polymeraseco

ORIGINALARTICLERNA-seqrevealscooperativemetabolicinteractionsbetweentwotermite-gutspirochetespeciesinco-cultureAdamZRosenthal1,EricGMatson1,AvigdorEldar2andJaredRLeadbetter11RonaldandMaxineLindeCenterforGlobalEnvironmentalScience,CaliforniaInstituteofTechnology,Mailcode138-78,Pasadena,CA,USAand2HowardHughesMedicalInstituteandDivisionofBiologyandDepartmentofAppliedPhysics,CaliforniaInstituteofTechnology,Pasadena,CA,USAThehindgutsofwood-feedingtermitestypicallycontainhundredsofmicrobialspecies.
Togetherwiththeirinsecthost,thesegutmicrobesdegradelignocelluloseintousablecatabolites.
Althoughpastresearchrevealedmanyfacetsofthestepwiseflowofmetabolitesinthisscheme,notmuchisknownaboutthebreadthofinteractionsoccurringbetweentermite-gutmicrobes.
Mostofthesemicrobesarethoughttodependon,andtohaveco-speciatedwith,theirhostandeachotherformillionsofyears.
Inthisstudy,weexploredtheinteractionsoftwospirochetespreviouslyisolatedfromtheverysametermitespecies.
Ashydrogen(H2)isthecentralfreeintermediateintermite-gutlignocellulosedigestion,wefocusedoninteractionsbetweentwocloselyrelatedtermite-gutspirochetespossessingcomplementaryH2physiologies:oneproducesH2,whiletheotherconsumesit.
Invitro,thesetwoTreponemaspeciesmarkedlyenhancedeachother'sgrowth.
RNAsequencingresolvedthetranscriptomesofthesetwocloselyrelatedorganisms,revealingthatco-cultivationcausescomprehensivechangesinglobalgeneexpression.
Theexpressionofwellovera100genesineachspecieswaschanged4twofold,withoveradozenchanged410-fold.
Severalchangesimplicatingsynergisticcross-feedingofknownmetaboliteswerevalidatedinvitro.
Additionally,certainactivitiesbeneficialtothehostwerepreferentiallyexpressedduringconsortialgrowth.
However,themajorityofchangesingeneexpressionarenotyetunderstandable,butindicateabroad,comprehensiveandmutualisticinteractionbetweenthesecloselyrelated,co-residentgutsymbionts.
Theresultssuggestthatstaggeringlyintricatenetworksofmetabolicandgeneinteractionsdrivelignocellulosedegradationandco-evolutionoftermitegutmicrobiota.
TheISMEJournal(2011)5,1133–1142;doi:10.
1038/ISMEJ.
2011.
3;publishedonline17February2011SubjectCategory:microbe–microbeandmicrobe–hostinteractionsKeywords:co-culture;RNASeq;symbiosis;termite-gutIntroductionAbacterium'scapacitytoperformphysiologicaltasksinisolationisnotnecessarilypredictiveofhowitperformsinthenaturalenvironment.
Takeforexampletheclassicmodelbacterium,Escherichiacoli,aversatileorganismthatdegradesmanysubstratesundermanygrowthconditionsinpureculture(Booneetal.
,2001).
Inenvironmentalcontext,muchlessisknownaboutitsphysiology(Changetal.
,2004).
E.
coliprobablydegradesonlyafewsubstratesinsitu,becauseitisoutnumberedandlikelyoutprocessedbyadiversityofotherspecialists.
Likewise,somespeciesdonotexpresscertaintraitswhengrownalone,andwhatappearstobeacrypticactivityonlybloomswhengrowthoccursinthepresenceofotherspecies(Straightetal.
,2007).
Lastly,host-associatedorganismsmayrespondtocuesfromtheirhost,andinresponsecatalyzeactivitiesthatimpactthehost(Palmeretal.
,2007).
Neitheroftheseactivitiesaretypicallyrecapitulatedduringstandardcultivationregimes.
Physiologicalanalysesofmodelorganisms,likeE.
coliabove,havegreatlybenefitedfromhigh-throughputtools,especiallymicroarrays(Changetal.
,2004).
However,arraysrequireextensivedesignandground-truthingbeforeeventhefirstexperimentcanbeperformed.
Theseconstraintsaresignificant,consideringthatthenumberofcandi-dateorganismsforexpressionstudieshasexplodedinthiseraofgenomesequencing.
Moreover,anysuccessfulinterpretationofarraydatabecomescomplicatedwheninvestigatingglobalgeneexpres-sionofcloselyrelatedspecies(sharinghighnucleicacidsimilarities)grownineitherdefinedco-culturesortheirnaturalhabitats.
Theprobe-bindingnatureofthemicroarrayapproachisnotwellsuitedtodistinguishbetweenconservedReceived26August2010;revised16November2010;accepted22December2010;publishedonline17February2011Correspondence:JRLeadbetter,RonaldandMaxineLindeCenterforGlobalEnvironmentalScience,CaliforniaInstituteofTechnology,Mailcode138-78,Pasadena,CA91125,USA.
E-mail:jleadbetter@caltech.
eduTheISMEJournal(2011)5,1133–1142&2011InternationalSocietyforMicrobialEcologyAllrightsreserved1751-7362/11www.
nature.
com/ismejnucleotidesequences.
Thus,array-basedstudiesonco-culturesandcommunitiesareessentiallylimitedtoexamplesinvolvingspecieshavingverydifferentgenomes(Saleh-Lakhaetal.
,2005).
Breakthroughsinnext-generationsequencingnotonlyallowthecomparisonofexpressionfromhighlysimilarsequences,butalsodramaticallyexpeditethepaceatwhichdatacanbegatheredfromless-well-studiedmicrobes(Holtetal.
,2008;Yoder-Himesetal.
,2009)andcomplexbiologicalenvironments(Gilbertetal.
,2008;Urichetal.
,2008).
Althoughtranscriptomiccalculationsarenowpossible,studiesofco-culturesandintactenvironmentsarechallenging,astheyentailinher-entdifficultieswhenmakingcomparisonsandcontrasts.
Forone,itisverydifficulttodirectlycomparethegeneexpressionpatternsofcultureswhenthegrowthconditionsareverydissimilar,andthisisthecasewhenenvironmentallyreliantsymbiontsareremovedfromtheirenvironment—orevenonefromanotherindefinedco-culture.
Furthermore,commonRNA-seqtechniquesdonotdifferentiatethedirectioninwhichatranscriptisoriented,andthereappearstobetimeswhenanti-codingRNAsarepresent,whichmaynotbefullyaccountedforbymanystudies(Dornenburgetal.
