Diarrhea in young children from low-income countries leads to large-scale alterations in intestinal microbiota composition

MISSING IMAGE

Material Information

Title:
Diarrhea in young children from low-income countries leads to large-scale alterations in intestinal microbiota composition
Physical Description:
Mixed Material
Language:
English
Creator:
Pop, Mihai
Walker, Alan W.
Paulson, Joseph
Lindsay, Brianna
Antonio, Martin
Hossain, M Anowar
Publisher:
Bio Med Central (Genome Biology)
Publication Date:

Notes

Abstract:
Background: Diarrheal diseases continue to contribute significantly to morbidity and mortality in infants and young children in developing countries. There is an urgent need to better understand the contributions of novel, potentially uncultured, diarrheal pathogens to severe diarrheal disease, as well as distortions in normal gut microbiota composition that might facilitate severe disease. Results: We use high throughput 16S rRNA gene sequencing to compare fecal microbiota composition in children under five years of age who have been diagnosed with moderate to severe diarrhea (MSD) with the microbiota from diarrhea-free controls. Our study includes 992 children from four low-income countries in West and East Africa, and Southeast Asia. Known pathogens, as well as bacteria currently not considered as important diarrhea-causing pathogens, are positively associated with MSD, and these include Escherichia/Shigella, and Granulicatella species, and Streptococcus mitis/pneumoniae groups. In both cases and controls, there tend to be distinct negative correlations between facultative anaerobic lineages and obligate anaerobic lineages. Overall genus-level microbiota composition exhibit a shift in controls from low to high levels of Prevotella and in MSD cases from high to low levels of Escherichia/Shigella in younger versus older children; however, there was significant variation among many genera by both site and age. Conclusions: Our findings expand the current understanding of microbiota-associated diarrhea pathogenicity in young children from developing countries. Our findings are necessarily based on correlative analyses and must be further validated through epidemiological and molecular techniques.
General Note:
Pop et al. Genome Biology 2014, 15:R76 http://genomebiology.com/2014/15/6/R76; Pages 1-12
General Note:
doi:10.1186/gb-2014-15-6-r76 Cite this article as: Pop et al.: Diarrhea in young children from low-income countries leads to large-scale alterations in intestinal microbiota composition. Genome Biology 2014 15:R76.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
© 2014 Pop et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
System ID:
AA00024688:00001

Full Text

PAGE 1

RESEARCHOpenAccessDiarrheainyoungchildrenfromlow-income countriesleadstolarge-scalealterationsin intestinalmicrobiotacompositionMihaiPop1,AlanWWalker2,JosephPaulson1,BriannaLindsay3,MartinAntonio4,MAnowarHossain5, JosephOundo6,BoubouTamboura7,VolkerMai8,IrinaAstrovskaya1,HectorCorradaBravo1,RichardRance2, MarkStares2,MyronMLevine3,SandraPanchalingam3,KarenKotloff3,UsmanNIkumapayi4,ChineloEbruke4, MitchellAdeyemi4,DilrubaAhmed5,FirozAhmed5,MeerTaifurAlam5,RuhulAmin5,SabbirSiddiqui5, JohnBOchieng6,EmmanuelOuma6,JaneJuma6,EuinceMailu6,RichardOmore6,JGlennMorris8, RobertFBreiman9,DebasishSaha4,JulianParkhill2,JamesPNataro10andOColinStine3*AbstractBackground: Diarrhealdiseasescontinuetocontributesignificantlytomorbidityandmortalityininfantsand youngchildrenindevelopingcountries.Thereisanurgentneedtobetterunderstandthecontributionsofnovel, potentiallyuncultured,diarrhealpathogenstoseverediarrhealdisease,aswellasdistortionsinnormalgut microbiotacompositionthatmightfacilitateseveredisease. Results: Weusehighthroughput16SrRNAgenesequencingtocomparefecalmicrobiotacompositioninchildren underfiveyearsofagewhohavebeendiagnosedwithmoderatetoseverediarrhea(MSD)withthemicrobiota fromdiarrhea-freecontrols.Ourstudyincludes992childrenfromfourlow-incomecountriesinWestandEastAfrica, andSoutheastAsia.Knownpathogens,aswellasbacteriacurrentlynotconsideredasimportantdiarrhea-causing pathogens,arepositivelyassociatedwithMSD,andtheseinclude Escherichia/Shigella ,and Granulicatella species,and Streptococcusmitis/pneumoniae groups.Inbothcasesandcontrols,theretendtobedistinctnegativecorrelations betweenfacultativeanaerobiclineagesandobligateanaerobiclineages.Overallgenus-levelmicrobiotacomposition exhibitashiftincontrolsfromlowtohighlevelsof Prevotella andinMSDcasesfromhightolowlevelsof Escherichia/Shigella inyoungerversusolderchildren;however,the rewassignificantvariationamongmanygenera bybothsiteandage. Conclusions: Ourfindingsexpandthecurrentunderstandingofmicrobiota-associateddiarrheapathogenicityin youngchildrenfromdevelopingcountries.Ourfindingsar enecessarilybasedoncorrelativeanalysesandmustbe furthervalidatedthroughepidemiologicalandmoleculartechniques.BackgroundDiarrhealdiseasescontinuetobemajorcausesofchildhoodmortality,rankingamongthetopfourlargestcontributorstoyearsoflifelostinsub-SaharanAfricaand SouthAsia[1].Theproportionofdeathsattributedto diarrheaamongchildrenagedunder5yearsisestimated tobeapproximately15%worldwide[2],andashighas approximately25%inAfricaand31%inSouthEastAsia [3].Morethantwodozenentericpathogens,belonging todiversebranchesofthetreeoflife,areknownto causediarrheaandcanbetestedforinaclinicalsetting. However,itislikelythatadditionalpathogensremainto beidentifiedamongtheentericmicrobiota. Inresponsetoimportantunansweredquestionssurroundingtheburdenandetiologyofchildhooddiarrhea indevelopingcountries,theWilliamandMelindaGates FoundationcommissionedtheGlobalEntericsMulticenterStudy(GEMS)[4],whichrecentlyreportedthepathogensresponsibleforcasesofmoderate-to-severediarrhea *Correspondence: ostin001@umaryland.edu3UniversityofMaryland,SchoolofMedicine,Baltimore,MD,USA Fulllistofauthorinformationisavailableattheendofthearticle 2014Popetal.;licenseeBioMedCentralLtd.ThisisanOpenAccessarticledistributedunderthetermsoftheCreative CommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),whichpermitsunrestricteduse,distribution,and reproductioninanymedium,providedtheoriginalworkisproperlycredited.TheCreativeCommonsPublicDomain Dedicationwaiver(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle, unlessotherwisestated.Pop etal.GenomeBiology 2014, 15 :R76 http://genomebiology.com/2014/15/6/R76

