Genome-wide DNA methylation pro

EPIDEMIOLOGY

Genome-wideDNAmethylationprofilingrevealsparity-associatedhypermethylationofFOXA1

SagarGhosh•FeiGu•Chou-MiinWang•Chun-LinLinJosephLiu•HowardWang•PeterRavdin•YanfenHu•TimH.M.Huang•RongLi

•

Received:8July2014/Accepted:9September2014/Publishedonline:19September2014ÓSpringerScience+BusinessMediaNewYork2014

AbstractEarlypregnancyinwomenbytheageof20isknowntohaveaprofoundeffectonreductionoflifelongbreastcancerriskascomparedtotheirnulliparouscoun-terparts.Additionalpregnanciesfurtherenhancethepro-tectionagainstbreastcancerdevelopment.Nationwidetrendofdelayedpregnancymaycontributetotherecentlyreportedincreaseintheincidenceofadvancedbreastcanceramongyoungwomeninthiscountry.Theunder-lyingmechanismfortheparity-associatedreductionofbreastcancerriskisnotclearlyunderstood.Thepurposeofthecurrentstudyistousewhole-genomeDNAmethylationprofilingtoexploreapotentialassociationbetweenparityandepigeneticchangesinbreasttissuefromwomenwithearlyparityandnulliparity.Breasttissuewascollectedfromage-matchedcancer-freewomenwithearlyparity(age\\20;n=15)ornulliparity(n=13).Themethyl-CpGbindingdomain-basedcapture-sequencingtechnology

wasusedforwhole-genomeDNAmethylationprofiling.Potentialparity-associatedhypermethylatedgeneswerefurtherverifiedbylocus-specificpyrosequencing,usinganexpandedcohortofparous(n=19)andnulliparous(n=16)womenthatincludedtheinitialsamplesusedintheglobalanalysis.Ourstudyidentifiedsixgenesthatarehypermethylatedintheparousgroup(P\\0.05).Pyrose-quencingconfirmedparity-associatedhypermethylationatmultipleCpGislandsoftheFOXA1gene,whichencodesapioneerfactorthatfacilitateschromatinbindingofestrogenreceptora.Ourworkidentifiesseveralpotentialmethyla-tionbiomarkersforparity-associatedbreastcancerriskassessment.Inaddition,theresultsareconsistentwiththenotionthatparity-associatedepigeneticsilencingofFOXA1contributestolong-termattenuationoftheestro-genicimpactonbreastcancerdevelopment.KeywordsParityÁFOXA1ÁDNAmethylationÁBreastcancerriskreduction

ElectronicsupplementarymaterialTheonlineversionofthisarticle(doi:10.1007/s10549-014-3132-2)containssupplementarymaterial,whichisavailabletoauthorizedusers.

S.GhoshÁF.GuÁC.-M.WangÁC.-L.LinÁJ.LiuÁY.HuÁT.H.M.HuangÁR.Li(&)

DepartmentofMolecularMedicine,CancerTherapyand

ResearchCenter,UniversityofTexasHealthScienceCenteratSanAntonio,SanAntonio,TX78229,USAe-mail:lir3@uthscsa.edu

PresentAddress:F.Gu

Children’sHospitalBoston,320LongwoodAve,Boston,MA02115,USA

H.WangÁP.Ravdin

DepartmentofMedicine,CancerTherapyandResearchCenter,UniversityofTexasHealthScienceCenteratSanAntonio,SanAntonio,TX78229,USA

Introduction

BesidestheBRCA1/2breastcancersusceptibilitygenes,severalendocrine-relatedfactorsareknowntosignificantlyinfluencetheriskforbreastcancer.Inparticular,womenwhohavetheirfirstfull-termpregnancybeforetheageof20haveapproximatelyhalfoftheriskfordevelopingbreastcancerascomparedtotheirnulliparouscounterparts,andadditionalpregnanciesfurtheraccentuatetheparity-associatedprotectionagainstbreastcancer[1].Thislife-longrisk-reducingeffectofearlyparityisincontrasttotherisk-enhancingeffectoflatepregnancyafterage35[2–4].Anumberofstudiesinhumanandrodentshavedemon-stratedthestrongandlifelongprotectiveroleofearly

