Regulation of MDM4 (MDMX) Function by P76mdm2: a New Facet in the Control of P53 Activity

Oncogene (2010) 29, 5935–5945 & 2010 Macmillan Publishers Limited All rights reserved 0950-9232/10 www.nature.com/onc ORIGINAL ARTICLE Regulation of MDM4 (MDMX) function by p76MDM2: a new facet in the control of p53 activity

S Giglio1,2,6, F Mancini1,2,6, M Pellegrino1, G Di Conza1,3, E Puxeddu4, A Sacchi5, A Pontecorvi2 and F Moretti1

1Institute of Neurobiology and Molecular Medicine, CNR/Fondazione Santa Lucia, Roma, Italy; 2Institute Medical Pathology, Catholic University, Roma, Italy; 3Department of Experimental-Clinical Medicine and Pharmacology, University of Messina, Messina, Italy; 4Department of Internal Medicine, University of Perugia, Perugia, Italy and 5Laboratory Molecular Oncogenesis, Regina Elena Cancer Institute, Roma, Italy

Under basal growth conditions, p53 function is tightly viability. P53 basal activity is therefore strictly con- controlled by the members of MDM family, MDM2 and trolled to avoid unwarranted anomalies of cell growth. MDM4. The Mdm2 gene codes, in addition to the full-length Moreover, levels of p53 basal activity control the p90MDM2, for a short protein, p76MDM2 that lacks the p53- magnitude of its stress-induced biological responses, binding domain. Despite this property and at variance with including those raised by oncogenic stimuli (Wang et al., p90MDM2, this protein acts positively toward p53, although 2009). In normal growing cells, p53 is regulated in a the molecular mechanism remains elusive. Here, we report non-redundant manner by MDM family members, that p76MDM2 antagonizes MDM4 inhibitory function. We MDM4 and MDM2 (Marine et al., 2006). Although show that p76MDM2 possesses intrinsic ubiquitinating and contribution of each of these proteins has not yet been degrading activity, and through these activities controls completely resolved, their activity results in the control MDM4 levels. Furthermore, the presence of p76MDM2 of p53 levels and function. An exception in the MDM decreases the association of MDM4 with p53 and group is represented by p76MDM2, an MDM2 protein p90MDM2, and antagonizes p53 degradation by the hetero- that acts positively toward p53 (Perry et al., 2000). dimer MDM4/p90MDM2.Thep76MDM2-mediated regulation P76MDM2 derives from alternative translation of full- of MDM4 occurs in the cytoplasm, under basal growth length MDM2 mRNA. Indeed, MDM2, both human conditions. Conversely, upon DNA damage, phosphorylation and mouse, gives rise to two major proteins starting of MDM4Ser403 dissociates p76MDM2 and prevents MDM4 from two different AUGs and differing in molecular degradation. The overall negative control of MDM4 by weight, p90MDM2 and p76MDM2 (Saucedo et al., 1999; p76MDM2 reflects on p53 function as p76MDM2 impairs Cheng and Cohen, 2007). P90MDM2, the more abundant MDM4-mediated inhibition of p53 activity. In agreement form, binds p53 through its amino-terminus and withthepositiveroleofp76MDM2 toward p53, the p76MDM2/ controls both p53 protein levels, through its E3 p90MDM2 ratio significantly decreases in a group of thyroid ubiquitin ligase activity, as well as p53 transcriptional tumor samples compared with normal counterparts. Overall, function (Marine and Lozano, 2010). P76MDM2 lacks the these findings reveal a new mechanism in the control of p53 first N-terminal 49 amino acids, in which part of the basal activity that may account for the distinct sensitivity of p53-binding domain resides, and cannot bind p53. tissues to stress signals depending on the balance among Nonetheless, it is able to affect p53 activity by stabilizing MDM proteins. Moreover, these data suggest an onco- its protein levels and favoring its transcriptional suppressive function for a product of the Mdm2 gene. function (Perry et al., 2000). Thus far, p76MDM2 function Oncogene (2010) 29, 5935–5945; doi:10.1038/onc.2010.324; has been attributed to its ability to inhibit p90- published online 9 August 2010 degradative function, although neither association of p90MDM2 with p53 nor total levels of p90MDM2 result Keywords: p53; MDM4; MDM2; p76MDM2; p90MDM2 decreased by p76MDM2 (Perry et al., 2000). Despite the lack of a clear molecular function, in vivo data suggest that p76MDM2 represents an important MDM2 Introduction factor in the control of p53 activity. Indeed, p90 is more abundant than p76MDM2 in the brain, heart and MDM2 MDM2 The activation of the oncosuppressor p53 leads to kidney, whereas p76 and p90 are roughly different outcomes, all strongly affecting cell growth and equivalent in the spleen, thymus and intestine, tissues where p53 rapidly accumulate and is active in response to DNA damage (Perry et al., 2000). In addition, a Correspondence: Dr F Moretti, CNR-Institute of Neurobiology and recent report has shown increased expression of p76MDM2 Molecular Medicine, Via del Fosso di Fiorano, Roma 64-00143, Italy. associated with enhanced activity of p53 in the E-mail: [email protected] 6These authors contributed equally to this work. epidermal vitiligo lesions (Salem et al., 2009), suggesting MDM2 Received 10 December 2009; revised 14 May 2010; accepted 28 June a function for p76 in some pathological condi- 2010; published online 9 August 2010 tions too. p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5936 MDM family includes MDM4 (also known as MDMX) too. MDM4 is an inhibitor of p53 activity under normal growth conditions (Marine et al., 2006; Wang et al., 2009), whereas it exerts a positive function toward p53-mediated apoptosis upon lethal DNA damage (Mancini and Moretti, 2009). Similar to p90MDM2, MDM4 inhibits p53 by controlling its protein levels and transcriptional function. However, MDM4 lacks E3 ubiquitin ligase activity and its ability to control p53 protein levels is mediated by heterodimer- ization with p90MDM2, which as a consequence acquires increased degradative potential toward p53 (Linares et al., 2003; Kawai et al., 2007; Linke et al., 2008; Okamoto et al., 2009). Although p76MDM2 shares with p90MDM2 the same protein domain at which levels MDM4 heterodimerizes, the function of p76MDM2 toward MDM4 in the regulation of p53 has not been investigated. In this work, we provide evidence that p76MDM2 antagonizes MDM4 activity. In particular, we show that p76MDM2 controls MDM4 protein levels by ubiquitinating and degrading it. Further, we showed that p76MDM2 impairs the association of MDM4 with p90MDM2 and p53, thereby reducing the degradative activity of the heterodimer toward p53. Most impor- tantly, the ratio of p76MDM2 to p90MDM2 levels was significantly decreased in a group of human thyroid tumors compared with matched normal tissues, supporting the role of p76MDM2-positive activity toward p53.

