Physical interaction with human tumor-derived p53 mutants inhibits p63 activities.

The p53 tumor suppressor gene is the most frequent target for genetic alterations in human cancers, whereas the recently discovered homologues p73 and p63 are rarely mutated. We and others have previously reported that human tumor-derived p53 mutants can engage in a physical association with different isoforms of p73, inhibiting their transcriptional activity. Here, we report that human tumor-derived p53 mutants can associate in vitro and in vivo with p63 through their respective core domains. We show that the interaction with mutant p53 impairs in vitro and in vivo sequence-specific DNA binding of p63 and consequently affects its transcriptional activity. We also report that in cells carrying endogenous mutant p53, such as T47D cells, p63 is unable to recruit some of its target gene promoters. Unlike wild-type p53, the binding to specific p53 mutants markedly counteracts p63-induced growth inhibition. This effect is, at least partially, mediated by the core domain of mutant p53. Thus, inactivation of p53 family members may contribute to the biological properties of specific p53 mutants in promoting tumorigenesis and in conferring selective survival advantage to cancer cells.

However, at variance with other tumor suppressor genes, cells with p53 mutations maintain expression of full-length protein. This may suggest that, at least, certain mutant forms of p53 can gain additional functions through which they contribute actively to cancer progression (6 -9). Such evidence is provided by several in vitro and in vivo studies (10 -17). We and others have reported that conformational defective p53 mutants can increase resistance of tumor cells to anticancer treatment and promote genomic instability (18 -20). The molecular mechanisms underlying such effects remain to be elucidated.
The recent identification of two p53-relatives, p63 and p73 holds new perspectives in studying gain of function of mutant p53 (21)(22)(23). p63 and p73 share a significant homology with each other and with p53. Indeed, they share the same modular organization, comprising an N-terminal transactivation domain, a central sequence-specific DNA binding domain, and a C-terminal oligomerization domain. Several p63 and p73 isoforms are present in cells (21,(23)(24)(25)(26)(27)(28). They result either from the use of a cryptic promoter that generates p63 isoforms (⌬N p63␣, p63␤, and p63␥) lacking N-terminal transactivation domain or by alternative splicing that generates p63 isoforms (p63␣, p63␤, and p63␥) with different C-terminal sequences (30,32). Exogenous expression of p63 or p73 causes growth arrest, apoptosis, and differentiation, recapitulating some of the most characterized p53 biological effects (9,21,23,30,31). These are mainly mediated by p73 and p63 through the activation of specific p53 target genes such as Bax, IGF-BP3, p21 waf1 , and cyclin G (16,21,23,24,(32)(33)(34)(35)(36)(37). The respective deficient mice have provided further insights into the physiological role of p73 and p63. p73-deficient mice exhibit severe defects in the development of nervous and immune systems (38). p63 Ϫ/Ϫ mice are born alive but show striking defects in development. Their skin does not progress from early stages of development, lacking stratification as well as expression of differentiation markers. The mammary glands, hair follicles, and teeth are absent in p63-deficient mice (39,40). In agreement with this phenotype, p63 was recently found mutated in patients affected by autosomal dominant disorder characterized by ectrodactyly, ectodermal dyspalsia, and facial clefts (41).
The possibility that p53, p73, and p63 form hetero-oligomers in cells has been indicated by recent reports (8,9). It was originally shown that p73␣ but not p73␤ can interact with p53 in two-hybrid screening (21). It was subsequently reported that human tumor-derived p53 mutants can engage in a physical association in vitro and in vivo either with p73 or p63 (33,34,42,43). Recent findings indicate that the association between mutant p53 and p73 can be, at least partially, governed by a common polymorphism at codon 72 of p53 that encodes Arg or Pro (34). The biological outcome of these interactions results in functional inactivation of p73 transcriptional activity as well as induction of apoptosis (33,34,42,43). Therefore, we have further investigated the physical association between human tumor-derived p53 mutants and p63. We found that this association occurs under physiological conditions as shown by reciprocal co-precipitation experiments performed either in T47D breast cancer cells and HaCat immortalized keratinocytes. The core domain of mutant p53 and the DNA binding domain of p63 mediate this association. We also report that binding of mutant p53 strongly impairs in vitro and in vivo the binding of p63 to its target gene promoters and consequently affects its transcriptional activity. Indeed, we found that in cells carrying endogenous mutant p53, such as T47D cells, p63 is unable to recruit some of its target gene promoters. Similar to wt-p53, overexpression of p63 strongly inhibits colony formation of SAOS2 cells. The concomitant expression of specific p53 mutants, such as p53His175 and p53His273, but not p53Gly281, counteracts the p63-mediated growth suppression. Such an effect requires the binding of p63 to mutant p53 and is not exerted by the coexpression of wt-p53. Of note, the core domain of mutant p53 can also exert such an impairment. Thus, protein-protein interactions between members of the p53 family might impact on growth and drug resistance of cancer cells.

