lncRNA-miRNA-ceRNA网络

SUBJECTAREAS:

GASTRICCANCERCANCERGENOMICS

LongnoncodingRNA

associated-competingendogenousRNAsingastriccancer

TianXia*,QiLiao,XiaomingJiang,YongfuShao,BingxiuXiao,YangXi&JunmingGuo

DepartmentofBiochemistryandMolecularBiology,ZhejiangProvincialKeyLaboratoryofPathophysiology,NingboUniversitySchoolofMedicine,Ningbo,Zhejiang,315211,China.

Received7May2014Accepted30July2014Published15August2014

Correspondenceandrequestsformaterialsshouldbeaddressedto

J.G.(junmingguo@

yahoo.com)

*Currentaddress:YuyaoEducationBureau,Yuyao,Zhejiang,315400,

China.

SomelongnoncodingRNAs(lncRNAs)playimportantrolesintheregulationofgeneexpressionbyactingascompetingendogenousRNAs(ceRNAs).However,therolesoflncRNAassociated-ceRNAsinoncogenesisarenotfullyunderstood.Here,basedonlncRNAmicroarraydataofgastriccancer,bioinformaticalgorithmmiRcodeandmicroRNA(miRNA)targetsdatabaseTarBase,wefirst

constructedanlncRNA-miRNA-mRNAnetwork.Then,weconfirmeditbydataofsixtypesofothercancerincludingheadandnecksquamouscellcarcinoma,prostatecancer,papillarythyroid

carcinoma,pituitarygonadotropetumors,ovariancancer,andchroniclymphocyticleukemia.Theresultsshowedaclearcancer-associatedceRNAnetwork.EightlncRNAs(AC009499.1,GACAT1,GACAT3,H19,LINC00152,AP000288.2,FER1L4,andRP4-620F22.3)andninemiRNAs(miR-18a-5p,miR-18b-5p,miR-19a-3p,miR-20b-5p,miR-106a-5p,miR-106b-5p,miR-31-5p,

miR-139-5p,andmiR-195-5p)wereinvolved.Forinstance,throughitsmiRNAresponseelements(MREs)tocompeteformiR-106a-5p,lncRNA-FER1L4regulatestheexpressionofPTEN,RB1,RUNX1,VEGFA,CDKN1A,E2F1,HIPK3,IL-10,andPAK7.Furthermore,cellularexperimentalresultsindicatedthatFER1L4-smallinterferingRNA(siRNA)simultaneouslysuppressedFER1L4andRB1mRNAlevel.TheseresultssuggestthatlncRNAsharborMREsandplayimportantrolesinpost-transcriptionalregulationincancer.

icroRNAs(miRNAs)playimportantrolesingeneexpressionregulation1.EachmiRNAsmayrepressuptohundredsoftranscripts,whileeachtranscriptmaybetargetedbymultiplemiRNAs2.Theirregulatorynetworksparticipateinavarietyofbiologicalprocesses,includingdevelopment,oncogenesisandtumor

metastasis3–5.

SalmenaandcolleaguespreviouslyproposedacompetingendogenousRNA(ceRNA)hypothesis6,whichwassupportedbynumerousevidences7–11.Thehypothesisdescribedacomplexpost-transcriptionalregulatorynetworkmediatedbymiRNAs:bysharingoneormoremiRNAresponseelements(MREs),protein-codingandnoncodingRNAscompeteforbindingtomiRNAsandthenregulateeachother’sexpression(Fig.1).Later,morestudiesprovidedconvincingevidencesforthishypothesis.Pandolfigroupfocusedonphosphataseandtensinhomologue(PTEN)anditsceRNAs12,13.TheydemonstratedtheexistenceofceRNAinteractionamongmRNAsinvitroandinvivo.Sumazinetal.analysedgeneexpressiondatainglioblastoma,andfoundmorethan7,000transcriptsactingasceRNAs14.FurtherstudyshowedthatceRNAsarenotlimitedtomRNAs.Linc-MD1,alongnoncodingRNA(lncRNA),regulatesmyoblastdifferentiationbycompetingforbindingtomiR-133andmiR-13515.

ToannotatetheassociationsbetweennoncodingRNAsanddiseases,severaldisease-associatedbioinformaticmethodsanddatabasesincludingLncRNADisease16,HMDD17anddbDEMC18havebeendeveloped.Otherwise,computationaltoolsformiRNA-targetinteractionpredictionhavebeendeveloped19.

