproﬁles.

Effectofcommunitycompositionondegreeofsubstrateoxidation.SincePCAgroupedourcommunitiesonthebasisofresponseswhichwereconsistentlyhigherinonecommunitythantheother,weexaminedtheunderlyingreasonforrela-tivelyhighresponses,askingwhethertheymightreﬂectthepresenceofseveralmicroorganismsusingthesamesubstrateorasinglehighlyactivestrain.Wefoundnocorrelationbe-tweenthenumberofstrainsabletoutilizeasubstrateandthedegreeofsubstrateoxidation.Forexample,whilethereweretwoCytophagaspp.incommunityIbutonlyoneCytophagasp.(strain47)incommunityII,therewasnopatterntothere-sponsesto␣-cyclodextrin,dextrin,andglycogenthatwouldsuggestgreaterutilizationincommunityIthanincommunityII.Similarly,inbothcommunityI(Corynebacteriumsp.strain5,Pseudomonassp.strain14,andCytophagasp.strain19)andcommunityII(Corynebacteriumsp.strain5,Pseudomonassp.strain31,andCytophagasp.strain47),therewerethreestrainsabletooxidizem-inositolandyetcommunityIconsistentlyexhibitedlowerresponsestom-inositolthancommunityII(somuchsothatm-inositolcontributedsigniﬁcantlytothePCA).Kineticsofsubstrateoxidationbycommunities.Althoughtherateandextentofsubstrateoxidationbycommunitieswerereproduciblecommunityparameters,theywerenotsimplythesumofthoseexhibitedbytheisolatesthatformedthecom-munity.Communitiescomposedofthreestrains(Corynebacte-riumsp.strain5,Cytophagasp.strain72,andPseudomonassp.

APPL.ENVIRON.MICROBIOL.

strain31)wereconstructed,andtherateofgalactoseoxidationbythecommunity,aswellastheviablecellcountforeachcommunitymemberateighttimesduringincubation,wasde-termined(Fig.6AandB).TherateofgalactoseoxidationforthecommunityofthreeisolateswasintermediatebetweentheproﬁlesofCytophagasp.strain72andPseudomonassp.strain31(Fig.6AandB).Neitherofthesestrainsexhibitedanypatterninviablecountswhichwouldexplaintheproﬁleofthecommunity(Fig.6B).Corynebacteriumsp.strain5,inoculatedatadensityof6ϫ107CFU/ml,couldnotbedetectedinthewells(detectionlimit,Ն104CFU/ml)atanysubsequenttimeintheexperiment.Themostpronouncedeffectofinoculumden-sitywasalagincolordevelopment(datanotshown).Inex-perimentsinwhicheachofthethreecommunitymemberswaspairedwithoneothermember(datanotshown),Corynebac-teriumsp.strain5hadnoeffectontherateofsubstrateoxi-dationexhibitedbyeitherCytophagasp.strain72orPseudo-monassp.strain31.However,whenCytophagasp.strain72waspairedwithPseudomonassp.strain31,theresultingproﬁlewasintermediatebetweentheproﬁlesofthetwoindividualstrainsandappearedsimilartothecommunityproﬁledepictedinFig.6B.Thisresultconﬁrmedthoseobtainedwithourmodelcommunitiesinthatthedegreeofsubstrateoxidationexhibitedbyacommunitydoesnotreﬂectthenumberofor-ganismsinthecommunitythatareabletoutilizethatsub-strate.Similarexperimentsweredonewithadditionaltestsub-strates(Fig.6CtoF).Colordevelopmentproﬁlesforbothisolatesandcombinationsweresubstratespeciﬁcandhighlyreproducible(Fig.6).