PAGE 2

(MSD)insevenimpoverishedcountriesofsub-Saharan AfricaandsouthAsia.Importantly,forapproximately60% ofMSDcasesinGEMS,noknownpathogencouldbeimplicatedbyconventionaldiagnosticmethods[5].These observationshighlightthepotentialpresenceofpreviously undiscoveredpathogens,and/orpossibleinteractionsbetweenpathogensandothermembersoftheintestinal microbiota(bothpathogenicandcommensal)thatmayeitherexacerbatetheclinicalmanifestationorprotectthe hostfromdisease. HereweapplymoleculartechniquestosurveytheintestinalmicrobiotainasubsetofGEMScasesandcontrols. Ourstudycomprises992childrenfromfourunderdevelopedcountriesinWestAfrica(TheGambiaand Mali),EastAfrica(Kenya),andSouthAsia(Bangladesh), representingasubsetoftheover25,000GEMSchildren enrolled.Ourresultsshedadditionallightonpotential mechanismsunderlyingMSDinchildrenofdeveloping countries.Priortopresentingtheseresultswewouldlike tostressthatouranalysesare,bynecessity,correlative andtheresultspresentedheremustbevalidatedthrough epidemiologicalandmolecularanalyses,severalofwhich arealreadyunderway.ResultsanddiscussionDescriptionofdataOurdatacompriseroughlyequalproportionsofcases andcontrols(0.51 vs. 0.49,respectively)fromfoursites: Bangladesh(N=206),TheGambia(N=269),Kenya(N= 305),andMali(N=212).Approximately55%ofthesubjectswereboys.Of992samples,508werefrompatients withMSD(Table1).Thechildrenrangedinagefrom newbornto59months.Westratifiedthemintofiveage categories:0to5months(N=112),6to11months(N= 308),12to17months(N=173),18to23months(N= 146),and24to59months(N=253).Therewereno significantdifferencesbetweentheproportionofcases andcontrolsineachcountryandfromeachagegroup (Table1).ThesequencingofPCRamplified16SrRNA genesresultedin3,584,096readspassingqualitychecks. Eachsamplehadatleast1,000reads,andtherewerean averageof3,613readspersample.ThereadswereclusteredusingDNAclust[6]into97,666operationaltaxonomicunits(OTUs).Ofthese,21,247passedchimera checking,weredetectedinmorethanfivesamples,orrepresentedatleast20sequencesinasinglesample,andwere includedinfurtheranalysis.ThenumberofOTUsper samplerangedfrom55to1252,withamedianof380and anaverageof412.ThemeanOTUsizewas138,ranging from5(bydefinition)to192,978(withmedianOTU size=15sequences).Representativesequencesfrom the21,247OTUsmatched728distincttaxafrom161 genera.Amongthese,4,730(22%)didnothavegood (>100bpexactmatch,>97%identity)matchestoisolate sequencesfromtheRibosomalDatabaseProject(RDP). Thesewereflaggedas ‘ unassigned ’ inouranalysisandare discussedfurtherbelow.Thesesequencesarenotsimply anartifactofourstringentalignmentcriteriaasevidenced bythefactthatare-analysisofthe6,879mostabundant OTUsusingthereference-basedOTUpickingalgorithm implementedinQiime[7]failedtoclassifyasimilarproportionofsequences(2,162or31%oftheabundant OTUs).MicrobiotavariationsbyageThewelldocumented[8-10]successionoftheintestinal microbiotaduringchilddevelopmentisapparentinour non-diarrhealcontrolsamples(Figure1A).Duringthe firstyearoflife,the ‘ healthy ’ gutmicrobiotainourinfant cohortsischaracterizedbycomparativelylowoveralldiversityandarelativelyhighproportionoffacultatively anaerobic,andpotentiallypathogenic,organisms(forexample,the Escherichia/Shigella group,whichcannotbe distinguishedfromeachotherby16SrRNAgenesequences),organismsthatarebelievedtoplayarolein thedevelopmentofthehostimmunesystem[11,12].In olderages,thedominanceoftheseorganismsisreduced, replacedbyacorrespondingincreaseinoveralldiversity (Figure1B),accompaniedbyaparticularlypronounced increaseintheproportionalabundanceofthebacterial Table1DemographicsofthechildrenDemographiccharacteristicsforsamples(N=992),N(%) MSD N=508 Controls N=484 P value Total N=992 Agegroupsbymonths0.1788 0to558(11)54(11)112(11) 6to11171(34)137(28)308(31) 12to1793(18)80(17)173(17) 18to2370(14)76(16)146(15) 24to59116(23)137(28)253(26) Country0.3622 TheGambia138(27)131(27)269(27) Mali110(22)102(21)212(21) Kenya165(32)140(29)305(31) Bangladesh95(19)111(23)206(21) Gender0.5785 Male286(56)264(54)550(55) Female222(44)220(46)442(45) Dysentericstools 140(28)7(1)<10-16147(15)Allagesareinmonths. P valuestestindependenceofMSDcasesandcontrols withregardstodemographicvariable. P valuesforageinmonths(treatedasa continuousvariable)computedbyindependentsamplest-test. P valuesfor categoricalvariablescalculatedusingchi-squaretest. MSD:Moderate-to-severediarrhea.Pop etal.GenomeBiology 2014, 15 :R76 Page2of12 http://genomebiology.com/2014/15/6/R76

PAGE 3

genus Prevotella .Thesechangesaremostevidentinour non-diarrhealcontrolsamples,wherethegenus Prevotella increasesfromapproximately12%toapproximately 48%proportionalabundanceduringthefirst5yearsof life,whilethe Escherichia genusdropsfromabout20% proportionalabundanceininfantsunder6monthsof agetoapproximately1%in2-to5-year-olds(Additional file1:TableS1).Twoothergenera, Veillonella and Streptococcus alsoexhibitsignificantdecreaseswithincreasingage.Ourdataalsoshowanincreasewithincreasing ageintheproportionofarangeoforganisms(labeled ‘ unassigned ’ inFigure1AandAdditionalfile1:TableS1) thathavenogoodqualitymatchestoculturedisolatesin publicdatabases,andwhichappeartobelongpredominantlytoobligateanaerobicbacteria(over60%canbe assignedbytheRDPclassifiertothe Ruminococcaceae and Lachnospiraceae familiesoftheFirmicutesphylum, whicharerelativelypoorlyrepresentedinculturecollections[13],aswellasthe Bacteroidaceae family).These previously-unculturedputativeobligateanaerobesincreaseinproportionalabundancefromapproximately 8%indiarrhea-freeyoungch ildrentoapproximately Figure1 Comparisonofdiarrhealandnon-diarrhealstool.(A) Proportionalabundanceofgenerainnon-diarrhealcontrolsandMSDcasesin differentagecategories.Eachcolorrepresentsadifferentgroup.Theorderandcolorforeachgroupisthesameforcontrols(patientswithout MSD)andMSDcases.Theeightgroupsmostfrequentlyfoundincontrols( Prevotella,Bacteroides,Escherichia/Shigella,Veillonella,Streptococcus, Lactobacillus,Faecalibacterium,Megasphaera ,plusunassignedandother)aredepicted. (B) Shannondiversityindexacrossagesanddiarrheal status.AverageShannondiversityindicesforthefivedifferentagestrataaswellasthecorresponding95%confidenceintervals.Bothcasesand controlsexhibitedhighermeanShannondiversityindexscoresathigheragegroupscomparedtoloweragegroups( P <0.001,one-wayANOVA). Thediversityofhealthysamplespositivelycorrelateswithageinthefirst2yearsoflife,aspreviouslyreported[12].Thediversityindexforcase sis significantlylessthanthatforcontrolswithineachcountry( P <0.02,Tukey ’ st-testcorrectedformultiplecomparisons).AlsoseeAdditionalfile2: TableS2andAdditionalfile3 : FigureS1. Pop etal.GenomeBiology 2014, 15 :R76 Page3of12 http://genomebiology.com/2014/15/6/R76

