Role of Mdm4 in Drug Sensitivity of Breast Cancer Cells

S Lam1, K Lodder1, AFAS Teunisse1, MJWE Rabelink1, M Schutte2 and AG Jochemsen1

1Department of Molecular Cell Biology, Leiden University Medical Center, Leiden, The Netherlands and 2Department of Medical Oncology, Josephine Nefkens Institute, Erasmus University Medical Center, Rotterdam, The Netherlands

The p53 tumor suppressor protein is frequently mutated in acquired mutations that impair the functionality of p53. human tumors. It is thought that the p53 pathway is All together, approximately 50% of all human malig- indirectly impaired in the remaining tumors, for example nancies harbor mutations in p53, whereas the p53 tumor by overexpression of its important regulators Mdm2 and suppressor pathway is thought to be functionally Mdm4, making them attractive targets for the develop- impaired in the remaining tumors, for example by ment of anti-cancer agents. Recent studies have suggested alterations in the main regulators of p53 (Toledo and that Mdm4 levels determine the sensitivity of tumor cells Wahl, 2006). for anti-cancer therapy. To investigate this possibility, we In unstressed normal cells, p53 is present at low levels studied the drug sensitivity of several breast cancer cell because of rapid degradation by the ubiquitin-depen- lines containing wild-type p53, but expressing different dent proteasome pathway (Brooks and Gu, 2006). Mdm4 levels. We show that endogenous Mdm4 levels can Mdm2 is an important regulator of p53 turnover by affect the sensitivity of breast cancer cells to anti-cancer binding p53 and acting as an E3 ubiquitin ligase (Haupt agents, but in a cell line-dependent manner and depending et al., 1997; Kubbutat et al., 1997). In addition, binding on an intact apoptotic response. Furthermore, treatment of Mdm2 to p53 disrupts the interaction between p53 with the non-genotoxic agent Nutlin-3 sensitizes cells for and the transcriptional machinery, thereby antagonizing doxorubicin, showing that activation of p53 by targeting p53 transcriptional activation (Momand et al., 1992). its regulators is an efficient strategy to decrease cell Importantly, Mdm2 overexpression has been found in viability of breast cancer cells. These results confirm a human cancers, in general correlating with the expres- function of Mdm4 in determining the efficacy of sion of wild-type p53 (Oliner et al., 1992). chemotherapeutic agents to induce apoptosis of cancer Mdm4, a structural homologue of Mdm2, is also a cells in a p53-dependent manner, although additional critical regulator of p53 activity (Shvarts et al., 1996, undetermined factors also influence the drug response. 1997). Unlike Mdm2, Mdm4 has no detectable ubiquitin Targeting Mdm4 to sensitize tumor cells for chemother- ligase activity and is, therefore, unable to target p53 for apeutic drugs might be a strategy to effectively treat degradation. The primary function of Mdm4 seems to tumors harboring wild-type p53. be inhibition of p53-mediated transcriptional activation. Oncogene (2010) 29, 2415–2426; doi:10.1038/onc.2009.522; In addition, it has been reported that Mdm4 can published online 8 February 2010 enhance Mdm2-mediated ubiquitination and degradation of p53 (Stad et al., 2000, 2001). Similar to Mdm2, Mdm4 Keywords: Mdm4; p53; cancer; Nutlin; doxorubicin has been found overexpressed in a significant number of various human tumors and tumor cell lines, in general correlating with the expression of wild-type p53, suggesting that Mdm4 contributes to p53 inactivation during Introduction tumorigenesis (Riemenschneider et al., 1999, 2003; Ramos et al., 2001; Danovi et al.,2004;Laurieet al., The p53 tumor suppressor protein has a major function 2006). As both Mdm2 and Mdm4 are overexpressed in in the maintenance of genome integrity. Activation of many tumors that retain wild-type p53, they are attractive p53 in response to various signals leads to the targets for the development of anti-cancer agents. transcriptional regulation of genes important for sup- Mdm2 and Mdm4 are both targeted by stress pressing tumorigenesis. In addition, p53 can induce signaling pathways that activate p53 (Meulmeester apoptosis in a transcription-independent manner et al., 2005). Genotoxic stress activates damage-respon- (Fuster et al., 2007). The critical function of p53 is sive kinases such as ATM, which activates p53 in part by reflected by the fact that it is frequently targeted for direct phosphorylation of p53 itself and also by inactivation and a majority of malignant tumors have destabilization and inhibition of its negative regulators (Pereg et al., 2005, 2006). Therefore, DNA damage Correspondence: Dr AG Jochemsen, Department of Molecular Cell potently stimulates p53 activity by eliminating interac- Biology, Leiden University Medical Center, PO Box 9600, Postal Zone tions with its two main negative regulators. S1-P, Leiden 2300 RC, The Netherlands. However, the use of genotoxic agents in chemo- E-mail: [email protected] Received 13 April 2009; revised 23 November 2009; accepted 3 therapy is associated with undesirable and severe side December 2009; published online 8 February 2010 effects (Shapiro and Recht, 2001). Other approaches Regulation of drug response by Mdm4 S Lam et al 2416 aim at disrupting the binding between Mdm2 and p53 to MCF-7, MDA-MB-175VII, MPE600, SK-BR-7, facilitate wild-type p53 activities. Recently, Nutlin-3 has UACC-812, ZR75-1, and ZR75-30 (Supplementary been identified in a high throughput screen for Figure 1; Wasielewski et al., 2006). From this panel of compounds capable of disrupting Mdm2–p53 interac- cell lines, protein expression profiles of the p53 and pRB tions (Vassilev et al., 2004). Nutlin-3 is a small molecule pathways have been analyzed by western blot (Figure 1). inhibitor that binds Mdm2 in its p53-binding pocket, Although p53 levels of breast cancer cell lines were thereby preventing binding to p53. Consequently, comparable with the normal breast epithelial cell line Nutlin-3 treatment results in stabilization of p53 and MCF-10A1, the expression of either Mdm2 or Mdm4 activation of the p53 pathway. Nutlin-3 was identified was increased in breast cancer cell lines. In addition to based on its ability to inhibit Mdm2–p53 binding. The suppression of the p53 pathway by overexpression of its p53-binding pockets of Mdm2 and Mdm4 show high negative regulators, the pRB tumor suppressor pathway homology and it has been shown that Nutlin-3 was able to disrupt the binding between Mdm4 and p53 by binding to Mdm4 (Laurie et al., 2006), although conflicting results have been reported (Hu et al., 2006; Patton et al., 2006; Wade et al., 2006). Similarly, MI-219 activates p53 by disrupting the Mdm2–p53 interaction. MI-219 was designed based on the crystal structure of the Mdm2–p53 complex, and its ability to inhibit Mdm4–p53 interactions was significantly less efficient (Shangary et al., 2008). Therefore, despite strong similarities between the p53-binding domains of Mdm4 and Mdm2, minor differences are sufficient to reduce the inhibitory effect of Nutlin-3 on Mdm4–p53 binding. Consistently, some studies have suggested that Mdm4 levels determine the sensitivity of tumor cells for anti- cancer therapy (Hu et al., 2006; Laurie et al., 2006; Patton et al., 2006; Wade et al., 2006). Importantly, Nutlin-3-induced activation of p53 is not associated with induction of DNA damage or damage-induced modifications of p53 (Vassilev et al., 2004). Owing to these characteristics, Nutlin-3 has been tested for its synergistic effects with existing therapies. It has been shown that treatment with Nutlin-3 is synergistic with genotoxic drugs and radiation in cancer therapy (Barbieri et al., 2006; Cao et al., 2006; Coll- Mulet et al., 2006; Patton et al., 2006). We have investigated the function of Mdm4 expression levels in determining the drug sensitivity in a number of breast cancer cell lines containing wild-type p53, but expressing different Mdm4 levels. Manipulation of Mdm4 levels showed that Mdm4 can affect the sensitivity of breast cancer cells to anti-cancer agents, but in a strong cell line-dependent manner and not correlating with initial or remaining Mdm4 levels. The effect of manipulating Mdm4 levels on the drug response of these breast cancer cell lines correlated with the effect on the apoptotic response. These results suggest that expression of proteins involved in the apoptotic pathway(s) together with Mdm4 levels determine drug sensitivity of cancer cells. Identification of such components, which affect drug sensitivity in tumor cells might provide additional targets for further improvement of anti-cancer therapies. Figure 1 Protein expression profiles of wild-type p53 breast cancer cell lines. Western blot analysis of proteins in the p53 and pRB pathways of a panel of eight wild-type p53 breast cancer cell lines Results (Wasielewski et al., 2006); MCF-10A1 was used as a control for normal breast epithelial cells. Proteins were extracted from cells in Protein expression profiles of wild-type p53 breast cancer exponential growth and cultured under standard conditions. HAUSP and g-tubulin expression were analyzed as loading cell lines controls. Protein bands were quantified and normalized for From a panel of 41 breast cancer cell lines, we have HAUSP expression. Relative expression was calculated by compar- selected cell lines, which express wild-type p53: DU4475, ing with DU4475 cells.

