Resistance to cisplatin (CDDP: cis-diaminedichloroplatinum)-based chemotherapy is a major cause of treatment failure in human ovarian cancer. Chemoresistance is a multifactorial phenomenon, the molecular mechanisms of which are poorly understood. While alterations in DNA platination do not appear to play a significant role in chemoresistance,1 induction of apoptosis is a key effect of CDDP chemotherapy,2 and alterations in the apoptotic capacity are frequently observed and are important determinants of chemosensitivity.3, 4, 5 Indeed, we and others have shown that key regulators of the apoptotic response to CDDP, including X-linked inhibitor of apoptosis protein (XIAP), FLIP, and the MAP kinases (p38, JNK, ERK), are dysregulated in chemoresistant ovarian cancer cells,1, 3, 4, 6, 7, 8 relative to their sensitive counterparts.
The normal function of the p53 tumor suppressor is associated with chemosensitivity and improved clinical outcome in ovarian cancer.3, 9, 10, 11 Indeed, our previous data suggest that p53 is a determinant of CDDP sensitivity in ovarian cancer cells. However, wild-type tumor protein 53 (TP53) alone is not a direct predictor of chemotherapeutic response,3, 4 suggesting that additional mechanisms, unrelated to TP53 genotype, play important roles in regulating CDDP sensitivity.
p53-mediated apoptosis occurs via transcriptional upregulation of gene products such as Bax,12 p53-Upregulated Modulator of Apoptosis (PUMA) and NOXA,13, 14 transcriptional repression of gene products such as Bcl-212 and survivin,15, 16 and via a transcription-independent mechanism whereby p53 directly binds to Bcl-2 and Bcl-XL at the mitochondria.17, 18 Recent evidence suggests that upregulation of PUMA is an important mechanism of CDDP-induced apoptosis.19 However, whether PUMA is sufficient for CDDP-induced apoptosis is not known. Additional transcription-independent mechanisms of p53-induced apoptosis have also been proposed.20 Moreover, p53-mediated apoptosis is dependent upon the phosphorylation of several residues, including Ser15, Ser20 and Ser37.21, 22, 23
Akt is a serine/threonine kinase that is activated in a phosphatidylinositol-3-OH-kinase (PI3K)-dependent manner by growth factors and cytokines, and is implicated in cell proliferation and survival. The PI3K/Akt pathway is frequently overexpressed/activated in ovarian cancers,24, 25, 26, 27 and activation of Akt promotes a chemoresistant phenotype, whereas inhibition of Akt sensitizes chemoresistant cells to CDDP-induced apoptosis.3, 5, 7, 28 However, our previous data also suggest a functional relationship between Akt and p53 in this regard.3
While Akt modulates p53 content by phosphorylating murine double minute-2 (MDM2), thus promoting the ubiquitin-dependent proteolysis of p53,29, 30, 31, 32 Akt can also regulate p53 function independently of changes in p53 content or subcellular localization.33 Since phosphorylation of p53 is required for its proapoptotic effects, it is possible that Akt may suppress apoptosis via this process.
CDDP, cis-diaminedichloroplatinum; DMEM, Dulbecco's modified Eagle medium; DN-Akt, dominant-negative Akt; GAPDH, glyceraldehyde phosphate dehydrogenase; MDM2, murine double minute-2; PI3K, phosphatidylinositol-3-OH-kinase; PMSF, phenylmethylsulfonyl fluoride; PUMA, p53-upregulated modulator of apoptosis; RPMI, Roswell Park Memorial Institute; RT-PCR, reverse transcriptase polymerase chain reaction; SDS-PAGE, sodium dodecyl sulfate–polyacrylamide gel electrophoresis; siRNA, small inhibitory RNA; TP53, tumor protein 53; Xiap, X-linked inhibitor of apoptosis protein.
We investigated the hypothesis that activated Akt inhibits CDDP-induced apoptosis and confers CDDP resistance in cultured ovarian cancer cells, in part, by attenuating p53 phosphorylation and activation of p53-responsive gene products, independently of p53 content. We show that p53 is required for CDDP-induced apoptosis in ovarian cancer cells, and that this is dependent upon the induction of PUMA. Moreover, CDDP induces phosphorylation of p53 on multiple residues in chemosensitive ovarian cancer cells, but not the respective chemoresistant variants. Akt attenuated both PUMA and phospho-p53 upregulation, and conferred chemoresistance. Finally, we show that p53 hypophosphorylation is associated with chemoresistance, and that phosphorylation of Ser15 and Ser20, but not of Ser37, are required for maximal CDDP-induced apoptosis.
Taken together, these data have important implications for our understanding of the molecular mechanisms of chemoresistance in ovarian cancer cells, and in particular, the role of Akt and p53 in this process. Since chemoresistance limits treatment success for this disease, it is critical to understand how resistant cells evade the normal execution of apoptosis.
Material and methods
Cisplatin (CDDP), Hoechst 33258, phenylmethylsulfonyl fluoride (PMSF), sodium orthovanadate (Na3VO4) and aprotinin were purchased from Sigma (St. Louis, MO). Mouse monoclonal antibodies to p53 (DO-1) and Bax (2D2) were from Santa Cruz Biotechnologies (San Diego, CA). Rat monoclonal anti-HA was purchased from Roche (clone 3F10, Palo Alto, CA). Mouse monoclonal anti-MDM2 was from Calbiochem (Ab-1, San Diego, CA). Mouse monoclonal anti-phospho-p53 (Ser15; clone 16G8), rabbit polyclonal anti-phospho-p53 (Ser6, Ser20, Ser33, Ser37, Ser46) and rabbit polyclonal anti-PARP antibodies were from Cell Signaling Technology (Beverly, CA). Rabbit polyclonal anti-PUMA was from Sigma. Mouse anti-glyceraldehyde phosphate dehydrogenase (GAPDH) (ab8245) was from Abcam (Cambridge, UK). Small inhibitory RNA (siRNA) to p53 was purchased from Cell Signaling Technology. siRNA to PUMA was purchased from Santa Cruz Biotechnologies. Control siRNA was from Dharmacon (Lafayette, CO). Ribojuice siRNA transfection reagent was from Novagen (San Diego, CA). Pre-stained sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE) standards were from BioRad (Hercules, CA). Adenoviral dominant-negative Akt was a generous gift from Dr. Kenneth Walsh (Cardiovascular Research, St. Elizabeth's Medical Centre, Boston, MA). HA-tagged wild-type p53 in pcDNA3.1 was a generous gift from Dr. Jin Q. Cheng, Moffitt Cancer Center, University of South Florida, Tampa, FL.
