Altered Expression of Insulin Receptor Isoforms in Breast

JiaqiHuang,ChrisMorehouse,KatieStreicher,BrandonW.Higgs,JinGao,MegganCzapiga,AnmarieBoutrin,WeiZhu,PhilipBrohawn,YongChang,JayeViner,TheresaLaVallee,LauraRichman,BahijaJallal,YihongYao*

MedImmune,LLC.,Gaithersburg,Maryland,UnitedStatesofAmerica

Abstract

Purpose:Insulin-likegrowthfactor(IGF)signalingthroughhumaninsulinreceptorisoformA(IR-A)contributestotumorigenesisandintrinsicresistancetoanti-IGF1Rtherapy.Inthepresentstudy,we(a)developedquantitativeTaqManrealtime-PCR-basedassays(qRT-PCR)tomeasurehumaninsulinreceptorisoformswithhighspecificity,(b)evaluatedisoformexpressionlevelsinmolecularly-definedbreastcancersubtypes,and(c)identifiedtheIR-A:IR-BmRNAratioasapotentialbiomarkerguidingpatientstratificationforanti-IGFtherapies.

ExperimentalDesign:mRNAexpressionlevelsofIR-AandIR-Bweremeasuredin42primarybreastcancersand19matchedadjacentnormaltissueswithTaqManqRT-PCRassays.Theresultswerefurtherconfirmedin165breastcancers.ThetumorsampleswereprofiledusingwholegenomemicroarraysandsubsequentlysubtypedusingthePAM50breastcancergenesignature.TherelationshipbetweentheIR-A:IR-Bratioandcancersubtype,aswellasmarkersofproliferationwerecharacterized.Results:ThemRNAexpressionlevelsofIR-Ainthebreasttumorsweresimilartothoseobservedintheadjacentnormaltissues,whilethemRNAlevelsofIR-Bweresignificantlydecreasedintumors.TheIR-A:IR-BratiowassignificantlyhigherinluminalBbreastcancerthaninluminalA.StrongconcordancebetweentheIR-A:IR-BratioandthecompositeOncotypeDXproliferationscorewasobservedforstratifyingthelattertwobreastcancersubtypes.

Conclusions:ThereductioninIR-BexpressionisthekeytothealteredIR-A:IR-Bratioobservedinbreastcancer.TheIR-A:IR-Bratiomayhavebiomarkerutilityinguidingapatientstratificationstrategyforananti-IGFtherapeutic.

Citation:HuangJ,MorehouseC,StreicherK,HiggsBW,GaoJ,etal.(2011)AlteredExpressionofInsulinReceptorIsoformsinBreastCancer.PLoSONE6(10):e26177.doi:10.1371/journal.pone.0026177

Editor:S.K.Batra,UniversityofNebraskaMedicalCenter,UnitedStatesofAmericaReceivedJune23,2011;AcceptedSeptember21,2011;PublishedOctober26,2011

Copyright:ß2011Huangetal.Thisisanopen-accessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense,whichpermitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalauthorandsourcearecredited.

Funding:AllauthorsareemployeesofMedImmune,thefunder,whichplayedaroleinstudydesign,datacollectionandanalysis,decisiontopublish,andpreparationofthemanuscript.

CompetingInterests:AlltheauthorsofthismanuscriptareMedimmuneemployeesandAstrazenecastockowners.TheworkpresentedisentirelyfundedbyMedImmune.MedImmuneisdevelopingananti-IGFmonoclonalantibody.Thisdoesnotaltertheauthors’adherencetoallthePLoSONEpoliciesonsharingdataandmaterials.

*E-mail:YaoY@MedImmune.com

Introduction

Thematurehumaninsulinreceptor(IR)hastwoisoforms,insulinreceptorisoformA(IR-A)andinsulinreceptorisoformB(IR-B).Bothisoformsresultfromalternativesplicingofthesameprimarytranscript.IR-AdiffersfromIR-Bbytheexclusionofexon11,whichencodesa12aminoacidfragment(residues717–728)oftheinsulinreceptora-subunit[1].TherelativeabundanceofmRNAsencodingtheIR-AandIR-Bisoformsisregulatedinatissue-specificmanner[2]andalsodiffersbystageofcelldevelopmentanddifferentiation.IR-Aisthepredominantisoforminfetaltissuesandcancercells[3].ThetwoIRisoformshavebeenreportedtoexhibitdistinctfunctionalproperties.WhileIR-Bhomodimersaretheclassicalreceptorsforinsulinwithmetaboliceffectsinmuscle,liverandadiposetissues,IR-AhomodimerscanbindIGF-IIand,toalesserextent,IGF-I,inadditiontoinsulin[4],[5],[6]andmaybeinvolvedinmediatingthegrowthpromotingandanti-apoptoticeffectsofthisgrowthfactorunderphysiologicalconditionslikeembryonicdevelopment[3].

PLoSONE|www.plosone.org

1

ThefineregulationoftheIR-A:IR-BexpressionratiomayberenderedmorecomplexbycoexpressionofthecognateIGF1R,whichmayformhybridreceptorswithIRs.SeveralstudieshavedemonstratedtheexistenceofIR-A/IGF1Rhybridreceptors.ThesehybridreceptorsareheterodimerscomprisedofanabchainfromtheIRandanabchainfromtheIGF1R,bothinnormal[7],[8],[9]andcancercells[10].IncreasedoccurrenceofthesehybridscanbeobservedinseveraltumorcellsastheresultofIRand/orIGF1Roverexpression[10],[11].ThepharmacologicalpropertiesofthesehybridsseemtodifferfromthoseoftheirhomodimercounterpartsanddependontheIRisoforminvolved.TheIR-A/IGF1RhybridreceptorsarestronglyactivatedbyIGF-IandIGF-IIandweaklyactivatedbyinsulin.Bycontrast,hybridreceptorscontainingIGF1RandIR-BaremuchlesssensitivetoIGF-IIandinsulin[12].ItissuggestedthatproliferativeIGFsignalingcanoccurthroughIGF1Rhomodimers,IGF1R/IR-Aheterodimers,andIR-Ahomodimers[6],[13],[14].

AsakeysignalingcomponentoftheIGFpathway,IGF1Risthetargetofseveralinvestigationalagentsinclinicaldevelopment.

October2011|Volume6|Issue10|e26177

InsulinReceptorIsoformsinBreastCancer

TheseagentsweredevelopedtospecificallyinhibitIGF1RwhilesparingIR-Aduetoconcernsthatco-targetingIGF1RandIRwouldresultinunacceptabletoxicity[15].However,clinicalfailureswithIGF1R-specifictherapysuggestthatitmaybecriticaltoinhibitaberrantIR-AsignalinginadditiontoIGF1R.ThisassertionhasbeensupportedbyUlanet,etal.[15],whoshowedthatIGF-IIsignalingthroughIR-Acontributestotumorigenesisandintrinsicresistancetoanti-IGF1Rtherapy.AdditionalworksupportingtheimportanceofIR-AsignalingwasreportedrecentlybyGaoetal.[16],whodescribedpreclinicaldatausingamonoclonalantibodythatsequestersIGF1andIGF2,therebyblockingsignalingthroughbothIGF1RandIR-A.ResultsshowedthatinresponsetoblockingIGF1andIGF2,dramaticantitumoractivitieswereobservedintumorsthatexpressbothIGF1RandIR-AorIR-Aalone.SimilarantitumoractivitieswerenotseeninthesetumorswheninhibitingIGF1Ralone.

