ebook img

Genomic and metabolic studies of the impact of probiotics on a model gut symbiont and host PDF

15 Pages·2015·2.15 MB·English
by  
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Genomic and metabolic studies of the impact of probiotics on a model gut symbiont and host

Washington University School of Medicine Digital Commons@Becker Open Access Publications 2006 Genomic and metabolic studies of the impact of probiotics on a model gut symbiont and host Justin L. Sonnenburg Washington University School of Medicine in St. Louis Christina T. Chen Washington University School of Medicine in St. Louis Jeffery I. Gordon Washington University School of Medicine in St. Louis Follow this and additional works at:http://digitalcommons.wustl.edu/open_access_pubs Part of theMedicine and Health Sciences Commons Recommended Citation Sonnenburg, Justin L.; Chen, Christina T.; and Gordon, Jeffery I., ,"Genomic and metabolic studies of the impact of probiotics on a model gut symbiont and host." PLoS Biology.4,12. 2213-2226. (2006). http://digitalcommons.wustl.edu/open_access_pubs/392 This Open Access Publication is brought to you for free and open access by Digital Commons@Becker. It has been accepted for inclusion in Open Access Publications by an authorized administrator of Digital Commons@Becker. For more information, please [email protected]. PLoS BIOLOGY Genomic and Metabolic Studies of the Impact of Probiotics on a Model Gut Symbiont and Host Justin L. Sonnenburg, Christina T. L. Chen, Jeffrey I. Gordon* CenterforGenomeSciences,WashingtonUniversitySchoolofMedicine,St.Louis,Missouri,UnitedStatesofAmerica Probioticsaredeliberatelyingestedpreparationsoflivebacterialspeciesthatconferhealthbenefitsonthehost.Many of these species are associated with the fermentation of dairy products. Despite their increasing use, the molecular details of the impact of various probiotic preparations on resident members of the gut microbiota and the host are generallylacking.Toaddress thisissue,wecolonized germ-freemice withBacteroides thetaiotaomicron,aprominent component of the adult human gut microbiota, and Bifidobacterium longum, a minor member but a commonly used probiotic.Simultaneouswholegenometranscriptionalprofilingofbothbacterialspeciesintheirguthabitatandofthe intestinal epithelium, combined with mass-spectrometric analysis of habitat-associated carbohydrates, revealed that the presence of B. longum elicits an expansion in the diversity of polysaccharides targeted for degradation by B. thetaiotaomicron(e.g.,mannose-andxylose-containingglycans),andinduceshostgenesinvolvedininnateimmunity. Although the overall transcriptome expressed by B. thetaiotaomicron when it encounters B. longum in the cecum is dependentuponthegeneticbackgroundofthemouse(asassessedbyamixedanalysisofvariance[ANOVA]modelof co-colonization experiments performed in NMRI and C57BL/6J animals), B. thetaiotaomicron’s expanded capacity to utilizepolysaccharides occurs independently of host genotype, and is also observed with a fermenteddairy product- associatedstrain,Lactobacilluscasei.Thisgnotobioticmousemodelprovidesacontrolledcasestudyofhowaresident symbiontandaprobioticspeciesadapttheirsubstrateutilizationinresponsetooneanother,andillustratesboththe generality and specificity of the relationship between a host, a component of its microbiota, and intentionally consumed microbial species. Citation:SonnenburgJL,ChenCTL,GordonJI(2006)Genomicandmetabolicstudiesoftheimpactofprobioticsonamodelgutsymbiontandhost.PLoSBiol4(12):e413. DOI:10.1371/journal.pbio.0040413 throughtheresidentmicrobiotaversusdirectsignalingtothe Introduction host,andtheimpactofhostgenotypeonprobioticresponses Over the millennia, our species has evolved to cope with, remain open questions. The answers have important impli- and perhaps benefit from, consuming microbes present in cations for designing and interpreting clinical trials that decomposing food. Some have postulated that ingestion of explore the specific and general effects of different pro- food-bornemicrobialspecieswascommoninthePleistocene biotics,andwhethertheirconsumptionshouldbedictatedby diet of hominins practicing subterranean food storage [1,2]. the microbial ecology or other properties of the host, Eventually these microbes were co-opted by humans to includingdiet andgenotype. transformfoodsviafermentation[3].Duringthelastcentury, The task of determining the mechanism of action of technological improvements, such as refrigeration and probioticsinhumansisdaunting.Ourintestinescontain10– pasteurization, have resulted in dramatic decreases in 100 trillion microbes. A recent comprehensive 16S rRNA consumption of food-borne microbes in many countries. sequence-basedenumerationhasprovideddetailedsnapshots Thissanitizationhascoincidedwithincreasedprevalenceofa of the composition of this community in a few healthy number of disorders, both within and outside of the gastro- individuals [8]. Our 21st century microbiota contains mem- intestinal tract, including inflammatory bowel disease and atopic states (e.g., asthma and food allergies). Although links AcademicEditor:JonathanEisen,UniversityofCaliforniaDavisGenomeCenter, between certain ‘‘western’’ diseases and decreased exposure UnitedStatesofAmerica tomicrobesarebeingconsideredatthepresenttime(e.g.,the ReceivedJune28,2006;AcceptedSeptember25,2006;PublishedNovember28, hygiene hypothesis in the case of atopic disorders) [4], the 2006 positive health effects of consuming fermented foods were DOI:10.1371/journal.pbio.0040413 notedmorethan100yagowhenElieMetchnikoffcorrelated Copyright: (cid:2)2006 Sonnenburg etal.Thisisanopen-accessarticledistributed