胰腺癌臨床級外泌體的生成和檢測Generation and testing of clinical-gradeexosomes for pancreatic cancer

JClinsIGHT

Generationandtestingofclinical-gradeexosomesforpancreaticcancer

MayelaMendt,l2SushrutKamerkar,'HikaruSugimoto,'KathleenM.McAndrews,

Chia-ChinWu,3MihaiGagea,?SujuanYang,'ElenaV.RodrigesBlanko,'QianPeng,'XiaoyanMa,5

JosephR.Marszalek,?AnirbanMaitra,?CassianYee,?KatayounRezvani,2ElizabethShpall,?ValerieS.LeBleu,'andRaghuKalluri

'DepartmentofCancerBiology,MetastasisResearchCenter,DepartmentofStemCellTransplantationandCellular

Therapy,DepartmentofGenomicMedicine,'DepartmentofVeterinaryMedicine&Surgery,CenterforCo-ClinicalTrialsandInstituteforAppliedCancerScience,DepartmentsofPathologyandTranslationalMolecularPathology,Ahmad

CenterforPancreaticCancerResearch,and/DepartmentsofMedicalMelanomaandStemCellTransplantationandCellularTherapy,UniversityofTexasMDAndersonCancerCenter,Houston,Texas,USA.

Exosomesareextracellularvesiclesproducedbyallcellswitharemarkableabilitytoefficiently

transfergeneticmaterial,includingexogenouslyloadedsiRNA,tocancercells.Here,wereportonabioreactor-based,large-scaleproductionofclinical-gradeexosomesemployinggoodmanufacturingpractice(GMP)standards.AstandardoperatingprocedurewasestablishedtogenerateengineeredexosomeswiththeabilitytotargetoncogenicKras(iExosomes).Theclinical-gradeGMPiExosomesweretestedinmultipleinvitroandinvivostudiestoconfirmsuppressionofoncogenicKras

andanincreaseinthesurvivalofseveralmousemodelswithpancreaticcancer.Weperform

studiestodeterminetheshelflife,biodistribution,toxicologyprofile,andefficacyincombinationwithchemotherapytoinformfutureclinicaltestingofGMPiExosomes.Collectively,thisreport

illustratestheprocessandfeasibilityofgeneratingclinical-gradeexosomesforvarioustherapiesofhumandiseases.

Authorshipnote:MMandSK

contributedequallytothiswork.

Confictofinterest:TheUniversityofTexasMDAndersonCancerCenterandRKarestockequityholdersinCodiak

Biosciences.RKreceivesresearch

supportfromCodiakBiosciencesandservesasamemberoftheboardof

directors.VSLservedasaone-timepaidconsultantforCodiakBiosciences.

Submitted:December14,2017

Accepted:March14,2018

Published:April19,2018Referenceinformation:

JCInsight.2018;3(8):e99263.https:///10.1172/jci.insight.99263.

Introduction

Exosomesarenanosizedvesiclesreleasedbycells,andtheyparticipateinintercellularexchangeofDNA,RNA,proteins,andothercellularcomponents.Thecapacityofexosomeshasspurredarenewedinterestintheirutilityasadeliverysystemforvarioustherapeutics(1-10).Unlikeliposomesandsyntheticnanoparti-cles,thenaturalfeaturesofexosomesmayofferuniqueadvantagesfortheefficientdeliveryoftherapeuticpayloadsintotumors.Theregulatorymachineryofexosomesthatisassociatedwithproductionandcellularuptakeremainslargelyunknown.Nonetheless,theengineeringofexosomesfortherapeuticcontrolofdiseas-es,includingcancer,indicatedpromisingresults(11-16).

WerecentlyreportedontheutilityofengineeredexosomesderivedfromfibroblastsforthedeliveryofsiRNAtargetingoncogenicKrasinthetreatmentofpancreasductaladenocarcinoma(PDAC)inmice(16).PreviousstudiesshowedthatoncogenicKras-controlledmacropinocytosisinpancreascancercellsincreasesexosomesuptake(17,18)andenhancestheefficacyofexosomesindeliveringtheirtherapeuticpayloadtopancreatictumors(16).SuchresultshighlightedthepotentialutilityofexosomeswiththeabilitytotargetoncogenicKras(iExosomes)forthetreatmentofpatientswithPDAC,anaggressiveandlethalcancerwithlimitedtherapeuticoptionsthatisontherise.