,2010).
However,studiesoftranscriptionaloutputindifferentconditionsdoallowonetoanalyzetheimpactofindividualmicrobesataparticularpointoftime,andtoproducetestablehypothesesaboutthefunctionsofdifferentmembersofamicrobialcommunity,issuesatthecruxofmolecularmicro-bialecology.
Inthisstudy,wecombinedtheuseofclassicmicrobiologicalcultivationtechniqueswithnext-generationsequencingtoexploregeneexpressionpatternsunderlyingthemetabolicinteractionsbetweencloselyrelatedbacteriathathadbeenisolatedfromthesamemicroliter-in-scaletermitehindgutenvironment.
Thegutcommunitiesofwood-feedingtermitesfermentthepolysaccharidefractionsoflignocellu-lose,ultimatelyprovidingthetermitewithcom-poundssuchascarbonandenergysubstrateacetate,aminoacidsandotheressentialgrowthfactors(OdelsonandBreznak,1983;Braumanetal.
,1992;Abeetal.
,2000).
Duringthecourseoflignocellulosefermentation,thefreeintermediatehydrogen(H2)isproducedandaccumulatestoconcentrationsnearsaturation(PesterandBrune,2007),beforebeingconsumedbyhomoacetogenicgutbacteriaandmethanogenicarchaea(BreznakandSwitzer,1986;Braumanetal.
,1992).
IthasbeenestimatedthatH2CO2acetogenicbacteriasupplytheirtermitehostwithuptoathirdofitsacetatesupply(BreznakandSwitzer,1986).
Together,thefermentationandCO2-reductiveacetogenesis-derivedacetatecancon-tributeupto70–100%ofthetermitehost'senergyrequirements(Drake,1994).
Inthisstudy,wehavesoughttorevealpossibleinteractionsoccurringbetweentwopreviouslyisolatedtermite-guttreponemes:TreponemaprimitiaZAS-2andTreponemaazotonutriciumZAS-9.
ThesetwospeciesresideincloseproximityinthehindgutofthedampwoodtermiteZootermopsisangusticolis(Graberetal.
,2004),wherein(asinmosttermites)adiversityofspirochetesconstitutealargeportionofthetotalgutmicrobiota(Breznak,2002).
Previousstudiessuggestthatthesebacteriahaveseparatephysiologicalrolesinthecomplexmutualism(Leadbetteretal.
,1999;Lilburnetal.
,2001;GraberandBreznak,2004;Graberetal.
,2004).
However,towhatextenttheyinteractwithandimpacteachotherandothergutspeciesislesswellexamined.
Certainly,T.
primitiaisknowntodependonanessentialgrowthfactor,folate,derivedfromothergutmicrobiota(GraberandBreznak,2005).
T.
azotonutriciumfixesnitrogen(Lilburnetal.
,2001)andproducesH2asafermentationproduct(Graberetal.
,2004),whereasT.
primitiaisanH2-consumingCO2-reducinghomoacetogen(Leadbetteretal.
,1999).
TheircomplementaryH2physiologiesandtheircloseproximitytoeachotherinthesmalltermite-gutenvironmentbegsthequestionofwhethertheymightengageinotherinteractions.
ThegenomesofT.
azotonutriciumandT.
primitiaarerelativelylarge(3.
86Mband4.
06Mb,respec-tively(Graberetal.
,2004)comparedwithothersequencedtreponemes(T.
denticola2.
84Mb(Sesha-drietal.
,2004);T.
pallidum1.
14Mb(Fraseretal.
,1998)).
Wesoughttoidentifygenesthatmayberelevanttointerspeciesinteractionsbyanalyzingglobalgeneexpressionduringconsortialgrowthofthetwo,viaIlluminatranscript-sequencingtechnol-ogy(Illumina,SanDiego,CA,USA).
MaterialsandmethodsBacterialstrainsandculturemediaT.
primitia(ZAS-2)andT.
azotonutricium(ZAS-9)culturesweregrownin4-YAComedia(4%yeastautolysate)supplementedwith20mMmaltoseand80%H2and20%CO2intheheadspace.
Culturesweregrownin5mlvolumesin25-mlBalchtubes,withcrimptopstoppersinthedarkatroomtemperature.
Vitamin,amino-acidandcofactorpreparationsusedtosupplementmediawereasfollows:VitaminB7:0.
3–0.
5mgbiotin(SigmaAldrich,StLouis,MO,USA)wasaddedperculturetube.
VitaminB6:120–200mgpyridoxal-HClandpyridoxal-phosphate(SigmaAldrich)wereaddedperculturetube.
B12andcorrinoids:30–50mgofoneofthefollowing:(1)hydroxocobalaminacetatesalt,(2)hydroxocobalaminhydrochloride,(3)methyl-cobalaminand(4)cyanocobalamin(SigmaAldrich),wasaddedperculturetube.
Tryptophan:60–100mgtryptophan(SigmaAldrich)wasaddedperculturetube.
Co-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1134TheISMEJournalRNAisolationandprocessingTotalRNAwasisolatedfromtwobiologicalrepli-catespergrowthcondition.
ProceduresforsamplepreparationfollowedfromthestandardIlluminaprotocolforRNA-Seqsamplepreparationavailablefromthemanufacturer(Illumina).
Inshort,RNAwasisolatedusingRNeasykit(Qiagen,Valencia,CA,USA)asperthemanufacturer'sprotocol.
SampleswererunthroughtheRNeasyproceduretwice,withtheoptionalDNasetreatmentperformedinbothinstances.
TotalRNAwasfragmentedusingtheAmbionRNAfragmentationkit(Ambion,Austin,TX,USA)andprotocol.
First-strandcomplementaryDNAwaspreparedfollowingtheSuperScriptIImethod(Invitrogen,Carlsbad,CA,USA),usingtheInvitrogenhexamerrandomprimerstoensurelowprimerbias.
Second-strandcomplementaryDNAwassubsequentlysynthesizedbyaddingtothefirst-strandreactionsecond-strandbuffer(500mMTris-HClpH7.
8,50mMMgCl2,10mMdithiothreitol),deoxyribo-nucleotidetriphosphate(0.
3mm),RNaseH(2Uml1;Invitrogen#18021-014),InvitrogenhexamerprimersandDNApolymeraseI(Invitrogen).
Thefinalreactionvolumeforsecond-strandsynthesiswas100ml,andreactionswerecarriedoutat161Cfor2.
5h.
ComplementaryDNAsequencingFragmentedsecond-strandcomplementaryDNAsamplesweresubmittedtotheCaltechSequencingCorefacility(Pasadena,CA,USA).
