p53/p21^CIP1 Cooperate in Enforcing Rapamycin-induced G₁ Arrest and Determine the Cellular Response to Rapamycin¹

The mammalian target of rapamycin, mTOR [also designated FRAP, RAFT1, or RAPT1 (1, 2, 3)], has been shown to link mitogen stimulation to protein synthesis and cell cycle progression. Both phosphatidylinositol 3′-kinase and, potentially, AKT/PKB considered to lie upstream of mTOR, can protect cells from apoptosis induced by stress. However, a role for mTOR in cell survival has not been established. To examine the potential role of mTOR in tumor cell survival, we used the macrocyclic lactone antibiotic rapamycin that potently inhibits mTOR. Rapamycin competes with a structural analogue, FK-506, for binding to a M_r 12,000 cytosolic protein designated FKBP-12. The FKBP-rapamycin complex inhibits the function of a serine/threonine kinase, mTOR, blocking growth factor stimulation of ribosomal p70S6 kinase, and phosphorylation of the eIF4E⁴ binding protein (also designated PHAS-I). 4E-BP1 is a direct substrate for mTOR kinase activity in cells (4, 5), whereas certain evidence indicates the potential for intermediate steps in the mTOR-mediated activation of p70S6 kinase (6, 7, 8). In growth factor-deprived cells, association of eIF4E with the multifunctional scaffolding protein eIF-4G is inhibited by 4E-BP1 and 4E-BPII (9, 10). Upon growth factor or serum stimulation, phosphorylation occurs leading to dissociation of 4E-BP proteins, assembly of the multisubunit complex (eIF-4F), and efficient translation of mRNA having highly structured 5′-untranslated regions (11). In many cell lines, exposure to rapamycin results in a relatively small decrease in overall protein synthesis (∼15–20%), but results in a specific G₁ cell cycle arrest. We have shown previously that rhabdomyosarcoma cells are highly sensitive to growth inhibition by rapamycin (12), and under serum-free conditions rapamycin induced p53-independent apoptosis (13).

The tumor suppressor p53 is a transcription factor that has been found to be mutated in >50% of human cancers (14, 15). In response to genotoxic damage, such as exposure to chemotherapeutic agents, γ-irradiation, or UV irradiation, p53 is up-regulated (16, 17, 18). Up-regulation of p53 either results in arrest of cells in G₁ phase and participates in DNA repair (18, 19, 20, 21) or drives cells toward apoptosis (19, 22, 23). The functions of p53 are dependent on cell type and developmental stage (24), but the mechanisms by which p53 exerts its functions is still not fully understood (25). Many reports show that p53 functions as a tumor suppressor by promoting apoptosis (15, 25). On the other hand, recent studies also demonstrate that p53 may retard tumor progression by another mechanism based on irreversible growth arrest which may lead to senescence (26, 27).

p53-dependent cell cycle arrest is in part a consequence of up-regulation of p21^CIP1, a p53-inducible gene (28, 29, 30). p21^CIP1 inhibits G₁ cdk which phosphorylate pRb and related family members, leading to a G₀-G₁ arrest of the cell cycle (31, 32, 33, 34). Interestingly, recent findings further reveal that p21^CIP1-induced cycle arrest in G₁ phase protects cells from apoptosis induced by ionizing radiation or chemical exposure (35, 36, 37, 38, 39, 40, 41, 42). Down-regulation of p21^CIP1 using antisense oligonucleotides has also been shown to radiosensitize cells by converting growth arrest to apoptosis (43). In contrast to the role of p53 in response to DNA damage, p53 appears also to act as a sensor, causing G₁ arrest of cells prior to damage. In cells treated with the antimetabolite N-(phosphonacetyl)-l-aspartic acid, an inhibitor of carbamoylphosphate synthetase that depletes cellular pools of both purines and pyrimidines p53 initiates a G₁ block prior to initiation of DNA replication (44, 45).

The observation that tumor cells with mutated p53 undergo apoptosis when exposed to rapamycin, in contrast to G₁ arrest, prompted us to investigate whether p53 acted as a sensor to prevent G₁ progression in drug-treated cells in which synthesis of a subset of proteins had been inhibited. Our results reveal that rapamycin-induced death is a consequence of continued cell cycle progression despite inhibition of mTOR function. The results suggest that p53 cooperates in blocking G₁ progression in rapamycin-treated cells and that protection from drug-induced apoptosis requires p21^CIP1.

Rh1 and Rh30 human rhabdomyosarcoma cell lines have been described previously (13, 46). Both express mutant p53 alleles (Rh1: Tyr²²⁰ → Cys²²⁰; Rh30 Arg²⁷³ → Cys²⁷³) and were grown in antibiotic-free RPMI 1640 supplemented with 10% FBS and 2 mm l-glutamine at 37°C and 5% CO₂. For experiments where cells were deprived of serum overnight, cell monolayers were washed with RPMI 1640 containing 2 mm l-glutamine and incubated in the same medium. For prolonged serum-free conditions, Rh30 cells were cultured in MN2E (DMEM/F12 supplemented with 1 μg/ml human transferrin, 30 nm sodium selenate, 20 nm progesterone, 100 μm putrescine, 30 nm vitamin E phosphate, 50 μm ethanolamine, and 1 mg/ml BSA). Rh1 cells were grown in MN2E with addition of fibronectin (10 μg/ml). Rh1 and Rh30 clones that stably express an AU-1-tagged mutant mTOR cDNA have been described previously (13). This mutant (designated mTOR-rr) has a single amino acid substitution (Ser²⁰³⁵ → Ile²⁰³⁵) in the FKBP-rapamycin binding domain that reduces the binding affinity of the FKBP-rapamycin complex (47). Wild-type and p53^−/− cells were obtained from Tyler Jacks (Massachusetts Institute of Technology, Cambridge, MA) and used within five passages. Wild-type and p21^−/− MEFs were kindly provided by Charles Sherr (Howard Hughes Medical Institute, St. Jude Children’s Research Hospital). For prolonged serum-free conditions, cells were cultured in MN2E as described above.