AmajorbottleneckintheadvancementofiExosome-basedtherapyintotheclinicisthedevelopmentofhighscaleandefficientproductionofclinical-gradeexosomes.Thiswouldrequiresterilegenera-tionofexosomeswiththerapeuticpayloads,producedinsufficientamountsforclinicaltesting,with-outbatch-to-batchvariationleadingtocompromisedefficacy.Therefore,wedevelopedaprocessforproductionofgoodmanufacturingpractice-grade(GMP-grade)exosomesderivedfrombonemarrowmesenchymalstem/stromalcells(MSCs).MSCshavebeeninfusedinpatientswithvariousdisorderswithoutanysignificantsideeffects(19).Additionally,bonemarrow-derivedMSCshaveemergedasanattractivecellularsourceforthegenerationofclinical-gradeexosomesforhumantherapies(20,21)Here,wereportontheproductionprocessandthepotentialofMSCsinthegenerationandengineering

RESEARCHARTICLE

/10.1172/jci.insight.99263

1

RESEARCHARTICLE

JClinSIGHT

ofiExosomesforhumantrialsandtestedtheefficacyofGMP-gradeMSCderived-iExosomesinseveralassaysandmousemodelsofpancreaticcancer.

Results

iExosomestargetoncogenicKrastosuppresspatient-derivedPDACxenograft.Weperformedglobalgeneexpres-

sionprofilingofuntreated/controlandiExosome-treatedpancreaticcancercellswithKrasG2Dmutation(Panc-1cells)orwild-typeKras(BxPC-3cells).Theexpressionprofilesofuntreatedcellswerecomparedwithexpressionprofilesofcellstreatedwithcontrolexosomes(nosiRNApayload),exosomeswithscram-bledsiRNAcontrol(siScrblExo),orexosomeswithKrasGI2DsiRNA(siKrasG12DiExo).Principalcompo-nentanalysesandhierarchicalclusteringanalysesofgeneexpressionrevealedaspecificsegregationofthetranscriptomicprofilesofPanc-1cellstreatedwithsiKrasG12DiExo,comparedwithallothercontrolgroups(SupplementalFigure1,A-C;supplementalmaterialavailableonlinewiththisarticle;https://doi.org/10.1172/jci.insight.99263DS1).ThiswasnotobservedwhenBxPC-3cellsweretreatedwithsiKrasG12DiExo,andthetranscriptomicprofileofsiKrasGl2DiExo-treatedBxPC-3cellsdidnotsegregatefromthatofBxPC-3controls(SupplementalFigure1A).VolcanoplotsofsignificantlyderegulatedgenesdepictasignificanteffectonthetranscriptomeofPanc-1cellstreatedwithsiKrasGi2DiExocomparedwithallothercontrols(Figure1AandSupplementalFigure1,BandC),whereasthetranscriptomeofBxPC-3cellswaslargelyunaffected(Figure1AandSupplementalFigure1,DandE).TheseresultsindicatethatsiKrasG12DiExospecificallyaffectedcellswithKrasG12Dmutationincontrastwithcellswithwild-typeKras.GenesetenrichmentanalysesofthederegulatedtranscriptomeofPanc-1cellswithsiKrasG12DiExocomparedwithallothercontrolsindicatedastrongupregulationofgenesassociatedwithproteasome,lysosome,andphagosomepathwaysandothers(SupplementalFigure1F).DownregulatedtranscriptswereassociatedwithresponsetomineralocorticoidandnuclearchromosomesegregationandpositiveregulationofSmadproteinsignaltransduction,chemokinereceptorsandchemokines,andresponsetocalcium(SupplementalFigure1F).TheseanalysesreflectpreviouslyreportedtranscriptomicchangesassociatedwithKrassignal-ingandoncogenicKrastargeting(22,23)andofferadditionalinsightsintomechanismsassociatedwithcelldeathbysiKrasGI2DiExotreatment.

