The p53 protein is an important tumour suppressor that is inactivated in many human cancers. Understanding how p53 is regulated and the downstream consequences of p53 function is helping us to devise novel therapies based on the reactivation of p53. Such approaches may be useful in the treatment of cancer, but a growing understanding of a role for p53 in other conditions suggests that modulation of p53 may have broader applications.
- murine double minute 2 (MDM2)
- tumour suppression
Although cancer occurs fairly frequently within the population, when considering the number of cells and cell divisions that occur during a human lifespan, the development of a malignant cell is a rather rare event. There are many mechanisms in place to prevent tumour development. One of these is the link between abnormal proliferation and the activation of cell death or a permanent cell-cycle arrest . This failsafe response ensures that cells acquiring changes that lead to deregulated proliferation, such as oncogene activation, are eliminated before they can form fully fledged cancers. One of the most important mediators of this and other tumour-suppressive responses is the p53 protein . p53 plays a critical role in protection against malignant development, and understanding how p53 functions will be key to the development of therapies based on activating or mimicking p53 functions.
Tumour-suppressor functions of p53
A number of lines of evidence show the importance of p53 in the prevention of tumour development. Mice lacking p53 can be born normally, showing that p53 function is not absolutely required for normal growth and development [3–5]. However, most of these mice develop cancer, most frequently lymphoma, within the first 6 months of life. In many experimental models, loss of p53 has been shown to co-operate with a wide variety of other oncogenic events to promote and accelerate cancer development. In humans, mutations within the p53 gene (TP53) are found in about half of all cancers, including the most common epithelial malignancies such as lung, colon and breast . There are rare exceptions to this, which have proved to be very informative. Cancers of the uterine cervix, for example, very rarely carry p53 mutations. The reason for this became apparent with the realization that this type of cancer results following infection with certain types of HPV (human papillomavirus). The principal oncoproteins encoded by these HPVs are E6 and E7, both of which function to inactivate host-cell-encoded tumour-suppressor proteins . More specifically, the HPV E6 protein was shown to bind to, and target the degradation of, p53 . So cells infected with these HPV types are depleted of p53, and the inactivation of p53 by the HPV E6 protein eliminates the need for mutational inactivation within the gene during the genesis of these cancers . Indeed, even in other cancers where there is no viral aetiology, the tumours that retain wild-type p53 often show alterations in other parts of the p53 pathway leading to a loss of the normal ability to activate p53 in response to oncogenic signals. So it seems likely that the p53 response is disabled in most human cancers, either by direct mutational inactivation of p53 or via an indirect mechanism . The mouse studies suggest that maintenance of p53 function is not essential for normal development, and rare families that inherit a mutant p53 allele have been described. Germline p53 mutations are associated with Li–Fraumeni syndrome, where patients develop a broad range of tumours very early in life . The families provide even more evidence to support the importance of p53 as a tumour suppressor.
The p53 pathway
In addition to oncogene activation, many other signals can induce a p53 response. These include stress signals to which nascent tumour cells might be exposed, including DNA damage, hypoxia, nutrient deprivation and loss of stromal support . There are numerous consequences of the induction of p53 in response to these stress signals, which can result in the activation of apoptotic cell death or the permanent arrest of cell-cycle progression through senescence [2,13]. However, p53 can also participate in the recovery from some forms of damage, for example by participating in DNA repair . Cells in which p53 is mutated, or where p53 is not properly activated in response to stress signals, survive and continue to divide, and show significant acquisition of further DNA damage and genomic instability. Each of these defects can contribute to the development of cancer, either by allowing proliferation or by contributing to further oncogenic alterations within the cell. The ability of p53 to drive cell death or senescence raises the hope that reactivation of p53 in established tumours would have a therapeutic effect, and recent studies in mice have supported such an approach [15–17]. The reactivation of p53 in a number of mouse tumour models led to the rapid regression of the tumours, through the activation of either senescence or apoptosis, depending on the type of tumour. Interestingly, although senescence might have been predicted to be only cytostatic, the regression of tumours following p53 induced senescence was associated with an effective immune response .
