Inactivation of the MDM2 RING domain enhances p53 transcriptional activity in mice

The MDM2 RING domain harbors E3 ubiquitin ligase activity critical for regulating the degradation of tumor suppressor p53, which controls many cellular pathways. The MDM2 RING domain also is required for an interaction with MDMX. Mice containing a substitution in the MDM2 RING domain, MDM2C462A, disrupting MDM2 E3 function and the MDMX interaction, die during early embryogenesis that can be rescued by p53 deletion. To investigate whether MDM2C462A, which retains p53 binding, has p53-suppressing activity, we generated Mdm2C462A/C462A;p53ER/− mice, in which we replaced the endogenous p53 alleles with an inducible p53ER/− allele, and compared survival with that of similarly generated Mdm2−/−;p53ER/− mice. Adult Mdm2-null mice died ∼7 days after tamoxifen-induced p53 activation, indicating that in the absence of MDM2, MDMX cannot suppress p53. Surprisingly, Mdm2C462A/C462A;p53ER/− mice died ∼5 days after tamoxifen injection, suggesting that p53 activity is higher in the presence of MDM2C462A than in the absence of MDM2. Indeed, in MDM2C462A-expressing mouse tissues and embryonic fibroblasts, p53 exhibited higher transcriptional activity than in those expressing no MDM2 or no MDM2 and MDMX. This observation indicated that MDM2C462A not only is unable to suppress p53 but may have gained the ability to enhance p53 activity. We also found that p53 acetylation, a measure of p53 transcriptional activity, was higher in the presence of MDM2C462A than in the absence of MDM2. These results reveal an unexpected role of MDM2C462A in enhancing p53 activity and suggest the possibility that compounds targeting MDM2 RING domain function could produce even more robust p53 activation.

As a transcription factor, p53 induces the transcription of several downstream targets in the presence of both intrinsic and extrinsic cellular stress signals; many of these targets regu-late the cell cycle, apoptosis, DNA repair, senescence, angiogenesis, and other cellular pathways (1)(2)(3)(4). The activation of p53-dependent pathways is thought to prevent and resist tumor development and growth by preventing the proliferation of damaged cells with oncogenic potential. For that reason, p53 has been called "the guardian of genome" because of its indispensable contribution to genome integrity (5)(6)(7). However, because of its global and heterogeneous functions, endogenous p53 activity must also be carefully regulated.
MDM2 (murine double minute 2) and MDM4 (murine double minute 4, also known as MDMX) are widely considered to be the primary negative regulators of p53 (8,9). Considering that Mdm2 and MdmX knock-out mice exhibit p53-dependent early embryonic lethality, MDM2 and MDMX play unique and nonredundant roles in the control of p53 activation (10 -12). Although MDM2 and MDMX are structurally similar, they exhibit differing activities toward p53 regulation. Through its RING domain, MDM2 not only functions as an E3 ubiquitin ligase for p53 (13)(14)(15) but also binds to MDMX in a heterodimer that is required for p53 regulation during embryogenesis (16,17). Recent evidence has demonstrated that the RING finger motif of MDM2 is indispensable for the suppression of p53 activity (18,19), and we have previously demonstrated that the MDM2 RING domain plays a crucial role in p53 regulation during embryonic development (14). Interestingly, two functions of the MDM2 RING domain, E3 ubiquitin ligase activity and MDMX binding, have been found to be independently dispensable in unstressed adult mice (15,17). Thus, the function of RING domain toward the regulation of physiological p53 transcriptional activity in the adult mouse still remains inadequately understood. In this study we sought to address the following questions: 1) Does the MDM2 RING domain function to regulate p53 transcriptional activity in adult mice? 2) Is MDM2-p53 binding sufficient for p53 regulation and adult organismal survival? 3) In the absence of MDM2 or MDM2 RING domain function, does MDMX play a role in regulating p53 transcriptional activity in adult mice?
To study the role of the MDM2 RING domain in vivo, we crossed mice bearing a single residue substitution in the MDM2 RING domain (Mdm2 C462A/C462A ) that abolishes several key RING domain functions, including MDM2 E3 ligase activity and MDM2-MDMX binding. Because Mdm2 C462A/C462A single-residue substitution results in the embryonic lethality, we substituted endogenous p53 with p53 ER/Ϫ , an inactive p53 estrogen receptor fusion allele, which allows for tamoxifen mediated p53 functional restoration (20). At the same time, we generated Mdm2 ϩ/ϩ ;p53 ER/Ϫ and Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ as control groups to define the function of the MDM2 RING domain on regulation of physiological p53. Strikingly, following p53ER functional restoration, mice bearing the MDM2 C462A mutation die earlier than mice bearing MDM2 deletion, and this early lethality correlates with increased p53 activity. These results provide evidence to suggest that loss of MDM2 RING domain function through MDM2 C462A mutation stimulates spontaneous p53 activation beyond what occurs when MDM2 is deleted.

