Microphthalmia-Associated Transcription Factor Controls the DNA Damage Response and a Lineage-Specific Senescence Program in Melanomas

Published OnlineFirst April 13, 2010; DOI: 10.1158/0008-5472.CAN-09-2913

Tumor and Stem Cell Biology Cancer Research Microphthalmia-Associated Transcription Factor Controls the DNA Damage Response and a Lineage-Specific Senescence Program in Melanomas

Sandy Giuliano1,2, Yann Cheli1,2, Mickaël Ohanna1,2, Caroline Bonet1,2, Laurent Beuret1,2, Karine Bille1,2, Agnès Loubat2,4, Véronique Hofman2,5, Paul Hofman2,5, Gilles Ponzio2,4, Philippe Bahadoran3, Robert Ballotti1,2, and Corine Bertolotto1,2

Abstract Apoptosis and senescence are cellular failsafe programs that counteract excessive mitogenic signaling observed in cancer cells. Melanoma is known for its notorious resistance to apoptotic processes; therefore, senescence, which remains poorly understood in melanomas, can be viewed as a therapeutic alternative. Microphthalmia-associated transcription factor (MITF), in which its M transcript is specifically expressed in melanocyte cells, plays a critical role in melanoma proliferation, and its specific inhibition is associated

with G0-G1 growth arrest. Interestingly, decreased MITF expression has been described in senescent melanocytes, and we have observed an inhibition of MITF expression in melanoma cells exposed to chemotherapeutic drugs that induce their senescence. All these observations thereby question the role of MITF in controlling senescence in melanoma cells. Here, we report that long-term depletion of MITF in melanoma cells triggers a senescence program characterized by typical morphologic and biochemical changes associated with a sustained growth arrest. Further, we show that MITF-silenced cells engage a DNA damage response (DDR) signaling pathway, leading to p53 upregulation, which is critically required for senescence entry. This study uncovers the existence of a lineage-restricted DDR/p53 signaling pathway that is inhibited by MITF to prevent senescence and favor melanoma cell proliferation. Cancer Res; 70(9); 3813–22. ©2010 AACR.

Introduction advanced melanoma tumors, regulates MITF expression and activity (2). Further, BRAF cooperates with MITF to The M isoform of microphthalmia-associated transcrip- transform immortalized melanocytes (2). In line with this, tion factor (MITF), which is specifically expressed in the me- MITF expression has been associated with resistance to lanocyte lineage, is critical for melanocyte homeostasis and chemotherapy and correlates with unfavorable prognosis melanoma development. Several proteins that contribute to (2, 3). These observations suggest that MITF may function cell migration/invasion (DIA1 and c-MET), survival (BCL2, as a lineage-specific “addictive oncogene” in melanoma cells. HIF1a, c-MET,andML-IAP), and proliferation [cyclin-dependent Melanoma cells exposed to chemotherapeutic drugs reveal kinase 2 (CDK2)] have been identified as MITF target genes a reduction in MITF expression, suggesting that MITF brings (1). Further, a genetic amplification of MITF has been found important melanoma proliferative and survival advantages in 10% to 20% of melanomas and has been associated with (4). Consistently, inhibition of MITF expression induces a a decreased 5-year survival (2). Additionally, BRAF, whose V600E G -G growth arrest of melanoma cells (5–7). MITF has been gene exhibits oncogenic mutation (BRAF ) in >60% of 0 1 also involved in the control of melanoma survival through the control of BCL2 or BIRC7 (8, 9), but MITF silencing only Authors' Affiliations: 1Institut National de la Sante et de la Recherche marginally causes apoptosis (8, 10). In addition to apoptosis, Medicale U895, Team 1, Biology and pathology of melanocytes: from cutaneous pigmentation to melanoma; 2University of Nice Sophia- senescence is another cellular failsafe program that counter- Antipolis, UFR médecine; 3CHU NICE, Department of Dermatology; acts excessive mitogenic signaling observed in cancer cells. 4Institut National de la Sante et de la Recherche Medicale U634; The role of MITF in the control of the senescence program 5ERI21/EA 4319 and Human Biobank, Nice, France has never been studied thus far. However, a decreased MITF Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/). expression has been observed in conditions that promote senescence of melanocytes (11). These observations prompted M. Ohanna and Y. Cheli contributed equally to this work. us to investigate whether MITF might control senescence of Corresponding Author: Corine Bertolotto, Institut National de la Sante et de la Recherche Medicale U895, 151 Route Saint Antoine de Ginestière, melanocytic cells. Nice 06204, France. Phone: 33-4-93-37-76-99; Fax: 33-4-93-37-76-98; Here,wereportthatlong-termsuppressionofMITFin E-mail: [email protected]. melanoma cells is associated with typical hallmarks of senes- doi: 10.1158/0008-5472.CAN-09-2913 cence, including enlarged and flattened cell morphologies, in- ©2010 American Association for Cancer Research. crease in cell granularity, acidic β-galactosidase staining,

www.aacrjournals.org 3813

Giuliano et al.