Soilcommunities.Onthebasisoftheresultsobtainedwithmodelcommunities,weanticipatedthatwhole-communitysubstrateutilizationpatternscouldbeusedtodistinguishbe-tweenrhizosphereandbulksoilcommunities.Inexperimentswithsoilmicrobialcommunitysamplestakenfromgreenhouseplantings,wewereunabletogroupsuchcommunitiesonthebasisofthepresenceorabsenceofapositiveresponsetospeciﬁcsubstratesortheirkineticsofsubstrateoxidation.Thelackofsimilarityamongsamplereplicatessuggestedmi-croscalespatialheterogeneityinthesesoils.Figure7depictstheseresultsandshowsthat(i)therewasnopatternofsub-strateutilizationorkineticproﬁledistinguishingrhizospherefrombulksoil,eitherwithinorbetweenpots,and(ii)therewasnoreproduciblekineticproﬁle,especiallyoncellobioseorN-acetylglucosamine,withinrhizosphereorbulksoilsamplesfromthesamepot.Patternswereequallyvariableforallothersamples.Althoughwedidnotdeterminetheinoculumdensityforeachsoilsuspension,onetestedrhizospheresample(pot1,Rhiz1,Fig.7)contained7.5ϫ105CFU/mlatthetimeofinoculationϫ108andexhibitedvaluesrangingfrom1.25ϫ108to9CFU/mlineverywell,includingtheblank,atthetimecolordevelopmentbegan(20hafterinoculation).Therewasnosigniﬁcantcolordevelopmentintheblankwellassociatedwiththisincreaseincelldensity.Allsamplesinitiatedcolordevelopmentataboutthesametime(Fig.7),whichsuggestsequivalentinocula.Readings(OD590)ofall36samplesat30h(approximatemidpointofcolordevelopmentproﬁle)werean-alyzedbyPCA,butnoseparationofsampleswithregardtoorigin(potnumber)ortypeofsample(rhizosphereversusbulksoil)waspossible.

DISCUSSION

Community-levelcarbonsourceutilization(Biolog)proﬁleshaverecentlybeenintroducedasameansofclassifyingmicro-bialcommunitiesonthebasisofheterotrophicmetabolism(2).Suchaclassiﬁcationsystemmightallowmicrobialecologiststo

Downloaded from http://aem.asm.org/ on January 9, 2014 by guestVOL.61,1995ANALYSISOFCOMMUNITYSUBSTRATEUTILIZATIONPROFILES1465

FIG.6.Triplicatecolordevelopmentproﬁles(OD590)exhibitedbythreebacterialstrains(72,Cytophagasp.strain72;31,Pseudomonassp.strain31;5,Corynebacteriumsp.strain5)inoculatedindependently(A,C,andE)orasacombinationofallthreestrains(B,D,andF)towellsofBiologMTmicroplatescontaining0.25%(wt/vol)galactose(AandB),asparagine(CandD),orcellobiose(EandF).PanelB(opensymbols)alsoindicatesthecelldensityofCytophagasp.strain72andPseudomonassp.strain31atselectedintervalsduringoneexperiment.Corynebacteriumsp.strain5,inoculatedatacelldensityofϳ6ϫ107cellsperml,wasnotdetectedatanyoftheindicatedintervalsduringthesameexperiment.

comparethemetabolicrolesofmicrobialcommunitiesfromsimplecommunities.BothGarlandandMills(2)andWindingdifferentenvironmentswithoutinvolvingtediousisolationand(5)notedacorrelationbetweeninoculumcelldensityandtheidentiﬁcationofcommunitymembers.Thesimplicityoftherateofcolordevelopmentanddealtwiththisproblemindif-methodandthecommercialavailabilityofBiologplatesareferentways.Winding(5)andZaketal.(6)recordeddataafterparticularlyattractive.However,notestsofthereproducibility3daysandafter60hofmicroplateincubation,respectively,ofcommunityBiologproﬁlesoroftheiraccuracyindepictingandneithermadeadjustmentsfordifferentinoculumcellden-communitystructureoractivityhavebeenreported.Ouruseofsities.GarlandandMills(2)adjustedforthisphenomenonbymodelsoilbacterialcommunitiesforwhichthetaxonomiccalculatinganaveragewellcolordevelopmentvalueforeachstructureandexpectedcarbonsubstrateutilizationpatternmicroplate.Thisvaluewasobtainedbycalculatingtherawwereknownaprioriallowedustodeterminethestrengthsanddifferencebetweentheopticaldensityineachwellandthatinweaknessesofthisapproachtomicrobialcommunityclassiﬁ-thecontrolwellandthensummingallthesevaluesanddividingcation.

by95.EveryvalueforeachsubstratewasthendividedbytheOurexperimentsconﬁrmthatmore-diluteinocularesultinaveragewellcolordevelopmentforthatmicroplatetonormal-protractedratesofcolordevelopmentforbothisolatesandizecomparisonsbetweensampleswithdifferentinoculumden-

Downloaded from http://aem.asm.org/ on January 9, 2014 by guest1466HAACKETAL.APPL.ENVIRON.MICROBIOL.

