proprietaryios11

RESEARCHARTICLEGWASforserumgalactose-deficientIgA1implicatescriticalgenesoftheO-glycosylationpathwayKrzysztofKiryluk1*,YifuLi1,ZinaMoldoveanu2,HitoshiSuzuki3,ColinReily2,4,PingHou5,JingyuanXie6,NikolMladkova1,SindhuriPrakash1,ClaraFischman1,SamanthaShapiro1,RobertA.
LeDesma1,DrewBradbury1,IulianaIonita-Laza7,FrankEitner8,9,ThomasRauen8,NicolasMaillard10,FrancoisBerthoux10,Ju¨rgenFloege8,NanChen6,HongZhang5,FrancescoScolari11,12,RobertJ.
Wyatt13,14,BruceA.
Julian2,4,AliG.
Gharavi1,JanNovak21Dept.
ofMedicine,Div.
ofNephrology,CollegeofPhysiciansandSurgeons,ColumbiaUniversity,NewYork,NewYork,UnitedStatesofAmerica,2Dept.
ofMicrobiology,UniversityofAlabamaatBirmingham,Birmingham,Alabama,UnitedStatesofAmerica,3DivisionofNephrology,Dept.
ofInternalMedicine,JuntendoUniversityFacultyofMedicine,Tokyo,Japan,4Dept.
ofMedicine,UniversityofAlabamaatBirmingham,Birmingham,Alabama,UnitedStatesofAmerica,5RenalDiv.
,PekingUniversityFirstHospital,PekingUniversityInstituteofNephrology,Beijing,China,6Dept.
ofNephrology,RuijinHospital,ShanghaiJiaoTongUniversitySchoolofMedicine,Shanghai,China,7Dept.
ofBiostatistics,MailmanSchoolofPublicHealth,ColumbiaUniversity,NewYork,NewYork,UnitedStatesofAmerica,8Dept.
ofNephrology,RWTHUniversityofAachen,Aachen,Germany,9KidneyDiseasesResearch,BayerPharmaAG,Wuppertal,Germany,10Nephrology,Dialysis,andRenalTransplantationDept.
,UniversityNorthHospital,SaintEtienne,France,11Div.
ofNephrology,AziendaOspedalieraSpedaliCiviliofBrescia,MontichiariHospital,UnivofBrescia,Brescia,Italy,12Dept.
ofMedicalandSurgicalSpecialties,RadiologicalSciences,UniversityofBrescia,Brescia,Italy,13Div.
ofPediatricNephrology,UniversityofTennesseeHealthSciencesCenter,Memphis,Tennessee,UnitedStatesofAmerica,14Children'sFoundationResearchInstitute,LeBonheurChildren'sHospital,Memphis,Tennessee,UnitedStatesofAmerica*kk473@columbia.
eduAbstractAberrantO-glycosylationofserumimmunoglobulinA1(IgA1)representsaheritablepatho-genicdefectinIgAnephropathy,themostcommonformofglomerulonephritisworldwide,butspecificgeneticfactorsinvolvedinitsdeterminationarenotknown.
WeperformedaquantitativeGWASforserumlevelsofgalactose-deficientIgA1(Gd-IgA1)in2,633subjectsofEuropeanandEastAsianancestryanddiscoveredtwogenome-widesignificantloci,inC1GALT1(rs13226913,P=3.
2x1011)andC1GALT1C1(rs5910940,P=2.
7x108).
ThesegenesencodemolecularpartnersessentialforenzymaticO-glycosylationofIgA1.
Wedemonstratedthatthesetwolociexplainapproximately7%ofvariabilityincirculatingGd-IgA1inEuropeans,butonly2%inEastAsians.
Notably,theGd-IgA1-increasingalleleofrs13226913iscommoninEuropeans,butrareinEastAsians.
Moreover,rs13226913representsastrongcis-eQTLforC1GALT1thatencodesthekeyenzymeresponsibleforthetransferofgalactosetoO-linkedglycansonIgA1.
ByinvitrosiRNAknock-downstudies,weconfirmedthatmRNAlevelsofbothC1GALT1andC1GALT1C1determinetherateofsecretionofGd-IgA1inIgA1-producingcells.
OurfindingsprovidenovelinsightsintothegeneticregulationofO-glycosylationandarerelevantnotonlytoIgAnephropathy,butalsoPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,20171/22a1111111111a1111111111a1111111111a1111111111a1111111111OPENACCESSCitation:KirylukK,LiY,MoldoveanuZ,SuzukiH,ReilyC,HouP,etal.
(2017)GWASforserumgalactose-deficientIgA1implicatescriticalgenesoftheO-glycosylationpathway.
PLoSGenet13(2):e1006609.
doi:10.
1371/journal.
pgen.
1006609Editor:AmandaJ.
Myers,UniversityofMiami,MillerSchoolofMedicine,UNITEDSTATESReceived:September19,2016Accepted:January27,2017Published:February10,2017Copyright:2017Kiryluketal.
ThisisanopenaccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited.
DataAvailabilityStatement:TheprimarydataareavailablefromdbGAP,accessionnumberphs000431.
v2.
p1.
Funding:ThisstudywassupportedbythefollowingNIHgrantsfromtheNationalInstituteforDiabetesandDigestiveKidneyDiseases(NIDDK):K23DK090207(KK),R03DK099564(KK),R01DK105124(KK),K01DK106341(CR),R01DK078244(JN),andR01DK082753(AGG,JN),andbytheCenterforGlomerularDiseasesatColumbiaUniversity.
ThefundershadnoroleintoothercomplextraitsassociatedwithO-glycosylationdefects,includinginflammatoryboweldisease,hematologicdisease,andcancer.
AuthorsummaryO-glycosylationisacommontypeofpost-translationalmodificationofproteins;specificabnormalitiesinthemechanismofO-glycosylationhavebeenimplicatedincancer,inflam-matoryandblooddiseases.
However,themolecularbasisofabnormalO-glycosylationinthesecomplexdisordersisnotknown.
WestudiedthegeneticbasisofdefectiveO-glycosyl-ationofserumimmunoglobulinA1(IgA1),thatrepresentsthekeypathogenicdefectinIgAnephropathy,themostcommonformofprimaryglomerulonephritisworldwide.
Wereportourresultsofthefirstgenome-wideassociationstudyforthistraitusingserumassaysin2,633individualsofEuropeanandEast-Asianancestry.
Inourgenomescan,weobservedtwosignificantsignalswithlargeeffects,onchromosomes7p21.
3andXq24,jointlyexplainingabout7%oftraitvariability.
