|Home | About | Journals | Submit | Contact Us | Français|
The p53 protein plays a critical role in inducing cell cycle arrest or apoptosis. Because p53 is inactivated in human gliomas, restoring p53 function is a major focus of glioma therapy. The most clinically tested strategy for replacing p53 has been adenoviral-mediated p53 gene therapy (Ad-p53). In addition to their therapeutic implications, investigations into Ad-p53 provide model systems for understanding p53’s ability to induce cell cycle arrest versus apoptosis, particularly because wild-type p53 cells are resistant to Ad-p53 – induced apoptosis. Here we use Ad-p53 constructs to test the hypothesis that simultaneous phosphorylation of p53 at threonine 18 (Thr18) and serine 20 (Ser20) is causally associated with p53-mediated apoptosis. Studies using phosphorylation- specific antibodies demonstrated that p53-induced apoptosis correlates with phosphorylation of p53 at Thr18 and Ser20 but not with carboxy-terminal phosphorylation (Ser392). To prove a causal relationship between apoptosis and Thr18 and Ser20 phosphorylation of p53, the effects of an adenoviral p53 construct that was not phosphorylated (Ad-p53) was compared with a Thr18/Ser20 phosphomimetic construct (Ad-p53-18D20D) in wild-type p53 gliomas. Whereas treatment with Ad-p53 resulted only in cell cycle arrest, treatment with Ad-p53-18D20D induced dramatic apoptosis. Microarray and Western blot analyses showed that only Ad-p53-18D20D was capable of inducing expression of apoptosis-inducing proteins. Chromatin immunoprecipitation assays indicated that the protein product of Ad-p53-18D20D, but not Ad-p53, was capable of binding to apoptosis-related genes. We thus conclude that phosphorylation of Thr18 and Ser20 is sufficient for inducing p53-mediated apoptosis in glioma cells. These results have implications for p53 gene therapy and inform other strategies that aim to restore p53 function.
The p53 protein plays a critical role in cell cycle arrest and apoptosis, and its loss or inactivation is common in human cancers.1–4 Consequently, therapies directed at restoring p53 function, particularly its ability to induce apoptosis, have been a major focus of anticancer therapy for a variety of tumor types, including glioblastoma multiforme, the most common malignant brain tumor in adults.5–8 Recent animal models have supported the therapeutic validity of restoring p53 function in cancer based on the concept of cellular “addiction” to p53 loss or inactivation.9–11 Although several approaches have been attempted for reactivating p53 function, the most clinically tested strategy has been adenoviral-mediated p53 gene therapy (Ad-p53), in which wild-type p53 is directly introduced into tumors via adenoviral vectors.5 Ad-p53 has been examined in several clinical trials,12 including a phase I trial in patients with recurrent malignant gliomas, which demonstrated that the approach is safe.13 However, efficacy was limited, at least in part, because the expression of exogenous p53 in glioma cells was only millimeters from the site of injection of Ad-p53.13 Therefore, the emphasis of much research on brain tumor gene therapy has been on improving the delivery of the p53 gene to tumor cells.14–16 More recently, strategies using small molecules for restoring p53 in mutant p53 and wild-type p53 tumors have been explored because of the potential of these drugs for better intratumoral distribution than gene therapy.17,18
An equally important but less emphasized outcome of investigations into p53 gene therapy in gliomas has been the recognition that tumors containing normal (wild-type) p53 alleles respond differently to Ad-p53 than do tumors containing mutant or null p53.19–21 Indeed, Gomez-Manzano et al.19 were the first to show that gliomas containing mutant p53 alleles undergo dramatic apoptosis after treatment with Ad-p53, whereas Ad-p53 has little cytotoxic effect on gliomas containing normal (wild-type) p53 alleles. Unlike gliomas containing mutant p53, when glioma cells harboring wild-type p53 alleles are treated with Ad-p53 the result is reversible growth arrest (G1 arrest) without evidence of apoptosis.20,22,23 Thus, these wild-type p53 cells are resistant to the apoptosis-inducing effects of Ad-p53, although they maintain p53-induced cell cycle arrest. This arrest-response of wild-type p53 cells to Ad-p53 is not restricted to gliomas and also occurs in other tumor types, including prostate and lung cancer.24,25 Indeed, enrollment in clinical trials of patients whose tumors contain wild-type p53 alleles undoubtedly contributed to the limited efficacy of Ad-p53 in these trials.13
Although the different responses of mutant p53 and wild-type p53 gliomas to Ad-p53 have been known for some time, the mechanism underlying these different responses has not been elucidated to date. Understanding this mechanism has obvious implications for improving p53 gene therapy as large numbers (50% – 70%) of human gliomas harbor wild-type p53 and are resistant to Ad-p53.6,8 More importantly, understanding why wild-type p53 gliomas undergo G1 arrest but not apoptosis after exposure to Ad-p53 may inform other approaches that attempt to restore p53 function in wild-type p53 cells and thus will have more broad implications beyond p53 gene therapy itself.18 Lastly, and potentially most important, understanding the mechanism underlying the different responses of mutant p53 and wild-type p53 tumor cells to Ad-p53 may provide insight into the biological mechanism underlying p53’s ability to induce cell cycle arrest compared with apoptosis in cancer cells. This is of interest as the presence of wild-type p53 in high-grade astrocytomas has been correlated with resistance to cytotoxic chemotherapy.7 Since inactivation of p53 is a sine qua non of cancer, the presence of wild-type p53 protein implies that mechanisms of inactivation of p53 function, other than gene mutation may impact p53’s ability to induce cell cycle arrest versus apoptosis.9 Indeed, induction of p53-mediated cell cycle arrest rather than apoptosis allows wild-type p53 cells to repair treatment-induced DNA damage, which imparts resistance to radiation and chemotherapy. Thus, defining the mechanism of Ad-p53 induced apoptosis in cancer cells harboring wild-type p53 may provide insights into ways of overcoming the resistance of wild-type p53 gliomas to DNA-damaging agents.
