SCH529074, a Small Molecule Activator of Mutant p53, Which Binds p53 DNA Binding Domain (DBD), Restores Growth-suppressive Function to Mutant p53 and Interrupts HDM2-mediated Ubiquitination of Wild Type p53

Abrogation of p53 function occurs in almost all human cancers, with more than 50% of cancers harboring inactivating mutations in p53 itself. Mutation of p53 is indicative of highly aggressive cancers and poor prognosis. The vast majority of mutations in p53 occur in its core DNA binding domain (DBD) and result in inactivation of p53 by reducing its thermodynamic stability at physiological temperature. Here, we report a small molecule, SCH529074, that binds specifically to the p53 DBD in a saturable manner with an affinity of 1–2 μm. Binding restores wild type function to many oncogenic mutant forms of p53. This small molecule reactivates mutant p53 by acting as a chaperone, in a manner similar to that previously reported for the peptide CDB3. Binding of SCH529074 to the p53 DBD is specifically displaced by an oligonucleotide with a sequence derived from the p53-response element. In addition to reactivating mutant p53, SCH529074 binding inhibits ubiquitination of p53 by HDM2. We have also developed a novel variant of p53 by changing a single amino acid in the core domain of p53 (N268R), which abolishes binding of SCH529074. This amino acid change also inhibits HDM2-mediated ubiquitination of p53. Our novel findings indicate that through its interaction with p53 DBD, SCH529074 restores DNA binding activity to mutant p53 and inhibits HDM2-mediated ubiquitination.

The tumor suppressor protein p53 integrates the cellular response to various stresses, including DNA damage, hypoxia, oncogenic transformation, viral infection, and metabolic changes. Activation of p53 in response to stress can trigger either cell cycle arrest and DNA repair, cellular senescence, or apoptosis (1). It accomplishes these activities by being at the center of a network of many proteins whose functions it regulates via transcription or direct interaction (2,3). Inactivation of the p53 pathway occurs in almost all tumors (4), with more than 50% of human cancers harboring mutations in p53 that abrogate its function. Mutation in p53 is often a characteristic of aggressive, treatment-refractory cancers (3,5). The majority of inactivating mutations in p53 are missense mutations that reside in the central core DNA binding domain (DBD) 4 (3,5). These mutations can be divided into two main classes as follows: contact point mutations, which alter a residue involved in contact with DNA, and structural mutations, which affect p53 function by distorting the structure of p53 and reducing its thermal stability (6,7). Both types of mutations alter the ability of p53 to bind to response elements in a variety of target genes and hamper the role of p53 in both transcriptional regulation (8) and transcription-independent apoptosis (9). Moreover, some mutant forms of p53 have been reported to impart a gainof-function phenotype, promoting cell growth and tumorigenicity (10,11).
Recent reports have suggested that the restoration of p53 function in mouse tumor models can lead to tumor regression and senescence, suggesting that the restoration of p53 function would be an effective target for therapeutic intervention (12,13). Several approaches have been taken to reactivate mutant p53, including treatments with Mab421, carboxyl-terminal peptide of p53, or small molecules such as CP-31398, PRIMA1, and the peptide CDB3 (14 -24). Both CP-31398 and PRIMA1 have been demonstrated to reduce tumor growth in animal models (16,17). The mechanism for CP-31398 has not been elucidated; however, a recent study indicates that PRIMA1 acts as an alkylating agent, covalently modifying p53 in vivo, but fails to bind p53 in vitro (25). However, the negatively charged nonapeptide CDB3 has been shown to bind p53 core domain and act as a chaperone, resulting in reactivation of mutant p53 (18,19).
Our current studies report on a small molecule (SCH529074) identified from our chemical library using a screen based on a p53 DNA binding assay (24). SCH529074 promotes the DNA binding activity of mutant p53 both in cell-free systems and in tumor cells, induces apoptosis in tumor cells, and reduces tumor growth in a xenograft model. Using radiolabeled [ 3 H]SCH529074, we have demonstrated the following: (i) SCH529074 binds to p53 core domain, and (ii) the compound, like a chaperone, is specifically displaced from p53 by cognate □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental "Materials and Methods" and Figs. S1-S8. 1  DNA oligonucleotide. In addition, we have shown that the binding of SCH529074 to p53 DBD inhibits HDM2-mediated ubiquitination and stabilizes wild type p53. We have also constructed a novel variant of p53 by changing a single amino acid in DBD, and we showed that the variant that maintains wild type DNA binding activity does not bind to SCH529074 and mimics the small molecule in inhibiting ubiquitination of p53 by HDM2.
Flow Cytometry Analysis-H1299, WiDr, DLD-1, and MB468 cells were treated for 24 h with either DMSO or 4 M SCH529074. The cells were then trypsinized, washed in phosphate-buffered saline, and stained according to the ApoAlert kit (BD Biosciences) and analyzed on a FACSCalibur instrument (BD Biosciences). H1299 tetracycline-inducible cells were treated with or without 1 g/ml doxycycline and either DMSO or 4 M SCH529074 for 24 h and then treated as above.
Cell Lysis and Western Blotting-Cell lysis of WiDr and H322 cells was carried out using the method of Foster et al. (16). 40 g of cell lysate was run either on a 10% polyacrylamide gel (for p53 and actin) or a 4 -20% polyacrylamide gradient gel (p21 and BAX). The gels were transferred to nitrocellulose paper and probed with antibodies to p53 (DO1), actin (Sigma), Bax (Santa Cruz Biotechnology), or p21 (Calbiochem).
Conformational Antibody Epitope Protection Assay-The conformation antibody epitope assay was performed according to the method of Foster et al. (16).
