abbr_fa154cf2cc2bbb52d86f0f62b97e8020

IntestinalBacteroidesSpeciesDeterminesDiet-InducedMicrobiotaAlterations

EricaD.Sonnenburg,1,3HongjunZheng,2,3PayalJoglekar,1StevenK.Higginbottom,1SusanJ.Firbank,2DavidN.Bolam,2,*andJustinL.Sonnenburg1,*

ofMicrobiologyandImmunology,StanfordUniversitySchoolofMedicine,Stanford,CA94305,USA

forCellandMolecularBiosciences,NewcastleUniversity,MedicalSchool,NewcastleuponTyne,NE24HH,UK

3Theseauthorscontributedequallytothiswork

*Correspondence:d.n.bolam@ncl.ac.uk(D.N.B.),jsonnenburg@stanford.edu(J.L.S.)DOI10.1016/j.cell.2010.05.005

2Institute1Department

SUMMARY

Theintestinalmicrobiotaimpactsmanyfacetsofhumanhealthandisassociatedwithhumandiseases.Dietimpactsmicrobiotacomposition,yetmecha-nismsthatlinkdietarychangestomicrobiotaalter-ationsremainill-defined.HereweelucidatethebasisofBacteroidesproliferationinresponsetofructans,aclassoffructose-baseddietarypolysaccharides.Structuralandgeneticanalysisdisclosedafructose-binding,hybridtwo-componentsignalingsensorthatcontrolsthefructanutilizationlocusinBacteroidesthetaiotaomicron.GenecontentofthislocusdiffersamongBacteroidesspeciesanddictatesthespeci-ficityandbreadthofutilizablefructans.BT1760,anextracellularb2-6endo-fructanase,distinguishesB.thetaiotaomicrongeneticallyandfunctionally,andenablestheuseoftheb2-6-linkedfructanlevan.ThegeneticandfunctionaldifferencesbetweenBacter-oidesspeciesarepredictiveofinvivocompetitivenessinthepresenceofdietaryfructans.Genesequencesthatdistinguishspecies’metaboliccapacityserveaspotentialbiomarkersinmicrobiomicdatasetstoenablerationalmanipulationofthemicrobiotaviadiet.

INTRODUCTION

Thetrillionsofmicrobialcellsthatresidewithintheintestineshapeaspectsofhostmetabolismandimmunefunctionandextendthephysiologicaldefinitionofhumans(Backhedetal.,2005;Hooper,2009;Louisetal.,2007).Whilethegeneralcompo-sitionoftheintestinalmicrobiotaissimilarinmosthealthypeople,withgreaterthan90%ofthecellsbelongingtotheFirmicutesorBacteroidetesphyla(Dethlefsenetal.,2008),thespeciescompo-sitionishighlypersonalized(Turnbaughetal.,2009).

Communitymembershipandfunctionofthemicrobiotacanchangeduetonumerousvariablesincludingantibiotictreat-ment,inflammation,orchangesindiet(Dethlefsenetal.,2008;Franketal.,2007;Jernbergetal.,2007;Leyetal.,2006).Pro-

tractedlossofthetypicalcompositionhasbeenassociatedwithseveraldisordersincludinginflammatoryboweldiseases(Franketal.,2007).Inaddition,changesincompositionhavebeenassociatedwithobesityandweightloss;however,factorsthatcausethesechangesarenotwelldefined(Duncanetal.,2008;Leyetal.,2006).Thealterationsincommunitymembership,whetherchronicorshort-term,areaccompaniedbychangesinthemicrobiota’scollectivegenome,ormicro-biome,andthepatternsandmetaboliccapabilitiesitspecifies(Turnbaughetal.,2009).Therefore,themechanismsthatlinkrelevantvariables,suchaschangesindiet,tochangesinthemi-crobiome,areintegraltounderstandinghowenvironmentalfactorsandbehaviorinfluencehumanbiology.

Manycomplexplantpolysaccharidesinthehumandietareresistanttohost-mediateddegradationduetoeitherinsolubilityorlackofhuman-encodedhydrolyticenzymes(Flintetal.,2008;Louisetal.,2007;Sonnenburgetal.,2005).Thesecarbohy-dratesarenotabsorbedintheuppergastrointestinaltractandserveasamajorsourceofcarbonandenergyforthedistalgutmicrobialcommunity.Polysaccharidedegradationisoneofthecorefunctionsencodedinthemicrobiome(Lozuponeetal.,2008;Turnbaughetal.,2007).Broadexpansionofthegenesandoperonsdedicatedtodegradingandconsumingpolysac-charideshasoccurredwithinthegenomesofmicrobiota-residentspecies(Xuetal.,2003,2007),alogicaloutcomeoftheintensecompetitionfortheseresources.Itis,therefore,expectedthatalterationsinthetypeandquantityofpolysaccha-ridesconsumedcanresultinchangesinthemicrobiotacommu-nitycompositionandfunction.

Inulin-andlevan-typefructans(homopolymersofb2-1orb2-6fructoseunits,respectively)arecommondietaryplantpolysac-charidesthatfeedtheintestinalmicrobiota(Roberfroidetal.,1993).Multiplebacterialtaxainthegututilizefructans,includingmembersofFirmicutes,Bacteroides,andBifidobacterium,(Duncanetal.,2003;Rossietal.,2005;VanderMeulenetal.,2006),anddietaryfructancanresultinexpansionofActinobac-teria,Firmicutes,orBacteroides(Kolidaetal.,2007;Menneetal.,2000;Ramirez-Fariasetal.,2008).Lackofpredictabilityinhowthemicrobiotarespondstosuchdietaryinterventionsreflectsourlimitedunderstandingofnutrientsensingandutilizationbymembersoftheintestinalmicrobiota.

Cell141,1241–1252,June25,2010ª2010ElsevierInc.1241

A

Bt1754SusGSusFSusESusDSusCSusBSusASusRBt1757Bt1758Bt1759Bt1760Bt1761Bt1762Bt1763Bt1765Figure1.Bt’sUseofFructose-ContainingCarbohydratesCorrespondstoInductionofthePolysaccharideUtilizationLocusBT1757-1763andBT1765

(A)GenomicorganizationofBt’sSuslocus(top)andputativefructanutilizationlocus(bottom).Genesofsimilarfunctionarecodedbycolor;inter-veningunrelatedgenesarewhite;geneswithoutcorrespondinghomologsaregray.

(B)Geneexpressionpatternsofdifferentiallyregu-latedsusCandsusDhomologsfromBtgrowninrichmedium(TYG)atfivetimepointsfromearlylog(3.5hr)tostationaryphase(8.8hr)induplicate.Colorsindicatestandarddeviationsabove(red)andbelow(green)agene’smeanexpression(black).(C)GrowthcurvesofBtinminimalmediumcon-tainingindicatedcarbonsourceat0.5%w/v.FOS,fructo-oligosaccharide.

(D)RNAabundanceforgenesrelevanttofructanuseincellsgrownindifferentcarbonsources,rela-tivetogrowthinminimalmediumplusglucose.Standarderrorsofexpressionlevelsfromthreebiologicalreplicateculturesareshown.

HTCSsensor/regulatorfructokinaseSusD-likeSusC-likeGH32 glycosideOuter membranehydrolasesSusEpolysaccharideInner membranepositioned

binding andmonosaccharide

gene

import proteinsimport

GH32glycosidehydrolaseB

3.5

Time (hours)

4.5

5.5

6.5

8.8

BT0317-BT0319BT0483/BT0484BT1042/BT1043BT1280/BT1281BT1619/BT1620BT1762/BT1763BT2460/BT2461BT2559/BT2560BT2625/BT2626BT2805/BT2806BT3310/BT3311BT3788/BT3789BT3854/BT3855BT3958/BT3959

C

Absorbance (OD 600)1.510.500

24

Time (hours)

48

FructoseSucroseFOSLevanInulin

-2-1012

D

Fold induction100000100001000100101

BT1754BT1757BT1763BT1765BT3082

Bacteroides,amajorgenerainthehumanmicrobiota,haveawidelyexpandedcapacitytousediversetypesofdietarypoly-saccharides(Xuetal.,2007).MuchoftheglycandegradingandimportmachinerywithinBacteroidesgenomesareencodedwithinclustersofcoregulatedgenesknownaspolysaccharideutilizationloci(PULs).B.thetaiotaomicron(Bt),aprototypicmemberoftheBacteroides,possesses88PULs,whichdifferinpolysaccharidespecificity(Martensetal.,2008).ThedefiningcharacteristicofaPUListhepresenceofapairofgeneshomol-ogoustoBtsusDandsusC,whichencodeoutermembraneproteinsthatbindandimportstarcholigosaccharides,respec-tively(Figure1A)(Martensetal.,2009;Shipmanetal.,2000).ThepairofsusCandsusDhomologsisusuallyassociatedwithgenesthatencodethemachinerynecessarytoconvertextracel-lularpolysaccharidesintointracellularmonosaccharides,suchasglycosidehydrolases(susA,susB,andsusGinFigure1A).Inadditiontomachineryforpolysaccharideacquisition,mostPULscontain,orarecloselylinkedto,ageneorgenesencod-inganinnermembrane-associatedsensor-regulatorsystem,includingthenovelhybridtwo-componentsystems(HTCS)(Sonnenburgetal.,2006).Bt’sgenomeencodes32ofthese

1242Cell141,1241–1252,June25,2010ª2010ElsevierInc.