FIG.7.Colordevelopmentproﬁles(ODsamples(RHIZ1toRHIZ3)andbulksoil590)forsoilsamplestakenfromtherhizosphereandbulksoilofgreenhouse-maintainedplantingsofcorn.Rhizospheresamples(BULK1toBULK3)fromtwodifferentpots(POT1orPOT2)wereinoculatedtoaBiologMTmicroplatecontainingsevencarbonsubstrates.ASP,aspartate;ASN,asparagine;FER,ferulicacid;GLU,glucose;CEL,cellobiose;CMC,carboxymethylcellulose;NAG,N-acetylglucosamine.Blankwellscontainedonlybuffer.Absorbancereadingswereinitiatedat20hpostinoculationandweretakenat2-hintervalsfor24h.

sities.OurresultssupporttheconclusionofGarlandandMillswereused(Fig.3),thisgrowthwasnotcorrelatedwiththerate(2)thatoptimaluseofthisapproachrequiressamplesofap-orextentofsubstrateoxidation.Inourexperiments,growthproximatelyequivalentinoculumdensities.Ifthiscriterioniswithoutattendantcolordevelopmentoccurredinblankwellsnotmet,communitycomparisonsbasedoneither(i)positiveforisolates,modelcommunities,andsoilsamples,andisolatesandnegativeresponsestothe95Biologsubstratesor(ii)thedisplayedequivalentgrowthbutdifferentcolordevelopmentdegreeofsubstrateoxidationofparticularsubstrateswillbeproﬁlesonthesamesubstrates.Insimplemodelcommunities,compromisedbytheslowresponseofdiluteinocula.Forex-substrateoxidationproﬁleswerenotdirectlycorrelatedwithample,Zaketal.(6)comparedthetotalnumberofsubstratesthegrowthofisolateswhichusedthesubstrate(Fig.6).

usedandthe‘‘activitylevels’’(summationofabsorbanceval-Althoughsubstrateoxidationproﬁlesarenotadirectrep-ues)forsixdifferentdesertsoilmicrobialcommunitiesat12-hresentationofbacterialgrowth,wenotethatBiologsubstrateintervalsovera72-hincubation.Onegroupofthreecommu-oxidationresponsesforbothisolatesandcommunitiesoftennitiesconsistentlyusedmoresubstratesandshowedhigherexhibitalagphase,anexponentialphase,andastationaryactivitylevelsthantheremainingthreecommunitiesateveryphase,asdobacterialgrowthcurves.Thisnonlinearityhasreadingtime,stronglysuggestinganinoculumdensityeffect.importantimplicationsfortheinterpretationofsingleﬁxed-AlthoughtheseauthorsnotedsomecommunitydifferencesintimereadingsofcommunityBiologresponses,asGarlandandsubstrateutilizationamongthethree‘‘active’’communities,Mills(2)ﬁrstnoted.Substratesextensivelyusedbyasingletheremaining(less-active)communitiesweregroupedsimi-communityafter,forexample,a72-hincubation,maynotbelarlybyeverystatisticalapproach.Unfortunately,Zaketal.(6)thesameasthoseusedmostextensivelyatpriorreadingtimes.didnotreportinoculumdensityfortheirsamples,sothesig-Forexample,inPCAofourmodelcommunities,therewereniﬁcanceoftheirﬁndingsisnotclear.

onlythreesubstrateswithhighcorrelationsontheﬁrstPCABothGarlandandMills(2)andWinding(5)notedthataxisatboth24and72h.Severalothersubstratesweresignif-growthoccurredinBiologmicroplatewells,andbothsug-icanttothePCAatonlyoneofthereadingtimes;conse-gestedthattheBiologcommunitysubstrateutilizationassayquently,thesubstratesinterpretedbymultivariateanalysistoreﬂectsthegrowthofaproportionofthecommunitywhichisbemostsigniﬁcantindistinguishingthecommunitieschangedcapableofutilizingBiologsubstrates.Wedeterminedthatoverthecourseoftheexperiment.Theoptimumapproachtowhilegrowthinwellsdidoccur,especiallywhendiluteinoculatheuseofBiologmicroplatesforcommunityanalysiswould

Downloaded from http://aem.asm.org/ on January 9, 2014 by guestVOL.61,1995ANALYSISOFCOMMUNITYSUBSTRATEUTILIZATIONPROFILES1467

seemtorequirecarefulmonitoringofcolordevelopmentsothateffectsofnonlinearitycanbeaccountedforintheﬁnalinterpretationofcommunityresponse.

