Does
the
proteasome
inhibitor
bortezomib
sensitize
to
DNA-damaging
therapy
in
gastroenteropancreatic
neuroendocrine
neoplasms? – A
preclinical
assessment
in
vitro
andin
vivo
✩
,
✩✩
FranziskaBriesta,b,1,∗;EvaJ.Koziolekc,d,e; JakobAlbrechtd;FränzeSchmidtc,d,e,f;
MoniqueR.Bernseng;JoostHaeckg;AnjaA.Kühlh; DagmarSeddinga,i;TeresaHartunga;
SamanthaExnerj;MartinaWelzelk; ChristianFischerk;CarstenGrötzingerc,j; WinfriedBrennerc,d,l;RichardP.Baumm,n; PatriciaGrabowskia,o,p
a Department of Hematology, Oncology and Tumor Immunology, Charité-Universitätsmedizin Berlin, Corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin and Berlin Institute of Health, Berlin, Germany
b Department of Biology, Chemistry, and Pharmacy, Institute of Chemistry and Biochemistry, Freie Universität (FU) Berlin, Berlin, Germany
c German Cancer Consortium (DKTK), Germany
d Department of Nuclear Medicine, Charité Universitätsmedizin Berlin, Corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin and Berlin Institute of Health, Berlin Germany e German Cancer Research Center (DKFZ), Heidelberg, Germany
f Institute for Biochemistry and Biotechnology, Martin-Luther-University (MLU) Halle-Wittenberg, Halle (Saale), Germany
g Department of Radiology, Erasmus MC, Rotterdam, The Netherlands
h iPATH.Berlin, Campus Benjamin Franklin, Charité – Universitätsmedizin Berlin, Corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Berlin
i Institute of Biology, Humboldt-Universität (HU) Berlin, Berlin, Germany
j Department of Hepatology and Gastroenterology and Molecular Cancer Research Center, Tumor Targeting Laboratory, Charité-Universitätsmedizin Berlin, Corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin and Berlin Institute of Health, Berlin, Germany k Experimental and Clinical Research Center, a joint cooperation between the Charité Medical Faculty and the Max-Delbrück Center (MDC) for Molecular Medicine, Berlin, Germany l Berlin Experimental Radionuclide Imaging Center (BERIC), Charité Universitätsmedizin Berlin, Corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin and Berlin Institute of Health, Berlin, Germany
m Department of Nuclear Medicine, Zentralklinik Bad Berka GmbH, Bad Berka, Germany n CURANOSTICUM Wiesbaden-Frankfurt, DKD Helios Clinic, Wiesbaden, Germany o Department of Gastroenterology and Endocrinology, Zentralklinik Bad Berka GmbH, Bad Berka, Germany
p Department of Medical Immunology, Charité Universitätsmedizin Berlin, Corporate member of Freie Universität Berlin, Humboldt-Universität zu Berlin and Berlin Institute of Health, Berlin, Germany
Abbreviations:AP-sides,apurinicorapyrimidinicsites;BER,baseexcisionrepair;BRCA1/2,breastcancerprotein1/2;BRIP1,BRCA1InteractingProteinC-TerminalHelicase1;CAM, chorioallantoicmembrane;CK2,caseinkinase2;d,day(s);DDR,DNAdamagerepair;DSB,double-strandbreaks;DSMZ,DeutscheSammlungfürMikroorganismenundZellen;ER, endoplasmicreticulum;ERK,extracellular-signal-regulatedkinase;F-18-FDG-PET,2-deoxy-2-(F-18)fluoro-D-glucosepositronemissiontomography;FOXM1,forkheadboxproteinM1; FOXO3a,forkheadboxproteinO3a;g,gram;G0/1/2phase,gapphase0/1/2;G1/2/3,grade1/2/3;Ga-68,Gallium-68;GEP-NEN,gastroenteropancreaticneuroendocrineneoplasms;GO, GeneOntology;h,hour(s);H2AX,histoneH2AX;HR,homologousrecombination;HTCA,humantumorcolonyassay;i.p.,intraperitoneal;i.v.,intravenous;IC50,doseofhalf-maximum inhibitoryeffect;JC-1,5,5,6,6-tetrachloro-1,1,3,3-tetraethylbenzimidazolcarbocyanineiodide;Lu-177,Lutetium-177;M,molar;Mphase,mitosisphase;(M)Bq,(Mega-)Becquerel; MMR,mismatchrepair;MRI,magneticresonanceimaging;n,number;NEN,neuroendocrineneoplasms;NER,nucleotideexcisionrepair;NHEJ,non-homologousendjoining;p70S6K, ribosomalproteinS6kinase;PARP1,poly(ADP-ribose)polymerase1;PI3K,phosphoinositide3-kinase;PRRT,peptidereceptorradionuclidetherapy;PTEN,phosphataseandtensin homolog;Rb,retinoblastomaprotein;ROS,reactiveoxygenspecies;RPA,replicationproteinA;(RT)-PCR,(real-time)polymerasechainreaction;Sphase,synthesisphase;SCLC,smallcell lungcancer;SKP2,S-PhaseKinaseAssociatedProtein2;SSB,single-strandbreaks;SSTR2,somatostatinreceptor2;TGF,transforminggrowthfactor;TLS,translesionsynthesis.
∗Correspondingauthor.
E-mailaddress:franziska.briest@charite.de(F.Briest).
✩ Funding:ThisworkwassupportedbytheTheranosticsResearchNetwork.TheprojectwasfurthersupportedbyaYoungInvestigatorsinNeuroendocrineNeoplasiaforGermany GrantfromNovartisPharmaGmbH.ThisworkwassupportedinpartbytheDeutscheForschungsgemeinschaft(DFG)forPET/MRIuse(INST335/454-1FUGG)andinpartbythe TechnologiestiftungBerlin(TSB)forSPECT/CTuse.Thefundingsourceshadnoinvolvementintheconductoftheresearchand/orpreparationofthearticle,instudydesign,inthe collection,analysisandinterpretationofdata,inthewritingofthereport,andinthedecisiontosubmitthearticleforpublication.
✩✩ConflictsofInterests:Theauthorsdeclarethattheyhavenoknowncompetingfinancialinterestsorpersonalrelationshipsthatcouldhaveappearedtoinfluencetheworkreportedinthis paper.FBandPGweresupportedbytheTheranosticResearchNetworkGermany.ThegroupofDr.PatriciaGrabowskireceivesfinancialsupportfromIpsenPharma,NovartisandPfizer. FBandPGreceivedatravelgrantandalecturesalaryfromIpsenPharma.
1Franziska Briest, Department of Hematology, Oncology and Tumor Immunology, Charité-Universitätsmedizin Berlin, Augustenburger Platz 1, Berlin, Germany, franziska.briest@charite.de.
Received7September2020;receivedinrevisedform1November2020;accepted3November2020
© 2020TheAuthors.PublishedbyElsevierInc.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/)
Abstract
Background: Well-differentiatedgastroenteropancreatic neuroendocrineneoplasmsareraretumorswitha slowproliferation.They are virtuallyresistanttomany DNA-damagingtherapeuticapproaches,such aschemo- and external beamtherapy,which might be overcomebyDNAdamageinhibitioninducedbyproteasomeinhibitorssuchasbortezomib.Methodsandresults:Inthisstudy, weassessed severalcombinedtreatmentmodalities invitroand invivo. Bycell-basedfunctional analyses, ina3Dinovo and an orthotopicmousemodel,wedemonstratedsensitizingeffectsofbortezomibcombinedwithcisplatin,radiationandpeptidereceptor radionuclidetherapy(PRRT).Bygeneexpressionprofilingandwesternblot,weexploredtheunderlyingmechanisms,whichresulted inanimpairedDNAdamagerepair.Therapy-inducedDNAdamagetriggeredextrinsicproapoptoticsignalingaswellastheinduction ofcellcyclearrest,leadingtoadecreasedvitaltumorvolumeandalteredtissuecompositionshownbymagneticresonanceimaging and F-18-FDG-PETinvivo,howeverwithnosignificantadditionalbenefitrelatedtoPRRTalone. Conclusions:Wedemonstrated thatbortezomibhasshort-termsensitizingeffectswhencombinedwithDNAdamagingtherapybyinterferingwithDNArepair in vitroandinovo.Nevertheless,duetohightumorheterogeneityafterPRRTinlong-termobservations,wewerenotabletoprovea therapeuticadvantageofbortezomib-combinedPRRTinaninvivomousemodel.
Neoplasia(2021)23, 80–98
Keywords:Bortezomib,Neuroendocrinetumors,Peptidereceptorradionuclidetherapy(PRRT),DNArepair,Combinationtherapy,Proteasomeinhibition
Introduction
Well-differentiated neuroendocrine neoplasms (NEN)of the pancreas and the intestineare raretumors, characterizedbyslowproliferationand the absenceofcommondrivermutations.Althoughmutationswithinthe phosphoinositide3-kinase(PI3K)pathwayareonlypresentinupto14%of pancreaticNENs [1],primarilyepigeneticdysregulationresultsintheloss of mTOR-associatedtumorsuppressorsandrendersthetumorsaccessible formTORinhibitors,suchaseverolimus [2,3].Moreover,ATRX,DAXXor MEN1mutationstriggeraberrantchromatinstructureandgeneticinstability, whichimpairsthefunctionofothertumorsuppressors,suchasTP53,ATM orCDKN2A [4–7].
Based on the NETTER-1 randomized clinical trial, peptide receptor radiotherapy (PRRT) using Lu-177-DOTATATE has been approved for somatostatin receptor 2 (SSTR2)-positive, nonresectable or metastatic, progressive, well-differentiated G1 and G2 NENs due to significantly increased progression-free survival (median time-to-progression was 36 months) [8–10].Thesuggestedmechanismisthatafterbindingofthe Lu-177labeledligandtoSSTR,theemittedbetaparticlesinduceDNA single-anddouble-strandbreaks(SSBandDSB)inSSTR2-positivetumorcells[11]. AccumulatedSSBsandDSBsblocktranscriptionand/orDNAreplication untilrepairedorfinallyinducecelldeathinthecaseofrepairfailure [12]. Although diseasecontrolratesarecomparablysatisfyingand canimprove patients’ quality of life, relatively low OR rates indicate mechanisms of baselineoracquiredresistancetoPRRTfromamolecular-biologicalpoint ofview.