123

654pregnancyonbreastcancer[1,5,6].Whiletheexactunderlyingmechanismbywhichearlyparityreduceslife-longbreastcancerincidenceisnotknown,itismostlikelyduetoenduringbiologicalchangesinthebreasttissue.Severalmodelshavebeenproposedbasedonparity-asso-ciatedchangesinbreastepithelialcellsanditssurroundingstromalcompartments[1,7].Forexample,ithasbeenhypothesizedthatearlyparity-induceddifferentiationofmammaryepithelialcellscouldrenderthemmoreresistanttooncogenesis[8].Inasimilarvein,earlyparitymaydecreasethepoolsofmammarystem/progenitorcells,thusdecreasingthenumberofputativecelloforiginforbreasttumors[9,10].Asearlyparitypredominantlyreducesriskofestrogenreceptora(ERa)-positivebreastcancer[11],itisalsoconceivablethatattenuationoftheestrogen-relatedpathwayscouldcontributetoparity-mediatedriskreduc-tion[12].Inaddition,therolesofhormonesandhormone-sensingcellsintheearlyparity-basedprotectiononbreastcancerhasbeenrecentlydocumented[13].Betterunder-standingofthecellularandmolecularbasisforthislong-standingbiologicalphenomenonwillgoalongwayinriskassessmentandbreastcancerprevention.

ERaisasite-specifictranscriptionfactorthatplaysakeyroleinnormalbreastductaldevelopmentandluminalbreastcancer[14].Traditionally,estrogen-stimulatedERaalonewasthoughttobesufficienttobindtothecognateestrogen-responsiveenhancersforhormone-stimulatedtranscrip-tionalactivation.However,wholegenome-basedstudiesinbreastcancercellsindicatethatFOXA1,anothersite-spe-cifictranscriptionfactorimportantinnormalandbreastcancerdevelopment[15],tendstoco-occupyERa-boundtranscriptionalenhancers[16,17].Importantly,ithasbeendemonstratedthatFOXA1servesasapioneerfactorthatfacilitatesERabindingtocompactedchromatinDNA.WithoutFOXA1,ERacannotbindtothecorrespondingenhancerseveninthepresenceofestrogens,andconse-quently,estrogen-mediatedtranscriptionandbreastcancercellproliferationareabrogated.MorerecentstudiesfurtherunderscoretheclinicalrelevanceoftheFOXA1–ERa–DNAinteractioncomplex[18,19].Forexample,itwasshownthatFOXA1iscapableofreprogrammingofERachromatinbindingandthisFOXA1-mediatedeventcorrelateswiththeclinicaloutcomeinbreastcancer[19,20].Furthermore,breastcancerrisk-associatedsinglenucleotidepolymor-phisms(SNPs)arefoundintheFOXA1/ERa-occupiedenhancers,whichalterFOXA1chromatinbindingandexpressionofFOXA1/ERadownstreamtargetgenes[21].ThesefindingshighlightthebiologicalandpathologicalimpactofthecooperativechromatinbindingofFOXA1-ERaonbreastcancerdevelopmentandprogression.

Inthecurrentstudy,wesoughttouncoverDNAepi-geneticsignaturesthatareassociatedwithbreasttissueofwomenwithearlyparity.Bycombininggenome-wide

123

BreastCancerResTreat(2014)147:653–659

MBDCap-sequencingandlocus-specificpyrosequencing,ourworkidentifieddistinctCpGislandsintheFOXA1genethatarepreferentiallyhypermethylatedintheearly-paritygroup.OurfindingsdrawattentiontosustainedFOXA1genesilencingandreducedestrogenactionsasapotentialcontributingfactortoearlyparity-associatedreductionofbreastcancerrisk.

