Use of p53 for Therapy of Human Cancer

Abstract

Tumor suppressor p53 is the central component of a system maintaining the genetic stability of animal and human somatic cells. Its gene is inactivated in almost all human cancers, allowing a tumor cell to rapidly accumulate additional mutations and progress toward a more malignant phenotype. Yet tumor cells are most sensitive to the suppressor effect of p53 when its function is restored. Hence, restoration of the p53 function is an appealing strategy of anticancer therapy. Various mechanisms inactivate p53 in cancer, including point mutations resulting in synthesis of an inactive mutant protein, deletion of the total gene or its portion, damage to the genes involved in regulating the p53 activity, and defects in p53 target genes. In addition, oncogenic viruses code for the specialized proteins that modify the p53 function to ensure optimal replication of the virus genome. These viral proteins are crucial for virus-induced carcinogenesis, in particular, in 95% of cervical carcinoma cases in women. The approaches to p53 activity restoration depend to a great extent on the defect in p53-dependent signaling. Introduction of exogenous p53 is effective in some case and is usually achieved with adenoviral vectors. The approaches under study are aimed at restoring the activity of mutant p53 or suppressing the viral inhibitors of p53. The review considers various schemes involving p53 in cancer therapy and prevention and discusses their potential efficacy and prospects of their clinical use.

INTRODUCTION

The past decade has brought a substantial progress towards a better understanding of the mechanisms controlling cell division and cell death. As these processes are most often affected in cancer cells, identification of their critical steps suggests potential targets for therapy. Cancer therapy is aimed at eliminating cancer cells with a minimal effect on normal cells and tissues. Standard therapeutic exposures take advantage mostly of the physiological features of cancer cells such as a high proliferation rate or changed metabolism. Such therapy often causes serious adverse events and is toxic to normal tissue components, especially to rapidly proliferating hematopoietic and intestinal epithelial cells. Cancer therapy is accompanied by substantial stress reactions, leading to fibrosis and premature aging of tissues. Identification of particular target genes facilitates the development of targeted therapy of cancer.

Cancer is uncontrolled proliferation of genetically changed cells, which have lost the ability to adequately respond to the demands of the body because of a distortion of many regulatory mechanisms [1]. It should be noted that the purpose of therapy is to selectively suppress cancer cell proliferation or to eliminate cancer cells rather than to convert them to normal cells.

Carcinogenesis is a multistep process that involves a gradual accumulation of genetic and epigenetic alterations and leads to selection of the most autonomous malignant cell variants [2]. As a result, cancer cells can have dozens of mutations, which alter many signaling pathways. Somatic cells normally have an extremely high genetic stability, which allows most cells of the body to remain genetically homogeneous throughout the life. When systems responsible for the genetic stability function properly, one cell can hardly accumulate many mutations. However, when these systems are themselves affected by mutations, the cell is unprotected from further accumulation of genetic lesions. It is this scenario that is observed in most cancer cases.

The p53 tumor suppressor gene plays the key role in maintaining the genetic homogeneity of somatic cells and is most often affected in cancer [3]. Inactivating mutations of p53 are found in half of all cancer cases. Inheritance of a mutant p53 allele leads to early-onset multiple cancer (Li-Fraumeni syndrome) [4]. A model of this syndrome is provided by mice with a p53 deletion: such mice display an extremely high genome instability and inevitably develop malignant lymphoma before 9 months of age [5].

Thus, functional incompetence of p53 destabilizes the genome and quickly leads to cancer. Is a restoration of the p53 function capable of suppressing the malignant phenotype in cancer cells having accumulated multiple genetic alterations? The available data indicate that elimination of defective cells is one of the main p53-dependent mechanisms that maintain the genetic stability by preventing proliferation of genetically modified cells. Reactivation of p53 may trigger these mechanisms in cancer cells, suppressing their proliferation or causing their death.

MAIN FUNCTIONS OF p53

It is known that p53 is the central component of the system that prevents the generation of genetically altered cells in the body. Although p53 is permanently transcribed in all cells of the body, its protein product does not accumulate under normal conditions but is rapidly degraded by the proteasome [6]. Degradation of p53 is regulated by its interaction with specific E3 ubiquitin ligases, the best known one being encoded by MDM2. This interaction is finely regulated via multiple modifications (phosphorylation, acetylation, sumoylation, methylation, etc.) of p53 and its competition with other proteins (e.g., p14ARF). Modifications arise in response to various signals and weaken the interaction of p53 with ubiquitin ligases, thus suppressing p53 degradation and increasing the p53 pool in the cell [7]. Apart from affecting the p53 content, modifications qualitatively modulate the functional properties of p53.

Dozens of modifications have been described to affect the p53 function in response to signals transmitted via numerous signal transduction pathways [8]. The p53 structure reflects the state of many processes occurring both within and beyond the cell. Thus, p53 variants with different modifications differ in activity and determine different fates of the cell.