p53 is a transcription factor
One of the key mechanisms by which p53 functions to induce responses like apoptosis and senescence is through its ability to regulate gene expression. p53 functions as a sequence-specific DNA-binding protein, driving transcriptional activation of genes that contain p53-responsive elements in their regulatory regions . A large number of p53-reponsive genes have been described, and it seems likely that the ultimate response to p53 activation represents the co-ordinated action of a specific selection of genes. Many studies have shown that the pattern of gene expression following activation of p53 can differ depending on factors such as tissue type, the activating signal, the presence of other cofactors and modifications or levels of p53 itself . Which genes are activated in response to p53 is subject to complex control that is not fully understood, but is likely to underpin the observed differences in response to p53. As mentioned above, activation of p53 in tumours can result in apoptosis or senescence depending on tissue type, and, among normal tissues, there are clear differences between those that die following p53 activation and those that are not sensitive to p53-induced death.
Despite the large numbers of genes induced by p53, it has still been possible to identify some key players in the mediation of the p53 response. A number of cell-cycle regulators are activated by p53, but among them, the cyclin-dependent kinase inhibitor p21 is particularly critical for the activation of G1 cell-cycle arrest . In many cell systems, p21 has also been shown to be required for the senescence response [21,22]. However, p21 is generally not found to be necessary for apoptosis, and indeed the induction of p21 may have some effect in protecting cells from the p53-driven apoptotic response . Many apoptotic genes are also targets for transcriptional activation by p53, including a number of so-called BH3 (Bcl-2 homology 3 domain)-only proteins, which play a role in regulating mitochondrial outer membrane permeability. Expression of the BH3-only proteins can lead to a release of apoptogenic factors such as cytochrome c from the mitochondria, and the subsequent induction of apoptosis though caspase activation . Of these BH3-only proteins, PUMA (p53 up-regulated modulator of apoptosis) has been shown to be critical for the induction of p53-dependent cell death in many tissue types, and seems to be a pivotal p53-inducible apoptotic target . Other p53-target genes play important roles in mediating other p53 responses, including the regulation of DNA repair and a recently described ability of p53 to control metabolism. There is also a group of p53-responsive genes that do not appear to mediate a p53 response, but rather contribute to the regulation of p53 itself. The best understood of these is MDM2 (murine double minute 2), a protein that plays a critical role in controlling p53 function, and which will be discussed in more detail below.
In addition to functioning as a transcriptional activator, p53 can repress the transcription of certain genes, and also carry out functions completely independent of the regulation of gene expression. There are some well-described activities of p53 in the cytoplasm, including an intriguing ability of p53 to function like a BH3-only protein in driving mitochondrial outer membrane permeability and apoptosis . Interestingly, this activity of p53 is promoted by PUMA expression, which releases cytoplasmic p53 from an inhibitory complex . Since PUMA expression is dependent on p53's transcriptional activity, both nuclear and cytoplasmic p53 may be necessary for a complete apoptotic response to oncogenic stress. Other cytoplasmic activities of p53 that have been recently described include an ability to inhibit autophagy , a function that might counteract the ability of p53 to promote autophagy by driving the expression of pro-autophagic proteins such as DRAM (damage-regulated autophagy modulator) .
Regulation of p53 stability by MDM2
As described above, p53 activation is generally incompatible with cell proliferation, so it is vitally important that p53 function is adequately restrained under conditions of normal growth and development. Indeed, p53 levels are extremely low in normal, unstressed, cells. Equally important is the ability to activate p53 rapidly in response to stress. As might be expected, there are a number of mechanisms by which p53 function is regulated within the cell. But key to the control of p53 function is the control of p53 stability, and the low levels of p53 maintained in normal cells are a reflection of the rapid turnover of the p53 protein. In response to the stress signals that activate a p53 response, the degradation of p53 is inhibited and the levels of p53 accumulate rapidly in the affected cell . Although various other post-translational modifications, such as acetylation , contribute to the regulation of the function of p53, the accumulation of p53 is an important and almost universal feature of p53 activation. The increased levels of p53 drive activation of p53-responsive genes, such as p21 and PUMA, and thereby promote tumour suppression.