Astherapeuticstrategiesforco-targetingIGF1RandIR-Amoveforwardinclinicaldevelopment,evaluatingIR-AlevelsinclinicaltumorsamplesbecomescriticalforidentifyingthemostappropriatepopulationtoreceiveIGF-targetedtherapy.Todate,themostcommonmethodtospecificallymeasurehumanIR-AandIR-BexpressionhasbeendescribedbyFrascaandcolleagues[3].ThismethodisbasedonPCRamplificationandgelseparation,followedbyqualitativemeasurementoftheresultingbands.Themethodistediousandlacksquantitativeaccuracy,whichlimitsitspotentialuseasabiomarkerforclinicaldevelopment.DuetothelackofanefficientandaccuratemethodtodetecthumanIR-AandIR-BmRNAlevels,theexpressionofIR-AandIR-Bincancertissues,particularlytheIR-A:IR-Bratio,hasbeendifficulttoevaluate.Inthispaper,wedescribethedesignandimplementationofTaqManqRT-PCR-basedassaystospecificallyquantifythelevelsofIR-AandIR-Btranscriptsinhumanbreastcancersamples.Furthermore,weevaluatedtheratioofIR-AtoIR-BexpressioninER+/PR+andHer2-breastcancersofdifferentmolecularsubtypes.Asthesesubtypeshavepreviouslybeenshowntodifferinproliferationindex,responsetohormonaltherapy,andoverallclinicaloutcome,theassociationoftheIR-A:IR-BratiowithaparticularsubtypemayidentifypatientsmorelikelytorespondtoIGF-targetedtherapy.

RNAExtraction

TotalRNAwasextractedfromsnap-frozentissuespecimensusingtheZRRNAMicroPrepkit(ZymoResearch,Orange,CA).RNApurityandconcentrationweredeterminedspectrophoto-metrically(260/280.1.9).RNAqualitywasassessedonanAgilent2100BioanalyzerusingtheRNA6000NanoLabChipH.

DesignReal-timeQuantitativePCRPrimerandProbesforInsulinReceptorIsoforms

Full-lengthmRNAtranscriptsequencesfortheIR-A(NM_001079817)andIR-B(NM_000208)isoformswereretrievedfromtheNCBIReferenceSequencesdatabase.FortheIR-Aassaydesign,wetargetedtheexon10/12junctionregionforthegenespecificprobe;theexon10codingregionfortheforwardprimerpairs;andtheexon12codingregionsforthereverseprimerpairs.FortheIR-Bassaydesign,wetargetedtheexon11interiorcodingregionforthegenespecificprobe;exon11/12junctionfortheforwardprimerpairs;andexon12forthereverseprimerpairs.Allprimer/probedesignswereimportedintothePrimerExpress(AppliedBiosystems,FosterCity,CA)softwaretooltoensureoptimaldesignforutilizationintheTaqManGeneExpressionassayprocedure.Allprobesweredesignedtoincorporateaminorgroovebindingmoiety(MGB),andwerelabeledwithafluorescentdye(FAM)fordetectionandanon-fluorescentquencher.PrimersandprobeswerecustomorderedfromAppliedBiosystems.Sequencesforallprimer/probecombinationsareasfollows:IR-A:Thesequenceoftheprobeis59-TCCCCAGGCCATCT-39;Thesequenceoftheforwardprimeris59-TGAGGAT-TACCTGCACAACG-39;

Thesequenceofthereverseprimeris59-ACCGTCA-CATTCCCAACATC-39.

IR-B:Thesequenceoftheprobeis59CCGAGGACCC-TAGGC-39;

Thesequenceoftheforwardprimeris59-CGTCCCCA-GAAAAACCTCTTC-39;

Thesequenceofreverseprimeris59-GGACCTGCGTTTCC-GAGAT-39.

MaterialsandMethodsPrimaryBreastTumorTissues

Forty-twogradeItoIIIinfiltratingbreastductalcarcinomaswerepurchasedfromILSbio(Chestertown,MD).Nineteenofthe42breasttumortissuesampleshadmatchednormaladjacentbreasttissue(NAT)samples.Inthecaseofbreastcancermastectomy,normaltissuefromtheoppositebreastwastakenandservesasthenormaladjacenttissue.Theagesofpatientsrangedfrom31to88years.Allthebreastcancersampleswereestrogenreceptor(ER)andprogesteronereceptor(PR)positiveandHER2negativeaccordingtoimmunohistochemistry.Allsampleswerefreshlyfrozenandcollectedbeforeinitiationofanytreatment.Tumorsamplesweremacrodissectedtoremovenormaltissue,andtumorpurityinallsampleswasgreaterthan85%.Normalsamplesweremacro-dissectedtoremovenon-glandulartissue.

PositiveandNegativeControlsforTaqManassayofIR-AandIR-B

CommerciallyavailablecDNAclonesthatcontainthefull-lengthcDNAclonesofIR-A(clonedinpCMV6-XL4)andIR-B(clonedinpCMV6-XL5)werepurchasedfromOriGeneTech-nologies,Inc(IR-A:SKU#.SC311328;IR-B:SKU#SC315880).ThesequenceverificationofeachIRisoformclonewasconductedin-houseatMedImmune.TheemptyplasmidspCMV6-XL4andpCMV6-XL5wereusedasnegativecontrolDNAforIR-AandIR-Bassays,respectively.

TaqManqRT-PCRGeneExpressionAnalysis

StandardTaqManqRT-PCRGeneExpressionassayswereconductedina384-wellformatforallprimer/probeandtemplatecombinations.Reactionsconsistedof5mLofTaqManUniversalMasterMix(AppliedBiosystems),0.5mLof10xGeneExpressionAssayMix,and4.5mLofvaryingcopynumbersofeithertheIR-BorIR-AcDNAclone,forafinalvolumeof10mLperwell.Eachprimer/probeandtemplatecombinationwasrepeatedatleast3times.AllassayplateswererunonanAppliedBiosystems7900HTdetectionsystemusingstandardsettings(cyclingprogramincluded10minincubationat95uCfollowedby40cyclesof95uCfor15secand60uCfor1min).Datavalues(CycleThreshold[Ct]values)wereextractedfromeachassaywiththeSDSv2.0softwaretool(AppliedBiosystems).

2

October2011|Volume6|Issue10|e26177

BreastCancerTissuecDNAArrays

FourbreastcancertissuecDNAarrays(BCRT101,BCRT102,BCRT103,andBCRT104)werepurchasedfromOriGeneTech-nologies(Rockville,MD).ThearrayscontaincDNAsfrom15normalbreasttissues(from10uniqueindividuals)and165uniquebreastadenocarcinomatissues.ThetumorstagerangedfromstageItoIVandthetissueswerecomprisedof50–90%tumor.