Next,weconfirmedtheefficacyofsiKrasG12DiExoinmiceorthotopicallyimplantedwithpatient-de-rivedxenograft(PDX)thatharborstheoncogenicKrasG12Dmutation(SupplementalFigure2A).MicewithPDXweremonitoredfortumorburdenbyultrasoundimagingandMRI;bothmodalitiesreflectedsimi-larmeasurementsintumorvolumes(SupplementalFigure2B).MRIwaschosenasthemodalityforfol-low-upimaging,andtheresultsindicatedthat,whilecontrolmicetreatedwithsiScrblExoprogressedwithincreasedtumorvolumesovertime,micetreatedwithsiKrasGl2DiExoshowedregressionoftumorvolumeovertime(Figure1,BandC).ThetumorburdencontrolachievedwithsiKrasG12DiExotreatmentwasasso-ciatedwithasignificantincreaseinsurvivalofmicecomparedwithcontrolsiScrblExotreatment(Figure

1D).siKrasG12DiExotreatmentwasstopped(Figure1D)whenallcontrolmicesuccumbedtopancreaticcancerandpresentedwithsubstantialtumorweightatendpoint(Figure1E).Notethatattheexperimentalendpoint,thepancreasweightofmicetreatedwithsiKrasGi2DiExowassignificantlyreducedcomparedwiththatofcontrols(Figure1E).Grossobservationsatnecropsyandhistopathologicalevaluationoftis-suesrevealedthatmicefromthesiKrasGi2DiExotreatmentgrouphadmarkedlyreducedtumorburdenandlessaggressivetumorsincomparisonwithmicefromsiScrblExotreatmentgroup(Figure1FandSupple-mentalFigure2,CandD).SomeofthemicewithsiKrasGI2DiExotreatmentdisplayedalifespansimilartothelifespanexpectedwithuntreatednudemice(Figure1D).Attheexperimentalendpoint,someoftheoldmicewithsiKrasG12DiExotreatmentpresentedwithminimalormildinflammationofthebowelandkidneys(SupplementalFigure2,CandD),whicharecommonage-relatedlesionsinnudemice,astheymaypresentwithincreasedsusceptibilitytoinfection.

MSC-derivediExosomessuppressPDACinmice.Totranslatethepreclinicalfindingsintotheclinic,we

optedfortheuseofGMP-gradecompliantnormaldonorbonemarrow-derivedMSCs,approvedforclin-icaluseinpatients,toproduceexosomes.WeevaluatedexosomeproductionofMSCsfrom3donorsandchoseMSCsfromdonor3forsubsequentexperimentsandGMPproductionofExosomesbasedonitssuperiorexosomeproductionrate(SupplementalFigure3A).Insomeofourpreclinicalstudies,weutilizedMSCsfromdonor2(foroneofthepreclinicalmodels,seeMethods)andhumanforeskinfibroblasts(BJfibroblasts)obtainedfromATCC(ref.16andFigure1,B-D).BothBJfibroblastsandMSCsdisplaysim-ilarmorphology(Figure2A)andsimilarsurfacemarkerexpression(MSCmarkersandputativeexosome

/10.1172/jci.insight.99263

2

RESEARCHARTICLE

JClinSIGHT

A

B

-log10(pvalue)

-log10(pvalue)

Tumorvolume(mm3)

log2foldchange

BxPC-3

log2foldchange

C

D—siScrbliExo

Percentsurvival

—siKrasG12D-1iExo-

]**

Days(Postbirth)

Tumorweight(g)

E

siKrasG12D-1iExosiScrbliExo

F

(410days)(321days)