Oncogene Regulation of drug response by Mdm4 S Lam et al 2417 was attenuated to varying extents by either homozygous induced degradation of Mdm4 has to be further deletion of pRB or p16, methylation of the p16 gene, or elucidated, but we did notice Ser15-phosphorylation of amplification/overexpression of Cyclin D1 (Hollestelle p53 on prolonged RITA treatment. Interestingly, the et al., 2009). cells showing relative resistance to RITA, ZR75-1 cells, hardly showed a decrease in Mdm4 levels. These results suggest that endogenous Mdm4 levels may affect the Drug sensitivity of breast cancer cells to various sensitivity to some anti-cancer drugs. anti-cancer drugs As all tested anti-cancer drugs activated p53, it was Several studies have suggested that Mdm4 levels can determined whether p53 activation would be required affect the sensitivity of tumor cells for anti-cancer for the drug-induced growth inhibition. Therefore, p53 therapy. To investigate this possibility in more detail, was knocked down using short hairpin RNA (shRNA) four cell lines expressing different Mdm4 levels were targeting p53 in these breast cancer cell lines (Figure 3). selected from our panel of eight breast cancer cell lines Nutlin-3-induced growth arrest/apoptosis was comple- expressing wild-type p53. These cell lines were tested for tely rescued in all cell lines with p53 knock down, which their drug sensitivity to six anti-cancer drugs, which can was accompanied by a strongly reduced or complete be divided in two groups: non-genotoxic experimental lack of induction of p53-target genes. In addition, knock compounds capable of disrupting the Mdm2–p53 down of p53 resulted in a partial, but significant rescue interaction (Nutlin-3 and RITA) and genotoxic agents of the doxorubicin-mediated growth inhibition in all used in chemotherapy (PALA, etoposide, doxorubicin, tested cell lines. Although the induction of p53 down- 5-FU). The cellular response to all tested drugs except stream targets by PALA, RITA, etoposide, and 5-FU for doxorubicin correlated to endogenous Mdm4 levels was strongly reduced on p53 knock down in all cell lines, in this panel of cell lines: ZR75-1 cells, which express rescue of drug-induced growth inhibition by p53 knock relatively low endogenous Mdm4 levels were most down was cell line dependent. sensitive to Nutlin-3, PALA, etoposide, and 5-FU, whereas breast cancer cells, which express high levels of Mdm4 were clearly less sensitive (Figures 2a–d). Effect of Mdm4 levels on sensitivity to Nutlin-3 In contrast, cells expressing high levels of Mdm4 were The results presented above suggest a correlation more sensitive to RITA treatment than cells with lower between endogenous Mdm4 levels in tumor cells, Mdm4 levels (Figure 2e). The response to doxorubicin inhibiting p53, and their sensitivity to Nutlin-3. To did not clearly correlate to endogenous Mdm4 levels, investigate this possibility in more detail, Mdm4 levels although again those cells expressing highest Mdm4 were manipulated by either knocking down endogenous levels were least sensitive (Figure 2f). Notably, the Mdm4 or by ectopic overexpression using lentiviral sensitivity to the Mdm2 inhibitor Nutlin-3 was inversely expression vectors. Ectopic overexpression of Mdm4 correlated with endogenous Mdm2 levels, that is cells hardly affected proliferation of these cell lines. In with relative high endogenous Mdm2 levels (ZR75-1) contrast, reducing Mdm4 levels significantly reduced were more sensitive to Nutlin-3 than cells with lower the proliferative capacity of all tested cell lines Mdm2 levels (ZR75-30) (Figure 2a). Endogenous Mdm2 (Supplementary Figure 2). In colony-forming assays, levels seemed to be less correlated to sensitivity to the we found that, as expected, growth inhibition by Mdm4 Mdm2 inhibitor RITA (Figure 2e). Treatment with each knock down is p53 dependent (data not shown). of these compounds resulted in induction of p53 and its Furthermore, reducing Mdm4 levels significantly ren- downstream target Mdm2, but the level of activation dered some cell lines more sensitive to Nutlin-3 varied depending on cell line and treatment (Figure 3). treatment, but in a cell line-dependent manner and not Nutlin-3 treatment resulted in increased levels of the correlating with initial Mdm4 levels. In addition, in cell anti-proliferation protein p21 and pro-apoptotic protein lines, which were sensitized for Nutlin-3 by Mdm4 PUMA in all tested cell lines, but importantly, no knock down, that is MCF-7 and ZR75-1, overexpres- detectable DNA damage response (Ser15 phosphoryla- sion of Mdm4 significantly reduced responsiveness to tion of p53) was observed. In contrast, PALA treatment Nutlin-3 (Figures 4a and c). Vice versa, Mdm4 over- had little effect on p21 and PUMA levels. RITA expression had no significant effect on Nutlin-3 sensi- treatment showed marginal effects on p53 downstream tivity of MPE600 and ZR75-30 cells, which also showed targets in ZR75-1 cells, which corresponds with the no effect on Mdm4 knock down (Figures 4b and d). The relative resistance in the growth assay. Surprisingly, observed differences in Nutlin-3-mediated effects on although ZR75-30 cells were strongly inhibited in manipulation of Mdm4 levels in proliferation assays growth by RITA, p21 and PUMA levels were not could not be directly correlated with a distinct regula- induced after treatment. Treatment with etoposide, tion of expression of the tested p53 targets. For example, doxorubicin, and 5-fluorouracil resulted in increased on knock down of Mdm4, p21 and PUMA levels were p21 and PUMA in all tested cell lines except in ZR75-30, both increased in MCF-7 and MPE600, although which might explain the relative resistance of this cell sensitization for Nutlin-3 in cell viability assays was line to these drugs. Decrease of Mdm4 levels was only observed in MCF-7 and not in MPE600 (Figures observed when cells were treated with genotoxic 4a and b). In ZR75-1, PUMA levels were increased on compounds and with RITA in most cell lines. Whether Mdm4 knock down (Figure 4c). Although ectopic the latter is also regulated through DNA damage- overexpression of Mdm4 resulted in decreased p21

