Wedemonstratelowcostsofsequencing,mate-pairedreads,highmultiplicities, - PDF document

Wedemonstratelowcostsofsequencing,mate-pairedreads,highmultiplicities,andhighconsensusaccuracies.TheseenableapplicationsincludingBAC(bacterialartificialchromosome)andbacterialgenomeresequencing,aswellasSAGE(serialanalysisofgeneexpression)tagandbarcodesequencing.Simulationssuggestthatthecurrentmate-pairedlibrariesarecom-patiblewithhumangenomeresequencing,pro-videdthatthereadlengthcanbeincreasedtocoverthefull17-to18-bptag(fig.S5).Whatarethelimitsofthisapproach?Asmanyas1billion1-mbeadscanpotentiallybefitintheareaofastandardmicroscopeslide(fig.S6).Weachieverawdataacquisitionrates400bp/s,morethananorderofmagnitudefasterthanconventionalsequencing.Fromanoth-erpointofview,wecollected786gigabitsofimagedatafromwhichwegleanedonlymegabitsofsequence.Thissparsity—oneusefulbitofinformationper10,000bitscollected—isaripeavenueforimprovement.Thenaturallimitofthisdirectionissingle-pixelsequencing,inwhichthecommonplaceanalogybetweenbytesandbaseswillbeatitsmostmanifest.ReferencesandNotes1.F.Sangeretal.,687(1977).2.F.S.Collins,M.Morgan,A.Patrinos,,2863.J.Shendureetal.Nat.Rev.Genet.,335(2004).4.I.Braslavsky,B.Hebert,E.Kartalov,S.R.Quake,Natl.Acad.Sci.U.S.A.,3960(2003).5.T.S.Seoetal.Proc.Natl.Acad.Sci.U.S.A.5926(2005).6.R.D.Mitra,J.Shendure,J.Olejnik,O.EdytaKrzymanska,G.M.Church,Anal.Biochem.320,55(2003).7.M.J.Leveneetal.,682(2003).8.M.Ronaghi,S.Karamohamed,B.Pettersson,M.Uhlen,P.Nyren,Anal.Biochem.242,84(1996).9.J.H.Leamonetal.,3769(2003).10.http://arep.med.harvard.edu/Polonator/Plone.htm11.R.D.Mitra,G.M.Church,NucleicAcidsRes.,e3412.P.M.Lizardietal.Nat.Genet.,225(1998).13.C.P.Adams,S.J.Kron,U.S.Patent5,641,658(1997).14.D.Dressman,H.Yan,G.Traverso,K.W.Kinzler,B.Proc.Natl.Acad.Sci.U.S.A.,881715.D.S.Tawfik,A.D.Griffiths,Nat.Biotechnol.,65216.F.J.Ghadessy,J.L.Ong,P.Holliger,Proc.Natl.Acad.Sci.U.S.A.,4552(2001).17.M.Nakanoetal.J.Biotechnol.,117(2003).18.A.M.Maxam,W.Gilbert,Proc.Natl.Acad.Sci.U.S.A.,560(1977).19.F.Barany,Proc.Natl.Acad.Sci.U.S.A.,189(1991).20.J.N.Housby,E.M.Southern,NucleicAcidsRes.4259(1998).21.S.C.Macevicz,U.S.Patent5,750,341(1998).22.S.Brenneretal.Nat.Biotechnol.,630(2000).23.N.B.Reppas,X.Lin,inpreparation.24.C.S.Barker,B.M.Pruss,P.Matsumura,J.Bacteriol.,7529(2004).25.R.H.Plasterk,P.vandePutte,EMBOJ.,237(1985).26.M.Mucke,S.Reich,E.Moncke-Buchner,M.Reuter,D.H.Kruger,J.Mol.Biol.,687(2001).27.D.Shore,R.L.Baldwin,J.Mol.Biol.,957(1983).28.Foradvice,encouragement,andtechnicalassistance,wearedeeplyindebtedtoJ.Zhu,S.Douglas,J.Chou,J.Aach,M.Nikku,A.Lee,N.Novikov,andM.Wright(ChurchLab);A.Blanchard,G.Costa,H.Ebling,J.Ichikawa,J.Malek,P.McEwan,K.McKernan,A.Sheridan,andD.Smith(Agencourt);S.Skiena(SUNY–StonyBrook)C.Felts(RPI);R.Fincher(Alcott);D.Focht(Bioptechs);andM.HotfelderandJ.Feng(Washing-tonUniversity).WethankB.Vogelstein,J.Edwards,andtheirgroupsforassistancewithemulsionPCR.ThisworkwassupportedbytheNationalHumanGenomeResearchInstitute–CentersofExcellenceinGenomicScienceandU.S.DepartmentofEnergy–GenomestoLifegrants.SupportingOnlineMaterialSOMTextFigs.S1toS814July2005;accepted27July2005Publishedonline4August2005;Includethisinformationwhencitingthispaper.PUMACouplestheNuclearandCytoplasmicProapoptoticFunctionofp53JerryE.Chipuk,LisaBouchier-Hayes,TomomiKuwana,DonaldD.Newmeyer,DouglasR.GreenTrp53tumorsuppressorgeneproduct(p53)functionsinthenucleustoregulateproapoptoticgenes,whereascytoplasmicp53directlyactivatespro-apoptoticBcl-2proteinstopermeabilizemitochondriaandinitiateapoptosis.Here,wedemonstratethatatripartitenexusbetweenBcl-xL,cytoplasmicp53,andPUMAcoordinatesthesedistinctp53functions.Aftergenotoxicstress,Bcl-xLsequesteredcytoplasmicp53.Nuclearp53causedexpressionofwhichthendisplacedp53fromBcl-xL,allowingp53toinducemitochondrialpermeabilization.MutantBcl-xLthatboundp53,butnotPUMA,renderedcellsresistanttop53-inducedapoptosisirrespectiveofPUMAexpression.Thus,PUMAcouplesthenuclearandcytoplasmicproapoptoticfunctionsofp53.Theantineoplasticfunctionofp53occurspri-marilythroughtheinductionofapoptosis(p53undergoesposttranslationalmodificationinresponsetooncogene-activatedsignalingpath-waysortogenotoxicstress;thisallowssta-bilizationofp53,whichaccumulatesinthenucleusandregulatestargetgeneexpression.Numerousgenesareregulatedbyp53,suchasthoseencodingdeathreceptorsforexample,FASandproapoptoticBcl-2proteins(forexample,,and).Inparallel,p53alsoaccumulatesinthecytoplasm,whereitdirectlyactivatesthepro-apoptoticproteinBAXtopromotemitochondri-alouter-membranepermeabilization(MOMP)8–10).OnceMOMPoccurs,proapoptogenicfactors(forexample,cytochromec)arereleasedfrommitochondria,caspasesareactivated,andapoptosisrapidlyensues().Thus,p53pos-sessesaproapoptoticfunctionthatisindepen-dentofitstranscriptionalactivity(Ifp53directlyengagesMOMPincoop-erationwithBAX,nofurtherrequirementforp53-dependenttranscriptionalregulationofadditionalproapoptoticBcl-2proteinswouldbeexpected.Nevertheless,PUMA(p53–up-regulatedmodifierofapoptosis),aproapoptoticBH3-onlyprotein,isadirecttranscriptionaltargetofp53.Furthermore,micedeficientinareresistanttop53-dependent,DNAdamage–inducedapoptosiseventhoughp53isstabilizedandaccumulatesinthecyto-plasm().Abetterunderstandingofthedistinctnuclearandcytoplasmicproapo-ptoticfunctionsofp53mayrevealstrategiesforthepreventionandtreatmentofcancer. DivisionofCellularImmunology,LaJollaInstituteforAllergyandImmunology,10355ScienceCenterDrive,SanDiego,CA92121,USA.UniversityofIowa,CarverCollegeofMedicine,DepartmentofPathology,IowaCity,IA52242,USA.*Presentaddress:DepartmentofImmunology,St.JudeChildren’sResearchHospital,332NorthLauderdaleStreet,Memphis,TN38105,USA.Towhomcorrespondenceshouldbeaddressed.E-mail:dgreen5240@aol.comFig.1.DNAdamage–inducedp53Bcl-xLandBcl-xLcomplexes.)Proteinsfromcy-tosolicextractspre-paredfromwild-typeorMEFstreatedwith5mJ/cmwereimmunoprecipi-tatedwithanagarose- conjugatedantibodytoBcl-xL,eluted,subjectedtoSDS-PAGE,andvisualizedbysilverstaining.Bandswereexcisedandsubjectedtotrypticdigestionandmassspectrometry.Theasterisk(*)indicatesafragmentofBcl-xLorp53.()Cytosolicextractsweretreatedasin(A),butproteincomplexeswereanalyzedbyWesternblot.mIgG(mouseimmunoglobulinG)isacontrolantibody.EPORTS9SEPTEMBER2005VOL309SCIENCEwww.sciencemag.org on August 30, 2012www.sciencemag.orgDownloaded from referencesequence;threewereonlypresentintheevolvedvariant(Table2).Ofthe100mockSNCs,53wereatpositionscalledwithhighconfidence.Allofthesewerecorrectlycalledassubstitutionsoftheexpectednucleo-tide(59of59onasecondsetofmockSNCs).Theabsenceofsubstitutionerrorsin3.3Mbofreferencesequencepositionscalledwithhighconfidencesuggeststhatweareachievingconsensusaccuraciessufficientforingapplications.PercentageofthegenomecoveredandmockSNCdiscoveryatvariouslevelsofcoverageareshowninTable1.Despite10coverageintermsofrawbasepairs,only91.4%ofthegenomehadatleastcoverage(fig.S4).Substantialfluctuationsincoveragewereobservedowingtothesto-chasticityoftheRCAstepoflibraryconstruc-tion.Wearecurrentlygeneratinglibrariesthataremorecomplexandmoreevenlydistributed.AGaussoiddistributionofdistancesbetweenmate-pairedtagswasobserved,consistentwiththesizeselectionduringlibraryconstruction(Fig.3,AandB).Notably,thehelicalpitchofDNA(10.6bpperturn)isevidentinthelocalstatisticsof1millioncircularizationevents(Fig.3B).Asafunctionofthenumberofbasessequenced,wegeneratedoveranorderofmagnitudemoremate-pairingdatapointsthananequivalentamountofconventionalsequenc-ing.Todetectgenomicrearrangements,weminedtheunplacedmate-pairsforconsistentlinksbetweengenomicregionsthatdidnotfallwithintheexpecteddistanceconstraints.InadditiontodetectingtheexpectedreplacementsoftheoperonswithandlambdaRedprophage(Fig