MaterialsandmethodsClinicalspecimens

Cancer-freebreasttissuewasprocuredfromwomenundergoingcosmeticreductionmammoplasty,followingaprotocolapprovedbytheInstitutionalReviewBoardattheUniversityofTexasHealthScienceCenteratSanAntonio.Alldonorssignedawrittenconsentformauthorizingtheuseofthespecimensforbreastcancer-relatedlaboratoryinvestigations.Theparticipantsalsofilledoutaquestion-naireconcerningpersonalmedicalhistoryandhealth-anddiet-relatedbehaviors.SupplementalTable1providesmoredetailedinformationabouttheage-matchedparousandnulliparousgroupsusedinthecurrentstudy.DNAisolationandMBDCap-seq

MBDCaplibrariesforsequencingwerepreparedfollowingstandardprotocolsfromIllumina(SanDiego,CA).Briefly,genomicDNAwaspreparedfromfrozentissueusingQiaAmpDNAextractionkit(Qiagen,USA)andsubse-quentlyfragmentedbysonicationtoreachanaveragesizeof250basepairs(bp).MethylatedDNAfragmentswerecapturedandelutedusingMethylMinerMethylatedDNAEnrichmentKit(Invitrogen,USA),followingstandardprotocolfromthemanufacturer.MBDCaplibrariesweresequencedusingtheIlluminaGenomeAnalyzerII(GAII)permanufacturer’sinstructions.Sequencingwasperformedupto25cyclesformappingtothehumangenomerefer-encesequence.Imageanalysisandbasecallingwereper-formedusingthestandardIlluminasoftware.BioinformaticanalysisofMBDCap-seqdata

Reads(upto50bp)weremappedtothehumanreferencegenome(hg18)usingtheBWAalgorithm,withuptotwobase-pairmismatches.Theuniquelymappedreadswereusedforadditionallinearnormalizationanddifferentialmethylationanalysisaspreviouslydescribed[22].NRead;i¼

URead;i

NU=10^ðINTðlog10NUÞÞ

BreastCancerResTreat(2014)147:653–659Table1Potential

hypermethylatedgenesintheearlyparousgroup

GenesymbolFOXA1GRIK5OLIG3MGST3BCAR3FOXB2

GeneID3,1692,901167,82,2598,412442,425

Chr#1466119

Regionstart371302401019495811378532241638630739391597378820390

Regionend371382401019575811378612241638710739392397378828390

Foldchange-0.47-0.47-0.57-0.6-0.71-0.73

655Pvalue0.03560.0470.00550.04620.03360.0446

Themethylationlevelwascalculatedbyaccumulatingthenumberofreads.NRead,iisthenumberofnormalizedreadsattheithbin;URead,iisthenumberofuniquelymappedreadsattheithbin,andNUisthenumberoftotaluniquelymappedreads.‘‘INT’’roundstheelementtothenearestintegerstowardminusinfinity,‘‘^’’meansthepoweroperator.

AP

MR;G¼GR;G

S;R¼ðbsþ0;bsþ1;...;bsþmÞ

GAregionofmethylationlevelwasrepresentedbytheaverageofthenormalizeduniquereads.ComparisonofgroupAandB(G=AorB),theaveragemethylationlevel(AVGR,G)wascalculatedseparatelyattwogroupsinagivenregionR(whichincludesmbinsize,andstartatthesthbin).ThenumberofsampleisSAforgroupA,andSBforgroupB.AVGR,GmeanstheaveragemethylationlevelofgroupGattheRregion.MR,GisthemethylationlevelsofeachsampleofgroupGattheRregion.Pyrosequencingassay

ThepyrosequencingtargetsequenceandlocationineachselectedCpGsiteforFOXA1werelistedinSupplementalFigure1.Analysisofeachsiteinvolvedoneforwardandonebiotin-labeledreversePCRprimer,aswellasonesequencingprimerforpyrosequencing.GenomicDNAwasbisulfite-convertedusingEZ-96DNAMethylationkit(ZYMOResearch,PND5001).Twentyngofbisulfite-convertedDNAwasusedforPCRamplificationandtheproductswereverifiedforsizeandoverallqualitybygelelectrophoresis.ThePCRconditionswereasfollows:95°Cfor5min,followedby50cyclesof95°Cfor1min,60°Cfor1min,72°Cfor1min,then72°Cfor7minusingtheTaqDNApolymerase(AppliedBiosystem,N12338).