Theabovestudiesshowedanewlayerofpost-transcriptionalregulation.However,tounderstandtherolesofceRNAnetworksinpathologicalconditions,moreworkisrequired.ItisfoundthatactivetranslationofmRNAimpedesinteractionofmiRNAsandtheirtargets20.ContrasttomRNAs,noncodingRNAsaremoreeffectiveceRNAswithoutanyinterferencefromtranslation6.Consideringthesefindingsandthefactthatgastriccancerisoneofthemostfrequentcausesofmortalityintheworld21,inthecurrentstudy,wefirstfocusedongastriccancerassociated-lncRNAsandconstructedaceRNAnetworkinsilico.Then,toconfirmthisnetwork,weperformedaregressionanalysisusingseveralsetsofgeneexpressiondataofothertypesofcancerincludinghead

1

M

SCIENTIFICREPORTS|4:6088|DOI:10.1038/srep06088

www.nature.com/scientificreportsFigure1|CompetingendogenousRNAs(ceRNAs)functionasmicroRNA(miRNA)spongessequestermiRNAstoregulateexpressionlevelofothertranscriptssharingcommonmiRNAresponseelements(MREs).(A)DownregulationoflongnoncodingRNAs(lncRNA)leadsmoremiRNAmoleculesfreetobindtomRNAthatcontainthesameMREs,thusitsproteinexpressionleveldecreases.(B)OverexpressionoflncRNAleadsfewermiRNAmoleculestobindtomRNA,thusitsproteinexpressionlevelincreases.

andnecksquamouscellcarcinoma,prostatecancer,papillarythyr-oidcarcinoma,pituitarygonadotropetumors,chroniclymphocyticleukaemia,andovariancancer.Finally,weusedseveralexperi-mentaldatatoprovetheceRNAnetwork.Thisnewapproachofpredictingcancerassociated-ceRNAnetworkmighthelpuseasilysearchingforcandidatecancer-associatedceRNAs.

Table1|AcollectionofdifferentiallyexpressedlncRNAsbetweengastriccancertissuesandparacanceroustissues

lncRNAAC009499.1GACAT1GACAT3H19

LINC00152RMRP

RP11-179G5.4RP11-187O7.3RPPH1

ABHD11-AS1AC073871.2AKR7A2P1AP000288.2FER1L4

RP1-15D23.2RP4-620F22.3RP4-740C4.4

GeneID

ENSG00000203386ENSG00000232991ENSG00000236289ENSG00000130600ENSG00000222041ENSG00000199916ENSG00000235082ENSG00000259124ENSG00000259001ENSG00000225969ENSG00000182648ENSG00000229020ENSG00000227757ENSG00000088340ENSG00000224228ENSG00000238081ENSG00000229813

ExpressionchangeUp-regulationUp-regulationUp-regulationUp-regulationUp-regulationUp-regulationUp-regulationUp-regulationUp-regulationDown-regulationDown-regulationDown-regulationDown-regulationDown-regulationDown-regulationDown-regulationDown-regulation

Foldchange

5.44.13.35.93.43.83.23.13.03.03.03.13.89.25.23.83.0

P-value0.0400.0080.0070.0250.0040.0430.0220.0310.0010.0270.0020.0010.0270.0470.0480.0210.039

2

SCIENTIFICREPORTS|4:6088|DOI:10.1038/srep06088

www.nature.com/scientificreportsSTAT3,VEGFA,E2F1,RB1,p21,MMP16,BCL2,CCND1,CDK6,CyclinD1,etc.Theirfunctionsareinvolvedincellproliferation,cellcycle,apoptosis,invasionandmetastasis.

ceRNAnetwork.Basedontheabovedatafromgastriccancer(Table2andTable3),weconstructedanlncRNA-miRNA-mRNAceRNAnetwork.Togetmorerobustresults,weassignedaP-valuetoeachrelationshipbyre-samplinganalysis(seeMethods).AsshowninFigure2,inthisceRNAnetwork,eightlncRNAsandninemiRNAswereinvolved.

PositivecorrelationbetweenceRNAs’expressionlevels.OurconstructedceRNAnetworkshowsthatlncRNAscouldinteractwithmRNAingastriccancer(Fig.2).Toconfirmthisfinding,weperformedaregressionanalysisusingseveralsetsofgeneexpressiondataofothertypesofcancerincludingheadandnecksquamouscellcarcinoma,prostatecancer,papillarythyroidcarcinoma,pituitarygonadotropetumors,chroniclymphocyticleukemia,andovariancancer.Theresultsrevealaverygoodorperfectpositivecorre-lationbetweenceRNAs’expressionlevels(Fig.3).Forexample,FER1L4interactedwithRUNX1mediatedbymiR-106a-5p(Fig.3AandB);LINC00152interactedwithTHBS1mediatedbymiR-18a-5p(Fig.3CandD);whileH19interactedwithMYCNmediatedbymiR-19a-3p(Fig.3EandF).

FER1L4-RB1isonepairofceRNAsassociatedbymiR-106a-5p.RB1isoneofmiR-106a-5p’stargets,andhasbeenvalidatedbydualluciferasereporterassay25.Inthisstudy,theinteractionbetweenFER1L4andmiR-106a-5pwasfirstpredictedbymiRcode(Table2).ThendualluciferasereporterassayshowedthattheluciferaseactivityofthemutantFER1L4plasmidwasabout56%higherthanthatofthewild-typeplasmid(Fig.4).ThisindicatedthatthemutationsintroducedintheseedmatchesimpairtheabilityofmiR-106a-5ptobindtoFER1L4.Taketogether,bothFER1L4andRB1aretargetsofmiR-106a-5p.ThisisthebasicconditionofceRNAs6.