.3D),wedetectedandcon-firmedtheabsenceofa776-bpIS1transposon(Fig.3C),apreviouslydescribedheterogeneityinMG1655strains().Wealsodetectedandconfirmeda1.8-kbregionthatwasheteroge-neouslyinvertedinthegenomicDNAusedtoconstructthelibrary(Fig.3E),owingtoactivityontheinvertiblePregion(Weobserveerrorratesof0.001forthebetterhalfofourrawbasecalls(Fig.2D).Al-thoughhighconsensusaccuraciesarestillachievedwithrelativelylowcoverage,ourbestrawaccuraciesarenotablyonetotwoordersofmagnitudelessaccuratethanmostrawbasesinaconventionalSangersequencingtrace.ThePCRamplificationsbeforesequencingarepotentiallyintroducingerrorsataratethatimposesaceilingontheaccuraciesachievablebythesequencingmethoditself.Onepotentialsolutionistocreatealibrarydirectlyfromthegenomicmaterialtobesequenced,suchthatthelibrarymoleculesarelinearRCAamplicons.Suchconcatemers,whereeachcopyisindependentlyderivedfromtheoriginaltemplate,wouldtheoreticallyprovideaformoferrorcorrectionduringePCR.Ouralgorithmswerefocusedondetectionofpointsubstitutionsandrearrangements.Increas-ingreadlengths,currentlytotalingonly26bpperamplicon,willbecriticaltodetectingawiderspectrumofmutation.Ahigherfidel-ityligase()orsequentialnonamerligations)mayenablecompletionofeach17-to18-bptag.EcoP15I,whichgenerates27-bptags,wouldallowevenlongerreadlengthswhileretainingthesamemate-pairingscheme(Weestimateacostof$0.11perrawkilo-baseofsequencegenerated(NoteS8),roughlyone-ninthasmuchasthebestcostsforelec-trophoreticsequencing.Rawdatainallse-quencingmethodsaregenerallycombinedtoformaconsensus.Eventhoughcostsaregenerallydefinedintermsofrawbases,thecriticalmetrictocomparetechnologiesiscon-sensusaccuracyforagivencost.Thereisthusaneedtodeviseappropriatecostmetricsforspecificlevelsofconsensusaccuracy.Iflibraryconstructioncostsarenotin-cluded,theestimatedcostdropsto$0.08perrawkilobase.Higherdensitiesofamplifiedbeadsareexpectedtoboostthenumberofbasessequencedperexperiment.Whileimaging,datawerecollectedatarateof400bp/s.Althoughenzymaticstepsslowedouroverallthroputto140bp/s,adualflowcellinstrument(suchthatthemicroscopeisalwaysimaging)willallowustoachievecontinuousdataac-quisition.Enzymaticreagents,whichdominateourcostequation,canbeproducedin-houseatafractionofthecommercialprice. Table2.Polymorphismdiscovery.PredictionsformutatedpositionsweretestedandverifiedascorrectbySangersequencing.Wefoundthreemutationsuniquetotheevolvedstrain—twoin,aporin,andonein,aglobalregulator.TypeGeneContextConfirmationComments10regionYesEvolvedstrainonly8-bpdeletionFrameshiftYesEvolvedstrainonlyAlaYesEvolvedstrainonly776-bpdeletionPromoterYesMG1655heterogeneityUTRYesMG1655heterogeneity-red,3274GlyYes-redheterogeneity-red,9846GluYes-redheterogeneity Fig.3.Mate-pairedtagsandrearrangementdiscovery.()Diagnostic6%polyacrylamidegelofthesheared,size-selectedgenomicDNAfromwhichthelibrarywasconstructed.Lanes1and4aremolecularsizemarkers.Lane2representsthematerialusedinthelibrarysequencedtogeneratethepaired-tagmappingsin(B),andlane3representsgenomicDNAforadifferentlibrary.()Histogramofdistancesbetween1millionmappedmate-pairsequences.TheprobabilityofcircularizationfavorsintegralsofthehelicalpitchofDNA,suchthattheFouriertransformofthedistribution()yieldsapeakat10.6bp().Consistent,aberrantmappingofunplacedmate-pairstodistalsequencesrevealedinformationaboutunderlyingrearrangements.Topandbottombluebarsindicategenomicpositionsforproximalanddistaltags,respectively.Greenconnectionsindicatemate-pairingsthatfallwithinexpecteddistanceconstraints,whereasredandblackconnectionsindicateaberrantconnections(redindicatesconnectionsbetweenthesamestrand,andblack,connectionsbetweenoppositestrands).(C)Detectionofa776-bpdeletioninthepromoter().(D)Detectionofthereplacementofthelocuswiththelambdaredconstruct.