Librariesweresequencedas37-mersusingthestandardSolexa(Illumina)protocolandpipeline.
SequencingdepthinformationissummarizedinSupplementaryTable3.
RNA-SeqdataanalysisIlluminarawdataprovidedbytheGERALD(Illumina)softwarepackagewerealignedtoaFASTAfilecontainingbothT.
azotonutricium(ZAS-9)andT.
primitia(ZAS-2)genomesusingtheMaqshortreadaligningprogram(WellcomeTrustSangerInstitute,Hinxton,UK).
Forinitialanalysisofmappingquality(seeTable1),zerooronemisseswereallowedperread.
Allsubsequentsampleswereanalyzedwithamaximumofonemismatchallowed.
Readsfrombiologicalreplicateswerefirstcom-paredwitheachother(a)graphicallyaftermappingontothetwogenomes,andthen(b)bylookingfordifferencesinfoldregulationwhencomparedinallpairwisecombinationsofotherreplicatesofinterest.
Biologicalreplicatesbroadlywerefarmoresimilartoeachotherthantoothersamples.
Readsfrombiologicalreplicatesweremergedandaveragedforallfurtheranalysis.
Toexcludereadsthatmayalignambiguously(thatis,toeithergenome),weassembledhypothe-ticaldatabasesofallpossible370-mersequencesforeachgenome,andmappedthemtotheopposinggenomewithonemismatchallowed.
Additionally,adatabaseofpooledIlluminasequencingdatafromeachpureculturewasalsoalignedtotheopposinggenome.
Readsthatambiguouslymappedwereexcludedfromtranscriptionalanalysis.
Geneexpressionvaluesweredeterminedbynormalizingthenumberofreadsmappedtoaparticulargene(excludingambiguousregions)dividedbythesizeofthegene(alsoexcludingambiguousregions).
Theresultingvalueisthenormalizedreadsperkilobase,inamannerconsistentwiththegeneexpressionindexcalculationsofpreviouslypublishedreports(Yoder-Himesetal.
,2009).
Inordertoadjustforintensitybetweensamples,theribosomalsignalfromeachsamplewasusedasastandard,andeachsample'sintensitywasmultipliedbyafactorthatwouldyieldanequalribosomalRNA(rRNA)signal.
Inconsideringup-ordownregulatedgenes,acutoffoftwofoldincreaseintranscriptionwasused.
Table1AhypotheticalshortsequencedatasetandexperimentalRNA-sequencingdatapreferentiallyalignedtothecognategenomeTotalDBsizeTotalhits(%rRNA)Non-rRNAhitsUniquelociaGeneswithahit(%)HypotheticalExact38556716518(45%)29232923109(2.
8%)1miss385567114086(27%)1028310283290(7.
6%)ActualbBeforemask109439941936998(99.
9%)1525575161(4.
1%)Aftermaskc542600646(0%)646340139(3.
6%)ActualdBeforemask151510142400846(99.
9%)1280409194(4.
8%)Aftermaskc427444380(0%)380204151(3.
8%)Abbreviation:rRNA,ribosomalRNA.
aUniquelocirefertothenumberofdistinctnon-ribosomalsequencesofT.
primitiathathadatleastonehit.
Anin-silico-generateddatasetofallpossible37base-pairsequencesinthegenomeofT.
azotonutricium(hypothetical)andtheRNA-seqdatafromasampleofT.
azotonutricium(denotedas'Actualb'inthetable)weremappedtothegenomeofT.
primitia.
Databasesizedescribesthenumberofshortsequencesinthedatasets.
ThetotalhitscolumndisplaysthenumberofshortsequencesthatmappedtotheT.
primitiagenome,andthepercentageofhitsthataligntoribosomal16Sor23S.
cAftermasksequencesarethoseremainingafterthemostsimilarsequencesbetweenthetwogenomeswereremovedfromsampling.
dReferstotheRNA-SeqdatafromasampleofT.
primitia,whenmappedtothegenomeofT.
azotonutricium.
Co-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1135TheISMEJournalAdditionally,exceptinthecaseoffunctionalgroupsandgeneclustersusedtogenerateSupplementaryTable1,onlygeneswithgreaterthan50adjustedhitsperkbofcodingDNAwereconsideredinanalyses.
SignalintensitieswerevisualizedgraphicallybyconvertingMaq-alignedreadsintoa.
BARfileusingtheCisgenomesoftware,andviewedontheCisgenomebrowser(StanfordUniversity,Stanford,CA,USA)(Jietal.
,2008).
QuantitativeRT-PCRofseveralT.
azotonutriciumgenes:clpX(TREAZ#737).
Endo-1,4-beta-xylanaseAprecursor(TREAZ#2717).
Endo-1,4-beta-xylanaseAprecursor(TREAZ#2718).
Glycoproteingp2,Endo-1,4-beta-xylanaseAprecursor(TREAZ#2716).
TheseqRTmeasurementsrevealexpressionratiossimilartothosereportedbyRNA-Seq.
QuantitativePCRInall,100mlsamplesfrombacterialculturesweretakendaily.
Sampleswerecentrifugedat13000gfor5min,aspirated,resuspendedinanequalvolumeofdH2O,andfrozenforlateruse.
QuantitativePCRonbacterialsamplesfollowed,usingprimersspecificfortheclpXgeneofeitherT.
primitiaorT.
azotonutricium.
QuantitativePCRprimersof20bplengthweredesigned,withthefollowingsequences:T.
azotonutriciumclpX(fwd):50-GGAACTTTTCGATGCTCTGC-30T.
azotonutriciumclpX(rev):50-GCGCTTAAGGTCTTCCCTCT-30T.
primitiaclpX(fwd):50-CTCCCGTTTCATTTCTTCCA-30T.
primitiaclpX(rev):50-GAAATGTTAGACGCCCTCCA-30Primersweredesignedtohavearoughlyequalamplificationproductlengthtoavoidlargediffer-encesinfluorescenceintensities.
Eachpairofprimerswastestedformeltingtemperature,speci-ficityandamplificationefficiency,andwasfoundtobespecificandsuitableforq-PCR(eachhavinganEfactorgreaterthan1.
75).
QuantitativePCRreactionswereperformedwiththeiTaqSYBRGreenPCRkitandaDNAenginechromo-4qPCRinstrument(Bio-Rad,Hercules,CA,USA).
AstandardcurveofknowndilutionsofT.
primitiaandT.
azotonutriciumgenomicDNAwasperformedwitheachsetofexperiments.
Triplicateq-PCRreactionswereusedforeachdatapoint.