Highlyproliferative cancersfrequentlydeveloptherapy-resistanttumor areasbyselectingsubclones,whichareabletocleartherapy-induceddamage atanincreasedpace.Incontrast,theslowproliferationofwell-differentiated neoplasmsperseprovidesextendedtimeforDNAdamagerepair(DDR). The maintenanceof DNAdamagebyinhibition ofdamagerepairmight therefore bea noveltreatment approachinboth,fastprogressingcancers and slowlyproliferating tumorssuchaswell-differentiated pancreaticand intestinalNENs.
Accordingly,chemotherapeuticagentshavelimitedapplicationpotential in well-differentiated NENs, where combined regimens containing streptozotocin (STZ) and 5-fluorouracil (5-FU), or capecitabine (CAP) and temozolomide (TEM) are deployed for the therapy of progressive or metastatic pancreatic NENs. Platinum-based regimens are only recommended for NEC and highly proliferative, progressive G3 tumors according tothe current guidelines [13,14], yet with verylimited effects inwell-differentiatedtumors [15].Althoughtherearecanonicalguidelines forthetreatmentofwell-differentiatedG1andG2NENsaswellaspoorly differentiatedG3NECs,thereisnoestablishedtherapyforG3NETssofar. ThisproducestheneedtoadaptregimensthatareusedforG2NENsorG3 NECs(suchasPRRTorChemotherapy)tofittothespecificfeaturesofhigh proliferativeandwell-differentiatedG3NETs.
Forinvitroexperiments,theuseofradiopharmaceuticalsislimiteddueto radiationsafetyissuesduringlongexperimentalhands-ontimes.Tomimic PRRT inthese experiments,we chosecisplatinand radiation, sincetheir expectedDNAdamageresponseprofilesweremostlikelytobecomparable tothoseafterLu-177-DOTATATE.Whereasantimetabolites(suchas5-FU and itsprodrug CAP) inhibit DNA and RNAmetabolism byinhibition ofthymidylatesynthetase,orthroughincorporationintoRNAandDNA, monofunctionalalkylatingagents(suchasSTZandTEM)generateDNA adducts,andbifunctionalplatinum-basedregimens(suchascisplatin)induce inter-andintrastrandDNAcrosslinks.Thereby,alkylatingandcrosslinking drugsinterferewithDNAreplicationbystericinterference,whichleadsto inhibitionofreplicationandDNAstrandbreaks,eventuallyrequiringDNA repairmechanismssimilartothoseofPRRTrepair.
In this study, we discuss the targeting of proteins that are involved inmechanisms,whichrepaircisplatin-andPRRT-inducedDNAdamage. Thisincludesaccumulation ofintermediates, suchassingle-strandbreaks (SSB). Accumulated double-strand breaks(DSB) and SSB finally induce apoptosisdirectlyorbyblockingtranscriptionand/orDNAreplication[12]. If celldeathisnotinducedandcell cycleprogresses,DNA repairdefects lead to severe replication failure and chromosomal instability. Besides a majorroleindegradationofdisposableproteins,the
ubiquitin-proteasome-system is highly involved in the regulation of DNA damage response. Onthe one hand,the stabilizationof wildtype p53 toward itsnegative regulator MDM2 (which is frequently upregulated in pancreatic NENs
[6,16])mightleadtoareactivationof“sensing” mechanismsandinductionof aproapoptoticdamageresponseinp53wildtypeneoplasms[7].Ontheother hand,(poly-) ubiquitinationstronglyregulates the activityandspecificity ofaveryhighnumberofDNA-damagerepairproteinsandpostreplicative bypassmechanisms[17–19].Furthermore,ubiquitinationmodulateshistone conditionsandtherebycontrolstheaccessibilityofDNAdamagelocitoDNA repairproteincomplexes [19].
Theproteasomeinhibitorbortezomibhasbeenshowntoinhibitadvanced stagesofDNA-damagerepairwhilemaintainingearlystagedamageresponse mechanisms,suchasphospho-H2AXandreplicationproteinA(RPA)foci accumulation. The preservation of early steps inthe context of damage repairinhibitionmightbeessentialfortheinductionofcontrolledcelldeath ratherthanuncontrolledevents [20].Bortezomibhasfurtherdemonstrated chemotherapy-sensitizing effects inseveral cancers in preclinical settings, wherewe were able to showan increased antitumoral effect incisplatin plusbortezomib-treatedsmallcelllungcancerinvivo [21,22].Nevertheless, bortezomibhasbeen withdrawnfrom clinicalassessmentasmonotherapy afteraverysmallclinicalphaseIIstudyincluding14patientswithmixed neuroendocrinetumorentitiesduetofailureofobjectivetherapyresponse
[23]in2004.Thus,itspotentialincombineddrugtherapyofGEP-NENshas notbeenanalyzedsofar.Inthestudypresentedherein,weanalyzedtheeffect oftheproteasomeinhibitorbortezomibinpreventingtherepairofcisplatin-, radiation- and Lutetium-177-induced DNA-damage in pancreatic and intestinalneuroendocrinetumorsindifferentpreclinical models.Thereby, weexploredclinicallyrelevantandpotentialagentswithcomparabledamage signaturesforcombinedtherapy.
MaterialandMethods
Celllines
ThefollowingGEP-NENcelllineswereusedforinvitroexperiments: pancreatic:BON [24]andQGP-1 [25](obtainedfromJapaneseCollection of Research Bioresources), ileal: KRJ-I [26] and colonic: LCC-18 [27]. BONcellswereagenerousgiftfromCMTownsend(UniversityofTexas, Galveston).KRJ-ImidgutNETcellsweregeneratedbyR.Pfragner(Medical UniversityofGraz)andkindlyprovidedbyI.Modlin(YaleUniversity,New Haven).TheneuroendocrineoriginandphenotypeofKRJ-Icellshasrecently been questioned [28]. However, these cells have been shown to share a numberofrelevantpropertieswithotherGEP-NENcelllines[29,30].
Cellswereculturedandhandledaspreviouslypublished [31]. Allcell lineswere authenticated (ifindicated as unique) bygenetic STR typing at the DSMZ, Braunschweig, Germany in 2012, 2013 and 2015 (in vitroexperiments werefinishedin2017).Cellswere notpassagedlonger than 20 passages after receipt. Cells were further tested periodically for maintainedcelllinespecificexpressionofneuroendocrineanddifferentiation markers(chromograninA,synaptophysin,cytokeratin,vimentin,syntaxin) byimmunofluorescencemicroscopy.
ImmunocytochemistryandestablishmentofSSTR2-positiveBONcells NativeBONcellsshowedalowSSTR2expression,whichwasverifiedby immunofluorescencemicroscopyusingaSSTR2-antibody(Supplementary materialand methods) and astandard immunocytochemistry protocolas previouslypublished [16].
BON cells were transfected with human SSTR2 to generate BON-SSTR2+,a stablyexpressingSSTR2-positiveBON cellline [32].Briefly, BON-SSTR2+ cellswereproduced bytransfectingwildtype BON cells usingtheplasmidpcDNA3.1-huSSTR2(#SSTR200000,cDNAResource
Center, Bloomsberg, PA, USA; www.cdna.org) and jetPEI transfection reagent(Polyplus-transfection,Illkirch,France).Positivecloneswereselected byadditionof 600μg/mL G-418tothe mediaand weretestedfortheir SSTR2 expression by qPCR [32], immunofluorescence and radioligand bindingassay.
Lu-177-DOTATOCradioligandbindingassay
Cellular uptake of Lu-177-DOTATOC by native BON and BON-SSTR2+cellswasmeasuredasdescribedpreviously[32].Briefly,cellswere seededin24-wellplates (2.5× 105 cells/well).Radioligand bindingassay
includingthewashingsteps,wereperformedinHanks’balancedsaltsolution (HBSS,Invitrogen),containing10mMHEPES,0.5%BSAatpH7.3.The HBSSincubationbufferfurthercontainedapprox.0.0074MBq/mL Lu-177-DOTATOC.AfteronehourofincubationwithHBSS/Lu-177-DOTATOC at37°C,wellswererapidlywashedwithice-coldwashingbuffer(50mM Tris-HClpH7.4,125mMNaCl,0.05%BSA),beforelysiswith1NNaOH.Cell lysatemeasurements(quintuplicates)wereperformedinγ -counter(Wallac 1470Wizard,PerkinElmer,USA)
Westernblot
SDSpageandwesternblotofNP-40lysedprimarytumororcellline material(cellsweresynchronizedin0.01%FBSfor24hbeforetreatment) was performed using a standard protocol and documented by ponceau Sstaining. Primary antibodies are listed in Supplementary material and methods.SecondaryantibodieswereobtainedfromDako(swineantirabbit IgG-HRP and goat anti-mouse IgG-HRP; Agilent, Santa Clara, USA). AntibodybindingwasdocumentedbyFujifilmLAS-4000luminescentimage analyzerusingECLprimeWesternBlottingdetectionreagent(Amersham GEhealthcare).Forreprobing,membranesweretreatedwithacidicglycine bufferaspublished [31].Chemiluminescencesignalsweredensitometrically analyzedwithMultiGaugeV3.1orImageJ1.53.Valuesof≥3independent experimentswere normalized to internal controlsand mean values were statisticallyassessedbyusingIBMSPSSStatisticsv22.0.0.0(basedonPython 2.7)orGraphpadPrism8.4.1.
Substancesandradiopharmaceuticals
Bortezomib(velcade)andcisplatinwereobtainedinphysiologicalsaline from the Charité Universitätsmedizin Berlin dispensary. DOTA-D-Phe1-Tyr3-octreotide(Lu-177-DOTATOC) was obtainedfrom ABX advanced biochemicalcompounds GmbH,Germany. Lutetium-177 waspurchased from ITG Isotope Technologies (Garching, Germany) and synthesized to Lu-177-DOTATOC in our Laboratories at the Berlin Experimental RadionuclideImagingCenter(BERIC).
Viabilityassay
WST-1 (Roche; Basel, Switzerland) or AlamarBlue (Thermo Fisher, Waltham, USA) assays were performed according the manufacturer’s instructionsinquintupletsafter treatment withdecreasingconcentrations ofthesubstancesortheircombination.Signaldensitywascolorimetrically quantified with a multi-well spectrophotometer (TECAN sunrise) and analyzedwithMSExcel2013andGraphPadPrism7.
Combinedconcentration-responsecurveswereanalyzedasrecommended bytheChou-Talalaymethod[33].DatawasanalyzedbyuseofPrism7and CompuSyn1.0 [34].