(E)DetectionoftheP-regioninversion().Detectionoftheinversiononabackgroundofnormallymate-pairedreadsindicatesthattheinversionisheterogeneouslypresent.www.sciencemag.orgSCIENCEVOL3099SEPTEMBER2005 on August 30, 2012www.sciencemag.orgDownloaded from WeselectedaderivativeofE.coliMG1655,engineeredfordeficienciesintryptophanbio-synthesisandevolvedfor200generationsunderconditionsofsyntrophicsymbiosisviacoculturewithatyrosinebiosynthesis–deficientstrain().Specificphenotypesemergedduringthelaboratoryevolution,leadingtotheexpec-tationofgeneticchangesinadditiontointen-tionallyengineereddifferences.Aninvitromate-pairedlibrarywascon-structedfromgenomicDNAderivedfromasinglecloneoftheevolvedTrpstrain.Tosequencethislibrary,weperformedsuccessiveinstrumentrunswithprogressivelyhigherbeaddensities.Inanexperimentultimatelyyielding30.1Mbofsequence,26cyclesofsequencingwereperformedonanarraycontainingampli-fied,enrichedePCRbeads.Ateachcycle,datawereacquiredforfourwavelengthsat20opticalmagnificationbyrasteringacrosseachof516fieldsofviewonthearray(Fig.1D).AdetaileddescriptionofthestructureofeachsequencingcycleisprovidedinNoteS6.Intotal,54,696images(14bit,10001000)werecollected.Cycletimesaveraged135minperbase(90minforreactionsand45minforimaging),foratotalof60hoursperinstrumentrun.Imageprocessingandbasecallingalgo-rithmsaredetailedinNoteS7.Inbrief,allimagestakenatagivenrasterpositionwerealigned.Twoadditionalimagesetswereac-quired:brightfieldimagestorobustlyidentifybeadlocations(Fig.2A)andfluorescentprimerimagestoidentifyamplifiedbeads.Ouralgo-rithmsdetected14millionobjectswithinthesetofbrightfieldimages.Onthebasisofsize,fluorescence,andoverallsignalcoherenceoverthecourseofthesequencingrun,wedeter-mined1.6milliontobewell-amplified,clonalbeads(11%).Foreachcycle,meaninten-sitiesforamplifiedbeadswereextractedandnormalizedtoa4Dunitvector(Fig.2,BandC).TheEuclideandistanceoftheunitvectorforagivenrawbasecalltothemediancen-troidofthenearestclusterservesasanaturalmetricofthequalityofthatcall.ThereferencegenomeconsistedoftheMG1655genome(GenBankaccessioncodeU00096.2)appendedwithsequencescorre-spondingtothecatgeneandthelambdaRedprophage,whichhadbeenengineeredintothesequencedstraintoreplacethetrpandbiooperons,respectively.Tosystematicallyassessourpowertodetectsingle-basesubstitutions,weintroducedasetof100randomsingle-nucleotidechangesintothereferencesequenceatrandomlyselectedpositions(mockSNCsCs)(Table1).Analgorithmwasdevelopedtoplacethediscontinuousreadsontothereferencesequence(NoteS7).Thematchingcriteriarequiredthepairedtagstobeappropriatelyorientedandlocatedwithin700to1200bpofoneanoth-er,allowingforsubstitutionsifexactmatcheswerenotfound.Ofthe1.6millionreads,wewereabletoconfidentlyplace1.16million72%)tospecificlocationsonthereferencegenome,resultingin30.1millionbasesofresequencingdataatamedianrawaccuracyof99.7%.Atthisstageoftheanalysis,thedatawerecombinedwithreadsfromapre-viousinstrumentrunthatcontributedanaddi-18.1millionbasesofequivalentquality(Fig.2D).Inthislatterexperiment,1.8mil-lionreadsweregeneratedfrom7.6millionobjects(24%),ofwhich0.8millionwereconfidentlyplaced(High-confidenceconsensuscallswerede-terminedfor70.5%oftheE.coligenomeforwhichsufficientandconsistentcoveragewasavailable(3,289,465bp;generallypositionsorgreatercoverage).Thereweresixpositionswithinthissetthatdidnotagreewiththereferencesequence,andthusweretargetedforconfirmationbySangersequenc-ing.Allsixwerecorrect,althoughinonecasewedetectedtheedgeofan8-bpdeletionratherthanasubstitution(Table2).ThreeofthesesixmutationsrepresentheterogeneitiesinlambdaRedorMG1655,orerrorsinthe Fig.2.Rawdataacquisitionandbasecalling.()Brightfieldimages(areashowncorrespondsto0.01%ofthetotalgelarea)facilitateobjectsegmentationbysimplethresholding,allowingresolutionevenwhenmultiple1-mbeadsareincontact.()False-colordepictionoffourfluorescenceimagesacquiredatthislocationfromasingleligationcycle.A,gold;G,red;C,lightblue;T,purple.