Figure1.SpecifictargetingofiExosomesforhumanPDACcellswithKrasC2e.(A)Volcanoplotsdepictinglog,foldchange(red,upregulatedgenes;

blue,downregulatedgenes;gray,genesthatwerenotsignificantlyderegulated)and-log,?(Pvalue)ofdifferentillyexpressedgenesbetween

sikras2D-1iExo-treatedPanc-cellandallontrolsandBxPC-3cells.(B)TumorvolumemeasuredbyMRIofsikrasGC2D1iExo(n=7)orsiScrbliExo(n=7)atbaseline(dayO=day62aftertumorinduction)andaftertreatment(day30,81,and228posttreatmentstart[PTS]).#,nomeasurementavailable;micedied.(C)RepresentativeMRlimagesofPDxpancreastumors;ayllowdashedlineenciclesthetumors.(D)Kaplan-Meiercurveindicatingthe

survivalofPDXmiceinthelistedtreatmentgroupsafterbith(sikrasOD-1iExo[n=7],siscrbliExo[n=7];log-rank(Mantel-Cox)test).Theapproximat-edtimeforthenaturallfespanofnudemiceiindicatedbythereddottedline.(E)TumorweightatendpointinstudieswithPDXmicesikrasC20iExo

/10.1172/jci.insight.99263

3

RESEARCHARTICLE

JClinsIGHT

[n=7],siScrbliExo(n=7).(F)RepresentativeH&EimageofaggressvelyinvasivepancreatictumorsinsiscrbliExo-treatedmice,incomparisonwithpredominantinflammationandmarkedlyreducedtumorsinpancreataofsikrasGR1Exo-treatedmice.Scalebars:100μm(left);50μm(right).The

dataarepresentedasthemean±SEM.Unlesotherwisestated,unpaired2-tailedttestwasusedtodeterminestatisticasignificance.*P<0.05,**P<0.01.SeeSupplementalSourceData1and2.

markers),asevaluatedbyflowcytometry(SupplementalFigure3,BandC).BothBJfibroblastsandMSCswerepositiveforCD44,CD90,CD105,CD29,CD73,andHLA-ABC(SupplementalFigure3,BandC)TheywerebothnegativeforCD45,CD31,HLA-DR,andCD34(SupplementalFigure3,BandC).TheybothexpressedexosomemarkersCD81,CD9,andCD63(SupplementalFigure3,BandC).BJfibroblastswerealsoCD10+,incontrastwithMSCs(SupplementalFigure3,BandC).BothcellsexpressedhighlevelsofCD47,previouslycharacterizedasimportantforimpactfuliExosometherapy(16)(SupplementalFigure3,BandC).BJfibroblastsandMSCsproducedhighlevelsofexosomes,withamodeandmeanof107nmand161nmforBJfibroblast-derivedexosomesand108nmand179nmforMSC-derivedexosomes,respectively,Figure2,BandC).TheMSCexosomeproductionlevelwassignificantlyhigherwhencom-paredwiththatofBJfibroblasts,withanoverallhighercountofexosomespercellovertime(Figure2,BandC).TheexosomesfrombothBJfibroblastsandMSCsexhibitedasimilarpresenceofexosomemarkers(CD9,CD63,CD81)andCD47,althoughthemesenchymalmarkers(CD29,CD90)werepredominantlynotedontheMSC-derivedexosomes(Figure2,DandE).

TogeneratelargeamountsofexosomesfromMSCs,thebioreactorcultureofbonemarrow-derivedMSCswasadaptedtoenabletheharvestof6different250-mlcollectionsofconditionedmedia(Supple-mentalFigure4A).Exosomeswereenrichedbyfltrationandultracentrifugation(SupplementalFigure4,BandC;seealsoMethods),andexosomenumbersrangedbetweenapproximately900billionandapproximately4,500billionperharvest(Figure3AandSupplementalFigure5A).Threedistinctbioreac-torexperimentswererun,andthesumnumberofexosomesgeneratedperbioreactorrunrangedbetween9.8and15.6trillionperbioreactorrun(SupplementalFigure5B).Theaveragesizeofexosomesfromharvestsofabioreactorrun(run1)showedamodeof165.9nmandameanof202.2nm(Figure3B).