Cell lines and cell culture
Chemosensitive ovarian cancer cells (A2780s and OV2008) and their respective chemoresistant variants (A2780cp and C13*) were cultured as previously reported3, 4, 7, 34 in Dulbecco's modified Eagle medium (DMEM)/F12 and RPMI 1640. OVCAR-3 cells were cultured in RPMI 1640 media. All media were supplemented with 10% FBS, streptomycin (50 g/ml), penicillin (50 U/ml), fungizone (0.625 g/ml; Life Technologies BRL, Carlsbad, CA) and nonessential amino acids (1%). Cells were plated at a density of 5 × 104 cells/cm2 on 6-well plates or 60-mm dishes 18 hr prior to the initiation of treatment. At the time of treatment, cell density was <85%.
HA-tagged wild-type p53 in pcDNA3 was used as a template for site-directed mutagenesis, using the QuickChange Site-Directed Mutagenesis Kit from Stratagene (La Jolla, CA). Primers used to mutate p53 to alanine at the indicated site(s) were as follows: Ser15: Forward 5′-GTCGAGCCCCCTCTGGCACAG GAAACATTTTCAGACC-3′, Reverse 5′-GGTCTGAAAATGTT TCCTGTGCCAGAGGGGGCTCGAC-3′; Ser20: Forward 5′-CT GAGTCAGGAAACATTTGCAGACCTATGGAAACTACTT-3′, Reverse 5′-AAGTAGTTTCCATAGGTCTGCAAATGTTTCCT GACTCAG-3′; Ser37: Forward 5′-TGTCCCCCTTGCCGGCACAAGCAATGGATGATTTG-3′, Reverse 5′-CAAATCATCCA TTGCTTGTGCCGGCAAGGGGGACA; Ser15/Ser20: Forward 5′-GTCGCACAGGAAACATTTGCAGACCTATGGAAACTAC TT-3, Reverse: 5′-AAGTAGTTTCCATAGGTCTGCAAATGTT TCCTGTGCCAG-3′. The template plasmid DNA was amplified using Pfu Polymerase (Fermentas, Hanover, MD) according to the manufacturer's instructions for 16 cycles, digested with DpnI (Fermentas) for 1 hr at 37°C, and then transformed into XL10 Gold cells (Stratagene) and plated onto LB plates containing 100 μg/ml ampicillin. The following day, colonies were picked, amplified overnight at 37°C in LB broth containing 100 μg/ml ampicillin and the plasmid DNA was extracted using the Plasmid Mini Kit from Qiagen (Valencia, CA). The presence of mutations was confirmed by direct sequencing at the OHRI sequencing facility.
Hoechst 33258 staining
At the end of the culture period, cells attached to the growth surface were removed by trypsin treatment [trypsin (0.05%), EDTA (0.53 mM); 37°C, 1 min]. Attached and floating cells were pooled, pelleted by centrifugation and resuspended in phosphate-buffered formalin (10%) containing Hoechst 33258 (12.5 ng/ml). Cells were spotted onto slides for microscopy. Nuclear staining was observed using a Zeiss fluorescence microscope (magnification 400×). Cells with typical apoptotic nuclear morphology (nuclear shrinkage, condensation and fragmentation) were identified and counted as previously reported,3, 4, 7, 35 using randomly selected fields. A minimum of 200 cells were counted in each treatment group. The counter was “blinded” to sample identity to avoid experimental bias. Data are expressed as the percentage of total cells showing apoptotic morphology.
Protein extraction and western blot analysis
Cells were pelleted and lysed in ice-cold lysis buffer (pH 7.4) containing 50 mM Hepes, 150 mM NaCl, 1.5 mM MgCl2, 1 mM EGTA, 100 mM NaF, 10 mM NaPPi, 10% Glycerol and 1% Triton X-100. Protease inhibitors PMSF (1 mM) and aprotonin (10 g/l), as well as 1 mM Na3VO4 were added to the lysis buffer freshly. Cell lysates were sonicated briefly, incubated on ice for 1 hr and pelleted by centrifugation (15,000g; 20 min). The supernatant was taken as whole-cell lysate and stored at −20°C for subsequent analyses. Protein concentration was determined using Bio-Rad DC protein assay kit. Equal amounts of proteins (30–70 μg) were loaded and resolved by 10% SDS-PAGE and electrotransferred (30 V, 16 hr) onto nitrocellulose membranes (Bio-Rad). Membranes were blocked (room temperature, 1 hr) with 5% Blotto [Tris-HCl (10 mM; pH 8.0), NaCl (150 mM), Tween 20 (0.05%, v/v; TBS-Tween 20) containing skim milk (5%; w/v)], then incubated overnight with primary antibodies [p53 (1:1,000), Bax (1:1,000), PUMA (1:1,000), MDM2 (1:2,000) or GAPDH (1:30,000)] and subsequently with the appropriate horseradish peroxidase (HRP)-conjugated secondary antibody [1:2,000 in 5% Blotto; room temperature, 1 hr; 1:20,000 for GAPDH]. Peroxidase activity was visualized with an Enhanced Chemiluminescence Kit (Amersham Pharmacia Biotech, Arlington Heights, IL) after 3 washes (15 min/wash) with TBS-Tween 20. Signal intensity was determined densitometrically using Scion Image software, version 4.02, from Scion Corporation (Frederick, MD). All quantified western blot data were corrected for loading using the anti-GAPDH blots. Western blots shown in figures are representative of at least 3 independent experiments.