The stability of p53 depends on its rate of degradation through the proteasome following ubiquitination, and a number of ubiquitin ligases that can target p53 for ubiquitination have now been described . Of these, the most important seems to be MDM2, a p53-binding protein that is transcriptionally regulated by p53 [33,34]. The importance of MDM2 in regulating p53 has been shown very clearly in animal models. Constitutive loss of MDM2 is incompatible with development, and MDM2-null embryos die very early owing to uncontrolled p53-induced apoptosis [35,36]. However, this lethality can be entirely rescued by simultaneous loss of p53, and MDM2/p53-null animals develop to become cancer-prone adults very similar in phenotype to p53-deficient animals. These results suggest very strongly that the main function of MDM2 is to keep p53 in check. MDM2 is also essential in the adult mouse, as shown by mice expressing a switchable version of p53 . After being bred to adulthood in the absence of MDM2 or p53, p53 can be turned on in selected tissues, leading to an extremely dramatic activation of a p53 response and death of the animal. Clearly, this observation needs to be kept in mind when considering the use of drugs that inactivate MDM2 as cancer therapies.
So if MDM2 is responsible for targeting the degradation of p53 in normal cells, what happens in response to stress? In most cases it would seem that stress signals to inactivate MDM2, and so allow for the stabilization of p53. In the case of DNA damage, this has been shown to result from both the enhanced degradation of MDM2  and phosphorylation of both p53 and MDM2 that inhibit the interaction between the two proteins . But other stress signals have different effects on MDM2. Oncogenes such as Myc or the deregulation of E2F, events that are associated with the deregulated proliferation exhibited by most cancers, activate p53 through the action of ARF (alternative reading frame), a small protein that binds to and inactivates MDM2 . Similarly, perturbations in ribosome biogenesis and nucleolar stress lead to the inactivation of MDM2 following interaction with ribosomal proteins such as L5 and L11 [41,42]. The mechanism through which proteins like ARF or L11 can inhibit MDM2 is still not clear, although it does not seem to involve inhibition of the interaction between MDM2 and p53.
Targeting MDM2 for therapy
Our understanding of how MDM2 is inactivated in response to stress has shed interesting light on the ways in which p53 function can be impeded in cancer cells that retain wild-type p53. It would appear that many of these tumours harbour defects in the pathways that allow for the inactivation of MDM2 in response to oncogenic stress, so, although p53 remains capable of tumour suppression, it is not properly activated. Examples include the mutation of stress-induced kinases such as ATM (ataxia telangiectasia mutated) that lead to the phosphorylation of p53 and MDM2, and loss of ARF. These defects in the pathways to activate p53 explain why cancers arise without the requirement for mutation of p53 itself, and also suggest that, if the wild-type p53 in these tumours could be reactivated, a therapeutically desirable p53 response would ensue. There is now significant effort to try to identify small-molecule inhibitors of MDM2, which might have use in the treatment of wild-type p53-expressing tumours . So far the most successful, at least in principle, has been the identification of Nutlin-3, a compound that binds within the N-terminal groove in the MDM2 protein that would normally form a binding pocket for p53 . Nutlin-3 prevents the interaction of p53 with MDM2, so very efficiently stabilizing p53, leading to the activation of a p53 response. However, preventing the p53–MDM2 interaction may not be the only mechanism of function for MDM2 inhibitors, as illustrated by the efficient function of ARF and the ribosomal proteins. Further analysis of the regions of MDM2 that are necessary for the ubiquitination of p53 has shown that the extreme C-terminal tail of MDM2 is important [45,46], and that this interacts with the RING (really interesting new gene) domain of the partner MDM2 (or MDMX) protein in a functional dimer . Deletion or mutation of the tail prevents the function of MDM2 as a ubiquitin ligase, raising the possibility that small molecules that somehow interfere with the RING–tail interaction may also inhibit MDM2 and stabilize p53. Other activities of MDM2, including the ability to deliver ubiquitinated p53 to the proteasome , may also be suitable targets for p53-activating drugs.
