Novel strategies for therapeutic design in molecular oncology using gene expression profiles

The sequence of the human genome is estimated to be available by the end of the year 2000 [1]. Pursuant to deciphering the genomic code, and the identification of the estimated 40,000 to 100,000 human genes, anticipated technological advances will make possible examination of global gene expression profiles. Despite the current inaccessibility to the entire genome, many fruitful gene expression profiling studies have been performed using less than 10% of the predicted suite of genes in human or mouse genomes. Even within the confines of this limited set of genes, many insights and discoveries have resulted and their applications to cancer research are particularly profound. This review will focus on recent applications of gene expression profiling that have benefited three major areas of research in molecular oncology: (i) discovery--applications which have found novel genes, families of genes, or pathways involved in cell growth deregulation and tumor development; (ii) diagnosis--applications that have refined, and in some cases, defined, diagnostic methodology; and (iii) therapeutic design--applications which hold potential for chemotherapeutic drug discovery.

KILLER/DR5, a novel DNA-damage inducible death receptor gene, links the p53-tumor suppressor to caspase activation and apoptotic death

TRAIL and its emerging receptors are the newest members of the TNF receptor super-family. The activation of TRAIL receptors by ligand binding leads to apoptosis through caspase activation through an as yet unclear signaling pathway that does not require the FADD adaptor. The TRAIL receptor KILLER/DR5, is induced by DNA damage and appears to be regulated by the tumor suppressor gene p53. Both the Fas receptor and KILLER/DR5 provide potential links between DNA damage-mediated activation of the p53 tumor suppressor and caspase activation. While further evaluation of the role of TRAIL receptors in human cancer is ongoing, initial studies suggest that both KILLER/DR5 and DR4 may be targets for inactivation and that these pro-apooptotic receptors may be tumor suppressor genes. Understanding the regulation of TRAIL and its receptors may thus be beneficial for the development of novel approaches for cancer treatment. TRAIL appears to be a cancer-specific cytotoxic agent and thus offers promise as a novel therapy for cancer either through replacement of the cytokine or potentially via gene replacement. Preliminary studies suggest the potential to combine TRAIL with classical cytotoxic chemotherapeutic drugs to achieve synergistic cell killing.

Molecular determinants of response to TRAIL in killing of normal and cancer cells

The tumor necrosis factor-related apoptosis-inducing ligand (TRAIL or Apo2L) is a potent inducer of death of cancer but not normal cells, which suggests its potential use as a tumor-specific antineoplastic agent. TRAIL binds to the proapoptotic death receptors DR4 and the p53-regulated proapoptotic KILLER/DR5 as well as to the decoy receptors TRID and TRUNDD. In the present studies, we identified a subgroup of TRAIL-resistant cancer cell lines characterized by low or absent basal DR4 or high expression of the caspase activation inhibitor FLIP. Four of five TRAIL-sensitive cell lines expressed high levels of DR4 mRNA and protein, whereas six of six TRAIL-resistant cell lines expressed low or undetectable levels of DR4 (chi 2; P < 0.01). FLIP expression appeared elevated in five of six (83%) TRAIL-resistant cell lines and only one of five (20%) TRAIL-sensitive cells (chi 2; P < 0.05). Two TRAIL-resistant lines that expressed DR4 contained an A-to-G alteration in the death domain encoding arginine instead of lysine at codon 441. The K441R polymorphism is present in 20% of the normal population and can inhibit DR4-mediated cell killing in a dominant-negative fashion. The expression level of KILLER/DR5, TRID, TRUNDD or TRID, and TRUNDD did not correlate with TRAIL sensitivity (P > 0.05). These results suggest that the major determinants for TRAIL sensitivity may be the expression level of DR4 and FLIP. TRAIL-resistant cells became susceptible to TRAIL-mediated apoptosis in the presence of doxorubicin. In TRAIL-sensitive cells, caspases 8, 9, and 3 were activated after TRAIL treatment, but in TRAIL-resistant cells, they were activated only by the combination of TRAIL and doxorubicin. Our results suggest: (a) evaluation of tumor DR4 and FLIP expression and host DR4 codon 441 status could be potentially useful predictors of TRAIL sensitivity, and (b) doxorubicin, in combination with TRAIL, may effectively promote caspase activation in TRAIL-resistant tumors.

Induction of the TRAIL receptor KILLER/DR5 in p53-dependent apoptosis but not growth arrest

The TRAIL death receptor KILLER/DR5 is induced by DNA damaging agents in wild-type p53-expressing cells. Here we show that, unlike the p53-target CDK-inhibitor p21WAF1/CIP1, the TRAIL death receptor KILLER/DR5 is only induced in cells undergoing p53-dependent apoptosis and not cell cycle arrest. Thus GM glioblastoma cells carrying an inducible MMTV-driven p53 gene undergo cell cycle arrest and upregulate p21 but not KILLER/DR5 expression upon dexamethasone exposure. WI38 normal lung fibroblasts undergoing cell cycle arrest in response to ionizing irradiation also induce p21 but not KILLER/DR5 gene expression. KILLER/DR5 upregulation is also deficient in irradiated lymphoblastoid cells derived from patients with Ataxia Teleangiectasia suggesting a role for the ATM-p53 pathway in regulating KILLER/DR5 expression after DNA damage. Inhibition of transcription by Actinomycin D blocks both KILLER/DR5 and p21 induction in cells undergoing p53-dependent apoptosis. Our results suggest that the p53-dependent transcriptional induction of KILLER/DR5 death receptor is restricted to cells undergoing apoptosis and not cells undergoing exclusively p53-dependent G1 arrest.

