文献1 (3).pdf

OPENORIGINAL ARTICLEDBC2/RhoBTB2 functions as a tumor suppressor protein viaMusashi-2 ubiquitination in breast cancerYM Choi1,2,KB Kim2,JH Lee3,YK Chun4,IS An2,S An1and S Bae1The gene encoding deleted in breast cancer 2(DBC2),also referred to as RHOBTB2(Rho-related BTB domain-containing protein 2),is classified as a tumor suppressor gene.DBC2 is a substrate-specific adaptor protein for a novel class of Cullin-3(CUL3)-based E3ubiquitin ligases;however,it is unclear if the substrate adaptor function of DBC2 is required for its tumor suppressor activity.Furthermore,the key substrates of DBC2-mediated ubiquitination have yet to be identified.In the present study,we established agenome-wide human cDNA library-based in vitro ubiquitination target screening assay and identified Musashi-2(MSI2)as a novelubiquitination target protein of DBC2.MSI2 directly interacted with DBC2,and this interaction promoted MSI2 polyubiquitinationand proteasomal degradation in breast cancer cells.Overexpression and knockdown experiments demonstrated that DBC2suppressed MSI2-associated oncogenic functions and induced apoptosis.Immunohistochemistry analysis of a breast cancer tissuemicroarray revealed that DBC2 and MSI2 protein levels are inversely correlated in both normal breast tissues and breast cancertissues.Taken together,these findings provide evidence that DBC2 suppresses tumorigenesis in breast cancer byubiquitinating MSI2.Oncogene advance online publication,12 December 2016;doi:10.1038/onc.2016.441INTRODUCTIONUbiquitin-mediated post-translational modifications are essentialfor nearly all biological processes,including cell growth,differ-entiation and apoptosis.1Dysregulation of this phenomenon leadsto irreversible changes in protein stability and can directly orindirectly promote a variety of pathological conditions,includingcancer.1The process of ubiquitination involves multiple stepsmediated by E1 ubiquitin-activating enzymes,E2 ubiquitin-conjugating enzymes and substrate-specific E3 ubiquitin ligases.2The most predominant class of E3 ligases is the family of reallyinteresting new gene(RING)-finger domain-containing proteins.These proteins are subdivided into two groups:the monomericRING-type E3 ligases and the multimeric RING-type E3 ligases,such as the Cullin-3(CUL3)-based E3 ligases.2The substratespecificity of CUL E3 ligases is primarily mediated by distinctadaptor proteins that contain an F-box,SOCS-box,or a broadcomplex,tramtrack and bric-a-brac(BTB)domain.2,3BTB domain-containing proteins comprise a new class ofsubstrate-specific adaptors of the CUL3-based E3 ubiquitin ligasecomplex.3,4Previous studies have demonstrated that BTB protein-dependent CUL3-based E3 ubiquitin ligases are present in bothmammals and non-mammalian species,including Drosophilamelanogaster,Schizosaccharomyces pombe,Caenorhabditis elegansand Arabidopsis thaliana.59BTBCUL3 ubiquitin ligase complexeshave essential roles in numerous biological processes,includingembryonic development,apoptosis,cytokinesis,cell movement,Hedgehog signaling and responses to oxidative stress.914Deleted in breast cancer 2(DBC2),also referred to as Rho-related BTB domain-containing protein 2(RhoBTB2),is a putativetumor suppressor protein that forms a complex with CUL315andhas a direct role in tumorigenesis.1618A downregulation in DBC2expression resulting from homozygous deletion or promotermethylation has been observed in breast cancer,16,19,20and aprevious study has reported that DBC2 expression is lost in 60%ofcases of breast cancers.21The antitumorigenic effects of DBC2 aremediated by the inhibition of cancer cell growth,proliferation,migration and invasion.16,22Furthermore,reduced DBC2 expres-sion is associated with distinct