EXPERIMENTAL PROCEDURES
Cell Culture-The H1299 cell line is derived from a human large cell lung carcinoma. H1299 cells were maintained in RPMI medium, supplemented with 10% fetal calf serum (FCS) (Invitrogen) (19). The H1299-p53His175#41 cell line was generated as previously reported (43). SAOS2 osteosarcoma cells (a gift from M. Fanciulli) and HaCat immortalized keratinocytes were maintained in Dulbecco's modified Eagle's medium supplemented with 10% FCS. T47D breast cancer cells were maintained in RPMI 10% FCS (43).
Transient transfections were done in Dulbecco's modified Eagle's medium plus 10% FCS by the calcium phosphate method in the presence of BES (N,N-bis(2-hydroxyethyl-2-aminoethanesulfonic acid, sodium salt) (Sigma). The precipitates were left for 12 h, after which the medium was changed again to RPMI plus 10% FCS. The cells were harvested at 36 or 48 h, respectively (43).
T47D and HaCat cells were lysed as previously described. Aliquots of cell extracts containing 2.5 mg of total proteins were immunoprecipitated with anti-p63 polyclonal antibody, with anti-IgG polyclonal antibody, or with a mixture of anti-p53 mAbs DO1 and 1801. For p53 detection, a mixture of p53 mAbs DO1 and 1801 was used at 1:40 dilution. For reciprocal co-precipitation experiments, T47D and HaCat cells were immunoprecipitated with anti-p53FL393 and with a mixture of anti-p53 mAbs DO1 and 1801, respectively. For p63 detection, an anti-p63 monoclonal antibody (4A4 that was raised against amino acids 1-205 of p63) (Santa Cruz Biotechnology) was used at 1:100 dilution.
Western blot analysis was performed with the aid of the enhanced chemiluminescence Supersignal West Pico Stable Peroxidase Solution (Pierce).
Recombinant Proteins and in Vivo Binding Assays-Recombinant proteins employed in the pull-down assay were produced as previously described (36,43). Pull-down assays were performed using 20 g of immobilized purified GST fusion proteins or wild type GST that were incubated with 2 mg of total cellular proteins prepared from H1299 cells transiently transfected with p53His273, p53Trp248, or p53His175-(74 -298), from H-175#41 and T47D cells. The immunoblots were probed with a mixture of anti-p53 mAbs DO1 and 1801 or with anti-p53 mAb 240 or with anti-MBP antibody. Detection was performed with the aid of the enhanced chemiluminescence Supersignal West Pico Stable Peroxidase Solution (Pierce).
