Whole-exome SNP array identifies 15 new susceptibility loci for psoriasis

ARTICLE

Received27Sep2014|Accepted28Feb2015|Published9Apr2015DOI:10.1038/ncomms7793OPEN

Whole-exomeSNParrayidenti?es15new

susceptibilitylociforpsoriasis

XianboZuo1,2,3,4,5,6,7,*,LiangdanSun1,2,3,4,5,6,7,*,XianyongYin1,2,3,4,5,6,7,*,JinpingGao1,4,5,6,7,*,YujunSheng1,4,5,6,7,*,JinhuaXu2,4,JianzhongZhang8,ChundiHe9,YingQiu10,GuangdongWen8,HongqingTian11,

XiaodongZheng1,4,5,6,7,ShengxiuLiu1,4,5,6,7,WenjunWang1,4,5,6,7,WeiranLi1,4,5,6,7,YuyanCheng1,4,5,6,7,LongdanLiu1,4,5,6,7,YanChang1,4,5,6,7,ZaixingWang1,4,5,6,7,ZenggangLi1,4,5,6,7,LongnianLi1,4,5,6,7,

JianpingWu1,4,5,6,7,LingFang1,4,5,6,7,ChangbingShen1,4,5,6,7,FushengZhou1,4,5,6,7,BoLiang1,4,5,6,7,GangChen1,4,5,6,7,HuiLi1,4,5,6,7,YongCui4,5,6,7,AieXu12,XueqinYang13,FeiHao14,LiminXu15,XingFan1,4,5,6,7,

YuzhenLi16,RinaWu17,XiuliWang18,XiaomingLiu19,MinZheng20,ShunpengSong21,BihuaJi22,HongFang23,JianbinYu24,YongxinSun25,YanHui26,FurenZhang11,RongyaYang27,SenYang1,4,5,6,7&XuejunZhang1,2,3,4,5,6,7Genome-wideassociationstudies(GWASs)havereproduciblyassociatedB40susceptibility

lociwithpsoriasis.However,themissingheritabilityisevidentandthecontributionsofcoding

variantshavenotyetbeensystematicallyevaluated.Here,wepresentalarge-scalewhole-

exomearrayanalysisforpsoriasisconsistingof42,760individuals.Wediscover16SNPs

within15newgenes/lociassociatedwithpsoriasis,includingC1orf141,ZNF683,TMC6,

AIM2,IL1RL1,CASR,SON,ZFYVE16,MTHFR,CCDC129,ZNF143,AP5B1,SYNE2,IFNGR2and

3q26.2-q27(Po5.00Â10À08).Inaddition,wealsoreplicatefourknownsusceptibilityloci

TNIP1,NFKBIA,IL12BandLCE3D–LCE3E.Thesesusceptibilityvariantsidenti?edinthecurrent

studycollectivelyaccountfor1.9%ofthepsoriasisheritability.ThevariantwithinAIM2is

predictedtoimpactproteinstructure.Our?ndingsincreasethenumberofgeneticriskfactors

forpsoriasisandhighlightnewandplausiblebiologicalpathwaysinpsoriasis.

ofDermatologyandDepartmentofDermatology,No.1Hospital,AnhuiMedicalUniversity,Hefei,Anhui230022,China.2Departmentof

Dermatology,HuashanHospital,FudanUniversity,Shanghai200040,China.3DepartmentofDermatology,No.2Hospital,AnhuiMedicalUniversity,Hefei,Anhui230022,China.4CollaborativeInnovationCenterofComplexandSevereSkinDisease,AnhuiMedicalUniversity,Hefei,Anhui230032,China.5StateKeyLabIncubationofDermatology,MinistryofScienceandTechnology,Hefei,Anhui230032,China.6KeyLabofDermatology,MinistryofEducation,Hefei,Anhui230032,China.7KeyLabofGeneResourcesUtilizationforSevereInheritedDisorders,Anhui230032,China.8DepartmentofDermatology,PekingUniversityPeople’sHospital,Beijing100044,China.9DepartmentofDermatology,No.1HospitalofChinaMedicalUniversity,Shenyang,Liaoning110001,China.10DepartmentofDermatology,JiningNo.1People’sHospital,Jining,Shandong272011,China.11ShandongProvincialInstituteofDermatologyandVenereology,Jinan,Shandong250022,China.12TheThirdPeople’sHospitalofHangzhou,Hangzhou,Zhejiang310009,China.13DepartmentofDermatology,GeneralHospitalofPLAAirForce,Beijing100036,China.14DepartmentofDermatology,SouthwestHospital,ThirdMilitaryMedical