Themeasurementoftotalexosomalproteincontentcorrelatedwiththeexosomescounts,asdeterminedbyNanoSightanalyses(Figure3C).Themetabolicreadoutsinthebioreactor(glucoseandlactose)duringthecourseofthebioreactorharvestofconditionedmediaforthe6differentcollectionsremainedcon-stant(SupplementalFigure5C),suggestingthattheMSCsremainedequallyviableduringtheentirepro-cess.Electronmicroscopyanalysesandflowcytometryanalysesofexosomesfromeachoftheharvestsrevealedthepresenceofexosomes(Figure3D),withconsistentpresentationofkeyexosomesbiomarkers

(CD9,CD63,CD81)andCD47(Figure3EandSupplementalFigure5D).All6bioreactorexosomeharvestswerefrozenaftereachcollection.Whenthefinalcollectionwascompleted,theharvestswerethawedandpooledforsubsequentisolationandelectroporationofsiRNAagainstKrasG12DandgenerationofiExosomesforfurthertesting(SupplementalFigure4,BandC).

TheefficacyofMSC-derivediExosomestotargetoncogenicKrasandinduceapoptosisofPanc-1cellswassimilartothatoffreshlypreparedBJfibroblast-derivediExosomes(Figure4,A-C).Forthisexperi-ment,MSC-derivedexosomeswerecollectedandfrozenat-80℃for2weeks.Afterthawing,theMSC-de-rivedexosomesandfreshlypreparedBJfibroblast-derivedexosomeswereelectroporatedaspreviouslydescribed(16)withsiRNAGI2D(siRNAGI2Dsource1[siRNAGI2D-];seeMethods).MSC-derivediExosomesinducedrobustapoptosisofPanc-1cells(Figure4,AandB)andsignificantlydownregulatedoncogenicKrasexpression(Figure4C)whencomparedwithrelevantcontrols(untreatedcells,Panc-1cellstreatedwithcontrolexosomeswithoutsiRNAcargo,andPanc-1cellstreatedwithsiScrbliExo).Previously,adefinedelectroporationbuffer(researchbuffer[RB])wasusedtointroducesiRNAintoexosomes,whichnecessitatedawashstepoftheiExosomespriortotreatmentofcellsormice(16).TheRBisnotapprovedforhumanclinicaltesting.ThewashstepwasassociatedwithalossofiExosomes(datanotshown).Toalleviatethishurdle,weemployedadiluentknownasPlasma-Lyte,referredtohereasclinicalbuffer(CB),thatenabledthesuccessfulelectroporationofthesiRNAintoexosomesanddirectuseforadministrationtocellsormicewithoutanadditionalwashstep.Plasma-Lyte(CB),anFDA-approveddiluentforhuman

use,hasbeenusedtodiluteMSCsforinfusionsintopatients.FollowingelectroporationofMSC-derivedexosomesinCB,electronmicroscopyanalysesconfirmedthepresenceofintactexosomes(Figure4D)BothRBandCBenabledthegenerationofiExosomeswithsimilarefficacy,inducedapoptosisofPanc-1cells(Figure4E),andsuppressedtheexpressionofoncogenicKras(Figure4F).

/10.1172/jci.insight.99263

4

JClinsSGHT

B

ABJfibroblastsMSCs

Numberofexosomes

C

Numberofexosomesperm

percellper48hour

Size(nm)

D

%ofMax

CD9CD63CD81

Figure2.ExosomeproductionbyMSCs.(A)Rep-resentativebright-fieldimagesofBJfibroblasts

andMSCs.Scalebar:100μm.(B)Comparisonof

thenumberofexosomesquantifiedbyNanoSight,producedbythesamenumberofBJfibroblasts

andMSCs,andcollectedfromtheconditioned

mediaoveraperiodof48hours(n=3independentcollections).(C)ParticlesizedistributionanalysisofBJfibroblastexosomesandMSCexosomesbyNanoSight.(DandE)Representativehistogram

offlowcytometryanalysisofexosomalmarkers

%ofMax

CD47CD29CD90

(CD9,CD63,CD81),CD47,andmesenchymalmarkers(CD29,CD90)onBJfibroblasts(red,D)exosomes

andMSCexosomes(blue,E).Numbersrepresentthepercentageofpositivebeads(gray,isotypecon-trol).Thedataarepresentedasthemean±SEM.Unpaired2-tailedttestwasusedtodetermine

statisticalsignificance.**P<0.01.SeeSupplemen-talSourceData1and2.