Humantumorclonogenicassay
Sterile methylcellulose solution (MC) was prepared as follows: 4% MC was dissolved in 50:50 sterile water/Iscove´s MDM. 0.06%
2-Mercaptoethanol (BME) solution in sterile PBS and 3% (w/v) agar solutioninsterilewaterwereprepared.Prewarmedagar-solutionwasdiluted 1:3680incellculturemedium.Cellswereharvested,countedanddiluted to30,000 cells/mL. MCsolution wasprewarmed and workingsolution waspreparedincluding57%MC(4%)solution,43%FBS,0.001%BME (0.06%) solution, 0.01% cell culture medium, 0.005% cell suspension. Workingsolutionwasmixedwithdilutedagarsolutionand1mL/discwas transferredinHumantumorclonogenicassay(HTCA)disksandincubated inthecellincubator.Colonieswerecountedanddatawasanalyzedbyuseof GraphPadPrism7software.
Radiation
KRJ-IandBONcellswereirradiatedwith0,5,or10Gyandtreatedwith increasingconcentrationsofbortezomib(Bon:1–50nM,KRJ-I0.1–20nM) for48h.ViableKRJ-I(floating)cellswerequantitatedbyalamarBlueCell ViabilityAssaybyThermoFisherScientific(Waltham,USA)accordingtothe manufacturer’sinstructions.BON(adherent)cellswerecountedafterDAPI staining.DatawasanalyzedusingGraphPadPrism7software.
Combinedconcentration-responsecurveswerestudiedasrecommended bytheChou-Talalaymethod [33].DatawasanalyzedbyuseofCompuSyn 1.0 [34].
Invivomodels
Chickenchorioallantoicmembranexenografts[35]
Fertilized pathogen free chicken eggs (VALO Biomedia, Cuxhafen, Germany)weremaintainedasdescribedearlier [36].
On day 7 of embryonic development, BON-SSTR2+ cells(1× 107
cellsinanequalvolumeofRPMI1640[Invitrogen,Germany]andmatrigel [basementmembranematrixhighconcentrationfromCorning,USA])were graftedontothechorioallantoicmembrane(CAM)andmaintainedfor5d toallowangiogenicconnectionofthetumorstotheCAManddegradation of matrigel [37].Fortreatment,CAMs wereeitherinjected intravenously with approximately20 MBqLu-177-DOTATOCortumors weretreated bydropping80μLbortezomib(25nM)orNaCl(0.9%,negativecontrol) ontomicrotumors.Aseconddoseof25nMbortezomibwasappliedtothe bortezomibonlyandcombinedtreatmenttumorsonday14.Onday18, tumor plaqueswereexplanted and analyzed byWesternblot. Fordetails onmultimodalimagingoftheinovomodelseeSupplementalmaterials& methods.
Finalconcentrationsofbortezomibwerecalculatedbyadoseescalation study (datanotshown)with respecttothetoxicityof bortezomibtothe chickenfetusassumingabloodvolumeofapproximately1.2mLontheday oftreatment[38].
OrthotopicpancreaticNETmousemodelandmultimodalimaging
Forthedevelopmentofanorthotopic,pancreaticNETxenograftmouse model,BON-SSTR2+cellswereimplantedintothepancreasof12-wk-old SCIDmiceaspreviouslydescribedbyAristizabalPradaetal. [39].
Tumorgrowthwasmonitoredonceaweekbyapreclinical1TPET/MRI (nanoScanPET/MRI,Mediso,Hungary)usingamousewholebodycoil.For tumordelineationandvolumetry,ahigh-resolutionT2w2Dfastspinecho (FSE)sequencewasestablished.
Whenthetumorsizereached∼1000 mm3,themetabolicactivityand
tumorheterogeneityoftheengraftedtumorswasdeterminedbysimultaneous F-18-FDGPET/MRI(approximately15MBqinjectedi.v.).Animalswere scannedafter30mintoensuretraceraccumulationinthetumortissue.
When tumors reached a size of ∼1400 mm3, the experiment was
terminated. Animal experimentswere performed inaccordance with the nationalandlocalguidelinesforanimalwelfareandapprovedbytheethics committeeofthestateBerlin(G0011/16).
Fordetailedimagingparameters,pleaserefertoSupplementarymaterial andmethods.
Dataanalysisassessingtumorheterogeneity
Theviabletumorareawasdeterminedaspreviouslydescribed[40],using PMODsoftwareversion3.505(PMODTechnologies,Zürich,Switzerland). Briefly,a wholetumorregionofinterest(ROI)wasplacedontothe T2w imagesand copiedtofusedPET/MRIimages.Usinga thresholdof30% of the hottestvoxel within the ROI, FDG-avidtissue wasautomatically delineated within the tumor. We defined viable tissue as the FDG-avid fractionofthewholetumor.T2imageswereusedtoanalyzetheamountof necrotictissuewithinthetumors.AcustomMatlabscript(Mathworksv15.0, USA)wasusedtoidentifythenecroticproportionatefromviable/undefined tissuebasedonSIpervoxel.AROI wasmanuallydrawntodelineatethe tumorfromothertissues.AsecondROIwasdrawnina regionof tumor toannotateviabletissueSI(vSI)asareferencevalueforthenecrotictissue. ThestandarddeviationofthevSIwascalculatedandusedtodeterminethe necrotictissue,wherevoxelvaluesof3standarddeviationslower(outsidethe 99.7%confidenceinterval)thanthevSIweredeemedtobenecrotic[41].The amountofvoxelscontainingnecrotictissuearerepresentedasapercentageof thewholetumor.
Treatmentregime
Mice weretreatedaccording to thefollowing treatment schedule:0.5 mg/kgbortezomibi.p.ondays0,3,7,and10inthefirsttreatmentcycle and0.3mg/kgbortezomibondays21,24,28,and31inthesecondcycle. The combined treatment group receivedadditional 30 MBq of Lu-177-DOTATOCi.v.ondays0and21.ThePRRTonlytreatmentgroupreceived 30MBqofLu-177-DOTATOCondays0and21aspreviouslypublished
[42].Vehiclecontrolwassterilesaline.Animalswereobservedformaximum 15weeksafterstartoftreatmentoreuthanizedatatumorvolumeof1400 mm3.MRIoftumorsizewasperformedonceperweek.
Immunohistochemistry
Paraffin sections (1–2 μm) of formalin-fixed tissues were prepared from 2 different tumor areas, dewaxed and subjected to heat-induced epitoperetrievalpriortoincubationwithanti-cleavedPARP1(cloneE51, Abcam), anti-53BP1 (polyclonal rabbit, Novusbio) or anti-Ki67 (clone MIB1, Agilent). For detection,Dako REAL EnVision DetectionSystem or for detection of MIB1, alkaline phosphatase-labeled streptavidin and chromogenRED(bothAgilent)wereemployed.Nucleiwerecounterstained with hematoxylin (Merck) and sections were coverslipped with glycerol gelatin(Merck).
ImageswereacquiredusingtheAxioImagerZ1microscope(CarlZeiss MicroImaging,Inc.).Allevaluationswereperformedinablindedmanner.
HIF1-alpha,cPARP,and53BP1immunohistochemistrywasquantitated bya12-pointimmunoreactivityscoring(IRS) [43]integratingtheproduct oftheexpressionlevel(0=no;1=low;2=intermediate;3=strong)with percentageofpositivecells(0=no;1=<10%;2=10%–50%;3=51%– 80%;4=>80%).NecrosisandKi-67weredisplayedaspercentageofcells. Cellcycleanalysis
For mitotic index flow cytometry, cells were treated with 50 nM bortezomib, 10 μM cisplatin or their combination for 24 hand stained followingapreviouslypublishedprotocol[31].ForJC-1staining,cellswere harvestedaftertreatmentwith50 nMbortezomib,10 μMcisplatinor50 nMbortezomib+10μMcisplatinandsuspendedinPBS.Positivecontrols wereincubated1:500withthegenericmitochondrialmembranedepolarizer Carbonylcyanidem-chlorophenylhydrazone(CCCP)for5min.Allsamples (negativecontrolsexcluded)wereincubatedwith1:500JC-1andincubated 30minat37°C.FlowcytometrywasconductedwithFACSCalibur(Becton Dickinson)byBDCellQuestProsoftwareandanalyzedwithFlowJo8.7 software,Prism7andIBMSPSSStatistics22.
nCountermultiplexgeneexpressionanalysis
BON cells were treated in triplets with 25 nM bortezomib, 10 μM cisplatinortheircombination versusDMSO for24 hortransfectedwith siRNA as explained below. Messenger RNA (mRNA) was isolated with RNeasyMini Kit (Qiagen) according to the manufacturer’s instructions. RNA was measured using nanodrop (Thermo Fisher Scientific). 60 ng RNAwas analyzed using the PanCancer pathway panel kit (NanoString Technologies,Seattle,WA,USA)accordingtothemanufacturersinstruction
[44,45]andnSolverv2.5(basedonRv3.1.1).Thedifferentiallyexpressed (DE)genesofthegeneexpressionanalysiswereanalyzedwithnSolverv2.5 software(NanoStringTechnologies)usingfirstprincipalcomponentanalysis andregressionanalysiswithandwithoutBenjamini-Yekutieliprocedure.Raw datawasnormalizedtothefollowinghousekeepinggenes:AGK,DDX50, EIF2B4,ZC3H14,CNOT10, MRPS5,PRPF38A,NUBP1,AMMECR1L, PIAS1,HDAC3,ACAD9,EDC3,RBM45,NOL7,USP39,COG7,ZNF384, SF3A3,VPS33B,SAP130,PIK3R4,TLK2,SLC4A1AP,ZKSCAN5,ZNF346, MTMR14,ERCC3,CNOT4,TMUB2,C10orf76.
PathwayvisualizationwasdonewiththenSolversoftwarebasedonthe KEGGontologypathwaysystem[46–48].
To compare bortezomib treated cells with cells under FOXM1 knockdown,weperformedRNAinterferencefor72hasdescribedbelow. Knockdownefficiencywasdeterminedas93%to94%FOXM1reduction. Messenger-RNAofFOXM1knockdownandcontrolcellswaspreparedand analyzed as described above. Raw data was normalized to the following housekeepers:CNOT10,SF3A3,DDX50,USP39,PIAS1,SLC4A1AP,FCF1, NOL7,SAP130, HDAC3,EIF2B4, ZC3H14,ERCC3,PRPF38A,EDC3, ZNF384, TMUB2, ACAD9, AMMECR1L, VPS33B, TLK2, C10orf76, CNOT4,AGK,PIK3R4,MTMR14,ZNF143,MRPS5,TRIM39,ZKSCAN5, DNAJC14.