To date, the efficacy of any of these approaches in humans has not been established. The dramatic phenotype of p53 in the absence of MDM2 in the adult mouse leads to some concerns that systemic inactivation of MDM2 using these kinds of drug may be unacceptably toxic to normal tissue. But small-molecule inhibitors are unlikely to completely eliminate MDM2 function in the same way as deletion of the gene, and initial studies of Nutlin-3 in mice have not reported any significant toxicity . Furthermore, mice with reduced MDM2 levels are resistant to tumour formation without showing obvious adverse responses such as premature aging . It is therefore possible that reduction, rather than inactivation, of MDM2 will be the key to successful therapy.
Other activities of p53
In addition to the ability to inhibit proliferation, p53 has recently been shown to carry out a number of other functions. Some of these may have effects beyond tumour suppression, and loss of p53 may have profound effects on other aspects of health and disease .
Not surprisingly, stress-induced activation of p53 can play a role in other disease pathologies. Unlike tumour suppression, activation of p53 in response to stress can be severely detrimental; for example, radiation sickness and the debilitating side effects of chemotherapy can be due to p53-induced apoptosis in haemapoetic systems, hair follicles or the gut . p53 can also play a role in the damaging effects of ischaemia, or neuronal damage associated with Parkinson's disease or Alzheimer's disease [52,53]. Slightly elevated p53 activity has also been associated with premature aging , although aging is associated with a decline in p53 function . There is clearly much to be learnt about the role of stress-induced p53 in other diseases. However, there is also evidence that basal levels of p53 play an important role in maintaining normal health, and that loss of p53 affects more than just the stress response. Analysis of p53-null mice provides some of the evidence for this: although they clearly can develop normally, in many cases they do not. Developmental abnormalities such as neural tube defects are strain- and sex-dependent, but can occur in a substantial proportion of the animals . Closer analyses of these mice have also revealed abnormalities in metabolism, fecundity and longevity, each of which appears to reflect a constitutive activity of p53 that is not dependent on activation by acute stress .
Regulation of metabolism by p53
Control of metabolism is key to normal cell growth and division, and so it is not surprising that p53 can play an important role in regulating various aspects of different metabolic pathways. p53 is induced in response to starvation , which represents another type of stress, so preventing proliferation of cells under conditions of nutrient deprivation. The activation of p53 can also drive the activation of AMPK (AMP-activated protein kinase), and so the promotion of a pathway to inhibit mTOR (mammalian target of rapamycin) signalling and reduce cell growth, thereby helping to balance growth and proliferation . This activity of p53 can also promote autophagy, as can other p53-target genes such as DRAM , a mechanism to allow survival during times of bioenergetic stress by enabling the cell to start consuming its own organelles . However, recent studies have also shown that cytoplasmic p53 can play a direct role in the inhibition of autophagy , although how these apparently opposing functions of p53 are balanced is not yet understood.
