Decontamination of 16S rRNA gene amplicon sequence ...

... and 0.5 μl each of 10 μM forward primer 341 F 5'-CCTACGGGNGGCWGCAG-3' and reverse primer 805R 5'-GACTACHVGGGTATCTAATCC-3' [14]. Skiptomaincontent Advertisement SearchallBMCarticles Search Decontaminationof16SrRNAgeneampliconsequencedatasetsbasedonbacterialloadassessmentbyqPCR DownloadPDF DownloadPDF Researcharticle OpenAccess Published:23April2016 Decontaminationof16SrRNAgeneampliconsequencedatasetsbasedonbacterialloadassessmentbyqPCR VladimirLazarevic1,NadiaGaïa1,MyriamGirard1&JacquesSchrenzel1  BMCMicrobiology volume 16,Article number: 73(2016) Citethisarticle 5026Accesses 41Citations 6Altmetric Metricsdetails AbstractBackgroundIdentificationofunexpectedtaxain16SrRNAsurveysoflow-densitymicrobiota,dilutedmockcommunitiesandculturesdemonstratedthatavariablefractionofsequencereadsoriginatedfromexogenousDNA.ThesourcesofthesecontaminantsarereagentsusedinDNAextraction,PCR,andnext-generationsequencinglibrarypreparation,andhuman(skin,oralandrespiratory)microbiotafromtheinvestigators.ResultsForinsilicoremovalofreagentcontaminants,apipelinewasusedwhichcombinestherelativeabundanceofoperationaltaxonomicunits(OTUs)inV3–416SrRNAgeneamplicondatasetswithbacterialDNAquantificationbasedonqPCRtargetingoftheV3segmentofthe16SrRNAgene.SeriallydilutedculturesofEscherichiacoliandStaphylococcusaureuswereusedfor16SrDNAprofiling,andDNAfromeachofthesespecieswasusedasaqPCRstandard.OTUsassignedtoEscherichiaorStaphylococcuswerevirtuallyunaffectedbythedecontaminationprocedure,whereasOTUsfromPseudomonas,whichisamajorreagentcontaminant,werecompletelyornearlycompletelyremoved.ThedecontaminationprocedurealsoattenuatedthetrendofincreaseinOTUrichnessinseriallydilutedcultures.ConclusionsRemovalofcontaminantsequencesderivedfromreagentsbasedonuseofqPCRdatamayimprovetaxonomicrepresentationinsampleswithlowDNAconcentration.Usingthedescribedpipeline,OTUsderivedfromcross-contaminationofnegativeextractioncontrolswerenotrecognizedascontaminantsandnotremovedfromthesampledataset. BackgroundThedevelopmentofPCRandnext-generationsequencingtechniqueshasfacilitatedstudyingmicrobialcommunitieswithoutitbeingnecessarytocultureindividualmembers.Becausegrowthrequirementsvarygreatlyamongdifferentspecies,andforsomespeciesgrowthconditionshavenotyetbeendetermined,molecularmethodsinmicrobiotainvestigationsareadvantageous.However,culture-freeapproachesmayintroducebiasesintheexperimentalpipeline,startingfromDNAextractionthroughthegenerationofsequencinglibrariestodataanalysis.Identificationofunexpectedtaxaindatasetsderivedfromlow-densitymicrobiota[1–3],dilutedmockcommunities[4],andcultures[5],demonstratedthatavariablefractionofsequencereadsoriginatedfromexogenousDNA.ThesourcesofthesecontaminantsarereagentsusedinDNAextraction,PCR,andnext-generationsequencinglibrarypreparation,andpossiblyhuman(skin,oral,andrespiratory)microbiotafromtheinvestigators[6].Sampledatasetscanbedecontaminatedbyremovingsequencereadsassignedtooperationaltaxonomicunits(OTUs)foundinnegativeextractioncontrols(NECs).BioinformaticspipelinesforperformingmicrobiomeanalysissuchasQiime[7]facilitateperformingthisstepinanautomatedmanner.SomeOTUsidentifiedascontaminantsacrossdifferentstudieswererepeatedlyassignedtothesamespeciesorgenera[5].However,anOTUthatcorrespondstothegenuinememberofthemicrobiotaofinterestmayalsobefoundinrelevantNECs.IthasbeensuggestednottoremoveOTUsidentifiedinNECsiftheyarebiologicallyexpectedinthegivensampletype[5].ThedistinctionbetweenexpectedandunexpectedOTUsinagivensampletypemaynotbealwaysstraightforward.Forexample,Propionibacteriumisaknownreagentcontaminantbutitisgenuinelypresentintheskinmicrobiotainproportionsthatvarybetweenindividuals[3].Similarly,Stenotrophomonas,anothercommonreagentcontaminant,emergedasanewairwaypathogen[4,8],thatmaycomplicatetheanalysisofrespiratorytractsamples.TheneedtorecognizeasmanycontaminantsaspossiblebasedondifferencesintherelativeabundanceofbacterialtaxabetweenNECs,low-densitysamples,andhigh-densitysampleshasbeenhighlighted[5,9].OthersproposedthatcontaminantOTUsexcludedshouldbethosewhoserelativeabundanceinNECsisaboveagiventhreshold[10].Inversecorrelationofataxonrelativeabundancewithbacterialloadasanindicatorofapossiblereagentcontaminantwasinitiallydescribedforbacterialgenerainamockcommunity[4]andsubsequentlyconfirmedattheOTUlevelinculturesand‘real’microbiotasamples[2,3,5,11].TheremovalofOTUswhosemeanrelativeabundanceinNECsishigherthanthatofmicrobiotasamplesofinteresthasbeenusedtodecontaminatedatasetsobtainedby16SrRNAgeneampliconsequencingofrelativelylow-densityskinandrespiratorytractbacterialcommunities[2,3].However,theabsoluteabundanceofcertainOTUsmaybesubstantiallyhigherinmicrobiotasamplesthaninrelevantNECs,eveniftheirrelativeabundanceshowstheoppositepattern.ItisadvisablenottoremovesuchOTUs,astheycorrespondtothemicrobiotaofinterest.Here,wefurtherdevelopthisapproachbycombiningrelativeabundanceofOTUswithbacterialloadinDNAextractsassessedbyquantitativereal-timePCR(qPCR).ResultsBacterialloaddeterminedbycultureandqPCR StaphylococcusaureusandEscherichiacoliovernightcultureswerewashedandconcentrated,andresultedin3.5x1010and5.3x109colony-formingunits(CFU)/ml,respectively.Serialdecimaldilutionsofthesemasterstockswerealiquotedintriplicateandfrozen.DNAwasextractedfromeachofthethreeidenticalseriesofaliquotsonseparateoccasions.Serialculturedilutionsdownto10−5correlatedwithdecreasingDNAyieldsinpurifiedextractsdeterminedbyqPCRtargetingtheV3segmentofthebacterial16SrRNAgenewithuniversalbacterialprimers(Fig. 1).Furtherdilutions(10−6–10−8)ofthemasterstockshadDNAquantityestimatessimilartothoseofNECsobtainedbysubstitutingcultureforwater(NEC_W)orlysisbuffer(NEC_B)inDNAextraction.ThelowestDNAconcentrationwasfoundforno-templatecontrols(NTC_W)inwhichwaterwasusedinsteadofDNAextract.BacterialloadsdeterminedbyqPCRbasedonS.aureus(Fig. 1a)orE.coli(Fig. 1b)referencecurvesshowedsimilarpatterns.Fig.1BacterialloadassessedbyqPCR.TheuniversalbacterialprimersusedinqPCRtargettheV3segmentofthe16SrRNAgene.BacterialloadsweredeterminedusingthestandardcurvesobtainedwithS.aureusMW2(a)orE.coliDH5α(b)genomicDNA.TheS.aureusandE.coligenomesweighapproximately2.9and4.8 fgandcontainsixandseven16SrRNAgenecopies,respectively.Eachsymbol(Exp1,Exp2andExp3)correspondstotheseriesofaliquotsprocessedatagivenpointandrepresentsthemeanofduplicatemeasurementswithrelativedeviationsfromthemean<2.5 %.BacterialloadisexpressedasthenumberofE.coliorS.aureusgenomeequivalentsin1 μlofDNAextract.Serialdecimaldilutionsofthemasterstockareindicatedfrom1E0(nodilution)to1E-8(10−8).SA,S.aureus;EC,E.coli.NEC_B,negativeextractioncontrolsobtainedbysubstitutingcultureforlysisbuffer;NEC_W,negativeextractioncontrolsobtainedbysubstitutingcultureforwater;NTC_W,notemplate(qPCR)controlreactionsperformedbysubstitutingDNAextractforwaterFullsizeimage TaxonomicanalysisofsamplesandnegativecontrolsThesequencedatasetgeneratedbyIlluminasequencingofV3–416SrRNAgeneampliconswasrepresentedby9042–176,345readspersampleafterqualityfilteringandOTUmapping.Theproportionofquality-filteredsequenceswithnohits(with≥97 %identity)intheGreengenesreferencedatabase[12]was0.63 ± 0.04 %(mean ± SD)formasterculturestocksand4.2 ± 1.9 %fornegativecontrols(NEC_W,NEC_B,andNTC_W).TheRDPclassifier[13](with≥80 %confidence)assignedthesesequencestoPseudomonadales(24.1 %),Parcubacteria(11.7 %),Actinomycetales(8.4 %),unclassifiedBacteria(14.6 %)andunclassifiedorganisms(7.8 %).Atotalof2673OTUswereidentifiedinthefinaldatasetofwhich1718werefoundonlyinsamples,276werespecifictonegativecontrols,and855werefoundinboth.OTUsrichnessinNEC_WandNEC_BweresimilartoeachotherbuthigherthanthatofNTC_W.Inthedatasetnormalizedtothesamenumberofsequencespersample(3500),OTUrichnessincreasedasbacterialcountsdecreased(Fig. 2).Fig.2OTUrichnessacrossthesamplesbeforeandafterinsilicodecontamination.R-OTU(ratiobetweenmean‘absolute’abundanceofOTUsinnegativeextractioncontrolsandculturesamples)cut-offsof1to0.001wereappliedfordecontamination.ThisratiowascalculatedfromtherelativeOTUabundanceandqPCRdataobtainedusingtheS.aureusstandardcurve.Dilutionsofthemasterstockareindicatedfrom1E0(nodilution)to1E-8(10−8).EC,E.coli;SA,S.aureus.NEC_W,negativeextractioncontrolsobtainedbysubstitutingcultureforwater;NEC_B,negativeextractioncontrolsobtainedbysubstitutingcultureforlysisbuffer;NTC_W,no-templatePCRcontrols;ND,nodecontaminationwasperformedFullsizeimage SerialculturedilutionswereassociatedwithadecreaseintheproportionofsequencereadsassignedtoEscherichiaorStaphylococcus,andanincreaseintherelativeabundanceofreadsthatderivedformcontaminants.Pseudomonaswasmajorcontaminant,andwasmostabundantinnegativecontrolsandhighlydilutedsamples(Fig. 3).Interestingly,OTUsassignedtoPseudomonashaddifferentprofilesinNECs(NEC_WandNEC_B)andNTC_W.Forexample,OTU4028110andOTU1566691,whichdominatedNECsandNTC_W,respectively,differedin11residuesinthesequencedV3–4region.TheseresultsshowspecificcontaminationofbothDNAextractionandPCRreagents.ThemostabundantOTUsinNTC_W,whichcorrespondedtothecontaminantsofthePCRreagents,wereidentifiedinmostNECs,wheretheywereoutnumberedbyOTUsfromcontaminantsfromDNAextraction.However,theOTUshighlyabundantinNECsweremostlyabsentfromNTC_W.ThesimilarityofNEC_WtoNEC_BanditsdifferencefromNTC_WOTUprofileindicatethatultrapurewaterwasnotthemajorsourceofDNAcontamination.Fig.3RelativeabundanceofpredominantOTUs.OTUswithameanrelativeabundance>1 %ineithersamples,negativeextractioncontrolsorNTC_Warepresented.Theproportionisindicatedbythescaleatthebottomoftheplot.Dilutionsofthemasterstockareindicatedfrom1E0(nodilution)to1E-8(10−8).ForEC,SA,NEC_BandNEC_W,thedataobtainedfromDNAextractionsperformedonthreeoccasions(Exp1–Exp3)arepresentedfromlefttoright.NTC_Wwereperformedinduplicateforeachofthethreeseries.EC,E.coli;SA,S.aureus.NEC_W,negativeextractioncontrolsobtainedsubstitutingcultureforwater;NEC_B,negativeextractioncontrolsobtainedbysubstitutingcultureforlysisbuffer;NTC_W,no-templatePCRcontrolFullsizeimage InthesequencedatafromhighlydilutedE.colicultures,weidentifiedasubstantialproportionofOTUsassignedtoStaphylococcusand,alsofoundEscherichiaOTUsinthesequencedatafromdilutedS.aureus.TheseresultsandthefactthatStaphylococcushasnotbeenpreviouslyrecognizedasreagentcontaminant[5]indicatecross-contaminationduringDNAextractionfromsampleswithhighbacterialload,notablybyE.coliinexperiment1andbyS.aureusinexperiment3(Fig. 3).However,theproportionofcross-contaminantswaslowerthanthatofreagentcontaminants. InsilicodecontaminationprocedureToobtainanapproximateestimationofthe‘absolute’abundanceofOTUs,expressedinarbitraryunits,wemultipliedtherelativeabundanceofeachOTUbythe16SrRNAgenecopynumberofagivensample(determinedbyqPCR).Wethencalculatedtheratio(designatedR-OTU)betweenmean‘absolute’abundanceofOTUsinNECsandculturesamples.ThedatasetwasdecontaminatedinsilicousingfourR-OTUcut-offvalues(1,0.1,0.01and0.001),byremovingOTUsforwhichthisratiowasexceeded.TheproportionofthebacterialgeneraEscherichia,StaphylococcusandPseudomonasbeforeandafterdecontaminationareshowninFig. 