PLoSONE|www.plosone.org

InsulinReceptorIsoformsinBreastCancer

AssessmentoftheExpressionlevelsofOtherGenesinBreastCancer

TaqManGeneExpressionassayswerepurchasedfromAppliedBiosystems.Theassaysinclude:INSR(AssayID:Hs009615_m1);ESR1(AssayID:Hs00174860_m1);PGR(AssayID:Hs01556707_m1);ERBB2(HER2,AssayID:Hs01001580_m1);tumorprolifer-ationgenes(PikeSetal2004):BIRC5(Survivin,AssayID:Hs00153353_m1),AURKA(STK15,AssayID:Hs01582073_m1),CCNB1(AssayID:Hs00259126_m1),MKI67(AssayID:Hs01032443_m1),MYBL2(AssayID:Hs009423_m1);andreferencegenes:ACTB(Hs99999903_m1),GUSB(AssyID:Hs99999908_m1),GAPDH(AssayID:Hs99999905_m1),RPLP0(AssayID:Hs99999902_m1),TFRC(AssayID:Hs99999911_m1).TheBioMarkynamicArray(FluidigmCorporation)microfluidicssystemallowsforhighthroughputreal-timePCR(upto2304individualreactionsperplate),producinghighqualitydatawithlowvariabilityandatightcorrelationwithconventionalRT-PCR.SinglestrandedcDNAwasgeneratedfromtotalRNAusingtheSuperScriptHIIIFirst-StrandSynthesisSuperMix(Invitrogen,Carlsbad,CA).cDNAsampleswerepre-amplifiedusingTaqManPre-AmpMasterMix,accordingtothemanufac-turer’sinstructions.Reactionscontained5mLofcDNA,10mLPre-AmpMasterMixand5mLof0.2Xgeneexpressionassaymix(comprisedofallprimer/probestobeassayed)forafinalvolumeof20mL.Reactionswerecycledwiththerecommendedprogramfor14cyclesandthendiluted1:5withTEbuffer.Pre-amplifiedcDNAwaseitherutilizedimmediatelyorstoredat220uCuntilprocessed.

Topreparesamplesforloadinginto488dynamicarraychips(Fluidigm),thereactionmixcontained2.5mL2XUniversalMasterMix(AppliedBiosystems),0.25mLSampleLoadingBuffer(FluidigmCorporation),and2.25mLpre-amplifiedcDNA.Topreparetheprimer/probes,thereactionmixcontained2.5mL20XTaqManGeneExpressionAssayand2.5mLAssayLoadingBuffer(FluidigmCorporation).Priortoloadingthesamplesandassayreagentsintotheinlets,thechipwasprimedintheIFCController.FivemLofsamplepreparedasdescribedwasloadedintoeachsampleinletofthedynamicarraychip,and5mLof10Xgeneexpressionassaymixwasloadedintoeachdetectorinlet.ThechipwasplacedontheIFCControllerforloadingandmixing.UponcompletionoftheIFCprimingstep,thechipwasloadedontheBioMarkTMReal-TimePCRSystemforthermalcycling(10minat95uCfollowedby40cyclesof95uCfor15secand1minat60uC).Thenumberofreplicatesandthecompositionofthesamplesvarieddependingontheparticularexperiment,butwereneverlessthantriplicate.AverageCtvalueswereusedtodeterminesensitivityandspecificityofthedesignedprobes.TheaverageCtvaluesofallavailablereferencegeneassayswithinasamplewereutilizedforDCtcalculation.

Software(GCOS)tool.Stratagene’s(LaJolla,CA)ArrayAssistHLitesoftwarewasusedtocalculateprobe-levelsummaries(GC-RMAandMAS5)fromthearrayCELfiles.

BreastCancerMolecularSubtypeClassification

AsubsetofER+/PR+andHER2-primarybreasttumors(n=40)andasubsetofmatchednormaladjacentbreasttissuesamples(n=15)wereprofiledonAffymetrixHumanGenomeU133Plus2.0GeneChipHarrays.TwooftheprimarybreastsamplesandfourofthenormaladjacentbreasttissuesamplesanalyzedontheFluidigmplatformwerenotprocessedonGeneChipduetoinsuf-ficientRNAquantity.Breastcancermolecularsubtypeclassification,withregardstoluminal-Aandluminal-Bsubtype,wasconductedutilizingourwholegenomearraydata.

Twomethodsfordeterminingputativesampleclassificationwereimplemented.ThefirstclassificationmethodutilizedapublishedPAM50-geneshrunkencentroidclassifier[17]forsamplesub-typing(normal,basal-like,HER2,luminal-A,orluminal-B)purposes.MAS5normalizedGeneChipdatawasusedforthisanalysisgiventhatthepublishedclassifierwasbuiltusingthistypeofscaleddata.ThesampleswereclassifiedaccordingtoaSpearman’srankcorrelation(50-geneintensityvectorvs.subtypecentroidclassifier),wherethesubtypewiththehighestcorrelationvaluewasassignedtoaparticularsample.ThesecondmethodutilizedGC-RMAnormalizedGeneChipdatatoidentifyapanelofdifferentiallyexpressedtranscriptsbyatwo-sampleWelch’st-testanalysis.Samplesweredividedintotwogroups(normalortumor)basedonpathologyassessmentpriortoconductingthestatisticalanalysis.Probesdisplayingafoldchangedifferential.3andP-value,1.0610212(q-value,1.0610211)(n=459probes)wereusedforanunsupervisedhierarchicalclusteringanalysis.Severaladditionalprobesetselectionthresholdswereutilizedandresultedinsimilarclusteringpatterns(datanotshown).

BioMarkynamicArrayMicrofluidicsSystem

Results

SensitivityandSpecificityofTaqManqRT-PCRAssaytoDistinguishIR-AandIR-B

ProbesensitivitywastestedbyperformingIR-AorIR-BTaqManqRT-PCRassaysstartingwith100pgtemplatestocksolutionsofIR-AorIR-B(approximately107copiesofDNAtemplate).TheDNAwasseriallydilutedto1024pg(approxi-mately10copiesDNAtemplate).Eachsamplewastestedinduplicate.Theslopeofthestandardtemplatedilutioncurvewasdeterminedbyplottingcycle-threshold(Ct)valuesasafunctionofthelogDNAcopynumber.TheresultsareshowninFigure1.Strongcorrelationswereobservedbetweenthelog[concentration]andresultantCtvaluesforeachassaytestedwithitsrespectivematchingstandardtemplate.Allregressioncoefficients(r2value)were$0.999(P#0.0001).LinearitiesweremaintainedintheDNAconcentrationrangesdescribedaboveinbothassays,demonstrat-ingawidedynamicrangeandyieldingaccurateCtvalues.TheresultsindicatethatboththeIR-AandIR-Bassayshavetheappropriatesensitivity,andareabletodetectthecorrespondingisoformtoapproximately35copiesofDNA.