Formaldehyde Cross-linking and Chromatin Immunoprecipitation-H1299 -175#41 cells were treated with ponasterone A to induce the expression of p53H175 for 24 h. DNA and proteins were cross-linked by the addition of formaldehyde (1% final concentration) 20 min before harvesting. Formaldehyde cross-linking was stopped by the addition of glycine, pH 2.5 (125 mM final concentration), for 5 min at room temperature. Cells were scraped off of the plates, resuspended in hypotonic buffer, and passed through a 26-gauge needle. Nuclei were spun down, resuspended in 300 l of SDS lysis buffer (1% SDS, 10 mM EDTA, 50 mM Tris-HCl, pH 8, and a protease inhibitor mixture), and sonicated to generate 500 -2000-bp fragments. After centrifugation, the cleared supernatant was diluted 10-fold with immunoprecipitation buffer (50 mM Tris-HCl, pH 8, 150 mM NaCl, 5 MM EDTA, 0.5% Nonidet P-40). The cell lysate was precleared by incubation at 4°C with 50 l of Protein A beads preadsorbed with sonicated single-stranded DNA and bovine serum albumin. The cleared lysates were incubated overnight with an anti-p63 polyclonal antibody (H-137, whose epitope corresponds to amino acids 15-151 of the DNA binding of ⌬Np63␣) (Santa Cruz Biotechnology) or with a control antibody. Immune complexes were precipitated with protein A beads preadsorbed with sonicated single-stranded DNA and bovine serum albumin. After centrifugation, the beads were washed and the antigen was eluted with 1% SDS, 100 mM sodium carbonate. DNA-protein cross-links were reversed by heating at 65°C for 4 -5 h, and DNA was phenol-extracted and ethanol-precipitated. Levels of Bax, p21 waf1 , 14-3-3, and p53AIP1 promoter DNAs were determined by PCR using oligonucleotides spanning the p53 binding sites. The following specific oligonucleotides were used: Bax (down, Electrophoretic Mobility Shift Assay-The electrophoretic mobility shift assay was performed in a 20-l DNA binding reaction, which contained 30 g of whole cell extract, 4 fmol of labeled duplex oligonucleotide, binding buffer (50 mM Tris-HCl, pH 7.5, 40% glycerol, 100 mM dithiothreitol, 2 g/ml bovine serum albumin, and 0.2% Triton X-100) and 300 g of salmon sperm. The reaction was carried out at room temperature for 10 min, and the protein-DNA complexes were subjected to native electrophoresis on 5% acrylamide, 0.5ϫ TBE gel. Anti-p73 polyclonal antibodies C-17 and C-20 (Santa Cruz Biotechnology) and anti-p63 polyclonal antibody H-129 (Santa Cruz Biotechnology) were added to the labeled oligonucleotide and incubated on ice for 15 min.
The following oligonucleotide was used: TCA CAA GTT AGA GAC AAG CCT GGG CGT GGG CAT TAT T. Luciferase Assays-H1299 cells (3 ϫ 10 5 /60-mm plate) were transfected with reporter plasmid together with the indicated combinations of plasmids. An equal number of pCMV-␤-gal plasmids was added to each transfection reaction mixture. 36 h later, cells were rinsed with cold phosphate-buffered saline, resuspended in cell lysis buffer (Promega Corp., Madison, WI), and incubated for 10 min at room temperature. Insoluble material was spun down, and luciferase activity was quantitated using a commercially available kit (Promega) with the aid of a TD-20E luminometer (Turner). The values were normalized for ␤-galactosidase and protein contents.
Indirect Immunofluorescence-Cells growing on glass coverslips were fixed with cold methanol and incubated at Ϫ20°C for 30 min. After rehydration with phosphate-buffered saline for 5 min, the cells were stained for 1 h with a mixture of p53 mAbs DO1 and 1801. Staining with the secondary antibody and with 4Ј,6-diamidino-2-phenylindole was performed as described before (19), followed by visualization under a fluorescence microscope.
Colony Suppression Assays-SAOS2 cells were plated and transiently transfected as reported above. 24 h after the transfection, cells were detached and replated in triplicate at 3 ϫ 10 4 /60-mm dish. Two days later, puromycin (2 g/ml) was added to each plate and maintained for 72 h. Two weeks later, colonies were fixed in methanol and stained with Giemsa solution followed by washing with water.