Matchingdifferentiallyexpressedgeneswereidentified.Agenepathway overrepresentation analysis was performed with the http://pantherdb.org/
databaseusingthePANTHEROverrepresentationTest(release20,170,413, referencelist: homosapiens)basedon the“panther pathway” annotation dataset(PANTHERversion12.0Released2017–07–10)withBonferroni correctionformultipletesting [49].Geneenrichmentwasestimatedbyuse oftheENRICHRplatform [50,51].
RNAinterference
Cells were transfected with 40 pmol/mL siRNA in appropriate amountsof Lipofectamine3000(Life Technologies,Carlsbad, CA,USA) accordingtothemanufacturer’sinstructionsfor72h.An endoribonuclease-prepared heterogeneous siRNA pool was used in order to enhance specificityandreduceoff-targeteffects:enhancedsiRNA(esiRNA)against FOXM1bySigma(EHU124431:NCBIreferencesequences:NM_021953, NM_202002, NM_202003) or negative control: Sigma Mission siRNA Universalnegativecontrol#1.KnockdownefficiencywasmeasuredbyqPCR. RealtimeqPCR
RNAwas isolated by useof the Qiagen RNeasy RNA isolation kit, quantifiedwithNanodrop2000andDNAdigestingwascarriedoutwith 0.1 μL/100 ng RNA DNase I peqGOLD (Peqlab, VWR International GmbH,Erlangen, Germany).1000ng RNAwastranscribedinto cDNA usingtheHigh-CapacityRNA-to-cDNAKitaccordingtothemanufacturer’s instructions(ThermoFisher,CarlsbadCA,USA).Realtimequantitative(q) PCRwasperformedwith25ngcDNAwiththeStepOneRealtimePCR System(ThermoFisher)andanalyzedwithStepOnev2.3SoftwareandMS ExcelusingtheCt-Method.FigurewaspreparedwithPrism7software. ThePCRprimerswereusedasfollows:FOXM1forward5GGAGCAGCG ACAGGTTAAGG3,FOXM1reverse5GTTGATGGCGAATTGTAT CATGG3;GUS(Housekeeper)forward5GAAAATATGTGGTTG GAGAGCTCATT3,GUSreverse5CCGAGTGAAGATCCCCTT TTTA3(allMetabion).
Statisticalanalyses
Data (excludingtheNanostringgeneexpressionanalysis)wasanalyzed usingIBMSPSS Statisticsv22.0.0.0(basedonPython2.7),Prism7and 8 (v7.04 and v8.4.1; GraphPad Software, San Diego, USA). Unpaired Students t test was used for 2 groups and one-way analysis of variance (ANOVA)or2-wayANOVAformultiplegroupsofnormallydistributed data,respectively.Kruskal-WallisTestwasappliedformultiplegroupdata withrejectedassumptionofnormality.Dunn’sorSidak’scorrectionwasused forcorrectionofmultipletesting.Thechoicewasdependentonthestructure ofthedata(numberofgroups,comparisonwithoneormultiplegroups),low numbersofsamples(<5)werealwaysanalyzedwithnon-parametrictests, parametrictestswerechosenaftertestingfornormalityindatawithn≥5. Normalitywasassumedafter Shapiro-Wilk normalitytest. Alldifferences wereconsideredtobesignificantwithalpha=0.05.
Results
BortezomibhasantiproliferativeeffectsGEP-NENandsensitizesto cisplatintherapybyinductionofG2/Marrestanddeathreceptor-driven apoptosisinGEP-NENcelllines
To analyze the effects of bortezomib monotherapy, we treated the cell lines BON, LCC-18 and KRJ-I with increasing concentrations of bortezomibfor24,50,and72handanalyzedthecellviability.Allcelllines respondedsignificantlyinatime-andconcentration-dependentmannerby thereductionofcellviability.TheIC50valuesfor72hoftreatmentranged between4.7nMinKRJ-Iand270nMinLCC-18cells(Figure1).
Inordertostudythechemosensitizingeffectsinneuroendocrinetumor celllines,wechosecisplatinduetoitslowmonotherapeuticeffectinthecell lines.Subsequently,weperformedviabilitystudiesafter24h(notshown)and 50hofcombinedtreatmentversusmonotherapycontrol.Here,bortezomib exhibitedadditiveandsynergisticeffectscombinedwithcisplatin,leadingto furtherdecreasesinIC50-values(Table1).InKRJ-I,representingtheonly p53wildtypecelllineinthisstudy,bortezomibaloneinducedaverystrong effect(includingthelowestIC50valuesofalltestedcelllines),whichwasnot furtherenhancedbycisplatininthemajorityof theexperimentalsettings. AsdemonstratedforBONcells,theantiproliferativeeffectwasprovenby HTCA.Here,bortezomibsignificantlyreducedthecell’scapacitytoform newcolonies and inducedstrong morphological changes (Supplementary Fig.1).
WhilecisplatininducedS-phasearrestinaccordancewiththenecessity to repair DNA crosslinks before entering the DNA replication step
[52], bortezomib induced G2 and G2/M arrest respectively, indicating thatcrucial G2/M checkpointmediators were affectedby the treatment. The combination of both substances significantly induced apoptosis in both cell lines with an earlier induction in KRJ-I cells (Supplementary Figure1).
Toclarifytheunderlyingsignaltransduction,weanalyzedBONcellsafter 24hofcombinedtreatmentbytargetedmRNAprofilingontheNanostring nCounterplatform [45].Theaffectedsignalingpathways(Figure2)showed onlyslightchangesuponcisplatinonlytreatment,includinganupregulation of DDR and cell cycle-associated genes. In contrast, bortezomib alone andthecombinedtreatmentexhibitedlargeeffects,especiallyoncellcycle regulatoryandtranscription-associatedgenes(Figure2A–C).The pathway-allocatedgenegroupsincludeboth,positiveandnegativeregulatorygenes. Here, the apoptotic signaling varied from an upregulation of primarily proapoptoticgenesuponbortezomibmonotherapytooveralldownregulation of anti-apoptotic genes in the combined treatment arm, resulting in an overall proapoptotic signaling in either case. Furthermore, transcripts of thedeathreceptor-relatedapoptotic pathways(encodingcaspase7and8,
Figure1. Therapeuticeffectofbortezomibinvitro:(A)BON,LCC-18andKRJ-Icelllinesweretreatedwithincreasingconcentrationsofbortezomibfor24h, 50h,and72h,respectively,andcellviabilitywasmeasuredbyWST-1proliferationassay.Dose-responsecurveandhistogramofthesamedata(representative dataoutofthreeindependentexperimentsisshown,barrepresentsmeansandSDofn=5replicates)isshownforamoreconcisepresentation.Allcell linesshowedasignificantantiproliferativeresponsetothetreatment(one-wayANOVA:P<0.0001;andHolm-Sidak’smultiplecomparisonstest;normality wasassumedafterboth,Shapiro-WilkandKolmogorov-Smirnovnormalitytest).Doseofrelativehalf-maximalinhibitoryeffect(IC50)wasdeterminedby non-linearregressionusingPrism7.Captions:∗P<0.05;∗∗P<0.0001.(B)Invitrocombinatoryeffectsofcisplatinandbortezomibafter50hoftreatment: celllinesweretreatedwithacombinationofbortezomibandcisplatin1:10(startingfrom100or50μMcisplatin+10or5μMbortezomib,respectively,and subsequentdilutions)versusthecorrespondingsingleagents(representativeexperimentofthreeindependentexperiments,barshowsmedianandinterquartile rangeofn=5replicates).DoseresponsecurveswereanalyzedaccordingtothemethodofChouandTalalay.BONcellsrespondedinallmeasureddilutions (excludingthestartingconcentrations)withsynergisticeffects.InLCC-18cells,bortezomibinducedsynergismindoses>10nM,whichislowerthanthe respectiveIC50.KRJ-Icells,whichshowedverystrongeffectsevenatlowconcentrationsofbortezomibalone,exhibitedaverystrongresponse,however, synergismonlysubstantiatedforintermediatedoses(500nMcisplatin+50nMbortezomiband5μMcisplatin+500nMbortezomib)(2-wayANOVAwith Sidak’smultiplecomparisonstest).∗P<0.05;∗∗P<0.01;∗∗∗P<0.001;∗∗∗∗P<0.0001.
tumornecrosisfactorreceptors,I-kappaBkinasesubunitgamma, calpain, BIDandFAS)wereoverrepresentedaftercombinedtreatment(Figure2B). Here, as proven by western blot, the strongly increased DNA damage, indicatedbyhistoneH2AXphosphorylation,didnotincreasetheexpression of BAX as mediator of intrinsic apoptotic processes (Figure 2D). In contrast,cisplatintreatmentaloneinducedaslight,butnotsignificant,BAX
upregulation.Thecombinationofbortezomibpluscisplatinratherinduced theexpressionofdeathreceptors(FasandTRAIL-receptors).Thesignalsof FasandTRAILreceptorsarecanonicallytransducedby Fas/TNF-receptor-associated proteins with death domain (FADD/TRADD) and cleavage of caspase 8, which in turn cleaves BID and activates effector caspases, which eventually results in apoptotic events such as in the cleavage of
Table1
IC50valuesaftercisplatin,bortezomibandcombinedtreatmentofcelllinesinvitro:95%confidenceintervals (CI)ofIC50valuesafter50hoftreatment.Forcombinedtreatmentwithafixratioofbortezomibtocisplatinof 1:10,therespectiveIC50valuesofthebortezomibcontentisshown.