Oncogene Regulation of drug response by Mdm4 S Lam et al 2418

Figure 2 Drug sensitivity of breast cancer cells to various anti-cancer drugs. MCF-7, MPE600, ZR75-1, and ZR75-30 were counted and seeded for proliferation assay. Cells were treated with (a)10mM Nutlin-3, (b) 100 mM PALA, (c)5mM etoposide, (d)10mM 5-ﬂuorouracil, (e) 0.1 mM RITA, or (f)1mM doxorubicin. Cell viability was measured 0, 24, 48, and 72 h after drug treatment using the WST-1 assay. Relative cell numbers were determined by calculating the OD450 of drug-treated cells relative to mock-treated cells of the same transduction from the same day of treatment. The results are expressed as mean±s.d. of triplicate cultures and representative of two independent experiments.

levels in all tested cell lines, PUMA levels were slightly suggest that Mdm4 protein levels can inﬂuence the decreased only in the two cell lines, which did not show sensitivity of cells to Nutlin-3 in a p53-dependent rescue of Nutlin-3-mediated growth inhibition on manner, but no direct correlation between Mdm4 levels Mdm4 overexpression. Although Nutlin-3-induced and drug sensitivity could be established. Mdm2 levels were enhanced in Mdm4 knock down cells, no correlation was found with Nutlin-3 hypersen- sitivity after Mdm4 knock down. Nevertheless, observed Effect of Mdm4 levels on Nutlin-3-induced apoptosis effects were dependent on p53, as speciﬁc knock down The distinct regulation of two main p53-targets p21 and of p53 protected cells against Nutlin-3. These results PUMA could not explain why sensitivity to Nutlin-3

Oncogene Regulation of drug response by Mdm4 S Lam et al 2419

Figure 3 p53-dependent response to various anti-cancer drugs. (a) MCF-7, (b) MPE600, (c) ZR75-1, and (d) ZR75-30 were infected with sh-control (open bars) or sh-p53 (closed bars) lentiviruses. Four days after transduction, cells were counted and seeded for total protein extraction and proliferation assay. Cells were treated with 10 mM Nutlin-3, 100 mM PALA, 0.1 6 mM RITA, 5 mM etoposide, 1 mM doxorubicin, or 10 mM 5-ﬂuorouracil. After 24 h, cells were harvested and analyzed for protein expression by western blotting. Alpha-tubulin expression was analyzed as loading control. Cell viability was measured 48 (doxorubicin) or 72 h after drug treatment using the WST-1 assay. Relative cell numbers were determined by calculating the OD450 of drug-treated cells relative to untreated cells of the same transduction from the same day of treatment. The results are expressed as mean±s.d. of triplicate cultures and representative of two independent experiments, *Po0.05, **Po0.01, ***Po0.001 compared with sh-control cells exposed to the same treatment.