The main, but not the only, function of p53 is related to its transcriptional activity: p53 activates or suppresses transcription of specific genes [9]. Transcriptional suppression is achieved via several mechanisms, including the recognition of specific nucleotide sequences in particular genes. To exert most of its effects, p53 acts as a specific transcription factor and induces transcription of more then three hundred genes [10]. The set of genes controlled by p53 depends on many factors, including the p53 level, cell type, cell state, etc. The regulatory activity of p53 changes according to these factors. Under physiological conditions (without stress), p53 occurs at a low level and activates transcription of a few genes such as the genes for regulators of glycolysis and mitochondrial respiration [11] or genes involved in controlling the cell pool of reactive oxygen species [12]. These activities are essential for maintaining energy homeostasis and protecting the genome from mutations resulting from DNA oxidation by peroxides, which are produced by mitochondria and membrane-associated NADPH-dependent oxidases in response to activation of some signaling pathways. Physiological modulation of the content and qualities of p53 changes the expression of some genes that are involved in tissue responses and affect the cell properties such as adhesion, mobility, chemotaxis, and secretion of certain factors. A greater increase in p53 content and activation of p53 via additional modification occur when the cell is exposed to various stress factors (irradiation or DNA damage, including breaks and base modification), transcription is distorted, the pool of DNA and RNA precursors is depleted, the level of growth factors is inadequate, the cell lacks necessary contacts with other cells or the extracellular matrix, the cytoskeleton assembly is distorted, chromosome segregation and the spindle integrity are altered, the expression of some genes is changed (e.g., in response to oncogene activation), the cell experiences hypoxia or hyperoxia, and in many other situations [13]. The response to an increase in p53 depends to a great extent on the lesion and the tissue where the cell belongs, because these factors determine the character of p53 modification and the availability of factors cooperating with p53.

Generally, p53 occurring at a relatively low level arrests cell division and simultaneously activates repair. This p53 activity allows the cell to eliminate the lesion; then, p53 modification is reverted, the intracellular pool of p53 decreases, and the cell recovers its normal functional state. The major role in the G1 arrest is played by p53-inducible CDKN1, which codes for the cyclin-dependent kinase inhibitor p21 [14]. In addition, p53 regulates several other genes (GADD45, 14-3-3-[sigma], REPRIMO, etc.) whose activity facilitates the cell cycle arrest in G1, G2, S, or M.

When stress factors are intense, DNA is dramatically damaged, or cell physiology is permanently altered by mutations or virus infection, distorting the regulatory processes, qualitative changes in p53 activate programmed cell death [15]. Acting as a transcription factor, p53 regulates many components of several apoptotic pathways, including the Apaf1 activator of the caspase cascade, the BH3 domain-containing proapoptotic proteins Bax and PUMA, apoptotic receptors APO/FAS and KILLER/DR5, endoplasmic reticulum stress-mediating proteins NOXA and scotin, and macroautophagy inductor DRUM. In addition, p53 activates the genes involved in production and excretion of large amounts of reactive oxygen species, which promote mitochondrial sensitization to proapoptotic effects. Apart from its effects on transcription, p53 can directly induce cell death by interacting with mitochondrial components and stimulating the conformational transition of Bax into an active proapoptotic state [16]. The multiplicity of cell death-inducing pathways is probably related to the fact that cells of different histogenetic lineages utilize different mechanisms of apoptosis, while the inductor role is always played by one protein, p53.

ALTERATIONS OF p53-DEPENDENT MECHANISMS IN CANCER

As evident from the above, p53-dependent mechanisms continuously monitor the cell state and prevent proliferation of abnormal cells. This activity provides a reliable barrier to an accumulation of mutations and induction of carcinogenesis. Yet the system protecting the genome and employing unique p53 as a central component is highly vulnerable: its function is nonredundant. When p53-dependent mechanisms are damaged, the cell loses the ability to regulate its state according to the programs acquired and fixed in the genome, which leads to a rapid accumulation of genetic lesions and causes pathology.

In half of all human cancer cases, p53 is affected by point mutations often leading to amino acid substitutions. In the other half, p53-dependent mechanisms are also altered, but the character of alterations is highly diverse. Such alterations may involve the genes that control p53 induction, the target genes of p53, or the components affected by the products of these genes. Damage to p53-dependent mechanism is often caused by overexpression of MDM2, which codes for a p53-regulating protein. As a result, the level of active p53 is decreased and p53 induction in response to stress is weakened. The mdm2 activity increases when MDM2 is amplified, its transcriptional regulation is distorted, or some factors alter the production of p19ARF, which regulates the binding of mdm2 and p53. Many DNA viruses code for specialized proteins inhibiting the p53 activity. For instance, p53 is inhibited by the SV40 large T antigen or the product of adenoviral E1B. The E6 product of human papillomaviruses (HPV) 16 and 18 accelerates proteolytic degradation of p53, which is observed in more than 95% of all human cervical carcinomas.

Thus, damage to p53 functions is the most common property that differentiates cancer and normal cells. It seems promising to take advantage of the difference between cancer and normal cells in cancer therapy. Attempts have been made to selectively eliminate the cells devoid of p53 activity or to restore this activity in cancer cells in order to induce their apoptosis. These issues are considered in detail below.

SELECTIVE ELIMINATION OF CELLS DEVOID OF p53 ACTIVITY

The lack of p53 activity in most cancer cells can be used for their selective elimination. For instance, normal cells can be protected by transformation with a p53-dependent gene conferring resistance to a certain drug. It is also possible to introduce a construct that determines synthesis of a toxic product along with expression of its antagonist from a p53-dependent promoter [17]. In this case, normal cells with the intact p53 function are protected from the toxic effect, while cancer cells are sensitive. Unfortunately, such strategies are unfeasible for the total body, requiring preliminary transformation of all cells with a p53-dependent gene for a transcriptional suppressor.