HTCS,whichmaymediatetherapidandspecificresponsesrequiredinthedynamicnutrientenvironmentoftheintes-tine.Here,wedissectaBtPULrequiredforutilizationoffructanstobetterunder-standhowBacteroidesspeciesacquire

Fructose

andprocessthiscommonclassofdietary

Sucrose

carbohydrates.Inaddition,weprovideFOS

evidencethattheassociatedHTCScon-LevanInulintrolstheexpressionofthefructanPUL

andthatmonomericfructoseistheacti-vatingsignalthatbindsdirectlytotheperiplasmicsensordomainoftheregula-toryprotein.Thesedataprovidean

exampleofawell-definedligandforamemberofthisclassofsensorregulators.

ThefructanPULisconservedtovaryingextentsamongBacter-oidesspecies,correspondingtoarangeoffructanutilizationcapabilityacrossthegenus.Usingmodelintestinalmicrobiotaslivingwithingnotobioticmice,wedemonstratethatdietaryfructancanhavedisparateeffectsoncommunitycomposition,depend-inguponthefructandegradingcapacityofmembersofthemicro-biota.Thesestudiessuggestthatwithinpersonalmicrobiomicdatasets,wewillbeabletoidentifygeneticbiomarkersofdiscretefunctions.Inferenceoffunctionfromthesebiomarkersshouldprovidepredictivepowerindetermininghowanindividual’smi-crobiotawillrespondtochangesindietandotherinterventions.RESULTS

BT1757-BT1763andBT1765FormaPutative

PolysaccharideUtilizationLocusthatIsTranscribedEarlyinBt’sGrowthinRichMedia

BT1757-BT1763andBT1765encodeseightopenreadingframesonthenegativestrandoftheBtgenome,includingone

susC/susDhomologpair(BT1763andBT1762),aputativeoutermembranelipoprotein(BT1761),aputativeinnermembranemonosaccharideimporter(BT1758),aputativefructokinase(BT1757),andthreeputativeglycosidehydrolases(BT1759,BT1760,BT1765)(Figure1A).TheseglycosidehydrolasesaremembersofGlycosideHydrolaseFamily32(GH32),afamilyofenzymesspecificforfructans(Cantareletal.,2009).Oneofthese,BT1760,possessesaN-terminallipidationmotifandispredictedtoresideonthecellsurface;theothertwo,BT1759andBT1765,arepredictedtobeperiplasmicandintracellular,respectively(www.cbs.dtu.dk/services/LipoP/andwww.cbs.dtu.dk/services/SignalP/).Directlyadjacenttothelocusisaputativeinnermembrane-associatedsensorregulatoroftheHTCSfamily,BT1754.ThesedatasuggestthatthisPULencodestheproteinsrequiredforBt’suseoffructans.

ExpressionprofilingofBtinrichmediumhasrevealedtheupregulationofseveralPULs,eachofwhichisconfinedtoadiscretephaseofgrowth(Sonnenburgetal.,2006).AnalysisofBttranscriptionalprofilesatfivetimepointsthatspannedfromearlylogtostationaryphaseinvitroinrichmedium,comparedtobasalexpressioninminimalmedium(MM)contain-ingglucoseasthesolecarbohydrate,revealedthat14pairsofsusC/susDhomologswereinducedgreaterthan20-foldatoneormoretimepointsduringthegrowth(Figure1B)(GeneExpres-sionOmnibusdatabase,www.ncbi.nlm.nih.gov/geo/;accessionnumbers,GSM40897–40926).TheputativefructanPULshowedupregulationearlyinBt’sgrowthsuggestingitisresponsivetoahighprioritysubstrateaccessedearlyingrowthonrichmedium(Figure1B).GeneswithinthisPULarecoexpressedbothinvitroinrichmediumandinvivoinBtmono-associatedgnotobioticmicefedapolysaccharide-richdiet(FigureS1Aavailableonline),consistentwiththefunctionalrelatednessofadjacentgenesandoperonpredictionsinBt(Westoveretal.,2005).BtincreasesexpressionofthisPULinvivowhiledownregulatingthevastmajorityofotherPULswhenbi-associatedinthegnotobioticmouseintestinewiththemethanogenicarcheon,Methanobrevi-bactersmithii(SamuelandGordon,2006).TheupregulationoftheputativefructanPULisconcomitantwithincreaseddensitiesofBtinvivo,suggestingthatexpressionofthislocusisassoci-atedwithgrowthpotentiationofBt.

BtUpregulatesItsPutativeFructanPULWhenGrownonFructose-ContainingCarbohydrates

Weinoculatedminimalmediumcontainingspecificfructose-basedcarbohydratesastheonlycarbonandenergysourcewithBttotestifthebacteriumiscompetenttogrowonfructans.Btgrewonabroadrangeoffructose-basedglycans,includingfreefructose,sucrose,levan(highMWfructosepolymerwithpredominantlyb2-6-linkages),andfructo-oligosaccharides(FOS;short-chainb2-1polymersof2–10fructoseunits)(Figure1C;seeFigureS2forcarbohydratestructures).However,Btgrewpoorlyoninulin(b2-1fructosepolymerwithanaveragedegreeofpolymerizationof$25),withgrowthonlyapparentthreedaysafterinoculation.Doublingtimesonsimplemono-saccharidesanddisaccharideweresimilartooneanother(TableS1).Incontrast,growthratesofBtbetweenthedifferentfructansshowedlargelinkage-dependentdifferences:b2-6levanresultedinthefastestdoublingtime(2.7hr),whileb2-1

FOSandinulinweresignificantlyslower(doublingtimesof5.6hrand96.4hr,respectively)(TableS1).

Todeterminewhetherthesefructose-basedsubstratesinducedexpressionofgenesassociatedwiththeputativefruc-tanPUL,Btwasgrownineitherglucoseoroneoffivefruc-tose-containingsubstrates(fructose,sucrose,levan,FOS,orinulin)asthesolecarbohydrate.Cellswereharvestedatmid-logphaseforquantitativeRT-PCR(qPCR)analysis,andRNAlevelsofthe30andthe50endsoftheoperon,BT1757(encodingthefructokinase)andBT1763(encodingtheSusC-likeprotein),respectively,wereusedasanindicatorofPULexpression(Figure1D).BothBT1757andBT1763weredramaticallyupregu-latedinallmediacontainingfructose,whetherasafreemono-saccharideoringlycosidiclinkage.Acrossallconditions,expressionofBT1757,BT1763,andBT1765showedcoordi-natedincreasesconsistentwiththepredictedoperonstructure.However,BT1754(thePUL-associatedputativeHTCS)showednosignificantinductionunderallconditionstested.Therefore,theoperonthatencodesthestructuralgenesofBt’sputativefructanPUListranscriptionallyresponsivetofructose-contain-ingcarbohydrates.PublishedsurveysofBtgeneexpressioninnumerouscarbohydratessupportthatupregulationofthefruc-tanPULisspecifictofructose-containingsubstrates(Martensetal.,2009;Sonnenburgetal.,2005).

TwogeneswithinBt’sgenomethatarenotphysicallyassoci-atedwiththeputativefructanPUL,asecondputativeperiplas-micGH32(BT3082)andasecondputativefructokinase(BT3305),werelikelycandidatestobeinvolvedinfructanutiliza-tion.AnalysisofBT3082andBT3305expressionbyqPCRrevealedthatBT3082wasinducedinallfructose-containingmediaandshowedapatternofinductionconsistentwiththoseseenforBT1757,BT1763,andBT1765(Figure1D);however,BT3305showednochangeinexpressionoraslightlyreducedexpressioninallconditions(datanotshown).Thesedatasug-gestthatthefructosidase,BT3082,butnottheputativefructoki-nase,BT3305,ispartoftheregulonoftheputativefructanPUL.TheHybridTwo-ComponentSystemBT1754IsRequiredforEfficientFructanUtilizationbyBt

WeassessedtheabilityofanisogenicmutantofBtlackingtheBT1754genetogrowinapaneloffructose-basedminimalmediatotestifupregulationofthePULwasdependentupontheHTCSsignalingsensor.Anin-frame,unmarkeddeletionofBT1754wasconstructedusingastandardcounter-selectableallele-exchangeprocedure.Bt-DBT1754exhibitednormalcolonymorphologyonsolidmediumandgrewwithasimilardoublingtimetowild-typeinMM-glucose(2.6hr);however,Bt-DBT1754failedtogrowinanyofthethreefructans(FOS,inulinandlevan)andshowedretardedgrowthinfructoseandsucrose(Figure2AandTableS1).Additionally,Bt-DBT1754doesnotexhibitprioritizedupregulationoftheputativefructanPULduringgrowthinrichmedia(FigureS1B).Complementationofthismutantwasachievedbyintroducingthegenomicfragmentcon-tainingBT1754andits50intergenicupstreampromoterregionintrans.GrowthoftheDBT1754::BT1754complementedmutantrestoredgrowthinallfructose-basedmediatolevelscomparabletowild-type(Figure2AandTableS1).ThesedatademonstratetheHTCSencodedbyBT1754isrequiredforBt’suseoffructans.

Cell141,1241–1252,June25,2010ª2010ElsevierInc.1243

A

Absorbance (OD 600)GlucoseFructose

1.510.50

24

48

Sucrose

1.510.50

0

24

48

0

FOS

10.50

24

48

Levan

1.51.5

11

0.50.5

00

0!!\"!!\"!!#0!!\"!!\"!!#24!!\"!!\"!!#48

WTBT1754BT1754::BT1754Figure2.BT1754HTCSBindsFructoseandIsRequiredforGrowthonFructose-ContainingCarbohydrates

(A)GrowthcurvesofBt-DBT1754comparedtowild-typeBt(WT)andthecomplementedmutant(DBT1754::BT1754)onfructose-basedcarbonsources.(B)DomainorganizationofBT1754.