theiringestionwithincreasedlifespaninBulgarianpeasants under the terms of the Creative Commons Attribution License, which permits [5]. More recently, the effects of consuming live microbial unrestricted use, distribution, and reproduction in any medium, provided the originalauthorandsourcearecredited. supplements with presumptive health benefits on human physiologyhavebecomethesubjectofvariousclinicalstudies Abbreviations:ANOVA,analysisofvariance;CAZy,CarbohydrateActiveenZymes database; CFU, colony forming units; COG, cluster of orthologous groups; FDR, (reviewedin[6,7]).However,theimpactofthese‘‘probiotic’’ falsediscoveryrate;GC-MS,gaschromatography–massspectrometry;LCM,laser speciesonthecompositionandfunctioningofthehumangut capturemicrodissection;qRT-PCR,quantitativereversetranscriptasePCR microbiota, the degree to which their effects are mediated *Towhomcorrespondenceshouldbeaddressed.E-mail:[email protected] PLoSBiology|www.plosbiology.org 2213 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions bersofatleastnineofthe70describeddivisionsofBacteria bacteria was determined in the cecum of each mouse at the [9], although it is dominated by just two: the Firmicutes and timeoftheirsacrifice.Thisguthabitatwaschosenasthesite the Bacteroidetes. One methanogenic archaeon, Methanobre- for our analyses because it is a readily defined anatomic vibacter smithii, which belongs to the Euryarchaea, is very structure, interposed between the distal small intestine and prominent,comprisingupto10%ofmicrobesinthecolonic colon, that normally houses a very dense and diverse microbiota [10]. Variations in the microbial composition of microbiota (.1012 organisms/ml luminal contents; seven the gut exist between individuals, primarily at the level of divisions of bacteria in a recent comprehensive 16S rRNA speciesandsubspecies[8].Moreover,withinagivenintestinal gene sequence-based survey) [16]. The cecal microbiota is habitat (address), some microbial members appear to be sustained, at least in part, by its ability to access large entrenched‘‘residents’’(autochthonouscomponents),where- quantitiesofplant-derivedpolysaccharidesthataredelivered as others act more like hitchhikers (allochthonous members) from the small intestine in an undigested state because the that pass through together with food, water, and other mouse(andhuman) hostlackstherequisiteglycoside hydro- components ofthe environment[11]. lases to break them down (e.g., xylans, arabinogalactans, and These considerations emphasize the importance of creat- polygalacturonates/pectins) [17,18]. ing simplified and defined model systems for studying the B. thetaiotaomicron and B. longum colonized the cecum in a functional properties and operating principles of our gut highlyreproduciblefashion.Moreover,microbialdensitywas microbial communities, and the consequences of adding not significantly different for either species when they were probiotics. In this report, we colonize germ-free mice with introduced alone (1010–1011 CFU per milliliter of cecal Bacteroides thetaiotaomicron with or without Bifidobacterium contents;p.0.05accordingtoStudentt-test).Furthermore, longum. B. thetaiotaomicron was chosen because it is a theco-colonization(bi-association)experimentrevealedthat prominentsaccharolyticmemberofourdistalgutmicrobiota thetwobacterialspeciescanco-existinthececumwithouta (6%ofallbacteriaintheenumerationstudydescribedabove) statistically significant impact on one another’s population [8], that possesses the largest arsenal of glycoside hydrolases size(unpublished data). and polysaccharide lyases among sequenced human gut To gain a genome-wide view of the effects of co- symbionts [12]. Thus, its highly evolved ‘‘glycobiome’’ helps colonization on each organism’s transcriptome, RNA was us to process otherwise indigestible glycans in our diet. B. isolated from the cecal contents of each animal in each longum is a normal, albeit less prominent member of our treatment group. The resulting RNA preparations were microbiota [8,13] that belongs to the Actinobacteria and is highly enriched for microbial RNA (.90% as judged by commonly used as a probiotic [14]. Since both bacterial quantitative reverse transcriptase PCR [qRT-PCR] assays of genomes have been sequenced, we were able to use a B. 16S and 18S rRNAs) [17]. Transcripts expressed in B. thetaiotaomicron/B. longum ‘‘community’’ GeneChip to simulta- thetaiotaomicron and B. longum were profiled using a custom neously examine how each organism affects the other’s ‘‘community’’GeneChip(Affymetrix,SantaClara,California, transcriptome in a specified intestinal habitat of a host United States) that contains oligonucleotides representing consuming a consistent and defined diet. Using a combina- 4,719 of B. thetaiotaomicron’s 4,779 predicted chromosomal tion of in silico reconstructions of microbial metabolism protein-coding genes (98.7%) [17] and 1,722 of B. longum’s based on these transcriptional profiles, mass spectroscopic 1,730 genes (99.5%). Each known or predicted gene in each analysis of glycan utilization, and whole genome transcrip- genome was represented by an average of 13 perfect match tional profiling of laser capture microdissected intestinal and control mismatch 25-mer ‘‘probe pairs’’ (together epitheliumfromgerm-freeandcolonizedmice,wewereable designated as a gene’s ‘‘probe set’’) (Table S1). Probe pairs to ascertain how a resident symbiont and probiotic species were designed to minimize cross-reactivity between genes affect one another’s carbohydrate substrate range and host containedwithinandbetweenthetwobacterialgenomes:this biology. The generality of