change in chromatin structure associated with formation of was used to histochemically detect β-galactosidase activity heterochromatin foci, and cessation of proliferation. We at pH 6. Senescence-associated β-galactosidase (SA-β-Gal) show that the senescence program triggered by MITF silenc- activity results in a cytoplasmic blue staining that can be ing operates via a DNA damage response (DDR), leading to visualized by light microscopy. The percentage of means ataxia-telangiectasia mutated (ATM), H2AX, p53-binding and SDs were derived from counting 100 cells in duplicate protein 1 (53BP1), and CHK2 activation and ultimately re- plates after 96 hours. sulting in p53 phosphorylation and stabilization. Finally, Immunofluorescence and confocal microscopy. Immu- our findings show that the DDR machinery is critically re- nofluorescence experiments were carried out as previously quired, as pharmacologic inhibition and genetic ablation of described (14) and examined with a Zeiss Axiophot micro- ATM, CHK2, or p53 efficiently bypass the senescence pro- scope equipped with epifluorescence illumination or under gram mediated by MITF depletion. oil immersion with a Leica TCS SP2 confocal microscope. Therefore, MITF is critically required to prevent senes- Representative experiments are shown. cence and favor melanoma cell proliferation. Our results, Western blot assays. Western blots were carried out as which uncover the existence of a melanocyte-specific se- previously described (15). Briefly, cell lysates (30 μg) were nescence program, make MITF and p53 potential therapeu- separated by SDS-PAGE, transferred onto a polyvinylidene di- tic targets and open new perspectives in the treatment of fluoride membrane, and then exposed to the appropriate melanoma. antibodies. Proteins were visualized with the enhanced chemiluminescence system. The Western blots shown are Materials and Methods representative of at least three independent experiments. For measurement of secreted PAI-1, the culture medium Cell cultures. Human 501 mel, WM9, and Skmel28 mela- was replaced by serum-free medium for 48 hours and cells noma cells and mouse B16 melanoma cells were grown in were next precipitated with acetone. DMEM supplemented with 7% FCS and 100 units/mL peni- Chromatin immunoprecipitation assay. Chromatin cillin/50 μg/mL streptomycin at 37°C in a humidified atmo- immunoprecipitation (ChIP) assay was performed using sphere containing 5% CO2. anti-p53 antibody and the ChIP assay kit (Upstate). Anti- Transient transfection of small interfering RNA. Briefly, mouse IgG was used as controls. The primers for PAI-1 a single pulse of 50 nmol/L of small interfering RNA (siRNA) were 5′-GACACAGGCAGAGGG-3′ (forward) and 5′- was administrated to the cells at 50% confluency by transfec- GTGGGCCACTGCCTCC-3′ (reverse), with size for PCR pro- tion with 5 μL Lipofectamine RNAiMAX in Opti-MEM for ducts of 272 bp. The primers for glyceraldehyde-3-phosphate the time indicated in the figure legends. Control and MITF dehydrogenase (GAPDH) were 5′-TACTAGCGGTTT- siRNAs were previously described (12). A second MITF siRNA TACGGGCG-3′ (forward) and 5′-TCGAACAGGAGGAGCAGA- sequence (siMi2) described elsewhere was also used to rule GAGCGA-3′ (reverse), with size for PCR products of 160 bp. out the risk of off-target effects (5). p53 siRNA was from Statistical analysis. Data are presented as the average ± Santa Cruz Biotechnology. SDandwereanalyzedbyStudent'st test using Microsoft Growth curves. Cells (20 × 103) were seeded in six-well Excel software. P ≤ 0.05 was considered significant. dishes and transiently transfected the following day with control or MITF siRNA. Cells were next detached from days Results 2 to 6 and manually counted in triplicate with a hemocytom- eter to assess cell proliferation. The experiment was per- MITF suppression triggers senescence of melanoma formed three times. cells. We observed that hydroxyurea, a chemotherapeutic Luciferase reporter assays. 501 mel melanoma cells drug, blocked human 501 mel melanoma cell growth and were transiently transfected as previously described using led to morphologic changes that were associated with a stim- the Lipofectamine reagent (Invitrogen; ref. 13). Briefly, cells ulation of the SA-β-Gal activity, a widely used biomarker of were transiently transfected with 0.3 μg PG13 (provided by cellular senescence (Fig. 1A). Noteworthy, hydroxyurea M. Oren, Weizmann Institute of Science, Rehovot, Israel) caused a dramatic reduction in the level of MITF without ap- or plasminogen activator inhibitor-1 (PAI-1) promoter plas- optosis induction as observed by the absence of sub-G1 cells mids (provided by D. Rifkin, Department of Pharmacology and caspase-3 activation (data not shown). Although hy- and Systems Therapeutics, Mount Sinai School of Medi- droxyurea is able to induce senescence independently of cine, New York, NY) and 0.05 μgpCMVβGal to control MITF by interfering with DNA synthesis, the correlation in the variability in transfection efficiency. The transfection melanoma cells between the reduction in MITF level and medium was changed 6 hours later, and when siRNA co- the senescence phenotype prompted us to hypothesize that transfection was performed, cells were transfected with MITF may play an active role in controlling senescence in 50 nmol/L of control or MITF siRNA as described above. melanocytic cells. Cells were assayed for luciferase and β-galactosidase activ- Indeed, we observed that sustained MITF silencing did not ities 48 hours later. Transfections were repeated at least promote apoptosis, as shown by the absence of sub-G1 cells by three times. flow cytometry and the lack of caspase-3 activation by West- Senescence-associated β-galactosidase assay. The senes- ern blot contrary to tumor necrosis factor–related apoptosis- cence β-galactosidase staining kit (Cell Signaling Technology) inducing ligand, which caused apoptosis of melanoma cells

3814 Cancer Res; 70(9) May 1, 2010 Cancer Research

MITF Controls a DDR/p53-Dependent Senescence Program

Figure 1. MITF depletion triggers a senescence program in melanoma cells. A, 501 mel cells were exposed to DMSO (Ct) or 150 μmol/L hydroxyurea (HU) for 96 h and then harvested for flow cytometric analysis (left) or stained for SA-β-Gal activity. Cells were visualized under phase-contrast microscopy. Middle, percentage of means and SDs of SA-β-Gal–positive cells; right, cell lysates from 501 mel cells exposed to DMSO or hydroxyurea (50 or 150 μmol/L) for 96 h were assessed by Western blot for the antibodies indicated. B, left, 501 mel cells were transfected with control (siC) or MITF (siMi) siRNA, stained for SA-β-Gal activity, and observed by phase-contrast microscopy. Percentage of means and SDs of SA-β-Gal–positive cells are indicated. Right, immunofluorescence experiments to HP1β- of MITF-depleted cells versus control cells were analyzed by confocal microscopy. Cell nuclei were counterstained with DAPI. C, cells were counted in triplicates from days 2 to 6. Lysates were analyzed by Western blotting with MITF and extracellular signal-regulated kinase 2 (ERK2) antibodies. D, human WM9 and mouse B16 melanoma cells were transfected with control or MITF siRNA and stained for SA-β-Gal activity. Numbers on the panels indicate percent of means and SDs of SA-β-Gal–positive cells.

www.aacrjournals.org Cancer Res; 70(9) May 1, 2010 3815

Giuliano et al.