PAGE 4

23%intheolderagegroup,consistentwithincreasein diversitywithintheintest inalmicrobiotaandtheknown expansionofthesegroups,whichareabletocolonize theintestineingreaternumbersasthecomplexpolysaccharidestheyutilizeforgrowthbecomeagreater featureofthehostdiet[14]. Theseobservationsbroadlyholdwhenstratifyingby countryoforigin;however,country-specificeffectsare alsoapparent.Forexample,thesamplesfromBangladesh aredifferentfromtheAfricancountries,particularlyin theyoungeragegroups,andarecharacterizedbyalower proportionof Prevotella sequencesandahigherproportionoforganismsfromthe Escherichia/Shigella and Streptococcus genera(Figure1AandAdditionalfile1:TableS1). Thepatternsobservedwithincontrolsampleswere significantlydifferentfrompatternsfrompatientswith MSD;however,someoverallage-relatedtrendswere similar.Forexample, Prevotella abundancecorrelates withage,albeitreachingamuchlowerpeak,withonly 23%abundanceintheoldestagegroup( vs. 48%incontrols, P <1016).Otherobligateanaerobicmicrobeshave lowerproportionalabundanceamongcasescompared tocontrols: Bacteroides andtheunclassifiedputative anaerobesareboth5%lowerincases,consistentwith previousobservationsthatindicateintestinaldysbiosisis associatedwithadecreaseintheproportionalabundanceofobligateanaerobes[15].Amongcases, Escherichia/Shigella and Streptococcus spp.maintainahigh proportionacrossallagegroups,thoughtheirpreponderancedropssignificantly(41%to13%and18.5%to 7.5%,respectively)aschildrenage.Furthermoreitappearsthat Prevotella and Escherichia/Shigella arenegativelycorrelatedinMSDcases(Spearmanrho=-0.55, P <0.0001).Thedisruptionassociatedwithdiarrheais alsoreflectedinlowerdiversityvaluesinMSDcasesin everyagegroup(Figure1B,Additionalfile2:TableS2, Additionalfile3:FiguresS1A-D). Country-specificeffectswerealsoobservedindiarrhealstool;forinstance,inKenya,diarrheaappearedto havealessmarkedeffectonthemicrobiota(Figure1A andAdditionalfile1:TableS1). Escherichia/Shigella spp. weremostcommoninMali,accountingfor34%ofthe sequences,nextmostcommoninBangladesh(24%)and leastcommoninTheGambia(15%). Prevotella spp. werefoundinhighproportionalabundancesinThe Gambia(18%)andKenya(19%).Thegenus Streptococcus isfoundinrelativelyhighabundancesinBangladesh (21%)andTheGambia(13%)withlowerabundancesin Mali(10%)andKenya(9%).Asexpected,thetaxonomic diversity(Shannondiversityindex)issignificantlydifferentbetweencasesandcontrolsinallcountries( P <0.005, pairwiset-test).Ofnote,where Prevotella ismorecommon(TheGambiaandKenya),thediversityishigher (Figure1B).Taxonomicgroupsstatisticallyincreasedordecreasedin diarrheaMultidimensionalscalinganalysiscouldnotseparate thediarrheaanddiarrhea-fr eebacterialcommunities duetohighinter-personalvariation(Additionalfile3: FigureS3).Weestimatedtheassociationofindividual OTUswithdiseaseusingstatisticaltestsaddressing bothpresence-absencestatistics(Fisher ’ sexacttestand logisticregression)andabundance-dependentstatistics (usinggeneralizedlinearmodels)thataccountforthe numberofOTU-specificsequencesineachstool,and potentialconfounderssuchassamplingdepth,age,and country(seeAdditionalfile4:TableS3forafullsummary).Theformeraddresssimilarquestionstothose commonlytargetedbythetraditionalculture-basedepidemiologicalstudies,whilethelatterallowustoassess howpathogenproportionalabundancecorrelateswith morbidity. TenOTUswerefoundtobepositivelyassociatedwith diarrheabyallstatisticaltests.TheOTUsassociated withMSDhavehigh-similaritymatchesagainstdatabase sequencesfrombacterialtaxainthe Escherichia/Shigella Granulicatella spp ,and Streptococcusmitis/pneumoniae groups.Whenonlyabundance-dependentstatisticsare usedtodeterminesignificance,anadditional18OTUsare foundtobehighlyassociatedwithdiarrhea,corresponding tothebacterialspecies Escherichia/Shigella Campylobacterjejuni, and Streptococcuspasteurianus .Whenonly consideringpresence/absencestatistics,43additional OTUsarefoundtobeassociatedwithdiarrhea,comprisingthebacterialgroupsalreadydiscussedabove aswellasmembersofthegenera Lactobacillus Neisseria Citrobacter Erwinia, and Haemophilus .Itisnoteworthy thatalloftheseorganismsareeitherfacultativelyanaerobicormicroaerophilic. Ontheotherhand,therewerenoOTUspositivelyassociatedwithhealthystoolsbybothstatisticalmethods, reflectingthehigherdegreeofinter-individualvariation inmicrobiotacontentinhealthyindividuals.Consideringonlypresence/absencestatistics,thereare43 OTUsassociatedwithnon-diarrhealcontrolsamples. Thegeneraassociatedwiththesecontrolsamplesincludemembersoftheclostridialfamilies Peptostreptococcaceae Eubacteriaceae, and Erysipelotrichaceae ,and thegenera Clostridiumsensustricto Dialister Enterococcus Prevotella Ruminococcus, and Turicibacter .When consideringonlyabundancestatistics,anadditional19 OTUsaresignificantlyassociatedwithnon-diarrheasamplesandhavehighqualitymatchestodatabasesequences correspondingto Bacteroidesfragilis Dialister Megasphaera Mitsuokella/Selenomonas Prevotella spp.,and Clostridiumdifficile .Thus,itcanbeseenthatmanyobligateanaerobicbacteriallineagescorrelatewithhealthy status.Pop etal.GenomeBiology 2014, 15 :R76 Page4of12 http://genomebiology.com/2014/15/6/R76

PAGE 5

FunctionaldifferencesbetweencasesandcontrolsThebroadstatementsmadeaboveaboutoxygentolerance inthediseasedmicrobiotaaresupportedbyPICRUST [16]analysesofourdata.Specifically,thisshowedputative signaturesofobligateanaerobicgutlineagestobeenrichedinthediarrhea-freesamples(forexample,glycolysis, P =10-9;pyruvatemetabolism, P =10-7;shortchain fattyacidbiosynthesis, P =10-3;xylenedegradation, P = 10-7;andsoon;all P valuesbyWelch ’ st-testascomputed bySTAMP[17]),whileoxygendependentpathways(for example,theTCAcycle, P <10-15)areenrichedindiseased samples.TaxonomicgroupscorrelatedwithdysenteryWesegregateddiarrhealstoolbasedondiagnosisofdysentery(presenceofblood)andfoundatotalof30OTUs thatwerestronglycorrelatedwithdysenterywhencomparingwithnon-dysenterydiarrhealstool(metagenomeSeq[18], P <0.05).Theseincludeseveralwell-known pathogenssuchas Enterococcusfaecalis Campylobacter jejuni Bacteroidesfragilis Clostridiumperfringens Enterobactercancerogenus ,andmembersofthe Granulicatella Haemophilus Klebsiella, and Escherichia/Shigella genera.Alsoassociatedwithdysenteryweremembersof the Streptococcuspasteurianus and Streptococcussalivarius groups.AsingleOTU,correspondingto Lactobacillusruminis ,wasfoundtobenegativelyassociated withdysentery.Agenus-levelrepresentationofthese findingsisshowninFigure2.NetworkviewofdiarrhealillnessTheoverallresultspresentedabovearealsoborneout incorrelationnetworksconstructedfromthedata (Additionalfile3:FigureS7).Atthebroadlevel,inboth MSDcasesandcontrols,itcanbeseenthattheretendto benegativecorrelationsbetweenfacultativeanaerobiclineagesandobligateanaerobiclineages.Themostobvious exampleisthenegativecorrelationofthepotentiallyprotective Prevotella genuswiththatofpotentialpathogens suchas Escherichia/Shigella .Similarly,therearealso positivecorrelationswithinthesetwophenotypicsubgroupings,suchthatobligateanaerobicgenerasuchas Prevotella Roseburia, and Dialister arecorrelatedwith eachother,whilefacultativeanaerobicormicroaerophilic generasuchas Streptococcus Lactobacillus Escherichia/ Shigella ,andotherProteobacteriaarealsocorrelatedwith eachother.Thediarrhea-freenetworkappearstobemore tightlyconnectedthanthediarrhealnetwork,consistent withecologicaltheoriesthatequateenvironmentdiversity andconnectednesswithecosystemstability/health[19,20]. Atthesametime,wewouldliketonotethatourdatado notallowareliablequantitativeassessmentofsuchphenomenaduetothelargelevelofinter-personalvariation.DiscussionOuranalysisofthe16SrRNAgene-basedtaxonomic profileofdiarrhealandcontrolstoolsampleshasdemonstratedastrongassociationbetweenacutediarrheal diseaseandtheoveralltaxonomiccompositionofthe stoolmicrobiotainyoungchildrenfromthedeveloping world.Wehaveidentifiedstatisticallysignificantdisease associationswithseveralorganismsalreadyimplicatedin diarrhealdisease,suchasmembersofthe Escherichia/ Shigella genusand C.jejuni .Inaddition,wehaveuncoveredanassociationwithdiarrhealdiseaseforseveral organismsnotwidelybelievedtocausethisdisease,such as Streptococcus and Granulicatella .StreptococcalOTUs associatedwithdiseaseprimarilybelongtoeitherthe Streptococcuspneumoniae/mitis group(indistinguishable withinthe16SrRNAgeneregionstargetedbyourstudy), whichcontainsseveralimportanthumanpathogens,or the Streptococcuspasteurianus group.Theseresultsmerit furtherexplorationasrecentstudiesprovideevidenceof Streptococcus -relateddiarrhealcases[21,22].Itisimportanttostressthatpathogenicityisonlyoneofmanypossibleexplanationsforthesefindingsandtheorganisms associatedwithdiseasestatusmayalsoeither:(1)usually inhabittheupperGItractandbecomeapparentindiarrhealstoolduetodislodgingandreducedtransittimeduringdisease;(2)thriveindisturbedgutenvironments;(3) maybebetterabletopersist/resistdislodgementduringa diarrhealpurge;or(4)acombinationofpathogensmay causediseaseinthesechildren[23].Priorevidencecertainlysuggeststhatfacultativeanaerobes(manyofwhich wefindassociatedwithdiarrhea)tendtoflourishinavarietyofperturbedgutenvironments,possiblybecausethe reducingpowerofthemicrobiotaisaffectedbythelossof obligateanaerobesfollowingperturbation[15].Anycausalitywouldneedtobedemonstratedthroughfurtherexperimentation.Atthesametime,streptococciarealso foundinourstudytobeassociatedwithmoresevere formsofdiarrhea(dysentery),therebystrengtheningthe caseforapossiblecausalconnection.Despiteuncertainty regardingthecausesandeffectsofmicrobiotaperturbationsinthesettingofMSD,dissectingthephysiologic implicationsiswarranted.Forexample,anincreasein streptococcalorotherspeciesinthesettingofdiarrhea mayconferorexacerbatediarrhealeffects. S.mutans has recentlybeenpostulatedtohavearoleinhumanenteritis. Ourworkrepresentsanimportantfirststepinunderstandingthecomplexinteractionbetweenmicrobiotaand diarrhealpathogensindevelopingcountrysettings. Ourstudyhasalsorevealedahighprevalenceofmembersofthe Prevotella genus(primarily Prevotellacopri ) inthestoolofdevelopingworldchildren,aswellasthe negativecorrelationofthisgenuswithdisease.Theseorganismsareprevalentinthedevelopingworld[14],yet arerelativelypoorlystudiedduetofairlylowprevalencePop etal.GenomeBiology 2014, 15 :R76 Page5of12 http://genomebiology.com/2014/15/6/R76