RESULTS
The Association between Human Tumor-derived p53 Mutants and p63 Occurs under Physiological Conditions-To determine the biological relevance of the association between mutant p53 and p63, we performed coprecipitation experiments using T47D breast cancer cells and HaCat skin-derived immortalized keratinocytes carrying endogenous mutant p53Phe194 and mutant p53Y179/W282, respectively, as well as endogenous p63 (34,44). Following a preclearing, equal portions of cell extract were taken for immunoprecipitation with control anti-IgG polyclonal serum (Fig. 1, A-C, lane 3), with anti-p63 polyclonal antibody (Fig. 1, A-C, lane 2), or with a mixture of anti-p53 mAbs DO1 and 1801 (Fig. 1A, lane 4). Immunoprecipitates were subjected to immunoblot with a mixture of anti-p53 mAbs DO1 and 1801. Aliquots of total cell lysate from control cells (Fig. 1, A and B, lane 1) were directly applied on the gel. As shown in Fig. 1, A-C, p53 mutants were detected only in the immunoprecipitates with p53 or with p63 (lanes 2 and 4) and not in the anti-IgG immunoprecipitates (lane 3). Reciprocal co-precipitation experiments of T47D and HaCat were performed by immunoprecipitation with anti-p53FL393 and a mixture of anti-p53 mAbs DO1 and 1801, respectively. Equal aliquots of lysates derived from the above mentioned cells were immunoprecipitated with anti-IgG serum or with anti-HA monoclonal antibody. Immunoblot was probed with a monoclonal anti-p63 antibody. As shown in Fig. 1, B-D, coprecipitated p63 was detected only in the p53-immunoprecipitates (lane 2). Of note, p63␥, which is quite abundant in T47D cells as verified by reverse transcription-PCR analysis (data not shown), is involved in the association with mutant p53 (Fig. 1B) Thus, these results indicate that the association between mutant p53 and p63 occurs under physiological conditions.
The DNA Contact-defective p53 Mutants, p53His273 and p53Trp248, Engage in a Physical Association with Endogenous p63␣-In an attempt to verify whether DNA contact-defective p53 mutants, such as p53His273 and p53Trp248 can also associate with endogenous p63␣, we performed coprecipitation experiments. To this end, H1299 cells were transiently transfected with a vector encoding p53His273 or p53Trp248 or with an empty vector. Cell lysates derived from these cells were processed as previously reported (Fig. 1C). As seen in Fig. 2A, coprecipitated mutant p53His273 and p53Trp248 were brought down only in anti-p63 immunoprecipitates (lanes 8 and 9). The specificity of these interactions was further confirmed by a pull-down assay in which identical cell lysates were incubated with GST-p63␣ or GST alone. Specifically bound mutant p53 was detected only when cell lysates (1.5 mg/lane) of H1299 overexpressing p53 mutants were incubated with GST-p63␣ (Fig. 2B, lanes 6 and 9). Aliquots of unprocessed lysates (100 g/lane) were processed and probed for each of the above reported experiments.
Taken together, these results indicate that other human tumor-derived p53 mutants distinct from those carried by T47D and HaCat cells can engage in a physical association with endogenous p63. Furthermore, they show that the two most frequent p53 mutations in human cancers, such as p53His273 and p53Trp248, can interact with p63.
The Core Domain of Mutant p53 Is Sufficient for the Association with p63␣-We and others have previously shown that the core domain of mutant p53 is sufficient for the association with different isoforms of p73 (42,43). To further characterize the association between mutant p53 and p63, we aimed to identify which p53 domain is involved in that association.
To this end, we transiently co-transfected H1299 cells with a vector encoding p53His175 or the indicated double mutants together with a vector encoding a c-Myc-tagged version of p63␣ (Fig. 3B). In agreement with previously reported data, we found that p53His175 and p63␣ can associate in reciprocal coprecipitation experiments (Fig. 3A, lane 4, and Fig. 3B, lane  2). Furthermore, as shown in Fig. 3B (lanes 3-6), p53 double mutants were present in anti-p63 immunoprecipitates. We checked the efficiency of the p63 immunoprecipitation by reprobing the blot with anti-c-Myc monoclonal antibody (Fig. 3B,  middle panel). Protein levels of p53His175, its double mutants, and c-Myc-tagged p63␣ reached in each transient co-transfection are shown in Fig. 3B (lower panels). The reported results clearly suggest that the core domain of mutant p53 might be sufficient for the association with p63␣. To investigate this issue, we employed an in vivo binding assay. Total cell lysates of H1299 cells transiently transfected either with pEGFPp53His175-(74 -298) vector or with its related empty vector were incubated with GST-p63␣ or with GST alone. Specifically bound mutant p53 was detected by probing the immunoblot with anti-p53 mAb 240 (Fig. 4A, lane 4). To verify whether the core of mutant p53 can also engage in a physical interaction with endogenous p63␣, H1299 cells were transiently transfected with the above reported vectors. Cell lysates derived from these cells were immunoprecipitated with anti-p63 and anti-IgG polyclonal sera and subjected to immunoblot with anti-p53 mAb 240. As shown in Fig. 4B, p53His175 (74 -298) was brought down only from extracts of cells transfected with pEGFPp53His175 (74 -298) and immunoprecipitated with anti-p63 polyclonal serum (lane 3).
