miR-Synth a computational resource for the design of multi-site multi-target synthetic miRNAs

NucleicAcidsResearch,20141

doi:10.1093/nar/gku202

miR-Synth:acomputationalresourceforthedesignofmulti-sitemulti-targetsyntheticmiRNAs

AlessandroLagan`a1,*,†,MarioAcunzo1,†,GiuliaRomano1,AlfredoPulvirenti2,DarioVeneziano1,2,LucianoCascione3,RosalbaGiugno2,PierluigiGasparini1,DennisShasha4,AlfredoFerro2andCarloMariaCroce1

1

DepartmentofMolecularVirology,ImmunologyandMedicalGenetics,ComprehensiveCancerCenter,TheOhioStateUniversity,Columbus,OH,43210USA,2DepartmentofClinicalandMolecularBiomedicine,UniversityofCatania,95100Italy,3IOR-InstituteofOncologyResearch,Bellinzona,6500Switzerlandand4DepartmentofComputerScience,CourantInstituteofMathematicalSciences,NewYorkUniversity,NewYork,NY10012USA

ReceivedSeptember13,2013;RevisedFebruary20,2014;AcceptedFebruary25,2014

ABSTRACT

Inlightofthisevidence,thefocusofdrugtherapymayneedRNAiisapowerfultoolfortheregulationofgeneex-toshiftfromsingle-tomulti-targetapproaches(1).

pression.ItiswidelyandsuccessfullyemployedinThisapproachisfurtherjustifiedbythefactthatmostfunctionalstudiesandisnowemergingasapromis-cancersreflectadysfunctionalityinmultiplepathwaysandingtherapeuticapproach.SeveralRNAi-basedclin-anaccumulationofnewoncogenicmutationsasthedis-icaltrialssuggestencouragingresultsinthetreat-easeprogresses.Thus,avalidstrategycancomefromtarget-ingmultiplegenesinvolvedinalteredpathwaysratherthanmentofavarietyofdiseases,includingcancer.Heresinglegenes,potentiallyassuringgreaterandmoredurablewepresentmiR-Synth,acomputationalresourcetherapeuticbenefits(2).

forthedesignofsyntheticmicroRNAsabletotar-RNAiisnowemergingasapromisingtherapeuticap-getmultiplegenesinmultiplesites.Theproposedproach(3,4).Selectivegenesilencingthroughsmallinterfer-strategyconstitutesavalidalternativetotheuseofingRNAsiswidelyandsuccessfullyemployedinfunctionalsiRNA,allowingtheemploymentofafewernumberofstudiesandiscurrentlybeinginvestigatedasapotentialtoolmoleculesfortheinhibitionofmultipletargets.Thisforthetreatmentofvariousdiseases,includingcancer,skinmayrepresentagreatadvantageindesigningthera-diseasesandviralinfections.siRNA,shRNAandtheiropti-piesfordiseasescausedbycrucialcellularpathwaysmizedchemicalmodificationsaretheactivesilencingagentsalteredbymultipledysregulatedgenes.ThesystemandareintendedtotargetsinglemRNAsinaspecificwayhasbeensuccessfullyvalidatedontwoofthemost(5).

SeveralongoingandalreadycompletedRNAi-basedprominentgenesassociatedtolungcancer,c-METclinicaltrialssuggestencouragingresults(6).siRNA-andEpidermalGrowthFactorReceptor(EGFR).(SeemediatedcleavageofatargetmRNA,withaconsequenthttp://microrna.osumc.edu/mir-synth).

reductionofproteinexpressionlevel,wasobtainedinthefirstin-humanphaseIclinicaltrialinwhichsiRNAwereadministeredsystemicallytosolidcancerpatients(4).

INTRODUCTION

Thegoaloftargetingmultiplegenesanddisruptingcom-plexsignalingpathwayscanbereachedbyco-expressionofManydiseases,suchascancerandneurologicalpathologies,multiplesiRNAorshRNAwhichenablemultipletargetin-occurastheresultofmultiplealterationsingeneswhicharehibition,alongwiththetargetingofmultiplesitesonaspe-partofcrucialcellularpathways.

cificgene(7).

Uptothepresentday,drugdevelopmenthasgenerallyAnimportantexperimentinantiviraltherapyresearchbeenfocusedontherapeuticaltargetingofindividualgeneshasshownthatstableexpressionofasingleshRNAtarget-orgeneproducts.Thisstrategy,however,hasproventobeingtheHIV-1Nefgenestronglyinhibitsviralreplication,limitedbecausetheinhibitionofsinglemoleculesmaynotbuttheshRNAdoesnotmaintainsuchinhibitionduetobesufficienttoeffectivelycounteractdiseaseprogressionmutationordeletionoftheneftargetsequencewhichal-andoftenleadstodrugresistancewithconsequentrelapse.

lowsthevirustoescape.Adelayinvirusescapeisobserved

*To

whomcorrespondenceshouldbeaddressed.Tel:+16142927278;Fax:+16142923558;Email:alessandro.lagana@osumc.edu

CorrespondencemayalsobeaddressedtoCarloMariaCroce.Tel:+16142924930;Fax:+16142923558;Email:carlo.croce@osumc.edu†

Theseauthorsequallycontributedtothework.

󰀅CTheAuthor(s)2014.PublishedbyOxfordUniversityPressonbehalfofNucleicAcidsResearch.

ThisisanOpenAccessarticledistributedunderthetermsoftheCreativeCommonsAttributionLicense(http://creativecommons.org/licenses/by-nc/3.0/),whichpermitsnon-commercialre-use,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited.Forcommercialre-use,pleasecontactjournals.permissions@oup.com

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 20142NucleicAcidsResearch,2014

Figure1.(a)ThefourdifferentkindsofcanonicalmiRNAseedsaredepicted.Theyallsharea6mercore(bases2–7).7mer-A1sitesfeatureanAoppositeofthefirstbaseofthemiRNA,7mer-m8sitesarefull7mers(bases2–8)and8mersitesare7mer-m8withanAoppositeofthefirstbaseofthemiRNA.(b)Inputsequencesarescreenedforrepeated6mer/7mersubsequencesthatwillconstitutethebindingsitesforthesyntheticmiRNAseeds.(c)RepeatedpatternsareusedasanchorsforthealignmentofthebindingsitesofsyntheticmiRNAs.miRNAsaredesignedbymaximizingcomplementaritytotheconsensustargetsequence(seealsoSupplementaryFigureS1).TargetbasescomplementarytomiRNAbasesareindicatedinblueandtheseedmatchisindicatedinred.(d)ThetreegeneratedbythelearningsystemM5P.Sixdifferentsetsofweightsforthesixconsideredfeaturesarecalculatedbasedonthevaluesofthethreediscriminantfeatures‘seedtype’,‘nucleotidecomposition’and‘AUcontent’.ThewhiteboxcontainsthesetofweightG1.SeeSupplementaryTableS1forthecompletelistofweight’ssets.(e)ThetreegeneratedbythelearningsystemCTree.ThesystemassignseachmiRNAtooneoffourdifferentscoreclasses,basedonthevaluesofthediscriminantfeatures‘seedtype’and‘nucleotidecomposition’.