CellLine 95%CI(IC50Cisplatin) 95%CI(IC50Bortezomib) 95%CI(IC50Combination,Givenc[Bor])
BON 4.960to11.54μM 0.01963to0.05009μM 0.01148to0.01921μM LCC-18 9.056to31.38μM 0.03125to0.1359μM 0.05334to0.07376μM KRJ-I 2.386to6.931μM 0.003470to0.008452μM 0.001668to0.006319μM
Poly-(ADP-ribose)-polymerase(PARP)inthenuclei[53].Thewesternblot onlyconfirmedaslightupregulationofFADD(notsignificant),butverified the significantly increase in caspase and PARP cleavage. The additional cleavageofBIDcouldbeonlydemonstratedforKRJ-I,QGP-1andLCC-18 cells(SupplementaryFigure2)
The cell cycle-related effects of bortezomib were further explored by westernblot(Figure2D).Here,wefoundasignificantdownregulationof
S-phaseKinaseAssociatedProtein2(SKP2).TheFOXM1targetSKP2is a subunit of the SCF E3 ubiquitin-proteinligase complex and a crucial cellcycleregulator, whichrecognizesphosphorylatedp27 andisinvolved in regulation of G1/S transition [54]. Loss of p27 and overexpression of FOXM1have been described asfrequentevents inGEP-NENbefore
[31,55]. FollowingslightlyreducedSKP2expression,wedemonstratedan upregulationofp27andp21,independentfromRb(whichcantransactivate
p21 expression [56]and inhibitSKP2-mediatedp27ubiquitination [57]) and without effect on Rb Serin-780 phosphorylation. In particular, the phosphorylation of Rb did notchange after bortezomib(andcombined) treatment, whereas the total protein abundance showed a tendency to decrease(althoughnotsignificantly),whichcouldbeduetotheenrichment of G2/M arrested cells (as demonstrated in Supplementary Figure 1). Accordingly,theM-phase proteinaurorakinaseA (auroraA)wasslightly more abundant in the bortezomib or combination treated cells without reachingsignificance.ComparableresultscanbeshownforotherGEP-NEN celllinesanduponbortezomib-combinedradiationtherapy(Supplementary Figs.2and3).
BortezomibshowssensitizingeffectstoPRRTinachickenegg-basedin vivomodel
ToadapttoamoreclinicallyrelevantpancreaticNENmodelbyexploiting PRRT-induced signaling changes as therapeutic targets, we established a short-term preclinical in ovo modelby implanting a pancreatic, SSTR2-positive cell line-derived 3D spheroid, sensitive to Lu-177-DOTATOC PRRT,ontothechorioallantoicmembrane(CAM)offertilizedchickeneggs (Figure3A).SincenativeBONcellslackfunctionalSSTR2expression,we usedahumanSSTR2-transfectedBONcellline[32].ThoseBON-SSTR2+ were positive for anti-SSTR2 immunofluorescent staining and showed a ∼60-fold higherLu-177-DOTATOCbindingcompared tonativeBONs (Figure3B,C).
To visualize SSTR2 expression in an in vivo setting we inoculated the BON- SSTR2+ 3D spheroids onto the CAM of 7 d old fertilized chicken eggs. Following an incubation time of 3 to 5 d, during which BON-SSTR2+ spheroids connected tothe embryonic vasculature while the matrigel was degraded, we verified functional SSTR2 expression in thosevascularizedmicrotumorsbySPECT/CTimaging.Infact,wecould
demonstrateanincreaseduptake(0.8%–2.2%ID)ofLu-177-DOTATOC 3hpostinoculation (Figure3D).Multiparametric invivoimagingusing PET/MRIallowedformonitoringindividualtumordevelopmentof BON-SSTR2+microtumorsundertreatmentinanoninvasivemanner.Fivedays afterinoculationofthemicrotumors,treatmentstartedwithbortezomib,or Lu-177-DOTATOC,oracombinationofbothorsaline(vehicletreatment). Whilethesizeofvehicle-treatedtumorsincreasedby∼35.2%within7d, confirmingsignificantgrowthof themicrotumors inthe in ovosetup,all treatmentarmsinducedashrinkagebelowtheinitialtumorvolumeatthe beginning ofthe treatment(day 0). Lu-177-DOTATOCandbortezomib monotherapyshowedapproximately13%and28%tumorvolumereduction, respectively.Thestrongesteffectwasseenaftercombinedtreatment,as BON-SSTR2+microtumorsnotonlystoppedtogrow,butdecreasedinvolume by approximately 50% (Figure 3F). Treatment effects were analyzed on molecularlevelaftertumorexcision.
Bortezomib-dependentapoptosisisinducedbyDNAdamagerepair insufficiency
Afterdemonstratingtheinductionofcellcyclearrestandapoptosisupon bortezomib-combinedDNA damagingtherapyof NENcellsinvitro and delayoftumorgrowthinvitroandwhenimplanted“inovo” (CAMmodel) weaimedtoclarifywhatcausedtheinductionofapoptosisinvitroandinthe chickenCAMmodel.Therefore,wereanalyzedtheinvitromRNAprofiling resultsanddeterminedeffectsofbortezomibpluscisplatinonDNAdamage responsegeneexpressionandsubsequentlyanalyzedtheexpressionofmajor candidatetargetproteinsinvitroandinthebortezomibplusPRRTtreated chickenCAMmodel.
Invitro,cisplatinalonecausedtheupregulationofanotablenumberof DNAdamagerepairassociatedgenesinBONcells(Figure4;Supplementary Table 1). In contrast, bortezomib alone triggered the downregulation
Figure2. Geneexpressionafterbortezomib,cisplatinandcombinatorytherapydemonstratedinductionofapoptosis: BONcellsweretreatedwith25nM bortezomib,10μMcisplatinorbothversussalinefor24handanalyzedbynCountergeneexpressionarray.(A)Genesetanalysesofaffectedpathwaysafter cisplatin,bortezomibandcombinationrelatedtocontrol.Thedirectedanalysisheatmapdisplayseachtreatment’sdirectedglobalsignificancescores.Directed globalsignificancestatisticsmeasuretheextenttowhichapathway’sgenesareup-ordownregulatedaftertreatment.Reddenotespathwayswhosegenesexhibit extensiveover-expressionupontreatment,bluedenotespathwayswithextensiveunder-expression.Cisplatinupregulatedgenesassociatedwithpositiveregulation ofDNArepairandcellcycleregulation. Aftercombinedtreatment, genesofallanalyzedpathwayswerefoundpredominantlydownregulated.Thedetailed analysisofapoptosis-relatedgeneexpression,color-codedasupregulatedinthebortezomibarmanddownregulatedinthecombinationarm,revealedthatthe bortezomibtreatmentaffectedthreegenes(ENOG,BCL2,andNFKB1)inanegativewayand5genes(CASP7,BID,IKBK,CAPN2,andMAP3K14)inapositive manner.Thecombinationofbortezomibandcisplatinchangedtheexpressionofelevengenes(ENDOG,PRKAR1B,BCL2,NFKB1,FASLG,PIK3CB,PPP3CB, PIK3R2,IL1RAP,PPP3CAandAKT2)negativelyandupregulatedninegenes(CASP8,CAPN2,CASP7,BID,IKBKG,TNFRSF10A,MAP3K14,TNFRSF10D, undFAS).Althoughtheresultingscorewouldleadtotheassumptionthatbortezomibregulatesapoptosisinamorepositivewaythanthecombinedtreatment, thecombinationaffectedmoreantiapoptoticgenes,includingBCL2,AKT2,subunitsofNF-kappaB,PI3K,PKAandBcl-2(toalowerextent),inanegativeway. Therefore,thesignalingchangesresultinaproapoptoticsignaling.Significantlydifferentiallyexpressedgeneshavebeenfilteredbyminimumfoldchange≥1.5-fold beforedetailedanalysis.GeneexpressionrawdatacanbefoundinSupplementaryTables1-3.(B)Pathwayanalysisofapoptosis-relatedgenesafterbortezomibplus cisplatintreatment.VisualizationbyKEGGontologypathwaysystem.Althoughtheoverallpathwayscoreforapoptosis-relatedgeneswasdeterminednegative, thedifferentialanalysisofthedistincttranscriptsdemonstratedpro-apoptoticmechanisms:Genesofpro-apoptoticdeathreceptor-mediatedpathways,suchasof Fas,TRAIL-R,caspases7and8,BIDandNIKwereupregulated(red),whereaspredominantlycrucialantiapoptoticgenes,includingthoseofBcl-2,NF-kappaB andgenesofthePI3Kpathway,weredownregulated(green).Thisresultedinanoverallnegativepathwayscore.(C)GOenrichmentanalysisusingtheENRICHR dataanalysistoolverifiedastrongmodulationofgeneexpressionaffectingapoptoticprocesses.(D)Forverificationofthegeneexpressionprofiling,BONcells weretreatedwith25nMbortezomib,10μMcisplatin,combinationofbothorcontrolfor24handanalyzedbywesternblot.Bortezomibandcombinedtherapy inducedmassiveDNAdamage(DNAdoublestandbreaksindicatedbyH2AXphosphorylation),butdidnotinduceinternalapoptosisviaBAX.Bothtreatment armspresenteddeathreceptor-dependentextrinsicapoptoticsignalingasdetectedbyFADDupregulationandcleavageofcaspase8,caspase6andfinallyPARP. Therelativeeffectswerestrongerinthecellsthatweretreatedwiththecombinationofboth,cisplatinandbortezomib.Expressionofthep27regulatorandDNA repairassociatedproteinSKP2wasreducedafterbortezomibandcombinedtherapyinassociationwithanupregulationofthep27cellcycleregulator.Independent fromRbproteinabundance(whichisrepresseduponbortezomibandcombinedtreatment),p21isupregulatedfollowingbortezomiborcombinedtreatment. CK2-alphaandauroraAwerealsoupregulateduponbortezomibandthecombination.WesternBlotdatawasdensitometricallyquantifiedbyuseofImageJand statisticallyevaluatedwithPrism8.NormaldistributionofthedatawasassumedafterpassingShapiro-Wilktestandaccordingtotheresults,datawastestedby Kruskal-Wallisorone-wayANOVAanalysis.GraphsshowmeanwithSD.Apop,apoptosis;CC,cellcycle;ChromMod,chromatinmodifications;HH,hedgehog; TXmisReg,transcriptionalmisregulation;Bo,bortezomib;Cis,cisplatin,ctrl,control;NA,notapplicable(duetotoolittlecontrastofthesignalanderror-prone densitometricmeasurements); ∗P<0.05; ∗∗P<0.01; ∗∗∗P<0.001; ∗∗∗∗P<0.0001.(Colorversionoffigureisavailableonline.)