Mice expressing a functionally inactive Mdm2 RING domain exhibit early lethality following p53 activation
In vitro, MDM2 has been shown to modulate p53 transcriptional activity by direct inhibitory binding to the p53 transactivation domain (21,22). The RING domain of MDM2, which is important for MDM2-mediated p53 ubiquitination and degradation and MDM2-MDMX heterodimer formation, is also critical for p53 regulation in vivo. Several studies using mouse models have suggested that each of these RING domain functions are important to p53 regulation during embryogenesis; however, MDM2-MDMX binding and MDM2 E3 ligase activity have been shown to be independently dispensable in the adult mouse (14,15,17). To confirm whether MDM2 retains p53 inhibitory capabilities in the absence of either of the canonical MDM2 RING domain functions in adult mice, we utilized mice bearing the MDM2 C462A mutation (Mdm2 C462A/C462A ), which disrupts both MDM2-MDMX binding and MDM2 E3 ligase function. Because Mdm2 C462A/C462A mice exhibit p53-dependent embryonic lethality (14), we also utilized mice bearing the inactive p53 estrogen receptor fusion allele (hereafter referred to as p53 ER ), which allows for tamoxifen-mediated p53 functional restoration (20). We crossed Mdm2 C462A/C462A mice with p53 ER/Ϫ mice to generate Mdm2 C462A/C462A ;p53 ER/Ϫ mice. We also generated the mice with the following genotypes as positive and negative controls for MDM2 function, respectively: Mdm2 ϩ/ϩ ;p53 ER/Ϫ and Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ . Previous studies have shown that Mdm2 ϩ/ϩ ;p53 ER/Ϫ mice tolerate daily injections of tamoxifen (20). Our study replicated these results. Following daily injection of tamoxifen, Mdm2 ϩ/ϩ ;p53 ER/Ϫ mice survived the duration of the study (up to 40 days) with no apparent abnormalities. Surprisingly, under the same conditions the maximum survival time of Mdm2 C462A/C462A ;p53 ER/Ϫ mice was ϳ4 -5 days (Fig. 1A). This dramatic difference between Mdm2 C462A/C462A ;p53 ER/Ϫ and Mdm2 ϩ/ϩ ;p53 ER/Ϫ mice suggests that a functional MDM2 RING domain is integral for maintaining homeostasis in adult mice. Further, mice lacking MDM2 (Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ ) have been shown to perish ϳ6 days after a single tamoxifen injection (23). We also compared the survival of Mdm2 C462A/C462A ;p53 ER/Ϫ mice with that of Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ mice following tamoxifen injection. Surprisingly, we found that Mdm2 C462A/C462A ;p53 ER/Ϫ mice lived an evidently shorter lifespan than Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ mice (Fig.  1B), which suggests that loss of MDM2 RING domain function through the MDM2 C462A mutation may alter MDM2 functions toward p53 regulation and enhance p53 function. Because the MDM2 C462A mutation disrupts MDM2-MDMX interaction and MDMX has been shown to be important for MDM2mediated p53 regulation (24 -26), we sought to determine whether the differences in survival between Mdm2 C462A/C462A ; p53 ER/Ϫ mice and either Mdm2 ϩ/ϩ ;p53 ER/Ϫ mice or Mdm2 Ϫ/Ϫ ; p53 ER/Ϫ mice could be dependent on differential MDMX protein stability or p53 regulation. To this end, we generated each of these mice in the background of MdmX deletion. We found that following tamoxifen injection, the median survival time of Mdm2 C462A/C462A ;MdmX Ϫ/Ϫ ;p53 ER/Ϫ mice was significantly shorter than that of Mdm2 Ϫ/Ϫ ;MdmX Ϫ/Ϫ ; p53 ER/Ϫ mice (Fig.  1C) and was equivalent to Mdm2 C462A/C462A ;p53 ER/Ϫ mice (Fig.  1D), indicating that changes in MDMX-mediated p53 regulation or protein stability are unlikely to contribute to survival differences present in Mdm2 C462A/C462A ;p53 ER/Ϫ mice. Further, the loss of MDMX did not contribute to changes in survival upon Mdm2 deletion, as Mdm2 Ϫ/Ϫ ;MdmX Ϫ/Ϫ ;p53 ER/Ϫ mice and Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ mice display similar survival times following tamoxifen injection (Fig. 1E), suggesting that in the absence of MDM2, MDMX plays a minimal role in contributing to organismal survival during basal levels of p53 activity. This notion is in line with previous studies in MdmX Ϫ/Ϫ ; p53 ER/Ϫ mice, which tolerate daily injections of tamoxifen for a median of ϳ29 days (27), which is a significantly longer period of survival than Mdm2 Ϫ/Ϫ ; p53 ER/Ϫ mice, which survive approximately 7 days.