Chromatin Immunoprecipitation Assay-WiDr cells were plated at 10 7 cells/ml and then treated for 24 h with either 1% DMSO (untreated) or 2.5 g/ml SCH529074 in 1% DMSO final. The chromatin immunoprecipitation assay was performed as described by Frank et al. (27). Immunoprecipitation was performed in the presence or absence of p53 antibody DO1. Reverse transcription-PCR was performed with 10 l of DNA, 800 nM primers, and fluorescent probe diluted in a final volume of 30 l. The accumulation of fluorescent products were monitored on an ABI Prism 7700 sequence detector (Applied Biosystems). The primer/probe pairs used were described in Demma et al. (24).
Calcein AM Proliferation Assay-SCH529074 was diluted to the appropriate concentrations in complete media in a 96-well tissue culture plate, for a total volume of 100 l/well. Cells were then seeded at a density of 5000 cells/well bringing the total volume per well to 200 l. The plates were incubated at 37°C for 72 h. The media were removed, and 50 l of calcein AM (Molecular Probes) at a concentration of 10 g/ml was added to each well and incubated for 10 -20 min at room temperature in the dark. The plate was read at absorbance of 485/535 nm.
Bromodeoxyuridine Proliferation Assay-Bromodeoxyuridine proliferation assay (Roche Applied Science) was performed by plating 4000 cells per well in 96-well plates, and then the cells were treated with a titration of SCH529074 from 0 to 20 M. Plates were incubated for 72 h and then processed per the manufacturer's directions.
Tetracycline-inducible Cell Lines-Tetracycline-inducible cell lines were constructed using the tetracycline operator cited in Meyer-Ficca et al. (28) as adapted by Invitrogen. The coding sequence of each p53 mutant (wild type p53, p53 R249S, p53 R273H, and p53 R175H) was cloned into the ApaI and PstI sites of pCDNA4/TO. The plasmids were then cotransfected with pCDNA6.0, which expresses the tetracycline repressor protein, and subjected to selection with 3 g/ml blasticidin and 250 g/ml Zeocin (Invitrogen). Clones were selected and analyzed for protein expression by inducing expression with 1 g/ml doxycycline.
Quantitative DNA Binding Assay and Electrophoretic Mobility Shift Assay (EMSA)-The quantitative DNA binding assay and EMSA were performed as described in Demma et al. (24).
Scintillation Proximity Assay (SPA)-SPA was performed using glutathione-yttrium silicate SPA beads from GE Healthcare (31). At assay start, 22.5 l of buffer was added to each well. To this, 5 l of [ 3 H]SCH529074 or [ 3 H]CDB3 peptide was added per well, and the plate was spun down at 1000 rpm for 5 min. Then, 5 l (5 g) of protein was added, followed by 12.5 l (31.25 g) of glutathione-yttrium silicate SPA beads. The plate was then shaken at setting 6 on a Lab-Line Titer plate shaker. At 1 h, the plate was spun at 1000 rpm for 5 min and read on a TopCount machine from Packard (PerkinElmer Life Sciences). [ 3 H]SCH529074 was prepared in three steps with the label incorporated via sodium borotritide reduction of p-chlorobenzophenone. The resulting 3 H-labeled alcohol was converted to chloride by treatment with thionyl chloride and coupled with the piperazine intermediate to generate the target compound at a specific activity of 16.7 Ci/mmol. [ 3 H]CDB3 peptide was prepared via Pd/C-catalyzed tritium gas reduction of the dehydroleucine intermediate at a specific activity of 59 Ci/mmol.
In Vivo Efficacy and Pharmacokinetic Studies-All animal studies were carried out in the animal facility of Schering-Plough Research Institute in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. After a week of acclimation, female nude mice (Crl: Nu/Nu-nu Br; Charles River Laboratories, Wilmington, MA), 5-7 weeks of age, received subcutaneous inoculation of DLD-1 human colorectal carcinoma cells, 5 million per mouse, on day 0. For every 12-h oral gavage of 20% hydroxylpropyl-␤-cyclodextran as vehicle, 30 or 50 mg/kg SCH529074 was started on day 3 after randomization (10 mice per group) until day 31. Change in tumor size was measured by an electronic caliper connected to the computer application program LABCAT twice a week. Tumor volume was measured in three dimensions and calculated with the formula of V ϭ 1/6 ϫ ϫ L ϫ W ϫ T, where L, W, and T represent length, width, and thickness respectively. On day 31, at different times (2, 7, and 12 h) after the final dosing, three mice were euthanized at each time point with carbon dioxide, and blood was collected through cardiac puncture. Plasma was prepared from the collected blood. Quantification of SCH529074 levels was achieved using acetonitrile precipitation, followed by high performance liquid chromatography-atmospheric pressure chemical ionization tandem mass spectrometry.

SCH529074 Restores DNA Binding Activity and "Wild Type"
Conformation to Mutant p53-To identify potential small molecules that restore DNA binding activity and reactivate mutant p53, we used a highly sensitive p53 DNA binding assay that we described previously (24). In this assay, we used recombinantly produced p53 core DNA binding domain (amino acids 92-312) of a p53 mutant (R273H) fused to glutathione S-transferase (GST) and measured its ability to bind a biotinylated oligonucleotide that contained a consensus p53 DNA-binding site The biotinylated oligonucleotide was then captured by a magnetic streptavidin-coated bead, and bound p53 was detected by an antibody specific to the protein followed by a ruthenium-labeled secondary antibody. The complex was then analyzed with an electrochemiluminescence detection system. Using this assay, we screened a directed chemical library of quinazolines and identified a series of small molecules that restore DNA binding activity in the mutant p53. One of these compounds is SCH529074, which belongs to a class of piperazinyl-quinazolines (Fig. 1a) and restores DNA binding activity to a recombinant GST fusion mutant p53 DNA binding domain (R273H) in an electrophoretic mobility shift assay (Fig. 1b).