To further investigate whether the association between mutant p53 or its core domain and p63 occurs directly, we employed an in vitro binding assay. Bacterially expressed and purified MBP-p63␣ protein was incubated with GST-p53His175 or with GST-p53His175 (74 -298) or with GST alone. The immunoblot was probed with anti-MBP antibody. Specifically bound p63␣ was detected only when MBP-p63␣ was incubated with mutant p53 or with its core domain (Fig.  4C, lanes 3 and 4). GST fusion proteins employed in this experiment are shown in the lower panel.
Thus, the reported results demonstrate that the core domain of mutant p53 can engage in vitro and in vivo in a physical interaction with p63␣. The association between human tumorderived p53 mutants and p63␣ might occur directly.
The DNA Binding Domain of p63 Is Involved in the Association with Mutant p53-We have previously reported that the region of p73␣ including the sequence-specific DNA binding and the oligomerization domains is sufficient for the interac-tion with mutant p53 (43). In an attempt to identify the minimal domain of p63 involved in that association, we analyzed whether the DNA binding domain of p63␥ is sufficient for the association with mutant p53. To this end, we transiently co- transfected H1299 cells with a vector encoding an HA-tagged version of full-length p63␥ or its DNA binding domain together with a plasmid encoding mutant p53His175. As seen in Fig. 5, p53His175 was detected in immunoprecipitates derived either from cells overexpressing full-length p63␥ or the p63 DNA binding domain. Similar data were obtained employing an in vivo binding assay in which GST-p63␥, GST-p63␥-(140 -321), and GST alone were incubated with cell lysates of H1299 cells overexpressing p53His175 (data not shown).
Thus, the DNA binding domain of p63 is sufficient for the association with human tumor-derived p53 mutants.
Functional Inactivation of p63 by Mutant p53-We have previously reported that the association with mutant p53 interferes with transcriptional activity of p73␣, -␤, -␥, and -␦ (43). To address the functional relevance of mutant p53 binding to p63␣, we performed DNA binding and transactivation assays. To this end, DNA binding reactions were employed using GST-p63␣, radiolabeled oligonucleotide resembling the p53 binding site of Bax promoter with or without cell lysates derived from the indicated cell lines (Fig. 6A). As shown in Fig. 6A, the binding of p63␣ to DNA was specific, because it was supershifted by the addition of anti-p63 antibody but not by anti-p73 antibody (Fig. 6A, lanes 7 and 8). Furthermore, a 200-fold molar excess of unlabeled probe specifically inhibited this binding (Fig. 6A, lane 5) but not a 1000-fold molar excess of an unrelated probe (Fig. 6A, lane 6). Mutant p53His175 strongly reduced the binding of p63␣ to DNA (Fig. 6A, lane 4). Of note, a similar effect was also seen in the presence of cell extracts derived from H1299 cells overexpressing the core domain of mutant p53 (Fig. 6A, lane 12).
Transactivation assays were performed by cotransfection of H1299 cells with p63␣ together with mutant p53His175 or with mutant p53His175 (74 -298) and a luciferase reporter gene driven by the p53-responsive Bax promoter. In agreement with DNA binding results, mutant p53His175 as well as its core domain markedly reduced the transcriptional activity of p63␣ (Fig. 6B).
To determine whether mutant p53 can also interfere with the activation of endogenous target genes by p63␣, we analyzed the levels of p21 waf1 protein in cells overexpressing p63␣ alone or together with a plasmid encoding mutant p53His175 as well as its core domain. As seen in Fig. 7, overexpression of p63␣ caused a clear accumulation of p21 waf1 protein (lanes 2 and 6). This is markedly reduced when mutant p53 or its core domain is concomitantly overexpressed (lanes 3 and 8).
Taken together, these results demonstrate that human tumor-derived p53 can interact with p63 not only physically but also functionally.