Forpyrosequencing,10lLofthePCRproductwasmixedwithstreptavidin-coatedagarosebeads(GEHealth-careUK)andprocessedthroughtheQiagenVacuumPrepToolforpurificationofthebiotinylatedsingle-standDNAfragments.PyroMarkGoldQ96CDTReagentswereusedforsequencingreactions.Theautomatedpyrosequencinginstrument,PyroMarkQ96MD(Qiagen),wasemployedforthesequencing-by-synthesismethodtodetectthe

methylationstatusofeachCpGsiteinaspecificregionusingthesequencingprimerat500nM.QuantificationofeachCpGsitewasperformedusingthemethylationSoftwarePyroMarkCpG1.0software.Thebuilt-ininternalqualitycontrolforbisulfitetreatmentandnon-specificbackgroundwassetat6.5%.MethylatedDNA(UniversalDNAwithmethylatedenzymetreatment)servedasthepositivecontrol.Statisticalanalysis

Welchttestandtheanalysisofvariance(ANOVA)wereappliedtodeterminewhetherthepercentagesofCpGislandsweredifferentacrossgroupsinvariousfactorsofinterest.Welchttestwasusedtoexaminewhethertwogroupsinafactorweresignificantlydifferentfromeachother.Itcomparedthemeanvaluesoftwogroupswiththeassumptionthatthevarianceswereunequalinthegroups.Pvaluelessthan0.05suggestedthattherewasastatisti-callydifferenceintwogroups.ANOVAwasconductedtoanalyzegroupdifferenceswhenthereweremorethantwogroupsinafactor(ethnicityinthecurrentstudy).Varianceinafactorwaspartitionedintotwocomponents:variancebetweengroupsandvariancewithingroups.FtestwasusedinANOVAtoexaminetwocomponentsbycompar-ingtheratioofvariancebetweengroupsandvariancewithingroupstoacriticalvalue.WhenPvalueislessthan0.05,theresultimpliesthatatleastonegroupinthefactorisdifferentfromtheothers.

Resultsanddiscussion

MBDCap-seq,ahighthroughputtechnologyforsurveyinggenome-wideDNAmethylationpatterns,combinescap-turingofmethylatedgenomicDNAbythemethyl-CpG-bindingdomain(MBD)ofMeCP2withnext-generationsequencing.TherobustprocedureallowscomprehensiveandunbiaseddetectionofDNAmethylationacrosstheentiregenome.Inthecurrentstudy,weappliedthistech-nologytocomparethemethylationstatusinthebreasttissueofwomenwithearlyparityversusnulliparity.Tothisend,wecollectedbreasttissuefromtwoage-matchedcohortsofwomenwhounderwentelectivereductionmammoplasty(SupplementalTable1).Thefirstgroup

123

656A

Breast tissue

DNA extracted and sheared

CHBiotin3CH3CHMBD3beadmagnet

Streptavidin

GST-MBD pull-down

Next-generation sequencing

Cohort 1Cohort 2

Differential methylation between the two cohorts

B

Early ParousNulliparous

hypermethylation

enrooitCal GyhptCemhypomethylation

enrooihtSal yGhtpeCmTSS

1kb

TSS

1kb

Fig.1Methylationprofilingofearlyparousandnulliparouswomen.aSchematicoutlineoftheexperimentalprocedureofMBDCap-seq.DNAfromthetissuesamplesweresubjectedtoMBD2capture.Afterlibraryconstructionandnext-generationsequencing,datawereanalyzedtoidentifydifferentiallymethylatedloci.bMethylationlevelsofloci,4kbupstreamanddownstreamfromthetranscriptionstartsiteofthedifferentiatedmethylatedgenesinearlyparousandnulliparoussamples

(n=15;averageageof37.2yearsold)consistedofthosewhohadthefirstfull-termpregnancybytheageof19yearsold.Thesecondcohort(n=13;averageageof37.8yearsold)wasfromwomenwhohadnohistoryofchildbearing.Thesameageframesofthesetwocohortsallowedustoexcludeanyage-relatedepigeneticdifferences.

