1

1 Exosome-dependentimmunesurveillanceatthemetastaticnicherequiresBAG6andCBP/p300-2

dependentacetylationofp533 4 MaximilianeSchuldner1,2,BastianDörsam1,OlgaShatnyeva2,KatrinS.Reiners2,3,AndriyKubarenko3,Hinrich5 P.Hansen2,FlorianFinkernagel4,KatrinRoth5,SebastianTheurich6,AndreaNist7,ThorstenStiewe7,Annette6 Paschen8, Gero Knittel9a-d, Hans C. Reinhardt9a-d, Rolf Müller4, Michael Hallek9a-d, Elke Pogge von7 Strandmann*1,28 9 1ExperimentalTumorResearch,CenterforTumorBiologyandImmunology,ClinicforHematology,Oncology10 andImmunology,PhilippsUniversity,Marburg,Germany11 2Innate Immunity Group, Department I of Internal Medicine, University Hospital of Cologne, Cologne,12 Germany13 3InstituteofClinicalChemistryandClinicalPharmacology,UniversityHospitalBonn,Bonn,Germany14 4InstituteofMolecularBiologyandTumorResearch(IMT),CenterforTumorBiologyandImmunology(ZTI),15 PhilippsUniversity,Marburg,Germany16 5CenterforTumorBiologyandImmunology(ZTI),CoreFacilityCellularImaging,PhilippsUniversity,Marburg,17 Germany18 6CancerandImmunometabolismResearchGroup,DepartmentIofInternalMedicine,UniversityHospitalof19 Cologne,Germany(currentaddress:DepartmentofMedicineIII,UniversityHospitalMunich,LMU,Germany)20 7Institute ofMolecular Oncology and Genomics Core Facility, Center for Tumor Biology and Immunology21 (ZTI),PhilippsUniversityMarburg,Germany22 8Department of Dermatology, University Hospital Essen, Essen and German Cancer Consortium (DKTK),23 PartnersiteUniversityHospitalEssen,Essen,Germany24 9aClinicIofInternalMedicine,UniversityHospitalofCologne,Cologne,Germany.25 9bCologneExcellenceClusteronCellularStressResponseinAging-AssociatedDiseases(CECAD),Universityof26 Cologne,Cologne,Germany.27

9cCenterofIntegratedOncology(CIO),UniversityHospitalofCologne,Cologne,Germany.28 9dCenterforMolecularMedicineCologne(CMMC),UniversityofCologne,Cologne,Germany29 30 Financialsupport:31 ThisstudywasfundedbygrantsoftheDeutscheForschungsgemeinschaft(KFO325,1408/14-1;1408/13-1)32 andtheWilhelmSanderStiftung(2015.145.1)toEPvS.33 34 Leadcontact/Correspondingauthor:ElkePoggevonStrandmann35 email:poggevon@staff.uni-marburg.de36 Tel.: 06421/28-2164037 Fax: 06421/28-6892338 PhilippsUniversityMarburg,Hans-Meerwein-Straße3,39 35043Marburg,Germany.40 41 42 43 44 45 46 47

2

ABSTRACT48 49 Extracellular vesicles released by tumor cells contribute to the reprogramming of the tumor50 microenvironment and interfere with hallmarks of cancer including metastasis. Notably, melanoma cell-51 derived EVs are able to establish a pre-metastatic niche in distant organs, or on the contrary, exert anti-52 tumoractivity.However,molecular insights intohowvesiclesareselectivelypackagedwithcargodefining53 theirspecificfunctionsremainelusive.54 Methods: Here,we investigated the role of the chaperone Bcl2-associated anthogene 6 (BAG6, synonym55 Bat3) for the formation of pro- and anti-tumor EVs. EVs collected fromwildtype cells andBAG6-deficient56 cellswerecharacterizedbymassspectrometryandRNAseq.Theirtumorigenicpotentialwasanalyzedusing57 theB-16Vtransplantationmousemelanomamodel.58 Results:WedemonstratethatEVsfromB-16Vcellsinhibitlungmetastasisassociatedwiththemobilization59 ofLy6Clowpatrollingmonocytes.Theformationoftheseanti-tumor-EVswasdependentonacetylationofp5360 by the BAG6/CBP/p300-acetylase complex, followed by recruitment of components of the endosomal61 sortingcomplexesrequiredfortransport (ESCRT)viaaP(S/T)APdoublemotifofBAG6.Geneticablationof62 BAG6anddisruptionof thispathway led to the releaseof adistinct EV subtype,which failed to suppress63 metastasisbutrecruitedtumor-promotingneutrophilstothepre-metastaticniche.64 Conclusion:WeconcludethattheBAG6/CBP/p300-p53axisisakeypathwaydirectingEVcargoloadingand65 thusapotentialnovelmicroenvironmentaltherapeutictarget.66 67 Keywords:extracellularvesicles,exosomes,BAG6/CBP/p300-p53,metastasis,melanoma68 69

70 71 BAG6/CBP/p300-mediatedacetylationofp53isacrucialstepintheformatonofanti-metastaticEVs.Lossof72 BAG6resultsinthereleaseofEVswithpro-tumorigeniccargo.73 74 75 76 77 78 79 80 81

3

INTRODUCTION82 83 Extracellular vesicles (EVs) are known to stimulate tumor progression and to interfere with hallmarks of84 cancerincludingimmuneevasionandmetastasis[1].Importantly,therecruitmentofinnateimmunecellsto85 distantorgansemergedasakeyprocessintheEV-dependentpromotionofmetastasis[2].Acriticalstepin86 EV-mediatedlivermetastasisofpancreaticductaladenocarcinomasistheMIF-dependentattractionofbone87 marrow-derivedmacrophages [3].Anotherexample is theTLR3-mediatedactivationof lungepithelialcells88 throughexosomalRNAscausingchemokinereleaseandneutrophilrecruitment[4].Onthecontrary,several89 studiesproposeananti-tumoractivityofEVsmainlyviaengagementofNKcells[5-8].Giventheir immune90 activiating potential, extracellular vesicles and exosomes are promising tools for potential clinical91 interventions [9-11]. Themolecular basis for different and opposite activities of EVs either counteracting92 tumorprogressionorsupportingtumorinitiation,developmentandmetastasis[12,13]islargelyunknown.93 Here,weinvestigatedtheroleofthechaperoneBcl2-associatedanthogene6(BAG6,synonymBat3)forthe94 formationofpro-andanti-tumorEVs.BAG6was identifiedasaregulatorofTcell [14]andNKcellactivity95 [15] via interaction with the TIM-3 and the NKp30 receptor, respectively. Regulation of NK cell activity96 dependson thepresentationofBAG6on the surfaceofEVs released in response to cellular stress,which97 triggersNKcell-mediatedanti-tumoractivity invitroand invivo [5-7].However,NKp30 isapseudogene in98 themurinesystemandnotexpressedasafunctionalreceptor[16],suggestingthattheanti-tumoractivityof99 BAG6-presentingEVsinmousemodelsdependsonNKp30-independentacitivities.100 Given the functionof intracellularBAG6 inmanycrucialbiologicalprocesses suchasproteintargetingand101 degradation[17],wehypothesizethatitsanti-tumoractivityreliesonBAG6-dependentrecruitmentofother102 effectormoleculestoEVs.103 NotmuchisknownaboutthemechanismsthatregulatetheformationofEVs,butESCRT(endosomalsorting104 complex required for transport)-dependentandESCRT-independentpathwaysaredescribed.Thecytosolic105 ESCRTproteincomplexesenableendosomalmembraneremodelingthatresults inmembranebuddingand106 vesicleformation[18].107 UsinganexperimentalmetastasismelanomamodelweshowthatBAG6isindispensablefortheformationof108 anti-metastaticEVs,whicharecharacterizedby thepresenceofTIMP3,an inhibitorofmetalloproteinases109 involved inthere-modellingoftheextracellularmatrix.TheseEVsfromBAG6-proficientcellswereableto110 inhibit lungmetastasisassociatedwiththeaccumulationofpatrollingmonocytes,whichcontrolmetastasis111 [19]. The release of these anti-tumor EVs required the BAG6/CBP/p300-dependent acetylation of p53112 followedbytherecruitmentoftheESCRTmachinery.LossofBAG6triggeredaswitchinthereleaseofEVs113 from anti- to pro-tumorigenic EVs, which contained among others the mRNA encoding the oncogene α-114 catulin.Ofnote,theseEVsgeneratedintheabsenceofcellularBAGinducedaneutrophilgenesignaturein115 the lungs of mice, which is indicative for the establishment of a pre-metastatic niche [20]. These data116 provideapreviouslyunidentifiedpathway,whichregulatestheformationofEVs.117 118 119 120 121 122 123 124 125 126

4

RESULTS127 CellularstressinducesthereleaseofBAG6-positiveEVs.128 The enhanced release of EVs was previously shown to be associated with cellular stress, such as DNA129 damage [21] or with the inhibition of cellular de-acetylases [6]. To analyze the underlying molecular130 pathways,themodelcelllinesHEK293andB-16Vweretreatedwithnon-lethaldosesofthecytostaticdrug131 doxorubicin (doxo) and with the histone de-acetylase inhibitor (HDACi) panabinostat (LBH) prior to132 purification of EVs from the cell supernatant (methods and further characterization of purified samples133 Figure S1). The treatment with these drugs resulted in a higher protein yield of EV preparations, which134 correlatedwithan increaseofparticlenumber inbothcell lines (Figure1A).Previousworksuggestedthat135 theEV-associatedligandBAG6isenrichedonEVsreleasedbycellsuponcellularstress[5-7].Inline,EVsfrom136 doxo- or LBH-treated cellswere characterized by an increased amount of BAG6, irrespective of the basal137 expressionlevel(Figure1B).TheinducibilityofBAG6onEVsinresponsetohypoxiaasamelanoma-relevant138 stress factor has not been analyzed before. Culturing B-16V cells under 1% O2 for 48h increased the139 expressionofBAG6onEVs,whiletheoverallsecretionofEVswasnotsignificantlyenhanced(Figure1Cand140 S2C).141 Thesedataindicatethatcellularstressinducedbydoxo,theHDACiLBHandhypoxiatriggeredthereleaseof142 BAG6-enrichedEVs.TheseEVsarereferredtoasstress-induced,BAG6-presentingEVs.143 mRNAandproteincargoofBAG6-proficientand-deficientEVsdefinedifferentEV-subtypes.144 ToinvestigatethefunctionalroleofBAG6forEVformation,EVswerepurifiedfromwild-type(WT-EVs)and145 BAG6-deficient (BAG6KO-EVs) B-16V cells, which were cultured under hypoxia (1% O2 for 48h)146 (characterizationofcelllinesseeFigureS2).147 Interestingly,aslightbutsignificantreductionofmeanparticlesizewasobservedforBAG6KO-EVscompared148 toWT-EVs asmonitored by Nanoparticle Tracking analysis (NTA) (Figure 1D), indicating that BAG6might149 impact on the release of EV subpopulations. Western blot analysis of cell and EV lysates confirmed the150 absence of cellular and EV-associated BAG6 in BAG6KO samples, while EV markers including ESCRT151 components TSG101, ALIX, FLOTILLIN-1, and processed ADAM10 were detectable in both preparations152 (Figure1E).153 Next-generation sequencing revealed under-representation of mRNAs and over-representation of 418154 mRNAs inBAG6KO-EVs compared toWT (Figure S2D). Functional geneenrichment analyses revealed that155 genes encoding mRNAs that were significantly decreased, or not detectable in BAG6KO-EVs, clustered156 predominantlyinmitochondria-associatedmetabolicpathways(FigureS2E).Incontrast,mRNAsenrichedin157 BAG6KO-EVs were characterized by functional clusters specifically associated with immune responses,158 melanoma and tumour progression, the latter including vascular wall signaling, extracellular matrix159 remodellingandplateletactivation(FigureS2EandmRNAlistinTableS1).160 Selected mRNAs encoding disease-related, tumor-promoting factors that were enriched in BAG6KO-EVs161 were analyzed by qRT-PCR (Figure 2A). This analysis confirmed a higher abundance of these mRNAs in162 BAG6KO-EVs than inWT-EVs e.g. for Ctnnal1 (encodingα-catulin, Catenin alpha-like protein 1), which is163 known to contribute critically to the invasive behavior ofmetastatic cells [22,23]. Collectively, these data164 indicatethatthepresenceofBAG6intheEV-releasingcellimpactsontheEVmRNAcargo.165 Next,massspectrometry(MS)wasappliedtocomprehensivelycharacterizetheEVproteomeunderhypoxic166 conditions.Asdepictedinavolcanoplotandaheatmap(Figure2B,C),315ofabout3000detectedproteins167 were significantly (p<0.01) up- or downregulated and 33 were only detectable in either WT or BAG6KO168 samples(amongthesewasBAG6asexpected)(MSdataseeTableS2).169 Of note, differentially expressed proteins, either decreased (subcluster WT up) or increased (subcluster170 BAG6KO up) in BAG6KO-EVs clustered into distinct endosomal vesicle signatures (DAVID, functional171