Cultured cells were transfected with HA-tagged p53 constructs for 24 hr (as indicated) and then treated with CDDP for a further 24 hr. The cells were lysed in standard lysis buffer (as earlier) for 1 hr on ice, then transferred to a 1.5-ml microcentrifuge tube and centrifuged for 20 min at 14,000g to remove cellular debris. The supernatants were analyzed for total protein content, and 300 μg of total protein was incubated with 15 μl of agarose-immobilized goat polyclonal anti-HA antibody (Bethyl Laboratories, Montgomery, TX) in a final volume of 300 μl, adjusted with lysis buffer. Immunoprecipitation was carried out with gentle rocking, overnight at 4°C. The agarose beads were pelleted by centrifugation at 500g for 2 min, and then washed 3 times with 1 ml lysis buffer, with each wash followed by a 2-min centrifugation at 500g. After the final wash, 30 μl of 2× SDS sample buffer was added to the beads, the samples were boiled and then loaded onto 12% SDS-PAGE gels. Following protein transfer to nitrocellulose, phospho-p53, total p53 and exogenous HA-p53 were detected by western blotting as described earlier.
Reverse transcriptase polymerase chain reaction
Cultured cells were harvested as earlier, and cell pellets were stored at −80°C until further processing. Total RNA was extracted using the RNeasy Mini Kit from Qiagen. An aliquot of total RNA from each sample was subjected to DNase I treatment to remove genomic DNA contamination using the DNA-free kit from Ambion (Austin, TX). mRNA was reverse-transcribed using oligo(dT) primers with Moloney murine leukemia virus (M-MuLV) reverse transcriptase via the Retroscript kit from Ambion. PCR primers were from Invitrogen (Burlington, ON) as follows: PUMA sense: 5′-TGTGACCACTGGCATTCATT-3′; PUMA antisense: 5′-CCTGTAAGATACTGTATATGCGCTGC-3′; β-actin sense: 5′-GGACTTCGAGCAAGAGATGG-3′; β-actin antisense: 5′-CACCTTCACCGTTCCAGTTT-3′. To ensure linearity of the results, the cycle number was optimized by performing PCR reactions at 20–45 cycles. All subsequent PCR reactions were performed within the linear range of amplification. PCRs were performed using HotStarTaq Polymerase from Qiagen. PCR conditions following activation (15 min; 95°C) were: denaturation (94°C) for 45 sec, annealing (PUMA: 56°C, β-actin: 54°C) for 45 sec, extension (72°C) for 30 sec, for 37 or 25 cycles for PUMA and β-actin, respectively. Ten microliters of each PCR product was separated on a 1.5% agarose–ethidium bromide gel and visualized by ultraviolet transillumination using a BioRad GelDoc system.
Genomic DNA sequencing
To confirm TP53 genotype, total genomic DNA from each cell line was extracted using the DNeasy Tissue Kit (Qiagen) and was subjected to exon-specific PCR amplification using primers from the SNPCapture p53 Mutation Screening Kit from Panomics (Fremont, CA). PCR products, corresponding to TP53 genomic DNA from exons 5 to 9 (which encode the DNA binding domain of p53), were visualized by 2% agarose gel electrophoresis and cloned into the pCR4-TOPO sequencing vector from Invitrogen, according to the manufacturer's instructions. Following the transformation and growth of chemically competent E. coli, the plasmid DNA was purified by the Plasmid Miniprep Kit from Qiagen. Isolated plasmid DNA was sequenced by the Dye Terminator method at the Ottawa Genome Centre. Sequences were compared with the published human TP53 sequence using NCBI Blast2. TP53 status was as follows: A2780s-WT; A2780cp-V172F, R260S; OV2008-WT; C13*-WT; OVCAR-3-Q317R.
Cells were infected with an HA-tagged “triple-A” (K179A, T308A, S473A) dominant-negative Akt (DN-Akt) or LacZ cDNA at various multiplicities of infection (MOI) as indicated in the text and previously described.8 As previously reported, adenovirus infection efficiency at MOI of 5, as determined by an X-gal staining assay against LacZ construct infected cells, was >90%.3, 4 DN-Akt expression was confirmed by western blot analysis against the HA epitope tag.
Six microliters of transfection reagent (Novagen) was added to 244 μl of DMEM/F12 without serum. The mixture was vortexed and incubated for 5 min at room temperature. Following incubation, 7.5 μl of 10 μM stock siRNA construct was added. The mixture was incubated at room temperature for a further 15 min. During this period, the culture media was removed from the cells and the cells were washed once with phosphate-buffered saline. The siRNA mixture was added to each well with an additional 1,250 μl of complete (10% FBS) media. The cells were returned to the incubator and the media was removed 6 hr later and replaced with fresh, complete media for the duration of the culture (24–48 hr). Downregulation was confirmed by western blot analysis.
Cells were cultured overnight in 6-well plates and then transfected with 1 μg of pcDNA3.1-derived vectors (empty vector control) using Lipofectamine Plus (Invitrogen) in 1 ml serum-free medium according to the manufacturer's instructions. Three hours post-transfection, each well was supplemented with 1 ml of medium containing 20% FBS. Twenty-four hours post-transfection, media were removed and the cells were harvested or treated as required for a further 24 hr.
DNA binding assay
For the assessment of p53-DNA binding capacity, nuclear lysates of CDDP-treated cells were obtained using the NE-PER Nuclear/Cytoplasmic Extraction kit from Pierce (Rockford, IL), according to the manufacturer's instructions and as previously reported in our laboratory,36 and total nuclear protein contents were determined (as earlier). To assess p53-DNA binding, we used the p53 TransAM kit from ActiveMotif (Carlsbad, CA). Briefly, equivalent amounts of nuclear protein were incubated for 1 hr at room temperature in separate wells of a 96-well plate containing an immobilized p53 consensus oligonucleotide (5′-GGACATGCCCGGGCATGTCC-3′). After 1 hr, plates were washed 3× in fresh 1× wash buffer (supplied with kit), and then incubated for 1 hr at room temperature with anti-p53 antibody. The plates were washed a further 3× with 1× wash buffer, and then incubated 1 hr with HRP-conjugated anti-rabbit antibody. After a further wash step, HRP was detected using 1× developing solution (supplied with kit), and the reaction was stopped with 1× stop solution (supplied with kit) when a light blue color was observed. Plates were immediately read using a UV spectrophotometer at 450 nm. In addition to sample wells, the assay was performed on 3 wells containing only lysis buffer as a negative control. Assay specificity was confirmed by preincubation with free mutant or wild-type oligonucleotide.