University,Chongqing400038,China.15DepartmentofDermatology,TianjinChangzhengHospital,Tianjin300106,China.16DepartmentofDermatology,SecondAf?liatedHospitalofHarbinMedicalUniversity,Harbin,Heilongjiang150000,China.17DepartmentofDermatology,TheAf?liatedHospitalofInnerMongoliaMedicalCollege,Huhehot,InnerMongolia010050,China.18ShanghaiSkinDiseasesandSTDHospital,Shanghai200050,China.19DepartmentofDermatology,TheFirstAf?liatedHospitalofDalianMedicalUniversity,Dalian,Liaoning116011,China.20DepartmentofDermatology,TheSecondAf?liatedHospital,ZhejiangUniversitySchoolofMedicine,Zhenjiang310009,China.21DepartmentofDermatology,DalianDermatosisHosptial,Liaoning116011,China.22DepartmentofDermatology,YijishanHospitalofWannanMedicalCollege,Wuhu,Anhui241000,China.23DepartmentofDermatology,TheFirstAf?liatedHospitalofZhejiangUniversitySchoolofMedicine,Zhenjiang310006,China.24DepartmentofDermatology,TheFirstAf?liatedHospitalofZhengzhouUniversity,Zhengzhou,Henan450052,China.25DepartmentofDermatology,AnshanTanggangzihosptial,Liaoning210300,China.26DepartmentofDermatology,FirstAf?liatedHospitalofXinjiangMedicalUniversity,Xinjiang830054,China.27DepartmentofDermatology,GeneralHospitalofBeijingMilitaryCommand,Beijing100010,China.*Theseauthorscontributedequallytothiswork.CorrespondenceandrequestsformaterialsshouldbeaddressedtoX.Z.(email:ayzxj@http://wendang.chazidian.com).

NATURECOMMUNICATIONS|6:6793|DOI:10.1038/ncomms7793|http://wendang.chazidian.com/naturecommunications1Institute1

&2015MacmillanPublishersLimited.Allrightsreserved.

ARTICLE

soriasisisachronicin?ammatoryhyperproliferativecutaneousdiseasewithdynamicinteractionsbetweentheimmunesystemandtheepidermisthataffectsupto3%ofthepopulationworldwide1–3.Morethan40susceptibilitygenesorlocihavebeenidenti?edforpsoriasisindiversepopulations,mostlythroughgenome-wideassociationstudies(GWASs)4–17.However,eachoftheseidenti?edgenesorlocihasasmallormoderateeffect,whichonlycollectivelyexplainasmallproportionofthegeneticvariationinpsoriasis.The‘missingheritability’inpsoriasisisevident18.Furthermore,mostofthesepreviouslyidenti?edvariantsarelocatedinnon-codinggenomicregions19andthusprovidefewcluesastothefunctionalmechanismthroughwhichthesevariantsaffectthedisease.Codingvariantswithhighpenetrance,whichwerepoorlycoveredinconventionalGWASsmaycontributeto?ndingthe‘missingheritability’incomplexdisorders20–22.Recenttechnologicaladvancesinhigh-throughputsequen-cing23provideanopportunitytoresequencemultiplegeneticregionsandhavegeneratedcompellingevidencethatcodingvariantscontributetothemechanismsofpsoriasis5,24andothercomplexdisorders25–34.However,theseeffortstoinvestigatecodingvariantsarestilllimitedduetosamplesizeandthusthestatisticalpower.Studiesusingnewexomechipsshowtheirabilitytocomprehensivelyidentifycodingvariantsforseveralcomplextraits34–36.Here,weperformthe?rstexome-wideassociationstudyinlarge-scaleindividuals(17,614casesand25,216controls)tosystematicallyinvestigatethecodingvariantsinpsoriasisbyusingIlluminaHumanExomeAsianBeadChip(Exome_AsianArray)andIlluminaHumanExomeFineMappingBeadChip(Exome_-FineArray).Athree-stagecase–controldesignwasimplementedinthepresentstudy,and15newgenes/lociwereidenti?edtoassociatewithpsoriasis.Wealsoreplicatefourknownsuscept-ibilitylociandallthese23susceptibilityvariantsidenti?edincurrentstudycollectivelyaccountfor1.9%ofthepsoriasisheritability.These?ndingsnotonlyincreasethenumberofgeneticriskfactorsforpsoriasis,butalsohighlightnewandplausiblebiologicalpathwaysinvolvedinthisdisease.

Results

Exome_AsianArrayandExome_FineArray.Inthe?rsttwostages,morethan260,000markersweregenotypedintwocohorts(SupplementaryTables1and2),including8,949individuals(4,179casesand4,770controls)and13,473individuals(7,066casesand6,407controls),byusingExome_AsianArrayandExome_FineArray,respectively.Afterqualitycontrol?lteringandprincipalcomponentanalysis(PCA,onlineMethods;SupplementaryFig.1),108,576and108,099variantswerequa-li?edintheExome_AsianandExome_FineArrays,respectively.In11,245casesand11,177controls,87,827non-MHCvariants(77,641nonsynonymous,nonsenseorsplice-sitevariants)werevariable(SupplementaryFig.2).Aquantile–quantileplotandManhattanplotweregeneratedusingCochran–Armitagetestfortrend(SupplementaryFigs3and4).Acleardeviationfromtheexpectednulldistributionwasobservedinthequantile–quantileplot(SupplementaryFig.3).Tovalidatethequalityofourapproach,wetestedforassociationatknownGWASlociinExome_AsianArrayandExome_FineArraystages,respectively.Amongthe88previouslydescribedpsoriasis-associatedSNPs(Po5.00Â10À8),39SNPsweredirectlygenotypedontheExome_Asianarraywith24SNPspassthequalitycontrols.Amongthese24SNPsexamined,weidenti?ed16SNPswithsigni?cantornominalassociation(Po0.05,SupplementaryTable3)throughlogisticregression(additivemodel).AsforExome_FineArraystage,38SNPswere