5

annotation clustering; https://string-db.org(24)) (Figure 2B,C). In these subclusters, 12 out of 28 proteins172 (43%)werespecificallyupregulatedinWT-EVs(butnotinWTcomparedtoBAG6KOwholecelllysates),while173 22outof32proteins(69%)werespecificallyenrichedinBAG6KO-EVs(butnotincorrespondingwholecell174 lysates)(FigureS2F-H).ThisreferstoaregulatedEVcargoloadingpathwaydependentonBAG6insteadof175 anunspecificenrichmentofcellularproteinsinEVs.176 WT-EVs-associatedproteinssuchasRAB27,MREG,MYO5aorMYO7aareindicativeforspecializedvesicles,177 termed melanosomes, which are released by differentiated melanocytes [25], whereas these proteins178 were less abundant or not detectable in BAG6KO-EVs. Instead, specific proteins of the ESCRT complex,179 including ESCRT-I (TSG101), ESCRT-II (SNF8), as well as the ESCRT-III (CHMP2a, CHMP2b, CHMP4b and180 CHMP5) were significantly upregulated in BAG6KO-EVs. These proteins were previously identified in181 associationwithbonafideexosomesisolatedfromhumandendriticcells[26](Figure2C).182 Finally, further analysis revealed that proteins enriched in BAG6KO-EVs clustered into the processes of183 migration and angiogenesis. Upregulated angiogenesis-related proteins included SEMA3C, EDIL3, ADCY9,184 SRC, PRKCA, and SPHK1. On the contrary, TIMP3, an inhibitor of metalloproteinases involved in the re-185 modelling of the extracellular matrix via targeting ADAM10, ADAMts1 and ADAMts4 [27] and thereby186 counteractingmetastasiswassignificantlyhigherinWT-EVscomparedtoBAG6KO-EVs.187 Thus,RNAseqandMSanalysis revealedthatbothWT-EVsandBAG6KO-EVsoriginate fromtheendosomal188 compartmentbutrepresentdistinctEVsubtypes,probablygeneratedviaaBAG6-dependentpathway.189 BAG6-presentingbutnotBAG6-deficientEVsinhibitmetastasistothelung.190 Given thedifferences inmRNAandprotein cargo loading of EVs isolated fromWTandBAG6KO cells,we191 analyzedthe invivoactivityoftheseEVsusinganexperimentalB-16Vlungmetastasismodel. Initially,WT-192 EVsorBAG6KO-EVs(20μg,i.v.,tailvein,protein/particleratioseeFigureS3B)orPBS(ctrl)wereinjectedinto193 tumorfreeC57BL/6miceweeklyoverafive-weekperiod(Figure3A).24hoursfollowingthelast injection,194 thelungtissuewasisolatedandsubjectedtonextgenerationRNAsequencing.195 A significant upregulation of 50 transcriptswas observed in the lungs from animals treatedwithWT-EVs196 comparedtountreatedcontrolsandamongthese6transcriptswerealsosignificantlyupregulatedcompared197 toBAG6KO-EVtreatedanimals(TableS3A).TheseenrichedmRNAswereinvolvedinimmunologicaldefense198 pathways(DAVID,functionalannotationclustering)andincludedthetranscriptionfactorSpic,whichcontrols199 the development of red bone marrow and pulp macrophages. In line, the macrophage soluble defense200 receptorCd5l; the interleukin IL10; thechemokineCxCl13,which ischemotactic forBcellsandweakly for201 monocytesandmacrophagesandtheFcreceptor-likescavengerreceptorFcrlsweresignificantlyincreased.202 Secondly, transcripts for theNK cell lectin receptorsKlri1 andKlrc3were significantly higher expressed in203 lungs of WT-EVs-treated animals compared to PBS controls, although the comparison to BAG6KO-EV204 treatmentdidnotreachsignificance(FigureS3AandTableS3A).205 The analysis of animals treated with BAG6KO-EVs showed a significant upregulation of 27 transcripts in206 comparison to both untreated and WT-EVs-treated animals (Supplementary Table S3B). Most strikingly,207 almosthalfof theupregulated transcriptswere indicative for theaccumulationandactivityofneutrophils208 (Figure3BandTableS3B).Neutrophilsareknowntopromotemetastasisformationbeforecancercellarrival209 [28,29],althoughanti-tumoractivitywasobeservedinaimmunodeficientmodel[15].Depictingthesehits210 in an interaction enrichment network (https://string-db.orgstring) confirmed specific and functionally211 meaningful interactions (PPI enrichment p value of 1x10-16) (Figure 3B). The upregulation of selected212 neutrophil-associated hits (S100a9, Elane, MPO, Prtn3) was confirmed by qRT-PCR and213 immunohistochemistry(FigureS3C-D).214 Taken together, EVsderived fromWTorBAG6KOcells causeddistinct geneexpression changes inmouse215 lungseithercorrespondingtoananti-orpro-tumorigenicphenotype.216

6

217 To investigate the impact of EVs on lung metastasis, animals were pre-treated with either WT-EVs or218 BAG6KO-EVs(10µgeverydayfor14days)orinjectedwithPBSfollowedbythei.v.injectionofwildtypeB-219 16Vtumorcells(Figure3C).220 EVs fromWT cells inhibited lungmetastasis significantly compared to untreated animals (Figure 3D). The221 inhibitionofmetastasiscorrelatedwiththeaccumulationofbonemarrowmonocytes,whichcorresponded222 to theLy6Clowphenotypeofanti-tumorpatrollingmonocytes [19]andadecreaseofmetastasis-promoting223 neutrophilsinthebonemarrow(Figure3E,gatingstrategypresentedinFigureS3E).224 Surprisingly, treatmentwith BAG6KO-EVs had no significant effect comparedwith untreated controls. An225 increaseofmetastasiswasnotobserved,althoughBAG6KO-EVsinducedapre-metastaticnichephenotype226 inthelungs(Figure3B).ThismightbeduetoaninterferenceofBAG6KO-EVswithWT-EVsreleasedfromthe227 injectedWTB-16Vmelanomacells(whichwerenotpresentintheexperimentpresentedinFigure3A,B).228 TIMP3 was identified as a top hit among the proteins specifically enriched in WT-EVs (Table S3A) and229 thereforeacandidatetoexecutetheanti-metastaticfunctionofWT-EVscomparedtoBAG6KO-EVs.TIMP3is230 aninhibitorofthematrixmetalloproteinases involvedinmigration,degradationoftheextracellularmatrix231 andmacrophagepolarization[27].HumancancersincludingmelanomashowconsistentlythatahighTIMP3232 expression level is associated with a favourable patient survival (proteinatlas.org). B-16Vmelanoma cells233 incubatedwithBAG6KO-EVsexhibitedanincreaseinwoundclosureascomparedtoWT-EVincubationand234 thePBScontrol (FigureS3F),consistentwiththepro-tumorcharacterisationoftheEVcontent(Figure2A).235 ThisphenotypewasreversedbyoverexpressionofTIMP3,whichisinlinewithapreviousstudyshowingthat236 TIMP3suppressesmelanomacellmigration[31].Inaddition,invitrodifferentiationofbonemarrow-derived237 monocytes in the presence of WT-EVs polarized towards an inflammatory M1 macrophage phenotype238 characterised by increased mRNA expression of IL-12, Nos2 and Cxcl10 and characteristic dendritic239 morphology (Figure S3G). This phenotypewas abrogatedby incubationwithBAG6KO-EVs and rescuedby240 overexpressionofTIMP3.These results suggest that theanti-metastaticactivitymightbeat leastpartially241 mediatedthroughBAG6-dependentrecruitmentoffunctionalTIMP3toEVs.242 p53iscriticallyinvolvedintheinduciblereleaseofBAG6-presentingEVs.243 Thenext setof experimentswasperformed to further analyze themolecularmechanisms responsible for244 distinctcargoloadingandthusfunctionaldifferencesbetweenWT-andBAG6KO-EVs.ThereleaseofBAG6-245 positive EVs could be induced by the inhibition of cellular de-acetylases (Figure 1A), suggesting that246 acetylationmightbeinvolvedinthesecretion.ItisknownthatBAG6isanessentialco-enzymeforthemajor247 CBP/p300acetyltransferase,firstreportedfortheacetylationofp53inamousemodel[32].Thisprompted248 us to test whether a BAG6/CBP/p300 complex formation was involved in the regulation of EV release.249 Immunoprecipitation(IP)ofBAG6fromHEK293celllysatestreatedwithdoxoallowedtheco-precipitationof250 p53 and vice versa (Figure 4A). Moreover, CBP/p300 antibodies co-precipitated both BAG6 and p53,251 suggestingthatanacetyltransferasecomplexwasformedupondoxotreatment(Figure4A).IPswithBAG6-252 deficient cells confirmed the specificity of the antibodies used (Figure S4A). The interaction of in vitro253 translated proteins obtained in a cell free system confirmed binding of p53 and BAG6 (Figure 4B).254 Speculatingthatp53anditsacetylationinresponsetoDNAdamagewascrucialfortheinduciblereleaseof255 BAG6-presentingEVs,wetestedthebasaland induciblerelease fromHCT116cellsandtheirp53-deficient256 (p53KO)derivative.Asexpected,p53KOcellsfailedtoreleaseincreasedamountsofEVsinresponsetoboth257 doxo- or LBH-treatment (Figure 4C). This phenotype could be rescued with p53WT but not with an258 acetylation-deficient p53 mutant (p53K372R/K373Rp53lys372/lys373) indicating that acetylation of p53 is259 indispensablefortheEVrelease(Figure4D;FigureS4C).Similarly,theinduciblereleaseofBAG6-positiveEVs260 upondoxotreatmentwasabrogated inHEK293cellswithasiRNA-mediatedp53knockdown(FigureS4B).261