RESEARCHARTICLE

E

%ofMax

CD9CD63CD81

%ofMax

CD47CD29CD90

WenexttestedtheefficacyofiExosomesfromBJfibroblastsandMSCs,producedusingRBorCB,inPDACmodels.iExosomesinducedapoptosisofKPC689cells(SupplementalFigure6A).IntheKPC-derivedorthotopicmodel(KPC689)(16),wetestedBJfibroblast-(freshlyisolatedandsubjectedtoelectroporation)andMSC-derivedexosomes(frozenfor45daysafterisolation,thawed,andthensubjectedtoelectropo-ration)electroporatedwithsiKrasGl2D-1usingRB(Figure5,A-G).BothBJfibroblast-andMSC-derived

/10.1172/jci.insight.99263

5

RESEARCHARTICLE

JClinsIGHT

A

B

Numberofexosomesperml

Numberofexosomes(10?)

C

Size(nm)

Protein(mg)

D

Bioreactorharvest1Bioreactorharvest2Bioreactorharvest3Bioreactorharvest4Bioreactorharvest5Bioreactorharvest6

Harvest1

%ofMax

E

CD9CD63CD81CD47

Harvest6

%ofMax

CD9CD63CD81CD47

Figure3.GMP-gradeproductionofMSC-derivedexosomes.(A)Thenumberofexosomes,quantifiedbyNanosight,producedby6consecutive

48-hourisolations(harvests)ofMSC-conditionedmediafromthebioreactor.(B)ParticlesizedistributionanalysisusingNanoSight.(C)Correlationbetweenthenumberofexosomesandexosomalproteinfromthebioreactorharvests(Pearsoncorrelationtest).DatashowninA-Carerepresenta-tiveofthesamedata,obtainedfromthesamebioreactorexperiment(Bioreactorrun1)(seealsoSupplementalFigure5A).(D)RepresentativeTEMofexosomesfromeachofthe6consecutivebioreactorharvests.Scalebar:100nm.(E)Representativehistogramofflowcytometryanalysisof

exosomalmarkers(CD9,CD63,CD81)andCD47onexosomesfrombioreactorharvests1and6(seealsoSupplementalFigure5D).Numbersrepresentthepercentageofpositivebeads(gray,isotypecontrol).SeeSupplementalSourceData1and2.

/10.1172/jci.insight.99263

6

RESEARCHARTICLE

JClinSIGHT

A

MSCsCtrlExo

MSCssiScrbliExoMSCssiKrasG12D-1

iExo

LIVE/DEADTM

FixableAquaStain

BJsiKrasG12D-1iExo

AnnexinV

AnnexinV

AnnexinVAnnexinV

Untreated

AnnexinV

B

C

expression(foldchange)

RelatveKRASe120

LVE/DEADTM

FixableAquaStain

EUntreatedResearchBuffer(RB)ClinicalBuffer(CB)

AnnexinVAnnexinVAnnexinV

ClinicalBufferResearchBuffer

D

(CB)(RB)

Percentageofapoptoticcells

RelativeKRASG120

expression(toldchange)