BRCA1 signals ARF-dependent stabilization and coactivation of p53

The hereditary breast and ovarian tumor suppressor BRCA1 can activate p53-dependent gene expression. We show here that BRCA1 increases p53 protein levels through a post-transcriptional mechanism. BRCA1-stabilized p53 has increased sequence-specific DNA-binding and transcriptional activity. BRCA1 does not stabilize p53 in p14ARF-deficient cells. A deletion mutant of BRCA1 which inhibits p53-dependent transcription confers resistance to topoisomerase II-targeted chemotherapy. Our results suggest that BRCA1 may trigger the p53 pathway through two potentially separate mechanisms: accumulation of p53 through a direct or indirect induction of p14ARF as well as direct transcriptional coactivation of p53. BRCA1 may also enhance chemosensitivity and repair of DNA damage through binding to and coactivation of p53.

Regulation of p53 downstream genes

The p53 tumor suppressor is the most commonly mutated gene in human cancer. p53 protein is stabilized in response to different checkpoints activated by DNA damage, hypoxia, viral infection, or oncogene activation resulting in diverse biological effects, such as cell cycle arrest, apoptosis, senescence, differentiation, and antiangiogenesis. The stable p53 protein is activated by phosphorylation, dephosphorylation and acetylation yielding a potent sequence-specific DNA-binding transcription factor. The wide range of p53's biological effects can in part be explained by its activation of expression of a number of target genes including p21WAFI, GADD45, 14-3-3 sigma, bax, Fas/APO1, KILLER/DR5, PIG3, Tsp1, IGF-BP3 and others. This review will focus on the transcriptional targets of p53, their regulation by p53, and their relative importance in carrying out the biological effects of p53.

p21waf1/cip1 and transforming growth factor beta 1 protein expression correlate with survival in non-small cell lung cancer

p21waf1/cip1 encodes a cyclin-dependent kinase inhibitor that is transcriptionally activated by the p53 tumor suppressor gene, transforming growth factor beta 1 (TGF-beta 1), AP2, and other pathways. Because p21waf1/cip1, p53, and TGF-beta 1 all regulate apoptosis and the cell cycle, we tested the hypothesis that their relative protein levels would correlate with biological features including the survival of non-small cell lung cancer (NSCLC) patients. We conducted an immunohistochemical analysis of p21waf1/cip1 and TGF-beta 1 and identified four patient groups with distinct survival outcomes. Concordant p21waf1/cip1 and TGF-beta 1 expression (i.e., either high p21waf1/cip1 and high TGF-beta 1 expression or low p21waf1/cip1 and low TGF-beta 1 expression) predicted 70% disease-free survival at 2000 days of follow-up. Discordant p21waf1/cip1 and TGF-beta 1 expression (i.e., either high p21waf1/cip1 and low TGF-beta 1 expression or low p21waf1/cip1 and high TGF-beta 1 expression) predicted 35% disease-free survival (P = 0.0003; log-rank test). These survival relationships were not attributable to differences in grade, stage, or p53 status. Although current models do not fully explain these complex interactions, most of these data fit a paradigm whereby TGF-beta 1 regulation determines NSCLC survival. In addition to the survival correlation, we found that high p21waf1/cip1 protein expression correlated with high tumor grade (P = 0.014). There is little evidence that p21waf1/cip1 protein levels accurately predict p53 mutation status in NSCLC; specifically, 20 of 48 (42%) tumors with p53 mutations contained high levels of p21waf1/cip1 protein. These findings indicate that p21waf1/cip1 immunohistochemical analysis may provide useful information concerning the biological properties of NSCLC.

p53-dependent and -independent regulation of the death receptor KILLER/DR5 gene expression in response to genotoxic stress and tumor necrosis factor alpha

The death receptor (DR) KILLER/DR5 gene has recently been identified as a doxorubicin-regulated transcript that was also induced by exogenous wild-type p53 in p53-negative cells. KILLER/DR5 gene encodes a DR containing cell surface protein that is highly homologous to DR4, another DR of the tumor necrosis factor (TNF) receptor family. Both DR4 and KILLER/DR5 independently bind to their specific ligand TRAIL and engage the caspase cascade to induce apoptosis. TRID (also known as TRAIL-R3) is an antiapoptotic decoy receptor that lacks the cytoplasmic death domain and competes with KILLER/DR5 and DR4 for binding to TRAIL. In this study, we demonstrate that the DR KILLER/DR5 gene is regulated in a p53-dependent and -independent manner during genotoxic and nongenotoxic stress-induced apoptosis. Just like other p53-regulated genes, ionizing radiation induction of KILLER/DR5 occurs in p53 wild-type cells, whereas methyl methanesulfonate regulation of KILLER/DR5 occurs in a p53-dependent and -independent manner. However, unlike other p53-regulated genes, KILLER/DR5 is not regulated following UV irradiation. TNF-alpha, a nongenotoxic cytokine, also induced the expression of KILLER/DR5 in a number of cancer cell lines, irrespective of p53 status. TNF-alpha did not alter the KILLER/DR5 mRNA stability, suggesting that the TNF-alpha regulation of KILLER/DRS expression appears transcriptional. We also provide evidence that KILLER/DR5 is regulated in a trigger and cell type-specific manner and that its induction by TNF-alpha, p53, or DNA damage is not the consequence of apoptosis induced by these agents. Unlike KILLER/DR5, none of the other KILLER/DR5 family members, including DR4, TRID, or the ligand TRAIL, displayed genotoxic stress or TNF-alpha regulation in a p53 transcription-dependent manner. Thus, KILLER/DR5 appears a bona fide downstream target of p53 that is also regulated in a cell type-specific, trigger-dependent, and p53-independent manner.