our findings was explored by was confirmed experimentally by hybridizing the GeneChips testing the effect of two different host genetic backgrounds tocDNAtargetspreparedfromeachoffivececalRNAsfrom onthebehaviorofthismodelbi-componentmicrobiota,and one or the other group of mice colonized with a single bycomparingtheresponsesofB.thetaiotaomicrontotwoother bacterialspecies(‘‘mono-associated’’animals).Whenthefive probiotic strains added to a variety of fermented milks: B. thetaiotaomicron mono-association–derived cDNA targets Bifidobacterium animalis and Lactobacillus casei. werehybridizedtotheB.thetaiotaomicron/B.longumGeneChip, a‘‘present’’call(definedbyGeneChipsoftware)wasobtained fromanaverageof67.7%oftheB.thetaiotaomicronprobesets, Results/Discussion butfromonly4.3%ofB.longumprobesets.Similarly,cDNAs Bi-associationofGerm-FreeMicewithB.thetaiotaomicron generatedfromthececalRNAsofB.longummono-associated and B. longum mice produced ‘‘present’’ calls for 63.2% of B. longum probe Three groups of 7-wk-old, germ-free male mice belonging setsversus 5.6%ofB. thetaiotaomicron probesets.B. thetaiotao- to the NMRI inbred strain and maintained on a standard micronprobesetscalled‘‘present’’inatleastthreeofthefive polysaccharide-rich chow diet, were colonized with a single GeneChipshybridizedwithB.longumtargets(153probesets) gavage of 108 colony forming units (CFU) of either B. were designated as cross-reacting, and eliminated from thetaiotaomicron (ATCC 29148; isolated from the fecal micro- further analysis using an electronic probe mask. The biota of a healthy adult human) [15], B. longum (NCC2705; reciprocal analysis resulted in electronic masking of 84 recoveredfromaninfant)[14],orbothbacterialspecies(n¼5 cross-reactingB. longumprobe sets. animals/group). Mice were sacrificed 10 d later, a period of ThefiveGeneChipdatasetsproducedfromthegroupofco- timesufficientfortheintestinalepitheliumanditsoverlying colonized animals were then referenced to five GeneChip mucuslayertoturnoverseveraltimes.Thenumberofviable datasets from the B. thetaiotaomicron mono-associated group, PLoSBiology|www.plosbiology.org 2214 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions Figure1.ImpactofCo-colonizationonB.thetaiotaomicronandB.longumGlycosideHydrolaseandPolysaccharideLyaseGeneExpression (A) and (B) Bacterial glycoside hydrolase and polysaccharide lyase genes (rows) exhibiting differential expression in mono-associated versus co- colonizedmice.Colorsindicatethedeviationofagene’ssignalabove(red)andbelow(green)itsmeanexpressionvalueacrossalltensamples.Thirty- three B. thetaiotaomicron enzymes and seven B. longum enzymes meet the selection criteria for differential expression described in Materials and Methods(FDR,0.01).Fold-differencesinaverageexpressionlevelsbetweenthetwocolonizationstatesareindicatedinparenthesisforeachgene. (C)and(D)Averageaggregatesignalintensityinthetwocolonizationstatesforthesevengroupsofglycosidehydrolases/polysaccharidelyasesthatare commontoB.thetaiotaomicron(C)andB.longum(D).Thenumberofenzymesineachgroupisindicatedinparenthesis.Valuesobtainedfrommono- associatedmiceareshownwithblackbars,andfromco-colonizedmicewithwhitebars.SeeTablesS14andS15foralistoftheB.thetaiotaomicronand B.longumgenes,respectively.SeeMaterialsandMethodsforadescriptionofhow‘‘%maximumaggregatesignal’’wascalculated.Meanvalues6 standarderror(S.E.)offivesamplesareshown:anasterisk(*)indicatesp,0.05,anddoubleasterisks(**)indicatep,0.001accordingtoStudentt-test. DOI:10.1371/journal.pbio.0040413.g001 yieldingacohortofB.thetaiotaomicrongenesshowingchanged involved liberating oligo- and monosaccharides from poly- expressionasafunctionofcolonizationstate.Atotalof681B. saccharides (classified in Table S3 according to the nomen- thetaiotaomicron genes were designated as differentially ex- clature employed by the Carbohydrate Active enZymes pressed upon co-colonization using selection criteria that database [CAZy; http://afmb.cnrs-mrs.fr/CAZY/]). The pres- maintainedafalsediscoveryrate(FDR)oflessthan0.01:359 ence of B. longum produced statistically significant (FDR , (53%) of these genes were expressed at higher levels in the 0.01)changesinexpressionof31B.thetaiotaomicron glycoside presence of B. longum, whereas 322 were expressed at lower hydrolases and two polysaccharide lyases when compared to levels.Functionalcategorizationofthedifferentiallyexpressed B. thetaiotaomicron residing in the ceca of mono-associated genesusingclustersoforthologousgroups(COGs)revealeda mice: the vast majority of these genes (29/33) were up- ‘‘metabolism’’-biased response to co-colonization, including regulated(Figure1A).Thesubstratespecificitiesrepresented statisticallysignificantover-representationofup-regulatedB. inthe29up-regulatedglycosidehydrolases(20CAZyfamilies; thetaiotaomicrongenesinvolvedin‘‘carbohydratetransportand Figure 1A and Table S3) suggests that B. thetaiotaomicron metabolism’’ and ‘‘energy production and conversion’’ (p , undergoesanexpansioninthediversityofpolysaccharidesit 0.01basedonabinomialdistributiontest;FigureS1AandS1B; targetsfordegradation wheninthe presence of B. longum. seeTableS2foralistofgenessortedbyCOGs). B.thetaiotaomicroncontains163paralogsoftwocomponents Co-colonization Prompts B. thetaiotaomicron to Expand of a well-characterized starch utilization system (Sus): 106 Expression of Genes Involved in the Acquisition and paralogs of the gene encoding SusC, a TonB-dependent, b- Breakdown of Polysaccharides barrel-type outer membrane porin that participates in The B. thetaiotaomicron genome encodes 226 known or energy-dependent transport of malto-oligosaccharides into predicted glycoside