()Four-colordatafromeachcyclecanbevisualizedintetrahedralspace,whereeachpointrepresentsasinglebead,andthefourclusterscorrespondtothefourpossiblebasecalls.Shownisthesequencingdatafromposition(1)oftheproximaltagofacomplexE.coli–derivedlibrary.()Cumulativedistribu-tionofrawerrorasafunctionofrank-orderedqualityfortwoindependentexperiments(redtri-angles,18.1-Mbrun;bluesquares,30.1-Mbrun).Theaxisindicatespercentilebinsofbeads,sortedonthebasisofaconfidencemetric.Theaxis(logarithmicscale)indicatestherawbase-callingaccuracyofeachcumulativebin.EquivalentPhredscoresareQ20,Q30{Phredscore10[log(rawper-baseerror)]}.Cumulativedistributionofrawerrorwithsequencingbyligationcyclesconsideredindependentlyisshowninfig.S8. Table1.GenomeCoverageandSNCprediction.Baseswithconsistentconsensuscoveragewereusedtomakemutationpredictions.Toassesspower,theoutcomeofconsensuscallingforthemockSNCpositionswithvariouslevelsofcoveragewasdetermined.DatafromtwoindependentsetsofmockSNCsareshown.‘‘86of87,’’forexample,meansthat87ofthe100mockSNCswerepresentinthesequencethatwascoveredwith1ormorereads,and86ofthesewerecalledcorrectly.PercentofgenomeCorrectlycalledmocksubstitutionsorgreater91.4%86of8788of90orgreater83.3%78of7875of76orgreater74.9%67of6768of68orgreater66.9%58of5862of62EPORTS9SEPTEMBER2005VOL309SCIENCEwww.sciencemag.org on August 30, 2012www.sciencemag.orgDownloaded from Conventionalshotgunlibrariesarecon-structedbycloningfragmentedgenomicDNAofadefinedsizerangeintoanEscherichiacolivector.Sequencingreadsderivedfromoppositeendsofeachfragmentaretermedmate-pairs.te-pairs.ToavoidbottlenecksimposedbyE.colitransformation,wedevelopedamultiplexed,cell-freelibraryconstructionprotocol.Ourstrategy(Fig.1A)usesatypeIIsrestrictionendonucleasetobringsequencesseparatedonthegenomeby1kbintoproximity.Each135–basepair(bp)librarymoleculecontainstwomate-paired17-to18-bptagsofuniquege-nomicsequence,flankedandseparatedbyuni-versalsequencesthatarecomplementarytoamplificationorsequencingprimersusedinsubsequentsteps.Theinvitroprotocol(NoteS1)resultsinalibrarywithacomplexityofmillionunique,mate-pairedspecies.Conventionally,templateamplificationhasbeenperformedbybacterialcoloniesthatmustbeindividuallypicked.Polymerasecolony,orpolony,technologiesperformmultiplexampli-ficationwhilemaintainingspatialclusteringofidenticalamplicons(10).Theseincludeinsitupolonies(),insiturollingcircleamplification(RCA)(),bridgepolymerasechainreaction(PCR)(),picotiterPCR(),andemulsionPCR().InemulsionPCR(ePCR),awater-in-oilemulsionpermitsmillionsofnoninteract-ingamplificationswithinamilliliter-scalevolume(15–17).Amplificationproductsofin-dividualcompartmentsarecapturedviain-clusionof1-mparamagneticbeadsbearingoneofthePCRprimers().Anysinglebeadbearsthousandsofsingle-strandedcopiesofthesamePCRproduct,whereasdifferentbeadsbeartheproductsofdifferentcompartmentalizedPCRreactions(Fig.1B).ThebeadsgeneratedbyePCRhavehighlydesirablecharacteristics:highsignaldensity,geometricuniformity,strongfeatureseparation,andasizethatissmallbutstillresolvablebyinexpensiveoptics.Providedthatthetemplatemoleculesaresufficientlyshort(fig.S1),anoptimizedversionoftheePCRprotocoldescribedbyDressmanetal.()robustlyandreproduciblyamplifiesourcomplexlibraries(NoteS2).Inpractice,ePCRyieldsempty,clonal,andnonclonalbeads,whicharisefromemulsioncompartmentsthatinitiallyhavezero,one,ormultipletemplatemolecules,respectively.IncreasingtemplateconcentrationinanePCRreactionbooststhefractionofamplifiedbeadsatthecostofgreaternonclonality().Togeneratepopulationsinwhichahighfractionofbeadswasbothampli-fiedandclonal,wedevelopedahybridization-basedinvitroenrichmentmethod(Fig.1C).Theprotocoliscapableofafivefoldenrichmentofamplifiedbeads(NoteS3).IterativeinterrogationofePCRbeads(Fig.1D)requiresimmobilizationinaformatcompat-iblewithenzymaticmanipulationandepifluo-rescenceimaging.Wefoundthatasimpleacrylamide