Pinkeyedm.2828dy.com

GeneExpressionAnalysisofZebrafishMelanocytes,Iridophores,andRetinalPigmentedEpitheliumRevealsIndicatorsofBiologicalFunctionandDevelopmentalOriginCharlesW.
Higdon*,RobiD.
Mitra,StephenL.
Johnson*DepartmentofGenetics,WashingtonUniversity,St.
Louis,Missouri,UnitedStatesofAmericaAbstractInordertofacilitateunderstandingofpigmentcellbiology,wedevelopedamethodtoconcomitantlypurifymelanocytes,iridophores,andretinalpigmentedepitheliumfromzebrafish,andanalyzedtheirtranscriptomes.
Comparingexpressiondatafromthesecelltypesandwholeembryosallowedustorevealgeneexpressionco-enrichmentinmelanocytesandretinalpigmentedepithelium,aswellasinmelanocytesandiridophores.
Wefound214genesco-enrichedinmelanocytesandretinalpigmentedepithelium,indicatingthesharedfunctionsofmelanin-producingcells.
Wefound62genessignificantlyco-enrichedinmelanocytesandiridophores,illustrativeoftheirshareddevelopmentaloriginsfromtheneuralcrest.
Thisisalsothefirstanalysisoftheiridophoretranscriptome.
Geneexpressionanalysisforiridophoresrevealedextensiveenrichmentofspecificenzymestocoordinateproductionoftheirguanine-basedreflectivepigment.
Wespeculatethecoordinatedupregulationofspecificenzymesfromseveralmetabolicpathwaysrecyclestherate-limitingsubstrateforpurinesynthesis,phosphoribosylpyrophosphate,thusconstitutingaguaninecycle.
Thepurificationprocedureandexpressionanalysisdescribedhere,alongwiththeaccompanyingtranscriptome-wideexpressiondata,providethefirstmRNAsequencingdataformultiplepurifiedzebrafishpigmentcelltypes,andwillbeausefulresourceforfurtherstudiesofpigmentcellbiology.
Citation:HigdonCW,MitraRD,JohnsonSL(2013)GeneExpressionAnalysisofZebrafishMelanocytes,Iridophores,andRetinalPigmentedEpitheliumRevealsIndicatorsofBiologicalFunctionandDevelopmentalOrigin.
PLoSONE8(7):e67801.
doi:10.
1371/journal.
pone.
0067801Editor:YukFaiLeung,PurdueUniversity,UnitedStatesofAmericaReceivedFebruary4,2013;AcceptedMay23,2013;PublishedJuly9,2013Copyright:2013Higdonetal.
Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited.
Funding:ThisworkwasmadepossiblebygrantsfromtheNationalInstitutesofHealth(NIH)(RO1-GM56988)theW.
M.
KeckFoundation(http://www.
wmkeck.
org).
Thefundershadnoroleinstudydesign,datacollectionandanalysis,decisiontopublish,orpreparationofthemanuscript.
CompetingInterests:Theauthorshavedeclaredthatnocompetinginterestsexist.
*E-mail:cwhigdon@wustl.
edu(CWH);sjohnson@genetics.
wustl.
edu(SLJ)IntroductionPigmentcellsserveasusefulmodelsforunderstandingmanyaspectsofdevelopmentalandcellbiology.
Forexample,melano-cytesarepigmentcellsstudiedtounderstandcellspecification,migration,differentiation,survival,regeneration,organelletrans-port,secretion,anddisease[1–15].
Melanocytesproducemelanin,whichinhumansservesasaUVprotectantinskin[16].
Melanocytesalsohaverolesinotherorganssuchastheear,brain,heart,andadiposetissue[17,18].
Theincidenceofmelanomas,adiseaseofmelanocytesandthemostlethalformofskincancer,isalsoincreasing[19].
Methodstoisolateandculturemelanocytesforinvitrostudieshavebeeninformativeforunderstandingmelanocytebiology[20,21].
Invivostudiesofmelanocytebiologyandmelanomadynamicshavebeenaidedbytheidentificationofmutantsinmiceandzebrafish[22–25].
Giventheutilityofzebrafishmelanocytestounderstandcellbiologyanddisease,thetranscriptome-widecharacterizationofgenesex-pressedinzebrafishpigmentcellswouldbeasignificantresource.
Inmammaliansystems,melanocytesaretheonlyneuralcrest-derivedpigmentcelltypefoundinthedermis.
Incontrast,severalneuralcrest-derivedpigmentcellsarefoundinzebrafishandotherpoikilotherms,includingreflectiveiridophores[5].
Severalre-quirementsforiridophoredevelopmentfromtheneuralcrestareknown[5,26,27].
However,itisunknownifmarkersofneuralcrestidentitypersistiniridophoresfollowingdevelopment,andwhetherthesemarkersaresharedbyotherneuralcrest-derivedpigmentcells,suchasmelanocytes.
Afurtherquestioniniridophorebiologyishowtheguanine-basedpigmentisproduced[28,29].
Zebrafishbearingmutationsinthedenovopurinesynthesisenzymesgartandpaicshaveiridophoredefects,indicatingpurinesynthesisisimportantforiridophorepigmentation[30].
Identify-ingapossiblemechanismbywhichiridophoresproduceanabundanceofguanineforpigmentformationwhilemaintainingadequatesuppliesofpurinesforDNAandRNAproductionwillbeinformativeforcellbiology.
Anotherpigmentcelltypesharedbymammalianandpoikilo-thermicvertebratesistheretinalpigmentedepithelium(RPE).
TheRPEisagroupofmelanin-producingcellsfoundinthevertebrateeye.
TheRPEdevelopsfromtheeyeprimordium,andiscontinuouswiththelayerofcellsthatformstheiris[31].
TheRPEiscriticalforeyedevelopmentandretinalhealth.
Itprovidestrophicsupportandrecycleswastesfromthephotoreceptorsoftheretina[32].
TheRPEformspartoftheblood-retinabarrier,providingtheeyewithanimmune-privilegedstatus[33].
DefectsintheRPEcontributetodiseasessuchasmaculardegenerationPLOSONE|www.
plosone.
org1July2013|Volume8|Issue7|e67801andretinitispigmentosa,whichresultinvisionproblems[34,35].
PreviousdescriptionsofgeneexpressionintheRPEofzebrafish,chicken,andhumanhaveelucidatedmanyofthegenesplayingrolesinRPEbiology[36–38].
Manyofthesegenesareresponsibleforproducingmelanin,anddefectsinmelaninproductionareoftenassociatedwithreducedvisualfunction[39].
However,itisunknownwhetherRPEandmelanocytesusedifferentpathwaysofmelaninproduction,oriftheyareessentiallyidentical.
ThisinformationwouldbeusefulforunderstandingRPEbiology,andwouldalsoinformfutureexaminationsofregulatorycontrolforgenesexpressedinoneorbothcelltypes.
Inordertofacilitateunderstandingofthesepigmentcells,wedevelopedarobustmethodtoisolatethesethreepigmentcelltypesfromzebrafishembryos,followedbymRNAsequencingandtranscriptomeanalysis.
Thispurificationprocedurereliesontheinherentdensitiesofmelaninandguanine-filledcells;henceitcanbeusedwithoutothercomplicatedlineagemarkers.
Here,wereporttheco-enrichmentandcell-typespecificgeneexpressionprofilesofmelanocytes,iridophores,andRPEfromembryoniczebrafish.
WhileRPEandiridophoresdonotexhibitsignificantoverlapofenrichedgeneexpression,ouranalysisrevealsconsiderableoverlapamongpairsofpigmentcelltypesindicativeoftheircommonoriginorfunction.
GenesenrichedinboththemelanocyteandRPElineagescontaingenesinthemelanin-productionpathway,andsuggestamorecompletepictureofmelanin-producingmachineryiscontainedwithinthisset.
Similarly,expressionco-enrichmentinmelanocytesandirido-phoresarereflectiveoftheirneuralcrestorigin,andsuggestgenesinthissetmaybespecifictoneuralcrestidentity.
Furthermore,thisisthefirstcharacterizationoftheiridophoretranscriptome.
Wefoundthatiridophoresspecificallyupregulatetheguanineportionofdenovopurinesynthesis,aswellasspecificenzymesfromothermetabolicpathwaysthataidinproducingtheiridophoreguanine-basedreflectivepigment.
Inadditiontotheanalysispresentedhere,thisprocedureandaccompanyingdataprovideasignificantresourceforfurtherbiologicaldiscoveryinpigmentcells.
MaterialsandMethodsZebrafishStrainandSampleCollectionTimePointsThisstudywascarriedoutinaccordancewiththeWashingtonUniversityAnimalUseCommitteeguidelinesunderapprovedprotocol#20110236.
Zebrafishwererearedandbredaccordingtostandardprotocols[40].
Thefishusedinthisstudywerehomozygousforatemperaturesensitivealleleofmicropthalmiatranscriptionfactor(mitfavc7)[41].
ThismutantfacilitatedthecollectionofRPE,asmitfaisnotrequiredforRPEdevelopmentinzebrafish[41].
Melanocytes,iridophores,andRPEdevelopnormallyat25uCinmitfavc7.
Whenheldat32uC,theneuralcrest-derivedmelanocytesdonotdevelop,butRPEandiridophoresdevelopnormally.
Allmelanocyteandiridophoresampleswereincubatedat25uCpriortocollection.
Wenotethepossibilitythatmitfamaybepartiallycompromisedbyaberrantsplicingproductsinthismutantatpermissivetemperatures,butfollowingdevelopmentat25uC,melanocytenumbers,morphol-ogy,andpigmentationareindistinguishablefromwild-typezebrafish[41].
EmbryosusedforRPEsampleswereincubatedateither25uCor32uCuntiltheequivalentof3–5daysat28.
5uC,asindicatedinTableS1[42].
Itshouldbenotedthatatstageslaterthan5dpf,choroidalmelanocytesmaybepresentadjacenttotheRPE.
OnlyoneofthefiveRPEcDNAlibrarieswaspreparedlaterthan5dpf,asindicatedinTableS1.
Wedidnothaveseparatemarkerstoidentifycontaminationfromeye-associatedneuralFigure1.
Purificationprocedureformelanocytes,iridophores,andretinalpigmentedepithelium.
Zebrafisharegrowntothedesiredtimepoint;shownin(A)isasixdayoldfish.
(B)Fisharedissociatedtoasinglecellsuspension;blackmelanocytes(arrows)andreflectiveiridophores(arrowheads)arevisibleasasmallpercentageofallcells.
(C)CellsareplacedatopaPercolldensitygradientandcentrifuged.
(D)TheresultingcellpelletisresuspendedandanalyzedbyFACS.
ShownisacharacteristicFACSplotdemonstratingtherelativepositionsofmelanocyteandiridophoregates(ovals).
Sortediridophoresareshownontheupperleftof(D)underincidentlight.
Sortedmelanocytesareonthelowerrightusingtrans-illumination.
doi:10.
1371/journal.
pone.
0067801.
g001TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org2July2013|Volume8|Issue7|e67801Figure2.
SchematicofcDNAlibrarypreparation.
PolyA-selectedmRNA(inred)isreversetranscribedusingapolyTprimertailedwithauniversalprimer(A).
SeeTableS3forprimersequences.
MMLVreversetranscriptaseaddscytosinestothe39endofthe1ststrandcDNA(inblack),allowingfortemplateswitchingandadditionofthe39universalprimer(B).
PCRamplificationofthelibraryisfollowedbyRsaIandAluIenzymaticdigestionofcDNAs(C),followedbythestandardIlluminalibrarypreparationstepsofend-repair,asingleadenineaddition,Y-adapterligation(D),PCRenrichment,andsizeselection(mockgelshowninEwithyellowboxindicatingareaofgelremovedforDNAextraction),priortoflowcellgenerationandsequencing.
doi:10.
1371/journal.
pone.
0067801.
g002TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org3July2013|Volume8|Issue7|e67801crest-derivedmelanocytes,orchoroidalmelanocytesinourRPEpreparations.
Ifpresent,weexpectthecontributiontototalgeneexpressionintheRPEsamplestoexhibithighvariance,andfoundtobeinsignificantbyStudent'sT-testuponcomparisontotheothercelltypes.
However,wefindthelaterstageRPElibrarytobehighlycorrelatedoverallwiththeearlierstages,indicatingthatextensivecontaminationfromchoroidalmelanocytesisunlikely(r=0.