Figure3. Invivoanalysisofproteinexpressionafterbortezomib-combinedpeptidereceptorradionuclidetherapy(PRRT)inthechickenchorioallantoic membrane(CAM)model:(A)Schematicdepictionof theSSTR2+chicken CAMmodel:cellsweremixedwith matrigeltoform atumorplaque and inoculatedontheCAMofafertilized,chickeneggond6ofembryonaldevelopment.(B-D)GenerationofSSTR2-positiveNENcelllineandanalysisfor receptorfunctionalityinvitroandinvivo:(B)BONcellsweretransfectedwithhumansomatostatinreceptor2(SSTR2).ImmunofluorescentstainingofSSTR2 ofnativeandSSTR2+-BONcellsdemonstratedalownativeexpressionofSSTR2innativeBONcellsundapositivemembranestaininginthegenetically modifiedcellline.Green:SSTR2orpan-cytokeratincontrol;blue:DAPI.(C)QuantificationofthefunctionalSSTR2expressionusingligandbindingassay: BON-SSTR2+cells∼60-foldhigherthaninnativeBONcells(P< 0.0001;unpairedttest;n=6).BONandBON-SSTR2+ cellswereincubatedin vitrowith0.0074MBq/mLLu-177-DOTATOC.Afterwashingandcelllysis,theradiopharmaceuticaluptakewasmeasuredusingagammacounter.Graph (barsindicatemeanwithSD)showsrepresentativeresultsof3independentexperiments.(D)SPECT/CTofCAMmodel:tumorplaquewasimaged3h afteri.v.injectionof20MBqLu-177-DOTATOCshowinga1.6%uptakeofthetotalinjecteddose.Representativeimage,leftside:coronalSPECT/CTof Lu-177-DOTATOC,rightside:coronalCT.(E)T2wMRimagingofSSTR2+BONtumors.Tumors(circle)wereeithertreatedoncewithsaline,25nM bortezomib,20MBqLu-177-DOTATOCoracombinationofbortezomibandLu-177-DOTATOConday12andtumorsizemeasuredbefore(d12)and aftertreatment(d18).Representativeimagesofn>3independentexperiments.(F)Statisticalevaluationofthetumorplaquegrowthinthefourtreatment groupsond18.PRRTaloneresultedinanoverall13%relativemassreductionrelatedtobaselinevolume,bortezomibinhibitedtumorgrowthwithaverage 28%massreduction.Thecombinedtreatmentreducedrelativetumormasssignificantlyby50%(P=0.041;ANOVAwithDunnett’stest).pos.ctrl.,positive control;CK,cytokeratin;∗P<0.05;∗∗P<0.01;∗∗∗P<0.001;∗∗∗∗P<0.0001;SSTR2+:SSTR2-transfected,d,day;MRI:magneticresonanceimaging, PET:positronemissiontomography;T2w,T2weighted;Bo,Bortezomib,Lu-177,Lu-177-DOTATOC;Bo+Lu-177,combinationtherapybortezomibplus Lu-177-DOTATOC.(Colorversionoffigureisavailableonline.)
Figure 4. Regulation of DNA damage repair related genes by bortezomib-combinedchemotherapy in vitro :BONcells wereeithertreatedwith25nM bortezomib,10μMcisplatin,combinationofbothorcontrolfor24h in vitro andanalyzedbynCountermRNAexpressionarray.(A)Cisplatininducedthe upregulationofseveralDNArepairgeneswhereas(B)Additionofbortezomib reversed the effect and triggered the downregulation of identical genes and genesassociatedwiththesamerepairpathways,respectively.Arrowsexemplarily mark DDB2, RAD51, FEN1 , and BRCA1 . (C)BON cellswere treatedwith 25nM bortezomib,10μMcisplatin,combinationofbothor controlfor 24 h and analyzed bywestern blot. Crucial DNA damage repair proteins were downregulatedafterbortezomibandcombinedtreatment.Theseincludeproteins of the HR (RPA, SKP2), FA (FANCA), MMR (RPA,RFC), BER (FEN1) and primarily ofthe NER (RPA,DDB2, XPA,XPD, ERCC1) pathwaysof DNA repair. These effects strongly induced DNA damage stress, indicated byhistoneH2A.Xphosphorylation.Representative dataofthreeindependent experiments are shown.Correspondingdataobtained fromKRJ-Iand LCC-18 cells are presented in Supplementary Figure 2. (D) Bortezomibaffected importantmoleculesofthehumanresponsetoDNAdamage:Alkylatingagents, ionizing radiation(IR), ultraviolet (UV) radiation, crosslinking agents(e.g., platin-basedchemotherapeutics),inhibitorsoftopoisomerases(TOPOi),PRRT ornucleosideanalogsinducedifferentDNAlesions(listisnotexhaustive).Major pathways ofDNArepair,whichdepend onthetype ofdamage andthe cell cyclephase,are directrepair(e.g.,bymethyltransferases),baseexcisionrepair (BER),nucleotideexcisionrepair (NER),interstrandcrosslinkrepair(ICLR), nonhomologous end joining (NHEJ), homologous recombination(HR) and mismatchrepair(MMR).Boldedproteinshavebeenshowntobedownregulated bytheproteasomeinhibitorbortezomib.BER,BaseExcisionRepair;FA,Fanconi Anemia; HR, Homologous Recombination; MMR,Mismatch Repair; NER, NucleotideExcisionRepair.
of identical or associated repair genes (Supplementary. Table 2). Under treatmentwithbothsubstances,thecellsretainedthebortezomib-controlled pattern,resultinginamassivedownregulationofDNAdamagerepairgenes (including a large number of those upregulated by cisplatin) and a vast accumulationofDNAdamage(Figure4;SupplementaryTable3).Underthe combinedbortezomib/cisplatintreatment,allcommonDNAdamagerepair pathwayswerenegativelyaffectedinBONcells(Figure4;relateddatafrom othercelllinesrefertoSupplementaryFigure2).Thisindicatesa versatile DNArepairinhibitoryeffectofbortezomibtreatmentlettingustoassume thatthecombinationofPRRTwithDNAdamagingtreatmentcouldhave syntheticlethaleffectsirrespectivefromthesourceoftheDNAdamage.To provethistheory,weanalyzedtheeffectofcombinationwithPRRTinstead ofcisplatinontheDNArepairproteinexpressioninthetissuesobtainedfrom thechickenCAMmodel.
Quantitative analysis of the respective protein expression upon bortezomib-combined PRRT in the tissues derived from the chicken CAM in ovo model showedsignificant protein expression alterationsfor severalDNA damageand apoptosis-relatedproteins according towestern blotdensitometry data(Figure5and rawdataSupplementaryFigure 4). Despite relatively high variances within the biological replicates of one treatment arm––which reduced the statistical power on a singleprotein level––theoverallproteinexpressionrevealeda generaldownregulationof DNArepairrelatedgenesinthecombinedtreatmentgroup.Here,especially thecandidateeffectorproteinsoftheBERpathwaywereaffected.
While bortezomibpredominantlyrepressed the expression of proteins acting in late stages of DNA repair, leaving the sensing and initiating proteinsunaffectedorupregulated,PRRTreducedtheexpressionofsome early recognition proteinssuchas ATM.Thiseffectmight interfere with apoptosis induction since DNA lesion-sensing proteins are involved in apoptosisregulation.However,incontrasttotheinvitrodata,wecouldnot clearlydistinguishwhetheranextrinsicorintrinsicinductionof apoptosis occurredafterthecombinedtreatment,sinceproteinsofbothpathwayswere affected.Thisvariabilityismostlikelycausedbythefactthatbortezomibwas droppedontotheengraftedtumorplaquesinovoresultinginaheterogeneous distributionofthepharmaceutical.
TheroleofFOXM1
TheforkheadboxtranscriptionfactorFOXM1 isa crucialregulator of G2/Mtransitionandmitosisandcanbesuppressedbyproteasomeinhibition
[58,59].Inneuroendocrinetumors,werecentlyclarifiedtheroleofFOXM1 asfrequentlyexpressed proliferationmarker [21,31]. FOXM1hasfurther beenattributedtothecontrolofDNAdamageresponseandresistanceto chemotherapy [60–63].Takentogether,weassumedthatbortezomibmight exertitscellcycleregulatoryandDNAdamagepreservingeffectbyinhibition ofFOXM1.
WethereforetreatedGEP-NENcelllineswithbortezomibandanalyzed the FOXM1 expression by western blot and real time qPCR. All cell lines responded with a downregulation of FOXM1 mRNA and protein (SupplementaryFigure5). Toprovethatbortezomibexerts itsanticancer effect by inhibition of FOXM1,we silenced the FOXM1 translation by RNAinterference inBON cellsand comparedthe differential expression upon bortezomib(25 nM for 24 h) with the expression upon FOXM1 knockdown (Supplementary table 4) by Nanostring nCounter pathway gene expression array. Knockdown efficacy was determined by qPCR (SupplementaryFigure5E).Wefound39correspondinggenes.Interestingly, those genes werepredominantly associatedwith cell cycle and apoptosis, accordingtothePANTHERpathwayoverrepresentationanalysis[49].There wasnooverrepresentationofDNAdamagerepair-relatedgeneswithinthe 39correspondinggenesindicatingthatFOXM1isinvolvedinthecellcycle regulationuponbortezomibtreatmentratherthaninDNAdamageresponse.
Figure5. RegulationofDNAdamagerepairrelatedproteinsafterbortezomib-combinedPRRTinvivo:ChickenCAMxenograftedtumorsweretreatedi.v. withasingledoseof∼20MBq177-Lu-DOTATOCwithandwithout25nMbortezomibversussalinecontrolandexplanted6daftertreatment.Protein abundanceofproteinsinvolvedinthesensingofDNAdamage,initiation,signaltransductionandexecutionofDNArepairandapoptosiswasanalyzedby westernblotanddensitometrywasperformedusingImageJv1.53.Datawasnormalizedtodensitometricdataoftotalproteinstain(ponceauS).(A,B)We identifiedsignificantproteinexpressionalterationsinthecombinedtreatmentgroupsforDDB2,ERCC1,cleavedBIDandp38,p41/p43cleavedcaspase 8(Fisher’sLeastSignificantDifferenceTest)andanoveralldownregulationofDNArepair-relatedgenes,especiallyoftheBERpathway(RNApolymerase II,DDB2,XPD,XPA,RPA,ERCC1,FEN1).(B)Heatmapshowsmeanof≥3independentexperimentspertreatmentgroupnormalizedtototalprotein stainandrelativetosalinecontrol.RawdataisshowninSupplementaryFigure4.BER,BaseExcisionRepair,;Bo,bortezomib;Lu-177,Lu-177-DOTATOC; Rpb1,RNApolymeraseIIsubunitB1c-terminaldomain,∗P<0.05;∗∗P<0.01.