Wehavedemonstratedfortheﬁrsttimethatthepatternofpositiveandnegativeresponsesandsubstrateoxidationrateandextentwerehighlyreproducibleforsimplemicrobialcom-munities(Fig.6),ifinoculumdensitywascontrolled.ReplicatemicroplatesinoculatedwiththesamemicrobialcommunityandplatesinoculatedwithtwoindependentconstructionsoftheidenticalmodelcommunitiesexhibitednearlyequivalentBiologGNproﬁles(Fig.4and5).Intheseexperiments,themicroplateswereinoculatedwithcommunitieshavingsimilarcelldensities.Withthepotentialcomplicationofdifferentin-oculumcelldensitycontrolledinourexperiments,weobtainedsimilarresultswith24-and72-hreadings.PCAreﬂectedthereproducibilityanduniquenessofthemodelcommunitypro-ﬁles.ReplicatemicroplatesinoculatedwiththesamemicrobialcommunitytendedtoclusterontheﬁrstPCAaxis,whichtypicallyaccountedforϾ60%ofthevariationinthedataset.Furthermore,thetwodifferentcommunitieswereclearlysep-aratedontheﬁrstPCAaxis.

Ourresultsimplythatifdifferentsamplesofsimilarinocu-lumdensityexhibitdifferencesinthepatternofpositiveandnegativeresponsestothe95substratesinBiologGNmicro-plates,thisprobablyreﬂectsrealdifferencesincommunitycomposition.Moreover,positiveandnegativecommunityre-sponsestoparticularBiologsubstratesmaybeinterpretedwithregardtowhetherutilizersofthosesubstratesarepresentinthecommunity.Ourexperimentsdonotaddresstheimpactofrarecommunitymembersoncommunitysubstrateutilizationproﬁles.However,wewouldanticipatethatdramaticdiffer-encesinresponsetoasubstrateacrossreplicates(e.g.,negativeinmostreplicatesbutstronglypositiveinone)mightindicatethepresenceofrarecommunitymembersintheinoculum.IfcommunitiesexhibitdifferencesintherateandextentofoxidationofparticularBiologsubstratesandthesecommuni-tiesaresegregatedinmultivariateanalysesonthebasisofdifferencesinthedegreeofoxidationofparticularsubstrates,doesthisindicate,assuggestedbyothers(2,5,6),realdiffer-encesincommunityfunctionormetabolicpotential?Ourre-sultssuggestthatinterpretationofsubstrateutilizationproﬁleswithregardtoeithercommunitytaxonomiccompositionornumbersandactivitiesofsubstrateutilizerswillbecomplicatedfortworeasons.First,thefailureofsomebacterialstrainstoproduceanyresponseinBiologmicroplatessuggeststhatthemicroplateenvironmentis‘‘selective’’andthereforewilldefactofailtorepresentthemetabolicactivityofallmemberswheninoculatedwithwhole-communitysamples.BacteriawhichproducenoresponseinBiologmicroplatesmaydosobecausethecellsfailtomaintainviability(asnotedforthenegativeresponseofCorynebacteriumsp.strain5togalactoseinFig.3)orbecausetheydonotusethesubstrate.Thepres-enceofCorynebacteriumsp.strain5incombinationwithCy-tophagasp.strain72andPseudomonassp.strain31hadnoeffectontheirratesofsubstrateoxidation,andBacillusstrains,whichgavenoresponseinBiologGNplates,hadnoeffectonthecommunityresponsestoall95substratesinBiologGNmicroplates.Onthebasisoftheseresults,weassumethatthepresenceofnonactivebacteriainacommunitymayhavenodetectableeffectonthecommunityproﬁle.ThisphenomenonwillcomplicateanyattempttoextrapolatefromcommunityBiologproﬁlestoinsitucommunityactivities.