The resistance of wild-type p53 gliomas to Ad-p53 – induced apoptosis is not simply a result of inhibition of cellular apoptotic pathways downstream of p53, because this resistance to Ad-p53 – mediated apoptosis can be overcome by combining Ad-p53 with DNA-damaging agents (radiation or chemotherapy).22–25 Moreover, because Ad-p53 results in very high levels of p53, the inability of Ad-p53 to induce apoptosis and the preference for cell cycle arrest in wild-type p53 cells indicate that p53-mediated apoptosis requires more than just increases in p53 protein levels. In this context, previous studies have suggested that N-terminal phosphorylation of p53 is important for p53’s functions.1–4 The p53 protein is capable of being phosphorylated at multiple amino acids (Ser6, Ser9, Ser15, Thr18, Ser33, Ser37, and Ser46) within the N-terminal transcriptional activation region of the protein, although the precise role of phosphorylation of each of these sites in the function of p53 remains unclear (Fig. 1A). Although several studies have suggested that N-terminal phosphorylation is not important for p53-mediated apoptosis,26,27 other studies have suggested that N-terminal phosphorylation of p53 is causally related to apoptosis.28–33
Based on these findings, we recently investigated the role of N-terminal phosphorylation in the response of mutant and wild-type p53 gliomas to Ad-p53. We found that N-terminal phosphorylation of exogenously delivered p53 occurs in mutant p53 gliomas that undergo apoptosis after treatment with Ad-p53, whereas N-terminal phosphorylation is not detected in the apoptosis-resistant wild-type p53 glioma cells that undergo G1 arrest after treatment with Ad-p53.23 Interestingly, the induction of apoptosis in wild-type p53 gliomas that occurs when Ad-p53 is combined with DNA-damaging agents (radiation or cisplatin) is associated with N-terminal phosphorylation of exogenous p53. Thus, these studies suggested that N-terminal phosphorylation may be a critical mechanism underlying the ability of Ad-p53 to induce apoptosis versus cell cycle arrest.23 In related studies aimed at defining the roles of phosphorylating specific N-terminal sites of p53, our group recently investigated the impact of phosphorylating p53 at Thr18 and Ser20.34–36 In these studies, we showed that phosphorylation of both Thr18 and Ser20 had significantly greater impact on the biochemical activity of p53 than did phosphorylation of either site alone, resulting in reduced interaction of p53 with MDM-2 (the primary negative regulator of p53).35,36 These studies, however, did not specifically address the biological consequence of phosphorylating Thr18 and Ser20 of p53, namely, whether phosphorylating these sites induces apoptosis.
Consequently, in the present study we determine the extent to which phosphorylation of Thr18 and Ser20 of p53 is causally related to the induction of apoptosis in cancer cells and whether this phosphorylation is a mechanism underlying p53’s ability to induce apoptosis. We reasoned that if phosphorylation of Thr18 and Ser20 of p53 was causally related to p53-mediated apoptosis, then an adenoviral vector containing a p53 gene engineered to mimic constitutive phosphorylation at Thr18 and Ser20 by site-specific substitution of Asp for Thr18 and Ser20 (Ad-p53-18D20D) should be capable of inducing apoptosis in wild-type p53 glioma cells that are resistant to the apoptotic effects of an adenoviral vector containing a wild-type (unphosphorylated) p53 gene (Ad-p53). Hence, we compared the effects of Ad-p53 to that of Ad-p53-18D20D to determine the extent to which Ad-p53-18D20D is capable of inducing apoptosis in wild-type p53 gliomas that normally undergo G1 arrest after treatment with Ad-p53. To prove that phosphorylation of Thr18 and Ser20 of p53 has a role in inducing apoptosis in cancers other than gliomas, the effect of Ad-p53-18D20D on colon cancer cells was also explored. Lastly, we tested the hypothesis that the proapoptotic effects of Ad-p53-18D20D are due to the capacity of this construct’s phosphorylated p53 protein product to induce the expression of apoptosis-related genes compared with that of Ad-p53.
The cell lines U251, U87, HT29, RKO, HCT116, and CCD32-Lu were obtained from the American Type Culture Collection (Manassas, VA, USA). D54 was generously provided by Dr. Darell Bigner (Duke University, Durham, NC, USA). Normal human astrocyte (NHA) was generously provided by Dr. Fueyo (The University of Texas M. D. Anderson Cancer Center, Houston, TX, USA). Glioma cells (U251, U87, and D54) were maintained in minimal essential medium (MEM) supplemented with 10% fetal bovine serum (FBS); colon cancer cells (HT29, RKO, and HCT116) were maintained in high-glucose Dulbecco’s modified Eagle’s medium supplemented with 10% FBS; and fibroblast cells (CCD-2-Lu) were maintained in α-MEM supplemented with 20% FBS. NHA cells were maintained with DMEM supplemented with astrocyte growth media (AGM) SingleQuots, 20 ng/ml human epidermal growth factor, 50 μg/ml transferrin, 25 μg/ml insulin, and 25 ng/ml progesterone (Lonza, Allendale, NJ, USA). Cells were incubated at 37°C in a humidified atmosphere containing 5% CO2/95% air. U87, D54, RKO, and HCT116 cells had homozygous wild-type p53 alleles based on sequence evaluations, whereas U251 and HT29 cells contained a mutant p53 allele at codon 273 (CGT/CAT; Arg/His).19
The generation and features of the Ad-p53 construct have been reported previously.19,20,22,23 The cytomega-lovirus-p53-18D20D expression vector36 was used as a template for generating the Ad-p53-18D20D vector.37 The E1A-deleted adenovirus vector (Ad-βgal) was used as a control.
Cells (2.2 × 105) were plated on 10-cm dishes. Twenty-four hours after plating, the cells were washed with phosphate-buffered saline (PBS) and incubated with medium, vector (Ad-βgal), Ad-p53, or Ad-p53-18D20D in 1 ml medium without serum for 1 h at 37°C in 5% CO2/95% air, with brief agitation every 10 min. Multiplicity of infection (MOI) was based on the original cell number plated in all experiments and was established at 100 plaque-forming units/cell. After 1 h, fresh medium supplemented with FBS was added to each dish.