Mdm2 RING mutation enhances p53 target gene transcription
The accelerated organismal lethality and increased apoptosis in tissues of Mdm2 C462A/C462A ;p53 ER/Ϫ mice are consistent with increased p53 activity. To assess the effect of the MDM2 C462A mutation on p53 transcriptional activity, we isolated mouse embryonic fibroblasts (MEFs) from each of the five genotypes and treated these cells with 4-hydroxytamoxifen (4-OHT). We then utilized qRT-PCR to detect changes in the transcription of several p53 target genes including p21, puma, and bax (Fig. 3A) (28 -30). Consistent with previous studies, p21 transcription was elevated in Mdm2 ϩ/ϩ ;p53 ER/Ϫ MEFs and was increased to a greater degree in Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ MEFs (23). Although puma and bax transcription were nearly unchanged following 4-OHT treatment in Mdm2 ϩ/ϩ ;p53 ER/Ϫ MEFs, an increase in their transcription was apparent in Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ MEFs. Interestingly, following 4-OHT treatment, p21, bax, and puma transcription were significantly increased in Mdm2 C462A/C462A ;p53 ER/Ϫ MEFs to an even greater extent than in Mdm2 Ϫ/Ϫ ;p53 ER/Ϫ or Mdm2 Ϫ/Ϫ ; MdmX Ϫ/Ϫ ;p53 ER/Ϫ MEFs, suggesting that p53 activity is greatest in Mdm2 C462A/C462A ;p53 ER/Ϫ MEFs. Further, the deletion of MDMX did not significantly alter p53 activity in the presence of the MDM2 C462A mutation. To confirm these observations in vivo, we harvested small intestine and lung from the p53ER mice after the daily administration of tamoxifen for 3 days. The trend that we observed in MEFs was also reflected in mouse tissue, including small intestine and lung, with Mdm2 C462A/C462A ;p53 ER/Ϫ and Mdm2 C462A/C462A ;MdmX Ϫ/Ϫ ; p53 ER/Ϫ mice exhibiting the highest expression of each target gene (Fig. 3, B and C). Intriguingly, because mice bearing the MDM2 C462A mutation exhibit greater p53 activity than mice lacking MDM2 altogether, these data suggest the possibility that MDM2-mediated p53 activity suppression is dependent on MDM2 RING function, and in the absence of this function MDM2 may have alternate p53 regulating capabilities in vivo.
Previous studies have shown that p53 activation requires its own acetylation (35). The acetylation of p53 through interaction with co-factors p300/CBP is required for p53 to bind to DNA and promote the transcription of p21 and other downstream targets (36,37). Consequently, we sought to ascertain whether the MDM2 C462A mutation affects p53 acetylation, thereby elevating p53 transcriptional activity. To normalize for the variations in p53ER protein levels in each of the different MEFs, we calculated the ratio of acetylated p53ER to total p53ER protein levels (Fig. 4C). Although p53ER protein levels were consistently highest in cells lacking MDM2 (Fig. 4A), relative p53ER acetylation levels were highest in cells harboring the MDM2 C462A mutation (Fig. 4C).