Identification of a novel AP-2 consensus DNA binding site

Activator Protein (AP)-2 is a transcription factor that is required for mouse development. AP-2 activates expression of positive and negative growth regulators including erbB-2 and p21 WAF1/CIP1. Induction of p21 has been correlated with cell cycle and growth inhibition of human cancer cells. Because several endogenous AP-2 binding sites do not fit the known consensus sequences well, we sought to define AP-2's interaction with DNA more precisely. Using Cyclic Amplification and Selection of Targets (CAST'ing) of random oligonucleotide sequences and recombinant human AP-2 protein, we identified 17 novel AP-2 binding sites. Mobility shift assays showed significant AP-2 binding of the novel sites as compared to p21, erbB-2 and hMtIIa sites. Several sites that bound with high specificity and affinity did not fit known AP-2 consensus sequences. A sequence comparison based on several of the novel sequences yielded a putative consensus binding sequence of 5'-TAGAAAGNYCYNG-3'. These DNA binding sites may help identify novel targets of AP-2 and aid in further understanding AP-2 function.

Bypass of abnormal MDM2 inhibition of p53-dependent growth suppression

Oncoprotein MDM2 inhibits p53-dependent cell cycle arrest and apoptosis. MDM2-overexpressing human cancer cell lines (n = 3) were found to be resistant to growth inhibition after infection by p53-expressing adenovirus (Ad-p53), as compared to low MDM2-expressing tumors (n = 3), in vitro. The growth of MDM2-overexpressing tumors, however, was inhibited by p21-expressing adenovirus (Ad-p21) infection, and the cyclin-dependent kinase-inhibitory region of p21 was sufficient to bypass the MDM2-p53 feedback loop. The phosphorylation state of Rb correlated with the response to either p53 or p21 gene therapy. MDM2-overexpressing cancer cells infected by Ad-p21 also developed a quiescent large-cell morphology. The results suggest that MDM2-mediated resistance to p53 may be bypassed by p21 and that the Rb phosphorylation state may predict the effects on growth after Ad-p53 or Ad-p21 infection.

Decreased immunoglobulin deposition in tumors and increased immature B cells in p53-null mice

Recent studies have hinted that there may be a relationship between p53 and the immune response. In preliminary experiments, we found significantly decreased levels of immunoglobulin deposition in 13 of 16 p53-null tumors compared with 2 of 17 tumors derived from p53 +/- mice. We further explored the effect of p53 on B-cell development and function. p53-null mice contained more splenic white pulp and more immature B cells in the bone marrow compared with p53 +/- mice. p53-null B cells were hyperresponsive to proliferative challenge but were not more resistant to signal-induced apoptosis. Several p53 DNA-binding sites were localized to the regulatory regions of immunoglobulin heavy and light chain genes, including the KII site, which serves as an enhancer for rearrangement of the mouse kappa chain J cluster genes. Levels of p53 protein and the kappa chain sterile transcript increased after exposure of pre-B cells to the DNA damaging agents etoposide and Adriamycin. Our observations suggest that p53 may be involved in B-cell maturation and may relay certain stress signals to affect B-cell function.

AP2 inhibits cancer cell growth and activates p21WAF1/CIP1 expression

The 52-kD Activator Protein (AP2) is a DNA-binding transcription factor implicated in signalling terminal differentiation. Profound developmental abnormalities have been recently observed in AP2-null mice. The molecular events by which AP2 promotes differentiation or development are, however, unknown. Increased expression of the universal cell cycle inhibitor p21WAF1/CIP1 occurs in growth-arrested terminally differentiating cells. In a search for cellular factors that could activate p21 during phorbol ester (TPA)-induced differentiation, we identified AP2 as a regulator of p21 expression. Mutagenesis of an AP2 DNA-binding site within a p21 promoter-luciferase reporter inhibited its activation by either AP2 transfection or TPA stimulation. Endogenous p21 protein levels were elevated and DNA synthesis was inhibited in AP2 versus control vector-transfected cells. Overexpression of AP2 in HepG2 human hepatoblastoma and SW480 human colon adenocarcinoma cells inhibited cell division and stable colony formation. These results link the differentiation-associated factor AP2 to negative cell cycle and growth control, possibly through p21 activation.