hydrolases and 15 polysaccharide lyases theperiplasmicspace,and57paralogsofthegeneencoding PLoSBiology|www.plosbiology.org 2215 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions SusD, a secreted starch-binding protein with an N-terminal lipidtailthatallowsittoassociatewiththeoutermembrane [12]. These paralogs likely confer the ability to bind and import diverse carbohydrate substrates encountered in the gut[19].Expressionof31B.thetaiotaomicronSusC/Dgeneswas significantly altered (1.4–6.7-fold) by B. longum: nearly two thirds (19 of 31) display higher expression with co-coloniza- tion;12ofthesearepresentinthegenomeasadjacentpairs ofSusC/Dparalogs(Figure S2). Sixty-one B. longum genes showed significantly changed expression upon co-colonization with B. thetaiotaomicron (26 up-regulated, 35 down-regulated; FDR , 0.01). Together, these genes represent a much smaller proportion of the B. longum genome than does the cohort of B. thetaiotaomicron genesaffectedbyB.longum(3.5%versus14.2%,respectively). Comparison of B. longum genes up- versus down-regulated upon co-colonization revealed that the only significant difference in COG representation involved down-regulated genes associated with ‘‘carbohydrate transport and metabo- lism’’ (p , 0.01; Figure S3; Table S4). Moreover, all down- regulatedgenesinthisCOGarelocatedinoneregionofthe B. longum genome (BL1327–BL1337) that is dedicated to the degradation, import, and metabolism of mannosylated glycans.Allfourglycosidehydrolaseswithreducedexpression arepositioned withinthis genecluster(see below). Co-colonizationResultsinDifferentialExpressionofGenes Encoding Mannosidases and Related Enzymes Mannose is highly coveted as a source of energy for many saccharolytic bacteria: it is a ubiquitous monosaccharide foundinbothplantandanimalglycans,andiseasilyshunted intotheglycolyticpathwayviaanisomerizationofmannose- 6-phosphate to fructose-6-phosphate. The desirability of mannose as a substrate is evidenced by the fact that the B. thetaiotaomicron genome dedicates over 15% of its repertoire of glycoside hydrolases to the liberation of glycosidically linked mannose (30 mannosidases; ten mannanases). B. thetaiotaomicron’s 40 mannose-liberating enzymes contrasts withthe three mannosidases presentinB. longum. ThreeofB.thetaiotaomicron’smannosidasesaremorehighly expressedwithbi-association(BT2199,BT3530,andBT4093; 1.9-,2.3-,and1.7-fold,respectively),whereastwoofB.longum’s mannosidasesshowreducedexpression(BL1327andBT1329; 2.8-and6.7-fold,respectively)(Figure1).ThesetwoB.longum a-mannosidasesarepositionedinthegeneclusterreferredto above that contains two linked permease genes (BL1331 and Figure2.CarbohydrateUtilizationbyB.thetaiotaomicronandB.longum BL1332; down-regulated 7.4- and 10.1-fold, respectively) and inthePresenceorAbsenceofOneAnother two other glycoside hydrolase activities (BL1333 [substrate (A)and(B)Averageexpressionlevelsforfivececalsamplesareshownfor specificity unknown]; BL1335 [endo-b-N-acetylglucosamini- genesencodingenzymesthatshunttheindicatedmonosaccharidesinto dase]; down-regulated 8.8- and 14.1-fold, respectively; Figure the glycolytic or pentose phosphate pathways. Values obtained from 1Band1D).Expression ofnineofthe tengenesencoding B. mono-associatedmiceareshownwithblackbars,andfromco-colonized mice with white bars. Significant differences in expression levels were longumenzymesthatdelivermonosaccharidestotheglycolytic determinedusingStudentt-test;doubleasterisks(**)indicatep,0.01, and pentose phosphate pathways is unchanged by the andtripleasterisks(***)indicatep,0.001. presence of B. thetaiotaomicron (Figure 2). The lone exception (C)GC-MSdeterminationofmonosaccharidecompositioninthececaof germ-free (GF), B. thetaiotaomicron mono-associated (Bt), B. longum is another component of the gene cluster, a mannose-6- mono-associated(Bl),orB.thetaiotaomicron/B.longumco-colonized(Bt/ phosphate isomerase (BL1337) that shows 4.2-fold lower Bl) mice. Mean values 6 S.E. of biological triplicates are plotted. expressionwithco-colonization(p,0.001;Figure2B).Thus, Significant changes in the abundance of a monosaccharide were B. thetaiotaomicron and B. longum exhibit changes in their identified using a one-way ANOVA followed by a Tukey post test; a singleasterisk(*)indicatesp,0.05,doubleasterisks(**)indicatep, transcriptomethatsuggestdifferencesineithertheextentor 0.01,andtripleasterisks(***)indicatep,0.001. efficiencyofmannosideutilizationoccurduringco-coloniza- DOI:10.1371/journal.pbio.0040413.g002 tioncomparedtowheneitherspeciesisthesoleoccupantof the cecum. The apparent decreased catabolism of mannose PLoSBiology|www.plosbiology.org 2216 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions up-regulation of xylose-degrading enzymes by both B. by B. longum is consistent with its decreased expression of thetaiotaomicron and B. longum, xylose was depleted to a mannosidases. Nonetheless, despite the relative disparity in significant degree when both bacterial species were present mannoside-degrading enzymes between the two organisms inthe cecalhabitat (Figure2C). (40inB.thetaiotaomicronversusthreeinB.longum),wemustbe To determine if depletion of xylose in the co-colonized cautiousbeforeconcludingthatB.longumisarelativelypoor state could be attributed to differential usage by either competitorformannosylatedglycans.Forexample,intheco- bacterial species, we performed in silico metabolic recon- colonized cecum, other substrates may be available that B. structionstovisualizeexpressionofgenesencodingenzymes longumprefersovermannose,resultingindown-regulationof that shunt monosaccharides liberated by the glycoside its mannosidase