4.Thedecontaminationprocedureimprovedthetaxonomicprofileoflow-abundance(diluted)culturesamples.Forexample,ata10−5dilution,StaphylococcusandEscherichiacorrespondedto27.5and37.5 %ofreads,respectively,butafterdecontaminationusinganR-OTUcut-offof0.01,theyincreasedto>80 %ofreads.Fig.4Effectofinsilicodecontaminationontaxonomicprofilesofculturedilutionsandnegativeextractioncontrols.MeansforthreesamplesobtainedinseparateDNAextractionexperimentsaregiven.TheR-OTU(theratiobetweenmean‘absolute’abundanceofOTUsinnegativeextractioncontrolsandculturesamples)cut-offsof1to0.001wereappliedfordecontamination.ThisratiowascalculatedfromtherelativeOTUabundanceandqPCRdataobtainedusingtheS.aureusstandardcurve.Dilutionsofthemasterstockareindicatedfrom1E0(nodilution)to1E-8(10−8).EC,E.coli;SA,S.aureus.NEC_W,negativeextractioncontrolsobtainedbysubstitutingcultureforwater;NEC_B,negativeextractioncontrolsobtainedbysubstitutingcultureforlysisbuffer;ND,nodecontaminationwasperformedFullsizeimage Veryfewreads(0–0.28 %persample)assignedtoEscherichiaorStaphylococcuswereaffectedbythedecontaminationprocedureusingR-OTUof0.01.Pseudomonas,whichrepresented37–63 %ofreadsin10−5to10−8dilutions,wascompletelyornearlycompletelyremovedbydecontamination(exceptinoneE.coli10−8dilution,whereitwasfoundataproportionof1.4 %).DecontaminationofNECsresultedindatasetsthatcontained99.1–100 %readsassignedtoacombinationofStaphylococcusandEscherichia(Fig. 4andAdditionalfile1:FigureS1).TheprofilesofOTUabundance(Fig. 3)pointtocross-contaminationbybothE.coliandS.aureusduringsamplehandling.ThedecontaminationprocedurealsoattenuatedthetrendofincreaseinOTUrichnessinseriallydilutedcultures(Fig. 2).DiscussionOurpipelineforremovalofreagentcontaminantsinsequencedatasetscombinestherelativeabundanceofOTUsinV3–416SrRNAgeneamplicondatasetswithbacterialDNAquantificationbasedonqPCRtargetingoftheV3segmentofthe16SrRNAgene.BothPCRandqPCRmayintroducebiasesbypreferentialamplificationofcertaintargets.However,ourresultsshowedonlyalimitedimpactoforganismschosentogeneratethestandardcurveinqPCRexperiment.DecontaminationproceduresbasedonE.coliandS.aureusstandardcurvesresultedincomparableresultswithR-OTUthresholdsof0.1,0.01,and0.001(Fig. 4andAdditionalfile2:FigureS2).Thesetwobacteriaarephylogeneticallyrelativelydistant,astheybelongtotwodifferentphyla(E.coli,Proteobacteria;S.aureus,Firmicutes),havedifferentgenomesizes(E.coli,4.7Mbp;S.aureus,2.8Mbp)anddifferinthenumberof16SrRNAgenecopynumber(E.coli,seven;S.aureus,six).TheR-OTUcut-offsof0.1,0.01and0.001hadasimilareffectontheremovalofcontaminantsforthetwopurebacterialculturesinourstudy.However,theR-OTUcut-offthatbestseparatescontaminantsfromexpectedtaxamaybeinfluencedbythetaxonomiccompositionandbacterialloadofthesamplesanalysedinagivenstudy.Definingthethresholdthatremovesasmuchcontaminanttaxaaspossiblewhilenotaffectingthetaxaofinterestremainsanarbitrarychoiceinbothdecontaminationprocedureweusedandotherapproaches.Althoughdecontaminationproceduresbothinsilicoandinthewetlabarenotyetfullydeveloped,theyimprovethetaxonomicprofilesofmicrobiota,thusprovidingbenefitswhensamples(e.g.clinicalspecimens)areavailableinlimitedamountand/orhavelowbacterialload.Longer16SrRNAgenesequences,lowersequencingerrorrates,andOTUclusteringatidentitythresholds>97 %maycontributetobetterdistinctionbetweencontaminantsand‘real’OTUs.Inadditiontofurtherdevelopmentofbioinformaticsandstatisticalapproachesfordecontaminationaftersequencing,itmayalsobeadvantageoustoreduceDNAcontaminationoflaboratoryreagentsduringtheirmanufacturingandreducetheriskofsample-to-samplecontaminationduringtheexperiments.ConclusionsWeshowthatremovalofcontaminantOTUsderivedfromreagentsbasedonthecombinationofqPCRdataandrelativeabundanceofOTUsmayimprovetaxonomicrepresentationinsampleswithDNAconcentrationsclosetothoseofNECs.Usingthedescribedapproach,OTUsderivedfromcross-contamination,incontrasttothosederivedfromreagents,werenotrecognizedascontaminantsandnotremovedfromthedataset.TheapproachweusedinthisstudymayproveusefulinsituationswhereOTUsidentifiedinnegativecontrolshavehigherrelativeabundancebutlowerabsoluteabundancecomparedwithmicrobiotasamples.MethodsBacterialstrainsandcultureFreshcoloniesofE.coliDH5α(Invitrogen,Carlsbad,CA,USA)andS.aureusMW2(strainNRS123obtainedfromtheNetworkofAntibioticResistanceinS.aureus(NARSA))wereinoculatedin20 