Toobtaingenomiccellequivalentunits,theribosomalcopynumberwasobtainedfromthestandardcurvecalculations,anddividedby2toaccountforthenumberof16Scopiesineachgenome.
ResultsanddiscussionFullandaccurateanalysisoftranscriptpoolsfromtwocloselyrelatedorganismsgrowntogetherreliesontheabilitytodistinguishbetweenhighlyhomo-logoussequences.
Therefore,beforeanalyzingtheexperimentaldata,wedeterminedthefrequencyofcross-identificationbetweenthetwotreponemesusedinthisstudy.
Todothis,adatasetofallpossibleshortnucleotidesequences(37bp,theexactlengthofthereads)fromeachgenomewasgeneratedinsilicoandalignedtothegenomeoftheotherspecies(Materialsandmethods).
Theresults(Table1)demonstratethatonlyasmallnumberofallpossible37-bpsequences(6518,ca.
0.
15%)areidenticalinbothbacteria.
Alargepercentageofthese,ca.
45%,arer16Sandr23SRNAs,whereasgreaterthan97%ofallnon-rRNAgenesarecompletelyunique(thatis,donotcontainasinglesequencereadhavinganexactmatchtotheothergenome).
Relaxingthestringencytoallowonemismatchper37-bpsequencedoesnotyielddramaticallylessfavorableresults,withover92%ofthegenesbeingcompletelyunique.
Moreover,innocasewasasinglegeneineithergenomefullymasked;allgenesinbothorganismshadatleastoneunique37-bpidentifier.
Thisresolutionallowscompletetranscriptdiscriminationandgenome-wideanalysisinco-cultureexperiments.
Experimentalconsiderationscompelledustoanalyzethebackgroundnoiseexpectedinactualdata.
Tothisend,adatabasetotalingnearly11millionRNA-SeqreadsfromT.
azotonutriciumwasmappedtothecombinedgenomesofbothbacteria.
Aspredicted,themajorityofreads(9million;82.
5%)alignedtoT.
azotonutricium,andpracticallyall(over99.
9%)ofthereadsthatdidmaptoT.
primitiawererRNA(Table1)Thisresultwasexpected,asribosomaltranscriptiontypicallycomprisesover90%ofthetotaltranscriptpool(NeidhardtandCurtiss,1996),andrRNAsaretypicallythemostconservedsequencesbetweenthegenomes(thesmallsubunitrRNAhas93%nucleotidesimilarity;(Graberetal.
,2004)).
WhenallrRNAsequences,excepthighlyvariableregionsandsequenceswithuptoonemismatchsimilarity,weremaskedfromtheanalysis,only646readsfromonespeciesmappedtotheothergenome.
Thisledustoconcludethatthismethodwaseffectiveatdiscri-minatingthesignalsfromtwosimilargenomes.
Thereciprocalanalysesyieldedsimilartrends,andarepresentedinTable1.
Forgeneexpressionprofiling,bacterialculturesofT.
primitia,T.
azotonutriciumandtheirco-cultureweregrownon4YAComediumsupplementedwithH2andmaltose(Graberetal.
,2004).
Thegrowthkineticsofeachspecieswasfollowedasafunctionofgenomecopynumber,thatis,usingqPCRofuniquechromosomalDNAmarkersfromeachgenome.
Thesemeasurementsrevealthatinitiallyeachspeciesgrewatsimilarrates,whetherinpureorco-culture(shadedregion,Figure1).
However,afterthisinitialgrowthphase,thegrowthofeachspecieswasmarkedlystimulatedinbothrateandyieldinco-cultures,relativetogrowthinCo-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1136TheISMEJournalmonoculture.
ThegreatereffectwasobservedforT.
azotonutricium(Figure1).
Importantly,althoughthesetwospeciesmightcompeteforanabolicnutrientsinthemedium'syeastautolysatebase,atnopointwerethetwospeciesobservedtohaveanynegativeimpactoneachother'sgrowth.
Allsamplesforgeneexpressionanalysiswerecollectedduringaperiodwithinthelater,synergism-dominatedgrowthphase(OD600E0.
35)(Figure1).
Overall,alargesetofgenes,consistingofapproximately7%ofallT.
azotonutriciumand4.
5%ofallT.
primitia-codingregions,weredifferentiallyregulatedasafunctionofco-cultivation(SupplementaryTable2).
Inagreementwithgrowthmeasurements,severalgenesrelatedtocelldivisionandotheraspectsofgrowthareupregulatedinco-cultures(Supplemen-taryTable2).
Theobservationthatinthelatergrowthphaseco-culturesmaintainnear-logphasegrowthrates(whilethepureculturesarenearingorenteringstationaryphase)canalsobeusedtoexplainthedifferencesbetweentranscriptpoolsintheco-cultures.
ThisisespeciallytrueforT.
azotonutricium,forwhichsamplesweretakeninalatestagetoallowforsufficientcellgrowth.
Thepositiveco-culturegrowtheffectonT.
primitiamightbeanticipatedwhenconsideringtheH2physiologiesofthetwospecies.
Fermentinggutmicrobes,includingT.
azotonutricium,providehomoacetogenslikeT.
primitiaasourceofH2(PesterandBrune,2007).
Intheculturemedia,aninitialamountofH2gas(80%),whichroughlymimicsphysiologicalconditionswithinthegut,wassup-pliedintheheadspacebutwasnotreplenishedduringgrowth.
Previously,thegrowthofT.
primitiahasbeenobservedtobenoticeablyrestrictedifH2isnotinitiallysuppliedinthemedium(GraberandBreznak,2004).
Thus,onepredictedaspectofsynergybetweenthetwostrainsisthecontinuedproductionofH2byT.
azotonutricium,asthisgasisconsumedbyT.
primitia.
Indeed,T.
primitiagenesassociatedwithH2utilizationweredifferentiallyregulatedinco-cultivation:amongthemosthighlyupregulatedT.
primitiageneswerethosethatencodedseveralhydrogenase-likeproteins;inturn,thegenesforseveralotherhydrogenase-likeproteinsaremarkedlydownregulatedduringco-cultivation(Table2,SupplementaryTable1).
ThesefindingsareconsistentwithpreviouslypublishedreportsofhydrogenaseexpressioninmethanogenicH2con-sumers(Desulfovibriovulgaris),whenculturedtogetherwithahydrogenproducer(Walkeretal.
,2009).
Whenthetwotreponemeswereco-cultured,anadditionaleffectonT.
primitiagenesassociatedwithCO2-reductiveacetogenesiswasobserved.