Figure 2. MITF silencing engages the DDR signaling pathway. A, images of control siRNA–transfected or MITF siRNA–transfected 501 mel cells showing DAPI labeling. DAPI staining is depicted in green for better contrast. High-magnification images display lagging chromosome. B, control siRNA–transfected or MITF siRNA–transfected 501 mel cells were analyzed by immunofluorescence or by Western blot for H2AX phosphorylated on Ser139 (γH2AX). C, typical fluorescence image of control siRNA–transfected or MITF siRNA–transfected 501 mel cells labeled with an antibody to 53BP1. Cell nuclei were counterstained with DAPI.

(Supplementary Fig. S1; ref. 16). In agreement with our hy- The senescence phenotype caused by MITF depletion was pothesis, MITF silencing induced dramatic morphologic also observed in other human melanoma cells such as WM9 changes associated with a G0-G1 growth arrest and the ac- and in mouse B16 cells as well as in melanoma cells freshly cumulation of cells positive for SA-β-Gal staining (∼50% isolated from tumor specimens (Fig. 1D; Supplementary after 3 days and >70% after 4 days; Fig. 1B). Noteworthy, Fig. S4). On the other hand, MITF siRNA did not promote SA-β-Gal staining was observed very occasionally in 501 SA-β-Gal staining in nonmelanoma cells (data not shown). mel cells transfected with control siRNA (2% of cells). Our findings therefore show that MITF silencing triggers a Quantification of the morphologic changes by flow cytome- lineage-restricted growth arrest and senescence program of try showed a 30% to 50% enlargement of the cell size and melanoma cells. up to a 60% increase in cell granularity, two parameters of MITF depletion promotes mitotic defects and causes cellular senescence (Supplementary Fig. S2). Senescence is DNA damages. Data from the literature indicate that senes- also associated with changes in chromatin structure, and cence operates in a signal- and context-dependent manner, we indeed found senescence-associated heterochromatin and therefore, it remained to elucidate which pathways were foci (SAHF) using heterochromatin protein 1β (HP1β) la- engaged during the process of MITF-induced senescence in beling of melanoma cells depleted for MITF (Fig. 1B). Con- melanoma cells. We found that MITF silencing promoted an sistently, MITF depletion triggered long-term growth arrest increase in p16INK4A and p27KIP1 and a decrease in CDK2, but of melanoma cells (Fig. 1C). To rule out the possibility that loss-of-function experiments revealed that none of them the effect of MITF silencing resulted from nonspecific cel- were key mediators of the senescence program mediated lular responses, we used a second siRNA duplex containing by MITF invalidation (Supplementary Fig. S5). a different MITF targeting sequence, which showed a com- On the other hand, careful examination of DNA using 4′,6- parable efficiency in inhibiting MITF and in triggering se- diamidino-2-phenylindole (DAPI) staining of nuclei showed nescence of melanoma cells (Supplementary Fig. S3). that cells transfected with the MITF siRNA encountered

3816 Cancer Res; 70(9) May 1, 2010 Cancer Research

MITF Controls a DDR/p53-Dependent Senescence Program

several chromosome segregation defects with a high preva- were positive for 53BP1 foci compared with control cells, and lence of lagging chromosome (anaphase chromatin forming the number of 53BP1 foci per nucleus was more important continuous bridges connecting the two sets of separating than those found occasionally in control cells (Fig. 2C). chromosomes) compared with control cells (Fig. 2A). Note- These observations therefore show that MITF depletion worthy, for better contrast, DAPI staining has been depicted causes mitotic defects that culminate in DNA damages, most in green. Anaphase bridges have been associated with DNA likely DNA DSBs. double-strand breaks (DSB; ref. 17). Further, in mammals, his- Activation of the DDR signaling cascade is required for tone H2AX phosphorylation (γH2AX) and 53BP1 are rapidly the senescence phenotypes mediated by MITF silencing. recruited to sites of DSB and act as key mediators in the DNA damages, through activation of the DDR signaling path- transduction of DNA damage signals (18, 19). In line with this, way, trigger cellular senescence (20). Thus, we next wished to immunofluorescence studies showed a weak and diffuse delineate the role of the DDR signaling cascade in the senes- γH2AX staining all through the nucleus in control cells, where- cence program triggered by MITF suppression. as 70% of cells transfected with MITF siRNA exhibited γH2AX CHK2 has a fundamental role in the network of genome foci at the site of DNA damages. Western blot analysis con- surveillance pathways by regulating cell cycle progression. firmed the phosphorylation of H2AX in MITF-depleted cells We observed that MITF silencing promoted a time-dependent (Fig. 2B). Additionally, a large number of MITF-depleted cells phosphorylation of CHK2 on Thr68 that is localized in discrete