(C)InteractionoftheN-terminalperiplasmicdomainofBT1754(BT1754-PD)withfructoseorlevanbioseassessedbyisothermalcalorimetry,showingtherawinjectionheats(upperpanel)andintegrateddata(lowerpanel)fittoasinglesitebindingmodel(fructoseonly).

ValuesareaveragesandSDsofthreeindependenttitrations.

02448

Time (hours)

B

periplasmic sensorNtransmembranephospho-acceptorhistidine kinaseAraC-type

receiverDNA bindingCtransmembranecytoplasmicC

0.0Time (min)0 20 40 60 80 100 1200.0 0.0Time (min)Time (min)0102030405060700 10 20 30 40 50 60 70µcal/sec -0.5µcal/sec-0.1 -0.1 -1.0 -1.5rimetrydatarevealthatBT1754-PDbinds

specificallytofructose,withaKdof$2mMandastoichiometryof1:1,butdoesnotinteractwitheitherb2-1-orb2-6-linkedfructooligosac-charidesoranyothermonosaccharides,includingglucoseandribose(Figure2C).

StructureofBT1754PeriplasmicSensorDomain

Tounderstandthemechanismofsignalpercep-tioninmoredetail,wedeterminedthestructureofBT1754-PDincomplexwithfructoseto

˚.Theclosesthomologwithknownstruc-2.66A

ture,aribose-bindingPBPfromThermoanaero-bactertengcongensis(TtRBP),PDB2IOY,wasusedasamolecularreplacementsearchmodel.Successfulmolecularreplacementresultedinadimerintheasymmetricunit.Aleast-squaresalignmentofthefinalmodelwithTtRBPgave

˚for269arootmeansquaredeviationof1.2A

alphacarbonsdespitetherelativelylowsequenceidentity,indicativeofthehighstructuralconservationofthisfamily.TheBT1754-PDstructurecomprisesatypicaltwo-subdomainPBP-fold,witheachsubdomainconsistingofacoreofsixbstrandsflankedbytwoorthreeahelices(Figure3A).Thepoly-peptidechainformsahingebycrossingbetweenthetwosubdomainsthreetimesalongoneside,thelastoftheseexitingthePBP-foldandthenformingalonga-helix,whichextendsbackalongthelengthoftheproteintotheN-terminalregion(Figure3A).

TheC-terminalhelixofBT1754-PDprovidesthepredominantinterfaceforhomo-dimerizationandisthemainstructuraldiffer-encebetweenclassicalsolublePBPssuchastheTtRBPandBT1754-PD(Figure3B).ThoughthereareseveralhydrogenbondstoretaintheturnbetweenthePBP-foldandthehelix,oncethepolypeptidehasprogressedbeyondthefirstresidueofthehelix(Asn306),theremainderofthecontacts,bothinter-andintramolecular,arenonpolar.Thedimer,generatingaburied

˚2,appearstobebiologicallyrelevantassurfaceareaof2640A

boththeN-andC-terminiofeachmoleculeareorientedsuchthattheyfaceinthesamedirectionand,therefore,bothmole-culesarepositionedcorrectlyforinsertionintothemembrane(Figure3A).

cal/sec-0.2 -0.2Levanbiosekcal/mole of injectant -2.0 -2.5 0 -2 -4 -6 -8 -10 -12 -14 -16Fructose-0.05 -0.05-0.10-0.10-0.15-0.15kcal/mole of injectant-0.20 -0.20kcal/mole of injectant0.00 0.00024681012140 2 4 6 8 10 12 14Molar RatioMolar Ratio0.0 0.5 1.0 1.5 2.0 Molar RatioKa = 4.6 ±1.1 105 M-1G = -7.7 ±0.2 kcal.mol-1H = -15.4 ±0.8 kcal.mol-1TS = -7.7 ±1.0 kcal.mol-1n = 0.9 ±0.3ThePeriplasmicDomainoftheHybridTwo-ComponentSystemBT1754BindstoMonomericFructose

OneofthekeyunansweredquestionsconcerningtheHTCSfamily,andmanyotherextracellularsensorysystems,istheidentityofthemoleculartriggersforsignalingevents.Thepre-dictedinner-membranelocalizationofBt’sHTCSfamilymembers,includingBT1754,suggeststhattheperiplasmicregionlikelyservesasthesensor/receptor,similartoclassictwo-componentsystems.AnalysisofthesequenceofBT1754revealedatypicalHTCSarchitecturewithanN-terminalpre-dictedperiplasmicsensordomainflankedbytwotransmem-braneregionsandaC-terminalcytoplasmichistidinekinasedomain,aphosphoacceptordomainandaresponseregulator(includingareceiverandanHTH_AraC-typeDNAbindingdomain)(Figure2BandFigureS3).UniquelywithinBt’sHTCS,thesensordomaindisplayshomologytoTypeIbacterialperi-plasmicbindingproteins(PBPs)(DwyerandHellinga,2004).AsPBPsareknowntobindsmallmoleculessuchassugars,weex-pressedtheperiplasmicdomainofBT1754(BT1754-PD;resi-dues29–343)inarecombinantformandtestedforbindingtoarangeofmonosaccharidesandfructan-derivedoligosaccha-ridestoseeifdirectinteractionwithaspecificcarbohydrateisthemeansofsignalperceptioninBT1754.Theisothermalcalo-1244Cell141,1241–1252,June25,2010ª2010ElsevierInc.

AB

Figure3.StructureofComplexwithFructose

BT1754-PDin

N

C

C

Arg224

CN

D

Asn222

Trp196

Asp248

Arg172(A)RepresentationofthehomodimerofBT1754-PDpresentintheasymmetricunit,witheachmonomerseparatedbyadottedline;moleculeoffructose(pink);theflexiblehingebetweenthetwosubdomains(circle).

(B)OverlayofBT1754-PD(green)withTtRBP(blue);theextendedC-terminalhelixinBT1754-PD(bracket)isuniquetoBT1754.

(C)Sideviewofthebindingsiteillustratinghydro-phobicinteractionsofBT1754-PDandfructose.Fo-Fcelectrondensitypriortomodelingthesinglemoleculeoffructoseintheb-furanoseformisshown(bluemeshcontouredat3s).

(D)TopviewofthebindingsiteofBT1754-PDillus-tratingthenumerousH-bonds(dottedblacklines)withfructose.

Asp43

Arg224

Pro168

Trp45

Tyr271

Asp43

Crystalsweregrowninthepresenceoffructose,andelectrondensityindicativeofafructosemoleculeintheb-furanoseformwasobservedinthecleftbetweenthetwosubdomains,thetypicalbindingsiteofPBPfamilyproteins(DwyerandHellinga,2004)(Figure3A).Thesugarringissandwichedbetweentwotryptophanresidues,onefromeachsubdomain(Trp45andTrp196),withTyr271andPro168alsoforminghydrophobiccontactsalongtheC4-C6edgeofthefructosering(Figure3C).Allremaininginteractionswiththesugararepolar,withmultipleH-bondsformedbetweensidechains,mainlyArgandAsp,andthehyroxylsandringoxygenofthefructose(Figure3D).IncommonwithotherPBPs,solventisexcludedfromthebindingsiteitself.ThestructuralandbiochemicaldataforBT1754-PDbindingtofructoseareconsistentwiththeincreasedexpressionofthefructanPULobservedinminimalmediumcontainingonlyfructose(Figure1D).Furthermore,thestructuraldatasuggestthatsignaltransductioninBT1754isdrivenbyaconformationalchangeoftheperiplasmicdomainonfructosebindingthatistransmittedacrossthemembraneviaa‘‘piston-like’’movementoftheTMhelices,similartothatpostulatedforothersensorkinases(FalkeandErbse,2009).

GeneticandBiochemicalBasisofb2-6FructanSpecificityofBt

ThelackoflinkagerecognitionbytheHTCSsensorsuggestedthattheb2-6-linkagespecificityofBt’sfructanusewasencodedwithinthestructuralgenesofthefructanPUL.WefirstfocusedongenesencodingGH32enzymes,themainfamilyofhydrolasesthatcatalyzethedepolymerizationoffructans(Cantareletal.,2009).ThreeGH32enzymes(BT1759,BT1760,andBT1765)areencodedwithinthefructanPUL;theotherGH32familymember,BT3082,isnotencodedwithinthePUL,butiscoregulated(Figure1D).

TotestwhethertheonlyputativecellsurfaceGH32inBt,BT1760,isrequired

Asp120

forlevanutilization,anin-frame,un-markeddeletionofBT1760wascon-Arg121

structed.Bt-DBT1760exhibitednormalArg46

colonymorphologyonsolidmediumandgrewwithanormaldoublingtimein

MM-glucose.Bt-DBT1760didnotgrowonlevan,butshowednormalgrowthonallothermediatestedincludingb2-1-linkedFOS(Figure4A),withdoublingtimescomparabletowild-typeinfructose,sucrose,andFOS(TableS1).Complementationintransofthismutantwasachievedbyfusingtheupstreaminter-genicpromoterregionofBT1765tothe50endofthegenomicfragmentcontainingBT1760.Levangrowthwasrestored,albeitatareducedrate,inthecomplementedBt-DBT1760::BT1760strain(Figure4A),confirmingtherequirementofthisgycosidehydrolaseforutilizationoftheb2-6linkedfructan.