7

Toverifythe invivorelevanceofthispathway,p53WTorp53KOsplenocytesisolatedfromtheestablished262 Tcl1-drivenCLLmousemodelwereused[33].Inlinewiththeinvitrodata,aninductionofthesecretionof263 BAG6-positiveparticleswasrestrictedtop53WTcells(Figure4E)confirmingtheimpactofp53onEVrelease.264 BAG6andCBP/p300arecrucialfortheacetylationofp53andregulatethereleaseofEVs.265 ToaddresstheimpactofBAG6/CBP/p300forp53-acetylationandsubsequentEVrelease,weusedHEK293266 andB-16VBAG6KOandCBP/p300doubleKO(dKO)cells(characterizationofCBP/p300dKOcelllines[34].As267 expected,dKOoftheacetyltransferasesCBP/p300abrogatedtheacetylationofp53atLysine373(p53K373)268 nearly completely and the depletion of BAG6 affected this acetylation in a similar manner (Figure 5A)269 confirmingthatBAG6wascrucial forp53K373acetylation.Thep53-acetylationcouldberestoreduponre-270 expressionofWTBAG6inBAG6KOcells,whichwasnotobservedupontransfectionofanN-terminaldeleted271 BAG6mutant(constitutivenuclearmutant,nucBAG6aa555-1132)(Figure5BandS5A,B).Subsequently,the272 releaseofEVsfromhumanHEK293andmouseB-16VBAG6cellswasanalyzed.Surprisingly,weobserveda273 significantly higher basal release of EVs in the BAG6KO cell line compared to their WT parental cells,274 indicatingthatBAG6inhibitedEVreleaseundernon-stressconditions(Figure5C,whitebars).275 ThisphenotypecouldberescuedinbothBAG6KOcelllinesuponexpressionofanN-terminalBAG6-deletion276 mutant(nucBAG6),whichreducedtheEVreleasesignificantly(Figure5C,blackbars).ThenucBAG6mutantis277 strictly nuclear and, unlike theWT BAG6 protein, does not translocate to the cytoplasm e.g. upon doxo278 treatment[15,35](Figure5DandS5B,C).279 Inline,theoverexpressionofWTBAG6,whichisabletoshuttlebetweenthenucleusandthecytoplasm,did280 notreducetheEVrelease(Figure5C,greybarsandcartooninFigureS5D).281 Toaccount for thepossibilityofdifferences inEVreleaseduetosinglecellclonesheterogeneity,wehave282 generatedseveralWTandBAG6KOB-16Vclones (FigureS2A)andconsistentlyobservedan increase inEV283 release fromBAG6KOcell clones compared toWT clones supporting amechanistic rather than stochastic284 regulation.285 BAG6waspreviously identifiedasanuclear-cytoplasmashuttle forCBP/p300 [35].To investigatewhether286 theinhibitionoftheEVreleasebynuclearBAG6wasassociatedwiththecaptureofCBP/p300inthenucleus,287 we analyzed the CBP/p300 translocation in HEK293WT and BAG6KO cells.Western blots of cytoplasmic288 HEK293 cell fractions revealed an increase of cytoplasmic CBP/p300 upon doxo treatment, which was289 diminishedinBAG6KOcells(Figure5E).290 TheseresultssuggestthattheBAG6/CBP/p300acetyltransferasecomplexdirectsthereleaseofEVs.Under291 basal conditions BAG6 localized predominantly in the nucleus and the release of EVs was low. Doxo292 treatment induced the nuclear export of BAG6 and CBP/p300 to allow the acetylation of p53, which293 triggeredthereleaseofBAG6-positiveexosomeswithanti-metastaticactivity.DepletionofBAG6abolished294 the inhibitory activity of nuclear BAG6 and allowed an increased release of BAG6-negative EVswith pro-295 tumorigenic cargo (Figure 5C), which in turn was blocked by overexpression of a nuclear BAG6 mutant296 unabletoshuttletothecytoplasm.297 DoxotreatmentinducescomplexformationbetweenBAG6andESCRTproteins.298 Wenext addressed themolecular link of the BAG6-acetylation complex and EV release. The best-studied299 pathwayinvolvedinEVbiogenesisandsecretionistheESCRTpathwayformedbyfourcomplexes,ESCRT-0,300 ESCRT-I,ESCRT-II,andESCRT-IIIactingsuccessivelyandinconcertwithaccessoryproteins[36].301 Co-immunoprecipitation experiments revelead an association of BAG6 with endogenous ESCRT-proteins302 HRS,TSG101andAlixandthisinteractionwasincreasedinresponsetodoxotreatment(Figure6A).Inline,303 sequenceanalysisoftheBAG6generevealedalateendosomalmotifP(S/T)APwithinthesequence,whichis304 knowntorecruitTSG101 (FigureS6A).LateendosomalmotifsLYPXnLandtheP(S/T)APmotifwere initially305 discovered in HIV-1 (human immunodeficiency virus) and EIAV (equine infectious anaemia virus) and are306

8

requiredforviralrecruitmentoftheESCRT-complextoensuretheefficientreleaseofnascentvirionsfrom307 the infected cells [37]. Direct binding of BAG6 to TSG101 was in fact mediated and dependent on this308 P(S/T)AP motif as shown by yeast two hybrid experiments with a WT BAG6 and a PSAP/PTAP-deletion309 mutant, respectively (Figure 6B). Altogether, these experiments suggest the involvement of the ESCRT310 complexintheBAG6-mediatedEVrelease.311 RelevanceofBAG6-dependentEVformationinhumanmelanoma.312 Based on our in vitro and in vivo mouse data, we speculate that late, metastatic melanoma stages are313 characterizedbythepresenceofpro-metastaticEVsresemblingBAG6KO-EVs isolatedfromB-16VBAG6KO314 cells. Consistent with this hypothesis, analysis of TCGA data deposited at proteinatlas.org315 (/ENSG00000204463-BAG6/pathology/tissue/ melanoma#12000001big) showed that the expression of316 BAG6declinedinmelanomaswithtumorprogressionandwassignificantlylowerinmetastaticstagesIII/IV317 comparedtoearlystages(Figure7A).Moreover,transcriptsfortheoncoproteinCTNNAL1(α-Catulin)which318 is known to drive malignant invasion and metastasis [22, 23] were clearly detectable and significantly319 enriched in EVs from the plasmaofmetastatic late stage patients compared to EVs from stage I patients320 (Figure7B).AsimilarphenotypewasobservedinEVscollectedfromBAG6KOmousecells(Figure2A).321 Mutations in the BAG6 gene, which we consider a novel tumor suppessor gene, are rarely detected in322 melanoma and other tumors, but we investigated a specific mutation described in a melanoma and a323 pancreatic cancer patient (uniprot/BAG6_HUMAN, stop gainedBAG6R786*). Thismutation at codon 786324 generatesastopcodonandresultedintheexpressionofatruncatedproteinupontransfection(FigureS7B325 and S7C). BAG6R786*wasunable to rescue thep53K373 acetylationnull phenotypeupon transfection in326 BAG6KOcells. Instead itdiminished thep53-acetylation inWTBAG6cells significantly (Figure7C)andcan327 thusbeconsideredasadominantnegativemutantofBAG6,which is known to formdimers [38] (control328 experiment in Figure S7A). In line with our finding of the requirement for p53 acetylation in EV release329 (Figure4),transfectionofBAG6R786*resultedinareducednumberofparticlessecretedbyWTcells(Figure330 7D). Taken together, these data suggest that a diminished expression or defects in the activity of BAG6331 impactontheformationandreleaseofanti-versuspro-metastaticEVs.332 Discussion333 Recentresearchontumorcell-released,andspecificallyonmelanomacell-derivedEVs,highlighttheircrucial334 rolefortumorprogressionandfortheformationofpre-metastaticniches[39-41].Onthecontrary,several335 studiesreportedtheexistenceofEVspromotingmelanomatumorcellclearancepredominantlyviaimmune336 cellactivation[7,8].However, little isknownaboutthemechanismsthatare involved inEVformationand337 cargoloading,whichregulatedistinctfunctionsofEVs.HerewepresentevidencethatthechaperoneBAG6338 isamolecularswitchdirectingthereleaseofEVswitheitheranti-orpro-metastaticactivity(seesummary339 model,FigureS9).340 ThisisinitiallybasedonthefindingthatEVsfromBAG6-proficientandBAG6-deficientmelanomacellswere341 characterized by distinct protein and mRNA cargo loading (Figure 2). Both EV types changed the gene342 expression profile in the lungs ofmice (Figure 3), which is compatible with the detected similar level of343 integrin-6, known to be associated with an exosomal tropism to the lung [42]. However, we observed344 distinctmolecularchangesinthelungsofmiceconditionedwiththesespecificEVsubtypesreflectingeither345 anti-orpro-metastaticactivity.WT-EVsnotonlyinducedtheexpressionofgeneswithreportedanti-tumor346 activity but also caused accumulation of patrolling Ly6Clow monocytes, which are known to cause tumor347 clearanceatthemetastaticnicheinaNKcell-dependentmanner[8,19].Moreover,EVscollectedfromthe348 B-16VWTcellssuppressedlungmetastasiswhereasEVsfromBAG6KOcellsfailedtoinhibitlungmetastasis349 intheB-16Vexperimentaltransplantationmodel (Figure3). Incontrast,BAG6KO-EVs inducedaneutrophil350 genesignature indicative for the formationofapre-metastaticniche [20].Recently, theoverexpressionof351