Using the quantitative DNA binding assay, we have shown that SCH529074 restores DNA binding activity to two mutant forms of p53, the contact point mutant R273H and the structural mutant R249S (Fig. 1, c and d). In both cases, SCH529074 significantly increased the amount of p53 protein that binds to DNA (B max ) as well as the binding affinity (K d ) for DNA. For wild type p53, SCH529074 stimulates the DNA binding activity by modestly increasing B max and affinity (Fig. 1e). These results are consistent with SCH529074 acting as a noncompetitive agonist to restore and stimulate DNA binding activity in mutant and wild type p53, respectively.
To assess if SCH529074 could protect wild type p53 conformation from thermal denaturation, we immobilized GST-fused wild type p53 DNA binding domain to microtiter wells in a similar manner to Foster et al. (16). In this assay, preservation of p53 active conformation is measured by the binding of monoclonal antibody 1620 to p53, which recognizes an epitope previously shown to be sensitive to denaturation. SCH529074 was able to maintain the 1620 antibody epitope from thermal denaturation when compared with treatment with DMSO alone (Fig. 1f). In addition, SCH529074 was able to protect the conformation of wild type p53 to a much greater extent than CP-31398, a previously described small molecule reported to protect p53 conformation from thermal denaturation ( Fig. 1f) (16).
To test the cellular activity of SCH529074, WiDr cells, a colorectal tumor cell line that harbors the p53 contact point mutation R273H, and KLE cells, an endometrial cancer cell line that has an R175H structural mutation, were treated with 4 M SCH529074 for 24 h. Nuclear extracts from the treated cells were used to study p53 DNA binding activity in the electrophoretic mobility shift assay. Binding to the 33 P-labeled deoxyoligonucleotide was observed only with the extract from the compound-treated cells but not from vehicle-treated cells (Fig. 2a). Binding to the labeled deoxyoligonucleotide was competed by unlabeled deoxyoligonucleotide. In addition, we similarly treated 2774 cells, an ovarian tumor cell line with an R273H contact point mutation with 4 M SCH529074. As seen with WiDr and KLE cells, binding to the labeled deoxyoligonucleotide was only observed with the extracts from treated cells, and binding was competed with an unlabeled deoxyoligonucleotide and not by an unlabeled deoxyoligonucleotide that contained a mutated p53 DNA-binding site (supplemental Fig. S1a). We also used the antibody supershift assay to further confirm the specificity of the gel shift assay (supplemental Fig. S1a, last lane).
We performed a chromatin immunoprecipitation assay to determine whether SCH529074 could restore the binding of mutant p53 to the promoters of p53-regulated genes in cells. As shown in Fig. 2b, mutant p53 isolated from SCH529074treated WiDr cells binds to the promoter region of the p21 and BAX genes but not to the ribosomal S9 gene. These results clearly show that SCH529074 has the ability to restore promoter-specific DNA binding activity to mutant p53 in tumor cells.
To demonstrate that the restoration of DNA binding activity was due to a conformational change of mutant p53 in cells, we immunoprecipitated p53 from cells treated with DMSO or 4 M SCH529074 using the monoclonal antibody 1620 that recognizes the wild type p53 conformational epitope (Fig. 2c). Both WiDr cells and H322 cells, which contain contact point mutants of p53 reported to be recognized by the Mab1620, contain a small amount of wild type conformation p53 in the absence of SCH529074. Upon compound treatment, the amount of p53 with a wild type conformation increased significantly. Similar results were also observed in DLD-1 cells, which harbor a p53 mutant that is not recognized by the 1620 antibody. These results clearly suggest that SCH529074 can induce a conformational change and restore DNA binding activity to mutant p53. Varying concentrations of SCH529074 were incubated with GST-p53DBD (R273H) for 15 min, and then 32 P-labeled consensus deoxyoligonucleotide was added for an additional 15 min. The sample was then electrophoresed on a 6% TBE polyacrylamide gel run in 0.5ϫ TBE buffer. oligo, oligonucleotide. c-e, effects of SCH529074 on wild type (wt) and mutant R273H and R249S at recombinant GST-p53 DNA binding domains. The DNA binding assay was performed with 10 nM recombinant p53 DNA binding domain and increasing concentrations of biotinylated p53 consensus deoxyoligonucleotide in the presence or absence of 500 nM SCH529074. f, protection of conformation-dependent epitopes on p53 DNA binding domain by SCH529074. GST-p53 DNA binding domain was immobilized in microtiter wells and then treated with DMSO or 10 g/ml of either SCH529074 or CP-31398 and treated as described under "Materials and Methods." The remaining percent of epitope for Mab1620 or for the monoclonal antibody to GST is shown as a percentage of the unheated control. RLU, relative luciferase units.

SCH529074 Restores Transcriptional Activation and Growth
Suppression Functions to Mutant p53-To examine the effects of SCH529074 on p53-regulated gene expression, we treated WiDr, H322, and DLD-1 cells with 4 M SCH529074 for 24 h, prepared cell lysates, and Western-blotted for p53, p21, BAX, PUMA, and actin. Treatment with SCH529074 did not alter the levels of p53 or actin but significantly increased the levels of p53-regulated proteins p21 and BAX in all three cell lines (Fig.  3a). Expression of PUMA was significantly increased in both DLD-1 and H322 cells, although there was only a slight increase in the level of PUMA in WiDr cells. To further demonstrate that these effects of SCH529074 were specific for p53, WiDr cells were transiently transfected, either with a plasmid expressing siRNA against p53 (pSuperp53) or with the control empty vector plasmid (pSuper), and then treated with 0 or 4 M SCH529074 for 24 h. Cell lysates were prepared and Westernblotted for p53, p21, BAX, and actin. In pSuper (empty vector)transfected cells, there was a significant increase in the levels of p21 and BAX protein in the presence of SCH529074, although the level of p53 and actin were unaltered. Cells transfected with the p53-siRNA plasmid had significantly lower levels of p53, although levels of actin were unaffected (Fig. 3b). Importantly, induction of p21 or BAX was severely reduced or absent in these cells, demonstrating that the effects of SCH529074 were dependent on the presence of p53.