Another, more realistic, approach is based on selective virus-dependent lysis of cancer cells. Many viruses inhibit p53 to achieve a more efficient control over the cell processes. For instance, human adenoviruses stimulate quiescent cells to enter the S phase, which is necessary for replication of the viral genome. Such stimulation is due to viral E1A, which blocks the pRB suppressor. E1B-55kDa simultaneously inhibits p53: this is essential for adenovirus replication, because E1A induces p53-dependent apoptosis. E1B-55kDa acts together with the E4-ORF6 product to accelerate proteolytic degradation of p53. It has been assumed in view of this that the dl1520 adenovirus mutant [18], which has a partial deletion from E1B and does not produce the 55-kDa protein, is not replicated in normal cells but does reproduce and cause death in cancer cells with a functional defect in p53 [19]. This assumption has been verified in several works. Indeed, the mutant (commercial name Onyx-015 [19]) is incapable of reproducing in primary cells and reproduces well in some p53-negative cancer cells, causing their death [20]. There is evidence, however, that Onyx-015 similarly well reproduces in cancer cells synthesizing the wild-type p53 [21]. This finding poses a question as to whether the Onyx-015 effect depends on the p53 status. Cancer cells are poorly differentiated and produce a greater amount of the surface-associated coxsackie- and adenovirus receptor (CAR) [22]: their infection with these viruses is more efficient, which explains selective oncolysis [23]. Another possibility is that p53-positive cancer cells are sensitive to Onyx-015 because their p53-dependent mechanisms are defective regardless of the state of p53 [24]. For instance, a defect in p14ARF leads to a loss of p53 function [25]. Onyx-015 cannot induce p53 in p14ARF-deficient cells and, consequently, is efficiently reproduced. In addition, the selective effect of Onyx-015 can be related to the fact that cancer cells efficiently sustain the nuclear export of late adenoviral RNA, which is essential for adenovirus expression, while this function needs induction with E1B in normal cells [26].

Whichever the case, experiments with cell cultures and preclinical trials have demonstrated the selectivity and safety of Onyx-015, allowing a phase III clinical study in patients with head-and-neck cancer. The study results are promising: the virus selectively replicates in cancer cells to cause their death [27]. Onyx-015 is even more effective when combined with standard chemotherapy with 5-fluorouracil (5FU), cisplatin [28], or doxorubicin and taxol [29]. Owing to selective infection of cancer cells, recombinant adenoviruses with an E1B deletion can be used to simultaneously deliver the gene for uracil phosphoribosyltransferase, which helps to overcome 5FU resistance. This variant of the virus in combination of 5FU proved highly effective in pancreatic cancer [30].

Another protocol of selective adenoviral oncolysis of p53-negative cells takes advantage of the p53-dependent system protecting normal cells from adenovirus infection. Expression of the adenoviral genome depends on the E2F cell transcription factor. This factor is especially active in the presence of adenoviral E1A, which inactivates the pRB tumor suppressor and releases E2F from its complex with pRB. Ramachandra et al. [31] have constructed adenovirus 01/PEME, which contains a gene for an artificial E2F antagonist under the control of a p53-dependent promoter, which is expected to decrease replication of the adenovirus in normal (p53-positive) cells without affecting its tropism to cancer cells. In addition, 01/PEME has a higher capability of killing cells owing to an artificial increase in synthesis of E3 (11.6 kDa), which functions to release the virus progeny from the cell at the late stages of the infection cycle. A testing with cell cultures has confirmed that 01/PEME is highly selective for cancer cells. The clinical efficacy and safety of 01/PEME is to be evaluated.

OVERPRODUCTION OF EXOGENOUS p53 IN CANCER THERAPY

Not only do cancer cells have a defect in p53-dependent mechanisms, but they also acquire a higher sensitivity to exogenous p53. Dramatic alterations of normal processes in cancer cells continuously generate pathological signals, which fail to arrest the cell cycle in the absence of p53 activity. When constructs expressing the wild-type p53 are introduced in cancer cells, such signals exert a massive suppressor effect, arresting the cell cycle or inducing apoptosis. This property is widely used to develop therapeutic strategies, wherein the main problem is to deliver p53-expressing construct to cancer cells.

The highly selective delivery of a construct to cancer cells is a difficult problem, which is unsolved as of yet. Approaches to this problem are developed taking advantage of some properties of cancer cells. For instance, hepatic cancer cells are characterized by a high-level expression of the epidermal growth factor receptor (EGFR). To deliver a plasmid for p53 expression in a targeted manner, the plasmid was introduced in complex with two specific peptides, which were adsorbed on DNA with polylysine. One peptide corresponded to the EFG region binding to EGFR and the other corresponded to the hemagglutinin region involved in endosmolysis. The complex preferentially bound to hepatic cancer cells and facilitated the release of vector DNA from endosomes, efficiently suppressing the growth of human tumors in athymic mice [32]. Another strategy utilized the transferrin receptor, whose synthesis is also increased in cancer cells. Plasmids were delivered with cationic liposomes, which are known to act as safe but low-specific transporters [33]. The introduction of transferrin in liposomes substantially increased the specificity of delivery, leading to lysis of tumor xenografts in athymic mice [34]. This strategy was used in combination with radiotherapy in human prostate cancer and demonstrated a high efficiency of delivery and an improved therapeutic effect [35]. A variant of the strategy utilized cationic liposomes incorporating single-chain antibodies to the transferrin receptor: their combination with a p53-expressing plasmid and the chemotherapeutic drug doxetacel was highly effective in treating xenografts of human breast cancer in athymic mice [36]. Promising approaches are based on the use of magnetic liposomes [37]; photochemical stimulation of liposome transfer [38]; and nanoparticles, which allow a prolonged and stable release of p53-expressing plasmids [39]. An interesting approach to p53 delivery into cells takes advantage of the ability of some proteins (single-chain antibody scFv and HIV protein tat) to penetrate into the cell and accumulate in the nucleus. Fused with such proteins, p53 is efficiently delivered into cancer cells and induces apoptosis by triggering the expression of p53-dependent genes [40]. VP22 of the herpes simplex virus is an especially efficient p53 transporter: this protein ensures not only the penetration of p53 into the cell, but also its release into the extracellular space and cell-to-cell transfer [41].

Virus-mediated p53 delivery is now best developed in the context of practical application. Many viruses are capable of foreign gene transduction and are used to construct various vectors. Attempts of a therapeutic p53 delivery have been made with retroviruses [42], the Semliki Forest virus [43], the herpes simplex virus [44], the vaccinia virus [45], and even an insect baculovirus [46], each producing a considerable effect in cell cultures and athymic mice with xenografts.

Each of the above viruses has its advantages (e.g., the possibility to simultaneously deliver several genes [43]) as well as many drawbacks, requiring further improvement. The most promising vectors are replication-defective adenoviruses, which proved to efficiently and safely deliver the p53 cDNA in cancer cells [47]. Such modified adenoviruses have passed multistep trials and are now commercially available for clinical use. Human type 2 and 5 adenoviruses with a low oncogenic potential are mostly used for this purpose, but certain advantages are offered by vectors based on animal adenoviruses, such as the avian adenovirus CELO [48]: humans have no immunity to such viruses, while those who have a history of adenovirus infection are immune to human adenoviruses.

Recombinant adenoviruses efficiently deliver p53 to cancer cells and display sufficiently low nonspecific toxicity in cell cultures and animal models. Introduced by this means, p53 induces its target genes, suppresses the growth of cancer cells, triggers their apoptosis, and causes regression of various tumors, including head-and-neck carcinoma [49], non-small cell lung cancer [50], ovarian cancer [51], bladder cancer [52], prostate cancer [53], osteosarcoma [54], and glioma [55]. Since only some cells of a tumor are efficiently infected with adenoviruses, the strategy of virus delivery still needs improvement to achieve an optimal effect, especially when the target site is difficult to access. However, infection of a minor portion of tumor cells may already have a substantial therapeutic effect. This is due to the fact that cells producing p53 to a high level affect the adjacent cancer cells via cell-to-cell contacts [56] or certain secreted factors, which affect various tissue responses. For instance, such factors can suppress angiogenesis, thus impairing the blood supply to the tumor, or stimulate the activity of natural killers [57].

Early clinical studies of p53-expressing adenoviruses yielded relatively modest results [58]. Though suppressing the tumor growth, monotherapy with p53 did not eliminate most cancer cells and, consequently, failed to exert a prolonged therapeutic effect. It has been noted, however, that the therapeutic effect is especially high when p53 is combined with other treatments, as p53 increases the sensitivity of tumors to radio- and chemotherapy [59]. Modern strategies of p53 therapy are mostly based on combined approaches [60]. For instance, a combination of p53 with cisplatin is effective in non-small cell lung cancer, arresting tumor progression for a longer period [61]. Protocols are developed to combine p53 therapy with expression of some other genes, such as p14ARF [61], p33ING1 [63], p16 [64], IL2 [65], and GM-SCF [66], or with repression of the genes for telomerase [67], cyclin D1 [68], PCNA [69], or clusterin [70]. Attempts are made to combine p53 with photodynamic therapy [71], thermotherapy [72], HDAC inhibitors [73], and radiotherapy [74]. Surprisingly, the therapy had no adverse effects expected from the background production of exogenous p53 in normal cells. Possible explanations are that the p53 activity is strongly regulated in normal cells even when the p53 dosage is increased [75] and that normal cells are far less efficiently infected owing to a low, if any, level of adenovirus receptors [76].

Since clinical studies of p53 yielded encouraging results, attempts have been made to “improve” p53, adapting it for therapeutic purposes. The protein should be maximally active in cancer cells and should predominantly trigger their apoptosis rather than arrest their division. In addition, a therapeutic p53 should be insensitive to the negative regulatory effects of proteins regulating the p53-dependent pathways, such as mdm2.

A modified p53 has a short deletion from the N-terminal region, involved in mdm2 binding [77]. The adenovirus producing this protein more efficiently induces apoptosis in cancer cells. An artificial p53 variant insensitive to mdm2-mediated ubiquitination has been constructed by substituting Ala for several Lys residues in the C-terminal region [78]. An additional advantage is that this p53 variant is capable of spreading into neighbor cells owing to the polyarginine tract, which stimulates the transfer of the protein across the plasma membrane [80]. The introduction of a fragment of influenza virus hemagglutinin 2 improves the nuclear transport of p53; the modified protein more efficiently suppresses proliferation and induces death in malignant glioma [81]. It is important that the exogenous protein be not inhibited via the dominant negative effect of mutant p53, which is often produced to a high level in cancer cells. To prevent such inhibition, we have constructed a hybrid p53 whose C-terminal domain, responsible for oligomerization of the human protein, is replaced by its chicken counterpart. The hybrid protein has a high transcriptional activity, which is not suppressed even at high concentrations of mutant p53 produced in SW480 carcinoma cells [82].

More radical changes have been introduced in the artificial p53-like protein CTS-1, which preserves only the p53 DNA-binding domain, recognizing the specific responsive elements in p53-inducible genes. The p53 N-terminal domain, which is responsible for transcriptional activation and p53-inhibiting interactions with mdm2, viral EBNA5, and E1B, has been replaced by the transactivation domain of herpes virus VP16. The p53 oligomerization domain, which is close to the C end and is important for the dominant negative effect of mutant p53 and for the negative regulation by adenoviral E4orf6, has been replaced by a highly efficient artificial leucine zipper. The C-terminal domain, negatively regulating the DNA-binding activity, has been deleted [83]. Despite these dramatic changes, artificial CTS-1 is as efficient as p53 in binds to specific DNA sequences and occurs as a dimer in solution. CTS-1 efficiently induces the p53-dependent genes and exerts the suppressor and proapoptotic effects when introduced in cancer cells. Moreover, CTS-1 is virtually insensitive to the negative effects of mdm2 and viral p53 inhibitors, including HPV E6. It is of interest that CTS-1 is even more effective than the wild-type p53 in inducing the p53-dependent genes involved in apoptosis [84]. Introduced with the help of an adenovirus, CTS-1 induces death of malignant glioma cells regardless of the functional state of their p53 [85]. This cell death is unusual: it does not involve caspases or a release of cytochrome c, nor is it blocked by the antiapoptotic protein bcl2. Although many aspects need investigation for the clinical use of such artificial proteins, they still seem promising for therapy.

RESTORATION OF ENDOGENOUS p53 ACTIVITY

When p53 is introduced with a vector, the expression construct cannot be delivered exclusively in tumor cells but enters the adjacent normal tissue as well. Although normal cells are less sensitive to the p53 effect, the mere introduction of expression DNA cassettes in cells is undesirable and involves risk of long-term adverse effects. Hence, an appealing idea is to rescue the function of the inactive endogenous p53 by eliminating the inhibiting effect of regulatory (in particular, viral) proteins or improving the conformation of the mutant p53. This approach is aimed at correcting the primary defect and, consequently, implies a minimal effect on normal cells.

Restoration of Mutant p53 Activity

Mutations of the p53 coding region are found in half of all human cancer cases. Thus, p53 is the most mutable gene in carcinogenesis. Missense mutations, causing amino acid substitutions, are most frequent in p53. Substitutions lead to a loss of function of the protein product. With the majority of genes, a loss of function of one allele just decreases the level of gene expression or halves the activity of the protein product. Since p53 is active as a homooligomer, the product of the mutant allele is capable of a dominant negative effect; i.e., the mutant p53 binds with the product of the intact allele to suppress its function. This explains why missense mutations account for 75% of all p53 mutations. Mutant proteins usually accumulate to large amounts in cancer cells, which is partly explained by a loss of function of p53, including a loss of the ability to induce its negative regulator mdm2 [86]. In heterooligomers, the mutant p53 acts as a dominant negative inhibitor and simultaneously stabilizes the wild-type p53 [87]. It is probably for this reason that the wild-type p53 allele is under selective pressure in cancer cells and is eventually eliminated, usually by a deletion of a chromosome region [88]. Apart from a loss of function, the mutant p53 accumulates in the cell in a large amount and exerts additional effects, suggesting mutations with a gain of function [89]. New functions can be gained via the interaction of the abundant mutant p53 with p53-related proteins, p63 and p73 [90]. Mutants can vary in the spectrum of new functions, which are possibly subject to selection in tumor progression, usually tending to enhance the transformed phenotype of cells or to increase the cell resistance to therapeutic exposures [91].

More than 20,000 p53 mutants representing about 2300 mutant variants have been examined to date [92]. One-third of all mutations occur in hot-spot codons 175, 245, 248, 249, 273, and 282, which is explained first and foremost by the key role of the corresponding amino acid residues in the p53 function. Other mutations often affect the central one-third of the protein-coding region, i.e., the protein sequence that is responsible for the recognition of the specific DNA elements in p53-dependent genes. As a result, mutant p53 loses the capability of binding to the responsive elements of genes or its binding is weakened, changing the set of p53-activated genes.

Missense mutations of p53 can be conventionally divided into three classes, which are characterized by a loss of oligomerization (I), an impaired binding to DNA (II), or a distorted folding of the central (core) domain (III) [93]. Mutations of the last two classes are more frequent. Class I mutations, or contact mutations, cause amino acid substitutions in the p53 region directly contacting DNA (e.g., mutation hot spots Trp248 and His273), which changes the strength of p53 binding to DNA sequences. The amino acid substitutions that distort the folding of the core domain dramatically change the protein structure so that the mutant is virtually incapable of binding to DNA. Such mutations (e.g., those affecting His175, Ser249, and Trp282) are assigned to class II (structural mutations). Owing to stable denaturation of the protein molecule, such mutants are tightly associated with heat shock proteins in the cell [94] and acquire a new antigenic structure recognizable by monoclonal antibodies [95]. Structural mutations decrease the p53 denaturation temperature [96] and some p53 variants act as temperature-sensitive mutants, displaying a normal activity at a lower temperature and being inactivated at the body temperature.

High-level synthesis of mutant p53 in the tumor cell has been regarded as a loaded gun [97], since factors restoring the p53 activity can exert a potent selective effect on the cancer cell. It should be noted, however, that it is far more difficult to restore the activity of a damaged protein than to inhibit it and that conformational correction of a denatured protein is especially intricate. The problem is further complicated by the multiplicity of mutant p53 variants: treatments activating some variants may be ineffective with some others. Many approaches have recently been proposed to rescue the activity of mutant p53, some being rather sophisticated. One employs ribozymes to correct the mutant p53 mRNA via trans-splicing [98]. The possibility of such a correction was demonstrated with several cancer cell lines: up to 10% of all p53 mRNA molecules were corrected, which was accompanied by the appearance of active p53, induction of the p53-dependent genes for p21 and Bax, and p53-dependent apoptosis [99]. A practical application of this approach is as yet problematic.

Another approach is based on compensatory substitutions in the mutant p53 gene and is similarly far from applicable. Certain substitutions allow p53 to fold into an active conformation even in the presence of a mutation [100]. This makes it possible to expect that some external factors may act on mutant p53 to facilitate its conformational activation.

The p53 activity can be regulated via C-terminal modification that suppresses the inactivating effect of the region including the 30 C-terminal residues. Monoclonal antibodies interacting with this region appreciably increase the DNA-binding activity of p53 and partly restore the capability of some p53 mutants to bind to DNA [101]. Surprisingly, a similar effect has been observed in the presence of a short peptide corresponding to the C-terminal region of p53 [102, 103]. The peptide binds to the core domain of p53 [104] and rescues its suppressor properties to induce apoptosis in cells expressing mutant p53 [103]. A considerable therapeutic effect of p53 peptides has been achieved in mouse tumor models. The peptide was fused with HIV tat to allow its delivery into the cell [105].

The DNA-binding properties of p53 are enhanced by the 9-mer peptide CDB3, which corresponds to a region of the p53-binding protein 53BP2 or ASPP [106]. Interacting with p53 mutants, CDB3 stabilizes their active conformation and partly restores their ability to induce apoptosis of cancer cells [107]. Thus, short peptides can act as chaperones, facilitating the folding of mutant p53 into an active conformation.

Small molecules have been extensively tested for rescuing the activity of mutant p53. A screening of a chemical library comprising 100,000 compounds has yielded the small molecule CP-31398, which substantially increases the denaturation temperature of the p53 core domain in vitro [108]. CP-31398 efficiently stabilizes some p53 mutants and restores their activity in cell cultures and experimental mice. The mechanism of its action remains unclear. CP-31398 does not bind with p53 in the cell and acts as a DNA-intercalating agent [109], exerting a nonspecific cytotoxic effect. It is possible that CP-31398 interacts only with newly synthesized p53 molecules to promote their correct folding or its effects observed with cell and animal models are mediated by p53-independent mechanisms. Experiments with CP-31398 have shown again that the results obtained with small molecules in vitro are difficult to extrapolate to in vivo models, where potential targets are far more numerous.

Derivatives of the alkaloid ellipticine, whose antitumor activity is well known, are capable of rescuing the activity of some p53 mutants with structural and contact mutations [110]. Ellipticine derivatives rescue the normal conformation of p53 and p53-dependent induction of p21 and MDM2. In particular, these effects have been observed in human tumors xenografted to nude mice [110].

Another antitumor compound proved to activate some p53 mutants is WR1065, a derivative of the cytoprotective agent amifostine [111]. Amifostine is a thiol compound that is clinically used to protect normal cells of the body from antitumor radiotherapy and chemotherapy. The mechanism of WR1065 action on the mutant p53 is unclear. WR1065 induces p53 without exerting a cytotoxic effect in normal cells and causes apoptosis in cells producing mutant p53. WR1065 is capable of reducing cysteine residues and reactivation of mutant p53 possibly involves a redox component [112].

A screening of a chemical library for compounds that selectively induce apoptosis in cancer cells expressing mutant p53 has yielded PRIMA-1, which restores the activity of a broad range of p53 mutants and induces mass p53-dependent apoptosis of cancer cells in vitro and in xenografts [113]. As revealed by retrospective analysis of the results of testing a cancer cell panel for PRIMA-1 sensitivity, the cytotoxic effect of PRIMA-1 possibly depends on mutant p53 synthesis [114]. The mechanism of PRIMA-1 action is poorly understood. PRIMA-1 affects the existing inactive p53 molecules [113] and its effect is enhanced by the drugs that further increase the production of mutant p53 (adriamycin, cisplatin, and fludarabin) [115]. One of the PRIMA-1 targets is probably Hsp90, which acts as a chaperone to facilitate a refolding of the mutant protein into an active conformation [116]. PRIMA-1 activates caspases in anuclear cytoplasts in a p53-dependent manner, suggesting the capability of reactivating the transcription-independent functions of p53 in mitochondria [117].

MIRA-1, another compound obtained by screening a chemical library, is a maleimide derivative [118]. In contrast to PRIMA-1, MIRA-1 reactivates a narrower range of p53 mutants to induce apoptosis in cancer cells. Maleimide compounds can modify cysteine residues, suggesting similar mechanisms of action for MIRA-1 and WR1065. It is possible that these compounds affect the redox state of p53 and protect it from aggregation via disulfide bonding.

Activity Restoration in Wild-Type p53

One of the common mechanisms inhibiting p53 is based on a higher activity of mdm2, the main negative regulator of p53 [119]. MDM2 is amplified or expressed to a higher level in some tumors. As a result, p53 ubiquitination is accelerated, p53 is rapidly degraded, and p53 activity is lacking. Suppression of mdm2 activity seems promising for cancer therapy. However, inhibition of the ubiquitin ligase activity of mdm2 may increase the mdm2 level, because this activity is essential for autoubiquitination and degradation of mdm2. A more appealing approach is to suppress the mdm2 production, for instance, with antisense oligonucleotides. An experimental introduction of antisense oligonucleotides decreases the mdm2 level and induces p53-dependent apoptosis of cancer cells both in vitro and in vivo, in human tumors xenografted in athymic mice [120]. The effect of antisense nucleotides directed to MDM2 is substantially enhanced by simultaneous administration of chemotherapeutic drugs activating p53 [121]. Another possible approach to p53 activation is to suppress the mdm2-mediated export of p53 from the nucleus into the cytoplasm. In some tumors such as neuroblastoma and breast cancer, p53 often occurs in an inactive bound form in the cytoplasm [122]. Nitric oxide (NO) has been found to suppress the mdm2-mediated nuclear export, thus leading to an accumulation of active p53 in the nucleus and induction of apoptosis in neuroblastoma cells [123]. Thus, NO donors may be effective in the cases where p53 activity is low because of a defect in p53 translocation into the nucleus.

The most straightforward approach to reactivation of endogenous p53 is to prevent mdm2 from interacting with p53. Many attempts have been made to displace mdm2 from its complex with p53. Intracellular injection of monoclonal antibodies binding to mdm2 in the contact region appreciably activates p53 [124]. Peptide libraries have been screened for the peptides that bind to the hydrophobic pocked located on the mdm2 surface and directly involved in the interaction with p53. When such peptides were fused with the carrier thioredoxin and the resulting complex was introduced in cells, p53 was activated to a substantial extent [125]. These experiments have demonstrated the principal possibility of destroying the mdm2 complex with p53, allowing a search for small molecules exerting a similar effect [126].

Pharmacological destruction of protein complexes is far more difficult to achieve than enzyme inhibition, because extended surface regions of both proteins are usually involved in their interaction [127]. Screening of chemical libraries is inefficient: the probability to find a molecule with necessary structural parameters is negligibly low in random selection. A certain activity is characteristic of chalcone derivatives [128] and the compound RITA-1, which has been reported to bind to the N-terminal region of p53 to prevent its binding with mdm2 [129]. RITA-1 induces p53-dependent apoptosis of cancer cells with a comparatively low nonspecific toxic effect on normal cells.

It is possible more rationally approach the construction of a compound specifically inhibiting the mdm2-p53 interaction via a predetermined mechanism. Hoffman La Roche (Nutley, United States) has developed and tested imidazole derivatives designated nutlins [130]. These small molecules have been designed on the basis of X-ray data obtained for the mdm2-p53 complex and the shape of the mdm2 hydrophobic pocket accommodating the three p53 residues--Phe19, Trp23, and Leu26--that make the greatest contribution to the interaction of the two proteins. Exactly fitting the pocket, nutlins enter it and thereby destroy the complex. Incubation of various tumor cells with nutlins increases the p53 level, induces the p53 targets MDM2 and p21/Waf1, and leads to p53-dependent antiproliferative and proapoptotic effects. Cancer cells with amplified MDM2 are especially sensitive to nutlins, but a p53-dependent toxic effect has been observed for cancer cells with a normal mdm2 production as well [131]. Nutlins predominantly affect cancer cells and exert only a minor toxic effect on normal stromal cells [132]. This is explained by a higher sensitivity of cancer cells to p53 and a low nonspecific toxicity of nutlins. Nutlins exert some additional effects associated with p53 activation; for instance, they suppress the development of a vascular network in the tumor [133]. Nutlins are promising as antitumor drugs. In contrast to the available chemotherapeutic agents, nutlins induce p53 without damaging DNA, which prevents most adverse effects usually accompanying anticancer therapy.

REACTIVATION OF p53 IN CERVICAL CARCINOMA

Cervical carcinoma, along with a few cases of pharyngeal cancer, provides a rare example of human malignancies with a predominantly viral etiology. About 95% of all cervical carcinoma cases are associated with the presence of a HPV genome in tumor cells. The virus is HPV-16 in three-fourths of these cases, HPV-18 in the other one-fourth, and HPVs of other types in single cases. HPVs are sexually transmitted and cause chronic changes in the cervical epithelium, including its erosion. Although malignant transformation is rare, the wide spread of HPV infection determines a considerable incidence of cervical carcinoma, especially in developing countries. The small circular HPV genome harbors eight genes, of which two, E6 and E7, are involved in cell malignant transformation. The products of these genes serve to create the optimal conditions for viral DNA replication and further propagation of the virus. The genes stimulate cell division and alter differentiation of keratinocytes in the stratified epithelium of the cervix and pharynx. The viral proteins interact with numerous cell targets to modify their activity [134]. In particular, E7 binds and inactivates pRB, thus releasing the active transcription factor E2F1. As a result, the cell cycle arrest in G1 is overcome and the cell enters the S phase [135]. E6 binds with p53 and ubiquitin ligase E6AP, which leads to ubiquitination and rapid degradation of p53 [136]. This prevents the p53-dependent cell response to a regulatory defect in the pRB module, which normally leads to p53-dependent apoptosis.

Owing to these mechanisms, HPV can efficiently replicate its DNA in S-phase cells devoid of the p53-mediated control. The conditions are favorable not only for HPV but also for an accumulation of additional mutations in the cell genome, leading to malignant transformation of HPV-infected cells. The HPV genome remains in malignant cells, being often rearranged and integrated in a cell chromosome. HPV-harboring tumor cells usually lack defects in p53- and pRB-dependent pathways and, what is more, are free from selective pressure with respect to such defects, because the relevant mechanisms are already efficiently suppressed. Hence, it is possible to reactivate these mechanisms via eliminating the inhibitory effect of the viral proteins. Cell treatment with small interfering RNAs specific to the polycistronic HPV RNA coding for both E6 and E7 induces p53 accumulation, apoptosis [137], and an increase in sensitivity to chemotherapeutic agents [138]. A similar effect has been observed for antisense oligonucleotides [139] and ribozymes [140] designed to suppress the E6-E7 mRNA.

Inhibition of the viral proteins can be used in therapy of cervical carcinoma and other HPV-containing tumors. Several experimental protocols have been tested. Based on the fact that viral E2 blocks the promoter region of E6 and E7, bovine papillomavirus E2 was synthesized in HeLa cells and proved to reduce the level of the E6 and E7 mRNA and to arrest division [141]. Another approach involved E6 inhibition by peptide aptamers, short peptides selected from a random peptide library [142]. Production of such aptamers caused specific p53-dependent apoptosis in HPV-positive tumor cells. Small molecules can also be used to specifically inhibit viral proteins. In particular, the known antitumor drug Cidofovir suppresses synthesis of the HPV RNA to induce p53, p53-dependent apoptosis, and p53-dependent antitumor effect in nude mice [143]. We screened a chemical library and selected the small molecules that activate a p53-dependent transcriptional reporter construct introduced in HeLa cervical carcinoma cells. Some of these molecules efficiently suppressed viral transcripts, which was accompanied by p53 activation and induction of p53-dependent apoptosis [144]. However, it is unlikely that the small molecules exclusively suppress transcription of viral genes, since viral genome expression employs the cell transcription system. A more selective approach is to obtain the small molecules that inhibit viral E6 or E7. Since both of these proteins exert their effects via protein-protein binding, inhibitors should prevent the formation of such complexes or neutralize the consequences of their formation. One small molecule can hardly suppress both E6 and E7, because specific features of protein-protein interactions should be taken into account to degrade each particular complex. As in the case of agents inhibiting the mdm2-p53 interaction, a rational approach based on the structural details of a target complex is more expedient. Another important task is to select a proper target for small molecules. Selective inhibition of the E7-pRB complex can be expected to arrest division but can hardly induce cell death, because p53 inhibition by E6 is preserved. In contrast, selective inhibition of E6 can release p53 from its complexes and exert a potent antiapoptotic effect, because pRB inhibition is preserved to increase the level of E2F1, which stimulates the proapoptotic activity of p53 [145]. There is evidence that zinc-binding compounds can displace zinc from E6 and to prevent E6-E6AP complexation [146]. The compounds only slightly activate p53 and are unsuitable for testing in vivo because of the high toxicity. It seems that detail structural data on the E6-p53 complex are necessary to design its highly specific inhibitors. Such compounds will be most promising for therapy of HPV-positive malignancies.

CONCLUSION

As damage to p53-dependent signaling pathways proved to play the key role in cancer, p53 reactivation in tumors was suggested as a potentially efficient approach to anticancer therapy. Ample evidence supporting this idea has accumulated in the past decade. Yet many problems are to be solved before therapeutic p53 rescue is introduced in clinics. Hence, approaches to p53 therapy still have to pass a long way of research, preclinical studies, and limited clinical trials. The approach based on recombinant adenoviruses producing p53 in tumors is most promising and its limited use in combination with standard therapeutic exposures can be expected for the nearest future. It is of primary importance to identify the defect in each particular case, because the defect determines the choice of particular therapy. When the tumor preserves intact p53, the greatest effect is possibly achievable with p53-mdm2 interaction inhibitors, such as nutlins, which should be combined with standard treatments. When the tumor produces mutant p53, compounds rescuing some p53 activities are potentially applicable, but the adequate choice of the drug should be based on the type of the p53 mutation. As for HPV-associated tumors, E6-p53 interaction inhibitors, which seem optimal for therapy, have not been found as of yet. Such agents will be sought independently at least for HPV-16 and HPV-18, because E6 substantially differs between these two viruses.

Successful development of new-generation drugs aimed at p53 reactivation will supplement the range of chemotherapeutic agents and their combined use will improve the efficacy of therapy. It should be noted, however, that it is extremely difficult to eliminate all cancer cells from the body, especially in metastatic cancer. Cancer cells are highly variable and readily acquire drug resistance, and p53 therapy cannot radically prevent this process. Prevention of cancer also depends on the state of p53-dependent mechanisms; modulation of these mechanisms may substantially reduce the risk of cancer. Predisposition to cancer is determined by hereditary defects in many genes, including those controlling the p53 activity. For instance, 12% of people are homozygous and 40% are heterozygous for the MDM2 allele characterized by a higher expression level. Carriers of this allele have a lower level of p53 and an appreciably higher risk of cancer [147]. Chronic insufficiency of p53 weakens its preventive function, which includes, in particular, a decrease in the cell pool of reactive oxygen species and, consequently, reduces the rate of spontaneous mutations [148]. On the other hand, an increase in reactive oxygen species can be efficiently compensated for by alimentary antioxidants. Identification of the above allele in combination with high-dose antioxidants as food additives is an appealing means to prevent cancer.