ThespecificityoftheassayswasalsoassessedbytestingtheIR-AassayinthepresenceoftheIR-BDNAtemplateor,alternatively,theIR-BassayinpresenceoftheIR-ADNAtemplate.ThedetailsofthetestingresultsareshowninTableS1.TheIR-AassaydoesnotamplifyIR-BDNAtemplateintherangetested(10to107copiesofIR-BDNA).Likewise,theIR-BassaydoesnotamplifytheIR-ADNAtemplateintherangetested.Incontrast,wefoundthecommercialAppliedBiosystemsassays(IR-A:Hs00965956_m1;

3

October2011|Volume6|Issue10|e26177

MicroarrayProcessing

Generationofbiotin-labeledamplifiedcRNAfrom75ngoftotalRNAwasaccomplishedwiththeMessageAmpTMPremierRNAAmplificationKit(Ambion,Austin,TX).TheconcentrationandpurityofthecRNAproductweredeterminedspectrophoto-metrically.Fifteenmicrogramsofeachbiotin-labeledcRNAwasfragmentedforhybridizationonAffymetrixHumanGenomeU133Plus2.0GeneChipHarrays.AllGeneChipHwashing,staining,andscanningprocedureswereperformedwithAffyme-trixstandardequipment.DatacaptureandinitialarrayqualityassessmentswereperformedwiththeGeneChipOperating

PLoSONE|www.plosone.org

InsulinReceptorIsoformsinBreastCancer

Figure1.IR-AandIR-Bassayserialdilutioncurves.ProbesensitivitywastestedbyperformingIR-AorIR-BTaqManqRT-PCRassayswithaserialdilutionofapproximately107to10copiesofplasmidDNAofeitherIR-A,IR-B,oranemptyvectorcontrol.TheYaxisrepresentscycle-threshold(Ct)valuesandtheXaxisrepresentslogDNAcopynumber.Theslopeandregressioncoefficient(r2value)ofthestandarddilutioncurvefortheIR-Aassayare-3.259and0.9992,respectively.Theslopeandregressioncoefficient(r2value)ofthestandarddilutioncurvefortheIR-Bassayare-3.155and0.99,respectively.

doi:10.1371/journal.pone.0026177.g001

IR-B:Hs00169631_m1)hadstrongcross-interactionandcouldnotmeasureeachisoformaccurately(datanotshown).

Theassayefficiencywasassessedbytheslopesofthestandarddilutioncurvesforbothassays(Figure1).Theslopeis-3.259forIR-Aand-3.155forIR-B.Thetwoslopesareverysimilar,suggestingonlyminordifferencesinprobeefficiency.

TheseTaqMan-basedassayshavebeensuccessfullyimplement-edontheBioMarkynamicArray(Fluidigm),facilitatingtheiruseinahigherthroughputsystemanddemonstratingtheirperformanceacrossmultipleplatforms.

Figure2.RelativemRNAexpressionlevelsofinsulinreceptoranditsisoformsinprimarybreastcancercomparedtonormalbreasttissues.TaqMangeneexpressionassaydeterminedfoldchangedifferentials(log2-basescale)ofINSR(total),IR-A,andIR-Bbetweennormal(n=19)andtumor(n=42)breasttissuesamples.AveragenormalDCtvalueswereusedforcalculationoffoldchangedifferentialsforeachsample.Atwo-tailedWelch’st-testanalysisidentifiedasignificantdifferencebetweennormalandtumorsamplesforbothINSRandIR-B(P,0.001),whereasnodifferencewasobservedforIR-A(P=0.450).Blackbarsrepresentthemedianlog2foldchangevaluewithinaparticulargenetargetandtissue-typecombination.doi:10.1371/journal.pone.0026177.g002

ExpressionofIR-AandIR-BinBreastCancer

In

ordertoassesstheIR-AandIR-BmRNAexpressioninbreastcancer,42primarybreasttissuesamplesthatareER+,PR+andHER2-,and19matchednormaladjacentbreasttissueswereevaluated.RandomhexamerprimedcDNAswerepre-amplifiedandassayedforexpressionlevelsoftotalIR,IR-A,andIR-BtranscriptsbyTaqManqRT-PCR(Fluidigm).Sampleswerenormalizedtotheaverageoffivehousekeepinggenesasdescribedinthemethodssection.TheresultsareshowninFigure2.Themeanfoldchangedifferentials(log2-basescale)695%CIofIR,IR-A,andIR-Binnormal(n=19)were0.0060.20,0.0060.27,and0.0060.20,respectively.Themeanfoldchangedifferentials(log2-basescale)695%CIofIR,IR-A,andIR-Bintumor(n=42)were-0.8860.29,-0.0760.33,and-2.0860.29,respectively.Atwo-tailedWelch’st-testanalysisindicatedthatthemRNAlevelsofIRandIR-Baresignificantlylowerinbreastcancerwhencomparedtonormalbreasttissue(P,0.0001).NosignificantdifferenceswereobservedinthemRNAlevelsofIR-Ainbreastcancerwhencomparedwithnormalbreasttissue(P=0.450).

IncreasingtherelativeproportionofIR-Aintumors.WecalculatedtheproportionofIR-Arelativetototalinsulinreceptorcomposition(IR-A+IR-B)inmatchedtumorandnormalpairsasdeterminedbya2(2DCt)calculation.Theresultsareshownin

DecreasingIR-BmRNAexpressionlevelintumors.

PLoSONE|www.plosone.org

4

Figure3.ThemeanIR-Atranscriptproportions(%)695%CIwere46.60%.74%and75.24%65.02%forthenormalpanel(n=19)andmatchedtumorsamples,respectively.Apairedsamplet-testanalysisindicatedthatasignificantincreaseinthecalculatedIR-Aproportionintumorsamplesexistswhencomparedtomatchednormaltissue(P,0.0001).TheresultssuggestthatthesignificantlydecreasedIR-BlevelsintumorcontributetoanoverallincreaseintheproportionofIR-Aintumorsamplescomparedtonormaltissue.

RatiosofIR-AtoIR-Binbreastcancer.InordertoassessthemRNAtranscriptratiosofIR-AandIR-B,wecalculatedDCtdifferentialsofIR-AandIR-Binnormalandprimarybreasttumorsamples.TheDCtdifferential(IR-BDCt–IR-ADCt)valueswerecalculatedforallsamplesutilizingthewithin-samplereferencegenepanel(averageCt)fornormalizationpurposes.ThemeanIR-A:IR-BDCt695%CIwere-0.2060.27and1.8160.31inthenormal(n=19)andprimarytumors(n=42),respectively.Atwo-tailedWelch’st-testanalysisidentifiedasignificantdifferencebetweennormalandtumorsamplesinrelationtoobservedIR-A:IR-BDCt(P,0.0001)(Figure4A).TheresultsindicatethatthereisasignificantincreaseintheratioofIR-AtoIR-Binbreasttumors.

