plete789se.com

RESEARCHARTICLEOpenAccessDivergenceincis-regulatorysequencessurroundingtheopsingenearraysofAfricancichlidfishesKellyEO'Quin1,DanielSmith1,ZanNaseer1,JaneSchulte1,SamuelDEngel1,Yong-HweeELoh2,JToddStreelman2,JeffreyLBoore3,4andKarenLCarleton1*AbstractBackground:Divergencewithincis-regulatorysequencesmaycontributetotheadaptiveevolutionofgeneexpression,butfunctionalallelesintheseregionsaredifficulttoidentifywithoutabundantgenomicresources.
AmongAfricancichlidfishes,thedifferentialexpressionofsevenopsingeneshasproducedadaptivedifferencesinvisualsensitivity.
Quantitativegeneticanalysissuggeststhatcis-regulatoryallelesneartheSWS2-LWSopsinsmaycontributetothisvariation.
Here,wesequenceBACscontainingtheopsingenesoftwocichlids,OreochromisniloticusandMetriaclimazebra.
Weusephylogeneticfootprintingandshadowingtoexaminedivergenceinconservednon-codingelements,promotersequences,and3'-UTRssurroundingeachopsininsearchofcandidatecis-regulatorysequencesthatinfluencecichlidopsinexpression.
Results:Weidentified20conservednon-codingelementssurroundingtheopsinsofcichlidsandotherteleosts,includingoneknownenhancerandaretinalmicroRNA.
Mostconservedelementscontainedcomputationally-predictedbindingsitesthatcorrespondtotranscriptionfactorsthatfunctioninvertebrateopsinexpression;O.
niloticusandM.
zebraweresignificantlydivergentintwoofthese.
Similarly,wefoundalargenumberofrelevanttranscriptionfactorbindingsiteswithineachopsin'sproximalpromoter,andidentifiedfiveopsinsthatwereconsiderablydivergentinbothexpressionandthenumberoftranscriptionfactorbindingsitessharedbetweenO.
niloticusandM.
zebra.
WealsofoundseveralmicroRNAtargetsiteswithinthe3'-UTRofeachopsin,includingtwo3'-UTRsthatdiffersignificantlybetweenO.
niloticusandM.
zebra.
Finally,weexaminedinterspecificdivergenceamong18phenotypicallydiversecichlidsfromLakeMalawiforoneconservednon-codingelement,two3'-UTRs,andfiveopsinproximalpromoters.
WefoundthatallregionswerehighlyconservedwithsomeevidenceofCRXtranscriptionfactorbindingsiteturnover.
WealsofoundthreeSNPswithintwoopsinpromotersandonenon-codingelementthathadweakassociationwithcichlidopsinexpression.
Conclusions:Thisstudyisthefirsttosystematicallysearchtheopsinsofcichlidsforputativecis-regulatorysequences.
Althoughmanyputativeregulatoryregionsarehighlyconservedacrossalargenumberofphenotypicallydiversecichlids,wefoundatleastninedivergentsequencesthatcouldcontributetoopsinexpressiondifferencesincisandstandoutascandidatesforfuturefunctionalanalyses.
BackgroundAdaptivephenotypicevolutionmayresulteitherfromprotein-codingmutationsthatmodifythestructureandfunctionofgenes,orfromregulatorymutationsthatalterthetiming,location,orexpressionofgenes[1-3].
Althoughexamplesofprotein-codingmutationsthatcontributetophenotypicevolutionarewellknown(e.
g.
,[4-6]),examplesofregulatorymutationsthatalsoaffectphenotypicadaptationarelesswellknown,butnolessimportant(e.
g.
,[7-9]).
Oneclassofregulatorymuta-tions,cis-regulatorymutations,arefoundincloseproxi-mitytothegenestheyregulateandfunctionbyalteringthebindingoftranscriptionfactorsnecessaryforgeneexpression.
Cis-regulatorymutationsexhibitseveralfea-turesthatmakethemideallysuitedforadaptivepheno-typicevolution,includingcodominance[10]andmodularity[8].
Thesefeaturesmakecis-regulatory*Correspondence:kcarleto@umd.
eduContributedequally1DepartmentofBiology,UniversityofMaryland,CollegePark,MD20742,USAFulllistofauthorinformationisavailableattheendofthearticleO'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/1202011O'Quinetal;licenseeBioMedCentralLtd.
ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(http://creativecommons.
org/licenses/by/2.
0),whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.
mutationsefficienttargetsfornaturalselection[11]andlimitthenegativeconsequencesofpleiotropythatpre-sumablyaffectmanytrans-regulatoryandprotein-cod-ingmutations.
Finally,sincecis-regulatorymutationsmayunderliemanyoftheadaptiveanddiseasepheno-typesfoundinnature,identifyingtheseallelesremainsanimportantgoalofevolutionarygenetics.
However,identifyingcis-regulatorymutationscanbechallengingwithoutabundantfunctionalgenomicresources,sincethetranscriptionfactorbindingsites(TFBS)theyaffectaresmall,lackstrictconservation,andarefoundindiffi-cult-to-annotateregionsofthegenome[2,3].
Thelocationofcis-regulatorysequencescanbenear-toorfar-fromthegenestheyregulate.
Promotersequencesfounddirectlyupstreamofgenescanharborcis-regulatoryalleles[12,13],ascanenhancerorrepres-sorelementslocatedmanykilobasesaway[14,15].
Cis-regulatorysequencescanevenresidewithintheuntranslatedregions(UTRs)ofgenes,wheretheyalterthebindingofmicroRNAs(miRNAs)thatregulategeneexpressionfollowingtranscription[16,17].
Butwhereevertheirlocation,twomethodscommonlyusedtoidentifycis-regulatorysequencesandallelesarephyloge-neticfootprintingandphylogeneticshadowing[18].
Inphylogeneticfootprinting,onecomparesDNAsur-roundingsomegene(s)ofinterestamongnumerousdivergenttaxainhopesofidentifyingnon-codingregionsthatarehighlyconserved.
Bytheverynatureoftheirconservation,theseconservednon-codingelements(CNEs)standoutascandidateregulatorysequences,sinceconservationisoftenusedtoindicatefunction.
Oncecandidateregulatorysequenceshavebeenidenti-fiedviaphylogeneticfootprinting,themethodusedtoidentifyputativecis-regulatoryalleleswithinthemisdif-ferentialphylogeneticfootprinting,orphylogeneticsha-dowing[18,19].
Inphylogeneticshadowing,onecomparesputativeregulatorysequencesamongcloselyrelatedtaxainhopesofidentifyingsequencepoly-morphismscorrelatedwiththedivergentexpressionofsometargetgene(s).
Followingtheirapplication,func-tionalgenomicanalysesarenecessarytovalidatethefunctionofanycandidatesequencesorallelesidentifiedbythephylogeneticfootprintingandshadowingmeth-ods;butevenbythemselves,bothmethodscanprovidevaluableinsightsintothelocationofpotentialcis-regula-torysequencesandthetranscriptionfactorsthatbindthem.
Thegoalofthisstudyistoidentifycandidatecis-regu-latorysequencesthatcontrolopsingeneexpressioninAfricancichlidfishes.
OpsinsareagroupofGprotein-coupledreceptorsthatconfersensitivitytolightandmediatecolorvision[20].
Africancichlidscompriseadiversecladeoffreshwater,teleostfishfoundthrough-outthelakesandriversofAfrica,includingthethreeAfricanGreatLakes,LakesTanganyika,Malawi,andVictoria[21,22].
CichlidsfromLakesTanganyikaandMalawiexhibitdramaticvariationintheirsensitivitytocoloredlight[23-25].
Speciesfromtheselakesexhibitretinalsensitivitiesthataremaximallysensitivetoshort,middle,orlong-wavelengthspectra;insomecases,clo-selyrelatedspeciescandifferintheirmaximalretinalsensitivitybyover100nm[25-27].
Thisstrikingvaria-tionmakesthecichlidvisualsystemoneofthemostdiversevertebratevisualsystemssofaridentified.
Mostvariationincichlidcolorsensitivityisduetochangesintheregulationoftheirconeopsingenes[26,27].
Cichlidshavesevenconeopsingenesusedforcolorvision;theseopsinsareSWS1(ultraviolet-sensitive),SWS2B(violet-sensitive),SWS2A(blue-sensitive),RH2B(blue-green-sensitive),RH2AandRH2A(green-sensitive),andLWS(red-sensitive)[28].
Additionally,theseopsinsarelocatedinthreeregionsofthecichlidgenome:SWS1isfoundoncichlidlinkagegroup(LG)17;RH2B,RH2AandRH2AarefoundtogetherinatandemarrayonLG5;andSWS2A,SWS2B,andLWSformasecondtandemarrayonLG5(Leeetal.
2005)(Figure1).
Amongdif-ferentcichlidspecies,theseopsinsarealternativelyco-expressedinthreepredominantgroups,orpalettes,toproducethethreecommonvisualpigmentsets:SWS1-RH2B-RH2A(shortwavelength-sensitive),SWS2B-RH2B-RH2A(middlewavelength-sensitive),andSWS2A-RH2A-LWS(longwavelength-sensitive)[26].
Cichlidsexhibitseveralcorrelationsbetweentheexpressionoftheiropsinsandimportantecologicalvariables,includ-ingforagingpreferenceandambientlightintensity[26,27].
Thesecorrelationssuggestthatopsingeneexpressionvariesadaptivelyincichlids,especiallysincesomeexpression-ecologycorrelationshaveevolvedinde-pendentlyamongcichlidsindifferentlakes[27].
ArecentquantitativegeneticanalysisofopsinexpressionintwoLakeMalawicichlidsfoundaquantitativetraitlocus(QTL)locatedneartheopsingenes[29].
TheproximityofthisQTLtotheopsinssuggeststhatmuta-tionswithinoneormorecis-regulatorysequencesmaycontributetovariationincichlidopsinexpression.
Butlikemanynon-modelsystems,fewgenomicresourcesarecurrentlyavailableforcichlids,makingitdifficulttoidentifypotentialcis-regulatoryallelesandtesttheirassociationwithopsingeneexpression.
Here,wesequenceandanalyzebacterialartificialchromosome(BAC)clonescontainingtheopsingenesoftwoAfricancichlidspecies,Oreochromisniloticus[30]andMetriaclimazebra[31].
Oreochromisniloticus(theNiletilapia)isariverinecichlidthatexpressesthelongwavelength-sensitiveopsinpaletteasadultsbutalsoexpressestheotherpalettesasfryandjuveniles[32].
O.
niloticusisanoutgrouptothediversehaplo-chrominecichlidsendemictoLakesTanganyika,O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page2of23Malawi,andVictoria.
Metriaclimazebra(the'classic'Zebracichlid)isonesuchhaplochrominecichlidfoundinLakeMalawi.
M.
zebraexpressestheshortwave-length-sensitiveopsinpaletteasanadultandduringalldevelopmentalstages[32].