Results are expressed as the mean ± SEM of at least 3 independent experiments. Statistical analysis was carried out by one- or two-way ANOVA or by Student's t test (where appropriate) using PRISM software (Version 3.0; GraphPad, San Diego, CA). Differences between multiple experimental groups were determined by the Bonferroni or Tukey post-hoc tests. Statistical significance was inferred at p < 0.05.
p53 is required for CDDP-induced apoptosis in human ovarian cancer cells
While wild-type TP53 is frequently associated with chemosensitivity in human ovarian cancer cells and tumors, ovarian tumors and cell lines bearing wild-type TP53 are often chemoresistant, suggesting that p53 is necessary, but not sufficient, for chemosensitivity, and that additional mechanisms are likely.3, 4, 10
To determine the involvement of p53 in CDDP-induced apoptosis, p53 wild-type, chemosensitive (OV2008) ovarian cancer cells were transfected with siRNA targeting p53 (or control; 0–100 nM; 24 hr) and treated with CDDP (0, 5 μM; 24 hr). As shown in Figure 1a, in the presence of the control siRNA, CDDP upregulated p53 and the p53-responsive gene product PUMA (lane 1 vs. lane 2) and significantly increased the percentage of cells undergoing apoptosis (1.24% ± 0.25% vs. 29.9% ± 3.97%, p < 0.001). By contrast, p53 siRNA downregulated p53 and attenuated the CDDP-induced upregulation of PUMA. Furthermore, CDDP-induced apoptosis was significantly inhibited by p53 siRNA (p < 0.05). This was also observed in the unrelated wild-type p53 chemosensitive ovarian cancer cell line A2780s, in which p53 siRNA downregulated p53 in both DMSO- and CDDP-treated cells and significantly inhibited CDDP-induced apoptosis (10.6% ± 0.63% vs. 3.89% ± 0.54%, p < 0.05; Fig. 1b).
To further examine the role of p53 in CDDP sensitivity, A2780s and their mutant p53 (V172, R260S) chemoresistant counterparts, A2780cp, were cultured in the presence of CDDP (0–10 μM; 24 hr). As shown in Figure 1c, CDDP upregulated p53 and PUMA and induced apoptosis in the A2780s cells, whereas this was significantly attenuated in the A2780cp cells (p < 0.001). The differential induction of apoptosis was confirmed by examining the CDDP-induced cleavage of the caspase-3 substrate poly(ADP) ribose polymerase (PARP).37 A2780s and A2780cp cells were cultured in the presence or absence of CDDP (0, 10 μM; 24 hr). CDDP induced cleavage of intact 116 kDa PARP into the 89 kDa form in the chemosensitive A2780s cells, but not in the chemoresistant A2780cp cells (Fig. 1c; bottom panel). Consistent with the presence of mutant p53 in the latter cell line, total p53 levels were higher in the absence of CDDP compared with the wild-type A2780s cells, likely due to loss of MDM2-mediated negative feedback against p53, which has been previously reported.4, 38
To evaluate whether PUMA is required for CDDP-induced apoptosis, we treated A2780s cells with PUMA or control siRNA (0–50 nM; 48 hr) and CDDP (0, 10 μM; 24 hr). As shown in Figure 1d, PUMA siRNA markedly downregulated PUMA content, without affecting p53, and significantly attenuated CDDP-induced apoptosis (18.1% ± 0.48% vs. 4.68% ± 0.81%, p < 0.001, all effects). These data suggest that PUMA is necessary for CDDP-induced apoptosis.
Taken together, these results demonstrate that p53 is required for sensitivity to CDDP-induced apoptosis in ovarian cancer cells, and that PUMA is necessary for this effect.
PUMA is not sufficient for CDDP-induced apoptosis
To evaluate whether PUMA is sufficient for CDDP-induced apoptosis, we treated chemosensitive OV2008 cells and their chemoresistant, wild-type p53 counterparts, C13*, with CDDP (as earlier) and examined p53 and PUMA contents as well as apoptosis. As shown in Figure 2a, basal p53 and PUMA levels were elevated in the C13* cells, relative to the isogenic, chemosensitive OV2008 cells. However, CDDP upregulated p53 and PUMA and induced apoptosis in the OV2008 cells but not in the C13* cells. We then confirmed this differential induction of apoptosis by examining the CDDP-induced cleavage of PARP. CDDP induced the cleavage of PARP in OV2008 cells but not in C13* cells (Fig. 2a; bottom panel). These results suggest that the presence of PUMA is not sufficient for CDDP-induced apoptosis.
To further evaluate the basal and CDDP-induced nuclear function of p53, we assessed p53-DNA binding. As shown in Figure 2b, CDDP caused the nuclear accumulation of p53 in OV2008 (p < 0.01) and C13* cells (p < 0.05) in response to CDDP, although basal nuclear p53 content was higher in the C13* cells (p < 0.05). Likewise, basal p53-DNA binding was ∼ 5-fold higher in the C13* cells relative to the OV2008 cells, which is consistent with the high basal levels of PUMA in these cells. However, CDDP upregulated p53-DNA binding to a significantly greater extent in OV2008 than in C13* (p < 0.01). Thus, the change in p53-DNA binding was far greater in the chemosensitive cells than in their resistant variants. The specificity of the assay for p53 was confirmed by introduction of free wild-type p53 consensus oligonucleotide, which competed out p53 binding to the immobilized oligo. By contrast, mutated p53 consensus oligonucleotide did not affect p53 binding to the immobilized oligo (Fig. 2c). Thus, C13* cells are resistant to CDDP, despite having high basal p53-DNA binding activity and PUMA content, suggesting that p53 nuclear activity and PUMA upregulation are insufficient to support a CDDP-sensitive phenotype. Thus, it is likely that additional mechanisms of CDDP-induced, p53-dependent apoptosis are at play in these cells.