PAGE 6

intheindustrializedworld[24].Samplescontaininghigh proportionsofmembersofthe Prevotella genusalso havehigheroverallbacterialdiversity,potentiallydriven bythelevelofcomplexpolysaccharides/starchyfiberin thediet.Recentevidencesuggeststhat Prevotella spp. areparticularlyabundantinruralAfricanchildrenconsumingahighfiberdiet[25].Thisisinstarkcontrastto Westernchildren,whotypicallyhavemuchhigherabundancesof Bacteroides spp.,andverylittle Prevotella ,a differencethatisbelievedtobelinkedtodiet[26]. Figure2 Comparisonofdysentericandnon-dysentericstool.(A) Genus-levelcomparisonofdysentericandnon-dysentericdiarrhealstool (top)stratifiedbyage;(bottom)stratifiedbycountry. (B) Proportionalabundanceboxplotsof Prevotella,Lactobacillus ,and Streptococcus in dysentericandnon-dysentericdiarrhealstoolsbyagecategory.Theupperwhiskerextendsfromthe75thpercentiletothehighestvaluethatis within1.5*IQRofthehinge,whereIQRistheinter-quartilerange,ordistancebetweenthefirstandthirdquartiles.Thelowerwhiskerextends fromthehingetothelowestvaluewithin1.5*IQRofthehinge.Databeyondtheendofthewhiskersareoutliersandarenotplotted.Asterisks abovethewhiskerindicateastatisticallysignificantdifference(byt-test)betweendysentericandnon-dysentericstoolsplacedinthepanelwit h themoreabundantmean.Asingleasteriskindicates P <0.05;doubleasterisksindicate P <0.01. Prevotella issignificantlyassociatedwith non-dysentericcasesoverall( P =0.0003)andinagegroups0to6months( P =0.01),12to17months( P =0.03),and24to59months( P =0.001). Lactobacillus issignificantlyassociatedwithnon-dysentericcasesoverall( P =0.0002)andinchildren6to11months(0.02)and12to17months ( P =0.003),whilethegenus Streptococcus isassociatedwithdysenteryoverall( P =0.007),particularlyinchildrenaged12to17months( P =0.01). Pop etal.GenomeBiology 2014, 15 :R76 Page6of12 http://genomebiology.com/2014/15/6/R76

PAGE 7

Ourco-occurrencenetworkanalyses(Additionalfile3: FigureS7)andproportionalabundanceanalysis(Additional file1:TableS1)suggestpotentialnegativeinteractions between Prevotella andentericpathogens,suchasmembersofthe Escherichia/Shigella genus,raisingthepossibilityforthedevelopmentofnovel Prevotella -based therapeuticstrategies.Anotherpossibleprobioticorganismidentifiedinourstudyis Lactobacillusruminis .This organismwasfoundtobeassociatedwithnon-diarrheal stoolandalsowithlesssevereformsofdiarrheawhen comparingdiarrhealtodysentericstool.Althoughtheincreaseinfrequencyofthesetaxaindiarrheacouldbedue toshortenedintestinaltransittime,thedifferenceinprevalenceof Lactobacillus betweencasesofMSDanddysenteryarelesslikelytorepresentthiseffect. Lactobacillus ruminis hasimmunomodulatorypropertiesandhasbeen previouslysuggestedasapotentialprobiotic[27]. AmongOTUsfoundassociatedwithnon-diarrheal stoolaresequencesclassifiedas Clostridumdifficile ,a surprisingfindinggiventhatthisorganismisacommon causeofentericdisease,primarilyinhospitalizedelderly patients.However,although C.difficile canbeanimportantpathogen,itisactuallycarriedasymptomaticallyby around60%ofinfants[28].Wealsofoundaconflicting associationofOTUsassignedas Bacteroidesfragilis with boththediarrhea-freestatusanddysentery,afinding thatcanperhapsbeexplainedbystrain-to-strainvariation.Enterotoxigenic B.fragilis strainsarewellcharacterizeddiarrhealagentsinchildren[29]whereas,in contrast,non-toxigenic B.fragilis hasbeenlinkedto anti-inflammatoryprotectiveeffectsinmousemodels [30].Itisthereforepossiblethatdifferentstrains,which cannotbedifferentiatedthrough16SrRNAgenesequencing,mightaccountfortheseopposingresults. Ourstudyidentifiedmanysequencesthatdonothave goodmatchesagainstculturedorganismsincurrent16S rRNAgenedatabases.Manyofthesesequencesonlyhave high-qualitymatchestootheruncultivatedanduncharacterizedintestinalmicrobes,highlightingthepresenceofa largereservoirofuncharacterizedmicrobesintheintestinaltractofchildrenwithinthedevelopingworld,as reportedbefore[31].ManyoftheunknownsequencesappeartobelongtoobligateanaerobiclineagesoftheFirmicutesphylum,whichareunder-representedinculture collectionscomparedtootherintestinaldwellinggroups suchas Bacteroides andbifidobacteria.Theprevalenceof such ‘ unknown ’ sequencesishigherincontrolsandseveral oftheseuncharacterizedorganismsexhibitstrongassociationswithdiarrhea-freesamples,highlightingtheirpotentialroleinthemaintenanceofahealthygutmicrobiota, andsuggestingtheneedforabetterin-depthcharacterizationofthegutmicrobiotaofchildrenwithinthedevelopingworld,complementingresourcesrecentlydeveloped inEurope[32]andtheUS[33]. Ourobservationsrelatedtothemicrobialsuccession inthedevelopinginfantgutmicrobiotacarryseveralcaveats.Asinglesamplewascollectedfromeachchildata singlepointintime,andwelackextensivedataonprior historyofdiarrhea.Whilethedataaresuggestiveofa progressioninmicrobiotastructure,monitoringofa birthcohortwillbenecessarytofullyunderstandthe progressionofgutmicrobiota,andassesstheimpactof diarrhea(including,potentially,multipleepisodesof diarrhea)onthisprocess.Atatechnicallevel,wewould alsonotethattheprimersetsusedinthisstudy(targetingtheV1-V2hypervariableregionsofthe16SrRNA gene)donoteffectivelyamplifybifidobacteria[34,35], knowntobedominantmembersoftheintestinalmicrobiotaofbreast-fedinfants,butthisbiasislikelytobe uniformbetweencasesandcontrols.Wepurposefully selectedaprimersetbettertargetedtowardsbacterial groupscontainingknownandpotentialpathogens,such as Enterobacteriaceae ,toimproveourchancesofdetectingnovelpathogensatthecostofobtaininglessinformationaboutthealreadywell-establishedearlydominance bybifidobacteria. Ourstudyrevealedthelimitationsofexistingmolecularandbioinformaticsapproachesemployedinaclinical settingforperformingtaxonomicsurveysofstoolsamples.Theuseofthe16SrRNAgene,forexample,does notaffordasufficientdiscriminationwithintaxonomic groupscontainingknownorputativepathogens( Escherichia/Shigella Streptococcus ,andsoon)indicatingthe pressingneedforthedevelopmentofnewcost-effective andrelativelyunbiasedmolecularapproaches[36]forincreasingtheresolutionofepidemiologicalsurveyssuch asours.Relatedly,theaccuratetaxonomicassignmentof sequencesgeneratedinstudiessuchasoursishampered bynumerouserrorsinpublicdatabasesandbytheuse ofsimplistic‘ lowestcommonancestor ’ heuristicsbysoftwaretoolsfacedwithambiguoustaxonomicinformation. Theresultspresentedinthispaperwereobtainedthrough thecarefulmanualannotationofalltheOTUsfound tobeassociatedwithdiseasestate(seeAdditionalfile5: TableS4).Finally,wehadtodevelopanovelstatistical method[18]foridentifyingdiseaseassociationinorderto appropriatelyaddressdatararefactionaswellastocontrol forthehighinter-personalvariability,atypicalfeatureof thehealthygutmicrobiota[37],andotherconfounding factors.ConclusionsOverallourstudydemonstratesthatthemajordifferencesinthemicrobiotabetweendiarrhealandnormal stoolsarequantitativedifferencesintheproportionsof themostprevalenttaxa.Suchquantitativedifferences werealsoobservedinourpreviousqPCR-basedstudy wherewefoundthat80%(1,665/2,072)ofcontrolsandPop etal.GenomeBiology 2014, 15 :R76 Page7of12 http://genomebiology.com/2014/15/6/R76