The Binding of Mutant p53 to p63 Interferes in Vivo with the Recruitment of Its Target Genes-To further investigate whether the physical association with mutant p53 can interfere in vivo with the binding of p63 to a specific DNA binding site, we performed chromatic immunoprecipitation (ChIP) and coprecipitation experiments in the same cell context such as H1299 cells whose mutant p53His175 expression is tightly regulated by ponasterone A (H1299-p53His175#41) (43). In order to induce mutant p53 expression, the cell line was grown in the presence of ponasterone A (2.5 M/ml) for 24 h.
To perform ChIP experiments, cells with or without ponasterone A addition were treated with formaldehyde to cross-link proteins to DNA. The cross-linked chromatin derived from equivalent numbers of cells was immunoprecipitated by using either an anti-p63 or an unrelated anti-IgG antibody. Following immunoprecipitation, the cross-linking was reversed, and the amount of endogenous Bax, p21 waf1 , 14-3-3, and p53AIP1 promoters was monitored in each sample by PCR amplification using internal primers (33,35,(45)(46)(47). We found that Bax, p21 waf1 , and 13-3-3 promoters were present in the chromatin immunoprecipitates with anti-p63 antibody (Fig. 8A, upper  panels). Of note, these promoters were not present in identical chromatin immunoprecipitates derived from cells overexpressing p53His175 upon the addition of ponasterone A (Fig. 8A,  upper panels). As expected, anti-IgG immunoprecipitates derived from the above mentioned cells did not contain the above mentioned promoters. As shown in Fig. 8A, p63 does not bind to the p53AIP1 promoter. This may be related to target specificity or to the lack of p63 modifications that make it able to recruit directly such a p53 target gene. To evaluate whether the binding of p63 to the Bax, p21 waf1 , and 14-3-3 promoters was specific, we applied the chromatin immunoprecipitation assay to the GADPH promoter that does not contain the binding site for p63. Thus, the glyceraldehyde-3-phosphate dehydrogenase promoter was present neither in anti-p63 nor anti-IgG immunoprecipitates (Fig. 8A, lower panel).
To perform co-precipitation experiments, cell extracts derived from H1299-p53His175#41 with or without ponasterone A stimulation were immunoprecipitated either with anti-p63 and anti-IgG polyclonal sera and subjected to immunoblot with a mixture of anti-p53 mAbs DO1 and 1801. As shown in Fig.  8B, mutant p53His175 was only present in immunoprecipitates derived from H1299-p53His175#41 cells upon induction with ponasterone A (lane 6). The predominant nuclear localization of mutant p53His175 upon ponasterone A induction is shown in Fig. 8C.
To further verify whether the binding of p63 to its target gene promoters was impaired in cells carrying endogenous mutant p53, we performed ChIP experiments in T47D cells. To this end, cross-linked chromatin derived from T47D cells was immunoprecipitated and processed as reported in Fig. 8A. We found that the binding of p63 to Bax, 14-3-3, and cyclin G promoters (upper panels) is impaired, whereas the binding to Mdm2 promoter (lower panel) is still present (Fig. 9) (48).
Taken together, the results of the ChIP (Figs. 8A and 9) and the co-precipitation experiments (Figs. 1 (A and B) and 8B) clearly indicate that, only upon binding to mutant p53, endogenous p63 is unable to recruit specific target genes.
Overexpression of Specific Human Tumor-derived p53 Mutants Markedly Reduces p63-mediated Growth Suppression-To further investigate the biological relevance of the association between mutant p53 and p63, we investigated whether exogenous expression of mutant p53 interferes with p63␣-mediated growth suppression. To this end, SAOS2 cells were cotransfected with p63␣ together with a plasmid encoding p53His175, or p53His273, or p53His175-(74 -298) and as a control with an empty vector. Overexpression of p63␣ suppressed colony formation of SAOS2 cells as compared with that of cells transfected with empty vector (Fig. 10A). Conversely, SAOS2 cells regained colony formation when mutant p53 was overexpressed (Fig. 10A). Interestingly, the core domain of mutant p53 could also promote such an effect (Fig. 10A). In agreement with the above reported observations, we found that transient overexpression of p63␣ strongly inhibited cell growth of SAOS2 cells, whereas coexpression of p53His175 restored such property to a similar extent as control cells (Fig. 10B). Unlike mutant p53, wt-p53 is per se able to inhibit colony formation of SAOS2 cells, but when co-expressed with p63␣ it does not interfere with p63-mediated growth suppression (Fig. 10A).