F

Figure4.ValidationofGMP-gradeiExosomeefficacyinvitro.(A)Representativedotplotand(B)quantificationofflowcytometryanalysisofapop-tosisinPanc-1cellinducedbyMSCCtrlExo,MSCsiscrblExo,MSCsikrasG2-1Exo,orBJsikrasG20-1Exoafter48hours,comparedwithuntreated

cells.Numbersrepresentthepercentageofpositivecells(n=4independentexperiments,1-wayANOVAcomparedwithuntreated).(C)KRASGRtran-scriptlevelsinPanc-1celltreatedwithMSCCtrExo,MSCsiscrbliExo,MSCsikras201Exo,orB/C21Exoafter3hours,comparedwithuntreated

cells(n=4independentexperiments,1-tailedunpairedttest).(D)RepresentativeTEMofMSCexosomes,afterelectroporation,usingeitherresearchbuffer(RB)orclinicabuffer(CB).Scalebar:100nm.(E)RepresentativedotplotofflowcytometryanalysesandquantificationofapoptosisinPanc-1cellsuntreatedortreatedfor48hourswithMSCsikrasO201iExoelectroporatedusingeitherRBorCB.Numbersrepresentthepercentageofpositivecells(n=4independentexperiments,1-wayANOVAcomparedwithuntreated).(F)KRASCDtranscriptlevelsinPanc-1cells(n=3independentexper-iments,1-tailedunpairedttest).Themean±SEMisdepicted.*P<0.05,**P<0.01,****P<0.0001.SeeSupplementalSourceData1and2.

/10.1172/jci.insight.99263

7

RESEARCHARTICLE

JClinsIGHT

B

A

Days

CControlExoBJsiKrasG12D-1iExoMSCssiKrasG12D-1iExo

DControlBJsiKrasG120-1MSCsiKrasG12D-

ExoiExoiExo

Day51Day28

1.0

0.8

0.6

x105

-0.4

0.2

◎

Radiance

(p/sec/cm2/sr)

F

E

TotalFluxat

Day51(p/sec)

RelativeKRASG12D

expression(foldchange)

G

H

Survival(days)

Days

ControlExoBJsiKrasG12D-1iExo(RB)MSCssiKrasG12D-1iExo(RB)MSCssiKrasG12D-1iExo(CB)

Figure5.ValidationofGMP-gradeiExosomeefficacyinvivo.(A)Kaplan-MeiercurveindicatingsurvivalaftertumorinductionofmicewithKPC689

orthotopictumorsinthelistedtreatmentgroups(ControlExo[n=4],BJsikrasCO20-iExo[n=6],MSCSssiKrasO201Exo[n=6];log-rank[Mantel-Cox]test).(B)SurfacelungnodulesofKPC689mice(ControlExo[n=4],BJsikras?-1iExo[n=6],MSCssikrasC-1iExo[n=6]).(C)RepresentativeH&E-stainedlung

/10.1172/jci.insight.99263

8

RESEARCHARTICLE

JClinsIGHT

sectionsfromKPC689mice.Tumormetastasisisindicatedbyadashedyllwine.Scalebar:100μm.(D)RepresentativeimagesofluciferaseactivityofKPC689tumorsatday28andday51aftertumorinduction(ControlExo[n=4],BJsikrasC201iExo[n=6],MSCssikrasC20-iExo[n=6]).(E)KPC689orthot-opictumorgrowth(bioluminescence)andtotalfluxatday51aftertumorinduction(ControlExo[n=4],BJsikrasCD-1Exo[n=6],MSCssikrasC2-1Exo

[n=6]).(F)KrasSDtranscriptlevelsinKPC689tumorsatendpointinthelistedexperimentalgroups(ControlExo[n=4],BJsikrasT?1Exo[n=5],MSCssikrasSC2D-1iExo[n=5];1-tailedunpairedttest).(C)Corelationbetweensurvivaland1/dCTforKrasSC20transcriptlevelsinKPC689tumors(Pearsoncor-

relationtest)(blackdots,ControlExo[n=4];reddots,BJsikrasDiExo[n=5];bluedots,MSCssikrasC21Exo[n=5).(H)Kaplan-MeiercurveindicatingthesurvivalofPanc-1tumor-bearingmiceaftertumorinductioninthelistedtreatmentgroups(ControlExo[n=5],MSCssiKrasO201Exo,CB[n=5],MSCssikrasCOD-1iExo,RB[n=4],BJsikKrasO201Exo,RB[n=5];log-rank(Mantel-Cox)test).#,pancreaswasnormalandmicewerenotmoribund(seeSupple-

mentalFigure7Afordetails.()Represent