Bothspecieslastsharedacommonancestor~18MYA,whereasM.
zebradivergedfromotherphenotypicallydiverseLakeMalawicichlidslessthan2MYA[33].
Aftersequencingtheopsin-containingBACclonesfromthesespecies,weusedtheresultingsequencesforseveralanalyses,including:(1)Annotationandcomparisonoftheopsin-con-tainingregionsfromthegenomeassembliesofsev-eralmodelteleosts.
Weperformphylogeneticfootprintingbycomparingtheopsin-containingregionsofO.
niloticusandseveralmodelfishgen-omes.
WeusethiscomparisontolocateconservedABCConservednon-codingsequenceProximalpromotersequenceRepetitivesequence1234580k100k82k84k86k88k90k92k94k96k98kG.
aculeatusO.
latipes74k76k78kD.
rerio72kSWS1TNPO350%100%1CALUAO.
latipesG.
aculeatusT.
nigroviridis123456LWS140k142k144k146k148k150k152k1234512345SWS2aSWS2b122k124k126k128k130k132k134k136k138kD.
rerioHCFC150%100%abLWS-LCR2345678910G.
aculeatusO.
latipesT.
nigroviridis0k20k2k4k6k8k10k12k14k16k18kD.
rerioRH2B1234520k22k24k26k28kSLC6A13-likeLTRtransposon50%100%111213141516G.
aculeatusO.
latipesT.
nigroviridisD.
rerioRH2AαRH2Aβ1234540k28k30k32k34k36k38k1234542k44k46k48k50k52k54k50%100%17181920Figure1ConservationbetweenO.
niloticusopsin-containingBACregionsandfourfishgenomes.
A)SWS1opsin-containingregion.
B)SWS2-LWSopsin-containingregion.
C)RH2opsin-containingregion.
ToplinerepresentsO.
niloticusBACsequence.
Conservednon-codingelements(CNEs)arenumberedandhighlightedinred;repetitivesequencesarehighlightedingreen;promotersequenceslaterexaminedforinterspecificpolymorphismarehighlightedinblue.
O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page3of23non-codingelements(CNEs)thatserveascandidatecis-regulatorysequencesfortheopsins.
(2)Computationalpredictionofbindingsitesfor12transcriptionfactorsimportantforvertebrateopsinexpression[34-41](Table1).
WeperformthissearchineachCNEaswellaswithintheproximalpromo-terofeachopsin.
WealsoperformananalogoussearchformiRNAtargetsiteswithinthe3'-UTRofeachopsin.
(3)PhylogeneticshadowingbetweenO.
niloticusandM.
zebrausingtheTFBSandmiRNAtargetsiteprofilesfoundineachCNE,promoter,and3'-UTRsequence.
IneachregionwecomparetheproportionofdivergentTFBS/miRNAtargetsiteswiththeamountexpectedgiventheover-allsequencediver-genceoftheopsinBACsandintrons(ameasureofneutralevolutionarydivergence[42,43]).
Thesecom-parisonsareusedtoidentifyputativecis-regulatorysequencesthathaveundergonesignificantevolution-arydivergenceamongAfricancichlids.
(4)Followingphylogeneticshadowing,were-sequencethemostdivergentregionsinapanelof18phenotypicallydiversecichlidsfromLakeMalawi.
Wesearchthesesequencesforpolymorphismsthatmayindicatethepresenceofcis-regulatoryalleles.
ThisfinalanalysisallowsustodeterminewhetherthedivergentregionsweidentifybetweenO.
niloti-cusandM.
zebraalsocontainpolymorphismscorre-latedwithopsinexpressioninthemorecloselyrelatedcichlidsofLakeMalawi.
WeusethefinalresultsofthisstudytoexaminewhichregulatoryregionsaremostlikelytocontainfunctionalregulatoryallelesthatdetermineopsinexpressioninAfricancichlids.
Wefindthatmanynon-codingregionsarehighlyconservedbetweenO.
niloticusandM.
zebra,aswellasamongthecloselyrelatedcichlidsofLakeMalawi.
However,wefindatleasttwoCNEs,fiveproximalpromoters,andtwo3'-UTRsthatexhibitsignificantdivergenceinthenumberandtypeofTFBSandmiRNAtargetsfoundbetweenO.
niloticusandM.
zebra.
WealsoidentifyatleastthreeallelesthatareweaklyassociatedwithSWS2A,RH2B,andLWSexpression-threeopsinsthatshowstrongdifferentialexpressionamongcichlidspecies.
Theseresultssuggestthatcis-regulatorysequencesmaycontributetoopsinexpressiondifferencesamongAfricancichlids,andpro-videnumerouscandidatesforfuturefunctionalstudies.
ResultsandDiscussionBACSequencingandAnalysisBACidentification,sequencing,assembly,andcomparisonWithinthecichlidgenome,theopsinsarefoundinthreeseparatetandemarrays.
SWS1isfoundaloneoncichlidlinkagegroup(LG)17;SWS2A,SWS2B,andLWSarefoundtogetherinatandemarrayonLG5[44];andRH2B,RH2Aa,andRH2Abarefoundinasec-ondtandemarrayonLG5approximately30cMfromtheSWS2-LWSarray(KLCarleton,unpublisheddata)[44].
Weidentifiedopsin-containingBACclonesforO.
niloticusbyPCRscreening[30]andforM.
zebrabyfil-terhybridization[31].
WethenshotgunsequencedeachcloneusingABISangeror454LifeSciencestechnology.
CloneIDs,estimatedsizes,sequencingmethods,assem-blystatistics,finalcontiglength,andGenBankaccessionnumbersforresultingcontigsarelistedinTable2.
TheaveragereadlengthforABI-generatedsequenceswas~700bp,whiletheaveragereadlengthfor454-gener-atedsequenceswas~110bp.
FortheO.
niloticusSWS1-containingclone,weusedacombinationofABIandTable1ListofcandidatetranscriptionfactorssurveyedinthisstudyTranscriptionFactorSymbolOMIM1#TESS2#(mice)Opsin(s)affectedRef(s)ActivatorProtein1AP-1165160T00032SWS1[37]Cone-rodhomeobox-proteinCRX/OTX602225T03461SWS2[41]NuclearFactorkappaBNFB164011T00588SWS1[37]Photoreceptor-specificnuclearreceptorPNR604485T03723*SWS[39]RetinoicAcidReceptoraRARa180240T01327SWS1[35]RetinoicAcidReceptorbRARb180220T01328SWS1[35]RetinoicAcidReceptorgRARg180190T01329SWS1[35]RetinoidXReceptoraRXRa180245T01331--RetinoidXReceptorbRXRb180246T01332--RetinoidXReceptorgRXRg180247T01333SWS[40]ThyroidHormoneReceptoraTHRa190120T01173SWS1[36]ThyroidHormoneReceptorbTHRb190160T00851*SWS1,RH2[36,38]1OnlineMendelianInheritanceinMan(http://www.
ncbi.
nlm.
nih.
gov/omim)2TranscriptionElementSearchSystem(http://www.
cbil.
upenn.
edu/cgi-bin/tess/tess)*TESS#forhumansequencesO'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page4of23454sequencessincetheassembliesbasedonABI-gener-atedreadsalonewerepoor.
Forallotherclones,weusedadditionalSangerreadstofillinthegapsandjoinallcontigsintotheirfinalBACassemblies(Table2).
Overall,thefinalassembliesofeachclonebasedonABIand454technologyjoinedanaverageof85%ofreadsintoasinglecontigthatwaswithin10-40kboftheestimatedclonesize(Table2).
Allassembliessuccess-fullycoveredtheopsin-containingregionsinO.
niloticusandM.
zebra.
WealignedeachBACassemblyfromO.
niloticusandM.
zebraandfoundthemtobehighlysimilar.
Theonlysignificantdifferencewasa6.
1kbinsertionintheM.
zebraRH2-containingBAC,locatedbetweentheRH2Aa,andRH2Abopsins(Additionalfile1).
Thisinsertionislikelyatransposon.
TheaveragepairwiseJukes-Cantor-correctedsequencedivergence(Dxy)acrosseachBACassemblywas8.
4%(±3.
1%s.
e.
).
Thisrateofsequencedivergenceisconsistentwithcomparisonsofothergenesbetweenthesespecies,anditisoneofthefirstlarge-scaleestimatesofsequencedivergencebetweenO.
niloticusandM.
zebra.
WethensubdividedeachBACassemblyintoopsinprotein-coding(CDS)andintronic(INT)sequences.
ForO.
niloticusandM.
zebra,themeanDxyacrossallopsinCDSwas3.
8%(±0.
3%),whilethedivergenceacrossallINTwas9.
5%(±1.
9%).
(Weexcludedboththefirstintronaswellasthefirstandlastsixbasesofeachintronsincetheseregionsmaycontainregulatorysequencesandsplicesitesthataremorehighlyconservedthanotherintronicregions[43]).
ComparisonoftheaverageDxyacrossallregionsrevealsthatthemeandivergenceofthefunctionallyimportantopsinCDSissignificantlylowerthanDxyacrosseithertheBACsorINTsequences(t-tests:CDSvs.
BAC,t8,0.
05=2.
60,p=0.
032;CDSvs.
INT,t27,0.
05=2.
17,p=0.
039),butthatDxybetweenBACandINTsequencesdonotdiffer(t23,0.
05=0.
08,p=0.
935).
Inadditiontoevaluatingwhichregionsofeachopsin-containingBACretainthehighestconservationandaremostlikelytobefunctional,thesedivergenceestimatesalsoprovideanimportantnullhypothesisforoursubse-quentanalysesusingphylogeneticshadowing:ingeneral,weexpectO.
niloticusandM.
zebratoshare(e.
g,exhi-bitorthologyin)~92%oftheirTFBSandmiRNAtargetsites,andexhibitdivergencein~8%.
Divergenceingreaterthan8%oftheTFBSandmiRNAtargetsitesidentifiedmayindicatesignificantcis-regulatorysequenceevolutionintheregionsexamined.
BACannotationandtheopsinrepertoireofteleostfishesInordertoperformphylogeneticfootprintingacrosstheopsinarraysofcichlids,wefirstinvestigatedthesyntenyofeachopsinarrayofO.
niloticusrelativetoseveralmodelfishspeciesusingPipMaker[45]andMultiPip-Maker[46].
Wefoundconsiderablesyntenyintheopsin-containingregionsamongO.
niloticus(tilapia),Gasterosteusaculeatus(stickleback),Oryziaslatipes(medaka),Tetraodonnigroviridis(tetraodon),andDaniorerio(zebrafish)(Figure1;Additionalfile2A).
Theclear-estexampleofthissyntenywastheSWS2-LWSopsinarray.
ThisarrayisflankedbythegenesHCFC1andGNL3Landisessentiallyco-linearinallfivefishgen-omes(Figure1;seeAdditionalfile3forthepositionandorientationofflankinggenes).
