p53 defies convention again: a p53 mutant that has lost tumor suppression but still can kill

Abstract

Loss of tumor suppression by the p53 protein involves altered or abrogated transcriptional activity resulting in a failure to mediate wild‐type cellular responses including cell cycle arrest, senescence, and apoptosis. Timofeev et al (2019) make the fascinating finding that a novel p53 cooperativity mutation devoid of DNA binding results in no tumor suppression but surprising retention of an apoptotic response to chemotherapy and other treatments. This shows a need for rethinking how mutant p53‐driven tumors are treated in the clinic.

The p53 protein has been well characterized for its tumor‐suppressing role in human cancer. p53 acts primarily as a transcriptional factor to regulate a wide variety of cellular responses including cell cycle arrest, senescence, and apoptosis (Fig 1A). Accordingly, an extensive transcriptome has been associated with p53 activation. Notably, this includes the CDKN1A gene encoding the cyclin‐dependent kinase inhibitor p21, for cell cycle effects, and the BBC3 gene that expresses the BH3‐only Puma protein, for apoptosis (Kastenhuber & Lowe, 2017). Several compelling studies have also implicated a non‐transcriptional role for p53 in cell death responses involving cytoplasmic or mitochondrial localization and protein–protein interactions with a subset of Bcl‐2 family members. The contribution of this non‐nuclear function of p53 to tumor suppression has been unclear as tumor‐derived mutants, which lose transcriptional activity, also lack this cytosolic activity that triggers apoptosis (Vaseva & Moll, 2009).

Figure 1. Missense mutations within the DNA binding domain of p53 result in loss of tumor suppression.

Although wild‐type p53 (A) is transcriptionally active and can mediate diverse responses including cell cycle arrest and apoptosis, tumor‐derived missense mutations have typically lost the ability to bind directly to DNA in a sequence‐specific manner and a failure to mediate those cellular outcomes (B‐C). These mutants either merely lose tumor suppression (B) or in addition gain other oncogenic activities (C). Occasional mutants retain some selective function on a subset of targets (D), but nevertheless promote tumorigenesis in mouse models. Timofeev et al (2019) expand this paradigm by identifying a novel engineered p53 mutation (E) that is transcriptionally dead, loses tumor suppression, but remarkably can mediate an apoptotic response that is not dependent on effects on gene expression.

Previous studies by the Stiewe laboratory have shown that the transcriptional activity of p53 not only relies on the ability of tetrameric p53 to bind to its response elements in a sequence‐specific manner, but those oligomers need to engage in cooperative interactions, which are needed for full transcriptional potency (Schlereth et al, 2010). An engineered mutation in human p53, R181E, was shown to lose cooperativity. Timofeev et al extend and validate those findings by generating an in vivo mouse model with the equivalent murine mutation, R178E (Timofeev et al, 2019). In so doing, they confirm that the 178E mutation is transcriptionally inactive, and, as expected, has lost tumor‐suppressing activity either for spontaneous tumorigenesis or in the context of two oncogenic activating mutations, Eμ‐myc for B‐cell lymphoma, or AML‐ETO9a and oncogenic N‐Ras for acute myeloid leukemia. In this regard, the 178E mice are, as expected, indistinguishable from p53 null animals.

The surprise came when the 178E mice were crossed to Mdm2‐null mice. Mdm2 is a key negative regulator of p53. Elegant studies from the Lozano and Bradley laboratories showed that loss of Mdm2 is embryonic lethal, but this phenotype could be rescued by simultaneous p53 knockout (Jones et al, 1995; Montes de Oca Luna et al, 1995). Since then, many laboratories had used the ability to complement Mdm2 loss in mouse development as a means to probe various functions of p53. In contrast to the classic experiments with p53 nullizygosity, the 178E allele failed to rescue Mdm2 loss, but rather showed embryonic lethality associated with apoptosis (Timofeev et al, 2019). As this allele lacks transcriptional function, this hinted that the non‐nuclear activity of p53 may be retained in this mutant. Indeed, Timofeev et al (2019) demonstrate that 178E p53 can localize to mitochondria and interact with Bcl‐2 family members. Further validation is needed, but it is likely that Timofeev et al (2019) may have finally identified the elusive mutation in p53 that distinguishes its cytosolic from its nuclear activities. If this indeed proves to be the case, the observation that 178E is no longer tumor‐suppressing would support the idea that the transcriptional function of p53 is essential for its ability to fully block oncogenesis.