In an attempt to verify whether each p53 mutant can bind and counteract p63 activities, we assessed the effects induced by mutant p53Gly281 overexpression on the p63-mediated growth suppression. As shown in Fig. 11A, mutant p53Gly281 does not counteract p63-mediated suppression of SAOS2 colony formation. The requirement of the physical association of mutant p53 to p63, in order to impair the activities of the latest, is clearly indicated by the finding that mutant p53Gly281 does not interact with p63␣ (Fig. 11B), whereas, under the same experimental conditions, p53His175 binds to p63␣ (Fig. 11C).
To explore whether the counteracting effect of mutant p53 on p63-mediated growth suppression occurs through the reduction of p63 protein levels, we looked at the protein levels of p63␣ upon overexpression of p53His175 or p53Gly281 in SAOS2 cells. We found that such overexpression does not impact on the protein levels of p63␣ (Fig. 11D).
Taken together, these results contribute to further define a functional role for the association between specific human tumor-derived p53 mutants and p63. DISCUSSION Recent reports have clearly indicated the existence of a network of protein-protein interactions between the members of the p53 family in cancer cells (33,34,42,43). Here we provide further evidence on the physical association between human tumor-derived p53 mutants and p63. In agreement with the data reported by Gaiddon et al. (42), we show that the interaction between mutant p53 and p63 occurs under physiological conditions. The simplest way to interpret such in vivo interaction could be that in tumors bearing mutant p53 growth inhibition, apoptosis or differentiation induced by p63 is impaired. In support of this, we show that overexpression of mutant p53 markedly reduces transcriptional activity of p63 as well as FIG. 8. Mutant p53 interferes in vivo with the binding of p63 to target gene promoters. A, cross-linked chromatin from H175#41 with or without ponasterone A treatment was immunoprecipitated with antibodies to p63 and IgG and analyzed by PCR with primers specific for the indicated promoters (see "Experimental Procedures"). Input corresponds to nonimmunoprecipitated cross-linked chromatin. B, H1299-inducible cell line was generated as reported under "Experimental Procedures." Cell extracts were prepared from H1299-p53His175#41 24h after the addition of 2.5 M of ponasterone A (lanes 1-3). Identical cell extracts were prepared from untreated cells. Aliquots of 1.5 mg of protein were subjected to immunoprecipitation with anti-p63 or with anti-IgG polyclonal sera. C, subcellular localization of mutant p53His175 is shown. Cells were stained with 4Ј,6-diamidino-2-phenylindole to visualize nuclei and a mixture of anti-p53 mAbs DO1 and 1801 to visualize mutant p53.
FIG. 9. The binding in vivo of p63 to some of its target gene promoters is impaired in T47D cells. Cross-linked chromatin from T47D cells was immunoprecipitated with antibodies to p63 and IgG and analyzed by PCR with primers specific for the indicated promoters (see "Experimental Procedures"). Input corresponds to nonimmunoprecipitated cross-linked chromatin. counteracting growth suppression by p63. The tumor suppressor function of p63 and p73 has been challenged by several observations (49). To date, both p63 and p73 have been found rarely mutated in human tumors. Indeed, several studies have reported that overexpression of p73 or ⌬Np63 is present in neuroblastoma, colorectal cancer, bladder cancer, nasopharingeal carcinomas, squamous-cell carcinoma of head and neck, and hepatocellular carcinoma (4, 50 -57). Thus, it is possible that overexpression of p63 and p73 in tumors is tolerated by the presence of mutant p53. p73 and possibly p63 can be activated in response to DNA damage. Indeed, it has been reported that p73 can be stabilized and tyrosine-phosphorylated in response to cis-platin and ␥-irradiation, respectively (58 -60). These post-translational modifications of p73 require a competent kinase-active c-Abl (58 -60). A more recent work has re-ported that, in response to DNA damage, p73 can be acetylated and consequently driven to specific target genes (61). We have recently shown that the co-activator YAP depicts specificity in binding and enhancing transcriptional activity of p53 family members (36,62,63). YAP binds to p63 and p73 but not p53 and promotes their ability to activate proapoptotic genes such as Bax (36). 