Covalent Rescue of Mutant p53

Mutation in the TP53 gene is the most common specific genetic alteration found in human cancer (1). The protein normally acts as a highly inducible transcription factor. In response to a wide range of stress signals, it transcribes at least 100 downstream genes that in turn limit cellular growth by inducing cell-cycle arrest, apoptosis, and senescence. Germline mutations in TP53 are responsible for the Li-Fraumeni cancer family syndrome (2), whereas missense mutations are associated with many cancers. Unusually for a tumor suppressor gene, the majority of these mutations result in single amino acid substitutions and the mutant proteins are highly expressed in cancer cells. Among these, the Y220C mutation is a common mutant. It is temperature sensitive, and the reduction in the volume of sidechain from Tyr to Cys creates a cavity/pocket that is solvent exposed. In this issue, Guiley and Shokat (3) exploit this specific cysteine to develop molecules that fit this pocket and additionally make a covalent bond with Y220C, stabilizing the mutant p53 in the wild-type conformation and resulting in promising chemical and biological activity.

An enormous frustration in cancer research has been the discovery of frequent specific genetic alterations in cancer cells that have not proved amenable to drug development. Principal among these undruggable targets have been the products of the MYC, RAS, and TP53 genes. Remarkable progress pioneered by Kevin Shokat's lab (4) has changed this dismal picture. By exploiting the presence of reactive cysteine groups present only in the G12C mutant KRAS, Shokat's team was able to develop covalent inactivators of mutant ras protein. A clinical compound (sotorasib) based on this work was recently approved by the FDA (5).

Here Guiley and Shokat (1) extend the concept of covalent binding drugs to mutant p53 based on the pioneering work from Alan Fersht's lab in Cambridge (6), which first identified a binding pocket in the Y220C mutant of p53 and developed a series of specific but noncovalent small-molecule (notably PhiKan083) reactivators of this mutant. The company PMV pharma separately has announced the results of a clinical trial of a Y220C-binding noncovalent compound at this year's American Society of Clinical Oncology Annual Meeting (7). The p53 target is more complex than RAS for both biological and structural reasons. First, the objective is to restore function to an inactive tumor suppressor protein rather than inhibit the activity of an oncoprotein. Second, missense mutations occur throughout the central DNA-binding domain of TP53 (that is required for its specific function as a transcription factor), so while the Y220C mutant of p53 is quite common, it still represents only 1.5% of the mutations found in the protein in The Cancer Genome Atlas database. The first set of compounds, which used the published structure of the Y220C protein complexed to PhiKan083 (6) as a guide, was based on a carbazole backbone and used an acrylamide or chloroacetamide as the electrophile that would react covalently with the C220 sulfur. To identify compounds that would react directly with Y220C, a cysteine light variant (Y220C-CL) of the DNA binding domain was designed and produced in which the surface cysteines present on p53 were replaced by serine. The new compounds, KG1, KG3, KG4, and KG5, covalently labeled the Y220C-CL protein but not the control p53 wild-type cysteine light (WT-CL). The reactivity of the compounds with Y220C p53 in the cellular environment was demonstrated using a click chemistry approach in which the TAMRA-azide modification of the alkyne group of KG5 complexed to p53 was clearly seen by western blotting analysis of the cell extract. The core DNA-binding domain of p53 is very unstable, melting and forming aggregates at 44°C, whereas this happens at only 37°C for the Y220C mutant. Disappointingly, this first set of compounds only marginally stabilized the mutant protein. Further rounds of chemical iteration, using high-resolution crystallography to guide new substitutions, eventually led to a breakthrough with the best compound KG13, which was able to stabilize the Y220C mutant to wild-type levels with a shift in temperature of aggregation to 43°C. Valuable clues for further development of new compounds emerged during the development of KG13, in particular a stabilizing charge interaction with an aspartic acid at position 228 (Fig. 1); this was confirmed when it was shown that D228A mutants interacted more weakly with KG13. In addition, the binding of the methacrylamide indole KG6 stabilized a new closed conformation of the core domain associated with the L6 region of p53, presenting potential novel binding surfaces.

The KG13 compound was shown in biochemical assays to be specific for the Y220C mutant, and no modifications of any cysteine in wild-type p53 were detected. As with the studies of the mechanism of action of the G12C modifying compounds, the KG13 molecule is already providing fascinating and unexpected insights into the properties of its target. The Y220C protein has been reported previously (8) to represent an unusual form of mutant p53 because it is found in the cytoplasm of H1229 cells at 37°C rather than the nucleus of the cell. It shares this property with only a few other p53 mutants, whereas most mutant p53s are located in the nucleus. Because the nuclear localization sequences defined in p53 are all present in the Y220C molecule, this may represent either a masking of those import sequences or perhaps more likely an enhanced nuclear export rate (9). The temperature-sensitive phenotype is very striking, and as confirmed here, the protein is produced at high levels at the permissive temperature of 28°C. At this temperature, antibody analysis shows the protein to be predominantly in the native confirmation, as shown by immunoprecipitation with the PAb1620 antibody, and the protein is active at inducing p53 downstream target gene transcription and binding to p53 target site on chromatin, as shown by a chromatin immunoprecipitation assay. A time course shows strong induction of the p21 and MDM2 downstream genes within 12 hours of shifting from 37°C to 28°C. This strongly suggests that like the V135A protein, the Y220C protein is nuclear at the permissive temperature. The expectation from the biochemical data gathered around KG13 is that it binds to the wild-type–like conformation of Y220C and stabilizes that conformation at elevated temperatures. The authors, therefore, examined the ability of KG13 to activate the functions of Y220C in temperature shift experiments in which Y220C-expressing cells were grown at 28°C and exposed to KG13 and then shifted to 37°C looking for a longer persistence of p53 solubility and activity at the nonpermissive temperature. A cellular thermal shift assay (CETSA) showed that KG13 inhibited aggregation on shift to 37°C, yielding a Tagg similar to the wild-type protein. To measure the biological consequence of this retained activity, a time point at the nonpermissive temperature was chosen in which the nonliganded Y220C induced expression of MDM2 and CDKNIA and BBC3 (Puma) gene expression had fallen to low levels. A clear though modest effect of the compound was apparent in both p53-null and p53 wild-type cells that had been stably transformed with the Y220C mutant. The effects were specific because they did not affect the activity of a control mutant p53 protein (R273C). A much more convincing response was seen in cell lines expressing endogenous levels of Y220C in which the persistence of p53 signaling up to 8 hours after temperature shift was readily detected by western blots. The wild-type p53 protein is strongly negatively regulated by the MDM2 E3 ligase pathway, and this has led to the idea that inhibitors of MDM2 binding to p53, such as Nutlin, might enhance the activity of p53 rescue compounds like KG13. The situation in the Y220C mutant may differ; however, as at 28°C when levels of MDM2 are high, the mutant protein levels are also very high, perhaps implying that MDM2 cannot target the protein for degradation. If that is the case, then no enhancement by Nutlin inhibitors could be expected.

A critical question for future development will, of course, be whether enough Y220C protein is present in the native conformation in tumors growing at human body temperature to be modified by KG13 and mimics. Noncovalent binders such as PhiKan and those from PMV give much hope, as they show cellular activity at 37°C. There is still skepticism that covalent modifiers will have off-target toxicity from potential reactivity with other proteins, but the clinical approval of the KRas molecule is cause for cautious optimism (10).