At 48 or 72 h after infection, cells were trypsinized with 0.05% trypsin/1 mM EDTA solution and stained with 0.04% trypan blue solution. Only negative-stained cells were counted using a hemocytometer. Survival curves were generated by combining data from three independent experiments.
The p53 protein and phospho-p53 (Ser15, 20, or 46), MDM-2, p21/WAF1, Fas, Bax, PUMA, PIG3, IGFBP-3, and PERP levels were determined by Western blotting as described previously.23,35 The following antibodies were used: p53 monoclonal antibody (mAb; PAb1801), α-tubulin mAb, MDM-2 mAb, p21/WAF1 mAb, Noxa mAb, and PIG3 polyclonal antibody (Oncogene Research Products, Cambridge, MA, USA); phospho-p53 mAb (Ser15), phospho-p53 polyclonal antibody (Ser20 and Ser46), and Bax polyclonal antibody (Cell Signaling Technology, Beverly, MA, USA); phospho-p53 polyclonal antibody (Thr18 and PAbThr18P, generously provided by Dr. Jabbur, The University of Texas M. D. Anderson Cancer Center); Fas polyclonal antibody and IGFBP-3 polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA); PUMA polyclonal antibody (Abcam, Cambridge, MA, USA); and PERP polyclonal antibody (Sigma, St. Louis, MO, USA).
Annexin V-PE assay was performed according to the manufacturer’s instructions (BD Pharmingen, San Diego, CA, USA). Specimens were analyzed using flow cytometry. An analysis region was set based on the negative control, and the percentage of labeled cells was calculated from this region.
At 36 h after infection, cells were washed three times with cold PBS and fixed with 4% paraformaldehyde in PBS for 15 min at room temperature. Fixed cells were incubated in blocking buffer (2% bovine serum albumin and 1% Triton X-100 in PBS) for 1 h followed by overnight incubation with the primary antibody (rabbit antihuman Fas polyclonal antibody, 1:2,000; Santa Cruz Biotechnology) in blocking buffer at 4°C. Cover-slips were incubated with Texas red – conjugated secondary antibody (1:200; Molecular Probes, Eugene, OR, USA) for 30 min. For staining of nucleus and mitochondria, coverslips were incubated with the fluorescein-conjugated antibodies (Dapi and Mitotracker Green FM; Molecular Probes). Stained cells were viewed by laser scanning epifluorescence microscopy using a Leica DMLB microscope equipped with a ×5 or ×20 lens. Epi-fluorescent images were captured with a charged-couple device camera and processed using IP Lab software and Adobe Photoshop 7.0 (Adobe Systems Inc., San Jose, CA, USA).
Microarray analysis was carried out as described previously38,39 using the pathway array containing 1,500 functionally well-characterized genes in duplicate (http://www.mdanderson.org/~genomics). Hybridized arrays were scanned with a GeneTAC LSIV scanner (Genomic Solutions, Ann Arbor, MI, USA). The images were quantified using the imaging software ArrayVision from Imaging Research, Inc. (St. Catherines, Ontario, Canada), and the significantly differentially expressed genes were identified as previously described.40 For differentially expressed genes, a smoothed t-value cutoff of 3.0 was used.
To determine changes in the expression of specific genes, quantitative real-time PCR (RT-PCR) was performed on the ABI Prism 7700 using the commercially available gene expression assays for Fas/TNFRSF6, Bax, p21/CDKN1A, PERP/PIGPC1, IGFBP3, and PUMA/BBC3 (Hs00234753_m1, Hs00163653_m1, Hs00751844_s1, Hs00355782_m1, Hs00560402_m1, Hs00366272_m1, Hs00751717_s1, Hs00169255_m1, Hs00426287_m1, Hs00248075_m1, Hs00223141_m1, and Hs00388035_ m1, respectively) and the cylophilin Vic-Labeled Pre-Developed Assay Reagent (Applied Biosystems, Foster City, CA, USA) without multiplexing. Initial experiments were performed to determine the valid range of RNA concentrations and to demonstrate the similarity of PCR efficiencies for each gene of interest compared to the endogenous control gene cyclophilin. Expression of Fas/TNFRSF6, Bax, p21/CDKN1A, and PERP/PIGPC1 was assessed using standard one-step chemistry and cycling conditions. Briefly, in triplicate, 50 ng total RNA was amplified for each sample for each assay in a 50-μl reaction containing 1× TaqMan Universal PCR Master Mix, 1× gene expression assay, and 1× Multiscribe RT with RNase inhibitors (Roche, Branchburg, NJ, USA) with the following cycling conditions: 30 min at 48°C and 10 min at 95°C, then 40 cycles of 95°C for 15 s and 60°C for 1 min. Expression of IGFBP-3 and PUMA/BBC was assessed using standard two-step chemistry and cycling conditions. Briefly, in triplicate, 5 – 50 ng cDNA was amplified for each sample for each assay in a 50-μl reaction containing 1× TaqMan Universal PCR Master Mix without AmpErase UNG and 1× gene expression assay with the following cycling conditions: 10 min at 95°C, then 40 cycles at 95°C for 15 s and 60°C for 1 min. Calculations were performed using the Δ ΔCt method to determine fold difference of each gene.