GenomicDNAwasisolatedfromthefrozenbreasttis-sue.MethylatedDNAfragmentswereboundtotheGST-MBDresin,eluted,andsubjectedtoMBDCap-seqbytheIlluminaHiSeq2000sequencingsystem(Fig.1a).Atotalof42billionbpfromthe28clinicalsampleswere

123

BreastCancerResTreat(2014)147:653–659

processed,and75%ofthesequencedfragmentsweremappedtouniquegenomelocations(SupplementaryTable2).AsDNAmethylationtendstobefoundinGC-richregions,aminimumoftenmillionuniquereadsisusuallyconsideredsufficienttocoveradequatesequencedepthforwhole-genomemethylationprofiling[22].Upondatanormalization,weconductedapairwisecomparisonwithinan8kbwindowtoidentifydifferentiallymethylatedCpGislandlociintheearlyparousandnulliparoussamples.

Weanalyzedatotalof13,081knownpromoterCpGislandsandfoundthatonlysixgenes(0.046%)weresig-nificantlyhypermethylatedinparoussamplesrelativetothenulliparousones(Fig.1b).TheyareBCAR3,FOXA1,FOXB2,GRIK5,MGST3,andOLIG3(Table1).Twooftheparity-associatedhypermethylatedgenesexhibitedmethyl-ationoftheCpGcore(BCAR3andOLIG3),whereastherestshowedCpGshoremethylation(right,leftorbothsides).Incontrast,afargreaternumberofgenes(211or1.62%)werehypomethylatedintheparoussamples(Fig.1b,Supple-mentalTable2).Ofthehypomethylatedgenes,104exhib-itedCpGcoremethylation,whereas107showedCpGshoremethylation.Noknownbreasttumorsuppressorgeneswereidentifiedfromthehypomethylatedgenes.

Amongthesixhypermethylatedgenesintheparousgroup,FOXA1encodesasite-specifictranscriptionfactorthatactsasapioneerfactortopromoteanddictatechro-matinbindingofERa[16,17].Itsamplificationisassoci-atedwithdevelopmentofmultiplecancertypesincludingbreastcancer[23].TheroleofFOXB2,anotherFOXfamilygene,incancerisnotknown.BCAR3encodesasignalingmoleculethatplaysanimportantroleinbreastcancerinvasionandresistancetoanti-estrogentherapies[24,25].Therefore,itisconceivablethatepigeneticchangesattheselocicouldaffecthormonalresponsivenessofbreastepi-thelialcellsandtheirpropensityfortumorigenesis.

ToconfirmtheMBDCap-seqdata,weconductedpy-rosequencinganalysisoftheFOXA1,FOXB2,andBCAR3loci,usingtheoriginalcohortsfortheMBDCap-seqplusfouradditionalearlyparousandthreenulliparoussamplessubsequentlyprocured(SupplementalTable1).MBD-seqandpyrosequencingaretwodifferentandmutuallycom-plementarymethodsfordetectingDNAmethylation.WhileMBD-seqassessesgenome-wideDNAmethylationinarelativelylowresolution,pyrosequencingisparticuarlysuitablefordeterminingDNAmethylationlevelatthesinglebase-pairresolutioninarelativelysmallregion.PyrosequencingprimersweredesignedforthespecificCpGislandsthatdisplayedparity-associatedhypermethy-lationasrevealedbyMBDCap-seq(SupplementalFig-ure1–3).ThemethylationstatusoftheindividualCpGsitesatthesegenelociarepresentedinFig.2(FOXA1)andSupplementalFigure4(FOXB2andBCAR3).Amongthe