WefoundevidenceforalocalizedduplicationoftheSWS2opsinsinO.
latipesandO.
niloticus,sinceboththesespecieshavetwoadjacentSWS2opsingenes(Additionalfile4).
Clo-selyrelatedPoeciliidfishesalsopossessadjacentSWS2paralogs[47],suggestingthatthisduplicationeventprobablyoccurredatleast153-113MYAatthebaseoftheAcanthopterygii[48,49].
IncontrasttotheSWS2-LWSarray,weobservedcon-siderablevariationinopsingenecontentfortheRH2opsins.
O.
niloticusandM.
zebrapossessthreeRH2geneswhileD.
reriohasfour[50,51],G.
aculeatushastwo,andT.
nigrovirdishasonefunctionalRH2opsinandoneRH2pseudogene[52].
WethereforeusedTable2AssemblystatisticsfortheO.
niloticusandM.
zebraopsin-containingBACsSpeciesOpsinarrayCloneIDEstimatedclonesize(bp)SequencingmethodContigsize(bp)Readsassembled(%)GenBankaccessionnos.
O.
niloticusSWS1T4057DH09210,000ABI,454171,83877K+3K(95+49)JF262087SWS2-LWST4075AE05184,000ABI171,7423072(85.
1)JF262088RH2A-RH2BT4024BG04200,000ABI177,3663072(84.
2)JF262086M.
zebraSWS1Mz042C687,00045477,65279,892(95.
2)JF262085SWS2-LWSMz045P996,000454107,62443,135(93.
8)JF262084RH2A-RH2BMz088M22133,00045483,46321,758(94.
8)JF2620891EstimatedclonesizebasedonPulsedGelElectrophoresis.
O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page5of23phylogeneticanalysestoinvestigatetheorthologyoftheRH2andSWS2genesamongthesefishesandfoundthatmostRH2duplicationsarespecies-specific[53](AdditionalFile4).
Thus,syntenyintheregioncontain-ingtheRH2opsinarraywaslowerthanintheSWS2-LWSarray,butwasstilllargelyco-linearbetweenO.
niloticus,G.
aculeatus,andT.
nigroviridis(Additionalfile2B).
ThegenesSLC6A13-likeandSYNPRflanktheRH2opsinsinthesefishes(Figure1;Additionalfile3).
SyntenyintheregionsurroundingtheSWS1opsinwasdifficulttoassessduetospecies-specificdeletionsandpoorgenomeassembly.
TheT.
nigrovirdisgenomeassemblylackstheSWS1opsinaltogether,andthisregionisfoundwithinanunorderedchromosomeorultracontiginboththeG.
aculeatusandO.
latipesgen-omes.
ForG.
aculeatus,wefoundasmall92kbregioncontainingtheSWS1opsinthatwascollinearwiththeO.
niloticusBACsequence,butwhichcontainedonelargeinversion.
ForO.
latipes,wefoundanevensmaller60kbregionthatwassyntenicforonly11kbsurround-ingtheSWS1opsin.
SyntenywithD.
reriowasalsogen-erallylow(Additionalfile2C).
Therefore,despitethelackofSWS1duplicatescomparedtotheSWS2orRH2opsins,theSWS1regionisstillpoorlyassembledintheexistingannotationsofseveralteleostgenomes,poten-tiallycomplicatingdirectcomparisonsofsyntenyinthisregion.
Inthesespecies,theSWS1opsinappearstobeflankedbythegenesTNPO3andCALUA(Figure1;Additionalfile3).
AnalysisofConservedNon-CodingElements(CNEs)PhylogeneticfootprintingtoidentifyCNEsWeusedMultiPipMaker[46]tohighlightnon-codingelementssurroundingeachopsingenearrayfromO.
niloticustoD.
rerio,representingnearly300MYoffishevolution[49].
Theresultingplotsillustrateatleast20conservednon-codingelements(CNEs)surroundingtheopsingenearraysofO.
niloticusandtheotherfishspe-ciesexamined(redbarsinFigure1).
Wealsofoundsixregionsofputativelyhighconservationthatarelargelycomposedofrepetitivesequence(greenbarsinFigure1),whichwedidnotanalyzefurther.
TheconservationoftheseCNEsoverseveralmillionyearsoffishevolu-tionsuggeststhattheycontainfunctionallyimportantregulatorymodulesnecessaryforgeneexpression.
AtleastoneCNEweidentifiedthroughphylogeneticfootprintingisorthologoustoothervertebratecis-regu-latorysequences.
CNE7(highlightedinFigure1andlocatedbetweentheSWS2BandLWSopsins)consistsoftwonon-contiguousregionsofhighconservationinpufferfish,stickleback,medaka,swordtails,andcichlids[47](Figure1).
Thefirstregion,CNE7a,wasalsoiden-tifiedfollowingacomparativeanalysisofopsin-contain-ingBACsfromswordtails(Xiphophorushelleri)[47].
ThroughBLASTandmirbase[54],wefoundthatCNE7aismostsimilartozebrafishmiRNAdre-miR-726(score173.
3,e-value=0.
006),andthesamegenomicregionfromzebrafishisidenticaltothismiRNA(Figure2).
Dre-miR-726isexpressedintheretinaoflarvalandadultzebrafish[55].
SincemanymiRNAsaretranscribedalongwiththegenestheyregulate,theproximityofmiR-726totheSWS2andLWSopsinssuggeststhatitcouldplayaroleinopsinregulation.
The~90bpCNEencodingmir-726isconservedinnumerousothertaxaaswell,includingadditionalfishes,frogs,andlizards[47,56].
Thesecondhighlyconservedregion,CNE7b,isposi-tionallyandstructurallyorthologoustothemammalianLWSlocuscontrolregion(LWS-LCR;Figure2B)[47,56,57].
Thisenhancerislocated~3.
8kbupstreamoftheLWSopsininO.
niloticusandothervertebrates,includinghumans.
TheLWS-LCRishypothesizedtoenhanceLWSexpressionineutherianmammalsbyloopingandbindingtotheLWSproximalpromoter[57-59].
Wangetal.
[59]demonstratedthatthehumanorthologofthissequencecanfunctionasanenhancerofbothLWSandMWSopsinexpressioninmice.
Addi-tionally,arecentstudyofLWSregulationinzebrafishalsoidentifiedasimilarsequenceatthispositionthatmodulatesLWSexpressioninthatspecies,whichtheynamedtheLWSactivatingregion(LAR)[60].
Compari-sonofthemammalianLWS-LCR,thezebrafishLAR,andCNE7bfromcichlidsandotherteleostsrevealsahighdegreeofsequencesimilarityamongtheseregions(Figure2B).
InFigure2B,wealsohighlightseveralcon-servedtranscriptionfactorbindingsitescommontoeachsequence,includingsitesforCRX,THR,andAP-1(Figure2B;seealsoTable1).
Thus,ourresultsdemon-stratetheeffectivenessofthephylogeneticfootprintingmethodforidentifyingfunctionalcis-regulatorysequencesnecessaryforvertebrateopsinexpression.
ItisthereforepossiblethattheremainderoftheCNEsweidentifyalsoencodecis-regulatorysequencesnecessaryforthecorrectspatialanddevelopmentalexpressionoftheopsinsincichlids.
Wenotethatourpresentstudyfocusesonsmallregionsofhighconservationwithina~30kbwindowofnon-codingsequencesurroundingtheopsinarrays,butthatcis-regulatorysequencesmayoftenresidetensorhundredsofkilobasesfromthegenestheyregulate.
However,tworecentanalysesofgeneraltranscriptionfactorbindingsitesfoundthatfunctionalbindingsitesgenerallyclusterinregions1kbaroundtheproximalpromoterofeachgene[61,62].
Thisobservationsug-geststhatafocusedstudyofconservedelementswithinorneartheopsinsisareasonablestrategyforthisinitialstudy.
AFASTAfileofallCNEsequencesfromO.
nilo-ticusandM.
zebraisprovidedinAdditionalfile5.
O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page6of23TFBSsearchandphylogeneticshadowingofCNEsWecomparedthe20CNEsidentifiedbetweenO.
niloti-cusandM.
zebraandfoundmanytobehighlycon-served;however,wefoundnoidentifiableorthologsbetweenO.
niloticusandM.
zebraforCNEs6or19.
FortheremainingCNEs,theaveragepairwisesequencedivergencebetweenO.
niloticusandM.
zebrawas4.
2%(±0.
5%),whichissignificantlylessthanthemeanDxyofintrons(9.
5%,t-test:t38,0.
05=2.
99,p=0.
005).
Thisresultsuggeststhattheconservednon-codingregionsidentifiedamongO.
niloticusandotherfisheshaveremainedconservedamongAfricancichlidsaswell.
WeusedtheTranscriptionElementSearchSystem[63]tocomputationallysearchallorthologousCNEsforbindingsitescorrespondingtotwelvetranscriptionfac-torsthathavebeenassociatedwithopsinexpressioninfishesandothervertebratesincludingthyroidhormoneandretinoicacidreceptors[34-37,39,41,64,65].
Acom-pletelistofthesetranscriptionfactorsandtheirasso-ciatedopsinsispresentedinTable1.
Wefoundcomputationally-predictedbindingsitesforthesefunc-tionallyimportanttranscriptionfactorsinallbutoneoftheCNEssurveyed(Table3;seeAdditionalfile6fordetailedcountsofallTFBS).
OnlyCNE10lackedbind-ingsitesforanyofthetwelvetranscriptionfactorsineitherspeciesexamined.
Withintheremainingsequenceswefoundbindingsitesforalltwelvetran-scriptionfactorsexceptPNRandRXRg.
Afterrelaxingourmatchingcriteria,westillfailedtofindbindingsitesforthesetwotranscriptionfactors(datanotshown).
InbothO.
niloticusandM.
zebra,bindingsitesforAP-1andCRXwereextremelyabundant,althoughbindingsitesforeachofthreeretinoicacidreceptors(RARs)andTHRbwerealsocommon(Additionalfile6).
WefoundseveralCNEswithahighdensityoftranscriptionfactorbindingsitesgiventhetotalsequencelengthsur-veyed-generally9TFBSormore(seeAdditionalfile6).
ForO.
niloticusthesehigh-densityCNEsareCNEs2,3,13,15,19,and20,andforM.
zebratheseareCNEs2,8,11,13,15,and20.
DuetotheirpotentialenrichmentforfunctionalTFBSsrelativetootherCNEs,webelievetheseeightCNEsrepresentthemostlikelycandidatesforfunctionalcis-regulatorsofopsinexpressioninfishes.
Consistentwiththehighsimilarityoftheirsequences,theresultsofourTFBSsearchdifferedverylittlebetweenO.
niloticusandM.
zebra.
Weusedexactbino-mialteststocomparetheproportionofsharedanddivergentTFBSsobservedbetweenO.
niloticusandM.
zebratothenullratioof92:8(seeabove).