Chromatin immunoprecipitation (ChIP) was carried out with slight modifications as described previously.41 Briefly, 24 h after treatment, cells were treated with formaldehyde to give a 1% solution in the media and incubated with rocking for 10 min at room temperature. After addition of glycine (0.125 M) and incubation for 5 min, cells were washed twice with cold PBS containing protease inhibitors (Roche Diagnostics, Indianapolis, IN, USA) on ice. Cells were then scraped and centrifuged for 4 min at 2,000 rpm at 4°C. After resuspending pellet in 200 μl of lysis buffer (1% sodium dodecyl sulfate [SDS], 10 mM EDTA, and 50 mM Tris [pH 8.1]) containing protease inhibitors, DNA was sonicated to a range of 200- to 1-kb fragments on ice, and lysates were cleared by centrifugation for 15 min at 14,000 rpm at 4°C. The cross-linked chromatin preparations were diluted 10-fold in buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 16.7 mM Tris-HCl [pH 8.1], and 167 mM NaCl) containing protease inhibitors, and 1% of samples were saved as total input DNA. After preclearing with 75 μl of salmon sperm DNA/protein A-agarose-50% slurry (Upstate Biotechnology, Lake Placid, NY, USA) for 1 h at 4°C, immunoprecipitation (IP) was performed with rotating overnight at 4°C using 5 μg of anti-p53 antibody (PAb1801). Then, 60 μl of above slurry was added and incubated by rotating for 4 h at 4°C. After washing pellet five times on ice with each washing solution (Upstate), the complex was eluted twice by rotating for 30 min at room temperature in 250 μl of elution buffer (1% SDS, 0.1 M NaHCO3, and 10 mM dithiothreitol). Then, 20 μl of 5 M NaCl was added and incubated overnight at 65°C to reverse cross-linking. Samples were digested with 20 μg of proteinase K (Roche), 10 μl of 0.5 M EDTA, and 20 μl of 1 M Tris-HCl (pH 6.5) for 1 h at 45°C. DNA fragments were recovered by phenol-chloroform extraction and ethanol precipitation with 20 μg of glycogen (Invitrogen, Carls-bad, CA, USA). Each sample was resuspended in 100 μl of 10 mM Tris (pH 7.5) and 1 mM EDTA.
PCR reactions contained 4 μl of IP sample or 4 μl of a 1:100 dilution of input sample; 0.5 μM of each primer; 200 μM each of dATP, dGTP, dCTP, and dTTP (Roche); 1× PCR buffer (Roche); and 1.25 units of FastStart Taq DNA polymerase (Roche) in 20 μl of total volume. After 35 cycles of amplification, PCR products were electrophoresed on a 1.3% agarose gel, and DNA was stained with ethidium bromide and visualized under UV light. Signals were quantitated using ImageQuant v5.0 (Molecular Dynamics, Sunnyvale, CA, USA). Fold changes were calculated by first dividing the IP signal intensity by the input signal intensity (to control for different cell numbers by each treatment). Then, each value was divided by the media value. The sequences of the PCR primers are as follows: for p21, GTGGCTCTGATTG-GCTTTCTG and CTGAAAACAGGCAGCCCAAG; for Fas, GACAGGAATTGAAGCGGAAG and CGTCT-GAGAACTGCCAGAAA; for Bax, AGGCTGAGACG-GGGTTATCT and GCGCAGAAGGAATTAGCAAG; for PUMA, GCGAGACTGTGGCCTTGTGT and CGT-TCCAGGGTCCACAAAGT; for PIG3, CACTCCCA-ACGGCTCCTTT and GCCCATCTTGAGCATGGGT; and for IGFBP3, ACAGCCAGCGCTACAAAGTT and GGAGAGAGGTAGGGCACAGA.
Statistical differences were assessed by Student’s t-test. Differences were determined to be statistically significant (p < 0.05) by using two-tailed procedure. The data for all PCR results were presented as mean ± SEM for at least three replicate determinations for each PCR and were observed in at least two sets of experiments.
In order to verify the role of N-terminal phosphorylation of p53 in Ad-p53–mediated apoptosis, specifically at Thr18 and Ser20, we treated glioma cell lines harboring either mutant p53 alleles (U251) or wild-type p53 alleles (U87) with medium, control vector (MOI 100), or Ad-p53 (MOI 100). Cells were collected at increasing times and assayed for levels of total p53 protein and levels of phosphorylated p53 at Thr18 and Ser20 using phosphorylation-specific antibodies. Levels of phosphorylation at carboxy-terminal sites (Ser392) were also assayed. Unphosphorylated and phosphorylated p53 levels were correlated with the level of apoptosis. Compared with treatment with medium or empty vector, infection with Ad-p53 produced high levels of total p53 protein in both U251 and U87 cells (Fig. 1B). However, only the U251 cells, which undergo apoptosis after Ad-p53 treatment, demonstrated phosphorylation of exogenous p53 at Thr18 and Ser20, whereas no phosphorylation of these sites was seen in U87 cells that are resistant to Ad-p53 – mediated apoptosis (Fig. 1B). In contrast, Ser392 was phosphorylated in both the apoptotic U251 cells and the apoptosis-resistant U87 cells. In U87 cells, which do not undergo apoptosis after treatment with Ad-p53 alone, combined treatment of Ad-p53 and radiation therapy (9 Gy), induced apoptotic cell death; importantly, this apoptotic response was associated with phosphorylation of Thr18 and Ser20. No change in the phosphorylation status of Ser392 was seen with the combination therapy (Fig. 1B). Taken together, these results confirm and extend our previous results23 that N-terminal phosphorylation, specifically at Thr18 and Ser20, correlates with apoptosis induction, whereas phosphorylation of the carboxy terminus (Ser392) does not. Similar results were seen for the mutant p53 (HT29) and wild-type p53 (RKO and HCT116) colon cancer cell lines (data not shown), suggesting that this correlation between apoptosis and N-terminal phosphorylation of p53 at Thr18 and Ser20 is not unique to gliomas.
These results suggest that there is a correlation between N-terminal phosphorylation of p53 and apoptosis induction. However, these results do not specifically indicate whether phosphorylation of these sites is causally related to the induction of apoptosis. To test the hypothesis that phosphorylation of specific N-terminal sites is sufficient for inducing apoptosis, we constructed an adenovirus vector containing a p53 gene in which Thr18 and Ser20 were mutated to aspartic acid (D) so as to mimic constitutive phosphorylation at these sites (Ad-p53-18D20D). Substitution of aspartic acid is a generally accepted method for mimicking phosphorylation due to the introduction of a negative-charged functional unit. We developed a vector with mutation at both Thr18 and Ser20 because previous studies from our group indicated that simultaneous phosphorylation of these two sites acts synergistically relative to p53 phosphorylated at each site alone.35,36 Because wild-type p53 gliomas do not undergo apoptosis after treatment with Ad-p53, we compared the survival of wild-type p53 cell lines (U87 and D54) treated with Ad-p53 to that of U87 and D54 cells treated with Ad-p53-18D20D. Consistent with previous reports,22,23 treatment of these wild-type p53 cell lines (U87, D54) with Ad-p53 resulted in only a slowing of cell growth compared to treatment with medium or vector control (Fig. 2A). In contrast, treatment with Ad-p53-18D20D resulted in a significant decrease in cell survival based on a dye exclusion assay and morphological analyses (Fig. 2A).