9

theserineproteaseinhibitorandtumorsuppressorPEDF(althoughnotdetectableinthemurineB-16VEVs)352 wasshowntoturnmetastasis-promotingEVs intotumor-inhibitingmelanomaEVs[8].Notably,thetargets353 ofPEDF,elane(elastase)andcathepsinG,whichareindispensibleformelanomametastasis[43]wereboth354 significantly induceduponB-16VBAG6KO-EVs treatment inmouse lungs (butnot in response toWT-EVs).355 We observed direct uptake of B-16V EVs by neutrophils in the lung of reporter mice using the Cre-loxP356 methodofEVtransfer(FigureS8,[44]andexpectthateffectsmediatedbyWT-andBAG6KO-EVsmightbe357 bothdirectandindirectasobservedinotherstudies[45].358 AnalyzingthetophitenrichedinWT-EVs,weidentifiedTIMP3asoneofthestrongcandidates,whichmight359 contributetotheiranti-metastaticactivity(FigureS3F,G).However,itseemsunlikelythattheanti-metastatic360 activityismediatedexclusivelybytheactionofasinglecandidatemolecule(suchasBAG6orTIMP3),butis361 ratherdeterminedbytheinterplayofthespecificEVcargocompositiondefinedbythebiogenesispathway.362 The EV content and immune-alert activity changes in response to cellular stress stimuli including363 doxorubicin,HDAC-inhibitionandhypoxia(thisstudy),receptoractivation(RIGI)orcytokines(IFNg).Herewe364 establishedadirectmolecularlinkofBAG6totheformationofsuchstress-inducedEVs.Mechanistically,the365 releaserequiredtheconcertedand inducible translocationofBAG6andCBP/p300 into thecytoplasmand366 theacetylationofp53(Figure4and5).TheacetyltransferasecofactoractivityofBAG6regulatingCBP/p300367 is involved indifferentcellularprocesses includingmetabolic regulation [46]orapoptosis [32].Recently,a368 roleforBAG6andp300-dependentacetylationinautophagywasreported[35].NuclearimportofBAG6and369 p300inresponsetostarvationallowedtheacetylationofnuclearp53,whichtriggeredautophagy,whereas370 basalcytoplasmicBAG6 levelwassufficient toacetylateATG7, in turn inhibiting thisprocess.Thenuclear-371 cytoplasmicshuttleofBAG6togetherwith theassociatedacetyltransferasesmightexplain thatautophagy372 andthereleaseofexosomesareconsideredasmutuallyexclusivecellularresponses[47].Inline,cytoplasmic373 p53,butnotnuclearp53wasshowntoinhibitautophagy[48],althoughtheunderlyingmechanismsarestill374 discussedcontroversely[49].375 A role for p53 in exosome secretion was already suggested [50] and recent work confirmed that p53376 deficiency ormutations lead to the enrichment of tumor-promoting proteins in released EVs [51]. In line377 withalinkofBAG6andp53inEVbiogenesisitwasreportedthatp53deficiencyreducedtheamountofthe378 ESCRTcomponentHRS(whichinteractswithBAG6,Figure5)inexosomesofcoloncarcinomacells.Notably,379 geneenrichmentanalysisusingCPDBperformedontheproteomicEVdataofHCT116p53nullandwildtype380 cells showed that p53null EVs also up-regulated pathways associated with tumor progression (e.g.381 angiogenesisp=0.000352andTGFβp=0.00399).Inaddition,exosomesfromp53nullcellsofthisstudywere382 smaller in size compared with their WT counterparts, a phenotype reminiscent to the BAG6KO-EVs383 phenotype shown in our study (Figure 1). Interestingly, previous studies have also shown that different384 stressstimuli likeheatshockandcisplatintreatmentinducedthereleaseofEVswithsmallersizebasedon385 EM measurements and these EVs exhibited enhanced ability to induce invasive behavior [52, 53].386 Furthermore,acidity,knowntocontributetotheimmuneescapeofcancercells[54,55]hasbeenreported387 asanotherimportantstressfactorwithinthetumourmicroenvironmenttoinduceincreasedamountsofEVs388 frommelanomacells[56,57].389 The ability of BAG6 to recruit to the exosomal pathway via association with components of the ESCRT-390 complex(Figure6)mayexplainitsregulatoryroleonEVcargoloading.Wefoundthelateendosomalmotif391 P(S/T)AP of BAG6 to be important for the direct association with TSG101, an interaction mechanism392 reminiscenttotheESCRTrecruitmentduringvirusassembly[37].Recently,therecruitmentofgalectin-3to393 exosomes was shown to be dependent on a P(S/T)AP motif [58] suggesting that this sequence can be394 regardedasanexosomaltargetingsequencedirectingthepackagingofsolubleproteinsinoronexosomes.395 Recently, BAG6 was reported to control the activity of Rab GTPases, which are known regulators of396

10

intracellular vesicle trafficking [59]. Notably, the regulation of specific Rab GTPases is an attractive397 hypothesistoexplainthesurprisingincreaseinEVreleaseintheabsenceofcellularBAG6undernon-stress398 conditionsandrequiresfurtherinvestigations.399 In view of the important role of BAG6 together with CBP/p300/p53 in tumor immune surveillance, an400 impaired function of BAG6 and CBP/p300 might be expected, as generally observed for p53. However,401 mutations in these genes in tumors are rather rare and defects might be related to posttranslational402 modificationsorepigeneticaberrations.Infact,re-expressionofBAG6andCBP/p300uponinhibitionofthe403 transcriptional repressor BCL6 rendered B cell lymphoma cells and xenografts susceptible to HDACi and404 HSP90 therapy [60]. Tumor suppressor features for BAG6 were previously described and BAG6405 polymorphismswereassociatedwithlungcancerriske.g.basedongenome-wideassociationstudies[61].In406 linewithtumor-suppressingactivity,weobservedthatarareBAG6mutationidentifiedinamelanomaand407 pancreatic cancer patient encoding a truncated protein had dominant negative activity affecting the408 acetyltransferaseactivityofCBP/p300/BAG6(Figure7).HighlightingtheimportanceofBAG6-regulatedEVs,409 we detected (tumor-promoting) vesicularα−catulinmRNA (enriched inmouse BAG6KO-EVs compared to410 WT-EVs)exclusivelyinlatestageEVsfrommetastaticmelanomapatients,andthiscorrelatedinverselywith411 thecellularBAG6transcriptexpression.MoreextensivedissectionoftheBAG6/CBP/p300-p53pathway,e.g.412 bystratifyingtumorpatientsandtheirp53statusaswellasanalysisofadditionalcellandpatientsettings413 willbenecessary in future studies.Wehypothesize that thedescribedpathwaymightbeaffectedbyp53414 mutations, but might also be influenced by mutations in other stress-related pathways, for instance a415 deregulatedRetinoblastomapathwaywhichisoftenobservedinmelanomapatients.416 In conclusion, our findings identify BAG6 as a crucial regulator of immune alert, anti-metastatic EVs via417 BAG6/CBP/p300-mediated acetylation of p53 and direct association with the ESCRT machinery via its418 P(S/T)AP motif. Loss of BAG6 allowed the release of EVs with pro-tumorigenic cargo and activity. This419 rendersBAG6apotential therapeutic tool tospecifically interferewith thebiogenesisofexosomes,which420 canre-programthebiologicalactivityoftumor-cellderivedEVsintoanti-cancermoieties.421 422 MaterialsandMethods423 (seeSupplementaryMethodsforextendedversion)424 Humansamples425 Plasmawasobtainedwithinformedwrittenconsentbythepatientsandapprovalbythelocalethics426 committeeoftheUniversityHospitalEssen[ref.no.11-4715].427 Invivotreatmentexperimentsandsamplepreparationfromexperimentalmice428 8-12weeksoldC57BL/6Jmice,housedandfedunderpathogen-freeconditions,were injectedwithEVsor429 cells intravenously into the tail vein. Dissected mouse lungs were either frozen in optimal-cutting-430 temperature compound (OCT Tissue Tec) on dry ice and stored at -80°C for immunohistochemistry or431 directlytakenupinRNAstabilizationsolutionandfrozenat-20°Cuntilfurtherprocessed.432 Celltreatment433 Celllinesweretreatedfor16hwith100nMdoxorubicinor100nMLBH,respectively.Forshortterm(5–120434 min)DNAdamageinduction,cellsweretreatedwith1or10µMdoxorubicin(asstatedinthefigurelegends).435 Plasmidsandtransfection436 pcDNA3.1wild-typeBAG6(BAT3)andnucBAG6aredescribed(15).pcDNA3.1BAG6R786*wasgeneratedby437 site-directedmutagenesis usingQuickChange II Site-DirectedMutagenesis Kit and the expression plasmid438 pcDNA3.1wild-typeBAG6asatemplate.HRSexpressionvectorpCS2wasobtainedfromAddgene(#29685)439 andpcDNA3.1forp53waskindlyprovidedbyDr.Pattingre(INSERM,France).440 Antibodies441

11

AllantibodiesusedarelistedinTableS4.442 Invitroexpressionassay443 Invitroexpressionexperimentswereperformedusingcell-freeproteinexpressionkitbasedonLeishmania444 tarentolae.RecombinantproteinswereobtainedbycloninginpLEXSY-invitrovector.445 Immunoprecipitation446 Cell lysates or in vitro expressed proteins were precipitated using specific antibodies against BAG6, p53,447 CBP/p300, HRS, ubiquitin or acetyl-lysine. Aminimum of 1 µg of antibodywere used for 100 µg of total448 proteinandProteinAmagneticbeadswereusedforpull-down.449 EVpreparation450 EVswerecollectedincellcultureforeither24or48hineitherEV-depletedmediumorprotein-freeCD293451 medium.Forpurifcation, consecutivepre-centrifugation stepsat300xg (10min),2,000xg (10min)and452 3,500 x g (20min) for clearance of cells and cellular debris was performed before ultracentrifugation at453 10,000xg(60min)and/or100,000xg(90min)usingSW41TirotororType45Tiforatleasttwotimeswith454 intermediate resuspension in PBS orHBSS and ultra-centrifugation at the respective g force using TLA-45455 rotor in the last centrifugation. EVs were resuspended in PBS or HBSS. The amount of EV protein was456 quantified by Nanodrop 1000 and/or using BCA assay. The number of particles was determined by457 NanoparticleTrackingAnalysis.458 QuantitativeRT-PCR459 RNAfromEVsorlungtissuewasreversetranscribedwithRevertAidFirstStrandcDNASysnthesisKitusing460 oligo-d(T)primersand/orrandomprimers.QuantitativePCRmeasurementswereperformedona7500real-461 timePCRsystemusingSYBRGreen.Initialheatinactivationwas95°Cfor15minand40cyclesof15secat462 94°C,30secat56°C,and30secat72°Cwereperformedfollowedbymeltingcurveanalysis.Allprimersused463 arelistedinTableS5.464 Massspectrometrybioinformaticsandstatisticalanalysis465 All mass spectrometric raw data were processed with Maxquant using default parameters. Briefly, MS2466 spectraweresearchedagainsttheUniprotMOUSEdatabase,includingalistofcommoncontaminants.False467 discoveryratesonproteinandPSMlevelwereestimatedbythetarget-decoyapproachto1%(ProteinFDR)468 and 1% (PSM FDR), respectively. The minimal peptide length was set to 7 amino acids and469 carbamidomethyolation at cysteine residues was considered as a fixed modification. Oxidation (M) was470 included as variable modification. The match-between runs option was enabled. LFQ quantification was471 enabled using default settings. Downstream data processing was conducted within the Perseus472 computational platform. Briefly, protein groups flagged as „reverse“, „potential contaminant“ or „only473 identified by site“ were removed from the data. LFQ data were log2 transformed. Statistical analysis of474 differentially regulated proteinswas performed using a two-sided t-test (fudge factor s0was adjusted to475 0.1).Resultingpvalueswerecorrectedformultipletestingusingapermutation-basedFDRapproach.476 RNAseq477 RNAqualitywasassessedusingtheExperionRNAStdSensAnalysisKit.RNA-seqlibrarieswerepreparedfrom478 totalRNAusing theTruSeqStrandedmRNALTkitaccording to themanufacturer’s instructions.Qualityof479 sequencinglibrarieswascontrolledonaBioanalyzer2100usingtheAgilentHighSensitivityDNAKit.Pooled480 sequencing librarieswerequantifiedwithdigitalPCR (QuantStudio3D)and sequencedon theHiSeq1500481 IlluminaplatforminRapid-Runmodewith50basesinglereads.RNAseqwasperformedfromRNAisolated482 usingtheRNAeasyminikitwiththeIlluminaTruseqmRNAkitv2onanIlluminaHiseq1500accordingtothe483 manufacturer’sinstructions.484 RNA-SeqBioinformaticAnalysis485 Raw transcriptome readswere aligned to themouse genome fromEnsembl 89using STARversion2.4.1a486 [62]. Gene expression was quantified on gene models that included only protein coding transcripts for487