2 Taken together, these results indicate that in response to DNA damage or diverse types of stress, p63 and/or p73 can promote apoptosis, being involved in alternative pathways to those recruiting p53. The identification of specific p63 or p73 target genes should highlight the downstream events of such pathways. Gain of function of mutant p53 could result in 2 S. Strano and G. Blandino, unpublished observations. FIG. 10. Mutant p53 overexpression rescues p63-induced growth suppression. A, SAOS2 cells were grown in 60-mm dishes and transfected with the indicated plasmids together with a selectable marker plasmid. The ratio between p63 expression plasmid (2 g/transfection) and the ones encoding for p53 mutants (6 g/transfection) and its core domain was 1:3. An equal amount of pBabepuro (0.5 g/transfection) was added to each transfection. Cells were replated and selected with puromycin as reported under "Experimental Procedures." The data shown represent the average of number of colonies formed relative to cells transfected with the marker alone. Error bars indicate S.D. of a representative experiment out of three performed in triplicate. B, SAOS2 cells were transiently transfected with the indicated plasmid combinations. The total amount of transfected DNA was maintained constant by the addition of an empty control vector. 48 h later, cells were fixed in paraformaldehyde for 10 min at room temperature. After rehydration with phosphate-buffered saline for 5 min, cells were visualized and photographed with the aid of phase-contrast microscopy.
binding and sequestering p63 from pathways, culminating in either growth arrest or apoptosis upon cancer treatment. As shown by chromatin immunoprecipitation experiments (Figs. 8 and 9), the binding of endogenous p63 to the Bax, p21 waf1 , 14-3-3, cyclin G, and p53AIP1 promoters is impaired in cells carrying either exogenous or endogenous mutant p53. This might represent a molecular basis to the impaired apoptotic activity of p63 in tumors bearing mutant p53.
We originally reported that the core domain of mutant p53 and a large region of p73, including the DNA binding and the oligomerization domains, mediate this association in vitro and in vivo (43). A more recent work has reported that the core domain of mutant p53 can either bind to p73 or to p63 (42). By deletion studies, we now report that the interaction between mutant p53 and p63 occurs through the respective DNA binding domains. The identification of the minimal stretch of residues of p63 and mutant p53 directly involved in this interaction may allow the design of small peptides aimed to disassemble the protein-protein complex and to make available p63 for anti-tumor effects. The core domain of mutant p53 has been regarded as inactive, since it cannot bind and activate wt-p53 target genes. Taken together, these observations with the previously reported data define the core domain of mutant p53 as a protein-protein interaction module that might contribute to gain of function of mutant p53 by sequestering and inactivating proteins required for anti-tumor functions. Further support for the potential role of the core domain in gain of function activity of mutant p53 might be provided by its ability to markedly reduce p63 transcriptional activity and to counteract p63-induced growth inhibition as found in transactivation and colony suppression assays. These findings do not exclude the possibility that the N terminus and/or C terminus (or termini) of mutant p53 can play a role in oncogenic activity of mutant p53 (64 -67). Indeed, mutant p53Gly281 that has been reported to exert, in vitro and in vivo, gain of function activity through its N-terminal transactivation domain does not associate with and counteract p63-mediated growth suppression (Fig. 11, A and  B). These findings might allow the definition of two classes of gain of function p53 mutants. The first one may account for mutants (p53His175 and p53His273) whose gain of function activity relies on protein-protein interactions with p63 and p73, whereas the second one includes p53 mutants (p53Gly281) that exert such activity independently from the physical association with the p53 family members. Thus, gain of function of mutant p53 might result from a combination of specific protein-protein interactions as well as from the activation or repression of specific target genes. In this case, a model based on the sequential role for the domains of mutant p53 can be considered. A rather speculative hypothesis might suggest that the core domain of mutant p53 plays the major role in sequestering and inactivating proteins required for biological activities of oncosuppressor proteins, such as p63 and p73, whereas both N-and C-terminal regions can mainly control activation or repression of specific target genes.