Analyses of apoptosis based on annexin-V assays demonstrated that treatment with medium, control vector, or Ad-p53 resulted in low levels of apoptosis (3% – 7% for U87 and 4% – 9% for D54 cells), whereas treatment with Ad-p53-18D20D significantly increased the amount of apoptosis (43.7% for U87 and 56.4% for D54) (Fig. 2B). Moreover, detailed examinations of changes in the cell cycle over time (Table 1) revealed that treatment of U87 cells with Ad-p53 resulted in G1 arrest that was detected 24 h after infection and persisted for at least 48 h, whereas treatment with Ad-p53-18D20D resulted in G2 arrest at 24 h followed by a significant increase in the sub-G1 (apoptotic) population between 36 and 48 h after infection (Table 1). As a positive control, mutant-p53 U251 cells were also treated with Ad-p53, Ad-p53-18D20D, or control vector; as expected, treatment with Ad-p53 (which is phosphorylated in these cells, see Fig. 1) induced significant cytotoxicity and apoptosis that was slightly augmented by treatment with Ad-p53-18D20D (Fig. 2). Similar results were seen for wild-type p53 (RKO and HCT116) and mutant-p53 (HT29) colon cancer cell lines (Fig. 3). Taken together, these results demonstrate that a p53 construct, which mimics constitutive phosphorylation at Thr18 and Ser20 (Ad-p53-18D20D), is capable of inducing apoptosis in cell lines that normally undergo G1 arrest after exposure to exogenously delivered unphosphorylated p53 (Ad-p53), indicating that phosphorylation at Thr18 and Ser20 of p53 is sufficient for inducing apoptosis in this model system.
In addition to Thr18 and Ser20, phosphorylation of other p53 N-terminal sites, particularly Ser15 and Ser46, has been implicated in the induction of apoptosis by p53.1,28,42 To determine whether N-terminal sites of p53 other than Thr18 and Ser20 are also important for apoptosis induction in our model, we investigated whether the protein product of Ad-p53-18D20D was also phosphorylated at Ser15 or Ser46 during apoptosis. U87 and D54 glioma cells were infected with Ad-p53 or Ad-p53-18D20D and analyzed by Western blotting using antibodies that recognize total p53 protein or p53 phosphorylated at Ser15 or Ser46 (Fig. 4). As a positive control, U251 (mutant p53) cells were also analyzed. Compared to controls, Ser15 or Ser46 of the exogenous p53-18D20D protein was not phosphorylated after treatment of U87 and D54 cells with Ad-p53-18D20D, indicating that phosphorylation of Thr18 and Ser20 is sufficient to induce apoptosis and that apoptosis can occur independent of Ser15 or Ser46 phosphorylation.
Given its function as a transcription factor, the ultimate effects of p53 are likely to be mediated through change in gene expression. Indeed, the p53 protein is known to regulate the expression of a large number of target genes that control cell cycle arrest and apoptosis. We hypothesized that the conversion of the biological response of wild-type p53 glioma cells from growth arrest (G1 arrest) after treatment with Ad-p53 to apoptosis after treatment with Ad-p53-18D20D occurred because Ad-p53-18D20D increased the expression of genes that mediate apoptosis compared with Ad-p53.
In order to broadly assess the influence of the phosphorylation of Thr18 and Ser20 of p53 on the expression of apoptosis-related p53 targets, we used a pathway microarray that contains many known p53-regulated genes to analyze the gene expression alterations in wild-type p53 glioma and colon cancer cells after infection with Ad-p53 compared with Ad-p53-18D20D. Figure 5 shows the expression profile of the wild-type p53 glioma cell lines U87 and D54 and colon cancer cell lines RKO and HCT116 treated with Ad-p53 compared with Ad-p53-18D20D at two different time points. We detected multiple genes that showed increased expression after Ad-p53-18D20D treatment compared with Ad-p53 treatment (Fig. 5 and Table 2). The cell cycle control gene p21/WAF1 was increased after treatment with Ad-p53, consistent with the ability of Ad-p53 to induce cell cycle arrest (see Supplemental Table 1, which compares Ad-p53 with control vector). Ad-p53-18D20D resulted in further increase in p21/WAF (Table 2). Most importantly, several genes had undetectable expression after Ad-p53, but highly significantly increased expression after Ad-p53-18D20D (Supplemental Tables 1 – 3). These genes included many apoptosis-related genes, specifically Fas (tumor necrosis factor receptor superfamily, member 6), APAF-1 (apoptosis protease activating factor), and IGFBP-3 (insulin-like growth factor binding protein 3), as well as several caspases. In addition, it was of interest that expression of the G2-arrest gene, GADD45, increased markedly in the Ad-p53-18D20D – treated cells compared to Ad-p53 – treated cells, a finding that may explain the occurrence of G2 arrest prior to apoptosis in these cells (see Table 1).43,44 Taken together, these patterns suggest a shift toward apoptosis-related genes after cellular exposure to constitutively phosphorylated p53 protein (Fig. 5B).