Results

P76MDM2 counteracts MDM4 activity on p53 independently of p90MDM2 P76MDM2, which does not bind p53, acts positively on p53 activity (Perry et al., 2000; Cheng and Cohen, 2007). P76MDM2 activity has been related to its ability to antagonize the degradative function of p90MDM2. Its relationship with MDM4 has not been investigated. Here, we tested the consequences of p76MDM2 expression on MDM4-mediated inhibition of p53 transcriptional activity in Mdm2À/ÀMdm4À/Àp53À/ÀMEFs (TKO- MEFs). In agreement to that described previously, MDM4 inhibits p53 transactivating function, although less strongly than p90MDM2 (Figure 1a). The expression MDM2 of p76MDM2 per se did not alter p53 activity; however, it Figure 1 76 retrieves MDM4 inhibitory activity toward p53. (a) Transient transfection of Mdm4À/ÀMdm2À/Àp53À/ÀMEFs (TKO- partly prevented MDM4-mediated inhibition of p53 MEFs) with the indicated plasmids plus PG13Luc reporter and activity, indicating its ability to antagonize MDM4 b-gal plasmids. Cells were assayed 24 h after transfection. The data repression, independently of the presence of p90MDM2 are representative of two experiments performed in duplicate. (Figure 1a). In accordance with Perry’s work, p76MDM2 (b) qRT–PCR of p21 and Noxa mRNAs upon transient transfec- MDM2 tion of indicated plasmids. The data are representative of three antagonizes p90 -mediated inhibition of p53 too experiments. Bars represent the mean and line the standard (Perry et al., 2000). Similar results were obtained in deviation. (c) Western blot (WB) of indicated proteins in cell Mdm2À/Àp53À/ÀMEFs (DKO-MEFs) (data not shown). lysates collected from TKO-MEFs transiently transfected with Luciferase data were conﬁrmed by quantitative real-time equimolar amounts of indicated plasmids. PCR (qRT–PCR) analysis of endogenous p21 and Noxa, two p53-induced targets in MEFs (Francoz The antagonism of p76MDM2 toward p90MDM2 has been et al., 2006). MDM4 represses moderately p53 activa- attributed to the inhibition of p90MDM2-mediated degra- tion of p21 and Noxa and p76MDM2 counteracts such dation of p53. However, MDM4 does not possess repression (Figure 1b). degradative function. Indeed, p53 protein levels did not

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5937 vary upon co-expression of p76MDM2 and MDM4 in fected with HA-Ub, MDM4, and p76MDM2 or p90MDM2, comparison with MDM4 alone (Figure 1c, lanes 6 and 5). and cell lysates immunoprecipitated with anti-HA anti- Conversely, MDM4 levels showed a strong decrease body (Figure 2b). In the presence of p76MDM2, there was associated with the expression of p76MDM2 independently a definite increase in MDM4 ubiquitinated forms of p53 (Figure 1c, compare lanes 6 and 5 with 7 and 1), associated with the downregulation of its levels. This suggesting that p76MDM2 activity toward p53 may be ubiquitination pattern was further enhanced by the mediated by downregulation of MDM4 protein levels. addition of proteasome inhibitor MG132, indicating that p76MDM2 possesses intrinsic ubiquitinating activity and that mediates downregulation of MDM4. MDM4 P76MDM2 degrades and ubiquitinates MDM4 ubiquitinated forms were more pronounced in the To ascertain the ability of p76MDM2 to control MDM4, presence of p76MDM2 than of p90MDM2, indicating higher independently of p90MDM2, MDM4 protein stability was efficiency of p76MDM2. assessed by cycloheximide experiments in the absence or To further confirm p76MDM2 activity toward MDM4, presence of p76MDM2. The expression of p76MDM2 caused we analyzed whether the presence of the deubiquitinating a strong decrease of MDM4 steady-state levels enzyme HAUSP could antagonize MDM4 downregula- (Figure 2a, left and right panels), indicating that tion by p76MDM2 (Meulmeester et al., 2005). Indeed, p76MDM2 affects MDM4 stability. co-expression of HAUSP counteracted p76MDM2-mediated P90MDM2 is a RING finger E3 ubiquitin ligase able to degradation of MDM4, confirming the role of ubiquitina- degrade different targets, including MDM4. P76MDM2 tion in MDM4 degradation (Figure 2c). In comparison, maintains an intact RING finger domain identical to p90MDM2 does not degrade MDM4 and co-expression of that of p90MDM2 and may therefore exert the same HAUSP does not alter MDM4 protein levels. function, although this has never been proven. To These data indicate that p76MDM2 possesses intrinsic ascertain this hypothesis, p76MDM2 was challenged in an ubiquitination function and that through this activity in vivo ubiquitination assay. DKO-MEFs were trans- controls MDM4 protein levels.

Figure 2 P76MDM2 ubiquitinates and destabilizes MDM4 protein levels. (a) Left panel: WB analysis of DKO-MEFs transfected with the indicated plasmids and treated after 24 h with cycloheximide (50 mg/ml) for the indicated times. Right panel: Quantiﬁcation of MDM4 levels related to tubulin levels as present in the right panel. (b) DKO-MEFs were transfected with MDM4 and HA-ubiquitin, plus p90MDM2 or p76MDM2 plasmids. After 16 h, they were treated with MG132 (10 mM) for 8 h and harvested. The upper panels show the ubiquitination pattern obtained after immunoprecipitation with anti-HA antibody and immunostaining with anti-MDM4 antibody. The lower panels show the total levels of indicated proteins in whole-cell extract (WCE). (c) WB analysis of WCEs collected from DKO-MEFs transiently transfected with equimolar amounts of indicated plasmids.

