Mutant p53 Gain-of-Function in the Spotlight: Are We Suffering a GOF Delusion?

Mutations in the TP53 gene are a common unifying feature of human cancers occurring in more than 50% of all cancers. In some tumor types, such as ovarian and uterine cancers, TP53 mutation is present in more than 90% of cases whereas in others the rate is much lower. Analysis of the Cancer Genome Atlas (TCGA) database finds that mutation of TP53 is associated with a worse prognosis and shorter survival time (1). The causative role of TP53 mutation in human cancer is established by the extraordinary lifetime risk associated with germline mutations in the gene, which cause the Li-Fraumeni cancer family syndrome (2). Loss of TP53 or TP53-mutant expression are in widespread use in many genetically engineered mouse models of human cancer in which deletion of TP53 or expression of a point mutant p53 protein lead to increased cancer rates. The massive increase in the number of cancer genomes sequenced has led to a comprehensive database of TP53 mutation types. In the TCGA, about 20% of TP53-mutant cases show complete loss of p53 expression whereas 80% show expression of a point mutant protein (1). The mutations are unusual in that they are found in a large number of different locations predominantly in the DNA binding domain of the protein. Certain “hot spot” mutations are more common than others, and indeed, the R175H mutation is the fourth most common specific mutation in human cancer (3). Three properties have been assigned to these mutant proteins. The first and most strongly supported is a loss of function. The mutant proteins lack the specific DNA-binding function and transcription activating functions of the wild-type protein (4). Although this is most simply interpreted as a direct effect on the DNA-binding domains’ ability to bind to specific p53-binding elements in the DNA, it is important to remember that in early studies the mutations were shown to inhibit the N terminal function of p53 as a transcription activating domain. The data obtained using p53 Gal4 fusion proteins implied that the mutations have allosteric effects (5). Not all mutations appear equal, a number of the rarer mutations retain residual wild-type activity, and some are temperature sensitive showing wild-type activity at lower temperatures (3). The second property of the mutant proteins is their ability to act as dominant negative inhibitors of wild-type p53. This is readily understood as p53 is a dimer of dimers. The dimers form on the polysome, so in a cell expressing both mutant and wild-type p53, the inhibited form is a 2M/2WT tetramer. Genetic studies establish that the dominant-negative function is of limited penetrance and depends on the level of expression of the mutant proteins. This helps to explain why loss of heterozygosity at the TP53 locus is such a frequent event in human cancers (1). The presence of an essential gene (6) close to TP53 on chromosome 17p is also a major reason why loss of heterozygosity in the presence of a point mutant allele is more common than homozygous deletion of both alleles of the TP53 gene. The final property ascribed to mutant p53 is a gain of function. In this model, the mutant proteins exert novel functions that drive tumorigenesis and mutant p53 acts as a driver oncogene. If correct, this suggests that mutant p53 is the most frequent driver mutation found in human cancer and a potential target for therapy (3). The high levels of expression of mutant p53 proteins readily detected by IHC in human cancers support such an idea but do not establish causality. Many reports have established novel functions for mutant p53 ranging from forming complexes with p53 family members p73 and p63 to acting directly to alter metabolism by direct binding to the AMPK enzyme (3). Common findings are that mutant p53 promotes cell migration, invasion, and metastasis and makes cells resistant to chemotherapy. Another finding has been that mutant p53 affects metabolic regulation and drives the glycolytic pathway. However, the wide variety of properties ascribed to the mutant proteins and the fact that many experiments involve the use of high-level expression constructs has created some questions. Indeed, one study suggests that all of the gain-of-function properties can be explained by the high level of aneuploidy in the p53-mutant cells tested (7). The most compelling support for the gain-of-function hypothesis comes from mouse genetic studies. Genetically modified mice are bred such that they express either no p53 or just mutant p53. This approach avoids complications that might derive from dominant-negative effects and allows the effect of the mutant protein alone to be examined. In two seminal studies, both groups found that although the expression of the mutant protein was at very low levels in normal tissues, it accumulated to high levels in the tumors that arose. Most significantly, although no change in overall survival was seen, the mutant expressing mice developed a much wider range of tumors including carcinomas compared with the TP53-null mice. Somehow the mutant protein was changing the type of cell that would act as a precursor for cancer development (3).