Raf-1/bcl-2 phosphorylation: a step from microtubule damage to cell death

Recent studies have shown that paclitaxel leads to activation of Raf-1 kinase and have suggested that this activation is essential for bcl-2 phosphorylation and apoptosis. In the present study, we demonstrate that, in addition to paclitaxel, other agents that interact with tubulin and microtubules also induce Raf-1/bcl-2 phosphorylation, whereas DNA-damaging drugs, antimetabolites, and alkylating agents do not. Activation of Raf-1 kinase by paclitaxel is linked to tubulin polymerization; the effect is blunted in paclitaxel-resistant cells, the tubulin of which does not polymerize following the addition of paclitaxel. In contrast, vincristine and vinblastine, drugs to which the paclitaxel-resistant cells retain sensitivity were able to bring about Raf-1 phosphorylation. The requirement for disruption of microtubules in this signaling cascade was strengthened further using paclitaxel analogues by demonstrating a correlation between tubulin polymerization, Raf-1/bcl-2 phosphorylation, and cytotoxicity. Inhibition of RNA or protein synthesis prevents Raf-1 activation and bcl-2 phosphorylation, suggesting that an intermediate protein(s) acts upstream of Raf-1 in this microtubule damage-activating pathway. A model is proposed that envisions a pathway of Raf-1 activation and bcl-2 phosphorylation following disruption of microtubular architecture, serving a role similar to p53 induction following DNA damage.

Role of oncogenes in resistance and killing by cancer therapeutic agents

Chemotherapeutic drug resistance is a major clinical problem and cause for failure in the therapy of human cancer. One of the goals of molecular oncology is to identify the underlying mechanisms, with the hope that more effective therapies can be developed. Several mechanisms have been suggested to contribute to chemoresistance: 1) amplification or overexpression of the P-glycoprotein family of membrane transporters (eg, MDR1, MRP, LRP) which decrease the intracellular accumulation of chemotherapy; 2) changes in cellular proteins involved in detoxification (eg, glutathione S-transferase pi, metallothioneins, human MutT homologue, bleomycin hydrolase, dihydrofolate reductase) or activation of the chemotherapeutic drugs (DT-diaphorase, nicotinamide adenine dinucleotide phosphate:cytochrome P-450 reductase); 3) changes in molecules involved in DNA repair (eg, O6-methylguanine-DNA methyltransferase, DNA topoisomerase II, hMLH1, p21WAF1/CIP1; 4) activation of oncogenes such as Her-2/neu, bcl-2, bcl-XL, c-myc, ras, c-jun, c-fos, MDM2, p210 BCR-abl, or mutant p53. An overview of these resistance mechanisms is presented, with a particular focus on the role of oncogenes. Some current strategies attempting to reverse their effects are discussed.

Proteasome-dependent regulation of p21WAF1/CIP1 expression

Proteasome-dependent degradation of regulatory proteins is a known mechanism of cell cycle control. We found that the proteasome-specific inhibitor lactacystin (LC) induced expression of the cell cycle inhibitor p21WAF1/CIP1 in human cancer cells regardless of their p53 status. Both wild-type (wt) p53 and p21 protein levels increased by two hours in wt p53 containing cells, whereas mutant (mt) p53 levels decreased and the increase in p21 levels was delayed to 6 hr following inhibition of proteolysis by LC in mt p53 expressing cells. We found that wt but not mt p53 expressing cells increased p21 mRNA and p21-promoter reporter levels following LC exposure, suggesting transcriptional induction of p21. Inhibition of protein synthesis by cycloheximide demonstrated increased p21 protein half-life in the presence of LC in mutant p53 containing cells. p21 induction was correlated with the cytostatic effects of LC. The results suggest that p21 protein expression could be increased by transcriptional mechanisms as well as inhibition of proteolysis by LC.

Expression of p21waf1/Cip1, MDM2 and p53 in vivo: analysis of cytological preparations

The aim of this study was to test the hypothesis that expression of p21waf1/Cip1 and MDM2 could be used as indicators of the activity of wild-type p53 which can transcriptionally activate the p21waf1/Cip1 and mdm2 genes. In cytological preparations of serous fluids, the expression of p53, p21waf1/Cip1 and MDM2 protein was assessed by immunohistochemistry. A series of 50 cases was assessed for both p53 and p21waf1/Cip1 expression and a subset of 37 cases had sufficient material for analysis of MDM2. In samples in which there were reactive mesothelial cells (n = 48) there was general concordance between p53 and p21waf1/Cip1 expression, but in nine cases p21waf1/Cip1 was expressed in the absence of detectable p53. Similarly, MDM2 expression was not correlated with p53 in 15 of 31 cases. p21waf1/Cip1 was correlated with MDM2 in 24 of 31 cases, while in the remaining seven, MDM2 was expressed without detectable p21waf1/Cip1 immunoreactivity. In samples with neoplastic cells (n = 18) the presence of p21waf1/Cip1 and MDM2 expression was always associated with p53 expression. Polymorphonuclear leucocytes frequently showed p21waf1/Cip1 immunoreactivity, and this was confirmed by immunoblotting of peripheral blood polymorphonuclear leucocytes. These data indicate that in general p21waf1/Cip1 expression correlates with p53 expression in reactive mesothelial cells, consistent with a known mechanism of regulation. However, in reactive mesothelial cells, MDM2 expression is perhaps dissociated from p53 expression, contrary to current models of MDM2 regulation. Finally, in addition to many normal tissues, it is likely that in reactive mesothelial cells and some tumours p21waf1/Cip1 expression is not dependent on the presence of wild-type p53 protein. In conclusion, p53 status cannot be reliably predicted based only on p21waf1/Cip1 or MDM2 expression.