Oncogene Regulation of drug response by Mdm4 S Lam et al 2420 was affected by manipulation of Mdm4 levels in some growth arrest and apoptosis, PARP cleavage was breast cancer cell lines, but not in others. As the cell investigated as a measurement for caspase activity proliferation assay could not distinguish between (Figure 5). PARP cleavage was signiﬁcantly increased

Oncogene Regulation of drug response by Mdm4 S Lam et al 2421 after Nutlin-3 treatment in all tested cell lines except In addition, reducing Mdm4 protein levels also did not MPE600. Interestingly, Mdm4 knock down resulted in a sensitize cells for 5-fluorouracil (data not shown). significant increase in the amount of cleaved PARP only in those cell lines, which were sensitized for Nutlin-3 in the growth assay, that is MCF-7 and ZR75-1. These Effect of Nutlin-3 and doxorubicin on cellular growth results suggest that Mdm4 levels can affect sensitivity to As knock down of p53 was only able to significantly Nutlin-3 by preventing activation of caspase-mediated rescue the growth inhibitory effects of Nutlin-3 and apoptosis. doxorubicin (Figure 3), these drugs were used for further study. Treatment with Nutlin-3 completely inhibited cell growth of MPE600, ZR75-1, and ZR75- Effect of Mdm4 levels on sensitivity to genotoxic drugs 30 cells, whereas growth of MCF-7 cells was severely The data presented in Figure 2 also suggest some slowed down (Supplementary Figure 3). Treatment with correlation between endogenous Mdm4 levels in tumor doxorubicin strongly reduced cell viability in all cell cells and their sensitivity to genotoxic agents. To lines. Similar to most chemotherapeutic agents, doxor- investigate this possibility in more detail, Mdm4 levels ubicin treatment can be accompanied by severe side were reduced by shRNA, and cells were treated with effects. This favors a reduction of the dose of doxorubicin. The growth of sh-ctrl cells is inhibited by doxorubicin to prevent the side effects without losing doxorubicin, but doxorubicin treatment of Mdm4 its effect as anti-cancer drug. As the action of knock down cells hardly resulted in further growth doxorubicin is partially dependent on activation of p53 inhibition (Figure 6). Although doxorubicin inhibited and the non-genotoxic agent Nutlin-3 is a potent cell growth stronger than Nutlin-3, no PARP cleavage inducer of p53, cells were treated with a combination was detected in MCF-7 and MPE600 after doxorubicin of both compounds. Combining Nutlin-3 and doxor- treatment (Figures 6a and b). In contrast, PARP ubicin indeed resulted in increased cell death, even with cleavage was observed after doxorubicin treatment in concentrations of doxorubicin that by itself had very ZR75-1 and ZR75-30. Interestingly, in ZR75-30 cells, little effect on cell survival (Supplementary Figure 3). PARP cleavage was even decreased in Mdm4 knock Moreover, the combined effect of Nutlin-3 and doxor- down cells compared with sh-ctrl cells, both after ubicin on cell viability was synergistic in MCF-7 and Nutlin-3 and doxorubicin treatment (Figures 6c and d). ZR75-1 (Figures 7a and c); in MPE600 and ZR75-30, This correlates with the relative resistance of Mdm4 the effect of the combination was additive (Figures 7b knock down ZR75-30 cells for doxorubicin treatment. and d). The observed effects on cellular level were

Figure 5 Effect of Mdm4 levels on PARP cleavage. MCF-7, MPE600, ZR75-1, and ZR75-30 were infected with sh-control, sh-Mdm4, or sh-p53 lentiviruses. Four days after transduction, cells were treated with 10 mM Nutlin-3. After 24 h of incubation, cells were harvested for total protein extracts. PARP cleavage was analyzed by western blot. Protein bands were quantiﬁed and normalized for g-tubulin expression. The results are expressed as mean±s.d. of three independent experiments, *Po0.05, **Po0.01, ***Po0.001 compared with mock-treated sh-control cells, NS ¼ not signiﬁcant.

Figure 4 Effect of Mdm4 levels on sensitivity to Nutlin-3. (a) MCF-7, (b) MPE600, (c) ZR75-1, and (d) ZR75-30 were infected with sh-control, sh-Mdm4, sh-p53, or Mdm4 expression lentiviruses. Four days after transduction, cells were counted and seeded for proliferation assay. After 24 h, cells were treated with 10 mM Nutlin-3. Cell viability was measured 0, 24, 48, and 72 h after drug treatment using the WST-1 assay. Relative cell numbers were determined by calculating the OD450 of drug-treated cells relative to mock-treated cells of the same transduction from the same day of treatment. The results are expressed as mean±s.d. of triplicate cultures and representative of at least three independent experiments, *Po0.05, **Po0.01, ***Po0.001 compared with sh-control cells at the same day. For the analysis of protein levels by western blot, cells were harvested after 24 h of treatment with 10 mM Nutlin-3. HAUSP expression was analyzed as loading control.

Oncogene 2422 Oncogene euaino rgrsos yMdm4 by response drug of Regulation Lam S tal et

Figure 6 Effect of Mdm4 levels on sensitivity to doxorubicin. (a) MCF-7, (b) MPE600, (c) ZR75-1, and (d) ZR75-30 were infected with sh-control, sh-Mdm4, sh-p53, or Mdm4 expression lentiviruses. Four days after transduction, cells were counted and seeded for proliferation assay. After 24 h, cells were treated with 1 mM doxorubicin or 10 mM Nutlin-3. Cell viability was measured 48 h after drug treatment using a WST-1 assay. Relative cell growth was determined by calculating the OD450 of t ¼ 48 h relative to t ¼ 0. The results are expressed as mean±s.d. of triplicate cultures and representative of at least three independent experiments, *Po0.05, **Po0.01, ***Po0.001 compared with sh-control cells of the same treatment. For the analysis of protein levels by western blot, cells were harvested after 24 h of drug treatment. HAUSP expression was analyzed as loading control. Regulation of drug response by Mdm4 S Lam et al 2423