WenextassessedwhetherBT1760isab2-6-specificfructa-nase.ActivityofarecombinantformofBT1760wastestedagainstarangeofb2-6andb2-1fructanoligo-andpolysaccha-rides.ThedatashowthatBT1760isindeedab2-6-fructanspecificenzymewithnodetectableactivityagainstb2-1fructansorfructooligosaccharides(TableS2).TLCanalysisoflevandigestionbyBT1760revealedthatamixtureofdifferentsizedoligosaccharideswasproduced.Mono-,di-,tri-,andtetra-levanoligosaccharidesaccumulatedasthemainproductsasthereactionproceeded(Figure4B).ThesedatademonstratethatBT1760isab2-6-specificendo-actingfructanase.

Todeterminewhethertheb2-6fructosidehydrolaseactivityofBT1760couldbedetectedonthecellsurface,wemeasuredtheactivityofwashedwholeBtcellsagainstlevanandinulin.Fructose-grownwild-typecellscoulddegradetheb2-6polymerbuthadnodetectableactivityagainstinulin,mirroringthespec-ificityofrecombinantBT1760(Figure4Canddatanotshown).ThislevanaseactivitywascompletelylostintheBt-DBT1760strainandwaslargelyrestoredinthecomplementedBt-DBT1760::BT1760strain(Figure4C).Moreover,cellsgrownonglucosedisplayed$100-foldlowerlevanactivity,confirmingthatthelevan-specifichydrolysisisinduciblebyfructose(data

Cell141,1241–1252,June25,2010ª2010ElsevierInc.1245

A

Absorbance (OD 600)0.8

LevanWTBT1760BT1760::BT1760C

Frc equivalents (µg/ml) 400 300 200 100 0WTBT1762BT1760BT1760::BT1760Figure4.BT1760EncodesanExtracellularEndo-LevanaseRequiredforBtGrowthinLevan

(A)GrowthcurvesofBt-DBT1760comparedtothecomplementedmutant(DBT1760::BT1760)inle-van(top)orFOS(bottompanel).

(B)TLCanalysisoftheproductsoflevandigestionbytheBtGH32enzymes,BT1760,BT1759,BT1765,andBT3082.Frc,fructose;L2,levan-biose;L3,levantriose;L4,levantetraose.

(C)DegradationoflevanbyBtcellsgrowninminimalmediumplusfructose.

ErrorbarsshowtheSDsfromthreeindependentexperiments.

0.4

01.2)'0.80.400

FOS*('(+'0 1 2 3 4 5Time (hours)

24

48

Time (hours)

B

FrcL2L3

BT1760BT1759BT1765BT3082

Frc

RecentstudieshaveindicatedthatwithinmanyBacteroidesPULs,the

L4

genefounddownstreamofthesusDhomologalsoencodesapolysaccha-ride-bindinglipoprotein(Martensetal.,2009).Althoughtheproductsofthese‘‘susE-positioned’’geneshaveno0 1m 10m 1h 5h0 1m 10m 1h 5h0 1m 10m 1h 5h 0 1m 10m 1h 5h

obvioussequencehomologytooneTime

anothertheyappeartobefunctionallyconserved.Toexploretheroleofthe

notshown).CytoplasmicandperiplasmicmarkerenzymesusE-positionedgenefromtheBtfructanPUL,BT1761,weas-assaysdemonstratedthatnocelllysisorleakageoccurredinsessedtheabilityofarecombinantformoftheproteintointeracttheassayconditionsused;therefore,thehydrolaseactivitywithinulinandlevan.ThedatarevealthatBT1761boundspecif-detectedcouldonlybeextracellular(datanotshown).Theseicallytolevan(FigureS4).ReducingsugarandTLCassayswithdataindicateBT1760isindeedlocalizedonthesurfaceofBT1761andBT1762againstinulinandlevanrevealedthatthebacterium.Thelocalizationandactivityareconsistentwithneitherproteinhadanydetectabledegradativecapacity(datathehydrolaseservingasakeystepforconvertinglong-chainnotshown).Together,thesegeneticandbiochemicaldatalevanintooligosaccharidesforSusC/SusD-homolog-mediatedshowthatthecellsurfacecomponentsofBt’sfructanPUL

exhibitb2-6linkagespecificity.import.

StructuralinsightintothenatureofSusDandaSusDhomolog

bindingtooligosaccharidessuggeststhatlinkageisanimpor-BtHasThreeGH32EnzymesthatAreNotLinkagetantdeterminantincellsurfacestructuralrecognitionofoligo-Specific

saccharides(Koropatkinetal.,2008,2009).WetestedwhetherTounderstandthepatternoffructandegradationinBtinmoreBT1760wasthesolespecificitydeterminantinBt’sefficientdetailwebiochemicallycharacterizedthethreeotherGH32suseoflevan,orwhethertheSusDhomologwithinthefructanexpressedduringgrowthonfructose-containingmedia,thePUL,BT1762,alsoexhibitedspecificityfortheb2-6linkage.predictedperiplasmicBT1759andBT3082andthepredictedWeconstructedaBtmutantinwhichBT1762wasdeleted,intracellularBT1765.Thedatarevealedthatallthreeoftheseandwetestedtheabilityofthismutanttogrowinminimalmediaenzymesareexo-actingfructosidasesthatreleasefructoseinwhichlevanisthesolecarbonsource.Bt-DBT1762showedfrombothb2-1andb2-6fructans,althoughsomedifferencessignificantlyretardedgrowthonlevancomparedwithwild-typeintheirkineticcharacteristicswereobserved(Figure4BandandgrowthofthismutantinlevanwaslargelyrestoreduponTableS2).BT1759andBT3082actequallywelloninulinandBT1762complementation.(Figure5AandTableS1).Absencelevan,aswellasoligosaccharidesofthesepolymers,althoughofBT1762,however,didnotaffectextracellularlevandegrada-BT3082appearstobeoverallamoreefficientenzymewithtion,supportingthatBT1760isresponsibleforcellsurface$2-to4-foldhigherkcat/KMvaluesthanBT1759formostlevandegradation(Figure4C).Todeterminethespecificityofsubstrates,drivenmainlybyitshigherturnovernumber.Consid-BT1762directly,theproteinwasexpressedinarecombinanteringtheb2-6fructanpreferenceofBt,itisinterestingthatbothformlackingitssignalpeptideandlipidationmotif,anditsenzymesdisplaylowerKMvalues($2-to8-fold)forb2-1oligo-interactionwithlevanandinulinwasassessedbyisothermalsaccharidescomparedtotheirb2-6equivalents(TableS2).calorimetry(Figure5B).ThedatashowthatBT1762bindstoBT1759andBT3082alsocleavesucrose,atraitsharedwiththeb2-6fructosepolymerbutdisplaysnoaffinityfortheb2-1otherbacterialfructosidases,althoughbothhaveahigherKMequivalent.BT1762displaysaKdof$40mMforlevan,similarforthedisaccharidethanforlargerb2-1kesto-oligosaccharides.totheaffinityoftheprototypicSusDforcyclodextrins(Koropat-Bycontrast,BT1765muchpreferssucroseoveranyoftheother

oligo-orpolysaccharidestested,althoughtheenzymeisalsokinetal.,2008).

1246Cell141,1241–1252,June25,2010ª2010ElsevierInc.

A

Absorbance (OD 600)0.60.40.20

Levan (2-6 fructan)

1.20.80.40

FOS (2-1 fructooligosaccharides)

Figure5.TheSusD-HomologEncodedbyBT1762IsRequiredforEfficientBtUtilizatonofLevanandBindsb2-6butNotb2-1Fructan

(A)Growthcurvesofwild-typeBt,Bt-DBT1762,andBt-DBT1762::BT1762inlevan(left)orFOS(right).

(B)InteractionofBT1762withfructansasassessedbyisothermalcalorimetry.Levanbindingdataintegratedandfittoasinglesitebindingmodel(bottomleft).

ValuesareaveragesandSDsofthreeindependenttitrations.

WT

BT1762BT1762::BT1762#%$!!$!!\"0!\"Time (hours)

12#\"24#\"0!$!!$!!\"12

Time (hours)

!$!!$!!\"24

twoB.fragilisorthologs(BF4326)displayed

only36%identitywithBT1754-PD,andthisTime (min)Time (min)

domainwasuniqueinitslackoffullyconserved0 20 40 60 80 100 120 1400 20 40 60 80 100 120 140 160fructosebindingresidues(FigureS3).Regions 0.0

0.0

adjacenttotheHTCSineachgenomewere -0.5

-0.1

analyzedandfoundtodisplaylocalsynteny

-1.0 -0.2

withtheBtlocus(Figure6,leftpanel),including

-1.5 -0.3

thepresenceofopenreadingframesthat

-0.4 -2.0arepredictedtoplayaroleinutilizationof -0.5fructose-containingcarbohydrates.Inallsix 0 0.0

Bacteroidesspecies,theHTCSisadjacent

-0.1

toapredictedfructokinase,aputativeinner -2

-0.2membranemonosaccharideimporter,and

-4Ka = 2.3 ±0.1 1010 M -0.3GH32-familyglycosidehydrolases.Ineach G = -5.9 ±0.0 kcal.mol

H = -7.8 ±0.3 kcal.molgenome,exceptthatofB.vulgatus,thesyntenic -0.4TS = -1.9 ±0.3 kcal.mol -6

n = 1.0 ±0.0 regionsalsocontainasusC/susDhomologous -0.5

0 10 20 30 40 0.0 0.5 1.0 1.5 2.0 2.5 3.0pair.