TreatingeachTFBSindependently,wecountedeachnon-orthologousordivergentTFBSasasuccess,eachorthologousorsharedTFBSasafailure,thentestedthehypothesisthatthetrueprobabilityofsuccess(proportionofdivergentTFBS,Pdiv)was>8%.
Of17testableCNEs,wefoundthatO.
niloticusandM.
zebradifferedsignificantlyfromthisnullexpectationatfourCNEs:CNEs3,4,15,and18(Table3).
AfterBonferronicorrectionformultiplecomparisons,however,onlytheresultsforCNE3remainedsignificant(exactbinomialtest:divergentTFBS=7,totalTFBS=8,Pdiv=87.
5%,p25%standoutasstatisticaloutliers,andonlythosewithPdiv>80%remainsignificantaftercorrectionformultiplecomparisons.
Inthefutureweaimtoperformmorenuanced,sequence-basedtestsofcis-regulatorydiver-genceincichlids.
Wepresentthesetestsforcis-regula-torydivergenceasafirststepinthisprocess.
AnalysisofProximalPromoterregionsPhylogeneticfootprintingofopsinproximalpromotersTheMultiPipplotsshowninFigure1reveal20CNEsupstreamoftheopsins,butalsoshowseveralregionsofhighconservationwithinthe5'proximalpromoterofmultipleopsinsaswell.
Inparticular,SWS2A,SWS2B,andLWSallexhibitregionsofhighconservationinthefirst1kbofsequenceupstreamoftheirtranslationstartsite(TSS).
FortheLWSopsin,thisregionofconserva-tionspansnearly0.
7kboftheproximalpromoterinmultiplefishspecies,includingG.
aculeatus,O.
latipes,andT.
nigroviridis(Figure1B).
RH2AandRH2AalsoexhibitsomesmallregionsofhighconservationjustupstreamoftheirTSSs,whichprobablyreflectthe5'-UTRregion.
Additionally,thepromoterupstreamofRH2Balsocontainssomeconservedregionsofrepetitivesequence(Figure1C).
Itiscompellingthatmanyoftheopsinsexhibitstrongconservationofsequenceswithin1kboftheirTSSs,whichweusetodefinetheproximalpromoter,becausethetruepromoterregionsforthesegenesareunknownincichlids.
However,importantcis-regulatorysequenceshavebeenidentifiedincloseproxi-mitytotheopsingenesinotherfishspecies.
Inparticu-lar,severalCRXtranscriptionfactorbindingsitesfoundwithin500bpoftheSWS2opsinregulatetheexpressionofthisgeneinD.
rerio[41].
Therefore,theconservationweobserveupstreamoftheSWS2A,SWS2B,andLWSopsinsmayindicatethepresenceofadditionalcis-regu-latorysequenceswithintheproximalpromotersofthesegenesaswell.
AFASTAfileofallopsinandnon-opsinpromotersequences(seebelow)fromO.
niloticusandM.
zebraispresentedinAdditionalfile5.
TFBSsearchandphylogeneticshadowingofopsinproximalpromotersThedistributionandnumberofTFBSsfoundwithintheproximalpromoterregionofeachopsinwassimilartothosefoundintheCNEs.
Withineachopsin'sproximalpromoter,wefoundthatAP-1andCRXbindingsiteswerenearlyubiquitous(Figure3).
BindingsitesforNFB,RARa,RARb,RXRbandTHRbwerealsocom-mon,andweonceagainfoundnobindingsitesforPNRandRXRg.
TheabsenceofbindingsitesforPNRandRXRginboththeCNEsandpromotersmayrule-outthesefactorsascandidatetrans-regulatorsofcichlidopsinexpressiondifferences;howeverthelackofthesefactorscouldalsobeduetobiasesinthewayTESSidentifiesbindingsites.
Interestingly,wefoundseveralCRXbindingsitesdirectlyupstreamoftheSWS2AandSWS2Bopsins(Figure3).
ThesebindingsitescouldpotentiallyfunctionasregulatorsofSWS2opsinexpres-sionincichlidsastheydoinzebrafish[41].
Pairwisesequencedivergenceintheproximalpromo-terregionswasgreaterthanfortheotherregionsexam-ined.
TheaverageDxyoftheproximalpromoterswas10.
2%(±3.
2%),whichdifferedsignificantlyfromthemeanofCNEs(4.
2%,t-test:t23,0.
05=2.
48,p=0.
021),butnottheintrons(9.
5%,t-test:t27,0.
05=0.
14,p=0.
89).
Thisresultsuggeststhattheopsinpromoterregionsofcichlidsmayexhibitgreaterdivergenceinputativecis-regulatorysequencesthantheCNEs.
Indeed,wefoundthatO.
niloticusandM.
zebraexhibitedsignif-icantdivergenceintheirTFBSprofilesforsixofthesevenproximalpromotersexamined(Figure3);how-ever,followingcorrectionformultiplehypothesistest-ing,onlyfiveoftheseremainedsignificant:SWS1,SWS2A,RH2B,RH2AandRH2A(Figure3;seealsoTable3).
O.
niloticusandM.
zebradifferdramaticallyintheexpressionofeachofthesegenes[32],suggestingthattheirdivergenttranscriptionfactorprofilescouldexplainthesedifferences.
AcomparisonofwhichTFBSdifferbetweenO.
niloticusandM.
zebrarevealsaslightover-representationofCRXsitesinO.
niloticus(17vs.
7),andofTHRasitesinM.
zebra(4vs.
0)(Figure3).
Usingphylogeneticshadowing,weidentifiedfivecichlidopsinswithpromotersequencesthatexhibitsig-nificantdivergenceintheirbindingsiteprofilesfor12transcriptionfactors.
Wenote,however,thatbyfocusingononlytheseTFBSs,wepotentiallymissmanyinterest-ingpatternsofdivergenceintranscriptionfactorsthathavenotalreadybeenassociatedwithvertebrateopsinexpression.
AcomprehensivesearchofallTFBSsidenti-fiedbyTESScouldpotentiallypickupthesemissedpat-terns,butsuchasearchwouldbeextremelycumbersomeandsubjecttomanyfalsepositives[66].
O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page9of23-953-562-897-796-429-156CRXCRX-501AP-1THRα-47-971-696-784CRXTHRβCRXTHRβRARAP-1CRXCRXTHRβ-297CRX-251RXRαCRX-165THRβAP-1RH2B(Pdiv=68.
1%,p0.
05)B-397-808-999THRα-289CRX-96CRXCRX-725RARα-593NFκBAP-1RXRβRARNFκBAP-1AP-1-937-543-734-429-144AP-1CRXCRX-19-842-755CRXCRXAP-1AP-1CRXAP-1THRβCRXRXRαCRXNFκBCRXAP-1AP-1AP-1CRXAP-1RH2Aα(Pdiv=47.
6%,p0.
05)GAP-1AP-1SWS1(Pdiv=64.
3%,p0.
05SWS2B-208C*T0.
4170.
05SWS2B-551bpindel0.
4440.
2400.
789>0.
05SWS2A-224*C*T0.
2220.
1271.
037>0.
05SWS2A-217*8bpindel0.
1940.
3921.
8410.
087RH2B-308C*G0.
167-0.
245-0.
893>0.
05RH2B-161C*T0.
1110.
2633.
4470.
004LWS-208C*T0.
1670.
3551.
002>0.
05CNE-7183A*T0.
2220.
055-0.
673>0.
05CNE-7570C*T0.
4170.
6082.
2370.
041SWS2B-UTR197A*C0.
3060.
3491.
264>0.
05*ThesepolymorphismsinterruptCRXtranscriptionfactorbindingsitesO'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page15of23eachpromoterversus~400bpforeachCNEandUTR),andtheincreasedpowertodetectsignificantdivergencefromnullexpectationsaffordedbythelargenumberofTFBSfoundwithintheproximalpromoters(wefound~22TFBSwithineachpromoterversus~6TFBS/miRNAtargetsiteswithineachCNEandUTR).
IftheoverallproportionofdivergentTFBS/miRNAtargetsites(Pdiv)isusedtoidentifythosenon-codingregionsmostlikelytocontainfunctionalcis-regulatoryalleles,theproximalpromoterregionsstillexhibitthehighestproportionofdivergentregulatoryregions,althoughtheadvantageisonlyslight.
Onlyabout55%ofTFBSaresharedbetweenO.
niloticusandM.
zebrapro-moters,while45%aredivergent(Figure5B).
Incontrast,theCNEsand3'-UTRsexhibitlower(andverysimilar)proportionsofsharedversusdivergentTFBS/miRNAtargetsites(~67%sharedand~33%divergent;Figure5B).
Inthiscase,itisdifficulttoconfidentlyconcludethat5'promoterregionsaremorelikelytocontainfunctionalallelesthatregulateopsinexpression,althoughthedataaresuggestive.
WhenbothpairwisedivergenceandtheproportiondivergentTFBS/miRNAtargetsitesaretakenintoaccount,wefindthatregionsthatexhibitstatisticallysignificantdivergencearenotnecessarilythoseregionsthatexhibitgreaterpairwisesequencedivergence(Figure5C).
Infact,theregionswiththehighestPdivalsoexhibitsomeofthelowestDxyvalues.
Thisresultsuggeststhattheincreasednumberofstatisticallydivergentpromoterregionsweobserveisnotafunctionofsequencedivergence,butratherincreasedstatisticalpoweraffordedbythegreaterlengthofthesequencessurveyedandthegreaternumberofTFBSfound.
Additionally,ourresultsshowthatthemajorityofthenon-codingregionsexaminedexhibitPdivvaluesnear37%,withamedianof30%(Figure5C).
Thisobserva-tionsuggeststhatthe8%divergencecriterionweusedasnullmodelforevolutionarydivergenceislikelytoolowandalsosuggeststhatourpowerformanyregionswasinadequateduetothesmallnumberofTFBSormiRNAtargetsitesidentified(seeabove).
Butevenwhenamoreliberalnulldivergencevalueof30%isused,ourresultslargelyremainconsistent:O.
niloticusandM.
zebrastillexhibitsignificantdivergenceintheirTFBSandmiRNAtargetprofilesforCNEs3and4(locatedneartheSWS2Aopsin),theproximalpromotersforRH2BandSWS1,andthe3'-UTRforSWS2B(p5kb)contigstothefinishedO.
niloticusBACsequencesinSequencherv4.
9andonceagaindesignedPCRprimerstosequenceacrossthegaps.
WeannotatedtheBACsequencesforbothO.
niloticusandM.
zebrausingBLAST[104].
Finally,weperformedaglobalalignmentofeachBACfromO.
niloticusandM.
zebraintheprogramwgVISTA[105].
WemeasuredsequencesimilarityanddivergenceacrosseachBACusingthephylipprogramdnadist,implementedintheMobyleonlinebioinfor-maticsserver[106].
Whenmeasuringpairwisesequencedivergence(Dxy),weusedtheJukes-Cantornucleotidemodeltocorrectformultiplehits.
WerepeatedthesemeasurementsforeachoftheCNEs,promoterregions,and3'-UTRs.