Next,weobservedwhetherFER1L4andRB1expressionlevelsareassociated.Quantitativereversetranscription-polymerasechainreaction(qRT-PCR)analysisrevealedthattransfectionofsmallinterferingRNA(siRNA)againstFER1L4notonlyreducedFER1L4levelsinGES-1,AGS,MGC-803andSGC-7901,butalsoreducedRB1levelsinallcellstested(Fig.5).SinceFER1L4isthetargetofmiR-106a-5p(Table2andFig.4),FER1L4knockdownledtomoremiR-106a-5pfreetobindtoothertargets;andRB1mRNAisoneofthem.

Table2|PutativemiRNAstargetinglncRNA

lncRNAAC009499.1GACAT1GACAT3H19LINC00152ABHD11-AS1AP000288.2FER1L4RP4-620F22.3

miRNAs

miR-18a-5p,miR-18b-5pmiR-106a-5p

miR-195-5p,miR-497-5p

miR-17-5p,miR-18a-5p,miR-18b-5p,miR-19a-3p,miR-20a-5p,miR-20b-5p,miR-106a-5p,miR-106b-5pmiR-18a-5p,miR-18b-5p,miR-31-5p,miR-139-5p,miR-195-5p,miR-497-5pmiR-133bmiR-19a-3p

miR-18a-5p,miR-18b-5p,miR-106a-5p,miR-133b,miR-139-5p,miR-195-5p,miR-497-5pmiR-195-5p,miR-497-5p

Results

AberrantlyexpressedlncRNAsingastriccancer.Fromthelnc-RNAsexpressionprofiles(dataaccessibleatNCBIGEOdatabase,accessionGSE47850,Guo,2013),weselected53lncRNAsdiffe-rentiallyexpressedbetweengastriccancertissuesandparacan-ceroustissues(foldchange$3.022,P-value,0.05).Toenhancethedatareliability,weonlyretainedlncRNAsincludedinEncyclopediaofDNAElements(ENCODE)23.Finally,17lncRNAs(9,up-regulated;8down-regulated)wereselected(Table1).

Gastriccancerassociated-lncRNAs’MREspredictedbymiRcode.Ourpreviousstudyhasfoundseveralgastriccancer-associatedmiRNAsthatincludemiR-17-5p,miR-18a-5p,miR-18b-5p,miR-19a-3p,miR-20a-5p,miR-20b-5p,miR-21-5p,miR-31-5p,miR-106a-5p,miR-106b-5p,miR-133b,miR-139-5p,miR-195-5p,miR-340-3p,miR-378a-3p,miR-421,miR-497-5pandmiR-65821.Here,wefocusedonwhetherthesemiRNAsaretargetedonabovelncRNAs.SincemiRNAsareinteractingwithlncRNAsthroughtheirMREswithinceRNAnetwork,thefirstthingweshoulddowastosearchforthepotentialMREsinlncRNAs.MREspredictedbymiRcodeshowedthat13miRNAsmayinteractwithnineofseventeenlncRNAs(Table2).

mRNAstargetedbymiRNAs.ToestablishlncRNA-miRNA-mRNAnetwork(ceRNAnetwork),thenextstepweshoulddowastosearchformiRNAs’mRNAtargets.BasedonthosemiRNAsthatmightinteractwithlncRNAs(Table2),wesearchedmiRNAs’mRNAtargetswithexperimentalsupportusingTarBase24.TheresultsshowedthatninemiRNAsincludingmiR-18a-5p,miR-18b-5p,miR-19a-3p,miR-20b-5p,miR-106a-5p,miR-106b-5p,miR-31-5p,miR-139-5p,andmiR-195-5pmaybeincluded(Table3).Mostoftheirtargetsarecancer-associatedgenessuchasPTEN,TNF-a,Table3|ValidatedmRNAs’targetsfromTarBase

miRNAmiR-18a-5p

Discussion

StudiesrevealedthatlncRNAsplayimportantregulatoryrolesingeneexpressionregulationandcontributetooncogenesisandtumor

mRNAstargetedbymiRNAs

ATM,BCL2L11,CA12,CA13,CCNL1,CKAP5,CREBL2,CTDSPL,CTGF,DICER1,ERa,ESR1,HIF1A,HOXA9,MID1,PTEN,RAB23,RAB5A,SERTAD3,Smad4,TGFBR2,THBS1,TNFSF11,VIL2

miR-18b-5pERa,ESR1

miR-19a-3pBCL2L11,Bim,CCND1,CTGF,DPYSL2,ERBB4,ESR1,HOXA5,MECP2,MYCN,NR4A2,PRMT5,PTEN,RAB14,Smad4,SOCS1,

TGFBR2,THBS1,TNF-a,VPS4B

miR-20b-5pARID4B,BAMBI,CDKN1A,ESR1,HIF1A,HIPK3,MUC17,PPARG,STAT3,VEGFAmiR-31-5pCASR,CXCL12,ETS1,FOXP3,FZD3,HOXC13,ITGA5,KLF13,LATS2,MMP16,MPRIP,NFAT5,NUMB,PPP2R2A,RET,RHOA,SELE,