-basedgelsystemdevelopedforinsitupolonies()waseasilyappliedtoePCRbeads,resultinginaarrayofdis-ordered,monolayered,immobilizedbeads(NoteS4,Fig.2A).Withfewexceptions(),sequencingbio-chemistriesrelyonthediscriminatorycapaci-tiesofpolymerasesandligases(19–22).Weevaluatedavarietyofsequencingprotocolsinoursystem.Afour-colorsequencingbyligationscheme(degenerateligationligation)yieldedthemostpromisingresults(Fig.2,BandC).Adetailedgraphicaldescriptionofthismethodisshowninfig.S7.Webeginbyhybridizingananchorprimerrtooneoffourpositions(immediately5or3tooneofthetwotags).Wethenperformanenyzmaticligationreactionoftheanchorprimertoapopulationofdegen-eratenonamersthatarelabeledwithfluorescentdyes.Atanygivencycle,thepopulationofnonamersthatisusedisstructuredsuchthattheidentityofoneofitspositionsiscorrelatedwiththeidentityofthefluorophoreattachedtothatnonamer.Totheextentthattheligasediscrim-inatesforcomplementarityatthatqueriedpo-sition,thefluorescentsignalallowsustoinfertheidentityofthatbase(Fig.2,BandC).Afterperformingtheligationandfour-colorimaging,theanchorprimer:nonamercomplexesarestrippedandanewcycleisbegun.WithT4DNAligase,wecanobtainaccuratesequencewhenthequerypositionisasfarassixbasesfromtheligationjunctionwhileligatinginthe5direction,andsevenbasesfromtheligationjunctioninthe3Thisallowsustoaccess13bppertag(ahexamerandheptamerseparatedbya4-to5-bpgap)and26bpperamplicon(2tags13bp)(fig.S7).Althoughthesequencingmethodpresentedherecanbeperformedmanually,webenefitedfromfullyautomatingtheprocedure(fig.S3).Ourintegratedliquid-handlingandmicroscopysetupcanbereplicatedwithoff-the-shelfcom-ponentsatacostofabout$140,000.AdetaileddescriptionofinstrumentationandsoftwareisprovidedinNotesS5andS7.Asagenomic-scalechallenge,wesoughtamicrobialgenomethatwasexpected,relativetoareferencesequence,tocontainamodestnum-berofbothexpectedandunexpecteddifferences. Fig.1.Amultiplexapproachtogenomesequencing.()Sheared,size-selectedgenomicfragments(yellow)arecircularizedwithalinker(red)bearingMmeIrecognitionsites(NoteS1).Subsequentsteps,whichincludearollingcircleamplification,yieldthe134-to136-bpmate-pairedlibrarymoleculesshownatright.()ePCR()yieldsclonaltemplateamplificationon1-mbeads(NoteS2).()Hybridizationtononmagnetic,low-density‘‘capturebeads’’(darkblue)permitsenrichmentoftheamplifiedfraction(red)ofmagneticePCRbeadsbycentrifugation(NoteS3).Beadsareimmobilizedandmountedinaflowcellforautomatedsequencing(NoteS4).()Ateachsequencingcycle,four-colorimagingisperformedacrossseveralhundredrasterpositionstodeterminethesequenceofeachamplifiedbeadataspecificpositioninoneofthetags.Thestructureofeachsequencingcycleisdiscussedinthetext,NoteS6,andfig.S7.www.sciencemag.orgSCIENCEVOL3099SEPTEMBER2005 on August 30, 2012www.sciencemag.orgDownloaded from 3.S.Georgi,www.batteriesdigest.com/id380.htm(accessedJune2005).4.R.M.Alexander,J.Exp.Biol.,55(1991).5.G.A.Cavagna,N.C.Heglund,C.R.Taylor,Am.J.,R243(1977).6.J.Drake,,90(2001).7.S.Stanford,R.Pelrine,R.Kornbluh,Q.Pei,inProceedingsofthe13thInternationalSymposiumonUnmannedUntetheredSubmersibleTechnologyUnderseaSystemsInstitute,Lee,NH,2003).8.T.Starner,J.Paradiso,inLowPowerElectronicsDesign(CRCPress,BocaRaton,FL,2004),p.45–1.9.G.A.Cavagna,M.Kaneko,J.Physiol.,647(1977).10.S.A.Gard,S.C.Miff,A.D.Kuo,Hum.Mov.Sci.597(2004).11.Supportingmaterialisavailableon12.Becauseitisaprototype,therehasbeennoattempttoreducetheweightofthebackpack—indeed,itissubstantially‘‘overdesigned.’’Further,the5.6kgincludestheweightofsixloadcellsandone25-cm-longtransducer,eachwithaccompanyingbracketsandcables,aswellasothercomponentsthatwillnotbepresentonatypicalpack.Infutureprototypes,weestimatethattheweightwillexceedthatofanormalbackpackbynomorethan1to1.5kg.13.Underhigh-powerconditions(5.6kmhour20-and29-kgloadsand4.8kmhourwitha38-kgload),powergenerationontheinclinewasthesameasontheflat.Underlow-powerconditions(4.8kmwith20-and28-kgloads),electricitygener-ationontheinclinewasactuallysubstantiallygreaterthanthatontheflat(tableS1).14.R.Margaria,BiomechanicsandEnergeticsofMuscular(Clarendon,Oxford,1976).15.R.A.Fergusonetal.J.Physiol.,261(2001).16.G.A.Cavagna,P.A.Willems,M.A.Legramandi,N.C.J.Exp.Biol.,3413(2002).17.A.Grabowski,C.T.Farley,R.Kram,J.Appl.Physiol.,579(2005).18.J.M.Donelan,R.Kram,A.D.Kuo,J.Exp.Biol.3717(2002).19.J.M.Donelan,R.Kram,A.D.Kuo,J.Biomech.,11720.J.S.Gottschall,R.Kram,J.Appl.Physiol.,1766(2003).21.Becausethissavingsinmetabolicenergyrepresentsonly6%ofthenetenergeticcostofwalkingwiththebackpack(492W)(tableS3)(),accuratede-terminationsofthepositionandmovementsofthecenterofmass,aswellasthedirectionandmagnitudeofthegroundreactionforces,areessentialtodiscernthemechanism.Thiswillrequiretwin–force-platformsingle-legmeasurements,aswellasacompletekine-maticsandmechanicalenergyanalysis().Theenergyanalysisismademorecomplexbecausethepositionoftheloadwithrespecttothebackpackframeandtheamountofenergystoredintheback-packspringsvaryduringthegaitcycle.Finally,elec-tromyogrammeasurementsarealsoimportanttotestwhetherachangeineffectivemusclemomentarmsmayhavecausedachangeinthevolumeofactivatedmuscleandhenceachangeinmetaboliccost(22.K.Schmidt-Nielsen,AnimalPhysiology:AdaptationandEnvironment(CambridgeUniv.Press,Cambridge,ed.3,1988).23.Thisassumesthatelectronicdevicesarebeingpow-eredinrealtime.Iftherewereapowerlossof50%associatedwithstorage(suchasinbatteries)andre-coveryofelectricalenergy,thenthesefactorswouldbehalved.24.Whennotwalking,therackcanbedisengagedandthegeneratorcrankedbyhandorbyfoot.Electricalpowersof3Wareachievablebyhand,andhigherwattagecanbeachievedbyusingthelegtopowerit.25.R.Kram,J.Appl.Physiol.,1119(1991).26.A.E.Minetti,J.Exp.Biol.,1265(2004).27.A.A.Biewener,C.T.Farley,T.J.Roberts,M.Temaner,J.Appl.Physiol.,2266(2004).28.T.M.Griffin,T.J.Roberts,R.Kram,J.Appl.Physiol.,172(2003).29.ThisworkwassupportedbyNIHgrantsAR46125andAR38404.SomeaspectsoftheprojectweresupportedbyOfficeofNavalResearchgrantN000140310568andagrantfromtheUniversityofPennsylvaniaRe-searchFoundation.TheauthorsthankQ.Zhang,H.Hofmann,W.Megill,andA.Dunhamforhelpfuldis-cussions;R.Sprague,E.Maxwell,R.Essner,L.Gazit,M.Yuhas,andJ.Milliganforhelpingwiththeexperimen-tation;andF.Letterioformachiningthebackpacks.SupportingOnlineMaterialwww.sciencemag.org/cgi/content/full/309/5741/1725/MaterialsandMethodsSOMTextFigs.S1andS2TablesS1toS414February2005;accepted25July2005AccurateMultiplexPolonySequencingofanEvolvedBacterialGenomeJayShendure,GregoryJ.Porreca,NikosB.Reppas,XiaoxiaLin,JohnP.McCutcheon,AbrahamM.Rosenbaum,MichaelD.Wang,KunZhang,RobiD.Mitra,GeorgeM.ChurchWedescribeaDNAsequencingtechnologyinwhichacommonlyavailable,inexpensiveepifluorescencemicroscopeisconvertedtorapidnonelectrophoreticDNAsequencingautomation.WeapplythistechnologytoresequenceanevolvedstrainofEscherichiacoliatlessthanoneerrorpermillionconsensusbases.Acell-free,mate-pairedlibraryprovidedsingleDNAmoleculesthatwereamplifiedinparallelto1-micrometerbeadsbyemulsionpolymerasechainreaction.Millionsofbeadswereimmobilizedinapolyacrylamidegelandsubjectedtoautomatedcyclesofsequencingbyligationandfour-colorimaging.Costperbasewasroughlyone-ninthasmuchasthatofconventionalsequencing.Ourprotocolswereimplementedwithoff-the-shelfinstrumentationandreagents.TheubiquityandlongevityofSangersequenc-ing()areremarkable.Analogoustosemicon-ductors,measuresofcostandproductionhavefollowedexponentialtrends().High-throughputcentersgeneratedataataspeedof20rawbasesperinstrument-secondandacostof$1.00perrawkilobase.Nonetheless,optimizationsofelec-trophoreticmethodsmaybereachingtheirlim-its.Meetingthechallengeofthe$1000humangenomerequiresaparadigmshiftinourunder-lyingapproachtotheDNApolymer(Cyclicarraymethods,anattractiveclassofalternativetechnologies,areareinthattheyleverageasinglereagentvolumetoenzymaticallymanipulatethousandstomil-lionsofimmobilizedDNAfeaturesinparal-lel.Readsarebuiltupoversuccessivecyclesofimaging-baseddataacquisition.Beyondthiscommonthread,thesetechnologiesdi-versifyinapanoplyofways:single-moleculeversusmultimoleculefeatures,orderedversusdisorderedarrays,sequencingbiochemistry,scaleofminiaturization,etc.