92,FigureS1a).
ItwasafurtherpossibilitythatmanylargechangesingeneexpressionwouldbepresentinRPEsamplesheldatrestrictiveandpermissivetemperaturesformitfa.
Wefoundthatsamplesheldatthelowandhightemperaturesexhibitedahighdegreeofsimilarity(r=0.
89),indicatingthatmostgenesarenotsignificantlydifferentatthetwotemperatures(FigureS1b).
Wealsoexpectsomegenestochangeduringdevelopmentbetween3and5dayspostfertilization.
Uponinspectionofmelanocytesamplespreparedatseveraltimepoints,wefindcorrelationsincreasewithincreasingdevelopmentalageoftheembryos(FigureS2).
Wedonothavethestatisticalpowertoconfidentlytrackdevelopmentalchangesingeneexpressionwithinspecificcelltypes,butwemakethedataavailableforallindividuallibrariestofacilitatefurtherinvestigation(TableS2).
CellDissociationandPigmentCellEnrichmentFigure1depictsthegeneralprocedureforpigmentcellpurification.
FishwereanesthetizedwithTricaine,rinsedwithCa-,Mg-DPBS(Sigma,D8537),andimmersedin100mLTrypLEExpress(Invitrogen,12604039)per1000fish.
Fishwereincubatedat37uCandshakenat100rpmfor15–20minutes,followedbytriturationwithaPasteurpipettetoremoveeyesfromlarva.
Afterseparationofeyesandlarva,eachgroupwasplacedinTrypLEExpressandshakenat100rpmat37uCfor1–1.
5hr.
Dissociatedcellswerefilteredthrougha120uMscreeninto50mLtubes.
Remainingintacttissuewastriturated10–20times,andagainfilteredthrougha120uMscreenintothedissociatedcells.
Dissociatedcellswerepelletedinaswingingbucketrotor(Eppendorf5810R)at500relativecentrifugalforce(rcf)for5minutesat4uC,thenresuspendedin1mLcoldisotonicPercoll(Sigma,P1644)bygentlepipetting.
IsotonicPercollwaspreparedbymixing1part10XPBSwith9partsPercoll.
Resuspendedcellsweretransferredto1.
6mLEppendorftubesandspunat2000rcffor5minutesat4uCinaswingingbucketrotorforisopycnicseparation.
Pigmentcellsinthepelletwerethenresuspendedin400mLoficecoldDPBSwith2%fetalcalfserum(FCS),andplacedontopreformedPercolldensitygradients.
Preformedgradientswerepreparedviacentrifugationof1mLaliquotsofisotonicPercollin1.
6mLtubesat10,000rcffor15minutesat4uCinafixedanglemicro-centrifuge(Eppendorf5415R).
TubescontainingpreformedPercollgradientswithoverlyingcellsuspensionswerecentrifugedinaswingingbucketrotorat2000rcffor10minutesat4uC.
Followingcentrifugation,overlyingPercollwasaspirated,leavingthefinal100mLcontainingthepigmentcellpellet.
Cellswereresuspendedwith50mLofcoldDPBSwith2%FCSandtransferredtoaclean1.
6mLtubecontaining500mLofcoldDPBSwith2%FCS,andkeptoniceuntilmRNAextractionorFACS.
FACSWeusedtheinherentpropertiesofthepigmentedcellstoperformFluorescence-ActivatedCellSorting(FACS).
Followingresuspensionin500mLcoldDPBSwith2%FCS,theenrichedTable1.
Candidatecontrolgenesaredifferentiallyexpressed.
RPKMCountsStudent'sT-testp-ValuesMelanocyteRPEIridophoreMelanocytevs.
RPEMelanocytevs.
IridophoreRPEvs.
IridophoreMelanocyteGenesgch254.
259.
811.
580.
0070.
0020.
113mlphb187.
847.
961.
390.
0040.
0030.
251kita3.
780.
390.
020.
0190.
0110.
076MelaninSynthesispmela15177.
254244.
20148.
900.
0510.
0110.
031dct14134.
126406.
98402.
470.
0250.
0000.
028tyrp1b12354.
093769.
95200.
990.
0260.
0040.
009IridophoreGenesatic7.
609.
63467.
160.
6640.
0100.
010ednrb14.
441.
7228.
630.
1100.
0180.
013ltk0.
020.
204.
020.
1380.
0020.
002NeuralCrestsox107.
571.
7613.
000.
0100.
0780.
005foxd33.
030.
356.
410.
0120.
2370.
063snai24.
260.
794.
080.
0160.
9040.
020RPEGenespax6a10.
0447.
480.
730.
0980.
0040.
055nr2e10.
752.
410.
050.
1890.
0410.
086myo7ab1.
893.
320.
150.
0500.
0010.
003Selectedgenesindicativeofpigmentcellidentityorsharedfunctionsareshown.
doi:10.
1371/journal.
pone.
0067801.
t001TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org4July2013|Volume8|Issue7|e67801cellpopulationswerescreenedthrougha30uMcellfilter(Partec,04–0042–2316).
CellswereanalyzedandsortedwithaDakoMoFlocellsorterusinga120uMnozzleatadropdrive(DD)frequencyof22390Hz.
Cellswereilluminatedusinga488nmlaser.
Cellsweregatedontwoattributestoseparatecellsfromeachotherandfromcellulardebris.
Cellulardebriswasdetectedusingforwardandsidescatter,selectingagainstthesmallestparticles(,1mmorless).
Cellsweresortedbasedondetectionusing510–530nmand575–595nmfilters,correspondingtoFL1andFL2inFigure1D,respectively.
Whenexcitedbythe488nmlaser,theautofluorescenceofiridophoresisclearlydetectableinthesechannelsasagroupofcellsextendingata45degreelineintheupperrightquadrant.
MelanocytesandRPEdonotautofluorescewiththisintensitywhenexcitedbythe488nmlaser,andclusteratthelowerleftoftheFACSplot.
Cellswerecollectedintoice-coldDPBSwith2%FCSandkeptoniceuntilmRNAextraction.
mRNAExtraction,cDNASynthesis,andIlluminaLibraryPreparationPigmentcellcDNAlibraryconstructionwasasfollows.
FormRNAextractiontheDynabeadsHmRNADIRECTKit(In-vitrogen)wasusedpermanufacturer'sinstructions.
FollowingmRNAelutionfromtheDynabeads,firststrandcDNAsynthesiswasperformedusingMMLVreversetranscriptase(Clontech)usingananchoredpolyTprimertailedwithauniversalprimersequence(SeeTableS3forprimersequencesandFigure2forpigmentcellcDNAlibraryconstructionoverview.
)Auniversalprimersequencewasalsoaddedtothe39endofthefirststrandbytemplateswitching,allowingforPCR-amplificationoftheresultantcDNA[43,44].
FollowingPCRamplificationusingthehighfidelitypolymeraseLATaq(TaKaRa,PCRcycle:95Cfor1minute,followedby20cyclesof98Cfor25seconds,60Cfor1minute,68Cfor20minutes),cDNAwasdigestedwithAluIandRsaIrestrictionenzymes(NEB).
Blunt-endenzymaticfragmenta-tionofcDNAwasusedinsteadofsonicationandgelextractiontominimizelossofsamplematerialandeliminatetheend-repairstepofIlluminalibrarypreparation.
SincethisreducedrepresentationstrategymightmissshortcDNAsthatlackbothrestrictionsites,wesoughttoavoidthisbyincludingenzymerecognitionsiteswithinthecDNAamplificationprimers.
ThisallowsfortheinclusionofshortcDNAsinourlibraries.
StandardIlluminalibraryprepara-tionsfollowed,performedbytheGenomeTechnologyAccessCenter(GTAC)atWashingtonUniversityinSt.
Louis(http://Table2.
Sharedgeneexpressionamongmelanocyte,RPE,andiridophore.
GeneMelanocyteRPEIridophoreEmbryoNotesrpl2622354.
2619692.
7423528.
0687.
23Ribosomalproteinrps1712677.
8210420.
7316940.
1759.
88Diamond-BlackfanAnemia[90]rps212516.
829306.
5712379.
3189.
39Ribosomalproteinrpl27a6506.
975852.
658537.
2748.
50Ribosomalproteinslc45a24415.
882376.
162801.
480.
20albinolocus[54,55]rps26l2338.
292362.
002596.
7823.
08Ribosomalproteinppp1r21263.
06227.
24202.
381.
77Proteinphosphatasecrfb5237.
93329.
44301.
821.
19Jak-STATcytokinereceptorLOC100535047217.
00212.
13135.
050.
40Uncharacterizeddhdh201.
46185.
72161.
641.
52dihydrodioldehydrogenasecyhr1150.
55157.
36138.
950.
87cysteine/histidine-rich1igf2bp2b133.
8197.
93128.
560.
75mRNA-bindingproteinghitm133.
71119.
78162.
370.
99BAXinhibitorproteinfamilyher9126.
2485.
17120.
900.
71NOTCHpathwayfam168a89.
2678.
2173.
660.
44Chemoresistance[91]comtb81.
1155.
0570.
610.
32Dopaminedegradationfkbp375.
2659.
4879.
660.
01Rapamycinbindingproteinmtbl65.
9436.
5848.
310.
32Heavymetalresistance[92]pard3b53.
4762.
5746.
530.
38Cell-cyclehbp141.
9350.
6152.
080.
12SOX-TCF-HMGfamilytranscriptionfactorzgc:15834531.
6743.
9259.
380.
31Tyrosinephosphatase,PTENC2domainccdc85al31.
2634.
9940.
960.
13Uncharacterized,coiled-coilproteingrma17.
1718.
2628.
830.
10Glutamatereceptor,GPCRmbd216.
5716.
0316.
310.
13BindsmethylatedDNAtriobpl14.
2816.
8612.
850.
01TRIOBP-like,actinorganization[93]rnd214.
2310.
6612.
260.
00RhoGTPase,neuritebranching[94]LOC10033499110.
177.
8710.
080.
03Uncharacterizedzgc:1365645.
949.
005.
300.
04C9orf64homologue,unknownfunctionShownareRPKMvaluesforgenesco-enrichedamongthethreepigmentcelltypesatalevel100-foldgreaterthanwholeembryos,withina2-foldchangeofeachother,withaminimumRPKMof4.
doi:10.
1371/journal.
pone.
0067801.
t002TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org5July2013|Volume8|Issue7|e67801gtac.
wustl.
edu).
Inbrief,asingleAwasaddedtothe39endofeachstrand,Y-adaptersligated,andlibraryenrichmentPCRper-formed,followedbygelextractionsize-selectionforfragmentsrangingfrom200–400basepairsinlength.
Illuminalibraryconstructionofpooled3dpfembryoswasperformedbyGTACfromtotalRNAextractedwithTrizolreagentaspreviouslydescribed[45].
NoPCRamplificationofwholeembryocDNAwasperformedpriortoIlluminaadapterligationandlibraryenrichment.
SequencingwasperformedontheGAIIXandHiSeq2000Illuminaplatforms.
Technicalsequencingreplicatesofthesamelibrariesonseparatelaneswereessentiallyidentical(FigureS3).
SequenceAnalysisWeusedNovoalign(www.
novocraft.
com),toassignresultantexpressionsequencetagstoacustomizednon-redundantdatabaseofcDNAsequencesconsistingof25,102knownandpredictedgenes(TableS4).
InitialinspectionoftheNCBIzebrafishmRNAdatabaseof28,286zebrafishcDNAs(ftp://ftp.
ncbi.
nih.
gov/refseq/D_rerio/mRNA_Prot/)viaanall-by-allBLASTsearchTable3.
SharedgeneexpressionamongmelanocyteandRPE.
GeneMelanocyteRPEIridophoreEmbryoNotespmela15177.
254244.
20148.
900.
57Silvermouse,fadingvisionzebrafish[95]dct14134.
126406.
98402.
470.
65Melaninsynthesistyrp1b12354.
093769.
95200.
991.
13Melaninsynthesistyrp12546.
871474.
0771.
460.
03Melaninsynthesisrlbp1b1181.
621212.
8648.
190.
26Retinaldehydebindingproteinmitfa1122.
44448.
5316.
220.
00Melanocytemasterregulatorpah1081.
84485.
2535.
7923.
60Melaninsynthesisstra6703.
09539.
1532.
813.
00Retinolmetabolismrgra300.
56504.