BreastCancerResTreat(2014)147:653–659Fig.2Pyrosequencingvalidationoftheparity-associatedhypymethylationatFOXA1.aGraphicalrepresentationofthe%methylationatvariousCpGsitesintheFOXA1geneusingthreesetsofprimers(FOXA1a,FOXA1bandFOXA1c).SeeSupplementalFigure1forthegenomiclocationsoftheCpGsites.Theboxescoversamplevaluesfromthe25percentiles(inthebottom)to75percentiles(inthetop),thewhiskersextendingfromtheboxes

stretchtotheminimumandthemaximumvaluesofthesamples.WelchttestswereusedtodeterminethePvaluesbetweentwogroups.bGraphicrepresentationofthe

methylationpatterninthefirstsixCpGsitesanalyzedfortheFOXA1locus.Thecolor

denotationisshownonthetop

657

A5kbB

0-5%5-9.99%10-14.99%15-19.99%

TSS

20-24.99%25-29.99%>30%

*1 2 3 4 5 6

FOXA1a

% methylationP=.03

Parous

Nulliparous

BSC-02BSC-16BSC-23BSC-26BSC-29BSC-34BSC-38BSC-57BSC-61BSC-BSC-67BSC-74BSC-76BSC-78BSC-19BSC-28BSC-39BSC-41BSC-53BSC-04BSC-12BSC-18BSC-27BSC-31

CpG site:123456

FOXA1b

% methylationCpG site:7

101112

NulliparousEarly ParousFOXA1c

% methylationBSC-32BSC-33BSC-47BSC-69BSC-71BSC-73BSC-35BSC-36BSC-48BSC-50BSC-54

P=.05

CpG site:131415161718

18CpGsitesinterrogatedattheFOXA1locus,sites2and18werefoundtobesignificantlyhypermethylatedintheparousgroup(P\\or=0.05,Fig.2),thusconfirmingthegenome-wideMBDCap-seqdata.Incontrast,nostatisti-callysignificant,parity-associateddifferencewasdetectedbypyrosequencingattheFOXB2orBCAR3loci,althoughseveralCpGsitesatthesetwogenelocidisplayedatrendofhypermethylationintheparousgroup(SupplementalFigure4).Moresampleanalysiswillbeneededtoresolvethisdiscrepancybetweenthetwomethods.

Ashigherlevelsofmethylationareusuallyassociatedwithsilencedtranscription,ourresultraisesthedistinctpossibilitythatparity-associatedepigeneticsilencingofFOXA1mayresultinattenuatedERafunctionandreducedriskoftumorigenesisforthecorrespondingbreastepithe-lialcells.Onecaveatinthecurrentworkwasourinability

123

658ofdirectlymeasuringtheFOXA1mRNAlevelsinthesamebreasttissuesamplesusedforthemethylationstudy,whichwaslikelyduetothesub-optimalpreservationoftheclinicalsamples.WearealsoawareofthepossibilitythatepigeneticchangesandERafunctionscouldbeinfluencedbymultipleknownriskfactorsincludingmenopausalsta-tus,bodyweightindex(BMI),comorbiditywithmetabo-lism-relateddisease,andethnicity,etc.However,usingthecurrentcohortofrelativelysmallsamplesize,wedidnotfindanysignificantcorrelationbetweenthemethylationpatternatthetwoCpGsites(#2and#18)ofFOXA1andtheotherconfoundingfactors(SupplementalTable3).Lastly,thesizeofthecohortsusedinthecurrentstudyisrelativelysmall.Itisthereforeimportanttovalidatethefindingsinfuturewithlargerpoolsofclinicalsamples.Uponconfirmation,thehypermethylationsitesidentifiedinthecurrentstudycouldserveasquantifiablemetricsforassessingparity-associatedbreastcancerrisk.

