Report

In Vivo Activation of the p53 Pathway by Small-Molecule Antagonists of MDM2

See allHide authors and affiliations

Science  06 Feb 2004:
Vol. 303, Issue 5659, pp. 844-848
DOI: 10.1126/science.1092472

Abstract

MDM2 binds the p53 tumor suppressor protein with high affinity and negatively modulates its transcriptional activity and stability. Overexpression of MDM2, found in many human tumors, effectively impairs p53 function. Inhibition of MDM2-p53 interaction can stabilize p53 and may offer a novel strategy for cancer therapy. Here, we identify potent and selective small-molecule antagonists of MDM2 and confirm their mode of action through the crystal structures of complexes. These compounds bind MDM2 in the p53-binding pocket and activate the p53 pathway in cancer cells, leading to cell cycle arrest, apoptosis, and growth inhibition of human tumor xenografts in nude mice.

The tumor suppressor p53 is a potent transcription factor that controls a major pathway protecting cells from malignant transformation (1, 2). As such, it is the most frequently inactivated protein in human cancer (3). In response to stress, the cellular level of p53 is elevated by a posttranslational mechanism, leading to cell cycle arrest or apoptosis. Under nonstressed conditions, p53 is tightly controlled by the MDM2 protein through an autoregulatory feedback loop (48). p53 can activate MDM2 expression which, in turn, leads to the repression of p53 by three mechanisms. First, MDM2 binds p53 at its transactivation domain and blocks its ability to activate transcription. Second, it is involved in the nuclear export of p53. Third, MDM2 serves as a ubiquitin ligase that promotes p53 degradation (9).

The mdm2 gene has been found amplified or overexpressed in many human malignancies (10, 11). Therefore, activation of the p53 pathway through inhibition of MDM2 has been proposed as a novel therapeutic strategy (1214). Several recent studies have shown that disruption of the p53-MDM2 interaction by different macromolecular approaches or by the suppression of MDM2 expression can lead to the activation of p53 and tumor growth inhibition (1519). Here, we report the identification of potent and selective small-molecule inhibitors of the MDM2-p53 interaction with in vitro and in vivo antitumor activity.

Historically, it has been difficult to develop small-molecule inhibitors of non-enzyme protein-protein interactions. However, the crystal structure of MDM2 bound to a peptide from the transactivation domain of p53 (20) has revealed that MDM2 possesses a relatively deep hydrophobic pocket that is filled primarily by three side chains from the helical region of the peptide. The existence of such a well-defined pocket on the MDM2 molecule raised the expectation that compounds with low molecular weights could be found that would block the interaction of MDM2 with p53.

To identify compounds that could inhibit p53-MDM2 binding, we screened a diverse library of synthetic chemicals. Several lead structures were identified and optimized for potency and selectivity. One such class was a series of cis-imidazoline analogs that we named Nutlins (for Nutley inhibitor) (Fig. 1A). These compounds displaced recombinant p53 protein from its complex with MDM2 with median inhibitory concentration (IC50) values in the 100 to 300 nM range. The imidazoline compounds were synthesized as racemic mixtures from which the enantiomers could be separated with the use of chiral columns. As shown with Nutlin-3 (Fig. 1A), only one of them, arbitrarily called enantiomer-a, possessed potent binding activity, whereas the other one, enantiomer-b, was 150 times less active.

Fig. 1.

Structure and mode of binding of MDM2 inhibitors. (A) The inhibition of MDM2-p53 binding was analyzed with Biacore's surface plasmon resonance technology in a solution competition format (26). A concentration series of each compound was incubated with recombinant human MDM2 and then injected onto a chip surface with captured recombinant p53 protein (29). Binding was measured at equilibrium and calculated as the percentage of maximum binding. Cisimidazolines Nutlin-1 and Nutlin-2 were used as racemic mixtures. The enantiomers of Nutlin-3 (a and b) were separated with the use of a chiral column and tested individually. (B) A Molscript (26) rendition of MDM2 with the inhibitor Nutlin-2 located in the p53 binding pocket. (C) An overlay of Nutlin-2 (carbon atoms drawn as white spheres, nitrogen in blue, oxygen in red, and bromine in brown) with the peptide side chains of Phe19, Trp23, and Leu26. (D) Surface rendition of the p53 binding pocket of MDM2 (buried regions, green; exposed portions, red) showing one bromophenyl group buried deep in the Trp pocket. Color scheme is the same as in (C). Panels (C) and (D) were drawn with molecular operating environment (MOE) (26).