insteadinHIV-1infectedcellsthatwerepreviouslytrans-cancer,c-METandEGFR.AscoringfunctionrankstheducedwithadoubleshRNAviralvector(8).

designedmiRNAsaccordingtotheirpredictedrepressionOptimizationsforco-expressionofsiRNAhavealsobeenefficiency.

proposed.Inarecentwork,dual-targetingsiRNAwithtwoOurexperimentalvalidationsofthescoringfunctionactivestrandswerespecificallydesignedtotargetdistinctshowthatadownregulationofupto70%wasobtainedmRNAtranscriptswithcompletecomplementarity.Thisre-bytop-rankingmiRNAs.miR-Synthisavailableathttp:sultedineasierRISCentrysinceonlytwostrands,instead//microrna.osumc.edu/mir-synth.offour,werecompetingforit(9).

AnalternativeapproachfortargetingmultiplegenesissuggestedbytheendogenousmicroRNA(miRNA)wayofMATERIALSANDMETHODSaction(10).miRNAs,indeed,arenaturallyintendedtotar-getmultiplegenes,ofteninmultiplesites,duetothepar-Cellculture,transfectionandchemicals

tialcomplementaritytheyexhibittotheirtargets(11).ThisHeLaandHEK-293AcellswereseededandgrowninstrategywouldalsoenjoytheadvantagethatcomesfromRoswellParkMemorialInstitutemedium(RPMI)(HEK-involvingfewernumberofmolecules.

293A)orDulbecco’smodifiedEagle’smedium(DMEM)Inlightoftheseconsiderations,wehavedevelopedmiR-(HeLa)with10%fetalbovineserum(FBS),L-glutamineSynth,abioinformaticstoolavailablethroughawebinter-andantibiotics(Invitrogen,Carlsbad,CA,USA).AllfaceforthedesignofsyntheticmiRNAsabletotargetmul-thetransfectionswereperformedbyusingLipofectaminetiplegenesinmultiplesites.Wehavevalidatedoursystem2000(Invitrogen)assuggestedbythemanufacturer.HEK-bydesigningandtestingsingle-anddouble-targetmiRNAs293Acellstransfectionforluciferaseassayisdescribedbe-fortwoofthemostprominentgenesassociatedwithlung

low.HeLacellswereculturedto80%confluenceinp60plateswithaserum-freemediumwithoutantibiotics,trans-

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 2014fectedwith100nmolofartificialmiRNA(a-miR)oligonu-cleotidesornegativecontrolandharvestedafter48h.Western-blotanalysis

HeLacellswereseededandgrowninDMEMwith10%FBSinsix-wellplatesfor24hbeforethetransfection.48haftertransfection,cellswerewashedwithcoldphosphate-bufferedsalineandsubjectedtolysisinlysisbuffer(50mMTris-HCl,1mMEDTA,20g/lSDS,5mMdithiothre-itol,10mMphenylmethylsulfonylfluoride).Equalamountsofproteinlysates(50mgeach)andrainbowmolecularweightmarker(Bio-RadLaboratories,Hercules,CA,USA)wereseparatedby4–20%SDS–PAGEandthenelectro-transferredtonitrocellulosemembranes.Themembraneswereblockedwithabuffercontaining5%non-fatdrymilkinTris-bufferedsalinewith0.1%Tween-20for2handincubatedovernightwithantibodiesat4◦C.Afterasec-ondwashwithTris-bufferedsalinewith0.1%Tween20,themembraneswereincubatedwithperoxidase-conjugatedsecondaryantibodies(GEHealthcare,Amersham,Pitts-burg,PA,USA)anddevelopedwithanenhancedchemi-luminescencedetectionkit(Pierce,Rockford,IL,USA).Antibodyusedforwestern-blotanalysis

␤EGFR-Actinantibodies(Sigma)waswereusedfromasCellaloadingSignalingcontrol.Technologies.METandRNAextraction

TotalRNAwasextractedwithTRIzolsolution(Invitrogen,Carlsbad,CA,USA),accordingtothemanufacturer’sin-structions.Q-real-timePCR

Forthedetectionofsingle-targeta-miRs,weperformedquantitativereversetranscriptase-polymerasechainreac-tion(qRT-PCR)byusingastandardTaqManPCRKitprotocolonanAppliedBiosystem7900HTSequenceDe-tectionSystem(AppliedBiosystems,Carlsbad,CA,USA).FortheTaqManqRT,the10mlPCRreactionincluded0.67mlRTproduct,1mlTaqManUniversalPCRMasterMix(AppliedBiosystems,Carlsbad,CA,USA),0.2mMTaq-Manprobe,1.5mMforwardprimerand0.7mMreverseprimer.Thereactionswereincubatedina96-wellplateat95◦◦Cfor10min,followedby40cyclesof95◦Cfor15sand60Cfor1min.Allreactionsranintriplicate.Thethresholdcycle(Ct)isdefinedasthefractionalcyclenumberatwhichthefluorescencepassesthefixedthreshold.Thecompara-tiveCtmethodforrelativequantizationofgeneexpression(AppliedBiosystems,Carlsbad,CA,USA)wasusedtode-terminea-miRexpressionlevels.They-axisrepresentstherelativeexpressionofthedifferenta-miRs(Figures2c,3cand4c).a-miRexpressionwascalculatedrelativetoU44andU48rRNA.Experimentswerecarriedoutintriplicateforeachdatapoint,anddataanalysiswasperformedbyusingsoftwaretools(Bio-RadLaboratories,Hercules,CA,USA).

NucleicAcidsResearch,20143

ItwasnotpossibletosynthesizeTaqMancustomprimersforthedetectionofthea-miRstargetingbothEGFRandc-METsoweperformedSYBRGreenPCRassay.cDNAwasobtainedfrom1␮goftotalRNAwithmiScriptII(Qiagen,Venlo,TheNetherlands)usingrandomprimers.cDNAwasthentreatedwithRNaseH,inordertoremoveanyresidualRNA.ForSYBRGreenqPCR,amplificationofcDNAwasperformedwithSYBRGreenPCRkit(Qiagen,Venlo,TheNetherlands),andnormalizedusingthe2−ΔCtmethodtoU6rRNA.They-axisrepresentstherelativeexpressionofthedifferenta-miRs.Luciferaseassay

Weusedtheluciferasereporterconstructsdescribedinotherworks(12,13).Mutationsina-miRbindingsitesinMETandEGFRconstructswereintroducedbyusingtheQuikChangeMutagenesisKit(Stratagene,LaJolla,CA,USA).HEK-293AcellsweretransfectedwithLipofec-tamine2000(Invitrogen,Carlsbad,CA,USA),1.2mgofpGL3controlcontainingEGFR,METorMETandEGFRmutants,200ngofRenillaluciferaseexpressionconstruct.After24h,cellswerelysedandassayedwithDualLu-ciferaseAssay(Promega)accordingtothemanufacturer’sinstructions.Mutagenesis’primersarereportedinSupple-mentaryTableS9.

