QUORUMSENSINGANDTHESTABILIZATIONOFCOOPERATIVEBEHAVIORINVIBRIOBACTERIAByEricLeeBrugerADISSERTATIONSubmittedtoMichiganStateUniversityinpartialful˝llmentoftherequirementsforthedegreeofMicrobiologyandMolecularGenetics-DoctorofPhilosophyEcology,EvolutionaryBiology,andBehavior-DoctorofPhilosophy2016ABSTRACTQUORUMSENSINGANDTHESTABILIZATIONOFCOOPERATIVEBEHAVIORINVIBRIOBACTERIAByEricLeeBrugerThisthesisstudiestheconnectionsbetweentwoubiquitousbiologicalphenomenon:co-operationandcommunication;andonemaydebatewhetherthelatterisaspeci˝cexampleoftheprior.Morespeci˝cally,thequestionsexaminedinthisdocumentcenteronbacteriaandtheircommonformofchemicalcommunicationknownasquorumsensing.Mybroadinterestsandopinionsregardingquorumsensingarethattheyplayanimmenseroleinprovidingfeedbacksthatadjustcellularbehaviorandgeneexpressiontooptimizegrowthinresponsetoaplethoraofenvironmentalcuesandsignals.Themorespeci˝ctopicsexaminedinthisdocumentrelatetothetendencyofquorumsensingtoregulatecooperativebehaviorssuchaspublicgoodsproduction.Suchcooperativebehaviorsaresusceptibletocheatingbynon-participantsiftheycanenjoyabene˝tthatagivenbehaviorproduceswithoutcontributingtothatbehavior.Weexaminedthestabilityofquorumsensinginavarietyofconditions,particularlywherepublicgoodspositivelycontributeto˝tness,andpursueexplanationsforhowcooperativegoodsproductionandtheoverlayingregulationofquorumsensingcanbemaintainedandstabilizedinthepresenceofpotentialcheats,andhowsuchbehaviorholdsupoverevolutionarytime.Afewinterestingresultswererevealedfromthisresearch.Firstly,itwasde˝nitivelyshownthatquorumsensingregulationprovidesadistinctadvantageinthefaceofcheatscomparedtounconditionalcooperators.Quorumsenserswereabletoappropriatelyregulatecellphysiology,includingpublicgoodsproduction,insuchawaythatthisstrategyhadequivalent˝tnesstoanobligatedefectorintheenvironmenttested,whereasunconditionalcooperatorspaidheavycostsintermsofgrowthrate,andwerethusoutcompetedbyotherstrainsatlowcelldensities,andalsocheatedbydefectorsathighcelldensities.Together,thisprovidesthestrongestreportedexperimentalevidencetodatethatquorumsensingregulationitselfcanstabilizecooperativebehaviors,evenintheabsenceofotherlayeredmechanismssuchaspolicingorpositiveassortmentbyclustering.Additionally,cooperativestrategieswerefoundtobeabletoinvademetapopulationsofmostlydefectorswhensu˚cientassortmentofcompetingtypeswasachieved.ThiswasfoundtobetrueforbothVibrioharveyiandVibriocholerae.Maximumevasionofdefectorswasachievedwhencooperatorstrainswereabletobothformstructurestoseparatefromcompetitors(bywayofdispersalbymotility),andtheabilitytodispersefromothercellsthroughmotilephenotypes.Again,themostsuccessfulstrategyinthisregardcamefromthefunctionalwildtypequorumsenserascomparedtoanunconditionallycooperatingstrategy.Lastly,experimentalevolutionofVibrioharveyipopulationswasconductedinminimalmediafor2000generations.Thesepopulationsunderwentavarietyofadaptationstotheexperimentalenvironment,whichwasexaminedindetailthroughhigh-throughputgenomicsequencing.Itwasdiscoveredthatwildtypepopulationssustainedhigherfractionsofbiolu-minescentcooperatorsaswellasahigherdegreeofphenotypicdiversityintermsofquorumsensing,whilenearlyallunconditionalcooperatorpopulationsexperiencedrapidevolutionandsweepsbydenovodefectors.Additionally,manyoftheevolvedformsinwildtypepopulationsappeartobeoptimizingandexhibitintermediatebioluminescentandproteaseproductionphenotypes.Tomyfamily:ToSilas,mychronometerandencyclopediaToElla,mygoofballandspit˝reToJess,myrockandcompassivACKNOWLEDGMENTSThisdissertationwasenabledandenhancedbyalargenumberofpeople,andwouldlookunrecognizablewithouttheirpresenceandin˛uence.Firstinallthings,thecompletionofthisdocumentservesasatestamenttomyfamilyandtheirdedicationtomypursuitofthisgoal.Mychildren,SilasandElla,havehardlyknownalifewheretheirfatherhasnotbeeninschoolpursuinghisPh.D.Theyareextremelyjoyfulandenergetickids,andhelpmetoforgetmytroublesandtorememberallthatisgoodandimportantinlife.WordscannotpayjusticetohowmuchgratitudeIowemywife,Jessica,forcompletingthisdegree.ShehasbeenceaselesslyandferociouslydedicatedtomeandourfamilyanddeservesallthepraiseIcangive.Thisdegreemaynothavestartedandde˝nitelywouldnothavecompletedwithoutherpresence.IamforevergratefulforherwillingnesstocomealongonthisridewithmeandformaintainingherfaithinmetoseethisPhDtoitsend.IonlyhopeIcanrepaythefavortoyousomedayandalsothatIcanliveandworkinawaytobeworthyofthatgoodfaith.TomyparentsforneverdiscouragingwhatIassumewasaprecociousandoftenannoy-inglycuriouschildwhowasneversatis˝edwiththesimpleanswertoaquestion.Inowknowwhatitfeelsliketobeontheothersideofthisinteraction,anditmakesmeappreciateyourpatienceallthemore.AchildisfortunatetohavetheloveandencouragementthatIhavefeltthroughoutmylife.IhopeIcanprovidethissortofmodelformyownchildren.Tomycommitteemembers,Dr.Dworkin,Dr.Kroos,andDr.Lenski,Ithankyouforyourguidanceoverthecourseofmydissertation,whichhashelpedmetobecomeabetterscientist.Thankyou,Ian,forprovidinganevolutionarybiologist'sperspectivetosomeonewhocametoMSUwithoutmuchofabackgroundinevolution,andforintroducingmetovmanynewwaystoanalyzedata.Thankyou,Lee,foryourattentiontodetailandyourcomprehensiveknowledgeofbacterialphysiologyandgenetics.Thankyou,Rich,foryouextensiveknowledgeofeverythingmicrobialevolutionandgoodinstinctsonthedirectiontotakemyprojectsandpublications.AlsothankyoutoDr.GemmaReguera,whomIappreciateforlettingmerotatewithinmy˝rstyearandforchairingmypreliminaryexam.TomyPI,Dr.ChrisWaters,Iamappreciativeofyourencouragingmanner,openap-proachtoscience,yourabilitytodeciphertheinterestingfromthefutile,andyourtoleranceforalltheinterestsandtangentialpathsIhavewandereddownduringthecourseofmytimeinyourlab.Althoughbuildingthisprojectwasachallenge,Iamveryproudofwereithasled,andhopefulaboutwhereitwillcontinuetolead.Tothemanylab-matesintheWaterslabIhaveknownovertheyears,thankyouforkeepingmehappyandsane,whileonverynormaloccasionsnearlydrivingmeinsane.Thankyouforbeingmyveryidiosyncraticworkfamily,andhelpingremindmethatworkandsciencecanbefunwithoutsacri˝cingrigor.Thanks˝rstandforemosttoDishaSrivastavaandBenKoestlerformakingmefeelwelcomewithoutreservationsfromtheverybeginning.ThankstoJohnShookforhelpingmeholdlabbarnighttogetherwhenitotherwisewouldhaveassuredlycollapsed.ThankstoAlessandraAgostinho(Hunt)forbeingmymostconstantbay-mateduringmytimehere.Bay1forever!Lastly,thankstothemanysmart,talented,interesting,fun,helpful,drivenpeopleIhavemetinavarietyof˝eldsandcontextsduringmyPhDatMSU,whetherthatbeotherstudentsintheMMG,EEBB,orBMSprograms,FASTfellows,BEACONites,orGraduateSchool.Thankstothemanyfaculty,sta˙,andstudentswhohavehelpedmenavigatethisprocess.ThankstothefundingsupportIhavereceivedatMSUfromtheGraduateSchool,EEBB,MMG,FAST,BEACON,andCollegeofNaturalScience.IhavefeltincrediblesupportandviasenseofcommunityatMSU,andfeelingveryblessedtohavecompletedmyPhDinsucharichacademicenvironment.viiTABLEOFCONTENTSLISTOFTABLES....................................xiLISTOFFIGURES...................................xiiKEYTOSYMBOLS..................................xivChapter1Evolutionarymechanismsthatmaintainbacterialcooperation....................1Preface..........................................21.1Introduction....................................41.2Facultativecooperation..............................51.3Spatialstructureandassortment........................91.4Policing......................................121.5DivisionofLaborinBacteria..........................141.6Conclusion.....................................15Chapter2Bacterialquorumsensingstabilizescooperationbyoptimizinggrowthstrategies............................17AbstractandImportance................................182.1Introduction....................................202.2MaterialsandMethods..............................222.2.1BacterialStrainsandMediaandGrowthConditions..........222.2.2CompetitionFitnessAssays.......................232.2.2.1CompetitionDesign......................232.2.2.2FitnessCalculations......................232.2.3ProteaseAssays..............................242.2.4StatisticalAnalyses............................242.3Results.......................................252.3.1QSisrequiredformaximumgrowthofVibrioharveyiutilizingcasein..............................252.3.2FunctionalandmutantQSstrainsdi˙erindensity-dependentproteaseactivity..................................262.3.3AfunctionalQScircuitpreventsdefectorinvasion...........282.3.4AfunctionalQSsystempreventsdefectorinvasionbymodulatinggrowthstrategiesatdi˙erentcelldensities...................372.4Discussion.....................................39Chapter3DispersalMaintainsQuorumSensinginBacteriaviaaSimpson'sParadox........................47viiiAbstract.........................................483.1Introduction....................................493.2MaterialsandMethods..............................523.2.1StrainsandMedia............................523.2.2MetapopulationGrowth.........................523.2.3Motilityplates..............................533.3Results.......................................543.3.1ComparinggrowthofVibriocooperatorsanddefectorsinM9-caseinmedia...................................543.3.2AmetapopulationapproachselectsforincreasesofVibriocooperatorstrains...................................563.3.3CombineddispersalbymotilityandstronggrowthincreasescooperatorfrequencyinM9-casein..........................603.4Discussion.....................................64Chapter4ExperimentalevolutionofquorumsensinginVibrioharveyi.71Preface..........................................724.1Introduction....................................734.2MaterialsandMethods..............................754.2.1Competitionswithsupplementedautoinducers.............754.2.2EvolutionExperiments..........................754.2.3Productivityestimates..........................764.2.4Phenotypemeasurements........................764.3Results.......................................774.3.1Exogenousautoinducerincreasesselectionfordefectors,especiallycooperator-deriveddefectors.................774.3.2Long-termexperimentalevolutionoftheWTandUCQSstrains...834.3.3Defectorlevelshavestronge˙ectsonpopulationlevelphenotypesinM9-casein.................................884.3.4Evolvednon-luminescentdefectorspossessmultiplephenotypicdi˙er-encesfromancestralcooperatorstrains.................904.4Discussion.....................................93Chapter5QuorumsensinginVibrioharveyiandgrowthpromotedbylowcelldensityregulation.......................100Preface..........................................1015.1Introduction....................................1025.2MaterialsandMethods..............................1035.3Results.......................................1035.4Discussion.....................................109Chapter6ConcludingRemarks..........................1126.1Introduction....................................1136.2StabilizationofVibriocooperativeproteaseproductionbyQS........1136.2.1Conclusions................................113ix6.2.2Futuredirections.............................1166.3Simpson'sParadoxinVibriobacteria......................1206.3.1Conclusions................................1206.3.2FutureDirections.............................1226.4EvolutionaryoutcomesofVibrioharveyi....................1246.4.1Conclusions................................1246.4.2FutureDirections.............................1256.5Low-cell-densitybehaviors............................1296.5.1Conclusions................................1296.5.2Futuredirections.............................1306.6Summary.....................................131APPENDIX........................................133REFERENCES......................................135xLISTOFTABLESTable2.1:StrainsofV.harveyiusedinthisstudy.................22TableA.1:StrainsofV.harveyiandV.choleraeusedinthiswork........134xiLISTOFFIGURESFigure1.1:Mechanismsthatacttomaintaincooperation.............7Figure2.1:V.harveyirequiresQSformaximalgrowthinM9-caseinmedia...27Figure2.2:ExtracellularProteaseactivityisregulatedbyQSinV.harveyi...29Figure2.3:V.harveyirequiresQSformaximalsignalproductioninM9-casein.30Figure2.4:Fitnessoutcomesfromsinglegrowthcyclecompetitionsbetweendif-ferentV.harveyigenotypesvarygreatlyinM9-caseinbutnotinM9-tryptone...............................32Figure2.5:V.harveyiQSprovidesequivalent˝tnessagainstdefectorsinM9-caseinoverarangeofdensityconditions................33Figure2.6:Growthperformanceofdefectorsandmixedpopulationsincompeti-tioninM9-casein.............................34Figure2.7:TheluxRdefectorinvadesluxOUbutnotWT...........35Figure2.8:TheWTstrainoptimizesgrowthstrategiesatlow-andhigh-celldensity.36Figure2.9:GrowthperformanceofluxRdefectorinthepresenceandabsenceofcompetingstrains...........................38Figure3.1:GrowthofV.harveyiandV.choleraestrainsinM9-caseinmedia..55Figure3.2:CompetitiveoutcomesofmixedV.harveyicooperatorsanddefectorsinunmixedM9-caseinmedia.......................57Figure3.3:DemonstrationofSimpson'sparadoxinmetapopulationsofV.har-veyiandV.choleraegrowninM9-casein................59Figure3.4:Phenotypicpro˝lesofsubpopulationsofV.harveyi..........61Figure3.5:Vibriostrainsexhibitacolonization-dispersaltradeo˙inM9-caseinmedia...................................62Figure3.6:Competitiveoutcomesinmotilityplatecompetitions.........64xiiFigure3.7:Competitiveoutcomesof˛rAinmotilityplatecompetitions....65Figure3.8:DispersalofV.harveyipopulationsinmonocultureandincompeti-tioninM9-caseinmotilityplates.....................67Figure4.1:CompetitionsofV.harveyistrainsinM9-caseinmediainthepresenceandabsenceofsupplementedautoinducers...............79Figure4.2:Frequencyandgrowthresultsfromexperimentallyevolvedpopula-tionsofV.harveyi............................82Figure4.3:ExperimentallineagesofV.harveyistrainsevolvedinM9-caseinme-diaover2000generationsexaminedonpetriplates..........85Figure4.3:GrowthandbioluminescenceofV.harveyipopulationsevolvedinM9-caseinmedia.............................89Figure4.4:ExaminationofgrowthandproteaseproductionofevolvedclonesinM9-casein.................................92Figure5.1:GrowthofV.harveyistrainsinM9mediawithdi˙erentsugaralco-holsasthelimitingcarbonsource....................104Figure5.2:GrowthofV.harveyistrainsinM9mediawithglutamateasthelimitingcarbonsource..........................106Figure5.3:GrowthofV.harveyistrainsinM9-glucosemediasupplementedwithdi˙erentphosphorussources.......................107Figure5.4:GrowthofV.harveyistrainsinM9mediawithchitinorNAGsup-plementedasthesolecarbonsource...................108Figure6.1:GrowthofV.harveyistrainsinM9-caseinmediainthepresenceandabsenceofglucose.............................119xiiiKEYTOSYMBOLSQSQuorumsensingHCDHighcelldensityLCDLowcelldensityWTWildtypeUCUnconditionalcooperatorAHLAcylhomoserinelactonesignalmoleculesCAACasaminoacidsM9-caseinM9mediacontainingcaseinasthesolecarbonsourceM9-CAAM9mediacontainingcasaminoacidsasthesolecarbonsourceLBLysogenybroth(LB),arichmediumforcellulargrowthPGPublicgoodSNSignalnegativestrain,atriplemutantforQSsignalsynthasegenesAIAutoinducer,signalmoleculethatinducesquorumsensingluxRVibrioharveyistrainwithluxRgenedeletedluxOUVibrioharveyistrainwithluxOUgenesdeletedluxOVibriocholeraestrainwithluxOgenedeletedhapRVibriocholeraestrainwithhapRgenedeletedNAGN-AcetylglucosamineNACnon-ancestral-cooperator,evolvedvariantwithreducedbioluminescencephenotypexivChapter1Evolutionarymechanismsthatmaintainbacterialcooperation1PrefaceThecontentsofthischapterwereadaptedfromanarticlepreviouslypublishedintheF1000Researchjournalin2015(Citation:BrugerEandWatersC.Sharingthesandbox:Evo-lutionarymechanismsthatmaintainbacterialcooperation[version1;referees:2approved].F1000Research2015,4(F1000FacultyRev):1504(doi:10.12688/f1000research.7363.1)).Someslightmodi˝cationshavebeenmadefromtheoriginalpublishedtext.TheoriginalmanuscriptwaspublishedunderaCCBY4.0license(detailsviewableonlineathttps://creativecommons.org/licenses/by/4.0/legalcode).Copyright:c2015BrugerEandWatersC.ThisisanopenaccessarticledistributedunderthetermsoftheCreativeCom-monsAttributionLicence,whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.Disclaimer:CreativeCommonsCorporationeisnotalaw˝rmanddoesnotprovidelegalservicesorlegaladvice.DistributionofCreativeCommonspubliclicensesdoesnotcreatealawyer-clientorotherrelationship.CreativeCommonsmakesitslicensesandrelatedinformationavailableonanâ•œas-isâ•šbasis.CreativeCommonsgivesnowarrantiesregardingitslicenses,anymateriallicensedundertheirtermsandconditions,oranyrelatedinformation.CreativeCommonsdisclaimsallliabilityfordamagesresultingfromtheirusetothefullestextentpossible.Inthischapter,Iexamineavarietyofpotentialproximatemechanismsthatcouldacttostabilizeandpreservecooperationinmicrobes.Quorumsensingisacentralfocusinthischapter,aswesuspectithascriticalrolestoplayformanycooperativebehaviorsinbacteria,bothonitsownandinconcertwithothermechanismsdiscussedinthechapter.Whilewe2predictthatmanyinstancesofsuchQS-mediatedstabilizationexistinnature,fewexampleshavebeenreportedatthispointintime.Microbesarenowknowntoparticipateinanextensiverepertoireofcooperativebehav-iorssuchasbio˝lmformation,productionofextracellularpublic-goods,groupmotility,andhigher-orderedmulticellularstructures.Afundamentalquestionishowthesecooperativetasksaremaintainedinthefaceofnon-cooperatingdefectorcells.Recently,anumberofmolecularmechanismsincludingfacultativeparticipation,spatialsorting,andpolicinghavebeendiscoveredtostabilizecooperation.Oftenthesedi˙erentmechanismsworkinconcerttoreinforcecooperation.Inthisreview,Idescribebacterialcooperationandthecurrentunderstandingofthemolecularmechanismsthatmaintainit.31.1IntroductionBacteriawereoncethoughttobesolitaryindividuals,butitisnowclearthattheyleadcomplexsociallives(Crespi,2001,Velicer,2003).Multicellularbacterialcommunitiestermedbio˝lmsarenowconsideredanormalformofbacterialgrowth.Bacterialchemicalcommu-nication,includingquorumsensing(QS),isubiquitous(PlattandFuqua,2010,WatersandBassler,2005),andthemolecularunderpinningsofmulticellularbacterialstructuressuchasMyxobacteriafruitingbodiesandStreptomyces˝lamentsarealsobeingelucidated(Kroos,2007,PetrusandClaessen,2014).Withourincreasedunderstandingofbacterialsocialitycomesafurtherappreciationoftheroleofcooperationinmanybacterialprocesses.Mi-crobialcooperativebehaviorshaveimportantimpactsonourownlives,includingantibioticresistance(Leeetal.,2010),bio˝lmformationinchronicinfections(Percivaletal.,2010),andvirulenceduringacuteinfections(Diardetal.,2013,2014).Explainingtheevolutionofcooperativetaskshaslongchallengedevolutionarybiology,asthesesystemsappearripeforexploitationbynon-cooperatingdefector/cheatercellsthatreceivethebene˝tsofcoop-erationwithoutpayingthecostofproduction(Westetal.,2007).Becauseoftheirshortgenerationtimes,largepopulationsizes,smallgenomes,andasexualreproduction,bacteriaarenowrecognizedasidealmodelsystemstounderstandthefactorsleadingtotheevolu-tionandpersistenceofcooperativebehaviors(BarrickandLenski,2013,ElenaandLenski,2003,Westetal.,2006).Inthisreview,wewillsummarizefrombothaconceptualandamechanisticperspectiveourunderstandingofhowcooperationismaintainedinbacteria.41.2FacultativecooperationBacteriahaveevolvedcomplexregulatorycircuitrytorespondande˙ectivelyacclimatetodi˙erentenvironments,soitisnotsurprisingthatthis˛exibleregulatorycircuitrycanalsobeutilizedtocontrolcooperativetraits.Cooperativebehaviorsinbacteria,suchastheproductionofextracellulargoomolecules,de˝nedasresourcesthatcanbeutilizedbyboththeproducersandthenon-producersinthecommunity,areexploitablebynon-producingcheater/defectorcells.Oneapproachtolimitcheaterinvasionisfacultativecooperation.Engagingincooperationatlimitedtimes,particularlywhenthebene˝tisthegreatest,orinenvironmentalconditionswherethecostofcooperationislowcanlimitorpreventcheaterinvasion(Heilmannetal.,2015,HenseandSchuster,2015).Inthisway,bacteriamaypreservecooperationinconditionsthatwouldotherwisefavoritscollapse(Cornforthetal.,2012).Itisnotable,however,thatfacultativeparticipationonlypartlymediatestheproblemofcooperationbylimitingthetimeswhenacellmustmaintainit.Othermechanisms,suchasrelatedness,arelikelyrequiredinconjunctionwithoptionalparticipationtopreservecooperation.Forpublicgoodstobee˙ective,theyoftenmustexceedathresholdconcentrationintheextracellularenvironment(Heilmannetal.,2015).Therefore,theremustbeasu˚cientnumberofproducingcellscontributingtothepublicgood.Forthisreason,productionofmanypublicgoodssuchasexoenzymes,proteases,chitinases,andsiderophoresareregulatedbyQS(Fig.1A)(WatersandBassler,2005).Thisprocessreliesonthesecretionanddetectionbybacteriaofsmallchemicalsignalsknownasautoinducersintotheextracellularenvironment.Asthecelldensityofagrowingcultureincreases,sodoestheconcentrationofautoinducers.ThisisreinforcedbythepositivefeedbackofmanyQSsystemsonautoinducer5synthesis(NgandBassler,2009,Williamsetal.,2008).Ataspeci˝cconcentrationofsignal,receptorsbindtoandsensetheseautoinducers,allowingthebacteriatoswitchfromalow-tohigh-celldensitystate.Thisisoftenseenasatransitionfromnon-productiontoproductionofcooperativetraitssuchasextracellularpublicgoods.QSitselfisanexploitablecooperativebehavior,asQS-speci˝ccheatersthatdonotsignal,overproducesignal,ordonotrespondtosignalcanevolve(Diggleetal.,2007).QSregulationofcooperationhasbeenwellstudiedinthebacteriumPseudomonasaerug-inosa.Thisbacteriumupregulatesextracellularproteasesinthehigh-celldensitystate(BrintandOhman,1995).Theseproteasesdegradeextracellularproteins,liberatingsmallerpep-tidesthatcanbeusedforgrowth.Thus,thegrowthofP.aeruginosainminimalmediawiththeproteincaseinasacarbonsourceisdependentuponafunctionalQSsystem.MutantsintheQSpathwaydonotsecretehighlevelsofproteasesandcannotgrowinthisenvironment,buttheyreceivenegativefrequency-dependent˝tnessbene˝tswhenmixedwithacooperat-ingstrain(Diggleetal.,2007).Inotherwords,QSmutantscaninvadethewild-typestrainwhenrarebutlosetheir˝tnessbene˝tswhencommon,afeaturelikelytooccurinmostpublicgoodsscenarios(Diggleetal.,2007,Ross-Gillespieetal.,2007,Sandozetal.,2007).Therefore,althoughQSlimitsmaximumpublicgoodproductiontohigh-celldensity,itisnotsu˚cientinthiscasetocompletelypreventcheaterinvasion,althoughitmaymediatetheextenttowhichthisoccurs.SimilarresultsinacaseingrowthmediumwererecentlydescribedforVibriocholerae(Katzianeretal.,2015)andVibrioharveyi(unpublishedre-sults,Waterslaboratory),whichalsosecreteproteasesathigh-celldensityinaQS-dependentmanner.ThedegreeofresistancetocheatingQScanprovidelikelydependsuponthecostandbene˝tfunctionsofthebehavior,themannerinwhichregulationisimposed,andthegeneticarchitectureoftheQSsystem.6Figure1.1:Mechanismsthatacttomaintaincooperation.A.QuorumSensing.Thecooperativebehaviorisinducedonlywhenasu˚cientamountofsignalhasaccumulated(left).B.SpatialStructure.Whencellsareabletoassortwithkininspace,particularlyinthecaseofbio˝lmformation(bottomleft),dispersalofcellsanddi˙usionofpublicgoodsarelimitedandpromotethemaintenanceofcooperativebehavior.C.Policing.Thismechanismmayactthroughdirectedharm(left)orrestraintofbene˝ts(right).D.MetabolicConstraint.Producersofacooperativebehaviorsuchasapublicgoodalsoproduceanindividuallyretainedprivategoodthatisbene˝cialorrequiredforsurvivalandgrowthinthefocalenvironment.E.MetabolicPrudence.Cellsdetectnutrientsandothercuesintheirenvironmenttodeterminewhetheritiscoste˙ectivetocooperate.AsQSitselfdoesnotappearsu˚cienttocompletelywardo˙cheaters,atleastinthesesystems,additionalmechanismsforcheaterpreventionarerequired.Onesuchmechanism,referredtoasmetabolicconstraint,foundthattheutilizationofadenosinewasalsopositivelycontrolledbyQS(Dandekaretal.,2012)(Fig.1D).Unlikeproteasesecretion,whichisapublicgood,adenosineutilizationprimarilybene˝tstheproducingindividual,making7thisfunctionasagood.TheadditionofadenosinetothemediuminhibitedtheevolutionofQSmutants,asthesewerenotabletoe˙ectivelyusethisresource.Therefore,thecooperatorswereabletoaccessabene˝tthatwasunavailabletothedefectors.ItwasproposedthatmutuallyregulatingpublicandprivategoodsundercontrolofQScouldbeamechanismtostabilizeQS-controlledcooperativetasks.However,ifutilizationofadenosineisaprivategoodthatbene˝tstheproducerinadensity-independentmanner,itisnotclearhowthissystemwouldhaveevolved.Analternatehypothesisisthatinitsnaturalenvironmentadenosineisprimarilypresentinhigh-celldensitysituations.QScouldfunctionasanenvironmentalcuetoprimeP.aeruginosatoutilizelikelynutrientsourcesthatmaybeencounteredatdi˙erentcelldensities.WhileQScanexplainvaryingbehavioraldi˙erenceswithdensity,theunderlyingcooperativebehaviorsarestillpromotedwhencellshaveuni˝edinterests,suchasrelatedness.Becauseofautoinducerspeci˝city,QSprovidesnotjustevidenceofgeneraldensitybutalsothedegreeofrelatednessofthesurroundingcommunity.Itisthereforeourviewthatmetabolicconstraintsystemswillonlyevolvewhenutilizationoftheprivategoodinadensity-dependentmannerisfavored.Onabroadernote,anyconditioninwhichco-regulationofthepublicandprivategoodismostbene˝cialcouldexhibitmetabolicconstraint.Anothermechanismtolimitcheaterinvasionistoonlyproducepublicgoodswhentheircostisminimal,anideatermedmetabolicprudence(Fig.1E).P.aeruginosaalsoproducescarbon-richmoleculescalledrhamnolipidsthatallowlargenumbersoftheseorganismstoovercertainsurfaces,suchassoftagarplates.Rhamnolipidsareapublicgoodthatcanbeexploitedbynon-producerstoswarm.Xavierandcolleaguesnoticed(Xavieretal.,2011),however,thatanon-producerdidnotexhibithigher˝tnessthanarhamnolipidproducerduringaswarmassayofchimericpopulations,eventhoughasigni˝cantportion8ofcarbonwasbeingdirectedtowardssynthesisofthispublicgood.TheauthorsdeducedthatP.aeruginosaonlyproducedrhamnolipidswhenexperiencinganexcessofcarboninrelationtonitrogenlevels.Thus,thecostofproductionforthispublicgoodwasminimizedanddidnotleadtoreduced˝tnessversusthenon-producer.Interestingly,rhamnolipidproductionisalsoregulatedbyQS,indicatingthatproductiononlyoccursathighdensityand/orrelatednessaswell(Lati˝etal.,1995).Metabolicprudenceisthusafacultativecooperationmechanism,whichillustratesthatbacteriaintegratetherelativecostassociatedwithcooperativetraits.Thoughithasnotyetbeenwidelydemonstrated,itislikelytooccurforadditionalmicrobialcooperativebehav-iors.Itiscommonforcomplextraitstobecontrolledbymultipleregulatoryinputs.Forexample,thecataboliterepressorproteinCRPwhichrespondstothepresenceofphospho-transfersugars,atleastinEscherichiacoli,regulates300genes,whichis7%ofitsgenome(Zhengetal.,2004).ItisouropinionthatregulatoryconnectionsbetweenQS(orothersignalingsystems),centralmetabolism,andthecontrolofcooperationwillbecommon,and˝ndingothersystemsthatdemonstratemetabolicprudencewillbeanexcitingnewavenueofresearch(SotoandNishiguchi,2014).1.3SpatialstructureandassortmentSpatialstructuringofrelatedcooperatorsisakeymechanismbywhichcooperationislikelyevolvedandmaintained(Gri˚netal.,2004,LionandBaalen,2008,Westetal.,2007)(Fig.1B).Onecriticalexampleofcellsactivelystructuringthemselvesinanenviron-mentthatisproposedtoencouragecooperationistheproductionofbio˝lms(Kreft,2004).Bio˝lmsaremulticellularcommunitiesofbacteriaencasedinanextracellularmatrix.A9costlyandpotentiallycheatablebehavioritself(Popatetal.,2012),bio˝lmformationpro-videsaframeworkforcellstosituatethemselvesinspaceanddirectcooperativebene˝tspreferentiallytowardsclonalo˙springandotherrelatedkin.Bio˝lmsalsorestrictdi˙usionsothatpublicgoods,suchasextracellularenzymes,remainneartheproducingcellratherthanbeingdispersedby˛oworotherforces(Drescheretal.,2014,Persatetal.,2015