To investigate the mode of binding of these compounds, we determined the crystal structure of the human MDM2–Nutlin-2 complex (21). The crystals diffracted to 2.3 Å resolution and the structure verified that the inhibitor binds to the p53 binding site on MDM2 (Fig. 1B). The inhibitor mimics the interactions of the p53 peptide to a high degree, with one bromophenyl moiety sitting deeply in the Trp pocket (Fig. 1, C and D), the other bromophenyl group occupying the Leu pocket, and the ethyl ether side chain directed toward the Phe pocket. In essence, the imidazoline scaffold replaces the helical backbone of the peptide and is able to direct, in a fairly rigid fashion, the projection of three groups into the pockets normally occupied by Phe19, Trp23, and Leu26 of p53 (20). Thus, this cis-imidazoline may represent a useful small-molecule scaffold to project functional groups similar to a helical peptide.

According to the model for p53 regulation by MDM2 (79), the treatment of cells with an inhibitor of MDM2-p53 binding should result in (i) stabilization and accumulation of the p53 protein resulting from the blockage of its nuclear export and degradation, (ii) activation of MDM2 expression, and (iii) activation of other p53-regulated genes and the p53 pathway. At the cellular level, these molecular events should cause cell cycle arrest in G1 and G2 phases and/or apoptosis. Furthermore, all of the above should occur only in cells with wild-type p53 and not in cells with transcriptionally inactive mutant p53. To determine if the inhibition of p53-MDM2 binding by the imidazoline analogs can translate into activation of the cellular p53 pathway, we studied human cancer cells with wild-type p53 (such as colon cancer lines HCT116 and RKO) that previously have been shown to activate p53 signaling in response to genotoxic stress (22). Cell lines in which p53 is disabled by mutations or deletions (SW480: colon; MDA-MB-435: breast; and PC3: prostate) served as negative controls.

We first examined the effect of MDM2 inhibitors on the cellular levels of p53, MDM2, and p21Waf1/Cip1, a major transcription target of activated p53 (23). Incubation of exponentially growing HCT116 cells with Nutlin-1 for 8 hours led to a dose-dependent increase in the levels of all three proteins (Fig. 2A). In contrast, SW480 cells exposed to the same conditions showed high basal levels of p53 but no detectable MDM2 or p21 (24). Treatment with Nutlin-1 yielded analogous results in other cancer cell lines with wild-type p53 (RKO and SJSA-1) and mutant p53 (MDA-MB-435), respectively (25). These results confirmed that wild-type p53 accumulates in cells treated with Nutlin-1 and leads to an elevation in the levels of MDM2 and p21 proteins in a manner that is consistent with activation of the p53 pathway.

Fig. 2.