Activation of Akt attenuates CDDP-induced p53 nuclear function and apoptosis
The PI3K/Akt pathway is frequently activated in human ovarian cancers24, 25, 26, 27 and is a determinant of chemoresistance in human ovarian cancer cells.3, 5, 7, 28 Furthermore, our previous data suggest that Akt-mediated chemoresistance may, in part, be mediated through alterations in p53-mediated apoptosis.3 Thus, we next examined the effects of Akt activation on CDDP-induced p53-mediated gene expression, using A2780s cells stably transfected with empty pcDNA3 vector (A2780-CTL) or pcDNA3 containing constitutively active Akt2 (A2780-AAkt2). The phenotype of these cells has been extensively characterized.3, 5, 28 As shown in Figure 3a, CDDP markedly upregulated p53 and PUMA and induced apoptosis in the control-transfected cells. In contrast, CDDP only slightly upregulated p53 in the A2780-AAkt2 cells, although basal p53 levels were higher in these cells. Despite the presence of p53, active Akt attenuated the CDDP-induced upregulation of PUMA, and significantly inhibited CDDP-induced apoptosis (15.5% ± 0.97% vs. 1.53% ± 0.22% at 10 μM CDDP, p < 0.001). To confirm the antiapoptotic effects of Akt activation, we examined CDDP-induced PARP cleavage in A2780-CTL and A2780-AAkt2 cells. As shown in Figure 3a (bottom panel), CDDP induced PARP cleavage in A2780-CTL cells, and this was completely attenuated in A2780-AAkt2 cells.
To evaluate the kinetics of CDDP-induced PUMA upregulation and the effects of Akt activation on this parameter, we next performed a time-course analysis on PUMA protein content in response to CDDP in A2780-CTL or A2780-AAkt2 cells. We also examined the protein content of MDM2 over the same duration. As shown in Figure 3b, PUMA upregulation occurred between 3 and 6 hr post-CDDP in the A2780-CTL cells. Upregulation of MDM2 occurred even earlier, with marked upregulation within 1–3 hr post-CDDP. Upregulation of these gene products was markedly lower in the A2780-AAkt2 cells, which is consistent with an inactivation of p53-mediated gene transcription in these cells.
We next evaluated the effects of Akt activation on PUMA mRNA content by reverse transcriptase polymerase chain reaction (RT-PCR). A2780-CTL and A2780-AAkt2 cells were cultured in the presence of CDDP (10 μM; 6 hr) and PUMA and β-actin mRNA contents were assessed. As expected, CDDP significantly upregulated the PUMA mRNA in A2780-CTL cells (Fig. 3c; p < 0.05) but not in the A2780-AAkt2 cells. Six hours of culture in the absence of CDDP (DMSO alone) in A2780-CTL cells did not change PUMA mRNA content, relative to the 0 hr group (lane 1 vs. lane 2). This result is consistent with the hypothesis that Akt attenuates the CDDP-induced upregulation of PUMA gene transcription.
To further evaluate the effects of Akt activation on p53 biological function, we evaluated p53 nuclear accumulation and DNA binding in A2780-AAkt2 cells and their control-transfected counterparts. CDDP significantly increased nuclear p53 localization (p < 0.05) and this was not significantly affected by activation of Akt. Furthermore, CDDP induced p53 binding to its consensus oligonucleotide in the A2780-CTL cells, and this was significantly attenuated in the A2780-AAkt2 (Fig. 3d, p < 0.001). Taken together, these data suggest that Akt may in part confer resistance through inhibition of the nuclear functions of p53, including the expression of p53-dependent gene products such as MDM2 and PUMA.
CDDP-induced p53 phosphorylation is attenuated in chemoresistant cells
p53-mediated upregulation of PUMA is necessary, but not sufficient, for CDDP-induced apoptosis. We therefore evaluated the hypothesis that phosphorylation of p53, which is required for p53-mediated apoptosis,21 is dysregulated in chemoresistant cells in response to CDDP. While p53 is phosphorylated on numerous residues, the precise function of these phosphorylations remains unclear. Several reports suggest that phosphorylation is required for protection of p53 from MDM2-mediated ubiquitination,39, 40, 41, 42, 43 and phosphorylation of Ser15 and Ser20 plays an important role in the transduction of p53-mediated apoptosis,21 although Ser15/Ser20 phosphorylation is dispensible for p53-dependent apoptosis in some systems.44 Thus, we next examined the hypothesis that alterations in CDDP-induced p53 phosphorylation are important determinants of chemoresistance in ovarian cancer cells. OV2008 and C13* cells were treated with CDDP (as earlier) and we determined total and phospho-p53 contents by western blot using phospho-specific antibodies targeted to Ser6, Ser15, Ser20, Ser33, Ser37 and Ser46. As shown in Figure 4a, and consistent with results shown in Figure 1, basal p53 was higher in C13* cells and CDDP upregulated p53 in both OV2008 and C13* cells, although the effects were much more pronounced in the chemosensitive OV2008 cells. However, when phospho-p53 content was analyzed, we found that while CDDP effectively induced phosphorylation of all examined residues except Ser6 in OV2008 cells, phosphorylation on these sites was markedly attenuated in C13* cells (except for Ser6), despite the presence of comparable total p53 levels in both cell lines (Fig. 4a, top panel). Ser46 phosphorylation was undetectable in both OV2008 and C13* cells. This result suggests that CDDP-induced p53 phosphorylation is attenuated in chemoresistant cells, and is consistent with the hypothesis that aberrant regulation of p53 phosphorylation may be a causative factor in chemoresistance in human ovarian cancer cells. To confirm that this was not a cell-type specific effect, we further examined the CDDP-induced p53 phosphorylation in A2780s cells and their chemoresistant counterpart, A2780cp. CDDP effectively induced p53 phosphorylation (all sites examined) in the sensitive A2780s cells. By contrast, while Ser6, Ser33 and Ser37 were phosphorylated in A2780cp cells to similar extent as in the chemosensitive A2780s cells, phosphorylation of Ser15 and Ser20 was markedly attenuated in the resistant cells, despite high total p53 content (Fig. 4, bottom panel). Ser46 phosphorylation was slightly, although not completely, attenuated in the chemoresistant cells. Taken together, these results demonstrate that p53 phosphorylation, particularly of Ser15 and Ser20, is markedly attenuated in chemoresistant ovarian cancer cells, and support the hypothesis that CDDP resistance may, in part, be mediated through aberrant regulation of phosphorylation at these sites.
Akt attenuates CDDP-induced p53 phosphorylation
While Akt activation can attenuate p53-mediated gene expression and promote a chemoresistant phenotype, whether p53 phosphorylation is affected is unknown. To test this possibility, we cultured A2780-CTL and A2780-AAkt2 cells in the absence and presence of CDDP (as earlier) and phospho- and total p53 contents were analyzed by western blot. As shown in Figure 5a, while CDDP upregulated total and phospho-p53 contents (Ser6, Ser15, Ser20, Ser33 and Ser37) in a concentration-dependent manner in the control A2780-CTL cells, phosphorylation at these sites (except Ser6) was markedly attenuated in the A2780-AAkt2 cells. This result demonstrates that Akt activation inhibits p53 phosphorylation, suggesting a novel mechanism of Akt-mediated cell survival. Furthermore, these findings are consistent with the hypothesis that downregulation of p53 phosphorylation may, in part, contribute to Akt-mediated chemoresistance in ovarian cancer cells.