genes, increases in mannose levels, and hydrolases/polysaccharide lyases into either the glycolytic or subsequentup-regulationofB.thetaiotaomicron’smannosidase pentosephosphatepathways.Withco-colonization,B.thetaio- genes. Alternatively, co-colonization could result in changes taomicronincreasesexpressionofonlyoneofitstengenesthat intheenvironment,suchaspHorosmolarity,thatrenderB. can be assigned to these pathways: a xylulokinase (BT0792) longum’s mannosidases more stable, or more efficient, result- that catalyzes a key step in D-xylose catabolism (3.1-fold ingin theirsubsequent down-regulation. increase with co-colonization, p , 0.01; Figure 2A). In B. thetaiotaomicron Up-Regulates Machinery Involved in contrast, neither of the two enzymes assigned to xylose- specific metabolism in B. longum—a xylose isomerase Xyloside Degradation and Xylose Catabolism in the (BL1704) and xylulokinase (BL1709)—showed differences in Presence of B. longum theirexpressionbetween thetwo colonizationstates. In contrast to their disparate responses to mannose- The observed up-regulation of genes involved in both containing glycans, expression of enzymes dedicated to liberation and catabolism of xylose by B. thetaiotaomicron degradationofxylose-containingglycansisenhancedinboth indicates that it is responsible for at least some of the bacterial species when they co-exist in the cecum. B. depletion that occurs during co-colonization. The up- thetaiotaomicron increases expression of four xyloside-degrad- regulation of xyloside-degrading machinery by B. thetaiotao- ing glycoside hydrolases: an endo-xylanase predicted to be micronandB.longummaybearesultofbothbacterialspecies secreted (BT3662; 1.8-fold), an exo-xylanase (BT3663; 2.4- turning to an available, but less-coveted substrate that is fold),andtwosecretedxylosidases(BT4095andBT4185;1.8- neglected in mono-association, but accessed when their and2.3-fold, respectively) (FDR ,0.01; Figure 1Aand1C). preferred substrates are depleted in the more competitive Two of the three glycoside hydrolases that B. longum up- environmentofco-colonization.Althoughlessparsimonious, regulates are secreted xylanases that occur in a putative an alternative possibility is that each bacterial species, when operon:BL1543,anendo-xylanase(2.3-fold),andBL1544,an present alone in the cecum, possesses a complement of exo-xylanase (1.8-fold), both of which contain sortase motifs glycosidehydrolasesthatisinsufficientforefficientdegrada- [20], indicating that they are located on the surface of this tion of available xylose-containing glycans. However, when Gram-positive bacterium (Figure 1B). Xylose-liberating hy- their repertoires of glycoside hydrolases are combined, drolasesaretheonlygroupofB.longumglycosidehydrolases complementary enzymes activities allow the two bacterial that show significantly higher expression when B. thetaiotao- speciestoparticipateina‘‘synergistic’’harvestofthisclassof micronis present(p, 0.001;Figure 1D). glycans, a phenomenon observed in other mixed microbial communities that degrade cellulose [21,22]. A third, formal VerificationofBacterialGlycanUtilizationBehaviorswithin possibility is that in mono-associated mice, each species the Ceca of Colonized Mice restrains expression of these genes (despite the presence of The coincident regulation of glycoside-degrading machi- substrate), but with co-colonization, in response to some nery by B. thetaiotaomicron and B. longum suggests that when signal from the other species, they up-regulate the genes to both components of this model community are present in thecombinedbenefitofbothspecies.Thisiscooperation(the their cecal habitat, utilization of a subset of available sacrificeofindividualfitnessforthefitnessofthegroup)and polysaccharide substrates changes. To examine this notion is unlikely to be taking place because of the inevitability of further, gas chromatography–mass spectrometric (GC-MS) ‘‘cheating’’ [23]. analysis was employed to establish which monosaccharide componentsofpolysaccharidesweresignificantlydepletedin B. longum Induces an Expansion of B. thetaiotaomicron’s each colonization state. When compared to germ-free Substrate Range Independent of Host Genotype controls, mono-association with B. thetaiotaomicron produced MicebelongingtotheNMRIinbredstrainwereselectedas apreferentialdepletionofthethreemostprominenthexoses the initial host for our experiments because of the ease of (glucose, galactose, and mannose), whereas there were no rearingtheminagerm-freestate.Todeterminethegeneral- significant changes in the most prominent pentoses (arabi- ity of B. thetaiotaomicron’s response to co-colonization with B. noseandxylose).B.longummono-associationalsoresultedin longum, we also performed a series of experiments, identical significant depletion of galactose, glucose, and mannose: to those described above (including use of the same however, unlike B. thetaiotaomicron mono-association, the autoclaved chow), but in animals belonging to the C57BL/6J pentose arabinose was also depleted (Figure 2C). This latter (B6)inbredstrain.Tenday–longmono-andbi-associationsof observation is consistent with B. longum’s considerable germ-free 6–8-wk-old male B6 mice produced levels of arabinoside-degrading capability (11 arabinosidases, ex- colonization similar to that documented in their NMRI ceeded only by glucosidases as the most highly represented counterparts (p.0.05;unpublished data). class of glycoside hydrolases in its genome; Table S3). Weidentifiedgeneswhoseexpressionchangessignificantly Consistent with the documented co-colonization–associated asafunctionofhostgenotype,byapplyingatwo-stagemixed