ThesesignalsimplicatetwogenesthatencodemolecularpartnersessentialforenzymaticO-glycosylationofIgA1andmucins,andrepresentpotentialnewtargetsfortherapy.
IntroductionN-andO-glycosylationarefundamentalpost-translationalmodificationsofproteinsinmam-maliancells.
Abnormalitiesinglycosylationhavebeenlinkedtoabroadrangeofhumandis-eases,includingneurologicdisorders,immune-mediatedandinflammatorydiseasesaswellascancer.
Proteinglycosylationismediatedbyalargefamilyofenzymesthathavecell-andtissue-specificactivity,andcangeneratehighlydiverseglycanstructuresthatareimportantforsignal-ing,cell-cellandcell-matrixinteractions.
Thecombinatorialpossibilitiesofglycanstructuresimpartedbythelargenumberofglycosylationenzymescomplicateasystematicanalysisofpro-teinglycosylationpatternsandidentificationofcriticalstepsinvolvedintheactivity,concentra-tion,andregulationinanygivencellortissue.
Insuchasetting,geneticstudiesofcongenitaldefectsofglycosylationinhumanshaveprovidedsignificantinsightintonon-redundantregula-torynodesinthispathway[1].
ThemajorityoftheseMendeliandisordersarisefromlossoffunctionmutationsthatseverelyperturbproteinglycosylationacrossarangeoftissuesandpro-duceawiderangeoforgandysfunctioninearlylife.
However,lesspronouncedabnormalitiesinproteinglycosylationhavealsobeendetectedincomplexdisorderssuchasautoimmunityandcancer,suggestingthatmoresubtledefectsinthispathwaycanhaveimportantconse-quencesforhumanhealth.
IgAnephropathy(IgAN),themostcommoncauseofglomerulonephritisandacommoncauseofkidneyfailureworldwide,isaprototypicalexampleofanimmune-mediateddisordercharacterizedbyabnormalglycosylation[2].
Inhumans,thehinge-regionsegmentsoftheheavychainsofimmunoglobulinA1(IgA1)have3to6O-glycans,resultinginavarietyofIgA1glyco-formsincirculation.
Inhealthyindividuals,theprevailingO-glycansincludetheN-acetylgalacto-samine(GalNAc)-galactosedisaccharideanditssialylatedforms.
InIgAN,galactose-deficientIgA1(Gd-IgA1)glycoformsaresignificantlymoreabundantcomparedtothoseofhealthycon-trols[3].
Theseunder-galactosylatedglycoformsaresecretedbyIgA1-producingcellswhilegalac-tosylationofothercirculatingO-glycosylatedproteinsispreserved,suggestingaspecificdefectwithinIgA1-producingcells[4].
ThepathogeneticmechanismofIgANinvolvesanautoimmuneGeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,20172/22studydesign,datacollectionandanalysis,decisiontopublish,orpreparationofthemanuscript.
Competinginterests:Theauthorshavedeclaredthatnocompetinginterestsexist.
responseresultinginproductionofIgAorIgGautoantibodiesagainstcirculatingGd-IgA1,andformationofimmunecomplexes(Gd-IgA1complexedwithautoantibodies)thatdepositinthekidneyandcausetissueinjury[2,5].
Consistentwiththismechanism,Gd-IgA1isthepredomi-nantglycoformincirculatingimmunecomplexesandintheglomerularimmunedepositsinpatientswithIgAN[6–9]andelevatedserumlevelsofGd-IgA1(autoantigen)andanti-glycanantibodies(autoantibody)areassociatedwithmoreaggressivediseaseandacceleratedprogres-siontoend-stagekidneyfailure[10,11].
Thedesignofasimplelectin-basedELISAassay,usingaGalNAc-specificlectinfromHelixaspersa(HAA),enablesscreeningofseratoquantifythelevelsofcirculatingGd-IgA1[3].
Usingthisassay,wehavedemonstratedthattheserumlevelsofGd-IgA1representanormallydistributedquantitativetraitinhealthypopulations,butuptotwothirdsofIgANpatientshavelevelsabovethe95thpercentileforhealthycontrols.
Examiningfamilymembersofpro-bandswithfamilialandsporadicformsofIgAN,wealsoshowedthatelevatedserumGd-IgA1levelssegregateindependentlyofserumtotalIgAlevelsandhavehighheritability(estimatedat50–70%)[12,13].
Moreover,manyhealthyfamilymembersexhibitedveryhighGd-IgA1levels,identifyingelevatedGd-IgA1asaheritableriskfactorthatprecedesthedevelopmentofIgAN.
Todate,multiethnicgenome-wideassociationstudiesinvolvingover20,000individualshaveidentified15risklocipredisposingtoIgAN,highlightingtheimportanceofinnateandadaptiveimmunityinthisdisorder.
Poweranalysesindicatedthatdiscoveryofadditionalrisklociusingthecase-controldesignwillrequiresignificantexpansioninsamplesize.
However,asystematicanalysisofquantitativeendophenotypesthatarelinkedtodiseasepathogenesis,suchasGd-IgA1,hasnotbeenconductedtodateandmayprovidetheopportunitytodiscoveradditionalpathogenicpathwaysusingasmallersamplesize.
Inthisstudy,weperformedthefirstGWASforserumGd-IgA1levels,andsuccessfullymappednewlociwithsurprisinglylargecontributionstotheheritabilityofthecirculatinglevelofGd-IgA1independentlyofIgAlevels.
ResultsInordertotestifserumlevelsofGd-IgA1remainstableovertime,wefirstperformedmea-surementsoftotalserumimmunoglobulinlevelsalongwithGd-IgA1levelsatbaselineandatfouryearsoffollow-upin32individualsofEuropeanancestryfollowedlongitudinally(Fig1).
WhileserumtotalIgGandIgAlevelsvariedwithtime,Gd-IgA1levels(normalizedfortotalIgA)remainedremarkablystableovera4-yearperiod(r2=0.
92,P=1.
8x1013),demonstrat-ingthatO-glycosylationofIgA1isminimallyaffectedbyrandomenvironmentalfactors.
WenextusedHAAlectin-basedELISAtoanalyzesingletime-pointseraof1,195individu-alsinourdiscoverycohortscomposedof950individualsofEast-Asianancestry(483biopsy-documentedIgANcasesand467controls)and245individualsofEuropeanancestry(141biopsy-documentedIgANcasesand104controls,Table1).
Aspreviouslyreported,serumGd-IgA1levelswerepositivelycorrelatedwithage(EastAsiansr=0.
13,P=8.