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5938

Figure 3 P76MDM2 impairs MDM4/p90MDM2/p53 association. (a) WB analysis of cell lysate collected from DKO-MEFs transiently transfected with equimolar amounts of the indicated plasmids. (b) Immunoprecipitation and WB analysis of cell lysates collected from DKO-MEFs transiently transfected with the indicated plasmids; 1 Â and 2 Â indicate the same and double amount of p76MDM2 with MDM2 MDM2 MDM2 1 respect to p90 .(c) DKO-MEFs were transiently transfected with equimolar amounts of MDM4, p90 and p76 and 4 molar amounts of p53 plasmids. Left panel: 500 mg of WCEs were immunoprecipitated with anti-p53 antibody and co- immunoprecipitated proteins were analyzed by WB. Right panel: WB analysis of indicated proteins in the WCEs used for immunoprecipitation.

P76MDM2 impairs MDM4/p90MDM2/p53 association p76MDM2 or p90MDM2 (Figure 3a). As p54MDM4 lacks the Various studies have reported the control of MDM4 RING ﬁnger domain through which MDM4 interacts protein levels by p90MDM2 upon DNA damage (Kawai with the MDM2 proteins, these data indicate the et al., 2003; Pan and Chen, 2003; Chen et al., 2005; requirement of the association between MDM4 and Okamoto et al., 2005; Pereg et al., 2005). As p76MDM2 p76MDM2 for the activity of this last. Similar results were and p90MDM2 are present simultaneously in the cell, we obtained in TKO-MEFs (data not shown). analyzed whether p76MDM2 activity results in cooperation These data indicate that p76MDM2 controls MDM4 with p90MDM2. To this aim, the effects of p76MDM2 toward independently of p90MDM2 and more efﬁciently. Further- MDM4 with respect to p90MDM2 or in combination with more, they suggest that the presence of p90MDM2 it were compared. As reported previously, expression of antagonizes rather than cooperates with p76MDM2. p76MDM2 in DKO-MEFs strongly decreased MDM4 Accordingly, binding of p76MDM2 and p90MDM2 to protein levels (Figure 3a). In comparison, expression of MDM4 was mutually exclusive as increased amounts p90MDM2 did not do so. Co-expression of p90MDM2 and of p76MDM2 lead to a decrease of p90MDM2 associated with p76MDM2 did not lead to a further decrease of MDM4 MDM4 (Figure 3b). In agreement to that described levels, rather to an increase, indicating that the two previously, the 30-tagged MDM4-HA was resistant to MDM2 proteins do not synergize in controlling MDM4 degradation (Pan and Chen, 2003). amount. In comparison with full-length protein, levels Recent reports hypothesize that the association of the MDM4 carboxy-terminus truncated form, between MDM4 and p90MDM2 stabilizes their complex p54MDM4 (Gentiletti et al., 2002), were not altered by and that heterodimer MDM4/p90MDM2 is the prevalent

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5939

Figure 4 Endogenous p76MDM2 controls cytoplasmic levels of MDM4. (a) WB analysis of nuclear and cytoplasmic fractions of cell lysate collected from DKO-MEFs transiently transfected with equimolar amounts of indicated plasmids. Numbers below the blot indicate the ratio of densitometric values of cytoplasmic MDM4/tubulin signal. (b) WB analysis of cytoplasmic fractions of DKO-MEFs transfected as in (a) and after 16 h treated with MG132 (10 mM) for additional 8 h. (c) WB analysis of cell lysate collected from MCF10A cells transfected with siCTRL (À) or siMDM2 ( þ ) for 24 h and then treated with MG132 (25 mM) for additional 6 h. Asterisk marks an unspecific band. degrading factor of p53 (Kawai et al., 2007). We et al., 2002; Mancini et al., 2009b) and its localization is therefore investigated whether the decreased association not modified by the presence of p76MDM2 or p90MDM2.As of MDM4 to p90MDM2, in the presence of p76MDM2, observed previously, its levels are decreased upon co- interferes with the association of MDM4 to p53 as well. expression with p76MDM2, but not with p90MDM2 Analysis of p53 complexes indeed showed that p76MDM2 (Figure 4a), and recovered by MG132 treatment reduced the amount of MDM4 associated with p53 in (Figure 4b). These data indicate that p76MDM2 does not the presence of p90MDM2 (Figure 3c, left panel, lanes substantially change MDM4 intracellular localization 3 and 6). This in turn results in increased p53 levels and that its activity occurs in the cytoplasm. They (Figure 3c, right panel, lanes 3 and 6), suggesting that further exclude that the inefficacy of p90MDM2 is owing to squelching MDM4 out of the complex decreases the a different localization. degradative function of the heterodimer. Strikingly, the At this point, we ascertained whether endogenous expression of sole p76MDM2 does not affect association of p76MDM2 actually controls MDM4 levels. To this MDM4 with p53 in comparison with p90MDM2 that purpose, we interfered with p76MDM2 expression by small conversely strongly stabilizes it (Figure 3c, left panel, interfering RNA (siRNA) in MCF10A, a non-trans- lanes 2 and 3), confirming the destabilizing effect of formed cell line, and analyzed MDM4 protein levels. p76MDM2 toward MDM4/p53/p90MDM2 complex. In con- Given the common mRNA, siRNA targeted both trast, as reported previously (Perry et al., 2000), p76MDM2 p76MDM2 and p90MDM2. The decrease of p76MDM2,as did not decrease the total amount of p90MDM2 (Figure 3c, specifically detected by comparison of 2A10 antibody right panel, lanes 4 and 5) or its fraction associated with signal (able to detect p76MDM2) with that of 4B2 (unable p53 (Figure 3c, left panel, lanes 4 and 5), supporting the to detect p76MDM2), correlated with a strong increase of hypothesis that impairment of p90MDM2-mediated MDM4 protein levels (Figure 4c). This increase was degradation of p53 is owing to the displacement of reduced in the presence of MG132, confirming that the MDM4 from the complex. modulation occurs at protein levels. The siRNA reduced the levels of p90MDM2 as well; however, on the basis of P76MDM2 controls cytoplasmic MDM4 basal levels previous and published data (Meulmeester et al., 2005), Given the ability of p76MDM2 to affect MDM4 but not we exclude a relevant activity of p90MDM2. Interestingly, p90MDM2, and the inefficacy of p90MDM2 toward MDM4, p53 protein levels result in an increase, as expected from we asked whether p76MDM2 function may be mediated by the downregulation of p90MDM2, but this did not result in a different localization with respect to p90MDM2.We a detectable increase of its targets p21 or Bax, suggesting analyzed nuclear and cytoplasmic fractions of DKO- that its transcriptional function is impaired in agreement MEFs expressing MDM4, p76MDM2 and p90MDM2. with the simultaneous increase of MDM4 levels. P90MDM2 is present in both nuclear and cytoplasmic Overall, these data confirm that p76MDM2 is an compartments similar to p76MDM2 (Figure 4a). MDM4 important factor in the positive control of p53 activity localizes almost exclusively in the cytoplasm (Migliorini through the regulation of MDM4.