p21 (WAF1/CIP1) expression is induced in newly nondividing cells in diverse epithelia and during differentiation of the Caco-2 intestinal cell line

We examined the relationship between expression of the p21 (WAF1/CIP1) inhibitor of cyclin-dependent kinases, cessation of proliferation, and terminal differentiation in the epithelia of the gastrointestinal tract. Using in situ hybridization, we performed a detailed study of patterns of p21 mRNA expression in different regions of the stomach, along the length of the intestine, and in tongue, cervix, and hair follicle. We detected strong hybridization only in cells that had ceased proliferation and begun the process of terminal differentiation. Induction of p21 transcription may serve as a useful marker for dissection of differentiation programs in these diverse epithelia. To determine the relative levels of p21 expressed in various regions of the gastrointestinal tract from the esophagus to the colon, we used quantitative RT-PCR with endogenous and exogenous sequences as internal standards. The highest levels of p21 expression were detected in the distal small intestine. To further investigate the role that cell cycle regulation may play during differentiation of intestinal epithelial cells, we examined the expression of p53, p21, cyclin D1, cyclin E, and E2F1 in the Caco-2 colon carcinoma cell line, which differentiates spontaneously after reaching confluence. p21 and p53 mRNA and protein levels increase as Caco-2 cells differentiate. In both undifferentiated and differentiated Caco-2 cells, p53 protein was not inducible by DNA damaging agents, suggesting the absence of functionally wildtype protein. Caco-2 cells should provide a useful model system for studying regulation of p21 and determining if it plays a role during intestinal epithelial cell differentiation.

In vitro evaluation of a p53-expressing adenovirus as an anti-cancer drug

Deficiency in p53-mediated cell death is common in human cancer, contributing to both tumorigenesis and chemoresistance. In an attempt to restore p53, we evaluated in vitro infectivity and cytotoxicity of a wild type (w.t.) p53-expressing adenovirus (Ad-p53) toward a panel of human cancer cell lines (n = 19). At a multiplicity of infection of 30, both Ad-p53 and adenovirus expressing beta-galactosidase (Ad-LacZ) infected greater than 99% of cells derived from brain, lung, breast, ovarian, colon, and prostate cancer, but failed to infect leukemia or lymphoma cells. Ad-p53, but not Ad-LacZ, infection of cancer cells was followed by nuclear accumulation of the CDK inhibitor p21WAFI/CIPI, cell cycle arrest and loss of viability. Ad-p53 induced apoptotic death in cancer cells that express mutant p53, including multi-drug resistant cells, but fewer deaths were observed in some w.t. p53 expressing cells. Ad-p53-infected SKBr3 breast cancer cells were more sensitive to cytotoxicity of the DNA damaging drugs mitomycin C or Adriamycin, but not the M-phase specific drug vincristine. Our results suggest that Ad-p53 is capable of infecting and killing cancer cells of diverse tissue origins (including multi-drug resistant cancer cells), that p21WAFI/CIPI may be a useful marker of p53 infectivity and that there may be synergy between Ad-p53 and either mitomycin C or Adriamycin induced cell death in tumors with p53 mutations.

In vitro evaluation of a p53-expressing adenovirus as an anti-cancer drug

Deficiency in p53-mediated cell death is common in human cancer, contributing to both tumorigenesis and chemoresistance. In an attempt to restore p53, we evaluated in vitro infectivity and cytotoxicity of a wild type (w.t.) p53-expressing adenovirus (Ad-p53) toward a panel of human cancer cell lines (n = 19). At a multiplicity of infection of 30, both Ad-p53 and adenovirus expressing beta-galactosidase (Ad-LacZ) infected greater than 99% of cells derived from brain, lung, breast, ovarian, colon, and prostate cancer, but failed to infect leukemia or lymphoma cells. Ad-p53, but not Ad-LacZ, infection of cancer cells was followed by nuclear accumulation of the CDK inhibitor p21WAFI/CIPI, cell cycle arrest and loss of viability. Ad-p53 induced apoptotic death in cancer cells that express mutant p53, including multi-drug resistant cells, but fewer deaths were observed in some w.t. p53 expressing cells. Ad-p53-infected SKBr3 breast cancer cells were more sensitive to cytotoxicity of the DNA damaging drugs mitomycin C or Adriamycin, but not the M-phase specific drug vincristine. Our results suggest that Ad-p53 is capable of infecting and killing cancer cells of diverse tissue origins (including multi-drug resistant cancer cells), that p21WAFI/CIPI may be a useful marker of p53 infectivity and that there may be synergy between Ad-p53 and either mitomycin C or Adriamycin induced cell death in tumors with p53 mutations.