clinicopathological features ofbreast cancer,including human epidermal growth factor receptor2(HER2)status and p53 mutations.19In addition,downregulatedDBC2 expression has been observed in lung,gastric,bone andbladder carcinomas.2326These findings indicate that DBC2functions as both a tumor suppressor and a putative substrate-recruiting adaptor protein of the CUL3-based ubiquitin ligasecomplex;however,the relationship between these two distinctfunctions remains unclear.A better understanding of the multi-functional nature of DBC2 and identification of DBC2 ubiquitina-tion substrates might inform the development of promisinganticancer therapies.RESULTSIdentification of MSI2 as a novel substrate for DBC2-dependent E3ubiquitin ligasesIn contrast to monomeric RING-finger E3 ubiquitin ligases,CUL3-based multimeric E3 ubiquitin ligases require BTB domain-containing proteins to provide substrate specificity.6AlthoughDBC2 has been shown to interact with CUL3,15the substratestargeted by DBC2/CUL3-E3 ubiquitin ligase activity,including thetarget protein that potentially mediates the tumor suppressor1KU Center for Integrated Science and Technology,Konkuk University,Seoul,South Korea;2Korea Institute of Dermatological Sciences,2nd Enterprise Research Building,Chungcheongbuk-do,South Korea;3Laboratory of Molecular Oncology,Cheil General Hospital and Womens Healthcare Center,Dankook University,College of Medicine,Seoul,South Korea and4Department of Pathology,Cheil General Hospital and Womens Healthcare Center,Dankook University,College of Medicine,Seoul,South Korea.Correspondence:Dr S Bae,KU Center for Integrated Science and Technology,Konkuk University,120 Neungdong-ro,Gwangjin-gu,Seoul 05029,South Korea.E-mail:sbaekonkuk.ac.krReceived 31 March 2016;revised 10 September 2016;accepted 20 October 2016Oncogene(2016), of DBC2,have yet to be identified.Therefore,we soughtto identify candidate substrates of DBC2-dependent E3 ubiquitinligase complexes.To this end,we adapted our previously reportedin vitro genome-wide screening system of a human cDNAlibrary.2729To isolate selectively E3 ligase-specific substrates,weincorporated recombinant E1 proteins,E2 proteins,GST-DBC2,aCUL3-ROC1 protein complex and His-ubiquitin into this system(Supplementary Figure 1a and Supplementary Materials andMethods).The in vitro ubiquitination assay was conducted asdescribed previously,15,30and the ubiquitin E3 ligase activity of theDBC2-CUL3 ligase complex was confirmed using western blot withan antibody against ubiquitin(data not shown).Using thisapproach,we isolated several highly ubiquitinated cDNA clones,one of which represented a novel ubiquitination target protein(SupplementaryFigure1b).Theclonesofinterestweresequenced,and we conducted an extensive literature review ofthe relevant genes.The cDNA clone encoding Musashi-2(MSI2),aprotein with two RNA recognition motif domains,was selected forfurther analysis as its role in the oncogenesis of multiple cancerswas the most well-studied compared with the other putativetarget proteins.3134We evaluated the potential role of DBC2 in MSI2 ubiquitinationusing in vitro binding assays with recombinant GST-MSI2 and 35Smethionine-labeled DBC2.We observed a direct interactionbetween DBC2 and MSI2(Figure 1a)but not between CUL3 andMSI2(Supplementary Figure S2),suggesting that MSI2 ubiquitina-tion is dependent on its interaction with DBC2.These findingswere confirmed using co-immunoprecipitation and western blotassays with MDA-MB-231 breast cancer cells co-transfected withplasmids expressing DBC2 and MSI2(Figure 1b).Immunofluores-cence staining and confocal microscopy analysis revealed thatDBC2 colocalized with MSI2 in the cytosol(Figure 1c).In addition,endogenous DBC2 and endogenous MSI2 