A number of studies have now also shown that p53-target genes play a role in controlling glycolysis and oxidative phosphorylation, by inhibiting the former and promoting the latter . These functions of p53 could contribute to tumour suppression, since enhanced glycolysis and a reduction in mitochondrial respiration play a role in cancer development. One p53 target that has been shown to participate in the regulation of glycolysis is TIGAR (TP53-induced regulator of apoptosis and glycolysis) . The TIGAR protein shows similarity to the phosphoglycerate mutase family of glycolytic enzymes. In particular, TIGAR shows structural and functional similarity to the bisphosphatase domain of 6-phosphofructo-2-kinase/fructose-2,6-bisphosphatase and has recently been shown to hydrolyse both fructose 2,6-bisphosphate and fructose 1,6-bisphosphate . Through these activities, TIGAR can dampen flux through the glycolytic pathway and promote use of the pentose phosphate pathway, an alternative route for the metabolism of glucose 6-phosphate. This may have several beneficial effects on the prevention of malignant progression. First, enhanced glycolysis and the promotion of this pathway through high levels of 6-phosphofructokinase activity (the opposite function to TIGAR) has been shown to contribute to transformation of primary cells by oncogenes such as activated Ras, and inhibition of this activity can suppress tumour growth . Secondly, one of the consequences of the pentose phosphate pathway is the generation of reduced glutathione for the removal of ROS (reactive oxygen species). ROS function as important regulators of many cellular processes, but at high levels can promote both genotoxic damage and cell death . The antioxidant activities of TIGAR may certainly help to limit cancer development, and would collaborate with other p53-inducible genes that have been shown to lower oxidative stress [65,66]. Interestingly, in addition to controlling cancer development, these antioxidative functions of p53 may also play a role in preventing aging and promoting longevity. It seems possible that the age-related decrease in p53 function may promote even further aging by removing some protection from elevated ROS. So p53-induced expression of TIGAR can prevent ROS-induced oncogenic damage, but induction of TIGAR also provides some survival function, through the inhibition of ROS . The apoptotic response to p53 is ROS-dependent, to some degree, and many of the apoptotic targets of p53 generate ROS [67–70], which contributes to the full and rapid apoptotic elimination of the affected cell. The ability of p53 to promote survival signalling is not limited to TIGAR, with several p53-target genes showing some anti-apoptotic function. These activities of p53 seem to reflect the complexity of balancing the survival and repair functions with the elimination of damaged cells. In some cell types, a switch in the p53 response from the induction of expression of survival genes to the activation of expression of apoptotic genes has been shown to correlate with the switch in response from cell-cycle arrest to cell death, and may reflect the severity of persistence of the stress signal.
Modulating the p53 response
Taken together, our growing understanding of the contribution of p53 to tumour suppression and other aspects of health and disease rather complicate the predictions on how best to modulate p53 for therapeutic gain. As mentioned above, there is much interest in finding mechanisms to reactive p53 in tumours, either by induction of endogenous wild-type p53, or through the use of drugs that might re-establish some wild-type function in mutant p53. Animal studies suggest that the reactivation of p53 will be helpful in tumour therapy, but there is also a possible role for drugs that inhibit p53. There is evidence from both animal studies and humans, showing that the presence of wild-type p53 may be detrimental to the efficacy of some types of therapy , possibly due to the activation of survival functions of p53 that protect tumour cells from cytotoxic drugs. Elegant mouse studies looking at the effects of radiation treatment showed that the absence of p53 substantially protected the animals from the initial toxicity without loss of tumour suppression, provided that p53 was turned on once the immediate response to treatment was resolved . This suggests that short-term systemic treatment with p53 inhibitors might allow patients to with-stand much higher doses of irradiation or chemotherapy, without the danger of triggering further malignancies in normal tissue . A similar protection from toxicity induced by genotoxic drugs was also shown using a small-molecule p53 inhibitor in mice .
Moving away from cancer, the advantages of inhibition of the p53 response to ischaemia or neuronal diseases are evident. Understanding when we do and when we do not need p53 is a high priority if we are to most efficiently harness the modulation of p53 function for therapy.
Functions of mutant p53
There is much compelling evidence that loss of wild-type p53 function contributes to cancer development. However, the mechanism by which p53 activity is lost through mutation in most human cancers is unusual, compared with other tumour-suppressor genes. In the case of p53, the most common type of inactivating mutation is a point mutation, leading to the expression of a mutant protein . The location of the resulting amino acid substitution is usually within the central DNA-binding domain of the p53, resulting in a loss of DNA-binding activity and so a failure to activate gene expression. This leads to the loss of p53's normal functions, including the ability to inhibit proliferation or induce cell death. However, it has become very clear that the expression of these forms of mutant p53 is not simply the equivalent of loss of wild-type p53, such as would be seen following deletion or loss of expression of TP53 .