Figure 3. Functional Mdm2 RING inactivation correlates with enhanced p53 target gene transcription in vivo.
A, MEFs of the indicated genotypes in a p53 ER/Ϫ background were treated with 4-OHT for the indicated times and harvested. Then qRT-PCR was used to measure the mRNA levels of p21, puma, and bax relative to those present in untreated Mdm2 ϩ/ϩ ;p53 ER/Ϫ MEFs. B and C, mice of each indicated genotypes (n ϭ 3) in a p53 ER/Ϫ background were treated with daily tamoxifen injection for 3 days and harvested. Then qRT-PCR was used to measure the mRNA levels of p21, puma, and bax present in small intestine (B) and lung (C) relative to those present in tissues from Mdm2 ϩ/ϩ ;p53 ER/Ϫ mice. The data are means Ϯ S.D. calculated from three independent experiments and normalized to Actin. *, p Ͻ 0.05; **, p Ͻ 0.01.

Discussion
The results of this study have provided some clarification of the role of the MDM2 RING domain in the regulation of p53 post-embryogenesis. First, whereas the two key functions of this domain (E3 ligase activity and MDMX binding) can be individually lost with little consequence in adult, unstressed mice (15,17), at least one of these functions is required to maintain tissue homeostasis (Fig. 2). When both E3 ligase activity and MDMX binding capabilities are lost, MDM2-p53 binding is not sufficient to control basal p53 transcriptional activity and prevent p53-induced apoptosis and organismal lethality (Fig.  1). This is principally demonstrated by the early lethality of Mdm2 C462A/C462A ;p53 ER/Ϫ mice compared with Mdm2 ϩ/ϩ ; p53 ER/Ϫ mice (Fig. 1A).
Second, the presence of the MDM2 C462A mutation correlated with both increased p53 transcriptional activation and acetylation compared with the complete absence of MDM2 (Fig. 5). There are several possibilities that could explain this observation. First, studies in our lab have revealed that the MDM2 C462A mutation facilitates greater interaction of p53 with histone acetyl transferase CBP/p300 (15,38), an important co-activator of p53 that potentiates p53 acetylation and transcriptional activity (37). Previous in vitro studies have also shown that disruption of the MDM2 RING domain disrupts MDM2-MDMX binding but not MDM2 homodimer formation (39). Although the MDM2 C462A mutation clearly disrupts MDM2-MDMX binding, whether or not it affects MDM2 homodimerization in vivo remains unknown. It is possible that MDM2-MDMX binding serves to inhibit MDM2 p300-mediated acetylation of p53 but that monomeric or homodimeric MDM2 could enhance p300 -p53 interaction and functional activation. Because we have recently shown that the MDM2 RING domain is required for MDM2-MDMX heterodimerization but not for MDM2-MDM2 homodimerization (39), it is possible that the MDM2 C462A mutant forms a homodimer, which may have a stronger binding affinity to acetyltransferases. This would suggest that MDM2-mediated inhibition of p53 acetylation by p300 requires MDM2-MDMX interaction, which is supported by the correlation of the MDM2 C462A mutation with increased p53 acetylation in mouse tissues (Fig. 5F). Although overexpressed MDMX has been shown to inhibit p300-mediated p53 acetylation (40), our data could suggest that MDMX alone cannot inhibit p300 -p53 interaction in vivo, because the loss of MDMX does not appear to increase p53 acetylation or activity beyond that which is observed in the presence of the MDM2 C462A mutation (Fig. 5), although this hypothesis remains to be tested in vivo.
It is also possible that the MDM2 RING domain is required for the inhibitory functions of MDM2-p53 binding. For example, in vitro assays using an overexpressed human equivalent HDM2 C464A mutation have demonstrated that a functional RING domain is necessary for MDM2-mediated p53 export (41). In this context, MDM2 C462A could tether p53 to the nucleus, allowing for increased interaction with DNA and B, the levels of p21 from three independent experiments as in A were quantified using densitometry and normalized to untreated Mdm2 ϩ/ϩ ; p53 ER/Ϫ MEFs. C, the ratio of acetyl-p53ER to total p53 ER protein from three independent experiments as in A were quantified using densitometry and normalized to untreated Mdm2 ϩ/ϩ ;p53 ER/Ϫ MEFs. The data are represented as means Ϯ S.D. calculated and were normalized to Actin. *, p Ͻ 0.05; **, p Ͻ 0.01.