Developmentofthedesigntoolandwebinterface

ThemiR-SynthdesigntoolwaswritteninRubyv1.9.3.TheprogramusestheexternalsoftwaretoolsRNAplfoldfromtheViennaRNAPackagev1.8.4forthecomputationofthestructuralaccessibilityandthestatisticalpackageRv3.0.1forthecomputationoftheconditionalinferencetrees(CTree)score.RisexecutedfromtheRubyscriptbyusingthegemrinruby.TheM5Pweightsandscoreswerecom-putedbyusingtheM5PimplementationavailableinthesoftwaretoolWekav3.7.9.

ThemiR-SynthwebinterfacewasdevelopedinRubyonRailsv2.3.5,aframeworkbasedontheMVC(Model-View-Controller)designpattern.AlltranscriptsequencedataforthespeciesprovidedbymiR-Synth,alongwithalluserspecifieddata,arecollectedandmaintainedinaMySQLdatabasev5.1runningonanApacheserverv2.2.15.Thequeriesthatthedatabaseallowstoperformwerecodedleveragingontheassociationmechanismsbetweenmodelsthattheframeworkprovides.TheinterfacemakesuseofthejQueryv1.7technologytoimprovetheusabilitythroughafastandagileclient-sideupdateofselectionsandresults.RESULTS

ThemiR-Synthalgorithmandthedesignfeatures

miR-Synthisatoolforthedesignofa-miRsfortherepres-sionofsingleormultipletargets.Theproblemofdesigningeffectivea-miRsisstrictlyconnectedtothepredictionofmiRNAbindingsites.Themainissueisthattargetpredic-tiontoolsyieldmanyfalsepositives(14).Nevertheless,theremarkableprogressmadeinrecentyearshasidentifiedkeyfeaturestocharacterizemiRNAfunctionaltargetsites.

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 20144NucleicAcidsResearch,2014

Figure2.(a)pGL3-MET3󰀁UTRconstructwasco-transfectedwitha-miRsornegativecontrolinHEK-293Acellsandluciferaseassaywasperformed(errorbars:±SEM,P<0.05).(b)c-METexpressionwasassessedbywesternblotinHeLacellstransfectedwitha-miRsornegativecontrolandharvestedafter72h.a-miR-M-60anda-miR-M-176enforcedexpressiondecreasesendogenouslevelsofthec-METprotein.Loadingcontrolwasobtainedbyusinganti-␤-actinantibody.(c)qRT-PCRofthetransfecteda-miRsinHeLacells.(d)qRT-PCRofthec-METmRNAaftera-miRsenforcedexpressioninHeLacells.(e)Representationofthec-MET3󰀁UTRsbindingsitesfora-miR-M-60.Inthefigure,pairingoftheseedregionofa-miR-M-60withthethreec-METbindingsitesisshown.Thedeletedbindingsitesareindicatedinred.(f)c-MET3󰀁UTRisatargetofa-miR-M-60.pGL3-METluciferasewild-typeandmutatedconstructswereco-transfectedwitha-miR-M-60ornegativecontrolinHEK-293Acellsandluciferaseassaywasperformed(errorbars:±SEM,P<0.05).

Wehavecombinedwell-establishedknowledgeonatedwithgreatertargetrepression.Amongcanonicalsites,miRNAtargetingtogetherwithsiRNAdesignrulesand7mer-m8and8mersitesyieldthestrongestrepression,whileempiricalobservationsonvalidatedmiRNA/targetinter-6mersitesareassociatedwithmildtoverymildefficacy.Inactionsintoapipelinewhichconsistsofthreesteps:(i)ordertoachieveasignificantrepressionofthetargets,weidentificationandfilteringofrepeatedpatterns;(ii)designhavechosentoconsideronlycanonicalsites,especiallyfa-andfilteringofa-miRsequences;and(iii)scoringandvoring7mer-m8and8mermatches.

rankingofthedesigneda-miRs.

Inordertoestimatethenumberofhuman3󰀁UTRse-ThefirststepmainlyreliesontheconceptofmiRNAquencesthatshareatleastacommon7ntpattern,wecol-seed,whichisthe5󰀁regionofthemiRNA,centeredonlectedgeneexpressiondataassociatedwithdistinctdiseasesnucleotides2–7(Figure1a).ThemiRNAseedisthemostfromtheGeneExpressionAtlas(GEA)(16)andfocusedonconservedportionofmetazoanmiRNAsandallowsthetheupregulatedgenes,thusmimickingaplausiblescenariocharacterizationofmiRNAfamilies.Theseedgenerallyfortheemploymentofa-miRs.Foreachdisease,wecalcu-matchescomplementary,oftenconserved,canonicalsiteslatedallthepossiblecombinationsoftwoandthreeupreg-onthe3󰀁UTRs(UnTranslatedRegions)ofregulatedtar-ulatedgenesandcountedhowmanyofthemshareatleastgets(11,15).Thereisevidencethatthelackofperfectseeda7mer3󰀁UTRsite.WefilteredoutpolyA-signalmotifs,ho-pairinginfunctionalbindingsitesis,attimes,balancedbymopolymermotifsandsitesmatchingtheseedsofendoge-thepresenceofcenteredor3󰀁compensatorysites.How-nousmiRNAs.Weperformedthisanalysisonallupregu-ever,thesecasesaremuchlessabundantthancanonicalsiteslatedgenepairsandtripletsasdetectedin83differentdis-whichrepresentthepredominantinteractionmodelassoci-eases,revealingthat97.3%ofpairsand81.32%oftriplets

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 2014NucleicAcidsResearch,20145