).ThiswasrecentlydemonstratedasV.choleraebio˝lmsattachedtochitinsurfacesretainsu˚cientamountsoftheextracellularenzymeschitinasestometabolizethisnutrient(Drescheretal.,2014).Cellsthatdonotformbio˝lmslosehigherportionsofthesepublicgoodsduetoincreasedlossviadi˙usiveandadvectiveforces,whichlikelyre˛ectsconditionsencounteredinnaturalenvironments(Drescheretal.,2014,Persatetal.,2015).Thismeansthatpublicgoodsbene˝tsremaindistributedoveranarrower,morelocalrangeinspacethatfavorstheirdiversiontowardneighboringkincells.Additionally,bio˝lmformersmayevenbeabletoexcludenon-producersfromcolonizednutrientsourcesurfaces(Schluteretal.,2015).However,thisstructuringalsocomeswiththepotentialformorecompetitionbetweenkin,especiallyifcellsdon'talsopossessthecapacitytodisperseandcolonizenewpatchesintheirhabitat.Forexample,experimentallyevolvedlineagesthatproducemorebio˝lmthroughanevolvedwrinklyspreaderlifestyleareabletobindtightlytoneighboringcellsduetoenhancedproductionofextracellularmatrixmaterials,butthistypicallycomeswithatradeo˙forgrowthpotential(Spiersetal.,2002,Spiers,2007).Thetradeo˙maybeinpartrestrainedbythereducedabilityofcellstodispersefromacluster.Asimilarexampleofatradeo˙betweencolonizationanddispersalisseeninV.cholerae,wherebio˝lmproducerscompeteandgrowbetteronasurfacebutarelesse˙ectiveatdispersingtonewlocationsintheirhabitat(NadellandBassler,2011),andnaturalpopulationsofVibriocyclitrophicusdemonstratedi˙erentialspecializationforcolonizationontoanddispersalfromparticles,10signifyingthatthisphenomenoncouldbemorewidespread(Yawataetal.,2014).Thissuggeststhatmultipleselectivepressuresarenaturallyactinguponmicrobesthatcaneitherreinforceoractagainstothercooperativebehaviors.Intheexamplesdescribedthusfar,noassumptionshavebeenmadeabouttheabilityofcellstorecognizethepresenceoridentityofneighboringcells,andthisisnotalwaystheoreticallynecessaryforspatialstructuretoenablecooperation(Oliveiraetal.,2014,VanBaalenandJansen,2006).However,itmaybeimportanttobeabletorecognizeneighborsandrestraincompetitionifsurroundedprimarilybyrelatives.Therearemanyexamplesofcellsbeingcapableofe˙ectivelydistinguishingbetweenselfandnon-selfandadjustingbehavioraccordinglyinwaysthatimpactgrowthoutcomes(Mehdiabadietal.,2006,Renduelesetal.,2015,Strassmannetal.,2011).QSisonesuchsysteminbacteria,butothercontact-dependentrecognitionsystemssuchasthecontact-dependentgrowthin-hibition(CDI)systemofBurkholderia,typeVIsecretionsystems,and˛occulationinyeasthavebeendescribed(Andersonetal.,2014,LeRouxetal.,2015,Smukallaetal.,2008).Thisabilitytocorrectlydecipheramongstneighborsandthecompositionofthesurroundingcommunitycouldgreatlyencouragethesuccessofcooperation,particularlyifproductionofcooperativebehaviorsispredicateduponsensingmembersofacell'sowngenotype.Itisworthnotingthatproducercellsatlowfrequencycaninsomecasespreferentiallygainthebene˝tofpublicgoodsproductioncomparedwithnon-producers,evenintheabsenceofhigherorderedstructure(Goreetal.,2009).Thiscreatesasnowdriftscenariowherebytheraretypehasanadvantage.Thissituationis,however,highlydependentontheparametersofthespeci˝ccooperativebehavior,butitmaycontributetomaintenanceofcooperativetasks.111.4PolicingPolicingandrelatedformsofpunishmentareproposedasanothermechanismtostabilizecooperationinthefaceofpotentialcheaters(Fig.1C).Inthisscenario,anaggressiveac-tionthatnegativelyimpacts˝tnesstargetscheatersrelativetocooperators(TravisanoandVelicer,2004).Policinghasbeencommonlyobservedamongmanyeukaryoticorganisms(Clutton-Brocketal.,1995),includinginsectworkers,birds,andsocialprimates,buttheprevalenceanddiversityofmolecularmechanismsunderlyingbacterialpolicingarenotwellcharacterized.Punishmentcouldbeenactedbyeitherrestrainingbene˝tsdirectedtoanon-contributingpartnerorbydirectharm.Inthesecondcase,thisbehaviormaybecostlytotheenactingindividualbutstillstabilizeothercooperativebehaviorsforcooperatingkininthepopulationaslongasthepunishmentandresultingcooperationarepositivelycorrelated(EldakarandWilson,2008,Hamilton,1970).Oneexampleofpolicingistheenforcementofsanctions,seenintheinteractionoftheroot-noduleformingbacteriaRhizobiawithitshostplant.Symbiontsthatdonotsu˚cientlycontribute˝xednitrogentotheirassociatedhostreceivealimited˛owofoxygenandnu-trientsinreturn(Kiersetal.,2003,Westetal.,2002).Limitingbene˝tsorimposingcoststolessornon-cooperativepartnersinthismannershouldfavormorecooperativepartnersinmutualisms(Westetal.,2002,BullandRice,1991).Insomesystems,suchassquid-Vibriosymbioses,thehostisableto˝lterandselectivelyfavorsuitablepartnersinsuchamanner,andhost-enforcedbottlenecksarealsolikelytoplayastrongroleinmaintaining˝delityintheinteraction(Kochetal.,2014,LeeandRuby,1994).Bottlenecksmaymoregenerallyacttostabilizecooperativebehavior,regardlessofhostassociation(Brockhurst,2007,Chuangetal.,2009,Cremeretal.,2012).Inthisway,restrainingabene˝tinthe12faceofnon-reciprocatingpartnerscanhavethee˙ectofmaintainingtheinteraction.Thesesanctionsneednotberestrictedtointer-speciesinteractionsandcouldbeimaginedtooccur,forinstance,inpopulationswherecellsareexchangingmetabolites(Estrelaetal.,2012).Recently,QSinP.aeruginosahasalsobeenshowntoinducepolicingthattargetsQSdefectorsbyregulatingcyanideproductionandresistance(Wangetal.,2015).Asdescribedabove,QSinductionofextracellularproteasesisnecessaryformaximumgrowthinamin-imalmediaenvironmentwithcaseinasthecarbonsource.Inthiscase,certainclassesofQSdefectorsareunabletoproduceco-regulatedcompoundsthatcounteractthee˙ectsofcooperator-producedcyanideandarethusunabletocompletelyinvadeaQS-pro˝cientpop-ulationcooperatingviaextracellularenzymeproductionthatcouldotherwisebeexploitedbythepotentialcheats.Inalllikelihood,thesetypesofpolicingmechanismsmaybedi˚culttomaintainandunlikelytobecommonformaintainingbacterialcooperation(Westetal.,2007).Becausepolicingbehaviorsarecostlytoperformandmaytargetrelatedkin,theymayconveya˝tnessdisadvantageundermanyconditions.However,thise˙ectmaybesomewhatalleviatedifsuchtraitsareonlyexpressedconditionally,asshownintheQS-regulatedexample.Forcooperativepartners,fromeitherthesameordi˙erentspecies,sanctionsmayarisemorenaturally.Duetonegativee˙ectsontheproductivitycausedbyapoorpartner,reciprocalsanctioninge˙ectswillmorenaturallyemerge,asfewerpartnerswillbepresenttorepaythefavor(Oliveiraetal.,2014).131.5DivisionofLaborinBacteriaApenultimateformofcooperationthatisarequirementforthedevelopmentofhigherorderedmulticellularityisoflabDivisionoflaborcanbede˝nedascooperatingindividualsthatperformdiscretetasksthatarethemselvescostlytotheindividuals,butthesumtotalofthistaskdistributionisbene˝cialtothelargercommunity.Divisionoflaborisclearlyevidentincomplexmulticellulareukaryotes.Aheartcellhasdi˙erentiatedtoperformverydi˙erenttasksthanalivercell.Intheseorganisms,developmentandter-minaldi˙erentiationarekeystodrivingandmaintainingphenotypicheterogeneity.Divisionoflaborhasalsoclearlybeenobservedinbacteria.AclassicexampleissporeformationbyMyxobacteria(Shimkets,1990).Uponstarvation,thispredatorybacteriumaggregatesintomulticellularmounds,whichultimatelyformstructurescalledfruitingbodiescoatedwithenvironmentallyresistantMyxobacteriasporesthatriseabovethelocalsurface.Thesestructuresarethoughttoaidindispersalofthesporestonewenvironments.Divisionoflaborisalsoproposedtobeacommonfeatureofbio˝lms,althoughthisisacontroversialideathatasofyethaslittleexperimentalsupport.Indeed,˝vespeci˝cBacillussubtiliscelltypescanbeobservedinamonospeciesbio˝lm(vanGesteletal.,2015).Thesesubtypeslocalizetodistinctregionsofthebio˝lm,buttheadaptivefunctionofthesecelltypesisnotclear,asalockedmatrix-producingcelltypeissu˚cienttoproducearobustbio˝lminthelab.However,itislikelythatahomogenousbio˝lmwouldbemaladaptiveinthenaturallifecycleofB.subtilis,asthelaboratoryenvironmentismissingkeyaspectsofthenaturalworld.Italsoseemsunlikelythatbio˝lmswillremainasgeneticallyhomogenousasliquidculturesduetolimiteddispersalleadingtorestrictionofcompetitiontolocalscales.Understandingtheevolutionaryfactorsthatdrivetheemergenceofphenotypic14heterogeneitymustrelyonabetterunderstandingoftheecologyofthesebacteriaandideallyguideexperimentsontheseorganismsinmorenaturallyrelevantsystems.Moreover,likeallformsofcooperation,identifyingthestrategiesandmechanismsthatmaintaintheseinteractionsinthefaceofdefectorswillbeanintriguingareaofresearch.Theaboveexamplesrepresentdivisionoflaborinmonospeciessystems,butinterspeciesdivisionoflaborcananddoesoccuraswell.Incomplexmicrobialcommunities,wepredictthatdivisionoflaborwillbemostevidentinthecooperationofindividualsthroughmetabolicexchanges(Taso˙etal.,2015,Werneretal.,2014).Indeed,thishasbeenobservedinclinicalcystic˝brosisisolatesofStaphylococcusaureus(Hammeretal.,2014).Thisphenomenonhasbeenshowntobepossibleinsyntheticaswellasnaturalcommunities(Oliveiraetal.,2014,Morrisetal.,2012,2014,Pfei˙erandBonhoe˙er,2004).Inmixedcommunities,membersthatarenotdirectlyinvolvedintheexchangemaystillimpactitprovidingadegreeofseparationorassortmentofindividualpartners,aprocesslabeledsocialinsulation(Oliveiraetal.,2014,Mitrietal.,2011).Withtheincreasingimportanceofthehumanandanimalmicrobiomeinhealthanddisease,understandingcooperativedivisionoflaborinteractionsinthesecommunities,andpotentiallywiththehost,willbeanincreasinglyimportantareainmicrobialevolution.1.6ConclusionAswehaveseen,microbialcooperationoccursindiversemanners,andthemechanismsguidingitsmaintenancearelikewisediverse.Thesemechanismsallfundamentallyacttomediatethe˝tnesscostsimposedbyexpressingcooperativebehaviororbyalteringthewaythatbene˝tsareadministered.Fitnessgainsmaybedirectedtotheactingparty(direct),15torelatedkin(indirect),orboth,andthusacttoincreasetheorganism'soverallinclusive˝tness.Alsoevidentthroughoutthisreview,thesemechanismsmayandoftendoworkinconcertwithoneanother.Asresearchersdiscovernovelexamplesofmicrobialcooperation,moremechanismsthatdirecttheirmaintenancewilllikelycometolight.Weareparticularlyoptimisticthatmanymechanismsofcooperationutilizingoptionalparticipationguidedbycommunicationwillbediscoveredinexperimentalandnaturalcommunities.Thesurfacehasonlybeenscratchedinthisareaofresearch,anditwillbeexcitingtoseewhatnewmechanismsareuncovered,howtheymaypotentiallybeusedforindustrialandmedicalapplications,aswellashowtheymayinformwhatweknowaboutbiologyatboththemicroandmacroscale.16Chapter2Bacterialquorumsensingstabilizescooperationbyoptimizinggrowthstrategies17AbstractandImportanceCommunicationhasbeensuggestedasamechanismtostabilizecooperation.Inbacteria,chemicalcommunicationtermedquorumsensing(QS)hasbeenhypothesizedtoful˝llthisrole,andextracellularpublicgoodsareofteninducedbyQSathighcelldensity.HereweshowwiththebacteriumVibrioharveyithatQSprovidesstrongresistanceagainstinva-sionofaQS-defectorstrainbymaximizinggrowthrateatlowcelldensitieswhileachievingmaximumproductivitythroughproteaseupregulationathighcelldensity.Alternatively,QSmutantsthatactasdefectorsorunconditionalcooperatorsmaximizeeithergrowthrateoryield,respectively,andarethusless˝tthanthewildtypeQSstrain.Our˝ndingsprovideexperimentalevidencethatregulationmediatedbymicrobialcommunicationcanoptimizegrowthstrategiesandstabilizecooperativephenotypesbypreventingdefectorinvasion,eveninwell-mixedconditions.Thise˙ectisduetoacombinationofresponsivenesstoenviron-mentalconditionsprovidedbyQS,loweringofcompetitivecostswhenQSisnotinduced,andpleiotropicconstraintsimposedondefectorsthatdonotperformQS.Cooperationisafundamentalproblemforevolutionarybiologytoexplain.Conditionalparticipationthroughphenotypicplasticitydrivenbycommunicationisapotentialsolutiontothisdilemma.Thus,amongbacteria,QShasbeenproposedtobeaproximatestabilizingmechanismforcooperativebehaviors.Here,weempiricallydemonstratethatQSinV.har-veyipreventscheatingandsubsequentinvasionbynon-producingdefectorsbymaximizinggrowthrateatlowcelldensityandgrowthyieldathighcelldensity,whereasanuncon-ditionalcooperatorisrapidlydriventoextinctionbydefectors.Our˝ndingsprovidethe˝rstexperimentalevidencethataddingQSregulationpreventstheinvasionofcooperativepopulationsbyQSdefectorseveninunstructuredconditionsandstronglysupporttherole18ofcommunicationinbacteriaasamechanismthatstabilizescooperativetraits.192.1IntroductionCooperativebehaviorisawidespreadphenomenonthatpervadesalllevelsofbiologicalorganizationandhashelpedtocatalyzeallmajortransitionsinthehistoryoflife(SzathmáryandSmith,1997).Extensivecooperationisalsoprevalentinmicrobesandplaysfundamentalrolesinmanybacterialprocessesincludingbio˝lmformation,virulence,bioenergy,host-microbeinteractions,aswellasformingstablecommunitiesthatmaintainessentialecosystemfunctions(Crespi,2001,Velicer,2003,Diardetal.,2013,Percivaletal.,2010,Popatetal.,2012,Westetal.,2002).Oneclassofmicrobialcooperativebehaviorsistheproductionofpublicgoods(PG):productsthatprovidebene˝tstobothproducersanddefectorswithinacommunity.InthecontextofPG,individualsthatcontributebyproducingthegoodarede˝nedascooperators,whilenon-producersarede˝nedasdefectors.Importantly,defectorsarenotalwaysinherentlycheaters,butarecapableofcheatingiftheyreap˝tnessadvantagesbyexploitingsocialbehaviors(ZhangandRainey,2013,Jonesetal.,2015,Ghouletal.,2014).However,conditionalparticipationthroughphenotypicplasticitydrivenbycommunicationcouldprovideapotentialsolutiontothisdilemma(Kümmerlietal.,2009).Quorumsensing(QS)isaubiquitousformofchemicalcommunicationinbacteriathatallowscellstosensechangesinlocalcelldensitytogloballyaltergeneexpression(WatersandBassler,2005,Westetal.,2012).ManyQSregulonsappeartobeenrichedforsecretedPGproducts(Popatetal.,2015a),andithasbeenhypothesizedthatQScouldactasamechanismtostabilizecooperation(TravisanoandVelicer,2004).Additionally,QShasbeenshowntointeractwithothermechanismsthatfunctiontostabilizecooperativebehaviors,suchasmetabolicconstraint,metabolicprudence,orpolicing(BrugerandWaters,2015,Dandekaretal.,2012,Xavieretal.,2011,Wangetal.,2015).20However,thepostulatedroleofQStoinnatelystabilizecooperativetraitshasnotbeenwidelydemonstrated.MostempiricalstudiesoflaboratoryandnaturalpopulationssuggestthatQSsystemscanbeinvaded,andinsomecasesdestabilized,includinginvasionofbothPseudomonasandVibriospecies,byQS-defectors(Sandozetal.,2007,Diggleetal.,2007,Katzianeretal.,2015).Infact,therangeofconditionsunderwhichcooperationisfavoredmaybequitelimited(Oliveiraetal.,2014).ThisraisesthequestionofwhyQSfrequentlyregulatestheexpressionofPGandwhetheritisinfactadvantageousformaintainingcoop-erativetraits.Here,weaddressedthisquestionbyexaminingaPGcommonlyregulatedbyQS:extra-cellularproteaseproductioninthebioluminescentmarinebacteriumVibrioharveyi.ThisQSsystemisidealtostudytheroleofcommunicationinstabilizingcooperationbecausebothdefectorsandunconditionalcooperators(UC)oftheQSpathwaycanbeeasilygener-ated.Byassessingthecompetitiveoutcomesofcooperatorsanddefectorsinanenvironmentwhere˝tnessisdependentonQS-regulatedproteaseproduction,wedemonstratethatawildtype(WT)strainwithafunctionalQSsystempreventsdefectorinvasionwhereasanUCmutantstrainisdriventoextinction.QSpreventscheatingbyregulatingaswitchingrowthstrategiesfrommaximizinggrowthrateatlowcelldensitybyrestrainingfromproteasepro-ductiontoincreasinggrowthyieldathighcelldensitywhenavailableresourcesbecomemorescarce.WhereastheWTstraincanuseQStotakeadvantageofthesedi˙erentstrategiesattheappropriatedensities,thedefectorandUCmaximizeonlyrateoryield,respectively,witharesultingtradeo˙intheothercapacity.OurresultssuggestthatprudentregulationofmetabolismbyQSisamechanismbywhichthissystempromotesthemaintenanceofcooperation.21Table2.1:StrainsofV.harveyiusedinthisstudy.2.2MaterialsandMethods2.2.1BacterialStrainsandMediaandGrowthConditionsBacterialstrainsusedinthisstudyincludeVibrioharveyistrainATCCBAA-1116(Bassleretal.,1997),recentlyreclassi˝edasV.campbellii(Linetal.,2010),andderivatives(2.1).BacteriaweregrowninLB(Accumedia)brothorinM9salts(Sigma-AldrichorBec-ton,DickinsonandCompany)supplementedwithsodiumchloride(MacronFineChemicals)toa˝nalconcentrationof2%(w/v)and0.5%(w/v)ofsodiumcaseinate(Sigma-Aldrich),casaminoacids,ortryptone(Becton,DickinsonandCompany)assolecarbonsources.Bac-teriawereroutinelygrownina30Celsiusshakerat250rpmin16mmborosilicateglasstubescontaining3mLmedia.ExperimentalcultureswereinitiallygrowninliquidLBcul-turesandsubsequentlyacclimatedbypassaginginM9liquidmedia.Cellswerecollectedbycentrifugation(10,000xg)andwashedintheequivalentgrowthmediumpriortocompetitionexperiments.SupernatantswereobtainedfromculturesgrowninM9-caseinmediafor24hours.Cellswereremovedby˝rstpelletingat10,000xgandthen˝lteredwith0.22m˝lters(Millipore).Supernatantswereaddedbackatthepercentageof˝nalculturevolumeindicated.222.2.2CompetitionFitnessAssays2.2.2.1CompetitionDesignCompetitorswereassessedfortheabilitytoinvadewhenrarebyinitiatingpopulationswitharangeofrelativefrequenciesofthedefectorandrelevantcooperatorstrainsforsin-glegrowthphasecompetitions.Populationsweregrown24hours,platedonLBagar,andpopulationcompositionswereassessed.Forserialcompetitionassays,populationswereini-tiatedbymixingstrainsataratioof99:1.Populationsweregrownfor24hours,platedonLBagartoassesspopulationcompositionandproductivity,andasubsetwasdiluted1/1000tonewM9liquidmediafollowedby24hoursofgrowth(10generations).Duringcompetitions,bioluminescencephenotypesweredeterminedinthedarkusinganAlphaIm-agerHPimagingsystem(ProteinSimple).Ancestralcooperatorstrainsuniformlyproducebrightbioluminescentcolonies,whilealldefectorstrainsinvestigatedwerenon-luminescentinappearance.2.2.2.2FitnessCalculationsWheredetermined,defectorrelative˝tness(w)wascalculatedastheratioofMalthusianparameters(m)forstrainsinpairwisecompetitions,whichalsoequatestotheratioofcom-petitordoublingsovertheexperimentaltimeintervalof24hours(Lenskietal.,1991).TheMalthusianparameteriscalculatedasmi=ln(x1=x0)23forcompetitori,wherex0andx1arethedensitiesofthatcompetitoratthestart(0)orend(1)oftheexperimentalperiod,andw=m1=m2;whencomparingcompetitors1and2.2.2.3ProteaseAssaysExtracellularproteaseactivitywasassessedusingaprotease˛uorescenceassay(Sigma-Aldrich)with˝ltered(0.45mpore˝lters,Ambion)supernatantsfromculturesgrowninM9-caseinmedia.Conjugatedcaseinmoleculesreleasethe˛uorophore˛uoresceinisothiocyanate(FITC)whencleaved,allowingsensitiveassessmentofproteaseactivity.FluorescencewasmeasuredinanEnVisionMultilabelPlateReader(PerkinElmer)withtheexcitationwave-lengthat485nmandtheemissionwavelengthat535nm.Measurementswerenormalizedtothedilutionusedforsupernatantsamplesandcelldensityoftheculture,determinedbyplatecounting.2.2.4StatisticalAnalysesStatisticalanalyseswereconductedusingR3.2.2.Growthofdi˙erentstrainsinvaryingM9mediasourceswascomparedusingANOVA[PopulationDensityStrainType(cooperatorordefector)*CarbonSource]withTukey'sHonestlySigni˝cantDi˙erence(HSD)post-hoctesttoaccountformultiplecomparisons(Fig.2.1).Proteaseproductionwascomparedacross3strainsat24hoursofgrowthinM9-caseinmediausingANOVA[Per-capitaproteaseactivityCellDensity+StrainType](Fig.2.2,toppanel).Thee˙ectofsupplementing24di˙erentsupernatantsonluxRgrowthinM9-caseinmediawascomparedusingANOVA[Growth(measuredbyOD600)Supernatantsource*Supernatantconcentration]withTukey'sHSDpost-hoctest.Alinearmodelwasconstructedtoanalyzee˙ectsofrelative˝tnessvs.frequencyestimates[Competitorrelative˝tnessCompetitionpairing*Startingcompetitorfrequency](Fig.2.4,toppanel).Inexperimentsexaminingthedynamicsofgrowth(Figs.4-6),datapointsareboundedwith95%con˝denceintervalsforerrorbarstoallowstatisticalcomparison.2.3Results2.3.1QSisrequiredformaximumgrowthofVibrioharveyiutilizingcaseinToevaluatethestabilityofQSandthecooperativebehaviorsitregulatesagainstdefectorinvasioninV.harveyi,we˝rstestablishedconditionsinwhich˝tnesswasdependentuponQSactivationofPGproduction.WefoundthatgrowthinM9-caseinmediawashighlydependentontheabilitytoinduceQSathighcelldensity.Inthisenvironment,caseinbreakdownbyextracellularproteaseproductionisrequiredtoreachamaximumgrowthyield(Natrahetal.,2011,Wilderetal.,2011,Popatetal.,2015a).InV.harveyi,theaccumulationofautoinducerincreasesexpressionofthemastertranscriptionfactorLuxR,switchinggeneexpressionfromthelow-tohigh-cell-densitystate.Weobservedthatonlycooperatorstrains,includingtheWTconditionalcooperator,whichinducesluxRinresponsetoautoinducerascelldensityincreases,andtheUCluxOUmutant,whichconstitutivelyexpressesluxRandsubsequentlythehigh-cell-densityQSregulon,reachmaximumdensities25inthismediaafter24hoursofgrowth(Fig.2.1).Meanwhile,defectors,includingtheluxRstrain(cannotproducetheLuxRmasterregulator),grewtoasigni˝cantlylowerdensityofaround4%oftheWTinM9-caseinmedia(Fig.2.1,p<0.001).ThisoutcomewasobservedforalldefectorstrainstestedthatcannotinduceluxRexpressionincludingtheluxOD47Emutantandanautoinducer'signalnegative'productionmutant('SN',Fig.2.1).TheproductionofbioluminescencedidnotimpactgrowthinthisconditionastheluxABdarkmutantgrewidenticallytotheWTstrain.Incontrast,usingpredigestedcarbonsourcessuchascasaminoacidsortryptonesigni˝cantlydiminishesthedi˙erencesingrowthbetweencooperatoranddefectorstrains(Fig.2.1).2.3.2FunctionalandmutantQSstrainsdi˙erindensity-dependentproteaseactivityTocon˝rmthatthesegrowthresultswereduetoextracellularproteaseactivity,wequan-ti˝edandmonitoredactivityduringgrowthofculturesstartedatlowcelldensity(Fig.2.2,toppanel).Asexpected,theluxOUstraindemonstratedhighproteaseactivityatallcelldensitiesasthisstrainislockedinthehigh-cell-densityQSstate.TheWTstrainexhibitedQSregulationofproteaseproduction,wherecellsinitiallyexhibitedhighproteaseactivityimmediatelyuponback-dilutionfromhighcelldensity.However,proteaseactivitydecreaseduntilaquorumwasreached,butitultimatelymatchedtheproteaseactivityobservedintheluxOUstrainathighcelldensity(Fig.2.2,toppanel).TheluxRstrainexhibitedloworundetectablelevelsofproteaseactivityatalldensitiesexamined.NotethattheluxRmutantgrowspoorlyinthismediumanddoesnotreachthecelldensityofthecooperatorstrains(Fig.2.1).OurdatashowingthatWTandluxOUstrainsofV.harveyihavehigher26Figure2.1:V.harveyirequiresQSformaximalgrowthinM9-caseinmedia.SixstrainsofV.harveyigrowninM9mediawithvaryingsolecarbonsourcesweregrown24hoursandenumeratedthroughviablecellcounts(CFU/ml).Top.Barsdisplaytheproductivityofagivengenotypeattheculminationofasinglegrowthcycle(24hours)offourseparatebiologicalreplicates,anderrorbarsdenotethe95%con˝denceintervalsforthemeanestimates.Bottom.ImagesofcooperatoranddefectorgenotypesinM9-caseinmediaafter24hoursofgrowth.Picturesarealigneddirectlybeneaththeircorrespondinglabelinthetoppanel.'WT'=WildTypestrain,'SN'=SignalNegativestrain,'D47E=luxOD47Estrain.Di˙erencesthatwerestatisticallysigni˝cantfromWTinthegivenmediatypearereported.*p<0.05,***p<0.001.27levelsofproteaseactivitythanQS-defectorstrainsagreewithearlierworkinthisandotherQSsystems(Natrahetal.,2011,Wilderetal.,2011,Popatetal.,2015a).ThepoorgrowthofQS-defectorluxRstrainoncaseinisnotduetoaninabilitytousepeptidesliberatedfromcaseinasthegrowthofluxRinthisenvironmentcanberescuedbytheadditionofcooperatorsupernatantsfromtheWTorluxOUstrains,butgrowthisnotimprovedbyitsownsupernatant(Fig.2.2,bottompanel,p<0.001).ItisalsoworthnotingthatsupernatantsfromallexamineddefectorstrainsinducedbioluminescenceexpressionofanautoinducermutantofV.harveyinearly100-1,000-foldlessthancooperatorsupernatants,alevellowerthanpredictedbygrowthdi˙erencesinM9-caseinalone,suggestingthatthesedefectorsproducedlowlevelsofautoinducerandwouldbeunlikelytocheatbyinducingcooperatorstoproducePG(Fig.2.3).2.3.3AfunctionalQScircuitpreventsdefectorinvasionAlthoughhigherlevelsofproteaseproductionbytheWTandluxOUstrainsleadtoincreasedgrowthinM9-caseinmedia,thisbehaviorispotentiallysusceptibletocheatingbydefectorsthatcouldbene˝tfromthenutrientsliberatedbytheseextracellularproteases.Totestwhethercooperatorsthatproducehigherlevelsofproteasecouldbecheated,wecompetedWTandluxOUstrainsagainsttheluxRdefector.Thecompetitionexperi-mentswereperformedinM9-caseinmedia,andfrequenciesofthecompetingstrainswerevaried.Thepopulationswerestartedatdensitiessigni˝cantlylowerthanthequorum.Thecompetitionsconsistedofonegrowthperiod,andtheQSphenotypeofisolatedcolonieswasdeterminedbycolonybioluminescencephenotypestoquantifydefectorsandcooperators.Astrikingdi˙erenceindefectorinvasionwasobserveddependingonwhichcooperatorstrainwasused.WhileluxRincreasedinfrequencyagainstluxOUatallstartingfre-28Figure2.2:ExtracellularProteaseactivityisregulatedbyQSinV.harveyi.Toppanel.ProteaselevelsbetweencooperatoranddefectorgenotypesgrowninM9-caseinmediaareplottedversusthegrowthasmeasuredbyCFU/mL(n=3).Errorbarsre˛ectstandarderrorofthemean.Bottompanel.luxRculturesweresupplementedwithsupernatantsattheindicatedconcentrationfromthestrainsindicatedonthex-axis(n=4).Errorbarsre˛ect95%con˝denceintervals.***P<0.001.N.S.:notsigni˝cant.29Figure2.3:V.harveyirequiresQSformaximalsignalproductioninM9-casein.SupernatantsfromsixstrainsofV.harveyigrowninM9-caseinmediaindicatedonthex-axisweresupplementedtogrowingculturesofthesignalnegative(SN)strain(n=9).Barsdisplaythebioluminescenceinducedbythesupernatantofagivenstrain,normalizedtoculturegrowthdensitymeasuredbyOD600,asapercentageoftheinductionprovidedbywildtypesupernatants.Colorsdi˙erentiatestrainsconsideredtobecooperators(darkgray)anddefectors(lightgray).'Media'treatmentdenotesnosupernatantaddedtotheculture.Errorbarsdenotethe95%con˝denceintervalsforthemeanestimates.'WT'=WildTypestrain,'SN'=SignalNegativestrain,'D47E'=luxOD47Estrain.Di˙erencesthatwerestatisticallysigni˝cantfromtheMediatreatmentarereported.