To examine the effects of SCH529074 on the expression of p53-regulated genes in cells with the same genetic background, we constructed a series of tetracycline-inducible mutant p53 genes (R273H, R249S, and R175H) in the p53 null cell line H1299 (Fig. 3c). Treatment of cells with both 4 M SCH529074 and doxycycline increased the expression level of mutant p53 proteins and caused a significant increase in the expression of p53 targets, p21, BAX, and PUMA in these cells as compared with treatment with DMSO alone (Fig. 3c). In the H1299-Tet-p53 cell lines, treatment with either SCH529074 or doxycycline alone also led to induction of the p53-responsive genes but to a much lower level than observed with the two compounds together. Due to the leakiness from the Tet-On promoter used, FIGURE 2. SCH529074 restores DNA binding activity and wild type conformation to mutant p53 in cells. a, DNA binding activity of a nuclear extract from WiDr and KLE cells treated with 0 or 4 M SCH529074. 1 g of nuclear extract was incubated with 33 P-labeled p53 consensus deoxyoligonucleotide in the presence or absence of unlabeled p53 consensus deoxyoligonucleotide and then subjected to electrophoresis on a 6% TBE gel. comp, SCH529074; oligo, oligonucleotide. b, quantitative PCR of p21, BAX, and S9 ribosomal promoter from chromatin immunoprecipitation assay of WiDr cells treated with 0 or 4 M SCH529074 and processed as described under "Materials and Methods." The assay was performed using an antibody (ab) that recognizes p53 (closed bars) or with no antibody as a control (open bars). c, immunoprecipitation of wild type conformation p53 using monoclonal antibody 1620 to p53. Cell lysates from WiDr, H322, and DLD-1 cells were treated with DMSO or 4 M SCH529074 and then immunoprecipitated with either control mouse IgG or monoclonal antibody 1620 and then blotted for p53 using anti-p53 antibody DO1. IP, immunoprecipitation.
there may be a small amount of p53 that is produced in the absence of doxycycline that is responsive to treatment with SCH529074 and results in p53-mediated transactivation of p53-mediated genes (28). In addition, after treatment with doxycycline, a large amount of mutant p53 is produced, along with an induction of some p53-mediated genes (Fig. 3c). In the case of the R273H mutant, there is some residual DNA binding activity that may mediate the induction of p53-responsive genes when highly overexpressed, as well as potential "gain-offunction" effects that are seen with p53 mutants in vivo (7,10,11,18,19).
To confirm the effects seen with SCH529074 in these engineered cell lines were due to induction of wild type conformation in mutant p53 by SCH529074, we immunoprecipitated doxycycline-induced p53 from SCH529074-treated and untreated cells with the monoclonal antibody Mab1620 (Fig. 3d). In the absence of SCH529074, only the p53 from cells containing the R273H mutant demonstrated any binding to Mab1620. Upon treatment with SCH529074, all three mutant p53 proteins showed strong binding to Mab1620, indicating that SCH529074 significantly induced wild type conformation in these engineered cell lines, consistent with results from the tumor cell lines (Figs. 2c and 3d).
In addition, we examined the expression profile of genes reported by Wei et al. (32) to have p53-binding sites in their promoters in the presence of SCH529074 in WiDr and H1299 cells. Cells were treated with 4 M SCH529074 for 4, 8, and 12 h, and RNA was isolated and assayed for expression of p53 target genes by quantitative-PCR (Fig. 4a). In WiDr cells, we observed an increase in the expression of BAX and CDKN1A (p21), similar to what was observed at the protein level (Figs. 3a and 4a), as well as increased expression of PMIAP, PRKCA, TNFR10b, and CCNG2 (Fig. 4a) at all time points following compound treatment. Little effect of the compound on the expression profile of these genes was observed in p53 null H1299 cells. Interestingly, PUMA expression was not significantly increased at either the protein or mRNA level in WiDr cells (Figs. 3a and 4a).
To directly compare the effects of SCH529074 to promote reactivation of mutant p53 with expression of wild type p53, we used tetracycline-inducible H1299 cell lines that express either mutant (R175H, R249S, and R273H) or wild type p53 and assayed for the induction of several p53-dependent genes (CDKN1A, BAX, BBC3, PMAIP, and CCNG1) by quantitative-PCR. The cells were treated with doxycycline (1 g/ml) or SCH529074 (4 M) or a combination of both for 8 h, and RNA was isolated from all cell lines. The induction of wild type p53 in the cell line was confirmed by Western blotting (supplemental Fig. S2). Reintroduction of wild type p53 into the H1299 null cells leads to a significant induction of p53-dependent genes (Fig. 4b). In the case of mutant p53 cell lines, treatment with doxycycline or SCH529074 alone did not result in a significant induction of any of the p53-dependent genes assayed. When cells were treated with both doxycycline and SCH529074, however, there was an up-regulation of all the p53- Cell lysates were then Western-blotted using anti-p53, anti-actin, anti-Bax, and anti p21 antibodies. c, effects of SCH529074 on p53 cellular activity on isogenic cell lines and Western blotting of tetracycline-inducible cell line lysates treated as indicated for 24 h. Immunoblots were probed with antibodies to p53, BAX, PUMA, p21, and actin. Dox, doxycycline. d, immunoprecipitation of wild type conformation of p53 using monoclonal antibody 1620. Cell lysates were treated as indicated, immunoprecipitated (IP) with either control mouse IgG or monoclonal antibody 1620, and then blotted for p53 using anti-p53 antibody DO1. dependent genes. In each mutant p53 cell line, the induction of the p53-dependent genes assayed in the presence of doxycycline and SCH529074 was slightly lower than what was observed with the expression of wild type p53.