In order to more precisely assess the influence of the phosphorylation status of p53 on the expression of apoptosis-related p53 targets, we analyzed the effects of Ad-p53 and Ad-p53-18D20D on specific p53-regulated genes known to mediate G1 arrest (p21/WAF1) and apoptosis (Fas, Bax, PERP, PUMA, PIG3, IGFBP-3).45
To assess the expression of these p53-target genes at the protein level, monolayers of U87 cells were treated with medium, Ad-βgal (control vector), Ad-p53 (MOI 100), or Ad-p53-18D20D (MOI 100), and cell lysates were subjected to Western blotting using antibodies to MDM-2 (as a control protein not involved in the cell cycle), the cell cycle control protein p21/WAF1, and the p53-regulated, proapoptotic proteins Bax, Fas, PUMA, PIG3, PERP, and IGFBP-3. Treatment of U87 with Ad-p53 significantly increased the expression of p21 compared to the vector and media controls, and treatment with Ad-p53-18D20D produced a slight increase compared to Ad-p53 (Fig. 6A). In contrast, the expression of the proapoptotic proteins PUMA, Fas, PIG3, PERP, and IGFBP-3 was not increased after treatment with Ad-p53 compared with control vector, whereas treatment with Ad-p53-18D20D resulted in highly significant increases in the expression of these proapoptotic proteins (Fig. 6A). Because Fas is a surface receptor, we also analyzed the surface expression of Fas by indirect immunofluorescence with laser scanning microscopy in U87 cells. Whereas Fas was essentially undetectable on the surface of U87 cells treated with Ad-p53, significant expression of Fas on the cell surface was seen exclusively in Ad-p53-18D20D – treated cells (Fig. 6B). Likewise, cell surface analysis using flow cytometry revealed a significant increase in surface expression of Fas in cells treated with Ad-p53-18D20D compared to Ad-p53 – treated cells (Fig. 6C). In contrast to all these proteins, levels of the proapoptotic protein Bax did not change in these cells after either Ad-p53 or Ad-p53-18D20D. Taken together, these data suggest that Ad-p53-18D20D induces apoptosis by increasing the expression of specific p53-regulated proapoptotic proteins.
To assess the extent to which changes in levels of cell cycle and proapoptotic proteins were reflected at the mRNA level, U87 and D54 cells were treated with medium, Ad-βgal (control vector), Ad-p53 (MOI 100), or Ad-p53-18D20D (MOI 100), and cell lysates were analyzed by real-time RT-PCR for gene expression of p21, Bax, Fas, IGFBP-3, and PERP. Treatment of U87 cells with Ad-p53 resulted in high transcriptional expression of the cell cycle control gene p21, which increased further after treatment with Ad-p53-18D20D (Table 3). In contrast, expression levels of the proapoptotic genes PERP, Fas, and IGFBP-3 were similar to the media-treated controls after treatment with Ad-p53 (0.80- to 1.27-fold change relative to media control) but were markedly increased after treatment with Ad-p53-18D20D (2.3- to 6.7-fold change relative to media control). Consistent with the protein expression data, no increase in Bax was detected.
To determine the extent to which phosphorylation of Thr18 and Ser20 increases DNA binding of p53 to proapoptotic gene promoters, U87 cells were treated with medium, Ad-βgal (control vector), Ad-p53 (MOI 100), or Ad-p53-18D20D (MOI 100) and ChIP assays were undertaken to assess binding of exogenous p53 to p21, Bax, Fas, PUMA, IGFBP-3, and PIG3. MDM-2 represented a non – cell cycle protein and was used a control. Whereas binding to p21 was detected both qualitatively and quantitatively after treatment with either Ad-p53 or Ad-p53-18D20D, binding to the proapoptotic proteins Fas, PUMA, IGFBP-3, and PIG3 occurred only after treatment with Ad-p53-18D20D (Fig. 7). No binding to Bax promoter was detected (Fig. 7). Taken together, these results indicate that phosphorylation of exogenous p53 at Thr18 and Ser20 specifically increases the binding and expression of the apoptosis-related genes.
We next examined whether nontumoral wild-type p53 cells, including both fibroblasts and astrocytes, were also sensitive to its apoptotic effects. Most normal fibro-blasts have low infectivity with adenovirus, thus providing selectivity to adenovirus-mediated cell kill approach. However, to evaluate the underlying mechanism of p53-18D20D, we used the CCD32-Lu fibroblast line, which has higher infectivity and can achieve 90% – 100% infectivity of cells with high titers (MOI 400) of vector. For NHAs, 90% – 100% infectivity was achieved with 200 MOI. Similar to wild-type p53 tumor cells, treatment of both fibroblasts and NHAs with Ad-p53 resulted in a slowing of cell growth, whereas treatment with Ad-p53-18D20D induced significant cell killing (Fig. 8). Thus, it appears that nontumoral wild-type p53 cells respond similarly to phosphorylated p53 as do wild-type p53-containing tumor cells, which suggests that phosphorylation of p53 may be a critical mechanism underlying the cellular choice between apoptosis and cell cycle arrest in both tumor cells and normal cells. Thus, therapeutic approaches aimed at increasing p53 phosphorylation must take this into account.
The data presented herein show that Ad-p53-18D20D is able to overcome the resistance of wild-type p53 gliomas to Ad-p53 – mediated apoptosis. Specifically, treatment of wild-type p53 glioma or colon cancer cells with Ad-p53, whose protein product is not phosphorylated at Thr18 and Ser20, results only in cell cycle (G1) arrest, whereas treatment with Ad-p53-18D20D, a construct that mimics phosphorylation at Thr18 and Ser20 of p53, converts this cell cycle arrest response to an apoptotic response. Moreover, the protein product of Ad-p53-18D20D is capable of binding to and trans-activating p53-regulated proapoptotic genes (e.g., Fas, PUMA, Noxa, and IGFBP3), whereas the protein product of Ad-p53 is not.