Figure 7 Effect of Nutlin-3 and doxorubicin combination treatment on cell viability. (a) MCF-7, (b) MPE600, (c) ZR75-1, and (d) ZR75-30 were counted and seeded for proliferation assay. After 24 h, cells were treated with increasing concentrations of doxorubicin (0–1 mM) with or without 10 mM Nutlin-3. Cell viability was measured 72 h after drug treatment using a WST-1 assay. Relative cell number was determined by calculating the OD450 of doxorubicin-treated cells relative to mock-treated cells (line). For cells treated with a combination of drugs, relative cell number was compared with single drug treatments with Nutlin-3 (bars). The results are expressed as mean±s.d. of triplicate cultures and representative of at least three independent experiments, *Po0.01, **Po0.001 compared with cells treated with the same dose of doxorubicin without Nutlin-3. (e) For the analysis of protein levels by western blot, cells were harvested after 24 h of drug treatment. Protein bands were quantiﬁed and normalized for a-tubulin expression. Relative expression was calculated by comparing with mock-treated cells; pSer15-p53 and pSer46-p53 were compared with total p53 levels; ND ¼ not detectable.

accompanied with stronger induction of p53 and its Discussion downstream targets Mdm2, p21, and PUMA (Figure 7e). In addition, phosphorylation of p53 at In this study, we have investigated a putative function of Ser15 and Ser46, indication of, respectively, DNA Mdm4 protein levels on drug sensitivity in breast cancer damage and apoptotic responses, was increased in cells. With the use of a panel of breast cancer cell lines combination of Nutlin-3 and doxorubicin compared containing wild-type p53, but expressing different levels with single drug treatment. Moreover, in normal breast of Mdm4, we could show that endogenous Mdm4 levels epithelial cells, we did not observe any synergistic effect can affect the sensitivity of breast cancer cells to anti- of Nutlin-3 and doxorubicin on cell survival (Supple- cancer agents, which was dependent on an intact mentary Figure 4). Thus, in combination with Nutlin-3, apoptotic response of the cell lines. Furthermore, a low dose of doxorubicin was sufficient to achieve the treatment with the non-genotoxic agent Nutlin-3 can same effect as a higher single dose of doxorubicin. sensitizes cells for doxorubicin, showing that activation Moreover, treatment with this combination can result of p53 by targeting its regulators is an efficient strategy in a more efficient decrease in cell viability of breast to decrease cell viability of breast cancer cells. These cancer cells. results confirm an effect of Mdm4 protein levels on the

Oncogene Regulation of drug response by Mdm4 S Lam et al 2424 efficacy of chemotherapeutic agents to induce apoptosis p53 cells on Nutlin-3 treatment is dependent on pRB in a subset of cancer cells in a p53-dependent manner, expression (Kitagawa et al., 2008). As treatment with but additional, yet undetermined factors determine the Nutlin-3 strongly reduced E2F-1 levels in the breast final outcome of the treatment. cancer cells tested here (data not shown), an involve- Activation of p53 would be beneficial in tumors ment of E2F-1 activity in the distinct sensitivity of the harboring wild-type p53. Conventional chemotherapeu- cell lines for Nutlin-3 seems unlikely. tic agents cause DNA damage, thereby targeting Mdm2 In general, the p53 downstream targets p21 and and Mdm4 resulting in activation of p53 (Meulmeester PUMA were upregulated similarly on Mdm4 knock et al., 2005). Treatment of various breast cancer cell down, which can explain the growth inhibition of these lines with conventional genotoxic chemotherapeutic cells. However, this effect could not explain the agents indeed resulted in increased p53 levels with differential effect of Mdm4 manipulation on the concomitant upregulation of two important down- sensitivity to Nutlin-3. It has been suggested that high stream targets, PUMA and p21, thereby reducing cell levels of Mdm4 inhibit the p53-mediated induction of growth and viability. pro-apoptotic target genes, in that way preventing the Another approach to activate p53 in tumor cells is induction of apoptosis by Nutlin-3 (Wade et al., 2008). based on the fact that a large proportion of tumors still In this study, PARP cleavage was analyzed as marker expressing wild-type p53 show increased levels of Mdm2 for caspase-mediated apoptosis (Nicholson et al., 1995; or Mdm4, making those proteins attractive candidates Tewari et al., 1995; Oliver et al., 1998). PARP cleavage for the development of anti-cancer drugs. Recently, was indeed increased after Nutlin-3 treatment in three of Nutlin-3, RITA, and MI-219 have been identified in the four cell lines, suggesting that apoptosis was drug discovery screens for low molecular weight induced. Interestingly, knock down of Mdm4 increased compounds capable of disrupting the Mdm2–p53 PARP cleavage only in those cell lines, which were interaction (Issaeva et al., 2004; Vassilev et al., 2004; sensitized for Nutlin-3. These results suggest that Mdm4 Shangary et al., 2008). Whether Nutlin-3 can also indeed can protect cells from p53 dependent, Nutlin-3- disrupt the Mdm4–p53 interaction is still a matter of induced apoptosis, but in a cell line-dependent manner. debate. We show here, by manipulating Mdm4 levels in Conventional chemotherapy is associated with un- a number of breast cancer cell lines expressing wild-type desirable and severe side effects (Shapiro and Recht, p53, that high Mdm4 levels can blunt the sensitivity to 2001). This favors a reduction of the dose of the Nutlin-3, but in a strong cell line-dependent manner. chemotherapeutic agent to prevent the side effects Mdm4 protein levels per se have no predictive value without losing its effect as anti-cancer drug. Earlier regarding Nutlin-3 sensitivity. Clearly, factors addi- studies have shown that Nutlin-3 is a potent sensitizer tional to Mdm4 have a function in determining for standard radio- and chemotherapy (Barbieri et al., sensitivity to Nutlin-3. Interestingly, in this panel of 2006; Cao et al., 2006; Coll-Mulet et al., 2006; Patton cell lines, the sensitivity to RITA seemed to be inversely et al., 2006). Therefore, Nutlin-3 was combined with correlated with endogenous Mdm4 levels, but more doxorubicin, which resulted in increased cell death, even experiments are needed to validate this conclusion. with concentrations of doxorubicin that by itself had It has been reported that basal levels of Mdm2 are very little effect on cell survival. Thus, combined important in determining the Nutlin-3 response, that is treatment of doxorubicin with Nutlin-3 may allow dose cells highly expressing Mdm2 are more prone for reduction of the genotoxic compound to reduce side Nutlin-3-induced apoptosis, because of a more efficient effects, whereas retaining the same anti-cancer effect or degradation of Mdm4 (Tovar et al., 2006). Indeed, we may even result in a more efficient decrease in cell found that cells expressing highest levels of Mdm2 are viability of breast cancer cells. most sensitive to Nutlin-3 treatment, whereas cell lines In conclusion, we have shown that high Mdm4 levels with lower Mdm2 expression were less responsive. may render a subset of breast cancer cells more resistant However, knock down of Mdm4 resulted in a stronger to various anti-cancer agents, but additional, Mdm4- Mdm2 induction after Nutlin-3 treatment, whereas independent factors also influence the drug response. Mdm4 overexpression reduced Nutlin-3-induced Even so, targeting Mdm4 to sensitize tumor cells Mdm2, especially in cell lines in which Mdm4 manip- for conventional chemotherapeutic drugs might be a ulation had no influence on Nutlin-3 sensitivity. strategy to effectively treat tumors harboring wild- Furthermore, no correlation was found between down- type p53. regulation of Mdm4 and sensitivity on Nutlin-3 treatment. These findings confirm a recent report in which it was shown that neither endogenous Mdm2 nor Mdm4 Materials and methods levels in cancer cells correlate with Nutlin-3 sensitivity (Kitagawa et al., 2008). These authors also proposed Cell culture and drug treatments that E2F-1 activity is a critical determinant for Nutlin-3- MCF-7 cells were maintained in Dulbecco’s Modified Eagle’s Medium (DMEM) and MPE600, ZR75-1, and ZR75-30 were induced apoptosis, which would explain why retino- maintained in RPMI 1640 medium (Invitrogen, Carlsbad, CA, blastoma cell lines are sensitive to Nutlin-3-induced USA). All culture media were supplemented with 10% FCS apoptosis in spite of expressing high levels of Mdm4 and antibiotics. MCF-10A1 cells were maintained in DMEM/ (Laurie et al., 2006). Furthermore, these authors show F-12 supplemented with 5% horse serum (both purchased at that the downregulation of E2F-1 observed in wild-type Invitrogen), 20 ng/ml epidermal growth factor (Upstate