Second,thekineticproﬁleforacombinationofisolatesisnotasimplesummationoftheirindividualproﬁles;therefore,thedegreeofoxidationofparticularsubstratescannotbein-terpretedwithregardtothenumberoractivityofcommunitymemberscarryingoutthatoxidation.Whilekineticproﬁlesforisolatesandcommunitieswerehighlyreproducible,therewasnomeanstopredictisolateorcommunityresponses.Forcom-munitiesoftwoorthreeisolates,wefoundkineticpatternsinwhichtherateandextentofcolordevelopmentwerelessthanorequaltothoseofthemostactivemember.Importantly,wenotednoadditivecommunityresponse,andtherewasnoevi-denceforsynergismbetweenisolatesinthekineticanalyses.Thismeansthatthedegreeofsubstrateoxidationisnotafunctionofthenumberofutilizersoroftheirindividualactiv-ities.Thisconclusionwasfurthersubstantiatedbyourexperi-mentswithmodelcommunities,forwhichwefoundnocorre-lationbetweenthenumberofutilizersandthedegreeofsubstrateoxidation.Infact,forsomesubstrateswhichweresigniﬁcantinthedifferentiationofthetwocommunitiesbyPCA,therewerethesamenumberofutilizersinbothcom-munities.

Ingeneral,PCAseparatedourtwomodelcommunitiesonaxeswhichcouldbeinterpretedinlogicalandexpectedways,typicallyonthebasisofresponseswhichwereveryhighinonecommunityandcorrespondinglylowornegativeintheothercommunity.Theemphasisonnegativeresultsisimportant,sinceourdataindicatethatnonutilizationofasubstratebyacommunityaccuratelyreﬂectsthelikelihoodthatno,orrare,communitymembersintheinoculatedsamplewouldproduceaBiologresponseonthatsubstrate.ThedependenceofPCAseparationondegreeofsubstrateoxidationismoreproblem-aticbecause(i)thedegreeofsubstrateoxidationataparticulartimemaybeafunctionofinoculumcelldensity,(ii)differentmicroorganismswhichusethesamesubstratesoxidizethosesubstratestodifferentextents,and(iii)thesubstrateoxidationproﬁlesofcommunitiesarenotsimplesummationsoftheindividualproﬁlesoftheirconstituentmembers.Environmen-talsamplesmaycontainseveralpopulationscapableofutiliz-ingagivensubstrate,allatundeterminedcelldensities,andexhibitingavarietyofkineticresponsesinBiologmicroplates.Ourresearchwithisolatesandsimplecombinationsofisolatesclearlyindicatesthatwhilewhole-communitysubstrateutiliza-tionproﬁlesconstitutereproduciblesignaturesusefulinclas-sifyingmicrobialcommunities,differencesintherateandex-tentofcolordevelopmentassociatedwithsuchsamplescannotbeinterpretedwithregardtothenumberofutilizersortothemetabolicpotentialorfunctionofthecommunity.

Wefoundthatthepatternsandkineticsofsubstrateutiliza-tionvariedbetweenreplicatesamplesofrhizosphereorbulksoil.Thisinitselfmaybesigniﬁcant,sincewehaveclearlydemonstratedthatbothwithin-wellsubstrateoxidationratesandthepatternofresponsestomultiplesubstratesarehighlyreproducibleforbothisolatesandcommunities.Wehavealsoshownthatdifferencesincommunitykineticproﬁlesonagivensubstratearemostprobablyduetodifferencesincommunitycompositionifinoculumdensityisnotafactor.Asaresult,differencesinsubstrateutilizationproﬁlesforreplicatesam-plesfromsoilmayindicaterealdifferencesincommunitycom-position.Ifthisisthecase,whole-communitysubstrateutili-zationproﬁlesmightbeusedtoassesstheﬁnestructureofmicrobialcommunitydistributionacrossrelativelysmallenvi-ronmentalgradients.Winding(5)reportedtheabilitytodis-criminatebetweendifferentsizefractionsofthesamesoilbyusingBiologGNproﬁles,butsincenotallhersamplesweresubjectedtothesamedilution,thereweredifferentinitialdensitiesofbacteriainthesoilsamples,andonlyasinglereadingwastakenafter3daysofincubation,itwasdifﬁculttodeterminethebasisfortheobserveddifferences.