Tofurthervalidatetheaboveresults,weassessedmRNAexpressionratiosofIR-AandIR-Binanadditionalbreastcancertissuepanel.PCRarrayscontainingcDNAsfrom15normalbreasttissues(obtainedfrom10independentdonors)and165unique

October2011|Volume6|Issue10|e26177

InsulinReceptorIsoformsinBreastCancer

breastadenocarcinomatissueswereused.EqualamountsofcDNAwerepre-amplifiedandassayedforexpressionlevelsofIR-A,IR-B,andERbyTaqManqRT-PCR(Fluidigm).TheDCtdifferential(IR-BDCt–IR-ADCt)wascalculatedforallsamplesutilizingthewithin-samplereferencegenepanel(ACTB,GUSB,andGAPDH)fornormalizationpurposes.TheresultsareshowninFigure4B.ThemeanIR-A:IR-BDCt695%CIwas-0.5760.34innormaltissues(n=15);and1.2060.17acrossallbreastcancersexamined(n=165).Wethencollectedthebreastcancersamplesthatdisplayeda2-foldoverexpressionofestrogenreceptorrelativetonormalbreasttissue,andcomparedtheirIR-A:IR-BDCtdifferentialstonormaltissueandtoallbreastcancersamples.TheresultsareshowninFigure5.ThemeanIR-A:IR-BDCt695%CIwas1.3960.23inER+breastcancers(n=83),whichissimilartothatobservedacrossthewholebreastcancerdataset.Atwo-tailedWelch’st-testanalysisidentifiedasignificantdifferencebetweennormalandtumorsamplesinrelationtoobservedIR-A:IR-BDCtdifferential(P,0.0001).

CorrelatingIR-A:IR-BRatiowithGenesInvolvedinBreastCancerProliferation

Figure3.TheproportionofIR-Aexpressioninmatchednormalbreastandbreastcancerspecimens.Withinsampleproportionofinsulinreceptorisoform-A(IR-A)relativetototalinsulinreceptorcomposition(ie,IR-A+IR-B)asdeterminedby2(-DCt)calculation.Apairedsamplet-testanalysisindicatedthatasignificantdifferenceexistsforcalculatedIR-Aproportionbetweenmatchednormalandtumorsamples(P,0.001).BlackbarsrepresentthemedianIR-Aproportion(%)withinthenormal(46.60.74,mean%695%CI)andtumor(75.2465.03,mean%695%CI)tissue.doi:10.1371/journal.pone.0026177.g003

AURKA,BIRC5,CCNB1,MKI67andMYBL2arewellcharacterizedgenesinvolvedinbreastcancerproliferation[18].ThecompositeexpressionscoreofthesegeneshasbeenusedinOncotypeDXandcellproliferationisthemostimportantfactorcontributingtobreastcancerrecurrenceinmanypatients.Therefore,weinvestigatedtherelationshipoftheIR-A:IR-Bratiowiththeproliferationscoreinprimarybreastcancersamplesetsusingregressionandcorrelationanalyses.LinearregressionanalysiswasconductedtoquantifytherelationshipbetweenthecalculatedIR-A:IR-BDCtdifferentialandapooledpanelof

Figure4.IncreasingIR-A:IR-BratioinprimarybreastcancerandbreastcancersamplesfromqPCRcDNAarray.(A)CalculatedDCtdifferentialsofinsulinreceptorisoformsIR-AandIR-Binprimarybreastcancersamples:normal(n=19)andtumor(n=42)breastsamples.(B)CalculatedDCtdifferentialsofinsulinreceptorisoformsIR-AandIR-BbreastcancersamplesfromqPCRcDNAarray:normal(n=15),breastcancercDNApanels(n=168),andbreasttumorsfromthesecDNApanelswith.2-foldestrogenreceptorover-expression(n=83).DCtdifferentials(IR-BDCt-IR-ADCt)valueswerecalculatedforallsamplesutilizingthewithin-samplereferencegenepanel(averageCt)fornormalizationpurposes.Atwo-tailedWelch’st-testanalysisidentifiedasignificantdifferencebetweennormalandtumorsamplesinrelationtoobservedIR-A:IR-BDCtdifferential(P,0.001).BlackbarsrepresentthemedianIR-A:IR-BDCt.doi:10.1371/journal.pone.0026177.g004

PLoSONE|www.plosone.org5October2011|Volume6|Issue10|e26177

InsulinReceptorIsoformsinBreastCancer

Figure5.CorrelationofIR-A:IR-BDCtdifferentialwiththeexpressionofproliferationgenes.LinearregressionanalysisoftherelationshipbetweencalculatedIR-A:IR-BDCtdifferential(Xaxis)andapooledpanelofproliferationmarkers(AURKA,BIRC5,CCNB1,KI67,andMYBL2)(Yaxis).Proliferationpanelsummaryvalueswerecalculatedbytakingthemean–(DCt)acrossallmarkersforaparticularsample.Summaryresultsforbothnormalandtumorsamplesarepresented.Thelinearregressionanalysisresultssuggestapositivecorrelationbetweenthetwosummaryvalues(r=0.78).doi:10.1371/journal.pone.0026177.g005

proliferationmarkers(AURKA,BIRC5,CCNB1,MKI67,andMYBL2).ProliferationpanelsummaryvalueswerecalculatedbytakingtheaverageDCtacrossallmarkersforaparticularsample.SummaryresultsforbothnormalandtumorsamplesarepresentedinFigure5.Thelinearregressionanalysisresultssuggestamoderatepositivecorrelationbetweenthetwosummaryvalues(r=0.78).TheseresultssuggestthattheincreasedproportionofIR-AobservedintumortissuemaycontributetoenhancedproliferationinER+/PR+andHER2-breastcancer.

IR-A:IR-BDCtDifferentialinBreastCancerSubtypes

UsingmRNAexpressionprofiles,ER+breastcancerscanbefurtherclassifiedbyhierarchicalclusteranalysisintoluminal-Aandluminal-Bsubtypes[19].Luminal-Acancersareusuallyhistologicallylow-gradeandsensitivetoendocrinetherapy[20].Bycontrast,luminal-Bcancersareoftenhigh-gradeandarelesssensitivetoendocrinetherapy,andhavehigherproliferationandpoorerprognosis[20],[21].Creighton,etal.[22]reportedthatanIGF-Isignatureismanifestedinluminal-Bbreastcancers.Thissignatureishighlycorrelatedwithpoorprognosticfactorsandoneofthestrongestindicatorsofdiseaseoutcome.ToexaminetheroleofadditionalcomponentsoftheIGFsignalingpathwayinbreastcancer,weinvestigatedthehypothesisthatdifferencesinIR-A:IR-Bratiosmaybeevidentwhencomparingluminal-Aandluminal-Bbreastcancers.