Oncogene Regulation of drug response by Mdm4 S Lam et al 2425 Biotechnology, Lake Placid, NY, USA), 100 mg/ml cholera non-fat dry milk, and subsequently incubated with the enterotoxin, 5 mg/ml hydrocortisone, 10 mg/ml insulin (all appropriate primary antibody. Membranes were washed three purchased at Sigma-Aldrich, St Louis, MO, USA). Cells were times with TBST and incubated with either horseradish seeded 24 h before treatment with Nutlin-3 (Cayman Chemi- peroxidase-conjugated goat anti-mouse or goat anti-rabbit cal, Ann Arbor, MI, USA), doxorubicin (Sigma-Aldrich), secondary antibody (Jackson Laboratories, West Grove, PA, N-phosphonacetyl-L-aspartate (PALA) (NCI), RITA (kind USA). After three times washing with TBST, proteins were gift of Dr G Selivanova), etoposide (Sigma-Aldrich), or 5- detected by enhanced chemiluminescense (Super Signal West fluorouracil (5-FU) (Sigma-Aldrich) at indicated concentra- Dura, Peirce Biotechnology, Rockford, IL, USA) and tions and time points. visualized by exposure to X-ray film (Fujifilm Europe GmbH, Du¨sselford, Germany) or with the use of the Chemigenius XE3 Plasmids and lentiviral infections (SynGene, Cambridge, UK). The following primary antibodies For the generation of a lentiviral expression vector for Mdm4, were used: rabbit anti-Mdm4 (BL1258, Bethyl Laboratories, HA-Mdm4 (de Graaf et al., 2003) was cloned into pLV-CMV- Montgomery, TX, USA), mouse anti-Mdm2 (mixture of 4B2 bc-GFP (Oruetxebarria et al., 2004) using the SalI and SnaBI (kind gift of Dr A Levine) and SMP14 (Santa Cruz restriction sites. To generate shRNA expression vectors Biotechnology, Santa Cruz, CA, USA), mouse anti-p53 (DO-1, against Mdm4 and p53 (Brummelkamp et al., 2002), oligonu- Santa Cruz Biotechnology), rabbit anti-phospho-p53 (Ser15 cleotides were cloned into the BglII and HindIII sites of the and Ser46, Cell Signaling Technologies, Beverly, MA, USA), pTER vector (van de Wetering et al., 2003) (Supplementary rabbit anti-PUMA (Sigma-Aldrich), mouse anti-p21 (CP74, Table 1 for sequences). Subsequently, fragments containing the Upstate Biotechnology), mouse anti-a-tubulin (Sigma- H1-promoter and cloned nucleotides were recloned into the Aldrich), mouse anti-g-tubulin (Sigma-Aldrich), rabbit anti- PstI site of the lentiviral pRRL-CMV-GFP vector (Carlotti HAUSP (BL851, Bethyl Laboratories), rabbit anti-PARP (Cell et al., 2004). Production of lentiviruses by transfection into Signaling Technology), mouse anti-pRB (G3-245, BD Phar- 293T cells has been described earlier (Carlotti et al., 2004). For mingen, San Diego, CA, USA), mouse anti-E2F-1 (clone lentiviral infections, cells were seeded 24 h before. To obtain KH95, Santa Cruz Biotechnology), rabbit anti-Cyclin D1 (06- Mdm4 overexpression, cells were infected with HA-tagged 137, Upstate Biotechnology), mouse anti-CDK4 (DCS-35, Mdm4-expressing lentiviruses. To obtain specific knock down, Abcam, Cambridge, MA, USA), and mouse anti-p16 (JC8, cells were infected with Mdm4- or p53–shRNA-expressing kind gift of Dr G Peters). The intensity of protein bands lentiviruses; shRNA specifically targeting murine Mdm4 was was determined by densitometry using GeneTools software used as negative control. For all infections, a multiplicity of (SynGene). Data analysis was performed by calculating infection of 1.5 was used. Cells were transduced overnight in protein levels relative to those of HAUSP or a-tubulin. the presence of 8 mg/ml polybrene (Sigma-Aldrich). The next day, the medium was replaced with fresh medium. Transduced Statistical analysis cells were seeded for experimental purposes 3 to 4 days post- Data are presented as means±s.d. of triplicate cultures and transduction. are representative of at least two independent experiments. Differences between various conditions were evaluated by Proliferation assay ANOVA with a Bonferroni correction for multiple compar- Cell viability was measured using a WST-1 assay (Roche ison. In some experiments, relative differences were tested with Biochemicals, Indianapolis, IN, USA). Briefly, cells were a two-tailed paired t-test. A value of Po0.05 was considered to seeded in 96-well flat-bottom plates (Greiner Bio-One, Frick- represent a significant difference. enhausen, Germany) in 100 ml of culture medium 24 h before drug treatments. Cell densities were as follows: MCF-7: 1500 cells/well, MPE600 and ZR75-1: 5000 cells/well, ZR75-30: 10 000 cells/well, and MCF-10A1: 1000 cells/well. After 0, 24, Conflict of interest 48, and 72 h of drug treatment, 10 ml of WST-1 reagent was added and plates were further incubated up to 4 h at 37 1C. The authors declare no conflict of interest. Absorbance was measured at 450 nm using a plate reader (Victor3 Multilabel Counter 1420-042, Perkin-Elmer, Well- esley, MA, USA). Acknowledgements Immunoblotting and antibodies Cells were lysed in Giordano buffer (50 mM Tris–HCl (pH 7.4), We thank Dr G Selivanova for providing RITA, Dr A Levine 250 mM NaCl, 0.1% Triton X-100, and 5 mM EDTA) with and Dr G Peters for providing anti-Mdm2 and anti-p16 15% glycerol and protease and phosphatase inhibitors. Protein antibodies, respectively. This study was supported by grants content was determined using Bradford Reagent (Biorad, from the Association for International Cancer Research (grant Hercules, CA, USA) and equal amounts of protein were 05-273) and by EC FP6 funding (contract 503576) to AG separated by SDS–PAGE. Proteins were transferred onto Jochemsen. This publication reflects the authors’ views and not polyvinyldene difluoride membranes (Immobilon-P, Millipore, necessarily those of the European Community. The EC is not Bedford, MA, USA). Membranes were blocked in Tris- liable for any use that may be made of the information buffered saline, 0.2% Tween-20 (TBST) containing 10% contained.