TIAM1,YY1

miR-106a-5pCDKN1A,E2F1,HIPK3,IL-10,PAK7,PTEN,RB1,RUNX1,VEGFA

miR-106b-5pAFP,AKAP11,BRMS1L,CABP2,CASP7,CD34,CDC37L1,CDK5R2,CDKN1A,CLOCK,DNAJB6,E2F1,EIF5A2,ELK3,HMGB3,

IFNAR2,JAK1,KDR,KIF23,LIMK1,MXI1,p21,PCAF,PKD2,PTEN,RB1,RBL2,RUNX1T1,Smad2/3,Smad9,SSX2,TbRII,TLR2,VEGFA

miR-139-5pHOXA10

miR-195-5pBCL2,BCL2L11,CCND1,CDK6,CyclinD1,E2F3,KRT7,MECP2,SKI,VEGFA,WEE1SCIENTIFICREPORTS|4:6088|DOI:10.1038/srep06088

3

www.nature.com/scientificreportsRecently,twogroupshaveidentifiedacircularRNA(circRNA)thatcontains,70MREswithmiR-733,34.ThecircRNAsequestersmiR-7awayfromitstargetsandsuppressesitsfunction.Thuscir-cularRNAsareanewclassofceRNAs35,36.Ontheotherhand,topredictceRNAs’effectontranscriptionalnetwork,Alaandcollea-guesdevelopedacomputationalmodelofceRNA37.Furtherinvesti-gationsofceRNAhaveimplicationsinmanyfields.Forinstant,toblockthefunctionsofmiRNAs,Tangetal.constructedanartificialceRNAnamedshorttandemtargetmimic(STTM)38.STTMissim-ilartomiRNAsponge,butcontainsdifferentkindsofMREsthatbindvariousmiRNAs.BasedontheceRNAhypothesis,Liuetal.furtherpresentedalongintergenicnoncodingRNA(lincRNA)data-basenamedLinc2GO39.Lietal.alsodevelopedadatabasenamedstarBasev2.0forpredictingmiRNA-ceRNAinteractions40.Moreover,ceRNAmightbeanewtherapeuticapproachtodiseases41.Inconclusion,weproposedanewapproachtoconstructceRNAnetwork.ItmighthelpustoeasilysearchingforceRNAsinvolvedincancer-associatedlncRNAs.OurfindingssuggestthatlncRNAsmayharborMREsandparticipateinacomplexceRNAnetwork.

Methods

Figure2|ceRNAnetworkingastriccancer.ColourednodesrepresentmiRNAs;blacknodesrepresentlncRNAs;graynodesrepresentmRNAs;colourededgesindicatemiRNA-targetinteractions.mRNAs’namesarenotshown.

Themethodswerecarriedoutin‘‘accordance’’withtheapprovedguidelines.Tissuesamples.BiopsysamplesofgastriccancertissuesandparacanceroustissueswereobtainedfromtheAffiliatedHospitalofNingboUniversitySchoolofMedicine.Theparacanceroustissueswere5cmfromtheedgeoftumor;andtherewerenoobvioustumorcells,asevaluatedbyanexperiencedpathologist.Therewasnoradiotherapyorchemotherapypriortotheendoscopyexamination.TheHumanResearchEthicsCommitteefromNingboUniversityapprovedallaspects.Writteninformedconsentwastakenfromallsubjects.

RNAextraction.TotalRNAwasextractedbyusingTRIzolReagent(Life

Technologies,Carlsbad,CA,USA)accordingtothemanufacturer’sprotocol.RNAquantitywasmeasuredbyaSmartSpecPlusspectrophotometer(Bio-Rad,Hercules,CA,USA).RNApuritywasevaluatedbytheA260/A280ratio.

lncRNAmicroarrayandcomputationalanalysis.ForlncRNAexpressionprofiling,theHumanLncRNAArray(Arraystar,Rockville,MD,USA)whichcovered18,534lncRNAswasused.Afterhybridizationandwashing,thearrayswerescannedbyanAxonGenePix4000BMicroarrayScanner(MolecularDevices,Sunnyvale,CA,USA).RawdatawereextractedandnormalizedbyNimbleScanv2.5softwarepackage(RocheNimbleGen,Madison,WI,USA).

ConstructionoftheceRNAnetwork.TheconstructionofceRNAnetworkincludedthreesteps(Fig.6):(i)lncRNAscreening:lncRNAsthatwereup-ordown-regulatedfoldchange$3.0andP-value,0.05werefirstretained;thentoenhancethedatareliability,lncRNAwhosesequenceshavenotbeenrecordedinENCODEwereremoved;(ii)lncRNA-miRNAinteractionswerepredictedbymiRcode(http://www.mircode.org/);(iii)mRNAsthatweretargetedbymiRNAswithexperimentalsupportwerefromTarBase(http://www.microrna.gr/tarbase).