Figure 3. Inhibition of the ATM-dependent DDR cascade prevents senescence engagement. A, fluorescence images of control siRNA– or MITF siRNA–transfected 501 mel cells stained with antibody to phospho-Thr68-CHK2 (pT68-CHK2). Nuclei were counterstained with DAPI. Kinetic study of the expression of MITF, CHK2 (total and phospho-Thr68), and ERK2 by Western blot. B, 501 mel cells were treated with DMSO (Ct) or a pharmacologic inhibitor of CHK2 (InhChk2, 10 μmol/L) for 96 h. Efficiency of the inhibitor was monitored by Western blot on autophosphorylation of CHK2 on residue Thr387 (pT387-CHK2). SA-β-Gal staining and percentage of means and SDs of SA-β-Gal–positive cells are shown. C, control siRNA– or MITF siRNA–transfected cells were exposed to Ku55933 (10 μmol/L) or Nu7026 (10 μmol/L) for 96 h. Lysates were analyzed by Western blot for phospho-Thr68-CHK2. SA-β-Gal staining and percentage of means and SDs of SA-β-Gal–positive cells are shown. D, immunofluorescence microscopy images of control siRNA–transfected or MITF siRNA–transfected 501 mel cells stained with phospho-Ser1981-ATM (pS1981-ATM) or MITF antibodies.

www.aacrjournals.org Cancer Res; 70(9) May 1, 2010 3817

Giuliano et al.

Figure 4. MITF silencing increases the level of transcriptionally active p53. A, fluorescence images show MITF and p53 labeling in control siRNA–transfected or MITF siRNA–transfected cells. Nuclei were counterstained with DAPI. B, left, Western blot revealed increased phosphorylation at Ser15 (pS15-p53) and increased protein levels of p53 after 36 h of MITF depletion; middle, cells were transfected with PG13-Luc, a p53-dependent firefly luciferase reporter gene, and then with control or MITF siRNA; right, supernatants from these cells were examined by Western blot for PAI-1. Inset, efficiency of siMi. C, right, luciferase assay using 501 mel cells transfected with PAI-1 promoter, pCMV-β-galactosidase, and various siRNA constructs or empty or p53-encoding plasmid. D, ChIP was performed using 501 mel cells and specific anti-p53 antibody and mouse IgG as control. Primers spanning the PAI-1 promoter region were used for PCR amplification. A control of PCR amplification was performed on nonimmunoprecipitated extracts (Input). Another control was performed using a primer pair to the human GAPDH promoter.

foci within the nucleus (Fig. 3A). In cells transfected with Noteworthy, debromohymenialdisine, another pharmacologic control siRNA, low level of CHK2 phosphorylation and almost CHK2 inhibitor, gave comparable results in inhibiting no foci were observed. Noteworthy, in Western blot assay, senescence phenotypes (Supplementary Fig. S6B). CHK2 phosphorylation was clearly visible at 24 hours after The DDR pathway activates members of the phosphoino- transfection and therefore was observed before SA-β-Gal sitide 3-kinase–related protein kinase (PIKK) family (i.e., detection that appeared at 48 hours, thereby indicating that DNA-PK, ATM, and ATR). We therefore assessed their rela- CHK2 phosphorylation was not a consequence of senescence tive involvement in the senescence program mediated by induction by MITF silencing. Autophosphorylation at Thr387, MITF depletion by using specific inhibitors, namely, KU- which is required for full CHK2 activity, was also stimulated by 55993, an inhibitor of ATM (21) and NU-7026, an inhibitor MITF suppression and was abrogated by a cell-permeable in- of DNA-PK (22). Phosphorylation of CHK2 at Thr68 by PIKK doloazepine that specifically inhibits CHK2 (InhChk2; Fig. 3B). was compromised in the presence of the pharmacologic in- The CHK2 inhibitor did not affect MITF expression. Interest- hibitors showing their efficiency (Fig. 3C). Noteworthy, the ingly, CHK2 inhibition dramatically impaired induction of the inhibition mediated by KU-55993 was stronger than that senescence phenotypes mediated by MITF siRNA, as shown by caused by NU-7026. Consistently, compared with the DNA-PK theblockadeofSA-β-Gal staining and by the decrease of inhibitor, the ATM inhibitor KU-55993 severely compromised cell size and granularity (Fig. 3B; Supplementary Fig. S6A). the number of senescent cells induced by MITF silencing.

3818 Cancer Res; 70(9) May 1, 2010 Cancer Research

MITF Controls a DDR/p53-Dependent Senescence Program

Consistently, cell size and granularity returned to control Phosphorylation of p53 on Ser15 is crucial for its stabilization levels in the presence of KU-55993, whereas NU-7026 barely but also stimulates p53 transcriptional activity (23, 24). We affected these two parameters (Supplementary Fig. S7). Fur- indeed observed in MITF-depleted cells an enhanced activity thermore, aggregation of phosphorylated ATM (Ser1981)in of a p53-responsive reporter plasmid, showing that MITF de- discrete nuclear foci was more important in MITF-depleted pletion induced the expression of a transcriptionally active cells compared with control cells (Fig. 3D). We therefore p53 (Fig. 4B). Consistently, MITF silencing stimulated the se- identify ATM and CHK2, two components of the DDR path- cretion of PAI-1, a p53 target gene, which has been reported way, as key molecules in the senescence program triggered by to be a key mediator of cellular senescence (25). Ponceau MITF depletion. staining revealed equal loading of the two lanes (data not MITF suppression promotes p53 accumulation and shown). Furthermore, MITF depletion increased the mRNA activation. Upregulation of p53 protein is also a key feature level of PAI-1 (data not shown) and stimulated the transcrip- of DNA damages and a pivotal player in cellular senescence. tional activity of the PAI-1 promoter to a level comparable Interestingly, MITF silencing clearly induced a nuclear accu- with that of p53 (Fig. 4C). p53 siRNA, which abolished both mulation of p53 (Fig. 4A). Western blot analysis confirmed basal and MITF-induced p53, prevented PAI-1 transcription the accumulation of p53. Noteworthy, activation of CHK2 (Figs. 4C and 5A). ChIP experiments revealed that, in mela- and increased expression of p53 were also observed in other noma cells, p53 bound to the promoter of PAI-1, thereby in- melanoma cells (Supplementary Fig. S8). dicating that MITF silencing increased, through upregulation Consistently, an increased phosphorylation of p53 on Ser15, of p53, the transcription of PAI-1 (Fig. 4D). In conclusion, a site dominantly phosphorylated by ATM, was observed. stimulation of H2AX, 53BP1, CHK2, and p53, three ATM