References

Barbieri E, Mehta P, Chen Z, Zhang L, Slack A, Berg S et al. chemotherapy-induced apoptotic cell death. Mol Cancer Ther 5: (2006). MDM2 inhibition sensitizes neuroblastoma to 2358–2365.

Oncogene Regulation of drug response by Mdm4 S Lam et al 2426 Brooks CL, Gu W. (2006). p53 Ubiquitination: Mdm2 and beyond. Patton JT, Mayo LD, Singhi AD, Gudkov AV, Stark GR, Jackson Mol Cell 21: 307–315. MW. (2006). Levels of HdmX expression dictate the sensitivity Brummelkamp TR, Bernards R, Agami R. (2002). A system for stable of normal and transformed cells to Nutlin-3. Cancer Res 66: expression of short interfering RNAs in mammalian cells. Science 3169–3176. 296: 550–553. Pereg Y, Lam S, Teunisse A, Biton S, Meulmeester E, Mittelman L Cao C, Shinohara ET, Subhawong TK, Geng L, Woon KK, Albert et al. (2006). Differential roles of ATM- and Chk2-mediated JM et al. (2006). Radiosensitization of lung cancer by nutlin, an phosphorylations of Hdmx in response to DNA damage. Mol Cell inhibitor of murine double minute 2. Mol Cancer Ther 5: 411–417. Biol 26: 6819–6831. Carlotti F, Bazuine M, Kekarainen T, Seppen J, Pognonec P, Maassen Pereg Y, Shkedy D, de Graaf P, Meulmeester E, Edelson-Averbukh JA et al. (2004). Lentiviral vectors efficiently transduce quiescent M, Salek M et al. (2005). Phosphorylation of Hdmx mediates its mature 3T3-L1 adipocytes. Mol Ther 9: 209–217. Hdm2- and ATM-dependent degradation in response to DNA Coll-Mulet L, Iglesias-Serret D, Santidrian AF, Cosialls AM, de Frias damage. Proc Natl Acad Sci USA 102: 5056–5061. M, Castano E et al. (2006). MDM2 antagonists activate p53 and Ramos YF, Stad R, Attema J, Peltenburg LT, van der Eb AJ, synergize with genotoxic drugs in B-cell chronic lymphocytic Jochemsen AG. (2001). Aberrant expression of HDMX proteins in leukemia cells. Blood 107: 4109–4114. tumor cells correlates with wild-type p53. Cancer Res 61: 1839–1842. Danovi D, Meulmeester E, Pasini D, Migliorini D, Capra M, Frenk R Riemenschneider MJ, Buschges R, Wolter M, Reifenberger J, Bostrom et al. (2004). Amplification of Mdmx (or Mdm4) directly contributes J, Kraus JA et al. (1999). Amplification and overexpression of the to tumor formation by inhibiting p53 tumor suppressor activity. MDM4 (MDMX) gene from 1q32 in a subset of malignant gliomas Mol Cell Biol 24: 5835–5843. without TP53 mutation or MDM2 amplification. Cancer Res 59: de Graaf P, Little NA, Ramos YF, Meulmeester E, Letteboer SJ, 6091–6096. Jochemsen AG. (2003). Hdmx protein stability is regulated by the Riemenschneider MJ, Knobbe CB, Reifenberger G. (2003). Refined ubiquitin ligase activity of Mdm2. J Biol Chem 278: 38315–38324. mapping of 1q32 amplicons in malignant gliomas confirms MDM4 Fuster JJ, Sanz-Gonzalez SM, Moll UM, Andres V. (2007). Classic as the main amplification target. Int J Cancer 104: 752–757. and novel roles of p53: prospects for anticancer therapy. Trends Mol Shangary S, Qin D, McEachern D, Liu M, Miller RS, Qiu S et al. Med 13: 192–199. (2008). Temporal activation of p53 by a specific MDM2 inhibitor is Haupt Y, Maya R, Kazaz A, Oren M. (1997). Mdm2 promotes the selectively toxic to tumors and leads to complete tumor growth rapid degradation of p53. Nature 387: 296–299. inhibition. Proc Natl Acad Sci USA 105: 3933–3938. Hollestelle A, Nagel JH, Smid M, Lam S, Elstrodt F, Wasielewski M Shapiro CL, Recht A. (2001). Side effects of adjuvant treatment of et al. (2009). Distinct gene mutation profiles among luminal-type and breast cancer. N Engl J Med 344: 1997–2008. basal-type breast cancer cell lines. Breast Cancer Res Treat (in press). Shvarts A, Bazuine M, Dekker P, Ramos YF, Steegenga WT, Merckx Hu B, Gilkes DM, Farooqi B, Sebti SM, Chen J. (2006). MDMX G et al. (1997). Isolation and identification of the human homolog overexpression prevents p53 activation by the MDM2 inhibitor of a new p53-binding protein, Mdmx. Genomics 43: 