WecomparedDxyamongeachoftheseregionsandtheentireBACsequencesusingt-testsimplementedinthestatisticalsoftwarepackageRv2.
10.
0[107].
Priortoperformingalltests,wetrans-formedtheDxyscoresbylog10inordertomeettheassumptionofnormalityoferrors.
PhylogeneticanalysesWegeneratedphylogeniesoftheteleostRH2andSWS2opsinsinordertoidentifyorthologousopsinsamongthefocalfishgenomesexamined.
Weaccessedallrele-vantopsinsequencesfromthegenomeassemblieslistedaboveviaBLAT.
WealignedbothopsindatasetsusingtheE-INS-istrategyofthemultiplealignmentprogramMAFFTv6.
0[108]andthenchoseanappropriatemodelofnucleotidesubstitutionviatheprogramjMo-delTestv0.
1.
1[109].
ThismodelwasTIM3ef+GforboththeRH2andSWS2alignments.
WethenusedthismodelandthecorrespondingparametersestimatedbyjModelTesttogenerateNeighbor-JoiningtreesfortheopsinswithMaximumLikelihood-correcteddistances.
FortheRH2/SWS2datasets,theseparametersincludedthenucleotidesubstitutionratematrix(A-C:0.
601/0.
617;A-G:1.
470/1.
734;A-T:1.
00/1.
00;C-G:0.
601/0.
617;C-T:2.
729/2.
877;G-T:0.
599/0.
155)andtheshapeofthegammadistribution(0.
507/0.
577).
Wemea-suredthenodalsupportofthesetreeswith1000boot-strapreplicates.
WerootedbothtreesusingtheLWS-1opsinofzebrafish.
Identificationofconservednon-codingelementsWeusedphylogeneticfootprinting[18]toidentifyputa-tivecis-regulatoryelementsbysearchingforconservednon-codingelements(CNEs)surroundingtheopsingenearrays.
Todothis,weidentified100-300kbregionsoforthologybetweentheO.
niloticusBACsequencesandthegenomeassembliesoffourteleostfishesusingBLATandtheUCSCgenomebrowser.
Theadditionalgenomeswerestickleback(Gasterosteusacu-leatus,BroadInstitutev1.
0,February2006),medaka(Oryziaslatipes,NationalInstituteofGeneticsandtheUniversityofTokyov1.
0,October2005),pufferfish(Tet-raodonnigroviridis,GeoscopeandBroadInstitutev7,February2004),andzebrafish(Daniorerio,TrustSangerInstitutezv8,December2008).
WethendeterminedthelocationofknownopsingenesandexaminedsyntenyacrosstheseregionsviaDOTplotsgeneratedintheprogramPipMaker[45](foranexampleseeAdditionalFile2).
RegionsofhighsyntenysurroundingtheopsinswerethenidentifiedusingMultiPipMaker[46].
WedefinedaCNEasanyregion≥50bplongthatwascon-served(>60%sequenceidentity)betweenOreochromisniloticusandatleastoneotherteleostspecies(Oryziaslatipes,Gasterosteusaculeatus,andTetraodonnigroviri-dis).
Ineachcase,weattemptedtoanalyzeasmanyCNEsaspossible,butacknowledgethatsomesmallregionsmayhavebeenmissed.
ProfilingoftranscriptionfactorbindingsitesandPhylogeneticshadowingWeidentifiedbindingsiteswithineachCNEaswellastheproximalpromoterslocatedapproximately1kbupstreamofeachopsin'stranslationstartsiteusingtheTranscriptionElementSearchSystem,TESSv6.
0[63].
WealteredthedefaultsearchparametersofTESSbychangingtheminimumlog-likelihoodratioscorefrom12to9.
WethenlimitedoursearchresultstohighqualitymatchesbyacceptingonlythosehitsthatmetO'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page18of23threecriteria:(1)alog-likelihood(La)score≥9.
0,(2)aratiooftheactuallog-likelihoodscoretothemaximumpossiblelog-likelihood(Lq)score≥80%,and(3)aprob-abilityvalueforthelog-likelihoodscore(Lpv)<0.
05.
AlthoughTESScanpotentiallyidentifybindingsitesformanydifferenttranscriptionfactors,wewereprimarilyinterestedinthosefactorsthathavebeenshowntoinfluenceopsinexpressioninfishandothervertebrates(Table1).
FollowingtheautomatedsearchinTESS,wemanuallysearchedthelistsforduplicatesitesateachposition,andremovedthempriortofurtheranalysis.
Forphylogeneticshadowing,weanalyzedthenumberofsharedanddivergenttranscriptionfactorbindingsitesfoundineachCNEandopsinproximalpromoterfromO.
niloticusandM.
zebra.
Wecountedthetotalnumberofbindingsitesorthologousinbothspecies,aswellasthosethatwerefoundinonlyonespeciesortheother.
WecalculatedtheproportionofdivergentTFBSs(Pdiv)as(D/(D+S))*100,whereDisthenumberofdiver-gentTFBSandSisthenumberofsharedsites.
WecomparedtheobservedproportionofdivergentsitestothenullproportionsuggestedbytheglobalsequencesimilarityoftheO.
niloticusandM.
zebraBACs(92%versus8%).
Wetestedtheindependencebetweentheseobservedandexpectedproportionsusingexactbinomialtests[110]implementedintheRstatisticalsoftwarepackage.
TocontroltheTypeIerrorrateforeachregionexamined,wecalculatedBonferroni-correctedp-valuesforalltestsinR.
ForphylogeneticshadowingbetweenO.
niloticusandM.
zebra,thecorrectedsignifi-cancethresholdwasa=0.
05/31=0.
0016.
Finally,wealsocomparedtheaveragenumberofbindingsitesforeachtranscriptionfactorbetweentheproximalpromotersoftheO.
niloticusopsinsandsevenrandomlychosen,non-opsingenesfromadraftassem-blyoftheO.
niloticusgenome(availableathttp://www.
BouillaBase.
org;accessedOctober2010).
ThesegeneswereACTG1,AMPD3,DHCR7,ENSGAC000000020282,IGFALS,KCNJ9,andREEP1.
ProximalpromotersfromtheserandomlychosensequenceswereidentifiedbasedoncomparisonoftheO.
niloticusgeneswithortholo-gousregionsfromthesticklebackgenome.
ComparisonoftheaveragenumberofbindingsitesacrossallopsinsandtranscriptionfactorswasperformedusingaWil-coxonpairedsigned-ranktestcomputedinR.
ComparisonofopsinexpressionandTFbindingsiteprofilesWeevaluatedthecorrelationbetweenthetranscriptionfactorbindingsitesintheproximalpromoterofeachopsinandtheexpressionofeachopsinamongdevelop-ingO.
niloticusfryusingMantel'stestoftwodistancematrices.
WegeneratedEuclideandistancematricesofthetotalnumberofbindingsitesfor12transcriptionfactorswithintheproximalpromoterregionofeachopsinaswellasthepercentoftotalopsinexpressionfromdevelopingO.
niloticusfry,reportedinCarletonetal.
[32].
WecalculatedMantel'stestusingthe'mantel.
randtest'functionfromtheRpackageade4[111].
Approximatep-valueswerecalculatedfollowing500randomizationsofeachmatrix.
Alltranscriptionfactornumbersandexpressionvalueswerestandardizedpriortoclustering.
Wealsoexpandedthisanalysistotheentireproximalpromoterregionaftercalculatingasequencesimilaritymatrixfortheentireproximalpro-moterusingthephylipprogramdnadist.
ProfilingofmicroRNAstargetsitesWesearchedthe3'-UTRsofeachopsinforbindingsitesmatchingthetargetseedofknownmiRNAs(miRNA)viatheSeedMatchalgorithmpreviouslyusedtoidentifymiRNAtargetsincichlidUTRs[72].
ThisalgorithmissimilartotheTargetScanSalgorithmusedinotherstu-diestoidentifymiRNAtargets[112].
Briefly,non-redun-dantfishmiRNAtargetswereobtainedfrommiRBase(http://www.
mirbase.
org[54];accessedJune2010)andsupplementedwithseveralmiRNAtargetsequencesidentifiedincichlids[72].
Wesearchedeachopsin3'-UTR–definedasthe~500bpregionbetweenthetran-scriptionendsiteandthepolyadenylationsite(AATAAA)–forsequencesmatchingtheseedsofmiR-NAsfromthisnon-redundantlibrary.
Inordertoaccountforthehighrateoffalse-positivesgeneratedbysimplysearchingformatchingseedsites,wealignedthe3'-UTRofeachcichlidopsinwiththosefromG.
aculea-tus,O.
latipes,T.
nigrovirdis,theJapanesepufferfish(Tetrapdonrubripes),andD.
rerioinordertoidentifysitesthatwereconservedacrossmultiplefishspecies.
Forthispurposewedefinedthefirst1kbofsequencedownstreamoftheselatterspecies'opsinsasthe3'-UTRandalignedthesetothecichlidsequenceswithMLagan[113].
Toaccountforerrorsinthealignmentofortholo-gous3'-UTRs,wecountedasconservedthesamemiRNAtargetsitefoundwithin50bpofeachotheracrossspecies.
Forcichlidopsinsthatlackedorthologsintheotherspecies,weusedthenearestparalog(seeAdditionalfile4).
ResequencingofputativeregulatorysequencesinLakeMalawicichlidsWegeneratedapanelof18LakeMalawicichlidsthatvaryinopsingeneexpression.
Inoneindividualperspe-cies,wesequencedapproximately1kbofDNAupstreamofthetranslationstartsiteforfiveopsinsandCNE7,aswellas0.
5kbdownstreamoftheSWS2BandLWSopsins.
WegeneratedprimersfortheseregionsbasedontheO.
niloticusandM.
zebraBACassemblies.
ThetaxasampledarelistedinAdditionalfile11alongO'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page19of23withtheirGenBankaccessionnumbers;theprimersusedtogeneratethesesequencesarelistedinAdditionalFile12.
Wemeasuredopsinexpressionforeachindivi-dualfollowingtheprotocolsdescribedinSpadyetal.
[28]andHofmannetal.
[26].
Describedbriefly,wedis-sectedwholeretinasfromindividualfishandextractedwholeRNAfromthemusingQiagenQiashredderandRNeasyRNAextractionkits(Valencia,CA).
Wequanti-fiedeachRNAsampleviaspectralabsorption,andthenreversetranscribed0.
5μgusingSuperscriptIII(Invitro-gen).
WeusedpreviouslydevelopedTaqmanprimersandprobestoindividuallyquantifytheexpressionofeachopsininthesesamples;however,asinourpreviousstudies[26,28],wequantifiedtheexpressionofthetwoRH2Aparalogsjointly.
ReactionefficienciesforeachopsinwerestandardizedrelativetoaninternalconstructdevelopedespeciallyforthispurposeanddescribedinSpadyetal.
[28].