Molar RatioMolar RatioThepresenceofanapparentfructanPULin

multipleBacteroidesspeciessuggestedthatfructanutilizationissharedbetweenmembers

abletoefficientlyhydrolyselevanbiose(TableS2).Thepredictedofthisgenus.Testingforgrowthonfructose-basedglycanscytoplasmiclocationofBT1765andsubstratespecificityrevealedthatallsixspeciesarecompetentforgrowthonfruc-suggestthatsomeofthedisaccharideproductsoflevan(andtose(Figure6,rightpanel),sucroseandFOS(TableS1).AllBac-possiblyFOS)digestionaretransportedacrosstheinnerteroidesspeciestested,exceptB.vulgatus,wereabletogrowmembranebeforetheyaredegradedbytheperiplasmicfructo-efficientlyusingoneofthelong-chainfructans,inulinorlevan.

TheinabilityofB.vulgatustogrowonlong-chainfructansissidases(seeFigureS5).

consistentwiththeabsenceofasusC/susD-likepairwithinitslocus.B.caccae,B.ovatus,B.fragilisandB.uniformiscanutilizeTheFructanPULIsVariablyConservedinSequenced

inulinwithefficiencysimilartotheiruseofglucose.ThiscontrastsBacteroides,whichHaveDifferingCapacitytoUtilize

withBtinulinuse,whichisonlyobservedafterthreedaysFructan

WeperformedacomparativegenomicanalysisfocusedonBt’s(Figure6).

Btistheonlyspeciestestedabletouselevan,whichwasfructanutilizationlocusbetweenfivesequencedspeciesof

Bacteroidestogainfurtherinsightintothemechanismoffructanparticularlystrikingwhenconsideringtheoverallsimilarityinuseforthismajorgroupofgutresidentmicrobes.UsingthePULstructurebetweenBt,B.caccae,andB.ovatus.However,N-terminalfructose-bindingdomainoftheHTCSBT1754toexaminationofPULgenecontentofthetwoinulin-utilizingqueryaBLASTdatabaseconsistingoftheBacteroidesspeciesspeciesrevealedgenesencodingPL19enzymes,afamilythatB.caccae,B.vulgatus,B.uniformis,B.fragilis,andB.ovatus,isknowntoincludememberscapableofdegradingtheb2-1wehaveidentifiedasingleorthologousHTCSineachspecies,fructan.Additionally,Bt’sextracellularb2-6-specificGH32,withtheexceptionofB.fragilis,whichharborstwoBT1754-likeBT1760,doesnotpossessanorthologousgeneintheothergenes.Sequenceidentitybetweentheperiplasmicsensorspecies(FigureS6).Notably,twoothersequencedBtstrainsdomainsoftheBT1754orthologswashighforallbutone,utilizelevanmoreefficientlythaninulininvitro(datanotshown),rangingfrom93%fortheB.ovatusproteinto58%fortheB.vul-similartothetypestrain.Bothofthesestrainspossessorthologsgatusdomain.Furthermore,theresiduesinvolvedinfructosetothetypestrain’sBT1760(FigureS6).TogetherthesedatabindinginBT1754arealmostcompletelyconservedamongdemonstratethatdifferencesinfructanspecificityofBacteroidesorthologs,consistentwithconservationoftheligandsensedbyspeciescorrespondtodifferencesinthegenecontentoftheireachHTCS(FigureS3).TheperiplasmicdomainofoneoftherespectivefructanPULs.

Levan (2-6 fructan)Inulin (2-1 fructan)

µcal/seckcal/mole of injectant4

-1-1-1-1

B

kcal/mole of injectantµcal/secCell141,1241–1252,June25,2010ª2010ElsevierInc.1247

Figure6.ComparativeGenomicandFunctionalAnalysisofFructanUtilizationamongBacteroidesSpecies

Fructan-utilizationlocifromBacteroidesspecies(left).Commonpredictedfunctionsarecolorcoded,interveningunrelatedgenesarewhite.PL19,polysaccha-ridelyasefamily19;GH32,glycosidehydrolasefamily32.Growthcurves(right)ofeachBacteroidesspeciesinfructose-basedcarbohydrates.

GenomicContentofBacteroidesSpeciesPredictsChangesinMicrobiotaCompositionInducedbyanInulin-BasedDiet

ThedifferencesinabilitytoutilizefructansbetweentheBacter-oidesspeciesimpliesthattherelativesuccessofaspecieswithinagutecosystemmaybedetermined,inpart,bytheabun-danceandtypeoffructaninthehostdiet.Furthermore,thecomparisonofgenomicsequencesanddifferencesinfructanusebetweenspeciessuggeststhatpersonalizedpredictionsofmicrobiotaresponsetospecificdietarypolysaccharidesmaybemadebasedonmetagenomicmicrobiomesequencedata.Weconstructeddefinedtwo-membercommunitiesofBacter-oidesspecieswithintheintestinesofgnotobioticmicetotesthowmodelmicrobiotasrespondinvivotodietaryinulin,which,unlikelevan,isavailableinpureforminquantitiessufficienttoconductsuchastudy.Ourinvivoexperimentaimedtotesthowdifferingfunctionalitiesembeddedwithinthegenomesoftwodifferenttwo-speciesmodelmicrobiotasinfluenceinulin-inducedchangesincommunitycomposition.

DuetoB.caccae’ssuperiorabilitytouseinulincomparedtoBt,wetestedwhetherB.caccaewouldbecomedominantoverBtwithintheintestinesofmicefedaninulin-supplementeddiet.Conversely,Bt’spoorgrowthoninulinisbetterthanB.vulgatus,whichisunabletoutilizeinulin,suggestingthatBtmightbenefitfrominulinwhencolonizedwithB.vulgatus.Twogroupsof8–12-week-old,germ-freemicewerecolonizedwithequivalentquantities(108colonyformingunits,CFU)ofBtand

1248Cell141,1241–1252,June25,2010ª2010ElsevierInc.

B.caccaeorBtandB.vulgatus.Eachmousewasmaintainedonastandardpolysaccharide-richdietforthefirst7daysofcolonizationandthenswitchedtoadietinwhichthesolepolysaccharidewasinulin(10%w/w)foranadditional14days(Figure7A).Micewereindividuallyhousedthroughouttheexper-imenttoensurenocrossinoculationcouldoccurandbeddingwaschangedeverytwodays.TotalbacterialcolonizationdensitywasdeterminedbyassessingtheCFUsinfecesover21days.Thechangeineachspecies’relativeabundancebeforeandafterdietaryinulinsupplementationwasassessedusingspecies-specificprimersinaquantitativePCRassay.

Ourresultsdisclosedthattotalfecalbacterialdensitiesoverthecourseoftheexperimentdidnotdiffersignificantlyupondietaryshift(totaldensitiesrangedfrom1010–1011bacteria/mloffecalmaterial).Relativedensitiesweredeterminedondays4and6(standarddiet)andondays13and21(6and14daysafterdietaryswitch).IntheBt/B.caccae,bi-associatedmice,beforethedietswitch(day6postcolonizationinmicefedastandarddiet),Btcomprised87±3%ofthecommunity,indi-catingthatBtisbetteradaptedthanB.caccaetotheseinvivoconditions.Sixdaysafterachangetotheinulin-baseddiet,Btlevelsdroppedto80±4%,andB.caccaeincreasedto20±4%.Aftertwoweeksconsumingtheinulindiet,therelativeproportionofthetwospeciesshowedamoredrasticshiftinfavorofB.caccae:Btrepresentationdecreasedtoapproxi-mately49±6%versus51±6%B.caccae(p=8310À5,day21versusday6;n=7mice;Student’sttest)(Figure7B).In

GF

colonizestandard

7 days

diet

diet change14 days

Difference in % representation of B. caccaeAD

50%40%30%20%10%0%Bt/B. caccaeFigure7.EffectofDietaryFructansonBac-teoridesCompetitionwithintheIntestine

(A)Experimentaldesignforinvivoexperiments.GF,germ-free.

(B)Averagerelativefecalproportion(%totalbacteria)ofBtandB.caccaeat4,6,14,and21daysaftercolonization;n=7mice.

(C)Averagerelativefecalproportion(%totalbacteria)ofBtandB.vulgatusat4,6,14,and21daysaftercolonization;n=3mice.

(D)Increaseinproportion(%)ofB.caccaeoverBtfromday6(1daypriortodietchange)today21(14daysafterdietchange).Allgroupsreceivedastandarddietondays1–7;typeofdietandwhetherthemicereceivedinulinintheirwaterondays7-21isindicated;n=3-7individuallyhousedmice.

(E)Averagerelativefecalproportion(%totalbacteria)ofinulin-utilizingBt(In+)andB.caccaeat4,6,14,and21daysaftercolonization;n=7individuallyhousedmice.

Valuesareaveragesandstandarderrors.

B

% representationcolonizationstandard100%75%50%25%0%

0

5

10

15

20

25

inulinBtInulinStandard-./0123'Standard diet-./0\"3$0452'Polysaccharide-#-0\"3$0452'dietdietInulin waterdeficient dietInulin waterE

B. caccae% representationcolonizationstandard100%1.0075%0.750.5050%0.2525%0.000%0

5

10

15

20

25

inulinBt(In+)Time (days)

C

% representationcolonizationstandard100%75%50%25%0%0

5

10

15

20

25

B. caccaeinulinTime (days)

BtB. vulgatusTime (days)

contrast,theBt/B.vulgatusbi-associatedmicedidnotexhibit

anysignificanttrendinchangedcommunitycompositionafter6daysonaninulin-baseddiet,butBtincreasedinabundancefrom74±3%onday6to84±5%onday21(p=0.1;n=3mice)ontheinulin-enricheddiet(Figure7C).ThedelayedandmodesteffectofdietinfluencingthecompositionoftheBt/B.vul-gatusbi-associationisconsistentwithpoorinulinusebyBtandnoinulinusebyB.vulgatus(Figure6).Together,thesedataareconsistentwithdietarypolysaccharide-inducedchangesinthemicrobiotacompositionthatarepredictablebasedontheresi-dentspecies’abilitytousethatpolysaccharide.