Inhibition of MDM2-p53 binding by Nutlin-1 activates the p53 pathway in cells with wild-type p53. (A) SW480 (mutant p53) and HCT116 (wild-type p53) cells were incubated with the indicated concentrations of Nutlin-1 for 8 hours and p53, MDM2, and p21 proteins were analyzed in the cell lysates by Western blotting. (B) Nutlin-1 treatment induces the expression of the p21 gene but not the p53 gene. Cells with wild-type p53 (HCT116, RKO, and H460a) were treated with Nutlin-1 for 8 hours, and the change in the level of transcription was measured by quantitative PCR and expressed as fold induction compared with the untreated control. (C) Nutlin-1 arrests cell cycle in the G1 and G2 phases. HCT116 and SJSA-1 cells were incubated with 4 μM Nutlin-1 or an equivalent amount of solvent for 22 hours and an additional 2 hours with 10 μM BrdU, and cell cycle distribution was analyzed after propidium iodide/fluorescein isothiocyanate–antibody to BrdU staining (30). Cells within the rectangles are in S phase. (D) Antiproliferative and cytotoxic activity of Nutlin-1. Exponentially growing cancer cells with wild-type p53 (HCT116, RKO, and SJSA-1) or mutant p53 (MDA-MB-435 and SW480) were incubated with a range of concentrations for 5 days and the cell mass and viability were measured by the MT T assay. (E) Inhibition of clonogenic cell growth. Cancer cells with wild-type p53 (HCT116 and RKO) or mutant p53 (MDA-MB-435, SW480, and PC3) were seeded at a low cell density and their ability to form colonies was measured after 5 days of incubation with Nutlin-1. (F) p53 activation by Nutlin-1 does not involve Ser15 phosphorylation on p53. Cancer cells were treated with doxorubicin (1 μM), etoposide (10 μM), or Nutlin-1 (6 μM) for 20 hours, and the amount of total p53 and p53 phosphorylated on Ser15 was determined by Western blottingin aliquots of cell lysates normalized for total protein.

To confirm that the accumulation of p53 was due to decreased degradation of the protein rather than elevated expression of the p53 gene, we treated three wild-type p53 cancer cell lines with Nutlin-1 for 8 hours and monitored the expression of the p53 and p21 genes by real-time polymerase chain reaction (PCR) (Fig. 2B). The transcription of p21 increased in a dose-dependent manner in all cell lines consistent with accumulation of its transcriptional activator p53 (Fig. 2A). By contrast, transcription of the p53 gene itself was unaffected by Nutlin-1, even at the highest concentration tested (16 μM), which caused an 8- to 10-fold induction of p21 mRNA. These data indicate that the Nutlins up-regulate p53 by means of a posttranslational mechanism.

One of the main cellular consequences of p53 activation in proliferating cells is cell cycle arrest in G1 and G2 phases. The cyclin-dependent kinase inhibitor p21 plays a major role in this arrest (22). Cell cycle analysis of bromodeoxyuridine (BrdU)–labeled HCT116 and SJSA-1 cells revealed increased G1 and G2/M phase fractions and nearly complete depletion of the S-phase compartment after 24 hours of treatment with 4 μM Nutlin-1 (Fig. 2C). Arrested cells did not show an increase in the mitotic index, indicating that the G2/M fraction consists of G2 phase cells. Similarly, G1 and G2 blocks were observed in RKO and SJSA-1 cells but not in the mutant p53 lines SW480 and MDA-MB-435 (25), confirming that Nutlin-1 activates a major cellular function of the p53 pathway.

Next, we examined the effect of the MDM2 inhibitor on the growth and viability of cultured cancer cells. Five exponentially growing cell lines, three with wild-type and two with mutant p53, were incubated with Nutlin-1 for 5 days and their cell mass and viability were measured with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay (26). The results showed a dose-dependent antiproliferative and cytotoxic activity that differed between cell lines depending on their p53 status (Fig. 2D). The IC50 values in the cells with wild-type p53 (1.4 to 1.8 μM) were substantially lower compared with those of cells with mutant p53 (13 to 21 μM), reflecting the fact that the p53 pathway can be activated only in cells with wild-type p53 (27). A similar clear separation of activity along the lines of p53 status was observed when five cancer cell lines were tested for the ability to form colonies in the presence of Nutlin-2 (Fig. 2E).