Re-samplinganalysisofceRNAnetwork.TogetmorerobustresultsofceRNAnetwork,weassignedaP-valuetoeachrelationshipbyre-samplinganalysis42.First,were-sampledthesetofabovemiRNAsfromthepoolofallhumanmiRNAsinmiRBasedatabaseandre-inferredtheceRNArelationships.Thenweobserved

whetherthepreviouslypredictedceRNArelationshipsoccurredinthenewlyinferreddataset.Werepeatedthere-sampling10,000timesandcountedthenumberoftimesthateachceRNArelationshipoccurred.TheP-valuewasdefinedasthenumberofoccurredtimesover10,000.Finally,weobtainedtheceRNArelationshipsusingthecutoffof0.01.

RegressionanalysisofceRNAs’expressionlevel.Severalsetsofgeneexpressiondataincancerswereanalysed(dataaccessibleatNCBIGEOdatabase,accessionGDS1667,Slebosetal,2005;accessionGDS4159,Vitarietal,2011;accessionGDS1665,Heetal,2005;accessionGDS4275,Michaelisetal,2011;accessionGDS4168,Gutierrezetal,2010;accessionGDS4066,Spillmanetal,2011).RegressionanalysisandcorrelationcoefficientweregeneratedusingSPSSStatisticsv20.0softwarepackage(IBM,Armonk,NY,USA).

Cellsandcultureconditions.Humangastricepithelialcellline,GES-1,wasobtainedfromCancerInstituteandHospital,ChineseAcademyofMedicalSciences(Beijing,China).Humangastriccancercelllines,AGS,MGC-803andSGC-7901,wereobtainedfromShanghaiInstitutesforBiologicalSciences,ChineseAcademyofSciences(Shanghai,China).Humanembryonickidneycellline,HEK293T,wasobtainedfromGeneChemCo.,Ltd.(Shanghai,China).GES-1,AGS,MGC-803andSGC-7901weregrowninRPMIMedium1640(LifeTechnologies)plus10%fetalbovineserum(FBS).HEK293TwasgrowninDMEM(LifeTechnologies)plus10%

metastasis26,27.LncRNAscanfunctionasendogenousmiRNAspongesasapartofceRNAnetwork15,28,29.Incurrentstudy,basedonlncRNAmicroarrayandmiRNAmicroarraydata,weconstructedtheceRNAnetworkjointedbylncRNAs,miRNAsandmRNAs.Previousworkshaveproposedseveraldifferentstrategies,suchasRNA22andmiRanda12,30,topredictmiRNA-targetinteractions.However,thesealgorithmsweremainlyusedinthepredictionofmiRNA-mRNAinteraction.SincethatlncRNAshaverecentlyemergedasmajorplayersingoverningfundamentalbiologicalpro-cessesanddiseaseoccurrence;andthat,18%oftheprotein-codinggenesthatproducelncRNAsareassociatedwithcancers,whereasonly9%ofallhumanprotein-codinggenesareassociatedwithcan-cers31,here,wefocusedonsearchingforlncRNA-miRNAinterac-tions.WeusedmiRcode,anlncRNA-miRNAinteractionpredictionalgorithm,topredictMREsinlncRNAs.Inaddition,toenhancethedatareliability,validatedmRNAtargetswereacquiredfromTarBaseonlywitheffectiveexperimentalsupportingdata.ThesedefinedrulesarebeneficialtoscreeningforcandidateceRNAs.

OuranalysissuggestedlncRNAsharborpotentialMREsandpar-ticipateinacomplexceRNAnetwork(Fig.2).ThenetworkbringstolightanunknownmiRNAregulatorynetworkingastriccancer.ItalsosuggeststhatlncRNAsmayplaycrucialrolesingastriccanceroccurrencesanddevelopment(Table1).AsSumazinetal.proved,ceRNAnetworkisascale-freenetworkinwhichmiRNAsmediatedinteractionsexhibitsymmetricbehavior14.WefoundthatmanyoncogenesandtumorsuppressorsparticipateintheceRNAnetworkingastriccancer(Table3).Forinstance,mediatedbymiR-106a-5p,PTEN,RB1,RUNX1,VEGFA,CDKN1A,E2F1,HIPK3,IL-10,orPAK7mayinteractwithlncRNA-FER1L4,GACAT1andH19.PreviousstudyhasshowedthatmiR-106a-5p,thecoreelementinthisnetwork,isoneofthetypicalonco-miRNAs32.FER1L4andRB1aretargetedbymiR-106a-5p.siRNAsilencingofFER1L4ledtoRB1mRNAleveldecrease(Fig.5).

Inaddition,weanalysedceRNAs’expressioninseveralsetsofdatafromcancersincludingsolidcancersandleukemia.Wefoundthattherewereexcellentpositivecorrelationsbetweentheirexpressionlevels(Fig.3).ThiscoexpressioncharactercorrespondswithceRNAhypothesis.