Both Xue et al. (12) and Ventura et al. (13) observed that restoration of wild type p53 led to senescence and tumor regression in vivo. We tested to see if SCH529074 could effect proliferation in cells harboring mutant p53. In a variety of tumor cells that express different mutants of p53, treatment with SCH529074 led to a reduction of proliferation as measured in a bromodeoxyuridine proliferation assay (Table 1). We also observed a reduction in proliferation of wild type p53 tumor cells, although they were not as sensitive to the compound as the tumor cells with mutant p53. To confirm the effect on wild type p53 tumor cells, we treated H460 lung carcinoma cells for 48 h with 0 or 4 M SCH529074, which leads to an induction of p53-dependent genes p21 and BAX (Fig. 5), as well as an increased level of p53 protein, indicating stabilization of wild type p53 by the compound.
Because the expression data indicated that apoptotic genes such as BAX were being up-regulated, we tested to see if SCH529074 could induce apoptosis in p53 mutant tumor cell lines. We used annexin staining and FACS analysis to assay for cell death in MB-468 breast tumor cells (R273H), DLD-1 colon tumor cells (S241F), WiDr colon tumor cells (R273H), and H1299 lung carcinoma cells (p53 null) treated with 4 M SCH529074 for 24 h. All p53 mutant cell lines showed an induction of apoptosis as seen by an increase in early and late apoptotic cells (Fig. 6a). p53 null H1299 cells were generally unaffected by SCH529074.
We also examined the ability of SCH529074 to induce apoptosis in the H1299 cells engineered to express mutant p53 proteins. Cell lines were treated with doxycycline (1 g/ml) and SCH529074 (4 M) individually or in combination for 24 h. Treated cells were stained with annexin and analyzed by FACs and compared with DMSO-treated cells (Fig. 6b). Apoptosis was observed in all cell lines only in the presence of both doxycycline and SCH529074.
When WS-1 cells (normal human skin fibroblasts) were treated with a 3-fold higher amount of SCH529074, there were no significant effects on the cell cycle profile compared with untreated cells (supplemental Fig. S3a). As further evidence for the induction of apoptosis, both WiDr cells and MB-468 cells showed a greater than 2-fold increase in caspase 3/7 activity after 24 h of treatment with SCH529074, although H1229 null cells showed no increase in caspase activity (supplemental Fig. S3b). Treatment of wild type p53 containing H460 cells with SCH529074 leads to an accumulation of cells undergoing either a G 2 arrest or apoptosis (supplemental Fig. S3c), similar to what is seen in mutant cell lines, but requiring a longer drug treatment (48 versus 24 h).
In vivo anti-tumor activity was assayed in a DLD-1 xenograft model as described previously (16). DLD-1 cells were implanted into nude mice and then treated after oral administration of SCH529074 twice daily for 4 weeks. Following 4 weeks of treatment, a 79 and 43% reduction of tumor growth was achieved with 50 and 30 mg/kg doses, respectively (Fig. 7). The degree of tumor inhibition correlated with the plasma exposure of the compound (0.26 -0.55 M at 30 mg/kg and 0.39 -0.79 M at 50 mg/kg, 2-12 h post final dosing) and was consistent with the in vitro EC 50 of 0.27 M in the DLD-1 proliferation assay (supplemental Fig. S4).
SCH529074 Binds to p53 Core and Acts as a Chaperone-In an effort to demonstrate that SCH529074 interacts directly with the p53 core domain, we developed an SPA that is used to study molecular interactions in a homogeneous system (31). Binding of [ 3 H]SCH529074 to GST-p53 DBD was measured using glutathione-coupled SPA beads to capture the recombinant GST-p53 DBD along with any bound SCH529074. [ 3 H]SCH529074 interacts with GST-p53 but does not interact with a control GST fusion protein, GST-VEGFR (vascular endothelial growth factor receptor), or to glutathione-SPA beads alone (Fig. 8a). Binding of [ 3 H]SCH529074 to GST-p53

TABLE 1 Inhibition of proliferation for various tumor cell lines by SCH529074
The effect of SCH529074 on various tumor cell lines was measured by bromodeoxyuridine incorporation of the proliferation assay after 72 h of treatment with a titration of SCH529074. can be displaced by the nonradioactive compound (Fig. 8b). In addition, the binding is saturable with a K d of 1-2 M (Fig. 8c).

Cell line
To see the effect of cognate and noncognate deoxyoligonucleotides on the binding of both SCH529074 and CDB3 to p53DBD, we utilized the SPA assay. The cognate deoxyoligo-nucleotides were derived from a p53 consensus binding sequence (consensus) and from the BAX promoter p53 binding sequence (bax), which have different affinities for p53 (8). The consensus sequence was altered to generate a noncognate sequence (Fig. 8d). Both consensus and bax deoxyoligonucle-  50 value that is approximately the respective K d value for deoxyoligonucleotide binding, although the noncognate deoxyoligonucleotide does not have any effect (Fig. 8d). Additionally, SCH529074 is able to increase the affinity of the mutant R273H for DNA by severalfold, close to the values observed for wild type p53 protein, in a quantitative DNA binding assay ( Table 2).