***p<0.001.quenciestested(Fig.2.4,toppanel,redline),itwasunabletoinvadetheWTstrainatanyexaminedfrequency(Fig.2.4,toppanel,blackline,Fig.2.6,toppanel).Asaresult,theWTstrainhadequivalent˝tnesstothedefectorstrainacrossallstartingdefectorfrequenciestested,whereastheluxOUstrainhadamuchlowerrelative˝tnessthanthedefector.WesimilarlycompetedtheWTstrainagainstluxOUatmultiplefrequencies.LiketheluxRdefector,WTwasabletoinvadetheluxOUstrainleadingtohigher˝tnessatallfrequen-ciesexamined(Fig.2.4,toppanel).AlthoughWT-mixedpopulationswerenotinvadedby30luxRatanyfrequencytested(i.e.therelative˝tnessofthestrainsdidnotvary),thesepopulationsexhibitednearlyidenticalfrequency-dependencedecreaseinpopulationyieldastheluxOU-mixedpopulations(Fig.2.6,bottompanel),indicatingthatabsolute˝tnessdecreasedwithincreasingdefectorfrequency.Furthermore,thisphenomenonpersistsoverawiderangeofdilutions(andthusawiderangeofstartingdensities)whereWTperformswellagainstluxR,whileluxOUperformsincreasinglyworseasthestartingdilution(andthusthenumberofgenerationsthatoccurbeforereachingtheenvironment'scarryingcapacity)increases(Fig.2.5).Basedonthesedata,wepredictedthatluxRshoulddriveluxOUtoextinctionoverlongertimescales,butitwouldbeunabletoinvadetheWTstrain.WetestedthispredictioninextendedpairwisecompetitionexperimentsintheM9-caseinenvironmentbetweeneithercooperativegenotype(WTorluxOU)againstthedefector(luxR),performedwithonecompetitorseededatalowstartingfrequencyof1%toevaluateitsabilitytoinvade.Thecompetitionswereextendedoverseveralgrowthcycleswithongoingdailydilution.Fromeitherstartingcondition,defectorsrapidlysweptanddominatedluxOUinmixedpopu-lations(Fig.2.7,toppanel).Thisresultedinanunderutilizationoftheavailablenutrientresourcesandaconsequentofthewheretheproductivityofthepopula-tiondropstolevelsofthedefectorstraingrowninisolation(Fig.2.7,bottompanel,Hardin,1968,Rankinetal.,2007,MacLean,2008).Thelackofstablecoexistenceineitherfre-quencyconditiontested(defectorrarevs.defectorcommon)suggeststhatthesecompetingstrategies(luxOUorluxR)areunlikelytocoexistinthisenvironment.WhentheWTcooperatorwascompetedagainsttheluxRdefectorincasein-limitedM9media,neitherstrainsigni˝cantlyinvadedtheother,asbothgenotypesmaintainedtheirstartingfrequencies(Fig.2.7,toppanel).UnlikecompetitionsagainsttheluxOUUC,the31Figure2.4:Fitnessoutcomesfromsinglegrowthcyclecompetitionsbetweendif-ferentV.harveyigenotypesvarygreatlyinM9-caseinbutnotinM9-tryptoneTop.FitnessoutcomesinM9-casein.Inpairwisecompetitionsbetweendi˙erentV.harveyistrains,therangeofstartingcompetitorfrequencywasalteredandcompetitionexperimentswerecompletedinM9-caseinmediafor24hours.Fitlinesconsistofregression˝t(solidlines)boundedby95%con˝denceintervals(dashedlines)foreachcompetitionpairingac-cordingtoalinearmodel.Foreachpairing,thestrainwhoserelative˝tnessisreportedonthey-axisandstartingpercentageonthex-axisisunderlined.Bottom.FitnessoutcomesinM9-tryptone.Inpairwisecompetitionsbetweendi˙erentV.harveyistrains,therangeofstartingcompetitorfrequencywasalteredandcompetitionexperimentswerecompletedinM9-tryptonemediafor24hours.Regression˝tlinesforthedi˙erentcompetitionpairsaredisplayedforeachcompetitionpairingaccordingtoalinearmodel.Foreachpairing,thestrainwhoserelative˝tnessisreportedisunderlined.32Figure2.5:V.harveyiQSprovidesequivalent˝tnessagainstdefectorsinM9-caseinoverarangeofdensityconditions.Growthresultsfromsinglecycle24hourcompetitionexperimentsinM9-caseinofpopulationsinitiatedatapproximately99%coop-eratorsand1%defectorsatawiderangeofdensitiesarereportedastherelative˝tnessofluxRinrelationtothegivencooperatorstrain.Foraseriesofdilutionsfrominitialcompetitionmixesusedtoachievearangeofdensities,foreachdilutionintheseries,N=4biologicalreplicatesperdensitylevel.Fitlinesconsistofregression˝t(centerlines)boundedby95%con˝denceintervals(dashedlines)foreachcompetitionpairingaccordingtoalinearmodel.resultingyieldoftheWT-mixedpopulationsmirroredtheinputfrequenciesandwasstableoverthistimeperiod(Fig.2.7,bottompanel).Infact,theresultingcelldensityarrivedatinWTversusdefectorcompetitionsseemstorelystronglyupontheinitialWTfrequency,andahighstartingfrequencyofWTcooperatorswasabletopreventatragedyofthecommonsoverthistimescale.33Figure2.6:Growthperformanceofdefectorsandmixedpopulationsincompeti-tioninM9-casein.Top.PerformanceofluxRinmixedpopulationsinM9-caseindependsheavilyonthecompetingcooperatorstrain.Densityofthedefectorstrain(luxR)atthecompletionofasingle24hourcycleofcompetitionisplottedasafunctionofthestartingdefectorfrequencyinthepopulation.Opencircles:luxOUvs.luxR,Filledcircles:WTvs.luxR.Theblacklinedepictsasmoothedregression˝tfortheexpecteddefectordensityiftherewasnoincreaseindefectorfrequency,basedonthe˝naltotalpopulationdensities.Bottom.GrowthperformanceofmixedpopulationsincompetitioninM9-caseindependsondefectorfrequencies.Growthresultsfromsinglecycle24hourcompetitionexperimentsinM9-caseinmediaarereportedastotal˝nalpopulationdensities(bothcompetingstrains)inCFU/mlasafunctionofthestartingfrequencyofthestrainunderlinedinthe˝gurekey.DatainbothpanelsarefromthesameexperimentshowninFigure2.4,toppanel.34Figure2.7:TheluxRdefectorinvadesluxOUbutnotWT.Populationscontainedoneofthetwocooperatorstrains,WTorluxOU,andtheluxRdefectorata99:1mixtureorviceversa.Culturesweregrownfor24hoursinM9-caseinmedia,diluted1000-foldandrepeatedovermultiplegrowthcycles.ThelegendinBreferstobothgraphs.Toppanel.Defectorfrequencyasdeterminedbynon-luminescentcoloniesisplottedversusgenerationsofgrowth.Bottompanel.Populationproductivitymeasuredbyplatecounts(CFU/ml)isplottedversusgenerationsofgrowth.Errorbarsre˛ect95%con˝denceintervalsfor3biologicalreplicatespertreatment.35Figure2.8:TheWTstrainoptimizesgrowthstrategiesatlow-andhigh-cellden-sity.Top.BioluminescenceinductionoftheWTandluxOUinmonocultureoverthecourseofagrowthcycleinM9-caseinmedia.Middle.DefectorfrequencyoverasinglegrowthcycleforluxOUvs.luxR,luxOUvs.WT,andWTvs.luxRwiththede-fectorstrainsunderlinedstartedat1%andcompetedinM9-caseinmedia.Bottom.GrowthoftheWT,luxR,andluxOUinmonocultureoverthecourseofasinglegrowthcycle.Errorbarsrepresent95%con˝denceintervalsonfourbiologicalreplicatespertreatment.362.3.4AfunctionalQSsystempreventsdefectorinvasionbymodulatinggrowthstrategiesatdi˙erentcelldensitiesAsQSisamechanismthatcanaltergeneexpressioninresponsetochangesincelldensity,wesoughttodeterminethedynamicsofQS-inductionandthekineticsofdefectorinvasionintheM9-caseinenvironment.TodeterminetheQSinductionstateatdi˙erentdensitiesintheM9-caseinenvironment,wemonitoredper-capitabioluminescenceasaproxyforactivationofQS.Asexpected,theWTstrainexhibitsastrongdropinbioluminescenceatlowcelldensity,butbeyond6hoursofgrowthaquorumisreachedandbioluminescenceisultimatelyinducedtomaximallevelsmatchingorevenexceedingluxOU(Fig.2.8,toppanel).Alternatively,luxOUmaintainsanearconstantlevelofbioluminescence,indicatingitisactivatingQSinanunconditionalfashion(Fig.2.8,toppanel).Theregulationofbioluminescenceinthesestrainsmirrorsthatseenforproteaseactivity,aswouldbeexpectedfortwophenotypessimilarlyinducedinthehigh-cell-densitystate(Fig.2.2,toppanel).Todetermineatwhatpointduringgrowthdefectorsinvadecooperators,wemeasuredthefrequencyofdefectorsstartedat1%ofthreepairwisecompetitions(luxOU/WT,WT/luxR,andluxOU/luxR,thedefectorstrainsareunderlined)overthecourseofonegrowthcycle.Weobservedthatsigni˝cantinvasionofluxOUbybothWTandluxRbegantooccurasthepopulationreachedquorum,atwhichpointthesestrainsrapidlyin-creasedtohighfrequencies(Fig.2.8,middlepanel).Athighercelldensities,theinvasionofbothofthesestrainsplateauedandluxOUmaintainedaminorfrequencyinthepop-ulation.Thisindicatesthatinvasionprimarilyoccurredatthetransitionfromlowtohighcelldensities.Asseeninearlierresults,theluxRdefectorcouldnotinvadeWTatanydensitiesexamined(Fig.2.8,middlepanel,Fig.2.9).37Figure2.9:GrowthperformanceofluxRdefectorinthepresenceandabsenceofcompetingstrains.GrowthresultsfrommonocultureandcompetitionexperimentsinM9-caseinmediaoverthecourseofasinglegrowthcycle.Forcompetitionmixedpopulations,dataarefromthesameexperimentasFigure2.8,andthenumbersreportedaretheluxRportionofthosepopulations.Formonocultureresults,matchingdensitiesofluxRalonewereinitiatedandtrackedinM9-caseinmedia.Tofurtherunderstandthesedynamics,wemeasuredthegrowthofWT,luxR,andluxOUoverthecourseofonegrowthcycleinmonoculture.BoththeWTandluxRstrainsexhibitedequivalentrapidgrowthatlowcelldensitycomparedwiththeunconditionalluxOUmutant(Fig.2.8,bottompanel).Thisdi˙erencewasespeciallyevidentatthe5-hourtimepoint,matchingthetimeperiodduringwhichluxOUisinvadedandanincreaseindefectorfrequencybecomesreadilyapparentincompetitions(Fig.2.8,middlepanel).AstheWTstraingrowstohighercelldensities,ittransitionstoanewgrowthstrategy,allowinggrowthtoresumeandcellnumberstoincreaseuntilthecarryingcapacityisreached.ThisportionofgrowthmorecloselyresemblestheluxOUmutant,whichexperiencesanincrease38ingrowthrateathighcelldensityandultimatelyreachesthesameyieldastheWTstrain(Fig.2.1).This,incombinationwiththeobservationthatluxRplateausingrowthbeforereachinghighcelldensity,suggeststhatthisphaseofgrowthislargelydependentuponQS-regulatedPG.BecausetheluxRmutantisunabletoalteritsgrowthstrategyliketheWTtoproducePGnecessaryforadditionalgrowth,itobtainsasigni˝cantlylowerpopulationyieldthaneithercooperatorstrain.TheslowgrowthrateoftheluxOUatlowcelldensityisspeci˝ctothisenvironmentasluxOUexhibitsequivalent˝tnesstotheWTandluxRmutantwhengrowninM9minimalmediumsupplementedwithtryptone(Fig.2.4,bottompanel).Furthermore,thehigher˝tnessofluxRcomparedtoluxOUisnotsimplyduetoitshigherindividualgrowthrate,butratherissocialinnatureasluxRgrowstohigherdensitieswhencompetedwithluxOUthanitcanbyitselfinmonoculture,demonstratingcheatingisoccurringintheseconditions(Ross-Gillespieetal.,2009;Fig.2.6,toppanel,Fig.2.9).Conversely,luxRdoesslightlyworseinmixedpopulationswithWTthaninmonocultureinbothM9-caseinandM9-tryptoneenvironments(Fig.2.9,Fig.2.4,bottompanel).2.4DiscussionQShasbeenproposedasamechanismtostabilizecooperation,butonlymodestexper-imentalevidenceexiststosupportthisproposition(Popatetal.,2012,BrugerandWaters,2015,Allenetal.,2016,PaiandYou,2009,Heilmannetal.,2015,HenseandSchuster,2015).Becausenon-producingdefectorsdonotpaythecosttoproducePGbutcanstillreaptheresultingbene˝ts,afundamentalquestionishowQSsystems,andthecooperativebehaviorstheyregulate,arestablymaintainedinbacterialpopulations.Mostevidenceaddressingthis39questionisbasedontheoryandsimulationswithlittleexperimentalpopulationdynamicdatasupportingit.AlthoughithasrecentlybeenshownthatQScanservetomodulateinvestmentincoop-erativegoods,thisstudydemonstratesthatQSallowscooperatorstomatchthe˝tnessofdefectorseveninwell-mixedenvironmentsintheabsenceofanyapparentspatialstructureorpolicingbycooperators(Allenetal.,2016).Morespeci˝cally,whilepreviousstudieshavesuggestedthevalueoffunctionalQSincomparisontoamodeledorsimulatedconstitutiveproducer(Allenetal.,2016,Schluteretal.,2016),thisisthe˝rststudytoexplicitlyandde˝nitivelyshowhowvastthedi˙erenceisbetweenQS-producersandUCproducersbyex-perimentalanalysis.Moreover,comparingQSP.aeruginosaversusasimulatedconstitutivecooperator,Allenet.al.concludedthatthereareregimeswhereallformsofcooperationaredisfavored(Allenetal.,2016).However,ourexperimentalresultsdemonstratethisisnotthecaseinV.harveyi.Thegeneralityofthise˙ectisunknown,andanydi˙erencesarelikelytodependonspeci˝csofthesystem,suchassignalredundancyandstability,within-populationheterogeneity,orQSarchitecture(Cornforthetal.,2014,Anetzbergeretal.,2009,NgandBassler,2009,Even-Tovetal.,2016).Inthiscase,conditionalPGproductionprovidedbyQSwouldactasaproximatemechanismtostabilizecooperation,allowingmoreultimatestabilizingmechanismsofcooperationsuchaskinselectiontoact(Scott-Phillipsetal.,2011).Indeed,QShasbeensuggestedtoidentifytheproportionofkininmixed-speciescommu-nities,andour˝ndingthatdefectorstrainsproducelesssignalcorroboratetheabilitytoidentifythenumberofcooperatorsinmixedsingle-speciespopulations(Schluteretal.,2016,Fig.2.3).We˝rstestablishedM9-caseinmediaasanenvironmentwhere˝tnessoutcomesweredependentupontheabilityofV.harveyitoactivatethecell'sQSsystemandproduce40regulatedPG(Fig.2.1).Interestingly,thoughcontrarytoourexpectations,whileluxRconsistentlyandextensivelyexploitedpopulationsoftheUCluxOUmutant,itwasunabletodothesameagainstWT(Fig.2.4andFig.2.7).Theseresultsbearecologicalsigni˝canceasnon-luminescent,luxRdefectors,andunconditionalluxOmutantshaveallbeenidenti˝edfromnaturalVibriopopulations(O'GradyandWimpee,2008,Wangetal.,2011,NealsonandHastings,1979).Importantly,thisoccurredinwell-mixedconditionsinwhichnobio˝lmsorcellaggregateswereobservedandneitherassortmentofcompetitorsnorprivatizationofgoodswouldbelikely.Theequivalent˝tnessoftheWTstraincomparedtothedefectoratallfrequenciesexamineddemonstratesthatfunctionalQScanpromotethemaintenanceofcooperativebehaviors,providingasolutiontothePGdilemmawheninmixedcommunities(Brown,1999).AnimportantconsiderationofourresultsistheroleofpleiotropyastheQSregulonofV.harveyiisextensive(Nealsonetal.,1970,Moketal.,2003,vanKesseletal.,2013).Pleiotropyhasbeenshownasapotentialmechanismtoinhibitcheatingduetocostsimposedagainstdefectors(Dandekaretal.,2012,Fosteretal.,2004,Asfahletal.,2015,MitriandFoster,2016,Jiricnyetal.,2010).V.harveyidefectorsdopaysuchcosts,asalldefectorstrainstestedexhibitedgrowthyielddefectsthatspannedmultipleenvironmentalconditions,includingthosewherenoproteolyticbreakdownofnutrientswasrequired(Fig.2.1,p<0.001,Fig.2.4,bottompanel,blackline).ThisisalikelyresultoftheQSsystem'sregulationofalargeportionofthecell'sgenomeandinterconnectednesswiththemetabolismofthecelltoappropriatelytunegrowthacrossawiderangeofenvironments(Dandekaretal.,2012,PaiandYou,2009,MellbyeandSchuster,2014).CompetitionresultsbetweenWTandluxRinM9-tryptoneandatlowerdensitiesinM9-caseinsuggestthatWTmayindeedbeabetternutrientcompetitor(Fig.2.9,Fig.2.4,bottompanel).Thee˙ectmaybeexplainedbythe41regulationofV.harveyiQSofmultiplegenesrelatedtotransport(vanKesseletal.,2013).AndwhileWTappearedtobeneutrallycompetitiveagainstluxRintheshakenconditionstestedandwasunabletoinvadewhenrare(Fig.2.7,toppanel),itmightbepredictedthataddingavariablesuchasstructurethatallowsassortmentbetweencompetinggenotypescouldfosteranadvantageforWT(Killingbacketal.,1999).PleiotropyislikelytoplayaroleinthediminishedgrowthrateoftheluxOUstrainaswell.ProteaseproductionisnottheonlytraitupregulatedbyQS,andexpressionofotherpublicandprivategoodsarealsolikelytoimpactgrowthperformance.TheWTstrainmaynotonlybeappropriatelyexpressingproteases,butothergoods,bothpublicandprivate.Indeed,regulationofcooperationbyQSmayevenbeanexaptationwiththeoriginalselectiononregulatedprivategoods,butregardlessitissigni˝cantforthepersistenceofthesenow-regulatedcooperativebehaviors.However,theluxOUmutantisnotmaladaptiveinallselectiveconditions,asthismutantexcelsinmonocultureorwhenhighnutrientlevelsarefreelyavailable.TheWTstrainwasalsoabletoexploitandinvadetheunconditionalluxOUcooperator(Fig.2.4,bottompanel,andFig.2.8,middlepanel).Thisdemonstratesthatsuchuncon-ditionalcooperationcanalsobeexploitedbymorerestrainedformsofcooperation.Indeed,theWTstrainresemblesthedefectorstrategyatlowdensities.CompetitiveexclusionofluxOUbybothluxRandWTexhibitsnegativefrequency-dependence,demonstratingsomediminishingreturnsaslessluxOUispresent.Thisislikelyduetoreducedopportu-nitiesforcheating,althoughthisdependencedoesnotelicitcoexistence(Fig.2.4,toppanel;Foster,2004,Ross-Gillespieetal.,2007).Instead,atnopointdoesthischeatingceasetobefavored,characteristicofaprisoner'sdilemma(TurnerandChao,1999,GreigandTrav-isano,2004).SimilarexperimentswithPseudomonasaeruginosaobservedanintermediate42frequencyatwhichthecooperatorsanddefectorshadequivalent˝tness(Diggleetal.,2007),illustratingtheseQSsystemsarenotfunctioningequivalently.Alternatively,theWTstrainappearstoperforminafrequency-independentmannerwhencompetedwiththeluxRdefector,suggestingthedefector'sperformancehereisnotaresultofcheatingonPG(Fig.2.4,toppanel,Fig.2.9).ThisisalsosupportedbytheinabilityofluxRtogrowbeyondmonocultureyieldsinthepresenceofWT(Fig.2.6,toppanel,Fig.2.9).Allsignsoffrequency-dependencewereremovedwhenthesesamestrainswerecompetedinM9-tryptonemediawherecaseinhasbeenenzymaticallydigested(Fig.2.4,bottompanel).Intheseconditions,luxOUexhibitedequivalent˝tnesstothestrainsthatinvadeditinM9-caseinmedia.Therefore,ourresultsarenotsimplyduetoluxOUexhibitingpoor˝tnessinallenvironmentsexamined.ThecompetitionresultsobtainedfromtheV.harveyiQSsystemaredi˙erentfromthosepreviouslydescribedinothersystems,especiallywithregardtoitsapparentabilitytowithstandinvasionbyadefectorstrainthatisunabletoinducethehigh-densityQSstate.Forexample,alasRmutant(anotherQS-defector)ofP.aeruginosawasabletosuccessfullyinvadeaWTQSstrainunderconditionsthatrequiredproteolysis(Dandekaretal.,2012,Sandozetal.,2007).InvasionofWTbydefectorshasevenbeenreportedforVibriocholerae,asadeletionstrainofhapR,ahomologtotheV.harveyiluxR,isabletoinvadeandsupplanttheWTstraininanM9-caseinenvironment(Katzianeretal.,2015).WhereasintheseothersystemstheWTstrainsarereadilyinvadedbythecorrespondingdefector,andinsomecasesswepttoextinction,theV.harveyiWTstrainisabletowithstandanalogousinvasion(Fig.2.4,toppanel,andFig.2.7).TheresultsofthisstudydemonstratethatQSsystemsandenvironmentalconditionsexistwhereQSbacteriacanresisttheinvasionofQS-defectormutants.ThusQS-regulatedcooperationishereshowntobeamoree˚cient,43robustalternativetounconditionalcooperation,allowingpopulationstoresistinvasionbywithin-speciesdefectors(Heilmannetal.,2015,HenseandSchuster,2015).ExamininggrowthdynamicsrevealedthatafunctionalQSsystemallowedtheWTstraintoconditionallyswitchbetweenlowinvestmentinQS-regulatedgoodswhennutrientsarereadilyavailable,andallowcooperatorstoonlyinvestintheproductionofPGsuchasproteaseswhenitisprudentduetolargernumbersofcooperatorcells.Thishasthee˙ectofmaximizinggrowthrateatlow-cell-densitywhilealsoachievingmaximumgrowthyieldsbyproducingproteasePGathigh-cell-density,presumablywhensurroundedbyalargenumberofkin.Thisrelativelyhigh˝tnessa˙ordedinbothstatespreservescooperativebehaviorsassociatedwithQS.Suchphenotypicplasticityhasbeensimilarlyshowntoprovide˝tnessbene˝tsforothercooperativegoods(Kümmerlietal.,2009).AlthoughtheluxRdefectorisabletorapidlyinvadetheUCinpairwisecompetitionsbecauseitisabletoachievethehighergrowthratestrategy,itisunabletomaximizeyieldinmonocultureslikeacooperator.Alternatively,theluxOUmutantmaximizesgrowthyieldattheexpenseofaloweroverallgrowthrate,particularlyatlowcelldensity,allowingittoperformwellinmonoculturebutrenderingitsusceptibletodefectorinvasion.Thiscontrastinggrowthstrategyhasrelevancetopreviousinvestigationsintotradeo˙sbetweengrowthrateandyield,andtheplasticityconferredbyQSmayallowcellstoevadethistradeo˙bymatchinggrowthstrategiesappropriatelytotheprevailingenvironmentconditions(MacLeanandGudelj,2006,Pfei˙eretal.,2001,Novaketal.,2006).ThedelayobservedinactivatingcooperationbyQS-regulationintheWTstrainisconsis-tentwithproposedformsofrestraintsuchasmetabolicprudenceandgeneralizedreciprocity,inwhichinvestmentincooperativetraitsexpressionistunedtobemoreeconomicalfortheproducingorganism(Xavieretal.,2011,Allenetal.,2016,Darchetal.,2012).Inourcase,44costisloweredforWTduringearlygrowthwhensignaldensityistoolowtoinduceQS.Inthisstate,productionofthePGisnote˚cientastheconcentrationofproteasewillbetoolowtoyieldanetbene˝t(Heilmannetal.,2015).ThissituationcouldnaturallyoccurwhenapurecultureofV.harveyiisatlowcelldensityorifV.harveyiisinamixedcommunitysurroundedbynon-kinordefectors,wheninvestmentincompetitivestrategiesismoreben-e˝cialthanhighlevelsofcooperation.However,wealsonotethatinvasionofluxRintotheunconditionalluxOUceasesathigherdensities(Fig.2.8,toppanel).Followingthereasoningofmetabolicprudence(Xavieretal.,2011),itispossiblethatunderthesedensityconditionsitislesscostlytoexpressandproduceproteases.Itisalsopossiblethatproteaseenzymesmaypersistintheenvironmentandthusmeasuringtheirlevelsre˛ectspastaswellascurrentproduction,causinganoverestimationofrealcoststocooperatorstrainsathighcelldensity.Insummary,wehaveexperimentallyshowninV.harveyithatQSlimitscooperativegoodsproductiontowhenthebene˝tofmakingtheseproductsoutweighsproductioncosts,andcouldallowcooperationtoresistdefectorinvasioninenvironmentswhereinitialdensitiesarelow,whicharelikelycommoninitsnaturalaquatichabitat(Paietal.,2012).QS-regulationmaybeparticularlyadvantageousforgoodsthatexhibitacceleratingbene˝tsandwouldbestbeexpressedathighercelldensities(Heilmannetal.,2015,Cornforthetal.,2012).Therefore,itisadvantageoustodelaycooperationuntildensityreachessu˚cientlyhighlevels,whereitisknownthatproductionismorebene˝cial(Darchetal.,2012),orparticularlyatthetransitionfromlowtohighdensity,preciselywhereQSispredictedtoplayanimportantrole(HenseandSchuster,2015).However,asfunctionalquorumsensersappeartoperformwellagainstdefectorsacrossawiderangeofdensityconditions(Fig.2.5),QSmayprovidetheenvironmentalsensitivitytoallowcellstoactasdensitygeneralists,45respondingparticularlytokinnumbersinagivenenvironment.Theshiftinresponsestodensityweobservedisresemblantofshiftsbetweenindividualandgroup˝tnessmaxima,andtheexpectationthatorganismswillbehaveasiftomaximizetheirinclusive˝tness(Westetal.,2012,BrownandJohnstone,2001).QScouldplayanimportantroleine˙ectivelymaximizinginclusive˝tness.Together,theseresultssuggestthatcellswithafullyfunctionalQSsystempossessasigni˝cantadvantageoversystemsthatdonotappropriatelyregulatecooperativebehaviorsandaremoreresistanttocheatingbyoptimizinggrowthstrategiesinadensity-dependentmanner,providingsupportfortheideathatcommunicationcanactasaformofcheatingcontrol(Velicer,2003,TravisanoandVelicer,2004).46Chapter3DispersalMaintainsQuorumSensinginBacteriaviaaSimpson'sParadox47AbstractQuorumsensing(QS)isamechanismofbacterialchemicalcommunicationthatregulatesmanygenesincludingthosethatencodecooperativebehaviors.AlthoughmechanismsexistthatcanhelptostabilizeQS,itisnotalwayscleariforhowcooperatorsmayincreaseinapopulationofprimarilydefectors.Wehypothesizedthatdiscontinuitiesbetweenlocalandglobal˝tnesstrendscouldallowQScooperatorstoinvadeaQSdefectorpopulation,eventhoughthecooperatorsdonoteverindividuallyexhibithigherrelative˝tnessesinlocalsubpopulations.Thistypeofdiscontinuityisclassi˝edasaSimpson'sparadox.Totestthishypothesis,mixedmetapopulationsofconditional,unconditional,anddefectorQSstrainsofVibrioharveyiandVibriocholeraewerediluted,fragmentedintosubpopulations,andgrowninanenvironmentthatrequiresQSinductionofpublicgoodsproductionformaximumgrowth.Wefoundthatseparationofthesestrainsintosmallpopulationsviastrongbottlenecksledtoaglobalinvasionofbothfacultativeandunconditionalcooperatorsintoamajoritydefectorpopulation.Examinationofthemotilityofmixedpopulationsrevealedthatcooperatorsareabletodisperseandoutcompetedefectorsattheedgeofmotilecolonies.OurresultssuggestthatnotonlydoesspatialassortmentofcooperatoranddefectorgenotypesstabilizeQScooperators,thedispersalofcooperativecellstocolonizenewenvironmentscanalsoincreaseoverallcooperative˝tness.Therefore,theabilitytomodulatecolonizationanddispersalmayprovideamechanismforthemaintenanceofQSsystems.483.1IntroductionExplainingtheongoingpersistenceofcooperativebehaviorhasbeenalong-standingchallengeforevolutionarybiology(Hamilton,1963,1964,Axelrod,2006).Cooperativebe-haviorsarenotexpectedtobeevolutionarilystable,andthusrequireadditionalmechanismstoexplaintheirexistenceandpersistenceinnature(Nowak,2006,BrugerandWaters,2015).Nonetheless,cooperationiswidelyfoundatalllevelsofbiologicalorganization.Onepar-ticularclassofcooperativebehaviorsistheproductionofpublicgoods;theseare˝tness-enhancingproductsthatrequireindividualinvestmentbutwhoseresultingbene˝tscanbesharedacrossmembersofapopulation.Publicgoodproductionprovidesanopportunityfornon-producingdefectorstocheatcontributingcooperatorswithinamixedpopulationofbacterialcells,increasingtheirfrequencywhilesimultaneouslydestabilizingproductivityofthewholepopulationleadingtoatragedyofthecommons(Hardin,1968,Rankinetal.,2007).Onemechanismcontributingtothelimitationofdefectorsisfacultativeproductionofpublicgoodswhenthebene˝tisincreasedorthecostisreduced(Foster,2004).Bacteriaoftenregulatepublicgoodproductioninadensity-dependentmannerbyusingextracellu-larchemicalcommunicationknownasquorumsensing(QS).InVibrioharveyiandVibriocholerae,theQScircuitconsistsofaphosphorelaypathwaythatiscontrolledbyexogenouschemicalsignalsknownasautoinducers(Waters&Bassler,Ng&Bassler).Atlowcelldensity,whenautoinducersarelow,membraneboundkinasesfunnelphosphateintothispathway,leadingtophosphorylationoftheresponseregulatorLuxO,whichultimatelyre-pressesexpressionofthehigh-cell-densitymasterregulatorsLuxRforV.harveyiandHapRforV.cholerae.Athighcelldensity,theautoinducersbindtotheircognatereceptorsand49activatetheirphosphataseactivity.ThisleadstodephosphorylationofLuxOandexpressionofLuxR/HapR,whichinducesgenesexpressedathighcelldensity.TheVibrioQScircuitisapowerfulsystemtostudytheevolutionofQSasaconstitutivecooperatingstraincanbegeneratedbymutationtoluxO,whichlocksthecellsintothehigh-cell-densitystate,whileadefectorgenotypelockingthecellsintothelow-cell-densitystatecanbegeneratedbydeletingluxR/hapR.Wepreviouslyshowedthatthecooperativeproductionofextracellularproteases,pub-licgoodssusceptibletocheatingbynon-producingdefectors,wasstabilizedinthemarinebacteriumVibrioharveyibyfacultativeexpressionviaQSregulation(Chapter2).Thisregulationa˙ordedcooperatorcellstheabilitytoeliminateproductioncostsatlowdensitiesandnotexpresscostlycooperativebehaviorsuntilasu˚cientnumberofcooperativebacteriawaspresent.Alternatively,aluxOUmutantstrainwhichisanunconditionalproducerofproteaselockedinthehigh-densityQSstatewasrapidlyinvadedbynon-cooperatingcells(luxR)anddriventoextinction(Chapter2).However,eventhoughQSregulationstabilizedthiscooperativebehavior,cooperatorcellswereunabletoinvadeapopulationofmostlydefectorsinanenvironmentthatrequiredproteolyticactivityformaximumproductivity(M9mediawithcaseinasthelimitingcar-bonsource).Inaddition,V.choleraelockedhigh-cell-densitystrainsthatarepresumablyunconditionalcooperators(UC)havebeenisolatedfromnaturalenvironments(Wangetal.,2011).ThisraisesthequestionofhowQSispreservedinnaturalenvironments,andwhetherconditionsexistthatenablecooperatorstosucceedandinvadeintoapopulationofmostlydefectors.WehypothesizedthatQS-pro˝cientstrainscouldinvadeapopulationofdefectorsviaaSimpson'sparadox(Simpson,1951,Blyth,1972).Generallyspeaking,thistermdescribes50astatisticalphenomenoninwhichagiventrendisevidentinmultiplegroups(e.g.A>B),butthereversetrendisobservedwhenallofthosegroupsarecombined(i.e.B>A).Withregardtocooperativetraits,thisparadoxdescribesasituationwheredefectorsoutperformcooperatorsonlocalscalesduetohigherrelative˝tness,butcooperatorsultimatelyoutper-formdefectorsonglobalscalesduetoahigherabsolute˝tness,eitherduetoadvantagesingrowthrateorgrowthyield.TotesttheroleofdispersalandpopulationbottlenecksontheperformanceofQSandQS-regulatedcooperativebehaviors,weadoptedanapproachthatincludedcyclesofglobalcellmixingasametapopulation(Levins1969),highlevelsofdilution,fragmentationintosubpopulationsandsubpopulationregrowth,followedbycombinationofthesubpopulationsintoanewmetapopulation.Wehypothesizedthatdefectorsshouldbefavoredovercoopera-torswithinmixedsubpopulations,butsubpopulationsthatbychancehadmorecooperatorspresentwouldbefavorablyweightedinthemetapopulationduetohighergrowthyields(Chuangetal.,2009,Cremeretal.,2012).Ifthedi˙erenceisgreatenough,thiswouldacttoovercometherelativedisadvantageofcooperatorstopreserveorevenincreasetheirtotalproportioninthemetapopulation.Underthisregime,weobservedthatboththeWTstrainandtheUCstraininvadedefectorsinmetapopulations.Additionally,althoughexperimentalresultsfromtherelatedbacterialpathogenV.choleraehaveshownthathapRdefectorscane˙ectivelyinvadeanddominatetheWTstraininamixedpopulation(Katzianeretal.,2015),weobservedinvasionofdi˙erentV.choleraecooperatorgenotypesintothehapRdefectorwithourexperimentalapproach.Moreover,competingcooperatoranddefectorstrainsinalow-agarmotilityassay(Kearns,2010)generatedanassortmentofcooperatorgenotypesandimprovestheircompetitiveoutcomes.OurresultsdemonstratethefrequencyofQSinmixedVibriopopulationscanbeincreasedbyaSimpson'sparadoxthroughpopu-51lationbottlenecks,dispersalevents,andself-structuringthatleadstogenotypicassortment,illustratingthathighgrowthyieldcooperativestrategiescanoutcompetehighgrowthratedefectorstrategiesonglobalscalesgiventheappropriateecologicalconditions(Pfei˙erandBonhoe˙er,2004,KreftandBonhoe˙er,2005).3.2MaterialsandMethods3.2.1StrainsandMediaV.harveyistrainsBB120(WT,ATCCBAA-1116),andderivativestrainsJAF78(luxOU),andKM669(luxR)weregrownat30Casindicated.V.choleraestrainsusedincludedC6706(WT),andderivativestrainsSLS349(luxO),andBH1543(hapR)andweregrownat35C.Strainsweregrownin'M9casein',aliquidmediabaseduponM9salts(Sigma-AldrichorBeckton,Dickinson,andCompany)supplementedwithsodiumchloride(Macron)to2%(w/v)˝nalconcentrationandsodiumcaseinate(Sigma-Aldrich)to0.5%(w/v)toformM9caseinmedia.Strainswerestreakedfromfrozenstocks,andindividualcolonieswerepas-sagedfor24hoursinLBbeforepassageinM9-caseinmedia.ForgrowthinstaticM9-caseincultures(Fig.3.2),cultureswereallowedtogrow48hoursunshakenat30degreesCelsiusbetweentransfers.3.2.2MetapopulationGrowthStrainswerepassagedinmonocultureinM9caseinmediafor24hoursbeforemixing.Duringinitialmixing,competingstrainswerewashed,dilutedtoequivalentdensities,andthencombined.Uponmixing,cellsweredilutedanddividedintosubpopulationsina96-52wellplateandpassagedforgrowth.DilutionestimatesweremadeonthebasisofOD600,suchthatthetargetdensitywas1-2cellspersubpopulation.Estimatesofaveragestartingsubpopulationwereadditionallycorroboratedbycountingthenumberofemptywells(de-terminedaswellswithnosigni˝cantincreaseinOD600overbackground)presentand˝ttingthatnumbertoaPoissondistribution(Chuangetal.,2009).Platesweregrownfor3daysin30degreesCelsiusforV.harveyiandfor2daysin35degreesCelsiusforV.choleraebetweentransfers.Plateswereincubatedonorbitalshakerstofacilitateevenmixingwithinsubpop-ulations.Aftertheperiodofgrowth,subpopulationsweremeasuredforindividualgrowthinaSpectraMaxM5platereader,mixed,andtheresultingmetapopulationwasmeasuredforoverallgrowthusingbothOD600andviablecellcountsonLBagarplates.ThefrequencyofcooperatorswasdeterminedbyquantifyingbioluminescentcoloniesforV.harveyiwhilethefrequencyofdefectorsforV.choleraewasdeterminedbyquantifyingrugose/smoothcoloniesindicativeofthehapRstrainandothercompetitorstrains.3.2.3MotilityplatesStrainsweregrown24hoursinLB,thentransferredandgrowninM9-caseinfor24hourstoacclimate.Competingstrainsweremixedina1:1ratioandapproximately106cellswereplatedonthesurfaceofM9-caseinpetriplatescontaining0.3%agar.PlatescontainingstrainsofV.harveyiwereincubatedat30degreesCelsiusandplatescontainingV.choleraestrainswereincubatedat35degreesCelsius.Plateswereimagedandsampledfromdaily.Edgesampleswereobtainedfromthevisibleedgeboundaryofthecolonies.Whenwholecoloniesweresampled(Fig.3.6),theentiredispersedcolonywasharvested,suspendedinM9-caseinliquidmedia,vigorouslyvortexeduntilanevenconsistencywasobtained,thendilutedandplatedtogetcountestimates.Cellswerephenotypedaccordingtocolonymorphology53(smooth/rugose)forV.choleraeandcolonybioluminescencephenotype(luminescent/non-luminescent)forV.harveyi.3.3Results3.3.1ComparinggrowthofVibriocooperatorsanddefectorsinM9-caseinmediaWehavepreviouslyshownthatwhencooperatorsanddefectorsofV.harveyiQSaremixedandcompetedinwell-mixedconditions,luxOUsu˙ersfromaprisoner'sdilemmaandisinvadedbyboththeWTstrainandluxRdefectoratallstartingfrequencies,whiletheWTexhibitsequivalent˝tnesstoluxRatallstartingdefectorfrequencies(Chapter2).Importantly,however,theWTstrainwasunabletoinvadetheluxRdefectorevenwhenpresentinlowabundance.Therefore,wewonderedhowtheWTQSsystemmaintainsitsabundanceinthenaturalworld.The˝rststeptoaddressthisquestionwasexaminingcooperatoranddefectorgrowthinmonocultureinaminimalmediuminwhichcaseinwasthesolecarbonsource(M9-casein).Aswehavepreviouslyreported,V.harveyicooperatorstrainsthatcanexpressthehigh-cell-densityQSstate(WTandluxOU)areabletogrowtohighyieldsinthisenvironment.ThisisduetotheinductionofextracellularproteaseproductionbyQSinV.harveyi,makingenzymesthatarerequiredtodigesttheextracellularcaseinliberatingnutrientsforgrowth.Alternatively,theluxRmutantisunabletogrowtoahighyieldinM9-caseinduetoaninabilitytoproduceextracellularproteases.Here,weextendedthetimeofthisexperimentandstillobservedpooryieldoftheluxRat48and72hourscomparedtotheWTandUC54(Fig.3.1).AnalogousQSstrainsofV.choleraeweresimilarlyexamined:WT(facultativeFigure3.1:GrowthofV.harveyiandV.choleraestrainsinM9-caseinmedia.StrainsweregrowninM9-caseinmediaandmeasuredatdailyintervals,usingabsorbanceat600nminaspectrophotometer.Cooperatorstrainsshowednofurtherchangesinabsorbanceafter24hours.'WT':wildtypestrain;'UC':unconditionalcooperatorstrain;'DR':QSresponsedefectorstrain.Errorbarsrepresent95%con˝denceintervalsonN=3biologicalreplicates.cooperator),luxO(UC),andhapR(defector).LikeV.harveyi,strainsofV.choleraethatcouldactivatethehigh-densityQSstateexhibitedhigherabsolute˝tness,demonstratedbyincreasedgrowthyieldcomparedtothedefectorstrains(Fig.3.1).AlthoughthehapRdefectorexhibitedpoorgrowthat24hours,onedi˙erencewenotedbetweenthetwoVibriospecieswasthatthehapRdefectorofV.choleraewasabletogrowtosigni˝cantcelldensitiesat48and72hours.553.3.2AmetapopulationapproachselectsforincreasesofVibriocooperatorstrainsWehypothesizedthatenvironmentsthatallowedsu˚cientassortmenttopreventex-ploitationofcooperatorscouldallowthefrequencyofcooperatorstrainstoincrease.Onesuchecologicalconditionthatcouldallowthiswouldbelocalassortmentofcooperatinggenotypea˙ordedbyspatialstructure,eitherimposedbythephysicalenvironmentorpro-ducedbycellsthemselves.Totestthishypothesis,weperformedcompetitionexperimentsbetweeneitherthefacultativecooperatorortheUCversusthedefectorinM9-caseinminimalmediainstaticliquidculturesorincultureswithlowconcentrationsofagaraddedtopro-videgreateropportunitiesforspatialstructuretoemerge.Thisexperimentwasperformedwiththecooperatorgenotypesatbothrare(1%)andcommon(99%)startingfrequencies.Underconditionsofunshakenmediatoallowmorepossibleassortment(Kassenetal.,2000,Srivastavaetal.,2004,RaineyandTravisano,1998),westillobservedthattheUCwascom-petitivelyexcludedatbothfrequencytreatmentconditionstested,andthatneitherluxRnorWTcouldinvadetheotherwhenrareinamixedpopulation(Fig.3.2).Identicalresultswereobtainedwithculturesinwhichalowconcentrationofagarwasaddedtotheliquidmediumtopreventfreedi˙usionofthebacteria.Therefore,intheconditionsthatwetested,parametersthatenabledincreasedassortmentwerenotsu˚cientforQScooperatingpopula-tionstoincreaseinfrequency.WenextconsideredthepossibilityofaSimpson'sparadoxinwhichcooperatorfrequencycouldincreasewithouteverhavingarelative˝tnesshigherthanthedefectorsifmostinteractionsbetweencellsarewiththeirowngenotype.Inthiscase,wehypothesizedthatdispersionofmixedmetapopulationsintomultiplesmallsubpopulationsthroughbottleneckscouldincreasethefrequencyofcooperators.Thisisbecausepopulations56Figure3.2:CompetitiveoutcomesofmixedV.harveyicooperatorsanddefectorsinunmixedM9-caseinmedia.CooperatorstrainsofVibrioharveyiweremixedwithluxRat99:1%or1:99%ratiosandallowedtogrowforaseriesofsteps.Eachtransfergrowsfor2-3daysbeforedilutingintonewmedia.Errorbarsrepresent95%con˝denceintervalson3biologicalreplicates.randomlyenrichedforcooperatorgenotypeswouldhavehigheryieldsrelativetopopulationsrandomlyenrichedfordefectorgenotypes.Thusweadjustedourexperimentalapproachtotestthishypothesis(Cremeretal.,2012).ThisconsistedofmixingeithertheconditionalorUCcooperatorstrainsintoapredominantly(˘90%)defectorpopulationinM9-caseinme-dia,followedbydilutinganddividingthemixtureintosubpopulationsin96-wellplatestopromoteassortment.Thedilutionswerecarriedoutsothateachpopulationwasseededwithanaverageof1-2foundingcells(seeMaterialsandMethods).Afteraperiodofregrowth,thesubpopulationswererecombined,andtheratioofcooperatorgenotypesversusdefectorswas57determinedbyplatingforviablecellcountsandquantifyingthenumberofbioluminescentcolonies.Thiscyclewasrepeated7times.Indeed,intheseexperimentalconditionsweobservedadistinctanduniformincreaseinthefrequencyofbothfacultativeandUCstrainsofV.harveyicomparedtothedefectorgenotype,e˙ectively˝xingforbothcooperatorgenotypes(Fig.3.3,toppanel).TheseresultsdemonstratedthateventheUCthatwasdistinctlydisadvantagedingrowthrateanddriventoextinctionbydefectorsinwell-mixedconditionswasfavoredintheseconditions.WethenexaminedthegeneralityofthisresultbyperforminganalogousexperimentswithV.cholerae.WeconsideredV.choleraeamorestringenttestofthegeneralityofourexperimentalapproachfortworeasons.First,previousresultsdemonstratedtheWTstrainofV.choleraewasinvadedbyahapRdefectormutant.Incontrast,wehaveshownthatWTV.harveyiexhibitsequivalent˝tnesstotheluxRmutant(Katzianeretal.,2015).Second,eventhoughV.choleraedefectorssu˙erfromagrowthde˝ciencyinM9-caseinmedia,theyrecoverasubstantialamountofgrowthby48hours,unliketheV.harveyidefectorsthatneversigni˝cantlyrecoveredfromtheirgrowthdefectat24hoursintheM9-caseinenvironment(Fig.3.1).Nonetheless,forbothV.choleraeWTandluxOUC,wesawastronginvasionofcooperativestrainsintoahapRdefector-majoritymetapopulationswithincreasesfrom10%to80%or95%fortheWTandluxOmutants,respectively(Fig.3.3,bottompanel).TomorecloselyexaminethedynamicsofWTandluxOUinvasion,wedeterminedQSactivity(measuredinproxybybioluminescencenormalizedtocellgrowth)andgrowth(measuredbyOD600)ofallsubpopulationsofV.harveyi.Theresultsforthe˝rstand˝naltransferareshown(Fig.3.4).Interestingly,eventhoughbothcooperatorstrainsareabletoe˙ectively˝xinthesemetapopulations,theWTstrainexhibitsmorehomogenousoutcomeswithmanymoresubpopulationsreachinghigherdensitiesandhigh(butnotmaximumatthetimeof58Figure3.3:DemonstrationofSimpson'sparadoxinmetapopulationsofV.harveyiandV.choleraegrowninM9-casein.Simpson'sParadoxresultsinincreasingfractionsofcooperatorsinmetapopulationsofVibriobacteria.Toppanel.ForV.harveyi,growthtimebetweentransfersis3days,andBottompanel.forV.choleraeitis2days.luxRistheV.harveyidefectorstrain,andhapRistheV.choleraedefectorstrain.Linesfollowthetrajectoriesofindividualmetapopulationsanddatapointsindicatethemeancooperatorfrequencyofthe3replicatemetapopulationspertreatment.59measurement)bioluminescencewithrelativelyfewpopulationspresentinintermediatestates.Thisisconsistentwithprevious˝ndingsthattheWTstrainismoree˙ectiveatmaximizinggrowthbyappropriatelyregulatingproductionofa˝tness-impactingpublicgood(Chapter2).Alternatively,theluxOUsubpopulationsdisplayacontinuumofgrowthphenotypicpro˝les,withmanypopulationsatlowtointermediatepopulationdensitiesyetwithveryhigh(nearmaximum)bioluminescence.Evenafterthelastgrowthcycle,whencooperatorsinbothtreatmentsarenear˝xationintheirmetapopulations,subpopulationsfromtheluxOUtreatmentmeasuredalongthiscontinuumwhilethemajorityofWTsubpopulationsreachedasaturatingdensity(Fig.3.4).Thisdi˙erencebetweenstrainsisalikelyresultofWTutilizingsuperiorgrowthstrategiesinresponsetoQSsignalingtomorerapidlyreachhighdensities,whileluxOUine˚cientlyinducesQSatlowdensities,whichinturndelaysitsgrowthrateandcausesaslowerreboundincellnumbersfromtheverysmallstartingsizesexperiencedinthesesubpopulations(Chapter2).3.3.3CombineddispersalbymotilityandstronggrowthincreasescooperatorfrequencyinM9-caseinTheobservedincreasesofcooperatorsinourmetapopulationexperimentsdemonstratedthatthecooperatorstrainsinthestudycouldinvadedefector-majoritypopulationsiftheyencounteredsu˚cientassortmentfromdefectorstorealizethegrowthyieldadvantagestheyenjoyedinM9-casein(Fig.3.1,Fig.3.3).Weadditionallymeasuredthedispersalcapabili-tiesofVibriostrainsinisolationbyplatingmonoculturesonM9-caseinmotilityplatesandmeasuringthetotalvisibledispersalareaovertime.Wefoundthatforbothspeciestested,thedefectorstrainhadthelargestdispersalarea,althoughitwasnotnearlyasdenseas60Figure3.4:Phenotypicpro˝lesofsubpopulationsofV.harveyi.Phenotypicpro˝lesofsubpopulationsofV.harveyioverthecourseoftheexperiment,shownonthe˝rsttransferday(toppanel)andthe˝naldayofgrowth(bottompanel),demonstratedi˙erentpatternsintherelationshipbetweenQSactivity(measuredinproxybybioluminescencenormalizedtocellgrowth)andgrowth(measuredbycultureabsorbance(OD600)).Interestingly,eventhoughbothcooperatorstrainsareabletoe˙ectively˝xinthesemetapopulations,theseresultsdisplaymoreofcontinuumwithanegativecorrelationforluxOUandmoreofasaturatingrelationshipforWTwithmanymoresubpopulationsreachinghigherdensities.ThisislikelyaresultoftheabilityofWTtoresistcheatingevenwheninmixedpopulationswithdefectors(Chapter2).61thecoloniesproducedbyeithercooperatorstrain(Fig.3.5).ThiswascloselyfollowedbytheWTstrain,andinbothcasestheUCstrainperformedworst,despiteclaimsthatQSpositivelyimpactsmotility(YangandDefoirdt,2015,WatersandBassler,2005,SrivastavaandWaters,2012).BecausetheV.harveyiUCstrainhasbeenpreviouslyshowntoberela-tivelyslow-growinginM9-casein(Chapter2),thisgrowthdefectwouldmaskanypotentialadvantageithasinmotility.ThecombinedgrowthandmotilityresultsledustopredictFigure3.5:Vibriostrainsexhibitacolonization-dispersaltradeo˙inM9-caseinmedia.Strainswereseededonmotility(0.3%)agarplateswithstartingpopulationsizesequalto˘1/1000ofthedefectorstraincarryingcapacityinM9-caseinmedia(˘105-106cells).Motilitywascalculatedbymeasuringthetotalareaoftheregionoverwhichcellshadtraversedfromtheirinitialinoculationspot,determinedbythefringeofvisiblecellgrowth,at24hourspost-inoculation.Errorbarsrepresent95%con˝denceintervalsfor3biologicalreplicates.thatdispersalandcolonizingofnewenvironmentalpatchescouldfavorcooperatinggeno-typesthatwerepro˝cientatbothmotilityandgrowthinagivenenvironment.Todothis,wemixedpopulationsofcooperatoranddefectorstrains(1:1mixtures)andallowedthemto62competeonM9-caseinmotilityplates,wherebothgrowthofcellsanddispersalwouldbeal-lowed.Indeed,weobservedthatQSpositivelyimpactsgrowthofV.harveyiandV.choleraecooperatorsindispersingcoloniescomparedtothecorrespondingdefectorgenotypes,par-ticularlyatthecolonyedges(Fig.3.6).Becausemuchlargerseedingpopulationsareusedinthisexperiment(around106cells),thisisamuchmorestringenttestofthepotentialfortheSimpson'sparadoxtoaidcooperativebehaviorslikeproteaseproductioninnature.Inthecolonycenters,bothWTstrainsperformbetterinthepresenceofdefectorsthanthecor-respondingUCstrain(Fig.3.6).Thisdi˙erencewasmostpronouncedbetweenV.harveyistrains,consistentwithourearlierresultsinwell-mixedconditions(Chapter2).TheWTstrainsalsoconstitutemajoritiesofthecellpopulationsonthecolonyedges,whiletheUCstrainsstruggletomakeittothecolonyexterior(Fig.3.6).Thus,whileVibriocooperatorsperformedsimilarlywellinexperimentalmetapopulations,WTandUCstrainsperformverydi˙erentlyfromoneanotherwhencompetedinmotilityplates.HarvestingentiredispersedcoloniesfrommotilityplatesrevealedthatitispossibleforV.choleraestrainscompetedagainsthapRtoincreaseinfrequency,butonlyincertaincases(Fig.3.6).OnlytheWTstrainofV.choleraeshowedclearandconsistentincreasesintheirfrequencieswithinthesecolonies,whileallstrainsthatlackedeitherfunctionalluxOor˛rAwereveryde˝cientincompetitionandexperienceduniformlylow˝tnessesagainsthapR(Fig.3.6,Fig.3.7).Di˙erenceswerealsodetectedbetweenV.harveyicooperatorstrainscompetedagainsttheluxRmutant,withWTshowingnosigni˝cantdropfromitsstartingfrequencyof50%,whiletheluxOUcompetitorwaslargelyoutcompetedwithindispersedcoloniesbytheendoftheexperiment(Fig.3.6).Together,theseexperimentsprovideevidencethatthee˙ectsofanunimpairedQSsystemprovidesupongrowthanddispersalmakeWTstrainsofeitherspeciesmorecompetitiveintheenvironmentstested,evenallowingtheV.choleraeWT63Figure3.6:Competitiveoutcomesinmotilityplatecompetitions.DuringgrowthinM9+caseinmotility(0.3%agar)plates,samplesweretakenfromtheedgeofcolonygrowth,dilutedandplatedtodeterminethecompositionofthesepopulations.Atthecompletionoftheexperiment,theentirecolonywasharvested,homogenized,diluted,andplatedtodeterminethecompositionofthesepopulations.CooperatorstrainswerecompetedagainstluxRforV.harveyi,andhapRforV.cholerae,respectively.TheUCstrainsrefertoluxOUforV.harveyi,andluxOforV.cholerae,respectively.Errorbarsrepresent95%con˝denceintervalsforN=4biologicalreplicatesperstrainmix.straintoincreaseinmixedpopulationswithdefectors.3.4DiscussionHerewehavedemonstratedthataSimpson'sparadoxoccurstoincreasethefrequencyoffunctionalQSsystemsinmetapopulationsofmixedcooperatinganddefectinggenotypeswhentheabilitytocooperateprovidesanadvantageincellyield.Previousresultshaveshown64Figure3.7:Competitiveoutcomesof˛rAinmotilityplatecompetitions.TheV.cholerae˛rAstrainwascompetedagainsthapRinM9+caseinmotility(0.3%agar)plates.Duringgrowth,samplesweretakenfromtheedgeofcolonygrowth,dilutedandplatedtodeterminethecompositionofthesepopulations.Atthecompletionoftheexper-iment,theentirecolonywasharvested,homogenized,diluted,andplatedtodeterminethecompositionofthesepopulations.Errorbarsrepresent95%con˝denceintervalsforN=4biologicalreplicatesperstrainmix.thatSimpson'sparadoxcanbeobservedinasyntheticsystemduetodi˙erencesbetweenthegrowthratesofproducersandnon-producersofasyntheticgeneticcircuitthatregulatesapublicgood(Chuangetal.,2009).Inthatsystem,asignalmoleculeitselfisconsideredthepublicgoodanditsreceptionregulatesagenethatallowsgrowthinthepresenceofanantibiotic.Here,wetestedifaSimpson'sparadoxstabilizescooperationusinganon-engineeredcooperativesystem,focusingonproteaseproductionundertheregulationofQSasthepublicgoodofinterest.Theseresultsareecologicallyrelevantasthisinteractionbetweenbacteriaisfoundinnaturalenvironmentsandisbasedupondi˙erencesincarryingcapac-65ity/productivity,whicharelikelytocommonlyoccuramongstrainsthatelicitcooperativebehaviorstodi˙eringdegrees(Fosteretal.,2007).Inaddition,weextendedourinvestigationstoamorestringentexperimentthatincludedlargestartingpopulationsofmixedcooperatoranddefectorstrainsinwhichdispersalwasdrivenbythebacteriathemselves.InthisM9-casein-agarenvironment,functionalquorumsensersofbothVibriospeciesusedmotilitytoperformsubstantiallybetterthanthecorre-spondingUCstrain,leadingtoincreasesofcooperatorsattheedgeofanexpandingmotilepopulation(Fig.3.6).InthecaseofV.cholerae,thecombinede˙ectsofQSongrowthandmotilityevenincreasedWTabundancethroughtheentirecolony(Fig.3.6).Thefactthatthisincreaseislessextensivethanthoseseeninthemetapopulationexperimentsisconsistentwithothermetapopulationexperimentsthatmanipulatedstartingpopulationsizes(Chuangetal.,2009)andprevious˝ndingsthatcelldensitya˙ectsthedistributionofcellsandextentofcooperationfoundwithinbio˝lms(vanGesteletal.,2014).Examiningoutcomesinmotilityplatescon˝rmedoursuspicionsthatwhileUCstrainsperformedquitewellinearliermetapopulationexperiments,theywouldnotdoaswellinalarger,moremixedpopulation.Indeed,forbothV.harveyiandV.cholerae,theWTdidsig-ni˝cantlybetterinthecolonycenterandthisimprovementincompetitivegrowthimpacteditsabilitytoreachandsucceedatthecolonyedges(Fig.3.6).Thisresultisconsistentwithobservationsinothersystemsthatcooperatorscanoutpacedefectorsattheleadingedgeofrangeexpansions(Morganetal.,2012).VisualexaminationofbioluminescenceinV.harveyicoloniesalsodemonstratesthisdi˙erence(Fig.3.8).WhilebothWTandluxOUV.harveyistrainsexhibitstrongbioluminescenceinmonoculturedispersingcolonies,theyreactverydi˙erentlywhencompetedinmotilitycoloniesagainstluxR.Whileneitherap-peartofullyovertakesluxRatthecolonyedges(Fig.3.6),theoutcomesareverystark,66withluxOUbeingsweptwithinthe˝rstdayofthecompetition,whileWTappearstogainanadvantageatthecolonyedgeagainstluxRdefectors(Fig.3.8,Fig.3.6).Thisisalsocorroboratedbydatafromtheentiremixedcolonies,whereWTretainssimilarabundancesthroughouttheentiredispersingpopulationoverthecourseoftheexperiment,whileluxOUwasoutcompetedanddrasticallyunderrepresentedinthesepopulationsbytheexperiment'send(Fig.3.6).FurtherexaminationoftheimportanceofmotilityinV.choleraecooperatorsFigure3.8:DispersalofV.harveyipopulationsinmonocultureandincompetitioninM9-caseinmotilityplates.Picturesweretakenofmotilityplatesafter24hoursofgrowth.Imagesareoverlaysofpicturestakenwithandwithoutvisiblelightappliedtodetectbioluminescenceofcolony.Intheseimages,astrongeryellowindicatesmorebioluminescenceandstrongerblueindicateslessbioluminescence.duringthesecompetitionswithdefectorsrevealedthatfunctionalmotilitywasnecessaryforcompetitivesuccessinthemotilityagarplates,as˛rAmutantsalsodidpoorlyinmotilityplatecompetitionassays(Fig.3.7).While˛rAperformedbetteragainsthapRthantheluxOstraininmixedcolonies,itstillsexperiencedsigni˝cantdropsinpopulationfrequen-ciesoverthecourseofthecompetitions,anexpectedresponseduetoitsinabilitytodispersefurtheroutinthegrowingcolonies.Theobservedpatternsofgrowthandmotilityofstrainsinisolation(Figs.3.1and3.5)and67incompetition(Fig.3.6),suggestthattheVibriostrainsexaminedprovideademonstrationofatradeo˙intheirpotentialsforcolonizationanddispersal,andmaypredictcoexistencebetweenthecompetingtypesintheexperimentalenvironment.Previouspopulationgeneticresultsforproductionofsiderophores,anotherpublicgood,inVibriosdemonstratedco-existencebetweenproducersandnon-producers,andourresultsfrommotilityplatesalsodemonstratemixedpopulationsatthecolonyfrontiers(Corderoetal.,2012,CorderoandPolz,2014,Fig.3.6).Wehaveshownthatstronggrowthandtheabilitytodispersearebothrequisitesforsuccessincompetitionswithdefectorsinmotilityplates(Fig.3.6,Fig.3.8).Whilepreviousreportsoftradeo˙sbetweencolonizationanddispersal,eveninVibrios,hasbeenreported,the˛exibilityingeneregulationandresultinggrowthresponsesa˙ordedtotheWTstrainsbyQSmayhelpsomewhatmediatethistradeo˙andprovidestrongcom-petitivenessunderavarietyofselectivepressures(NadellandBassler,2011,Yawataetal.,2014,Chapter2).Theexperimentsinmotilityagarstillconstituteacontinuousresource-richenvironment.Apatchier,morevariedenvironment(suchasparticlesinnaturalaquaticenvironments)shouldhavethee˙ectofplacingmoreweightontheabilitytobothdisperserapidlytoreachnewpatchesandalsostronglygrowonpatchesofresourceoncecolonized(Cadotteetal.,2006,Yuetal.,2004,Muller-Landau,2010).ThissuggeststhatthesedefectorstrainscouldbeactingasaespwhichpossessedstrongerdispersalcapabilitiesinM9-caseinplates(Fig.3.5),whilethecooperatorstrainswouldrepresentacompetitivelysuperiorspeciesduetotheirincreasedcapacityforgrowth(Hutchinson,1951).Thiswouldalsosuggestthatfugitivesandstrongercompetitorscouldcoexistinmorecomplexenvironmentsthatcansupportmultipleniches,thefugitivebydispersingmoree˙ectivelytoopenpatchesandtheothercooperatorstrainsbybeingmorefecundinthisenvironment.68Together,wehavedemonstratedthataSimpson'sparadoxcanbeobservedinexperimen-talmetapopulationsthatfavorscooperatorsutilizingQSwhencompetingstrainsundergostrongpopulationbottlenecks.Whilebottleneckshavebeenappreciatedaspotentiallyim-portanttomaintainingcooperativebehaviorsinpopulations(Brockhurst,2007,Waiteetal.,2015),thisshowsthateveninlargepopulations,cooperationcanbepromotedifthereissu˚cientassortmentbetweencooperatorsandnon-cooperators.Onemechanismbywhichthisassortmentcanbeachievedisbyactivedispersalthroughcellularmotility.Indeed,strainsutilizingQSinsteadofunregulatedstrategiesachievebettercombinedgrowthandmotilityoutcomesthatallowsthemtosucceedinthepresenceofhighlevelsofdefectorsinexpandingcolonies.ThisdispersalcombinedwithfacultativecooperationcontrolledbyQSenablesincreased˝tnessofcooperatorseveninthepresenceofnon-cooperatingdefectors.Thepotentialimpactsofthese˝ndingsarenotlimitedtoproteaseproductioninVibriosexaminedinthisstudy,butcouldplayaroleinpreservingmanycooperativebehaviorsthatpositivelyimpactgrowth.Thesephenomenacouldbeespeciallyimportantwithinbio˝lmswheretheassortmentneededtoallowenhancedcooperatorgrowthisrealized,orinhetero-geneousenvironmentswheredispersalfromandcolonizationofdi˙erentpatchesishighlybene˝cial(Drescheretal.,2014,Mágorietal.,2003).PreviousinvestigationshavesuggestedthattheSimpson'sparadoxdoesn'tfunctiontostabilizecooperationwithinbio˝lms(Pennetal.,2012),butitispossiblethatdi˙erentoutcomesmayoccurinbio˝lmscomposedofalternativespeciessuchasVibrios.Bio˝lmshavebeenshowntopromotecooperativebehav-iorsandfunctiontolocalizepublicgoodsforV.cholerae(Kreft,2004,KreftandBonhoe˙er,2005,NadellandBassler,2011,Drescheretal.,2014).Additionally,Vibriosalsobegintorepressbio˝lmformationandactivatemotilityathighcelldensities,whichshouldpromotedispersal.However,itisalsoimportantthatacompetingcooperatorstrainexperiencessuf-69˝cientgrowthtoseedthatdispersalandsucceedinitsenvironment.Inthismanner,itispossiblethattheVibriostrainswithQSinthisstudyundergocyclesofgrowthanddisper-sal,thusmakingthemstrongcompetitorsinavarietyofenvironmentsandevenincreasingcooperativebehaviorsundertheirregulation.70Chapter4ExperimentalevolutionofquorumsensinginVibrioharveyi71PrefacePreviouschaptershavefocusedonrelativelyshort-termoutcomesofselectionandhaveexaminedpre-de˝nedcompetingstrainsofV.harveyiandV.cholerae.ThedefectorstrainsusedinthosestudieswereconstructedbyengineereddeletionsoftheQSmasterregulatorgenesluxR/hapRfromthegenome,conveyingacompletelossoftheexpressionofhigh-cell-densitygenessuchasproteaseproduction.Likewise,theunconditionalcooperator(UC)strainswereengineeredbymutationtothegeneencodingtheresponseregulatorluxO.WhilenaturalmutantsofQShavebeendetected,theyarelikelytobedi˙erentinsourcethanthegeneticallyengineeredstrainsthatwereutilizedinChapters2and3.Thischapterseekstoevaluatewhetherdefectorscanevolvefromparentalcooperatinggenotypesdenovoinexperimentalpopulationsandhowtheevolutionofdefectorsdependsonboththegenotypefromwhichitevolvedandthecooperatorgenotypesitencountersinitsenvironment.724.1IntroductionThedegreetowhichquorumsensing(QS)isunderselectioninnatureisanopenques-tion.Ithasbeensuggestedthatinvestmentinsignalingshouldbehighestinintermediatefrequencystatesofcooperatorsanddefectors(BrownandJohnstone,2001),andthatselec-tionshouldfavordecreasedinvestmentinQSovertime(Popatetal.,2015b).Ithasalsobeenobservedthatnaturalpopulationsofbacteriaoftenhavemixedpopulationsofproduc-ersandnon-producersofpublicgoodsandQScooperatorsanddefectors(Wangetal.,2011,Corderoetal.,2012),suggestingthatselectionfavorstheevolutionofdefectingtypesandmayalsofosterthecoexistenceofthesedi˙erentstrategists.IhavesofarshownthatfunctionalQSinVibriosplaysimportantrolesinstabilizingcooperativebehaviorsinthepresenceofengineereddefectorsinmixedpopulations(Chap-ter2)andcanalsopromotetheincreaseofcooperativebehaviorswhencompetingstrainsundergodispersaleventsorpopulationbottlenecks(Chapter3).However,itispossiblethatdi˙erent,˝tterdefectorsthantheluxRmutantweusedastheprototypicaldefectorinthosestudiescouldevolveifenoughtimewereprovided.Ourpreviousinvestigationsintocooperationanddefectionprimarilyutilizedstagedcom-petitionsofmixedpopulationswiththeluxRstrain.IndividualcellsofthismutantstrainshouldbehaveasdefectorsforcooperativebehaviorspositivelyregulatedbyQS,suchasextracellularproteaseproduction,andcouldpossiblyexploitcooperatorsexpressingthisbe-haviorintheirpresence.ThistypeofcheatingbehaviorbyQS-defectorshasbeenreportedinotherexperimentalsystemssuchasPseudomonasaeruginosaandVibriocholerae(Sandozetal.,2007,Katzianeretal.,2015,Diggleetal.,2007).However,masterregulatorknockoutmutantsarealsolikelytobearsigni˝cantpleiotropic˝tnesscostsastheyoftenregulatelarge73numbersofgenes,andtheestimatedregulonofluxRinV.harveyiisaround10%ofthegenome(vanKesseletal.,2013,SchusterandGreenberg,2006).Therefore,wehypothesizedthat˝tterdefectorscouldevolveandinvadeV.harveyicoop-eratorstrainsduringthecourseofextendedgrowthinM9-casein,anenvironmentinwhichabsolute˝tnessisdependentontheextentofQS-regulatedproteaseproduction.Ourpre-dictionwasthatdefectorswouldevolveandbecomedetectableinM9mediawithlimitedcarbonsourcesifgivensu˚cienttime.DuetothedramaticsuccessofluxRwhencompetedagainstluxOUinM9-casein,wealsopredictedthatitwouldalsobemorecommontoseedefectorsevolveandincreaseinthepresenceofUCthanthoserestrainedbyQS.Toexaminethis,weinitiatedreplicatepopulationsofV.harveyicellsfromtwocooperatorgenotypes,aWTstrainwithfunctionalQSandanUC,luxOU,thatexpressesQSasthoughinducedathighdensityregardlessoftheactualsignalordensityconditionsinwhichitispresent.ThesestrainswereevolvedinM9mediaenvironmentsfor2000generations.Asexpected,wefoundthatafunctionalQSsystemdelayedtheinvasionofdefectorgenotypes.WhereastheluxOUUCgenotypewasrapidlyinvadedbynon-luminescentdefectorsthatalsodonotproduceprotease,signi˝cantlyhighernumbersofcooperatorspersistedinthemajorityofWTpopulationsprovidingfurtherevidencethatQScontrolofcooperationstabilizesthesephenotypes.ManyofthesepersistentcooperatorsexhibitedadimphenotypeinwhichQSwasmaintainedbutdampened,similartoourearliercompetitionexperimentswithaddedsignals.Overall,theseresultsprovidefurthersupportthatmaintainingfunctionalQShasthee˙ectofpreservingcooperativebehaviorsunderitsregulation,evenoverlongerevolutionarytimescales.744.2MaterialsandMethods4.2.1CompetitionswithsupplementedautoinducersCompetitorswereassessedfortheabilitytoinvadewhenrarebyinitiatingmixedpop-ulationsofcooperatoranddefectorstrains,andpopulationcompositionwasassessedatdailyintervals.Fortheseserialcompetitionassays,populationswereinitiatedbymixingstrainsat99%ofacooperatorstrainand1%ofapotentialdefectorstrain(defectorrare),orviceversa(cooperatorrare).PopulationsweregrowninM9minimalmediacontain-ing0.5%(w/v)sodiumcaseinate(Sigma)asthesolecarbonsourcefor24hours,platedonLBagartoassesspopulationcompositionandproductivity,andasubsetwastrans-ferred(a1000-folddilution,allowingapproximately10generationsdaily)tonewM9liquidmedia.Competitionstreatmentsincludingexogenousautoinducerwereconductedinthesamemannerasotherserialcompetitionexperiments,exceptthatmediawassupplementedwiththeacylhomoserinelactone(AHL)signalmolecule3-hydroxybutanoyl-L-homoserinelactone(HAI-1)andthefuranosylboratediestersignalmolecule3A-methyl-5,6-dihydro-furo(2,3-D)(1,3,2)diox-aborole-2,2,6,6A-tetraol(AI-2,providedintheformof4,5-dihydroxy2,3pentanedione(DPD),generouslycontributedbyBonnieBassler)toa˝nalconcentrationof10M.4.2.2EvolutionExperimentsTwelvereplicatepopulationswereinitiatedfromstocksofwildtype(BB120)andluxOU(JAF78)andcorrespondinglacZ-taggedstrains(ELB714andELB738)V.harveyistrainsandpassagedinM9saltsbasedmedia(SigmaorDifco)withtwopercentsodiumchloride75(Sigma)andcasaminoacids(Difco)orsodiumcaseinate(Sigma)usedasthesolecarbonsource(M9-caseinandM9-CAA).Cultureswerepassagedin1mLculturesin96deep-wellplates.Culturesweredailydiluted1000-foldintohomologousmedia,allowing˘9.97(i.e.approximately10)generationsperday.Stocksweremadeperiodicallythroughthecourseoftheexperimentandstoredat-80degreesCelsius.Populationswereperiodicallysampledtodetermineabsorbance,cellcountsandQSphenotypes.Ifanycontaminationofcultureswasexperienced,sampleswerereinitiatedfromthemostrecentfrozenstockavailable.4.2.3ProductivityestimatesMeasuresofcellgrowthweremadebyperiodicallytakingsamplesofthereplicatepopu-lations,diluting,andplatingonLBagarplates,andenumeratingthroughviablecellcounts.MorefrequentmeasuresofgrowthwerealsomadebymeasuringabsorbancesofpopulationsubsamplesinaBeckmanCoulterspectrophotometeratawavelengthof600nm.4.2.4PhenotypemeasurementsProteaseproductionwasdeterminedbyFITC-caseinaseassays(Sigma)aspreviouslydescribed(Chapter2).InFigure4.4lowerpanel,proteaseconcentrationswereestimatedbynormalizingmeasurementstoatrypsinstandardcurve.Theclonesthatwereanalyzedweretakenfromplatedsamplesoftheevolvedpopulationsatgeneration870oftheexperi-ment.Monoculturesofthesecloneswere˝rstpassagedinLBfor24hours,diluted1000-foldintoM9-caseinmedia,grownfor24hours,andthenextracellularproteaselevelsweremea-suredfromculturesupernatants.Bioluminescencewasmeasuredatthepopulationlevelbytakingculturesubsamples(100Lundiluted,ordilutedifluminescencelevelsapproached76saturation)andmeasuringinanEnvisionMultilabelPlateReader(PerkinElmer).Biolumi-nescenceattheclonallevelwasqualitativelyassessedbyexaminingcoloniesonLBagarinanAlphaImagerHPlightbox(ProteinSimple)throughthechemiluminescence˝lterandnovisiblelightappliedtoobserveonlylightproducedbycellsgrowingontheplates.Clonalbioluminescencewasalsomeasuredquantitativelybypropagatingisolatedcoloniesfrompop-ulationsinM9-caseinandmeasuringlightoutputinanEnVisionMultilabelPlateReader(PerkinElmer).Todeterminethestrainbackgroundofscreeneddefectorsfromtheshort-termselectionexperiment(Fig.4.1),luxOUandluxRcellsortheirevolveddescendantsweredeter-minedbyscreeningforresistantcoloniesonLBwithchloramphenicol(10M).TheWTstrainissensitivetochloramphenicol.TheluxOUstrainwasdistinguishedfromluxRbycolonyPCRofselectclones.4.3Results4.3.1Exogenousautoinducerincreasesselectionfordefectors,especiallycooperator-deriveddefectorsOurpreviousresults(Chapter2)indicatedthatafunctionalQSsystemcouldpreventinvasionbydefectorcellswhereasaUCstrainwasrapidlyinvaded.Oneprimaryexplana-tionofthisresultisthatprematureproductionofproteasebytheluxOUunconditionalcooperatoratlowdensitiesenhanceddefectorinvasion.Likewise,timelyrestraintofpublicgoodsproductionbytheWTstrainwaspredictedtopreventdefectorinvasion.Totestthispossibility,wereasonedthatprematureinductionofthehigh-densityQSstateintheWT77strainbyadditionofexogenousautoinducer(AI)signalmoleculesthatactivatetheQSre-sponseshoulddecreaseitsabilitytopreventdefectorinvasionbyprematurelyactivatingQSinamannerdecoupledfromtheactualcelldensityconditions,whichisresemblantoftheQSphenotypeoftheluxOUstrain.Therefore,theWTstrainwassupplementedwith10MconcentrationsofHAI-1andAI-2,thetwopredominantAIsinV.harveyiattheinitiationoftheculture(WatersandBassler,2006).AdditionoftheseAImoleculesresultedinWTbioluminescencebeingvirtuallyindistinguishablefromtheluxOUstrainatalldensitiesduringgrowth,demonstratingthatQSwasinducedbythesesignals.Werepeatedthecom-petitionsofWTwiththeluxRdefectorinthepresenceandabsenceofexogenousHAI-1andAI-2andobserveddi˙erentoutcomesuponAIsupplementation(Fig.4.1,topleft).Asobservedpreviously,thedefectorstrainrapidlyinvadedtheluxOUstrain,andthisinvasionwasaccompaniedbyadropintheproductivityofthepopulationasawhole(Fig.4.1,toppanels,solidredlines).InthecaseoftheWTtreatmentwithaddedsignal,nosigni˝cantinvasionoccurredforthe˝rst20generations,atwhichpointnon-luminescentdefectorsand(variantsthatmaintainreducedbutstilldetectablebioluminescence)thatwelabelasNACs(foropbegantoincreaseandultimatelybecamenearly90%ofthepopulation(Fig.4.1,topleftpanel,bluelines).Bytheendoftheexperiment,theWTtreatmentwithaddedAIsalsoexhibitedasigni˝cantdropinpopulationproductiv-ity,butstillremainedmoreproductivethantheluxOUtreatmentpopulations(Fig.4.1,toprightpanel,blueversusredline).Inthisparticularexperiment,theWTpopulationswithoutaddedsignalalsoexperiencedamodestinvasionbyNACsaswellbytheendoftheexperiment,althoughtoamuchlesserextentthantheotherconditionsexamined(Fig.4.1,topleftpanel,solidblackline).Additionally,thisinvasionwasnotenoughtoupsettheproductivityofthepopulations(Fig.4.1,toprightpanel,solidblackline).Theseresults78Figure4.1:CompetitionsofV.harveyistrainsinM9-caseinmediainthepresenceandabsenceofsupplementedautoinducers.V.harveyicooperatorstrainsWTandluxOUwerecompetedagainstluxRinM9-caseinmedia,withorwithouttheadditionof10MautoinducerHAI-1andAI-2.Toppanels.Defectorfrequency(topleftpanel)andpopulationproductivity(toprightpanel)weremeasuredonadailybasis.Defectorback-groundwasdeterminedbychloramphenicolphenotype.Fortoppanels,errorbarrepresent95%con˝denceintervals.Bottompanels.Attheconclusionoftheselectionexperiment,clonesfromdi˙erentphenotypicclasses(classi˝edbycolonybioluminescence)wereisolatedfromtheexperimentalpopulationsandanalyzedforlightproduction(bottomleftpanel)andgrowthinM9-caseinmedia(bottomrightpanel).Growthwasmeasuredbyabsorbanceat600nminaspectrophotometer.BioluminescencewasestimatedusingaEnVisionPlateReader.Errorbarsforthelowerpanelsrepresent95%errorbarsforeachphenotypicclass.Theancestralnon-luminescentcontrolstrainusedwasluxRandtheancestralluminescentcontrolstrainsincludedbothWTandluxOU.supportourhypothesisthatprematureinductionofproteasecanpromotedefectorinvasion,althoughitappearsthateveninthepresenceofsupplementedAIsatconcentrationsbeyondthatrequiredfortypicalQSinduction,theWTstrainsu˙erslessinvasionthantheluxOUUC.79DefectorstookmuchlongertoinvadeincompetitionswithWTthanincompetitionswithluxOU,suggestingdefectorsevolvedduringtheexperimentratherthanre˛ectinganinvasionoftheseededluxRdefector.Therefore,wehypothesizedthatthecompetitionswithWTselectedforadi˙erentsubsetofdefectorsthanthecompetitionsbetweenluxOUandluxR.BecausetheluxRstrainischloramphenicol-resistant,wedistinguishedifthedefectorsthataroseincompetitionswithWTwithorwithoutAIswerederivedfromtheancestralWTstrainortheluxRmutantstrain.Indeed,inbothcases,themajorityofNACswerechloramphenicol-sensitive,suggestingtheseevolvedfromtheWTancestorratherthantheluxRdefector(Fig.4.1,topleftpanel,blackandbluedottedlines).Alternatively,therapidsweepofluxOUincompetitionsuggestedthatthesedefectorsaredescendedfromtheseededluxRmutant,andindeed,PCRcon˝rmedthistobethecaseforeverydefectortestedinthiscondition(datanotshown).TheseresultssuggestthattheWTstrainmaintainingtheabilitytocommunicatehasmoreadaptivepotentialtoevolve˝tterdefectorvariantsthatoutcompeteboththeWTancestorandtheseededluxRdefectorsinmixedpopulations,whereastheunconditionalluxOUstrainismoreevolutionarilyconstrained.Whenwedidacloseranalysisoftheinvadingstrains,weobservedaqualitativedi˙erencebetweenthedi˙erenttreatmentgroups.WhereasalldefectorsintheluxOUtreatmentdisplayedanon-luminescentphenotype,aswouldbeexpectedasthesearederivedfromtheluxRmutant(Fig.4.1,topleftpanel),theevolvedtypesintheWTwithaddedAIconditionandthosethatevolvedwithoutAIadditionweremarkedlydi˙erent.Thedynamicsproceededassuch:luxRlevelsintheWTwithaddedAIpopulationsbegantoriseafter20generationsofgrowthinM9-caseinmediaandreachedmaximumfrequenciesaroundgeneration50(Fig.4.1,topleftpanel,dashedblueline).Atthispoint,weobservedtheemergenceofaWT-evolvedsubpopulation,thatdisplayedadimphenotype,whichbeganto80appearathigherlevelsandwereselectedtobecomethemajormemberofthepopulationsbygeneration70(bluedottedline,Fig.4.1,topleftpanel).BecauseWTpopulationswithaddedsignaldonotreachthemaximumpopulationdensityobservedintheexperimentuntildaystwothroughfouroftheexperiment(Fig.4.1,toprightpanel,blueline),wehypothesizedthatthisreduceddensitydelayedinvasionofluxRdefectorsduringtheinitialgenerationsoftheexperiment,asitisathighdensitieswheredefectorsaremostlikelytobeabletoexploitWTcells.WhenluxRdefectorsreachedtheirmaximumfrequenciescorrespondsdirectlywithwhenWT-deriveddefectorsanddimsbegantorapidlyincreaseinfrequencyinthepopulations,displayingdynamicsresemblingclonalinterference(HillandRobertson,1966,KaoandSherlock,2008,Fogleetal.,2008,Langetal.,2013).ThisincreaseofWT-deriveddimcellsindicatesthatthesemutantsweremore˝tthanluxRlineagestheyencounteredinthisenvironment.Whilethesewavesofinvasionservedtoreducetheancestralstrain,itwasnotcompletelysweptfromthepopulationandwaspresentat10%atgeneration70.Duringthisperiodoftime,thedensityofthesepopulationsdroppedroughlythree-tofour-fold,butstillmaintainedaroundaseven-toeight-foldincreaseoverthedensitiesoftheluxOUpopulationsthatweresweptbyluxR(Fig.4.1,toprightpanel).Importantly,theWT-deriveddimsmaintainedtheabilitytoQSastheyexpresseddimluminescence,andeveninthepresenceofexogenousAI,80%ofthecellsintheWTwithaddedAIconditionexpressedQStosomedegree.Intotal,thissuggeststhatthedimvariantsthatevolvedfromtheWTgenotypeare˝tterthanluxRdefectorsinM9-caseinandthattheirevolutionpreventsthemoreextensivetragedyofthecommonsthatwasobservedinluxOUlineages.Asubsetofclonesthatmaintainedabrightphenotype,dimphenotype,ordarkphenotypewereisolatedfromtheWTwithAIselectionexperimentandanalyzedforbioluminescenceandgrowthinmonoculture(Fig.4.1,bottompanels).The81resultsindicatedi˙eringlevelsofbioluminescencewiththedimsexhibitingnearly3-ordersofmagnitudelessbioluminescencethantheevolvedbrights.ThisdecreasedinvestmentinQSwasfurtherexhibitedbyseverelyreducedgrowthinmonoculture,indicatingthatthesedimmutantsarelikelydefectingrelativetothecooperatorstrains.Wealsoobservedreducedbioluminescencefortheevolvedbrightclonescomparedtotheparentalstrain,suggestingthatevenpopulationmemberslackingastrongdimphenotypemightbeunderselectionfordecreasedQSinvestmentunderthehighsignallevelcondition(Fig.4.1,bottompanels).Figure4.2:Frequencyandgrowthresultsfromexperimentallyevolvedpopula-tionsofV.harveyi.The˝guredepictsobserveddefectorfrequency(leftpanels)andpopulationdensityachievedat24hoursofgrowth(rightpanels)asmeanestimatesacrosstreatmentgroups(toppanels)orforindividualpopulationsovertime(bottompanels).Allfourtreatmentgroupsaredepictedinmeanestimateplots,butonlyM9-caseinpopulationsaredisplayedinthetrackingofindividualpopulations.Errorbarsintoppanelsrepresent95%con˝denceintervalspertreatment.824.3.2Long-termexperimentalevolutionoftheWTandUCQSstrainsBasedontheresultsoftheseexperimentsindicatingthatalternatedefectorstrategiescouldevolveinourexperimentalconditions,weinitiatedalonger-termevolutionexperimenttostudytheevolutionofQSsystemsindi˙erentbackgroundsandgeneticconditions.Twelvereplicatepopulationseachwerestartedfromtwostrains:WTV.harveyiwhichhasafunc-tionalQSsystem,andtheluxOUUCthatconstitutivelyexpressestheQShigh-densityregulon.Thisexperimentwasperformedintwodi˙erentenvironments:M9-caseininwhichQSisrequiredforahighgrowthyield,andM9mediawithcasaminoacids(M9-CAA)inwhichQSisnotrequiredforahighgrowthyield.Thesepopulationswerepassagedfor2,000totalgenerations.Weexpectedtoobservetheevolutionofdefectors,de˝nedascoloniesthatdonotpro-duceanyvisiblebioluminescence,inM9-caseinbutnotinM9-CAA.However,duringthecourseoftheexperiment,defectorsandNACsevolvedinallfourtreatmentgroups,althoughthepatternsofevolutionofthesedefectorswasquitedi˙erentbetweengroups(Fig.4.2,topleftpanel).Weadditionallysawtheevolutionofdimvariantsoccurinmanyoftheexperimentalpopulations.Becausethedynamicpatternswereclearerandofmoreinter-estregardingQSinM9-casein,wefocusedthemajorityofouranalysesontheselineages.Thesedefectorsbegantoappearatdetectablefrequenciesearlyintheexperiment,within100generationsforM9-caseinpopulations(Fig.4.2,leftpanels).Asexpected,completelynon-luminescentdefectorsevolvedrapidlyinluxOUM9-caseinpopulations,increased,and˝xedinallluxOUpopulations,withtwoexceptionsdiscussedbelow(Fig.4.2,leftpan-els,redlines,Fig.4.3).Alternatively,shortlyaftertheappearanceofdefectorsintheUC83lineages,non-luminescentdefectorsbecamedetectableinWTM9-caseinpopulationsaswell(Fig.4.2,leftpanels,blacklines).However,defectorsevolvedtoonaverage46%membershipoftheWTM9-caseinpopulations,andalthoughtherewashighvariabilitybetweenexper-imentalpopulations,mostpopulationswerenotsweptbythesedefectors.Thedi˙erencebetweenthetwogenotypetreatmentsinM9-caseinisalsoquitevisuallystrikingwhenthesepopulationsaredilutedandplatedonpetriplatestoallowgrowthofindividualpopulationmembersinisolation.WhileluxOUpopulationsrapidlyandirreversiblyloseallbiolumi-nescentcooperators,mostWTpopulationscontainsigni˝cantfractionsofcooperatorsevenatgeneration2000(Fig.4.3).8485Figure4.3:ExperimentallineagesofV.harveyistrainsevolvedinM9-caseinmediaover2000generationsex-aminedonpetriplates.Picturesweretakenofplatedsubsamplesoftheexperimentalpopulationsaftersu˚cientgrowthtimetoallowvisiblecoloniestoform,usually24-48hoursofgrowth.Imagesareoverlaysofpicturestakenwithandwithoutvisiblelightappliedtodetectbioluminescenceofcolony.Intheseimages,astrongeryellowindicatesmorebioluminescenceandstrongerblueindicateslessbioluminescence.Eachcolumncorrespondstoasinglepopulation,nameattop,sampledatmultipletimepointsduringtheexperimentandobservedforbioluminescentcoloniesonpetriplates.Thenumbersontheleftsidecorrespondtothegenerationaltimepointatwhichthesamplingtookplace.PopulationsCAS01-12(toppanel)wereinitiatedfromWTclonesandpassagedinM9-caseinmediaforthedurationoftheexperiment;populationsCAS13-24(bottompanel)wereinitiatedfromluxOUclonesandpassagedinM9-caseinmediaforthedurationoftheexperiment.86Figure4.3(cont'd)86BecausedefectionimpactsnutrientbreakdownintheM9-caseinenvironment,theriseofdefectorsinM9-caseinpopulationshasthee˙ectofdroppingthepotentialproductivityofthatpopulation.ThisisindeedwhatwasseenintheM9-caseintreatment,withpopulationsfrombothgenotypebackgroundsdroppinginpopulationdensitiesachievedoncedefectorsbecamedetectable(Fig.4.2,rightpanels).ThisdropindensityintheWTpopulationswasdi˙erentthanwhatwasobservedwhenshort-termexperimentswereperformedwithWTwithaddedAI,inwhichpopulationdensityremainedhigh(Fig.4.1,toprightpanel).However,thedi˙erentcooperatinggenotypesintheM9-caseintreatmentsrespondedverydi˙erentlytothisdropinpopulationdensity.WepredictthattheluxOUUCpopulationsshouldcontinuetoexpressproteaseproductionandotherbehaviorspositivelyregulatedbyLuxRregardlessofactualcelldensitiesbecausetheyarelockedinthehigh-cell-densityQSstate,thusleadingtotheirbeingrapidlyoutcompetedbydefectors,whichwasobserved(Fig.4.1,topleftpanel,Fig.4.2,leftpanels,Fig.4.3).WhiledefectorsdidinvadetodetectablelevelsintheWTM9-caseinpopulations,theyonlyappearedtosweeponeofthem,withelevenofthetwelvereplicatelineagesmaintainingdetectablelevelsofcooperatorsatthe2000generationtimepointoftheexperiment(Fig.4.2,leftpanels,Fig.4.3).Rather,inmostcasescooperatorsanddefectorsphenotypicvariantscoexistedtogether,whichhasbeenobservedinotherexperimentalsettings(Goreetal.,2009,MacLeanandGudelj,2006,Pollaketal.,2016,TurnerandChao,2003).Indeed,inmanyWTlineages,therewashardlyanydefectorinvasionexperiencedoverthecourseoftheexperiment,andtheextenttowhichtheevolveddefectorsdidinvadevariedgreatlybetweenlineages(WTM9-caseinpopulationsgeneration2000defectorfrequencyCOV=0.793,Fig.4.2,lowerleftpanel).874.3.3Defectorlevelshavestronge˙ectsonpopulationlevelphenotypesinM9-caseinBecausethesepopulationswerestartedfromclonalcooperators,theystartedoutwithsimilarlyhighlevelsofbioluminescenceandreachedhighdensitiesinbothexperimentalmediaconditions.InFigure4.3,weshowthatalltheexperimentalpopulationsnearthestartoftheexperiment(10generationsintoexperiment)hadsimilarhighlevelsofbothgrowthyield(inCFU/mL)andinbioluminescence(inRLU/mL),whichweinterpretasaproxyforhighlevelsofQSactivityinthepopulations(Fig.4.3,triangles).Butoverthecourseoftheexperiment,thepopulationsdivergedinbothproperties.Allpopulationsexperienceddropsindensityfromstartingpoints,particularlyinM9-caseinwheredefectorshadstrongere˙ectsonproductivity(Fig.4.2,rightpanels).Aftertheseearlydropsindensity,luxOUM9-caseinpopulationsremainfairlylevelinpopulationdensitiesfortheremainderoftheexperiment,withtheexceptionoftwopopulationswhichexperiencedlargeincreasesinpopulationyields,withendingpopulationyieldsreachedatthe2000generationpointamongthehighestoftheexperimentalpopulations(Fig.4.2,rightpanels).Thesetwopopulations(CAS16andCAS19)alsohavehigherlevelsofbioluminescencethananyoftheotherluxOUM9-caseinpopulations;thisbioluminescenceisdi˚cultandinsomecasesnotpossibletodetectonagarplates,butitcanbedetectedabovebackgroundlevelsinasensitiveplatereader(Fig.4.3).At2000generations,theothertenluxOUM9-caseinpopulationsdidnothavedetectablebioluminescencebyeitherplatephenotypesorbyplatereadermeasurements.ThesepopulationswerephenotypicallyindistinguishableinbothgrowthandbioluminescencefromourcontroldefectorstrainsluxR,luxOD47E,andSN(APPENDIXTableA.1).88Figure4.3:GrowthandbioluminescenceofV.harveyipopulationsevolvedinM9-caseinmedia.Populationsweresampledatgenerations10and2000andmeasuredforgrowthbyviablecellcountingandforbioluminescenceinaplatereader.Eachpointcorrespondstoasinglepopulation.M9-caseinpopulations(CAS01-CAS24)aredisplayed.Trianglesrepresentsamplesfromgeneration10andcirclesrepresentsamplesfromgeneration2000.Meanwhile,aspreviouslymentioned,elevenofthetwelveWTM9-caseinpopulationsstillcontaineddetectablelevelsofluminescentcooperatorsattheendoftheexperiments(Fig.4.2,lowerrightpanel,Fig.4.3).Asinglepopulationcompletelylostdetectablebiolumi-nescentcooperatorsfairlyearlyintheexperiment(CAS09),althoughoneotherappearedtonearlyhavelostcooperatorsat2000generationsaswell(CAS07)(Fig.4.2,lowerrightpanel).However,thoughtheotherWTpopulationsmaintaincooperators,theydisplayadi-verserangeofpopulation-levelbioluminescenceat2000generations(Fig.4.3,closedcircles).Therewasalsoastrongpositiverelationshipbetweenpopulationbioluminescence,whichwetakeasameasureofQSactivity,andgrowthoutcomesforWTM9-caseinpopulations(cor-relationbetweengrowthandluminescenceinWTM9-caseinpopulationsatgeneration200089=0.9,Fig.4.3,closedcircles).ThiswasnotthecaseforluxOUM9-caseinpopulations,wheremostpopulationsexhibitedlowbioluminescenceandgrowthyieldre˛ectiveoftheevolveddefectorsthatovertookthosepopulations.Thee˙ectsseeninM9-CAApopulationsweredi˙erentthanthoseseeninM9-casein,especiallyintermshowdefectorimpactdependedongenotypicbackground.Foreithergenotype,QSdidnotcontributeasmuchtogrowthintheM9-CAAenvironment,anddefectorsincreasedinfrequencyinmostpopulationsofbothtreatmentgroups(Fig.4.2,upperleftpanel).ThepositiverelationshipbetweenQSactivityandgrowthyieldinM9-CAAisalsoabsent(datanotshown),furthersuggestingthatQSactivitydoesnotbolstergrowthinthisenvironmenttotheextentitdoesinM9-casein.WhileQSdidnotimpactgrowthyieldsasstronglyinM9-CAA,itshouldbenotedthatM9-CAApopulationsdidexperiencedropsingrowthyieldsduringtheexperiment,againnotasseverelyasintheM9-caseinpopulations(Fig.4.2,upperrightpanel).Thisispredictedtobeduetoslight(2-3x)dropsinproductivitythatQS-defectorsexperienceinM9-CAAandbecauseselectionisstrongerupongrowthratesthangrowthyields.4.3.4Evolvednon-luminescentdefectorspossessmultiplephenotypicdi˙erencesfromancestralcooperatorstrainsAsnoted,theriseofnon-luminescentdefectorsinM9-caseinpopulationsledtolargedropsinproductivity(Fig.4.2,upperrightpanel).Becausebioluminescenceandproteaseproductionaresimilarlyinducedathighcelldensity,andbaseduponpreviousmeasurementsofproteaseproductionandbioluminescenceindefectors(Fig.2.2inChapter2),wehypothe-sizedthatthedecreasedpopulationproductivityresultedbecausethesedefectorsalsoexhib-90itedlowtonon-measurablelevelsofproteaseproduction,similartoourengineeredluxRstrain.However,itispossiblethatbioluminescenceevolvedindependentlyfromproteaseproduction.Totestthispossibility,singleclonesfromeachofthedi˙erentbioluminescencephenotypicclasses(bright,dim,dark)presentinagivenpopulationatgeneration870oftheexperimentwerechosenfromall48experimentalpopulationsbasedontheircolonybiolumi-nescentphenotype.TheclonesweresubsequentlygrowninM9-caseinmedia,andmeasuredforproteaseactivity.Thesemeasurementsofproteaseproductiontakenafterculturesweregrown24hourstoallowthemtoreachtheircarryingcapacityinM9-caseindidindeedcon-˝rmthatallnon-luminescentclonestestedproducedsimilarlowlevelsofproteasetotheluxRdefectorcontrol(Fig.4.4).Thelowlevelsofproteaseproductionalsocon˝rmedthattheyareinfactgeneralQS-defectors,andnotonlyharboringmutationsspeci˝ctobiolumi-nescence.Weinterprettheseresultstomeanthatdefectorsdisplayingcorrespondingdropsinbothpropertiesareduetogeneticchangesa˙ectingtheQSregulatorymachineryratherthanmultipleseparatemutationsspeci˝ctobothindividualprocesses.Thoughnotspeci˝-callymeasuredhere,signalmeasurementsofothercontrolstrainsshowedthatQSdefectorstendtoalsoproducelesssignalmolecules,andthistrendcouldalsobetruefortheseevolveddefectors(Chapter2Fig.2.3).Todetermineifvaryinglevelsofproteaseactivityimpactedgrowth,wecharacterizedgrowthphenotypesofthesesclonesinM9-caseinwhengrowninmonoculturesandcomparedittotheirproteaseproductionphenotypes(Fig.4.4,lowerpanel).WealsomeasuredseveralreplicatesofthecontrolgenotypesWT,luxOUandluxRforcomparison.Wefoundthattherewasasaturatingrelationshipbetweenthetwoproperties,whereclonesthatproducedmoreproteasealsogrewbetterinM9-caseinmedia.Acertainamountofextracellularpro-teaseproduction,measuringataround50nMpresentinculturesupernatants,corresponded91Figure4.4:ExaminationofgrowthandproteaseproductionofevolvedclonesinM9-casein.Clonesfromdi˙erentphenotypicclasses(gaugedbycolonybioluminescence)weretakenfromexperimentalpopulationsatgeneration870.Theclonesweregrownfor24hoursinM9-casein.Growthwasmeasuredbyabsorbanceat600nminaspectrophotome-ter.ProteaseproductionwasestimatedusingaFITC-caseincleavageassaywithculturesupernatants.Fluorescencemeasurementswerebinnedbyphenotypicclass(toppanel),ornormalizedtoatrypsinstandardcurveandplottedagainstgrowthmeasurements(bottompanel.)Errorbarsintoppanelrepresent95%errorbars.Inbottompanel,circlesrepresentevolvedclonesandtrianglesrepresentcontrolstrains.Thenon-luminescentcontrolstrainusedwasluxRandluminescentcontrolstrainsincludedbothWTandluxOU.92tonearlymaximumgrowthyieldat24hours,andbeyondthatlevelofproductionadditionalbene˝tsingrowthyieldwereminimal.Asexpected,theluxRstrain(Fig.4.4,redtrian-gles)displayedrelativelylowlevelsofbothquantitiesandtheancestralcooperatorstrains(Fig.4.4,orangetriangles)bothpossessedhighlevelsofproteaseproductionandgrowthinM9-caseinmedia.Likewise,allevolveddefectorclonesassayedhadlowerlevelsofgrowthinthemediaandproducedlowerlevelsofprotease.Interestingly,evolvedclonescharacterizedasbrightsordimswerefoundtohaveawiderangeofgrowthandproteasephenotypes.Somehadsimilarlevelsofproteaseproductionandgrowthastheancestralcooperatorstrains,somedisplayedevenhigherlevelsthantheancestralcooperatorstrains,andothershaddecreasedcapacitiesforboth(Fig.4.4,lowerpanel).Thosethathaddecreasedlevelsofeachseemedtoproduceasmallerfractionofproteaseproductionbutexperiencethemajorityofoverallgrowthincomparisonwiththeyieldsoftheancestorstrains.TheclonesthatexhibitedthiscombinationofphenotypesweremainlyclonesevolvedfromtheWTM9-caseintreatment,andpossessedalargeproportionofdimclones.4.4DiscussionInChapter2,weshowedthatQSpreventsdefectorinvasiononshorttimescales.Inthischapter,IextendedthesestudiestoaddresswhetherQSstabilizescooperationinthepresenceofexogenousAIandonlongerevolutionarytimescales.TheseexperimentsfurtheraddressedquestionsaboutthelinksbetweenQSandthecooperativebehaviorsthatitoftenregulates.WhiletheWTstrainofV.harveyihaspreviouslybeenseentoresistinvasionbytheluxRdefector,thatparticularstrainwasalsoobservedtohavepleiotropiccostsandmayhavebeenconstrainedinitsabilitytofurtheradapttotheM9-caseinenvironment(Chapter2).Even93whenweattemptedtocoercetheWTstrainintobehavinglikeaconstitutivecooperatorbyaddingexogenoussignaltotheenvironment,wesawthatWTwasnotoutcompetedinthesamefashionasluxOU,andtheenvironmentevenelicitedselectionfornovelintermediatephenotypes(Fig.4.1).Wewondered,ifdefectorsweregiventheopportunitytoevolvedenovo,ratherthanbegeneticallyengineeredandforciblyintroducedintoexperimentalpopulations,wouldtheyevolveandwouldthisoccuralongthesamepathsandimpactthesamemutationaltargets?Althoughcooperationisatheoreticalchallengeforevolutionarybiologytoexplain(Hamil-ton,1963,1964,Westetal.,2007),weconsideredthepossibilitythatdefectorswouldbeunlikelytoevolveinthepresenceofWTduetotheresultsfoundinChapter2.Thispredic-tionwasonlypartlycorrect,asdefectorsdidevolveandincreaseinthepresenceofWT,butinreducedproportioncomparedtotheluxOUancestor(Fig.4.2,upperleftpanel).AllluxOUpopulationswererapidlyinvadedandultimatelysweptbyevolvedvariantswithreducedQSactivity(Fig.4.2,lowerleftpanel).Intenofthetwelvereplicatelines,thesedefectorswereindistinguishableinphenotypefromtheengineereddefectorstrainsusedascontrols.However,intwootherlineages(CAS16andCAS19),afterappearingtobecom-pletelysweptbydarkdefectors,extremelydimvariantsemergedandproceededtodominatethepopulation.