3427.
160.
21GPCRmsnb254.
08217.
3511.
100.
45FERMfamily,RDXhomologuefam213ab217.
86160.
1612.
320.
28Antioxidantenzymerbp1a148.
55137.
028.
741.
79Retinolbindingproteinslc24a4a141.
0889.
647.
631.
26NeuralcrestandRPEexpression[96]zgc:114181122.
41128.
266.
710.
58PutativeCNDP1homologuemab21l2120.
20176.
482.
980.
82Eyedevelopment[97]LOC100004225105.
7852.
373.
461.
32Uncharacterizedlratl105.
0996.
403.
520.
39Lecithinretinolacyltransferaseoca2102.
5656.
961.
630.
07Pinkeyed-dilutionmouse[98]kif21al96.
64104.
624.
160.
82Kinesinfamilymember,LOC100537698dhrs11al95.
71130.
886.
868.
45Dehydrogenase/reductaseSDRfamilycadm387.
46116.
234.
880.
12Nectinfamilycelladhesionproteincdh286.
6999.
726.
171.
10Celladhesionslc24a583.
8134.
681.
560.
00goldenlocus[99]s1pr183.
1474.
374.
781.
20RhodopsinfamilyGPCRctgf62.
1562.
194.
200.
32Secretedmitogenfads649.
5284.
552.
760.
08Fattyacidbiosynthesisfoxp439.
9684.
031.
780.
25Neuronalarborizationmaintenance[100]foxg1b39.
7164.
022.
380.
10Forkheadboxtranscriptionfactorkif21al37.
4758.
203.
210.
37Kinesinfamilymemberabcg2d34.
7248.
811.
580.
00whitefamilymember[101]LOC10014932433.
0831.
151.
550.
14Uncharacterizedphospholipaseefcab4b30.
8938.
651.
710.
00CalciumsensingGTPasecol11a1a27.
3927.
111.
220.
75Collagenalphachainprecursorrdh1324.
3031.
040.
810.
00Retinoldehydrogenasecam4l23.
3837.
541.
340.
29Celladhesionmolecule4-likesrcrb4l21.
1420.
881.
180.
80ScavengerreceptorCys-richgroupB-likecol4a520.
8629.
391.
690.
49TypeIVcollagendao.
220.
2722.
391.
231.
39D-amino-acid-oxidaseRPKMvaluesforgenesco-expressedinmelanocyteandRPEatleast10-foldgreaterthaniridophoresandwholeembryos,withaminimumof10RPKM.
doi:10.
1371/journal.
pone.
0067801.
t003TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org6July2013|Volume8|Issue7|e67801revealedmultiplenearlyidenticalsequencesforcDNAsthatwouldconfoundtheunambiguousassignmentofsequencetags.
Inordertogenerateanon-redundantcDNAdatabaseweselectedsinglerepresentativesforeachgeneasfollows.
IninstanceswhereacDNAinthedatabaseresultedinaBLASThitofgreaterthan94%identitytomorethan70%thelengthofanothertranscript,weexcludedthesmallerofthetwocDNAs.
Theresultantnon-redundantzebrafishcDNAdatabasecontained25,102uniquegenerecords.
Wefurtheranalyzedournon-redundantdatabasebymappingallcDNAsontotheUCSCzebrafishbrowser(ZV9).
Manualexaminationof6Mbsarbitrarilychosenalongchromo-some13revealed130annotatedgenes(combinedRefSeqandEnsemblgenetracks).
Ofthese,ournon-redundantdatabaseidentified126.
The4genesnotrepresentedinourdatabaseincluded35SribosomalRNAgenesandEN-SDARG00000086970,anannotatedgenewithapredictedORFbutnoclearorthologytootherspecies.
Thus,thistestshowsthatourmethodtogenerateanon-redundantcDNAlibraryresultsinadatabasethatidentifies,99%ofannotatedgenes(excludingribosomalRNAgenes).
Furthermore,manualexaminationofournon-redundantlibrarymappedontothis6Mbofchromosomalsequencerevealed10sequencesthatwerenotannotatedasRefSeqorEnsemblgenes.
ComparisontorepeattracksontheUCSCbrowserrevealedthatmost(8/10)oftheseunannotatedmatchesidentifiedORFsfromrepetitive,orretrotransposonDNA.
ItisnotclearhowthesesequenceswereinitiallyincludedintheNCBIcDNAdatabase.
Insummary,oureffortshavegeneratedanon-redundantzebrafishcDNAdatabasethatidentifiesmost(,99%)ofannotatedgenes,andincludes7%(10/126)recordsofdubiousutility,butwouldtendnottoconfoundRNA-seqanalysissuchasreportedhere.
SequencingresultsforeachcDNAlibraryaresummarizedinTableS1.
ForeachcDNAlibrary,thenumberoftagsalignedtoeachgenewasnormalizedbythelengthofthegeneandthetotalnumberofuniquelyaligningreadsforthatlibraryusingcustomPerlscripts(readsperkilobaseofcDNApermillionmappedreads-RPKM[45].
StatisticalcalculationswereperformedinR(www.
r-project.
org).
AllsequencingdatausedinthisstudyarepubliclyavailableatNCBI'sGEOdatabaseunderaccessionGSE46387(http://www.
ncbi.
nlm.
nih.
gov/geo/query/acc.
cgiacc=GSE46387).
QuantitativeRT-PCRPrimersusedforexpressionanalysisarelistedinTableS3,designedwithNCBI'sPrimer-BLASTtobeseparatedbyatleastoneintron(http://www.
ncbi.
nlm.
nih.
gov/tools/primer-blast/).
Expressiondatawasnormalizedrelativetobetaactinforeachsample,usingcDNAproducedasdescribedabove,withoutPCRamplification.
Q-PCRwasperformedinaPerkin-Elmerthermo-cyclerwiththefollowingconditions:95C1min,98C20sec,60C1min,repeatedfor40cycles.
ResultsandDiscussionPurificationofMelanocytes,Iridophores,andRPEfromWholeEmbryostoGenerateCell-specificGeneExpressionDataAtthreedayspostfertilization(dpf),melanocytes,iridophores,andtheRPEarereadilyvisibleinzebrafish.
Melanocytesareextensivelydendritic,andidentifiableduetothepresenceofblackmelanin.
Iridophoresareroundreflectivecells,easilyseenwithamicroscopeusingincidentlight.
TheRPEispresentintheeyesasahexagonallypackedlayerofmelanizedcells.
However,thepercentageofmelanocytes,iridophores,andRPEcomparedtoallothercellsinthefishislessthan1%,makingcell-specificgeneexpressionanalysisfromthewholeorganismdifficult.
AidedbyapreviouslyreportedmethodusingdensitygradientcentrifugationtoisolatemelanocytesfromthecaudalfinsofBlackmoorgoldfish[20],wedevelopedasimpleproceduretorapidlypurifythesepigmentedcellsfromzebrafishembryos.
Thisprocedurereliesontheinherentdensitiesofthemelanin-filledmelanocytesandRPE,andtheguanine-fillediridophores.
Inbrief,cellsareenzymaticallydissociatedfromintactfish,enrichedforpigmentcellsviadensitygradientcentrifugation,andsortedbyflowcytometryusingtheautoflourescentpropertyofiridophores(Figure1).
Theotherpigmentcellsinzebrafish,thepteridine-containingxanthophores,Table4.
Sharedgeneexpressionamongmelanocytesandiridophores.
GeneMelanocyteIridophoreRPEEmbryoNotessyngr2l86.
58450.
7455.
7281.
471Unclearfunction,transmembraneproteintuba8l378.
779132.
46511.
7250.
397pumalocus[102]pcdh10a59.
63954.
8005.
1430.
423Neuralcrestexpression[64]crestin16.
92225.
6462.
2140.
145Expressedrepetitiveelement[60]LOC55921614.
65227.
2121.
3500.
689UncharacterizedRhoGEFsi:dkey-72l14.
711.
65740.
2471.
0040.
000RhodopsinfamilyGPCRcdk158.
4926.
4750.
2910.
112Cyclindependentkinaseemp3l6.
85917.
6440.
2930.
046Tumorsuppressor[103]lamb1b5.
4866.
1380.
6300.
427Laminin-typeEGF-likedomainszgc:1583284.
3558.
5530.
7760.
004Uncharacterized,EMIandvWBFdomainsopn53.
8241.
7960.
1930.
000RhodopsinfamilyGPCRppfia23.
5213.
8590.
4430.
037Axonguidance[104]rab27bl3.
26313.
3310.
2780.
118Melanosometransport[105]mc1r2.
1823.
9580.
0760.
000GPCRformelanocytestimulatinghormonebirc71.
8182.
9600.
2330.
000Inhibitorofapoptosisprotein[106]RPKMvaluesforgenesexpressedinmelanocytesandiridophoresatleast5-foldgreaterthanRPEandwholeembryos.
doi:10.
1371/journal.
pone.
0067801.
t004TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org7July2013|Volume8|Issue7|e67801donotpassthroughthedensitygradientandarenotisolatedbythismethod.
Followingcellpurification,mRNAisisolatedandIlluminalibrariesareconstructedfromcDNA(Figure2).
Wepurifiedmelanocytes,iridophores,andRPEasdescribedaboveandperformedmRNAsequencingfor11,5,and5independentlyisolatedsamples,respectively.
Inordertoconfirmthequalityofourcellspecificlibraries,weassembledalistofgenesknowntobeexpressedinmelanocytes,iridophores,andRPE.
WhenwecomparedexpressionofthesegenesinourcDNAlibrariestothiscontrollist,weobservedexpressioncorrespondingtoeachcelltype(Table1).
Furthermore,wealsofoundgoodcorrelationforfoldchangesdeterminedbyqPCRofindependentlypreparedbiolog-icalsamplesandthosedeterminedbyourRNA-seqdata,indicatingourRNA-seqfoldchangecalculationsbetweenpigmentcelltypesarerealisticvalues(FigureS4).
Forinstance,ltk,knowntobeamarkerofiridophores,ishighlyenrichediniridophorescomparedtomelanocytes(redsquarewithxinFigureS4)[27].
Similarly,wefinddct,aknownrequirementformelanogenesis,tobehighlyenrichedinmelanocytesandRPErelativetoiridophores.
Interestingly,wealsofindrpe65a,well-knowntobeexpressedinRPE,toalsobeexpressedbymelanocytes.
Thisisnotentirelysurprising,asRPE65isknowntobepresentinkeratinocytes,melanocytes,andmelanoma,inadditiontotheRPE[46,47].
Thus,usingthismethod,cDNAfrommelanocytes,iridophores,andRPEcanbegeneratedconcomitantlyfromthousandsofwholezebrafishembryosinasingleday.
Thedevelopmentofafastandrobustmethodforpurifyingthesepigmentcellsgreatlysimplifiestheproductionofmultiplebiologicalreplicatesneededforinformativeanalysisofhigh-throughputexpressiondata.
AnalysisofGeneExpressioninMelanocytes,Iridophores,andRetinalPigmentedEpitheliumWefirstaimedtouseourRNA-seqdatatoidentifythetotalnumberofgenesexpressedineachofthepigmentcelltypes.
Toeliminatelow-levelbackgroundexpression,weappliedabaselineexpressionthresholdof1readperkilobaseoftranscriptpermillionreads(1RPKM).
Wechosethisthresholdbasedontheabilitytodetectexpressionofknownpigmentationandneuralcrestgenes.
Forexample,wefoundmc1r,kita,andfoxd3tohaveRPKMvaluesTable5.
Iridophoreenrichedgenes.
GeneIridophoreMelanocyteRPEEmbryoNotesifi30l2138.
176.
5527.
161.
23Melanomaantigenprocessing[107]fhl31003.
294.
6420.
971.
12Actinorganization[108]slc23l592.
0510.
649.
701.
19Nucleobasetransport[109]gpnmb423.
581.
025.
281.
70Glycoprotein[70]LOC100538040396.
381.
529.
301.
85Uncharacterized,likelyGPI-linkedglycoproteinLOC100334697330.
191.
064.
780.
09Crystalprotein-liketpd52l1252.
615.
697.
751.
02Membranetrafficking[110]LOC100535932221.
280.
984.
640.
18Uncharacterizedpltp145.
301.
072.
020.
02Lipoproteinmetabolism[111]LOC795494127.
940.
432.
170.
11NucleotidemetabolismdomainLOC10033098799.