2NATURECOMMUNICATIONS|DOI:10.1038/ncomms7793Pincludedand23oftheSNPspassthequalitycontrols.Finally,16SNPswerereplicated(Po0.05)inthisstage(SupplementaryTable3).TheseresultssuggestedthatthegenotypedatafrombothofExome_AsianArrayandExome_FineArraywerehighreliabilityforourdownstreamanalyses.Meta-analysisofthe?rsttwostages.Weperformedameta-analysisofthe?rsttwostagesonthe87,827sharedvariantswithinthenon-HLAregionandidenti?ed?venovelsusceptibilitylocithroughlogisticregression(additivemodel),includingMTHFR(rs2274976,P¼2.33Â10À10,oddsratio(OR)¼0.79),IL1RL1(rs1420101,P¼1.71Â10À10,OR¼0.88),ZNF143(rs10743108,P¼1.70Â10À8,OR¼1.14),ZNF683(rs10794532,P¼4.18Â10À8,OR¼1.11),andTMC6(rs12449858,P¼2.28Â10À8,OR¼1.12)genesatgenome-wideassociationsig-ni?canceanddiscoveredfournewvariantswithinthreeknownsusceptibilityloci,includingLCE3D–LCE3E(rs10888501,P¼6.48Â10À13,OR¼0.86),IL12B(rs1473247,P¼5.63Â10À11,OR¼0.88;rs10076782,P¼4.11Â10À11,OR¼0.88)andNFKBIA(rs12884468,P¼1.05Â10À8,OR¼0.88)(Table1andSupplementaryFigs5and6).Weperformedconditionalandlinkagedisequilibrium(LD)analysestoevaluatewhetherthesesigni?cantlyassociatedvariantswereindependentoftheestablishedGWAS-identi?edSNPsintheChinesepopulation.Conditionalanalysisprovidednoevidenceofassociationforthesethreelocicontainingsecondarysignals(SupplementaryTable4).Forlow-frequencyvariants(0.01oMAFo0.05),wedidnotidentifyanyofthemreachinggenome-widesigni?cancethroughsingle-variantanalysis(SupplementaryData1).Genotypingvalidation.Toevaluateadditionalsusceptibilitygeneticfactors,weselectedthetop76SNPswith5.00Â10À8oPmetao1.00Â10À4(Table2andSupplementaryData2)forfurthergenotypinginanindependentreplicationcohortof6,369casesand13,969controls(SupplementaryTable1).Meta-analysisofthe76SNPsinthediscovery(Exome_AsianArraystageandExome_FineArraystage)andreplicationstagestudiesidenti?ed11variantslocatedin11diversenewsusceptibilitygenesinthenon-HLAregionthroughlogisticregression(additivemodel),includingNPPA(rs5063,P¼3.51Â10À9,OR¼0.85),C1orf141(rs72933970,P¼1.23Â10À8,OR¼1.16),AIM2(rs2276405,P¼3.22Â10À9,OR¼0.83),CASR(rs1042636,P¼1.88Â10À10,OR¼0.91),GPR160(rs6444895,P¼1.44Â10À12,OR¼1.11),ZFYVE16(rs249038,P¼2.14Â10À8,OR¼0.84),CCDC129(rs4141001,P¼1.84Â10À11,OR¼1.11),AP5B1(rs610037,P¼4.29Â10À11,OR¼1.11),SYNE2(rs2781377,P¼4.21Â10À11,OR¼0.85),IFNGR2(rs9808753,P¼2.75Â10À8,OR¼0.92)andSON(rs3174808,P¼1.15Â10À8,OR¼1.10;Table2andSupplementaryFigs5and6).Notably,twosusceptibilitycodingvariantswereidenti?edat1p36andlocatedinNPPA(rs5063)andMTHFR(rs2274976),respectively.LDanalysisrevealedthatthesetwovariantsareinamoderateLD(D0¼0.70,r2¼0.49)andfurtherconditionalana-lysisindicatedthattheyarenotindependentsignalsfromeachother(rs5063,Pcondition¼2.83Â10À1,OR¼0.94;rs2274976,Pcondition¼2.60Â10À5,OR¼0.80;SupplementaryTable4).Similarly,wealsoidenti?edtwosusceptibilitycodingvariantsat21q22.11,whichlocatedinIFNGR2(rs9808753)andSON(rs3174808),respectively.ConditionalandLDanalysesshowedthatthesetwovariantsareinverymildLD(D0¼0.76,r2¼0.15)andindependentfromeachother(rs9808753,Pcondition¼1.22Â10À4,OR¼0.94;rs3174808,Pcondition¼9.03Â10À4,OR¼1.06;SupplementaryTable4).Inaddition,weidenti?edamissensevariant(rs72933970)withinC1orf141.Atthisregion,multiplevariantsinornearIL23Rhavebeenidenti?edtobeNATURECOMMUNICATIONS|6:6793|DOI:10.1038/ncomms7793|http://wendang.chazidian.com/naturecommunications