Further experiments were performed to exclude the possibility that the activation of p53 by Nutlins is independent of their effect on MDM2. The mechanism by which known genotoxic drugs activate p53 involves phosphorylation of the protein on specific serine residues near the MDM2 binding domain. Of these, Ser15 is phosphorylated most frequently (28). We therefore analyzed Ser15 phosphorylation in p53 from lysates of HCT116 and RKO cells treated for 24 hours with Nutlin-1 and two genotoxic drugs, doxorubicin and etoposide (Fig. 2F). Western analysis revealed that all three compounds induced the accumulation of p53, but only doxorubicin and etoposide caused phosphorylation of Ser15. These results suggest that a genotoxic mechanism is unlikely to contribute to activation of p53 by Nutlin-1.

To investigate the specificity of these inhibitors, we used the separated enantiomers of Nutlin-3. As shown earlier, although the two molecules have nearly identical physical and chemical properties, enantiomer-b is 150 times less potent in binding to MDM2 than enantiomer-a (Fig. 1A). Therefore, the inactive enantiomer serves as a valuable control for non–MDM2-related cellular activities. We treated HCT116 and SW480 cells with the active or inactive enantiomer of Nutlin-3 for 8 hours and monitored the expression of p53, MDM2, and p21 genes by real-time PCR (Fig. 3A). As previously shown with Nutlin-1 (Fig. 2B), enantiomer-a induced the expression of MDM2 and p21 (but not p53) only in cells with wild-type p53 (HCT116). Enantiomer-b had no effect regardless of the p53 status of the cells. This result was confirmed at the protein level by Western blotting (Fig. 3B). Similar results were obtained in the MTT assay (Fig. 3C). Only the active enantiomer showed a potent antiproliferative activity and clear separation of potency between the cells harboring wild-type p53 and those harboring mutant p53. The potency of enantiomer-b was much lower in the wild-type p53 cells and nearly identical to the potency of enantiomer-a against the mutant p53 cells. This indicated that the low level of cytotoxicity of the imidazoline compounds in cells with mutant p53 is both p53 and MDM2 independent.

Fig. 3.

Activation of p53 and apoptosis induction by MDM2 inhibitors is enantiomer specific. (A) Only the active enantiomer of Nutlin-3 activates the expression of p21 and MDM2 genes. Cells were incubated with the active (a) or inactive (b) enantiomer for 8 hours and the level of transcription was analyzed as in Fig. 2B. (B) p53 and p21 proteins are elevated by the active but not the inactive Nutlin-3 enantiomer. HCT116 cells were treated with enantiomer-a or -b of Nutlin-3 for 24 hours and aliquots from cell lysates containing 10 μg of protein were analyzed by Western blotting. (C) Effect of the enantiomers of Nutlin-3 on the growth and viability of cancer cells with wild-type p53 (HCT116, RKO, and SJSA-1) and mutant p53 (MDA-MB-435 and SW480). Cells were grown, treated, and analyzed as in Fig. 2D. (D) Active enantiomer of Nutlin-3 induces apoptosis in SJSA-1 cells. Cells were incubated with 10 μM enantiomer-a (green) or -b (blue) for 24 and 48 hours, and the TUNEL-positive cell fractions were measured by flow cytometry and compared with untreated controls (C, control). Bars represent the average value from two independent experiments. Error bars show the SEM. (E) Apoptosis in SJSA-1 cells treated with Nutlin-3. Cells were incubated with 10 μM enantiomer-a (a) or enantiomer-a plus 100 μM Z-VAD-FMK (b) or equivalent concentration of solvent (c) for 48 hours and apoptotic cells were stained by the TUNEL assay (green). Cell nuclei were stained blue with 4′,6′-diamidino-2-phenylindole. Scale bars, 50 μm.