otides displace [ 3 H]SCH529074 from p53 DBD with an EC
Friedler et al. (18) described a nonapeptide, CDB3, derived from p53-binding protein 2 (p53BP2) that restored DNA binding activity to mutant p53 by directly interacting with p53 core domain. CDB3 displaced [ 3 H]SCH529074 from p53 DBD with an EC 50 value similar to the K d value of the peptide for p53 (18,19) (Fig. 8e and supplemental Fig. S5a). The p53 carboxyl-terminal peptide reported previously to rescue mutant p53 via an unknown mechanism (15) failed to displace [ 3 H]SCH529074 from p53 DBD (Fig. 8e). A similar SPA displacement assay using [ 3 H]CDB3 showed that the [ 3 H]CDB3 binding to p53 DBD could be displaced by either unlabeled CDB3 or SCH529074 with an EC 50 value similar to K d value of either molecule (supplemental Fig. S5b). In addition, CDB3 was displaced by cognate DNA but not noncognate DNA in a manner similar to SCH529074 (supplemental Fig. S5c). Taken together, these results suggest that the binding of SCH529074 and CDB3 peptide to p53 DBD is mutually exclusive and the two molecules function by a similar mechanism. Friedler et al. (18,19) suggested a chaperone mechanism for CDB3-mediated p53 reactivation, where CDB3 binds to mutant p53 and induces a conformational change allowing DNA binding. Once bound to DNA, p53 undergoes a further conformational rearrangement and releases CDB3, which is then available for another p53 molecule. This differs from the mechanism proposed for PRIMA-1, which acts as an alkylating agent, and has been suggested to covalently modify mutant p53 (25).
SCH529074 Binds Specifically to p53 DBD, and the Binding Is Conformation-dependent-Because DNA binding domains of p53 homologs p63 and p73 share considerable sequence homology with p53 DBD (33) and bind to p53 response elements in p53 target genes, we evaluated SCH529074 and CDB3 for binding to DBDs of p63 and p73. We expressed and purified p63 and p73 DBDs (24,34) as GST fusion proteins and used them in the SPA binding assay (Fig. 9a). There was no apparent binding of either [ 3 H]SCH529074 (Fig. 9a, top panel) or [ 3 H]CDB3 (bottom panel) to p63 and p73 DBDs. These results clearly suggest that both SCH529074 and CDB3 bind specifically to p53 DBD. As expected, SCH529074, as well as CDB3, did not have any effect on p63 or p73 DNA binding activity (supplemental Fig. S6, a-c) (24,34).
One major difference between the DBD of p53 and its homologs is that both p63 and p73 DBDs are thermodynamically more stable than p53 (28,34). Unlike p63 and p73, human p53 DBD is extremely susceptible to mutational inactivation that lowers its thermodynamic stability (35). A second site suppressor mutation in p53, N268D, restores DNA binding activity by increasing the thermodynamic stability in some oncogenic mutants (35)(36)(37). The amino acid asparagine (Asn-268) in human p53 is not conserved and is substituted by aspartic acid in rat and mouse p53 and by arginine in chicken, Xenopus, and other species (38). Arginine is also the corresponding amino acid in p63 and p73 (Fig. 9b). To study the effects of amino acid substitution at codon 268 on compound binding, we engineered a single amino acid change at residue 268 from asparagine to either aspartic acid (p53N268D) or arginine (p53N268R). To confirm that the modified proteins are still capable of binding to cognate DNA, an EMSA with both variants (N268D and N268R) and wild type p53 DBD was performed. Neither of these substitutions affected DNA binding activity (Fig. 9c). On examining their binding to SCH529074 and CDB3, we found that aspartic acid substitution (N268D) did not have any effect; however, arginine at position 268 (N268R) significantly reduced binding to both the small molecule and the peptide (Fig. 9d). When either aspartic acid or arginine is included in a quadruple mutant that includes other second site suppressor mutations, such as M133L, V203A, and N239Y (Quad268R or Quad268N), similar results are observed, where the N268R variant binds DNA as effectively as the N268D variant but fails to bind either [ 3 H]SCH529074 or [ 3 H]CDB3 (supplemental Fig. S7, a-f). These results indicate that p53 variants with arginine at the codon 268 (N268R and Quad268R) have DNA binding and compound binding activities similar to p63 and p73. These results suggest that conformational changes induced by arginine at position 268 impede the binding to SCH529074 and CDB3 without affecting DNA binding activity.
SCH529074 Binding Inhibits HDM2-mediated Ubiquitination and Stabilizes Wild Type p53 in Tumor Cells-Recent literature suggests that binding of HDM2 to the amino terminus of p53 induces a conformational change in both proteins leading to an interaction between the central domain of HDM2 and the p53 DBD. This interaction is reported to be necessary for HDM2-mediated ubiquitination and proteasomal degradation of p53 (39,40). To evaluate the effect of binding of the conformational stabilizers (SCH529074 and CDB3) on HDM2-mediated ubiquitination of p53, we adapted a cell-free ubiquitination assay (30). 35 S-Labeled p53 was produced in cell-free FIGURE 7. In vivo efficacy of SCH529074 in human DLD-1 colorectal cancer xenograft model. Female nude mice, 5-7 weeks of age, received subcutaneous inoculation of DLD-1 human colorectal carcinoma cells, 5 million per mouse, on day 0. Mice were randomized on day 3 (10 mice per group) and treated with an oral gavage of 20% hydroxylpropyl-␤-cyclodextran as vehicle and 30 or 50 mg/kg SCH529074 every 12 h until day 31. Tumor volume was measured in three dimensions and calculated with the formula of V ϭ 1/6 ϫ ϫ L ϫ W ϫ T, where L, W, and T represent length, width, and thickness, respectively. On day 31, percentage of tumor growth inhibition was derived using mean tumor volume from different treatment groups. Tumor growth inhibition ϭ 1-100% ϫ mean tumor volume of the treatment group/mean tumor volume of the vehicle group.