Inpureculture,bothaselenocysteine-containing(fdhFsec)andanon-selenocysteine(fdhFcys)variantoftheenzymeformatedehydrogenase,whichcata-lyzesanearlystepoftheWood–Ljungdahlpathway,weretranscribed(Matsonetal.
,2010).
Duringgrowthinco-cultures,theselenocysteineformoftheenzymewasupregulated,whereasthenon-seleniumformwasdownregulated(Table2,Supple-mentaryTables1and4).
Co-cultivationwithT.
primitiasignificantlyenhancedthegrowthrateandyieldofT.
azoto-nutricium(Figure1).
Examinationofthegeneexpressiondataandthegenomesequencesofthetwotreponemessuggestedseveralpotentialgrowthfactorsthatmightbeinvolvedintheinteraction.
Thedifferentialregulationofmanygenesinvolvedincorrinoidproductionortransport,vitaminB7(biotin),vitaminB6(pyridoxalphosphate)andcoenzyme-A(Table2,SupplementaryTable1),wasobserved.
InthecaseofvitaminB7,severalrelevantregulatoryortransportgenesaredownregulatedwhenT.
azotonutriciumisco-cultivatedwithT.
primitia(Table2,SupplementaryTable1),butonlythelatterhasacomplementofthegenestoTd1Td2YieldT.
primitia32.
21790.
43T.
primco-culture30.
978.
50.
69T.
azotonutricium64158.
80.
2T.
azotoco-culture4959.
50.
07100Hours1.
00.
10.
01109cells.
ml-1genomicequiv.
0300200Figure1Synergisticgrowtheffectsinco-cultureoftermite-guttreponemes.
QuantitativePCRmeasurementsaregivenforT.
primitiaandT.
azotonutriciumabundanceineitherpureculture(solidlines)orco-culture(dashedlines).
Thegrowthphaseinwhichbacteriainbothpureculturesandco-culturesgrowatroughlyequalratesismarkedbyashadedbox.
RNA-sequencingsamplesweretakenatapproximately160hours(co-culture),210h(T.
primitia)and260h(T.
azotonutricium).
ThesecollectiontimesareinthelatergrowthphaseandaremarkedbyarrowsontheXaxis.
Sampleswererunintriplicates,andthes.
d.
isdenotedbyerrorbars.
Growthratesandyieldofeachsamplearesummarizedbelowthecurve.
Yieldsareinnumberofgenomicequivalentsperml.
Doublingtimeswerecalculatedfromdatapointseitherwithinthefirst117hofgrowth(Td1)orwithintheproductivestageofgrowththatproceeds(Td2).
Co-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1137TheISMEJournalsynthesizeB7(SupplementaryFigure1).
Similarly,animportantgeneforthesynthesisofvitaminB6precursors(forexample,phosphoserineaminotrans-ferase)(LamandWinkler,1990)andseveralgenesthatrequirevitaminB6forfullactivitywerealsodownregulatedduringconsortialgrowthofT.
azotonutricium(Table2,SupplementaryTable1).
TheenzymeresponsibleforproducingtheactiveformofvitaminB6(pyridoxalkinase)isexclusivetothegenomeofT.
primitia(SupplementaryFigure1).
Finally,keygenesforcoenzyme-AbiosynthesisareabsentinT.
azotonutriciumbutpresentinT.
primitia(SupplementaryFigure1).
OneexplanationfortheupregulationofgenesthattransportvitaminB7orrequirevitaminB6foractivityisthatasthebacterialpopulationsizeincreases,theamountofavailablevitaminsisincreasinglyscarce,andthusbacteriaaremorereliantonimportingandusingwhatlittlevitaminsarepresent.
Toexamineifanyofthesecandidategrowthfactorsmightberelevanttotheobservedsynergisticgrowthofthespeciesduringtheirco-culture,thegrowthmediumofT.
azotonutriciumpurecultureswassupplemented.
Theresults(Figure2)demon-stratethatvitaminB6,andtoalesserextentB7andcoenzyme-A,haveapositiveimpactongrowth.
NoneofthecorrinoidorB12preparationstestedimprovedthegrowthofT.
azotonutricium,norweretheadditionsofanyofthesesupplementsobservedtohaveanybeneficialeffectonthegrowthofT.
primitia.
Inadditiontothecandidatecofactorsandvita-minsthatwereidentifiedandexploredabove,wenotethatanumberofhypotheticalgenesarealsoregulateddifferentlyinco-culture(Figure3,SupplementaryTable2).
Thelargenumberofthesegenessuggeststhatadditionalprocessesmaybeinvolvedincomplexmutualisticsymbiosis.
Theabundanceofgenesregulatedbythesecondi-tionsalsoservestounderscorethelargeeffectsthatgrowthinthepresenceofothermicrobeshasonthephysiologyandbehaviorofanotherbacterial02004006000.
010.
1110BasemediaB6/B7CoAB7B6B6/B7/CoATime(hours)OD600nmFigure2PositivegrowtheffectsofthevitaminsB6,B7andcoenzyme-AonT.
azotonutricium.
T.
azotonutriciumgrowthrateandyieldbenefitfromtheadditionofvitaminB6,coenzyme-AandvitaminB7workinconsortiawithB6tofurtherincreasetheyieldandrateoftheculturemedia.
Datafromduplicateswereaveragedandplotted.
Table2Majorgroupsoftranscriptionallyup-anddownregulatedgenesinco-cultureProcessFunctionUpinco-cultureDowninco-cultureTreponemaprimitiaMetabolismHydrogenandC1metabolism57VitaminsandcofactorsB12andcorrinoidrelated223Tryptophan/phenylalanine/tyrosinebiosynthesis10AminoacidsMethioninesynthesisandtransport3Isoleucine/leucine/valinetransport2TreponemaazotonutriciumVitaminsandcofactorsB12andcorrinoid-related8Biotintransport,regulation,metabolism3VitaminB6precursorsynthesis1EnzymesrequiringB6foractivity3AminoacidsIsoleucine/leucine/valinebiosynthesis4Serinea3Cysteinea2RegulatedgenesandgeneclustersofT.
primitiaandT.
azotonutriciumarelistedbymajorcellularpathways.
Includedaregeneswithclearannotationandfoldchangethatisabovebackground(seeMaterialsandmethods),andwhicharediscussedinthemanuscript.
Thevaluesintheupanddowncolumnsdescribethenumberofgenesassociatedwithaspecificprocess.
aSomegenesinvolvedintheserineandcysteinebiosynthesispathwaysrequirevitaminB6,andappearinbothcategories.
Co-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1138TheISMEJournalspecies.