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5940 18 MDM4 essentially following DNA damage and this x 95% CI Notched Outlier event is considered an important step in the activation of Boxplot 16 95% CI Mean Diamond p53 function (Kawai et al., 2003; Chen et al., 2005; x Okamoto et al., 2005; Pereg et al., 2005). We therefore P=0.040 Outliers>1.5 and <3 IQR 14 ascertained the effectiveness and the potential coopera- MDM2 MDM2 Outliers>3 IQR tion of p76 with p90 under these conditions. MDM2 12 DKO-MEFs expressing MDM4 and p76 or p90MDM2 were grown in the absence or presence of adriamycin (Adr) for 8 h and then analyzed for MDM4 10 + x protein levels. In the absence of Adr, MDM4 levels were strongly 8 reduced by the presence of p76MDM2 but not of p90MDM2, P1/P2 mRNA Ratio confirming previous results (Figure 6a). Conversely, in 6 ‡ the presence of Adr, p76MDM2 did not reduce to a similar extent MDM4 levels (Figure 6a, lanes 7 and 5), whereas 4 p90MDM2 did it efficiently (Figure 6a, lanes 6 and 5). Co-expression of the two proteins did not synergize 2 in MDM4 degradation, similar to that observed previously (see Figure 3a). These data indicate that 0 p76MDM2 is ineffective in the control of MDM4 upon CTRL Tumors DNA damage and suggest a distinct control of MDM4 Figure 5 Vertical box–whisker plots showing P1/P2 mRNA ratio by p76MDM2 and p90MDM2, depending upon cell growth in matched normal thyroid samples (CTRL) compared with tumor conditions. samples. Each plot shows graphically the central location and MDM2 scatter/dispersion of the values of each group: the line series shows The inefficacy of p90 to control MDM4 under parametric statistics (mean and confidence interval of mean), the normal growth conditions is correlated with its associa- notched box and whiskers show non-parametric statistics (median, tion to HAUSP that counteracts p90MDM2-mediated confidence interval of median and interquartile range). Crosses and degradation (Meulmeester et al., 2005). We therefore circles indicate possible outliers, between 1.5 and 3 interquartile asked whether the switch of MDM4 degradation range and over 3 interquartile range, respectively. P-value was MDM2 MDM2 calculated according to paired t-test. from p76 to p90 is owing to a different association of HAUSP with these proteins. However, the complexes between HAUSP and MDM4 are not substantially altered in the presence of p76MDM2 or Human tumors are usually characterized by reduced p90MDM2, nor any difference was found in the association p53 function. Given the p76MDM2-positive activity HAUSP/p76MDM2 or HAUSP/p90MDM2 (Figure 6b), sug- towards p53, we wondered whether p76MDM2 levels are gesting that HAUSP is not a determinant factor in altered in human tumors. To this aim, a set of human the switch of MDM4 degradation between p76MDM2 thyroid tumor samples characterized by wt-p53 were and p90MDM2. compared with matched normal thyroid tissues (Pro- To further understand the molecular mechanism that dosmo et al., 2008). Although p76MDM2 and p90MDM2 controls this switch, we analyzed the fraction of p76MDM2 share the same mRNA, transcription initiated at the and p90MDM2 bound to MDM4. NIH3T3 cells that promoter P1 of human MDM2 has the ability to endogenously express detectable levels of p76MDM2 and generate both MDM2 protein isoforms, whereas trans- p90MDM2 (Perry et al., 2000) and stably express a cription initiated at the promoter P2 produces prefer- doxycycline-inducible MDM4 (NIH3T3-MDM4) were entially full-length p90MDM2 (Cheng and Cohen, 2007). used (Gentiletti et al., 2002). During Adr treatment, As P1-driven transcript lacks exon 2, whereas P2-driven p90MDM2 levels were strongly upregulated, whereas those transcript lacks exon 1 (Cheng and Cohen, 2007), P1 of p76MDM2 increased to a lesser extent (Figure 6c, and P2 transcripts were analyzed by appropriate primers right panel), in accordance with Saucedo’s work (1999). sets and P1/P2 ratio calculated in tumor samples Upon immunoprecipitation of cytoplasmic MDM4, a compared with normal tissues. Interestingly, a signifi- change in the proportion of the two MDM2 proteins cant decrease of P1/P2 ratio was observed in thyroid associated with MDM4 occurred. In the absence of tumors compared with normal thyroid tissues (Figure 5), Adr, MDM4 is bound to p76MDM2 more abundantly indicating that p76MDM2 levels decrease in comparison than to p90MDM2, whereas in the presence of Adr, the with those of p90MDM2 in human tumors. These data levels of p76MDM2 bound to MDM4 decreased further support the hypothesis that p76MDM2 is a positive (Figure 6c, left panel). At the latest time point, 16 h, regulator of p53 function. an increase of the p76MDM2 levels bound to MDM4 was detected again. These observations suggest that the different regulation of MDM4 by p76MDM2 is associated P76MDM2 and p90MDM2 degrade MDM4 under different with a change in their association. On the contrary, growth conditions under the same conditions, the fraction of MDM4 Previous data indicate that p76MDM2 controls MDM4 associated with p90MDM2 does not substantially levels more efficiently than p90MDM2. P90MDM2 degrades change.