Figure3.(a)pGL3-EGFR3󰀁UTRconstructwasco-transfectedwitha-miRsornegativecontrolinHEK-293Acellsandluciferaseassaywasperformed(errorbars:±SEM,P<0.05).(b)EGFRexpressionwasassessedbywesternblotinHeLacellstransfectedwitha-miRsornegativecontrolandharvestedafter72h.a-miR-E-3anda-miR-E-106enforcedexpressiondecreasesendogenouslevelsoftheEGFRprotein.Loadingcontrolwasobtainedbyusinganti-␤-actinantibody.(c)qRT-PCRofthetransfecteda-miRsinHeLacells.(d)qRT-PCRoftheEGFRmRNAaftera-miRsenforcedexpressioninHeLacells.(e)RepresentationoftheEGFR3󰀁UTRsbindingsitesfora-miR-E-3.Inthefigurepairingoftheseedregionofa-miR-E-3withthethreeEGFRbindingsitesisshown.Thedeletedbindingsitesareindicatedinred.(f)EGFR3󰀁UTRistargetofa-miR-E-3.pGL3-EGFRluciferasewild-typeandmutatedconstructswereco-transfectedwitha-miR-E-3ornegativecontrolinHEK-293Acellsandluciferaseassaywasperformed(errorbars:±SEM,P<0.05).

shareatleastone7mersite.Onaverage,pairsandtripletsthatappearatleastonceinanyoftheprovidedsequencessharedabout136and247mersites,respectively(seeSup-(Table1).

plementarySectionS2fordetailsandadditionalanalysis).Thesecondstepofthealgorithmconsistsoftheactuala-Inlightofthis,andconsideringcasesinwhichasetofhighlymiRsequencedesign.Foreachrepeatedpatternidentifiedsimilarsequencesischosenfortargeting,wedecidedtosetinthepreviousphaseananticomplementarya-miRseedisamaximumthresholdofeighttargetsequencesthatuserscreated.Therestofthesequenceisconstructedbyaligningcanprovideasinputtothesystem.Thislimitationmakestheseed’sbindingsitesandmaximizingthematchoutsidesense,becauseeightisalreadyaconsiderablenumberoftar-theseedregionthroughasequenceprofiletechnique,asde-gets,unlikelytobepracticalinmostapplications.Thesese-pictedinFigure1c.Thea-miRsequencesthusobtainedwillquencesarescreenedforrepeatedpatternsofsixorsevenbe22ntlong.

nucleotides(dependingonuserchoice),whichwillconsti-Thedesigneda-miRsarethenfilteredbasedontheirnu-tutethebindingsitesfora-miRseeds(Figure1b).Thesecleotidecomposition,combiningwell-establishedsiRNAsitesarethenfilteredbasedonuser-providedspecifications,designruleswithendogenousmiRNAfeatures.Inpartic-e.g.asitemustappearinmultiplecopiesonthesametargetular,sequenceswithGCcontentoutoftheuser’sspeci-and/oritmustbepresentatleastonceineverytarget.More-fiedrange(23–78%bydefault)orcontainingstretchesofover,userscanalsoprovidealistofsequencesthatmustnotsixormorenucleotidesofthesamekindarediscardedbetargeted.Inthiscase,thesystemwillremoveallthesites(17,18).Theseparticularthresholdswerechosenaccording

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 20146NucleicAcidsResearch,2014

Figure4.(a)pGL3-MET3󰀁UTRandpGL3-EGFR3󰀁UTRwereco-transfectedwitha-miRsornegativecontrolinHEK-293Acellsandluciferaseassaywasperformed(errorbars:±SEM,P<0.05).(b)EGFRandc-METexpressionwasassessedbywesternblotinHeLacellstransfectedwitha-miRsornegativecontrolandharvestedafter72h.Loadingcontrolwasobtainedusinganti-␤-actinantibody.(c)qRT-PCRofthetransfecteda-miRsinHeLa.(d)qRT-PCRofthec-METandEGFRmRNAaftera-miRsenforcedexpressioninHeLacells.(e)Representationofthec-METandEGFR3󰀁UTRsbindingsitesfora-miR-ME-196.Inthefigure,pairingoftheseedregionofa-miR-ME-196withthec-MET/EGFRbindingsiteisshown.Thedeletedbindingsiteisindicatedinred.(f)MET3󰀁UTRistargetofa-miR-ME-196.pGL3-METluciferasewild-typeandmutatedconstructswereco-transfectedwitha-miR-ME-196ornegativecontrolinHEK-293Acellsandluciferaseassaywasperformed.(g)EGFR3󰀁UTRisatargetofa-miR-ME-196.pGL3-EGFRluciferasewild-typeandmutatedconstructswereco-transfectedwitha-miR-ME-196ornegativecontrolinHEK-293Acellsandluciferaseassaywasperformed(errorbars:±SEM,P<0.05).

towhathasbeenobservedintypicalendogenousmiRNAScoringandrankingofa-miRs

nucleotidecomposition(seeSupplementarySectionS2).Inthisphaseuserscanalsochoosetodiscarda-miRThethirdstepofthemiR-SynthpipelineconsistsinthesequencessharingaseedwithanyendogenousmiRNA.evaluationandrankingofthedesigneda-miRs.Wede-Moreover,userscanenablethepredictionofpotentialoff-velopedascoringfunctionbasedonsixdifferentfeaturestargetgenes.Afilterallowstheremovalofthosea-miRsofvalidatedendogenousmiRNA/targetinteractions:seedwhoseseedispredictedtobindmorethanauser-providedtype,pairingofthemiRNA3󰀁region,AUcontentofthemaximumnumberofoff-targetgenes.Alternatively,thebindingsiteanditssurroundingregions,miRNAnucleotideusercanrequestthetop10a-miRswiththesmallestnum-composition,structuralaccessibilityofthebindingsiteandberofoff-targethits.Thisisanimportantfeature,sinceapresenceofARE(AURichElement)andCPE(Cytoplas-singlea-miRmaytargeteventhousandsofdifferentgenes.micPolyadenylationElement)motifsupstreamofthebind-ThisissuewillbefurtherdiscussedintheValidationandingsites(15,19–22).Foranygivena-miR,eachfeatureisDiscussionsections.Moredetailsaboutthealgorithmandassignedascorerangingfrom0to1andatotalrepressionthefiltersaregivenassupplementaryinformation(seeSup-scoreiscalculatedbycombiningthetree-basedmultiplelin-plementarySectionS1).

earregressionlearningsystemM5PwithCTree(23,24).Wehavetrainedthesystemonasetofpubliclyavail-ablegeneexpressionprofilesfollowingtheover-expression

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 2014NucleicAcidsResearch,20147

Table1.DetailsaboutthetestedmiRNAs

SyntheticmiRNAsforc-METRankIDSequenceSites

Seedtypes

M5PscoreCTreescore160UUUGAAACGGAGGCUGUCUAGA38mer/8mer/8mer−0.261−0.2252118UUUAUAAAGUCGAUACGUGUUU38mer/8mer/8mer−0.260−0.2253181UUCUUUCUAAGGACGGGGCCGU28mer/8mer−0.253−0.2254176UCAGUACAAAACCUUGUGGCUU28mer/8mer