To further examine the role of Akt on p53 phosphorylation, chemoresistant, p53 wild-C13* cells were infected with adenoviral DN-Akt (MOI = 0–80, 48 hr; LacZ control). This adenovirus encodes a kinase-dead mutant form of Akt (K179A), which also cannot be activated by phosphorylation (T308A, S473A). We previously demonstrated that expression of DN-Akt sensitizes chemoresistant cells to CDDP-induced apoptosis in a p53-dependent manner.3 DN-Akt expression was confirmed by western blot against the HA epitope (Fig. 5b). Expression of DN-Akt upregulated total p53, and this p53 was phosphorylated on Ser15, Ser20 and Ser37. Moreover, while DN-Akt upregulated total p53 by 1.5-fold (1.5 ± 0.02 at MOI = 80 relative to MOI = 0), phospho-p53 (Ser15) was upregulated by nearly 10-fold (9.8 ± 1.2 at MOI = 80 relative to MOI = 0), which represents a significant increase in the ratio of phospho/total p53 of nearly 7-fold (6.8 ± 0.84 at MOI = 80 relative to MOI = 0, p < 0.0001).
To further confirm the effects of DN-Akt on phospho-p53 content, chemoresistant A2780cp cells were infected with DN-Akt (MOI = 0, 80; as earlier) and total and phospho-p53 (Ser15) contents were evaluated by western blot. Since Akt can induce an MDM2-dependent downregulation of p53,29, 30, 31, 32 and since mutant p53 does not transactivate the MDM2 gene, resulting in reduced MDM2 protein content,4 we surmised that DN-Akt should not increase total p53 in p53 mutant cells, thus providing an ideal system to study the effects of Akt inhibition on p53 phosphorylation. Expression of DN-Akt was confirmed by detection of the HA epitope. As expected, DN-Akt did not upregulate p53 in these cells, but increased phosphorylation of Ser15 was observed in cells infected with DN-Akt, relative to the LacZ-infected control cells (Fig. 5c, upper panel). Indeed, DN-Akt increased the ratio of phospho/total p53 by greater than 2.5-fold, relative to the LacZ-infected cells (Fig. 5c, lower panel, p < 0.01), demonstrating that suppression of endogenous Akt activity upregulates p53 phosphorylation.
Taken together, these results suggest that Akt inhibits p53 phosphorylation and offer a novel paradigm for Akt-mediated chemoresistance and cell survival.
Phosphorylation of Ser15 and Ser20 is required for optimal CDDP-induced, p53-dependent apoptosis
To test whether p53 phosphorylation is a determinant of CDDP-induced apoptosis, we mutated p53 such that it could not undergo phosphorylation (Ser to Ala) at key sites (Ser15, Ser20 or Ser37). We then transiently transfected CDDP-resistant, p53 mutant (Q317R)45 OVCAR-3 cells with HA-tagged WT, S15A, S20A, S37A-p53 or a combined S15A/S20A-p53 (pcDNA3.1 empty vector control; 24 hr) and then treated the cells with CDDP (10 μM; 24 hr). OVCAR-3 cells were chosen because a previous report suggested that reconstitution of wild-type p53 was sufficient to sensitize these cells to CDDP,46 whereas we have demonstrated that this is not the case in other p53 mutant ovarian cancer cell lines, such as A2780cp.3, 4 All of the p53 constructs were expressed to equivalent levels in the absence of CDDP, as measured by western blot against the HA epitope, and this was not affected by treatment with CDDP (Fig. 6a, top panel). CDDP alone failed to induce apoptosis in OVCAR-3 cells (Fig. 6a). Likewise, expression of the p53 constructs alone did not increase the number of apoptotic cells, relative to the control vector, providing further evidence that p53 is not sufficient for the induction of apoptosis in ovarian cancer cells. However, transfection with wild-type p53 significantly sensitized the cells to CDDP-induced apoptosis (p < 0.001). Furthermore, this effect of wild-type p53 was significantly, although not completely, blocked by mutation of Ser15 or Ser20 to alanine (p < 0.001), whereas mutation of Ser37 to alanine had no effect on p53-mediated sensitization. This result strongly supports the hypothesis that p53 phosphorylation is required for CDDP sensitivity, and suggests that Ser15 and Ser20 are key amino acids involved. Interestingly, double mutation of Ser15 and Ser20 to alanine did not significantly reduce CDDP-induced apoptosis relative to the S15A mutant (6.7% ± 0.85% vs. 4.7% ± 1.5%, p > 0.05) but did significantly attenuate CDDP-induced apoptosis relative to S20A (9.4% ± 1.0% vs. 4.7% ± 1.5%, p < 0.01), suggesting that Ser15 may be the more critical residue in this regard. Furthermore, while CDDP alone did not upregulate PUMA in OVCAR-3 cells, all of the HA-p53 constructs upregulated PUMA to similar levels, further suggesting that the observed dependence of CDDP-induced, p53-mediated apoptosis on p53 phosphorylation is not mediated through changes in PUMA expression, which is consistent with the results obtained in C13* cells (Fig. 2).
To confirm that these constructs were phosphorylated, and that mutation to alanine prevents these phosphorylations, we transfected OVCAR-3 cells with the HA-p53 constructs followed by treatment with CDDP (10 μM), and then immunoprecipitated HA-p53 using an agarose-immobilized goat polyclonal anti-HA antibody (goat IgG control), followed by western blot against total and phospho-p53, using specific antibodies. We detected both exogenous and endogenous p53 when the immunoprecipitates were probed for total p53 (Fig. 6b). While endogenous p53 was phosphorylated on Ser15 and Ser20 in all groups (Fig. 6b; lower bands; lanes 2–6), mutation to alanine blocked phosphorylation in the exogenous p53 (Fig. 6b; upper bands). This occurred for both Ser15 and Ser20. Taken together, these data strongly support the hypothesis that phosphorylation of p53 on Ser15 and Ser20 is required for CDDP-induced apoptosis.