Additionofbortezomibtoconventionallu-177-PPRTdoesnotaffect overalltumorvolumeorviabletumorcellmassaccordingtoinvivo
FDG-PET/MRIanalysesinanorthotopicmousemodel
To confirm the preliminary results obtainedfrom the chicken CAM model, we established an orthotopic xenograft mouse model bearing intrapancreaticBON-SSTR2+tumors[39].Whenthetumorsreached100 mm3,treatmentwasstartedaccordingtotheschemepresentedin Figure6
including2therapycycles of3weeks.Aspreliminary invivostudieshad
shownthatasingleinjection(i.e.,onecycle)ofLu-177-DOTATOCinduced agrowthplateauofabout2weeks,wedesignedthestudyinsuchawaythat thesecondcyclestartedoncethetumorintheLu-177DOTATOCtreated animalscommencedtoprogressagain.Asseenin Figure6,BON-SSTR2+ tumors responded to 177-Lu-DOTATOC monotherapy by showing a growthdelayofabout6.5weekscomparedtosaline-treatedBON-SSTR2+ xenograft tumors,while bortezomibmonotherapydelayed the growthby only3weeks.Interestingly,the combinedtreatment(bortezomib4hours beforePRRT)delayedtumorgrowthby4weeks.Hence,theKaplan-Maier
survivalanalysisofcumulativetimetoendpoint(definedbyatumorvolume of1400mm3)indicatesthatthecombinationofbortezomiband
Lu-177-DOTATOCdoesnotresultinalongeroveralltimetoendpointcompared toPRRTmonotherapy(SupplementaryFigure6).
Notably, 5 to6 weeks after the second cycle,the treatment-induced, reducedtumorgrowthterminatedandametabolicallyactivetumorfraction startedtodevelopinalltreatmentgroupsascomparedtothesalinegroup. This unexpected tumor outgrowth may be an indication that therapy-resistanttumorclonesareabletodevelopthroughoutthetreatment.Inorder tocheckwhethertheoutgrowingtumorcellsareactuallyvital,weperformed combinedFDG-PET/MRIscansoncetumorsreachedatumorvolumeof approximately1000mm3.TheT2-weightedMRIsequenceusedherewas
establishedinsuchawaythatviabletumortissueresultsinhyperintenseand necrotic/degradingtissueinhypointensesignalintensity(SI).Thisallowed toperform a3D-tumor textureanalysis forthisparticulartimepoint, in
whichFDG-PETidentified viable tissuewhereasT2w imaging identified truenecroticregionsbyanalyzingSIineachvoxel(Figure6).Theresulting thirdtumorsectioncanbedescribedasamixtureofdifferentcellularstates includingcellsintransitionfromviabletoapoptoticand/ornecrotictissue underhypoxiaornormoxia.
When PBS and Lu-177-DOTATOCtreated animals werecompared, PBS-treatedanimalsshowedfarlesstissuedamagethanLu-177-DOTATOC treatedtumorsorthosetreatedwithbortezomibaloneorincombinationwith Lu-177-DOTATOC.Thisconfirmedinductionofnecrosisasasecondary treatmenteffect.Thereducednutrientperfusion,whichtypicallyoccursin thetumorcenterandtherebyaffectsbothtreatedandcontrolgroups,triggers increasednecrosisinthe DNA-damagedtumorcore.Visual inspectionof theT2wimagesclearlyindicatedthatPRRT-treatedtumorsaswellastumor treatedwithbortezomibinmono-orcombinedtherapyshowedlargeareas oftissuedamage.Thistissuemostlikelyconsistedofnecrosis,celldebris,and
damagedofvariousintermediatestates,varyingfrom∼21%to∼33%ofthe tumormassinthethreetreatmentarmscomparedto11%inthecontrolarm (Figure6).
As expected, the saline treated tumors exhibited a high percentage of viable tumor tissue and only small parts of necrotic tissue (mean of 45.6%and11.4%respectively).TheLu-177-DOTATOCtreatmentgroup showed in 26.8% viable and 32.5% necrotic tumor tissue, respectively. Thebortezomibandbortezomib/Lu-177-DOTATOCgroupshowedalmost identicaldistribution of the individualtumortissuestates(approximately 28%–31%livingtumortissue versusapproximately21%necrosis).Here, bortezomib monotherapy resulted in a slightly higher portion of viable tumortissue.Thecombinationtreatmentresultedinlessvitaltumortissue, butahigher percentageof intermediatestatetumortissue atthetimeof investigation.
Immunohistochemicalanalysisofnecrosisinduction,proliferation,and apoptosismarkerssuchasnuclearassemblyof53BP1andPARP cleavage revealednostatisticallysignificantdifferencesbetweenthetreatmentarms. However,thehistologyresultsarestronglydeterminedbytheanalyzedtumor areaand onlyreflect a snapshot.These variancesindicate thatbiological heterogeneityresultingfromresponsiveand non-responsivetumorareasis veryhighand thatclonalselectionafter PRRTmay causedisease relapse (Figure6andSupplementaryFigure7).
Discussion
Targeting DNA-damage repair is an emerging therapy approach in cancers that are characterized by a high mutational burden, impairing genes responsible for DNA repair. Whereas PARP inhibitors induce syntheticlethalityincancers withdeleteriousgermlineorsomaticgenetic defects in homologous recombination (HR) repair, clinical development of combination with DNA-damaging chemotherapy was limited by the more-than-additive toxicity,in particular dose-limitingmyelosuppression. Multiple other DNA damage repair inhibitorshave been developed and enteredpreclinicalandclinicaltestinginthelasttwodecades[64].However, profoundmolecularknowledgeofagent-ormutation-specificvulnerabilities inDDRiscrucialtostratifypatientsforDDRinhibitoryregimens.
Because of the specific molecular features of neuroendocrine tumors, whichisdominatedbysomatostatin(SST)signalingandoveractivationof mitogenicsignaltransduction(reviewedin [3]),DDRinhibition hasnot beenafocusofbasicresearch inNENssofar.Furthermore,therelatively lowmutationalburdendecreasesthelikelihoodofpresenceofsyntheticlethal mutations.
However,DDRplaysaclinicallyrelevantroleinneuroendocrinetumors. A study published in 2017 found DDR defect-related gene signatures (BRCA,MUTYH,andAPOBEC)aswellasbiallelicmutationsintheDDR genesMUTYH,CHEK2orBRCA2in8%ofpancreaticNETs [65].Other studiesdemonstratedaconsiderablyhighprevalenceofaberrantregulation ofMMRgenesbymicrosatelliteinstability,promoterhypermethylationand lossofheterozygosityininsulinomasandotherpancreaticNENs [66,67], although contradictory studies exist [68]. Furthermore, PTEN, which is frequentlylostinNENsduetoepigeneticalterationsorcopyloss,isinvolved inDDR(reviewedin [69]).Finally,theexpressionofMGMTispredictive oftheresponsetotemozolomideinpancreaticneuroendocrinetumors [70]. Incontrast,predictivemarkersorbroadmechanisticknowledgeofDDR uponPRRTisrare(reviewedin [71]).PRRTusingLu-177asbetaemitter has been shown to induce mainly SSBs and, to a lower extent, DSBs
[11,72,73]. SSBs are primarily repaired by components of base excision repair(BER),nucleotideexcisionrepair(NER)andmismatchrepair(MMR) whereasDSB are clearedpredominantlybynon-homologous endjoining (NHEJ)orhomologousrecombination(HR),dependingontheavailablilty ofhomologousDNAmaterialandgeneticbackground.Althoughbothkinds of lesions induce different ways of damage repair, DSB repair generates intermediates,suchassingle-strandedDNA(ssDNA),whicharepresentin MMR,BER,andNERpathwaysofDNArepair(butnotindirectrepair). Accordingly,commoninteractingproteinsarerecruitedinspecificstepsof therepairpathwaystoprotectssDNA,polymerizesecondstrandsandcontrol ligation.Thoseversatileproteins,includingPARP1,RPA,FEN1arecrucial tomorethanonepathwayofDNArepair.
Since therewasnoclear evidence,whichof those componentsmight berelevanttowhichextentinPRRT(andbeyondthatalsotargetable),we decidedtousecisplatinasamodelforourbasicinvitroexperiments.This decisionwasmadeonthefactthatcrosslinkrepairintegratesalargenumber of components of DSB repair and thus,in contrast to alkylatingagents
Figure6. Analysis of treatment efficacy of Lu-177-DOTATOC plus bortezomib therapyin orthotopic pancreatic micexenograft tumors: Mice were xenotransplantedwithBON-SSTR2+cellsandtreatedwitheithersaline,Lu-177-DOTAOC,bortezomiborLu-177-DOTATOC/bortezomibwhentumor volumereached≥100mm3.(A)Thetherapeuticschemewasbasedonthehumanbortezomibtreatmentschedule,including2cyclesof3weekseach(2weeks
treatmentandoneweekpause).Micereceivedeitherasingledoseof∼30MBqLu-177-DOTATOC(i.v.),PBS(i.p.),i.p.injectionof0.5mg/kgbortezomibor acombinationofLu-177-DTATOCandbortezomibonday0andd21.Forcombinedtreatment,bortezomibwasinjected4hpriortoLu-177-DOTATOC toensurefullabsorption.Duetoitsrapidclearance,bortezomibtreatmentwasrepeatedondays3,7,and10inthefirstcycleandday24,28,and31during thesecondcycle.Duringweek3ofthefirstcycle,thetreatmentwaspaused,beforethesecondanalogouscyclestartedinweek4withareduceddosageof bortezomib.(B)Representativehigh-resolutionT2-weightedimagesofalltreatmentgroups(tumorcenter)2moafterthesecondcyclewasfinished,revealed morphologicheterogeneityduetotreatmenteffectsinthetumortissue.Viabletumortissueisrepresentedashyperintensesignalintensity(SI)withinthetumor, whiledegraded,necrotictumortissueisrepresentedbyhypointenseSI(whitearrows).FusedFDG-PET/T2wtumorimagesclearlyshowreducedFDG-avid fractionsofarepresentativetumorpertreatmentgroupconfirmingthetumortissueheterogeneityseenincorrespondingT2wimages.(C)MRI-andPET-based tumortextureanalysis,representativeimagesoftumorcenter(PBSgroup):BasedonT2wimagesawholetumorROIwasdefined.T2wMRI-analysisofSI basedonthresholdsettings:redareasrepresenttruenecrosis,lightblueareaslivingandintermediatetumortissue.FusedMRI/PET:FDG-PETanalysisbased onthresholdsettings:redROIrepresentsFDG-avidtumortissue.(D)ThetumorgrowthwasmonitoredbyMRIonceaweekuntilthetumorsreachedthe endpointof1400mm3intumorsize.TheLu-177-DOTATOCmonotherapyarmshowedthestrongestgrowthdelay.(E)FDG-PET/T2wdemonstrateda
lowerlivingtumortissuecontentinthetreatmentgroupsrelatedtothecontrolgroup,withoutreachingsignificanceineithertheviabletumororthenecrotic tumorcontentcomparison.One-WayANOVAandDunnett’smultiplecomparisonstest(afterassumingnormalitybyShapiro-Wilktest)demonstrateda tendencytodifferencesinvitaltumorvolumeintheLu-177-DOTATOCversusctrl.Group(p=0.1042).(F)ImmunohistochemicalanalysesofHIF1-alpha, Ki-67,53BP1,andcleavedPARPinformalin-fixedmicetumorrimtissueresultedinnosignificantdifferencesbetweenthefourtreatmentarmsduetohigh intratumoralheterogeneity.Kaplan-MeierplotandcorrespondingrawdataareshowninSupplementaryFigs.6and7.d,day;FDG,18F-fluorodeoxyglucose; MRI,magneticresonanceimaging;PET,positronemissiontomography;ROI,RegionofInterest;T2w,T2weighted;Captions:arrows:red=tumortissue; white=necrotic/damagedtumortissue;blue=kidneys/intestine;treatmentarms:B,bortezomibmonotherapy;B+L,combinedtreatment;ctrl.,control;L, Lu-177-DOTATOC.(Colorversionoffigureisavailableonline.)