PLoSBiology|www.plosbiology.org 2217 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions analysis of variance (ANOVA) model to our 20 GeneChip hydrolase family 20) and a-fucosidases (glycoside hydrolase datasets (i.e., five NMRI B. thetaiotaomicron mono-associated, family 95), the enzymes that liberate monosaccharides fiveNMRIB.thetaiotaomicron/B.longumco-colonized,fiveB6B. commonly found in host mucus glycans, which are coveted thetaiotaomicron mono-associated, and five B6 B. thetaiotaomi- substrates of B. thetaiotaomicron in vivo [17]. Together, these cron/B. longum co-colonized). The ANOVA model identified a data indicate that B. longum induces a shift in B. thetaiotaomi- totalof2,369B.thetaiotaomicrongenesthatweredifferentially cron substrate preferences away from B6 and NMRI mucus expressed in the two host mouse strains independent of glycans and towards arabinose- and xylose-containing plant colonizationstate(50.2%ofthe4,719genessurveyed;FDR, polysaccharides in the host diet. This change in glycan 0.01; see Figure S4, plus Table S5 for a full list of genes foraging behavior could also be responsible for altering the identified by the ANOVA model and the associated p-values functional relationships between B. thetaiotaomicron and its and q-values of each gene, as well as Tables S6 and S7, plus host(see below). Figure S5 for a COG-based functional categorization of the Expansion of the Carbohydrate Substrate Range of B. 1,168 B. thetaiotaomicron genes that exhibit higher expression thetaiotaomicron Can Be Induced by Multiple Probiotic in NMRI versus B6 mice, and the 1,201 genes that exhibit higherexpression inB6versus NMRI). Species Despite the large effect of host genotype on B. thetaiotao- Probioticbacterialspecies varyintheirabilitytoestablish micron gene expression, the B. longum–induced expansion in residency within the human intestinal microbiota. Another thecarbohydratesubstraterangeofB.thetaiotaomicronoccurs Bifidobacteriumspecies, B.animalis, andL.casei,amemberofa independently of host genotype. The ANOVA model identi- different bacterial division (the Firmicutes), are common fied1,806geneswhoseexpressionissignificantlychangedasa ingredientsoffermentedmilkproducts[24,25].Bothspecies, function of colonization state (mono-association or co- althoughlesswelladaptedthanB.thetaiotaomicronorB.longum colonization; Figure S6A). Comparison ofthe 801 genes that to take up residency in the human gut, can remain viable areexpressedathigherlevelsduringco-colonization,withthe during their transit through our gastrointestinal tracts [26– 1,005 genes expressed at reduced levels (Tables S8 and S9), 28]. revealed that ‘‘carbohydrate transport and metabolism’’ is TodeterminethespecificityofB.thetaiotaomicron’sresponse oneofthemostsignificantlyover-representedCOGs(15%of to B. longum, we performed expression profiling of B. the response, p , 10(cid:2)10; Figure S6B). Nearly twice as many thetaiotaomicron in the presence of either B. animalis DN-173 glycoside hydrolases/polysaccharide lyases exhibit higher 010 (Danone Vitapole, Jouy-en-Josas, France) or L. casei DN- expression during co-colonization compared to mono-asso- 114001(DanoneVitapole,CNCMnumber,I-1518).Toensure ciation (61 up-regulated versus 33 down-regulated; Figure continuous interaction between B. thetaiotaomicron and these S7A and S7B). Similarly, SusC/D paralogs whose expression two frequently ingested strains, B6 mice were mono- changesasafunctionofcolonizationstateshowastrongbias associated with each of the three species, or were first given toward up-regulation with co-colonization (43 up-regulated B. thetaiotaomicron followed by B. animalis or L. casei every 3 d versus 22 down-regulated; Figure S7A and S7B). In several overa10-dspan(fivemice/treatmentgroup).Importantly,at instances, co-regulated SusC/D paralogs and glycoside hydro- thetimeofsacrifice,thelevelofcecalcolonizationwassimilar lases/polysaccharide lyases are present together in putative inmono-orbi-associatedanimalsforallbacterialspecies(83 operons, a genomic feature that would allow B. thetaiotaomi- 1010 to331012;p.0.1). crontoefficientlycoordinateglycanbindingandimportwith Whole genome transcriptional profiling of cecal popula- glycan degradation (Figure S7Aand S7B). tions of B. thetaiotaomicron revealed that the impact of B. A comparison of the glycoside hydrolase/polysaccharide animalis and L. casei on the B. thetaiotaomicron transcriptome lyasefamiliesrepresentedinthegroupsofgenesthatincrease was modest when compared with the response elicited by B. ordecreasetheirexpressionduringco-colonizationirrespec- longum. Fifty-eight genes showed increased levels of expres- tive of host genotype further substantiated that B. thetaiotao- sion in the presence of B. animalis and 53 increased in the micron’s polysaccharide substrate preferences change in the presence of L. casei (FDR , 0.05; Figure 3A; see Tables S10 presence of B. longum. The 61 polysaccharide-degrading and S11 for gene lists). Moreover, the principal COG enzymesthatshowincreasedexpressionfallinto27different categories representing the response to B. animalis were glycoside hydrolase/polysaccharide lyase families (44 families ‘‘transcription’’ and ‘‘replication’’ (Figure 3B), whereas the are represented in the entire B. thetaiotaomicron genome), twoCOGcategoriesthatdominatetheresponsetoL.caseiare consistent with a broad diversification of glycan utilization. ‘‘carbohydratetransportandmetabolism’’and‘‘inorganicion The most highly represented glycoside hydrolase family, 43 transport and metabolism.’’ These latter two categories are (11members),providesevidencethatsubstratescontainingb- composed largely of glycoside hydrolase genes and SusC/D xylose and a-arabinose become a major focus of B. paralogs, respectively, and represent over 60% of B. thetaiotaomicron when B. longum is present (Figure S7C). thetaiotaomicron’s functionally categorized response (Figure Alpha-mannosidases and b-galactosidases (glycoside hydro- 3B). The enrichment of glycoside hydrolases (11 genes; four lase families 2 and 92, respectively) are also abundantly galactosidases,threehexoaminidases,twoarabinosidases,one represented (Figure S7C). amylase,andonesialidase)andSusC/Dparalogs(13genes)in The 33 glycan-degrading enzymes with decreased expres- the dataset of 53 B. thetaiotaomicron genes up-regulated by L. sionalsodisplaydiverseactivities:theyrepresent18glycoside casei (Figure 3C) is functionally similar to the B. longum– hydrolase/polysaccharide lyase families, and show a distribu- inducedresponse.ThisstandsincontrasttoB.animalis,which tion dissimilar to the up-regulated enzymes (Figure S7C). does not significantly impact expression of any glycoside Highly represented activities distinct from those present in hydrolaseorSusC/DparalogintheB.thetaiotaomicrongenome. theup-regulatedgroupincludeb-hexosaminidases(glycoside The L. casei up-regulation of three B. thetaiotaomicron PLoSBiology|www.plosbiology.org 2218 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions Figure3.ChangesintheB.thetaiotaomicronTranscriptomeinResponsetoCo-colonizationwithB.animalisorL.casei (A)B.thetaiotaomicrongenesthatshowsignificantlyincreasedexpressionlevelsuponco-colonizationwitheitherofthetwofermenteddairyproduct– associatedstrainsareshown.Selectioncriteria:(1)a1.2-foldorgreaterdifferenceinexpressionusingthe90%confidenceboundofthefold-change;(2) differenceinsignal(cid:3)25;(3)p-value(cid:4)0.025usingStudentt-test;(4)transcriptcalled‘‘present’’in66%ormoreoftheGeneChipsintheco-colonized group;and(5)amedianFDRlessthan0.05basedon50permutatedcomparisons. (B)COG-basedfunctionalclassificationofgenesidentifiedin(A).Thirty-fiveof53geneswithhigherexpressioninthepresenceofL.caseiand30of58 geneswithhigherexpressioninthepresenceofB.animaliscouldbeassignedtoCOGs.Functionalcategorieswithsignificantover-representationinthe dataset of differentially expressed genes, compared to their representation in the B. thetaiotaomicron genome, were determined using a hypergeometricdistribution;anasterisk(*)indicatesp,0.01. (C) All B. thetaiotaomicron SusC/D paralogs and glycoside hydrolases/polysaccharide lyases that show differential expression between the three colonizationstatesareup-regulatedinthepresenceofL.casei.Fold-differencesareindicatedinparenthesisandreflectaverageexpressionacrossthree cecalsampleswhenB.thetaiotaomicronisco-colonizedwithL.casei,comparedtotheiraverageexpressioninthreececalsamplesobtainedfromB. thetaiotaomicronmono-associatedanimals. DOI:10.1371/journal.pbio.0040413.g003 glycosidehydrolasefamily20hexosaminidasescontrastswith up-regulation of three enzymes that target arabinose-con- the down-regulation of this family’s hexosaminidases that taining polysaccharides (BT0360, endo-a-arabinosidase; occurs in the presence of B. longum (Figure S7C). Two of the BT0368,a-arabinofuranosidase;andBT4668,arabinogalactan induced hexosaminidases (BT0459 and BT0460) are present endo-b-galactosidase) suggests that accessing dietary plant- in a putative operon containing an up-regulated a-sialidase derived substratesis also partofthe responseto L. casei. andb-galactosidasededicatedtodegradingsialylatedglycans Insummary,comparisonofB.thetaiotaomicron’sresponseto fromhostmucus(BT0455–BT0461;Figure3C).Inadditionto the three bacterial species revealed that they are able to theexpandeddegradationofhostglycans,B.thetaiotaomicron’s induce distinctresponses inB. thetaiotaomicron, bothinterms PLoSBiology|www.plosbiology.org 2219 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions of the magnitude of the response (i.e., the number of B. Togainabroaderperspectiveofthehostresponsestoeach thetaiotaomicron genes that undergo significant changes in of the three colonization states, we categorized all up- expression)andthefunctionofaffectedgenes.Thedisparity regulated host genes using Ingenuity Pathways Analysis inidentityanddiversityofpolysaccharidesubstratestargeted software (http://www.ingenuity.com). A right-tailed Fisher by B. thetaiotaomicron in response to B. longum versus L. casei exact test was used to determine significantly over-repre- suggests that the nature of the expansion, the resulting sented(p,0.05)functionalcategoriescontainingaminimum impact on the intestinal nutrient environment, and ulti- of eight genes (Table S12). Gene interaction networks were mately the effect on the host, are likely to show significant then generated for each colonization state using these up- specificity with regardto themicro-organism consumed. regulated genes as ‘‘focus genes’’ among the thousands of genes called ‘‘present’’ for each condition (B. thetaiotaomicron Host Epithelial Response to B. thetaiotaomicron plus B. mono-association, 10,795 genes; B. longum mono-association, longum Is Characterized by a Synergistic Induction of 10,479 genes; and co-colonization, 12,138 genes). Unsuper- Interferon-Responsive Genes visedgenerationofmolecularinteractionnetworks,basedon The transcriptional responses of B. thetaiotaomicron and B. (1) focus genes, (2) genes called ‘‘present’’, and (3) unrepre- longumtoco-colonization,andtheresultingtransformationin sented but highly connected genes was performed by the the cecal nutrient environment raised the question of how Ingenuity Pathway Analysis software using a knowledge base host