(Fig.4.3).Itwillbeinterestingtodeterminethegeneticnatureofthesedefectors.ThisbroadtrendofsweepsoftheUCbyevolveddefectorscorroboratespastcompetitionresultsindicatingthatluxOUismuchlessadeptatpreventingcheatingbydefectionintheM9-caseinenvironmentthantheWTstrain.Ourresultsalsopredictthatthedimvariantsthatevolvedwere˝tteragainsttheWTancestorthantheluxRstrain,astheywereabletoinvadeWTpopulationsinmanycaseswhiletheluxRstraincouldnot(Chapter2results,Fig.4.1,upperleftpanel).94ThelevelofQSparticipationoftheexperimentalpopulationsmeasuredbypopulationbioluminescenceimpactedgrowththemostintheWTM9-caseintreatmentevolvedfor2000generations(Fig.4.3).ItwasalsothecasethatthetwoluxOUM9-caseinpopulationsthatexhibitedstrongerQSactivitygrewmuchbetterthantheothertenreplicates,whichmeasuredlowinbothbioluminescenceandgrowthinM9-casein.ThetrendbetweenQSparticipationandgrowthyieldswasnotobservedintheasocialM9-CAAcondition.Becauseofthisrelationshipbetweenourphenotypesofinterestandthestrongerdynamictrendsofdefectorevolution,wefocusedmoreofouranalysesonlineagespassagedintheM9-caseinenvironment.Speci˝cally,thestrongdi˙erencesobservedbetweendefectordynamicsinthetwogenotypicbackgroundsinM9-caseinsupportourpriorclaimthatfunctioningQScanhavestrongimpactsonstabilityofcooperativebehaviorsovertimescalesrangingto2000generations,andlikelymuchfurther.Theseresultswereobservedinwell-mixedpopulationsinwhichspatialstructuralwasunlikely(bio˝lmswerenotobserved)andmigrationwasnotpossible,variablesthatwouldbemorecommonplaceinnaturalenvironmentsandcouldfurtherextendthisstabilizinge˙ectQShasoncooperativebehaviors.WhendefectorsinvadedinM9-caseintreatments,populationproductivitywassacri˝ced(Fig.4.2,rightpanels),butfortwodi˙erentreasons.InluxOUpopulations,thisdecreasedproductivityisbecausecooperatorgenotypeswerereplacedbydefectortypeswithlowercarryingcapacitiesinthisenvironment.InWTpopulations,thedensitydecreasedtobelowlevelsneededtofullyinduceQS(Fig.4.2,rightpanels).AdropofdensityintheWTparentgrowninM9-caseinshouldhaveatwo-folde˙ect;itshoulda)leadtolessactivationofQSduetolowerdensities,andb)wehavepreviouslyseenthatinmanycasesQS-defectorsalsoproducelowerpercapitasignallevelsduetoregulatoryfeedbacks(NgandBassler,2009,Chapter2Fig.2.3).Thecapacitytoturno˙QSintheappropriateconditions,particularly95inresponsetodropsindensityand/orrisesindefectorfrequenciesprovidestheabilityforWTtowithstandevenevolveddefectorsfromovertakingapopulation.However,manyWTM9-caseinlineagesdidexperiencesomegainsinpopulationproductivityovertheremainderoftheexperiment(Fig.4.2,lowerrightpanel).Thiscouldbeduetofurtherbene˝cialmutationsbybothcooperatorsanddefectorswithinthesepopulations,andtheextenttowhicheitheroccurswillbeaquestionofinterestmovingforward.Whilewedon'texpectcompleteQSdefectorssuchasluxRtocoerceQSinductionofWTcooperatorsduetotheirloweroveralllevelsofsignalproduction(Haas,2006,CzáránandHoekstra,2009),wewereabletoarti˝ciallysimulatethisscenariobyaddingexogenoussignalmoleculesathighconcentrationstoM9-caseinenvironments.Inthisway,weattemptedtorecapitulatetheconstitutiveQSphenotypeobservedfromtheunconditionalluxOUstrain,whichexpressesQS-associatedgenesregardlessofactualsignalconcentrations,intheWTbackground(WatersandBassler,2006).Interestingly,evenwhenwedothis,theinvasiondynamicsofluxRintoWT-majoritypopulationsarequitedi˙erentfromthoseseenagainsttheUCstrain(Fig.4.1).ThoughwesupplementedthetwoautoinducersresponsibleforthemajorityofQSactivation(autoinducers1and2,HAI-1andAI-2),itappearsthatluxRonlybeginstoinvadeafter2dayswhenpopulationsreachtheircarryingcapacitybeforethe24-hourtransferpoint.Evenmoreinteresting,onceluxRreachesasizableportionofthepopulation,defectorsevolvedfromtheWTbackgroundwererapidlyselectedforandledtoadecreaseinthefrequencyofluxRinthepopulations(Fig.4.1,upperleftpanel).Manyoftheseevolvedvariantstestedharboredintermediatephenotypeswithregardtobothbioluminescence(indicativeofdecreasedQSactivity)andgrowthinM9-casein,mostlylikelyduetodecreasesinproteaseproduction(Fig.4.1,lowerpanels).ThefactthatthesevariantsdidnotcausethesamepopulationproductivitydropasluxR,evenwhenthesetypesdid96becomethemajorpopulationmemberalsosuggeststhattheyharboranintermediateQSphenotype,ratherthanbeingQS-defectorstothesameextentasluxR(Fig.4.1,toppanels).ThissuggeststhatevenifthereisadecreaseinthelevelsofinvestmentintoQSexpendedbythesestrains,thoselevelsarestillmarkedlyhigherthanfullQS-defectorstrains.Therewerealsoalargenumberofdimvariantsthatevolvedduringthelong-termevolu-tionexperiment,andthesewereespeciallyenrichedintheWTM9-caseintreatment,with8ofthose12lineagespossessingsomenotabledimmingorharboringadimtypewithinthesepopulationsbygeneration2000oftheexperiment.Thise˙ectwassubstantiallydiminishedamongluxOUlineages,withonly2lineagesdisplayinganybioluminescence,andthisbio-luminescencewasextremelydi˚culttovisuallydetect.GrowthinM9-CAAalsoappearstohavediminishedthise˙ect,withonly1luxOUand5WTpopulationsdisplayinganydimvariants,twooftheWTpopulationsinwhichagainthebioluminescencewasinconsistentlyobservable.ComparingthegrowthandproteaseproductionofevolvedclonesfromthesepopulationswasinformativeabouttherelationshipbetweenQSinvestmentandgrowthinM9-casein.Whenplottedagainsteachother,itformsanon-linearsaturatingcurve(Fig.4.4,lowerpanel).ThiscurveisreminiscentofdescriptionsofMichaelis-MentonorMonoddynamicsforresourceswelldescribedelsewhere(Tilman,1981).Theedgewheremanydimclonesarelocatedisintheareaofthecurvewhereincreasesingrowthobtainedasafunctionofproteaseproductionbeginstodecrease.ThiscurveisalsoresemblantofaParetofront(SatterwhiteandCooper,2015,Shovaletal.,2012)andsuggeststhattheevolvedWTlineages,especiallythedims,wereselectedfortypesthatweremoreoptimizedtotheexperimentalenvironmentthroughdecreasedinvestmentinproteaseproduction,likelythroughdiminishedQSinvest-ment,butwithoutcompletelylosingallofthebene˝tsofQS,asfullQS-defectorsdo.Inthis97way,evolveddimswithreducedbutnotfully-eliminatedQSactivityshouldbeabletobetterconserveresources,withstanddefectorexploitation,andmaximizegrowthratesinM9-caseinmediacomparedtotheirmorecooperativeancestors.Whilethisresemblesdefection,andthedimvariantscouldpotentiallyexploitmoreaggressivelycooperatinggenotypesinsimilarfashiontotheoutcomebetweenWTandUCstrains(Chapter2),thisprincipallyshouldactasaprocessofoptimizationtotheselectionregime,wherepopulationsthatbestbalancepressurestomaximizegrowthrates,minimizeexploitation(namely,maximizinginclusive˝tness,Hamilton,1964,Westetal.,2012),andmaximizethelikelihoodtogainadditionalbene˝cialmutationsarelikelytoperformwellovertime.Reducinginvestmentincostlycellularprocessessuchasproteaseproductionwouldaidinbothincreasingcellulargrowthratesandinminimizingexploitationbycheaters,butbynotcompletelyeliminatingthisproduction,italsominimizeslossesinpopulationyields,andthusincreasesthelikelihoodofthepopulationacquiringmorebene˝cialmutations(SniegowskiandGerrish,2010).Inthisway,dimsmaybeexpectedtoevolveandtakeoverallpopulationsgrowninM9-caseinifgivensu˚cienttime.Insum,IhavepresentedfurtherevidencethatQScansustaincooperativebehaviorsitregulatesinmixedpopulationswithdefectorsoverlongerevolutionarytimescales.Wesawthatdefectorsevolvedinallgenotypicbackgroundsandexperimentalenvironmentstested,althoughtheyweremoststronglyselectedforinM9-caseinmediaandintheluxOUgeneticbackground.WhiledefectorsdoevolveintheWTlineages,theirpathsaremorevariableandtheyaremuchlesslikelytosweeptheirresidentpopulations,andthegeneralrelationshipbetweenQSactivityandgrowthproductivityofthepopulationsismaintained.SomeofthisresistancetoexploitationmustberelatedtoourpreviousresultsshowingthatQSallowscellstomaximizegrowthratesatlowdensitiesandpostponeproductionofcostlysecreted98productsuntilsignalanddensitylevelsarehighenoughtohelpmediatethatcost.However,wealsosawthatinmanyWTM9-caseinlineages,whileQSwasmaintained(measuredbyluminescenceproxy),theapparentinvestmentlevelsinQSdecreasedthroughevolutionofvariantsthatexhibitreductionsinbioluminescence,proteaseproduction,andgrowthyieldphenotypes(Fig.4.2,Fig.4.3).ThesedimvariantsmaybeoptimizedtotheM9-caseinexperimentalenvironment.Importantly,though,thedimvariantsstillmaintainthecooperativebehaviorandcouldserveasareservoirforthisphenotypeovertime(Pollaketal.,2016).Overall,whiledefectionisthestablestrategyinM9-caseininthepresenceofluxOU,intheWTM9-caseintreatmentmixedpopulationsofcooperatorsanddefectorswerethedominanttrend,andQSandthecooperativetraitsitregulatesdecreasedinbothfunction-levelandfrequencyinthepopulations,butwerenotwhollyeliminated,againdemonstratinghowfunctionalQScanstabilizecooperativebehaviors.99Chapter5QuorumsensinginVibrioharveyiandgrowthpromotedbylowcelldensityregulation100PrefacePreviouschaptersinthisdissertationhaveprimarilyfocusedontheimpactsofVibrioQSonthestabilityofcooperativebehaviors.Thiswasmainlylimitedtostudiesofextracellularproteaseproduction,abehaviorthatispositivelyregulatedbyQSandismaximallyproducedathighercelldensities.However,V.harveyiQSregulateshundredsofgenesinthegenome,manyofwhicharenotpositivelyregulatedbyQS.ThisinincontrasttootherQSsystems,likePseudomonasaeruginosa,inwhichthevastmajorityofgenesregulatedbyQSareinducedathighcelldensity.Manyofthesegenesrepresentintracellularprivategoods,butinterestinglytherearealsogenesthatcontributetotheproductionofpublicgoodsthatarenegativelyregulatedbyQS(i.e.expressedatlowcelldensity).Givensomeofthesetrends,wewonderedifthereareenvironmentalconditionsthatwouldallowQS-defectivestrainstooutperformQS-functionalstrains.1015.1IntroductionAnumberofstudieshaveshowthatQSe˙ectivelyactstomaximize˝tnessbene˝tsathighdensity,e˚cientlytransitionbetweendi˙erentdensitystates,andinmanycasespreferentiallyregulatessecretedproducts(Darchetal.,2012,HenseandSchuster,2015,Heilmannetal.,2015,Popatetal.,2015a,Schluteretal.,2015).Secretedproductsareoftencheatablepublicgoods.However,itisalsotruethattheV.harveyiQSsystemhassigni˝cante˙ectsontheregulationofhundredsofgenes,estimatesrangingashighas10%ofthegenome(SchusterandGreenberg,2006,vanKesseletal.,2013r).Thissetofgenesiswide-spanninginpredictedfunctions,andonlyasmallportionofthesegenesa˙ectpublicgoodsproduction.Onthewhole,itappearsthatQSinVibriosregulatemoregenesathighdensitiesthanatlowdensities,asthelowdensitymasterregulatoraphAregulatesfarfewergenesthanthehighcelldensitymasterregulatorluxR(Rutherfordetal.,2011,vanKesseletal.,2013).Butbothcontributetolarge-scaledi˙erencesinregulationatlowandhighcelldensities.Thepreviousthreechaptersexaminedinteractionswithrespecttoasinglelimitingcarbonsource,casein,wheregrowthispositivelyimpactedbytheabilitytoturnontheQScircuit.ButbecausemanygenesaredownregulatedbyQS,wehypothesizedthatwecouldidentifylimitingcarbonsourceswheregrowthisnegativelycorrelatedwithastrain'sabilitytoturnontheQScircuit.Thisnete˙ectcouldoccurbyrepressingabehaviorathighdensitiesorbyactivatingitatlowdensities.Inenvironmentswithcarbonsourcesthatrequiredthesedownregulatedbehaviorstobeutilized,wehypothesizethatlockedlow-cell-densitymutantswouldactuallyoutperformlockedhigh-cell-densitymutantsinmonoculture,andifgrowthdependsonpublicgoodsdownregulatedbyQS,thentheremaybecheatinguponQS-102defectivestrainsbyQS-functionalstrains,thereverseofwhatwehavereportedinChapters2-4.Toprobethishypothesis,weexaminedanumberofdi˙erentcarbonsourcestoseewhethergrowthwasa˙ectedbytheabilitytoactivatetheQScircuit.5.2MaterialsandMethodsCulturesofVibrioharveyiweregrownM9liquidmediawitharangeofsinglecarbonsources.TheseweregrowneitherinBiolog96-wellformatplatesorinborosilicateglasstesttubesat30degreesCelsiuswith250rpmonanorbitalshaker.DNAprovidedforgrowthwasherringspermDNA(Invitrogen).Atdailyintervals,growthin96-wellplatesweremeasuredinaSpectraMaxM5platereaderatandgrowthintesttubeswasenumeratedbymeasuringopticaldensitiesinaBeckmanCoulterspectrometerorbyviablecellcountsonagarplates.5.3ResultsManycarbonsourceswereexamined,guidedbythoseavailable(viaBiologplates)andwithreasoningledbyknowndownregulatedtargetsfromresultsofmicroarrayandRNAseqstudies(Rutherfordetal.,2011,vanKesseletal.,2013).WewereabletosuccessfullyidentifyseveralcarbonsourcesthatshowedenhancedgrowthforQS-defectivestrainsoverQS-functionalstrains.ThispatternforsuchcarbonsourceswasthatQS-defectivestrainsgrewtoamaximumlevelamongtestedstrainsby24hoursofgrowth,WTgrewslightlyworsetoequallywellastheQS-defectivestrain,andthelockedhigh-cell-densityluxOUmutantgrewsigni˝cantlyworsethanallotherstrainstested.Thespeci˝ccarbonsourcesthatmetthisoutcomeincludedglycerol(Fig.5.1,toppanel),mannitol(Fig.5.1,bottompanel),andglutamate(Fig.5.2).Severalgenesrelatedtoglycerolmetabolismaredownregulatedby103Figure5.1:GrowthofV.harveyistrainsinM9mediawithdi˙erentsugaralcoholsasthelimitingcarbonsource.StrainsweregrowninM9mediawith0.5%glycerol(toppanel)ormannitol(bottompanel)supplementedasthesolecarbonsourcefor24hours.Cellgrowthwasmeasuredinaspectrophotometerbyabsorbanceoftheculturesat600nm.Errorbarsrepresent95%con˝denceintervalsonN=3biologicalreplicates.104V.harveyiQS,consistentwithourobservedresult(vanKesseletal.,2013).Interestinglythough,genesrelatedtomannitolwerepredictedtobeupregulatedbyQS,butgrowthonmannitolstillappearedtobemostfavoredbyQS-defectivestrains.MannitolrepressorproteinispositivelyregulatedbyQS,whichcouldexplainthisoutcome(vanKesseletal.,2013).Glutamateisalsoaninterestingcase,becausethoughitwasprovidedasthesolecar-bonsource,itcanalsobeusedbycellsasanitrogensource.Glutamateisalsousedtogenerateotheraminoacidsinthecell.Infact,ithasbeenestimatedthatasmuchas85%ofcellularnitrogenisderivedfromglutamate,makingitanincrediblyimportantnutrientsource(Magasanik,1993).Thisisbecausethemostimportantpathwayofassimilationistheubiquitousglutaminesynthetase/glutamatesynthetase(GS/GOGAT)pathway(Mer-rickandEdwards,1995).Together,glutamateandammonia,theothernitrogensourcepresentintheM9mediausedforgrowth,arecombinedtoformglutaminebytheenzymeglutaminesynthetase(GS).Thiscompoundisthencombinedwithalpha-keto-glutarateandNADPHtoformtwomoleculesofglutamatebyglutamatesynthetase(GOGAT).Alterna-tively,becauseofthelargeammoniaconcentrationspresentthatmayprovideafeedbackonthiscircuitpreventinge˙ectiveprocessingofglutamate,thiscouldhavenegativeimpactsonusingglutamateasacarbonsource.IthasbeensuggestedelsewherethatQScouldactasacontrolpointforcellse˙ectivelyprocessingnitrogenfromtheenvironment(DeAngelisetal.,2008).ConsistentwithexpectationsbasedonouranalysesofVibrioharveyiproteaseproduction,inBIOLOGplateswefoundthatQS-defectorswereuniversallyimpairedforgrowthondi-peptidesasthelimitingnitrogensourceincomparisonwitheithercoopera-torstrain.ThereweretwoexceptionswhereluxOUwasimpairedmorethantheothertwostrains,glycine-aspartateandglycine-glutamate.TheconstitutiveluxOUappeared105Figure5.2:GrowthofV.harveyistrainsinM9mediawithglutamateasthelim-itingcarbonsource.StrainsweregrowninM9mediawith0.5%glutamatesupplementedasthesolecarbonsourcefor24hours.Cellgrowthwasmeasuredinaspectrophotometerbyabsorbanceoftheculturesat600nm.Errorbarsrepresent95%con˝denceintervalsonN=3biologicalreplicates.inhibitedonanumberofnitrogensourcesrelevanttotheGS/GOGATcycle,includingL-glutamineandL-glutamate.Beyondthesecompounds,luxOUwasalsolesscompetentthantheluxOD47Edefectorstrainwheninorganicsaltslikenitrateornitritewereincludedasthelimitingnitrogensource,suggestingthatluxOUmayexperiencesomedefectsinprocessingnitrogen.Thisalsoincludedtoalesserextentammonia,thelimitingnitrogensourceinM9media.Lastly,luxOUwasde˝cientforgrowthontheDNA-relatedcom-poundsadenineandguanosine.BecauseitwassuggestedthatDNaseproductionwasalsodownregulatedbyQS(BlokeschandSchoolnik,2008),wealsoexaminedgrowthonDNAasasolephosphorussource.Withglucoseheldconstantasthecarbonsource,phosphatewasprovidedtoculturesofcellsfrominorganicphosphatesaltsororganic(DNA)sources.It106Figure5.3:GrowthofV.harveyistrainsinM9-glucosemediasupplementedwithdi˙erentphosphorussources.StrainsweregrowninM9mediawith0.5%glucosemediasupplementedwithDNAorinorganicphosphatesaltsprovidedasthephosphorussourcefor24hours.Cellgrowthwasmeasuredinaspectrophotometerbyabsorbanceofthecultures,anddisplayedastheopticaldensityat600nm(toppanel)orastheratioofabsorbancesbetweenthetwoconditions(growthwithorganicphosphate/growthwithDNA,bottompanel).Errorbarsrepresent95%con˝denceintervalsonN=3biologicalreplicates.107Figure5.4:GrowthofV.harveyistrainsinM9mediawithchitinorNAGsup-plementedasthesolecarbonsource.StrainsweregrowninM9mediawith0.5%(w/v)chitinorNAGsupplementedasthesolecarbonsourcefor24hours.Cellgrowthwasmeasuredbyviablecellcountsonpetriplates.V.harveyistrainsusedwereWT('QS+'),luxOU('HCD'),luxOD47E('LCD'),andluxR('QS-').Errorbarsrepresent95%con˝-denceintervalsonN=3biologicalreplicates.wasobservedthatluxOUexperiencedade˝nitedelayingrowthwhenprovidedDNAasthesolephosphorussource(Fig.5.3).ThisispresumablyduetothenegativeregulationoftwoextracellularDNases,dnsandxds,whichcouldbethoughtofaspublicgoods.ThisisinterestingbecauseifluxOUisde˝cientintheproductionofthesepublicgoods,itlaysoutthepotentialscenariowhereinpairwisecompetitions,luxOUcouldcheatotherstrainswhengrownonDNAasaphosphorussource,astarkchangefromtheoutcomeobservedinM9-casein.Insimilarfashionchitinases,whichcanalsoqualifyaspublicgoods(Drescheretal.,2014),arenegativelyregulatedbyQSaswell(Defoirdtetal.,2010).Chitinisanex-tremelycommonbioticpolymerfoundinaquaticenvironments,consistingofmanyunitsofN-acetyl-D-glucosamine(or'NAG').GrowthonchitingranulesandonNAGwereenhancedinQS-defectivestrains,andmostnotablyluxOUgrewworsethananyofthestrainstested108intheseconditions(Fig.5.4).5.4DiscussionTheresultsofthischapterrea˚rmthatQSisaglobalregulatorysystemthatcontrolsmanybehaviorsbeyondjusttheregulationofpublicgoods.Weidenti˝edenvironmentsandnutrientsourcesinwhichQSactivationisinhibitorytoreachingmaximumgrowthinthatenvironment.ThisnegativeregulationmustbeespeciallystronginsituationswhereWT,whichwethinkbehavesasadensitygeneralistthatcanacclimateandperformwellinmostenvironments(Fig.2.5;unpublisheddata,WillSoto,Waterslaboratory),isinhibitedforgrowthaswasseeninM9-glutamate.Itislikely,andperhapsunsurprising,thatagenecircuitthatimpactstheexpressionofsomanygenesfromthegenomewouldhavestrongimpactsonmetabolism(Davenportetal.,2015),andaswesawthenutrientsourcerevealingthehypothesizedgrowthpatternswerediverseinform.Myresultssuggestthatthelow-cell-densityresponseisbene˝cialfornitrogenmetabolismandcatabolismofseveralnitrogen-richpolymers.Thisstudyalsorevealshowtheparticularcontextinwhichalimitingnutrientisusedcanbeimportant.Forinstance,inearlierchapters,wefocusedonthedynamicsoccurringwhencaseinisusedasthelimitingcarbonsourceinanenvironment.Itispossiblethattheuseofcaseinasthesinglenitrogensourceorbothalimitingcarbonandnitrogensourcewouldshiftthetypesofgrowthoutcomesthatoccur.OfgreatestinteresttoemergefromthisworkistheideathatQSmightbeimpactingnitrogenmetabolism,somethingthathasnotbeendirectlyexaminedinotherchaptersofthisthesis.ItappearsthatthelockedhighcelldensitystraingrowsworsethanWTorQS-defectorstrainswhengrownwithmajorcomponentsoftheGS/GOGATcycleasthe109limitingnitrogensource:glutamine,glutamate,ammonia.Thisstrainalsoappearstogrowworseonnitrateandnitrite,suggestingoverallimpairmentfornitrogenprocessinginthecell.IndependentexperimentsshowedthatgrowthonglucosewithcaseinasthelimitingnitrogensourceledtoyielddecreasesforallstrainstestedexceptluxOU,suggestingthatsomedecouplingoffeedbacksbetweencarbonandnitrogenprocessingoccursinthisstrain.Growthandcompetitionexperimentsbetweendi˙erentQSstrainsgrowingonchitinwereperformedbutnotextensivelycharacterized.Growthongranularchitinpresentscomplicat-ingfactors,withmorespecialrestrictionsimposedinthegrowthmediaandslowergrowthratesoverallforallstrainsduetotheneedto˝rstadheretoachitinparticle.ThegeneralgrowthpatternappearssimilaronitssubunitNAG,withtheUCstrainperformingthepoorestinmonoculture.ThefactthatluxOUgrowspoorlyonthissubunitsuggeststhatitwouldnotbeabletocheatonotherstrainsinthepresenceofthislimitingnutrientsource.However,thisassayevaluatesgrowthproductivitymorethangrowthrate,andonlydirectcompetitionswillrevealwhethercheatingdynamicsoccur.ExtracellularDNAisuniqueamongthesecompoundsbecauseitcanbeutilizedforcarbon,nitrogen,orphosphorus.WehavesofaronlyconductedpreliminaryanalyseswithDNAasthesolephosphorussource.Theresultswerestrikingsupportingtheideathatthelow-cell-densitystateaidsinthebreakdownofpolymericDNAforuptakethroughDNaseproduction,andthatDNasecouldbeacheatablepublicgoodproducedbylowcelldensitycellsthatcouldbecheatedbycellsinahighcelldensitystate.Inthisrespect,itwillbeinterestingtoseewhatcompetitiveoutcomesresultwhenluxOUiscompetedagainstaQS-defectoronlimitingDNA.AnotherimportantcontrolistotestgrowthonnucleotidesratherthanaDNApolymertoevaluatewhetherstrainslikeluxOUareintrinsicallyrestrictedforgrowthonthesetypesofnutrients.BecauseluxOUdidrecoversomewhatingrowthon110DNAby48hours,itispredictedthatthisisanissueofDNaseproductionrateratherthanadefectfornutrientuptake.ImpactsongrowthbythenegativeregulationofQSinVibrioscouldhaveimportantimplicationsforcompetitioninnaturalenvironments,andincasesliketheDNAgrowthresults,shiftourideasaboutwhatgenotypeswillbehaveascooperatorsanddefectors,andthismayshiftgiventheresourceenvironment.Manydi˙erentpolymersshouldbeencoun-teredinheterogeneousenvironments,andamongthemostprevalentthatVibrioencounterinmarineenvironmentsischitin,andwehaveshownthatgrowthonchitinisreducedforstrainswithmoreactiveQSactivity(Fig.5.4).Intotal,theseresultsmayultimatelyhelpexplainthetremendousdiversityobservedamongenvironmentalbacterialiketheVibriosandhelpusworktowardaframeworkforpredictingwhichQSstrategieswillwinoutinvariousconditions.111Chapter6ConcludingRemarks1126.1IntroductionTheoverallaimofthisdissertationhasbeentouseVibriospeciesasamodelsystemtoexamineifandhowchemicalcommunicationstabilizescooperativesocialbehaviors.MycentralhypothesiswasthatQScanactasastabilizingmechanismforcooperativebehavior.The˝ndingspublishedhereextendourunderstandingoftherolesthatQSplaysincells,tonotonlymediatedensity-dependentchangesingeneexpression,buttoalsoappropriatelyaligngrowthstrategiestotheenvironmentofapopulationandtherebyprovide˝tnessgainsandstabilizationofQS-controlledbehaviors.MyresultsdonotimplythattheroleofQSisrestrictedtoregulatingcooperativebehaviors,oreventhatitevolvedforthispurpose(thoughitdoesseemtobeoneofthestrongadvantagesofhavingaQSsystem),butratherQSislikelya˙ectingawiderangeofbehaviors,bothpublicandprivate,inVibrios.6.2StabilizationofVibriocooperativeproteaseproductionbyQS6.2.1ConclusionsAsseenthroughoutChapters2ofthisthesis,intheVibriosexamined,havingfunctionalQSpositivelyimpactsthestabilityofthecooperativebehaviorofproteaseproduction,eveninconditionsthatwouldgenerallybepredictedtodisfavorthemaintenanceofcooperation.Moregenerally,italsoin˛uencescellular˝tnessonawiderangeofgrowthsubstrates.WedemonstratedthattheproteaseproductionofV.harveyiispotentiallyexploitable,butthatthefunctionalQSsystemofthewildtypestrainactstominimizecheatinginthesocialM9-113caseinenvironment,inmanyinstancescompletelyabolishingit,byrestrainingexpressionofproteaseproductionuntilhighdensityafterotheravailablenutrientshavebeendepleted.BecauseQSisquiteubiquitousamongbacteria(LaSarreandFederle,2013,vanKesseletal.,2013)andlargeQSregulonsarenotuncommon(Chuganietal.,2012,Majerczyketal.,2014,SchusterandGreenberg,2006),weanticipatethisresultisgeneralizable.Broadlyspeaking,furtherexamplesofcommunicationstabilizingcooperativebehaviorsinmanyotherorgan-ismswillemerge,inmanycasesbeingarequirementforthestableexistenceofcooperation.Whilehistoricallymodelersinevolutionarygametheoryandotherbiological˝eldstendedtothinkofcooperationanddefectionintermsof˝xedstrategies,thisislikelynotthecommonnormforthemajorityofnaturalsystems(Axelrod,2006,Doebelietal.,2004,Killingbacketal.,1999,KillingbackandDoebeli,2002).Especiallywhenagivenbehavioriscostly,us-ingmoreinformationtoguideexpressionofthatbehavior,suchassurroundingcelldensity,wouldbebene˝cialtorespondconditionallyinadirectorprobabilisticway(Anetzbergeretal.,2009).Inaddition,wedemonstratedforthe˝rsttimethatQSitselfisanadaptabletraitandcanbemodi˝ed,andnotlostfrompopulations,todecreaseinvestmentincoopera-tiontomoreappropriatelygrowandpreventdefectionoverlongtermevolution.Therefore,IconcludethatQShelpstostabilizecooperativebehaviorsoverlongertimescales(BrugerandWaters,2015).MosthistoricalworkdoneonQSandcooperationhasshownthatevenWTstrainswereinvadedbymasterregulatormutantstrainsanalogoustotheluxRdefectorstrainthatweemployed(Diggleetal.,2007,Katzianeretal.,2015,Sandozetal.,2007).Thismakesoursystemuniquewithrespecttohowresistantitwastodefectorinvasion.Theparticularreasonsforthisdi˙erencelikelyre˛ectthearchitectureoftheQSsystem,andotherdetailsofthegenome,andwillbeworthwhileforfutureexamination.However,wepredictthat114morecasesinwhichthebene˝tsofpossessingregulationbyQSoveranunregulatedstrategywillemerge.WhilesomemodelingandsimulationworkhavepredictedthatQScoulddiminishex-ploitationbycheaters(Allenetal.,2016,Schluteretal.,2016),thisisthestrongestempiricalevidencetosupportthisclaimtodate.Ourexperimentalsystemishighlyadvantageousbe-cause:a)thegeneticsunderlyinghavebeenwell-described,b)thetightlyQS-regulatedbioluminescencegeneexpressionservesasabuilt-inreporterofQSactivity,andc)wehaveanengineeredstrain(luxOU)thatactivatesQSconstitutively,butnotatmaximumlevelsdi˙erentfromtheQS-inducedWTstrain,whichallowsustoaskspeci˝cquestionsaboutthevalueofQS-regulationovertheexpressionofitsregulon.Asaresult,wewereabletotestourhypothesisinamannerthathasnotbeenpossibleinotherexperimentalsystemswhereap-propriateconstitutivestrainshavenotbeenidenti˝ed.AsQShasalsobeenreportedtoactasareporterofkinship(Schluteretal.,2016),ourresultsalsoagreewithlong-standingknowl-edgethatkinselectioncanstabilizecooperationinotherorganismssuchassocialanimals(Nowak,2006,Westetal.,2007,Hamilton,1964).BecauseQSisveryubiquitousinbacteria,ourresultsareverylikelygeneralizabletootherorganismsandcooperativebehaviorsbeyondproteaseproductiontoexplaintheirstabilityinnaturalcommunities.Furthermore,ourre-sultslendsupporttotheideathattheinformationprovidedbycommunicationiscriticaltostabilizingcooperativebehaviors.Inthefuture,itmaycometobeseenasthenormthatnotonlyarecooperativebehaviorscommonlyregulatedbycommunicationsystemssuchasQS,butthatcommunicationisrequiredtomakecooperationevolutionarilystable.1156.2.2FuturedirectionsAnumberofquestionsremainabouttheimpactsofsocialandnon-socialbehaviorsinpreventingcheatingontheWTstrain.Thiscouldbeapproachedbyperforming˝tnessassaysinawiderangeofenvironments,fromthosewhereabsolute˝tnessisheavilydependentontheQSsystem(e.g.M9-caseinorM9-BSA)tothosewhere˝tnessdependslessonQS(e.g.M9-tryptone,M9-CAA,orLB),andalongagradientwherethesemediasaremixedatdi˙erentratios.ParticularlyinM9-BSA,growthdoesnotrapidlyreachhighdensitiesandthereappearstobeverylowlevelsofavailablenutrientswithoutproteolysissothismayconstituteanenvironmentofextreme˝tnessdependenceonacooperativetrait.ItwouldalsobeinformativetodoexpressionanalysissuchasRNAseqonthedi˙erentstrainsusedinthesestudiesgrowninM9-caseinmediumtoseethedi˙erencesinexpressionandwhethertheyarelimitedtomainlycooperativegoodgenesorextendbroadly.Becausewehaveobservedpleiotropice˙ectswhentheQSsystemismutated,theexpectationisthattherewillbebroad-reachinge˙ectsingeneexpressionchangesthatwillalsoincludemanyintracellularprivategoodsanda˙ectothercentralcellularprocessessuchasmetabolism,anditwillbeparticularlyinterestingwhichimpartthestrongeste˙ectsinourexperimentalsocialconditions(M9-casein).Genesthatspeci˝callyimpactinteractionsbetweencellsinapopulationareanotherpotentialtarget.Forexample,TypeVIsecretionhasbeenstudiedinotherbacteria,includingotherVibriospecies,butithasnotbeenextensivelystudiedinV.harveyiandcouldplayaroleintheinteractionsthatoccurbetweencooperatorsanddefectorsifQSatalla˙ectstheexpressionoftheT6SSgenes(ShaoandBassler,2014,Majerczyketal.,2016).WhileQSprovidedstabilitytothemaintenanceofproteaseproductionoverarangeof116densitiesinclosed-systembatchcultures,thereasonsforthisabilitytopreventdefectorinvasionremainslightlyunclear.Ourunderstandingsuggeststhatatlowdensities,theWTstrainisabletoachieveequivalent˝tnessagainstdefectorsbecauseitsaveson˝tnesscostsbyturningQSo˙.However,thenullexpectationwasthatdefectorswouldstillreceiveaddedbene˝tsinthepresenceofcooperators,includingWT,athighcelldensities.Indeed,evenincompetitionswithWTinM9-casein,therearehintsthatinsomecompetitionreplicatesthedefectorstrainbeginstodoslightlybetterathigherdensities(Fig.2.5fromChapter2),thoughnotenoughsotomakegreatgainsbeforepopulationsreachcarryingcapacity.Testing˝tnessesinbatchculturesathighcelldensitywouldbedi˚cultbecauseveryfewdivisionscouldoccurbeforepopulationcarryingcapacitiesarereached.However,itwouldbeinformativetolookatanalogouscompetitionsinthesamemediaenvironmentsunder˝xeddensityconditionsbyuseofeitherchemostatsorturbidostats.MypredictionforsuchexperimentsisthatQSprovidesprotectionagainstcheating,especiallyatlowdensitieswhereQSisturnedo˙,andthatthisadvantagewoulddiminishasthedensityincreases.ParticularlyifconditionscouldbeestablishedwherethelongtermsteadystatedensityisabovethatrequiredforQS-induction(e.g.