12

protein coding genes using custom python scripts. Differential genes were called using edgeR [63] at a488 thresholdofFDR<=0.05;|log2FC|>=1andcountspermillion>=.3.489 Statisticalanalysis490 StatisticalanalyseswereperformedusingGraphpadPrismsoftware.Two-tailed,unpairedStudent’s t-tests491 wereperformedtoanalyzethesignificanceofmeanvaluesbetweentwovariables,ifnototherwisestated.492 StatisticalanalysisofEVseducationanimalexperimentswasperformedusingnon-parametricKruskal-Wallis493 test(meanrankscomparedwithWT-EVsgroup)andDunn’smultiplecomparisonstest.AnunpairedWelch’s494 t-testwasperformedtoanalyzeTCGAdatatoaccountforunequalsamplesizes.495 Datadeposition496 ThemassspectrometryproteomicsdatahavebeendepositedtotheProteomeXchangeConsortiumviathe497 PRIDE partner repository with the dataset identifier PXD010677. RNAseq data of mouse lungs were498 deposited at ArrayExpress accession E-MTAB-7119. RNAseq data of B-16V EVs were deposited at499 ArrayExpressaccessionE-MTAB-7119.500 501 Acknowledgements502 WethankClothildeThery,INSERMFrancefordiscussingtheproteomicsdataandSophiePattingre,INSERM,503 Franceforprovidingp53mutants.WethanktheCECADproteomicsfacility,Cologne,formassspectrometry504 measurementsand inparticularChristianFrese forhelpwith thedataanalysis.WethankGergardWeber,505 FFEServiceGmbH,Feldkirchen,forassistancewiththeFFEseparation.506 Conflictofinterest:Theauthorsdeclarethattheyhavenoconflictofinterest.507 508 509

510

13

References511 1. Ruivo CF, Adem B, Silva M, Melo SA. The Biology of Cancer Exosomes: Insights and New 512 Perspectives. Cancer Res. 2017;77(23):6480-8. 513 2. Becker A, Thakur BK, Weiss JM, Kim HS, Peinado H, Lyden D. Extracellular Vesicles in 514 Cancer: Cell-to-Cell Mediators of Metastasis. Cancer Cell. 2016;30(6):836-48. 515 3. Costa-Silva B, Aiello NM, Ocean AJ, Singh S, Zhang H, Thakur BK, et al. Pancreatic cancer 516 exosomes initiate pre-metastatic niche formation in the liver. Nat Cell Biol. 2015;17(6):816-26. 517 4. Fabbri M, Paone A, Calore F, Galli R, Gaudio E, Santhanam R, et al. MicroRNAs bind to 518 Toll-like receptors to induce prometastatic inflammatory response. Proc Natl Acad Sci U S A. 519 2012;109(31):E2110-6. 520 5. Besse B, Charrier M, Lapierre V, Dansin E, Lantz O, Planchard D, et al. Dendritic cell-521 derived exosomes as maintenance immunotherapy after first line chemotherapy in NSCLC. 522 Oncoimmunology. 2016;5(4):e1071008. 523 6. Reiners KS, Topolar D, Henke A, Simhadri VR, Kessler J, Sauer M, et al. Soluble ligands for 524 NK cell receptors promote evasion of chronic lymphocytic leukemia cells from NK cell anti-tumor 525 activity. Blood. 2013;121(18):3658-65. 526 7. Dassler-Plenker J, Reiners KS, van den Boorn JG, Hansen HP, Putschli B, Barnert S, et al. 527 RIG-I activation induces the release of extracellular vesicles with antitumor activity. 528 Oncoimmunology. 2016;5(10):e1219827. 529 8. Plebanek MP, Angeloni NL, Vinokour E, Li J, Henkin A, Martinez-Marin D, et al. Pre-530 metastatic cancer exosomes induce immune surveillance by patrolling monocytes at the metastatic 531 niche. Nat Commun. 2017;8(1):1319. 532 9. Viaud S, Thery C, Ploix S, Tursz T, Lapierre V, Lantz O, et al. Dendritic cell-derived 533 exosomes for cancer immunotherapy: what's next? Cancer Res. 2010;70(4):1281-5. 534 10. Robbins PD, Morelli AE. Regulation of immune responses by extracellular vesicles. Nat Rev 535 Immunol. 2014;14(3):195-208. 536 11. Pitt JM, Kroemer G, Zitvogel L. Extracellular vesicles: masters of intercellular 537 communication and potential clinical interventions. J Clin Invest. 2016;126(4):1139-43. 538 12. Zhang L, Yu D. Exosomes in cancer development, metastasis, and immunity. Biochim 539 Biophys Acta Rev Cancer. 2019;1871(2):455-68. 540 13. Wortzel I, Dror S, Kenific CM, Lyden D. Exosome-Mediated Metastasis: Communication 541 from a Distance. Dev Cell. 2019;49(3):347-60. 542 14. Rangachari M, Zhu C, Sakuishi K, Xiao S, Karman J, Chen A, et al. Bat3 promotes T cell 543 responses and autoimmunity by repressing Tim-3-mediated cell death and exhaustion. Nat Med. 544 2012;18(9):1394-400. 545 15. Pogge von Strandmann E, Simhadri VR, von Tresckow B, Sasse S, Reiners KS, Hansen HP, 546 et al. Human leukocyte antigen-B-associated transcript 3 is released from tumor cells and engages 547 the NKp30 receptor on natural killer cells. Immunity. 2007;27(6):965-74. 548 16. Hollyoake M, Campbell RD, Aguado B. NKp30 (NCR3) is a pseudogene in 12 inbred and 549 wild mouse strains, but an expressed gene in Mus caroli. Mol Biol Evol. 2005;22(8):1661-72. 550 17. Casson J, McKenna M, High S. On the road to nowhere: cross-talk between post-translational 551 protein targeting and cytosolic quality control. Biochem Soc Trans. 2016;44(3):796-801. 552 18. Babst M. MVB vesicle formation: ESCRT-dependent, ESCRT-independent and everything in 553 between. Curr Opin Cell Biol. 2011;23(4):452-7. 554 19. Hanna RN, Cekic C, Sag D, Tacke R, Thomas GD, Nowyhed H, et al. Patrolling monocytes 555 control tumor metastasis to the lung. Science. 2015;350(6263):985-90. 556 20. Tüting T, de Visser KE. CANCER. How neutrophils promote metastasis. Science. 557 2016;352(6282):145-6. 558 21. Yu X, Harris SL, Levine AJ. The regulation of exosome secretion: a novel function of the 559 p53 protein. Cancer Res. 2006;66(9):4795-801. 560

14

22. Cao C, Chen Y, Masood R, Sinha UK, Kobielak A. alpha-Catulin marks the invasion front of 561 squamous cell carcinoma and is important for tumor cell metastasis. Mol Cancer Res. 562 2012;10(7):892-903. 563 23. Kreiseder B, Orel L, Bujnow C, Buschek S, Pflueger M, Schuett W, et al. alpha-Catulin 564 downregulates E-cadherin and promotes melanoma progression and invasion. Int J Cancer. 565 2013;132(3):521-30. 566 24. Szklarczyk D, Morris JH, Cook H, Kuhn M, Wyder S, Simonovic M, et al. The STRING 567 database in 2017: quality-controlled protein-protein association networks, made broadly accessible. 568 Nucleic Acids Res. 2017;45(D1):D362-D8. 569 25. Chi A, Valencia JC, Hu ZZ, Watabe H, Yamaguchi H, Mangini NJ, et al. Proteomic and 570 bioinformatic characterization of the biogenesis and function of melanosomes. J Proteome Res. 571 2006;5(11):3135-44. 572 26. Kowal J, Arras G, Colombo M, Jouve M, Morath JP, Primdal-Bengtson B, et al. Proteomic 573 comparison defines novel markers to characterize heterogeneous populations of extracellular vesicle 574 subtypes. Proc Natl Acad Sci U S A. 2016;113(8):E968-77. 575 27. Jackson HW, Defamie V, Waterhouse P, Khokha R. TIMPs: versatile extracellular regulators 576 in cancer. Nat Rev Cancer. 2017;17(1):38-53. 577 28. Landsberg J, Tuting T, Barnhill RL, Lugassy C. The Role of Neutrophilic Inflammation, 578 Angiotropism, and Pericytic Mimicry in Melanoma Progression and Metastasis. J Invest Dermatol. 579 2016;136(2):372-7. 580 29. Park J, Wysocki RW, Amoozgar Z, Maiorino L, Fein MR, Jorns J, et al. Cancer cells induce 581 metastasis-supporting neutrophil extracellular DNA traps. Sci Transl Med. 2016;8(361):361ra138. 582 30. Lozupone F, Luciani F, Venditti M, Rivoltini L, Pupa S, Parmiani G, et al. Murine 583 granulocytes control human tumor growth in SCID mice. Int J Cancer. 2000;87(4):569-73. 584 31. Das AM, Bolkestein M, van der Klok T, Oude Ophuis CM, Vermeulen CE, Rens JA, et al. 585 Tissue inhibitor of metalloproteinase-3 (TIMP3) expression decreases during melanoma progression 586 and inhibits melanoma cell migration. Eur J Cancer. 2016;66:34-46. 587 32. Sasaki T, Gan EC, Wakeham A, Kornbluth S, Mak TW, Okada H. HLA-B-associated 588 transcript 3 (Bat3)/Scythe is essential for p300-mediated acetylation of p53. Genes Dev. 589 2007;21(7):848-61. 590 33. Knittel G, Rehkamper T, Korovkina D, Liedgens P, Fritz C, Torgovnick A, et al. Two mouse 591 models reveal an actionable PARP1 dependence in aggressive chronic lymphocytic leukemia. Nat 592 Commun. 2017;8(1):153. 593 34. Sauer M, Schuldner M, Hoffmann N, Cetintas A, Reiners KS, Shatnyeva O, et al. CBP/p300 594 acetyltransferases regulate the expression of NKG2D ligands on tumor cells. Oncogene. 595 2017;36(7):933-41. 596 35. Sebti S, Prebois C, Perez-Gracia E, Bauvy C, Desmots F, Pirot N, et al. BAT3 modulates 597 p300-dependent acetylation of p53 and autophagy-related protein 7 (ATG7) during autophagy. Proc 598 Natl Acad Sci U S A. 2014;111(11):4115-20. 599 36. Juan T, Furthauer M. Biogenesis and function of ESCRT-dependent extracellular vesicles. 600 Semin Cell Dev Biol. 2018;74:66-77. 601 37. Hurley JH, Odorizzi G. Get on the exosome bus with ALIX. Nat Cell Biol. 2012;14(7):654-5. 602 38. Binici J, Hartmann J, Herrmann J, Schreiber C, Beyer S, Guler G, et al. A soluble fragment of 603 the tumor antigen BCL2-associated athanogene 6 (BAG-6) is essential and sufficient for inhibition of 604 NKp30 receptor-dependent cytotoxicity of natural killer cells. J Biol Chem. 2013;288(48):34295-605 303. 606 39. Lobb RJ, Lima LG, Moller A. Exosomes: Key mediators of metastasis and pre-metastatic 607 niche formation. Semin Cell Dev Biol. 2017;67:3-10. 608 40. Peinado H, Aleckovic M, Lavotshkin S, Matei I, Costa-Silva B, Moreno-Bueno G, et al. 609 Melanoma exosomes educate bone marrow progenitor cells toward a pro-metastatic phenotype 610 through MET. Nat Med. 2012;18(6):883-91. 611