−0.246−0.225SyntheticmiRNAsforEGFRRankIDSequenceSitesSeedtypes

M5PscoreCTreescore13UGUGGCUUCACCUCCUGUAUCG38mer/8mer/7mer-m8−0.241−0.2252106UGUGUGACACUGCGUAAGGGGG28mer/8mer−0.238−0.225325CAAAUGCUCGAGAGUCCGAUGU28mer/7mer-m8−0.229−0.225483UAACAAUGCACUGGGGGCCCUG28mer/7mer-m8−0.228−0.225SyntheticmiRNAsforc-METandEGFRRankIDSequenceSitesSeedtypesM5PscoreCTreescore1141UUCCAAUUCGAGGGGAGGUGGG1+18mer/8mer−0.262−0.225223UCAAUUUCGGUCCCGAGUUCCA1+18mer/8mer−0.258−0.2253140UCCAAUUGGACGGGAGGUGGGU1+18mer/8mer−0.249−0.2254106UUUCAUGAGCCCUAGACUGGGG1+18mer/8mer−0.246−0.2255196UGAGUUUCUCAGCGACGGACCG1+18mer/8mer−0.241−0.225698UUUCUUAAGCACGCCGUUGGGG

1+1

8mer/8mer

−0.239

−0.225

Moredetailsaregivenassupplementaryinformation(SupplementaryTablesS2–S4).

ofnineindividualendogenousmiRNAs(15).Inparticular,choiceconstitutesagoodexampleofbeneficialemploy-bindingsiteswerepredictedforeachtransfectedmiRNAonmentofmulti-targeta-miRs,giventhereciprocalandcom-downregulatedgenes,thenfeaturevalueswerecalculated.plementaryrelationshipbetweenEGFRandc-METinac-Thegeneexpressionfoldchangewasusedasameasurequiredresistancetokinaseinhibitorsinlungcancer,andofthedegreeofrepressioninducedbythemiRNA.Thus,thenecessityofconcurrentinhibitionofbothtofurtherim-lowervaluesmeanstrongerdownregulationofthetarget.provepatientoutcomes(26).

OnlytranscriptswithsinglebindingsitesforthetransfectedWedesignedtwodifferentsetsofmulti-sitea-miRsexclu-miRNAswereconsidered,inordertoreducethechancesofsivelytargetingc-METandEGFR,respectively.Thesys-indirecteffects.

temreturned111a-miRsforc-METand59a-miRsforAccordingtotheM5Ptree(Figure1d,SupplementaryEGFR(SupplementaryTablesS6andS7).ForeachoftheTableS2),themostdiscriminantfeatureswerethenu-twogenes,wefocusedonthetopfoura-miRsasrankedcleotidecompositionofthemiRNA,thetypeofseedandbyourscoringsystem(SupplementaryTablesS3andS4).theAUcontentofthebindingsite.

SupplementaryTableS1(a)and(b)summarizesthemainDependingonthevaluesofthesethree,sixdifferentsetsfeaturesofthesea-miRs.Theeighta-miRsthustakenintoofweightswereassignedtoallofthefeatures.Onlytheseedconsiderationhadatleasttwobindingsitesontheirtar-typeandthenucleotidecompositionofthemiRNAweregets,withapredominantpresenceof8mermatches.Tover-consideredasdiscriminantfeaturesbyCTree(Figure1e).ifydirecttargeting,thewild-type3󰀁UTRsofc-METandThesetwomethodsareusedtoevaluatethedesigneda-EGFRwereclonedintopGL3controlvectorsdownstreammiRs.Inparticular,a-miRsarefirstrankedaccordingtooftheluciferaseopenreadingframe.a-miRsforc-METtheCTreescoreandsubsequentlybytheM5Pscore.CTreeandEGFRwereindividuallyco-transfectedwiththec-splitsthea-miRsintomajorclasses,whileM5PisusedtoMETandEGFR3󰀁UTRconstructs,respectively,inHEK-ranka-miRswithineachclass.

293Acells.ThisresultedinasignificantinhibitionoftheWevalidatedthisscoringfunctionbyusingadatabaseluciferaseactivityinducedbytwoc-METa-miRsandthreeofexperimentallyvalidatedhumanmiRNA/targetinterac-EGFRa-miRs,ascomparedtothenegativecontrol(Fig-tionscalledmiRTarBaseasatestset(25).Thisdatasetcon-ures2aand3a).Moreover,western-blotandqRT-PCRas-tains495casesofprovendirectinteractions,490casesforsaysshowedthatover-expressionofa-miRsinHeLacellswhichdirectbindingwasnotverifiedand71negativecases.stronglyreducedtheendogenousproteinandmRNAlev-Weconsidered1000randomlycreatedgroupswiththesameelsofc-METandEGFRascomparedtocontrol(Figuresnumberofprovendirectandprovennegativecases.Foreach2banddand3bandd),inagreementwiththeluciferasegroup,thetop10interactions,asrankedbyourapproach,assayresults.Expressionoftransfecteda-miRsinHeLaalwayscontainedahighernumberoftruedirectinteractionstransfectedcellswasconfirmedbyqRT-PCR(Figures2ccomparedtosetsof10casesrandomlychosen(P<0.0001).and3c).Amongthefivefunctionala-miRs,a-miR-M-60Amoredetaileddescriptionofthescoringfeatures,classi-anda-miR-E-3rankedfirstandyieldedstrongdownregu-ficationandvalidationprocessesisgivenassupplementarylationofc-METandEGFR3󰀁UTRsluciferaseactivity,re-information(seeSupplementarySectionS1).spectively(Figures2aand3a).Hence,asfurtheranalysis,weperformedmutagenesisofa-miR-M-60anda-miR-E-3Validationofsingle-targetmulti-sitea-miRs

bindingsiteswithintheMETandEGFR3󰀁UTRs,whichabolishedtheabilityofthesea-miRstoregulateluciferase

Oura-miRdesignsystemwasvalidatedonc-METandEGFR,twowell-knowngenesinvolvedinlungcancer.This

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 20148NucleicAcidsResearch,2014

expression,thusconfirmingthatthebindingsitesarefunc-tional(Figures2eandfand3eandf).Validationofmulti-targetsynthetica-miRs

Wesubsequentlydesigneda-miRsintendedtotargetbothc-METandEGFRconcurrently.Thealgorithmreturnedatotalof125a-miRswith7mer-m8/8mermatchesontheUTRsofbothgenes(SupplementaryTableS8).Weselectedthetopsixa-miRsasrankedbyourscoringfunction(Sup-plementaryTablesS1(c)andS5).Allofthemhadone8merbindingsiteoneachgene.