co-precipitated withone another(Figure 1d).DBC2 contains a GTPase domain and two BTB domains in its N-and C-terminal regions,respectively.15A previous study reportedthat the Y284D mutation in the first BTB domain of DBC2 disruptsits interaction with CUL3.15Co-immunoprecipitation and westernblot assays revealed that both the C-terminal region of DBC2(DBC2-C)and the DBC2 mutant protein with the Y284D mutation(DBC2-MT)interacted with full-length MSI2(Figures 1e and f),indicating that the Y284 residue of DBC2 is required for itsinteraction with CUL3 but not for its interaction with itsubiquitination substrate MSI2.Taken together,these resultsdemonstrate that DBC2 and MSI2 directly interact.DBC2 promotes polyubiquitination-mediated proteasomaldegradation of MSI2Next,we evaluated the potential role of DBC2 in MSI2 ubiquitina-tion.MSI2wasubiquitinatedinDBC2-overexpressingcells(Figure 2a,lane 2 versus lane 4),and this effect was enhancedin cells co-transfected with CUL3 and ROC1(Figure 2a,lane 4versus lane 8).In contrast,MSI2 was ubiquitinated at relativelylower levels in cells co-transfected with plasmids expressing CUL3and ROC1 only(Figure 2a,lane 2 versus lane 6).This result waslikelymediatedbyendogenousDBC2(Figure2aandSupplementary Figure S3),as MSI2 did not interact with CUL3in vitro(Supplementary Figure S2).DBC2 also promoted K48-linked polyubiquitination of MSI2(Supplementary Figure S4).AsK48-linked ubiquitination is associated with proteasomal degrada-tion of the target protein,35we investigated if DBC2 down-regulated MSI2 protein levels via proteasomal degradation.Wefound that DBC2 decreased endogenous MSI2 protein levels in aconcentration-dependent manner(Figure 2b),whereas DBC2 didnot influence MSI2 mRNA levels(Supplementary Figure S5).Inaddition,ROC1andCUL3overexpressionenhancedDBC2-mediated MSI2 degradation(Figure 2c,lane 2 versus lane 4).Incontrast,DBC2-MT did not significantly affect endogenous MSI2levels(Figures2bandc).Furthermore,MG132treatmentabolished the effect of DBC2 on MSI2 degradation(Figure 2d).Cellular ubiquitination assays demonstrated that polyubiquitina-tion of endogenous MSI2 increased in DBC2-overexpressing cellscompared with control cells,and CUL3-binding-deficient DBC2-MTdid not promote endogenous MSI2 ubiquitination(Figure 2e).Moreover,the half-life of MSI2 decreased in cells transientlyoverexpressing DBC2 compared with control cells treated with theprotein biosynthesis inhibitor cycloheximide(Figure 2f).Takentogether,these results indicate that MSI2 is ubiquitinated byDBC2/CUL3 E3 ligase complexes and that this modification targetsMSI2 for proteasomal degradation.A previous study demonstrated that heat-shock protein 90disrupts the intramolecular interaction between the GTPase andBTB domains of DBC2,thereby promoting the formation of theDBC2/CUL3 E3 ligase complex.36,37Consistent with these findings,MSI2 degradation was suppressed in MDA-MB-231 cells treatedwith the heat-shock protein 90-specific inhibitor geldanamycin(Supplementary Figure S6a).In addition,MSI2 inhibited theintramolecular interaction between the GTPase and BTB domainsof DBC2(Supplementary Figure S6b),indicating that the interac-tion between MSI2 and the BTB domains of DBC2 might relive theautoinhibitory conformation of DBC2.A previous study demon-strated that DBC2 is ubiquitinated by the CUL3 E3 ligase15and thiseffect was also observed in the present study.Interestingly,DBC2ubiquitination and degradation was markedly suppressed in MSI2-overexpressing cells(Supplementary Figures S6d and S6e).Inaddition,DBC2 protein levels,but not DBC2 mRNA levels,increasedinMSI2-overexpressingcells(SupplementaryFigure S6c),indicating that DBC2 primarily functions as asubstrate-specific adaptor for the CUL3 ligase