In the initial studies, two common hotspot mutations R175H (mouse R172H) and R273H (mouse R270H) were used. However, this property of mutant p53 is not universal as when the same study was repeated using the R249S (mouse R246S) mutation, which is induced by aflatoxin and is found frequently in liver cancers that develop in toxin-exposed populations, no gain of function could be detected and mice expressing the R246S protein had the same phenotype and tumor spectrum as the null mice. However, mutant p53 accumulation was seen specifically in the tumor tissue as in the earlier examples (8). Another mutation tested in the same system gave an even more provocative result. Here the mutant expressing mice developed tumors much later than the null mice, and the tumors were quite exceptional in being always diploid (9). One further powerful genetic experiment showed that expression levels of the mutant protein might be important for its gain of function. MDM2 is the major E3 ligase that regulates p53 level and activity. Loss of Mdm2 is embryonic lethal but is rescued by loss of p53. In mice that express only mutant p53, loss of Mdm2 is also tolerated, and here the levels of mutant p53 found in normal tissues are much higher. These mice develop cancer earlier than p53-null mice supporting the concept that mutant p53 protein levels drive its gain-of-function activity. The consensus emerged then that mutant p53 expression was driving cancer cell behavior and that the removal of the mutant protein might then inhibit cancer cell growth. Indeed, siRNA experiments and in vivo ablation studies supported this idea (3)

The experiments conducted by Wang and colleagues in this issue of Cancer Discovery (10) that rely on deletion of mutant TP53 using an inducible Cas9 system are then very challenging for this consensus. They unequivocally establish that removal of the mutant p53 has no effect on the growth of a large selection of mouse and human cancer cell lines that normally express the protein. Fittingly, as a set of studies designed to overturn an established model, they have been performed with admirable rigor. Not only do the knockout cells grow as well as the parent cells but they also show no difference in metabolism, migration, ability to grow in low serum medium, or drug resistance. In summary, all these properties that have been associated with the gain of function are not altered by removal of mutant p53. In the most challenging study of all, removal of mutant p53 expression does not inhibit the growth of mouse tumor cell lines in immunologically normal mice and nor did it effect metastasis. These are areas where gain-of-function experiments had looked very clear cut (3). In seeking confirmation of their results, the authors examined the DepMap database. Here again CRISPR knock out of TP53 in 391 human cancer cell lines carrying 158 different mutant p53s was seen to have no effect on growth of the cells. How can such conflicting data be reconciled? The simple idea that something about the CRISPR technology is the problem is easily dismissed as the growth-inhibitory effect of other bona fide oncogenic drivers such as RAS and BRAF are readily seen in the DepMap data base and have been the subject of detailed studies. Pertinently, knock out of wild-type TP53 gives a clear growth advantage to tumor cells expressing wild-type p53 in the same data base, and this work confirms that the tumor cells remain sensitive to growth inhibition if wild-type p53 function is restored to them (10). In the early considerations of oncoprotein function, two models were proposed, one the so called hit and run model in which the oncoprotein is needed to create a risk for other alterations (for example), it could be mutagenic or effect the accuracy of mitosis. The second model imagines the oncoprotein as a continuous driver required to maintain the transformed state. The use of temperature sensitive mutants of SV40 T antigen to transform cells established that T antigen was continuously required to maintain the transformed state. Using new switchable systems, the need for other oncoproteins to be continuously expressed has been established and drastically reinforced by the potent antitumor activity of drugs targeting oncogene function. It is this maintenance function of mutant p53 that fails the test in the work described here. Can one instead imagine a hit and run mechanism for mutant p53? A crucial step in the development of cancer is the loss of genetic stability, and a common route to this instability is whole genome doubling leading to aneuploidy on further rounds of division. For the tetraploid cells to divide, they must escape a p53 checkpoint. Point mutation to create a dominant negative allele is a single event that could promote, this hit, and if certain mutations were more potent in this checkpoint inhibition than others one can imagine that the mutant p53 proteins represent the legacy of such a rate-limiting event. Even this idea leaves a great deal of directly contradictory data in the field that can only be resolved by further analysis and discovery.

D.P. Lane reports grants from Karolinska Institute and personal fees from Chugai Pharmabody outside the submitted work. No other disclosures were reported.