Toverifymultipledirecttargetingofc-METandEGFR,thedesigneda-miRswereindividuallyco-transfectedwithbothwild-typec-METandEGFR3󰀁UTRconstructsintoHEK-293Acells.a-miR-ME-196anda-miR-ME-141in-ducedasignificantinhibitionoftheluciferaseactivityforbothconstructs,whilea-miR-ME-140anda-miR-ME-106yieldedasignificantrepressionofc-METonly,ascom-paredtothenegativecontrol(Figure4a).Moreover,over-expressionofthea-miRsinHeLacellsinducedastrongrepressionoftheendogenousc-METandEGFRproteinsandmRNAsinthreecasesandamilddownregulationinthethreeremainingcases,ascomparedtothecontrol(Fig-ure4bandd).Interestingly,althoughnotalltesteda-miRswerefunctionalattheluciferaselevel,theeffectsontheendogenousproteins,whoserepressionrepresentsourpri-marygoal,wasmuchstronger.Thiscouldbeduetotheintrinsiclimitationsoftheluciferaseassay,beingbasedonanartificialconstruct.Nevertheless,outofthesixtesteda-miRs,a-miR-ME-196waschosenforfurtherinvestigationbecauseofitsgreaterdownregulationatboththeproteinandtheluciferaselevel(Figure4a,bandd).Theexpressionofa-miR-ME-196inHeLatransfectedcellswasconfirmedbyqRT-PCR(Figure4c).Mutagenesisofthea-miRbind-ingsitewithinthec-METandEGFR3󰀁UTRseliminateditsabilitytoregulateluciferaseexpression,thusconfirmingthatthebindingsiteisfunctional(Figure4eandf).Inor-dertofurtherdemonstratetherobustnessofthemiR-Synthscoringfunctionandtheadditionalbenefitsofincorporat-ingfeaturesotherthantheseedmatch,wetestedthebottomsixa-miRsdesignedforc-METandEGFRandfoundthatthreeofthesea-miRsyieldedamildrepressionofEGFR,lowerthanobservedforthebesttopsixa-miRs,andthatnoneofthemwasabletosignificantlyrepressc-MET,de-spitetheirgoodseedmatches(7mer/8mer)(SupplementaryFigureS2).

Onafinalnote,inordertoassessthegeneralapplicabil-ityofourmethod,weadditionallyranmiR-Synthon14,325pairsofupregulatedgenesineightdiseasesretrievedfromtheGEAdatasetmentionedabove.miR-Synthwasabletodesignatleastana-miRfor95%ofpairsandatleastsixa-miRsfor86.9%ofpairs.Thefeatureandglobalscoresofthetopsixa-miRsfromGEAwereverycomparabletothescoresofthevalidatedc-MET/EGFRtopsixa-miRs.Inparticular,thisheldtrueforfeaturessuchasAUcontentandstructuralaccessibility,whichsolelydependonthetargetse-quence,thusconfirmingtheresultsobtainedwiththe7meranalysisdescribedabove.However,whenweappliedtheoff-targetfilter,wefoundthatonly43%ofgenepairssharedatleasta7merwithnomorethan2000off-targethits,andthe

percentagedroppedto5.6%whenweconsideredgenepairssharinga7merwithnomorethan1000off-targets.Thisisanintrinsicfactorofanya-miR,duetotheshortlengthoftheseedregion,whichwithnodoubtrequirespropercon-sideration.Ourexperimentsshowed,however,thataperfectseedmatchisnottheonlyindicatorofeffectivenessandthatotherfeaturesmustbetakenintoaccount.Inlightofallthis,ouroff-targetpredictionanalysisandfiltersconstituteausefultooltohelptheuserselectthebesta-miRs.Moredetailsaregivenassupplementaryinformation(seeSupple-mentarySectionsS3andS4).ThemiR-Synthwebinterface

miR-Synthisfreelyavailableforacademicusethroughawebinterface(http://microrna.osumc.edu/mir-synth).UserscanprovideuptoeightUTRsequencesorselectthemfromamenubytheirname,RefseqaccessionnumberorEn-trezgeneID.AlthoughthesystemwastrainedonhumanmiRNAs,itallowsselectionoftargetsfromotherspeciesaswell,suchasmouseandrat.Userscaneitherrequesttodesigna-miRssimultaneouslytargetingalloftheprovidedsequencesortoincludea-miRstargetingsubsetsofthemaswell.Alistofsequences(ortheirIDs)thatmustnotbedi-rectlytargetedbythedesigneda-miRscanalsobeprovided.Intheavailableoptionsuserscanspecifythekindofseedmatchesallowed(6merand/or7mer-m8/8mer),theGC%contentrange(defaultis23–78%)andwhethertheendoge-nousmiRNAfiltershouldbeapplied.Sequencemaskscanalsobeprovided,inordertospecifyportionsoftheinputsequencesthatshouldnotbetargeted.Thiscanbeause-fuloptionwhenthepresenceofSNPs(SingleNucleotidePolymorphism)orothermutationsinthetargetscouldneg-ativelyaffecta-miRbinding(27,28).

Finally,userscanchoosetoviewthelistofpotentialoff-targetgenes,whichisobtainedthroughthecomputationofseedmatchesonthewholedatabaseofUTRsequencesfromtheselectedspecies.

Thesystemisfast.Forexample,thedesignofa-miRsforapairoftargetswithdefaultparameterstakes30satmost.However,giventhevariabilityinthenumberofin-putsequencesandthedifferentoptionsthatcanbeselected,whichcouldsubstantiallyincreasecomputationtime,usersareprovidedwiththeresultspagelinkbye-mailoncethecomputationhascompleted.Foreachindividuala-miR,de-tailsaboutinteractionfeaturesandtheirbindingsitesaregiven,includingpartialandglobalscoresalongwiththelistofoff-targetgenesandthenumberoftheirpotentialbind-ingsites,ifrequested.

Technicaldetailsaboutthedevelopmentofthewebinter-faceareprovidedintheonlinemethods.DISCUSSION

RNAiconstitutesapowerfultoolfortheregulationofgeneexpression(29,30).RecentprogressinthedevelopmentofincreasinglyefficientcarriersfortheintracellulardeliveryofsmallRNAs,suchasnanoparticlesandviralsystems,hasmadetheestablishmentoftherapeuticsbasedonthispromisingtechnologyimaginable(31–33).Moreover,newstrategiesfororaldeliveryofantisensenucleotidesandre-centfindingssuggestingthatexogenousmiRNA,suchas

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 2014thoseofplantorigin,simplyintroducedthroughfoodin-takecouldbeactiveandfunctionalinrecipientcells,opensanewscenarioinwhichRNAicouldconstituteanappealingandconcretetherapeuticaltoolforcancer,viralinfectionsandotherdiseasescausedorprogressivelymaintainedbytheover-expressionofmultiplegenes(34–36).