Figure 5. p53 plays a critical role in the senescence program mediated by MITF depletion. A, co-knockdown of p53 and MITF prevented p53 expression and severely compromised SA-β-Gal staining. Percentage of means and SDs of SA-β-Gal–positive cells are indicated. B, p53-deficient SKMel28 cells were transfected with two different MITF siRNAs. Cells were next analyzed by Western blot for MITF and ERK2 and for SA-β-Gal activity. C, SKMel28 cells transfected with MITF siRNAs were also counted in triplicates from days 2 to 6. D, model for MITF silencing-induced senescence.

www.aacrjournals.org Cancer Res; 70(9) May 1, 2010 3819

Giuliano et al.

targets, together with p53 accumulation and increased time the existence of a lineage-specific cell-autonomous activity, reveals mobilization of the DNA damage detection senescence program. machinery on MITF silencing. Interestingly, knockdown of the c-myc oncogene, a tran- p53 is a key mediator of the senescence program scription factor of the bHLH-LZ family, such as MITF, also mediated by MITF depletion. We next investigated the im- caused senescence of melanoma cells (29). Therefore, MITF, portance of p53 and found that p53 suppression caused cells which was proposed to function as a melanocyte-specific on- to efficiently bypass the senescence program mediated by cogene, and c-myc may both function in convergent path- MITF depletion (Fig. 5A). Interestingly, in p53-mutant ways controlling cellular senescence, recently referred as SKMel28 melanoma cells, although the two MITF siRNAs pro- oncogene inactivation-induced senescence (OIIS; refs. 2, 29, moted a strong reduction in the level of MITF, only a small 30). However, unlike MITF, the effect of c-myc depletion proportion of cells became positive for SA-β-Gal staining was not restrained to melanoma cells because it also trig- (Fig. 5B). In these cells, MITF suppression slowed down gered senescence in lymphoma and osteosarcoma cells proliferation, but these cells kept growing (Fig. 5C). (30). Further, c-myc inactivation in these different cell types These findings thereby show that MITF silencing requires did not engage the same cell cycle regulators and checkpoint activation of the ATM-CHK2-p53 signaling cascade to reacti- proteins, thereby showing that OIIS, caused by c-myc sup- vate a senescence growth arrest of melanoma cells (Fig. 5D). pression, operates in a context-dependent manner. There- fore, it was important to unveil the molecular mechanism by which MITF depletion exerted its prosenescing effects. Discussion We also show that MITF suppression clearly promotes senescence in p16INK4A-proficient (501 mel) and p16INK4A- Cutaneous melanoma is a highly aggressive tumor recog- deficient melanoma cells (WM9 and B16) and that genetic nized for its notorious resistance to all current therapeutic ablation of p16INK4A in 501 mel cells does not abrogate the intervention. Numerous studies have ascribed this resistance senescence growth arrest phenotype in agreement with pre- to alterations in apoptotic processes (26). Less attention has vious reports (31). Additionally, CDK2 and p27KIP1, two other been paid to senescence, a process that limits cell prolifera- cell cycle regulators deregulated on MITF depletion and in- tion and oncogene-induced transformation. Alteration in se- volved in cellular senescence (32, 33), do not seem as key nescence processes might also play a key role in melanoma mediators of MITF siRNA prosenescing effect. Noteworthy, development. Therefore, deciphering the mechanism of a 25% increase in the number of senescent cells is observable senescence regulation in melanoma could offer new thera- in CDK2-depleted cells that could be explained by a reduced peutic strategies for the treatment of this disease. MITF expression (Supplementary Fig. S5). The decrease in MITF plays a crucial role in melanocyte homeostasis and MITF expression on CDK2 suppression may reflect a nega- melanoma development. Therefore, several attempts have tive feedback loop that remains to be investigated. been made to further elucidate how MITF exerts its effects. On the other hand, when examining MITF-silenced cells, Here, we show that long-term silencing of MITF in mela- nuclei reveal severe chromosome segregation defects with a noma cells induces a phenotype displaying all the hallmarks high prevalence of lagging chromosome at anaphase. None of senescence, such as enlarged and flattened cell, vacuoliza- of the MITF target genes described thus far have been in- tion, SA-β-Gal activity at acidic pH, and SAHF labeled by volved in this process. Therefore, our observations ascribe to HP1β. SAHF are subnuclear structures containing hetero- MITF novel functions in correct chromosome segregation. chromatinproteins,whichresultinstablerepressionof Anaphase bridges have been associated with DNA DSBs, promoters of proliferation-associated genes (27) that might suggesting the presence of DNA damages (17). These obser- explain the robust and sustained cell cycle arrest observed vations prompted us to investigate, in response to MITF in MITF-depleted melanoma cells. depletion, the involvement and the role of the DDR pathway MITF suppression triggers senescence-like phenotypes that has been already implicated in the control of senescence in several human and mouse melanoma cells of different (20, 34). In agreement with our hypothesis, we show here that genetic background (Supplementary Fig. S9), thereby show- the senescence program mediated by MITF silencing engages ing that this process is not restrained to species or to a a broad DDR signaling pathway marked by activation of ATM, unique melanoma cell line. In normal human melanocytes, γH2AX, 53BP1, CHK2, and p53, culminating in p53 accumula- MITF silencing displays weak senescence-promoting activity tion. Noteworthy, although we cannot rule out the existence of (data not shown). Importantly, MITF silencing mediates se- DNA single-strand breaks, engagement of the ATM/γH2AX/ nescence in melanoma cells harboring BRAFV600E, the most 53BP1/CHK2/p53 cascade in MITF-silenced cells strongly frequently mutated oncogene in this disease (28). Addition- indicates that the type of DNA damages is mainly DSB. ally, MITF silencing may also engage senescence of freshly Pharmacologic or genetic inhibition of ATM, CHK2, or p53 isolated melanoma cells, two of them being BRAFV600E. Alto- severely compromises the senescence program mediated by gether, these observations highlight the physiopathologic MITF suppression, showing the requirement of the DDR cas- relevance of our observations and provide a rational basis cade in this process. Notably, like ATM, DNA-PK has also for the use of MITF suppression in the treatment of mela- been reported to phosphorylate CHK2 (35) and to promote noma. More importantly, the control of senescence by MITF, p53 stabilization (36). However, using pharmacologic inhibi- a cell-specific transcription factor, evidences for the first tion of these kinases, we show that ATM plays a more active