9x10-5;Europe-ansr=0.
15,P=1.
7x10-2)andtotalIgAlevels(EastAsiansr=0.
75,P0.
05).
Inbothcohorts,Gd-IgA1levelswerealsosignificantlyhigherinIgANcasescomparedtocontrolsindependentlyofageandtotalIgAlevels(adjustedP95%(discovery)or>90%(replication)wereincludedintheanalysis.
**KASPTM:KompetitiveAlelleSpecicPCR(aproprietarySNP-typingtechnologybyLGCGenomics;accuracy>99.
8%).
doi:10.
1371/journal.
pgen.
1006609.
t001GeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,20174/22WefirstexaminedpotentialassociationswithknownIgANsusceptibilityloci,butfoundnostatisticallysignificantorsuggestivesignalsbetweenGd-IgA1levelsandknownIgANriskalleles(S1Table).
Inaddition,wefoundnoassociationbetweentheglobalpolygeneticriskscoreforIgAN,whichcapturesthecombinedeffectofallIgANriskloci,andGd-IgA1levels.
WealsodidnotdetectanyassociationsofGd-IgA1levelswithlocipreviouslylinkedtovariationintotalIgAlevels[14–16],IgAdeficiency[17],orN-glycosylationofIgG[18].
Atthesametime,werepli-catedpreviouslyreportedassociationoftotalIgAwithELL2(rs56219066,P=8.
5x10-3)[15],confirmingthatgeneticregulationofIgAlevelsisdistinctfromthatforGd-IgA1levels.
ThesedatathusindicatedthepresenceofyetundiscoveredlocicontrollingvariationinGd-IgA1levels.
Wenextexaminedgenome-widedistributionofP-valuesfromthediscoverystagetoidentifynovellociassociatedwithGd-IgA1levels.
Althoughnosignalreachedgenome-widesignifi-canceinthediscoverystage,weobservedseveralsuggestive(P0.
10),ancestryoutliers,andsampleswithadetectedGeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201713/22sexmismatch.
Afterallquality-controlsteps,theChineseDiscoveryCohortwascomposedof950individualstypedwith508,112SNPs,whiletheUSDiscoveryCohortwascomposedof245indi-vidualstypedwith531,778SNPs.
Intotal,468,781SNPsoverlappedbetweenthecohorts,andthissetofcommonmarkerswasusedforthediscoverymeta-analysis.
Toreduceanypotentialbiasfrompopulationstructure,weusedmodifiedPCA-basedancestrymatchingalgorithms(Spec-tral-GEMsoftware)[38,39],asdescribedinourpriorstudiesofthesecohorts[34,37].
PrimaryassociationtestingfortheGd-IgA1phenotype(expressedasstandardizedresiduals)wasper-formedforeachindividualcohortunderanadditivelinearmodelinPLINK[40].
Weincludedsignificantprincipalcomponentsofancestryascovariatesinlinearmodelsusedforassociationtesting.
Additionally,weperformedregressionanalyseswithandwithoutadjustmentforserumtotalIgAlevels.
WederivedadjustedeffectestimateswithstandarderrorsforeachSNP,andwecombinedtheseresultsusinganinversevariance-weightedmethod(METALsoftware)[41].
Wevisuallyexaminedgenome-widedistributionsofP-valuesusingQQ-plotsforeachindividualcohort,aswellasforthejointanalysisofbothcohorts.
Weestimatedthegenomicinflationfac-tors[42],thatwerenegligibleforeachindividualdiscoverycohort(lambda=1.
011and1.
013fortheChineseandUScohorts,respectively).
Theoverallgenomicinflationfactorwasestimatedat1.
010andthefinalmeta-analysisQQ-plotsshowednoglobaldeviationfromtheexpecteddistri-butionofP-values(S1Fig).
Stage2:Follow-upofsuggestivesignalsWenextprioritizedthetop50SNPsforreplicationamongthetopsuggestiveSNPswithP99.
8%accuracyrates[43].
Table1summarizestheethniccompositionofourreplicationcohortsalongwiththegenotypingmethodandaveragegenotypecallrates.
Wefirstcarriedoutassociationanalysesindividuallywithineachofthecohortsusingthesamemethodsasinthediscoverystudy.
Next,wecombinedtheresultsusingafixed-effectsmodel(S2Table).
ForeachofthegenotypedSNPs,wederivedpooledeffectestimatesandtheir95%confidenceintervals.
Todeclaregenome-widesignificance,weusedthegenerallyacceptedthresholdofP0.
5)ateachlocus.
WeannotatedthesevariantsusingANNOVAR[46],SeattleSeq[47],SNPNexus[48],FunciSNP[49],HaploReg4[50],andChroMos[51].
ThetranscriptswhoseexpressionswerecorrelatedwiththeleadSNPsincis-ortrans-werealsoidentifiedusingavailableeQTLdatasets,including:(1)peripheralbloodeQTLsbasedonmeta-analysisof5,311Europeans[52],(2)primaryB-cellandmonocyteeQTLsfrom288Europeans[53],and(3)thelatestreleaseofGTExdataacrossmultipletissuetypes[20,54].
Weutilized,automatedMEDLINEtextminingtoolstoassessnetworkconnectivitybetweengenesresidinginimplicatedGWASloci,includingGRAIL[55],e-LiSe[56],andFACTA+[57].
Wealsointerrogatedallknownprotein-proteinGeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201715/22interactionnetworksforconnectivitybetweencandidategenesusingtheDiseaseAssociationProtein-ProteinLinkEvaluator(DAPPLE)[58]andProteinInteractionNetworkAnalysisplat-form(PINA2)[59].
WeusedCytoscapev.
2.
8tovisualizenetworkgraphs.
siRNAknock-downstudiesinIgA1secretingcelllinesIgA1-secretingcelllinesfromfivepatientswithIgANandfivehealthycontrolsweretrans-fectedusingON-TARGETplusSMARTpoolsiRNAs(ThermoFisherScientific)specificforhumanC1GALT1,COSMC,orboth.
TheON-TARGETplusNon-targetingPoolsiRNAswasusedasacontrol.
WefollowedourpreviouslypublishedprotocolforAmaxanucleofectorII(Lonza)[60].
Twenty-fourhoursaftertransfection,theknock-downefficiencywasdeterminedbyqRT-PCRwithpreviouslydescribedprimers[9,60].
TheknockdownwasexpressedascDNAleveloftheindividualgenenormalizedtoGAPDHafterrespectivesiRNAtreatment,dividedbytherespectivevalueobtainedaftertreatmentbynon-targetingsiRNA.