Next, we investigated the ability of MDM2-p53 binding inhibitors to induce apoptosis in cancer cells with wild-type p53. SJSA-1 osteosarcoma cells were treated with 10 μM of the active or inactive enantiomer of Nutlin-3 for 24 and 48 hours and examined for apoptotic cells by terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining with flow cytometry (Fig. 3D) and fluorescence microscopy (Fig. 3E). Few dead cells were detected during the first 24 hours of incubation with both enantiomers. However, after 48 hours of exposure to the active enantiomer, 45% of the cell population became TUNEL positive (Fig. 3D). Cells treated with the inactive enantiomer were indistinguishable from the untreated controls. Treatment of SJSA-1 cells with 10 μM Nutlin-3 (active enantiomer) in the presence of the pancaspase inhibitor Benzyloxycarbonyl-Val-Ala-Asp (OMe) fluoromethylketone (Z-VAD-FMK) for 48 hours decreased the number of dead (propidium iodide–stained) cells from 47 to 5% (26) and substantially reduced the number of TUNEL-positive cells (Fig. 3E). This result suggests that caspase activation is involved in the Nutlin-induced apoptotic program. Apoptotic figures were also observed in HCT116 and RKO cells treated with the active Nutlin-3 enantiomer; however, these cells were less sensitive than SJSA-1 cells, which have an amplified MDM2 gene (25).

We also examined the effect of MDM2 inhibitors on the growth and viability of human and mouse normal diploid fibroblasts, which have a functional p53 pathway. In the MTT assay, the active enantiomer of Nutlin-3 inhibited the proliferation of exponentially growing human skin (1043SK) and mouse embryo (NIH/3T3) fibroblasts with 2.2 and 1.3 μM IC50, respectively, but the cells retained their viability after 1 week of incubation at concentrations as high as 10 μM. SJSA-1 cells, treated in parallel, were inhibited with comparable IC50 (1.5 μM) but lost their viability at concentrations above 3 μM.

Finally, we examined whether MDM2 inhibitors can suppress the growth of established tumor xenografts in nude mice. For this study, we chose Nutlin-3 and the human osteosarcoma cell line SJSA-1. Nutlin-3 was well tolerated upon oral administration of 200 mg/kg twice a day for 20 days and achieved steady-state plasma levels above its in vitro IC90 determined in SJSA-1 cells (3.5 μM). Nutlin-3 treatment of mice bearing tumors 100 to 300 mm3 in size resulted in 90% inhibition of tumor growth, relative to vehicle controls (Fig. 4). The mice did not lose significant weight and did not show any gross abnormalities upon necropsy at the end of the treatment. In a parallel experiment, doxorubicin administered intravenously at its maximal tolerated dose of 10 mg/kg inhibited the growth of SJSA-1 xenografts by 81%.

Fig. 4.

In vivo antitumor activity of MDM2 inhibitors. Nude mice (10 animals per dose group) bearing subcutaneous human cancer xenografts (SJSA-1) with mean volumes of 185 mm3 received 200 mg/kg of an oral dose of Nutlin-3 (racemic) twice daily or 10 mg/kg of intravenous doxorubicin once a week for 3 weeks. The tumor volumes were measured and recorded periodically duringthe course of the study. P < 0.001 for Nutlin-3 and doxorubicin compared with corresponding vehicle controls. Error bars show SEM.

Inhibition of MDM2 expression by anti-sense oligonucleotides has recently been shown to activate p53 and suppress tumor growth in mouse xenograft models (18, 19). However, this approach targeted the expression of the protein and could not distinguish between p53-dependent and p53-independent mechanisms because tumor cells with mutant p53 also responded to the treatment both in vitro and in vivo. The small-molecule inhibitors of MDM2-p53 interaction reported here have shown p53-dependent activity in multiple cellular models, strongly implying that their antitumor mechanism is derived from activation of the p53 pathway. However, the in vivo activity of the Nutlins may not be limited to tumor tissues. Our preliminary evidence suggests that p53 activation may affect xenograft growth by more than one mechanism, including effects on the tumor microenvironment (25). Although our data strengthen the notion that unleashing the powerful growth suppressive and proapoptotic activity of p53 by MDM2 antagonists is a potentially valuable strategy for treating cancer, further studies are needed to address the true therapeutic potential of the approach. This strategy may be applicable not only to tumors with aberrant MDM2 expression but also to tumors that have retained wild-type p53.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1092472/DC1

Materials and Methods

References

References and Notes

View Abstract

Navigate This Article