&2015MacmillanPublishersLimited.Allrightsreserved.

NATURECOMMUNICATIONS|DOI:10.1038/ncomms7793ARTICLETable1|Associationresultsfromthe?rsttwostages(Exome_AsianArrayandExome_FineArray)throughlogisticregression(additivemodel).

PP

*Thesesusceptibilitylocihavebeenreportedpreviously.Table2|Associationresultsfromeachofthethreestagesandcombinedanalysesthroughlogisticregression(additivemodel).*Thesesusceptibilitylocihavebeenreportedpreviously.

associatedwithpsoriasis.Torevealtherelationshipbetweenrs72933970andreportedvariants,weperformedconditionalandLDanalysesandindicatedthatrs72933970isanindependentsignalatthisregion(SupplementaryTable4).Wealsoconformedthreeknowngenes,suchasLCE3D–LCE3E(rs41268474,P¼5.99Â10À11,OR¼1.17;rs76337351,P¼1.71Â10À8,OR¼0.83),andTNIP1(rs10036748,P¼4.26Â10À9,OR¼1.10;Table2andSupplementaryFigs5and6).ConditionalandLDanalyseswerecarriedouttoevaluatewhethertheseSNPswereindependentsignalsfromtheestablishedGWAS-identi?edSNPs.Onlyonemakerhadlimitedimpactontheassociationsatrs10036748(Pcondition¼2.04Â10À3,OR¼1.07)in(SupplementaryTable4).Furthermore,threesuggestiveloci1q42.3,10q22.3and21q22.11werealsoidenti?edwithPmetao1.00Â10À06.DiscussionInthepresentstudy,weidenti?ed23SNPswithin19genes/lociassociatedwithpsoriasis,including16codingvariantsand7non-codingvariants.Fifteenof19arenewlyidenti?edgeneticriskgenes/loci,includingC1orf141,ZNF683,TMC6,AIM2,IL1RL1,CASR,SON,ZFYVE16,CCDC129,MTHFR,ZNF143,AP5B1,SYNE2,IFNGR2and3q26.2-q27,andremainingfourare3NATURECOMMUNICATIONS|6:6793|DOI:10.1038/ncomms7793|http://wendang.chazidian.com/naturecommunications

&2015MacmillanPublishersLimited.Allrightsreserved.