TheeffectofsiRNAknock-downonthephenotype(thedegreeofgalactose-deficiencyofIgA1)wasbasedonthereactivityofsecretedIgA1withalectinfromHelixaspersaspecificforterminalGalNAc,asdescribed[9,60].
SupportinginformationS1Fig.
Studydesignandquantile-quantileplotsforthediscoverymeta-analysis.
(a)Studyflowchartsummarizingthediscoverycohorts(stage1)andthereplicationcohorts(stage2)withfinalnumbersofindividualsafterphenotypeandgenotypequalitycontrolanalyses;(b)QQ-plotforthegenome-widediscoverymeta-analysis(N=1,195)ofserumGd-IgA1levelswithoutadjustmentforserumtotalIgAlevelsand(c)afteradjustmentforserumtotalIgAlev-els.
AllsignalswithP<5x10-4(horizontalline)frombothanalyseswereprioritizedforfollow-upinreplicationcohorts(stage2).
Lambda:genomicinflationfactor.
(PDF)S2Fig.
Thesuggestivelocusonchromosome7p13encodingHECW1.
(a)Meantraitvalues(+/-standarderrors)byrs978056genotype.
(b)RegionalplotoftheHECW1locusandthetopsignalrepresentedbyrs978056(P=3.
3x10-5);thex-axispresentsphysicaldistanceinkilobases(hg18coordinates),andthey-axispresentslogPvaluesforassociationstatistics.
(c)Thenet-workofknownprotein-proteininteractionsbetweenHECW1,C1GALT1,andC1GALT1C1-encodedproteins.
Eachnoderepresentsaproteinandeachedgerepresentsahighconfidencephysicalinteraction.
Theseedtermsarehighlightedingreenandtheircommoninteractorsinyellow.
TheproteininteractionswereanalyzedandvisualizedusingtheProteinInteractionNetworkAnalysis(PINA2)platform.
(PDF)S3Fig.
Thegene-phenotypeco-citationnetwork.
Theco-citationnetworkwasconstructedbasedonallPubMedabstractsforthequerytermsC1GALT1(61abstracts),C1GALT1C1(39abstracts),andHECW1(5abstracts).
Bothhumanandmousediseasephenotypes(circles)wereanalyzedforco-citation(edges)withthethreequeryterms(greendiamonds).
Commoninteractorsarehighlightedinyellow.
ThePubMedquerywasperformedonDecember15th,2015andthegene-phenotypenetworkwasvisualizedinCytoscapev.
2.
8.
IgAN:IgAnephropa-thy;HSPN:Henoch-Schoenleinpurpuranephritis;ALS:amyotrophiclateralsclerosis.
(PDF)S4Fig.
Conditionalanalysisofthetopthreelociusingallimputedmarkers(1000Genomesreference,version3).
Thetoprowdepictsunconditioneddiscoverymeta-analysisresultsforalltheimputedmarkersatthe(a)C1GALT1,(b)C1GALT1C1,and(c)HECW1loci.
TheGeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201716/22bottomrowdepictsthediscoverymeta-analysisresultsafterconditioningindividualcohortresultsfortheleadSNP(s)ateachlocus:(d)rs13226913andrs1008897attheC1GALT1locus,(e)rs5910940andrs2196262attheC1GALT1C1locus,and(f)rs978056attheHECW1locus.
ThereddottedlinecorrespondstoP=1x103andisprovidedforreference.
(PDF)S5Fig.
DensityplotsforthedistributionofadjustedandstandardizedGd-IgA1levelsbycase/controlstatus.
ThedistributionaldifferencesinGd-IgA1levelsbetweencasesandcon-trolsfor(a)allstudycohorts,(b)Europeancohorts,and(c)EastAsiancohorts.
TheGd-IgA1traitisexpressedasstandardizedresidualsofnaturallog-transformedserumGd-IgA1levelsafteradjustmentforage,sex,totalIgAlevels,andcohortmembership;eachstandarddeviationincreaseintheGd-IgA1endophenotypeisassociatedwithdiseaseOR(95%CI)of1.
53(1.
40–1.
68),1.
49(1.
31–1.
72),and1.
56(1.
37–1.
78)forAll,European,andEastAsiancohorts,respec-tively.
(PDF)S1Table.
AssociationofknownIgANsusceptibilitylociwithserumGd-IgA1levelsinthejointanalysisofthediscoverycohorts(totalN=1,195).
Theassociationresultswereadjustedforage,totalIgA,case-controlstatus,ancestry,andcohortmembership.
(PDF)S2Table.
Combinedassociationresultsforthe50lociselectedforreplication.
SerumGd-IgA1levelsbeforeandafteradjustmentforserumtotalIgAlevels.
(PDF)S3Table.
Studypower.
Thepowerwasestimatedforarangeofeffectsizesexpressedasfrac-tionoftotalvarianceofthequantitativetraitexplainedbyageneticvariant(columns).
Theassumptionsinclude:standardnormaltraitdistribution,additiveriskmodel,noheterogeneity,markerallelicfrequencyof0.
25,perfectLDbetweenamarkerandacausalallele,afollow-upsignificancethresholdofP<5*104(toprow)andajointsignificancelevelofP<5*108(bot-tomrow).
Shadedinredisthestudydetectionlimitcorrespondingtoallelesexplaining1.
5%oftotalvariance.
(PDF)S4Table.
Totalvarianceexplainedbygenome-widesignificantloci.
Thefractionoftotalvarianceexplainedwasestimatedbyregressingindividualgeneticpredictors(additivecoding)againsttheoutcomeofstandardizedresidualsforthetrait(Gd-IgA1levelsadjustedforage,case-controlstatus,andserumtotalIgAlevels)andderivingR2fortheregressionmodel.
Thetotalvarianceexplainedacrossmultiplecohortswascalculatedasanaveragefractionofexplainedvarianceforindividualcohortsweightedbycohortsize.
ThevarianceexplainedbytheC1GALT1locuswascalculatedbyincludingbothrs13226913andrs1008897intheregres-sionmodel.
ForC1GALT1C1locus,bothrs5910940andrs2196262wereincludedunderaddi-tivecoding.
ThetotalvarianceexplainedjointlybyC1GALT1andC1GALT1C1lociwascalculatedbyincludingallfourSNPpredictorsfromtheselociinasingleregressionmodel.
(PDF)S5Table.
Mutualconditioningacrossthegenome-widesignificantloci.
EachSNPthatreachedgenome-widesignificanceinourstudywasconditionedonallotherSNPsthatreachedgenome-widesignificance,oneatatime.