Delayedpregnancyisanationwidetrendamongwomenofchildbearingage.Thiscouldbeoneofthefactorscon-tributingtotherecentlyreportedincreaseintheincidenceofadvancedbreastcanceramongyoungwomen(age25–39)inthiscountryoverthepastthreedecades[26].Conversely,earlyandmulti-parityamongHispanicwomencouldpartlyaccountfortherelativelylowbreastcancerincidenceinthispopulation(www.cancer.org).Infact,tissueprocurementinourpilotstudywasfacilitatedbytherelativelyhighpercentageofwomenwithearly/multi-parityinSanAntonioandSouthTexas.Theaverageageoffirst-timemothersinourcurrentcohortis21.6years,ascomparedtonationalaverageof25years.Inaddition,closeto40%offirstbirthsinourcohortaretomothersunderage20,whereasonly21%aresonationwide.EstablishingacausalrelationshipbetweenFOX1Ameth-ylationandbreastcancerriskinfutureworkmayinformtargetedwaysofmimickingtheprotectiveeffectofearlyparityamonglate-parousandnulliparouswomen.

AcknowledgmentsThisworkwassupportedbytheNIHgrantstoR.L.(CA161349)andT.H.M.H(U54CA113001),Y.H.(CA118578),andaDepartmentofDefenseBreastCancerResearchgranttoR.L(W81XWH-14-1-0129).WealsothankthegeneroussupportbytheCancerTherapyandResearchCenteratUniversityofTexasHealthScienceCenteratSanAntonio(P30CA054174).ConflictofinterestTheauthorsdeclarethattheyhavenoconflict

ofinterest.

References

1.BrittK,AshworthA,SmalleyM(2007)Pregnancyandtheriskofbreastcancer.EndocrRelatCancer14:907–933

2.SchedinP(2006)Pregnancy-associatedbreastcancerandmetastasis.NatRevCancer6:281–291

123

BreastCancerResTreat(2014)147:653–659

3.PolyakK(2006)Pregnancyandbreastcancer:theothersideofthecoin.CancerCell9:151–153

4.HaricharanS,DongJ,HeinS,ReddyJP,DuZ,ToneffM,HollowayK,HilsenbeckSG,HuangS,AtkinsonRetal(2013)Mechanismandpreclinicalpreventionofincreasedbreastcancerriskcausedbypregnancy.eLife2:e00996

5.MacMahonB,ColeP,LinTM,LoweCR,MirraAP,RavniharB,SalberEJ,ValaorasVG,YuasaS(1970)Ageatfirstbirthandbreastcancerrisk.BullWorldHealthOrgan43:209–221

6.AlbrektsenG,HeuchI,HansenS,KvaleG(2005)Breastcancerriskbyageatbirth,timesincebirthandtimeintervalsbetweenbirths:exploringinteractioneffects.BrJCancer92:167–1757.RussoJ,BaloghGA,HeulingsR,MailoDA,MoralR,RussoPA,SheriffF,VanegasJ,RussoIH(2006)Molecularbasisofpreg-nancy-inducedbreastcancerprotection.EurJCancerPrev15:306–342

8.RussoJ,MoralR,BaloghGA,MailoD,RussoIH(2005)Theprotectiveroleofpregnancyinbreastcancer.BreastCancerRes7:131–142

9.SilvaramanL,MedinaD(2002)Hormone-inducedprotectionagainstbreastcancer.JMammaryGlandBiolNeoplasia7:77–9210.SiwkoSK,DongJ,LewisMT,LiuH,HilsenbeckSG,LiY

(2008)Evidencethatanearlypregnancycausesapersistentdecreaseinthenumberoffunctionalmammaryepithelialstemcells–implicationsforpregnancy-inducedprotectionagainstbreastcancer.StemCells26:3205–3209