Apoptosis is a key determinant of chemosensitivity in ovarian cancer.2, 3, 4, 5, 7 While p53 is an important regulator of apoptosis, its precise role in determining CDDP sensitivity in ovarian cancer cells is unclear. TP53 mutations are common in ovarian cancer and are often associated with chemoresistance in tumors and cultured cells.2, 9, 10, 47, 48, 49 While we previously demonstrated that p53 is an important regulator of cell fate in ovarian cancer cells,3, 4, 34 a direct correlation between p53 mutation and chemoresistance has not been demonstrated.50 Thus, it is of interest to determine if and how p53 contributes to the CDDP sensitivity of ovarian cancer cells.
CDDP upregulated p53 in chemosensitive ovarian cancer cells and downregulation of p53 attenuated CDDP-induced apoptosis in wild-type p53, chemosensitive ovarian cancer cells, suggesting that endogenous p53 is a key determinant of CDDP-induced apoptosis. Song et al. previously showed that reintroduction of wild-type p53 in mutant p53 ovarian cancer cells sensitized them to CDDP-induced apoptosis,51 while Vasey et al. showed that expression of dominant-negative p53 conferred a chemoresistant phenotype.52 However, these studies dealt with the effects of exogenous p53, and did not address whether physiological levels of p53 are required for CDDP-induced apoptosis. Thus, the current report builds upon these studies and provides evidence that endogenous p53 is required for the sensitivity of ovarian cancer cells to CDDP-induced apoptosis.
CDDP-induced apoptosis was associated with PUMA upregulation in 2 unrelated, p53 wild-type chemosensitive ovarian cancer cell lines (A2780s and OV2008). Furthermore, this was inhibited in cells treated with p53 siRNA and in the p53 mutant cell line A2780cp, which is consistent with a p53-dependent mechanism of upregulation. Downregulation of PUMA significantly attenuated CDDP-induced apoptosis. PUMA knockout mice are resistant to p53-dependent apoptosis and show a similar apoptotic phenotype to p53 knockout mice,13 and our results are consistent with a central role for PUMA in CDDP-induced apoptosis. We cannot exclude the possibility that other p53-responsive genes, such as Bax, may also contribute. However, Bax was not affected by CDDP in the chemosensitive OV2008 cells (data not shown), suggesting that its p53-mediated upregulation may not be a major contributing factor in these cells. Whether other p53-responsive gene products (e.g. p53AIP1, PTEN, Fas) play a role in these processes is not known.
A very recent study has shown that p53-induced PUMA is required for CDDP-induced apoptosis in renal cells,19 and our data suggest a similar requirement in ovarian cancer cells. Furthermore, our observations that PUMA upregulation is mediated through p53, and that PUMA or p53 downregulation significantly inhibits CDDP-induced apoptosis strongly suggest that p53 is a critical mediator of CDDP-induced apoptosis in ovarian cancer cells, a finding that has been disputed in some studies. However, p53/PUMA is not sufficient for CDDP-induced apoptosis in these cells, as previously shown in neurons,53 suggesting that this may be context-specific. We previously showed that sensitization of C13* cells to CDDP-induced apoptosis is dependent on p53.3, 4 Since chemoresistance was associated with a failure of CDDP to increase the DNA binding activity of p53 or to upregulate PUMA in these cells, one intriguing possibility is that cells expressing high levels of p53 activity, such as these, are selected during the development of chemoresistance, have adapted to living under these normally lethal conditions, and thus do not respond to CDDP by further activating p53. This is supported by data showing that p53 levels are generally much higher in chemoresistant cells.54 Taken together, these data suggest that while upregulation of PUMA is a mediator of CDDP-induced, p53-dependent apoptosis, additional mechanisms likely contribute.
Akt activation confers resistance of ovarian cancer cells to CDDP-induced apoptosis, whereas suppression of Akt sensitizes chemoresistant cells to CDDP in a p53-dependent manner,3, 5, 28 suggesting a functional link between Akt-mediated chemoresistance and p53. While several studies have shown that Akt can facilitate p53 degradation via phosphorylation of MDM2,29, 30, 31, 32 we observed a slight upregulation, rather than a reduction, in p53 content in cells expressing activated Akt relative to the control cells. This may have been due to the slight downregulation of basal MDM2 content in cells expressing activated Akt (Fig. 4b). However, a marked reduction of CDDP-induced p53-DNA binding capacity, upregulation of PUMA, and apoptosis was observed in response to Akt activation, suggesting that Akt attenuates p53 function rather than its content.33 It has recently been demonstrated that Akt activation blocks proteasomal degradation of p300,55 and p300 is an inhibitor of DNA damage-induced p53-dependent PUMA expression and apoptosis.56 This is one possible mechanism by which Akt may inhibit the activation of p53 in these cells, although this hypothesis has not been evaluated.
p53 is phosphorylated in response to various cell stresses, and p53 phosphorylation protects p53 from MDM2-mediated ubiquitination and proteasomal degradation. Phosphorylation of p53 on Ser15 and/or Ser20 is required for p53-induced apoptosis,21 although this does not hold true under all circumstances.44 CDDP induced phosphorylation on numerous p53 residues, including Ser15 and Ser20, in 2 chemosensitive ovarian cancer cell lines, but not in their respective chemoresistant variant cell lines, suggesting that altered p53 phosphorylation may contribute to the chemoresistance of these cells. Moreover, these sites are required for CDDP-induced apoptosis. To our knowledge, this is the first demonstration that p53 phosphorylation is altered in chemoresistant cells and is required for CDDP-induced apoptosis, and provides compelling evidence that aberrant regulation of phosphorylation may be a critical determinant of the sensitivity of human ovarian cancer cells to CDDP-induced apoptosis. We cannot exclude the possibility that phosphorylation of sites other than Ser15 and Ser20 (including those in the p53 C-terminal) may also be implicated in this process, although it appears that Ser37 does not play a major role in CDDP-induced apoptosis. Indeed, Ser46 is an important phosphorylation site for the regulation of p53-induced gene expression and apoptosis.57 However, Ser46 phosphorylation was undetectable in OV2008 or C13* cells exposed to CDDP. Moreover, while Ser46 phosphorylation was detected in A2780s cells, it was also observed in the chemoresistant A2780cp cells. These data suggest that Ser46 may not be required for CDDP-induced, p53-dependent apoptosis in ovarian cancer cells.