transcription and translation reactions using reticulocyte lysate. This in vitro translated p53 was incubated with either HDM2 or human papilloma virus (HPV) E6 protein (41) together with ubiquitin-activating enzyme (E1), ubiquitin-conjugating enzyme (E2), and ubiquitin in the presence and absence of conformation stabilizers followed by electrophoresis FIGURE 8. Direct binding of [ 3 H]SCH529074 to p53 and displacement from DNA. a-c, SPA assay showing that SCH529074 interacts specifically with GST-p53 and does not bind nonspecifically to protein, glutathione SPA beads, or the GST fusion tag on the recombinant p53 (a), binding is competitive with unlabeled SCH529074 (b), and binding is saturable (c). d, SPA assay was performed with GST-p53 and [ 3 H]SCH529074 in the presence of double-stranded DNA deoxyoligonucleotides containing either the p53 consensus (con) binding sequence, Bax binding sequence, or an altered sequence. e, displacement of [ 3 H]SCH529074 from GST-p53 by CDB3 peptide. wt, wild type. (Fig 10a). A ladder of ubiquitinated p53 was seen with wild type p53 in the presence of either HPV E6 or HDM2, with more high molecular weight ubiquitinated p53 seen with HPV E6 than with HDM2 (Fig. 10a, DMSO lane) indicating that the viral protein is more efficient in p53 ubiquitination in this cell-free assay. In the presence of SCH529074, HDM2-mediated ubiquitination was significantly reduced, but HPV E6-catalyzed ubiquitination was not affected. Similar results were seen with CDB3 peptide (supplemental Fig. S8a).
We also studied the effects of CP-31398 and PRIMA1 in the ubiquitination assay. CP-31398 was reported to protect p53 core from thermal denaturation (16) and was also shown to inhibit HDM2-mediated ubiquitination in H460 cells, a lung tumor cell line with wild type p53. PRIMA1 has been reported to alkylate p53, although its effects may require other cellular factors, such as HSP90, to exert an influence on p53 in vivo (25,41). In our ubiquitination assay, CP-31398 inhibits HDM2-mediated ubiquitination but not HPV E6-catalyzed ubiquitination, although PRIMA1 does not affect either HDM2 or E6 mediated ubiquitination (Fig 10a). These results suggest that binding of SCH529074, CDB3, or CP31398 to the p53 core domain interferes with conformational changes required for HDM2 docking and ubiquitination.
As described above (Fig. 5, Table 1, and supplemental Fig.  S2C), SCH529074 can stabilize wild type p53, can induce transcription of p53 target genes, and induce apoptosis in tumor cells. We then evaluated the effect of SCH529074 on p53 ubiquitination by HDM2 in wild type p53 tumor cells. Treatment of H460 cells with 5 g/ml (8.8 M) SCH529074 leads to accumulation of p53 over a 2-h period (supplemental Fig. S8b). We then treated H460 cells with 5 g/ml (8.8 M) SCH529074 for 1 h, followed by the proteasome inhibitor ALLN at 50 g/ml for an additional 4 h. The cells were lysed and Western-blotted for p53. In the presence of SCH529074 or the proteasome inhibitor ALLN, the p53 level was increased. Upon treatment of the proteasome inhibitor alone, a distinct ladder of ubiquitinated p53 was observed, which was absent when SCH529074 was added along with the proteasome inhibitor (Fig. 10b). General ubiquitination was not inhibited by compound treatment.
Because the conformational change in the core domain of p53N268R impedes binding to SCH529074 and CDB3, we evaluated the ubiquitination profile of p53 N268R by HDM2 and HPV E6 (Fig. 10c). 35 S-Labeled wild type p53 and both variants (N268R and N268D) were produced in a cell-free transcription/ translation reaction, and their ubiquitination profiles were followed as before in the absence of any compound or peptide.
Interestingly, arginine substitution (N268R) significantly inhibited HDM2-mediated ubiquitination of p53. The substitution (N268R), however, did not have any noticeable effect on HPV E6-mediated ubiquitination, indicating that the viral protein is less sensitive to the conformational change present in the N268R variant. On the other hand, substitution of aspartic acid for asparagine in N268D did not affect p53 ubiquitination by either HDM2 or HPV E6.

DISCUSSION
Restoration of p53 function by genetic manipulation in vivo has been demonstrated to lead to tumor regression and clearance (12,13), making pharmacological rescue of mutant p53 by small molecules an attractive strategy for developing cancer therapeutics. In this study, we report the identification of a small molecule (SCH529074) that binds to p53 DBD and restores wild type p53 function and conformation to various oncogenic mutants of p53, induces apoptosis in tumor cells harboring mutant p53, and reduces tumor growth in a xenograft model. The binding of SCH529074 to p53 DBD also interferes with HDM2-mediated ubiquitination and stabilizes wild type p53 protein in tumor cells. A single amino acid change in the p53 core domain (asparagine to arginine at codon 268) significantly reduces the binding of SCH529074, and mimics the compound by interfering with ubiquitination by HDM2. This is the first report that a single amino acid change in the core domain of p53 (N268R) impairs binding of conformation stabilizers and blocks ubiquitination by HDM2 without affecting p53 DNA binding activity.
There is growing evidence that human p53 has evolved to be a highly dynamic and conformationally flexible protein to allow rapid regulation of various functions critical for maintaining genomic integrity of cells in response to different stress signals (42,43). As a consequence, human p53 is an intrinsically unstable protein with the DNA binding domain being highly susceptible to a variety of oncogenic mutations. The DNA binding domains of these oncogenic mutants are conformationally compromised, unstable at physiological temperature, and lack DNA binding activity. In principle, these mutants can be rescued by increasing the thermodynamic stability of their DBD as suggested by structural studies (44,45) and as shown using suppressor mutations (34). Furthermore, the nonapeptide CDB3 was shown by NMR to bind at the same site in wild type and mutant p53 core domain and induce a conformational change that stabilizes the core domain and restores DNA binding activity (45). Molecules like SCH529074 and CDB3 may act via a "chaperone" mechanism, binding to the DBD and inducing a conformational change resulting in p53 functionally competent to bind DNA at physiological temperatures, and the chaperone molecules are then released when p53 binds to cognate DNA (19). In this study, we demonstrate that SCH529074 binds p53DBD directly and is displaced by cognate DNA, suggesting that SCH529074, like the CDB3, restores DNA binding activity in mutant p53 via a chaperone mechanism. This differs from the recent mechanism proposed for another p53-reactivating compound, PRIMA-1 (25). Unlike SCH529074 and other conformational stabilizers, PRIMA-1 cannot restore p53 DNA binding activity by itself in vitro and requires other cellular factors to restore p53 activity in vivo (20,24,25,46). Our results with the purified DBDs of both classes of mutant p53 (contact point R273H and structural R249S) clearly show that SCH529074 restores DNA binding activity in both mutant DBDs by increasing the amount of the protein capable of binding to DNA and the affinity for DNA. This suggests that conformation stabilizers (SCH529074 and CDB3) rescue both classes of mutant p53 by binding to p53 core domain and increasing their thermodynamic stability at physiological temperatures (20,45).