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5941

Figure 6 p76MDM2 and p90MDM2 bind differently MDM4 under stress conditions. (a) WB analysis of WCEs collected from DKO-MEFs transiently transfected with equimolar amounts of indicated plasmids, and grown for 8 h in the presence or absence of Adr. (b) DKO- MEFs were transiently transfected with equimolar amounts of plasmids expressing the indicated proteins. Five hundred micrograms of WCEs were immunoprecipitated with anti-HAUSP antibody and co-immunoprecipitated proteins were analyzed by WB. (c) NIH3T3 stably expressing inducible MDM4 for 24 h and treated with Adr (0.5 mM) were collected at the indicated time points. Right panel: 400 mg of cytoplasmic fraction were immunoprecipitated with anti-MDM4 antibody and proteins in the immunocomplex were analyzed by WB. The numbers below the blot represent the ratio of densitometric values of p76MDM2/p90MDM2 signals. Left panel: WB analysis of indicated proteins in WCEs.

Phosphorylation of MDM4 at Ser403 regulates association, and confirm that p76MDM2 is a relevant its association with p76MDM2 regulator of basal levels of MDM4. Upon DNA damage, phosphorylation of MDM4 at specific residues occurs. As S403 residue is an important phosphorylation site (Chen et al., 2005; Pereg et al., 2005), its role in the association of p76MDM2 to MDM4 Discussion was investigated. Under non-stress conditions, p76MDM2 co-immunoprecipitates wt-MDM4 and mutant Under normal growth conditions, p53 activity is tightly MDM4S403A in a similar way (Figure 7a). Conversely, controlled by MDM family members, MDM2 and following Adr treatment, immunoprecipitation of MDM4, and their derivative forms (Marine et al., p76MDM2 was able to co-immunoprecipitate MDM4S403A 2006; Toledo and Wahl 2007; Mancini et al., 2009a). All more efficiently than wt-MDM4 (Figure 7a). The MDM members exert a negative function toward p53, reciprocal co-immunoprecipitation confirmed these data by controlling its transcriptional function and/or protein (data not shown), indicating that phosphorylation of levels. An exception is represented by p76MDM2, a protein MDM4 at S403 have an important role in the dissociation deriving from translation of an internal AUG of the of MDM4 from p76MDM2. To confirm these data, we MDM2 full-length mRNA. For this reason, p76MDM2 tested the ability of p76MDM2 to degrade MDM4S403A lacks the first 49 amino acids and is unable to bind p53. upon DNA damage. Indeed, MDM4S403A levels were Despite this property, p76MDM2 acts positively on p53 decreased by p76MDM2 independently of cell growth and its presence has been associated to increased conditions, whereas in comparison wt-MDM4 levels were p53 activity in both normal tissues (Perry et al., 2000) decreased only in the absence of stress (Figure 7b). As and human pathological conditions (Salem et al., 2009), control, p90MDM2 degrades wt-MDM4 only in the presence suggesting a potential role for this factor in the of Adr, whereas loses its activity towards MDM4S403A, upregulation of p53 function. as reported previously (Chen et al., 2005; Pereg et al., Thus far, the molecular mechanism that underlies 2005). p76MDM2 function has been ascribed to the antagonism Overall, these results indicate that phosphorylation of toward its full-length partner, p90MDM2, particularly MDM4 at Ser403 is an important factor in mediating toward the degradative activity of this. However, this p76MDM2 activity toward MDM4 by controlling their antagonism remains to be elucidated since p76MDM2 does