epithelial responses differ between colonization states. that consists of more than 1,000,000 functional and physical Therefore, we performed comprehensive transcriptional interactionsformorethan23,900mammaliangenes(includ- profiling on cecal epithelia harvested from the same groups ing 8,200 for mouse). These networks reveal that the host of NMRI mice that were used to profile bacterial gene response to B. thetaiotaomicron is centered on tumor necrosis expression(i.e., B. thetaiotaomicron mono-associated; B. longum factor-a (TNF-a; Figure S8A), an LPS-responsive cytokine mono-associated, and B. thetaiotaomicron/B. longum co-colon- producedbynaturalkiller(NK)andTcellsandmacrophages. ized). A group of age-matched, untreated male germ-free This contrasts to the network generated by the B. longum NMRI mice that were given the same batch of autoclaved responsewhichiscenteredontheTcell–producedcytokine, chow served asadditional controls. interferon-c(IFN-c;Figure S8B). Laser capture microdissection (LCM) was performed on The network generated from the co-colonization gene set cecalcryosectionspreparedfromfourmice/treatmentgroup revealed that both TNF-a and IFN-c are central highly (;5,000epithelialcellsrecovered/cecum).RNAwasprepared connected nodes (Figure S8C), suggesting that the cytokine from epithelial cells harvested from the cecum of each milieu in the cecal epithelium of co-colonized mice shares individual mouse in each treatment group. Equivalent characteristics witheach mono-associationstate. amounts of RNA from each cecum were then pooled from Consistent with the Ingenuity Pathway Analysis network each mouse per group, targets were generated by two construction,the groupofgenes showing highestexpression independent amplifications of each pooled RNA sample, in the co-colonized state relative to germ-free controls and each amplified RNA preparation was hybridized to a includes three members of the interferon-inducible 47kDa separate Affymetrix mouse 430v2.0 GeneChip. GTPases (Igtp, 5.6-fold; Ifi47, 6.6-fold; and Iigp2, 3.7-fold) and Expression profiles from each of the three colonization three members of the interferon-inducible p56 family (Ifit1, states were referenced to datasets obtained from germ-free 12.3-fold; Ifit2, 16.0-fold; and Ifit3, 3.2-fold). The interferon- controls to identify significantly up-regulated genes using stimulated gene-15 (Isg15), a ubiquitin family member criteria described in Materials and Methods. These compar- implicated in mediating viral resistance [29,30], is induced isons yielded 50 genes in the cecal epithelium of B. 17.1-fold with co-colonization (20.5-fold by qRT-PCR) com- thetaiotaomicron mono-associated mice, 24 genes in B. longum paredto2.0-fold(2.1-foldbyqRT-PCR)and1.2-fold(1.8-fold mono-associatedhosts,and52genesinco-colonizedanimals, by qRT-PCR) with B. thetaiotaomicron and B. longum mono- whose expression was defined as significantly increased associations, respectively (Figure 4B and 4C). Two other relative to germ-free epithelium. Comparison of genes up- interferon-responsive genes, Zbp-1, a Z-DNA binding adeno- regulated in each colonization state revealed that the host sinedeaminase,andMuc4,amucincapableofinhibitingviral responsetoco-colonizationwasdistinctfromtheresponseto capsid binding to target glycans [31] are up-regulated in all mono-association with either organism (Figure 4A and 4B): colonization states but show highest expression with co- i.e.,ofthe52genesidentifiedintheepithelialresponsetothe colonization(Figure 4B). two-component B. thetaiotaomicron/B. longum model commun- In addition to inducing host cytokine-responsive genes, B. ity, only 15 genes are shared with the response to B. longum appears to possess the capacity to regulate host thetaiotaomicronalone,andonlysevengeneswiththeresponse antibacterial proteins. Reg3c (regenerating islet-derived-3c) toB. longum. isaC-typelectinthatissecretedbyanepithelialcelllineage SincetheseconclusionswerebasedonaGeneChipanalysis that plays an important role in innate immune responses. ofpooledRNAs,weperformedfollow-upqRT-PCRassaysof Reg3cbindswithhighaffinitytopeptidoglycan,andexhibits laser capture microdissected cecal epithelial RNAs isolated broad microbicidal activity against Gram-positive bacteria from the individual animals used to prepare the pooled [32]. We found that although Reg3c is induced in mice samples (n ¼ 4/colonization state). Reg3c and Isg15 were colonized with B. thetaiotaomicron alone (5.2-fold above germ selectedforverificationbecausetheyhadexhibiteddifferent free; 3.6-fold by qRT-PCR), it is significantly down-regulated expression profiles across the three colonization states. The when B. longum is present by itself (repressed 8.5-fold qRT-PCR results established that GeneChip data obtained compared to germ free; 9.7-fold by qRT-PCR) (Figure 4B from pooled RNA samples were representative of individual and4C).Pancreatitis-associatedprotein(Pap;Reg3b),another animals withineachgroup (Figure 4C). memberoftheRegfamilythatalsocontainsalectindomain, PLoSBiology|www.plosbiology.org 2220 December2006|Volume4|Issue12|e413 Probiotic-Symbiont-HostInteractions Figure4.GeneExpressionChangesintheCecalEpitheliumInducedbyColonization PLoSBiology|www.plosbiology.org 2221 December2006|Volume4|Issue12|e413

Description:
microbiota [8,13] that belongs to the Actinobacteria and is commonly used as a probiotic [14]. Since both bacterial genomes have been sequenced, we were able to use a B. thetaiotaomicron/B. longum ''community'' GeneChip to simulta- neously examine how each organism affects the other's.
See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.