MDM2 C462A mutation activates p53
greater transcriptional activity. The combination of increased p53 acetylation and activity with nuclear sequestration could explain why mice bearing the MDM2 C462A mutation demonstrate increased p53 activity compared with mice lacking MDM2 (Fig. 3).
This study suggests the following about in vivo MDM2-mediated p53 regulation: 1) the MDM2 RING domain contributes to the regulation of p53 transcriptional activity in adult mice; and 2) MDM2-p53 binding is not sufficient for p53 regulation and adult organismal survival; and 3) in the absence of MDM2 or MDM2 RING domain function, MDMX does not appear to play a role in regulating p53 transcriptional activity in adult mice. However, it remains possible that the observations made in this study could be specific to the MDM2 C462A mutation or p53ER. In future studies it would be useful to confirm these results using another model system, such as mice lacking the MDM2 RING domain, containing an alternate RING-disrupting mutation or an alternate inducible p53 model. The results of this study also have implications for p53-inducing therapies. Small molecule compounds that inhibit MDM2-p53 interaction, like nutlin-3a, have been widely used to facilitate p53 activation (42). Our study suggests the possibility that compounds targeting MDM2 RING domain function could produce even more robust p53 activation.

Genetically engineered mice and in vivo studies
To generate mice bearing the genotypes used in this study, we bred p53 ER/Ϫ mice with mice from several different backgrounds, including Mdm2 Ϫ/Ϫ , Mdm2 C462A/C462A , and MdmX Ϫ/Ϫ (10,14,20). All experimental procedures were approved by the University of North Carolina Institutional Animal Care and Use Committee. To restore p53ER function, mice were intraperitoneally injected with 100 mg/kg of tamoxifen (T5648; Sigma) according to body weight. , colon (C), and lung (D) tissues were harvested for Western blot analysis with the indicated antibodies. E, the levels of p21 from three independent experiments as in A-D were quantified using densitometry and normalized to Mdm2 ϩ/ϩ ;p53 ER/Ϫ mouse tissues. F, the acetyl-p53 ER protein level from three independent experiments as in A-D were quantified using densitometry and normalized to Mdm2 ϩ/ϩ ;p53 ER/Ϫ mouse tissues. The data are means Ϯ S.D. and were normalized to Actin. *, p Ͻ 0.05; **, p Ͻ 0.01. S.I., small intestine.

MDM2 C462A mutation activates p53
Cell culture MEF cells were isolated from embryonic day 13.5-16.5 embryos and cultured in a 37°C incubator with 5% CO 2 in DMEM supplemented with 10% fetal bovine serum (Gibco) and 100 IU/ml penicillin (Gibco). To restore p53ER function in vitro, cultured MEFs were exposed to 100 nM 4-OHT (H7904; Sigma) for the indicated amounts of time.

Immunoblot analysis
Protein was extracted from organs and cells as described previously (14). Protein lysates were fractionated on SDS-polyacrylamide gels and transferred onto nitrocellulose membranes. Procedures and conditions for immunoblotting were described previously. The following antibodies were purchased commercially: monoclonal mouse anti-MDMX (MDMX-82; Sigma), monoclonal mouse anti-actin (MAB1501; EMD Millipore), polyclonal rabbit anti-acetyl-p53 (K379; Cell Signaling Technology Inc.), and polyclonal goat anti-p21 (C-19; Santa Cruz Biotechnology). The monoclonal mouse anti-p53 (pAB122) and monoclonal mouse anti-MDM2 (2A10) were homemade by our laboratory. The intensity of the bands in the linear range of exposure was quantified using ImageJ 1.46r.

Hematoxylin and eosin staining and TUNEL assay
Tissues were dissected, flushed gently with cold PBS, fixed in 10% formalin overnight, dehydrated in 50% ethanol, and stored in 70% ethanol until they were transferred to the Histology Research Core Facility at University of North Carolina for paraffin embedding. The tissues were then cut into 4-m sections and then processed for H&E staining. TUNEL staining for apoptosis was performed using a Apoptag peroxidase in situ apoptosis detection kit according to the manufacturer's instructions. TUNEL stained tissue was analyzed using an Olympus IX-81 microscope fitted with a SPOT camera and software.

Statistical analysis
The results are represented as means Ϯ S.E. of the mean. Quantitative PCR data and immunoblotting quantifications were evaluated for significance using the unpaired t test. A p value Ͻ0.05 was considered significant for all analyses. Significant differences between experimental groups were as follows: *, p Ͻ 0.05; or **, p Ͻ 0.01. Calculations were performed using GraphPad Prism 5 software.

Ethics statement
This investigation has been conducted in accordance with ethical standards, the Declaration of Helsinki, and national and international guidelines and has been approved by the authors' institutional review board.