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5942

Figure 7 Phosphorylation of MDM4 at Ser 403 impairs its MDM2 MDM2 Figure 8 Proposed model of p76 function. The presence of binding to p76 and subsequent degradation. (a) DKO-MEFs p76MDM2 leads to p53 activation by both decreasing MDM4 protein were transiently transfected with equimolar amounts of murine MDM2 MDM2 MDM2 levels and squelching MDM4 out of the MDM4/p90 /p53 p76 , p90 and human wtMDM4 or the phosphorylation complex. mutant MDM4S403A, and grown in the presence of MG132 (10 mM) for 4 h and in the absence or presence of Adr (0.5 mM) for 8 h. WCEs (250 mg) were immunoprecipitated with anti-MDM4 antibody and co-immunoprecipitated proteins analyzed by WB. cytoplasmic fraction of MDM4, relief of p53 function Left panel: WB analysis of 1/15 of WCEs used in immunopreci- may result from both an increase of p53 nuclear fraction pitation. (b) WB analysis of cells transfected as in (a) and treated and from the global reduction of MDM4 levels. with Adr (0.5 mM) for 8 h. Interestingly, our data show that p76MDM2 does not control MDM4 levels following DNA damage. Upon not squelch p90MDM2 out from association with p53, nor stress conditions, phosphorylation of MDM4 at Ser403 decreases p90MDM2 levels. causes dissociation of p76MDM2 from MDM4 and Present data provide new molecular insights into decreased degradation of the last. Accordingly, the p76MDM2 activity. Indeed, we showed that p76MDM2 exerts phosphorylation-deficient mutant MDM4S403A is still a tight control toward the other member of MDM degraded by p76MDM2 following DNA damage. On the family, MDM4. In particular, p76MDM2 regulates contrary, the same mutation impairs the degradation of MDM4 protein levels and impairs its association with MDM4 by p90MDM2, indicating that p76MDM2 and p53/p90MDM2 complex. This ultimately results in de- p90MDM2 control MDM4 levels in a complementary creased p53 degradation and an increase of its tran- way: p76MDM2 during cell proliferation and p90MDM2 scriptional function. following DNA damage. A reason for this switch may In normal growing cells, control of p53 protein levels reside in the need of controlling MDM4 in different cell resides almost exclusively in the ubiquitin ligase compartments. In fact, although MDM4 is mainly a p90MDM2. The association of MDM4 with p90MDM2 is cytoplasmic protein, after DNA damage a fraction of it considered an important event in directing the p90MDM2 translocates into the nucleus (Li et al., 2002; LeBron activity specifically toward p53, pointing to the hetero- et al., 2006; Wang et al., 2007), in which endogenous dimer MDM4/p90MDM2 as the main controller of p53 p90MDM2 is more abundant. levels under normal growth conditions (Kawai et al., Overall, the consequences of p76MDM2-mediated regu- 2007). The ability of p76MDM2 to decrease both MDM4 lation of MDM4 are the attenuation of MDM4 and levels and its association with the p53/p90MDM2 complex MDM2 inhibitory activity toward p53, pointing to leads ultimately to imbalance in the function of the p76MDM2 as an important factor in the positive control of heterodimer and consequently to a decrease of p53 p53 function. In agreement with this hypothesis, analysis degradation (Figure 8). of mRNAs originating from MDM2 P1 and P2 Our data also show that p76MDM2-mediated control of promoters and giving rise to p76MDM2 þ p90MDM2 or only MDM4 levels contributes to increase the levels of p90MDM2, respectively, showed a significant decrease of transcriptional active p53, independently of the presence p90 þ p76/p90 ratio in a group of human papillary of p90MDM2. MDM4-mediated inhibition of p53 activity thyroid tumors compared with their matched normal has been attributed both to the masking of p53 tissues. P2 is a p53-sensitive promoter; however, transactivation domain (Shvarts et al., 1996) and to different reports have evidenced its activation by the subtraction of p53 out of the nucleus (Ohtsubo et al., transcription factors other than p53 (Dimitriadi et al., 2009). As we have shown that p76MDM2 controls the 2008; Pikkarainen et al., 2009; Zhou et al., 2009).

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5943 Interestingly, increased activity of the P2 promoter has the first ATG that gives rise to p90MDM2, to avoid any been associated with enhanced resistance to cell stress illegitimate translation from the mutated ATG. (Pikkarainen et al., 2009; Zhou et al., 2009). Further- Transient transfections were performed with Lipofectamine more, analysis of MDM2 proteins in human tumors has Plus (Invitrogen). In 60-mm plates, cells were transfected with MDM2 MDM2 MDM4 evidenced the low frequency of p76MDM2 expression in 1 mg of p76 , p90 , MDM4 and p54 ; 0.5 mgof comparison with other MDM2 forms, supporting the pCAGp53; and 1 mg of myc-HAUSP plus 0.3 mg of pEGFP (Invitrogen), as internal control of transfection efficiency. For hypothesis that its presence may not be beneficial for the the in vivo ubiquitination assay of MDM4, 2 mg of HA- oncogenic process (Bueso-Ramos et al., 1996; Ralhan ubiquitin-expressing plasmid was added to the transfection et al., 2000). mix. For transcriptional assays, 0.1 mg of p53, 0.1 mgof Conversely, the presence of p76MDM2 has been observed pG13luc, 0.1 mgofb-gal and 0.2 mg of p76MDM2, p90MDM2 and in human vitiligo lesion samples associated with an MDM4 were used. Luciferase