Lookingahead,itispossiblethattheaccumulationofexpressionstudiesofmicrobialcommunitiescanbeusedtohighlightuncharacter-izedgenesthatactasputativesymbiosisdetermi-nants.
Ahallmarkofseveralinsect–microbesymbiosesisthebacterialproductionoffactorsvitaltothehost,includingvitamins,cofactorsandessentialaminoacids(reviewedinMoran,2006).
VitaminsB6andB7,theproductionofwhichappearstoaidinthegrowthofT.
azotonutriciumwhenitisgrowninco-culturewithT.
primitia,havearoleinthesym-biosisbetweentse-tsefliesandtheirendosymbiontWigglesworthiaglossinidiabrevipalpis(Akmanetal.
,2002).
Genesforseveralamino-acidbiosynth-esispathwayswereupregulatedduringthecon-sortialgrowthofthetwotreponemes,includingthoseforaromaticaminoacidsandmethioninebiosynthesesinT.
primitia,andforbranched-chainaminoacidsinT.
azotonutricium(Table2).
Togethertheserepresentsixoutoftenaminoacidsknowntobeessentialininsects(Giletal.
,2003).
Curiously,thetryptophansynthesispathwayappearstobeabsentinthediazotrophT.
azotonutricium,buttheexpressionofthesegeneswashighlyinducedinT.
primitiawhenthetwoweregrowntogether.
PaststudieshadrevealedthatT.
azotonutriciumiscapableofgrowthasabona-fideN2-fixingbacterium(Lilburnetal.
,2001),yetrequiredthatthemediumbesupplementedwithyeastautolysatewhiledoingso,presumablyatleastinparttocomplementitsneedfortryptophan.
Thus,ourresultssuggestthatthefixationofN2,ultimatelyintoafullsuiteofessentialaminoacids,isaconsortialactivityintheN-limitedlignocellulose-degradingenvironmentofthetermitegut.
Inthecaseofthebranched-chainaminoacids(forwhichbothstrainsappeartohaveallrequiredbiosyntheticcomponents),co-cultiva-tionstimulatesexpressionofkeysynthesisgenesinT.
azotonutriciumandkeytransportgenesinT.
primitia(Table2,SupplementaryTable1).
Takentogether,thegeneexpressionpatternsobservedduringtheconsortialgrowthofthetwospeciessuggestastreamlineddivisionofbiosyntheticlaborandsharedproductionofkeynutrients.
Theupregulationofessentialamino-acidbio-synthesisgenesinco-culturesuggeststhattheirproductionmayalsobenefitthehostandothercommunitymembers,astheseessentialaminoacidsareoftenimplicatedininsect/bacteriasymbioses(Shigenobuetal.
,2000;Giletal.
,2003;Nakabachietal.
,2006).
Therecentlypublishedgenomesoftwoendosymbiontsoftermite-gutprotozoahavealsosuggestedtheimportanceofotheraspectsofamino-acidandvitaminproductioninthesymbiosis(Hongohetal.
,2008a,b).
Thegenomesofintracellularendosymbiontsofinsectsareoftenhighlyreduced(Shigenobuetal.
,2000;Akmanetal.
,2002;Giletal.
,2003;Nakabachietal.
,2006;Hongohetal.
,2008a,b)andunderlietheobligaterelianceofthesebacteriaonnutrientsandfactorssuppliedbythehost.
Inreturn,manysuchendosymbiontsactalmostexclusivelyasessentialaminoacidsandvitaminfactories,capableoflittleelse(Shigenobuetal.
,2000;Akmanetal.
,2002;Giletal.
,2003;Nakabachietal.
,2006).
Incontrast,thesetermite-gutspirochetesarenotintracellularresidentsoftheirhost'stissues,andhavecompara-tivelylargegenomes,suggestingthattheyarecapableofperformingmanymoretasksintheirspecies-richsymbioticenvironmentthantheirendosymbionticcounterparts.
Severalofthesetasks,asimplicatedbyexpressionandactivitydatafromthisandpreviousstudies,arepresentedinFigure4.
TheyincludeacetateproductionduringsugarfermentationandCO2-reductiveacetogenesis,H2cycling,nitrogenfixation,amino-acidbiosynthesisandvitaminandco-factorproduction(Leadbetteretal.
,1999;Lilburnetal.
,2001;Graberetal.
,2004).
Inreturn,thetermiteprovidesitsmicrobiotawithfinelygroundparticlesofrecalcitratcarbonandenergysource,lignocelluloseandacontrolledenvironment.
Inadditiontoprovidinginsightsintothesymbio-ticinteractionsofthesetwotreponemesinthecontextofthetermitegut,theRNA-Seqdatathat0200400246810Fold-regulationGenesFigure3Theexpressionofalargenumberofgenesinbothbacterialspeciesiseffectedbyco-culturing.
Thenumberofgeneswithexpressionlevelsregulatedbytwofold,fourfold,sixfold,eightfoldand10-fold(Xaxis)plottedforT.
primitia(solidlines),T.
azotonutricium(dashedline)andbothspecies(dottedline).
Best-fitlineswereaddedtoaidindatavisualization.
Theareabelowthedark-shadedboxshowsthenumberofgenesthatareonaveragetwofoldregulated,andalsohavethreeofthefourpossiblebiologicalreplicatecombinationsabove1.
7-foldregulated.
Thevaluesbelowthelightgrayshadearegenesthatareonaveragetwofoldregulated,andinwhichallfourbiologicalreplicatecombinationsareabove1.
7-foldregulated.
TermiteVitaminsAnaerobicenvironmentH2Acetate,Amino-Acids,VitaminspolysaccharidesT.
primitiaT.
azotonutriciumAcetateAmino-AcidsFigure4SchematicrepresentationofthesymbiosisbetweenT.
primitiaandT.
azotonutriciumandthetermitehost.
Co-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1139TheISMEJournalweremappedontothetwogenomesalsoprovidedinformationongenesandintergenictranscriptsthathadnotyetbeenpredicted.
Forexample,onscanningthetwogenomesforintergeniclociwithhighgeneexpression,wewereabletoidentifysevenhighlytranscribedintergenicregionsinthetwogenomes.
FiveofthesearebetweenpreviouslyannotatedgenesofT.
primitia(betweengenepairs175:176,2323:2324,2456:2457,2870:2871and3885:3886).
TheothertwointergenictranscriptswereinT.
azotonutricium(betweengenepairs486:487and953:954).
ThehightranscriptlevelsofthesegenesaredisplayedinSupplementaryTable4,andaregenerallyhigherthanthoseofmRNAsfromthepredicted,annotatedgenes.