15

41. Dror S, Sander L, Schwartz H, Sheinboim D, Barzilai A, Dishon Y, et al. Melanoma miRNA 612 trafficking controls tumour primary niche formation. Nat Cell Biol. 2016;18(9):1006-17. 613 42. Hoshino A, Costa-Silva B, Shen TL, Rodrigues G, Hashimoto A, Tesic Mark M, et al. 614 Tumour exosome integrins determine organotropic metastasis. Nature. 2015;527(7578):329-35. 615 43. El Rayes T, Catena R, Lee S, Stawowczyk M, Joshi N, Fischbach C, et al. Lung inflammation 616 promotes metastasis through neutrophil protease-mediated degradation of Tsp-1. Proc Natl Acad Sci 617 U S A. 2015;112(52):16000-5. 618 44. Zomer A, Maynard C, Verweij FJ, Kamermans A, Schafer R, Beerling E, et al. In Vivo 619 imaging reveals extracellular vesicle-mediated phenocopying of metastatic behavior. Cell. 620 2015;161(5):1046-57. 621 45. Boelens MC, Wu TJ, Nabet BY, Xu B, Qiu Y, Yoon T, et al. Exosome transfer from stromal 622 to breast cancer cells regulates therapy resistance pathways. Cell. 2014;159(3):499-513. 623 46. Jiang W, Wang S, Xiao M, Lin Y, Zhou L, Lei Q, et al. Acetylation regulates 624 gluconeogenesis by promoting PEPCK1 degradation via recruiting the UBR5 ubiquitin ligase. Mol 625 Cell. 2011;43(1):33-44. 626 47. Baixauli F, Lopez-Otin C, Mittelbrunn M. Exosomes and autophagy: coordinated 627 mechanisms for the maintenance of cellular fitness. Front Immunol. 2014;5:403. 628 48. Tasdemir E, Maiuri MC, Galluzzi L, Vitale I, Djavaheri-Mergny M, D'Amelio M, et al. 629 Regulation of autophagy by cytoplasmic p53. Nat Cell Biol. 2008;10(6):676-87. 630 49. Naidu SR, Lakhter AJ, Androphy EJ. PIASy-mediated Tip60 sumoylation regulates p53-631 induced autophagy. Cell Cycle. 2012;11(14):2717-28. 632 50. Yu X, Riley T, Levine AJ. The regulation of the endosomal compartment by p53 the tumor 633 suppressor gene. FEBS J. 2009;276(8):2201-12. 634 51. Sun Y, Zheng W, Guo Z, Ju Q, Zhu L, Gao J, et al. A novel TP53 pathway influences the 635 HGS-mediated exosome formation in colorectal cancer. Sci Rep. 2016;6:28083. 636 52. Samuel P, Mulcahy LA, Furlong F, McCarthy HO, Brooks SA, Fabbri M, et al. Cisplatin 637 induces the release of extracellular vesicles from ovarian cancer cells that can induce invasiveness 638 and drug resistance in bystander cells. Philos Trans R Soc Lond B Biol Sci. 639 2018;373(1737):20170065. 640 53. Bewicke-Copley F, Mulcahy LA, Jacobs LA, Samuel P, Akbar N, Pink RC, et al. 641 Extracellular vesicles released following heat stress induce bystander effect in unstressed 642 populations. J Extracell Vesicles. 2017;6(1):1340746. 643 54. Calcinotto A, Filipazzi P, Grioni M, Iero M, De Milito A, Ricupito A, et al. Modulation of 644 microenvironment acidity reverses anergy in human and murine tumor-infiltrating T lymphocytes. 645 Cancer Res. 2012;72(11):2746-56. 646 55. Pilon-Thomas S, Kodumudi KN, El-Kenawi AE, Russell S, Weber AM, Luddy K, et al. 647 Neutralization of Tumor Acidity Improves Antitumor Responses to Immunotherapy. Cancer Res. 648 2016;76(6):1381-90. 649 56. Parolini I, Federici C, Raggi C, Lugini L, Palleschi S, De Milito A, et al. Microenvironmental 650 pH is a key factor for exosome traffic in tumor cells. J Biol Chem. 2009;284(49):34211-22. 651 57. Logozzi M, Mizzoni D, Angelini DF, Di Raimo R, Falchi M, Battistini L, et al. 652 Microenvironmental pH and Exosome Levels Interplay in Human Cancer Cell Lines of Different 653 Histotypes. Cancers (Basel). 2018;10(10):370. 654 58. Banfer S, Schneider D, Dewes J, Strauss MT, Freibert SA, Heimerl T, et al. Molecular 655 mechanism to recruit galectin-3 into multivesicular bodies for polarized exosomal secretion. Proc 656 Natl Acad Sci U S A. 2018;115(19):E4396-E405. 657 59. Takahashi T, Minami S, Tsuchiya Y, Tajima K, Sakai N, Suga K, et al. Cytoplasmic control 658 of Rab family small GTPases through BAG6. EMBO Rep. 2019;20(4):e46794. 659 60. Cerchietti LC, Hatzi K, Caldas-Lopes E, Yang SN, Figueroa ME, Morin RD, et al. BCL6 660 repression of EP300 in human diffuse large B cell lymphoma cells provides a basis for rational 661 combinatorial therapy. J Clin Invest. 2010:4569-82. 662

16

61. Chen J, Zang YS, Xiu Q. BAT3 rs1052486 and rs3117582 polymorphisms are associated 663 with lung cancer risk: a meta-analysis. Tumour Biol. 2014;35(10):9855-8. 664 62. Dobin A, Davis CA, Schlesinger F, Drenkow J, Zaleski C, Jha S, et al. STAR: ultrafast 665 universal RNA-seq aligner. Bioinformatics. 2013;29(1):15-21. 666 63. Robinson MD, McCarthy DJ, Smyth GK. edgeR: a Bioconductor package for differential 667 expression analysis of digital gene expression data. Bioinformatics. 2010;26(1):139-40. 668 64. Garrison S, Hojgaard A, Patillo D, Weis JJ, Weis JH. Functional characterization of Pactolus, 669 a beta-integrin-like protein preferentially expressed by neutrophils. J Biol Chem. 670 2001;276(38):35500-11. 671 65. Hunold K, Weber-Steffens D, Runza VL, Jensenius JC, Mannel DN. Functional analysis of 672 mouse ficolin-B and detection in neutrophils. Immunobiology. 2012;217(10):982-5. 673 66. Lober C, Lenz-Stoppler C, Dobbelstein M. Adenovirus E1-transformed cells grow despite the 674 continuous presence of transcriptionally active p53. J Gen Virol. 2002;83(Pt 8):2047-57. 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713

17

Figures714 715

716 717 718 719

18

Figure1CellularstressinducesthereleaseofBAG6-positiveEVs.720 (A) DeterminationofEVproteinconcentration(normalizedtocellularproteinconcentration)and721

particlenumberbyNTA(normalizedtocellnumber)ofHEK293andB-16VEVsisolatedby722 ultracentrifugation(100kxg)ofsupernatantsfromcellseithernon-treatedortreatedwith100nM723 doxoor100nMLBHfor16hours.Bargraphsrepresentmean±SEMofthreeindependent724 experiments.725

(B) QuantificationofEV-associatedBAG6byELISA(normalizedtoEVproteinconcentration)purifiedby726 ultracentrifugationat100kxgfromHEK293andB-16Vcellsthatwereeithernon-treatedortreated727 with100nMdoxoor100nMLBHfor16hours.Bargraphsrepresentmean±SEMofthree(B-16V)and728 two(HEK293)independentexperiments.729

(C) AnalysisofEVsreleasedfromB-16Vcellsunderhypoxicconditions.Left:Conditionedmediumwas730 analyzedbyNTA(normalizedtocellnumbers)fromB-16Vcellsincubatedateithernon-treated(NT)731 orhypoxic(1%O2)conditions.Right:QuantificationofEV-associatedBAG6byELISA(normalizedto732 particles/mlbyNTA)purifiedbyultracentrifugationat100kxgfromconditionedmediumofB-16V733 cellseithernon-treated(NT)orculturedunderhypoxicconditions(1%O2).Datarepresentmean±734 SEMofthreeindependentexperiments(seeFigureS2Cforknockoutdata).735

(D) RepresentativeparticlesizedistributiongraphsbyNTAofB-16VWT-andBAG6KO-EVsandparticle736 sizequantificationofthemeanandmodesize.Thegraphdepictsmean±SEMofthreeindependent737 experiments.738

(E) ImmunoblotanalysisofB-16VcelllysatesandEVsusingtheindicatedantibodies.Ahigherexposure739 timewasrequiredforthedetectionofBAG6inEVs.Eachlanewasloadedwith20µgprotein.Data740 arerepresentativeoftwoindependentexperiments.741 Two-tailed,unpairedStudent’st-tests:*p < 0.05,**p < 0.01,***p < 0.001;SEM,standarderrorof742 themean;EVs,extracellularvesicles;kDa,kilodalton;NT,non-treated;doxo,doxorubicin;LBH,LBH-743 589/Panobinostat;WT,wild-type.744

19

745 746 747 748 749 750 751

Figure 2. mRNA and protein cargo of BAG6-proficient and -deficient EVs define distinct EV-subtypes.