activity was assayed at 24 h on increased transcriptional function of p53, particularly whole-cell extracts (Promega, Italia, Milano, Italy). the growth arrest function (Salem et al.,2009).We recently reported that cytoplasmic MDM4 facilitates p53 Immunoprecipitations and western blot analyses mitochondrial apoptosis (Mancini et al., 2009b). The Cells were lysed in Saimod buffer (50 mM Tris–HCl, pH 7.5, 5 mM downregulation of MDM4 levels by p76MDM2 in the EDTA, 150 mM NaCl, 0.5% Triton X-100). Immunoprecipita- tions and nuclear and cytoplasmic fractions were performed as cytoplasm may therefore also contribute to decrease p53 described previously (Mancini et al., 2009b). apoptotic potential, directing its function toward the The following primary antibodies were used: rabbit anti- growth arrest response as observed in the vitiligo samples. MDM4 polyclonal antibody R1 (against full-length human Our data may also concur to explain an unresolved MDM4), mouse anti-MDM4 monoclonal antibody MDMX- issue in the function of p53 in Mdm2 knockout mice 82 (Sigma-Aldrich, St Louis, MO, USA), mouse anti-MDM2 (Francoz et al., 2006). In this model, the absence of monoclonal antibodies 2A10, 4B2, Ab1 (Calbiochem, San MDM2 leads to increased p53 levels, but the transcrip- Diego, CA, USA), sheep anti-p53 polyclonal antibody Ab7 tional activity of p53, normalized to protein amount, (Oncogene, San Diego, CA, USA), rabbit anti-USP7 poly- does not increase, suggesting that the upregulation of clonal antibody (34A for western blot, 33A for immunopre- transcriptional inhibitors parallels the increase of p53 cipitation—Bethyl Lab, Montgomery, TX, USA), mouse anti-p21 F5 and mouse anti-HA monoclonal antibodies, rabbit levels. On the basis of our data, it may now be inferred MDM2 anti-Bax N-20 and rabbit anti-Sp1 sc-59 polyclonal antibodies that the absence of p76 in the knockout mice leads (Santa Cruz Biotechnology, Santa Cruz, CA, USA), mouse anti– to increased MDM4 levels and therefore to decreased tubulin monoclonal antibody (Sigma, St Louis, MO, USA) and p53 transcriptional function. rabbit anti-GFP polyclonal antibody (Invitrogen). Finally, our data indicate that p76MDM2 is endowed of intrinsic properties of ubiquitin ligase and degradation Small interfering RNA highlighting the potential existence of specific and/or MDM2 and control siRNAs were generated by Dharmacon more affine targets of this minor form of MDM2 in (Thermo Fisher Scientific, Lafayette, CO, USA). Target sequence comparison with the classical p90MDM2. of MDM2 mRNA is AACCACCTCACAGATTCCAGC and In conclusion, our data highlight a complex interplay control scramble sequence is AAGCTTTTGTCATGGAGAACG. Cells were transfected using RNAiMAX reagent (Invitrogen). among MDM family members and their relationship in the fine regulation of p53 function. They further suggest that Tissue samples and patients knowledge of their reciprocal expression pattern may be a Twenty-eight papillary thyroid carcinomas and 28 matched more precise predictor of their alteration in human tumors. normal thyroid tissue samples from the contralateral lobe from 28 patients were studied. All specimens were obtained from patients undergoing surgery at the University of Perugia from 1997 to 2007. Before the surgical procedure, all patients signed Materials and methods informed consent forms for collection of fresh thyroid samples for genetic studies. All specimens were sampled from the Cell culture, plasmids and transfections primary tumour at the time of surgery, snap-frozen and stored NIH3T3 mouse fibroblasts stably expressing MDM4 (A1-18 at À80 1C until use. Tumours containing at least 70% of clone) were cultured according to Gentiletti et al. (2002). tumour cells based on the ematoxylin/eosin staining were Mdm4À/ÀMdm2À/Àp53À/À MEFs (kindly provided by Dr JC selected. All normal thyroid tissue from the contralateral lobe Marine), derived from the triple knockout mice, were cultured were histopathologically analysed for the presence of tumour. in Dulbecco’s modified Eagle’s medium high glucose plus 10% fetal bovine serum (Cambrex, Charles City, IA, USA). All qRT–PCR experiments were performed between passages 4 and 10. qRT–PCR was performed according to Prodosmo et al. (2008). The EcoRI/HindIII fragment of cDNA encoding p76MDM2 The analysis of P1 and P2 transcripts and mouse p21 and Noxa was obtained by PCR using the following primers: p76FVECO, was driven by qRT–PCR using specific probes and SYBR 50-CCCCGAATTCACCATGAAAGAGATTATATTTTATA Master mix (Applied Biosystems, Life Technologies Corporation, TTGGCCAG-30; and p76RVSTOP, 50-TTTTAAGCTTCTA Carlsbad, CA, USA) with evaluation of dissociation curves (see GTTGAAGTAAGTTAGCACAATCATTTGGATTGG-30. Supplementary Table 1 for primers). Our results are expressed as The fragment was cloned into pcDNA3.1/Myc-His(À)B relative units (RU) of target mRNA, referred to a sample called expression vector (Invitrogen, Paisley, UK) not in frame with calibrator, chosen to represent 1 Â expression of the target gene. the myc and histidine tags and checked by sequencing. The Thecalibratorusedwasthelowestvalueinthetissuecollection plasmid coding for p76MDM2 was obtained by deleting the first under study. All samples express n-foldmRNArelativetothe 150 bp of full-length MDM2 cDNA rather than by mutation of calibrator. Each tissue sample mRNA was normalized relative to