Followingre-sequencingofthecandidatecis-regula-toryregions,weestimatedpolymorphismstatisticsfortheresultingsequences,andalsoperformedasliding-windowanalysisofnucleotidediversity(π),inthepro-gramDnaSPv5[114].
Forthesliding-windowanalysis,weignoredallgapsandspecifiedawindowlengthof50bpandastepsizeof10bp.
Finally,wecalculatedthestatisticalassociationbetweenpolymorphismsfoundinCRXbindingsitesandotherpeaksofnucleotidediver-sityamongthesampledtaxausinglinearregressionintheprogramgPLINKv1.
07[115].
Foreachtest,weesti-matedtheassociationofeachlocuswiththeexpressionofitsdownstreamopsinunderanadditivegeneticmodel,usingmembershipinoneoftwomajorphyloge-neticclades(mbunaandutaka;seeAdditionalfile8)asacovariate.
AdditionalmaterialAdditionalfile1:Synteny(Pipplots)ofO.
niloticusandM.
zebraopsin-containingBACsequences.
Additionalfile2:Synteny(Pipplots)ofO.
niloticusopsin-containingBACsagainstthegenomeassembliesoffiveteleostspecies.
Additionalfile3:Opsingenecontentoffiveteleostgenomes.
Phylogenyoftheteleosttaxaisrecreatedfrom[118].
Additionalfile4:OrthologyofRH2andSWS2opsinparalogsfromfiveteleostfishgenomes.
A)RH2phylogeny.
B)SWS2phylogeny.
Inbothcases,brokenlinesindicatebranchesleadingfromtheoutgroupthatwereshortenedtofiteachtreeintothefigure;thesedonotrepresentmissingorincompletebranchlengthinformation.
Additionalfile5:FASTAfileof20conservednon-codingelements(CNEs),promotersequences,3-UTRs,andsevennon-opsinpromotersfromO.
niloticusandM.
zebra(80sequencestotal).
Additionalfile6:Completetranscriptionfactorbindingsiteprofilesfor20CNEsinO.
niloticusandM.
zebra.
Additionalfile7:CompletelistofmiRNAtargetsitesidentifiedwithinthe3'-UTRofeachopsininO.
niloticusandM.
zebra.
Additionalfile8:Names,opsinexpressionvalues,andpolymorphismsfoundwithintheproximalpromotersof18LakeMalawicichlidspecies.
Additionalfile9:LengthandDxyscoresbetweenO.
niloticusandM.
zebraforeachcodingandnon-codingregionexamined.
Additionalfile10:Identificationofopsin-containingBACsfromFingerPrintedContigs.
A-C)BACsfingerprintedcontigcontainingtheSWS2A-SWS2B-LWS(A)RH2(B)andSWS1(C)genes.
ArrowsindicatePCRproductssuccessfullyamplifiedusingprimersdesignedtoBACendsequencesforcloneswhosenamesareshowninthecorrespondingcolor.
Coloredcirclesaretheapproximatelocatesofeachgene.
Additionalfile11:GenBankaccessionnumbersforallsequencesgeneratedinthisstudy.
Additionalfile12:Primersusedtoamplifyandsequencetheproximalpromoterregionsand3'-UTRofseveralopsinsfrom18LakeMalawicichlidspecies.
AbbreviationsBAC:bacterialartificialchromosome;CDS:protein-codingsequence;CNE:conservednon-codingelement;Dxy:pairwisesequencedivergence;HWE:Hardy-Weinbergequilibrium;INT:intronicsequence;LG:linkagegroup;MAF:minorallelefrequency;miRNA:microRNA;Pdiv:proportiondivergenceTFBS/miRNAtargetsites;PRO:proximalpromoterregion;QTL:quantitativetraitlocus;SNP:singlenucleotidepolymorphism;TFBS:transcriptionfactorbindingsite;TSS:translationstartsite;UTR:untranslatedregionAcknowledgementsWethankTakayukiKatagiriformakingtheOreochromisniloticusBACclonelibraryandBoYoungLeeforpoolingthislibraryforPCRscreening.
WealsothankFredericaDiPalmaforgeneratingtheMeteriaclimazebralibraryandCelesteKiddforscreeningthislibraryfortheopsin-containingBACs.
ThisworkwassupportedwithgrantstoKLCfromNSF(IOS-0841270),NIH(R15EY016721-01)andtheUniversityofMaryland.
KEOwassupportedbyaWayneT.
andMaryT.
HockmeyerDoctoralFellowshipandanAnnG.
WylieDissertationFellowshipfromtheUniversityofMaryland.
Authordetails1DepartmentofBiology,UniversityofMaryland,CollegePark,MD20742,USA.
2SchoolofBiology,PetitInstituteforBioengineeringandBioscience,GeorgiaInstituteofTechnology,Atlanta,GA30332USA.
3GenomeProjectSolutions,Hercules,CA94547,USA.
4DepartmentofIntegrativeBiology,UniversityofCalifornia,Berkeley,CA94720,USA.
Authors'contributionsKEOparticipatedinBACannotation,carriedoutthesurveyoftranscriptionfactorbindingsites,participatedinthesequencingofopsinproximalpromoters,participatedinthesurveyofmiRNAtargetsites,performedallstatisticalanalysis,andwrotethemanuscript.
DSparticipatedintheBACassemblyandannotation.
ZNandJSbothparticipatedinthesequencingofopsinproximalpromoters.
SDEsequencedtheLWSandSWS2B3'-UTRs.
YHLandJTSperformedthesearchofmicroRNAtargetsites.
JLBperformedtheBACsequencing.
KLCdesignedthestudy;aidedintheBACscreening,sequencing,andassembly;participatedinBACannotation;carriedouttheanalysisofopsingeneexpression,andparticipatedinthedraftingofthemanuscript.
Allauthorsreadandapprovedthefinalmanuscript.
Received:18January2011Accepted:9May2011Published:9May2011References1.
CarrollSB:Evolutionattwolevels:ongenesandform.
PLoSBiol2005,3:e245.
2.
HoekstraHE,CoyneJA:Thelocusofevolution:evodevoandthegeneticsofadaptation.
Evolution2007,61:995-1016.
3.
WrayGA:Theevolutionarysignificanceofcis-regulatorymutations.
NatRevGenet2007,8:206-216.
O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page20of234.
HoekstraHE,HirschmannRJ,BundeyRA,InselPA,CrosslandJP:ASingleAminoAcidMutationContributestoAdaptiveBeachMouseColorPattern.
Science2006,313:101-104.
5.
JessenTH,WeberRE,FermiG,TameJ,BraunitzerG:Adaptationofbirdhemoglobinstohighaltitudes:demonstrationofmolecularmechanismbyproteinengineering.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica1991,88:6519-6522.
6.
YokoyamaS,ZhangH,RadlwimmerFB,BlowNS:AdaptiveevolutionofcolorvisionoftheComorancoelacanth(Latimeriachalumnae).
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica1999,96:6279-6284.
7.
ChanYF,MarksME,JonesFC,VillarrealG,ShapiroMD,BradySD,SouthwickAM,AbsherDM,GrimwoodJ,SchmutzJ,etal:AdaptiveEvolutionofPelvicReductioninSticklebacksbyRecurrentDeletionofaPitx1Enhancer.
Science2010,327:302-305.
8.
JeongS,RebeizM,AndolfattoP,WernerT,TrueJ,CarrollSB:TheevolutionofgeneregulationunderliesamorphologicaldifferencebetweentwoDrosophilasisterspecies.
Cell2008,132:783-793.
9.
TishkoffSA,ReedFA,RanciaroA,VoightBF,BabbittCC,SilvermanJS,PowellK,MortensenHM,HirboJB,OsmanM,etal:ConvergentadaptationofhumanlactasepersistenceinAfricaandEurope.
NatGenet2007,39:31-40.
10.
LemosB,AraripeLO,FontanillasP,HartlDL:Dominanceandtheevolutionaryaccumulationofcis-andtrans-effectsongeneexpression.
ProcNatlAcadSciUSA2008,105:14471-14476.
11.
HartlDL,ClarkAG:PrinciplesofPopulationGenetics.
4edition.
SunderlandMA:SinaurAssociates,Inc.
;2006.
12.
BermanBP,NibuY,PfeifferBD,TomancakP,CelnikerSE,LevineM,RubinGM,EisenMB:Exploitingtranscriptionfactorbindingsiteclusteringtoidentifycis-regulatorymodulesinvolvedinpatternformationintheDrosophilagenome.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica2002,99:757-762.
13.
YuhCH,BolouriH,DavidsonEH:GenomicCis-RegulatoryLogic:ExperimentalandComputationalAnalysisofaSeaUrchinGene.
Science1998,279:1896-1902.
14.
EbertBL,FirthJD,RatcliffePJ:HypoxiaandMitochondrialInhibitorsRegulateExpressionofGlucoseTransporter-1viaDistinctCis-actingSequences.
JournalofBiologicalChemistry1995,270:29083-29089.
15.
TuanDY,SolomonWB,LondonIM,LeeDP:Anerythroid-specific,developmental-stage-independentenhancerfarupstreamofthehuman"beta-likeglobin"genes.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica1989,86:2554-2558.
16.
ChenK,RajewskyN:NaturalselectiononhumanmicroRNAbindingsitesinferredfromSNPdata.
NatGenet2006,38:1452-1456.
17.
KlocM,BilinskiS,Pui-YeeChanA,EtkinLD:TheTargetingofXcat2mRNAtotheGerminalGranulesDependsonacis-ActingGerminalGranuleLocalizationElementwithinthe3'UTR.
DevelopmentalBiology2000,217:221-229.
18.
GumucioDL,SheltonDA,ZhuW,MillinoffD,GrayT,BockJH,SlightomJL,GoodmanM:Evolutionarystrategiesfortheelucidationofcisandtransfactorsthatregulatethedevelopmentalswitchingprogramsofthebeta-likeglobingenes.
MolPhylogenetEvol1996,5:18-32.
19.
BoffelliD,McAuliffeJ,OvcharenkoD,LewisKD,OvcharenkoI,PachterL,RubinEM:PhylogeneticShadowingofPrimateSequencestoFindFunctionalRegionsoftheHumanGenome.
Science2003,299:1391-1394.
20.
WaldG:Themolecularbasisofvisualexcitation.
Nature1968,219:800-807.
21.
KocherTD:Adaptiveevolutionandexplosivespeciation:thecichlidfishmodel.
NatRevGenet2004,5:288-298.
22.
SeehausenO:Africancichlidfish:amodelsysteminadaptiveradiationresearch.
ProceedingsoftheRoyalSocietyB:BiologicalSciences2006,273:1987-1998.
23.
CarletonKL:Cichlidfishvisualsystems:mechanismsofspectraltuning.
IntegrativeZoology2009,4:75-86.
24.
CarletonKL,SpadyTC,KocherTD:VisualcommunicationinEastAfricancichlidfishes:diversityinaphylogeneticcontext.
InCommunicationinFishes.