ARTICLE

previouslyreportedloci,suchasTNIP1,NFKBIA,IL12BandLCE3D–LCE3E.These?ndingsprovideconvincingevidencethatcommongeneticvariationisanimportantcontributortotheriskofpsoriasis.Allthesenewlyidenti?edlociaremini-effectones,whichneedslarge-scalesamplestobedetected.Inthisstudy,17,614casesand25,216controlsareinvolved.Although11of26newlyidenti?edSNPs(Tables1and2)arecoveredinIlluminaHuman610-QuadBeadChipsusedinourpreviousstudy16,becauseofthelimitedsamplesizes(1,139casesand1,694controlsinourpreviousGWASandthecurrentstudyhasmuchbiggersamplesizeandthuspowerthanpreviousstudies),thePvaluearenotsigni?cantenough(SupplementaryTable5).Therefore,these11SNPswerenotchosentovalidateinpreviousstudies.For1p36,weidenti?edtwomissensevariantswithinNPPA(rs5063)andMTHFR(rs2274976).Conditionanalysisindicatedthattheyrepresentthesamesignalandrs2274976ismoresigni?cantthanrs5063(SupplementaryTable4).MTHFRencodesaproteinthatactsasaco-substrateforhomocysteineremethylationtomethionine,whichisimportantformaintainingthemethyldonorsforDNAmethylation,thusresultingingeneregulationandcellulardifferentiation37.ForAIM2andSYNE2,weidenti?edastop-gainedvariantateachgene,includingrs2276405(AIM2)andrs2781377(SYNE2).AIM2encodesacytosolicdouble-strandedDNA(dsDNA)receptor.Thisreceptorinteractswithapoptosisspeck-likeproteintoformacaspase-1-activatingin?ammasomeandplaysaputativeroleintumorigenicreversionandcontrollingcellproliferation38.TheproteinencodedbySYNE2isanuclearoutermembraneproteinthatbindscytoplasmicF-actinandisanovelnesprinisoformthatisexpressedinskin39.Apreviousstudydemonstratedthatthedepletionofnesprin-2reducesboththeamountofactiveb-catenininsidethenucleusandT-cellfactor/lymphoid-enhancingfactor-dependenttranscription40.At2q12.1,weidenti?edavariant(rs1420101)withinintron3ofIL1RL1.Theproteinencodedbythisgeneisamemberoftheinterleukin1receptorfamily,whichhavebeenproventobeinvolvedinthefunctionofhelperTcellsandthisreceptorcanbeinducedbyproin?ammatorystimuli41.At3q13,wediscoveredamissensevariantatCASR.Itencodesanendogenouscalcium-sensingreceptor(CaR),whichisessentialformediatingCa(2þ)signallingduringCa(2þ)(o)-induceddifferentiation42.ExtracellularCa(2þ)(Ca(2þ)(o))isacriticalregulatorthatpromotesdifferentiationinepidermalkeratinocytes.Thetransforminggrowthfactorbetasignallingpathwayisnecessaryforavarietyofnormalcellularprocesses43.ForZNF683,ZNF143andZFYVE16,weidenti?edamissensevariantoneachgene,includingrs10743108(ZNF143),rs10794532(ZNF683)andrs9808753(ZFYVE16).Ithasbeenimplicatedthattranscriptionalregulatoryproteinscontainingtandemlyrepeatedzinc?ngerdomainsarethoughttobeinvolvedinbothnormalandabnormalcellularproliferationanddifferentiation44,45.TheZFYVE16geneencodesanendosomeproteinthatbelongstotheFYVEzinc?ngerfamilyofproteins.Theencodedproteinfunctionsasascaffoldproteininthetransforminggrowthfactorbetasignallingpathwayandisinvolvedinpositiveandnegativefeedbackregulationofthebonemorphogeneticproteinsignallingpathway46.Inaddition,wealsodiscoveramissensevariantrs12449858inTMC6,whichencodesamemberoftheEVERproteinfamily,whichmaybeinvolvedintheregulationofcellularzinchomeostasisinlymphocytesandwhichhavebeenconsideredaskeycomponentsoftheactivation-dependentregulationofZn(2þ)concentrationinTcells47.At21q22.11,weidenti?edtwosusceptibilitycodingvariants,whichwerelocatedinIFNGR2(rs9808753)andSON(rs3174808),