010.
561.
800.
01Uncharacterizedzgc:7737594.
860.
442.
240.
01Haloaciddehalogenaseslc25a38a94.
771.
101.
630.
46MitochondrialcarrierproteinLOC10053497081.
870.
652.
480.
18Uncharacterizedtmem179bl79.
070.
320.
660.
14Transmembraneproteinfkbp1532.
470.
550.
510.
31Uncharacterizedpcolcel30.
460.
310.
220.
00Extracellularmatrixproteinsi:ch211-38m6.
628.
420.
120.
230.
00MajorFacilitatorSuperfamilytagln3b24.
880.
130.
740.
06Cytoskeleton-associated[112]alx4b24.
810.
100.
240.
08Skullossification[89]hsf521.
070.
050.
200.
00Transcriptionfactorosbpl1018.
670.
260.
550.
08Intracellularlipidreceptorsi:dkey-225f23.
414.
270.
140.
130.
00Uncharacterizedzgc:11205412.
990.
340.
400.
00bZIPtranscriptionfactorslc52a312.
730.
030.
120.
12Riboflavintransportercart19.
870.
190.
280.
00ALX1homologmyadm7.
810.
140.
230.
00Myeloidassociateddifferentiationznf8317.
040.
090.
100.
06Zinc-fingerdoubledomainsi:ch211-14k19.
85.
800.
060.
090.
00Uncharacterizednfascl5.
270.
070.
110.
00FibronectindomainRPKMvaluesforgenesexpressediniridophoresatleast30-foldgreaterthanmelanocytesandRPE,and100-foldgreaterthanembryos.
doi:10.
1371/journal.
pone.
0067801.
t005TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org8July2013|Volume8|Issue7|e67801inmelanocytesof2.
2and3.
8,and3.
0,respectively(Table1).
Furthermore,wedetectedmorethan95%ofthegenesexpressedat1RPKMwithasequencingdepthofapproximatelyonemillionreadsperlibrary,whichestablishedoursequencingdepththreshold(FigureS5).
Usingthisbaselinethresholdof1RPKM,wefound8,472genesareexpressedbymelanocytes,8,096byiridophores,and9,053bytheRPE(SeeTableS5fortheaveragedRPKMsandT-testvalues).
Toputthisintoperspective,wealsoalignedover700millioncDNAtagsgeneratedbytheStemplelaboratory[48]fromavarietyofembryonicstagesandadulttissues.
Wefoundthismixeddatasetrevealedexpressionofatleast1RPKMfor20,548geneentriesfromourdatabaseof25,102uniquecodingsequences(Notshown).
Theseresultsindicateournon-redundantdatabaseisareasonableapproximationoftheprotein-codingtranscriptomeinzebrafish,andconcludethat30–40%ofallgenesareappreciablyexpressed(.
1RPKM)bythesespecificcelltypes.
Melanocyte,Iridophore,andRPEGeneExpressionisCorrelatedComparedtoWholeEmbryosHavingobtainedgeneexpressiondataforthesepigmentcells,wesetouttoidentifysignaturesofpigmentcellfunctionsfromthedatasets.
Wereasonedthatgenesco-enrichedamongpigmentcelltypeswouldindicatesharedpigmentcellfunctions.
AsaninitialFigure3.
Theguaninesynthesiscycleishighlyenrichediniridophores.
ShownisamodelforguanineproductionbasedontranscriptomedataasgiveninTable6.
Genesthatarestatisticallyenrichedcomparedtomelanocytesareshowninbold,thosenotstatisticallydifferentareinnormalfont.
Thearrowthicknessescorrespondtothefoldchangesiniridophoresrelativetowholeembryos.
ChemicalstructuresarefromtheKEGGCompounddatabase.
doi:10.
1371/journal.
pone.
0067801.
g003TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org9July2013|Volume8|Issue7|e67801assessmentofthesimilarityofgeneexpressionbetweenmelano-cytes,RPE,andiridophores,weappliedPearson'sProduct-MomentCorrelationtesttothedatasets.
Becausecorrelationvaluescanbeartificiallyskewedbyoutlyingdatapoints[49],wecalculatedtheaveragecorrelationfrom24,102overlappingwindowsof1000genes,sortedbyincreasingwholeembryoexpression(FigureS6).
Usingthismetric,wefoundthemelanin-producingmelanocytesandRPEwerethemosthighlycorrelated(r=0.
90).
Theneuralcrest-derivedmelanocytesandiridophoresweremodestlycorrelated(r=0.
52).
RPEandiridophoreswerealsomodestlycorrelated(r=0.
49).
Weexpectedthecorrelationtobedramaticallylowerwhencomparingasinglepurifiedcelltypetowholeembryosthanwhencomparingtwopurifiedpigmentcelltypestoeachother.
WethusdeterminedthecorrelationsofeachpigmentcelltypetomRNA-seqdatafromwhole3dpfzebrafish.
Bythisgeneralassessment,melanocytes,RPE,andiridophoresaredistinctfromwholeembryos,withaveragecorrelationvalueslessthan0.
02.
Presumably,thislowvalue(0.
02)reflectsthebaselinecorrelationcomponentfromexpressionofhousekeepinggenessharedbyallcells,andgreatervalues(0.
4–0.
9)reflectsharedspecificgeneexpressionamongthepigmentcelltypes.
GeneExpressionIndicativeofaPigmentCellIdentityThesecorrelationsdescribedabovesuggesttherearegenesenrichedinallthreepigmentcelltypesthataregenerallyindicativeofpigmentcellidentity.
ItisnotclearaprioriwhichTable6.
Guaninesynthesis-relatedgeneexpressionenrichmentiniridophores.
RPKMP-ValueGeneIridophoreMelanocyteRPEEmbryoIridvs.
MelIridvs.
RPEslc2a15a(GLUT5)48.
550.
160.
730.
287.
66E-037.
89E-03pgm234.
000.
540.
731.
244.
64E-034.
68E-03dera3.
670.
980.
392.
763.
40E-041.
55E-04pfkp12.
250.
631.
282.
421.
31E-021.
53E-02aldoca6.
220.
040.
000.
832.
50E-032.
45E-03gapdhs1605.
92191.
73163.
319.
031.
92E-021.
79E-02pgk1281.
34146.
41125.
4638.
369.
05E-037.
22E-03pgam1a587.
26216.
58158.
8432.
461.
13E-034.
02E-04eno31503.
93215.
46307.
89100.
274.
30E-084.
09E-05pkm2a371.
83115.
9697.
9424.
944.
52E-042.
83E-04ldhba440.
6274.
5344.
4672.
722.
70E-042.
85E-04phgdh*1538.
41371.
96175.
608.
254.
21E-032.
22E-03psat1160.
857.
0311.
927.
624.
68E-033.
85E-03psph2602.
05146.
30102.
781.
361.
53E-031.
38E-03shmt286.
025.
375.
922.
839.
83E-039.
50E-03mthfd1177.
667.
988.
4113.
992.
71E-022.
72E-02fh562.
28150.
18109.
278.
652.
53E-021.
81E-02mdh1a*3948.
05388.
29432.
9527.
332.
55E-051.
48E-05pck21.
440.
100.
016.
091.
05E-028.
50E-03aclya106.
5742.
2939.
308.
192.
49E-031.
72E-03rpia11.
662.
492.
246.
524.
70E-033.
89E-03prpsap1*95.
2735.
4824.
123.
133.
13E-031.
88E-03ppat49.
232.
493.
727.
894.
22E-034.
26E-03gart167.
666.
414.
293.
466.
01E-035.
72E-03pfas34.
590.
530.
564.
531.
52E-031.
51E-03paics1465.
3757.
1255.
3511.
683.
63E-033.
50E-03adsl805.
7746.
0636.
4312.
452.
69E-032.
38E-03atic467.
167.
609.
6318.
341.
01E-021.
02E-02impdh1b816.
1911.
4521.
981.
061.
30E-031.
21E-03gmps102.
413.
865.
146.
351.
15E-031.
02E-03prtfdc1*309.
881.
563.
100.
001.
39E-031.
39E-03hprt1l21.
310.
251.
1315.
997.
28E-046.
03E-04adssl0.
731.
041.
756.
822.
61E-011.
08E-01ak187.
12228.
27149.
7440.
664.
24E-031.
75E-01RPKMvaluesforenzymesrelatedtoguaninesynthesisiniridophores.
Enzymesaregroupedwiththeircommonlyassociatedpathways.
Rate-limitingenzymesfromeachofthespecificpathwaysareindicatedwithanasterisk.
Notably,adsslisthefirstadenine-specificgeneinpurinesynthesis,andisnotupregulatediniridophores.
doi:10.
1371/journal.
pone.
0067801.
t006TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org10July2013|Volume8|Issue7|e67801genesshouldbesharedbydifferentpigmentcelltypes.
Inordertoconstructaninformativesetofgenesforthispurpose,wesearchedforthosethatareexpressedwithina2-foldchangeofeachotherataminimumof4RPKM,andatleast100-foldgreaterthanwhole3dpfembryos.
Wefound28genesfulfilledthesecriteria(Table2).
Remarkably,4ofthetop5highestexpressedgenesonthislistareribosomalproteins.
Pigmentcellenrichmentofribosomalcom-ponentsisnotsurprising,consideringseveralmousecoatcolormutantsareribosomalproteins,althoughtheyarenottheribosomalcomponentsenrichedhere[50–53].
Wealsofindthezebrafishalbinogene(slc45a2)tobeamongthehighestexpressedgenesineachofthesepigmentcelltypes[54,55].
Theexpressionofslc45a2iniridophoresisinteresting,consideringalbinofisharenotreportedtohaveaniridophoredefect.
OnepossibilityisthatSLC45A2performsacommonrolefororganellepHhomeostasisinpigmentcells.
However,insituanalysisrevealsnoenrichedexpressioninxanthophores,suggestingthatslc45a2expressionisnotsharedbyallpigmentcells[55].
Thislistofco-expressedgenesalsocontainsseveralotherunexpectedmembers,includingtheJak-STATcytokinereceptorcrfb5,andtheBAX-inhibitorproteinghitm.
Itisnotclearwhatrolesthesegenesplayinpigmentcells,butthislistprovidesastartingpointforunderstandingtheirsharedfunctionsinpigmentcellbiology.
GeneExpressionIndicativeofCellularFunctionorDevelopmentalOriginWewerealsointerestedtoidentifygenesthatwereenrichedinonlytwopigmentcelltypesthatwouldrevealsharedfunctionordevelopmentalorigin.
Forthisanalysis,wefilteredforsharedexpressionofgenesintwocelltypesatleast2-foldoverthethirdcelltype,wheredifferencesexceededasignificanceofp,0.
05.
UponvalidationofexpressiondifferencesviaqPCRbetweenpigmentcellgenesandwholeembryos,wefoundasystematicbiasofovercallingthefoldchangesbetweenpigmentcellvaluesandwholeembryos(FigureS7).
Basedonthisresult,wealsorequiredan8-foldgreaterthanembryoexpressionthresholdforco-enrichmentandcell-specificgenes,asdiscussedbelow.
Usingthesecriteria,wefound214geneswereenrichedinbothmelanocytesandRPEbutnotiridophores(TableS6),and62genesenrichedinmelanocytesandiridophoresbutnotRPE(TableS7).
GiventhattheRPEandiridophoresdonotshareapigment-typeproductionordevelopmentalorigin,weexpectedfewergenestobeco-enrichedinthesecells,butnotexpressedbyneural-crestmelanocytes.
Onlyonegene,alcama(alsodm-grasporneurolin-a),thetargetofthezn-5/8monoclonalantibody[56,57]wasenrichedinbothRPEandiridophoreswhencomparedtomelanocytesandwholeembryos(TableS8).
ThisisconsistentwiththenotiontherearefewspecificfunctionssharedonlybetweeniridophoresandRPE.
Itisintriguingthatalcamahasbeenfoundtomediateendothelin1signalingincartilagedevelopment[58].
Relatedly,endothelinreceptor(ednrb1)signalingisrequiredforiridophoredevelopmentinzebrafish[5].
ItwillbeinterestingtoknowwhetheralcamamutationshavefunctionalconsequencesintheiridophoreortheRPE.
Therefore,wesuggesttheselistsofco-expressedgenesarelikelyenrichedforcommonmetabolicfunctions,inthecaseofmelanocytesandRPE,ordevelopmentalorigins,inthecaseofmelanocytesandiridophores.