Studies in mice show the impact that deletion of p53 has on increasing tumour susceptibility, and loss of wild-type function is an important consequence of mutations in p53 seen in human cancers. However, the mutant p53 expressed in cancers appears to have additional functions that can contribute to promoting the malignant phenotype [76,77]. Since p53 functions as a tetramer , one possible function of mutant p53 is to inactivate co-expressed wild-type p53 through a dominant-negative activity. Certainly in experimental overexpression systems, mutant forms of p53 have been shown to inhibit wild-type p53. However, in more physiologically normal circumstances, where cells express p53 from one mutant and one wild-type allele (as seen in Li–Fraumeni disease patients), it is less clear that the mutant p53 can result in a profound impairment of wild-type p53 function under normal growth conditions. Under such circumstances, both the wild-type and mutant p53 are expressed at very low levels, and even in animals expressing only mutant p53, the protein remains unstable in normal tissue . This instability of mutant p53, like wild-type p53, depends on the function of MDM2, since deletion of MDM2 results in a stabilization of the mutant p53 and enhanced tumour development . The increase in level of mutant p53 seems to allow the manifestation of an activity that contributes to the development of cancers, and to the ability of these cancers to invade and metastasize. But what is mutant p53 doing, if not simply inactivating wild-type p53 function?
The answer to this question is far from clear yet, but there are two activities shown by mutant p53 that appear to contribute to the transforming function. Many studies have shown that, although mutant p53 loses the ability to activate the expression of genes that are responsive to wild-type p53, it is not inert when it comes to the regulation of transcription. Indeed, mutant p53 can acquire the ability to both positively and negatively regulate gene expression, and it is clear that this contributes to some of the pro-tumorigenic functions of mutant p53, such as enhanced survival and resistance to therapy . A second activity shown by mutant p53 is the ability to form an interaction with the p53 family members p63 and p73 . This is an interesting activity, since it is not exhibited by wild-type p53, and may be related to a change in the conformation of some mutant p53 proteins. The effect of this interaction has been shown to be the inactivation of p63 and p73 function, so a dominant-negative function of p53 against family members, rather than p53 itself. The interaction between mutant p53 and p63/p73 depends on the DNA-binding domain of the proteins, rather than the C-terminal oligomerization domain that controls the interaction of p53 with itself , and so the consequences of the binding might be quite different. Importantly, recent data suggests that p73 can also function as a tumour suppressor, suggesting that the ability of mutant p53 to inactivate its cousins may be important in the generation of a more aggressive and invasive tumour . Since around half of all human cancers express forms of mutant p53 that could show this activity, there is a great deal of interest in trying to understand in more detail how mutant p53 functions. These observations also introduce a note of caution when thinking about using drugs to systemically inhibit MDM2 . The drawback that they may lead to the activation of wild-type p53 in normal tissue is discussed above, but now we have to consider another possible problem. Each individual is likely to harbour a number of pre-malignant lesions that contain mutant p53 expressed at low levels. The in vivo data available show that MDM2 inhibition would result in the stabilization of mutant p53, and this is likely to enhance the manifestation of the oncogenic activities of these proteins. There is still much work to be done, but all of these concerns will need to be fully addressed before these approaches can be used as routine cancer therapy.
This work was supported by Cancer Research UK.
I thank Eric Cheung for helpful comments.
Sir Frederick Gowland Hopkins Memorial Lecture:
Abbreviations: ARF, alternative reading frame; BH3, Bcl-2 homology 3 domain; DRAM, damage-regulated autophagy modulator; HPV, human papillomavirus; MDM2, murine double minute 2; PUMA, p53 up-regulated modulator of apoptosis; RING, really interesting new gene; ROS, reactive oxygen species; TIGAR, TP53-induced regulator of apoptosis and glycolysis
- © The Authors Journal compilation © 2009 Biochemical Society