Regulation of p21WAF1/CIP1 expression by p53-independent pathways

The CDK-inhibitor p21WAF1/CIP1 has been implicated as a growth arrest mediator in p53-tumour suppression, cellular senescence and terminal differentiation. Cell type specific differences in p53-independent p21 expression and cell cycle arrest were found following treatment of human tumour cell lines with serum, 12-O-tetradecanoyl-phorbol-13-acetate (TPA), or okadaic acid (OA). TPA induced p21 in ML1, K562 and HL60 leukemia cells, whereas OA induced p21 in SW480 and GM4723 carcinoma cells as well as in leukemic cells. In addition, TPA- and serum- but not OA-induced cell cycle arrest was reversed upon return of p21 to basal levels. To further investigate the mechanisms underlying p53-independent regulation of p21, the transcription inhibitor, Actinomycin D (AMD), was used to block p21 expression. The results showed a complete inhibition of p21 mRNA and protein induction by TPA or adriamycin but little effect on p21 mRNA induced by OA in the presence of AMD. These results suggested that TPA-induced p21 expression requires transcription initiation, while a post-transcriptional mechanism may be involved in OA-induction as well. Transient transfection assays with p21 promoter-luciferase reporters and TPA or OA treatment further confirmed that TPA, and to a lesser extent, OA, initiated transcription of p21. Finally, the protein kinase C inhibitor, staurosporine, was found to interfere with p21 induction and prevent cell cycle arrest following treatment with TPA but not OA, suggesting a requirement for PKC in TPA activation of p21 expression.

BCR-ABL-mediated inhibition of apoptosis with delay of G2/M transition after DNA damage: a mechanism of resistance to multiple anticancer agents

A critical determinant of the efficacy of antineoplastic therapy is the response of malignant cells to DNA damage induced by anticancer agents. The p53 tumor-suppressor gene is a critical component of two distinct cellular responses to DNA damage, the induction of a reversible arrest at the G1/S cell cycle checkpoint, and the activation of apoptosis, a genetic program of autonomous cell death. Expression of the BCR-ABL chimeric gene produced by a balanced translocation in chronic myeloid leukemia, confers resistance to multiple genotoxic anticancer agents. BCR-ABL expression inhibits the apoptotic response to DNA damage without altering either the p53-dependent WAF1/CIP1-mediated G1 arrest or DNA repair. BCR-ABL-mediated inhibition of DNA damage-induced apoptosis is associated with a prolongation of cell cycle arrest at the G2/M restriction point; the delay of G2/M transition may allow time to repair and complete DNA replication and chromosomal segregation, thereby preventing a mitotic catastrophe. The inherent resistance of human cancers to genotoxic agents may result not only by the loss or inactivation of the wild-type p53 gene, but also by genetic alterations such as BCR-ABL that can delay G2/M transition after DNA damage.

Topological control of p21WAF1/CIP1 expression in normal and neoplastic tissues

The p53-regulated gene product p21WAF1/CIP1 is the prototype of a family of small proteins that negatively regulate the cell cycle. To learn more about p21WAF1/CIP1 regulation in vivo, monoclonal antibodies were developed for immunohistochemistry. These revealed that p21WAF1/CIP1 expression followed radiation-induced DNA damage in human skin in a pattern consistent with its regulation by p53. A detailed comparison of the human, rat, and mouse p21WAF1/CIP1 promoter sequences revealed that this induction was probably mediated by conserved p53-binding sites upstream of the transcription start site. In unirradiated tissues, p21WAF1/CIP1 expression was apparently independent of p53 and was observed in a variety of cell types. Moreover, there was a striking compartmentalization of p21WAF1/CIP1 expression throughout the gastrointestinal tract that correlated with proliferation rather than differentiation. As epithelial cells migrated up the crypts, the Ki67-expressing proliferating compartment near the crypt base ended abruptly, with the coincident appearance of a nonproliferating compartment expressing p21WAF1/CIP1. In colonic neoplasms, this distinct compartmentalization was largely abrogated. Cell cycle inhibitors are thus subject to precise topological control, and escape from this regulation may be a critical feature of neoplastic transformation.

Absence of WAF1 mutations in a variety of human malignancies

A newly cloned gene named wild-type p53-activated fragment 1 (WAF1; also known as p21, Pic-1, Cip-1, or SDI1) is directly regulated by p53 and can itself suppress tumor cell growth in culture. Induction of expression of WAF1 may be an important means by which cells with DNA injury arrest their growth to repair DNA or undergo apoptosis. Based on the hypothesis that mutations of this gene may play a role in carcinogenesis, we have studied 351 DNAs from 14 kinds of malignancies, as well as 36 human transformed cell lines, for alterations of WAF1 gene by single-strand conformation polymorphism analysis of polymerase chain reaction amplification of the DNA coding region of the WAF1 gene. No abnormal band shifts of WAF1 were noted in any of the samples or cell lines, but three major variants in exons 2 and 3 of the gene were found that are consistent with the existence of two different DNA polymorphisms. Sequence analysis of the amplified products producing these three variants in each exon from normal DNAs confirmed the presence of the polymorphisms in the WAF1 gene. Of 290 selected tumor samples previously evaluated for p53 mutations by single-strand conformation polymorphism, 90% had no detectable p53 alterations. In summary, mutations within the coding portion of the WAF1 gene were undetectable in a large series of human tumors, many of which had a normal p53 gene. This suggests that WAF1 alterations are generally caused indirectly, through p53 mutations rather than through intragenic mutation of the WAF1 itself.