4NATURECOMMUNICATIONS|DOI:10.1038/ncomms7793respectively.ConditionalandLDanalysesshowedthatthesetwovariantsareindependentfromeachother.IFNGR2encodesthenon-ligand-bindingbetachainofthegammainterferonreceptor,whichmayaffectthegenerationofTh17cellsfrommemoryTcells48,49.However,thebiologicalfunctionofSONisunclear.Inaddition,fournewsusceptibilitygenes(C1orf141,GPR160,CCDC129andAP5B1)withunknownfunctionsinthepathogenesisofpsoriasiswerealsoidenti?ed,indicatingthatadditionalmolecularmechanismscontributetotheriskofdevelopingpsoriasis.Therefore,furtherstudiesarerequiredtofullyunderstandhowvariationsinthesegenesareinvolvedinthepathogenesisofpsoriasis.Therestrictedmaximum-likelihoodmethod18indicatedthatthesusceptibilityvariantsidenti?edinthisstudytogetherexplained1.9%ofthevarianceinpsoriasisheritability.FifteenofthemarenonsynonymousvariantsandsevenwerepredictedtobedamagingbyeitherSIFTorPolyPhensoftware(SupplementaryTable6).Inaddition,aminoacidresidue32ofAIM2(rs2276405)wasfoundtobelocatedinthemiddleofanalpha-helixmotifburiedinsidetheproteinstructure(SupplementaryFig.7).Thewild-typeGluresidueisacidic,butthemutantLysresidueisbasic.AsthechemicalpropertiesofGluandLysarecompletelyopposite,thissubstitutionmaydestabilizethealpha-helixmotif.Fiveofthese?fteennewlyidenti?ednon-HLAgeneswereshowntobesigni?cantlyenrichedinthenetwork(MTHFR,NPPA,AIM2,CASRandIFNGR2;FDRo0.1)(SupplementaryFig.8).Geneexpressionanalysisonthebasisofpublicpsoriasisdatabases38indicatedthatmostofthesusceptibilitygenes,newlyidenti?edorcon?rmedinthisstudy,displayhighlydifferentialexpressioninskinfrompsoriaticpatientscomparedwithnormalcontrols(SupplementaryTable7).WealsoperformedfunctionalannotationsonthebasisoftheENCODEdatasetforthese26newlyidenti?edSNPs(SupplementaryData3),andshowedthatmostoftheseSNPsfallwithinpromoters,enhancers,Dnasehypersensitivesitesandtranscriptionfactorbindingsites.Thisstudywasdesignedtomaximizestatisticalpowerinacost-effectivemannerbyadoptingamulti-stageanalysisstrategyforalarge-scaleHanChinesepopulationresultingintheidenti?cationof15newsusceptibilitygenes/lociforpsoriasis.Our?ndingshighlightthegeneticcontributionsofcommonvariantstothepathogenesisofpsoriasisandincreasethenumberofknowngeneticriskfactorsforpsoriasis.Thisstudyalsohighlightsnewandplausiblebiologicalpathwaysinpsoriasis,therebysuggestingadditionalgeneticfactorsthatmaycontributetothegeneticheterogeneityofpsoriasisintheHanChinesepopulation.Furtherstudywillbeneededtounderstandthemolecularmechanismsunderlyingtheseriskvariantsidenti?edinthisstudyintheaetiologyofpsoriasis.MethodsStudydesignandstudysamples.Weimplementedathree-stagecase–controldesigninthisstudy(SupplementaryFig.2).Thesubjects,consistingof17,614psoriaticcasesand25,146healthycontrols,wereenrolledthroughacollaborativeconsortiuminChina(SupplementaryTable1).Allthecaseswerediagnosedbyatleasttwodermatologists,andtheirclinicalinformationwascollectedthroughacomprehensiveclinicalcheck-upbyprofessionalinvestigators.Inaddition,demographicinformationwascollectedfromalltheparticipantsthroughapre-viouslydescribedstructuredquestionnaire16.Allthehealthycontrolswereclinicallydeterminedtobewithoutpsoriasis,anyautoimmunedisordersandsystemicdisordersoranyfamilyhistoryofpsoriasisandotherautoimmune-relateddisorders(including?rst-,second-andthird-degreerelatives).Casesandcontrolswerewellmatchedforbothageandsex.Allsampleswereself-reportedHanChinese.Written,informedconsentwasgivenbyalltheparticipants.Thestudywasapprovedbytheinstitutionalethicscommitteeofeachhospital(TheSecondHospitalofAnhuiMedicalUniversity,TheFirstAf?liatedHospitalofAnhuiMedicalUniversityandHuashanHospitalofFudanUniversity)andwasconductedaccordingtotheDeclarationofHelsinkiprinciples.NATURECOMMUNICATIONS|6:6793|DOI:10.1038/ncomms7793|http://wendang.chazidian.com/naturecommunications

&2015MacmillanPublishersLimited.Allrightsreserved.

NATURECOMMUNICATIONS|DOI:10.1038/ncomms7793

ARTICLE

Statisticalanalyses.Single-variantanalysis.Single-markerassociationanalyseswereperformedtotestfordisease–SNPassociations,assuminganadditivealleliceffectandusinglogisticregressionineachstage.TheCochran–Armitagetrendtestwasconductedinthesetwo-stagesamples.Weperformedheterogeneitytests(I2andPvaluesoftheQstatistics)betweenthetwogroupsusingtheBreslow–Daytest52,andtheextentofheterogeneitywasassessedusingtheI2index53.To

improvethestatisticalpower,wecombinedtheassociationresultsinthe?rsttwostagesusingmeta-analysis.The?xedeffectmodel(Mantel–Haenszel)wasappliedwhenI2waso30%(ref.54).Otherwise,therandomeffectmodel(DerSimonian–Laird)wasimplemented55.

Conditionalanalysis.Wecarriedoutconditionalanalysestoidentifyadditionalassociationsignalsafteraccountingfortheeffectsofknownandnewlydiscoveredsusceptibilityloci.Toinvestigatemorethantwoassociationsignalsperlocus,weusedastepwiseprocedureinwhichadditionalSNPswereaddedtothemodelaccordingtotheirconditionalPvalue,asprogrammedinEMMAX56.We

estimatedtheLDmetricsr2andD’using9,633individualsfromMETSIMwhopassedgenotypingqualitycontrol.LDwithSNPsnotincludedontheexomearraywasdeterminedonthebasisofexomesequenceortargetedsequencedatafor21,309HanChineseindividuals.

Annotation.Annovar57wasusedtofunctionallyannotatetheSNPsaccordingtotheirlocationandtheirexpectedeffectonencodedgeneproductsonthebasisofinformationfromtheRefSeqdatabase.

Proteinstructureanalysis.Wesearchedforpublishedthree-dimensionalproteinstructuresintheResearchCollaboratoryforStructuralBioinformaticsProteinDataBank(RCSBPDB;http://wendang.chazidian.com/pdb/home/home.do)anddownloadedstructureforAIM2(3VD8).WeusedDeepView/Swiss-PdbViewer(http://wendang.chazidian.com/spdbv/)toviewtheproteinstructuresandtoexaminethesidechainsoftheoriginalandmutantresiduesattherelevantaminoacids.SIFT58wasusedtopredictthedamageevolutionandprogressionforassociatednonsynonymousvariants.