3820 Cancer Res; 70(9) May 1, 2010 Cancer Research

MITF Controls a DDR/p53-Dependent Senescence Program

role in the senescence program engaged by MITF depletion. As previously reported, we find no induction of cell death In SKMel28 melanoma cells, which harbor a mutant form of in response to MITF depletion (5, 12). The fact that increased p53, MITF silencing induces a modest accumulation of cells expression of p53 observed on MITF depletion causes senes- that stained positive for SA-β-Gal and does not completely cence instead of cell death may be linked to the genetic back- block cell growth, suggesting that p53 is a key actor of ground of melanoma cells. Indeed, alteration of several the senescence program mediated by MITF silencing. This molecules involved in the regulation and execution of apo- hypothesis was confirmed because siRNA to p53 prevents ptosis, such as BCL2 and APAF1, was reported (26). Further, almost completely the SA-β-Gal staining induced on MITF increased expression of p27KIP1 has been reported to exert depletion in p53-proficient melanoma cells. antiapoptotic activity (45). Therefore, it is likely that the high Previous reports indicated that p53 mediated G1 cell cycle ar- antiapoptotic background of melanoma cells associated with rest through transcriptional repression of c-myc (37) and c-myc the genetic program mediated by MITF depletion would fa- depletion promoted senescence of melanoma cells (29). Howev- vor the establishment of a senescence program rather than er, no change in the intracellular level of c-myc is observed, cell death. ruling out the possibility that MITF depletion acts through Several lines of evidence indicate that p53 signaling can c-myc suppression to trigger senescence (data not shown). suppress melanoma development, as inactivation of p53 fa- p21CIP1 is one of the most studied p53 targets facilitating vors melanoma development induced by oncogenic NRAS or the cell cycle arrest that characterizes senescence (38). Al- BRAF (46–48). However, unlike in other solid tumors, p53 is though MITF silencing causes p53 upregulation and activa- rarely mutated in melanomas. Therefore, it is crucial to un- tion, the expression of p21CIP1 seems rather inhibited in the derstand how melanoma cells could escape from the growth- different melanoma cell lines. The uncoupling between suppressive function of p53. According to our data, we can p21CIP1 and p53 is not unprecedented particularly in melano- propose that MITF, in which its expression is conserved in ma cells (39). In support of these observations, in melanoma 80% of melanomas, maintains a low level of p53, inactivates cells, MITF is a direct transcriptional activator of p21CIP1, the senescence barrier, and allows melanoma development. and therefore, MITF suppression would cause an inhibition This hypothesis is in complete agreement with data showing of p21CIP1 expression (40). Additionally, p21CIP1 expression is an absence of p53 staining in some melanoma biopsies and a repressed at the transcriptional level by TBX2, a transcrip- positive correlation between p53 detection and overall sur- tion factor of the T-box family overexpressed in melanoma vival of patients (49, 50). Therefore, our results might bring cells (41). Therefore, MITF and TBX2 could be responsible important clues to explain the enigmatic absence of p53 for the uncoupling of p53 and p21CIP1 in melanoma cells. Fur- mutations in a vast majority of melanoma and foster the re- ther, p21CIP1 can limit the DDR (42). Therefore, the decrease assessment of drugs targeting the p53 pathway in the treat- in p21CIP1 could take part to a genetic program activated by ment of melanoma. MITF silencing that would favor activation of the DDR, upregulation of p53, and induction of senescence. Disclosure of Potential Conflicts of Interest Kortlever and colleagues (25) recently showed that PAI-1 was necessary and sufficient to trigger, through p53, cellular No potential conflicts of interest were disclosed. senescence. Here, we observe a marked increase in the production of PAI-1 in MITF-silenced cells, showing the func- tional activation of p53. Noteworthy, our results, in melanoma Grant Support cells, indicate that p53 binds to the promoter of PAI-1 and Institut National de la Sante et de la Recherche Medicale, Institut National controls PAI-1 expression at the transcriptional level. Consis- du Cancer grant R08009AP, Ligue Nationale contre le Cancer “Equipe labelli- tently, PAI-1 has been already negatively associated with mel- sée,” and Association pour la Recherche sur le Cancer grant 3895. M. Ohanna is a recipient from the Institut National du Cancer and Y. Cheli from the Asso- anoma development. Indeed, PAI-1 was described to inhibit ciation pour la Recherche sur le Cancer. the activity of the secreted protease urokinase-type plasmin- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in ogen activator, which can cause cells to progress through G1 to S phase (43). Further, inhibition of proliferation and inva- accordance with 18 U.S.C. Section 1734 solely to indicate this fact. sive properties of melanoma cells was associated with an Received 08/05/2009; revised 01/18/2010; accepted 02/15/2010; published increased expression of PAI-1 (44). OnlineFirst 04/13/2010.