>108CFU/mL),defectorsmaybeabletoinvadeeventheWTstrain.Butagain,thismaybedi˚culttocarryout,asinvadingdefectorswouldbepredictedtocausepopulation-widedropsincelldensitiesinenvironmentswheregrowthyieldsdependonQS,suchasinM9-casein.Wehaveexamined˝tnessoutcomesoverafairlywiderangeoffrequencies,butoveranarrowerrangeofstartingdensities.Thedynamicsthatoccurredbetweencooperatorsanddefectorsatdi˙erentstartingpopulationsizeswereonlyjustexplored(Fig.2.5).Expandingsimilarexperimentstoawiderswathofdensitieswouldallowmorecompleteconstructionofa˝tnesslandscapeasafunctionofdensityandfrequencybetweencooperatoranddefector117strains,andadeeperunderstandingofalltheconditionsinwhichonestrategyortheotherisfavored.Again,theseoutcomesweremainlyrestrictedtoM9-caseinmediawhere˝tnessoutcomeswerecloselylinkedtoasocialbehavior;thiscouldalsobeexpandedtoexamineenvironmentsandgrowthsourceswhicharepredictedtodependless(ornotatall)onsocialtraits.Theseexperimentswereconductedusingasinglesubstrateasalimitingcarbonsource.Itwouldbeinterestingtoseeifsimilardynamicspersistwhenacooperativetraita˙ecting˝tnessisrelatedtoalimitingnitrogenorothernutrientsource.Caseinisauniquecompoundinthisrespectasitcouldbeemployedinagivenmediaasacarbonornitrogensource.Theseexperimentalregimescouldbeconductedinisolation,orevenbeoscillatedovertimetoshiftselectionpressures,suchasswitchingbetweencaseinasthelimitingcarbonsourceandthelimitingnitrogensourcebytheadditionofadditionalnutrients.Ofinteresttothispoint,whileweobservedsimilargrowthoutcomesbetweenWTandtheUCstrainsofV.harveyiinM9-caseinmedia,preliminaryresultssuggestdi˙erencesinM9mediawithglucoseasacarbonsourceandcaseinprovidedasthelimitingnitrogensource(Fig.6.1),withUCachievingdensitiesasmuchasanorderofmagnitudemorethantheWT(andmorethanwhencaseinisthesolecarbonsource).Alternatively,QS-defectorsdoworseinthisconditionthaninM9-casein.ThissuggestssomeinterferencebetweenuptakeofthesenutrientsourcesunderthelowcelldensityregulationprogramthatisalleviatedwithQSactivationathigherdensities.Ourinitialexaminationintothecompetitiveoutcomesbetweencooperatorsanddefectorsfocusedprimarilyondi˙erentQSgenotypesinM9-caseinmedia.Analternateapproachistoplacetheexpressionofthepublicgood,theextracellularprotease,outsideoftheQSregulon.IattemptedthisexperimentbyoverexpressingthemetalloproteasegeneVIBHAR_RS11785intheWTstraintomimictheconstitutiveluxOUmutant,butthis118Figure6.1:GrowthofV.harveyistrainsinM9-caseinmediainthepresenceandabsenceofglucose.StrainsweregrowninM9-caseinmediawithorwithout0.5%glucosesupplementedfor24hoursandenumeratedwithviablecellcountsonLBpetriplates.Errorbarsrepresent95%con˝denceintervalsonN=3biologicalreplicates.experimentledtoseveregrowthde˝ciencies.Analternatestrategycouldmovetheproteasegenetothegenomeundercontrolofaconstitutivepromoterinordertoavoidtheselectionandcopynumberissuesassociatedwithaplasmid.Additionally,itwouldbeusefultocreateadeletionstrainofVIBHAR_RS11785tohelpisolatethe˝tnesse˙ectsthatarespeci˝callyduetoproteaseproductionfromthosethatareduetoothergenesunderQScontrol.However,thereisanotherprotease(peptidaseM4)encodedintheV.harveyigenomewithhighidentitytotheproteaseencodedbyVIBHAR_RS11785(hemagglutinin),whichmaycontributetotheoverallproteasephenotypeofastrainandalsoexplainwhywedidnotobservespeci˝cprotease-negativevariantsinourevolutionexperiment.WhileinmostcasesIhadareadilydistinguishablephenotypetodi˙erentiatebetweencooperatorsanddefectors,suchasbioluminescenceorcolonymorphology,inothercondi-119tions,theremaybesubtlerorunobservablephenotypicdi˙erencesinthesestrains,orstainsmayhavesimilarstartingphenotypes.Onecommonexamplefromthisworkisindi˙er-entiatingWTandluxOUV.harveyistrains.BecausebothcapablyactivateQSathighdensities,theycannotbedi˙erentiatedbycolonybioluminescencephenotype.Historically,thesehavebeendi˙erentiableonplatesduetotheluxOUstrainharboringachlorampheni-colresistancecassetteinitsgenome.Duringthecourseofmywork,IhavealsointroducedalacZcassettetoaneutralgenomicsiteviaaTn7vectorsystemwhichcouldalsobeusedtodistinguishthesestrains.Thisisusefulbecauseitsimpli˝esidenti˝cationofcompetingstrains,butthissystemcouldbeextendedtomoreuses.Speci˝cally,itwouldbemostusefultotagdi˙erent˛uorescentproteinmarkersusingthissamesystem,whichwouldthenallowustodi˙erentiatelargenumbersofcompetingstrainsinhigh-throughputfashionvia˛owcytometry.6.3Simpson'sParadoxinVibriobacteria6.3.1ConclusionsInChapter3ofthisdissertation,weextendedpreviousstudiesconductedinwell-mixedliquidculturestoexaminewhetherwecouldidentifyconditionswherecooperationwouldbefavoredandcouldincreaseinthepresenceofdefectors.ItwaspreviouslydemonstratedusingmodelingandsyntheticallyengineeredsystemsthataSimpson'sparadoxcouldleadtodynamicsthatfavoredcooperationattheglobalpopulationleveldespitecooperatorshavinglowerrelative˝tnesseslocallyinthepresenceofdefectors(Chuangetal.,2009,Cremeretal.,2012).Thisoccurredbecausedi˙erencesbetweencooperatoranddefectorgrowthoutcomes,120wherecooperatorsexperiencedeitherincreasedgrowthrateorgrowthyieldthatresultedinenhancedgrowthforgroupsthatcontainedhigherproportionsofcooperators.Thesestudiesutilizedametapopulationapproachthatweadoptedforuseinoursystem.Wefoundthatwhenweappliedstringentbottleneckstoourmetapopulations,weobservedincreasesofcooperatorstrainsofbothV.harveyiandV.cholerae,evenfortheunconditionalcooperatingstrainsused.Thisservedascon˝rmationofthepreviousresults,andextendedthemfurtherbyapplyingtheapproachtoanaturallyevolvedQSsystem.These˝ndingsareinsupportofthewell-establishedideathatassortmentofcooperatorsfromdefectorscanaidinthestabilizationofcooperation(NowakandMay,1992,KillingbackandDoebeli,1998).ItwasstrikingthateventheUCstrainsdramaticallyincreased,even˝xinginsomeex-perimentalmetapopulations,andledustoexploretherangeofconditionswhereboththeWTandUCcouldinvade.Wenextexaminedanoppositeextremethatinvolvedmixingandseedinglargenumbersofcooperatorsanddefectorsinaviscousmediathatwouldallowdis-persalandpossibleseparationbetweenstrains.Underthisregime,wedidindeedseestrongdi˙erencesbetweencooperatorstrainsforbothVibriospeciesexamined.Inbothcases,theUCperformedworseagainstdefectorsthanthecorrespondingWTstrain,particularlywhenfrequenciesweremeasuredfromthebulkpopulation.BothWTstrainsperformedespeciallywellonthecolonyedge,wheretheycouldbefoundtobethedominantfractionofcellspresent.Whenentirecoloniesweremixedandenumerated,theWTV.choleraestrainwasfoundtohavesigni˝cantlyincreasedfromstartingfrequencies,demonstratingthatwecouldrecreatetheoutcomeseeninmetapopulationsontheseplates.TheWTV.harveyistraindidnotchangefromitsstartingfrequencyinthepopulation,possiblyasaresultoftheinabilityofV.harveyistrainstoformrobustbio˝lmstructuresthatcouldleadtoseparationofcompetingstrains.1216.3.2FutureDirectionsWhiletheinvestigationintotheSimpson'sparadoxsuggestedthatprovidingassortmentbetweencompetingstrainscouldenableanadvantageforcooperatorsinmetapopulations,thisstudyalsointroducedothercomplicationsthatcouldusefurtherclari˝cation.Specif-ically,whiletheexperimentsconductedinChapter2usedisolatedpopulationswithnomigrationbetweenthem,oneofthekeystepsinthemetapopulationapproachincorporatedinChapter3wasmixingofsubpopulations,whichintroducedmigrationbetweenlineagesinadditiontoimposingbottlenecks.Toseparatethesetwophenomenon,theseexperimentscouldbecompletedwithalteredmigrationregimes,byeitherallowingdecreased,non-globalmigrationorimposingnomigrationbyonlybottleneckingwithinindividualsubpopulationsandnomixingbetween.Thislastapproachwouldhavethedrawbackoflikelylimitingco-operatorincreasesbecausethemetapopulationswerestartedwithamajorityofdefectorsandsubpopulationsthatstartwithdefectorsarelikelytoremaindominatedbydefectors.Becauseofthegeneticnatureofthestrainsusedinthesestudies(mutantshavegenedele-tions),itispossibleforcooperatorstrainstospontaneouslymutatetoadefectorstate,butintheabsenceofhorizontalgenetransferitisimpossibletogetrevertantsofV.harveyiluxRorV.choleraehapRmutantstrainsthatswitchtoacooperativestate.Thus,thebestapproachtogetatthisquestionwouldbetousearangeofsub-globalmigrationrates.Wepredictthatthedi˙erencesbetweenV.choleraeandV.harveyiWTstrainsinmotilityplateswasdueinparttodi˙erencesintheabilitytoformbio˝lmstructures,andmorerigoroustestingofthispredictioncouldbeaninformativeexperimentalfollow-up.TheV.harveyistrainandderivativesuseddonotformdetectablebio˝lmsduetoinsertionsequencespresentingenesrequiredforbio˝lmformation(unpublisheddata,Waterslaboratory).To122examinethisprediction,afollow-upsetofcompetitionscouldbeperformedwithV.harveyistrainsthatareexpressingbio˝lmgenesfromV.cholerae,eitheronaplasmidorfromaneutralsiteinthegenome,whichcanbetaggedusingaTn7-derivedsystem(Choietal.,2005).AnalternativeapproachcouldbetouseaV.choleraedefectorstrainthatdoesnotproducebio˝lms.ThehapRstrainusedisknowntobeastrongbio˝lmformer(HammerandBassler,2003).Whilebio˝lmformationdidnotseemtomakeastrongdi˙erenceinthisenvironment,asvpsLstrainsthatdonotformbio˝lmsstillperformedwell,allofthesestrainswerecompetedagainstthehapRdefector.Togetmoredeeplyatthisquestion,theseexperimentswouldhavetoberepeatedwithahapRvpsLdefectorstrain.ThismutantstraindoesnothavethedistinctiverugosecolonymorphologyofthehapRstrain,andcurrentlywedonothaveamarkerinplacetodi˙erentiateitfromthecooperatorstrains.Inordertocarryouttheseexperiments,strainsmust˝rstbeengineeredtoharboragenethatwouldallowustodistinguishcompetitorsonpetriplates,suchaslacZortheV.harveyiluxgenes,orbyusing˛uorescentproteinencodinggenesthatwouldallowdi˙erentiationonanindividualcellbasisina˛owcytometeroratthebulkpopulationlevelinaplatereader.WhilewewereabletoobserveaSimpson'sparadoxwiththemetapopulationapproachused,andalsorecapitulatedthatresultforV.choleraeonmotilityplates,therangeofpa-rametersforwhichthisworksisstillnotwellde˝ned.WecorrectlypredictedandobservedthatwhileUCstrainsperformedwellinmetapopulations,theydidmuchworsethanWTstrainsinthelargemixedpopulationsinitiatedonmotilityplates.However,atwhichpop-ulationsizetheoutcomesbegintodi˙erbetweenWTandUCstrainsisstillunknown.Tomakepredictionsaboutwherethismightoccur,itwouldbeadvantageoustoinitiallyexam-inethisinasystemthatiseasiertomanipulateandawiderparameterrangecanmoreeasilybeinvestigated.WeproposecarryingoutanalogousexperimentsusingtheinsilicoAvida123evolutionaryplatformtobetterunderstandwherethedi˙erencesbetweenquorumsensingandunconditionallycooperatingstrategiesmayoccurandbemostimportantforcompetitiveoutcomes.6.4EvolutionaryoutcomesofVibrioharveyi6.4.1ConclusionsWhenwedidnotseetheinvasionofWTquorumsensersbyourengineereddefectorstrainthatwaspredictedfromthebroaderliteratureonmicrobialsocialevolution,wenextexaminedtheevolutionarytrajectoriesofVibrioharveyiintwolaboratoryenvironments,onemeanttobesocial(M9-casein)andanothermeanttobenon-social(M9-CAA)inanattempttoplacedi˙erentselectivepressuresonthefunctionofQS,andintheWTstrainandtheconstitutiveluxOUstrain.Toourknowledge,thisisthe˝rstlongtermevolutionexperimentthataimedtoprobethee˙ectsofQSontheregulationofcooperativebehaviors.Overtime,wesawthatQS-defectorsdidevolveinalltreatmentgroups,andthatthise˙ectwasmorestriking,consistent,andrevealedthestrongestdi˙erencesbetweenancestralgenotypesintheM9-caseinenvironment.WhiledefectorsalsoevolvedintheWTM9-caseinpopulations,thedynamicsweremorevariedbetweenreplicatepopulationsandallwereverydi˙erentfromtheluxOUM9-caseinpopulations.ThedefectorsweretoleratedbetterbyWTthanluxOUanddidnotrapidlysweeptheirresidentpopulations.ThusthedegreeofdiversityinQSfunctionwithintheexperimentalpopulationswasmaintainedathigherlevelsinWTpopulations,withmultiplelineageswithdi˙erentlevelsofQSactivityabletopersisttogetherovertime.Asaresult,themajorityofWTpopulationswerenotsweptby124defectors,andinfactmaintainedhighfrequenciesofcooperatorswithintheirlineages.Therewerealsoafewinterestingexceptions:oneWTlineagethatlostcooperators(CAS09),anotherthatseemedtobeintheprocessoflosingcooperators(CAS07),andtwoluxOUpopulationsexperiencedhighlevelsofgrowthandextremelylow,butstilldetectable,levelsofbioluminescence(CAS16andCAS19).Theseresultsdemonstratethat,whiletherewerede˝niteoverarchingtrendsobservedintheexperiment,therewasalsodiversityintherangeofpopulationoutcomesobserved.6.4.2FutureDirectionsThescopeofpotentialdirectionstoextendandfollow-uponthisinitialexperimentisextensive.Wholeresearchprogramshavebeenestablisheduponevolutionexperiments(Lenskietal.,1991),andtherearemanyuniquespeci˝cquestionsthatcouldbeaskedaboutQSandVibrios,andmoregeneralquestionsaboutmicrobialevolutionasawhole.Ourremainingquestionsrelatetoboththegenotypesandenvironmentsused(CooperandLenski,2010,Blountetal.,2008),andthedynamicsandinteractionsoccurringwithintheexperimentalpopulations(Renduelesetal.,2015,Fiegnaetal.,2015,Martinetal.,2016).WehaveconductedgenomicsequencesoftheevolvedM9-caseinpopulationsandareintheprocessofdeterminingwhatevolutionarypathsdi˙erentlineageshaveundertaken(BarrickandLenski,2013).Manyquestionsremainabouttheresultsofthisexperiment,including:1.Dodi˙erenttypesofmutationsunderliethedefectorsthatevolveinthedi˙erentgeno-typicbackgrounds?Determiningthiswouldtellusmoreaboutwhetherthegenotypesusedaredi˙erentlyconstrainedintermsofthebene˝cialmutationsthatcanoccur.2.Isthereparallelismthatoccurredintheevolutionofdefectorsthatoccurredwithinand125sweptluxOUM9-caseinpopulations,occurredwithinbutdidnotsweepWTM9-caseinpopulations,oramongthedimvariantsthatevolvedespeciallyoftenwithintheWTM9-caseinpopulations?Knowingthiswouldtellusmoreabouttherangeandtypesofgeneticchangesthatcangiverisetothesephenotypes,andpotentiallyexpandourknowledgeofwhatgenesareimpactingtheQSsystemofV.harveyi.3.Fromwhatwelearnofthesequencingprocess,wecanexaminemorecloselywhatisoccurringwithinparticularexperimentalpopulations.Thedimminge˙ectthathasoccurredinmanypopulationsdidnotobservablyhappeninallpopulations.Isthisamatterofthesepopulationsnotsamplingadimmutationyet,orbecausesome˝ttervariantevolvedinstead?ItmightbepredictedthatQSallowscellstoactasgeneralistsintermsofdensity-dependentstrategies.Doesthisturnouttobethecase?Dopopulationsthatlackdimvariantsinsteadhavetwocoexistingspecialists,adefectorandacooperator?Howdoestheoccurrenceofdimsimpacttheprevalenceofmorenon-functionaldefectorswithinthesepopulations?Forthedimsourcemutationsthatarediscovered,thesecanthenberemadeinourcontrolstrainstoseehowmuchofthechangeobservedin˝tnessisduetothesespeci˝cmutations.Dimscanalsobecompetedagainstengineereddefectorsandancestralcooperatorstrainstodeterminehowtheyperforminthepresenceofeither.MypredictionisthatdimswilloutcompeteallancestralcooperatorstrainsinM9-casein,andpossiblytheluxRdefectoraswell.Inanycase,IwouldnotexpecttheevolveddimvariantstoperformworseinthepresenceofdefectorsthantheWTstrain.WehavethusfarprimarilylimitedourgenomicanalysestotheM9-caseintreatments.ItwouldshedfurtherlightontheselectivepressuresontheV.harveyiQSsystemindi˙erentenvironmentalconditionsbyconductingsimilargenomicanalysesofthepopulationsthatwereevolvedintheM9-CAAenvironment.MypredictionthereisthatQSshouldbeunderlessstringentselectioninM9-CAAandthus126thatchangesthatoccurinthegenomewillbemoreoftenunrelatedtotheQSsystem,andwhentheyare,arelikelytobelessparallelandconsistentinnature.WehavemadetheclaimthatdimsareoptimizingtotheM9-caseinenvironment,andthisissupportedbythediminishedbutstilldetectablelevelsofproteaseproductionandbioluminescenceinthesestrains,andintheincreasedgrowthpotentialinM9-caseintheypossessovercompleteQS-defectors.Iftheexperimentwerecontinued,itispossiblethatdimscouldevolveinallpopulationsthatdidnotcompletelylosecooperation.Aninterestingextensionofthisquestionis:willdimscontinuetobeselectedfordecreasedinvestmentinQS,orwillthatlevelstabilize?Thiscouldpossiblybemodeledfromanadaptivedynamicsorgametheoreticalstandpointtomakeadvancedpredictionofwhatwecouldexpecttoseebyextendingtheevolutionexperiment.WealsomadetheclaimthattheevolveddimsanddefectorsexhibitedgeneraldropsinQS-regulatedbehaviorsandshoulddisplayacorrespondingdropinsignalproduction.Thisstillneedstobeexperimentallymeasured,eitherdirectlywithmassspectrometry,orindirectlybyseeinghowwellthesestrainsinducebioluminescencefromtheSNreporterstrain.Itispossible,butnotlikely,thatsomestrainscouldstillproducehighersignallevels,andtherebyelicitexpressionofcooperativebehaviorsfromothersintheirpopulationsinamannerre˛ectiveofdeceitfulcheaters(Ghouletal.,2014,Backwelletal.,2000).Di˙erentselectionregimesshouldalsoleadtodi˙erentevolutionarytrajectories,espe-ciallyrelatedtoQSandindi˙erentQSancestralgenotypebackgrounds.ThiscouldbeselectioninM9mediawithcaseinasasolenitrogensourceorasasolecarbonandni-trogensource,selectioninamorestringentconditiondependentonproteaseactivitysuchasM9-BSA,orinalternatingnutrientconditions.Particularlyinvariableandshiftingen-vironmentalconditions,wepredictthattheWTstrainwithfunctionalQSwillbemore127evolutionarilystablethananyengineeredmutantstrainswithalteredQSfunction.Theobservedevolvedvariantswereselectedforinthepresenceofabroaderpopulationofdi˙erenttypes.Itwouldbeinterestingtoseehowevolutionwouldproceediftheywereisolatedfromtheirresidentpopulationsandusedtoformnewexperimentalpopulations.TheselectionfordecreasedQSinvestmentmightbelightenedinthiscase,andfurtherdefectorvariantsmaynotrapidlyevolve,againsuggestingthatthesestrainsareoptimizingtotheexperimentalenvironmentandmightbeapproachinganevolutionarilystablestrategySmith,1982).Byalteringexperimentalenvironmentsfromtheoneinwhichpopulationswereevolved,forexamplebyevolvingM9-caseinpopulationsinM9-CAAandviceversa,wecouldexam-inewhetherevolutioninasocialcondition(M9-casein)leadsto˝tnesscostsinanasocialcondition(M9-CAA)andwhetherthisrelationshipisreciprocal.Thiscouldrevealwhethertrade-o˙soccurwhencellsinvestmoreinsocialbehaviorsversusnon-socialbehaviorsoverthecourseofevolution.Mostevolvedpopulationshaveamixedcomposition.Whencompetedagainstanances-torstrain,Ipredictthatthedefectorspresentwouldbene˝tthemost,butthisneedstobevalidatedthroughin-depthexaminationofcompetitionsofalltheevolvedlineages.Addi-tionally,itisunclearwhethertheselineagesdisplayaconstantlyincreasing˝tnessagainstancestors(Wiseretal.,2013),orwhetherthecooperator-defectordynamicspresentwithinthepopulationsmeanthat˝tnessisonlyincreasedrelativetotheimmediatelyprecedingpopulation.Thiscanbeexaminedbycompetingpopulationsamplesfromdi˙erenttimepointsagainsttheancestoraswellasagainstoneanother.Wecouldexaminepotentiallyevolvedinteractions,byexaminingthegrowthperformanceofisolatedphenotypescomparedtomixedpopulations.Evolvinginteractionscouldbeei-128thersynergistic,signalingtheinfancystagesofmutualisms,orantagonistic,suchasoftenobservedred-queendynamics(Pfei˙erandBonhoe˙er,2004,Ewald,1987,Dercoleetal.,2006).Whenmultipletypescoexistforextendedperiodsoftime,itispossiblefornovelformsofcooperationsuchascross-feedingtodevelopandenhancetheoverallgrowthofallpopulationsTheexceptionalpopulationcasesthatdonotre˛ectthemoregeneraltrendsarealsoquiteinteresting.ItispossiblethatintheCAS09lineagewhereWTcooperatorswentextincttherejusthappenedtobemorebene˝cialmutationsthatdefectorsacquired,orpossiblyadefectorvariantevolvedthatwasbetteratexploitingWTbutrarertoevolveduetoarestrictedmutationaltarget.ItisalsointriguingthattheluxOU-evolvedCAS16andCAS19populationsgrewverywellinM9-caseindespiteexhibitingextremelylow(butstilldetectable)levelsofQSinvestment.Identifyingandrecreatingthesevariantswillshedlightuponthecausesoftheirsuccessinthisexperiment,andperhapstellusmoreabouthowQSdoes(andsometimesdoesnot)provideadaptiveadvantagesinthisM9-caseinenvironment.6.5Low-cell-densitybehaviors6.5.1ConclusionsQShashistoricallybeenthoughtofasadensity-dependentformofcontrolovergeneexpression,withparticularemphasisonturninggenesonappropriatelyoncepopulationsreachhighcelldensities(Swiftetal.,2001,Fuquaetal.,1994,Darchetal.,2012).However,theQSregulonsareoftenextensive,andalsocausechangesingeneexpressionatlowcelldensities(vanKesseletal.,2013,Rutherfordetal.,2011).Thisregulationatlowcelldensities129shouldalsocontributetocellular˝tnessinnaturalenvironments.Weidenti˝edseveralconditionswhere,ratherthanobservingthatQS-defectorsexhibitnegativepleiotropice˙ectsasinenvironmentssuchasM9-caseinandM9-CAAmedia,itistheQS-cooperatorsthatappeartoexperiencegrowthdefects.Morespeci˝cally,theconstitutiveluxOUstrainofV.harveyithatislockedinahighcelldensitystateofQSregulationperformedpoorlyonsomelimitingcarbonsources,whiletheWTstraindoessimilarlywelltoQS-defectorstrains.Thesenutrientsourceswerediverse,rangingfromsugaralcoholsincludingglycerolandmannitol,theaminoacidglutamate,tovaryingpolymersthatrequirepublicgoodsforbreakdownintodigestiblesubunits.Especiallyforpolymerbreakdown,itwouldbeveryinterestingtocon˝rmthatthepublicgoodenzymesresponsibleforthisdegradationaredownregulatedbyQS.Ifcon˝rmed,thiswouldpredictthatinsomeenvironmentsopactuallybehaveasdefectorsandcouldcheatthatareactuallyproducingmoreofagivenpublicgoodmolecule.6.5.2FuturedirectionsWhileitisclearthatsuchenvironmentsexistwherebytheluxOUlockedhigh-cell-densitymutantdisplaysgrowthde˝ciencies,itisunclearwhythiswouldbe,particularlyonresourcesthatrequiredbreakdownbyextracellularpublicgoods.Ithasbeensuggestedelsewherethatcooperationshouldactuallynotbefavoredtooccurathighdensitynumberduetotheriskofcheating(ZhangandHui,2011),whichcouldexplainthispattern.Butitisnotclearwhysomepublicgoods(e.g.extracellularproteasesandamylases)seemtobefavorablyproducedathighercelldensitiesforV.harveyi,whileothersdonot(e.g.chitinasesandDNases).Togetabetterideaofwhythismightbe,moreknowledgeaboutthe˝tnessofcooperatorsanddefectorsinthepresenceofdi˙erentpolymersisneededbyperforming130competitionsonsinglepolymers,aswellasmixturesofstarch,DNA,chitin,andproteinsources.SimilartotheevolutionexperimentsconductedinM9-caseinmedia,selectionexperimentsofV.harveyistrainsinconditionswheregrowthisfavoredbytherepressionofQSwouldtellusmoreaboutthefavoredevolutionarypathsthatoccurintheseconditions,andthetypesofstrategiesthatemerge.ItispossiblethatluxOmutantscouldevolve,andfunctionasdefectors,underanextendedselectiveregimeinM9mediawithchitinasalimitingcarbonsourceorinM9mediawithDNAasalimitingnutrientsource.Inaddition,oscillatinggrowthinM9-caseinandinthepresenceofoneoftheseQS-repressedpolymersubstrates(chitinorDNA)mayrevealadditionalpopulationdynamics.Itispossiblethatshiftingbetweentheseenvironmentscouldgiverisetolongtermcoexistencebetweencooperatorsanddefectors,evenbetweenUCandQS-defectorstrains.However,wewouldalsopredictthattheWTstrain,whichcanaccessbothlowandhigh-cell-densitybehaviors,couldout-competeboththedefectorandUCinchangingorvariableenvironments.Additionally,becauseDNAcanserveasalimitingcarbon,nitrogen,orphosphorussource,itcouldbedi˙erentlyusedinmediaandpossiblyleadtodi˙erentcompetitivedynamicsdependentonwhichnutrientsource(s)itisproviding.6.6SummaryInconclusion,thisdissertationhasaddedtoourknowledgeoftheecologicalandevolu-tionarydynamicsfacingpopulationsofQSbacteria.ThoughQSoftenregulatescooperativegoodsproduction,directevidencethatQSisdirectlyneededtostabilizethesebehaviorswithinpopulationswaslacking.WehaveclearlyshownthatVibrioproteaseproduction131worksasanexploitablepublicgood.However,wedemonstratedthatQSprovidesaddedstabilityagainstnotonlyde˝nedobligatedefectors,butalsoagainstsubtlerdefectorsthatevolvedirectlyfromQS-pro˝cientstrains.Inmanycases,thestrainsthatultimatelyemergeintheconditionsexaminedhavebeenselectedforlowerinvestmentinQS,butnotacompletecessation.WeshowedthatQSallowscooperativebehaviorstobemaintainedinwell-mixedpopulations,eveninthepresenceofdefectors.Beyondthestalematethatoccursagainstdefectorsinwell-mixedconditions,QSalsoallowsthefrequencyofcooperationtoincreasewhenenvironmentalconditionsprovideassortmentofcooperatorsfromdefectors.Thisphe-nomenoncouldalsobeextendedoverthousandsofgenerations,wherewesawthatWTlineages,unlikeUClineages,couldpreservecooperationwithinexperimentalpopulations,oftenathighfrequencies.Together,weseethatinmultiplewaysregulationthroughQSprovides˝tnessadvantagesbeyondthoseprovidedbyunconditionalstrategies.132APPENDIX133TableA.1:StrainsofV.harveyiandV.choleraeusedinthiswork.134REFERENCES135REFERENCESRichardCAllen,LukeMcNally,RomanPopat,andSamPBrown.Quorumsensingprotectsbacterialco-operationfromexploitationbycheats.TheISMEJournal,16,2016.MelissaSAnderson,ErinCGarcia,andPeggyACotter.Kinddiscriminationandcom-petitiveexclusionmediatedbycontact-dependentgrowthinhibitionsystemsshapebio˝lmcommunitystructure.PLoSPathogens,10(4):e1004076,2014.ClaudiaAnetzberger,TorstenPirch,andKirstenJung.Heterogeneityinquorumsensing-regulatedbioluminescenceofVibrioharveyi.MolecularMicrobiology,2009.KyleLAsfahl,JessicaWalsh,KerriganGilbert,andMartinSchuster.Non-socialadaptationdefersatragedyofthecommonsinPseudomonasaeruginosaquorumsensing.TheISMEJournal,2015.RobertMAxelrod.Theevolutionofcooperation.Basicbooks,2006.PatriciaRYBackwell,JohnHChristy,StevenRTelford,MichaelDJennions,andJennionsPassmore.Dishonestsignallingina˝ddlercrab.ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,24,2000.Je˙reyEBarrickandRichardELenski.Genomedynamicsduringexperimentalevolution.NatureReviewsGenetics,2013.BonnieLBassler,EPeterGreenberg,andAnnMStevens.Cross-speciesinductionoflu-minescenceinthequorum-sensingbacteriumVibrioharveyi.JournalofBacteriology,1791997.MelanieBlokeschandGaryKSchoolnik.TheextracellularnucleaseDnsanditsroleinnaturaltransformationofVibriocholerae.JournalofBacteriology,2008.ZacharyDBlount,ChristinaZBorland,andRichardELenski.HistoricalcontingencyandtheevolutionofakeyinnovationinanexperimentalpopulationofEscherichiacoli.ProceedingsoftheNationalAcademyofSciences,2008.ColinRBlyth.OnSimpson'sparadoxandthesure-thingprinciple.JournaloftheAmericanStatisticalAssociation,1972.JMarkBrintandDennisEOhman.SynthesisofmultipleexoproductsinPseudomonasaeruginosaisunderthecontrolofRhlR-RhlI,anothersetofregulatorsinstrainPAO1withhomologytotheautoinducer-responsiveLuxR-LuxIfamily.JournalofBacteriology,1995.MichaelABrockhurst.Populationbottleneckspromotecooperationinbacterialbio˝lms.PLoSOne,2(7):e634,2007.136SamPBrownandRufusAJohnstone.Cooperationinthedark:signallingandcollectiveactioninquorum-sensingbacteria.ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,268(1470):9612001.SPBrown.Cooperationandcon˛ictinhost-manipulatingparasites.ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,266(1431):18991999.EricBrugerandChristopherWaters.Sharingthesandbox:Evolutionarymechanismsthatmaintainbacterialcooperation.F1000Research,4:1504,2015.JJBullandWRRice.Distinguishingmechanismsfortheevolutionofco-operation.JournalofTheoreticalBiology,1991.MarcWCadotte,AllisonMFortner,andTadashiFukami.Thee˙ectsofresourceenrichment,dispersal,andpredationonlocalandmetacommunitystructure.Oecologia,157,2006.PeterChesson.Mechanismsofmaintenanceofspeciesdiversity.AnnualReviewofEcologyandSystematics,2000.Kyoung-HeeChoi,JaredBGaynor,KimberlyGWhite,CarolinaLopez,CatharineMBo-sio,RoxAnnRKarkho˙-Schweizer,andHerbertPSchweizer.ATn7-basedbroad-rangebacterialcloningandexpressionsystem.NatureMethods,2005.JohnSChuang,OlivierRivoire,andStanislasLeibler.Simpson'sparadoxinasyntheticmicrobialsystem.Science,2009.SudhaChugani,ByoungSikKim,SomsakPhattarasukol,MitchellJBrittnacher,SangHoChoi,CarolineSHarwood,andEPeterGreenberg.Strain-dependentdiversityinthePseudomonasaeruginosaquorum-sensingregulon.ProceedingsoftheNationalAcademyofSciences,2012.BeatriceClaudi,PetraSpröte,AnnaChirkova,NicolasPersonnic,JanineZankl,NuraSchür-mann,AlexanderSchmidt,andDirkBumann.PhenotypicvariationofSalmonellainhosttissuesdelayseradicationbyantimicrobialchemotherapy.Cell,2014.TimHClutton-Brock,Geo˙reyAParker,etal.Punishmentinanimalsocieties.Nature,1995.TimFCooperandRichardELenski.ExperimentalevolutionwithE.coliindiverseresourceenvironments.I.