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5944 the GAPDH mRNA, and samples for p21 and Noxa were Acknowledgements normalized relative to TBP mRNA. Statistical analysis was carried out using the Analyse-it software for Microsoft Excel The work was supported by research grants from Associazione (Analyse-it Software Ltd, Leeds, UK). Italiana Ricerca sul Cancro (AIRC), Ministero della Sanita` . We are grateful to Dr Lozano for Mdm2À/Àp53À/ÀMEFs, to Dr JC Marine for Mdm4À/ÀMdm2À/Àp53À/ÀMEFs and to Dr Jochemsen for phosphorylation-defective human Conﬂict of interest MDM4S403A coding plasmid. We are also especially grateful to Dr Jochemsen for critical review and helpful comments to The authors declare no conﬂict of interest. the paper.

References

Bueso-Ramos CE, Manshouri T, Haidar MA, Yang Y, McCown P, Marine JC, Lozano G. (2010). Mdm2-mediated ubiquitylation: p53 Ordonez N et al. (1996). Abnormal expression of MDM-2 in breast and beyond. Cell Death Differ 17: 93–102. carcinomas. Breast Cancer Res Treat 37: 179–188. Meulmeester E, Maurice MM, Boutell C, Teunisse AF, Ovaa H, Chen L, Gilkes DM, Pan Y, Lane WB, Chen J. (2005). ATM and Abraham TE et al. (2005). Loss of HAUSP-mediated deubiquitina- Chk2-dependent phosphorylation of MDMX contribute to p53 tion contributes to DNA damage-induced destabilization of Hdmx activation after DNA damage. EMBO J 24: 3411–3422. and Hdm2. Mol Cell 18: 565–576. Cheng T-H, Cohen SN. (2007). Human MDM2 isoforms translated Migliorini D, Danovi D, Colombo E, Carbone R, Pelicci PG, Marine differentially on constitutive versus p53-regulated transcripts have JC. (2002). Hdmx recruitment into the nucleus by Hdm2 is essential distinct functions in the p53/MDM2 and TSG101/MDM2 feedback for its ability to regulate p53 stability and transactivation. J Biol control loops. Mol Cell Biol 27: 111–119. Chem 277: 7318–7323. Dimitriadi M, Poulogiannis G, Liu L, Ba¨cklund LM, Pearson DM, Ohtsubo C, Shiokawa D, Kodama M, Gaiddon C, Nakagama H, Ichimura K et al. (2008). p53-independent mechanisms regulate the Jochemsen AG et al. (2009). Cytoplasmic tethering is involved in P2-MDM2 promoter in adult astrocytic tumours. Br J Cancer 99: synergistic inhibition of p53 by Mdmx and Mdm2. Cancer Sci 100: 1144–1152. 1291–1299. Francoz S, Froment P, Bogaerts S, De Clercq S, Maetens M, Okamoto K, Kashima K, Pereg Y, Ishida M, Yamazaki S, Nota A Doumont G et al. (2006). Mdm4 and Mdm2 cooperate to inhibit et al. (2005). DNA damage-induced phosphorylation of MdmX at p53 activity in proliferating and quiescent cells in vivo. Proc Natl serine 367 activates p53 by targeting MdmX for Mdm2-dependent Acad Sci USA 103: 3232–3237. degradation. Mol Cell Biol 25: 9608–9620. Gentiletti F, Mancini F, D’Angelo M, Sacchi A, Pontecorvi A, Okamoto K, Taya Y, Nakagama H. (2009). Mdmx enhances p53 Jochemsen AG et al. (2002). MDMX stability is regulated by p53- ubiquitination by altering the substrate preference of the Mdm2 induced caspase cleavage in NIH3T3 mouse ﬁbroblasts. Oncogene ubiquitin ligase. FEBS Lett 583: 2710–2714. 21: 867–877. Pan Y, Chen J. (2003). MDM2 promotes ubiquitination and Kawai H, Lopez-Pajares V, Kim MM, Wiederschain D, Yuan ZM. degradation of MDMX. Mol Cell Biol 23: 5113–5121. (2007). RING domain-mediated interaction is a requirement for Pereg Y, Shkedy D, de Graaf P, Meulmeester E, Edelson-Averbukh MDM2’s E3 ligase activity. Cancer Res 67: 6026–6030. M, Salek M et al. (2005). Phosphorylation of Hdmx mediates its Kawai H, Wiederschain D, Kitao H, Stuart J, Tsai KK, Yuan ZM. Hdm2- and ATM-dependent degradation in response to DNA (2003). DNA damage-induced MDMX degradation is mediated by damage. Proc Natl Acad Sci USA 102: 5056–5061. MDM2. J Biol Chem 278: 45946–45953. Perry ME, Mendrysa SM, Saucedo LJ, Tannous P, Holubar M. LeBron C, Chen L, Gilkes DM, Chen J. (2006). Regulation of MDMX (2000). p76(MDM2) inhibits the ability of p90(MDM2) to nuclear import and degradation by Chk2 and 14-3-3. EMBO J 25: destabilize p53. J Biol Chem 275: 5733–5738. 1196–1206. Pikkarainen S, Kennedy RA, Marshall AK, Tham el L, Lay K, Li C, Chen L, Chen J. (2002). DNA damage induces MDMX nuclear Kriz TA et al. (2009). Regulation of expression of the rat orthologue translocation by p53-dependent and -independent mechanisms. Mol of mouse double minute 2 (MDM2) by H(2)O(2)-induced oxidative Cell Biol 22: 7562–7571. stress in neonatal rat cardiac myocytes. J Biol Chem 284: Linares LK, Hengstermann A, Ciechanover A, Mu¨ller S, 27195–27210. Scheffner M. (2003). HdmX stimulates Hdm2-mediated ubiquitina- Prodosmo A, Giglio S, Moretti S, Mancin F, Barbi S, Avenia N et al. tion and degradation of p53. Proc Natl Acad Sci USA 100: (2008). Analysis of human MDM4 variants in papillary thyroid 12009–12014. carcinomas reveals new potential markers of cancer properties. Linke K, Mace PD, Smith CA, Vaux DL, Silke J, Day CL. (2008). J Mol Med 86: 585–596. Structure of the MDM2/MDMX RING domain heterodimer Ralhan R, Sandhya A, Meera M, Bohdan W, Nootan SK. (2000). reveals dimerization is required for their ubiquitylation in trans. Induction of MDM2-P2 transcripts correlates with stabilized wild- Cell Death Differ 15: 841–848. type p53 in betel- and tobacco-related human oral cancer. Am Mancini F, Di Conza G, Moretti F. (2009a). MDM4 (MDMX) and its J Pathol 157: 587–596. transcript variants. Curr Genomics 10: 42–50. Salem MM, Shalbaf M, Gibbons NC, Chavan B, Thornton JM, Mancini F, Di Conza G, Pellegrino M, Rinaldo C, Prodosmo A, Schallreuter KU. (2009). Enhanced DNA binding capacity on up- Giglio S et al. (2009b). MDM4 (MDMX) localizes at the regulated epidermal wild-type p53 in vitiligo by H2O2-mediated mitochondria and facilitates the p53-mediated intrinsic-apoptotic oxidation: a possible repair mechanism for DNA damage. FASEB J pathway. EMBO J 28: 1926–1939. 23: 3790–3807. Mancini F, Moretti F. (2009). Mitochondrial MDM4 (MDMX): an Saucedo LJ, Myers CD, Perry ME. (1999). Multiple murine double unpredicted role in the p53-mediated intrinsic apoptotic pathway. minute gene 2 (MDM2) proteins are induced by ultraviolet light. Cell Cycle 8: 3854–3859. J Biol Chem 274: 8161–8168. Marine JC, Francoz S, Maetens M, Wahl G, Toledo F, Lozano G. Shvarts A, Steegenga WT, Riteco N, van Laar T, Dekker P, Bazuine (2006). Keeping p53 in check: essential and synergistic functions of M et al. (1996). MDMX: a novel p53-binding protein with some Mdm2 and Mdm4. Cell Death Differ 13: 927–934. functional properties of MDM2. EMBO J 15: 5349–5357.

Oncogene p76MDM2 regulates p53 through control of MDM4 S Giglio et al 5945 Toledo F, Wahl GM. (2007). MDM2 and MDM4: p53 regulators Wang YV, Wade M, Wong E, Li YC, Rodewald LW, Wahl GM. as targets in anticancer therapy. Int J Biochem Cell Biol 39: (2007). Quantitative analyses reveal the importance of regulated 1476–1482. Hdmx degradation for p53 activation. Proc Natl Acad Sci USA 104: Wang YV, Wade M, Wahl GM. (2009). Guarding the guardian: 12365–12370. Mdmx plays important roles in setting p53 basal activity Zhou JX, Lee CH, Qi CF, Wang H, Naghashfar Z, Abbasi S et al. and determining biological responses in vivo. Cell Cycle 8: (2009). IFN regulatory factor 8 regulates MDM2 in germinal center 3443–3444. B cells. J Immunol 183: 3188–3194.

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

Oncogene