References 1. Cheli Y, Ohanna M, Ballotti R, Bertolotto C. 15-year quest in search MITF expression via HDAC inhibitors in the melanocyte lineage. Pig- for MITF target genes. Pigment Cell Melanoma Res 2010;23:27–40. ment Cell Melanoma Res 2008;21:457–63. 2. Garraway LA, Widlund HR, Rubin MA, et al. Integrative genomic anal- 5. Carreira S, Goodall J, Denat L, et al. Mitf regulation of Dia1 con- yses identify MITF as a lineage survival oncogene amplified in malig- trols melanoma proliferation and invasiveness. Genes Dev 2006;20: nant melanoma. Nature 2005;436:117–22. 3426–39. 3. Halaban R, Krauthammer M, Pelizzola M, et al. Integrative analysis of 6. Du J, Widlund HR, Horstmann MA, et al. Critical role of CDK2 for epigenetic modulation in melanoma cell response to decitabine: clin- melanoma growth linked to its melanocyte-specific transcriptional ical implications. PLoS One 2009;4:e4563. regulation by MITF. Cancer Cell 2004;6:565–76. 4. Yokoyama S, Feige E, Poling LL, et al. Pharmacologic suppression of 7. Kido K, Sumimoto H, Asada S, et al. Simultaneous suppression of

www.aacrjournals.org Cancer Res; 70(9) May 1, 2010 3821

Giuliano et al.

MITF and BRAF V600E enhanced inhibition of melanoma cell prolif- 29. Zhuang D, Mannava S, Grachtchouk V, et al. C-MYC overexpression eration. Cancer Sci 2009;100:1863–9. is required for continuous suppression of oncogene-induced senes- 8. DynekJN,ChanSM,LiuJ,ZhaJ,FairbrotherWJ,VucicD. cence in melanoma cells. Oncogene 2008;27:6623–34. Microphthalmia-associated transcription factor is a critical transcrip- 30. Wu CH, van Riggelen J, Yetil A, Fan AC, Bachireddy P, Felsher DW. tional regulator of melanoma inhibitor of apoptosis in melanomas. Cellular senescence is an important mechanism of tumor regression Cancer Res 2008;68:3124–32. upon c-Myc inactivation. Proc Natl Acad Sci U S A 2007;104: 9. McGill GG, Horstmann M, Widlund HR, et al. Bcl2 regulation by the 13028–33. melanocyte master regulator Mitf modulates lineage survival and 31. Bandyopadhyay D, Curry JL, Lin Q, et al. Dynamic assembly of melanoma cell viability. Cell 2002;109:707–18. chromatin complexes during cellular senescence: implications for 10. Nakai N, Kishida T, Shin-Ya M, et al. Therapeutic RNA interference of the growth arrest of human melanocytic nevi. Aging Cell 2007;6: malignant melanoma by electrotransfer of small interfering RNA tar- 577–91. geting Mitf. Gene Ther 2007;14:357–65. 32. Collado M, Medema RH, Garcia-Cao I, et al. Inhibition of the phos- 11. Schwahn DJ, Timchenko NA, Shibahara S, Medrano EE. Dynamic phoinositide 3-kinase pathway induces a senescence-like arrest regulation of the human dopachrome tautomerase promoter by mediated by p27Kip1. J Biol Chem 2000;275:21960–8. MITF, ER-α and chromatin remodelers during proliferation and se- 33. Freedman DA, Folkman J. CDK2 translational down-regulation dur- nescence of human melanocytes. Pigment Cell Res 2005;18:203–13. ing endothelial senescence. Exp Cell Res 2005;307:118–30. 12. Larribere L, Hilmi C, Khaled M, et al. The cleavage of microphthal- 34. Mallette FA, Gaumont-Leclerc MF, Ferbeyre G. The DNA damage mia-associated transcription factor, MITF, by caspases plays an signaling pathway is a critical mediator of oncogene-induced senes- essential role in melanocyte and melanoma cell apoptosis. Genes cence. Genes Dev 2007;21:43–8. Dev 2005;19:1980–5. 35. Li J, Stern DF. Regulation of CHK2 by DNA-dependent protein 13. Bertolotto C, Abbe P, Hemesath TJ, et al. Microphthalmia gene kinase. J Biol Chem 2005;280:12041–50. product as a signal transducer in cAMP-induced differentiation of 36. Boehme KA, Kulikov R, Blattner C. p53 stabilization in response to melanocytes. J Cell Biol 1998;142:827–35. DNA damage requires Akt/PKB and DNA-PK. Proc Natl Acad Sci 14. Khaled M, Larribere L, Bille K, Ortonne JP, Ballotti R, Bertolotto C. U S A 2008;105:7785–90. Microphthalmia associated transcription factor is a target of the 37. Ho JS, Ma W, Mao DY, Benchimol S. p53-dependent transcriptional