AlthoughtherulesforthedesignofefficientsiRNAandshRNAarenowadayswellestablished,sequencedesignmethodologiescanneverthelessbefurtherimproved,espe-ciallytoreduceoff-targeteffects.

siRNAsaredesignedtoregulatespecifictargetsthroughperfectcomplementarity,butevidenceshowsthatthepres-enceofoneormoreperfectmatchesin3󰀁UTRsequenceswiththesiRNAseedregionisassociatedwithconsider-ableoff-targeteffectsandrepresentsawidespreadandunin-tendedconsequenceofsiRNA-mediatedsilencing(37,38).Thisphenomenon,whichreflectsthenaturalbehaviorofmiRNAs,suggestsapossibleapproachfordesigningfewermoleculesthatmayreducetheexpressionofmanytargets.Infact,ourexperimentsshowthatasinglea-miRmaybeabletorepressatleasttwounrelatedgenesatthesametime,whileitmaylikelytakeapoolofdifferentsiRNAs/shRNAstoobtainthesignificantinhibitionofasinglegene.Itisveryimportanttopointoutthat,inprinciple,thereisnodiffer-encebetweenasinglemulti-targeta-miRandasingle-targetsiRNAintermsofbasicseed-basedoff-targeteffects.Anyveryshortnucleotidesequence,suchasa7mer,islikelytoappearinasubstantialnumberofUTRs.Unsurprisinglytherefore,aninsilicotestconfirmedthatdouble-targetinga-miRsarelikelytohavefeweroff-targetsthanpairsofsingle-targetingsiRNAs(seeSupplementarySectionS3).Thisindicatesasubstantialadvantageintheemploymentofa-miRsinplaceofsiRNAs/shRNAs.

ThemiR-Synthpipelineallowstherationaldesignofa-miRsbytakingmultiplefactorsintoconsideration.Itin-tegratescurrentknowledgeregardingmiRNA/targetinter-actionandfeaturessimpleyetpowerfuloptionswhichal-low,forexample,toinvestigateoff-targeteffectsanddesignmoleculesvirtuallynotaffectedbySNPsandotherpoly-morphisms.

FutureworkincludesrefinementofthedesignprocessandfurtheranalysisofmiRNA/targetinteractions,inor-dertobetterunderstandthecausalconnectionbetweenthetargetingfeaturesandthedegreeofdownregulation,andimprovetheselectionofeffectivemolecules.SUPPLEMENTARYDATA

SupplementaryDataareavailableatNAROnline,includ-ing[39–45].

ACKNOWLEDGMENTS

WethankNicolaBombieri,AlessandroDaneseandFrancescoMartinellifortheirprecioushelpwiththetest-ingofmiR-Synth.Wethanktheanonymousreviewersfortheirhelpfulcommentsandsuggestions.FUNDING

NationalInstitutesofHealth/NationalCancerInstitute[R01-CA135030,P01-CA081534toA.L.,U01-CA152758,

NucleicAcidsResearch,20149

U01-CA166905toM.A.];U.S.NationalScienceFoun-dation[DBI-0445666,DBI-0421604,NSFIOS-0922738,MCB-0929339,N2010IOB-0519985];NationalInstitutesofHealth[2R01GM032877–25A1toD.S.].Conflictofintereststatement.Nonedeclared.

REFERENCES

1.Petrelli,A.andValabrega,G.(2009)Multitargetdrugs:thepresentandthefutureofcancertherapy.ExpertOpin.Pharmacother.,10,5–600.

2.McCarty,M.F.(2004)Targetingmultiplesignalingpathwaysasastrategyformanagingprostatecancer:Multifocalsignalmodulationtherapy.Integr.CancerTher.,3,349–380.

3.Sinha,G.(2012)InterestresparksinRNAi.Nat.Biotechnol.,30,1012.4.Davis,M.E.,Zuckerman,J.E.,Choi,C.H.J.,Seligson,D.,Tolcher,A.,Alabi,C.A.,Yen,Y.,Heidel,J.D.andRibas,A.(2010)Evidenceof

RNAiinhumansfromsystemicallyadministeredsiRNAviatargetednanoparticles.Nature,4,1067–1070.

5.Rao,D.D.,Vorhies,J.S.,Senzer,N.andNemunaitis,J.(2009)siRNAvs.shRNA:similaritiesanddifferences.Adv.DrugDeliv.Rev.,61,746–759.

6.Davidson,B.L.andMcCray,P.B.(2011)CurrentprospectsforRNAinterference-basedtherapies.Nat.Rev.Genet.,12,329–340.

7.Cheng,T.L.,Teng,C.F.,Tsai,W.H.,Yeh,C.W.,Wu,M.P.,Hsu,H.C.,Hung,C.F.andChang,W.T.(2009)Multitargettherapyofmalignantcancersbythehead-to-tailtandemarraymultipleshRNAsexpressionsystem.CancerGeneTher.,16,516–531.

8.terBrake,O.,Konstantinova,P.,Ceylan,M.andBerkhout,B.(2006)SilencingofHIV-1withRNAinterference:amultipleshRNAapproach.Mol.Ther.,14,883–2.

9.Tiemann,K.,Hohn,B.,Ehsani,A.,Forman,S.J.,Rossi,J.J.andSaetrom,P.(2010)Dual-targetingsiRNAs.RNA,16,1275–1284.10.Passioura,T.,Gozar,M.M.,Goodchild,A.,King,A.,Arndt,G.M.,

Poidinger,M.,Birkett,D.J.andRivory,L.P.(2009)Interfering

ribonucleicacidsthatsuppressexpressionofmultipleunrelatedgenes.BMCBiotechnol.,9,57.

11.Bartel,D.P.(2009)MicroRNAs:targetrecognitionandregulatory

functions.Cell,136,215–233.

12.Acunzo,M.,Visone,R.,Romano,G.,Veronese,A.,Lovat,F.,

Palmieri,D.,Bottoni,A.,Garofalo,M.,Gasparini,P.,Condorelli,G.etal.(2012)miR-130atargetsMETandinducesTRAIL-sensitivityinNSCLCbydownregulatingmiR-221and222.Oncogene,31,634–2.13.Acunzo,M.,Romano,G.,Palmieri,D.,Lagan`a,A.,Garofalo,M.,

Balatti,V.,Drusco,A.,Chiariello,M.,Nana-Sinkam,P.and

Croce,C.M.(2013)Cross-talkbetweenMETandEGFRinnon-smallcelllungcancerinvolvesmiR-27aandSprouty2.Proc.Natl.Acad.Sci.U.S.A.,110,8573–8578.