Editedby:LadichF,CollinSP,P.
M,G.
KB.
Enfield.
SciencePublishers;2006:487-515.
25.
JordanR,KelloggK,HoweD,JuanesF,StaufferJ,LoewE:PhotopigmentspectralabsorbanceofLakeMalaicichlids.
JFishBiol2006,68:1291-1299.
26.
HofmannCM,O'QuinKE,MarshallNJ,CroninTW,SeehausenO,CarletonKL:Theeyeshaveit:regulatoryandstructuralchangesbothunderliecichlidvisualpigmentdiversity.
PLoSBiol2009,7:e1000266.
27.
O'QuinKE,HofmannCM,HofmannHA,CarletonKL:ParallelevolutionofopsingeneexpressioninAfricancichlidfishes.
MolecularBiologyandEvolution2010.
28.
SpadyTC,ParryJW,RobinsonPR,HuntDM,BowmakerJK,CarletonKL:Evolutionofthecichlidvisualpalettethroughontogeneticsubfunctionalizationoftheopsingenearrays.
MolBiolEvol2006,23:1538-1547.
29.
CarletonKL,HofmannCM,KliszC,PatelZ,ChircusLM,SimenauerLH,SoodooN,AlbertsonRC,SerJR:Geneticbasisofdifferentialopsingeneexpressionincichlidfishes.
JEvolBiol2010,23(4):840-53.
30.
KatagiriT,AsakawaS,MinagawaS,ShimizuN,HironoI,AokiT:ConstructionandcharacterizationofBAClibrariesforthreefishspecies;rainbowtrout,carpandtilapia.
AnimGenet2001,32:200-204.
31.
DiPalmaF,KiddC,BorowskyR,KocherTD:ConstructionofbacterialartificialchromosomelibrariesfortheLakeMalawicichlid(Metriaclimazebra),andtheblindcavefish(Astyanaxmexicanus).
Zebrafish2007,4:41-47.
32.
CarletonKL,SpadyTC,StreelmanJT,KiddMR,McFarlandWN,LoewER:Visualsensitivitiestunedbyheterochronicshiftsinopsingeneexpression.
BMCBiol2008,6:22.
33.
GennerMJ,SeehausenO,LuntDH,JoyceDA,ShawPW,CarvalhoGR,TurnerGF:AgeofCichlids:NewDatesforAncientLakeFishRadiations.
MolecularBiologyandEvolution2007,24:1269-1282.
34.
AppleburyML,FarhangfarF,GlosmannM,HashimotoK,KageK,RobbinsJT,ShibusawaN,WondisfordFE,ZhangH:TransientexpressionofthyroidhormonenuclearreceptorTRbeta2setsSopsinpatterningduringconephotoreceptorgenesis.
DevDyn2007,236:1203-1212.
35.
BrowmanH,HawryshynC:RetinoicAcidModulatesRetinalDevelopmentintheJuvenilesofaTeleostFish.
JExpBiol1994,193:191-207.
36.
BrowmanHI,HawryshynCW:Thedevelopmentaltrajectoryofultravioletphotosensitivityinrainbowtroutisalteredbythyroxine.
VisionRes1994,34:1397-1406.
37.
DannSG,AllisonWT,VeldhoenK,JohnsonT,HawryshynCW:ChromatinimmunoprecipitationassayontherainbowtroutopsinproximalpromotersillustratesbindingofNF-kappaBandc-juntotheSWS1promoterintheretina.
ExpEyeRes2004,78:1015-1024.
38.
NgL,HurleyJB,DierksB,SrinivasM,SaltoC,VennstromB,RehTA,ForrestD:Athyroidhormonereceptorthatisrequiredforthedevelopmentofgreenconephotoreceptors.
NatGenet2001,27:94-98.
39.
PengGH,AhmadO,AhmadF,LiuJ,ChenS:Thephotoreceptor-specificnuclearreceptorNr2e3interactswithCrxandexertsopposingeffectsonthetranscriptionofrodversusconegenes.
HumMolGenet2005,14:747-764.
40.
RobertsMR,HendricksonA,McGuireCR,RehTA:RetinoidXReceptorγIsNecessarytoEstablishtheS-opsinGradientinConePhotoreceptorsoftheDevelopingMouseRetina.
InvestigativeOphthalmology&VisualScience2005,46:2897-2904.
41.
TakechiM,SenoS,KawamuraS:Identificationofcis-actingelementsrepressingblueopsinexpressioninzebrafishUVconesandpinealcells.
JBiolChem2008,283:31625-31632.
42.
HalliganDL,Eyre-WalkerA,AndolfattoP,KeightleyPD:PatternsofEvolutionaryConstraintsinIntronicandIntergenicDNAofDrosophila.
GenomeResearch2004,14:273-279.
43.
KeightleyPD,GaffneyDJ:FunctionalconstraintsandfrequencyofdeleteriousmutationsinnoncodingDNAofrodents.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica2003,100:13402-13406.
44.
LeeBY,LeeWJ,StreelmanJT,CarletonKL,HoweAE,HulataG,SlettanA,SternJE,TeraiY,KocherTD:Asecond-generationgeneticlinkagemapoftilapia(Oreochromisspp.
).
Genetics2005,170:237-244.
45.
SchwartzS,ZhangZD,SmitA,ReimerC,BouckC,GibbsRA,HardisonRC,MillerW:PipMaker–AwebserverforaligningtwogenomicDNAsequences.
GenomeResearch2000,10:577-586.
46.
SchwartzS,ElnitskiL,LiM,WeirauchM,RiemerC,SmitA,GreenED,HardisonRC,MillerW:MultiPipMakerandsupportingtools:AlignmentsandanalysisofmultiplegenomicDNAsequences.
NucleicAcidsRes2003,31:3518-3524.
O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page21of2347.
WatsonCT,LubienieckiKP,LoewE,DavidsonWS,BredenF:Genomicorganizationofduplicatedshortwave-sensitiveandlongwave-sensitiveopsingenesinthegreenswordtail,Xiphophorushelleri.
BMCEvolBiol2010,10:87.
48.
CarletonKL,KocherTD:ConeopsingenesofAfricancichlidfishes:tuningspectralsensitivitybydifferentialgeneexpression.
MolBiolEvol2001,18:1540-1550.
49.
SteinkeD,SalzburgerW,MeyerA:NovelrelationshipsamongtenfishmodelspeciesrevealedbasedonaphylogenomicanalysisusingESTs.
JMolEvol2006,62:772-784.
50.
ChinenA,HamaokaT,YamadaY,KawamuraS:Geneduplicationandspectraldiversificationofconevisualpigmentsofzebrafish.
Genetics2003,163:663-675.
51.
HofmannCM,CarletonKL:Geneduplicationanddifferentialgeneexpressionplayanimportantroleinthediversificationofvisualpigmentsinfish.
JIntCompBiol2009,49:630-643.
52.
NeafseyDE,HartlDL:Convergentlossofanancientlyduplicated,functionallydivergentRH2opsingeneinthefuguandTetraodonpufferfishlineages.
Gene2005,350:161-171.
53.
YokoyamaS,TadaT:Evolutionarydynamicsofrhodopsintype2opsinsinvertebrates.
MolBiolEvol2010,27:133-141.
54.
Griffiths-JonesS,SainiHK,DongenSv,EnrightAJ:miRBase:toolsformicroRNAgenomics.
NucleicAcidsResearch2007,gkm952.
55.
KloostermanWP,SteinerFA,BerezikovE,deBruijnE,vandeBeltJ,VerheulM,CuppenE,PlasterkRH:CloningandexpressionofnewmicroRNAsfromzebrafish.
NucleicAcidsRes2006,34:2558-2569.
56.
Opsinevolution:LWSPhyloSNPs[[http://genomewiki.
ucsc.
edu/index.
php/Opsin_evolution:_LWS_PhyloSNPs]].
.
57.
WakefieldMJ,AndersonM,ChangE,WeiKJ,KaulR,GravesJA,GrutznerF,DeebSS:Conevisualpigmentsofmonotremes:fillingthephylogeneticgap.
VisNeurosci2008,25:257-264.
58.
SmallwoodPM,WangY,NathansJ:Roleofalocuscontrolregioninthemutuallyexclusiveexpressionofhumanredandgreenconepigmentgenes.
ProcNatlAcadSciUSA2002,99:1008-1011.
59.
WangY,MackeJP,MerbsSL,ZackDJ,KlaunbergB,BennettJ,GearhartJ,NathansJ:Alocuscontrolregionadjacenttothehumanredandgreenvisualpigmentgenes.
Neuron1992,9:429-440.
60.
TsujimuraT,HosoyaT,KawamuraS:Asingleenhancerregulatingthedifferentialexpressionofduplicatedred-sensitiveopsingenesinzebrafish.
PLoSGenet2010,6:e1001245.
61.
BirneyE,StamatoyannopoulosJA,DuttaA,GuigoR,GingerasTR,MarguliesEH,WengZ,SnyderM,DermitzakisET,ThurmanRE,etal:Identificationandanalysisoffunctionalelementsin1%ofthehumangenomebytheENCODEpilotproject.
Nature2007,447:799-816.
62.
RozowskyJ,EuskirchenG,AuerbachRK,ZhangZD,GibsonT,BjornsonR,CarrieroN,SnyderM,GersteinMB:PeakSeqenablessystematicscoringofChIP-seqexperimentsrelativetocontrols.
NatBiotechnol2009,27:66-75.
63.
SchugJ:UsingTESStopredicttranscriptionfactorbindingsitesinDNAsequence.
CurrProtocBioinformatics2008,Chapter2,Unit26.
64.
SalbertG,FanjulA,PiedrafitaFJ,LuXP,KimSJ,TranP,PfahlM:RetinoicacidreceptorsandretinoidXreceptor-alphadown-regulatethetransforminggrowthfactor-beta1promoterbyantagonizingAP-1activity.
MolEndocrinol1993,7:1347-1356.
65.
SchuleR,RangarajanP,YangN,KliewerS,RansoneLJ,BoladoJ,VermaIM,EvansRM:RetinoicacidisanegativeregulatorofAP-1-responsivegenes.
ProcNatlAcadSciUSA1991,88:6092-6096.
66.
WassermanWW,SandelinA:Appliedbioinformaticsfortheidentificationofregulatoryelements.
NatRevGenet2004,5:276-287.
67.
LetovskyJ,DynanWS:MeasurementofthebindingoftranscriptionfactorSp1toasingleGCboxrecognitionsequence.
NucleicAcidsRes1989,17:2639-2653.
68.
StormoGD:DNAbindingsites:representationanddiscovery.
Bioinformatics2000,16:16-23.
69.
BakerDL,DaveV,ReedT,PeriasamyM:MultipleSp1bindingsitesinthecardiac/slowtwitchmusclesarcoplasmicreticulumCa2+-ATPasegenepromoterarerequiredforexpressioninSol8musclecells.
JBiolChem1996,271:5921-5928.
70.