PAGE 8

89%(1,307/1,461)ofMSDcaseshaddetectablelevelsof Shigella .Quantitativemeasurementsof Shigella abundancewerecriticaltoassessingattributablerisk[38]. Amongtheknowncausesofdiarrhea(rotavirus, Shigella Cryptosporidium ,Enterotoxigenic E.coli ,andsoon)the attributablefractionofdiarrheainyoungchildrenisestimatedtobejust43%[5].Ourstudyprovidesinitialevidencefortheexistenceofnovelpathogenicagents.The mostlikelycandidatesfromourstudyaremembersofthe Enterobacteriaceae andstreptococci,taxawhichalready containmanyknownhumanpathogens.Furtherexplorationoftheseorganismsisnecessarytobetterunderstand theirpathogenicpotentialandthelikelihoodoftheir emergenceasmajorpathogensthroughtheacquisitionof additionalpathogenicityfactors.Importantly,ourstudyrevealsapossibleprotectiveroleagainstdiarrheaforthe Prevotella genusand Lactobacillusruminis .Understandingsucheffectisimportant.Forexample,microbiological [39]ordietary[26]interventionsmaybepossibleinthe supportivetreatmentofdiarrheainchildrensimilartoapproachesusedinthemanagementofentericinfections inadults[39-41].Furthergenomicandepidemiologicalstudiesarenecessarytobettercharacterizethisgenusandtoassessthepotentialdevelopmentofdiet-or microbiological-basedtherapeutics.MaterialsandmethodsStudydesignandparticipantsStoolsampleswereselectedfromalargecase/control studyofmoderate-to-severediarrheainchildrenaged under5years[42].CaseswereenrolleduponpresentationtoahealthclinicreportingMSD.MSDeligibility criteriaincludedsunkeneyes,lossofnormalskinturgor, adecisiontoinitiateintravenoushydrationortohospitalizethechild,orthepresenceofbloodinthestool. Controlsweresoughtfollowingcaseenrollment,sampledfromademographicsurveillancedatabaseofthe area.Individualswereexcludediftheywereunableto produceasufficientamountofstoolvolumefortesting ortheywereunableorunwillingtoconsenttoinvolvementinthestudy.Everyparticipantwasconsentedprior tocollectionoftheirstoolandtheirdata.Consentwas givenbythecaregiver(usuallymother)becausethepatientsareallchildrenagedlessthan5years.Allsamples werecollectedbetweenMarchof2008andJuneof2009. Onesamplewascollectedforeachchildandnotimeseriesanalyseswereconducted.TheInstitutionalReview Boards(IRBs)atallcooperatinginstitutionshavereviewed andapprovedtheprotocol.TheIRBFederalWideAssurancenumbersforallthesitesareasfollows:Universityof MarylandBaltimoreFWA00007145,TheGambia,Medical ResearchCouncilLabsFWA00006873,KenyaMedical ResearchInstituteFWA00002066,UniversityofMaliFacultyofMedicinePharmacyandDentistryFWA00001769, andInternationalCentreforDiarrhoealDiseaseResearch, BangladeshFWA00001468.FurtherdetailsonstudydesignaredescribedbyKotloff etal. [42].MicrobiologymethodsStoolspecimenswerecollectedinsterilecontainersand examinedwithin24h.Stoolswerestoredat2to8C whileintransittothelaboratory.Eachfreshstoolspecimenwasaliquotedintomultipletubes.Allsampleswere analyzedbytraditionalmicrobiologicaltestsforknown bacterial,viral,andeukaryoticpathogens.Detailsof thesemethodscanbefoundinPanchalingam etal .[43] DNAwasisolatedusingabeadbeaterwith3mmdiametersolidglassbeads(sigmaLifeScience),andsubsequently0.1mmzirconiumbeads(BIO-SPECInc.)to disruptcells.Thecellslurrywasthencentrifugedat 16,000 g for1min,thesupernatantremovedandprocessedusingtheQiagenQIAampDNAstoolextraction kit.ExtractedDNAwasprecipitatedwith3Msodium acetateandethanolandtheDNAshippedtotheUSA.AmplificationandsequencingDNAwasamplifiedusing ‘ universal ’ primerstargeting theV1-V2regionofthe16SrRNAgene(smallsubunit oftheribosome)inbacteria(338R(5 ’ -CATGCTGCC TCCCGTAGGAGT-3 ’ and27F(5 ’ -AGAGTTTGATC CTGGCTCAG-3 ’ ).Bothforwardandreverseprimers hada5 ’ portionspecificforusewith454FLXsequencingtechnologyandtheforwardprimerscontaineda barcodebetweentheFLXandgenespecificregion,so thatsamplescouldbepooledtoamultiplexlevelof96 samplesperinstrumentrun(seeAdditionalfile6:TableS5 forbarcodeinformation).DataavailabilitySequencingdataandsamplemetadataareavailableat theNCBIarchiveunderprojectPRJNA234437. SourcecodeanddocumentationfortheanalysispipelineareavailableatGitHub:[44]. Abundancetableandmetadataareavailable,inBIOM [45]format,at[46]. Additionalinformationonthestudyaswellaslinksto allresourcesoutlinedabovearemadeavailableat[47].AnalysispipelineTheindividualreadswerefilteredforqualityusingcustomin-housescriptsthatperformthefollowingchecks suggestedinHuse etal. [48]:(1)sequencescontaining anyambiguitycodes(N)areremoved;(2)sequencesthat wereshorterthan75cyclesofthe454instrumentwere removed(eachcycleyieldsanaverageof2.5bpdependingonthesequencecomposition);(3)sequencesfor whichabarcodecouldnotbeidentifiedwereremoved. ThesechecksaresimilartothosethatcanbeperformedPop etal.GenomeBiology 2014, 15 :R76 Page8of12 http://genomebiology.com/2014/15/6/R76