Statisticalanalysisofnetworks.Toidentifytheproximalinteractors,weexpandedtheglobalnetworkbyincludingtheHumanNetproteininteractiondatabase59andliterature-curatedinteractionsfromSTRING60,61toderiveanexpandedglobalnetworkbasedonknownprotein–proteininteractionsusingthepreviouslypublishedcandidategene-basedandGWAS-baseddata.

Exomearrayandgenotypingin?rsttwostages.Inthisstudy,weusedtwoexomearraytypesasfollows:customIlluminaHumanExomeAsianBeadChip(Exome_AsianArray)andIlluminaHumanExomeFineMappingBeadChip(Exome_FineArray).Theformerplatformincludes242,102markersfocusedonputativefunctionalcodingvariantsfrom412,000exomeandgenomesequencesrepresentingmultipleethnicitiesandcomplextraitsinadditionto430,000Chi-nesepopulation-speci?ccodingvariants,identi?edbywhole-exomesequencingperformedin781psoriasiscasesand676controlsbyourgroup5.Thelatter

includesallmarkersintheIlluminaHumanExome-12v1_ABeadChipand28,139codingvariantsin185susceptibilitygenes,whichhavebeenreportedinimmune-relateddiseaseGWASs.ThedetailsoftheSNPcontentandselectionstrategiesaredescribedontheexomearraydesignwebpage(http://genome.sph.umich.edu/wiki/Exome_Chip_Design).

Inthisstudy,twocohorts,including8,949samples(4,179casesand4,770controls)and13,473samples(7,066casesand6,407controls),inadditionto100blindduplicatesamples,weregenotypedusingtheExome_AsianArrayandExome_FineArray,respectively.ThegenotypingwasconductedattheStateKeyLabIncubationBaseofDermatology,MinistryofNationalScienceandTechnology(AnhuiMedicalUniversity).ThegenotypecallingandtheclusteringofstudysamplegenotypeswereperformedusingIllumina’sGenTrain(version1.0)clusteringalgorithminGenomeStudio(version2011.1).

Qualitycontrols.Weexcluded204samples(90casesand114controls)withgenotypingcallrateso98%inindividualsduringthe?rsttwostages.WethenexaminedpotentialgeneticrelatednessonthebasisofpairwiseidentitybystateforallthesuccessfullygenotypedsamplesusingPLINK1.07software50.Onthe

identi?cationofa?rst-orsecond-degreerelativepair,weremovedoneofthetworelatedindividuals(thesamplewiththelowercallratewasremoved).Wede?nedcloserelativesasthoseforwhomtheestimatedgenome-wideidentity-by-descentproportionofallelessharedwas40.10.Intotal,87samples(33casesand54controls)wereremovedduetosampleduplicationandgeneticrelatedness.Theremainingsamplesweresubsequentlyassessedforpopulationoutliersand

strati?cationusingaPCA-basedapproach51.ForallPCA,allHLASNPsonchr.6:25–34MbandSNPsonnon-autosomeswereremoved(SupplementaryFig.1).Furthermore,weexcludedSNPswithacallrateo99%,aminorallelefrequency(MAF)o0.01and/orasigni?cantdeviationfromHardy–Weinbergequilibrium(HWE)inthecontrols(Po10À4)duringeachstage.WecomputedprincipalcomponentsofExome_AsianArray(including13,473individuals)and

Exome_FineArray(including8,949individuals)stagesusing108,576SNPsand108,099SNPs(MAF40.01,SNPswithHWEP410À4,SNPswithacallrate499%,andcarryingoutLDpruningusingthePLINKoption‘-indep-pairwise5050.20),respectively.Afterqualitycontrol,thegenotypedataof89,720overlappedautosomalvariantsin11,245casesand11,177controlswereincludedforfurtheranalysis.

Genotypingcomparison.Toevaluatethegenotypingquality,wecomparedtheconcordanceratesforthesamplesgenotypedinourstudyandeither(i)samplessequencedbywhole-exomesequencinginourpreviousstudy1or(ii)samplesgenotypedontheIlluminaHuman610-QuadBeadChip16.FortheExome_FineArraydata,thecomparisonswerebasedon89,720and12,320overlappingvariantswithin102and38individuals,respectively.Theconcordancerateswere99.985and99.978%forthewhole-exomesequencingdataandIlluminaHuman610-QuadBeadChipdata,respectively.Moreover,theconcordanceratesforthehomozygousandheterozygousgenotypeswere99.975and99.965%,respectively,forthewhole-exomesequencingdataand99.874and99.954%,respectively,fortheIlluminaHuman610-QuadBeadChipdata.FortheExome_FineArraydata,the

comparisonswerebasedon15,620and22,458overlappingvariantswithin348and159individuals,respectively.Theconcordancerateswere99.964and99.986%forthewhole-exomesequencingdataandtheIlluminaHuman610-QuadBeadChipdata,respectively.Moreover,theconcordanceratesforhomozygousand

heterozygousgenotypeswere99.978and99.968%,respectively,forthewhole-exomesequencingdataand99.865and99.976%,respectively,fortheIlluminaHuman610-QuadBeadChipdata.Theconcordancerateofthe100blindduplicatesampleswas99.988%.