Highlightedinredareindependenteffectsformarkerslocatedwithinthesamelocusafterconditioningontheothersignificantmarkerwithinthesamelocus.
Notably,conditioningwithineachlocusdemonstratesresidualeffects,whilemutualconditioningacrosslocistrengthenstheassociationsignalateachlocus.
BecauseGeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201717/22chromosomeXmarkersareincludedintheseanalyses,allmodelsweresub-stratifiedbasedonsex;theconditioningwasfirstperformedwithineachsub-cohort,thentheresultswerecom-binedusingfixedeffectsmeta-analysis.
Inallanalyses,markerswerecodedunderanadditivemodelandtheGd-IgA1-increasingallelewasusedasatestallele.
StdErr.
Standarderror.
(PDF)S6Table.
HaploRegregulatoryannotationsforvariantsinlinkagedisequilibrium(r2<0.
85)withrs13226913basedonRoadmapEpigenomesandENCODEdata:sortedbyr2withrs13226913;mostpromisingcandidateshighlightedinred.
(XLSX)S7Table.
ExpressionQTLeffectsofrs13226913acrossmultipletissuetypes.
(PDF)S8Table.
Explorationofalternativegeneticmodels.
Weexploredtwoalternativegeneticmodels(dominantandrecessive)andcomparedthesemodelsusingBayesianInformationCri-terion(BIC).
Thebestmodelishighlightedinred.
Whilethisanalysissuggestsanadditivemodelfor4outof5topmarkers,theeffectofrs5910940(C1GALT1C1locus)isbestexplainedbyaT-alleledominantmodel.
Allanalyseswerestratifiedbasedonsex,explainingslightdiffer-encesineffectestimatesandp-valuescomparedtoTable2.
StdErr:standarderror.
(PDF)S9Table.
HaploRegregulatoryannotationsforvariantsinlinkagedisequilibrium(r2<0.
85)withrs5910940basedonRoadmapEpigenomesandENCODEdata:sortedbyr2withrs5910940;mostpromisingcandidateshighlightedinred.
(XLSX)S10Table.
Ethnicity-specificassociationresultsforthesignificantandsuggestiveloci.
TheEastAsiansincludetheChineseDiscovery,theChineseReplication,andtheJapaneseReplica-tioncohorts(N=1,603).
TheEuropeansincludetheUSdiscoverycohort(100%self-identifiedWhites),German,French,andUSReplicationcohorts(N=1,030).
Theresultsforallethnic-ity-definedcohortswerecombinedusingfixedeffectsmeta-analysis.
Allelicfrequencieswereaveragedwithintheethnicity-definedcohorts.
(PDF)S11Table.
SamplesizesrequiredfortestingnewGd-IgA1lociforassociationwithIgAnephropathy.
Minimumsamplesizes(cases+controls)requiredtodetectassociationsofthenewlydetectedGd-IgA1lociwiththeriskofIgANinEastAsian,European,andbi-ethnicGWASassumingobservedeffectsizes,50%caseproportion,α=5x108,andpower(1-β)of80%,90%and99.
9%.
ThevarianceexplainedbyeachlocuswasderivedasinS4Table.
TheobservedORsofdiseaseperstandarddeviationofendophenotypewerecalculatedbasedonlogisticregressionwithcase/controlstatusasanoutcomeandstandardizedresidualsofGd-IgA1(afteradjustmentforage,sex,cohort,andtotalIgAlevels)asapredictor.
Separateesti-mateswereobtainedforourEastAsian,European,andbi-ethniccohorts.
ThecalculationswereperformedwithintheframeworkofMendelianRandomization,aspreviouslyproposedbyBrionetal.
IntJEpidemiol42,1497–501(2013)andimplementedintheonlinecalculatorathttps://cnsgenomics.
shinyapps.
io/mRnd/(PDF)AcknowledgmentsWearegratefultoallstudyparticipantsfortheircontributiontothiswork.
GeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201718/22AuthorContributionsConceptualization:KKAGGJN.
Datacuration:KKYLNMlPHJX.
Formalanalysis:KKAGGNMlIIL.
Fundingacquisition:KKAGGJN.
Investigation:YLZMHSCRPHJXNMlSPCFSSRALDB.
Methodology:KKAGGJNIIL.
Resources:ZMHSJXFETRNMaFBJFNCHZFSRJWBAJJN.
Supervision:KKAGGJN.
Validation:YLKKZMHSJN.
Visualization:KK.
Writing–originaldraft:KKAGGJN.
Writing–review&editing:KKAGGJNBAJ.
References1.
FreezeHH,ChongJX,BamshadMJ,NgBG.
Solvingglycosylationdisorders:fundamentalapproachesrevealcomplicatedpathways.
AmJHumGenet.
2014;94(2):161–75.
PubMedCentralPMCID:PMCPMC3928651.
doi:10.
1016/j.
ajhg.
2013.
10.
024PMID:245077732.
MagistroniR,D'AgatiVD,AppelGB,KirylukK.
Newdevelopmentsinthegenetics,pathogenesis,andtherapyofIgAnephropathy.
Kidneyinternational.
2015;88(5):974–89.
PubMedCentralPMCID:PMC4653078.
doi:10.
1038/ki.
2015.
252PMID:263761343.
MoldoveanuZ,WyattRJ,LeeJ,TomanaM,JulianBA,MesteckyJ,etal.
PatientswithIgAnephropathyhaveincreasedserumgalactose-deficientIgA1levels.
KidneyInt.
2007;71:1148–54.
doi:10.
1038/sj.
ki.
5002185PMID:173421764.
SmithAC,deWolffJF,MolyneuxK,FeehallyJ,BarrattJ.
O-glycosylationofserumIgDinIgAnephropa-thy.
JournaloftheAmericanSocietyofNephrology:JASN.
2006;17(4):1192–9.
doi:10.
1681/ASN.
2005101115PMID:165107645.
SuzukiH,FanR,ZhangZ,BrownR,HallS,JulianBA,etal.
AberrantlyglycosylatedIgA1inIgAnephropathypatientsisrecognizedbyIgGantibodieswithrestrictedheterogeneity.
JClinInvest.
2009;119(6):1668–77.
PubMedCentralPMCID:PMC2689118.
doi:10.
1172/JCI38468PMID:194784576.
AllenAC,BaileyEM,BrenchleyPE,BuckKS,BarrattJ,FeehallyJ.
MesangialIgA1inIgAnephropathyexhibitsaberrantO-glycosylation:observationsinthreepatients.
KidneyInt.
2001;60(3):969–73.
doi:10.