Inthepreviousexperiment,inulinwasthesolepolysaccharideinthediet.Wewonderedwhetherwewouldobservethesameinulin-inducedincreaseinB.caccaerelativetoBtifotherpolysaccharideswerealsopresentinthediet.Totestthis,gnotobioticmicewereco-colonizedwithBtandB.caccaeandmaintainedonthestandarddietwithinulinsupplementationinthewater(1%w/v).Overthe14daysthemiceingestedanaverageof117±6mgofinulindailyviathewater(comparedto355±7mg/daywiththeinulindiet).FecalsamplesweretestedbyqPCRoverthecourseofthe21-dayexperimentforrelativelevelsofBtorB.caccae.Thesedatarevealednostatisticaldifferenceinthechangeinrelativecolonizationbetweenmicefedinulin-supplementedwatercomparedtocontrolsthatreceivedthesamestandarddietfor21days,butreceivednoinulin(Figure7D).Thesedatasuggestthatwhenmicewerefedadietrichincarbohydrates,thepresenceofinulindidnotprovideenoughofanadvantagetoB.caccaetoallowittoout-competeBt;however,theamountofinulinsuppliedinthewater

(117mg/dayaverage)waslessthantheamountderivedfromtheinulindiet(355mg/dayaverage)potentiallycon-tributingtothelackoftheB.caccaeresponse.

Wedecidedtofeedmiceacustomdietdeficientinallpolysaccharidesand

supplementinulininthewatertodeterminewhetheralowerdoseofinulinintheabsenceofotherpolysaccharideswassufficienttoprovideB.caccaeacompetitiveadvantageoverBtinvivo.Underthisexperimentalparadigmthemiceconsumedanaverageof97mgofinulinperday.After14daysoninulin-watersupplementation,theproportionofB.caccaeincreasedby26±8%(Figure7D).Whilenotasrobustanincreaseasobservedintheinulin-onlydietexperiment(whichshoweda36±7%increaseinB.caccae),thesedatademonstratethatreducedinulinconsumptionintheabsenceofcompetingpolysaccharides,offersasignificantcompetitiveadvantagetoinulin-utilizingB.caccae,consistentwiththeflexiblenutrientforagingtheBacteroidesspeciesexhibit.Thewiderangeofpoly-saccharidespresentinthestandarddietallowsBttocompeteeffectivelywithB.caccaeeveninthepresenceofinulin.

Wefinallydemonstratetheimportanceofinulinutilizationforconferringacompetitiveadvantageinhostsfedaninulin-richdietusingageneticproofofthiseffect.TheregionoftheB.caccaefructan-utilizationlocusfromthesusC-likegenethroughtheGH32-encodinggene(BC02727-BC02731)wasclonedandexpressedinastrainofBtthatiscompromisedinitsabilitytoutilizelevan(Bt-DBT1763)underthecontroloftheBT1763promoter(datanotshown).Theresultingstrain,Bt(In+),exhibitsefficientgrowthinminimalmediumcontaininginulin,similartoB.caccae(FigureS7).RepeatingouroriginalinvivocompetitionexperimentwithBt(In+)revealedthatconferringinulinuseabilityuponBteliminatestheabilityofB.caccaetobecomedominantinthepresenceofaninulin-baseddiet(Figure7E).Thisresultconfirmsthatthespecificityofdietarypolysaccharideuseis

Cell141,1241–1252,June25,2010ª2010ElsevierInc.1249

thekeyfunctionalitythatdictatesthealterationsinthemodelmicrobiotathatweobserve.Theseresultssupportourhypoth-esisthatchangesinmicrobiotacommunitymembershipbroughtonbydietarychangecanbeinferredbasedongenomicandfunc-tionalknowledgeofresidentmicrobialpopulations.Theyalsosuggestthatdietcanbeadominantdeterminantindictatingchangesinmicrobiotacomposition.DISCUSSION

Inulin(b2-1fructan)andlevan(b2-6fructan)arepolysaccharidesthatareabundantinthehumandiet,butareresistanttohost-mediateddigestionintheuppergastrointestinaltract.Theseglycansinsteadserveasacarbonandenergysourceforthebacteriathatresideinthedistalintestine.Bacteroidesthe-taiotaomicron,aresidentofthehumanGItract,encodesafruc-tanutilizationlocus,BT1757-63andBT1765,thegeneproductsofwhichenableefficientacquisitionanduseoflevan-typecarbohydrates.

ThefructanPULisadjacenttoahybridtwo-componentsystemsensor-regulator,BT1754,whichbindsonlytomono-mericfructose,asignalsufficienttoinducetranscriptionofthelocus.Whiletheupregulationofpolysaccharideutilizationmachineryinresponsetoamonosaccharidemayseemunex-pected,thissignalisalikelyconsequenceoftheenvironmentinwhichBtresides.Withinthenaturalhabitatofthelargebowel,freefructoseandsimpledisaccharides,suchassucrose,donotoccuratappreciablelevelsasthehostabsorbssuchsugarswithinthesmallintestine.Therefore,theregulationofthislocusevolvedintheabsenceofselectivepressuretodiscriminatefreemonosaccharidefrompolysaccharides.Inaddition,unlikemanyothermonosaccharides,fructoseisfoundinonlyasingleclassofpolysaccharide,namelyhomopolymericfructans.Btappearstousetheliberatedfructoseasaproxy(i.e.,indicator)forfructan,whichresultsinupregulationofthemachinerytoutilizethepolysaccharide.ThisisconsistentwithpreviousdatathatdemonstrateBt’sconstitutive,low-levelexpressionofPULsinconditionslackingtherelevantsubstrates(Martensetal.,2009;Sonnenburgetal.,2005),aswellasthelow-levelcellsurfacelevanaseactivityweobservewithwholecellsgrowninglucose.TheconstitutiveexpressionsuggeststhatBtemploysastrategyofbeingpreparedtodegrademultiplepolysaccharidesimmediatelyupontheirarrivalintothedistalgutenvironment.SpecificliberatedcarbohydratesthatresultfromthedegradationserveassignalsthataugmentexpressionoftheappropriatePULviaaspecificsensor-regulatorsuchasaHTCS.

ThebindingofBT1754tomonomericfructosealsoresultsinafailureofthesensortodifferentiateb2-1andb2-6linkagesdespiteBtbeingmuchmoreefficientinuseofthelevan-typefructans.SpecificityofsignalisinsteadderivedfromthecellsurfacestructuralcomponentsofthePUL,whichserveasthe‘‘gateway’’forsubstratescrossingtheoutermembrane.ThecellsurfaceSusDhomolog,BT1762,thesusE-positionedgeneproduct,BT1761,andtheendo-levanase,BT1760,allcontributetothespecificimportofb2-6fructansintoBt’speriplasm.BT1754reliesuponthespecificityofthecellsurfacepolysaccha-ridedegradationandbindingmachinerytoprovidefructose

1250Cell141,1241–1252,June25,2010ª2010ElsevierInc.

derivedfromb2-6fructantotheperiplasmwherethesensorissequestered.

DespiteBt’sinabilitytoutilizeinulinefficientlyitisabletogrowwellonFOS,ashortchainb2-1fructan.Notably,thefructanPULofBtisupregulatedduringgrowthinvitroinminimalmediumcontainingFOSorinulin.Bt’sabilitytogrowinFOSataratethatissignificantlyfasterthaninulinislikelyduetothedifferenceindegreeofpolymerizationbetweenthetwosubstrates.Whethersmalloligosaccharidesfromthesesubstratesundergopassivediffusionintotheperiplasmorareaccessedviaanothermechanismrequiresfurtherinvestigation.