p53 gene mutations are associated with decreased sensitivity of human lymphoma cells to DNA damaging agents

The present study assessed the role of the p53 tumor suppressor gene in cell cycle arrest and apoptosis following treatment of Burkitt's lymphoma and lymphoblastoid cell lines with gamma-rays, etoposide, nitrogen mustard, and cisplatin. Cell cycle arrest was measured by flow cytometry; p53 and p21Waf1/Cip1 protein levels were measured by Western blotting; cell survival was measured in 72-96-h growth inhibition assays and by trypan blue staining, and apoptotic DNA fragmentation was assessed by either agarose gel electrophoresis or a modified filter elution method. We found that gamma-rays and etoposide induced a strong G1 arrest in the wild-type p53 lines while nitrogen mustard and cisplatin induced relatively little G1 arrest. All agents failed to induce G1 arrest in cells containing mutant p53 genes. The degree of G1 arrest observed with these agents correlated with the rate of p53 and p21Waf1/Cip1 protein accumulation: gamma-rays and etoposide induced rapid accumulation of both p53 and p21Waf1/Cip1; nitrogen mustard and cisplatin induced slow accumulation of p53 and no major accumulation of the p21Waf1/Cip1 protein. Despite differences in G1 arrest and kinetics of p53 or p21Waf1/Cip1 protein accumulation, all agents tended to decrease survival to a greater extent in the wild-type p53 lines compared to the mutant p53 lines. Cell death in the wild-type p53 lines was associated with intracellular DNA degradation into oligonucleosomal sized DNA fragments, indicative of apoptosis. We also observed an inverse sensitivity relationship between nitrogen mustard/cisplatin and etoposide in the mutant p53 lines and this was found to correlate with topoisomerase II mRNA levels in the cells. Our results suggest that p53 gene status is an important determinant of both radio- and chemosensitivity in lymphoid cell lines and that p53 mutations are often associated with decreased sensitivity to DNA damaging agents.

p53 tagged sites from human genomic DNA

The product of the tumor suppressor gene p53 binds to DNA and activates transcription from promoters containing its consensus binding site. This activity has been hypothesized to be responsible for its biological effects. However, the total number and nature of human genomic sites with which p53 can functionally interact is unknown. In this paper, we have used a Saccharomyces cerevisiae-based screen to identify human genomic sequences that activate transcription from an adjacent reporter gene in a p53-dependent manner (p53-tagged sites, PTS). Fifty-seven different PTS were identified, and the total number of such sites in the human genome was predicted to be between 200 and 300. Almost all contained two adjacent copies of the previously defined consensus 5'-PuPuPuC(A/T)(T/A)GPyPyPy-3'. Spacing between the copies was found to be critical for sequence-specific transcriptional activation in vivo. These results further refine the nature of the genomic sequences likely to be most important for p53-mediated tumor suppression.

Sequence-specific transcriptional activation is essential for growth suppression by p53

Although several biochemical features of p53 have been described, their relationship to tumor suppression remains uncertain. We have compared the ability of p53-derived proteins to act as sequence-specific transcriptional (SST) activators with their ability to suppress tumor cell growth, using an improved growth-suppression assay. Both naturally occurring and in vitro derived mutations that abrogated the SST activity of p53 lost the ability to suppress tumor cell growth. Additionally, the N- and C-terminal ends of p53 were shown to be functionally replaceable with foreign transactivation and dimerization domains, respectively, with concordant preservation of both SST and tumor-suppressive properties. Only the central region of p53, conferring specific DNA binding, was required to suppress growth by such hybrid proteins. The SST activity of p53 thus appeared to be essential for the protein to function as a tumor suppressor.

WAF1/CIP1 is induced in p53-mediated G1 arrest and apoptosis

The tumor growth suppressor WAF1/CIP1 was recently shown to be induced by p53 and to be a potent inhibitor of cyclin-dependent kinases. In the present studies, we sought to determine the relationship between the expression of WAF1/CIP1 and endogenous regulation of p53 function. WAF1/CIP1 protein was first localized to the nucleus of cells containing wild-type p53 and undergoing G1 arrest. WAF1/CIP1 was induced in wild-type p53-containing cells by exposure to DNA damaging agents, but not in mutant p53-containing cells. The induction of WAF1/CIP1 protein occurred in cells undergoing either p53-associated G1 arrest or apoptosis but not in cells induced to arrest in G1 or to undergo apoptosis through p53-independent mechanisms. DNA damage led to increased levels of WAF1/CIP1 in cyclin E-containing complexes and to an associated decrease in cyclin-dependent kinase activity. These results support the idea that WAF1/CIP1 is a critical downstream effector in the p53-specific pathway of growth control in mammalian cells.