The binding of SCH529074 and CDB3 to p53 DBD is mutually exclusive, and the IC 50 value for CDB3 displacement of SCH529074 is similar to the published K d value for the peptide (18,19), suggesting overlapping binding modes. In addition, the binding is specific for the p53 DBD, because neither SCH529074 nor CDB3 binds to the p53 homologs p63 and p73 in vitro. Because mutant p53 oligomerizes with p63 and p73 and negatively affects their function, it is possible that a conformational change in mutant p53 by SCH529074 may indirectly activate p63 and p73 resulting in the expression of p53 target genes that can also be activated by p63 and p73 (e.g. p21 and BAX) (47,48). However, results of the siRNA experiment demonstrate that depletion of mutant p53 prevents the activation of p53 genes after treatment with SCH529074, suggesting that the effects are solely due to reactivation of mutant p53 and not activation of p63 or p73.
The second site suppressor mutation asparagine to aspartic acid (N268D) has been shown to increase thermostability and restore wild type function to oncogenic p53 mutants (35,37). Accordingly, Joerger et al. (35) included this suppressor mutation when designing the superstable quadruple p53 mutant for structural studies, which revealed that the mutations de-crease intrinsic thermodynamic instability and increase rigidity of the protein. The suppressor mutation N268D, which is located within ␤-strand S10 of p53 DBD, results in an altered hydrogen bonding pattern linking S10 with ␤-strand S1. The newly formed hydrogen bond bridges two sheets of the ␤-sandwich scaffold in an energetically more favorable way (35). The N268D mutation retains binding to both SCH529074 and CDB3, although mutation of asparagine to arginine at this amino acid, which is the corresponding residue in found in p63 and p73 (34), abrogates binding of both SCH529074 and CDB3 (34). It is likely that arginine at amino acid 268, being larger than aspartic acid, may induce further alterations in the hydrogen bonding pattern. Because the mutation does not have any deleterious effect on DNA binding activity, it is possible that the arginine substitution increases stability and rigidity to a greater extent than the N268D substitution, thereby interfering with the binding of conformational stabilizers SCH529074 and CDB3.
Studies on p53 ubiquitination by HDM2 suggest that, in addition to binding p53 at the amino terminus, HDM2 must also interact with the p53 DBD in a region that contains amino acid 268 to mediate p53 ubiquitination (39,40). Our studies demonstrated that the N268R mutant is not ubiquitinated by HDM2, even though it retains DNA binding activity. The crystal structure of the p53 core-DNA complex suggests that the region is conformationally flexible, and the motif is constrained in wild type p53 (44) but is unfolded or misfolded in oncogenic mutants of p53 (49,50). Therefore, the conformational change induced by arginine at position 268 may prevent conformational changes required for HDM2 docking and thus inhibit HDM2-mediated ubiqitination of p53. Interestingly, HDM2 binds to p63 and p73 at their amino termini but fails to ubiquitinate these proteins (51,52). It is possible that the DBDs of p63 and p73, with the arginine at 268, are not conformationally flexible to allow HDM2 docking and subsequent ubiquitination.
It is likely that SCH529074 and CDB3 binding increases stability and rigidity of p53 DBD to an extent that it interferes with HDM2 docking and p53 ubiquitination, similar to the N268R mutation. Both SCH529074 and CDB3 may function in a similar role to facilitate additional hydrogen bonding that may link various ␤-strands in the p53 DNA binding domain, making the protein more thermodynamically stable. This would allow the compounds not only to rescue mutant p53 but to function as an activator of wild type p53 by inhibiting HDM2-mediated ubiquitination. Our data confirm that SCH529074 stabilizes wild type p53 in tumor cells with overexpression of HDM2 by interfering with HDM2-mediated ubiquitination, although not affecting ubiquitination in general. Wild type p53 tumor cell lines are not as sensitive to SCH529074 as mutant p53 tumor cells, and this may be due to the expression of HDM2 in the wild type cells, which may delay the effects of the compound by delaying the onset of p53 accumulation. Taken together, our results indicate that the N268R mutation mimics SCH529074 and CDB3, inducing the conformational change that interferes with HDM2-mediated ubiquitination. These results suggest that there is a conformational difference between DNA binding activity and HDM2 docking in the p53 DBD and also indicate the feasibility of developing therapeutic agents that will inhibit HDM2 docking to the DBD without abrogating p53 function.
Our studies for the first time clearly demonstrate that a small molecule, SCH529074, can bind to p53 DBD and restore wild type function to many oncogenic mutants, including ones at hot spot residues. These results provide a more complete biochemical understanding of the mechanism of reactivating mutant p53 by small molecules and may facilitate the development of novel cancer therapeutics that work via this mechanism. In addition, stabilizing wild type p53 by inhibiting HDM2-mediated ubiquitination presents an opportunity for these molecules to be used as a therapeutic approach for cancers with wild type p53 as well.