Toaddressthisquestionweconductedwholegenomearrayanalysison40ER+/PR+andHER2-negativebreasttumorsand15normalbreastsamples.WeinitiallyutilizedapublishedPAM50-geneshrunkencentroidclassifier[17]onourMAS5normalizedGeneChipdataasabenchmarkforamethodthathasbeenusedtopartitionER+/PR+andHER2-negativebreast

PLoSONE|www.plosone.org

6

tumorsamplesintoluminal-Aandluminal-Bsubtypes.SampleswereclassifiedaccordingtoaSpearman’srankcorrelation,wherethesubtypewiththehighestcorrelationvaluewasassignedtoaparticularsample.IR-A:IR-BDCtdifferentialsinthenormal,luminal-Aandluminal-Bsampleswerethencompared.ThescatterplotrepresentationofcalculatedIR-A:IR-BDCtdifferen-tialswithregardstosamplesubtype(normal,luminal-A,orluminal-B)areshowninFigure6A.ThemeanIR-A:IR-BDCt695%CIwas-0.2760.34innormal(n=15);1.0960.38inluminal-Abreastcancers(n=13);and2.1260.39inluminal-Bbreastcancers(n=27).Allsubtypepair-wisecomparisonsdisplayasignificantdifference(two-samplet-test,P,0.001).TheresultsindicatethatIR-A:IR-Bratiosaresignificantlyincreasedinluminal-Bcancersrelativetoluminal-Acancers,withthegreatestdifferentialobservedbetweenluminal-Bandnormalbreasttissue.Additionally,wefoundthattheseluminal-Bsamplesexhibitedasignificantincreaseinexpressionofaproliferationgenepanel(datanotshown).IncreasedproportionofIR-Ainluminal-Bpatientsmayplayanimportantroleininducingtheincreasedproliferationobservedinthisstudyandothers[21].

Inadditiontotheshrunkencentroidclassifier,weidentifiedanindependentpanelofdifferentiallyexpressedtranscriptsfromourGeneChipdatausingatwo-sampleWelch’st-testandunsuper-visedhierarchicalclusteringanalysis.Samplesweredividedintotwogroups(normalortumor)basedonpathologyassessmentsperformedpriortoconductingthestatisticalanalysis.Sub-populationsidentifiedwereclassifiedasnormal,luminal-A,orluminal-Basafunctionoftranscriptpanelcompositionforthethreeprimaryclustersidentified.Differentiationofluminal-Aandluminal-Bsub-typeswasprimarilydeterminedbyover-expressionofproliferationandcellcycleassociatedgenes(e.g.AURKB,BUB1,CCNA2,CCNE2,E2F7,GTSE1,MKI67,andPKMYT1)intheputativeluminal-Btumorsandover-expressionofextracellularmatrixassociatedgenes(e.g.ASPN,BGN,CILP2,MFAP2,andVCAN)intheputativeluminal-Atumors.IR-A:IR-BDCtdifferentialsinthenormal,luminal-Aandluminal-Bwerethencompared.TheresultsareshowninFigure6B.ThemeanIR-A:IR-BDCt695%CIwas-0.3260.28innormal(n=15);1.0560.20inluminal-Apredictedbreastcancers(n=18);and2.4260.37inluminal-Bpredictedbreastcancers(n=22).Allsubtypepair-wisecomparisonsdisplayasignificantdifference(two-samplet-test,P,0.001).

Overall,an80%concordancebetweenthetwoclassificationmethodsimplementedwasobserved,basedonourunbiased

¨vetothetruesub-typeassessments.Unfortunately,wearenaı

classificationofthetumorsamplessoitisdifficulttoresolvedifferencesinclassificationbetweenthetwomethodologies.However,theoverallconcordancebetweentheresultsfromcomparisonsoftheIR-A:IR-BDCtdifferentialwithboththepublishedPAM50classifierandanindependentgeneselectionandclusteringmethodfurthersupportsourobservationthattheIR-A:IR-Bratioissignificantlyincreasedintheluminal-Bsubtypeofbreastcancer.

Discussion

WehavedescribedspecificTaqManqRT-PCR-basedassaysthataccuratelymeasurethemRNAlevelsofhumanIRisoformsIR-AandIR-B.Intheabsenceofspecificreagentstoquantitativelydistinguishthetwoisoforms,previousresearchutilizedanRT-PCR/gelquantificationprotocoltodistinguishIR-AandIR-Bandmeasuretheirrelativeamountsinnormalandcancertissues[3].TheTaqManassayswehavedevelopedprovideasignificantimprovementinsensitivity,reproducibility,and

October2011|Volume6|Issue10|e26177

InsulinReceptorIsoformsinBreastCancer

Figure6.IR-A:IR-BDCtdifferentialinsubtypesofER+breastcancer.ScatterplotrepresentationofcalculatedIR-A:IR-BDCtdifferentialswithregardtosamplesubtype(normal,luminal-A,orluminal-B)classificationdeterminedbyashrunkencentroidclassifier-basedmethodology(A)andahierarchicalclusteringanalysis(B).Allsubtypepair-wisecomparisonsdisplayasignificantdifference(two-samplet-test,P,0.001).BlackbarsrepresentthemedianIR-A:IR-BDCtdifferentialwithinaparticularsamplesubtype.doi:10.1371/journal.pone.0026177.g006

quantitativecapacityoverpreviousmethods.TheassayshavebeensuccessfullytransferredtotheBioMarkynamicArray(Fluidigm)platformforhighthroughputapplications,suggestingthatmigrationtoanapplicableclinicaltestingplatformisalsopossible.Additionally,preliminaryresultsindicatethatwecansuccessfullyimplementtheseassaysonFFPEsamplesduetotherelativelysmallampliconsizeofeachassay(datanotshown).SinceFFPEistheclinicalstandardforcancerdiagnosis,examiningsamplesinthismannerwillallowforhigh-throughputandsensitivequantitationofIR-A:IR-BratiosoverwhatistypicallyalongcourseofclinicalcareandalsohelpidentifythesubsetofpatientswithalteredIR-A:IR-BratiosthatmightrespondtoIGF-targetedtherapy.

Over-expressionofhybridIGF1RandIR-Areceptorshasbeenreportedinbreastcancer[23],butowingtotechnicaldifficultiesinquantifyingIR-AandIR-Binlargenumbersofclinicalsamples,theoverallexpressionlevelsofIR-AandIR-Binbreastcancerrelativetonormalbreasttissuewerenotfullyknown.OuranalysisofprimarybreastcancersamplesindicatedthatIR-AmRNAexpressionlevelsinthebreasttumorsweresimilartothelevelsobservedinadjacentnormalbreasttissues,whilethemRNAexpressionlevelsofIR-Bwerelargelydecreasedwhencomparedtotheadjacentnormalbreasttissues.OurobservationofanincreasedIR-A:IR-Bratioinbreastcancertissuerelativetonormalisconsistentwithwhathasbeenseenpreviously;however,ourresultsaredistinctinthatweidentifiedareductioninIR-Bexpressionasthekeytothisalteredratio.GiventhelinkbetweenalteredexpressionofIRisoformsandresistancetotheEGFRinhibitorgefitinib[24],itisclearthataberrantexpressionlevelsofIR-AandIR-Bmayhavesignificantfunctionaleffectsoncancerprogression.