SCIENTIFICREPORTS|4:6088|DOI:10.1038/srep06088

4

www.nature.com/scientificreportsFigure3|LinearregressionofceRNAs’expressionlevel.Dashedlinesrepresent95%confidenceinterval.(A)FER1L4vsRUNX1(headandnecksquamouscellcarcinoma,n536).(B)FER1L4vsRUNX1(prostatecancer,n515).(C)LINC00152vsTHBS1(papillarythyroidcarcinoma,n518).(D)LINC00152vsTHBS1(pituitarygonadotropetumors,n523).(E)H19vsMYCN(chroniclymphocyticleukemia,n552).(F)H19vsMYCN(ovariancancer,n515).

FBS.Allcelllinesweregrownat37uCinahumidifiedatmospherewith5%CO2.CellswerecountedusingaTC10AutomatedCellCounter(Bio-Rad).

Dualluciferasereporterassay.miR-106a-5pexpressionplasmid(GV268)waspurchasedfromGeneChemCo.,Ltd.(Shanghai,China).Thewild-typeandmutant

DNAsequencesofFER1L4werecustomsynthesizedbyGeneChemCo.,Ltd.andclonedintoGV272Fireflyluciferaseplasmid(GeneChem).ThemiR-106a-5pMREofwild-typeFER1L4was59-GCACUU-39,whilethemutantsequencewas59-UACAGG-39.HEK293Tof80%confluencein24-wellplateswastransfectedusingLipofectamine2000Reagent(LifeTechnologies)accordingtothemanufacturer’s

SCIENTIFICREPORTS|4:6088|DOI:10.1038/srep060885

www.nature.com/scientificreportsFigure4|RelativeFirefly/Renillaluminescence(mean6SD,n53)mediatedbyluciferaseplasmidharboringthewild-typeormutantFER1L4sequenceupontransfecitonwithmiR-106a-5pexpressionplasmid.ItindicatesthedirectinteractionbetweenFER1L4andmiR-106a-5p.***P,0.001.

protocol.FireflyluciferaseplasmidandmiR-106a-5pexpressionplasmidwereco-transfectedwithpRL-TKRenillaluciferasevector(Promega,Madison,WI,USA)fornormalization.48hoursaftertransfection,luciferaseactivitywasmeasuredusingDual-GloLuciferaseAssaySystem(Promega).Allexperimentswereperformedintriplicate.

Transienttransfection.ForthetransfectionofsiRNA,GES-1,AGS,MGC-803andSGC-7901(23105)wereseededin6-wellplates.Thefollowingdaytheyweretransfectedwith120nMsiRNAsusingLipofectamine2000Reagent(LifeTechnologies).FER1L4siRNAwas:59-CAGGACAGCUUCGAGUUAATT-39

(sense)and59-UUAACUCGAAGCUGUCCUGTT-39(antisense).NegativecontrolsiRNAwas:59-UUCUCCGAACGUGUCACGUTT-39(sense)and59-ACGUGACACGUUCGGAGAATT-39(antisense).

Figure6|AflowchartofceRNAnetworkconstruction.(i)lncRNAsthatarefoldchange$3.0andP-value,0.05wereretained;(ii)lncRNAsthathavenotbeenrecordedinENCODEwereremoved;(iii)miRNA-lncRNAinteractionswerepredictedbymiRcode;(iv)mRNAsthattargetedbymiRNAswerecapturedfromTarBase;(v)ceRNAnetworkconstruction.

qRT-PCRanalysis.qRT-PCRwasperformedusingGoTaq2-StepRT-qPCRSystem(Promega)inaMx3005PQPCRSystem(Stratagene,LaJolla,CA,USA)accordingtothemanufacturer’sprotocol.Glyceraldehyde-3-phosphatedehydrogenase(GAPDH)wasusedasacontrol.Relativequantificationofgeneexpressionwasperformedwith2-DDCtmethod43.FER1L4primers:forward59-CCGTGTTGAGGTGCTGTTC-39andreverse59-GGCAAGTCCACTGTCAGATG-39.RB1primers:forward59-TGGCG-TGCGCTCTTGAGGTT-39andreverse59-AGAGCCATGCAAGGGATTCC-ATGA-39.GAPDHprimers:forward59-AAGGTGAAGGTCGGAGTCAA-39andreverse59-AATGAAGGGGTCATTGATGG-39.Allexperimentswereperformedintriplicate.

Statisticalanalysis.Datawerepresentedasmean6SD.Differencesamonggroupswereevaluatedbytwo-tailedStudent’st-testbyusingSPSSStatisticsv20.0softwarepackage(IBM).P,0.05wasconsideredtobestatisticallysignificant.

Figure5|qRT-PCRanalysisofexpressionlevels(mean6SD,n53)ofFER1L4andRB1inGES-1,AGS,MGC-803andSGC-7901treatedwithsiRNAagainstFER1L4.NC,negativecontrol.*P,0.05;**P,0.01;***P,0.001.