34–42. Nutlin. J Biol Chem 281: 33030–33035. Shvarts A, Steegenga WT, Riteco N, van Laar T, Dekker P, Bazuine Issaeva N, Bozko P, Enge M, Protopopova M, Verhoef LG, Masucci M et al. (1996). MDMX: a novel p53-binding protein with some M et al. (2004). Small molecule RITA binds to p53, blocks p53- functional properties of MDM2. EMBO J 15: 5349–5357. HDM-2 interaction and activates p53 function in tumors. Nat Med Stad R, Little NA, Xirodimas DP, Frenk R, van der Eb AJ, Lane DP 10: 1321–1328. et al. (2001). Mdmx stabilizes p53 and Mdm2 via two distinct Kitagawa M, Aonuma M, Lee SH, Fukutake S, McCormick F. (2008). mechanisms. EMBO Rep 2: 1029–1034. E2F-1 transcriptional activity is a critical determinant of Mdm2 Stad R, Ramos YF, Little N, Grivell S, Attema J, Der Eb AJ et al. antagonist-induced apoptosis in human tumor cell lines. Oncogene (2000). Hdmx stabilizes Mdm2 and p53. J Biol Chem 275: 27: 5303–5314. 28039–28044. Kubbutat MH, Jones SN, Vousden KH. (1997). Regulation of p53 Tewari M, Quan LT, O’rourke K, Desnoyers S, Zeng Z, Beidler DR stability by Mdm2. Nature 387: 299–303. et al. (1995). Yama/CPP32 beta, a mammalian homolog of CED-3, Laurie NA, Donovan SL, Shih CS, Zhang J, Mills N, Fuller C et al. is a CrmA-inhibitable protease that cleaves the death substrate (2006). Inactivation of the p53 pathway in retinoblastoma. Nature poly(ADP-ribose) polymerase. Cell 81: 801–809. 444: 61–66. Toledo F, Wahl GM. (2006). Regulating the p53 pathway: in vitro Meulmeester E, Pereg Y, Shiloh Y, Jochemsen AG. (2005). ATM- hypotheses, in vivo veritas. Nat Rev Cancer 6: 909–923. mediated phosphorylations inhibit Mdmx/Mdm2 stabilization by Tovar C, Rosinski J, Filipovic Z, Higgins B, Kolinsky K, Hilton H HAUSP in favor of p53 activation. Cell Cycle 4: 1166–1170. et al. (2006). Small-molecule MDM2 antagonists reveal aberrant Momand J, Zambetti GP, Olson DC, George D, Levine AJ. (1992). p53 signaling in cancer: implications for therapy. Proc Natl Acad Sci The mdm-2 oncogene product forms a complex with the p53 protein USA 103: 1888–1893. and inhibits p53-mediated transactivation. Cell 69: 1237–1245. van de Wetering M, Oving I, Muncan V, Pon Fong MT, Brantjes H, Nicholson DW, Ali A, Thornberry NA, Vaillancourt JP, Ding CK, van Leenen D et al. (2003). Specific inhibition of gene expression Gallant M et al. (1995). Identification and inhibition of the ICE/ using a stably integrated, inducible small-interfering-RNA vector. CED-3 protease necessary for mammalian apoptosis. Nature 376: EMBO Rep 4: 609–615. 37–43. Vassilev LT, Vu BT, Graves B, Carvajal D, Podlaski F, Filipovic Z Oliner JD, Kinzler KW, Meltzer PS, George DL, Vogelstein B. (1992). et al. (2004). in vivo activation of the p53 pathway by small-molecule Amplification of a gene encoding a p53-associated protein in human antagonists of MDM2. Science 303: 844–848. sarcomas. Nature 358: 80–83. Wade M, Rodewald LW, Espinosa JM, Wahl GM. (2008). BH3 Oliver FJ, de la RG, Rolli V, Ruiz-Ruiz MC, de Murcia G, Murcia activation blocks Hdmx suppression of apoptosis and cooperates JM. (1998). Importance of poly(ADP-ribose) polymerase and its with Nutlin to induce cell death. Cell Cycle 7: 1973–1982. cleavage in apoptosis. Lesson from an uncleavable mutant. J Biol Wade M, Wong ET, Tang M, Stommel JM, Wahl GM. (2006). Hdmx Chem 273: 33533–33539. modulates the outcome of p53 activation in human tumor cells. Oruetxebarria I, Venturini F, Kekarainen T, Houweling A, Zuijder- J Biol Chem 281: 33036–33044. duijn LM, Mohd-Sarip A et al. (2004). P16INK4a is required for Wasielewski M, Elstrodt F, Klijn JG, Berns EM, Schutte M. (2006). hSNF5 chromatin remodeler-induced cellular senescence in malig- Thirteen new p53 gene mutants identified among 41 human breast nant rhabdoid tumor cells. J Biol Chem 279: 3807–3816. cancer cell lines. Breast Cancer Res Treat 99: 97–101.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene