Restoring p53-mediated apoptosis in cancer cells: New opportunities for cancer therapy

1. Introduction 

During the past decade, intensive research on the p53 pathway has revealed many novel mechanistic insights through which its tumor suppressor function is achieved. Despite tremendous progress, the underling mechanism for p53 to induce apoptosis remains controversial in many ways. p53 has long been considered a prime target for therapeutic modulation (Lane and Lain, 2002, Mashima and Tsuruo, 2005), but its function is frequently inactivated due either to gene mutations or to alterations of other components of its signaling pathway.

Several approaches aimed at restoring the lost p53 function in tumors have been developed. Past research has focused mainly on its tumor suppressor function as a sequence-specific transcription factor. However, more recent investigations have shown that p53 can trigger apoptosis in a transcription-independent manner. Since many p53 mutants with loss of transcriptional activity might still retain its ability to activate apoptosis through the transcription-independent pathway, there is considerable excitement that this newly identified mechanism will ultimately lead to a new generation of cancer therapeutics.

2. The influence of p53 status on chemotherapy response in human cancer: a paradox 

Given the well-documented apoptotic function of p53 in various model systems, it is natural for p53 to become a target for cancer therapy. Loss of p53 can contribute not only to aggressive tumor behavior but also to therapeutic resistance (El-Deiry, 2003, Vogelstein et al., 2000). The role of p53 in chemosensitivity is well established in rodent model system both in vitro and in vivo (Lowe et al., 1993a, Lowe et al., 1993b). However, the influence of p53 on chemosensitivity in human cancers is by no means clear, and the literature is confusing and often contradictory.

A positive correlation between p53 and chemoresponse has been suggested by a robust “in-house database” analysis of 60 NCI cancer cell lines treated with 123 different anticancer drugs (O’Connor et al., 1997). The study revealed that cancer cell lines with a mutant p53 tended to be less sensitive than those with the wild-type p53. However, many individual studies have produced conflicting results that show that p53 inactivation in cancer cells leads to increased sensitivity to chemotherapeutic agents (Fan et al., 1995, Gupta et al., 1997, Hawkins et al., 1996, Wahl et al., 1996). To provide a systematic evaluation of the published articles, Parodi and colleagues performed a meta-analysis of 356 independent studies concerning the p53 status and drug response (Cimoli et al., 2004). After analyzing isogenic systems in which p53 function is altered, the authors conclude that p53 alone is insufficient to determine sensitivity or resistance to cytotoxic drugs.

Clinically, the effect of p53 mutations on the sensitivity of tumors to the induction of a chemotherapeutic response has also been disputed and so far p53 has failed to demonstrate a definite role in predicting treatment response. Conflicting results have been observed in various cancers. While p53 status appeared to be positively correlated to the cisplatin response in ovarian cancer (Righetti et al., 1996), it does not seem to play a role in response to chemotherapy in small cell lung cancer (SCLC) patients (Kawasaki et al., 1997, Kawasaki et al., 1998, Rodriguez-Salas et al., 2001). Further, no correlation between p53 status and tumor responsiveness to etoposide was observed in primary NSCLC tested ex vivo for chemosensitivity (Vogt et al., 2002). In an effort to study the effect of p53 mutation on the response of breast cancers to chemotherapy, Bertheau et al. observed that among advanced breast cancer patients treated with the combination of epirubicin/cyclophosphamide, those with complete response all had p53 mutations (Bertheau et al., 2002). Although this variation in outcome of studies could be caused by the heterogeneity in methodology for response assessment and the response is probably dependent on the type of tumor as well as type and dosage of drugs, it does raise the possibility that chemotherapy-induced p53 activation in some cancers impairs, rather than favors death of tumor cells.

3. The role of p53-dependent transcription in apoptosis: pro- or anti-apoptotic? 

The p53 protein acts as a powerful transcription factor and has been extensively studied. It binds to as many as 300 different promoter elements in the human genome (Wei et al., 2006) and broadly alters patterns of specific gene expression (Kho et al., 2004). p53 activation following genotoxic stress results in apoptosis and cell cycle arrest. Although there is little doubt that p53-transcriptional activation of target genes such as p21 contributes to p53-dependent cell cycle control (El-Deiry et al., 1993, Harper et al., 1993), the precise mechanism for p53 to regulate apoptosis is not clear. Genetic and biochemical evidence suggests that the p53 engagement in apoptotic activation mainly proceeds through the intrinsic, mitochondrial pathway and this pathway is primarily regulated by the BCL-2 family of proteins (Schuler et al., 2000, Schuler et al., 2003, Soengas et al., 1999). It has been thought that p53 induces apoptosis through transcriptional activation of apoptotic target genes, such as PUMA, BAX, NOXA, BID, PIG3, CD95, DR5 or p53AIP1 (Vogelstein et al., 2000, Vousden, 2002, Vousden and Lu, 2002). This theory is well-supported because the tumor-derived common mutations have no ability to transactivate downstream target genes and to induce apoptosis. Consistent with the role of p53 transactivation in apoptosis, knockout mice lacking these p53 targets have impaired ability of p53 to induce apoptosis (Ihrie et al., 2003, McCurrach et al., 1997, Shibue et al., 2003). However, the relevance of transcriptional activation in p53 apoptosis in human cancer remains highly controversial. Despite increasing numbers of p53 targets being implicated in apoptosis, no single target gene has been found to be consistently associated with the major apoptotic effect of p53.

Several lines of evidence now indicate that p53 transcriptional activity does not always correlate to its apoptotic activity. By characterizing the apoptotic activity of 179 mutant p53s with diverse transcriptional activities, Kakudo et al. found that there is no correlation between p53 transcriptional activities on any single target gene (Kakudo et al., 2005), suggesting that transactivation does not play a major role in p53-dependent apoptosis. Furthermore, breast cancer patients carrying transcription-deficient mutant p53 have better response to chemotherapy than the p53 normal cancer patients (Bertheau et al., 2002). In colon cancer HCT116 cells, activation of p53 leads to apoptosis in response to the genotoxic agent 5-fluorouracil (5-FU), but to cell cycle arrest in response to adriamycin (Bunz et al., 1999). However, microarray analysis of p53-responsive genes indicates that the two drugs induced very similar profiles of p53-mediated transcriptional program (Yu Q., unpublished results). So, p53 transcriptional activity alone is not sufficient to explain its overall functionality. Furthermore, in a recent study in mouse fibroblasts treated with different doses of UV- and γ-irradiation, the primary response mediated by p53 transcriptional activity was found to be anti-apoptotic, while the pro-apoptotic response requires an additional, transcription-independent mechanism (Speidel et al., 2006). These studies raised a striking possibility that at least in certain context the p53 transcriptional program may elicit an anti-apoptotic response.

4. p21 as an anti-apoptotic gene: manipulation of p21 as a strategy for increasing therapeutic response 

A key molecule that might mediate the anti-apoptotic response of p53 is p21 cyclin-dependent kinase inhibitor. In response to DNA damage and subsequent p53 activation, cells undergo growth arrest and/or apoptosis. In addition to the well-defined p21 function in the induction of growth arrest, a critical role of p21 as an anti-apoptosis molecule has also been uncovered in recent years. It has been observed by several investigators that depletion of p21 in cancer cells led to an enhanced apoptotic effect of chemotherapeutic agents both in vitro and in vivo (Bunz et al., 1999, Bunz et al., 2002, Fan et al., 1997, Waldman et al., 1996, Wouters et al., 1997). Furthermore, p21 also acts to inhibit the pro-apoptotic effectors, such as procaspase 3, caspase 8, and ASK1 (Gartel and Tyner, 2002, Huang et al., 2003). These studies provide support for a role of p21 as the executer of the p53 anti-apoptotic function. In support of this notion, tumors with high levels of p21 are prone to be resistant to DNA-damaging agents (Kralj and Pavelic, 2003, Liu et al., 2004), whereas cancer cells with Myc overexpression have increased sensitivity to DNA damaging agents, since Myc suppresses p21 transcription and results in the conversion of DNA damage-induced p53 response from growth arrest to apoptosis (Seoane et al., 2002). Thus, p21 appears to be critical in maintaining the balance between growth arrest and apoptosis in response to DNA damage.

The p21 level can also be regulated through protein expression. It has been shown that WISp39, a Hsp90 binding protein binds to and stabilizes p21 protein by preventing its proteasomal degradation (Jascur et al., 2005). Knockdown of WISp39 reduces p21 protein stability and results in the bypass of cell cycle checkpoint control in response to DNA damage. A more recent study has shown that the small molecule mTOR inhibitor RAD001 can reduce p21 protein translation and stability and dramatically sensitize cisplatin-induced cell death in cancer cell lines (Beuvink et al., 2005). Thus, manipulation of p21 level appears to be a feasible approach for modulating chemotherapeutic response. These studies have important implications in cancer therapeutics. Many tumor cells appear to be highly prone to DNA damaging agent-induced apoptosis (Weiss, 2003). However, the difficulty with such agents is general toxicity combined with a narrow therapeutic window: too low a dose has no effect, whereas too high a dose cause damage to all cells (El-Deiry, 2003, Khanna and Jackson, 2001, Weiss, 2003). Thus, any sensitizer that can direct even a mild DNA damage response towards an apoptotic program would have the potential to enhance the efficacy of DNA-damaging chemotherapeutic agents and thus reduce the toxic effects. Thus, p21 might be a rational target for drug development. Accordingly, p21 antisense therapy has been shown to enhance radiosensitivity in human colon cancer cells (Tian et al., 2000). However, targeting p21 as a therapeutic approach should be considered with caution, since p21 is a “universal inhibitor” of cyclin-dependent kinases (Xiong et al., 1993) and is not overexpressed in human cancer and thus inhibition of p21 might cause hyper-proliferative side effect.

5. Small molecule activation of p53 as potential therapeutics 

A number of small molecule compounds have been reported to activate p53 (Table 1). p53 is maintained at low levels in cells, largely through MDM2-mediated degradation. Thus, rescue of p53 function by disrupting MDM2-p53 interaction is considered to be an efficient approach for anti-cancer therapy (Klein and Vassilev, 2004). Such drugs include the potent and selective small molecule inhibitors RITA (Issaeva et al., 2004), Nutlin-3 (Vassilev et al., 2004) and HLI98 (Yang et al., 2005). These non-genotoxic compounds bind to the p53-binding site on the MDM2 molecule, displace p53 from its complexes with MDM2 and prevent MDM2 from targeting p53 for degradation. This leads to p53-dependent apoptosis and broad antitumor activities in tumors with wild-type p53 (Carvajal et al., 2005, Coll-Mulet et al., 2006, Secchiero et al., 2006). Since p53 remains wild-type in the majority of hematological malignancies, this approach is particularly attractive in this group of cancers (Coll-Mulet et al., 2006, Secchiero et al., 2006). However, the apoptotic effect of Nutlin-3 in many p53 normal solid tumor cell lines is not as evident as its effect on growth arrest (Tovar et al., 2006), indicating that the pathway for p53 to induce apoptosis might have accumulated defects in these cancer cells. It should be pointed out that although activation of p53 transcriptional apoptotic targets has been observed in these studies, unfortunately no study has clearly demonstrated that the induced apoptosis by these drugs is truly dependent on the expression of apoptotic targets.

Table 1. Small molecule p53 activators

		Structure	Dose range	Mechanism	References
	Nutlin3		μM	Inhibits MDM2 activity and blocks MDM2-p53 interaction	Vassilev et al. (2004); Coll-Mulet et al. (2006); Carvajal et al. (2005); Secchiero et al. (2006); Tovar et al. (2006)
	RITA		μM	Binds to p53 and blocks the MDM2 binding to p53 and ubiquitation	Issaeva et al. (2004)
	HLI98		μM	Inhibits MDM2 ubiquitin ligase activity	Yang et al. (2005)
	PRIMA-1		μM	Restores sequence-specific DNA binding and the active conformation to mutant p53 proteins	Bykov et al., 2002a, Bykov et al., 2002b, Bykov et al., 2003

A second approach is to reactivate mutant p53 and restore normal p53 functions. Several compounds with such activity have been identified by random screening and the model compound PRIMA-1 has been shown to induce apoptosis in human tumor cells with different p53 mutations (Bykov et al., 2002a, Bykov et al., 2003). This compound can restore native conformation and wild-type function to mutant p53 and activates p53 target genes. In vivo studies in mice revealed an antitumor effect of PRIMA-1 with no apparent toxicity. However, its mechanism of action remains unresolved, since PRIMA-1 also induces apoptosis in cells without p53.