MelanocyteandRPECo-ExpressionDuetotheirsharedfunctionofproducingmelaninandmelanosomes,weexpectedmanygenestobesharedbetweenmelanocytesandtheRPEwhencomparedtoiridophores.
Wefound214genesthatfitourparametersforsharedenrichment(TableS6).
Asexpected,therearemanygenespresentinvolvedinmelaninsynthesisandmelanosomebiogenesis,includingpmela,dct,tyrp1b,tyrp1,pah,andslc24a5.
Thereareseveraltranscriptionfactorsinthisenrichmentgroup,includingthreeforkheadbox(foxo1b,foxp4,andfoxg1b)andfivehomeobox-containingtranscrip-tionfactors(hmx1,hmx4,otx1a,otx2,andotx5).
AlthoughnotrequiredforRPEdevelopmentinzebrafish,mitfaisexpressedintheRPEatarelativelyhighlevel,consistentwiththereportedabilityofmitfatopromotepigmentedfateinzebrafishretinas[59].
However,aswefindmostoftheothertranscriptionfactorsco-enrichedinmelanocytesandRPEareexpressedatlowerlevels,itisinterestingtospeculatethatoneormoreofthesefactorsareabletocompensateformitfaspecificallyintheRPE.
Thus,thisco-enrichmentsetmayidentifyamorecomprehensivelistofgenesformelanocyteandRPEidentityandmelanosomebiogenesis.
Anadditionalfindingapparentfromthe38mosthighlyexpressedgenesco-enrichedinmelanocytesandRPEisthatiridophoresalsoexpressthesegenes,albeitatamuchlowerlevel(Table3).
Theclosedevelopmentallineagerelationshipbetweenmelanocytesandiridophoressuggestsapossible''leakiness''ofspecificitythatmayresultinweakexpressionofmelanocytegenesiniridophores.
Thispossibilityissupportedbytheobservationthattheparadezebrafishmutantcontainspigmentcellswithbothmelanocyteandiridophorecharacteristics[22].
MelanocyteandIridophoreCo-ExpressionBecauseofthecommondevelopmentaloriginfromtheneuralcrestformelanocytesandiridophores,wespeculatedthatneuralcrest-specificgeneexpressionwouldbereadilyidentifiable.
Wefound62genesthatweresignificantlyupregulatedinmelanocytesandiridophoresoverRPE(TableS7).
Alistofthe15genesexpressedatleast5-foldgreaterthanRPEand10-foldgreaterthanwholeembryosisshowninTable4.
Includedinthisgroupareseveralwell-knownregulatorsandmarkersofneuralcrestandpigmentcelldevelopment,includingthetranscriptionfactorssox10andtheexpressedrepetitiveelementcrestin[60].
Thewell-knownpigmentationgenemc1risalsoexpressediniridophoresandmelanocytes.
Alsoincludedisthecelladhesionmoleculepcdh10a,whichisexpressedbymigratingzebrafishneuralcrestcells,andactsasatumorsuppressorinseveralcancers[61-63].
Wealsofindaretinoicacidnuclearreceptorsubfamilymember(rxrga)previouslyreportedtobeexpressedinneuralcresttissues[64–67].
Becausewefindthisknownsetofneuralcrestgenesinthisco-enrichmentlist,otherunknownmarkersofneuralcrestidentityarelikelytobepresent.
Forinstance,severaltranscriptionfactorsareinthisenrichmentsetnotpreviouslyreportedinneuralcrest,includingtheforkheadboxtranscriptionfactorfoxo1a,aswellasthecellcycleregulatorcdk15.
Thus,thislistofgenesco-enrichedinmelanocytesandiridophoresmaymorebroadlyidentifymarkersofneuralcrestorigin.
Cell-TypeSpecificGeneExpressionWewerealsointerestedindetermininggeneexpressionspecifictoeachcelltype.
BecausemelanocyteandRPEgeneexpressionhavebeenpreviouslycharacterized,wedonotdiscussthemindetailhere,butpresentthedataasthesupplementaltableslistedbelow[36–38,68,69].
Forthisanalysiswerequiredanexpressionlevelatleast2-foldovertheothertwocelltypes(p,0.
05),aswellas8-foldgreaterexpressionthanwholeembryos.
Thisfilteringstrategyresultedinalistof108genesspecificallyenrichedinmelanocytes(TableS9)and24intheRPE(TableS11).
Toindependentlyvalidatetheseenrichmentsets,wegeneratedseparatebiologicalsamplesforeachcelltypeandcomparedexpressionofselectedgenesfromtheenrichmentsetsviaqRT-PCR.
WeobservedgoodcorrelationofRNA-seqandqRT-PCRTranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org11July2013|Volume8|Issue7|e67801relativeexpressionlevels,indicatingourRNA-seqdataisareliableindicatorofthegeneexpressioninthesepigmentcells(r2=0.
75,FigureS4).
ThesedatawillbeusefulforfurtherstudiesofmelanocyteandRPEbiology.
IridophoreGeneEnrichmentIncontrasttomelanocytesandRPE,theiridophoretranscrip-tomehasnotpreviouslybeenexplored.
Sinceiridophoreproduceaguanine-basedpigment,ratherthanthemelanincharacteristicofmelanocytesandRPE,weexpectedtofindmanygenestobespecificallyenrichedinthiscelltype.
Thisindeedturnedouttobethecase,with346genespassingourbaselinethresholdforenrichment(TableS10).
Includedinthisenrichmentlistareseveralfactorsknowntobeimportantforiridophoredevelopment,includingltk,ednrb1,andpnp4a[5,26,27].
Also,inordertoidentifypreviouslyunreportedgenesthatmayplayinterestingrolesiniridophores,wefilteredourlisttoincludegenesexpressedatleast30-foldgreaterthanmelanocytesandRPE,and100-foldgreaterthanwholeembryos.
Thirtygenesmetthesecriteria(Table5).
Thethirdhighestexpressedgeneonthislist,slc23l,mayactasaguaninetransporter.
Inmammals,theSLC23genefamilyhasrolesintransportingnucleobases,suchasguanine,aswellasvitaminC.
ItisnotclearhowthishighlyexpressediridophoregeneidentifiesauniqueroleforvitaminCintheiridophore,butitistemptingtospeculatearoleinguaninetransport,eithertotransportguanineintothecell,orperhapstotransportnewlysynthesizedguanineintothereflectingplateletorganelles.
Onesurpriseisthefindingthatgpnmbishighlyenrichedintheiridophore.
RolesforGPNMBhavebeendescribedformelano-cytes,melanoma,andthepigmentedirisinmammals,andithasbeensuggestedtoactbothasaplasmamembraneproteinandacomponentofthemelanosome[70].
Ourfindingthatgpnmbismorehighlyexpressedintheiridophorethanthemelanocyteraisesthepossibilityofasimilarfunctioniniridophorereflectingplateletorganellebiogenesis.
Alsointhisiridophore-specificenrichmentlistaresixuncharacterizedgenes.
Theirenrichmentiniridophoresmayaidinidentifyingfunctionsfortheseproteins.
Wespeculatethatmanyiridophore-enrichedgeneswillbeindicativeofnovelcell-specificbiologicalfunctions.
Manyofthegenespreviouslyknowntobeexpressedbyiridophoresarecomponentsofthepurinesynthesispathway,asiridophorepigmentlargelyconsistsofstacksofguanineplates[30,71].
Accordingly,wefoundadramaticenrichmentofenzymescomprisingthepathwayofguaninemetabolism,fromextracellularglucoseimportandglycolysis,throughdenovosynthesisandpurinesalvage(Table6).
WhencomparingiridophorestomelanocytesandRPE,wefind5facilitatedglucosetransporters,7/11stepsofglycolysis,and9/9enzymesfordenovopurinesynthesistobeenriched.
Giventhatiridophorepigmentconsistslargelyofguanine,onemightexpecttheguaninepathwaytobespecificallyupregulated.
Consistentwiththismodel,wefoundthesplitinthepurinesynthesispathwayatIMPtofavorguanineproductionratherthanadenine.
Thefirstguanine-specificenzyme,impdh1b,isexpressedatalevel71-foldgreaterthanmelanocytes.
Incontrast,thefirstadenine-specificenzyme,adssl,isnotsignificantlydifferentfrommelanocytesorRPE,at0.
7-foldthelevelofmelanocytes.
Uponinspectionoftheknownpathwayofguanineproduction,weobservedsynthesisofguaninefromGMPlikelyresultsintherecyclingof5-Phosphoribosyl1-Pyrophosphate(PRPP),therate-limitingsubstrateinpurinesynthesis(KEGGPathway:dre00230).
Theenzymeresponsibleforthisfinalstepofguaninesynthesis,prtfdc1,isenriched198-foldovermelanocytes(p,0.
01).
Fromourexpressiondata,wesuggestamodelofguaninepigmentproductioniniridophoresthatillustratesacycleofguaninesynthesisutilizingPRPPasarecycledcarriermolecule(Figure3).
Inthiscycle,specificenzymesfromglycolysis,thepentosephosphatepathway,serine/glycinemetabolism,andthecitratecycle,areupregulatedtocoordinatetheextensiveguaninesynthesisrequiredforthereflectiveiridophorepigment.
Anotherquestioniniridophorebiologyishowthemembranousplateletscontainingthereflectiveguaninecrystalsareformed.
OurdatasuggestsalikelycontributionfromADPRibosylationFactors(ARFs)andRabGTPases.
ARFsarealargefamilyofras-relatedGTPasesthatregulatemembranetraffickingandorganellestructure(Forreviewsee[72]).
WefindtwoARF-relatedgenestobesignificantlyenrichediniridophorescomparedtomelano-cytesandRPE,arf6andarfip1.
Interestingly,wealsofoundtwomembersoftheras-relatedoncogenefamilytobeenrichediniridophores;rab27bandrab38.
RabGTPasesregulatemanyaspectsofmembranousvesicleformationandtraffic[73].
Inhumans,Rab27mutationscausehypopigmentationassociatedwiththeimmunodeficiencydisorderGriscellisyndrometypeII[74].
Inaddition,theformationofCOPItransportvesiclesisknowntobemediatedbytheinteractionofGAPDHwithRAB2[75].
PhosphorylationofGAPDHoccursthroughtheSrc/PI3K/AKTpathway,oftendownstreamofreceptortyrosinekinaseactivation[76].
Wefoundasinglereceptortyrosinekinaseenrichediniridophores,ltk,whichwehavepreviouslyshowntoberequiredforiridophoredevelopment[27].
ItisthusinterestingtospeculatethatcombinedrolesexistforGAPDHinbothproducingreactivecarbonylspeciesduringglycolysisforguanineproduction,aswellasinformingtheorganelleswithinwhichiridophorepigmentiscontained.
Iridophoresarearelativelyless-studiedpigmentcellthanmelanocytesandRPE,andasidefromtherequirementsforfoxd3,ednrb1,andltk,notmuchisknownregardingthetranscriptionalregulationofiridophoreidentity.
Onemightexpectasinglemasterregulatorofiridophoreidentity,analogoustomitfainmelanocytes,whichishighlyexpressed,tobereadilyidentifiableinourdata.
However,wedidnotidentifyahighlyexpressedsinglecandidateforamasterregulatorofiridophoreidentity.
Instead,wefoundmanymoderatelyexpressedtranscriptionfactorswithknownmouseorhumanorthologuestobesignificantlyenrichediniridophores.
Ofthese,thereareseveralcandidatesthatstandoutaspossibleregulatorsofiridophoreidentity.
Onememberofthebasichelix-loop-helix(bHLH)familyoftranscriptionfactorsisspecificallyenrichediniridophores,tfec.
Knowntobeexpressediniridophores,TFECformsheteroandhomodimerswithotherbHLHmembersandcanfunctionasatranscriptionalactivatororrepressor[77–79].
Interestingly,thereisoneotherbHLHinenrichediniridophoreswhencomparedtomelanocytesandRPE,butitdidnotmeetthe8-foldrequirementoverwholeembryos.
Thisgene,mycl1a,isaparalogoftheclassiconcogenicproteinMYC.
Akeycomponentofiridophoreidentityistheupregulationofglycolysisandguaninesynthesisenzymes.
ItisestablishedthatMYCupregulatesglycolysis,DNA-synthesis,andnucleotidemetabolism[80–82].