SCIENTIFICREPORTS|4:6088|DOI:10.1038/srep06088

6

www.nature.com/scientificreports1.Pasquinelli,A.E.MicroRNAsandtheirtargets:recognition,regulationandanemergingreciprocalrelationship.NatRevGenet13,271–282(2012).

2.Thomas,M.,Lieberman,J.&Lal,A.DesperatelyseekingmicroRNAtargets.NatStructMolBiol17,1169–1174(2010).

3.Zhang,H.,Li,Y.&Lai,M.ThemicroRNAnetworkandtumormetastasis.Oncogene29,937–948(2010).

4.Gennarino,V.A.etal.IdentificationofmicroRNA-regulatedgenenetworksbyexpressionanalysisoftargetgenes.GenomeRes22,1163–1172(2012).5.Li,L.etal.3’UTRshorteningidentifieshigh-riskcancerswithtargeteddysregulationoftheceRNAnetwork.SciRep4,5406(2014).

6.Salmena,L.,Poliseno,L.,Tay,Y.,Kats,L.&Pandolfi,P.P.AceRNAhypothesis:theRosettaStoneofahiddenRNAlanguage?Cell146,353–358(2011).7.Ebert,M.S.,Neilson,J.R.&Sharp,P.A.MicroRNAsponges:competitiveinhibitorsofsmallRNAsinmammaliancells.NatMethods4,721–726(2007).8.Poliseno,L.etal.Acoding-independentfunctionofgeneandpseudogenemRNAsregulatestumourbiology.Nature465,1033–1038(2010).

9.Wang,J.etal.CREBup-regulateslongnon-codingRNA,HULCexpressionthroughinteractionwithmicroRNA-372inlivercancer.NucleicAcidsRes38,5366–5383(2010).

10.Lee,D.Y.etal.Expressionofversican3’-untranslatedregionmodulatesendogenousmicroRNAfunctions.PLoSOne5,e13599(2010).

11.Jeyapalan,Z.etal.ExpressionofCD443’-untranslatedregionregulates

endogenousmicroRNAfunctionsintumorigenesisandangiogenesis.NucleicAcidsRes39,3026–3041(2011).

12.Tay,Y.etal.Coding-independentregulationofthetumorsuppressorPTENbycompetingendogenousmRNAs.Cell147,344–357(2011).

13.Karreth,F.A.etal.Invivoidentificationoftumor-suppressivePTENceRNAsinanoncogenicBRAF-inducedmousemodelofmelanoma.Cell147,382–395(2011).

14.Sumazin,P.etal.AnextensivemicroRNA-mediatednetworkofRNA-RNAinteractionsregulatesestablishedoncogenicpathwaysinglioblastoma.Cell147,370–381(2011).

15.Cesana,M.etal.AlongnoncodingRNAcontrolsmuscledifferentiationbyfunctioningasacompetingendogenousRNA.Cell147,358–369(2011).

16.Chen,G.etal.LncRNADisease:adatabaseforlong-non-codingRNA-associateddiseases.NucleicAcidsRes41,D983–D986(2013).

17.Li,Y.etal.HMDDv2.0:adatabaseforexperimentallysupportedhuman

microRNAanddiseaseassociations.NucleicAcidsRes42,D1070–D1074(2014).18.Yang,Z.etal.dbDEMC:adatabaseofdifferentiallyexpressedmiRNAsinhumancancers.BMCGenomics11,S5(2010).

19.Jeggari,A.,Marks,D.S.&Larsson,E.miRcode:amapofputativemicroRNAtargetsitesinthelongnon-codingtranscriptome.Bioinformatics28,2062–2063(2012).

20.Gu,S.,Jin,L.,Zhang,F.,Sarnow,P.&Kay,M.A.BiologicalbasisforrestrictionofmicroRNAtargetstothe3’untranslatedregioninmammalianmRNAs.NatStructMolBiol16,144–150(2009).

21.Guo,J.etal.DifferentialexpressionofmicroRNAspeciesinhumangastriccancerversusnon-tumoroustissues.JGastroenterolHepatol24,652–657(2009).

22.McKiernan,P.J.,Molloy,K.,Cryan,S.A.,McElvaney,N.G.&Greene,C.M.LongnoncodingRNAareaberrantlyexpressedinvivointhecysticfibrosisbronchialepithelium.IntJBiochemCellBiol52,184–191(2014).

23.Harrow,J.etal.GENCODE:thereferencehumangenomeannotationforTheENCODEProject.GenomeRes22,1760–1774(2012).

24.Vergoulis,T.etal.TarBase6.0:capturingtheexponentialgrowthofmiRNAtargetswithexperimentalsupport.NucleicAcidsRes40,D222–D229(2012).25.Volinia,S.etal.AmicroRNAexpressionsignatureofhumansolidtumorsdefinescancergenetargets.ProcNatlAcadSciUSA103,2257–2261(2006).

26.Ying,L.etal.DownregulatedMEG3activatesautophagyandincreasescellproliferationinbladdercancer.MolBiosyst9,407–411(2013).