The mechanism by which Ser15/Ser20 phosphorylation alters the apoptotic capacity of p53 remains unclear, though alterations in the phosphorylation of these sites did not affect p53-induced PUMA upregulation. Unger et al. showed similar results with respect to Bax activation.21 However, the possibility that phosphorylation of these sites may alter the expression of additional p53-responsive genes, thereby promoting p53-mediated apoptosis, requires further investigation.
While our results suggest that Ser15 phosphorylation may not be required for p53-dependent transactivation in ovarian cancer cells, our results agree with the observation that Ser15 is not required for maintenance of p53 steady-state levels.58 We recently demonstrated that CDDP induces the direct targeting of p53 to the mitochondria in chemosensitive cells, but not in their resistant counterparts,59 and that p53 targeted to the mitochondria was able to induce apoptosis more rapidly than nuclear-targeted p53. Thus, phosphorylation may be implicated in the mitochondrial translocation of p53. While recent data suggest that phosphorylation is not the determining factor for the alternate translocation of p53 to the nucleus or the mitochondria,60 whether phosphorylation is required for mitochondrial translocation has not been evaluated. p53 also directly activates Bax, an event that requires its PUMA-dependent liberation from a p53-Bcl-XL complex.20 Thus, it is possible that this process requires phosphorylation of p53 on 1 or more residues. These hypotheses are currently under investigation in our laboratory.
Akt activation reduced the CDDP-induced phosphorylation of p53 on several residues while dominant-negative Akt upregulated p53 and induced p53 phosphorylation independently of changes in content. Interestingly, Yamaguchi et al. did not observe a change in p53 phosphorylation in response to Akt activation.33 However, they used a [32P]-orthophosphate labeling assay, which may not be sensitive enough to detect changes in phosphorylation at individual amino acids in polyphosphorylated molecules such as p53. In contrast, our western blot assay permits an analysis of phosphorylation status at specific residues. While this manuscript was in preparation, Limesand et al. reported that activated Akt1 in murine salivary acinar cells attenuates p53 phosphorylation at Ser18 (equivalent to human Ser15).61 We have significantly extended these findings by demonstrating that activated Akt inhibits the phosphorylation of numerous p53 residue (in addition to Ser15), and, importantly, that endogenous Akt inhibits p53 phosphorylation.
The mechanism by which Akt regulates p53 phosphorylation is unclear. CDDP-induced phosphorylation of Ser6 was not affected by Akt activation, suggesting that the actions of Akt in this respect may be distal to DNA damage itself. Moreover, the phosphorylations attenuated by Akt activation are regulated by diverse kinases, suggesting that Akt likely acts at a downstream step. One candidate target molecule is ATR, which is activated by CDDP, and which directly and indirectly facilitates the phosphorylation of p53 on numerous residues, including Ser15, Ser20 and Ser37. Intriguingly, ATR contains a putative Akt phosphorylation site (RRRLSS436). As such, Akt may phosphorylate ATR, thereby attenuating its activation, and reducing ATR-dependent p53 phosphorylation. We are currently evaluating this hypothesis in our laboratory.
The contribution of p53 to chemosensitivity remains unclear, although several studies2, 9, 10, 49, 62 have demonstrated a strong correlation between wild-type TP53 and sensitivity to CDDP and prolonged survival in human ovarian cancer. In addition, p53 is required for CDDP-induced apoptosis in ovarian cancer cell culture and xenografts.3, 11, 51 However, other studies fail to show a correlation between wild-type p53 and CDDP sensitivity.63, 64, 65 Many of these studies have relied upon nonspecific ablation of p53 function (e.g. expression of HPV-E6) and/or did not specifically distinguish between CDDP-induced apoptosis and necrosis. This is of concern since apoptosis is a primary effect of CDDP-centered chemotherapy in ovarian cancer patients and in xenograft models of human ovarian cancer.2, 48 Moreover, because specific TP53 mutations may affect p53 function differently, a detailed analysis of the relationship between particular mutations and chemoresistance is warranted, and may shed light upon the role of p53 in chemoresistance.
The current study suggests that CDDP upregulates PUMA via activation of p53, thereby facilitating apoptosis. Furthermore, CDDP induces p53 phosphorylation on several residues, including Ser15 and Ser20, which are not phosphorylated in chemoresistant cells and are required for efficient CDDP-induced apoptosis. Akt effectively blocks these processes, thereby conferring resistance to CDDP-induced apoptosis. Either mutation of the TP53 gene, which inhibits the upregulation of PUMA, or failure to phosphorylate wild-type p53, or both, results in a chemoresistant phenotype; both events are required for the full apoptotic response to CDDP. This is consistent with the observation that wild-type TP53 status is not always correlated with chemosensitivity. In addition, while other p53-independent cellular events, including the downregulation of Xiap, may be required for CDDP-induced apoptosis, our evidence suggests that effective activation and phosphorylation of p53 is essential. Thus, it will be of interest to study the effects of chemotherapy on total and phospho-p53 and PUMA contents in human ovarian tumors with respect to treatment outcomes. Moreover, since Akt attenuates both processes, it is important to study the relationship between activation/overexpression of Akt in ovarian tumors and sensitivity to CDDP.
In summary, we have demonstrated that p53 is essential for CDDP-induced apoptosis in human ovarian cancer cells, and that this is mediated, at least in part, through the upregulation of PUMA and the phosphorylation of p53 on Ser15 and Ser20. Furthermore, p53 phosphorylation is attenuated in chemoresistant cells, suggesting a novel mechanism of chemoresistance. Activation of Akt confers resistance by blocking p53-mediated transactivation and p53 phosphorylation. A more thorough understanding of the molecular mechanisms underling chemoresistance in human ovarian cancer may ultimately improve treatment outcomes for this disease.
The authors thank Dr. Mohammed Abedini for the immunoprecipitation protocols. B.K.T. is the recipient of a grant from the National Cancer Institute of Canada (013335), with funds from the Canadian Cancer Society, and from the Canadian Institutes of Health Research (MOP-15691). M.F. was the recipient of a Canada Graduate Scholarship Doctoral Research Award from the Canadian Institutes of Health Research.