WAF1, a potential mediator of p53 tumor suppression

The ability of p53 to activate transcription from specific sequences suggests that genes induced by p53 may mediate its biological role as a tumor suppressor. Using a subtractive hybridization approach, we identified a gene, named WAF1, whose induction was associated with wild-type but not mutant p53 gene expression in a human brain tumor cell line. The WAF1 gene was localized to chromosome 6p21.2, and its sequence, structure, and activation by p53 was conserved in rodents. Introduction of WAF1 cDNA suppressed the growth of human brain, lung, and colon tumor cells in culture. Using a yeast enhancer trap, a p53-binding site was identified 2.4 kb upstream of WAF1 coding sequences. The WAF1 promoter, including this p53-binding site, conferred p53-dependent inducibility upon a heterologous reporter gene. These studies define a gene whose expression is directly induced by p53 and that could be an important mediator of p53-dependent tumor growth suppression.

A mammalian cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in ataxia-telangiectasia

Cell cycle checkpoints can enhance cell survival and limit mutagenic events following DNA damage. Primary murine fibroblasts became deficient in a G1 checkpoint activated by ionizing radiation (IR) when both wild-type p53 alleles were disrupted. In addition, cells from patients with the radiosensitive, cancer-prone disease ataxia-telangiectasia (AT) lacked the IR-induced increase in p53 protein levels seen in normal cells. Finally, IR induction of the human GADD45 gene, an induction that is also defective in AT cells, was dependent on wild-type p53 function. Wild-type but not mutant p53 bound strongly to a conserved element in the GADD45 gene, and a p53-containing nuclear factor, which bound this element, was detected in extracts from irradiated cells. Thus, we identified three participants (AT gene(s), p53, and GADD45) in a signal transduction pathway that controls cell cycle arrest following DNA damage; abnormalities in this pathway probably contribute to tumor development.

Definition of a consensus binding site for p53

Recent experiments have suggested that p53 action may be mediated through its interaction with DNA. We have now identified 18 human genomic clones that bind to p53 in vitro. Precise mapping of the binding sequences within these clones revealed a consensus binding site with a striking internal symmetry, consisting of two copies of the 10 base pair motif 5'-PuPuPuC(A/T)(T/A)GPyPyPy-3' separated by 0-13 base pairs. One copy of the motif was insufficient for binding, and subtle alterations of the motif, even when present in multiple copies, resulted in loss of affinity for p53. Mutants of p53, representing each of the four "hot spots" frequently altered in human cancers, failed to bind to the consensus dimer. These results define the DNA sequence elements with which p53 interacts in vitro and which may be important for p53 action in vivo.

High expression of the DNA methyltransferase gene characterizes human neoplastic cells and progression stages of colon cancer

DNA methylation abnormalities occur consistently in human neoplasia including widespread hypomethylation and more recently recognized local increases in DNA methylation that hold potential for gene inactivation events. To study this imbalance further, we have cloned and localized to chromosome 19 a portion of the human DNA methyltransferase gene that codes for the enzyme catalyzing DNA methylation. Expression of this gene is low in normal human cells, significantly increased (30- to 50-fold by PCR analysis) in virally transformed cells, and strikingly elevated in human cancer cells (several hundredfold). In comparison to colon mucosa from patients without neoplasia, median levels of DNA methyltransferase transcripts are 15-fold increased in histologically normal mucosa from patients with cancers or the benign polyps that can precede cancers, 60-fold increased in the premalignant polyps, and greater than 200-fold increased in the cancers. Thus, increases in DNA methyltransferase gene expression precede development of colonic neoplasia and continue during progression of colonic neoplasms. These increases may play a role in the genetic instability of cancer and mark early events in cell transformation.

Mechanisms of error discrimination by Escherichia coli DNA polymerase I

The mechanism of base selection by DNA polymerase I of Escherichia coli has been investigated by kinetic analysis. The apparent KM for the insertion of the complementary nucleotide dATP into the hook polymer poly(dT)-oligo(dA) was found to be 6-fold lower than that for the noncomplementary nucleotide dGTP, whereas the Vmax for insertion of dATP was 1600-fold higher than that for dGTP. The ratio of Kcat/KM values for complementary and mismatched nucleotides of 10(4) demonstrates the extremely high specificity of base selection by DNA polymerase I and is in agreement with results obtained with a different template-primer, poly(dC)-oligo(dG) [El-Deiry, W. S., Downey, K. M., & So, A. G. (1984) Proc. Natl. Acad. Sci. U.S.A. 81, 7378]. Studies on the effects of phosphate ion on the polymerase and 3'- to 5'-exonuclease activities of DNA polymerase I showed that, whereas the polymerase activity was somewhat stimulated by phosphate, the exonuclease activity was markedly inhibited, being 50% inhibited at 25 mM phosphate and greater than 90% inhibited at 80 mM phosphate. Selective inhibition of the exonuclease activity by phosphate also resulted in inhibition of template-dependent conversion of a noncomplementary dNTP to dNMP and, consequently, markedly affected the kinetic constants for insertion of noncomplementary nucleotides. The mutagenic metal ion Mn2+ was found to affect error discrimination by both the polymerase and 3'- and 5'-exonuclease activities of DNA polymerase I.(ABSTRACT TRUNCATED AT 250 WORDS)