AnotheroncogeneshowntoregulateglycolysisandassociatedfeederpathwaysisEts-1[83].
Thev-ets-erythroblastosisvirusE26oncogenehomolog1a(ets1a)isalsohighlyenrichediniridophores.
Inmouse,tfectranscriptionisactivatedthroughmultipleets-bindingdomainsinitspromoterregion,suggestingaconservedregulatorymechanismforETS1Aintfectranscriptionalregulationaswell[84].
FurtherworkwillbenecessarytodeterminewhetherTFECandMYCl1Acoordinatetheupregulationofguaninesynthesisenzymesiniridophores.
Wealsofindfivehomeobox-containingtranscriptionfactorsareenrichediniridophores:gbx2,cart1,alx3alx4a,andalx4b.
KnowntohaveseveralrolesintheembryoandanearlyspecifierofTranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org12July2013|Volume8|Issue7|e67801posteriorneuralcrestinXenopus,gastrulationbrainhomeobox2(gbx2),ishighlyenrichediniridophores[85].
Remarkably,theotherfourhomeoboxtranscriptionfactorsweidentifiedareallaristaless-related;cart1,alx3,alx4a,andalx4b.
MembersoftheCart1/Alx3/Alx4familyofhomeodomainproteinsareknowntoregulateformationofskeletalelementsinorganismsrangingfromseaurchintomammals[86,87].
Inhumans,mutationandhaploinsufficiencyofALX4areassociatedwithskullossificationdefects[88,89].
Itisnotclearhowthesearistaless-relatedtranscriptionfactorsmightregulateiridophoreidentity,butrecognizingtheirexpressionherewillaidinunderstandingtheirfunction.
Together,theseiridophore-enrichedtranscriptionfactorslikelyplaykeyrolesinregulatingiridophoreidentity.
ConclusionTocontributetotheunderstandingofpigmentcellbiology,wehavedevelopedamethodforrapidlyandreliablypurifyingmelanocytes,iridophores,andretinalpigmentedepitheliumfromzebrafishembryos,followedbyglobalgeneexpressionanalysisbymRNAsequencing.
Thisworkrepresentsthefirstconcomitantcomparisonofthreepigmentcelltypesandwholezebrafishembryos,whichuniquelyallowedustoidentifyco-expressedgenesindicativeofsharedfunctionordevelopmentalorigin,aswellasthosethatarespecificallyenrichedinsinglecelltypes.
Wethushaveidentifiedmanygenespreviouslynotreportedtobeenrichedinthesepigmentcelltypes.
Inparticular,wediscoveredadramaticupregulationofspecificenzymesfromseveralmetabolicpathwaysthatcoordinateguaninesynthesisiniridophores,alongwithmanymembrane-traffickingcomponentsandtranscriptionfactorsthatarelikelycriticalforiridophoreidentity.
Thischaracterizationofglobalgeneexpressiondatafrommultiplepurifiedzebrafishpigmentcelltypeswillprovidearesourceforfurtherbiologicalanalysisofthesecells.
SupportingInformationFigureS1RPECorrelations.
Shownin(A)aretheRPKMsofRPE_5_143hpf_32,whichwascollectedat143hpf,comparedtotheaverageRPKMsofRPElibraries1–4,whichwerecollectedat77–86hpf,forthe9029genesdetectedatanaverageof1–5000RPKMacrossallfiveRPElibraries.
Shownin(B)aretheRPKMvaluesforallgenesdetectedbetween1and5000RPKMbyRPEsamplesheldat25uCand32uC.
Magentacirclesrepresentthe24genesdescribedasRPE-enrichedinouranalysiscomparedtoiridophores,melanocytes,andwholeembryos(r=0.
95).
(TIF)FigureS2MelanocyteTimePointCorrelations.
ShownarescatterplotsofRPKMvaluesforallgenesexpressedbetween1and5000RPKMbymelanocytescollectedatdifferenttimepoints.
(TIF)FigureS3TechnicalReplicateCorrelation.
ShownarescatterplotsofRPKMvaluesobtainedforthetechnicalsequencingreplicatesofsampleMel_3_77hpf.
Sequencingruns351and433weresingleendreadsof36and42nucleotides,respectively.
399s61and399s62representthetwoendsofapairedend101sequencingrun.
(TIF)FigureS4QuantitativeRT-PCRandmRNAsequencingexpressiondataarecorrelated.
Examinedgenesareindicatedbyobjectshape,asinthelegendonthelowerright.
Celltypecomparisonsareindicatedbycolorinthelegendonupperleft.
Thepurplelineindicatesthepositionofperfectcorrelation.
RNA-seqfoldchangesarecomputeddirectlyforeachcelltypecomparison[i.
e.
log2(melanocyteRPKM/iridophoreRPKM)].
QPCRfoldchangesarecalculatedbyfirstnormalizingexpressionrelativetobetaactin,followedbythelogtransforma-tion.
(TIF)FigureS5TranscriptomeCoverage.
Thenumberofgenesidentified(y-axis)pernumberofsequencereads(x-axis)obtainedisplottedforeachsampleusedinthisanalysis.
Technicalreplicatesareshownasindividuallines,coloredbylibrarytypeasindicatedonthelowerright.
Forexample,thethreegreenlinesrepresentthethreetechnicalsequencingreplicatesofthepooled3dpfwholeembryoscDNAlibrary.
Thedashedverticallineisatonemillionreads.
(TIF)FigureS6PearsoncorrelationsofRNA-seqexpressiondata.
Genesareorderedbyincreasingwholeembryoexpression(Y-axis).
Eachpointrepresentsthecorrelationvalueforthe1000gene-windowbetweentheindicatedcell-typecomparison,begin-ningatthatposition.
Themosthighlyexpressedgenesinwholeembryosareribosomalproteins,whichcorrespondtoaslightpeakincorrelationvalueswhencomparedtomelanocytes(lowerright).
TheaveragePearsoncorrelationsacrossallwindowsforeachcomparisonareindicatedontheright,withcorrespondinghorizontallines.
(PDF)FigureS7WholeembryosRNA-seqfoldchangebias.
QPCR-basedfoldchangevaluesdemonstrateasystematicovercallingoffoldchangevaluesuponcomparisonofrandomlyfragmentedwholeembryocDNAlibrarieswithreduced-represen-tationpigmentcelllibraries.
Thepurplelinerepresentsthepositionofperfectcorrelation.
(TIF)TableS1SequencingSummary.
GrosssequencingresultsforcDNAlibrariesusedinthisanalysis,withtheirrespectivedevelopmentaltimepointsat28.
5C.
Unlessindicated,fishusedforeachlibrarywereheldat25uC.
Thetypeofsequencingrunsareindicatedbythesuffixesoftechnicalreplicatenames:351:singleend36,433,436,and440:singleend42,399s61and399s62:pairedend101.
*UniquelymappedtotheWashingtonUniversitynon-redundantcDNAdatabase.
TagsthatwerenotuniquelyassignedtoagenetypicallymappedtopolyA,genomic,andIllumina-adaptersequences.
**TotalRNAwasusedforIlluminalibrarypreparationof20pooled3dpfembryos,resultinginalowerfractionofuniquelymappingsequencesduetoalargeproportionofribosomalRNAinlibrary.
Abbreviations:Mel-Melanocyte,Irid-Iridophore,RPE-RetinalPigmentedEpithelium,hpf-hourspostfertilization,dpf-dayspostfertilization.
(XLSX)TableS2AllLibrariesRPKM.
Normalizedexpressionvalues(RPKM)foreachofthelibrariesusedinthisanalysis.
(XLS)TableS3Primers.
PrimersusedforlibrarypreparationandqPCRanalysis.
(XLSX)TableS4Non-redundantcDNAdatabase.
Sequencesforthegenesusedinthisanalysis.
(XLSX)TableS5PigmentCellExpressionAvgRPKMpValues.
AveragedRPKMexpressionvaluesformelanocyte,RPE,iridophore,andwholeembryo.
AlsoshownareStudent'sT-testTranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org13July2013|Volume8|Issue7|e67801p-values,ensembltranscriptandgenereferencenumbers,andthecorrespondingzv9genomiclocationofeachgene.
(XLSX)TableS6Melanocyte&RPEShared.
GenesexpressedbybothmelanocytesandRPEwithanRPKM2-foldgreaterthaniridophores,and8-foldgreaterthanwholeembryos.
(XLSX)TableS7Melanocyte&IridophoreShared.
Genesex-pressedbybothmelanocytesandiridophoreswithanRPKM2-foldgreaterthanRPE,and8-foldgreaterthanwholeembryos.
(XLSX)TableS8RPE&IridophoreShared.
AsinglegeneisexpressedbybothRPEandiridophoreswithanRPKM2-foldgreaterthanmelanocytes,and8-foldgreaterthanwholeembryos(XLSX)TableS9MelanocyteEnrichedGenes.
Genesexpressedinmelanocytes2-foldgreaterthanRPEandiridophores,and8-foldgreaterthanwholeembryos.
(XLSX)TableS10IridophoreEnrichedGenes.
Genesexpressediniridophores2-foldgreaterthanRPEandmelanocytes,and8-foldgreaterthanwholeembryos.
(XLSX)TableS11RPEEnrichedGenes.
GenesexpressedinRPE2-foldgreaterthaniridophoresandmelanocytes,and8-foldgreaterthanwholeembryos.
(XLSX)AcknowledgmentsWethankMichaelBrooks,MaximSchillebeeckx,andDavidMayhewforhelpfuldiscussionsandcommentsonthemanuscript.
WethanktheGenomeTechnologyAccessCenteratWashingtonUniversityfortheirIlluminasequencingexpertise.
WealsothanktheAlvinJ.
SitemanCancerCenteratWashingtonUniversitySchoolofMedicineandBarnes-JewishHospitalinSt.
Louis,Mo.
,fortheuseoftheHighSpeedCellSorterCore,whichprovidedFACSservice.
AuthorContributionsConceivedanddesignedtheexperiments:CWHRDMSLJ.
Performedtheexperiments:CWH.
Analyzedthedata:CWHRDMSLJ.
Contributedreagents/materials/analysistools:CWHRDMSLJ.
Wrotethepaper:CWHRDMSLJ.
References1.
HenionPD,WestonJA(1997)Timingandpatternofcellfaterestrictionsintheneuralcrestlineage.
Development(Cambridge,England)124:4351–9.
2.
JinEJ,EricksonCA,TakadaS,BurrusLW(2001)WntandBMPSignalingGovernLineageSegregationofMelanocytesintheAvianEmbryo.
DevelopmentalBiology233:22–37.
3.
RawlsJF,JohnsonSL(2003)Temporalandmolecularseparationofthekitreceptortyrosinekinase'srolesinzebrafishmelanocytemigrationandsurvival.
DevelopmentalBiology262:152–161.
4.
LeeH,LevorseJM,ShinMK(2003)Theendothelinreceptor-Bisrequiredforthemigrationofneuralcrest-derivedmelanocyteandentericneuronprecursors.
DevelopmentalBiology259:162–175.
5.
ParichyDM,MellgrenEM,RawlsJF,LopesSS,KelshRN,etal.
(2000)Mutationalanalysisofendothelinreceptorb1(rose)duringneuralcrestandpigmentpatterndevelopmentinthezebrafishDaniorerio.
DevelopmentalBiology227:294–306.
6.
CarreiraS,GoodallJ,AksanI,RoccaSA,GalibertM,etal.
(2005)MitfcooperateswithRb1andactivatesp21Cip1expressiontoregulatecellcycleprogression.
Nature433:764–9.
7.
LangMR,PattersonLB,GordonTN,JohnsonSL,ParichyDM(2009)Basonuclin-2RequirementsforZebrafishAdultPigmentPatternDevelopmentandFemaleFertility.
PLoSGenetics5:e1000744.
8.
RawlsJF,JohnsonSL(2001)Requirementsforthekitreceptortyrosinekinaseduringregenerationofzebrafishfinmelanocytes.
Development(Cambridge,England)128:1–7.
9.
BotchkarevaNV,KhlgatianM,LongleyBJ,BotchkarevVA,GilchrestBA(2001)SCF/c-kitsignalingisrequiredforcyclicregenerationofthehairpigmentationunit.
FASEBJournal15:645–58.
10.
NishimuraEK,GranterSR,FisherDE(2005)MechanismsofHairGraying:IncompleteMelanocyteStemCellMaintenanceintheNiche.