1046/j.
1523-1755.
2001.
060003969.
xPMID:115320917.
HikiY,OdaniH,TakahashiM,YasudaY,NishimotoA,IwaseH,etal.
Massspectrometryprovesunder-O-glycosylationofglomerularIgA1inIgAnephropathy.
Kidneyinternational.
2001;59(3):1077–85.
doi:10.
1046/j.
1523-1755.
2001.
0590031077.
xPMID:112313638.
TomanaM,NovakJ,JulianBA,MatousovicK,KonecnyK,MesteckyJ.
CirculatingimmunecomplexesinIgAnephropathyconsistofIgA1withgalactose-deficienthingeregionandantiglycanantibodies.
JClinInvest.
1999;104(1):73–81.
PubMedCentralPMCID:PMCPMC408399.
doi:10.
1172/JCI5535PMID:103937019.
SuzukiH,MoldoveanuZ,HallS,BrownR,VuHL,NovakL,etal.
IgA1-secretingcelllinesfrompatientswithIgAnephropathyproduceaberrantlyglycosylatedIgA1.
JClinInvest.
2008;118(2):629–39.
Epub2008/01/04.
doi:10.
1172/JCI33189PMID:1817255110.
ZhaoN,HouP,LvJ,MoldoveanuZ,LiY,KirylukK,etal.
Thelevelofgalactose-deficientIgA1intheseraofpatientswithIgAnephropathyisassociatedwithdiseaseprogression.
Kidneyinternational.
2012;82(7):790–6.
PubMedCentralPMCID:PMC3443545.
doi:10.
1038/ki.
2012.
197PMID:2267388811.
BerthouxF,SuzukiH,ThibaudinL,YanagawaH,MaillardN,MariatC,etal.
Autoantibodiestargetinggalactose-deficientIgA1associatewithprogressionofIgAnephropathy.
JournaloftheAmericanGeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201719/22SocietyofNephrology:JASN.
2012;23(9):1579–87.
PubMedCentralPMCID:PMC3431415.
doi:10.
1681/ASN.
2012010053PMID:2290435212.
GharaviAG,MoldoveanuZ,WyattRJ,BarkerCV,WoodfordSY,LiftonRP,etal.
AberrantIgA1glyco-sylationisinheritedinfamilialandsporadicIgAnephropathy.
JournaloftheAmericanSocietyofNephrology.
2008;19(5):1008–14.
Epub2008/02/15.
doi:10.
1681/ASN.
2007091052PMID:1827284113.
KirylukK,MoldoveanuZ,SandersJT,EisonTM,SuzukiH,JulianBA,etal.
AberrantglycosylationofIgA1isinheritedinbothpediatricIgAnephropathyandHenoch-Scho¨nleinpurpuranephritis.
Kidneyinternational.
2011;80(1):79–87.
Epub2011/02/18.
doi:10.
1038/ki.
2011.
16PMID:2132617114.
YangC,JieW,YanlongY,XuefengG,AihuaT,YongG,etal.
Genome-wideassociationstudyidenti-fiesTNFSF13asasusceptibilitygeneforIgAinaSouthChinesepopulationinsmokers.
Immunogenet-ics.
2012;64(10):747–53.
doi:10.
1007/s00251-012-0636-yPMID:2286492315.
SwaminathanB,ThorleifssonG,JoudM,AliM,JohnssonE,AjoreR,etal.
VariantsinELL2influencingimmunoglobulinlevelsassociatewithmultiplemyeloma.
NatCommun.
2015;6:7213.
PubMedCentralPMCID:PMCPMC4455110.
doi:10.
1038/ncomms8213PMID:2600763016.
ViktorinA,FrankowiackM,PadyukovL,ChangZ,MelenE,SaafA,etal.
IgAmeasurementsinover12000SwedishtwinsrevealsexdifferentialheritabilityandregulatorylocusnearCD30L.
HumMolGenet.
2014;23(15):4177–84.
PubMedCentralPMCID:PMCPMC4082371.
doi:10.
1093/hmg/ddu135PMID:2467635817.
FerreiraRC,Pan-HammarstromQ,GrahamRR,GatevaV,FontanG,LeeAT,etal.
AssociationofIFIH1andotherautoimmunityriskalleleswithselectiveIgAdeficiency.
Naturegenetics.
2010;42(9):777–80.
doi:10.
1038/ng.
644PMID:2069401118.
LaucG,HuffmanJE,PucicM,ZgagaL,AdamczykB,MuzinicA,etal.
LociassociatedwithN-glycosyla-tionofhumanimmunoglobulinGshowpleiotropywithautoimmunediseasesandhaematologicalcan-cers.
PLoSgenetics.
2013;9(1):e1003225.
PubMedCentralPMCID:PMCPMC3561084.
doi:10.
1371/journal.
pgen.
1003225PMID:2338269119.
QinW,ZhouQ,YangLC,LiZ,SuBH,LuoH,etal.
PeripheralBlymphocytebeta1,3-galactosyltransfer-aseandchaperoneexpressioninimmunoglobulinAnephropathy.
Journalofinternalmedicine.
2005;258(5):467–77.
doi:10.
1111/j.
1365-2796.
2005.
01558.
xPMID:1623868320.
ConsortiumGT.
TheGenotype-TissueExpression(GTEx)project.
Naturegenetics.
2013;45(6):580–5.
PubMedCentralPMCID:PMC4010069.
doi:10.
1038/ng.
2653PMID:2371532321.
JuT,CummingsRD.
AuniquemolecularchaperoneCosmcrequiredforactivityofthemammaliancore1beta3-galactosyltransferase.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica.
2002;99(26):16613–8.
PubMedCentralPMCID:PMC139192.
doi:10.
1073/pnas.
262438199PMID:1246468222.
LyonsPA,RaynerTF,TrivediS,HolleJU,WattsRA,JayneDR,etal.
GeneticallydistinctsubsetswithinANCA-associatedvasculitis.
NEnglJMed.
2012;367(3):214–23.
PubMedCentralPMCID:PMCPMC3773907.
doi:10.
1056/NEJMoa1108735PMID:2280895623.
LiangL,Willis-OwenSA,LapriseC,WongKC,DaviesGA,HudsonTJ,etal.
Anepigenome-wideasso-ciationstudyoftotalserumimmunoglobulinEconcentration.
Nature.
2015;520(7549):670–4.
PubMedCentralPMCID:PMCPMC4416961.
doi:10.
1038/nature14125PMID:2570780424.
WeidingerS,GiegerC,RodriguezE,BaurechtH,MempelM,KloppN,etal.