14.Thomas,M.,Lieberman,J.andLal,A.(2010)Desperatelyseeking

microRNAtargets.Nat.Struct.Mol.Biol.,17,1169–1174.15.Grimson,A.,Farh,K.K.-H.,Johnston,W.K.,Garrett-Engele,P.,

Lim,L.P.andBartel,D.P.(2007)MicroRNAtargetingspecificityinmammals:determinantsbeyondseedpairing.Mol.Cell,27,91–105.16.Petryszak,R.,Burdett,T.,Fiorelli,B.,Fonseca,N.A.,

Gonzalez-Porta,M.,Hastings,E.,Huber,W.,Jupp,S.,Keays,M.,

Kryvych,N.etal.(2013)ExpressionAtlasupdate–adatabaseofgeneandtranscriptexpressionfrommicroarray-andsequencing-basedfunctionalgenomicsexperiments.NucleicAcidsRes.,doi:10.1093/nar/gkt1270.

17.Reynolds,A.,Leake,D.,Boese,Q.,Scaringe,S.,Marshall,W.S.and

Khvorova,A.(2004)RationalsiRNAdesignforRNAinterference.Nat.Biotechnol.,22,326–330.

18.Dykxhoorn,D.M.,Novina,C.D.andSharp,P.A.(2003)Killingthe

messenger:shortRNAsthatsilencegeneexpression.Nat.Rev.Mol.CellBiol.,4,457–467.

19.Betel,D.,Koppal,A.,Agius,P.,Sander,C.andLeslie,C.(2010)

ComprehensivemodelingofmicroRNAtargetspredictsfunctionalnon-conservedandnon-canonicalsites.GenomeBiol.,11(8),R90.20.Shao,Y.,Chan,C.Y.,Maliyekkel,A.,Lawrence,C.E.,Roninson,I.B.

andDing,Y.(2007)EffectoftargetsecondarystructureonRNAiefficiency.RNA,13,1631–10.

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 201410NucleicAcidsResearch,2014

21.Kertesz,M.,Iovino,N.,Unnerstall,U.,Gaul,U.andSegal,E.(2007)

TheroleofsiteaccessibilityinmicroRNAtargetrecognition.Nat.Genet.,39,1278–1284.

22.Sun,G.,Li,H.andRossi,J.J.(2010)Sequencecontextoutsidethe

targetregioninfluencestheeffectivenessofmiR-223targetsitesintheRhoB3’UTR.NucleicAcidsRes.,38(1),239-252.

23.Wang,Y.andWitten,I.H.(1997)Inducingmodeltreesforcontinuous

classes.In:vanSomeren,MandWidmer,G(eds).PosterPapersofthe9thEuropeanConferenceonMachineLearning(ECML97).Prague,CzechRepublic,128–137.

24.Hothorn,T.,Hornik,K.andZeileis,A.(2006)Unbiasedrecursive

partitioning:Aconditionalinferenceframework.J.Comput.Graph.Stat.,15,651–674.

25.Hsu,S.D.,Lin,F.M.,Wu,W.Y.,Liang,C.,Huang,W.C.,Chan,W.L.,

Tsai,W.T.,Chen,G.Z.,Lee,C.J.,Chiu,C.M.etal.(2010)miRTarBase:adatabasecuratesexperimentallyvalidatedmicroRNA-targetinteractions.NucleicAcidsRes.,39,D163–D169.

26.Suda,K.,Tomizawa,K.,Osada,H.,Maehara,Y.,Yatabe,Y.,Sekido,Y.

andMitsudomi,T.(2011)Conversionfromthe“oncogeneaddiction”to“drugaddiction”byintensiveinhibitionoftheEGFRandMETinlungcancerwithactivatingEGFRmutation.LungCancer,doi:10.1016/j.lungcan.2011.11.007.

27.Haas,U.,Sczakiel,G.andLaufer,S.D.(2012)MicroRNA-mediated

regulationofgeneexpressionisaffectedbydisease-associatedSNPswithinthe3’-UTRviaalteredRNAstructure.RNABiol.,9,924–937.28.Saunders,M.A.,Liang,H.andLi,W.-H.(2007)Human

polymorphismatmicroRNAsandmicroRNAtargetsites.Proc.Natl.Acad.Sci.U.S.A.,104,3300–3305.

29.Dykxhoorn,D.M.andLieberman,J.(2005)Thesilentrevolution:

RNAinterferenceasbasicbiology,researchtool,andtherapeutic.Annu.Rev.Med.,56,401–423.30.Rao,D.D.,Wang,Z.,Senzer,N.andNemunaitis,J.(2013)RNA

interferenceandpersonalizedcancertherapy.Discov.Med..,15,101–110.

31.Daka,A.andPeer,D.(2012)RNAi-basednanomedicinesfortargeted

personalizedtherapy.Adv.DrugDeliv.Rev.,,1508–1521.32.Kesharwani,P.,Gajbhiye,V.andJain,N.K.(2012)Areviewof

nanocarriersforthedeliveryofsmallinterferingRNA.Biomaterials,33,7138–7150.

33.Breitbach,C.J.,Burke,J.,Jonker,D.,Stephenson,J.,Haas,A.R.,

Chow,L.Q.M.,Nieva,J.,Hwang,T.-H.,Moon,A.,Patt,R.etal.(2011)Intravenousdeliveryofamulti-mechanisticcancer-targetedoncolyticpoxvirusinhumans.Nature,477,99–102.

34.Akhtar,S.(2009)OraldeliveryofsiRNAandantisense

oligonucleotides.JDrugTarget.,17,491–495.

35.Aouadi,M.,Tesz,G.J.,Nicoloro,S.M.,Wang,M.,Chouinard,M.,

Soto,E.,Ostroff,G.R.andCzech,M.P.(2009)OrallydeliveredsiRNAtargetingmacrophageMap4k4suppressessystemicinflammation.Nature,458,1180–1184.

36.Zhang,L.,Hou,D.,Chen,X.,Li,D.,Zhu,L.,Zhang,Y.,Li,J.,Bian,Z.,

Liang,X.,Cai,X.etal.(2012)ExogenousplantMIR168aspecificallytargetsmammalianLDLRAP1:evidenceofcross-kingdomregulationbymicroRNA.CellRes.,22,107–126.

37.Jackson,A.L.,Burchard,J.,Schelter,J.,Chau,B.N.,Cleary,M.,Lim,L.

andLinsley,P.S.(2006)WidespreadsiRNA‘off-target’transcriptsilencingmediatedbyseedregionsequencecomplementarity.RNA,12,1179–1187.

38.Birmingham,A.,Anderson,E.M.,Reynolds,A.,Ilsley-Tyree,D.,

Leake,D.,Fedorov,Y.,Baskerville,S.,Maksimova,E.,Robinson,K.,Karpilow,J.etal.(2006)3󰀁UTRseedmatches,butnotoverallidentity,areassociatedwithRNAioff-targets.Nat.Methods,3,199–204.

Downloaded from http://nar.oxfordjournals.org/ by guest on March 20, 2014