Multiple prior studies have demonstrated that not all tumor‐derived mutant p53 proteins are equivalent. The majority of such genetic alterations block the sequence‐specific binding of p53, rendering the protein transcriptionally dead. Mouse modeling of such mutants places them in two classes based upon the tumorigenesis phenotype (Kim & Lozano, 2018; Sabapathy & Lane, 2018). The first are those that are indistinguishable from p53 null animals, in that they display primary lymphomas and sarcomas with minimal metastatic spread (Fig 1B). The second have gained additional functions, including shortened tumor onset and aggressiveness, as well as increased metastasis (Fig 1C). Extensive biochemical studies have suggested multiple underlying mechanisms for these latter so‐called “gain‐of‐function” mutants (Kim & Lozano, 2018; Sabapathy & Lane, 2018). The basis at the molecular level for these distinctions remains elusive. The sequence‐specific DNA binding domain of p53 is largely a barrel of beta sheets with a small number of specific residues involved in direct DNA interactions. Mutations in the latter amino acids have been called “contact”, whereas those that disrupt the barrel are broadly categorized as “structural” (Bullock & Fersht, 2001). Less common are tumor‐derived mutations that result in a selective loss of transcriptional activity on some targets but not others. The best characterized of these is found at the 175 residue hot spot. While mutation to arginine (175R) renders p53 transcriptionally inactive (Fig 1C), the presence of a proline (175P) results in selective loss of function on apoptotic targets. In an engineered mouse model, 175P can suppress early‐onset tumors, but eventually these mice succumb to a much higher rate of spontaneous tumorigenesis than the wild type (Liu et al, 2004) (Fig 1D).

The studies from the Stiewe laboratory have challenged and expanded these notions. The 178E mutation appears to affect higher order cooperative interactions rather than directly influencing the structure of the DNA binding domain and its contact points with response elements (Schlereth et al, 2010). It thus is neither “contact” nor “structural”. There is no evidence for any gain‐of‐function activity for 178E as the tumor biology that is observed is indistinguishable from p53 null mice. Further, in contrast to these transcriptionally dead mutants, 178E appears to retain the ability to induce apoptosis, albeit under conditions in which Mdm2 expression is ablated or p53 expression is induced by ectopic chemotherapy (Fig 1E) (Timofeev et al, 2019).

The key caveat to these findings is that they rely on the use of a mutation that has been engineered in the laboratory. There is little evidence, as yet, that such mutants are found in actual human tumors. The authors show data with the tumor‐derived human R181L that support the notion that a mutant associated with human cancer retains apoptotic activity in response to Mdm2 inhibitors or chemotherapy. However, this mutant retains some transcriptional activity and thus does not fully mimic the effects seen with the engineered 178E allele, which appears dead in gene regulation.

Wild‐type p53 has been well characterized in the laboratory as being necessary for cellular responses that are triggered by a variety of cancer therapies. However, the use of p53 status as a determinant for clinical responsiveness has not proven to have the utility that would be expected (Sabapathy & Lane, 2018). Multiple laboratories have been exploring the possibility of targeting mutant p53, mainly by its destabilization, in combination with chemotherapy (Sabapathy & Lane, 2018). The studies of Timofeev and colleagues would argue that such an approach may be counterproductive, since human tumors harboring a mutant such as 181E may, in fact, be resistant to chemotherapy if the mutant is targeted. Given the diversity of ways in which loss of tumor suppression can be achieved (Fig 1), understanding the molecular mechanisms underlying specific mutants clearly is of paramount importance. The increased complexities revealed by Timofeev et al (2019) demonstrate the daunting need for further studies into the notion that individual p53 mutations may have their own peculiar properties. Nevertheless, their studies also provide the exciting prospect of effective treatments for, at least, a subset of mutant p53‐driven human tumors.

The EMBO Journal (2019) 38: e103322