Genome-widescanontotalserumIgElevelsidentifiesFCER1Aasnovelsusceptibilitylocus.
PLoSgenetics.
2008;4(8):e1000166.
PubMedCentralPMCID:PMCPMC2565692.
doi:10.
1371/journal.
pgen.
1000166PMID:1884622825.
FuJ,WeiB,WenT,JohanssonME,LiuX,BradfordE,etal.
Lossofintestinalcore1-derivedO-glycanscausesspontaneouscolitisinmice.
JClinInvest.
2011;121(4):1657–66.
PubMedCentralPMCID:PMC3069788.
doi:10.
1172/JCI45538PMID:2138350326.
Perez-MunozME,BergstromK,PengV,SchmaltzR,Jimenez-CardonaR,MarstellerN,etal.
Discor-dancebetweenchangesinthegutmicrobiotaandpathogenicityinamousemodelofspontaneouscoli-tis.
Gutmicrobes.
2014;5(3):286–95.
PubMedCentralPMCID:PMC4153765.
doi:10.
4161/gmic.
28622PMID:2466209827.
ChangD,GaoF,SlavneyA,MaL,WaldmanYY,SamsAJ,etal.
AccountingforeXentricities:analysisoftheXchromosomeinGWASrevealsX-linkedgenesimplicatedinautoimmunediseases.
PloSone.
2014;9(12):e113684.
PubMedCentralPMCID:PMC4257614.
doi:10.
1371/journal.
pone.
0113684PMID:2547942328.
TheodoratouE,CampbellH,VenthamNT,KolarichD,Pucic-BakovicM,ZoldosV,etal.
Theroleofgly-cosylationinIBD.
NaturereviewsGastroenterology&hepatology.
2014;11(10):588–600.
29.
JuT,CummingsRD.
Proteinglycosylation:chaperonemutationinTnsyndrome.
Nature.
2005;437(7063):1252.
doi:10.
1038/4371252aPMID:16251947GeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201720/2230.
RadhakrishnanP,DabelsteenS,MadsenFB,FrancavillaC,KoppKL,SteentoftC,etal.
Immaturetrun-catedO-glycophenotypeofcancerdirectlyinducesoncogenicfeatures.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica.
2014;111(39):E4066–75.
PubMedCentralPMCID:PMC4191756.
doi:10.
1073/pnas.
1406619111PMID:2511827731.
Lomax-BrowneHJ,ViscontiA,PuseyCD,CookHT,SpectorTD,PickeringMC,etal.
IgA1Glycosyla-tionIsHeritableinHealthyTwins.
JournaloftheAmericanSocietyofNephrology.
2017;28(1):64–68.
32.
AlexanderWS,VineyEM,ZhangJG,MetcalfD,KauppiM,HylandCD,etal.
ThrombocytopeniaandkidneydiseaseinmicewithamutationintheC1galt1gene.
ProceedingsoftheNationalAcademyofSciencesoftheUnitedStatesofAmerica.
2006;103(44):16442–7.
PubMedCentralPMCID:PMC1637601.
doi:10.
1073/pnas.
0607872103PMID:1706275333.
BrionMJ,ShakhbazovK,VisscherPM.
CalculatingstatisticalpowerinMendelianrandomizationstud-ies.
IntJEpidemiol.
2013;42(5):1497–501.
PubMedCentralPMCID:PMCPMC3807619.
doi:10.
1093/ije/dyt179PMID:2415907834.
KirylukK,LiY,ScolariF,Sanna-CherchiS,ChoiM,VerbitskyM,etal.
DiscoveryofnewrisklociforIgAnephropathyimplicatesgenesinvolvedinimmunityagainstintestinalpathogens.
Naturegenetics.
2014;46(11):1187–96.
PubMedCentralPMCID:PMCPMC4213311.
doi:10.
1038/ng.
3118PMID:2530575635.
LiM,FooJN,WangJQ,LowHQ,TangXQ,TohKY,etal.
IdentificationofnewsusceptibilitylociforIgAnephropathyinHanChinese.
NatCommun.
2015;6:7270.
PubMedCentralPMCID:PMCPMC4458882.
doi:10.
1038/ncomms8270PMID:2602859336.
MoldoveanuZ,WyattRJ,LeeJY,TomanaM,JulianBA,MesteckyJ,etal.
PatientswithIgAnephropa-thyhaveincreasedserumgalactose-deficientIgA1levels.
Kidneyinternational.
2007;71(11):1148–54.
doi:10.
1038/sj.
ki.
5002185PMID:1734217637.
GharaviAG,KirylukK,ChoiM,LiY,HouP,XieJ,etal.
Genome-wideassociationstudyidentifiessus-ceptibilitylociforIgAnephropathy.
Naturegenetics.
2011;43(4):321–7.
PubMedCentralPMCID:PMC3412515.
doi:10.
1038/ng.
787PMID:2139963338.
LeeAB,LucaD,RoederK.
ASpectralGraphApproachtoDiscoveringGeneticAncestry.
AnnApplStat.
2010;4(1):179–202.
Epub2010/08/07.
PubMedCentralPMCID:PMC2916191.
doi:10.
1214/09-AOAS281PMID:2068965639.
LeeAB,LucaD,KleiL,DevlinB,RoederK.
Discoveringgeneticancestryusingspectralgraphtheory.
GenetEpidemiol.
2010;34(1):51–9.
Epub2009/05/21.
doi:10.
1002/gepi.
20434PMID:1945557840.
PurcellS,NealeB,Todd-BrownK,ThomasL,FerreiraMA,BenderD,etal.
PLINK:atoolsetforwhole-genomeassociationandpopulation-basedlinkageanalyses.
AmJHumGenet.
2007;81(3):559–75.
Epub2007/08/19.
doi:10.
1086/519795PMID:1770190141.
WillerCJ,LiY,AbecasisGR.
METAL:fastandefficientmeta-analysisofgenomewideassociationscans.
Bioinformatics.
2010;26(17):2190–1.
Epub2010/07/10.
PubMedCentralPMCID:PMC2922887.
doi:10.
1093/bioinformatics/btq340PMID:2061638242.
DevlinB,RoederK,BacanuSA.
Unbiasedmethodsforpopulation-basedassociationstudies.
GenetEpidemiol.
2001;21(4):273–84.
Epub2002/01/05.
doi:10.
1002/gepi.
1034PMID:1175446443.
KirylukK,LiY,Sanna-CherchiS,RohanizadeganM,SuzukiH,EitnerF,etal.