11.ColditzGA,RosnerBA,ChenWY,HolmesMD,HankinsonEE

(2004)Riskfactorsforbreastcanceraccordingtoestrogenandprogesteronereceptorstatus.JNatlCancerInst96:218–22812.ThordarsonG,JinE,GuzmanRC,SwansonSM,NandiS,Tal-amantesF(1995)Refractorinesstomammarytumorigenesisinparousrats:isitcausedbypersistentchangesinthehormonalenvironmentorpermanentbiochemicalalterationsinthemam-maryepithelia?Carcinogenesis16:2847–2853

13.Meier-AbtF,MilaniE,RoloffT,BrinkhausH,DussS,Meyer

DS,KlebbaI,BalwierzPJ,vanNimwegenE,Bentires-AljM(2013)ParityinducesdifferentiationandreducesWnt/Notchsignalingratioandproliferationpotentialofbasalstem/progeni-torcellsisolatedfrommousemammaryepithelium.BreastCancerRes15:R36

14.DerooBJ,KorachKS(2006)Estrogenreceptorsandhuman

disease.JClinInvest116:561–570

15.BernardoGM,KeriRA(2012)FOXA1:atranscriptionfactor

withparallelfunctionsindevelopmentandcancer.BiosciRep32:113–130

16.CarrollJS,LiuS,BrodskyAS,LiW,MeyerCA,SzaryAJ,

EeckhouteJ,ShaoW,HestermannEV,GeistlingerTRetal(2005)Chromosome-widemappingofestrogenreceptorbindingrevealslong-rangeregulationrequiringtheforkheadproteinFoxA1.Cell122:33–43

17.AugelloMA,HickeyTE,KnudsenKE(2011)FOXA1masterof

steroidreceptorfunctionincancer.EMBOJ30:3885–3418.BarbieriCE,BacaSC,LawrenceMS,DemichelisF,BlattnerM,

TheurillatJP,WhiteTA,StojanovP,VanAllenE,StranskyNetal(2012)ExomesequencingidentifiesrecurrentSPOP,FOXA1andMED12mutationsinprostatecancer.NatGenet44:685–6

19.WattersRJ,BenosPV,OesterreichS(2012)Tobindornotto

bind–FoxA1determinesestrogenreceptoractioninbreastcancerprogression.BreastCancerRes14:312

20.HurtadoA,HolmesKA,Ross-InnesCS,SchmidtD,CarrollJS

(2010)FOXA1isakeydeterminantofestrogenreceptorfunctionandendocrineresponse.NatGenet43:27–33

21.Cowper-SallariR,ZhangX,WrightJB,BaileySD,ColeMD,

EeckhouteJ,MooreJH,LupienM(2012)Breastcancerrisk-

BreastCancerResTreat(2014)147:653–659

associatedSNPsmodulatetheaffinityofchromatinforFOXA1andaltergeneexpression.NatGenet44:1191–1198

22.HuangTT,GonzalesCB,GuF,HsuYT,JadhavRR,WangCM,

ReddingSW,TsengCE,LeeCC,ThompsonIMetal(2013)Epigeneticderegulationoftheanaplasticlymphomakinasegenemodulatesmesenchymalcharacteristicsoforalsquamouscellcarcinomas.Carcinogenesis34:1717–1727

23.KatohM,IgarashiM,FukudaH,NakagamaH(2013)Cancer

geneticsandgenomicsofhumanFOXfamilygenes.CancerLett328:198–206

24.vanAgthovenT,SieuwertsAM,Meijer-vanGelderME,Look

MP,SmidM,VeldscholteJ,SleijferS,FoekensJA,DorssersLC

659

(2009)Relevanceofbreastcancerantiestrogenresistancegenesinhumanbreastcancerprogressionandtamoxifenresistance.JClinOncol27:542–549

25.WilsonAL,SchrecengostRS,GuerreroMS,ThomasKS,Bouton

AH(2013)Breastcancerantiestrogenresistance3(BCAR3)promotescellmotilitybyregulatingactincytoskeletalandadhesionremodelingininvasivebreastcancercells.PLoSOne8:e65678

26.JohnsonRH,ChienFL,BleyerA(2013)Incidencerateofbreast

cancerinyoungwomen–reply.JAMA309:2435–2436

123