Inourexperiments,differencesbetweenthebulk-soilandrhizospherecommunitiesmighthavebeenelucidatedbyusing

Downloaded from http://aem.asm.org/ on January 9, 2014 by guest1468HAACKETAL.anextendedsuiteofcompoundsorBiologGNmicrotiterplates.Inexperimentsnotdescribedhere,wefoundthat42substratesweresufﬁcienttodistinguishbetweenourmodelsoilcommu-nitiesand21substratesweresufﬁcienttodistinguishbetweenexperimentallymanipulatedstreambioﬁlmcommunities(4).Therefore,thenumberofsubstratesisnotasimportantaswhethertherearedifferencesintheirusebycommunities.Weforeseethattheuseofresearcher-selectedcompoundsandtheadditionofkineticreadingswillincreasethepotentialappli-cationsandstrengthentheinterpretationofwhole-communitysubstrateutilizationpatterns.ProﬁlessuchasthoseshowninFig.7canthenbeobtainedtoprovideinformationusefulinchoosingtheappropriatetimetocomparereadingsbetweensamplesandallowtheinvestigatortoassesswhetherfeaturessuchasincreasedlagtimeassociatedwithdecreasedinoculumdensityshouldbeconsidered.Kineticproﬁlescanbecharac-terizedbyﬁttingthedatatoanappropriateequation,andtheresultingparametersdeterminedfromcurveﬁttingcouldbeenteredintoastatisticalanalysis,providinggreateranalyticalpowerforecologicalstudies.

ACKNOWLEDGMENTS

WeacknowledgecontributionstothedesignandanalysisoftheseexperimentsbyWilliamHolbenandDaveOdelson.Wearedeeply

APPL.ENVIRON.MICROBIOL.

indebtedtothemanyindividualswhoassistedatvarioustimesintheseexperiments,includingHalCollins,MichaelKaufman,SandyMarsh,andScottMcNabb.Weacknowledgethehelpfulcommentsoftwoanonymousreviewersofapreviousversionofthemanuscript.

ThisresearchwassupportedbyNationalScienceFoundationgrantBIR9120006andagranttotheCenterforMicrobialEcology,Mich-iganStateUniversity,bytheMichiganStrategicFund.

REFERENCES

1.Bochner,B.19.Breathprintsatthemicrobiallevel.ASMNews55:536–539.2.Garland,J.L.,andA.L.Mills.1991.Classiﬁcationandcharacterizationofheterotrophicmicrobialcommunitiesonthebasisofpatternsofcommunity-levelsole-carbon-sourceutilization.Appl.Environ.Microbiol.57:2351–2359.3.Haack,S.K.,H.Garchow,D.A.Odelson,L.J.Forney,andM.J.Klug.1994.Accuracy,reproducibility,andinterpretationoffattyacidmethylesterproﬁlesofmodelsoilbacterialcommunities.Appl.Environ.Microbiol.60:2483–2493.4.Haack,S.K.,H.Garchow,W.Sobczak,L.J.Forney,andM.J.Klug.1994.Analysisofchangesinstreambioﬁlmcommunitystructureandfunctionusingwhole-communityfattyacidmethylesterproﬁlesandcarbon-substrateutili-zationpatterns,abstr.N-206.InAbstractsofthe94thGeneralMeetingoftheAmericanSocietyforMicrobiology1994.AmericanSocietyforMicrobiology,Washington,D.C.

5.Winding,A.1993.FingerprintingbacterialsoilcommunitiesusingBiolog᭨microtiterplates,p.85–94.InK.Ritz,J.Dighten,andK.E.Giller(ed.),Beyondthebiomass:compositionalandfunctionalanalysisofsoilmicrobialcommunities.JohnWiley&SonsLtd.,Chichester,UnitedKingdom.

6.Zak,J.C.,M.R.Willig,D.L.Moorhead,andH.G.Wildman.1994.Func-tionaldiversityofmicrobialcommunities:aquantitativeapproach.SoilBiol.Biochem.26:1101–1108.

Downloaded from http://aem.asm.org/ on January 9, 2014 by guest

因篇幅问题不能全部显示，请点此查看更多更全内容

查看全文