GeographicdifferencesingeneticsusceptibilitytoIgAnephropathy:GWASreplicationstudyandgeospatialriskanalysis.
PLoSgenetics.
2012;8(6):e1002765.
PubMedCentralPMCID:PMCPMC3380840.
doi:10.
1371/journal.
pgen.
1002765PMID:2273708244.
DudbridgeF,GusnantoA.
Estimationofsignificancethresholdsforgenomewideassociationscans.
GenetEpidemiol.
2008;32(3):227–34.
Epub2008/02/27.
doi:10.
1002/gepi.
20297PMID:1830029545.
CarrelL,WillardHF.
X-inactivationprofilerevealsextensivevariabilityinX-linkedgeneexpressioninfemales.
Nature.
2005;434(7031):400–4.
doi:10.
1038/nature03479PMID:1577266646.
WangK,LiM,HakonarsonH.
ANNOVAR:functionalannotationofgeneticvariantsfromhigh-through-putsequencingdata.
Nucleicacidsresearch.
2010;38(16):e164.
PubMedCentralPMCID:PMC2938201.
doi:10.
1093/nar/gkq603PMID:2060168547.
NgSB,TurnerEH,RobertsonPD,FlygareSD,BighamAW,LeeC,etal.
Targetedcaptureandmas-sivelyparallelsequencingof12humanexomes.
Nature.
2009;461(7261):272–6.
PubMedCentralPMCID:PMC2844771.
doi:10.
1038/nature08250PMID:1968457148.
ChelalaC,KhanA,LemoineNR.
SNPnexus:awebdatabaseforfunctionalannotationofnewlydiscov-eredandpublicdomainsinglenucleotidepolymorphisms.
Bioinformatics.
2009;25(5):655–61.
PubMedCentralPMCID:PMC2647830.
doi:10.
1093/bioinformatics/btn653PMID:1909802749.
CoetzeeSG,RhieSK,BermanBP,CoetzeeGA,NoushmehrH.
FunciSNP:anR/bioconductortoolinte-gratingfunctionalnon-codingdatasetswithgeneticassociationstudiestoidentifycandidateregulatoryGeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201721/22SNPs.
Nucleicacidsresearch.
2012;40(18):e139.
PubMedCentralPMCID:PMC3467035.
doi:10.
1093/nar/gks542PMID:2268462850.
WardLD,KellisM.
HaploReg:aresourceforexploringchromatinstates,conservation,andregulatorymotifalterationswithinsetsofgeneticallylinkedvariants.
Nucleicacidsresearch.
2012;40(Databaseissue):D930–4.
PubMedCentralPMCID:PMC3245002.
doi:10.
1093/nar/gkr917PMID:2206485151.
BarenboimM,MankeT.
ChroMoS:anintegratedwebtoolforSNPclassification,prioritizationandfunc-tionalinterpretation.
Bioinformatics.
2013;29(17):2197–8.
PubMedCentralPMCID:PMC3740627.
doi:10.
1093/bioinformatics/btt356PMID:2378261652.
WestraHJ,PetersMJ,EskoT,YaghootkarH,SchurmannC,KettunenJ,etal.
SystematicidentificationoftranseQTLsasputativedriversofknowndiseaseassociations.
Naturegenetics.
2013;45(10):1238–43.
doi:10.
1038/ng.
2756PMID:2401363953.
FairfaxBP,MakinoS,RadhakrishnanJ,PlantK,LeslieS,DiltheyA,etal.
Geneticsofgeneexpressioninprimaryimmunecellsidentifiescelltype-specificmasterregulatorsandrolesofHLAalleles.
Naturegenetics.
2012;44(5):502–10.
PubMedCentralPMCID:PMC3437404.
doi:10.
1038/ng.
2205PMID:2244696454.
GamazonER,WheelerHE,ShahKP,MozaffariSV,Aquino-MichaelsK,CarrollRJ,etal.
Agene-basedassociationmethodformappingtraitsusingreferencetranscriptomedata.
Naturegenetics.
2015;47(9):1091–8.
PubMedCentralPMCID:PMC4552594.
doi:10.
1038/ng.
3367PMID:2625884855.
RaychaudhuriS,PlengeRM,RossinEJ,NgAC,InternationalSchizophreniaC,PurcellSM,etal.
Identi-fyingrelationshipsamonggenomicdiseaseregions:predictinggenesatpathogenicSNPassociationsandraredeletions.
PLoSgenetics.
2009;5(6):e1000534.
PubMedCentralPMCID:PMC2694358.
doi:10.
1371/journal.
pgen.
1000534PMID:1955718956.
GladkiA,SiedleckiP,KaczanowskiS,ZielenkiewiczP.
e-LiSe—anonlinetoolforfindingneedlesinthe'(Medline)haystack'.
Bioinformatics.
2008;24(8):1115–7.
doi:10.
1093/bioinformatics/btn086PMID:1832188457.
TsuruokaY,MiwaM,HamamotoK,TsujiiJ,AnaniadouS.
Discoveringandvisualizingindirectassocia-tionsbetweenbiomedicalconcepts.
Bioinformatics.
2011;27(13):i111–9.
PubMedCentralPMCID:PMC3117364.
doi:10.
1093/bioinformatics/btr214PMID:2168505958.
RossinEJ,LageK,RaychaudhuriS,XavierRJ,TatarD,BenitaY,etal.
Proteinsencodedingenomicregionsassociatedwithimmune-mediateddiseasephysicallyinteractandsuggestunderlyingbiology.
PLoSgenetics.
2011;7(1):e1001273.
PubMedCentralPMCID:PMC3020935.
doi:10.
1371/journal.
pgen.
1001273PMID:2124918359.
CowleyMJ,PineseM,KassahnKS,WaddellN,PearsonJV,GrimmondSM,etal.
PINAv2.
0:mininginteractomemodules.
Nucleicacidsresearch.
2012;40(Databaseissue):D862–5.
PubMedCentralPMCID:PMC3244997.
doi:10.
1093/nar/gkr967PMID:2206744360.
SuzukiH,RaskaM,YamadaK,MoldoveanuZ,JulianBA,WyattRJ,etal.
CytokinesalterIgA1O-glyco-sylationbydysregulatingC1GalT1andST6GalNAc-IIenzymes.
TheJournalofbiologicalchemistry.
2014;289(8):5330–9.
PubMedCentralPMCID:PMC3931088.
doi:10.
1074/jbc.
M113.
512277PMID:24398680GeneticsofIgA1O-glycosylationPLOSGenetics|DOI:10.
1371/journal.
pgen.
1006609February10,201722/22