Science307:720–724.
11.
McNivenMA,WangM,PorterKR(1984)Microtubulepolarityandthedirectionofpigmenttransportreversesimultaneouslyinsurgicallyseveredmelanophorearms.
Cell37:753–65.
12.
SheetsL,RansomDG,MellgrenEM,JohnsonSL,SchnappBJ(2007)Zebrafishmelanophilinfacilitatesmelanosomedispersionbyregulatingdynein.
CurrentBiology17:1721–34.
13.
LaskinJD,PiccininiL,EngelhardtDL,WeinsteinIB(1982)ControlofmelaninsynthesisandsecretionbyB16/C3melanomacells.
JournalofCellularPhysiology113:481–6.
14.
PattonEE,WidlundHR,KutokJL,KopaniKR,AmatrudaJF,etal.
(2005)BRAFmutationsaresufficienttopromoteneviformationandcooperatewithp53inthegenesisofmelanoma.
CurrentBiology15:249–254.
15.
Saldana-CaboverdeA,KosL(2010)Rolesofendothelinsignalinginmelanocytedevelopmentandmelanoma.
PigmentCell&MelanomaResearch23:160–70.
16.
HerrlingT,JungK,FuchsJ(2008)Theroleofmelaninasprotectoragainstfreeradicalsinskinanditsroleasfreeradicalindicatorinhair.
SpectrochimicaActaPartA:MolecularandBiomolecularSpectroscopy69:1429–1435.
17.
LevinMD,LuMM,PetrenkoNB,HawkinsBJ,GuptaTH,etal.
(2009)Melanocyte-likecellsintheheartandpulmonaryveinscontributetoatrialarrhythmiatriggers.
JournalofClinicalInvestigation119:3420–36.
18.
PlonkaPM,PasseronT,BrennerM,TobinDJ,ShibaharaS,etal.
(2009)Whataremelanocytesdoingalldaylong…ExperimentalDermatology18:799–819.
19.
LinosE,SwetterSM,CockburnMG,ColditzGA,ClarkeCA(2009)IncreasingBurdenofMelanomaintheUnitedStates.
JournalofInvestigativeDermatology129:1666–1674.
20.
ClarkCR,TaylorJD,TchenTT(1987)PurificationofBlackMoorgoldfishmelanophoresandresponsestoepinephrine.
InVitroCellular&Develop-mentalBiology23:417–21.
21.
NaGY,PaekSH,ParkBC,KimDW,LeeWJ,etal.
(2006)Isolationandcharacterizationofouterrootsheathmelanocytesofhumanhairfollicles.
TheBritishJournalofDermatology155:902–9.
22.
KelshRN,BrandM,JiangYJ,HeisenbergCP,LinS,etal.
(1996)Zebrafishpigmentationmutationsandtheprocessesofneuralcrestdevelopment.
Development(Cambridge,England)123:369–89.
23.
MayerTC(1970)Acomparisonofpigmentcelldevelopmentinalbino,steel,anddominant-spottingmutantmouseembryos.
DevelopmentalBiology23:297–309.
24.
SearleA(1968)Comparativegeneticsofcoatcolourinmammals.
LogosPressLimited.
308p.
25.
CeolC,HouvrasY,Jane-ValbuenaJ,BilodeauS(2011)TheSETDB1histonemethyltransferaseisrecurrentlyamplifiedinandacceleratesmelanoma.
Nature471:513–7.
26.
CurranK,RaibleDW,ListerJA(2009)Foxd3controlsmelanophorespecificationinthezebrafishneuralcrestbyregulationofMitf.
DevelopmentalBiology332:408–417.
27.
LopesSS,YangX,Mu¨llerJ,CarneyTJ,McAdowAR,etal.
(2008)LeukocyteTyrosineKinaseFunctionsinPigmentCellDevelopment.
PLoSGenetics4:e1000026.
28.
HitchingsG,FalcoE(1944)TheIdentificationofGuanineinExtractsofGirellaNigricans:TheSpecificityofGuanase.
ProcNatlAcadSciUSA30:294–7.
29.
RohrlichS,RubinR(1975)BiochemicalcharacterizationofcrystalsfromthedermaliridophoresofachameleonAnoliscarolinensis.
TheJournalofCellBiology66:635–45.
30.
NgA,UribeR,YiehL,NuckelsR,GrossJ(2009)Zebrafishmutationsingartandpaicsidentifycrucialrolesfordenovopurinesynthesisinvertebratepigmentationandoculardevelopment.
Development136:2601–11.
31.
BhartiK,NguyenMT,SkuntzS,BertuzziS,ArnheiterH(2006)Theotherpigmentcell:specificationanddevelopmentofthepigmentedepitheliumofthevertebrateeye.
PigmentCellRes19:380–94.
32.
StraussO(2005)Theretinalpigmentepitheliuminvisualfunction.
PhysiologicalReviews85:845–81.
33.
RunkleEA,AntonettiDA(2011)Theblood-retinalbarrier:structureandfunctionalsignificance.
MethodsMolBiol686:133–48.
TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org14July2013|Volume8|Issue7|e6780134.
DridiS,HiranoY,TaralloV,KimY,FowlerBJ,etal.
(2012)ERK1/2activationisatherapeutictargetinage-relatedmaculardegeneration.
ProcNatlAcadSciUSA109:13781–6.
35.
ThompsonDA,GalA(2003)GeneticdefectsinvitaminAmetabolismoftheretinalpigmentepithelium.
DevelopmentsinOphthalmology37:141–54.
36.
LeungYF,MaP,DowlingJE(2007)Geneexpressionprofilingofzebrafishembryonicretinalpigmentepitheliuminvivo.
InvestigativeOphthalmology&VisualScience48:881–90.
37.
RizzoloLJ,ChenX,WeitzmanM,SunR,ZhangH(2007)AnalysisoftheRPEtranscriptomerevealsdynamicchangesduringthedevelopmentoftheouterblood-retinalbarrier.
MolVis13:1259–73.
38.
WistowG,BernsteinSL,WyattMK,FarissR,BehalA,etal.
(2002)ExpressedsequencetaganalysisofhumanRPE/choroidfortheNEIBankProject:over6000non-redundanttranscripts,novelgenesandsplicevariants.
MolVis8:205–20.
39.
GrnskovK,EkJ,Brondum-NielsenK(2007)Oculocutaneousalbinism.
OrphanetJournalofRareDiseases2:43.
40.
SpragueJ,DoerryE,DouglasS,WesterfieldM(2001)TheZebrafishInformationNetwork(ZFIN):aresourceforgenetic,genomicanddevelop-mentalresearch.
NucleicAcidsResearch29:87–90.
41.
JohnsonSL,NguyenAN,ListerJA(2011)mitfaisrequiredatmultiplestagesofmelanocytedifferentiationbutnottoestablishthemelanocytestemcell.
DevelopmentalBiology350:405–13.
42.
KimmelCB,BallardWW,KimmelSR,UllmannB,SchillingTF(1995)Stagesofembryonicdevelopmentofthezebrafish.
DevelopmentalDynamics203:253–310.
43.
MatzM,ShaginD,BogdanovaE,BritanovaO,LukyanovS,etal.
(1999)AmplificationofcDNAendsbasedontemplate-switchingeffectandstep-outPCR.
NucleicAcidsResearch27:1558–60.
44.
ZhuYY,MachlederEM,ChenchikA,LiR,SiebertPD(2001)Reversetranscriptasetemplateswitching:aSMARTapproachforfull-lengthcDNAlibraryconstruction.
BioTechniques30:892–7.
45.
MortazaviA,WilliamsBA,McCueK,SchaefferL,WoldB(2008)MappingandquantifyingmammaliantranscriptomesbyRNA-Seq.
NatureMethods5:621–628.
46.
HinterhuberG,CauzaK,BruggerK,Dingelmaier-HovorkaR,HorvatR,etal.
(2004)RPE65ofretinalpigmentepithelium,aputativereceptormoleculeforplasmaretinol-bindingprotein,isexpressedinhumankeratinocytes.
JInvestDermatol.
Feb;122(2):406–13.
47.
AmannPM,LuoC,OwenRW,HofmannC,FreudenbergerM,etal.
(2012)VitaminAmetabolisminbenignandmalignantmelanocyticskincells:importanceoflecithin/retinolacyltransferaseandRPE65.
JCellPhysiol.
Feb;227(2):718–28.
48.
CollinsJE,WhiteS,SearleSM,StempleDL(2012)IncorporatingRNA-seqdataintothezebrafishEnsemblgenebuild.
GenomeResearch22:2067–78.
49.
FujitaA,SatoJR,DemasiMA,SogayarMC,FerreiraCE,etal.
(2009)ComparingPearson,SpearmanandHoeffding'sDmeasureforgeneexpressionassociationanalysis.
JournalofBioinformaticsandComputationalBiology7:663–84.
50.
OliverER,SaundersTL,TarleSA,GlaserT(2004)RibosomalproteinL24defectinbellyspotandtail(Bst),amouseMinute.
Development(Cambridge,England)131:3907–20.
51.
DraptchinskaiaN,GustavssonP,AnderssonB,PetterssonM,WilligTN,etal.
(1999)ThegeneencodingribosomalproteinS19ismutatedinDiamond-Blackfananaemia.
NatureGenetics21:169–75.
52.
SkarnesWC,RosenB,WestAP,KoutsourakisM,BushellW,etal.
(2011)Aconditionalknockoutresourceforthegenome-widestudyofmousegenefunction.
Nature474:337–42.
53.
HrabedeAngelisMH,FlaswinkelH,FuchsH,RathkolbB,SoewartoD,etal.
(2000)Genome-wide,large-scaleproductionofmutantmicebyENUmutagenesis.
NatureGenetics25:444–7.
54.
TsetskhladzeZR,CanfieldVA,AngKC,WentzelSM,ReidKP,etal.
(2012)Functionalassessmentofhumancodingmutationsaffectingskinpigmentationusingzebrafish.
PLoSONE7:e47398.
55.
DooleyCM,SchwarzH,MuellerK,MongeraA,KonantzM,etal.
(2012)Slc45a2andV-ATPaseareregulatorsofmelanosomalpHhomeostasisinzebrafish,providingamechanismforhumanpigmentevolutionanddisease.
PigmentCell&MelanomaResearchdoi:10.
1111/pcmr.
12053.
56.
FashenaD,WesterfieldM(1999)SecondarymotoneuronaxonslocalizeDM-GRASPontheirfasciculatedsegments.
JCompNeurol406(3):415–24.
57.
TrevarrowB,MarksDL,KimmelCB(1990)Organizationofhindbrainsegmentsinthezebrafishembryo.
Neuron4(5):669–79.
58.
ChoudhryP,JoshiD,FunkeB,TredeN(2011)AlcamamediatesEdn1signalingduringzebrafishcartilagemorphogenesis.
DevelopmentalBiology349(2):483–93.
59.
LaneBM,ListerJA(2012)Otxbutnotmitftranscriptionfactorsarerequiredforzebrafishretinalpigmentepitheliumdevelopment.
PLoSONE7:e49357.
60.
RubinsteinAL,LeeD,LuoR,HenionPD,HalpernME(2000)GenesdependentonzebrafishcyclopsfunctionidentifiedbyAFLPdifferentialgeneexpressionscreen.
Genesis26:86–97.
61.
ZhongX,ZhuY,MaoJ,ZhangJ,ZhengS(2012)FrequentepigeneticsilencingofPCDH10bymethylationinhumancolorectalcancer.
JournalofCancerResearchandClinicalOncologydoi10.
1007/s00432–012–1353–5.
62.
BertrandKC,MackSC,NorthcottPA,GarziaL,DubucA,etal.
(2011)PCDH10isacandidatetumoursuppressorgeneinmedulloblastoma.
Child'sNervousSyst27:1243–9.
63.
LiZ,LiW,XieJ,WangY,TangA,etal.
(2011)EpigeneticinactivationofPCDH10inhumanprostatecancercelllines.
CellBiologyInternational35:671–6.
64.
ThisseC,ThisseB(2005)HighThroughputExpressionAnalysisofZF-ModelsConsortiumClones.
ZFINDirectDataSubmission(http://zfin.
org).
65.
TallafussA,HaleLA,YanY,DudleyL,EisenJS,etal.
(2006)Characterizationofretinoid-Xreceptorgenesrxra,rxrba,rxrbbandrxrgduringzebrafishdevelopment.