These studies have obvious implications for adenoviral-mediated p53 gene therapy. Specifically, they indicate that Ad-p53-18D20D is potentially more effective against human gliomas than is Ad-p53 because it can induce apoptosis in tumors with wild-type p53 alleles, which make up 50% – 70% of human gliomas, while maintaining its ability to induce apoptosis in mutant p53 gliomas. This is important because p53 gene therapy, which is currently being tested in clinical trials, is limited at least in part by the resistance of wild-type p53 tumor cells to Ad-p53. In addition, because Ad-p53-18D20D induces apoptosis without adjuvant therapies, such as radiation and chemotherapy, it may be effective as a single-agent therapy, particularly in recurrent disease when radiation and standard cytotoxic chemotherapy have been exhausted. As strategies for increasing delivery of therapeutic genes (e.g., convection-enhanced delivery, armed conditionally replicative viruses, or nanoparticles for gene delivery) are developed, the potential therapeutic benefit of delivering Ad-p53-18D20D compared with Ad-p53 may be realized.14–16 However, because Ad-p53-18D20D may induce apoptosis in normal astrocytes and fibroblasts, its use may result in greater toxicity than Ad-p53, which heretofore has proved safe in multiple clinical trials. Although high titers of Ad-p53 (MOI 200 – 400) were needed to infect normal cells, making it unlikely that toxic levels will be reached, strategies for activating the p53 gene, specifically in tumor cells, such as the use of tumor-specific promoters, may also overcome the potential for toxicity to normal cells.
In addition to their implications for p53 gene therapy, the studies reported herein also provide insight into the biological mechanisms underlying p53’s ability to induce either cell cycle arrest or apoptosis. Despite continued research, what remains unclear is the extent to which p53 mediates the choice between cell cycle arrest and apoptosis and the mechanism underlying this choice, including the role of N-terminal phosphorylation of p53. More specifically, studies have both supported28–33,42,46–48 and refuted26,27,49,50 the concept that N-terminal phosphorylation of p53 integrates the choice between cell cycle arrest and apoptosis. In this context, we show that while treatment of glioma cells with Ad-p53, a construct whose protein product remains unphosphorylated at Thr18 and Ser20, results only in cell cycle arrest, treatment with Ad-p53-18D20D, a construct that mimics phosphorylation at Thr18 and Ser20, converts this cell cycle arrest response to an apoptotic response. Similar results were seen in colon cancer indicating that the findings are not unique to gliomas. Thus, our data suggest that p53 is capable of mediating the choice between cell cycle arrest and apoptosis and that N-terminal phosphorylation has a causal role in this choice.
Importantly, we show that other phosphorylation sites within the N-terminus of p53-18D20D protein, such as Ser15 and Ser46, are not differentially phosphorylated compared with Ad-p53 in this experimental system (Fig. 4). Consequently, the N-terminus of the protein product of Ad-p53-18D20D differs from that of Ad-p53 only by aspartate mutations at Thr18 and Ser20 and therefore indicates that Ad-p53-18D20D – mediated apoptosis occurs independent of other N-terminal phosphorylation sites. Thus, phosphorylation of Thr18 and Ser20 on p53 appears to be sufficient to induce apoptosis. Because several of these other N-terminal phosphorylation sites (specifically, Ser15 and Ser46) have been implicated in apoptosis,28,42 it will be of interest to test in the future whether these sites provide a redundant apoptosis functional switch or whether Ser15 and Ser46 play a modifier role for Thr18 and Ser20 in the apoptosis process. This potential cross-talk between different sites may provide a mechanism for fine-tuning apoptotic responses in response to graded amounts of DNA damage.
Consistent with our studies, others have also shown in various model systems an association between N-terminal phosphorylation of p53 and apoptosis induction.28–33 At least one of these studies also supports the notion that simultaneous phosphorylation of N-terminal sites of p53, rather than single-site phosphorylation, is critical to induction of apoptosis.29 Nevertheless, a substantial body of evidence also suggests that phosphorylation of N-terminal sites is not required for p53-mediated apoptosis.26,27,49,50 Ashcroft et al.26 reported that inactivating mutations of all N-terminus phosphorylation sites of p53 did not abrogate p53 transcriptional function. In therapeutically relevant studies, Thompson et al.27 explored the ability of nutlin-3, a recently developed MDM-2 antagonist, to induce apoptosis. Exposure of wild-type p53 cells to nutlin-3 resulted in an increased level of p53 and apoptosis induction, particularly in cells that overex-pressed MDM-2. However, nutlin-induced, wild-type p53 was not phosphorylated at any of the six serines tested. In contrast to our results, the nutlin-induced, unphosphorylated p53 was capable of activating p53-target genes, although the investigators did not specifically explore the expression of apoptosis genes. One explanation for the disparity between our results and those of Thompson et al. is that nutlins may induce apoptosis via p53 independent pathways. For example, nutlins have been shown to induce apoptosis in p53 mutant cells via activation of E2F-1.51 Alternatively, we and others have previously shown that phosphorylation of Thr18 and Ser 20 of p53 results in the important biochemical consequence of inhibiting the interaction between p53 and MDM-2.34–36,52–54 Therefore, treatment with nutlins, which inactivates MDM-2, may have the same effect as phosphorylating p53 at Thr18 and Ser 20. This would imply that release of p53 from MDM-2 is a critical consequence of N-terminal phosphorylation of Thr18 and Ser20. In our model system, this release of p53 from MDM-2 requires phosphorylation of p53. This is of interest as inactivation of p53 in wild-type p53 cells typically occurs through increases in MDM-2 or by loss of p14/ARF.9 Although we have shown that phosphorylation is sufficient for apoptosis induction, studies such as those of Thompson et al. and Ashcroft et al. imply that phosphorylation of Thr18 and Ser20 is not necessary for the induction of apoptosis. Since releasing MDM-2 (or other inhibitors of p53) may be a main consequence of phosphorylation, artificial methods such as deletion of the N-terminus or treatment with molecules such as nutlins may also achieve this goal. Thus, the disparity between the reports published to date may actually imply a common mechanism for p53-mediated apoptosis induction.