ThefunctionalconsequencesofanalteredIR-A:IR-Bratiooncancerprogressionmayalsobemediated,inpart,bychangesinreceptordimerizationpatternsandthesubsequenteffectoncellsignalingpathways.AnincreaseintherelativeabundanceofIR-AmightsignificantlyincreaseformationofIGF1RandIR-Ahybrid

PLoSONE|www.plosone.org

7

receptors,IR-A/IR-Bheterodimers,andIR-Ahomodimersinbreastcancertissues.ThesechangesindimerizationpatternscouldresultinanincreasednumberofbindingsitesforIGFs,leadingtoincreasesinmitogenicsignalinginbreastcancercells[23],aswellaschangesingeneexpressionstronglyassociatedwithcellproliferation,metabolism,andDNArepair[22].OfthegenesalteredinresponsetoIGFstimulation,themRNAexpressionlevelsofproliferationgenesaremostcloselycorrelatedwithclinicaloutcomes[18],[25],[26].OurdataillustrateapositivecorrelationbetweentheIR-A:IR-BDCtdifferentialandtheproliferationscore.ThesedataalsoprovidesupportingevidencethattheincreasingIR-Aproportionmaycontributetothetumorproliferationphenotypeofbreastcancers.

Geneexpressionprofilingstudieshaveconsistentlyrevealedbiologicallydistinctbreastcancersubtypeswithdifferentprogno-sesandtreatmentresponses[27].Atthemolecularlevel,ER+/PR+andHER2-breastcancercanbefurtherdividedintoluminal-Aandluminal-Bsubtypes[19].Theluminal-BsubtypeischaracterizedbyacellproliferationsignaturethatincludestheexpressionofMKI67,CCNB1,andMYBL2,whichhavebeenassociatedwithtamoxifenresistance[21],[28].Theluminal-Asubtype,bycontrast,ischaracterizedbylowerproliferation,responsetohormonaltherapyandrelativelygoodprognosis.InthisstudywefoundapositivecorrelationbetweenIR-A:IR-BDCtdifferentialandtheexpressionofmultipleproliferationgenes,aswellasamoreprominentIR-A:IR-BDCtdifferentialinluminal-Bbreastcancersclassifiedaccordingtoconventionalmethods.Therefore,ourfindingssupporttheassertionthatthesignificantincreaseofIR-A:IR-Bratioinluminal-Bpatientsmayberelatedtothehighproliferationandpoorprognosisobservedforthissubtype.Theluminal-Bsubtypehasbeenclinicallyassociatedwithhighergrade,largertumorsize,positivelymphnodeinvolvement,increasedlymphovascularinvasion,andpoorerrelapse-freesurvival[29].Furthermore,breastcancerpatientswithluminal-Btumorsareatincreasedriskforrelapseanddeath,despiteendocrinetherapy[29].TheassociationofincreasedIR-A:IR-B

October2011|Volume6|Issue10|e26177

InsulinReceptorIsoformsinBreastCancer

ratiowiththeluminal-Bsubtypesuggeststhatluminal-Bpatientscouldpotentiallyrespondtotherapythatco-targetsIGF1RandIR-A.Moreover,ithighlightsthepotentialimportanceoftargetingIR-AinthissubgroupofpatientsandsuggeststhattherapeuticstrategiesdesignedtotargetthegrowthfactorsIGF1andIGF2themselves,maybemoresuccessfulthanstrategiestargetingtheIGF1Ronly.

Previousworkhasshownthatproteinlevelsofthetotalinsulinreceptorareelevatedinbreastcancertissuecomparedtonormalbreasttissue[30]andthatinsulinitselfmayaffecttumorprogressionbyactingthroughitsownreceptor[31].Specifically,Papaetal.[30]reportedtheaverageproteinexpressionleveloftheinsulinreceptorwaselevatedapproximatelysix-foldin159breastcancertissueswhencomparedto27normalbreasttissues.Theseresultsappeartocontradictourresults,whichsuggestthattotalinsulinreceptortranscriptlevelsarelowerinbreastcancertissuewhencomparedtonormalbreasttissues.However,inasubsequentmanuscriptfromthesamegroup[31],totalinsulinreceptorproteinlevelswereshowntobeelevatedin2of3breastcancercelllines,butnoquantitativedifferenceininsulinreceptormRNAcontentwasobservedinthesesamecelllines.Thissuggeststhatpost-transcriptionalmechanismscouldplayaroleinregulatingthetotalproteinlevelofinsulinreceptorincancer.Incontrast,relativeincreasesinthemRNAexpressionratioofIR-AandIR-Bhasbeenobservedindedifferentiatedthyroidcarcino-mas[32]andassociatedwithcharacteristicsofstemnessandthedevelopmentofthyroidcancer[33].Therefore,alterationsintheIR-A:IR-Btranscriptratioobservedinthisstudycouldsignifi-cantlyincreaseourunderstandingoftheroleofinsulinreceptor

isoformsinbreastcancer,independentofthetotalinsulinreceptorcontentinthesetumors.

Insummary,thespecificTaqManqRT-PCR-basedassaysthatwedevelopedcaneasilybeincorporatedintoroutineclinicalpracticeandmeasuretherelativeabundanceofIR-AandIR-BmRNAsinhumantissueswithhighsensitivityandspecificity.StudiesbasedontheseassaysmayprovidevalueinevaluatingthetherapeuticbenefitoftargetingbothIGF1RandIR-ApathwaysbyinhibitingthegrowthfactorsIGF1andIGF2,aswellaspotentiallyidentifyingpatientsmorelikelytorespondtoanyanti-IGFtherapyunderclinicaldevelopment.Theresultsreportedhereprovideapromisingfoundationtosupportthesefutureefforts.

SupportingInformation

TableS1SpecificityofIR-AandIR-Bassays.Assay

specificitywastestedbyperformingIR-A,IR-B,andIRqRT-PCRassayswithaserialdilutionofapproximately107to10copiesofplasmidDNAofIR-AandIR-B.CrossreactionsofeitherIR-AassaytoIR-BDNAorIR-BassaytoIR-ADNAwerenotobserved.(PDF)

AuthorContributions

Conceivedanddesignedtheexperiments:JHCMYCJVTLLRBJYY.Performedtheexperiments:JHCMKSBHJGMCABWZPB.Analyzedthedata:JHCMKSBHWZYY.Wrotethepaper:JHCMKSBHYYJV.

References

1.SeinoS,SeinoM,NishiS,BellGI(19)Structureofthehumaninsulinreceptorgeneandcharacterizationofitspromoter.ProcNatlAcadSciUSA86(1):114–8.

2.MollerDE,YokotaA,CaroJF,FlierJS(19)Tissue-specificexpressionoftwoalternativelysplicedinsulinreceptormRNAsinman.MolEndocrinol3(8):1263–9.

3.FrascaF,PandiniG,ScaliaP,SciaccaL,MineoR,etal.(1999)InsulinreceptorisoformA,anewlyrecognized,high-affinityinsulin-likegrowthfactorIIreceptorinfetalandcancercells.MolCellBiol19(5):3278–88.

4.MosthafL,VogtB,HaringHU,UllrichA(1991)AlteredexpressionofinsulinreceptortypesAandBintheskeletalmuscleofnon-insulin-dependentdiabetesmellituspatients.ProcNatlAcadSciUSA88(11):4728–30.

5.SestiG,MariniMA,TullioAN,MontemurroA,BorboniP,etal.(1991)Alteredexpressionofthetwonaturallyoccurringhumaninsulinreceptorvariantsinisolatedadipocytesofnon-insulin-dependentdiabetesmellituspatients.BiochemBiophysResCommun181(3):1419–24.