GeneExpressionPatterns6:556–65.
66.
WaxmanJS,YelonD(2007)ComparisonoftheexpressionpatternsofnewlyidentifiedzebrafishretinoicacidandretinoidXreceptors.
DevelopmentalDynamics236:587–95.
67.
HeC,WangC,LiB,XieF,ChenY,etal.
(2009)Tissue-specificandembryonicexpressionoftheretinoidXreceptorsinSebastiscusmarmoratus.
ComparativeBiochemistryandPhysiology.
PartB,Biochemistry&MolecularBiology154:221–8.
68.
FlockhartR,WebsterD,QuK,MascarenhasN,KovalskiJ,etal.
(2012)BRAFV600EremodelsthemelanocytetranscriptomeandinducesBANCRtoregulatemelanomacellmigration.
GenomeResearch22:1006–1014.
69.
AnJ,WanH,ZhouX,HuD,WangL,etal.
(2011)AComparativeTranscriptomicAnalysisofUvealMelanomaandNormalUvealMelanocyte.
PLoSONE6:e16516.
70.
TomihariM,HwangS,ChungJ,CruzPD,AriizumiK(2009)Gpnmbisamelanosome-associatedglycoproteinthatcontributestomelanocyte/keratino-cyteadhesioninaRGD-dependentfashion.
ExperimentalDermatology18:586–95.
71.
CurranK,ListerJA,KunkelGR,PrendergastA,ParichyDM,etal.
(2010)InterplaybetweenFoxd3andMitfregulatescellfateplasticityinthezebrafishneuralcrest.
DevelopmentalBiology344:107–118.
72.
DonaldsonJG,HondaA(2005)LocalizationandfunctionofArffamilyGTPases.
BiochemicalSocietyTrans33:639–42.
73.
GrosshansBL,OrtizD,NovickP(2006)Rabsandtheireffectors:achievingspecificityinmembranetraffic.
ProcNatlAcadSciUSA103:11821–7.
74.
BahadoranP,AberdamE,MantouxF,BuscaR,BilleK,etal.
(2001)Rab27a:Akeytomelanosometransportinhumanmelanocytes.
TheJournalofCellBiology152:843–50.
75.
TisdaleEJ,KellyC,ArtalejoCR(2004)Glyceraldehyde-3-phosphatedehydrogenaseinteractswithRab2andplaysanessentialroleinendoplasmicreticulumtoGolgitransportexclusiveofitsglycolyticactivity.
JournalofBiologicalChemistry279:54046–52.
76.
HuangQ,LanF,ZhengZ,XieF,HanJ,etal.
(2011)AKT2suppressesGAPDHmediated-apoptosisinovariancancercellsviaphosphorylatingGAPDHatthreonine237anddecreasingitsnucleartranslocation.
JournalofBiologicalChemistry286:42211–20.
77.
ListerJA,LaneBM,NguyenA,LunneyK(2011)EmbryonicexpressionofzebrafishMiTfamilygenestfe3b,tfeb,andtfec.
DevelopmentalDynamics240:2529–38.
78.
ChungMC,KimHK,KawamotoS(2001)TFECcanfunctionasatranscriptionalactivatorofthenonmusclemyosinIIheavychain-Ageneintransfectedcells.
Biochemistry40:8887–97.
79.
ManskyKC,SulzbacherS,PurdomG,NelsenL,HumeDA,etal.
(2002)Themicrophthalmiatranscriptionfactorandtherelatedhelix-loop-helixzipperfactorsTFE-3andTFE-Ccollaboratetoactivatethetartrate-resistantacidphosphatase.
JournalofLeukocyteBiology71:304–10.
80.
YeungSJ,PanJ,LeeMH(2008)Rolesofp53,MYCandHIF-1inregulatingglycolysis–theseventhhallmarkofcancer.
CellularandMolecularLifeSciences65:3981–99.
81.
AlbajarM,Gomez-CasaresM,LlorcaJ,MauleonI,VaqueJP,etal.
(2011)MYCinchronicmyeloidleukemia:inductionofaberrantDNAsynthesisandassociationwithpoorresponsetoimatinib.
MolecularCancerResearch9:564–76.
82.
LiuY,LiF,HandlerJ,HuangC,XiangY,etal.
(2008)Globalregulationofnucleotidebiosyntheticgenesbyc-Myc.
PLoSONE3:e2722.
83.
VerschoorML,WilsonLA,VerschoorCP,SinghG(2010)Ets-1regulatesenergymetabolismincancercells.
PLoSONE5:e13565.
84.
RehliM,LichanskaA,CassadyAI,OstrowskiMC,HumeDA(1999)TFECisamacrophage-restrictedmemberofthemicrophthalmia-TFEsubfamilyofbasichelix-loop-helixleucinezippertranscriptionfactors.
TheJournalofImmunology162:1559–65.
85.
LiB,KuriyamaS,MorenoM,MayorR(2009)TheposteriorizinggeneGbx2isadirecttargetofWntsignallingandtheearliestfactorinneuralcrestinduction.
Development136:3267–78.
86.
EttensohnCA,IlliesMR,OliveriP,DeJongDL(2003)Alx1,amemberoftheCart1/Alx3/Alx4subfamilyofPaired-classhomeodomainproteins,isanessentialcomponentofthegenenetworkcontrollingskeletogenicfatespecificationintheseaurchinembryo.
Development130:2917–28.
87.
BeverdamA,BrouwerA,ReijnenM,KorvingJ,MeijlinkF(2001)SeverenasalcleftingandabnormalembryonicapoptosisinAlx3/Alx4doublemutantmice.
Development128:3975–86.
88.
MavrogiannisLA,AntonopoulouI,BaxovaA,KutlekS,KimCA,etal.
(2001)HaploinsufficiencyofthehumanhomeoboxgeneALX4causesskullossificationdefects.
NatureGenetics27:17–8.
TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org15July2013|Volume8|Issue7|e6780189.
WuytsW,CleirenE,HomfrayT,Rasore-QuartinoA,VanhoenackerF,etal.
(2000)TheALX4homeoboxgeneismutatedinpatientswithossificationdefectsoftheskull(foraminaparietaliapermagna,OMIM168500).
JournalofMedicalGenetics37:916–20.
90.
CmejlaR,CmejlovaJ,HandrkovaH,PetrakJ,PospisilovaD(2007)RibosomalproteinS17gene(RPS17)ismutatedinDiamond-Blackfananemia.
HumanMutation28:1178–82.
91.
GuY,FanS,LiuB,ZhengG,YuY,etal.
(2011)TCRP1promotesradioresistanceoforalsquamouscellcarcinomacellsviaAktsignalpathway.
MolecularandCellularBiochemistry357:107–13.
92.
WuSM,TsaiPR,YanCJ(2012)Maternalcadmiumexposureinducesmt2andsmtBmRNAexpressioninzebrafish(Daniorerio)femalesandtheiroffspring.
ComparativeBiochemistryandPhysiology.
Toxicology&Pharmacology156:1–6.
93.
KitajiriS,SakamotoT,BelyantsevaIA,GoodyearRJ,StepanyanR,etal.
(2010)Actin-BundlingProteinTRIOBPFormsResilientRootletsofHairCellStereociliaEssentialforHearing.
Cell141:786–798.
94.
FujitaH,KatohH,IshikawaY,MoriK,NegishiM(2002)RapostlinisanoveleffectorofRnd2GTPaseinducingneuritebranching.
TheJournalofBiologicalChemistry277:45428–34.
95.
SchonthalerHB,LampertJM,vonLintigJ,SchwarzH,GeislerR,etal.
(2005)Amutationinthesilvergeneleadstodefectsinmelanosomebiogenesisandalterationsinthevisualsysteminthezebrafishmutantfadingvision.
DevelopmentalBiology284:421–36.
96.
ThisseB,ThisseC(2004)FastReleaseClones:AHighThroughputExpressionAnalysis.
ZFINDirectDataSubmission(http://zfin.
org).
97.
KennedyBN,StearnsGW,SmythVA,RamamurthyV,vanEedenF,etal.
(2004)Zebrafishrx3andmab21l2arerequiredduringeyemorphogenesis.
DevelopmentalBiology270:336–49.
98.
RinchikEM,BultmanSJ,HorsthemkeB,LeeST,StrunkKM,etal.
(1993)Ageneforthemousepink-eyeddilutionlocusandforhumantypeIIoculocutaneousalbinism.
Nature361:72–6.
99.
LamasonRL,MohideenMA,MestJR,WongAC,NortonHL,etal.
(2005)SLC24A5,aputativecationexchanger,affectspigmentationinzebrafishandhumans.
Science310:1782–6.
100.
TamWY,LeungCK,TongKK,KwanKM(2011)Foxp4isessentialinmaintenanceofPurkinjecelldendriticarborizationinthemousecerebellum.
Neuroscience172:562–71.
101.
NetterS,FauvarqueMO,delCorralRD,DuraJM,CoenD(1998)white+transgeneinsertionspresentingadorsal/ventralpatterndefineasingleclusterofhomeoboxgenesthatissilencedbythepolycomb-groupproteinsinDrosophilamelanogaster.
Genetics149:257–75.
102.
LarsonTA,GordonTN,LauHE,ParichyDM(2010)DefectiveadultoligodendrocyteandSchwanncelldevelopment,pigmentpattern,andcraniofacialmorphologyinpumamutantzebrafishhavinganalphatubulinmutation.
DevelopmentalBiology346:296–309.
103.
FumotoS,HiyamaK,TanimotoK,NoguchiT,HiharaJ,etal.
(2009)EMP3asatumorsuppressorgeneforesophagealsquamouscellcarcinoma.
CancerLetters274:25–32.
104.
Serra-Page`sC,MedleyQG,TangM,HartA,StreuliM(1998)Liprins,afamilyofLARtransmembraneprotein-tyrosinephosphatase-interactingproteins.
TheJournalofBiologicalChemistry273:15611–20.
105.
ChenY,SamaraweeraP,SunT,KreibichG,OrlowSJ(2002)Rab27bassociationwithmelanosomes:dominantnegativemutantsdisruptmelano-somalmovement.
TheJournalofinvestigativedermatology118:933–40.
106.
MaL,HuangY,SongZ,FengS,TianX,etal.
(2006)LivinpromotesSmac/DIABLOdegradationbyubiquitin-proteasomepathway.
CellDeathandDifferentiation13:2079–88.
107.
RauschMP,IrvineKR,AntonyPA,RestifoNP,CresswellP,etal.
(2010)GILTacceleratesautoimmunitytothemelanomaantigentyrosinase-relatedprotein1.
JournalofImmunology185:2828–35.
108.
CoghillID,BrownS,CottleDL,McGrathMJ,RobinsonPA,etal.
(2003)FHL3isanactin-bindingproteinthatregulatesalpha-actinin-mediatedactinbundling:FHL3localizestoactinstressfibersandenhancescellspreadingandstressfiberdisassembly.
TheJournalofBiologicalChemistry278:24139–52.
109.
YamamotoS,InoueK,MurataT,KamigasoS,YasujimaT,etal.
(2010)Identificationandfunctionalcharacterizationofthefirstnucleobasetransporterinmammals:implicationinthespeciesdifferenceintheintestinalabsorptionmechanismofnucleobasesandtheiranalogsbetweenhigherprimatesandothermammals.
TheJournalofBiologicalChemistry285:6522–31.
110.
ThomasDD,MartinCL,WengN,ByrneJA,GroblewskiGE(2010)TumorproteinD52expressionandCa2+-dependentphosphorylationmodulateslysosomalmembraneproteintraffickingtotheplasmamembrane.
AmericanJournalofPhysiology.
CellPhysiology298:C725–39.
111.
SchollerM,WadsackC,MetsoJ,ManavalanAP,SreckovicI,etal.
(2012)PhospholipidtransferproteinisdifferentiallyexpressedinhumanarterialandvenousplacentalendothelialcellsandenhancescholesteroleffluxtofetalHDL.
TheJournalofClinicalEndocrinologyandMetabolism97:2466–74.
112.
DepazIM,WilcePA(2006)Thenovelcytoskeleton-associatedproteinNeuronalprotein22:elevatedexpressioninthedevelopingratbrain.
BrainResearch1081:59–64.
TranscriptomeAnalysisofZebrafishPigmentCellsPLOSONE|www.
plosone.
org16July2013|Volume8|Issue7|e67801