27.Tripathi,V.etal.LongnoncodingRNAMALAT1controlscellcycleprogressionbyregulatingtheexpressionofoncogenictranscriptionfactorB-MYB.PLoSGenet9,e1003368(2013).

28.Wang,Y.etal.EndogenousmiRNAspongelincRNA-RoRregulatesOct4,Nanog,andSox2inhumanembryonicstemcellself-renewal.DevCell25,69–80(2013).

29.Kallen,A.N.etal.TheimprintedH19lncRNAantagonizeslet-7microRNAs.MolCell52,101–112(2013).

30.Jalali,S.,Bhartiya,D.,Lalwani,M.K.,Sivasubbu,S.&Scaria,V.SystematictranscriptomewideanalysisoflncRNA-miRNAinteractions.PLoSOne8,e53823(2013).

31.Khachane,A.N.&Harrison,P.M.Miningmammaliantranscriptdataforfunctionallongnon-codingRNAs.PLoSOne5,e10316(2010).

32.Xiao,B.etal.DetectionofmiR-106aingastriccarcinomaanditsclinicalsignificance.ClinChimActa400,97–102(2009).

33.Memczak,S.etal.CircularRNAsarealargeclassofanimalRNAswithregulatorypotency.Nature495,333–338(2013).

34.Hansen,T.B.etal.NaturalRNAcirclesfunctionasefficientmicroRNAsponges.Nature495,384–388(2013).

35.Karreth,F.A.&Pandolfi,P.P.ceRNAcross-talkincancer:whence-blingrivalriesgoawry.CancerDiscov3,1113–1121(2013).

36.Tay,Y.,Rinn,J.&Pandolfi,P.P.ThemultilayeredcomplexityofceRNAcrosstalkandcompetition.Nature505,344–352(2014).

37.Ala,U.etal.IntegratedtranscriptionalandcompetitiveendogenousRNAnetworksarecross-regulatedinpermissivemolecularenvironments.ProcNatlAcadSciUSA110,7154–7159(2013).

38.Tang,G.etal.Constructionofshorttandemtargetmimic(STTM)toblockthefunctionsofplantandanimalmicroRNAs.Methods58,118–125(2012).

39.Liu,K.,Yan,Z.,Li,Y.&Sun,Z.Linc2GO:ahumanlincRNAfunctionannotationresourcebasedonceRNAhypothesis.Bioinformatics29,2221–2222(2013).40.Li,J.H.,Liu,S.,Zhou,H.,Qu,L.H.&Yang,J.H.starBasev2.0:decodingmiRNA-ceRNA,miRNA-ncRNAandprotein-RNAinteractionnetworksfromlarge-scaleCLIP-Seqdata.NucleicAcidsRes42,D92–D97(2014).

41.Almeida,M.I.,Reis,R.M.&Calin,G.A.DecoyactivitythroughmicroRNAs:thetherapeuticimplications.ExpertOpinBiolTher12,1153–1159(2012).

42.Li,J.etal.Identificationofhigh-qualitycancerprognosticmarkersandmetastasisnetworkmodules.NatCommun1,34(2010).

43.Yu,X.etal.GrowthinhibitoryeffectsofthreemiR-129familymembersongastriccancer.Gene532,87–93(2013).

Acknowledgments

ThisworkwassupportedbytheZhejiangProvincialNaturalScienceFoundationofChina(GrantNo.LY14C060003),theAppliedResearchProjectonNonprofitTechnologyofZhejiangProvince(GrantNo.2014C33222),theNationalNaturalScienceFoundationofChina(GrantNo.81171660and31301068),theNaturalScienceFoundationofNingbo(GrantNo.2013A610207),theScientificInnovationTeamProjectofNingbo(GrantNo.2011B82014),theOutstanding(Postgraduate)DissertationGrowthFoundationofNingboUniversity(GrantNo.PY2012004),andtheK.C.WongMagnaFundinNingboUniversity.

Authorcontributions

T.X.andJ.G.conceivedanddesignedtheexperiments.T.X.Q.L.X.J.Y.S.B.X.performedtheexperiments.T.X.Q.L.X.J.B.X.Y.X.J.G.analyzedthedata.T.X.J.G.wrotethepaper.

Additionalinformation

Competingfinancialinterests:Theauthorsdeclarenocompetingfinancialinterests.Howtocitethisarticle:Xia,T.etal.LongnoncodingRNAassociated-competingendogenousRNAsingastriccancer.Sci.Rep.4,6088;DOI:10.1038/srep06088(2014).

ThisworkislicensedunderaCreativeCommonsAttribution-NonCommercial-NoDerivs4.0InternationalLicense.Theimagesorotherthirdpartymaterialinthisarticleareincludedinthearticle’sCreativeCommonslicense,unlessindicatedotherwiseinthecreditline;ifthematerialisnotincludedundertheCreativeCommonslicense,userswillneedtoobtainpermissionfromthelicenseholderinordertoreproducethematerial.Toviewacopyofthislicense,visithttp://creativecommons.org/licenses/by-nc-nd/4.0/SCIENTIFICREPORTS|4:6088|DOI:10.1038/srep060887