6. Transcription-independent pathway for p53 apoptosis: an opportunity for novel therapeutics 

In addition to its transcription-mediated mechanism, p53 can also induce apoptosis through a transcription-independent mechanism. This emerged from a number of studies some time ago (Caelles et al., 1994, Haupt et al., 1995) but has not received much attention until recently. After 2000, interest in this area has been fueled by several studies showing that the transcription-independent p53 apoptosis is linked to the intrinsic mitochondrial apoptotic pathway (Chipuk et al., 2004, Leu et al., 2004, Mihara et al., 2003, Moll and Zaika, 2001). These studies demonstrated that p53 can translocate to the cytosol and mitochondria upon apoptotic stimuli, where it physically interacts with the Bcl-2 family member proteins Bcl-xL and Bcl-2 and antagonizes their anti-apoptotic stabilization of the outer mitochondrial membrane (Chipuk et al., 2004, O’Brate and Giannakakou, 2003, Mihara et al., 2003). Mitochondrial p53 also directly promotes the pro-apoptotic activities of Bak and directly induces Bak oligomerization (Leu et al., 2004). In this case, p53 acts like a BH3-only activator of Bax/Bax. Hence, mitochondrial p53 has a dual action of neutralizing anti-apoptotic members as well as activating pro-apoptotic members of Bcl2 proteins. Moll and colleagues recently provided the first in vivo evidence that the direct mitochondrial p53 program contributes to its tumor suppressor activity in a mouse model (Talos et al., 2005). These findings have established a foundation for a direct p53 mitochondrial cell death program.

However, it is not clear what the physiological impact is of the transcription-independent mechanism for p53 apoptosis induction and how the two functions of p53 are coordinated. A recent study monitored the kinetics of p53 translocation to mitochondria upon DNA damage in mice and demonstrated biphasic kinetics of p53 apoptosis (Erster et al., 2004). In this study, a rapid p53 translocation to the mitochondria was observed upon γ-irradiation, which was followed by a rapid first wave of caspase-3 activation. Apoptosis initiated in this stage does not seem to require the p53 apoptotic target protein PUMA and thus appears to be transcription-independent. Interestingly, the transcriptional p53 response, as monitored by up-regulation of p53 targets, occurred significantly later and was followed by a second wave of caspase-3 activation. This study indicates that upon stress p53 triggers apoptosis through both transcription-dependent and -independent mechanism in murine tissue and mitochondrial apoptotic pathway precedes the transcriptional mechanism and is responsible for the early events of cell death.

Establishing the p53 direct apoptotic mechanism thus paves the way for the new therapeutic opportunity. Pharmacologic activation of the transcription-independent apoptosis program has since been explored (Chipuk et al., 2003, Tan et al., 2005). PRIMA-1 is a small drug that can modulate the ability of mutant p53 to bind DNA and thereby restore its activity as a transactivator (Bykov et al., 2002b). It has been thought that PRIMA-1 induces apoptosis by reactivating p53-transcriptional program. However, Chipuk et al show that this drug can induce p53-dependent apoptosis in the absence of transcription (Chipuk et al., 2003), indicating that this drug can directly activate p53 direct apoptosis pathway. We have shown that a small molecule kinase inhibitor targeting GSK-3β (LY2119301) can trigger p53-dependent apoptosis upon DNA damage through the direct mitochondrial pathway (Tan et al., 2005). Importantly, this apoptosis does not require p53 target PUMA but requires Bax that undergoes conformational activation in this process. Interestingly, LY2119301 has another function to modulate p53 pathway: it abrogates p53 transcriptional activity upon DNA damage, leading to impaired p21 induction. Thus, the combinatorial effect of this dual function is striking: it reduces the anti-apoptotic transcriptional function of p53 but activates the pro-apoptotic direct mitochondrial pathway as outlined in Fig. 1. As a result, the p53 response to DNA damage is converted from growth arrest to apoptosis. Taken together, the above data support a model in which the p53 direct mitochondrial pathway for apoptosis is suppressed in cancer cells through a GSK-3β-mediated mechanism, which leads to the direction of p53 response primarily to growth arrest. Pharmacologic modulation of GSK-3β reverses this suppression and restores p53 apoptotic capability in response to DNA damage.