mlDifcoMueller-Hintonbrothmedium(BDDiagnostics,Sparks,MD,USA)andincubatedovernightwithshaking(180 rpm).Threeovernightculturesofthesamestrainwerepooled,centrifugedat1600 gfor10 minandwashedtwicewithNaCl0.9 %.Thecellsweresuspendedin6 mlddH2O(Sigma-Aldrich,Munich,Germany).Fromeachsuspension,fourserialdilutions10−1to10−8wereperformedbyadding100 μlinoculumto900 μlddH2O.OneserieswasusedimmediatelyforplatingontoMueller-HintonAgar(BDDiagnostics).CFUwerecountedafter24-hincubationat37 °C.Theotherthreeserieswereplacedat−20 °CandusedforDNAextractionwithinthefollowing6d.DNAextractionDNAwasextractedusingtheNucleoSpinSoilkit(Macherey-Nagel,Düren,Germany).Fivehundredmicrolitersofbacterialcellsuspensions,700 μloflysisbufferSL1,and100 μlofEnhancerSXwereshakeninaNucleoSpinBeadTubefor4 minatmaximumspeedonaVortex-Genie2withahorizontaltubeholder(ScientificIndustries,NewYork,USA).Thelysatewascentrifugedat11,000 gfor1 min.Then,wefollowedtheNucleoSpinSoilkitbookletprotocol(November2011/Rev.03).DNAwaselutedin50 μlofelutionbufferSE.PurifiedDNAwasstoredat−20 °C.OneE.coliandoneS.aureusdilutionseriesofsamples(100–10−8)wereprocessedinparallelonthreedifferentdays.Ineachbatch,anNECwasperformedusing500 μlSL1buffer(NEC_B)or500 μlddH2O(NEC_W)insteadofbacterialsuspensions.PCRandsequencingTheV3–4regionofthebacterial16SrRNAgenes(E.colipositions341–805)wasamplifiedusingtemplateDNAfromE.coliandS.aureuscultures,andfromNECs.PCRwasperformedina25 μlvolumethatcontained5 μlofDNAextract,12.5 μlKAPA2GRobustHotStartReadyMix(KapaBiosystems,Boston,MA,USA),6.5ddH2O,and0.5 μleachof10 μMforwardprimer341 F5’-CCTACGGGNGGCWGCAG-3’andreverseprimer805R5’-GACTACHVGGGTATCTAATCC-3’[14].TwoNTC_WwereperformedinparallelforeachseriesofbacterialsuspensionsandNECsusing5 μlddH2OinsteadofDNAextract.ThePCRconditionsincludedaninitialdenaturationat95 °Cfor3 min,followedby35 cyclesofdenaturationat95 °Cfor30 s,annealingat51 °Cfor30 s,andextensionat72 °Cfor30 s,withafinalextensionat72 °Cfor5 min.EachPCRwasperformedinduplicateandtheproductswerecombined.Thepooledsamplewasrunona2100Bioanalyzer(AgilentTechnologies,SantaClara,CA)forqualityanalysis.TheprimersfromthefirstroundofPCRwereremovedbydigesting5-μlsampleswith1unitExonucleaseI(NewEnglandBiolabs,Ipswich,MA,USA)inatotalvolumeof10 μlExonucleaseIReactionBuffer(NewEnglandBiolabs)at37 °Cfor30 min.Theenzymewasinactivatedat95 °Cfor15 min.Ampliconbarcodingwasperformedbyre-amplificationusing1 μlofExonucleaseI-treatedfirst-roundPCR,15pmoleachofforwardprimer5’-NNNNNNNNNNTCCTACGGGNGGCWGCAG-3’andreverseprimer5’-NNNNNNNNNNTGACTACHVGGGTATCTAAKCC-3’ina20-μLvolumeofMyTaqbufferthatcontained1.5unitsMyTaqDNApolymerase(Bioline,London,UK)and2 μlofBioStabPCRoptimizer(II)(Sigma-Aldrich,Munich,Germany).Foreachsample,theforwardandreverseprimershadthesame10-ntbarcodesequence.PCRswerecarriedoutusingthefollowingparameters:pre-denaturationfor2 minat96 °C,followedbyeightcyclesof96 °Cfor15 s,50 °Cfor30 s,and70 °Cfor90 s.DNAconcentrationofampliconsofinterestwasdeterminedbygelelectrophoresis.About20 ngampliconDNAofeachsamplewerepooledandpurifiedwithonevolumeAgencourtAMPureXPbeads(BeckmanCoulter,Nyon,Switzerland)toremoveprimerdimersandothersmallmisprimingproducts,followedbyanadditionalpurificationusingaMinElutePCRPurificationKit(Qiagen,Venlo,theNetherlands).About100 