AmongtheBacteroidesspeciestested,Btappearstobeuniqueinitsabilitytoutilizelevan,whereasotherspeciesareadeptatutilizingpolymericb2-1fructans.Suchphenotypicdifferences,combinedwithdietaryvariationbetweenindivid-uals,couldprovidethebasisforthestrikingperson-to-personvariabilityobservedforBacteroidetesinhumanmicrobiotaenumerationstudies(Eckburgetal.,2005).Ourinvivostudiesillustratethatspecieswell-adaptedtouseinulingainacompeti-tiveadvantagewhenhostsarefedaninulin-baseddiet.Althoughageneticloss-offunctionexperiment,inwhichinulinuseiscompromised,couldbeusedtotestwhethertheobservedchangesinspeciesabundanceareduetoinulinuse,wehaveusedagain-of-functionexperiment,inwhichinulinuseisconferreduponBt,toillustratethispointunequivocally.Theseresultssuggestthatsomeaspectsofdiet-inducedchangesinmicrobiotacompositionmaybepredeterminedbasedontheintrinsiccapacityofanindividualspeciestousethesubstratesthatarebeingconsumedbythehost.Wespeculatethatdietsen-richedindifferentpolysaccharides,orpolysaccharide-deficientdiets,couldresultinmicrobiotasofverydifferentspeciescomposition.Futurestudiesthatfollowspeciesandgenecompositionofthehumanintestinalmicrobiotaduringconsump-tionoflevan-orinulin-baseddietswillprovideinsightintotherapiditywithwhichmembersofacomplexcommunityadaptatafunctional,compositional,andgeneticlevel.Howsuchnichespecializationoccursoverthecourseofevolutionandtherolethatdietplaysindeterminingaspecies’glycanutilizationreper-toireremainimportantyetdifficultquestionstoaddress.Perspective

Astheageofpersonalgenomesapproaches,someaspectsofdietandmedicaltherapieswillbecustomizedbasedongeno-type.Dietcanalsobepersonalizedtooptimizemicrobiotafunctionandinteractionwiththehostbasedonthemetagenomicanalysisofanindividual’smicrobiota.Aprerequisiteforincorpo-ratingvastamountsofmicrobialgenomicdataintopersonalized,preventativemedicineistoattainamechanisticunderstandingofthemostdominantaspectsofmicrobiotafunction.Herewepresentacasestudyofhowunderstandingthemechanismsthatlinkthemicrobometomicrobiotafunctionmayenableindi-vidualizedpredictionsofmicrobiotaresponsetoperturbations.Wehavetakentwo-speciesmodelmicrobiotasthatcollectivelypossesscloseto10,000genesandpredictedhowtheywillrespondtoaspecificdietarycuebasedonafunctionalunder-standingofthe$20relevantgenes.Asimilardistillationoffullmicrobiomicdatasetsthatcontain>106genes,toarelevantsubset,willberequiredtomakemicrobiotamanagement

tractable.Withanever-increasingunderstandingofhowthebiologyofhostandmicrobiotaintegrate,wemaysoonbeabletousegenomicandmicrobiomicsequencedatatointentionallyprogramorreprogramtheemergentpropertiesofthehost-microbialsuperorganism.

EXPERIMENTALPROCEDURES

CulturingBacteria

BacteriawereculturedinTYGandMMasdescribedpreviously(Martensetal.,2008;Sonnenburgetal.,2005).Thefollowingbacteriawereused:Bt(VPI-5482),B.caccae(ATCC-43185),B.ovatus(ATCC-8483),B.fragilis(NCTC-9343),B.uniformis(ATCC-8492),andB.vulgatus(ATCC-8482).GrowthcurvesinMMwereobtainedusingaPowerwave(Biotek)readingOD600every30minfromanaerobicculturesat37󰀄C.

QuantitativeRT-PCRAnalysis

QuantitativeRT-PCRwasperformedusinggene-specificprimersasdescribedpreviously(TableS3)withSYBRGreen(ABgene)inaMX3000Pthermocycler(Strategene)(Martensetal.,2008).

GeneDeletionandComplementationinBt

In-frame(nonpolar)genedeletionsformutantsweregeneratedusingcounter-selectablealleleexchange(Martensetal.,2008).PCRamplifiedgenesforcomplementationwereligatedintothepNBU2-tetQbvectorandconjugatedintoBtviaE.coliS17.1l-pir(Martensetal.,2008).ResultingcloneswerescreenedbyPCRandsequencedtoconfirmisolates.

GeneCloning

GenesforexpressionwereamplifiedfromBtgenomicDNAusingtheprimersstatedinTableS3andclonedintopRSETA(Invitrogen)orpET22b(Novagen).ProteinExpressionandPurification

RecombinantproteinswereexpressedinE.coliC41orBL21cellsandpurifiedinasinglestepusingmetalaffinitychromatographyasdescribedpreviously(Bolametal.,2004).

SourcesandPreparationofCarbohydrates

Monosaccharides,sucrose,andchicoryinulinforenzymaticandbindingassayswereobtainedfromSigma.GrowthofBacteroidesstrains,qRT-PCR,andmouseexperimentsusedinulin,FOS(Beneo-Oraftigroup;OraftiHP,Oraf-tiP95,respectively)andlevan(Sigma;66674).KestooligosaccharideswerefromMegazyme.Levanoligosaccharideswereproducedbypartialacidhydro-lysis(1MHClat25󰀄Cfor20min-1hr)oflevan(MontanaPolysaccharides).NaOH-neutralizedsampleswereseparatedonBioGelP2(BioRad)sizeexclu-sionresin.

IsothermalTitrationCalorimetry

Measurementswerecarriedoutessentiallyasdescribedpreviously(Bolametal.,2004),exceptthataMicrocalVP-ITCmachinewasused,andproteinsweredialyzedinto20mMTris-HCl(pH8.0).Theassumptionthatn=1forBT1762bindingtolevanwasbasedonthestructureofthestarchbindingSusD(Koropatkinetal.,2008).

Thin-LayerChromatography

Sampleswerespottedontofoilbackedsilicaplatesandplacedinaglasstankequilibratedwithbutanol:aceticacid:H2O(2:2:1).Sugarswerevisualizedusingorcinol-sulphuricacid(sulphuricacid:ethanol:H2O3:70:20v/v,orcinol1%w/v),90󰀄Cfor5–10min.

EnzymeAssays

Allassayswerecarriedoutat37󰀄Cin20mMTris-HCl(pH8.0).ActivityofBT1760wasdeterminedbyquantifyingtheamountofreducingsugarreleasedusingtheDNSAassay(Miller,1959).Freefructosewasdeterminedusingamodifiedfructosedetectionkit(MegazymeInternational).Kineticparametersweredeterminedbyfittinginitialratesversussubstrateconcentration

(measuredatsixsubstrateconcentrationsthatspannedtheKM)totheMichae-lis-Mentenequationusingnonlinearregression(GraphpadPrism,v5.0).

EnzymeLocalizationStudies

Culturesgrownon0.5%(w/v)fructoseorglucosewereharvestedbycentrifu-gation(OD600$1.0).PBS-washedcellsand0.5%levanorinulinin20mMTris-HCl,pH8.0,wereincubatedat37󰀄C.ReducingsugarpresentwasquantifiedusingDNSAreagent(Miller,1959).Activitiesoftheperiplasmicmarkeralkalinephosphataseandcytoplasmicmarkerglucose-6-phophatedehydrogenasewerecomparedtolysedcellstoensurenocelllysis/leakageoccurred.

BacterialColonizationandDensityDeterminationofGerm-FreeMice

Germ-freeSwiss-Webstermiceweremaintainedingnotobioticisolatorsandfedanautoclavedstandarddiet(PurinaLabDiet5K67)orcustomdiet(Bio-Serv,http://bio-serv.com/),inaccordancewithA-PLAC,theStanfordIACUC.Micewerebi-associatedusingoralgavage(108CFUofeachbacterialspecies).RelativedensitiesofbacteriaweredeterminedbyqPCRusingstrain-specificprimers(TableS3)(Martensetal.,2008).

Crystallization,StructureDetermination,andRefinementofBT1754-PD

Crystalsformedin0.7MK/Naphosphate,0.1MHEPES(pH8.0)(proteinat8mg/mlwith5mMfructose).Diffractiondata,collectedatDiamondLightSource(Oxford,UK)onatiledADSCQ315CCDdetectorwereprocessedwithMOSFLM(Leslie,1992).Scalingofdata,searchmodelgeneration,molec-ularreplacementandstructurerefinementwerecarriedoutusingSCALA,CHAINSAW,MOLREPandREFMAC(CollaborativeComputationalProject,1994),respectively,withmodelrebuildinginCOOT(EmsleyandCowtan,2004).

ACCESSIONNUMBERS

ProteinDataBankcoordinateshavebeendepositedundertheaccessioncode2X7X.

SUPPLEMENTALINFORMATION

SupplementalInformationincludesfourtables,sevenfigures,andSupple-mentalReferencesandcanbefoundwiththisarticleonlineatdoi:10.1016/j.cell.2010.05.005.

ACKNOWLEDGMENTS

WethankKarlaKirkegaardandStanleyFalkowforvaluablecommentsandSaraFisherforeditingthemanuscript.InulinandFOSformouseexperimentswereakindgiftfromBeneo-Orafti.LevanwasakindgiftfromMontanaPoly-saccharides.WethankJeffreyGordonandmembersoftheGordonLabforvaluableadvice;CarlMorlandforexcellenttechnicalassistance;andEricMartensandAndrewGoodmanfordevelopmentofgenetictoolsusedinthispaper.SomeBacteroidesgenomicdatawereproducedbyTheGenomeCenteratWashingtonUniversitySchoolofMedicineinSt.Louis(genome.wustl.edu).ThisworkwasfundedinpartbygrantsfromNationalInstitutesofHealththroughtheNIHDirector’sNewInnovatorAwardProgram(DP2-OD006515)theNIDDK(K01-DK077053),theStanfordDigestiveDiseaseCenter(PO3-DK56339)andtheBBSRC(BB/F014163/1).Received:October15,2009Revised:January20,2010Accepted:April27,2010Published:June24,2010

Cell141,1241–1252,June25,2010ª2010ElsevierInc.1251

REFERENCES

Backhed,F.,Ley,R.E.,Sonnenburg,J.L.,Peterson,D.A.,andGordon,J.I.(2005).Host-bacterialmutualisminthehumanintestine.Science(NewYork,NY307,1915-1920.

Bolam,D.N.,Xie,H.,Pell,G.,Hogg,D.,Galbraith,G.,Henrissat,B.,andGilbert,H.J.(2004).X4modulesrepresentanewfamilyofcarbohydrate-bindingmodulesthatdisplaynovelproperties.J.Biol.Chem.279,22953–22963.