View Large Image Download to PowerPoint

Fig. 1. A dual role of GSK-3β inhibitor LY2119301 in modulating p53 response. DNA damaging chemotherapeutic agents activates p53 and induces growth arrest, through the induction of p21, or apoptosis through the transcription-independent Bax activation. In many cancer cells, p53 activation favors growth arrest instead of apoptosis. LY2119301 can suppress the p53 transcriptional activity and thus reduces p21 induction. In addition, LY2119301 can trigger p53-dependent Bax activation upon damage. This dual function of LY2119301 converts DNA damage induced p53 response from growth arrest to apoptosis.

p53 activation by anticancer drugs leads to the induction of apoptotic target genes and thus cell death in both cancer and normal tissues. In primary cells like mouse embryonic fibroblasts (MEFs) and rat fibroblast cells, it has been clearly demonstrated that p53-dependent transactivation by DNA damaging agents is required for p53-mediated cell death, which results in toxicity in normal tissues. Thus, compounds that can inhibit p53-induced transcription are expected to reduce the toxicity in normal tissues. This strategy has been preliminarily tested using a small molecule inhibitor of p53, pifithrin-α, which protects mice from lethal doses of IR through inhibition of p53-mediated transcription (Komarov et al., 1999). This treatment might be helpful in reducing the side-effects, but does not necessarily lead to improved efficacy of radio-or chemo-therapy in cancer cells. Modulating DNA damage-induced p53 pathway by LY2119301 could achieve both of the objectives, since the dual function of this compound might be able to maximize the therapeutic efficacy in cancer cells while protecting normal tissues from toxicity. Although highly speculative, the above hypothesis and the clinical relevance of the unique feature of LY2119301 deserve further studies.

7. Targeting mutant p53 for tumor-selective killing 

Many tumors overexpress mutant forms of p53. Despite the loss of transcriptional activity in the majority of p53 mutants, many tumor-derived p53 mutants retain their ability to induce apoptosis. It has been shown that expression of DNA-binding domain mutant, such as p53R175H, retargeted to the mitochondria, is still able to induce apoptosis (Marchenko et al., 2000). The p53QS transactivation-deficient mutant also shows apoptotic activity in response to certain forms of stress (Johnson et al., 2005). In rat mammary carcinoma cells, a DNA-binding mutation at cordon 246 causes deficiency in p53 transcriptional activity but gains enhanced ability to induce apoptosis (Hamaguchi et al., 2006). The small molecule compound PRIMA-1 has been shown to restore normal p53 transcriptional functions to mutant p53. However, this compound can also activate transactivation-deficient mutant p53 for apoptosis. These studies point to a possibility that at least some of the mutant p53s mediate their apoptotic activity through the transcription-independent pathway. Thus, targeting transcription-independent apoptosis pathway appears to be a reasonable therapeutic strategy for both p53 normal and mutant cancer cells. Moreover, given that mutant p53 is often expressed in significantly higher amounts than the wild type p53, there is a chance that the cells with mutant p53 might have a better response to this therapeutic strategy, resulting in potentially preferential killing in p53 mutant tumors.

8. Concluding remarks 

p53 has long been considered a prime target for therapeutic intervention. The role of p53 in apoptosis is complex, as it is able to promote and to suppress apoptosis. Although the p53 transactivation function can contribute to apoptosis, it might not be the primary mechanism at least in some human cancer cells. Instead, activation of the transcription-independent direct mitochondrial pathway might be a more attractive therapeutic strategy. Although some of the approaches outlined here are still in preclinical studies, they might have merits in the clinic. Further studies will elucidate how the direct p53 apoptosis pathway is regulated in cancer cells. Novel mechanisms for p53 to induce apoptosis will continue to emerge and new mechanism-driven therapeutic approaches will arise accordingly. This will demand the use of well-defined models, together with the enthusiasm to test the idea at the edge of possibility. With luck, the next decade will see new classes of cancer drugs emerging from targeting these new pathways.