Our results also indicate that phosphorylation of Thr18 and Ser20 actually drives p53 toward inducing apoptosis over G1 arrest because phosphorylation of Thr18 and Ser20 enhances p53’s ability to transactivate apoptosis-related genes. ChIP analyses indicate that the p53 protein product of Ad-p53, which is not phosphorylated at Thr18 and Ser20, binds primarily to p21/CIP/WAF with essentially no binding to proapoptotic proteins, whereas the Thr18/Ser20 phosphomimetic p53 protein product of Ad-p53-18D20D is capable of binding to both p21/CIP/WAF and the promoters of proapoptotic genes. Transcriptional studies using RT-PCR to quantify mRNA levels confirm that only Ad-p53-18D20D is capable of transactivating proapoptotic genes, whereas Ad-p53 is only able to activate the cell cycle protein p21/CIP/WAF. Microarray analyses comparing Ad-p53 with Ad-p53-18D20D demonstrated that the shift from G1 arrest to apoptosis that occurs after treatment with Ad-p53-18D20D correlates with a corresponding shift in the general expression pattern of p53-responsive genes toward those that are known to mediate apoptosis. Likewise, analyses of specific p53-regulated proteins indicate that whereas Ad-p53 and Ad-p53-18D20D are both capable of activating p21, a critical gene mediating G1 arrest, only Ad-p53-18D20D was capable of activating the proapoptotic proteins PUMA, Fas, IGFBP-3, PIG3, and PERP. These studies suggest that the phosphorylation status of p53 may dictate the capacity of p53 to bind to specific promoters with the transactivation of proapoptotic proteins depending, at least in this model system, on the presence of phosphorylated Thr18 and Ser20 of p53. These results are consistent with those of other investigators,42 who also reported that N-terminal phosphorylation (albeit at Ser46) dictated an apoptotic response. Thus, these studies support the notion that the phosphorylation status of p53 influences the choice of p53 target genes, with phosphorylation of Thr18 and Ser20 shifting the expression toward proapoptotic genes. Contrary to models in which p53 activation always results in a stereotypic apoptotic signal that is either attenuated or potentiated by other p53-independent survival or proapoptotic signals (e.g., Akt and nuclear factor κB), our results support the alternative proposal that p53 itself dictates the response by activating cell cycle arrest or apoptotic signal.4 Indeed, phosphorylation of Thr18 and Ser20 appears to be a mechanism whereby p53 alters the cellular fate toward apoptosis by specifically influencing gene expression.4 Moreover, because high-level expression of unphosphorylated p53, which is produced with adenoviral transfer, results only in expression of p21 and in G1 arrest, whereas expression of phosphorylated p53 transactivates proapoptotic proteins and induces cell death, these data suggest a model in which high levels of p53 (not phosphorylated at Thr18 and Ser20) result in G1 arrest and subsequent amino-terminal phosphorylation at Thr18 and Ser20 results in the induction of apoptosis.
Since phosphorylation of Thr18 and Ser20 led to increased expression of a large spectrum of proapoptotic genes, our data seem to indicate that phosphorylation of Thr18 and Ser20 generally activates all proapoptotic genes as a class with little specificity for each individual gene. However, it is important to note that the level of the proapoptotic gene Bax was not increased after transfer of Ad-p53-18D20D and that this phosphorylated form of p53 did not bind to the Bax promoter or increase Bax mRNA levels. This finding leaves open the possibility that specific phosphorylation events on the N-terminus of p53 may mediate expression of specific apoptotic genes. Oda et al.42 suggested a similar model based on their finding that phosphorylation of Ser46 resulted in the specific expression of p53AIP1. Although further studies are needed to validate this concept, such specificity may provide a mechanism for regulating the apoptotic response under complex stress signals.
An important caveat of our findings is that adenoviral vectors result in very high levels of p53, arguably supraphysiological levels. Nevertheless, this remains an interesting aspect of our studies; it is clear that these high levels of p53 alone are not sufficient for inducing apoptosis because treatment with Ad-p53 does not induce apoptosis in wild-type p53 cells. Only when p53 is phosphorylated is apoptosis induced. It appears that the high levels of exogenous p53 are not sufficient to result in binding of exogenous p53 to proapoptotic gene promoters, although high levels do result in binding to the cell cycle regulator, p21 promoter. Thus, the phosphorylation status of p53 is a critical factor in activation of specific p53-responsive elements and may be critical for inducing cell cycle arrest genes compared with apoptotic genes. Additionally, our studies did not evaluate the status of serines 6, 9, 33, and 37, and it is possible that phosphorylation of these sites is necessary for cell cycle arrest induction without apoptosis. Studies into these sites are currently under way. Moreover, one must recognize that substitution of Asp for Thr18 or Ser20 is only a first approximation of true phosphorylation. Nevertheless, within the limitations of these protein modifications, the results show a definite difference in the responsiveness of p53-regulated proapoptotic genes to Ad-p53 compared with Ad-p53-18D20D. Comparisons of different adenoviral phosphomimetics in this model system may shed light on the contribution of each N-terminal phosphorylation site in apoptosis induction.
Our studies raise the interesting possibility that attenuation of p53 phosphorylation may be an important mechanism whereby the apoptotic function of p53 is inactivated in tumor cells harboring wild-type p53 alleles. Because unphosphorylated p53 is capable of inducing cell cycle arrest but not apoptosis, inhibition of site-specific N-terminal phosphorylation of p53 would have the unique property of specifically inactivating p53-mediated apoptosis while preserving p53-induced cell cycle arrest. Preferential activation of cell cycle arrest after DNA damage would increase the resistance of tumor cells to radiation therapy and chemotherapy. This concept may explain why 70% of gliomas and high proportions of other solid tumors contain wild-type p53 alleles despite the fact that inactivation of the p53 protein is a sine qua non of cancer. It is well recognized that p53 protein can be inactivated by posttranslational mechanisms, such as overexpression of MDM-2.35,52,53,55 It is of interest that Thr18 and Ser20 are located at the important p53 domain that interacts with MDM-2. This interaction would prevent p53 N-terminal phosphorylation and thus inactivate the apoptotic function of p53 while potentially preserving the cell cycle arrest function. Whether proteins other than MDM-2 are capable of sequestering p53 from N-terminal phosphorylation or whether increases in levels or activity of phosphatases reduce p53 phosphorylation and thereby inhibit apoptosis without interfering with p53’s cell cycle arrest function are interesting hypotheses that remain to be explored.
This research is supported by grant RO1 CA098716A-01A1 from the National Institutes of Health/National Cancer Institute and by generous donations from the Anthony Bullock III Foundation and the Elias Family Fund for Brain Tumor Research. We thank Stephanie Jenkins for her assistance with preparation of the manuscript.