Uponfurtheranaly-siswewereabletopredictaputativeroleforthreeoftheseintergenics.
Allthreeofthesenewlyannotatedgenesappeartoberibozymes,withRnasePgenesbeingfoundinbothgenomesandwhatappearstobetmRNAinthegenomeofT.
primitia.
Thereremainsaninherentdifficultyinthein-silicoidentificationofgenesbasedonpredictionsofRNAfolding(asopposedtoproposingcanonicalopenreadingframesafterscanningallpossibletranslationread-ing-frames).
Thus,RNA-Seqprovidesawet-benchexperimentalmethodtoaidintheidentificationofnon-canonicalopenreadingframes.
Dataontheexpressionofoneofthesegenes,thetmRNAfromthegenomeofT.
primitia,areprovidedinFigure5a.
Apartfromintergenicsequences,wehavescruti-nizedgraphicaldisplaysoftranscriptstoidentifyatypicaltranscriptionpatternsinthetwogenomes.
Aninterestingexampleofanatypicaltranscriptpatternisagenethathadoverlyabundanttran-scriptstowardsthegenecenter,asopposedtothetwoends(Figure5b).
Inordertodeterminewhichtranscriptsforthisregionofthegenewereinthecodingdirectionornoncoding(antisense)direction,weusedqRT-PCRprimersspecificforeitherdirec-tion,targetingdifferentareasofthegene.
AsshowninFigure5b(andinset),whereasthecodingtranscriptsarerelativelyconstantthroughoutthegene,alargeamountoftranscriptsintheantisensedirectionarepresenttowardthegenecenter.
Thishighlytranscribedanti-senseregioncontainssixevenlyspaceddirectrepeatsofunknownfunction,butwhichmayhelptoexplaintheperiodicityoftranscriptdepthobservedinthecentralportionofthegene.
ThisfindingissimilartothosedetailedinarecentpublicationthatusesavariationoftheRNA-seqprotocoltocomposeagenome-widemapofbothcodingandantisensetranscripts(Dornenburgetal.
,2010).
ConclusionsThisworkrevealsthat,whengrowntogether,twotermite-gutTreponemaspeciesinfluenceeachother'sgeneexpressioninafarmorecompre-hensiveandnuancedmannerthanmighthavebeenpredictedbasedontheresultsofpreviousstudiesontherespectivepurecultures.
AlthoughH2-basedinteractionsarepredictedbyknownphysiologies,usingnewtechniqueswefindadivisionoflaborbetweenthebacteria,anduncoverunforeseensymbioticinteractions.
Ontheroadtounderstand-ingtheevenmorecomplexinteractionsthatoccuramongthehundredsofmicrobialspeciesresidingintermiteandothergutenvironments,webelievethatstudiesofdefinedmicrobialconsortiaofrepresen-tativestrainisolatesbecomeapromisingavenuetowardgainingabetterunderstandingofsuchsystems.
Despitetheinherentdifficultiesincompar-ingtranscriptsignalsfromdifferentorganismsandco-cultures(asinthisstudy),weexpectthatdeepRNAsequencingwillenableandexpeditesuch170017451351749601TREPR2458abTREPR2457TREPR24562455Genomelocation3476953482356TREAZ27172718GenomelocationTREAZ2716TREAZ271925MappedreadsFluorescence(Arbitraryunits)CodingNon-codingMappedreadsFigure5Transcriptionmappingofintergenicandatypicaltranscripts.
(a)AgraphicalrepresentationoftranscriptsmappedontoanannotatedportionoftheT.
primitiagenomecontainingoneofseveralidentifiedintergenictranscriptsinthisstudy.
ThesequenceissimilartothatofatmRNA.
Anotatedgenesaredisplayedasblockarrowsbelowthemappedreads.
(b)AnatypicaltranscriptpatterninaputativeglycohydrolasegeneofT.
azotonutricium(TREAZ2717,shadedblockarrow)showsahightranscriptdensityinthegenecenterwithrepetitivepattern.
DatafromqRT-PCRexperimentsdifferentiatingcoding(blackbars)fromantisense(whitebars)areshownintheinsetbox,andcorrespondtothesectionofthegeneanalyzed.
qRT-PCRwasperformedwithdirection-specificprimersintheamino-terminaldomain(NTD),centralportion(MID),andcarboxy-terminaldomain(CTD)ofthegene.
Co-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1140TheISMEJournalstudies,revealinggenesandinteractionsunder-lyingtheassociations,andaidinginthedevelop-mentofmanynewhypothesesandnewdirectionsofresearch.
ConflictofinterestTheauthorsdeclarenoconflictofinterest.
AcknowledgementsWethankourlaboratorycolleaguesandparticipantsinCaltech'sGECfacilityfortheirinsightsandadvice.
ThisresearchwassupportedbytheDOE(DE-FG02-07ER64484)andtheNSF(EF-0523267).
GenomesequencesforthetwobacterialspeciesaredepositedwithGenbankunderthefollowingaccessionnumbers:(CP001843)and(CP001841).
ReferencesAbeT,BignellDE,HigashiM.
(2000).
Termites:Evolution,Sociality,Symbioses,Ecology.
KluwerAcademicPublishers:DordrechtandBoston.
AkmanL,YamashitaA,WatanabeH,OshimaK,ShibaT,HattoriMetal.
(2002).
Genomesequenceoftheendocellularobligatesymbiontoftsetseflies,Wigglesworthiaglossinidia.
NatGenet32:402–407.
BooneDR,CastenholzRW,GarrityGM.
(2001).
Bergey'sManualofSystematicBacteriology.
GarrityGM(ed.
),2ndedn.
Springer:NewYork.
BraumanA,KaneMD,LabatM,BreznakJA.
(1992).
Genesisofacetateandmethanebygutbacteriaofnutritionallydiversetermites.
Science257:1384–1387.
BreznakJA,SwitzerJM.
(1986).
AcetatesynthesisfromH(2)plusCO(2)bytermitegutmicrobes.
ApplEnvironMicrobiol52:623–630.
BreznakJA.
(2002).
Phylogeneticdiversityandphysiologyoftermitegutspirochetes.
IntegrCompBiol42:313–318.
ChangDE,SmalleyDJ,TuckerDL,LeathamMP,NorrisWE,StevensonSJetal.
(2004).
CarbonnutritionofEscherichiacoliinthemouseintestine.
ProcNatlAcadSciUSA101:7427–7432.
DornenburgJE,DevitaAM,PalumboMJ,WadeJT.
(2010).
WidespreadantisensetranscriptioninEscherichiacoli.