PAGE 9

byMothur[49].Thehighqualitysequenceswereseparatedinto992sample-specificsetsaccordingtothemultiplexingbarcodes.ConservativeOTUswereclustered usingDNAclust[6]withparameters(-r1)(99%identity radius)thusensuringthatthedefinitionofanOTUis consistentacrossallsamples.Toobtaintaxonomicidentification,arepresentativesequencefromeachOTUwas alignedtoRibosomalDatabase(RDP)[50](rdp.cme. msu.edu,release10.4)usingblastnwithlongword length(-W100)inordertoonlydetectnearlyidentical sequences.Sequenceswithoutanearlyidenticalmatch toRDP(>100bpperfectmatchand>97%identity,as definedbyBLAST)weremarkedasbeing ‘ unassigned ’ andassignedanOTUidentifier.Theresultingdatawere organizedintoacollectionoftablesatseveraltaxonomiclevelscontainingeachtaxonomicgroupasarow andeachsampleasacolumn. Wenotethattheclusteringcriteriaweuse(<2%divergence,includinginsertionsanddeletions)aremoreconservativethancommonlyuseddefinitionsof ‘ specieslevel ’ OTUs(<2%divergenceexcludingindels).Weused conservativeclusteringbecausenouniversalcutoffappliestoallorganisms[51]andinordertoavoidmerging togetherorganismswithpotentiallydifferentphenotypes (forexample,closely-relatedstrains,seeAdditionalfile3: FigureS4foranexampleinclosely-related Escherichia/ Shigella OTUs).Similarconsiderationshaveledtothe developmentofspecializedsoftwarefortheanalysisofvaginal16SrRNAgenesurveydata[52].Ourapproachprovidesagoodtradeoffbetweenmitigatingtheeffectof errorsandallowinganunbiasedanalysisofthedata.Furthermore,anexplorationofincreasinglypermissiveclusteringthresholdsrevealsthatourconservativeclustering strategydoesnotlosestatisticalpower(seeAdditionalfile3: FiguresS5,S6). ChimeracheckingwasperformedwithUchime4.2.40 [53].PICRUSTanalysisThemostabundant6879OTUswerereprocessedusing QIIME[7]version1.8.0-devasrecommendedonthe PICRUSTwebsite(specificallyOTUswereconstructed withthepick_closed_reference_otus.pyscriptagainstthe latestversion(version13.5)oftheGreengenes[54]database)andtheresultinginformationwasprocessedwith PICRUST[16]version1.0.0-devusingtheKEGGanalysismoduleandaggregatingtheresultstolevel3.The resultswerefurtherexploredwithSTAMP[17]version 2.0.2,usingthetwo-groupanalysismodule,focusingon knownaerobicandanaerobicpathways.DatanormalizationInordertoavoidthebiasthatmaybeintroducedbypreferentialamplificationorsequencingofspecificsequences, wescaledthecountsbythe56thpercentileofthenumber ofOTUsineachsample.The56thpercentilewasempiricallydeterminedfromthedistributionofnon-zerocounts requiredtobehaveconsistentlyacrossoursamples.We normalizedwithaCumulativeSumScalingapproach, whichscalescountsbydividingthesumofeachsample ’ s countsuptoandincludingthe p thquantile(thatis,forall samples j Sp= i( cij| cij) qpj,where qpjisthe pthquantile ofsample j ).Normalizedcountsarethengivenby cijSpj1000. Thismethodconstrainscommunitieswithrespecttoa totalsize,butdoesnotplaceundueinfluenceonfeatures (OTUs)thatarepreferentiallysampled.Afulldescription ofthemethodologyisprovidedinPaulsonetal.[18].StatisticalapproachesTotestforpresenceandabsenceofanorganismweperformedFisher ’ steststratifyingbypositiveandnegative samples.Sampleswerestratifiedaspositiveforanorganismifthesamplehadoneormoresequencesoftheorganismwithasamplebeingnegativeiftherewasabsence ofsequences.Thetotalswerecalculatedforeachtaxa,a minimumof20positivesampleswasrequiredforastatisticaltesttobeattempted.Tocorrectformultiplecomparisonsweminimizedtheexpectedproportionoffalse positivesfollowingBenjaminiandHochberg[55]. Differentialabundancewasassessedwiththepackage metagenomeSeq[18]-astatisticalapproachthatmodels confoundingsuchasageandcountry,andalsotheeffect ofundersamplingontheobservedcounts.Significant findingswerereportedforOTUsthatsatisfiedthefollowingcriteria:(1)OTUwasabundant( 12normalized countspersample)incasesorcontrols;(2)OTUwas prevalent(presentin 10casesandcontrols);(3)OTU hadfoldchangeoroddsratioexceeding2ineithercases orcontrols;and(4)statisticalassociationwassignificant ( P <0.05)afterBenjamini-Hochbergcorrectionformultipletesting. AnalyseswereperformedusingtheRsoftwarepackage 3.0.1andpackages,Vegan2.0-7andmetagenomeSeq 1.2.21.CorrelationnetworkconstructionCorrelationnetworkswerec onstructedseparatelyon casesandcontrolstocharacterizethedependenciesbetween268differentiallyabundantOTUs(Additionalfile4: TableS3). EachnetworkwasbuiltusingSparCC[56],atoolspecificallydevelopedforassessingthecorrelationstructure withinmicrobialcommunities.Thestatisticalsignificance foreachOTU-OTU-interactionwasobtainedwithanempiricalnulldistributionusing1,000bootstrapiterations. The P valueswerefurtheradjustedformultiplecomparisonsusingtheBenjaminiandHochberg[55]correction.Pop etal.GenomeBiology 2014, 15 :R76 Page9of12 http://genomebiology.com/2014/15/6/R76

PAGE 10

AllOTU-OTU-interactionswithFDR<=0.05,wereconsideredsignificantandwererepresentedasedgesinthe network. Forsimplicityofvisualrepresentation,OTUswereaggregatedatgenusorlowertaxonomiclevelsusingthe mediannormalizedabundanceoftheaggregatedOTUs astheabundanceofthecorrespondingtaxonomicgroup. Weomittedalltaxonomicgroupswithmedianabundance lowerthan500normalizedcounts,aswellasalledges withSparCCcorrelationlowerthan0.09.Theplotswere drawninCytoscape3.0.1[57].AdditionalfilesAdditionalfile1:TableS1. Proportionalabundanceofdominant bacterialgeneraincasesandcontrols,bothoverallandstratifiedbyage stratumandcountry. Additionalfile2:TableS2. ComparisonofShannondiversityacross agesandcountries.P-valuescomputedwithTukey ’ shonestlysignificant differencetesttoaccountformultiplecomparisons. Additionalfile3: Additionalfigures(S1-S8)andfigurecaptions. Additionalfile4:TableS3. OTUsfoundtobesignificantlyassociated withdiarrheaorwithdiarrheafreecontrols. Additionalfile5:TableS4. Mappingoftaxonomicnamesusedin