LatchmanDS:Eukaryotictranscriptionfactors.
4edition.
Amsterdam;Boston:Elsevier/AcademicPress;2004.
71.
AndolfattoP:Adaptiveevolutionofnon-codingDNAinDrosophila.
Nature2005,437:1149-1152.
72.
LohYH,YiSV,StreelmanJT:EvolutionofmicroRNAsandthediversificationofspecies.
GenomeBiolEvol2010.
73.
AroraA,McKayGJ,SimpsonDA:PredictionandverificationofmiRNAexpressioninhumanandratretinas.
InvestOphthalmolVisSci2007,48:3962-3967.
74.
RyanDG,Oliveira-FernandesM,LavkerRM:MicroRNAsofthemammalianeyedisplaydistinctandoverlapingtissuespecificity.
MolVis2006,12:1175-1184.
75.
WienholdsE,KloostermanWP,MiskaE,Alvarez-SaavedraE,BerezikovE,deBruijnE,HorvitzHR,KauppinenS,PlasterkRHA:MicroRNAExpressioninZebrafishEmbryonicDevelopment.
Science2005,309:310-311.
76.
XuS,WitmerPD,LumayagS,KovacsB,ValleD:MicroRNA(miRNA)transcriptomeofmouseretinaandidentificationofasensoryorgan-specificmiRNAcluster.
JBiolChem2007,282:25053-25066.
77.
GuerinMB,McKernanDP,O'BrienCJ,CotterTG:Retinalganglioncells:dyingtosurvive.
IntJDevBiol2006,50:665-674.
78.
KatoM,PuttaS,WangM,YuanH,LantingL,NairI,GunnA,NakagawaY,ShimanoH,TodorovI,etal:TGF-betaactivatesAktkinasethroughamicroRNA-dependentamplifyingcircuittargetingPTEN.
NatCellBiol2009,11:881-889.
79.
LiQJ,ChauJ,EbertPJ,SylvesterG,MinH,LiuG,BraichR,ManoharanM,SoutschekJ,SkareP,etal:miR-181aisanintrinsicmodulatorofTcellsensitivityandselection.
Cell2007,129:147-161.
80.
TeraiY,SeehausenO,SasakiT,TakahashiK,MizoiriS,SugawaraT,SatoT,WatanabeM,KonijnendijkN,MrossoHD,etal:DivergentselectiononopsinsdrivesincipientspeciationinLakeVictoriacichlids.
PLoSBiol2006,4:e433.
81.
LisneyTJ,StuddE,HawryshynCW:ElectrophysiologicalassessmentofspectralsensitivityinadultNiletilapiaOreochromisniloticus:evidenceforvioletsensitivity.
JExpBiol2010,213:1453-1463.
82.
CalissanoM,DissJK,LatchmanDS:Post-transcriptionalregulationoftheBrn-3btranscriptionfactorindifferentiatingneuroblastomacells.
FEBSLett2007,581:2490-2496.
83.
DecembriniS,BressanD,VignaliR,PittoL,MariottiS,RainaldiG,WangX,EvangelistaM,BarsacchiG,CremisiF:MicroRNAscouplecellfateanddevelopmentaltiminginretina.
ProcNatlAcadSciUSA2009,106:21179-21184.
84.
ZhaoJJ,YangJ,LinJ,YaoN,ZhuY,ZhengJ,XuJ,ChengJQ,LinJY,MaX:IdentificationofmiRNAsassociatedwithtumorigenesisofretinoblastomabymiRNAmicroarrayanalysis.
ChildsNervSyst2009,25:13-20.
85.
O'QuinKE,SmithAR,SharmaA,CarletonKL:Newevidencefortheroleofheterochronyintherepeatedevolutionofcichlidopsinexpression.
Evolution&Development2011,13(2):193-203.
86.
HutvagnerG,ZamorePD:AmicroRNAinamultiple-turnoverRNAienzymecomplex.
Science2002,297:2056-2060.
87.
LlaveC,XieZ,KasschauKD,CarringtonJC:CleavageofScarecrow-likemRNAtargetsdirectedbyaclassofArabidopsismiRNA.
Science2002,297:2053-2056.
88.
BartelDP:MicroRNAs:genomics,biogenesis,mechanism,andfunction.
Cell2004,116:281-297.
89.
LewisBP,ShihIH,Jones-RhoadesMW,BartelDP,BurgeCB:PredictionofmammalianmicroRNAtargets.
Cell2003,115:787-798.
90.
StarkA,BrenneckeJ,RussellRB,CohenSM:IdentificationofDrosophilaMicroRNAtargets.
PLoSBiol2003,1:E60.
91.
KasowskiM,GrubertF,HeffelfingerC,HariharanM,AsabereA,WaszakSM,HabeggerL,RozowskyJ,ShiM,UrbanAE,etal:Variationintranscriptionfactorbindingamonghumans.
Science2010,328:232-235.
92.
LohYH,KatzLS,MimsMC,KocherTD,YiSV,StreelmanJT:ComparativeanalysisrevealssignaturesofdifferentiationamidgenomicpolymorphisminLakeMalawicichlids.
GenomeBiol2008,9:R113.
93.
RobertsRB,SerJR,KocherTD:SexualConflictResolvedbyInvasionofaNovelSexDeterminerinLakeMalawiCichlidFishes.
Science2009,326:998-1001.
94.
AllenderCJ,SeehausenO,KnightME,TurnerGF,MacleanN:DivergentselectionduringspeciationofLakeMalawicichlidfishesinferredfromparallelradiationsinnuptialcoloration.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica2003,100:14074-14079.
95.
MorleyM,MolonyCM,WeberTM,DevlinJL,EwensKG,SpielmanRS,CheungVG:Geneticanalysisofgenome-widevariationinhumangeneexpression.
Nature2004,430:743-747.
O'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page22of2396.
SungHM,WangTY,WangD,HuangYS,WuJP,TsaiHK,TzengJ,HuangCJ,LeeYC,YangP,etal:Rolesoftransandcisvariationinyeastintraspeciesevolutionofgeneexpression.
MolBiolEvol2009,26:2533-2538.
97.
WittkoppPJ,HaerumBK,ClarkAG:Evolutionarychangesincisandtransgeneregulation.
Nature2004,430:85-88.
98.
HsiaCC,McGinnisW:Evolutionoftranscriptionfactorfunction.
CurrOpinGenetDev2003,13:199-206.
99.
LevineM,TjianR:Transcriptionregulationandanimaldiversity.
Nature2003,424:147-151.
100.
SpadyTC,SeehausenO,LoewER,JordanRC,KocherTD,CarletonKL:Adaptivemolecularevolutionintheopsingenesofrapidlyspeciatingcichlidspecies.
MolBiolEvol2005,22:1412-1422.
101.
EwingB,HillierL,WendlMC,GreenP:Base-callingofautomatedsequencertracesusingphred.
I.
Accuracyassessment.
GenomeRes1998,8:175-185.
102.
Phrap:phragmentassemblyprogram[[http://www.
phrap.
org]].
.
103.
MarguliesM,EgholmM,AltmanWE,AttiyaS,BaderJS,BembenLA,BerkaJ,BravermanMS,ChenYJ,ChenZ,etal:Genomesequencinginmicrofabricatedhigh-densitypicolitrereactors.
Nature2005,437:376-380.
104.
AltschulSF,GishW,MillerW,MyersEW,LipmanDJ:Basiclocalalignmentsearchtool.
JMolBiol1990,215:403-410.
105.
CouronneO,PoliakovA,BrayN,IshkhanovT,RyaboyD,RubinE,PachterL,DubchakI:StrategiesandToolsforWhole-GenomeAlignments.
GenomeResearch2003,13:73-80.
106.
NéronB,MénagerH,MaufraisC,JolyN,MaupetitJ,LetortS,CarrereS,TufferyP,LetondalC:Mobyle:anewfullwebbioinformaticsframework.
Bioinformatics2009,25:3005-3011.
107.
R:Alanguageandenvironmentforstatisticalcomputing.
[http://www.
R-project.
org].
108.
KatohK,TohH:RecentdevelopmentsintheMAFFTmultiplesequencealignmentprogram.
BriefBioinform2008,9:286-298.
109.
PosadaD:jModelTest:Phylogeneticmodelaveraging.
MolecularBiologyandEvolution2008,25:1253-1256.
110.
SokalRR,RohlfFJ:BiometryNewYork:W.
H.
FreemanandCompany;1995.
111.
ChesselD,DuforAB,ThioulouseJ:Theade4package-I:One-tablemethods.
RNews2004,4:5-10.
112.
LewisBP,BurgeCB,BartelDP:Conservedseedpairing,oftenflankedbyadenosines,indicatesthatthousandsofhumangenesaremicroRNAtargets.
Cell2005,120:15-20.
113.
BrudnoM,DoCB,CooperGM,KimMF,DavydovE,GreenED,SidowA,BatzoglouS:LAGANandMulti-LAGAN:efficienttoolsforlarge-scalemultiplealignmentofgenomicDNA.
GenomeRes2003,13:721-731.
114.
LibradoP,RozasJ:DnaSPv5:asoftwareforcomprehensiveanalysisofDNApolymorphismdata.
Bioinformatics2009,25:1451-1452.
115.
PurcellS,NealeB,Todd-BrownK,ThomasL,FerreiraMA,BenderD,MallerJ,SklarP,deBakkerPI,DalyMJ,ShamPC:PLINK:atoolsetforwhole-genomeassociationandpopulation-basedlinkageanalyses.
AmJHumGenet2007,81:559-575.
116.
HacklerLJr,WanJ,SwaroopA,QianJ,ZackDJ:MicroRNAprofileofthedevelopingmouseretina.
InvestOphthalmolVisSci2010,51:1823-1831.
117.
KaraliM,PelusoI,MarigoV,BanfiS:IdentificationandcharacterizationofmicroRNAsexpressedinthemouseeye.
InvestOphthalmolVisSci2007,48:509-515.
118.
HoeggS,BooreJ,KuehlJ,MeyerA:ComparativephylogenomicanalysesofteleostfishHoxgeneclusters:lessonsfromthecichlidfishAstatotilapiaburtoni.
BMCGenomics2007,8:317.
doi:10.
1186/1471-2148-11-120Citethisarticleas:O'Quinetal.
:Divergenceincis-regulatorysequencessurroundingtheopsingenearraysofAfricancichlidfishes.
BMCEvolutionaryBiology201111:120.
SubmityournextmanuscripttoBioMedCentralandtakefulladvantageof:ConvenientonlinesubmissionThoroughpeerreviewNospaceconstraintsorcolorgurechargesImmediatepublicationonacceptanceInclusioninPubMed,CAS,ScopusandGoogleScholarResearchwhichisfreelyavailableforredistributionSubmityourmanuscriptatwww.
biomedcentral.
com/submitO'Quinetal.
BMCEvolutionaryBiology2011,11:120http://www.
biomedcentral.
com/1471-2148/11/120Page23of23