SNPselectionandgenotypingforreplication.Toreplicatetheassociationresultsofthemeta-analysisoftheAsiaarrayandtheFineMappingarray,wefurtheranalysedthe76topvariantsinanadditional20,338samples(6,369casesand13,969controls,SupplementaryTable1)usingtheSequenomMassARRAYsystem.AlloftheseselectedSNPsmetthefollowingqualitycriteria:(1)theMAFwashigherthan0.5%inboththecasesandcontrols;(2)HWEinthecontrolswasPZ0.01andtheHWEinthecaseswasP410À4;(3)SNPswithameta-associationofPo10À4afteradjustmentforgender;(4)proximitytoputativecandidategenes(immune-relatedorinvolvedinimmunecellproliferationanddifferentiation)orknownsusceptibilitylociforautoimmunediseases;and(5)ineachlocus,oneortwoofthemostsigni?cantSNPswereselectedforvalidation.Forallofthe76SNPsanalysedinthevalidationstudy,theclusterpatternsofthegenotypingdatafromtheIlluminaandSequenomanalyseswerecheckedtocon?rmtheirhighquality.ThegenotypedataareavailableinTable2andSupplementaryData2.

References

1.Bowcock,A.M.&Barker,J.N.Geneticsofpsoriasis:thepotentialimpacton

newtherapies.J.Am.Acad.Dermatol.49,S51–S56(2003).

2.Gottlieb,A.B.Psoriasis:emergingtherapeuticstrategies.Nat.Rev.DrugDiscov.

4,19–34(2005).

3.Grif?ths,C.E.&Barker,J.N.Pathogenesisandclinicalfeaturesofpsoriasis.

Lancet370,263–271(2007).

4.Stuart,P.E.etal.Genome-wideassociationanalysisidenti?esthreepsoriasis

susceptibilityloci.Nat.Genet.42,1000–1004(2010).

5.Tang,H.etal.Alarge-scalescreenforcodingvariantspredisposingtopsoriasis.

Nat.Genet.46,45–50(2014).

6.Tsoi,L.C.etal.Identi?cationof15newpsoriasissusceptibilitylocihighlights

theroleofinnateimmunity.Nat.Genet.44,1341–1348(2012).

7.Ellinghaus,E.etal.Genome-widemeta-analysisofpsoriaticarthritisidenti?es

susceptibilitylocusatREL.J.Invest.Dermatol.132,1133–1140(2012).8.Liu,Y.etal.Agenome-wideassociationstudyofpsoriasisandpsoriatic

arthritisidenti?esnewdiseaseloci.PLoSGenet.4,e1000041(2008).

9.Ellinghaus,http://wendang.chazidian.combinedanalysisofgenome-wideassociationstudiesfor

Crohndiseaseandpsoriasisidenti?essevensharedsusceptibilityloci.Am.J.Hum.Genet.90,636–647(2012).

10.Sheng,Y.etal.Sequencing-basedapproachidenti?http://wendang.chazidian.commun.5,4331(2014).

11.Ellinghaus,E.etal.Genome-wideassociationstudyidenti?esapsoriasissusceptibilitylocusatTRAF3IP2.Nat.Genet.42,991–995(2010).

12.Nair,R.P.etal.Genome-widescanrevealsassociationofpsoriasiswithIL-23andNF-kappaBpathways.Nat.Genet.41,199–204(2009).

13.Strange,A.etal.Agenome-wideassociationstudyidenti?esnewpsoriasissusceptibilitylociandaninteractionbetweenHLA-CandERAP1.Nat.Genet.42,985–990(2010).

14.Sun,L.D.etal.AssociationanalysesidentifysixnewpsoriasissusceptibilitylociintheChinesepopulation.Nat.Genet.42,1005–1009(2010).

15.Capon,F.etal.Identi?cationofZNF313/RNF114asanovelpsoriasissusceptibilitygene.Hum.Mol.Genet.17,1938–1945(2008).

16.Zhang,X.J.etal.Psoriasisgenome-wideassociationstudyidenti?es

susceptibilityvariantswithinLCEgeneclusterat1q21.Nat.Genet.41,205–210(2009).

17.Huffmeier,http://wendang.chazidian.commonvariantsatTRAF3IP2areassociatedwithsusceptibilitytopsoriaticarthritisandpsoriasis.Nat.Genet.42,996–999(2010).

5

NATURECOMMUNICATIONS|6:6793|DOI:10.1038/ncomms7793|http://wendang.chazidian.com/naturecommunications

&2015MacmillanPublishersLimited.Allrightsreserved.