ngofthepooledampliconDNAwasusedtoconstructasequencinglibraryusingtheOvationRapidDRMultiplexSystem1–96(NuGEN,SanCarlos,CA,USA).Thelibrarywassize-selectedbygelelectrophoresisandsequencedfrombothendsfor300 cyclesontheIlluminaMiSequsingMiSeqv3ReagentKit(Illumina,SanDiego,CA,USA)atLGCGenomics(Berlin,Germany).DemultiplexedFASTQfilesweregeneratedfrombase-callsusingIllumina’sbcl2fastqv1.8.4software.Readswithincorrectbarcodes,missingbarcodes,orconflictingbarcodepairswerediscarded.Amaximumofthreemismatchesperprimerwereallowed.AfterremovalofprimersequencesusingproprietaryLGCGenomicssoftware,forwardandreverse-complementedreversereadsweremergedusingBBMergeformtheBBMap_34.48package(http://sourceforge.net/projects/bbmap/)withminimumoverlapof12basesandamaximumof3mismatches.SequenceanalysisSequencefilteringwasperformedusingthecommandtrim.seqinMOTHURv1.35[15].Thereadsthatcontainedambiguousbasesorhomopolymerrunslongerthan12baseswereremoved.Then,sequencesweretruncatedatthebeginningofa20-basewindowwithanaveragePhredquality<30.Sequencesthat,aftertrimming,hadalength<300baseswerediscarded.Denoisingandclusteringof16SrDNAsequencesweremadebyOTUmappingwiththeGreengenesreferencedatabase[12]pre-clusteredat97 %identity(Greengenesfile97_otus.fastaasof17May2013)usingUSEARCH(−usearch_global–wordlength30–id0.97–query_cov0.9–top_hits_only)[16].Sequenceswithnohitswerediscarded.Forsequenceswithmultiplebesthits,thehitthatcorrespondedtotheGreengenesreferencesequencemostfrequentlyassignedintheentiredatasetwasretained.ThereadswereclassifiedusingnaïveBayesianmethodandtheRDPreferencedatabase[13]viaMOTHUR(commandclassify.seqswithoptions–method = wangand–cutoff = 80)andMOTHURfilestrainset10_082014.rdp.fastaandtrainset10_082014.rdp.tax.TheconsensustaxonomyofanOTUwasdefinedasthetaxonomythatrepresentedmostofthereadswithinthisOTU.qPCRqPCRassaywasperformedonanMx3005PqPCRsystem(AgilentTechnologies,SantaClara,CA,USA).Reactionmixturescontained12.5 μlof2×BrilliantIISYBRGreenQPCRMasterMix(AgilentTechnologies),0.75 μlof1/250dilutedreferencedye(AgilentTechnologies),0.3 μlofeach25 μMforward(5’-ACTCCTACGGGAGGCAGCAGT-3’)andreverse(5’-ATTACCGCGGCTGCTGGC-3’)primers[17],1 μlofDNAextract,and10.15 μlwater.TheprimersusedamplifytheV3regionofbacterial16SrRNAgenes(E.colipositions338–534).Thecyclingconditionsincludedinitialdenaturationof10 minat95 °Cfollowedby40 cyclesof95 °Cfor30 sand68°for1 min.No-templateqPCRcontrolswereperformedusing1 μlddH2OinsteadofDNAextract.ThereferencecurvesforDNAquantitationwereobtainedusingknownconcentrationsofgenomicDNAofE.colistrainDH5αandS.aureusstrainMW2.Allreactionswerecarriedoutinduplicate.EthicsapprovalandconsenttoparticipateNotapplicable.ConsentforpublicationNotapplicable.AvailabilityofdataandmaterialsThedatasetsupportingtheconclusionsofthisarticleisavailableintheMG-RASTrepository[18]undertheprojectID14505 (4647144.3–4647209.3). AbbreviationsNEC: negativeextractioncontrol NEC_B: negativeextractioncontrolobtainedbysubstitutingcultureforlysisbuffer NEC_W: negativeextractioncontrolobtainedbysubstitutingcultureforwater NTC_W: no-templatePCR(orqPCR)control OTU: operationaltaxonomicunit R-OTU: theratiobetweenmean‘absolute’abundanceofOTUsinNECandculturesamples ReferencesBiesbroekG,SandersEA,RoeselersG,WangX,CaspersMP,TrzcinskiK,etal.Deepsequencinganalysesoflowdensitymicrobialcommunities:workingattheboundaryofaccuratemicrobiotadetection.PLoSONE.2012;7:e32942.CAS  Article  PubMed  PubMedCentral  GoogleScholar  LazarevicV,GaiaN,EmonetS,GirardM,RenziG,DespresL,etal.Challengesintheculture-independentanalysisoforalandrespiratorysamplesfromintubatedpatients.FrontCellInfectMicrobiol.2014;4:65.Article  PubMed  PubMedCentral  GoogleScholar  TroccazM,GaiaN,BeccucciS,SchrenzelJ,CayeuxI,StarkenmannC,etal.Mappingaxillarymicrobiotaresponsibleforbodyodoursusingaculture-independentapproach.Microbiome.2015;3:3.Article  PubMed  PubMedCentral  GoogleScholar  WillnerD,DalyJ,WhileyD,GrimwoodK,WainwrightCE,HugenholtzP.ComparisonofDNAextractionmethodsformicrobialcommunityprofilingwithanapplicationtopediatricbronchoalveolarlavagesamples.PLoSONE.2012;7:e34605.CAS  Article  PubMed  PubMedCentral  GoogleScholar  SalterSJ,CoxMJ,TurekEM,CalusST,CooksonWO,MoffattMF,etal.Reagentandlaboratorycontaminationcancriticallyimpactsequence-basedmicrobiomeanalyses.BMCBiol.2014;12:87.Article  PubMed  PubMedCentral  GoogleScholar  KnightsD,KuczynskiJ,CharlsonES,ZaneveldJ,MozerMC,CollmanRG,etal.Bayesiancommunity-wideculture-independentmicrobialsourcetracking.NatMethods.2011;8:761–3.CAS  Article  PubMed  PubMedCentral  GoogleScholar  CaporasoJG,KuczynskiJ,StombaughJ,BittingerK,BushmanFD,CostelloEK,etal.QIIMEallowsanalysisofhigh-throughputcommunitysequencingdata.NatMethods.2010;7:335–6.CAS  Article  PubMed  PubMedCentral  GoogleScholar  SuretteMG.Thecysticfibrosislungmicrobiome.AnnAmThoracSoc.2014;11Suppl1:S61–5.Article  PubMed  GoogleScholar  WeissS,AmirA,HydeER,MetcalfJL,SongSJ,KnightR.Trackingdownthesourcesofexperimentalcontaminationinmicrobiomestudies.GenomeBiol.2014;15:564.Article  PubMed  PubMedCentral  GoogleScholar  FloresGE,HenleyJB,FiererN.AdirectPCRapproachtoaccelerateanalysesofhuman-associatedmicrobialcommunities.PLoSONE.2012;7:e44563.CAS  Article  PubMed  PubMedCentral  GoogleScholar  Jervis-BardyJ,LeongLE,MarriS,SmithRJ,ChooJM,Smith-VaughanHC,etal.Derivingaccuratemicrobiotaprofilesfromhumansampleswithlowbacterialcontentthroughpost-sequencingprocessingofIlluminaMiSeqdata.Microbiome.2015;3:19.Article  PubMed  PubMedCentral  GoogleScholar  McDonaldD,PriceMN,GoodrichJ,NawrockiEP,DeSantisTZ,ProbstA,etal.AnimprovedGreengenestaxonomywithexplicitranksforecologicalandevolutionaryanalysesofbacteriaandarchaea.ISMEJ.2012;6:610–8.CAS  Article  PubMed  PubMedCentral  GoogleScholar  ColeJR,WangQ,FishJA,ChaiB,McGarrellDM,SunY,etal.RibosomalDatabaseProject:dataandtoolsforhighthroughputrRNAanalysis.NucleicAcidsRes.2014;42:D633–42.CAS  Article  PubMed  PubMedCentral  GoogleScholar  HerlemannDP,LabrenzM,JurgensK,BertilssonS,WaniekJJ,AnderssonAF.Transitionsinbacterialcommunitiesalongthe2000 kmsalinitygradientoftheBalticSea.ISMEJ.2011;5:1571–9.CAS  Article  PubMed  PubMedCentral  GoogleScholar  SchlossPD,WestcottSL,RyabinT,HallJR,HartmannM,HollisterEB,etal.Introducingmothur:open-source,platform-independent,community-supportedsoftwarefordescribingandcomparingmicrobialcommunities.ApplEnvironMicrobiol.2009;75:7537–41.CAS  Article  PubMed  PubMedCentral  GoogleScholar  EdgarRC.SearchandclusteringordersofmagnitudefasterthanBLAST.Bioinformatics.2010;26:2460–1.CAS  Article  PubMed  GoogleScholar  HartmanAL,LoughDM,BarupalDK,FiehnO,FishbeinT,ZasloffM,etal.Humangutmicrobiomeadoptsanalternativestatefollowingsmallboweltransplantation.ProcNatlAcadSciUSA.2009;106:17187–92.CAS  Article  PubMed  PubMedCentral  GoogleScholar  MeyerF,PaarmannD,D’SouzaM,OlsonR,GlassEM,KubalM,etal.ThemetagenomicsRASTserver-apublicresourcefortheautomaticphylogeneticandfunctionalanalysisofmetagenomes.BMCBioinformatics.2008;9:386.CAS  Article  PubMed  PubMedCentral  GoogleScholar  DownloadreferencesAcknowledgementsWethankNARSAforprovidingthestrainMW2andStefanoLeoforassistancewithmanuscriptrevision. Funding Nofundingwasreceivedforthisstudy. AuthorinformationAuthorsandAffiliationsGenomicResearchLaboratory,DepartmentofGeneticsandLaboratoryMedicine,DepartmentofMedicalSpecialties,GenevaUniversityHospitals,RueGabrielle-Perret-Gentil4,CH-1211,Geneva,14,SwitzerlandVladimirLazarevic, NadiaGaïa, MyriamGirard & JacquesSchrenzelAuthorsVladimirLazarevicViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarNadiaGaïaViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarMyriamGirardViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarJacquesSchrenzelViewauthorpublicationsYoucanalsosearchforthisauthorin PubMed GoogleScholarCorrespondingauthorCorrespondenceto VladimirLazarevic.AdditionalinformationCompetinginterestsTheauthorsdeclarethattheyhavenocompetinginterests.Authors’contributionsVLandJSdesignedtheexperimentsanddraftedthemanuscript.MGperformedtheexperiments.NGandVLperformedthebioinformaticsandtaxonomicanalyses.Allauthorsapprovedthefinalmanuscript.Additionalfiles Additionalfile1:FigureS1.Relativeabundanceofbacterialgenerabeforeandafterthedecontaminationprocedure.Generawithmeanrelativeabundance>0.5 %innegativeextractioncontrolsarepresented.Theproportionisindicatedbythescaleatthebottomoftheplot.TheR-OTU(ratiobetweenmean‘absolute’abundanceofOTUsinnegativeextractioncontrolsandculturesamples)cut-offof0.01wasappliedfordecontamination.ThisratiowascalculatedfromtherelativeOTUabundanceandqPCRdataobtainedusingtheS.aureusstandardcurve.Foragivenculture/dilutionornegativeextractioncontrol,thedataobtainedfromDNAextractionsperformedatthreedifferenttimepoints(Exp1–Exp3)arepresentedfromlefttoright.Dilutionsofthemasterstockareindicatedfrom1E0(nodilution)to1E-8(10−8).NEC_W,negativeextractioncontrolsobtainedbysubstitutingcultureforwater;NEC_B,negativeextractioncontrolsobtainedbysubstitutingcultureforlysisbuffer.(PDF13kb)Additionalfile2:FigureS2.Effectofinsilicodecontaminationontaxonomicprofilesofculturedilutionsandnegativeextractioncontrols.MeansforthreesamplesobtainedinseparateDNAextractionexperimentsaregiven.TheR-OTU(ratiobetweenmean‘absolute’abundanceofOTUsinnegativeextractioncontrolsandculturesamples)cut-offsof1to0.001wereappliedfordecontamination.ThisratiowascalculatedfromtherelativeOTUabundanceandqPCRdataobtainedusingtheE.colistandardcurve.Dilutionsofthemasterstockareindicatedfrom1E0(nodilution)to1E-8(10−8).EC,E.coli;SA,S.aureus.NEC_W,negativeextractioncontrolsobtainedbysubstitutingcultureforwater;NEC_B,negativeextractioncontrolsobtainedbysubstitutingcultureforlysisbuffer;ND,nodecontaminationwasperformed.(PDF16kb)Rightsandpermissions OpenAccessThisarticleisdistributedunderthetermsoftheCreativeCommonsAttribution4.0InternationalLicense(http://creativecommons.org/licenses/by/4.0/),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedyougiveappropriatecredittotheoriginalauthor(s)andthesource,providealinktotheCreativeCommonslicense,andindicateifchangesweremade.TheCreativeCommonsPublicDomainDedicationwaiver(http://creativecommons.org/publicdomain/zero/1.0/)appliestothedatamadeavailableinthisarticle,unlessotherwisestated. ReprintsandPermissionsAboutthisarticleCitethisarticleLazarevic,V.,Gaïa,N.,Girard,M.etal.Decontaminationof16SrRNAgeneampliconsequencedatasetsbasedonbacterialloadassessmentbyqPCR. BMCMicrobiol16,73(2016).https://doi.org/10.1186/s12866-016-0689-4DownloadcitationReceived:24August2015Accepted:14April2016Published:23April2016DOI:https://doi.org/10.1186/s12866-016-0689-4SharethisarticleAnyoneyousharethefollowinglinkwithwillbeabletoreadthiscontent:GetshareablelinkSorry,ashareablelinkisnotcurrentlyavailableforthisarticle.Copytoclipboard ProvidedbytheSpringerNatureSharedItcontent-sharinginitiative KeywordsContaminantDNABacterialcommunities16SrRNAgenesequencingMicrobiome DownloadPDF Advertisement BMCMicrobiology ISSN:1471-2180 Contactus Submissionenquiries:[email protected] Generalenquiries:[email protected]