and antimetabolites, platin-based chemotherapy might have considerable functionaloverlapswithDNAdamageresponsetoPRRT.Platinum-induced DNA crosslinks are repaired by NER or by interstrandcrosslink repair, which makes use of components of NER, DSB repair pathways such as homologous recombination (HR), and translesion synthesis [74]. In contrast, TEM- and STZ-induced DNA adducts are primarily repaired throughdirectrepairbymethyltransferasessuchasO6-methylguanineDNA methyltransferase(MGMT),ALKBfamilydemethylasesandBER[75].Since wewantedtofocusspecificallyonPRRT-inducedDNArepairmechanisms, we excluded the more clinically relevant alkylating agents (TEM, STZ) andantimetabolites(5-FU,CAP)fromthestudy,despitethefactthatour preliminaryexperimentshavedemonstratedenhancementeffectswhenwe combinedthemwithbortezomibaswell(unpublished).
Not surprisingly, transcripts of BAX, CDKN1A, DDB2, FDXR,
GADD45A, MDM2, STAT5B and XPC showed time-dependent
upregulation in peripheral blood cells upon radionuclide exposure in a recent study [76]. This indicates induction of apoptosis and cell cycle arrest alongwith NERinduction. In ourstudy, weverified an activation ofNERcomponentsuponPRRT.Precisely,wedemonstratedasignificant upregulation of the endonuclease ERCC1 upon Lu-177-DOTATOC treatment inthe chicken CAM model,along with postexcision features, including upregulation of the Flap Endonuclease 1 (FEN1), which is involved in the final ligation step of NER and BER, and of replication protein A (RPA),whichrecruits proteincomplexes to ssDNAinvarious processeswheressDNAisanintermediatestructure [77].
Inthelastyears,PARPinhibitorshavebeendiscussedasPPRTsensitizers. PARPfamilyproteinsaremultifacedrepairproteins,involvedinBER,NER, NHEJandHRpathwaysandthusplayaroleinboth,SSBandDSBrepair
[78]. Accordingly, preclinical studies, combining PARP1 inhibition with PRRT,have shownpromisingresults [72,79,80]. Concomitantinhibition ofPARP1, leadstoaPARP1trapping,stallingofthereplicationforkand inductionof DSBs [81].However, inthe contextof HRdeficiency(e.g., byBRCA1or2mutations),whichispresent inasmallfractionofNENs
[65],thisapproachwouldinducesyntheticletality [81].InthecaseofHR proficiency, the cellscould bypass fork stalling by inductionof HRand thereforesurvive [81–83].Inourstudywepresentfurtherevidencethatthe DDRuponPRRTisverycomplexandinvolvesaspectsofseveralpathways. Hence, a versatileinhibiton of DDR might be necessary to reduce the inductionofbypassingprocesses.
Therefore, we hypnotized thata major effect, which orchestrates the complex interplayof repairproteins, suchas posttranslationalregulation, couldperturbDDRmoreefficientlythantargetingspecificcomponents.So far,multiplestudieshavedemonstratedthattheubiquitin-proteasome-system is involved inregulation of DNA-damage proteins [19,20]. Accordingly, proteasome inhibition has been showntosensitize tumorcellsto DNA-damagingtherapyinvitro [22,84–87].Althoughbortezomibmonotherapy has not been effective in phase II studies of metastatic neuroendocrine tumorsofdifferentoffspring [23]andforthetreatmentofneuroendocrine SCLC [88], the additionof a strong proapoptotictrigger, suchas DNA damage,presentedencouragingresultsconcerningresponseandtolerability ofbortezomib/chemotherapycombinationsinanumberofclinicalstudies
[89–98]. Recently, we have demonstrated a significant benefit from a bortezomib/cisplatinregimeninneuroendocrinelungcancertherapyinvivo models [21]. In the study presented herein, were able to to show that bortezomibaffectsgeneexpressionofallcommonDNArepairmechanisms, includingradiation/cisplatin-inducedNERandSSB-repairingBERinvitro and in vivo. Both cellular stress and apoptosis are induced in reponse to unfolded proteinand/or DNA damage accumulation.However, since the molecular mechanisms induced by bortezomib are not completely understoodandpresumablydependentoncellularcontext,weexpectedto detectevidenceofalteredproteindecayduetoproteasomeinhibitionrather than effecton genetranscription. Nevertheless, recent datasuggests that
bortezomibtargetsSpfamilytranscriptionfactorsincancercells[99],which inturn regulatethe expression of DDRgenes followinggenotoxic stress (reviewedin[100]).
Wefurtherhypothesizedthatbortezomibcouldbebeneficialtoovercome resistancetoDNAdamagingtherapyinTP53mutatedandwildtypetumors, sincebortezomibexertsitspro-apoptoticeffectmainlybydeath receptor-dependentpathwaysandthus,isfairlyindependentfromTP53mutations
[101–103]. Therefore it might have applicability in both, rarely TP53 -mutated PRRT-treatedG1/G2 NENsand frequently mutated G3NECs whichqualifyforplatinum-basedchemotherapy[65,104–106].Accordingly, BONcellswhichareTP53mutated[16,107]showedarelativelowinduction ofprimarilyp53-driven intrinsic(mitochondrial)apoptosisandthe major treatmenteffectofbortezomibintheanalyzedNENcelllineswasinduced byextrinsicapoptoticpathways.
CurrentdatasuggestthatPRRTmightbeeffectiveintheapproximately 20%to30%SSTR2-positiveG3NETs[108],whichiscurrentlydiscussedas therapeuticoptionforSSTRimaging-positivetumors [109,110].However, thereisstillagreatneedfortreatmentoptionsforG3NETs(whichproliferate fastand showbetter-differentiatedmorphology). As demonstratedinour preclinical study, bortezomib might be a novel approach to sensitize to cisplatin(whichexploitshighproliferationratesbutwithlimitedsuccessin G3NETs).ItmightalsoenhancetheeffectofPRRT(whichtargetsSSTR expression,butrequiresmoreresearchinG3NETs).Inthispreclinicalstudy wegivedetailedinsightintocommonresistance-associatedeffectsandprovide groundworkforsubsequenttranslationalconsiderations.
SeveralgroupshavedescribedFOXM1asacrucialregulatoroftheDNA damagereponse(reviewedin [60,111,112]).Thissuggeststhatinhibition of FOXM1 impairsDNAdamage repairand thatrepressionof FOXM1 uponbortezomibtreatmentmediatestheDNAdamagerepairinhibition.For instance,FOXM1wasdiscussedasregulatorofthe HRproteinsBRCA2, DNA polymerase delta, BRIP1 or Rad51 [60]. Surprisingly, we could demonstratethatFOXM1knockdownwasassociatedwithapoptosisandcell cycle arrest,butwe werenotabletodetectconsiderable effectsonDNA repair-relatedgeneexpression.Aswecould confirmsufficientknockdown by verification of FOXM1 decrease by western blot and real time PCR of FOXM1, weassumethattheseDNArepairgenesarenotregulatedby FOXM1 directly,butfollow the downregulationof interposedregulatory events.EvenFOXM1knockdowninpresenceofcisplatindidnotalterthe expressionofDNAdamage-relatedgenescomparedtocisplatinplusmock siRNA,suggestingthattheinhibitionofFOXM1aloneisnotsufficientto interfere with DNAdamageresponse (datanot shown).Thus,weassume thatFOXM1isnotinvoledintheDNAdamagecontroluponproteasome inhibition,butmediatessomeofit’seffectsviaapoptosisinductionandcell cyclearrest.
Thestrengthofourstudyisthebroad,multimodel-basedassessmentof detailedmechanisticdataabouttherapiesthathavealreadyenteredclincial useinNENs,butarestillnotfullycharaterizedfromamolecularpointof view.Weprovidefurtherinformationaboutweaknessesincellularresponse toDNAdamagingtherapythatmightbeexploitedfortherapeuticpurposes inthefuture.
However, our study also has some shortcomings. Most notably, the availability of proper preclinical models is limited in NEN research. In ourstudy weused twodifferent pancreaticcellline models:native BON andQGP-1 cellsweregeneratedfromwell-differentiatedpancreaticNET, however, their mutational pattern, lack in SSTR2 and Chromogranin A expression, their reduced SSTR5 expression and their relatively high proliferationratesrepresentfeatures,thatarenotcompletelyrepresentative of well-differentiated NETs [113], butmight be exploitedfor preclinical evaluation of therapy approaches thatare attributedtoless differentiated tumors.However,afterre-expressionofSSTR2inBONcells,themodelwas alsoapplicableforPRRT,whichrequiresspecificepitopessuchasSSTR2and haslimitedapplicabilityinpoorlydifferentiatedtumors [14].