A!

0

5

1 0

1 5

2 0

2 5

Rel

ativ

e ex

pres

sion

NFKBia Eno2

Ctnnal1 HDAC7

Rgs16

Efemp2

Nes

Ptpn3 ACTB

***

*

** * *

**

* * ns

WT BAG6KO B!

Melanosome-like

C

Row Z-Score

BAG6KO WT

-1 0 +1

Subcluster BAG6KO UP Subcluster WT UP Myo7a Ncstn Ppib Pdia3 Pdia4 Pdia6 Slc3a2 Snd1 Sytl2 Trpv2 Tyr Tyrp1 Ywhab Ywhae

Cnp Atp6v0a1 Atp6v1b2 Atp1b3 Rab27a Sec22b Anxa2 Calu Dct Gpnmb Gna13 Lamp1 Mreg Myo5a

Arfgef2 Cd68 Ehd3 Sh3gl1 Snf8 Sec31a Appl2 Bst2 Eea1 AI314180 Flot1 Grb2 Mvb12a Mvb12b

Pacsin2 Sort1 Snx2 Stx7

Chmp2a Chmp2b Chmp4b Chmp5 Tsg101 Vps28 Vps36 Vps37b Vps37c Vps4b Vta1 Vps4a Ifitm3 Lrp1

Exosome-like

!!!!!!!!

Alix CD63 Flot-2 CD81

Anxa7,11 Rab35,11 VAMP2,3,7,8 TSPAN9,10,15

p-va

lue

Log2 fold change (BAG6KO/WT) 0 1 2 3 4 5 6 -1 -2 -3 -4 -5

0.000001

0.01

0.0001

1 0 5

10 15 20 25

20

Figure2mRNAandproteincargoofBAG6-proficientand-deficientEVsdefineEV-subtypes.752 (A) qRT-PCRanalysisoftheindicatedpro-tumorigenicmarkersinEVsderivedfromhypoxia-stressedWT753

orBAG6KOB-16Vcells.EqualamountsofcDNAwereusedandthegraphdepictsrelativeexpression754 ofBAG6KO-EVCtvaluesnormalizedtotheWT-EVCtvalues.Bargraphsrepresentmean±SEMand755 qRT-PCRwasperformedinthreeindependentrunswiththreebiologicalEVpreparationreplicates.756

(B) Volcanoplotshowingthelog2foldchangeversusp-valueofthemassspectrometricanalysis757 comparingproteinsdetectedinEVsisolatedbyultracentrifugationfromthesupernatantofhypoxia-758 stressedB-16VWTandBAG6KOcells,eachmeasuredinthreeindependentbiologicalreplicates.The759 dottedlinesindicatecutofflinesatap-valueof0.01andlog2foldchangeof1.5usedforanalysis760 (significantlyup-anddownregulatedmarkedinpurpleandgreen,respectively).Alistofunchanged761 endocyticproteinmarkersisshown.762

(C) HeatmapshowingthehierarchicalclusteringofWT-andBAG6KO-EVproteomicdata(three763 independentbiologicalreplicateseach,LFQvalue-transformedZscores),usingSpearmanRank764 CorrelationandAverageLinkage.Thetablessummarizetheendocyticproteinsdifferentially765 enrichedinWT(melanosome-like)andBAG6KO(exosome-like)B-16VEVsasanalyzedusingDAVID766 softwareandcorrespondinginteractionnetworksgeneratedbySTRINGareshown.Proteinsinbold767 arederegulatedinEVsbutnotinthecorrespondingcelllysate.FigureS2Gshowsvalidationof768 selectedhitsbywesternblotting.AlistcomparingEVandwholecelllysateproteinlevelsbymass769 spectrometryisprovidedinFigureS2H.770 Two-tailed,unpairedStudent’st-tests:ns,notsignificant,*p < 0.05,**p < 0.01,***p < 0.001;SEM,771 standarderrorofthemean;WT,wild-type.772

773 774 775 776 777 778 779 780 781 782

21

783 784 785 786

787 788 789 790 791 792 793

Figure 3. In vivo activity of WT-EVs and BAG6KO-EVs.

A!

neutrophils 44 %

Total = 27

B!

C!

D!

PBS WT BAG6KO +EVs

0

2 0

4 0

6 0

8 0

Num

ber o

f m

etas

tase

s

PBS WT BAG6KO

* *

0

10

20

30

40

E!

6 0

6 5

7 0

7 5

8 0

8 5

% M

onoc

ytes

(L

y6C

+ Ly

6G –

)

% N

eutr

ophi

ls

(Ly6

C+

Ly6G

+)

+EVs

PBS WT BAG6KO +EVs PBS WT BAG6KO

+EVs

**** ** ns ns **** **

RBC 19 %

Others 37 %

0

5

1 0

1 5

PBS WT BAG6KO +EVs

* * ns

Time (in days) 0 7 14 21 28

I.v. EVs injection (20 µg/mouse) or PBS as control Lung necropsy + RNAseq

EV injection treatment plan!

C57BL/6J (8-12 weeks)

29

Time (in days) 0 7 14 21 28

I.v. EVs injection (10 µg/mouse/day) or PBS as control

Analysis

EV education treatment plan!

C57BL/6J (10 weeks)

35

Single I.v. injection of 1 x 106 B-16V WT cells

15

0 20 40 60 80

60 70 80

65 75 85

0

40

10 20 30

0 5

15 10

% P

atro

lling

mon

ocyt

es

(Ly6

Clo

w L

y6G

–)

22

Figure3BAG6-presentingbutnotBAG6-deficientEVsinhibitmetastasistothelung.794 (A) ExperimentaltreatmentplanofEVinjections(tumorcellswerenotinjected).795 (B) PiechartillustratingtheenrichmentanalysisofRNAsupregulatedinlungsofmicetreatedwith796

BAG6KOB-16VEVsascomparedtotreatmentwithWTB-16VEVsorPBS(treatmentplanprovidedin797 (A)withn=3micepergroup.Theprotein-proteininteractionnetworkofneutrophil-associatedhits798 wasconstructedbySTRING.Onlytwoproteinsdidnotshowadirectfunctionalinteractionwithin799 thisnetwork,i.e.neutrophilintegrinbeta2likeprotein(Itgb2),whichinteractswiththeextracellular800 matrix[64]andficolin-B(FcnB),aneutrophilenrichedlectindetectingpathogenassociated801 molecularpattern(PAMPs)[65].AtablesummarizingallhitsisprovidedinFigureS3A.802

(C) ExperimentaltreatmentplanofEVeducationpriortoi.v.injectionofB-16Vcellstoform803 experimentalmetastases.804

(D) Representativemacroscopicappearanceoflungmetastasesandquantificationoflungmetastases805 formedbyB-16VWTcellsintravenouslyinjectedintothetailveinofWTBJ/6miceeithereducated806 withWT-EVs,BAG6KO-EVsortreatedwithPBSascontrolaccordingtothetreatmentplanprovided807 in(C).n=8mice(PBScontrol),n=8mice(injectedwithWT-EVs),n=10mice(injectedwithBAG6KO-808 EVs).Statisticalanalysiswasperformedusingnon-parametricKruskal-Wallistest(meanranks809 comparedwithWT-EVsgroup)andDunn’smultiplecomparisonstest.Dataareshownasmean±810 SEM.811

(E) Flowcytometricanalysisofisolatedbonemarrowcells,gatedonCD11b+cellsandstainedfor812 neutrophils(Ly6C+Ly6G+)andmonocytes(Ly6C+Ly6G–andLy6ClowLy6G–)fromeducatedmice813 accordingtothetreatmentplanin(C)(analysisofn=8mice(PBScontrol),n=6mice(injectedwith814 WT-EVs),n=7mice(injectedwithBAG6KO-EVs)).Statisticalanalysiswasperformedusingnon-815 parametricKruskal-Wallistest(meanrankscomparedwithWT-EVsgroup)andDunn’smultiple816 comparisonstest.Dataareshownasmean±SEM.Thegatingstrategyofmyeloidcellsisprovidedin817 FigureS3E.818 ns,notsignificant,*p < 0.05,**p < 0.01,***p < 0.001,****p<0.0001;SEM,standarderrorofthe819 mean;EVs,extracellularvesicles;i.v.,intravenouslyinjected;WT,wild-type;kDa,kilodalton;RBC,red820 bloodcells.821 822

23

823 824 825 826 827 828 829 830 831 832 833 834 835 836

Figure 4. BAG6-p53 interplay regulates EV release.

BAG6-IP

:Doxo + - + - + - + - + - + - p53-IP p30

0-IP

ms-IgG

rb-Ig

G lys

ate

BAG6 p53

BAG6 + +

+ + + + + + :BAG6

:p53 BAG6-IP p53-IP

A! B!

p53

E!

0

1

2

3

4

5 mock +p53 WT +p53 K372R/ K373R

NT Doxo

* ns * ns

0

5

1 0

1 5

EVs

rele

ase

(x10

8 pa

rtic

les/

ml)

NT Doxo LBH

*** ***

*** C! D!

WT p53KO

0

2

4

6

8

p53 +/+ p53 -/-

** *

ns

**

NT Doxo

BA

G6

(µg/µl

)

0

10

5

15

0 1 2 3 4 5

EVs

rele

ase

(x10

8 pa

rtic

les/

ml)

0

4 2

6 8

55 130

55

130 kDa kDa

24

Figure4p53iscriticallyinvolvedintheinduciblereleaseofBAG6-presentingEVs.837 (A) ImmunoblotanalysisofBAG6andp53inHEK293celllysatesimmunoprecipitatedwithanti-BAG6,838

anti-p53oranti-p300specificantibodiesaftertreatmentwith1µMdoxofor1horleftuntreated.839 Immunoprecipitationwitheithermouse(ms)orrabbit(rb)IgGisotypecontrolswasperformedas840 controlandcelllysatewasloadedastheinput.Dataarerepresentativeof3threeindependent841 experiments.ThesameexperimentwasconductedusingHEK293BAG6KOcellsasacontrol(Figure842 S4A).HEK293cellsthatwereusedtoanalyzethephysicalinteractionofp53withBAG6and843 CBP/p300areAd-transformedcellsandcontainaccumulatedactivep53(66)whichisnotfurther844 stabilizedinreponsetodoxotreatmentasobservedforB-16V(FigureS4D).845

(B) ImmunoblotanalysisofinvitrotranslatedBAG6andp53mixedasindicatedandincubatedfor1hat846 25°followedbyprecipitationusinganti-BAG6oranti-p53antibodies.Dataarerepresentativeoftwo847 independentexperiments.848

(C) NTAanalysisofEVsderivedfromHCT116WTandp53-deficientcellsthatwereeithernon-treatedor849 treatedwith100nMdoxoorLBHfor16hours.Particlemeasurementswerenormalizedtoprotein850 concentrationofcorrespondingtotalcelllysatesandgraphsrepresentmean±SEMofthree851 independentexperiments.852

(D) NTAanalysisofEVsderivedfromHCT116p53-deficientcellsthatwereeithermocktransfectedorre-853 transfectedwitheitherWTp53oraK373R/K373Racetylation-sitemutantofp53(p53K372R/K373R)854 andnon-treatedortreatedwith100nMdoxofor16h.Immunoblotanalysisofthep53reconstitution855 isprovidedinFigureS4C.856

(E) DetectionofBAG6-associatedEVsthatwereisolatedfromEµ:Tcl1a;Cd19:Cre;Tp53WT(p53+/+)or857 Eµ:Tcl1a;Cd19:Cre;Tp53fl/fl(p53-/-)mousesplenocytes(n=5)andeithernon-treatedortreatedex858 vivofor16hwith100nMdoxorubicinusingaBAG6-specificELISA.Valueswerenormalizedtototal859 proteinamountdeterminedbyBCAassay.860 Two-tailed,unpairedStudent’st-tests:ns,notsignificant,*p < 0.05,**p < 0.01,***p < 0.001;SEM,861 standarderrorofthemean;EVs,extracellularvesicles;NT,non-treated;doxo,doxorubicin;LBH,862 LBH-589/Panobinostat;WT,wild-type;kDa,kilodalton;IP,immunoprecipitation;ms,mouse;rb,863 rabbit;IgG,immunoglobulinG.864

865 866 867 868 869 870 871 872 873 874 875 876 877

25

878 879 880 881 882 883 884 885 886 887 888 889 890

1 20

1

2

3

4

1 20

5

1 0

1 5

Figure 5. Subcellular localization of BAG6 with CBP/p300 is crucial for EV production.

HEK293 B-16V *** ***

*** ** C!

A!

0

1

2

3

1 0

2 0

3 0

p53

K37

3 ac

etyl

atio

n (R

elat

ive

geom

ean)

WT BAG6KO CBP/ p300KO

NT 3.5h 16h

* **

10

2

4

6

8

1 2 30

2

4

6

8

1 0NT LBH

mock +WT BAG6 +nucBAG6 BAG6KO

B!

WT

D!

*

1 2 30

1

2

3

4NT 5 min 10 min

WT BAG6KO CBP/ p300KO

p300

(fo

ld c

hang

e)

+doxo

merge

WT BAG6 nucBAG6

DAPI

BAG6-GFP

PKH26

E!WT

Actin

p300

CBP/ p300KO BAG6KO

0 5 10 0 5 10 0 5 10

p300

mock +WT BAG6 +nucBAG6 WT BAG6KO WT BAG6KO

+Doxo :(min)

0

3 2

10

1

20 30

0

4 2

6 8

10

0

4 2

6

0 1 2 3 4

0 5

10 15

Num

ber o

f par

ticle

s (fo

ld c

hang

e)

p53

K37

3 ac

etyl

atio

n (R

elat

ive

geom

ean)

8

40

kDa

26

Figure5BAG6andCBP/p300arecrucialfortheacetylationofp53andregulatethereleaseofEVs.891 (A) Intracellularflowcytometricanalysisofp53-acetylation(K373)inWT,BAG6KOorCBP/p300dKO892

HEK293cellseithernon-treatedortreatedwith100nMLBHfortheindicatedtime.Bargraphs893 representmean±SEMofthreeindependentexperiments.Aspecificanti-p53K373antibodyand894 DylightTM649secondaryantibodywereused.895

(B) Intracellularflowcytometricanalysisofp53-acetylation(K373)inBAG6KOHEK293cellseithernon-896 treatedortreatedwith100nMLBHfor16huponmock-transfectionortransfectionwithfull-length897 BAG6(+WTBAG6)orwithaN-terminaldeletedBAG6mutant(+nucBAG6).Bargraphsrepresent898 mean±SEMofthreeindependentexperiments.Aspecificanti-p53K373antibodyandDylightTM649899 secondaryantibodywasused.900

(C) NTAanalysisof24hEVreleasefromWTorBAG6KOHEK293andB-16Vcellsthatwereeithermock-901 transfected,transfectedwithfull-lengthBAG6(+WTBAG6)orwithanN-terminaldeletedBAG6902 mutant(+nucBAG6).Animmunoblotofthemyc-tagtransfectedBAG6isshownFigureS5A.A903 cartoonillustratingtheimpactofBAG6expressionandsubcellularlocalizationontheEVreleaseis904 showninFigureS5D.905

(D) ImmunofluorescencemicroscopicanalysisofHEK293cellstransfectedwitheitherfulllengthBAG6906 (+WTBAG6)orN-terminaldeletedBAG6mutant(+nucBAG6)andtreatedwith1µMdoxofor1h.907 TransfectedBAG6wasvisualizedbyaGFP-tag,thenucleusandcellmembranewerevisualizedby908 stainingwithDAPIandPKH,respectively,andmergedimagesareshown.Scalebar:2µm.The909 analoguousexperimentusingnon-treatedHEK293cellsisshowninFigureS5B.910

(E) ImmunoblotanalysisofCBP/p300incytoplasmicfractionsextractedfromWT,BAG6KOand911 CBP/p300dKO-HEK293cellsthatwereeithernon-treatedortreatedwith1µMDoxofortheindicated912 time.Thebargraphrepresentsquantificationoftwoindependentimmunoblotsfrombiological913 replicatesnormalizedtoactinastheloadingcontrol.914 Two-tailed,unpairedStudent’st-tests:*p < 0.05,**p < 0.01,***p < 0.001;SEM,standarderrorof915 themean;NT,non-treated;WT,wild-type;EV,extracellularvesicles;doxo,doxorubicin;kDa,916 kilodalton;AC,acetylation;DAPI,4ʹ,6-Diamidin-2-phenylindol,GFP,greenfluorescentprotein.917

918

27

919 920 921 922 923 924 925 926 927 928 929 930

A!

B!

pGBT9 (Prey)

pGADT7 (Bait)

DDO QDO

BAG6 TSG101 ! ! BAG6del TSG101 ! –

BAG6 empty ! –

TSG101 empty ! – BAG6 NKp30 ! !

0 10 30 60 lysate 0 10 30 60 α-BAG6

:Doxo (min) BAG6

HRS TSG101

ALIX Actin

Yeast two hybrid interaction

Figure 6. Doxo treatment induces complex formation between BAG6 and ESCRT proteins.

130 100

55

100

40

kDa

28

Figure6DoxotreatmentinducescomplexformationbetweenBAG6andESCRTproteins.931 (A) ImmunoblotanalysisofcelllysatesandBAG6immunoprecipitationsofHEK293cellsthatwereeither932

non-treatedortreatedwith1µMdoxorubicinfortheindicatedtimeprobedwithspecificantibodies933 againstBAG6,theESCRTproteinsHRS,TSG101andALIXandactinasaloadingcontrol.Theblot934 representsoneoutoftwoindependentexperiments.935

(B) AnalysisoftheinteractionofBAG6andTSG101usingtheYeasttwohybridmethod.BAG6,BAG6936 withoutTSG101bindingmotif(BAG6delwithdeletedPSAP/PTAPmotifsataminoacidpositions471-937 475and496-499,seeFigureS6A)andTSG101weresubclonedinpGADT7andpGBT9expression938 vectorstransformedintoyeastlinesY2HGoldandY187orcontrolvectorsandmated.NKp30and939 BAG6wereusedasapositivecontrolforinteractionanalysisandexperimentswereperformedin940 twoindependentruns.941 doxo,doxorubicin;kDa,kilodalton;DDO,MinimalMediaDoubleDropouts;QDO,MinimalMedia942 QuadrupleDropouts(QDO)943

944 945 946 947 948

29

949 950 951 952 953 954 955 956 957 958 959 960 961 962

0

5 0

1 0 0

1 5 0

2 0 0

2 5 0

Stage II Stage III/IV

BA

G6

RN

A ex

pres

sion

(F

PKM

) *

Figure 7. Relevance of BAG6 in melanoma patients.

A! B!

C!

0

2

4

6

WT BAG6KO

+BAG6R786* mock

* **

150 200 250

100 50

0 0

2

4

6

Stage I Stage IV

***

Ctn

nal1

(F

old

chan

ge)

0

6 4 2

0

2

4

6

p53

K37

3 (r

el. g

eom

ean)

Num

ber o

f par

ticle

s (fo

ld c

hang

e)

0

1

2

3

HEK293 cells HEK293 WT cells

+BAG6R786* mock +BAG6R786*

* D!

30

Figure7RelevanceofBAG6-dependentEVformationinhumanmelanoma.963 (A) TCGAdataanalysiscomparingBAG6RNAtissueexpressionlevelofstageII(n=71)versusstageIII/IV964

(n=29)melanomapatients.*indicatesp=0.0462ofunpairedt-testwithWelch’scorrection.965 (B) qRT-PCRanalysisofCtnnal1inEVsisolatedfromtheplasmaofmelanomapatientscomparingstageI966

andstageIVdisease.ThegraphdepictsthecomparisonofCtnnal1expressionlevelsasfoldchange967 betweenstageI(n=10)andstageIV(n=10)patients.*indicatesp=0.0002ofunpairedt-testwith968 Welch’scorrection.AcontrolexperimentusingcDNAmadefromisolatedpatientplasmaEV-RNAis969 showninFigureS7A.970

(C) Intracellularflowcytometricanalysisofp53-acetylation(K373)depictedforWTorBAG6KO-HEK293971 cellseithermock-transfectedortransfectedwiththeBAG6R786*mutantand24hlatertreatedwith972 100nMLBHfor16h.Bargraphsrepresentmean±SEMofthreeindependentexperiments.Statistical973 analysiswasperformedusingtwo-tailed,unpairedStudent’st-tests.Aschematicandanimmunoblot974 showingtheexpressionoftheBAG6R786*mutantproteinisshowninFigureS7BandS7C,975 respectively.976

(D) NTAanalysisof16hEVreleasefromWTHEK293cellsthatwereeithermock-transfectedor977 transfectedwiththeBAG6R786*mutant24hpriortothetreatmentwith100nMLBHduringEV978 collection.Bargraphsrepresentmean±SEMoffourindependentexperiments.Statisticalanalysis979 wasperformedusingtwo-tailed,unpairedStudent’st-tests.980 *p < 0.05,**p < 0.01,***p < 0.001;SEM,standarderrorofthemean;WT,wild-type;BAG6R786,981 BAG6mutantwithstopcodonatArgininposition786.982

983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014