MBio1:e00024-10.
DrakeHL.
(1994).
Acetogenesis.
Chapman&Hall:NewYork.
FraserCM,NorrisSJ,WeinstockGM,WhiteO,SuttonGG,DodsonRetal.
(1998).
CompletegenomesequenceofTreponemapallidum,thesyphilisspirochete.
Science281:375–388.
GilR,SilvaFJ,ZientzE,DelmotteF,Gonzalez-CandelasF,LatorreAetal.
(2003).
ThegenomesequenceofBlochmanniafloridanus:comparativeanalysisofreducedgenomes.
ProcNatlAcadSciUSA100:9388–9393.
GilbertJA,FieldD,HuangY,EdwardsR,LiW,GilnaPetal.
(2008).
Detectionoflargenumbersofnovelsequencesinthemetatranscriptomesofcomplexmarinemicrobialcommunities.
PLoSOne3:e3042.
GraberJR,BreznakJA.
(2004).
PhysiologyandnutritionofTreponemaprimitia,anH2/CO2-acetogenicspirochetefromtermitehindguts.
ApplEnvironMicrobiol70:1307–1314.
GraberJR,BreznakJA.
(2005).
Folatecross-feedingsupportssymbiotichomoacetogenicspirochetes.
ApplEnvironMicrobiol71:1883–1889.
GraberJR,LeadbetterJR,BreznakJA.
(2004).
DescriptionofTreponemaazotonutriciumsp.
nov.
andTreponemaprimitiasp.
nov.
,thefirstspirochetesisolatedfromtermiteguts.
ApplEnvironMicrobiol70:1315–1320.
HoltKE,ParkhillJ,MazzoniCJ,RoumagnacP,WeillFX,GoodheadIetal.
(2008).
High-throughputsequencingprovidesinsightsintogenomevariationandevolutioninSalmonellaTyphi.
NatGenet40:987–993.
HongohY,SharmaVK,PrakashT,NodaS,TaylorTD,KudoTetal.
(2008a).
CompletegenomeoftheunculturedTermiteGroup1bacteriainasinglehostprotistcell.
ProcNatlAcadSciUSA105:5555–5560.
HongohY,SharmaVK,PrakashT,NodaS,TohH,TaylorTDetal.
(2008b).
GenomeofanendosymbiontcouplingN2fixationtocellulolysiswithinprotistcellsintermitegut.
Science322:1108–1109.
JiH,JiangH,MaW,JohnsonDS,MyersRM,WongWH.
(2008).
AnintegratedsoftwaresystemforanalyzingChIP-chipandChIP-seqdata.
NatBiotechnol26:1293–1300.
LamHM,WinklerME.
(1990).
Metabolicrelationshipsbetweenpyridoxine(vitaminB6)andserinebiosynth-esisinEscherichiacoliK-12.
JBacteriol172:6518–6528.
LeadbetterJR,SchmidtTM,GraberJR,BreznakJA.
(1999).
AcetogenesisfromH2plusCO2byspirochetesfromtermiteguts.
Science283:686–689.
LilburnTG,KimKS,OstromNE,ByzekKR,LeadbetterJR,BreznakJA.
(2001).
Nitrogenfixationbysymbioticandfree-livingspirochetes.
Science292:2495–2498.
MatsonEG,ZhangX,LeadbetterJR.
(2010).
Seleniumcontrolstranscriptionofparalogousformatedehydro-genasegenesinthetermitegutacetogen,Treponemaprimitia.
EnvironMicrobiol12:2245–2258.
MoranNA.
(2006).
Symbiosis.
CurrBiol16:R866–R871.
NakabachiA,YamashitaA,TohH,IshikawaH,DunbarHE,MoranNAetal.
(2006).
The160-kilobasegenomeofthebacterialendosymbiontCarsonella.
Science314:267.
NeidhardtFC,CurtissR.
(1996).
EscherichiacoliandSalmonella:CellularandMolecularBiology,2ndedn.
ASMPress:Washington,DC.
OdelsonDA,BreznakJA.
(1983).
Volatilefattyacidproductionbythehindgutmicrobiotaofxylophagoustermites.
ApplEnvironMicrobiol45:1602–1613.
PalmerKL,AyeLM,WhiteleyM.
(2007).
NutritionalcuescontrolPseudomonasaeruginosamulticellularbehaviorincysticfibrosissputum.
JBacteriol189:8079–8087.
PesterM,BruneA.
(2007).
Hydrogenisthecentralfreeintermediateduringlignocellulosedegradationbytermitegutsymbionts.
ISMEJ1:551–565.
Saleh-LakhaS,MillerM,CampbellRG,SchneiderK,ElahimaneshP,HartMMetal.
(2005).
Microbialgeneexpressioninsoil:methods,applicationsandchal-lenges.
JMicrobiolMethods63:1–19.
SeshadriR,MyersGS,TettelinH,EisenJA,HeidelbergJF,DodsonRJetal.
(2004).
ComparisonofthegenomeofCo-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1141TheISMEJournaltheoralpathogenTreponemadenticolawithotherspirochetegenomes.
ProcNatlAcadSciUSA101:5646–5651.
ShigenobuS,WatanabeH,HattoriM,SakakiY,IshikawaH.
(2000).
GenomesequenceoftheendocellularbacterialsymbiontofaphidsBuchneraspAPS.
Nature407:81–86.
StraightPD,FischbachMA,WalshCT,RudnerDZ,KolterR.
(2007).
AsingularenzymaticmegacomplexfromBacillussubtilis.
ProcNatlAcadSciUSA104:305–310.
UrichT,LanzenA,QiJ,HusonDH,SchleperC,SchusterSC.
(2008).
Simultaneousassessmentofsoilmicrobialcommunitystructureandfunctionthroughanalysisofthemeta-transcriptome.
PLoSONE3:e2527.
WalkerCB,StolyarS,ChivianD,PinelN,GabsterJA,DehalPSetal.
(2009).
ContributionofmobilegeneticelementstoDesulfovibriovulgarisgenomeplasticity.
EnvironMicrobiol11:2244–2252.
Yoder-HimesDR,ChainPS,ZhuY,WurtzelO,RubinEM,TiedjeJMetal.
(2009).
MappingtheBurkholderiacenocepacianicheresponseviahigh-through-putsequencing.
ProcNatlAcadSciUSA106:3976–3981.
SupplementaryInformationaccompaniesthepaperonTheISMEJournalwebsite(http://www.
nature.
com/ismej)Co-cultureofsymbiotictermite-guttreponemesAZRosenthaletal1142TheISMEJournal