˛uctuatingenvironmentspromotedivergenceofreplicatepopulations.BMCevolutionarybiology,10(1):1,2010.OttoXCorderoandMartinFPolz.Explainingmicrobialgenomicdiversityinlightofevolutionaryecology.NatureReviewsMicrobiology,2014.OttoXCordero,Laure-AnneVentouras,EdwardFDeLong,andMartinFPolz.Publicgooddynamicsdriveevolutionofironacquisitionstrategiesinnaturalbacterioplanktonpopulations.ProceedingsoftheNationalAcademyofSciences,2012.137DanielMCornforth,DavidJTSumpter,SamPBrown,andÅkeBrännström.Synergyandgroupsizeinmicrobialcooperation.TheAmericanNaturalist,180(3):296,2012.DanielMCornforth,RomanPopat,LukeMcNally,JamesGurney,ThomasCScott-Phillips,AlasdairIvens,StephenPDiggle,andSamPBrown.Combinatorialquorumsensingallowsbacteriatoresolvetheirsocialandphysicalenvironment.ProceedingsoftheNationalAcademyofSciences,2014.JonasCremer,AnnaMelbinger,andErwinFrey.Growthdynamicsandtheevolutionofcooperationinmicrobialpopulations.Scienti˝cReports,2,2012.BernardJCrespi.Theevolutionofsocialbehaviorinmicroorganisms.Trendsinecologyandevolution,,2001.TamásCzáránandRolfFHoekstra.Microbialcommunication,cooperationandcheating:quorumsensingdrivestheevolutionofcooperationinbacteria.PloSOne,4(8):e6655,2009.AjaiADandekar,SudhaChugani,andEPeterGreenberg.Bacterialquorumsensingandmetabolicincentivestocooperate.Science,66,2012.SophieEDarch,StuartAWest,KlausWinzer,andStephenPDiggle.Density-dependent˝tnessbene˝tsinquorum-sensingbacterialpopulations.ProceedingsoftheNationalAcademyofSciences,2012.ManoshiSenDatta,KirillSKorolev,IvanaCvijovic,CarmelDudley,andJe˙Gore.Rangeexpansionpromotescooperationinanexperimentalmicrobialmetapopulation.ProceedingsoftheNationalAcademyofSciences,59,2013.PeterWDavenport,JulianLGri˚n,andMartinWelch.QuorumsensingisaccompaniedbyglobalmetabolicchangesintheopportunistichumanpathogenPseudomonasaeruginosa.JournalofBacteriology,2015.KristenMDeAngelis,StevenELindow,andMaryKFirestone.Bacterialquorumsensingandnitrogencyclinginrhizospheresoil.FEMSMicrobiologyEcology,2008.TomDefoirdt,HADarshaneeRuwandeepika,IndraniKarunasagar,NicoBoon,andPeterBossier.QuorumsensingnegativelyregulateschitinaseinVibrioharveyi.Environmentalmicrobiologyreports,2010.FDercole,RFerriere,AGragnani,andSRinaldi.Coevolutionofslopopulations:evolutionarysliding,evolutionarypseudo-equilibriaandcomplexRedQueendynamics.ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,2006.MédéricDiard,VictorGarcia,LisaMaier,MitjaNPRemus-Emsermann,RolandRRegoes,MartinAckermann,andWolf-DietrichHardt.Stabilizationofcooperativevirulencebytheexpressionofanavirulentphenotype.Nature,56,2013.138MédéricDiard,MikaelESellin,TamasDolowschiak,MarkusArnoldini,MartinAcker-mann,andWolf-DietrichHardt.AntibiotictreatmentselectsforcooperativevirulenceofSalmonellaTyphimurium.CurrentBiology,,2014.StephenPDiggle,AshleighSGri˚n,GenevieveSCampbell,andStuartAWest.Coop-erationandcon˛ictinquorum-sensingbacterialpopulations.Nature,2007.MichaelDoebeli,ChristophHauert,andTimothyKillingback.Theevolutionaryoriginofcooperatorsanddefectors.Science,62,2004.KnutDrescher,CareyDNadell,HowardAStone,NedSWingreen,andBonnieLBassler.Solutionstothepublicgoodsdilemmainbacterialbio˝lms.CurrentBiology,2014.OmarTonsiEldakarandDavidSloanWilson.Sel˝shnessassecond-orderaltruism.Proceed-ingsoftheNationalAcademyofSciences,986,2008.SantiagoFElenaandRichardELenski.Evolutionexperimentswithmicroorganisms:thedynamicsandgeneticbasesofadaptation.NatureReviewsGenetics,2003.SylvieEstrela,ChristopherHTrisos,andSamPBrown.Frommetabolismtoecology:cross-feedinginteractionsshapethebalancebetweenpolymicrobialcon˛ictandmutualism.TheAmericanNaturalist,180(5):566,2012.EranEven-Tov,ShiraOmerBendori,JulieValastyan,XiaoboKe,ShaulPollak,TasneemBareia,IshayBen-Zion,BonnieLBassler,andAvigdorEldar.SocialEvolutionSelectsforRedundancyinBacterialQuorumSensing.PLoSBiology,14(2):e1002386,2016.PaulWEwald.TransmissionModesandEvolutionoftheParasitism-MutualismContinuum.AnnalsoftheNewYorkAcademyofSciences,1987.FrancescaFiegna,AlejandraMoreno-Letelier,ThomasBell,andTimothyGBarraclough.Evolutionofspeciesinteractionsdeterminesmicrobialcommunityproductivityinnewenvironments.TheISMEJournal,2015.CraigAFogle,JamesLNagle,andMichaelMDesai.Clonalinterference,multiplemutationsandadaptationinlargeasexualpopulations.Genetics,73,2008.KevinRFoster.Diminishingreturnsinsocialevolution:thenot-so-tragiccommons.Journalofevolutionarybiology,2004.KevinRFoster,GadShaulsky,JoanEStrassmann,DavidCQueller,andChrisRLThomp-son.Pleiotropyasamechanismtostabilizecooperation.Nature,2004.KevinRFoster,KatieParkinson,andChristopherRLThompson.Whatcanmicrobialgeneticsteachsociobiology?TrendsinGenetics,2007.139WClaiborneFuqua,StephenCWinans,andEPeterGreenberg.Quorumsensinginbacteria:theLuxR-LuxIfamilyofcelldensity-responsivetranscriptionalregulators.JournalofBacteriology,176(2):269,1994.MelanieGhoul,AshleighSGri˚n,andStuartAWest.Towardanevolutionaryde˝nitionofcheating.Evolution,2014.Je˙Gore,HyunYouk,andAlexanderVanOudenaarden.Snowdriftgamedynamicsandfacultativecheatinginyeast.Nature,256,2009.DuncanGreigandMichaelTravisano.ThePrisoner'sDilemmaandpolymorphisminyeastSUCgenes.ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,271(Suppl2004.AshleighSGri˚n,StuartAWest,andAngusBuckling.Cooperationandcompetitioninpathogenicbacteria.Nature,430(7003):10242004.DieterHaas.Costofcell-cellsignallinginPseudomonasaeruginosa:whyitcanpaytobesignal-blind.NatureReviewsMicrobiology,2006.WDHamilton.ThegeneticalevolutionofsocialbehaviourIandII.JournalofTheoreticalBiology,1964.WilliamDHamilton.Theevolutionofaltruisticbehavior.TheAmericanNaturalist,971963.WilliamDHamilton.Sel˝shandspitefulbehaviourinanevolutionarymodel.Nature,2281970.BrianKHammerandBonnieLBassler.Quorumsensingcontrolsbio˝lmformationinVibriocholerae.MolecularMicrobiology,2003.NealDHammer,JamesECassat,MichaelJNoto,LisaJLojek,AshleyDChadha,JonathanESchmitz,CBuddyCreech,andEricPSkaar.Inter-andintraspeciesmetabo-liteexchangepromotesvirulenceofantibiotic-resistantStaphylococcusaureus.Cellhostandmicrobe,2014.GarrettHardin.Thetragedyofthecommons.Science,48,1968.SiljaHeilmann,SandeepKrishna,andBenjaminKerr.Whydobacteriaregulatepublicgoodsbyquorumsensing?Howtheshapesofcostandbene˝tfunctionsdeterminetheformofoptimalregulation.Frontiersinmicrobiology,6,2015.BurkhardAHenseandMartinSchuster.Coreprinciplesofbacterialautoinducersystems.MicrobiologyandMolecularBiologyReviews,2015.WilliamGHillandAlanRobertson.Thee˙ectoflinkageonlimitstoarti˝cialselection.Geneticalresearch,1966.GEvelynHutchinson.CopepodologyfortheOnithologist,1951.140NJiricny,SPDiggle,SAWest,BAEvans,GBallantyne,ARoss-Gillespie,andASGri˚n.Fitnesscorrelateswiththeextentofcheatinginabacterium.Journalofevolutionarybiology,2010.EmilyIJones,MichelleEAfkhami,ErolAkçay,JudithLBronstein,RedouanBshary,MeganEFrederickson,KatyDHeath,JasonDHoeksema,JoshuaHNess,MSabrinaPankey,StephanieSPorter,JoelLSachs,KlaraScharnagl,andMarenLFriesen.Cheatersmustprosper:reconcilingtheoreticalandempiricalperspectivesoncheatinginmutualism.EcologyLetters,2015.KatyCKaoandGavinSherlock.MolecularcharacterizationofclonalinterferenceduringadaptiveevolutioninasexualpopulationsofSaccharomycescerevisiae.NatureGenetics,2008.ReesKassen,AngusBuckling,GrahamBell,andPaulBRainey.Diversitypeaksatinter-mediateproductivityinalaboratorymicrocosm.Nature,2000.DavidSKatzianer,HuiWang,RyanMCarey,andJunZhu.QuorumsocialcheatinganddeceptioninVibriocholerae.Appliedandenvironmentalmicrobiology,812015.DanielBKearns.A˝eldguidetobacterialswarmingmotility.NatureReviewsMicrobiology,2010.ETobyKiers,RobertARousseau,StuartAWest,andRFordDenison.Hostsanctionsandthelegume-rhizobiummutualism.Nature,1,2003.TimothyKillingbackandMichaelDoebeli.Self-organizedcriticalityinspatialevolutionarygametheory.Journaloftheoreticalbiology,1998.TimothyKillingbackandMichaelDoebeli.Thecontinuousprisoner'sdilemmaandtheevo-lutionofcooperationthroughreciprocalaltruismwithvariableinvestment.TheAmericanNaturalist,2002.TimothyKillingback,MichaelDoebeli,andNancyKnowlton.Variableinvestment,thecon-tinuousprisoner'sdilemma,andtheoriginofcooperation.ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,266(1430):17231999.KarlEKloseandJohnJMekalanos.Distinctrolesofanalternativesigmafactorduringbothfree-swimmingandcolonizingphasesoftheVibriocholeraepathogeniccycle.MolecularMicrobiology,1998.EricJKoch,TimMiyashiro,MargaretJMcFall-Ngai,andEdwardGRuby.Featuresgoverningsymbiontpersistenceinthesquid-vibrioassociation.Molecularecology,23(6):2014.Jan-UlrichKreft.Bio˝lmspromotealtruism.Microbiology,2004.141Jan-UlrichKreftandSebastianBonhoe˙er.Theevolutionofgroupsofcooperatingbacteriaandthegrowthrateversusyieldtrade-o˙.Microbiology,1,2005.LeeKroos.TheBacillusandMyxococcusdevelopmentalnetworksandtheirtranscriptionalregulators.AnnualReviewofGenetics,2007.RKümmerli,NJiricny,LSClarke,SAWest,andASGri˚n.Phenotypicplasticityofacooperativebehaviourinbacteria.Journalofevolutionarybiology,2009.GregoryILang,DanielPRice,MarkJHickman,EricaSodergren,GeorgeMWeinstock,DavidBotstein,andMichaelMDesai.Pervasivegenetichitchhikingandclonalinterferenceinfortyevolvingyeastpopulations.Nature,2013.BreahLaSarreandMichaelJFederle.Exploitingquorumsensingtoconfusebacterialpathogens.MicrobiologyandMolecularBiologyReviews,2013.AmelLati˝,MichaelKWinson,MarylineFoglino,BarrieWBycroft,GordonSABStewart,AndréeLazdunski,andPaulWilliams.MultiplehomologuesofLuxRandLuxIcontrolexpressionofvirulencedeterminantsandsecondarymetabolitesthroughquorumsensinginPseudomonasaeruginosaPAO1.MolecularMicrobiology,1995.HenryHLee,MichaelNMolla,CharlesRCantor,andJamesJCollins.Bacterialcharityworkleadstopopulation-wideresistance.Nature,2010.Kyu-HoLeeandEdwardGRuby.E˙ectofthesquidhostontheabundanceanddistributionofsymbioticVibrio˝scheriinnature.AppliedandEnvironmentalMicrobiology,60(5):1994.RichardELenski,MichaelRRose,SuzanneCSimpson,andScottCTadler.Long-termexperimentalevolutioninEscherichiacoli.I.Adaptationanddivergenceduring2,000generations.Americannaturalist,pages1991.MicheleLeRoux,SBrookPeterson,andJosephDMougous.Bacterialdangersensing.Journalofmolecularbiology,3,2015.BaochuanLin,ZhengWang,AnthonyPMalanoski,ElizabethAO'Grady,CharlesFWim-pee,VarapornVuddhakul,NelsonAlvesJr,FabianoLThompson,BrunoGomez-Gil,andGaryJVora.ComparativegenomicanalysesidentifytheVibrioharveyigenomesequencedstrainsBAA-1116andHY01asVibriocampbellii.Environmentalmicrobiologyreports,22010.SebastienLionandMinusvanBaalen.Self-structuringinspatialevolutionaryecology.Ecol-ogyLetters,2008.RCraigMacLeanandIvanaGudelj.Resourcecompetitionandsocialcon˛ictinexperimentalpopulationsofyeast.Nature,1,2006.RCMacLean.Thetragedyofthecommonsinmicrobialpopulations:insightsfromtheoret-ical,comparativeandexperimentalstudies.Heredity,2008.142BorisMagasanik.Theregulationofnitrogenutilizationinentericbacteria.Journalofcellularbiochemistry,1993.KrisztiánMágori,BeátaOborny,UlfDieckmann,andGézaMeszéna.Cooperationandcompetitioninheterogeneousenvironments:theevolutionofresourcesharinginclonalplants.EvolutionaryEcologyResearch,2003.CharlotteMajerczyk,MitchellBrittnacher,MichaelJacobs,ChristopherDArmour,MathewRadey,EmilySchneider,SomsakPhattarasokul,RichardBunt,andEPeterGreenberg.GlobalanalysisoftheBurkholderiathailandensisquorumsensing-controlledregulon.Jour-nalofBacteriology,2014.CharlotteMajerczyk,EmilySchneider,andEPeterGreenberg.QuorumsensingcontrolofTypeVIsecretionfactorsrestrictstheproliferationofquorum-sensingmutants.eLife,5:e14712,2016.MarivicMartin,TheresaHölscher,AnnaDrago²,VaughnSCooper,andÁkosTKovács.Laboratoryevolutionofmicrobialinteractionsinbacterialbio˝lms.JournalofBacteriol-ogy,pages2016.NancyMcFarland,LindaMcCarter,StanleyArtz,andSydneyKustu.NitrogenregulatorylocusglnRofentericbacteriaiscomposedofcistronsntrBandntrC:identi˝cationoftheirproteinproducts.ProceedingsoftheNationalAcademyofSciences,1981.NatashaJMehdiabadi,ChandraNJack,Ti˙anyTalleyFarnham,ThomasGPlatt,SaraEKalla,GadShaulsky,DavidCQueller,andJoanEStrassmann.Socialevolution:kinpreferenceinasocialmicrobe.Nature,2006.BrettMellbyeandMartinSchuster.Physiologicalframeworkfortheregulationofquorumsensing-dependentpublicgoodsinPseudomonasaeruginosa.JournalofBacteriology,1962014.MJMerrickandRAEdwards.Nitrogencontrolinbacteria.Microbiologicalreviews,59(4):1995.MelissaBMiller,KarenSkorupski,DerrickHLenz,RonaldKTaylor,andBonnieLBassler.ParallelquorumsensingsystemsconvergetoregulatevirulenceinVibriocholerae.Cell,2002.SaraMitriandKevinRFoster.Pleiotropyandthelowcostofindividualtraitspromotecooperation.Evolution,2016.SaraMitri,JoãoBXavier,andKevinRFoster.Socialevolutioninmultispeciesbio˝lms.ProceedingsoftheNationalAcademyofSciences,108(Supplement2):2011.KennyCMok,NedSWingreen,andBonnieLBassler.Vibrioharveyiquorumsensing:acoincidencedetectorfortwoautoinducerscontrolsgeneexpression.TheEMBOjournal,2003.143BabakMomeni,AdamJamesWaite,andWenyingShou.Spatialself-organizationfavorsheterotypiccooperationovercheating.eLife,2:e00960,2013.AndrewDMorgan,BenjaminJZQuigley,SamPBrown,andAngusBuckling.Selectiononnon-socialtraitslimitstheinvasionofsocialcheats.EcologyLetters,2012.JJe˙reyMorris,RichardELenski,andErikRZinser.TheBlackQueenHypothesis:evolutionofdependenciesthroughadaptivegeneloss.mBio,3(2):e000362012.JJe˙reyMorris,SpiridonEPapoulis,andRichardELenski.CoexistenceofevolvingbacteriastabilizedbyasharedBlackQueenfunction.Evolution,71,2014.HeleneCMuller-Landau.Thetolerance-fecunditytrade-o˙andthemaintenanceofdiversityinseedsize.ProceedingsoftheNationalAcademyofSciences,2010.CareyDNadellandBonnieLBassler.A˝tnesstrade-o˙betweenlocalcompetitionanddispersalinVibriocholeraebio˝lms.ProceedingsoftheNationalAcademyofSciences,2011.FMINatrah,HADarshaneeRuwandeepika,SushantPawar,IndraniKarunasagar,PatrickSorgeloos,PeterBossier,andTomDefoirdt.RegulationofvirulencefactorsbyquorumsensinginVibrioharveyi.Veterinarymicrobiology,2011.KennethHNealson,TerryPlatt,andJWoodlandHastings.Cellularcontrolofthesynthesisandactivityofthebacterialluminescentsystem.JournalofBacteriology,1970.KHNealsonandJohnWHastings.Bacterialbioluminescence:itscontrolandecologicalsigni˝cance.Microbiologicalreviews,43(4):496,1979.Wai-LeungNgandBonnieLBassler.Bacterialquorum-sensingnetworkarchitectures.An-nualreviewofgenetics,43:197,2009.MajaNovak,ThomasPfei˙er,RichardELenski,UweSauer,andSebastianBonhoe˙er.ExperimentalTestsforanEvolutionaryTrade-O˙betweenGrowthRateandYieldinE.coli.TheAmericanNaturalist,2006.MartinANowak.Fiverulesfortheevolutionofcooperation.Science,2006.MartinANowakandRobertMMay.Evolutionarygamesandspatialchaos.Nature,3591992.ElizabethAO'GradyandCharlesFWimpee.MutationsintheluxoperonofnaturaldarkmutantsinthegenusVibrio.Appliedandenvironmentalmicrobiology,2008.NunoMOliveira,ReneNiehus,andKevinRFoster.Evolutionarylimitstocooperationinmicrobialcommunities.ProceedingsoftheNationalAcademyofSciences,17946,2014.144AnandPaiandLingchongYou.Optimaltuningofbacterialsensingpotential.Molecularsystemsbiology,5(1):286,2009.AnandPai,YuTanouchi,andLingchongYou.Optimalityandrobustnessinquorumsensing(QS)-mediatedregulationofacostlypublicgoodenzyme.ProceedingsoftheNationalAcademyofSciences,,2012.AlexandraSPenn,TimCRConibear,RichardAWatson,AlexRKraaijeveld,andJeremySWebb.CanSimpson'sparadoxexplainco-operationinPseudomonasaeruginosabio˝lms?FEMSImmunologyandMedicalMicrobiology,,2012.StevenLPercival,JohnGThomas,andDavidWWilliams.Bio˝lmsandbacterialimbalancesinchronicwounds:anti-Koch.Internationalwoundjournal,2010.AlexandrePersat,CareyDNadell,MinyoungKevinKim,FrancoisIngremeau,AlbertSirya-porn,KnutDrescher,NedSWingreen,BonnieLBassler,ZemerGitai,andHowardAStone.Themechanicalworldofbacteria.Cell,2015.MarloesLCPetrusandDennisClaessen.PivotalrolesforStreptomycescellsurfacepoly-mersinmorphologicaldi˙erentiation,attachmentandmycelialarchitecture.AntonievanLeeuwenhoek,9,2014.ThomasPfei˙erandSebastianBonhoe˙er.Evolutionofcross-feedinginmicrobialpopula-tions.TheAmericannaturalist,2004.ThomasPfei˙er,StefanSchuster,andSebastianBonhoe˙er.CooperationandcompetitionintheevolutionofATP-producingpathways.Science,292(5516):5042001.ThomasGPlattandClayFuqua.What'sinaname?Thesemanticsofquorumsensing.Trendsinmicrobiology,2010.ShaulPollak,ShiraOmer-Bendori,EranEven-Tov,ValeriaLipsman,TasneemBareia,IshayBen-Zion,andAvigdorEldar.Facultativecheatingsupportsthecoexistenceofdiversequorum-sensingalleles.ProceedingsoftheNationalAcademyofSciences,2157,2016.AudraJPompeani,JosephJIrgon,MichaelFBerger,MarthaLBulyk,NedSWingreen,andBonnieLBassler.TheVibrioharveyimasterquorum-sensingregulator,LuxR,aTetR-typeproteinisbothanactivatorandarepressor:DNArecognitionandbindingspeci˝cityattargetpromoters.MolecularMicrobiology,2008.RPopat,DMCornforth,LMcNally,andSPBrown.Collectivesensingandcollectiveresponsesinquorum-sensingbacteria.JournaloftheRoyalSocietyInterface,12(103):20140882,2015a.RomanPopat,ShanikaACrusz,MarcoMessina,PaulWilliams,StuartAWest,andStephenPDiggle.Quorum-sensingandcheatinginbacterialbio˝lms.ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,pagerspb20121976,2012.145RomanPopat,EricJGPollitt,FreyaHarrison,HardeepNaghra,Kar-WaiHong,Kok-GanChan,AshleighSGri˚n,PaulWilliams,SamPBrown,StuartAWest,etal.Con˛ictofinterestandsignalinterferenceleadtothebreakdownofhonestsignaling.Evolution,692015b.PaulBRaineyandMichaelTravisano.Adaptiveradiationinaheterogeneousenvironment.Nature,,1998.DanielJRankin,KatjaBargum,andHannaKokko.Thetragedyofthecommonsinevolu-tionarybiology.Trendsinecology&evolution,2007.OlayaRendueles,PeterCZee,IrisDinkelacker,MichaelaAmherd,SébastienWielgoss,andGregoryJVelicer.Rapidandwidespreaddenovoevolutionofkindiscrimination.Pro-ceedingsoftheNationalAcademyofSciences,1,2015.AdinRoss-Gillespie,AndyGardner,StuartAWest,andAshleighSGri˚n.Frequencydependenceandcooperation:theoryandatestwithbacteria.TheAmericanNaturalist,2007.AdinRoss-Gillespie,AndyGardner,AngusBuckling,StuartAWest,andAshleighSGri˚n.Densitydependenceandcooperation:theoryandatestwithbacteria.Evolution,63(9):2009.StevenTRutherford,JuliaCvanKessel,YiShao,andBonnieLBassler.AphAandLuxR/HapRreciprocallycontrolquorumsensinginVibrios.Genesanddevelopment,2011.KelsiMSandoz,ShelbyMMitzimberg,andMartinSchuster.SocialcheatinginPseudomonasaeruginosaquorumsensing.ProceedingsoftheNationalAcademyofSciences,104(40):2007.RebeccaSSatterwhiteandTimFCooper.ConstraintsonadaptationofEscherichiacolitomixed-resourceenvironmentsincreaseovertime.Evolution,2015.JonasSchluter,CareyDNadell,BonnieLBassler,andKevinRFoster.Adhesionasaweaponinmicrobialcompetition.TheISMEJournal,2015.JonasSchluter,ArminPSchoech,KevinRFoster,andSaraMitri.TheEvolutionofQuorumSensingasaMechanismtoInferKinship.PLoSComputationalBiology,12(4):e1004848,2016.MartinSchusterandEPeterGreenberg.Anetworkofnetworks:quorum-sensinggeneregulationinPseudomonasaeruginosa.Internationaljournalofmedicalmicrobiology,2962006.ThomasCScott-Phillips,ThomasEDickins,andStuartAWest.Evolutionarytheoryandtheultimate-proximatedistinctioninthehumanbehavioralsciences.PerspectivesonPsychologicalScience,2011.146YiShaoandBonnieLBassler.QuorumregulatorysmallRNAsrepresstypeVIsecretioninVibriocholerae.MolecularMicrobiology,2014.LawrenceJShimkets.SocialanddevelopmentalbiologyoftheMyxobacteria.Microbiologicalreviews,1990.OrenShoval,HilaSheftel,GuyShinar,YuvalHart,OmerRamote,AviMayo,ErezDekel,KathrynKavanagh,andUriAlon.Evolutionarytrade-o˙s,Paretooptimality,andthegeometryofphenotypespace.Science,1160,2012.EdwardHSimpson.Theinterpretationofinteractionincontingencytables.JournaloftheRoyalStatisticalSociety.SeriesB(Methodological),pages1951.JohnMaynardSmith.EvolutionandtheTheoryofGames.Cambridgeuniversitypress,1982.ScottSmukalla,MarinaCaldara,NathaliePochet,AnneBeauvais,StephanieGuadagnini,ChenYan,MarceloDVinces,AnJansen,MarieChristinePrevost,Jean-PaulLatgé,GeraldRFink,KevinRFoster,andKevinJVerstrepen.FLO1isavariablegreenbeardgenethatdrivesbio˝lm-likecooperationinbuddingyeast.Cell,2008.PaulDSniegowskiandPhilipJGerrish.Bene˝cialmutationsandthedynamicsofadapta-tioninasexualpopulations.PhilosophicalTransactionsoftheRoyalSocietyB:BiologicalSciences,365(1544):1252010.WilliamSotoandMicheleKNishiguchi.MicrobialexperimentalevolutionasanovelresearchapproachintheVibrionaceaeandsquid-Vibriosymbiosis.Frontiersinmicrobiology,5:593,2014.AndrewJSpiers.Wrinkly-Spreader˝tnessinthetwo-dimensionalagarplatemicrocosm:maladaptation,compensationandecologicalsuccess.PloSOne,2(8):e740,2007.AndrewJSpiers,SophieGKahn,JohnBohannon,MichaelTravisano,andPaulBRainey.AdaptivedivergenceinexperimentalpopulationsofPseudomonas˛uorescens.I.Geneticandphenotypicbasesofwrinklyspreader˝tness.Genetics,2002.DianeSSrivastava,JurekKolasa,JanBengtsson,AndrewGonzalez,SharonPLawler,ThomasEMiller,PabloMunguia,TamaraRomanuk,DavidCSchneider,andMKur-tisTrzcinski.Arenaturalmicrocosmsusefulmodelsystemsforecology?Trendsinecologyandevolution,,2004.DishaSrivastavaandChristopherMWaters.Atangledweb:regulatoryconnectionsbetweenquorumsensingandcyclicdi-GMP.JournalofBacteriology,93,2012.DishaSrivastava,Meng-LunHsieh,AtulKhataokar,MatthewBNeiditch,andChristopherMWaters.Cyclicdi-GMPinhibitsVibriocholeraemotilitybyrepressinginductionoftran-scriptionandinducingextracellularpolysaccharideproduction.MolecularMicrobiology,2013.147JoanEStrassmann,OwenMGilbert,andDavidCQueller.Kindiscriminationandcoop-erationinmicrobes.Annualreviewofmicrobiology,2011.SimonSwift,JAllanDownie,NeilAWhitehead,AnneMLBarnard,GeorgePCSalmond,andPaulWilliams.Quorumsensingasapopulation-density-dependentdeterminantofbacterialphysiology.Advancesinmicrobialphysiology,,2001.EörsSzathmáryandJohnMaynardSmith.Fromreplicatorstoreproducers:the˝rstmajortransitionsleadingtolife.Journaloftheoreticalbiology,1997.JoshuaTaso˙,MichaelTMee,andHarrisHWang.Aneconomicframeworkofmicrobialtrade.PloSOne,10(7):e0132907,2015.DavidTilman.TestsofresourcecompetitiontheoryusingfourspeciesofLakeMichiganalgae.Ecology,,1981.MichaelTravisanoandGregoryJVelicer.Strategiesofmicrobialcheatercontrol.Trendsinmicrobiology,2004.KimberlyCTuandBonnieLBassler.MultiplesmallRNAsactadditivelytointegratesensoryinformationandcontrolquorumsensinginVibrioharveyi.Genes&development,2007.PaulETurnerandLinChao.Prisoner'sdilemmainanRNAvirus.Nature,398(6726):1999.PaulETurnerandLinChao.EscapefromPrisoner'sDilemmainRNAphage'6.TheAmericanNaturalist,2003.MVanBaalenandVAAJansen.Kindsofkindness:classifyingthecausesofaltruismandcooperation.Journalofevolutionarybiology,2006.JordivanGestel,FranzJWeissing,OscarPKuipers,andAkosTKovács.Densityoffoundercellsa˙ectsspatialpatternformationandcooperationinBacillussubtilisbio˝lms.TheISMEJournal,2014.JordivanGestel,HeraVlamakis,andRobertoKolter.Divisionoflaborinbio˝lms:theecologyofcelldi˙erentiation.Microbiologyspectrum,3(2),2015.JuliaCvanKessel,StevenTRutherford,YiShao,AlanFUtria,andBonnieLBassler.IndividualandcombinedrolesofthemasterregulatorsAphAandLuxRincontroloftheVibrioharveyiquorum-sensingregulon.JournalofBacteriology,,2013.GregoryJVelicer.Socialstrifeinthemicrobialworld.Trendsinmicrobiology,2003.GünterPWagnerandJianzhiZhang.Thepleiotropicstructureofthegenotypypemap:theevolvabilityofcomplexorganisms.NatureReviewsGenetics,2011.148AdamJamesWaite,CarolineCannistra,andWenyingShou.Defectorscancreateconditionsthatrescuecooperation.PLoSComputationalBiology,11(12):e1004645,2015.MeizhenWang,AmyLSchaefer,AjaiADandekar,andEPeterGreenberg.QuorumsensingandpolicingofPseudomonasaeruginosasocialcheaters.ProceedingsoftheNationalAcademyofSciences,2015.YunduanWang,HuiWang,ZhigangCui,HailiChen,ZengtaoZhong,BiaoKan,andJunZhu.Theprevalenceoffunctionalquorum-sensingsystemsinrecentlyemergedVibriocholeraetoxigenicstrains.Environmentalmicrobiologyreports,2011.ZhengWang,BaochuanLin,AnahitaMostaghim,RobertARubin,EvanRGlaser,PimonsriMittraparp-arthorn,JanelleRThompson,VarapornVuddhakul,andGaryJVora.Vib-riocampbelliihmgA-mediatedpyomelanizationimpairsquorumsensing,virulence,andcellular˝tness.Vibrioecology,pathogenesisandevolution,page103,2014.ChristopherMWatersandBonnieLBassler.Quorumsensing:cell-to-cellcommunicationinbacteria.AnnualReviewofCellandDevelopmentalBiology,346,2005.ChristopherMWatersandBonnieLBassler.TheVibrioharveyiquorum-sensingsystemusessharedregulatorycomponentstodiscriminatebetweenmultipleautoinducers.Genes&development,,2006.ChristopherMWaters,WenyunLu,JoshuaDRabinowitz,andBonnieLBassler.Quorumsensingcontrolsbio˝lmformationinVibriocholeraethroughmodulationofcyclicdi-GMPlevelsandrepressionofvpsT.JournalofBacteriology,2008.GijsbertDAWerner,JoanEStrassmann,AniekBFIvens,DanielJPEngelmoer,ErikVerbruggen,DavidCQueller,RonaldNoë,NancyCollinsJohnson,PeterHammerstein,andETobyKiers.Evolutionofmicrobialmarkets.ProceedingsoftheNationalAcademyofSciences,2014.StuartAWest,ETobyKiers,EllenLSimms,andRFordDenison.Sanctionsandmutualismstability:whydorhizobia˝xnitrogen?ProceedingsoftheRoyalSocietyofLondonB:BiologicalSciences,2002.StuartAWest,AshleighSGri˚n,AndyGardner,andStephenPDiggle.Socialevolutiontheoryformicroorganisms.NatureReviewsMicrobiology,2006.StuartAWest,AshleighSGri˚n,andAndyGardner.Evolutionaryexplanationsforcoop-eration.CurrentBiology,2007.StuartAWest,KlausWinzer,AndyGardner,andStephenPDiggle.Quorumsensingandtheconfusionaboutdi˙usion.Trendsinmicrobiology,2012.CaraNWilder,StephenPDiggle,andMartinSchuster.CooperationandcheatinginPseu-domonasaeruginosa:therolesofthelas,rhlandpqsquorum-sensingsystems.TheISMEJournal,2011.149JoshuaWWilliams,XiaohuiCui,AndreLevchenko,andAnnMStevens.Robustandsensitivecontrolofaquorum-sensingcircuitbytwointerlockedfeedbackloops.Molecularsystemsbiology,4(1):234,2008.MichaelJWiser,NoahRibeck,andRichardELenski.Long-termdynamicsofadaptationinasexualpopulations.Science,,2013.JoaoBXavier,WookKim,andKevinRFoster.AmolecularmechanismthatstabilizescooperativesecretionsinPseudomonasaeruginosa.MolecularMicrobiology,179,2011.QianYangandTomDefoirdt.Quorumsensingpositivelyregulates˛agellarmotilityinpathogenicVibrioharveyi.Environmentalmicrobiology,2015.YutakaYawata,OttoXCordero,FilippoMenolascina,Jan-HendrikHehemann,Mar-tinFPolz,andRomanStocker.Compersaltradeo˙ecologicallydi˙erenti-atesrecentlyspeciatedmarinebacterioplanktonpopulations.ProceedingsoftheNationalAcademyofSciences,2014.DouglasWYu,HowardBWilson,andNaomiEPierce.Anempiricalmodelofspeciescoexistenceinaspatiallystructuredenvironment.Ecology,2001.DouglasWYu,HowardBWilson,MeganEFrederickson,WilyPalomino,RafaelDeLaCol-ina,DavidPEdwards,andAngelABalareso.Experimentaldemonstrationofspeciescoexistenceenabledbydispersallimitation.JournalofAnimalEcology,2004.FengZhangandCangHui.Eco-evolutionaryfeedbackandtheinvasionofcooperationinprisoner'sdilemmagames.PloSOne,6(11):e27523,2011.Xue-XianZhangandPaulBRainey.Exploringthesociobiologyofpyoverdin-producingPseudomonas.Evolution,4,2013.DonglingZheng,ChrystalaConstantinidou,JonLHobman,andStephenDMinchin.Iden-ti˝cationoftheCRPregulonusinginvitroandinvivotranscriptionalpro˝ling.Nucleicacidsresearch,,2004.150