ourpaperandnearesthitstothecorresponding16SrRNAsequence. Duetothepoorresolutionofthe16SrRNAregionusedinourstudywe manuallyassignedeachOTUtothemostprecisetaxonomiclevel possible.Insomecasesasameorganismappearsinmultiplegroups, reflectingerrorsintheunderlyingdatabaseused(RDPversion10.4).For brevity,onlyambiguoustaxonomicgroupsarelisted. Additionalfile6:TableS5. Mappingofbarcodeinformationto sampleIDs. Competinginterests Theauthorsdeclarethattheyhavenocompetinginterests. Authors ’ contributions Samplecollectionanddatamanagement:MP,BRL,MA,MAH,JO,BT, MML,SP,KK,UNI,CE,MA,DA,FA,MTA,RA,SS,JBO,EO,JJ,EU,RO.16S rRNAgenedatacollectionandanalysis:MP,AWW,JNP,BRL,VM,IA, HCB,RR,MDS,VM,JP,JPN,OCS.Studydesign:MP,AWW,MA,MAH,JO, VM,MML,RFB,JGM,DS,JP,OCS,JPN.Statisticalanalysis:MP,AWW,JNP, BL,IA,HCB,OCS.Writing:MP,AWW,JNP,BL,JPN,OCS.Allauthorsread andapprovedthefinalmanuscript. Acknowledgments ThisworkwasfundedinpartbytheWilliamandMelindaGatesFoundation, award42917toJPNandOCS;USNationalInstitutesofHealthgrants 5R01HG005220toHCB,5R01HG004885toMP;USNationalScience FoundationGraduateResearchFellowshipawardDGE0750616toJNP;AWW andJParefundedbyTheWellcomeTrust(GrantNo.WT098051). Authordetails1UniversityofMaryland,CollegePark,MD,USA.2WellcomeTrustSanger Institute,Hinxton,Cambridgeshire,UK.3UniversityofMaryland,Schoolof Medicine,Baltimore,MD,USA.4MedicalResearchCouncilUnit,Serrekunda, Gambia.5InternationalCentreforDiarrhoealDiseaseResearch,Bangladesh, Dhaka,Bangladesh.6KenyaMedicalResearchInstitute(KEMRI)-USCentersfor DiseaseControlandPreventionResearchCollaboration,Kisumu,Kenya.7CenterforVaccineDevelopment,Bamako,Mali.8UniversityofFlorida, Gainesville,FL,USA.9EmoryUniversity,Atlanta,Georgia,USA.10Universityof Virginia,Charlottesville,VA,USA. Received:14January2014Accepted:27June2014 Published:27June2014 References1.LozanoR,NaghaviM,ForemanK,LimS,ShibuyaK,AboyansV,AbrahamJ, AdairT,AggarwalR,AhnSY,AlvaradoM,AndersonHR,AndersonLM, AndrewsKG,AtkinsonC,BaddourLM,Barker-ColloS,BartelsDH,BellML, BenjaminEJ,BennettD,BhallaK,BikbovB,BinAbdulhakA,BirbeckG, BlythF,BolligerI,BoufousS,BucelloC,BurchM, etal : Globalandregional mortalityfrom235causesofdeathfor20agegroupsin1990and2010: asystematicanalysisfortheGlobalBurdenofDiseaseStudy2010. Lancet 2012, 380: 2095 – 2128. 2.BlackRE,CousensS,JohnsonHL,LawnJE,RudanI,BassaniDG,JhaP, CampbellH,WalkerCF,CibulskisR,EiseleT,LiuL,MathersC,ChildHealth EpidemiologyReferenceGroupofWHOandUNICEF: Global,regional,and nationalcausesofchildmortalityin2008:asystematicanalysis. Lancet 2010, 375: 1969 – 1987. 3.WalkerCL,AryeeMJ,Boschi-PintoC,BlackRE: Estimatingdiarrheamortality amongyoungchildreninlowandmiddleincomecountries. PLoSOne 2012, 7: e29151. 4.LevineMM,KotloffKL,NataroJP,MuhsenK: TheGlobalEntericMulticenter Study(GEMS):impetus,rationale,andgenesis. ClinInfectDis 2012, 55: S215 – S224. 5.KotloffKL,NataroJP,BlackwelderWC,NasrinD,FaragTH,PanchalingamS, WuY,SowSO,SurD,BreimanRF,FaruqueAS,ZaidiAK,SahaD,AlonsoPL, TambouraB,SanogoD,OnwuchekwaU,MannaB,RamamurthyT,Kanungo S,OchiengJB,OmoreR,OundoJO,HossainA,DasSK,AhmedS,QureshiS, QuadriF,AdegbolaRA,AntonioM, etal : Burdenandaetiologyof diarrhoealdiseaseininfantsandyoungchildrenindevelopingcountries (theGlobalEntericMulticenterStudy,GEMS):aprospective,case – control study. Lancet 2013, 382: 209 – 222. 6.GhodsiM,LiuB,PopM: DNACLUST:accurateandefficientclusteringof phylogeneticmarkergenes. BMCBioinformatics 2011, 12: 271. 7.CaporasoJG,KuczynskiJ,StombaughJ,BittingerK,BushmanFD,Costello EK,FiererN,PenaAG,GoodrichJK,GordonJI,HuttleyGA,KelleyST,Knights D,KoenigJE,LeyRE,LozuponeCA,McDonaldD,MueggeBD,PirrungM, ReederJ,SevinskyJR,TurnbaughPJ,WaltersWA,WidmannJ,YatsunenkoT, ZaneveldJ,KnightR: QIIMEallowsanalysisofhigh-throughput communitysequencingdata. NatMethods 2010, 7: 335 – 336. 8.PalmerC,BikEM,DiGiulioDB,RelmanDA,BrownPO: Developmentofthe humaninfantintestinalmicrobiota. PLoSBiol 2007, 5: e177. 9.KoenigJE,SporA,ScalfoneN,FrickerAD,StombaughJ,KnightR,Angenent LT,LeyRE: Successionofmicrobialconsortiainthedevelopinginfantgut microbiome. ProcNatlAcadSciUSA 2011, 108: 4578 – 4585. 10.MackieRI,SghirA,GaskinsHR: Developmentalmicrobialecologyofthe neonatalgastrointestinaltract. AmJClinNutr 1999, 69: 1035s – 1045s.11.CebraJJ: Influencesofmicrobiotaonintestinalimmunesystem development. AmJClinNutr 1999, 69: 1046S – 1051S. 12.SjogrenYM,TomicicS,LundbergA,BottcherMF,BjorkstenB, Sverremark-EkstromE,JenmalmMC: Influenceofearlygutmicrobiotaon thematurationofchildhoodmucosalandsystemicimmuneresponses. ClinExpAllergy 2009, 39: 1842 – 1851. 13.Rajilic-StojanovicM,SmidtH,deVosWM: Diversityofthehuman gastrointestinaltractmicrobiotarevisited. EnvironMicrobiol 2007, 9: 2125 – 2136. 14.YatsunenkoT,ReyFE,ManaryMJ,TrehanI,Dominguez-BelloMG,Contreras M,MagrisM,HidalgoG,BaldassanoRN,AnokhinAP,HeathAC,WarnerB, ReederJ,KuczynskiJ,CaporasoJG,LozuponeCA,LauberC,ClementeJC, KnightsD,KnightR,GordonJI: Humangutmicrobiomeviewedacross ageandgeography. Nature 2012, 486: 222 – 227. 15.WalkerAW,LawleyTD: Therapeuticmodulationofintestinaldysbiosis. PharmacolRes 2013, 69: 75 – 86. 16.LangilleMGI,ZaneveldJ,CaporasoJG,McDonaldD,KnightsD,ReyesJA, ClementeJC,BurkepileDE,VegaThurberRL,KnightR,BeikoRG, HuttenhowerC: Predictivefunctionalprofilingofmicrobialcommunities using16SrRNAmarkergenesequences. NatBiotech 2013, 31: 814 – 821. 17.ParksDH,BeikoRG: Identifyingbiologicallyrelevantdifferencesbetween metagenomiccommunities. Bioinformatics 2010, 26: 715 – 721. 18.PaulsonJN,StineOC,BravoHC,PopM: Differentialabundanceanalysisfor microbialmarker-genesurveys. NatMethods 2013, 10: 1200 – 1202.Pop etal.GenomeBiology 2014, 15 :R76 Page10of12 http://genomebiology.com/2014/15/6/R76