phosphatidylinositol-3-kinase pathway. J Invest Dermatol 2003; repression of c-myc is required for G1 cell cycle arrest. Mol Cell Biol 121:831–6. 2005;25:7423–31. 15. Hilmi C, Larribere L, Giuliano S, et al. IGF1 promotes resistance to 38. Ben-Porath I, Weinberg RA. The signals and pathways activating apoptosis in melanoma cells through an increased expression of cellular senescence. Int J Biochem Cell Biol 2005;37:961–76. BCL2, BCL-X(L), and survivin. J Invest Dermatol 2008;128:1499–505. 39. Gray-Schopfer VC, Cheong SC, Chong H, et al. Cellular senescence 16. Larribere L, Khaled M, Tartare-Deckert S, et al. PI3K mediates pro- in naevi and immortalisation in melanoma: a role for p16? Br J tection against TRAIL-induced apoptosis in primary human melano- Cancer 2006;95:496–505. cytes. Cell Death Differ 2004;11:1084–91. 40. Carreira S, Goodall J, Aksan I, et al. Mitf cooperates with Rb1 and 17. Acilan C, Potter DM, Saunders WS. DNA repair pathways involved in activates p21Cip1 expression to regulate cell cycle progression. Na- anaphase bridge formation. Genes Chromosomes Cancer 2007;46: ture 2005;433:764–9. 522–31. 41. Prince S, Carreira S, Vance KW, Abrahams A, Goding CR. Tbx2 di- 18. Rogakou EP, Pilch DR, Orr AH, Ivanova VS, Bonner WM. DNA rectly represses the expression of the p21(WAF1) cyclin-dependent double-stranded breaks induce histone H2AX phosphorylation on kinase inhibitor. Cancer Res 2004;64:1669–74. serine 139. J Biol Chem 1998;273:5858–68. 42. Viale A, De Franco F, Orleth A, et al. Cell-cycle restriction limits DNA 19. Ward IM, Minn K, van Deursen J, Chen J. p53 binding protein 53BP1 damage and maintains self-renewal of leukaemia stem cells. Nature is required for DNA damage responses and tumor suppression in 2009;457:51–6. mice. Mol Cell Biol 2003;23:2556–63. 43. De Petro G, Copeta A, Barlati S. Urokinase-type and tissue-type 20. Bartek J, Bartkova J, Lukas J. DNA damage signalling guards plasminogen activators as growth factors of human fibroblasts. against activated oncogenes and tumour progression. Oncogene Exp Cell Res 1994;213:286–94. 2007;26:7773–9. 44. Pasco S, Ramont L, Venteo L, Pluot M, Maquart FX, Monboisse JC. 21. Shiotani B, Zou L. Single-stranded DNA orchestrates an ATM-to- In vivo overexpression of tumstatin domains by tumor cells inhibits ATR switch at DNA breaks. Mol Cell 2009;33:547–58. their invasive properties in a mouse melanoma model. Exp Cell Res 22. Veuger SJ, Curtin NJ, Richardson CJ, Smith GC, Durkacz BW. 2004;301:251–65. Radiosensitization and DNA repair inhibition by the combined use 45. Eymin B, Haugg M, Droin N, Sordet O, Dimanche-Boitrel MT, Solary E. of novel inhibitors of DNA-dependent protein kinase and poly(ADP- p27Kip1 induces drug resistance by preventing apoptosis upstream ribose) polymerase-1. Cancer Res 2003;63:6008–15. of cytochrome c release and procaspase-3 activation in leukemic 23. Lakin ND, Jackson SP. Regulation of p53 in response to DNA dam- cells. Oncogene 1999;18:1411–8. age. Oncogene 1999;18:7644–55. 46. Bardeesy N, Bastian BC, Hezel A, Pinkel D, DePinho RA, Chin L. Dual 24. Zhao H, Traganos F, Darzynkiewicz Z. Phosphorylation of p53 on inactivation of RB and p53 pathways in RAS-induced melanomas. Ser15 during cell cycle caused by Topo I and Topo II inhibitors in Mol Cell Biol 2001;21:2144–53. relation to ATM and Chk2 activation. Cell Cycle 2008;7:3048–55. 47. Chudnovsky Y, Adams AE, Robbins PB, Lin Q, Khavari PA. Use 25. Kortlever RM, Higgins PJ, Bernards R. Plasminogen activator inhibitor-1 of human tissue to assess the oncogenic activity of melanoma- is a critical downstream target of p53 in the induction of replicative associated mutations. Nat Genet 2005;37:745–9. senescence. Nat Cell Biol 2006;8:877–84. 48. Yu H, McDaid R, Lee J, et al. The role of BRAF mutation and p53 26. Soengas MS, Lowe SW. Apoptosis and melanoma chemoresistance. inactivation during transformation of a subpopulation of primary Oncogene 2003;22:3138–51. human melanocytes. Am J Pathol 2009;174:2367–77. 27. Narita M, Nunez S, Heard E, et al. Rb-mediated heterochromatin for- 49. Brantley MA, Jr., Harbour JW. Deregulation of the Rb and p53 mation and silencing of E2F target genes during cellular senescence. pathways in uveal melanoma. Am J Pathol 2000;157:1795–801. Cell 2003;113:703–16. 50. Essner R, Kuo CT, Wang H, et al. Prognostic implications of p53 28. Garnett MJ, Marais R. Guilty as charged: B-RAF is a human onco- overexpression in cutaneous melanoma from sun-exposed and gene. Cancer Cell 2004;6:313–9. nonexposed sites. Cancer 1998;82:309–16.

3822 Cancer Res; 70(9) May 1, 2010 Cancer Research

Microphthalmia-Associated Transcription Factor Controls the DNA Damage Response and a Lineage-Specific Senescence Program in Melanomas

Sandy Giuliano, Yann Cheli, Mickaël Ohanna, et al.

Cancer Res 2010;70:3813-3822. Published OnlineFirst April 13, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-09-2913

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2010/04/13/0008-5472.CAN-09-2913.DC1

Cited articles This article cites 50 articles, 15 of which you can access for free at: http://cancerres.aacrjournals.org/content/70/9/3813.full#ref-list-1

Citing articles This article has been cited by 22 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/70/9/3813.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerres.aacrjournals.org/content/70/9/3813. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.