Cantarel,B.L.,Coutinho,P.M.,Rancurel,C.,Bernard,T.,Lombard,V.,andHenrissat,B.(2009).TheCarbohydrate-ActiveEnZymesdatabase(CAZy):anexpertresourceforGlycogenomics.NucleicAcidsRes.37,D233–D238.CollaborativeComputationalProject.(1994).TheCCP4suite:programsforproteincrystallography.ActaCrystallogr.50,760–763.

Dethlefsen,L.,Huse,S.,Sogin,M.L.,andRelman,D.A.(2008).Thepervasiveeffectsofanantibioticonthehumangutmicrobiota,asrevealedbydeep16SrRNAsequencing.PLoSBiol.6,e280.

Duncan,S.H.,Lobley,G.E.,Holtrop,G.,Ince,J.,Johnstone,A.M.,Louis,P.,andFlint,H.J.(2008).Humancolonicmicrobiotaassociatedwithdiet,obesityandweightloss.InternationalJournalofObesity32,1720–1724.

Duncan,S.H.,Scott,K.P.,Ramsay,A.G.,Harmsen,H.J.,Welling,G.W.,Stewart,C.S.,andFlint,H.J.(2003).Effectsofalternativedietarysubstratesoncompetitionbetweenhumancolonicbacteriainananaerobicfermentorsystem.Appl.Environ.Microbiol.69,1136–1142.

Dwyer,M.A.,andHellinga,H.W.(2004).Periplasmicbindingproteins:aversa-tilesuperfamilyforproteinengineering.Curr.Opin.Struct.Biol.14,495–504.Eckburg,P.B.,Bik,E.M.,Bernstein,C.N.,Purdom,E.,Dethlefsen,L.,Sargent,M.,Gill,S.R.,Nelson,K.E.,andRelman,D.A.(2005).Diversityofthehumanintestinalmicrobialflora.Science308,1635–1638.

Emsley,P.,andCowtan,K.(2004).Coot:model-buildingtoolsformoleculargraphics.ActaCrystallogr.60,2126–2132.

Falke,J.J.,andErbse,A.H.(2009).ThePistonRisesAgain.Structure17,1149–1152.

Flint,H.J.,Bayer,E.A.,Rincon,M.T.,Lamed,R.,andWhite,B.A.(2008).Poly-saccharideutilizationbygutbacteria:potentialfornewinsightsfromgenomicanalysis.Nature6,121–131.

Frank,D.N.,StAmand,A.L.,Feldman,R.A.,Boedeker,E.C.,Harpaz,N.,andPace,N.R.(2007).Molecular-phylogeneticcharacterizationofmicrobialcommunityimbalancesinhumaninflammatoryboweldiseases.Proc.Natl.Acad.Sci.USA104,13780–13785.

Hooper,L.V.(2009).Dosymbioticbacteriasubverthostimmunity?Nature7,367–374.

Jernberg,C.,Lofmark,S.,Edlund,C.,andJansson,J.K.(2007).Long-termecologicalimpactsofantibioticadministrationonthehumanintestinalmicro-biota.ISMEJ.1,56–66.

Kolida,S.,Meyer,D.,andGibson,G.R.(2007).Adouble-blindplacebo-controlledstudytoestablishthebifidogenicdoseofinulininhealthyhumans.Eur.J.Clin.Nutr.61,1189–1195.

Koropatkin,N.,Martens,E.C.,Gordon,J.I.,andSmith,T.J.(2009).StructureofaSusDhomologue,BT1043,involvedinmucinO-glycanutilizationinaprom-inenthumangutsymbiont.Biochemistry48,1532–1542.

Koropatkin,N.M.,Martens,E.C.,Gordon,J.I.,andSmith,T.J.(2008).Starchcatabolismbyaprominenthumangutsymbiontisdirectedbytherecognitionofamylosehelices.Structure16,1105–1115.

Leslie,A.G.W.(1992).RecentchangestotheMOSFLMpackageforprocess-ingfilmandimageplatedata.JointCCP4+ESF-EAMCBNewsletteronProteinCrystallography,26.

Ley,R.E.,Turnbaugh,P.J.,Klein,S.,andGordon,J.I.(2006).Microbialecology:humangutmicrobesassociatedwithobesity.Nature444,1022–1023.

1252Cell141,1241–1252,June25,2010ª2010ElsevierInc.

Louis,P.,Scott,K.P.,Duncan,S.H.,andFlint,H.J.(2007).Understandingtheeffectsofdietonbacterialmetabolisminthelargeintestine.J.Appl.Microbiol.102,1197–1208.

Lozupone,C.A.,Hamady,M.,Cantarel,B.L.,Coutinho,P.M.,Henrissat,B.,Gordon,J.I.,andKnight,R.(2008).Theconvergenceofcarbohydrateactivegenerepertoiresinhumangutmicrobes.Proc.Natl.Acad.Sci.USA105,15076–15081.

Martens,E.C.,Chiang,H.C.,andGordon,J.I.(2008).Mucosalglycanforagingenhancesfitnessandtransmissionofasaccharolytichumangutbacterialsymbiont.CellHostMicrobe4,447–457.

Martens,E.C.,Koropatkin,N.M.,Smith,T.J.,andGordon,J.I.(2009).Complexglycancatabolismbythehumangutmicrobiota:ThebacteroidetesSus-likeparadigm.J.Biol.Chem..284,24673–24677.

Menne,E.,Guggenbuhl,N.,andRoberfroid,M.(2000).Fn-typechicoryinulinhydrolysatehasaprebioticeffectinhumans.J.Nutr.130,1197–1199.Miller,G.L.(1959).UseofDinitrosalicylicAcidReagentforDeterminationofReducingSugar.Anal.Chem.31,426–428.

Ramirez-Farias,C.,Slezak,K.,Fuller,Z.,Duncan,A.,Holtrop,G.,andLouis,P.(2008).Effectofinulinonthehumangutmicrobiota:stimulationofBifidobac-teriumadolescentisandFaecalibacteriumprausnitzii.Br.J.Nutr.101,541–550.

Roberfroid,M.,Gibson,G.R.,andDelzenne,N.(1993).Thebiochemistryofoli-gofructose,anondigestiblefiber:anapproachtocalculateitscaloricvalue.Nutr.Rev.51,137–146.

Rossi,M.,Corradini,C.,Amaretti,A.,Nicolini,M.,Pompei,A.,Zanoni,S.,andMatteuzzi,D.(2005).Fermentationoffructooligosaccharidesandinulinbybifidobacteria:acomparativestudyofpureandfecalcultures.Appl.Environ.Microbiol.71,6150–6158.

Samuel,B.S.,andGordon,J.I.(2006).Ahumanizedgnotobioticmousemodelofhost-archaeal-bacterialmutualism.Proc.Natl.Acad.Sci.USA103,10011–10016.

Shipman,J.A.,Berleman,J.E.,andSalyers,A.A.(2000).CharacterizationoffouroutermembraneproteinsinvolvedinbindingstarchtothecellsurfaceofBacteroidesthetaiotaomicron.J.Bacteriol.182,5365–5372.

Sonnenburg,E.D.,Sonnenburg,J.L.,Manchester,J.K.,Hansen,E.E.,Chiang,H.C.,andGordon,J.I.(2006).Ahybridtwo-componentsystemproteinofaprominenthumangutsymbiontcouplesglycansensinginvivotocarbohy-dratemetabolism.Proc.Natl.Acad.Sci.USA103,8834–8839.

Sonnenburg,J.L.,Xu,J.,Leip,D.D.,Chen,C.H.,Westover,B.P.,Weatherford,J.,Buhler,J.D.,andGordon,J.I.(2005).Glycanforaginginvivobyanintestine-adaptedbacterialsymbiont.Science307,1955–1959.

Turnbaugh,P.J.,Hamady,M.,Yatsunenko,T.,Cantarel,B.L.,Duncan,A.,Ley,R.E.,Sogin,M.L.,Jones,W.J.,Roe,B.A.,Affourtit,J.P.,etal.(2009).Acoregutmicrobiomeinobeseandleantwins.Nature457,480–484.

Turnbaugh,P.J.,Ley,R.E.,Hamady,M.,Fraser-Liggett,C.M.,Knight,R.,andGordon,J.I.(2007).Thehumanmicrobiomeproject.Nature449,804–810.VanderMeulen,R.,Makras,L.,Verbrugghe,K.,Adriany,T.,andDeVuyst,L.(2006).InvitrokineticanalysisofoligofructoseconsumptionbyBacteroidesandBifidobacteriumspp.indicatesdifferentdegradationmechanisms.Appl.Environ.Microbiol.72,1006–1012.

Westover,B.P.,Buhler,J.D.,Sonnenburg,J.L.,andGordon,J.I.(2005).Operonpredictionwithoutatrainingset.Bioinformatics21,880–888.Xu,J.,Bjursell,M.K.,Himrod,J.,Deng,S.,Carmichael,L.K.,Chiang,H.C.,Hooper,L.V.,andGordon,J.I.(2003).Agenomicviewofthehuman-Bacter-oidesthetaiotaomicronsymbiosis.Science299,2074–2076.

Xu,J.,Mahowald,M.A.,Ley,R.E.,Lozupone,C.A.,Hamady,M.,Martens,E.C.,Henrissat,B.,Coutinho,P.M.,Minx,P.,Latreille,P.,etal.(2007).Evolu-tionofSymbioticBacteriaintheDistalHumanIntestine.PLoSBiol.5,e156.NoteAddedinProof

PL19enzymeshaverecentlybeenreclassifiedintoglycosidehydrolasefamily91(seewww.cazy.org/GH91.html).