PAGE 11

19.GirvanMS,CampbellCD,KillhamK,ProsserJI,GloverLA: Bacterialdiversity promotescommunitystabilityandfunctionalresilienceafter perturbation. EnvironMicrobiol 2005, 7: 301 – 313. 20.McCannKS: Thediversity-stabilitydebate. Nature 2000, 405: 228 – 233. 21.ShieldsTM,ChenKD,GouldJM: PediatricCaseReportofChronicColitis AssociatedWithanUnusualSerotypeofStreptococcuspneumoniae. InfectiousDiseasesinClinicalPractice 2012, 20: 357 – 358. 22.JinD,ChenC,LiL,LuS,LiZ,ZhouZ,JingH,XuY,DuP,WangH,XiongY, ZhengH,BaiX,SunH,WangL,YeC,GottschalkM,XuJ: Dynamicsoffecal microbialcommunitiesinchildrenwithdiarrheaofunknownetiology andgenomicanalysisofassociatedStreptococcuslutetiensis. BMC Microbiol 2013, 13: 141. 23.TaniuchiM,SobuzSU,BegumS,Platts-MillsJA,LiuJ,YangZ,WangXQ,Petri WAJr,HaqueR,HouptER: Etiologyofdiarrheainbangladeshiinfantsin thefirstyearoflifeanalyzedusingmolecularmethods. JInfectDis 2013, 208: 1794 – 1802. 24.ArumugamM,RaesJ,PelletierE,LePaslierD,YamadaT,MendeDR, FernandesGR,TapJ,BrulsT,BattoJM,BertalanM,BorruelN,CasellasF, FernandezL,GautierL,HansenT,HattoriM,HayashiT,KleerebezemM, KurokawaK,LeclercM,LevenezF,ManichanhC,NielsenHB,NielsenT, PonsN,PoultainJ,QinJ,Sicheritz-PontenT,TimsS, etal : Enterotypesof thehumangutmicrobiome. Nature 2011, 473: 174 – 180. 25.DeFilippoC,CavalieriD,DiPaolaM,RamazzottiM,PoulletJB,MassartS, ColliniS,PieracciniG,LionettiP: Impactofdietinshapinggutmicrobiota revealedbyacomparativestudyinchildrenfromEuropeandrural Africa. ProcNatlAcadSciUSA 2010, 107: 14691 – 14696. 26.WuGD,ChenJ,HoffmannC,BittingerK,ChenYY,KeilbaughSA, BewtraM,KnightsD,WaltersWA,KnightR,SinhaR,GilroyE,GuptaK, BaldassanoR,NesselL,LiH,BushmanFD,LewisJD: Linkinglong-term dietarypatternswithgutmicrobialenterotypes. Science 2011, 334: 105 – 108. 27.TaweechotipatrM,IyerC,SpinlerJK,VersalovicJ,TumwasornS: LactobacillussaerimneriandLactobacillusruminis:novelhuman-derived probioticstrainswithimmunomodulatoryactivities. FEMSMicrobiolLett 2009, 293: 65 – 72. 28.JangiS,LamontJT: AsymptomaticcolonizationbyClostridiumdifficilein infants:implicationsfordiseaseinlaterlife. JPediatrGastroenterolNutr 2010, 51: 2 – 7. 29.SearsCL,MyersLL,LazenbyA,VanTassellRL: EnterotoxigenicBacteroides fragilis. ClinInfectDis1995, 20: S142 – S148. 30.MazmanianSK,RoundJL,KasperDL: Amicrobialsymbiosisfactor preventsintestinalinflammatorydisease. Nature 2008, 453: 620 – 625. 31.LinA,BikEM,CostelloEK,DethlefsenL,HaqueR,RelmanDA,SinghU: Distinctdistalgutmicrobiomediversityandcompositioninhealthy childrenfromBangladeshandtheUnitedStates. PLoSOne 2013, 8: e53838. 32.QinJ,LiR,RaesJ,ArumugamM,BurgdorfKS,ManichanhC,NielsenT,Pons N,LevenezF,YamadaT,MendeDR,LiJ,XuJ,LiS,LiD,CaoJ,WangB, LiangH,ZhengH,XieY,TapJ,LepageP,BertalanM,BattoJM,HansenT, LePaslierD,LinnebergA,NielsenHB,PelletierE,RenaultP, etal : Ahuman gutmicrobialgenecatalogueestablishedbymetagenomicsequencing. Nature 2010, 464: 59 – 65. 33.HumanMicrobiomeProjectConsortium: Aframeworkforhuman microbiomeresearch. Nature 2012, 486: 215 – 221. 34.SimK,CoxMJ,WopereisH,MartinR,KnolJ,LiMS,CooksonWO,Moffatt MF,KrollJS: Improveddetectionofbifidobacteriawithoptimised16S rRNA-genebasedpyrosequencing. PLoSOne 2012, 7: e32543. 35.FrankJA,ReichCI,SharmaS,WeisbaumJS,WilsonBA,OlsenGJ: Critical evaluationoftwoprimerscommonlyusedforamplificationofbacterial 16SrRNAgenes. ApplEnvironMicrobiol 2008, 74: 2461 – 2470. 36.LindsayB,PopM,AntonioM,WalkerAW,MaiV,AhmedD,OundoJ, TambouraB,PanchalingamS,LevineMM,KotloffK,LiS,MagderLS,Paulson JN,LiuB,IkumapayiU,EbrukeC,DioneM,AdeyemiM,RanceR,StaresMD, UkhanovaM,BarnesB,LewisI,AhmedF,AlamMT,AminR,SiddiquiS, OchiengJB,OumaE, etal : Surveyofculture,goldengateassay,universal biosensorassay,and16SrRNAGenesequencingasalternativemethods ofbacterialpathogendetection. JClinMicrobiol 2013, 51: 3263 – 3269. 37.LayC,Rigottier-GoisL,HolmstromK,RajilicM,VaughanEE,deVosWM, CollinsMD,ThielR,NamsolleckP,BlautM,DoreJ: Colonicmicrobiota signaturesacrossfivenorthernEuropeancountries. ApplEnvironMicrobiol 2005, 71: 4153 – 4155. 38.LindsayB,OchiengJB,IkumapayiUN,ToureA,AhmedD,LiS, PanchalingamS,LevineMM,KotloffK,RaskoDA,MorrisCR,JumaJ,Fields BS,DioneM,MalleD,BeckerSM,HouptER,NataroJP,SommerfeltH,Pop M,OundoJ,AntonioM,HossainA,TambouraB,StineOC: QuantitativePCR fordetectionofshigellaimprovesascertainmentofshigellaburdenin childrenwithmoderate-to-severediarrheainlow-incomecountries. JClinMicrobiol 2013, 51: 1740 – 1746. 39.LawleyTD,ClareS,WalkerAW,StaresMD,ConnorTR,RaisenC, GouldingD,RadR,SchreiberF,BrandtC,DeakinLJ,PickardDJ,Duncan SH,FlintHJ,ClarkTG,ParkhillJ,DouganG: Targetedrestorationofthe intestinalmicrobiotawithasimple, definedbacteriotherapyresolvesrelapsingClostridiumdifficilediseaseinmice. PLoSPathog 2012, 8: e1002995. 40.UbedaC,BucciV,CaballeroS,DjukovicA,ToussaintNC,EquindaM,Lipuma L,LingL,GobourneA,NoD,TaurY,JengRR,vandenBrinkMR,XavierJB, PamerEG: IntestinalmicrobiotacontainingBarnesiellaspeciescures vancomycin-resistantEnterococcusfaeciumcolonization. InfectImmun 2013, 81: 965 – 973. 41.SeniorK: FaecaltransplantationforrecurrentCdifficilediarrhoea. Lancet InfectDis 2013, 13: 200 – 201. 42.KotloffKL,BlackwelderWC,NasrinD,NataroJP,FaragTH,vanEijkA, AdegbolaRA,AlonsoPL,BreimanRF,FaruqueAS,SahaD,SowSO,SurD, ZaidiAK,BiswasK,PanchalingamS,ClemensJD,CohenD,GlassRI,Mintz ED,SommerfeltH,LevineMM: TheGlobalEntericMulticenterStudy (GEMS)ofdiarrhealdiseaseininfantsandyoungchildrenindeveloping countries:epidemiologicandclinicalmethodsofthecase/controlstudy. ClinInfectDis 2012, 55: S232 – S245. 43.PanchalingamS,AntonioM,HossainA,MandomandoI,OchiengB,Oundo J,RamamurthyT,TambouraB,ZaidiAK,PetriW,HouptE,MurrayP,PradoV, VidalR,SteeleD,StrockbineN,SansonettiP,GlassRI,Robins-BrowneRM, TauschekM,SvennerholmAM,BerkeleyLY,KotloffK,LevineMM,NataroJP: DiagnosticmicrobiologicmethodsintheGEMS-1case/controlstudy. ClinInfectDis 2012, 55: S294 – S302. 44. AnalysispipelineforGEMSpathogendiscoveryproject. https://github. com/MihaiPop/GEMS-db. 45.McDonaldD,ClementeJC,KuczynskiJ,RideoutJR,StombaughJ,WendelD, WilkeA,HuseS,HufnagleJ,MeyerF,KnightR,CaporasoJG: TheBiological ObservationMatrix(BIOM)formator:howIlearnedtostopworrying andlovetheome-ome. Gigascience 2012, 1: 7. 46. DatageneratedinGEMSpathogendiscoveryproject. http://www.cbcb. umd.edu/datasets/gems-study-diarrheal-disease. 47. GEMSpathogendiscoveryproject:summarypage. http://www.cbcb.umd. edu/research/projects/GEMS-pathogen-discovery. 48.HuseSM,HuberJA,MorrisonHG,SoginML,WelchDM: Accuracyandquality ofmassivelyparallelDNApyrosequencing. GenomeBiol 2007, 8: R143. 49.SchlossPD,WestcottSL,RyabinT,HallJR,HartmannM,HollisterEB, LesniewskiRA,OakleyBB,ParksDH,RobinsonCJ,SahlJW,StresB, ThallingerGG,VanHornDJ,WeberCF: Introducingmothur:open-source, platform-independent,community-supportedsoftwarefordescribing andcomparingmicrobialcommunities. ApplEnvironMicrobiol 2009, 75: 7537 – 7541. 50.ColeJR,ChaiB,FarrisRJ,WangQ,KulamSA,McGarrellDM,GarrityGM, TiedjeJM: TheRibosomalDatabaseProject(RDP-II):sequencesandtools forhigh-throughputrRNAanalysis. NucleicAcidsRes 2005, 33: D294 – D296. 51.WhiteJR,NavlakhaS,NagarajanN,GhodsiMR,KingsfordC,PopM: Alignmentandclusteringofphylogeneticmarkers – implicationsfor microbialdiversitystudies. BMCBioinformatics2010, 11: 152. 52.RavelJ,GajerP,AbdoZ,SchneiderGM,KoenigSS,McCulleSL,KarlebachS, GorleR,RussellJ,TacketCO,BrotmanRM,DavisCC,AultK,PeraltaL,Forney LJ: Vaginalmicrobiomeofreproductive-agewomen. ProcNatlAcadSci USA 2011, 108: 4680 – 4687. 53.EdgarRC,HaasBJ,ClementeJC,QuinceC,KnightR: UCHIMEimproves sensitivityandspeedofchimeradetection. Bioinformatics 2011, 27: 2194 – 2200. 54.DeSantisTZ,HugenholtzP,LarsenN,RojasM,BrodieEL,KellerK,HuberT, DaleviD,HuP,AndersenGL: Greengenes,achimera-checked16SrRNA genedatabaseandworkbenchcompatiblewithARB. ApplEnviron Microbiol 2006, 72: 5069 – 5072. 55.BenjaminiY,HockbergY: Controllingthefalsediscoveryrate-apractical andpowerfulapproachtomultipletesting. JRoyStatSocBMet 1995, 57: 289 – 300.Pop etal.GenomeBiology 2014, 15 :R76 Page11of12 http://genomebiology.com/2014/15/6/R76

PAGE 12

56.FriedmanJ,AlmEJ: Inferringcorrelationnetworksfromgenomicsurvey data. PLoSComputBiol 2012, 8: e1002687. 57.ClineMS,SmootM,CeramiE,KuchinskyA,LandysN,WorkmanC, ChristmasR,Avila-CampiloI,Creec hM,GrossB,HanspersK,IsserlinR, KelleyR,KillcoyneS,LotiaS,MaereS,MorrisJ,OnoK,PavlovicV, PicoAR,VailayaA,WangPL,AdlerA,ConklinBR,HoodL,KuiperM, SanderC,SchmulevichI,SchwikowskiB,WarnerGJ, etal : Integration ofbiologicalnetworksandgeneexpressiondatausingCytoscape. NatProtoc 2007, 2: 2366 – 2382.doi:10.1186/gb-2014-15-6-r76 Citethisarticleas: Pop etal. : Diarrheainyoungchildrenfrom low-incomecountriesleadstolarge-scalealterationsinintestinal microbiotacomposition. GenomeBiology 2014 15 :R76. Submit your next manuscript to BioMed Central and take full advantage of: € Convenient online submission € Thorough peer review € No space constraints or color “gure charges € Immediate publication on acceptance € Inclusion in PubMed, CAS, Scopus and Google Scholar € Research which is freely available for redistribution Submit your manuscript at www.biomedcentral.com/submit Pop etal.GenomeBiology 2014, 15 :R76 Page12of12 http://genomebiology.com/2014/15/6/R76