complex when theinteraction between DBC2 and MSI2 is enhanced.DBC2 knockdown enhances MSI2 protein stabilityThe results of the ubiquitination assays prompted us to furtherevaluate the role of endogenous DBC2 in the regulation of MSI2protein levels.We infected MCF7 and MDA-MB-231 cells withlentiviral vectors expressing two different short hairpin RNAtargeting DBC2(shDBC2).As DBC2 mRNA levels were lower inshDBC2-expressing MCF7 cells compared with that in MDA-MB-231 cells(data not shown),we generated two independentstable shDBC2-expressing cell lines for further analysis.MSI2protein levels significantly increased in DBC2-depleted cells,whereas MSI2 mRNA levels were unaffected(Figures 3a and b).These results suggested that endogenous DBC2 specificallytargets the MSI2 protein for degradation.Consistent with thesefindings,a cellular ubiquitination assay demonstrated that MSI2ubiquitination was abolished and the half-life of MSI2 increased inDBC2-depleted cells(Figures 3c and d).Taken together,these datasuggest that MSI2 ubiquitination in breast cancer cells is primarilymediated by DBC2.DBC2 induces cell cycle arrest and apoptosis through MSI2degradationRecent studies reported that the oncogenic functions of MSI2 aremediated by p21,BAX,c-Myc and BCL-2.32,38,39Therefore,weinvestigated if DBC2 negatively regulates these proteins in breastcancer cells.DBC2 inhibited the MSI2-mediated downregulation ofp21 but did not influence p21 mRNA levels(Figure 4a andSupplementary Figure S7).In addition,DBC2,but not DBC2-MT,inhibited cell growth and proliferation in both control and MSI2-overexpressing cells(Po0.05;Figures 4b and c).Furthermore,theproportion of cells in the sub-G1 and G0/G1 phases of the cellcycle increased in cells transiently expressing DBC2 and in shMSI2-expressing cells(Figure 4d).Interestingly after 48 h post-transfection,only the proportion of cells in the sub-G1 phaseDBC2 leads to MSI2 ubiquitinationYM Choi et al2Oncogene(2016)111increased in cells expressing DBC2 and shMSI2(SupplementaryFigure S8),suggesting that the inhibition of cell proliferation wasdue to DBC2-mediated cell death.To evaluate this hypothesis,apoptotic cells were labeled with Annexin V-fluorescein isothio-cyanate and propidium iodide and analyzed using flow cytometry.The proportion of apoptotic cells increased by 11%in DBC2-overexpressing cells compared with control cells(Figure 4e),and asimilar result was observed in cells expressing shMSI2(Figure 4e).Previous studies demonstrated that MSI2 knockdown enhancedapoptosis,affected extracellular signal-regulated kinase(ERK)andp38mitogen-activatedproteinkinasephosphorylation,andaltered the expression of the proapoptotic protein BAX and theantiapoptotic protein BCL-2.39To investigate the relationshipbetween DBC2-mediated apoptosis and MSI2-associated signal-ing,we evaluated these effects in DBC2-overexpressing cells andshMSI2-knockdown cells using western blot assays.The levels ofphosphorylated p38,mitogen-activated protein kinase,phos-phorylatedextracellularsignal-regulatedkinaseandBCL-2decreased in DBC2-overexpressing cells,and p21 and BAX levelsincreased.Similar results were observed in shMSI2-knockdowncells(Figure 4f),indicating that the effects observed in DBC2-overexpressing cells were not off-target effects.Taken together,these findings suggest that DBC2-induced cell cycle arrest andapoptosis are mediated by DBC2-induced MSI2 degradation.Tumor-suppressive function of DBC2 is MSI2-dependentTo evaluate the hypothesis that MSI2 degradation is required forthe tumor suppressor activity of DBC in breast cancer,weevaluated clonogenic growth and cell migration.Clonogenicgrowth was markedly suppressed in DBC2-overexpressing MDA-MB-231 cells but not in DBC2-MT-overexpressin