6.KellererM,SestiG,SefferE,Obermaier-KusserB,PongratzDE,etal.(1993)Alteredpatternofinsulinreceptorisotypesinskeletalmusclemembranesoftype2(non-insulin-dependent)diabeticsubjects.Diabetologia36(7):628–32.

7.PandiniG,VigneriR,CostantinoA,FrascaF,IppolitoA,etal.(1999)Insulinandinsulin-likegrowthfactor-I(IGF-I)receptoroverexpressioninbreastcancersleadstoinsulin/IGF-Ihybridreceptoroverexpression:EvidenceforasecondmechanismofIGF-Isignaling.ClinCancerRes5(7):1935–44.

8.VellaV,SciaccaL,PandiniG,MineoR,SquatritoS,etal.(2001)TheIGFsysteminthyroidcancer:Newconcepts.MolPathol(3):121–4.

9.PandiniG,MedicoE,ConteE,SciaccaL,VigneriR,etal.(2003)Differentialgeneexpressioninducedbyinsulinandinsulin-likegrowthfactor-IIthroughtheinsulinreceptorisoformA.JBiolChem278(43):42178–.

10.HaringHU,KellererM,MosthafL(1994)Modulationofinsulinreceptor

signalling:Significanceofalteredreceptorisoformpatternsandmechanismofhyperglycaemia-inducedreceptormodulation.Diabetologia37Suppl2:S149–.

11.SoosMA,SiddleK(19)Immunologicalrelationshipsbetweenreceptorsfor

insulinandinsulin-likegrowthfactorI.Evidenceforstructuralheterogeneityofinsulin-likegrowthfactorIreceptorsinvolvinghybridswithinsulinreceptors.BiochemJ263(2):553–63.

12.PollakM(2008)Insulinandinsulin-likegrowthfactorsignallinginneoplasia.

NatRevCancer8(12):915–28.

13.SoosMA,NaveBT,SiddleK(1993)ImmunologicalstudiesoftypeIIGF

receptorsandinsulinreceptors:Characterizationofhybridandatypicalreceptorsubtypes.AdvExpMedBiol343:145–57.

14.SoosMA,WhittakerJ,LammersR,UllrichA,SiddleK(1990)Receptorsfor

insulinandinsulin-likegrowthfactor-Icanformhybriddimers.characterizationofhybridreceptorsintransfectedcells.BiochemJ270(2):383–90.

15.UlanetDB,LudwigDL,KahnCR,HanahanD(2010)Insulinreceptor

functionallyenhancesmultistagetumorprogressionandconveysintrinsicresistancetoIGF-1Rtargetedtherapy.ProcNatlAcadSciUSA107(24):10791–8.

16.GaoJ,ChesebroughJW,CartlidgeSA,RickettsSA,IncognitoL,etal.(2011)

DualIGF-I/II-neutralizingantibodyMEDI-573potentlyinhibitsIGFsignalingandtumorgrowth.CancerRes71(3):1029–40.

17.WeigeltB,MackayA,A’hernR,NatrajanR,TanDS,etal.(2010)Breastcancer

molecularprofilingwithsinglesamplepredictors:Aretrospectiveanalysis.LancetOncol11(4):339–49.

18.PaikS,ShakS,TangG,KimC,BakerJ,etal.(2004)Amultigeneassayto

predictrecurrenceoftamoxifen-treated,node-negativebreastcancer.NEnglJMed351(27):2817–26.

19.PerouCM,SorlieT,EisenMB,vandeRijnM,JeffreySS,etal.(2000)

Molecularportraitsofhumanbreasttumours.Nature406(6797):747–52.

20.SotiriouC,PusztaiL(2009)Gene-expressionsignaturesinbreastcancer.

NEnglJMed360(8):790–800.

21.MarcomPK,IsaacsC,HarrisL,WongZW,KommarreddyA,etal.(2007)The

combinationofletrozoleandtrastuzumabasfirstorsecond-linebiologicaltherapyproducesdurableresponsesinasubsetofHER2positiveandERpositiveadvancedbreastcancers.BreastCancerResTreat102(1):43–9.

22.CreightonCJ,CasaA,LazardZ,HuangS,TsimelzonA,etal.(2008)Insulin-likegrowthfactor-Iactivatesgenetranscriptionprogramsstronglyassociatedwithpoorbreastcancerprognosis.JClinOncol26(25):4078–85.

23.BelfioreA,PandiniG,VellaV,SquatritoS,VigneriR(1999)Insulin/IGF-I

hybridreceptorsplayamajorroleinIGF-Isignalinginthyroidcancer.Biochimie81(4):403–7.

24.JonesHE,GeeJM,BarrowD,TongeD,HollowayB,etal.(2006)Inhibitionof

insulinreceptorisoform-Asignallingrestoressensitivitytogefitinibinpreviouslydenovoresistantcoloncancercells.BrJCancer95(2):172–80.

25.van’tVeerLJ,DaiH,vandeVijverMJ,HeYD,HartAA,etal.(2002)Gene

expressionprofilingpredictsclinicaloutcomeofbreastcancer.Nature415(6871):530–6.

26.ParkerJS,MullinsM,CheangMC,LeungS,VoducD,etal.(2009)Supervised

riskpredictorofbreastcancerbasedonintrinsicsubtypes.JClinOncol27(8):1160–7.

27.FanC,OhDS,WesselsL,WeigeltB,NuytenDS,etal.(2006)Concordance

amonggene-expression-basedpredictorsforbreastcancer.NEnglJMed355(6):560–9.

PLoSONE|www.plosone.org8October2011|Volume6|Issue10|e26177

InsulinReceptorIsoformsinBreastCancer

28.OhDS,TroesterMA,UsaryJ,HuZ,HeX,etal.(2006)Estrogen-regulated

genespredictsurvivalinhormonereceptor-positivebreastcancers.JClinOncol24(11):1656–.

29.CheangMC,ChiaSK,VoducD,GaoD,LeungS,etal.(2009)Ki67index,

HER2status,andprognosisofpatientswithluminalBbreastcancer.JNatlCancerInst101(10):736–50.

30.PapaV,PezzinoV,CostantinoA,BelfioreA,GiuffridaD,etal.(1990)Elevated

insulinreceptorcontentinhumanbreastcancer.JClinInvest86(5):1503–10.

31.MilazzoG,GiorginoF,DamanteG,SungC,StampferMR,etal.(1992)Insulin

receptorexpressionandfunctioninhumanbreastcancercelllines.CancerRes52:3924–3930.

32.VellaV,PandiniG,SciaccaL,MineoR,VigneriR,etal.(2002)Anovel

autocrineloopinvolvingIGF-IIandtheinsulinreceptorisoform-Astimulatesgrowthofthyroidcancer.JClinEndocrinMetab87:245–2.

33.MalaguarneraR,FrascaF,GarozzoA,GianiF,PandiniG,etal.(2001)Insulin

receptorisoformsandinsulin-likegrowthfactorreceptorinhumanfollicularcellprecursorsfrompapillarythyroidcancerandnormalthyroid.JClinEndocrinMetab96:766–774.

PLoSONE|www.plosone.org9October2011|Volume6|Issue10|e26177