().Innovativeproof-of-conceptexperimentshavebeenre-ported,butaregenerallylimitedintermsofthroughput,featuredensity,andlibrarycom-plexity().Arangeofpracticalandtech-nicalhurdlesseparatethesetestsystemsfromcompetingwithconventionalsequencingongenomic-scaleapplications.Ourapproachtodevelopingamorematurealternativewasguidedbyseveralconsider-ations.(i)Anintegratedsequencingpipelineincludeslibraryconstruction,templateampli-fication,andDNAsequencing.Wethereforesoughtcompatibleprotocolsthatmultiplexedeachsteptoanequivalentorderofmagnitude.(ii)Asmoregenomesaresequenceddenovo,demandwilllikelyshifttowardgenomicrese-quencing;e.g.,tolookatvariationbetweenin-dividuals.Forresequencing,consensusaccuracyincreasesinimportancerelativetoreadlengthbecauseareadneedonlybelongenoughtocorrectlypositionitonareferencegenome.However,aconsensusaccuracyof99.99%,i.e.,theBermudastandard,wouldstillresultinhun-dredsoferrorsinamicrobialgenomeandhun-dredsofthousandsoferrorsinamammaliangenome.Toavoidunacceptablenumbersoffalse-positives,aconsensuserrorrateof1isamorereasonablestandardforwhichtoaim.(iii)Wesoughttodevelopsequencingchemistriescompatiblewithconventionalepi-fluorescenceimaging.Diffraction-limitedopticswithcharge-coupleddevicedetectionachievesanexcellentbalancebecauseitnotonlypro-videssubmicrometerresolutionandhighsen-sitivityforrapiddataacquisition,butisalsoinexpensiveandeasilyimplemented. DepartmentofGenetics,HarvardMedicalSchool,Boston,MA02115,USA.DepartmentofGenetics,HowardHughesMedicalInstitute,WashingtonUni-versity,St.Louis,MO63110,USA.*Theseauthorscontributedequallytothiswork.Towhomcorrespondenceshouldbeaddressed.E-mail:shendure@alumni.princeton.edu(J.S.),gregory_porreca@student.hms.harvard.edu(G.J.P.)EPORTS9SEPTEMBER2005VOL309SCIENCEwww.sciencemag.org on August 30, 2012www.sciencemag.orgDownloaded from DOI: 10.1126/science.1117389, 1728 (2005);309 Science et al.Jay ShendureGenome This copy is for your personal, non-commercial use only. clicking here.colleagues, clients, or customers by , you can order high-quality copies for yourIf you wish to distribute this article to others   here.following the guidelines can be obtained byPermission to republish or repurpose articles or portions of articles   ): August 30, 2012 www.sciencemag.org (this information is current as ofThe following resources related to this article are available online at http://www.sciencemag.org/content/309/5741/1728.full.htmlversion of this article at: including high-resolution figures, can be found in the onlineUpdated information and services, http://www.sciencemag.org/content/suppl/2005/10/12/1117389.DC1.html can be found at: Supporting Online Material http://www.sciencemag.org/content/309/5741/1728.full.html#relatedcan berelated to this article A list of selected additional articles on the Science Web sites 320 article(s) on the ISI Web of Sciencecited by This article has been http://www.sciencemag.org/content/309/5741/1728.full.html#related-urls83 articles hosted by HighWire Press; see:cited by This article has been registered trademark of AAAS. is aScience2005 by the American Association for the Advancement of Science; all rights reserved. The title CopyrightAmerican Association for the Advancement of Science, 1200 New York Avenue NW, Washington, DC 20005. (print ISSN 0036-8075; online ISSN 1095-9203) is published weekly, except the last week in December, by theScience on August 30, 2012www.sciencemag.orgDownloaded from