Replication-deficient (ΔE1A and E3) adenoviral recombinants (Genetic Therapy, Inc., Gaithersburg, MD) expressing β-galactosidase (Ad-β-gal) (48), wild-type p53 (Ad-p53; from Linda Harris, St. Jude Children’s Research Hospital), or wild-type human p21^CIP1 (Ad-p21; from Wafik El-Deiry, University of Pennsylvania, Philadelphia, PA) were prepared using 293 cells by conventional procedures. The adenoviral recombinants were titered by a plaque-forming assay after infection of 293 cells as described elsewhere (48). All infections were performed firstly in 2% FBS/DMEM for 2 h, then continued in 10% FBS/DMEM for 22 h. The infectivity of the cell lines was compared by infecting cells with Ad-β-gal and then staining with 5-bromo-4-chloro-3-indolyl-β-d-galactopyranoside as reported previously (48). The MOI was defined as the ratio of the total number of plaque-forming units used in a particular infection per total number of cells to be infected. Controls included Ad-β-gal infection and mock infection in which the cells were incubated with corresponding medium only.

Cells (Rh1, 8.5 × 10⁵; Rh30, 1.7 × 10⁶/162-cm² flask) were grown overnight in serum-free N2E medium. On day 1 rapamycin (100 ng/ml) was added, and cells were exposed for up to 6 days. Control cells in RPMI 1640 containing 10% FBS or N2E were grown for the corresponding periods without addition of rapamycin (100 ng/ml). Cells were trypsinized, washed with PBS, resuspended in 200 μl of binding buffer (Clontech Laboratories, Inc., Palo Alto, CA), and incubated with 10 μl of Annexin V^FITC (final concentration 1 μg/ml; Clontech Laboratories, Inc.) and 500 ng of propidium iodide in a final volume of 410 μl. Cells were incubated at room temperature in the dark for 10 min before flow cytometric analysis (FACSCalibur, Becton Dickinson, Mountain View, CA) as described previously (13).

Cultured cells were briefly washed with PBS and trypsinized. Cell suspensions were centrifuged at 1000 rpm for 5 min and pellets were resuspended in propidium iodide staining solution (50 ng/ml propidium iodide, 0.1% sodium citrate, and 0.1% Triton X-100) at 1 × 10⁶ cells/ml. The cells were then pretreated with RNase (DNase-free) solution (0.2 mg/ml RNase in 15 mm NaCl, 10 mm Tris-HCl, pH 7.5) and filtered through 40-μm diameter mesh to remove clumps of nuclei. Percentages of cells within each of the cell cycle compartments (G₀-G₁, S, or G₂-M) were determined by flow cytometry (FACSCalibur; Becton Dickinson).

Cultured cells were briefly washed with cold PBS and on ice lysed in RIPA buffer (150 mm NaCl, 1% sodium deoxycholate, 0.1% SDS, 1% Triton X-100, and 50 mm Tris, pH 7.2) containing 1 mm phenylmethylsulfonyl fluoride, 5 μg/ml aprotinin, and 5 μg/ml leupeptin. Lysates were cleared by centrifugation at 14,000 rpm for 15 min at 4°C. Protein concentration was determined by the bicinchoninic acid assay using BSA as the standard (Pierce, Rockford, IL). Equivalent amounts of protein were separated on a 12% SDS-PAGE and transferred to polyvinylidene difluoride membranes (Immobilon polyvinylidene difluoride; Millipore, Bedford, MA). Membranes were blocked with PBS containing 0.05% Tween 20 and 5% nonfat dry milk and probed with rabbit polyclonal anti-p21^CIP1 (1:800; Santa Cruz Biotechnology, Inc., Santa Cruz, CA), polyclonal anti cyclin A or E (1:500; Santa Cruz Biotechnology, Inc.) followed by incubation with goat anti-rabbit IgG-conjugated horseradish peroxidase (1:16,000; Sigma, St. Louis, MO). For detection of cyclin D1 a monoclonal antibody (1:500; PharMingen, San Diego, CA) was used, followed by goat antimouse IgG-conjugated horseradish peroxidase (1:20,000; Pierce). Immunoreactive bands were visualized using Renaissance chemiluminescence reagent (NEN; Life Science Products, Inc., Boston, MA). To check the protein loading, the immunoblots were treated with stripping solution (62.5 mm Tris buffer, pH 6.7, containing 2% SDS and 100 mm β-mercaptoethanol) for 30 min at 50°C and reprobed with mouse monoclonal anti-β-tubulin antibody (1:2,000; Sigma), followed by incubation with horseradish peroxidase-coupled rabbit antimouse IgG (1:8,000; Sigma).

Rh30 cells (4 × 10⁴/well) were seeded on 2-well glass chamber slides (Nunc, Naperville, IL) in 10% FCS-RPMI 1640 and infected with Ad-β-gal or Ad-p53, followed at 24 h by treatment with or without rapamycin (100 ng/ml) in MN2E medium for 24 h. BrdUrd (20 μm; Sigma) was then added and cells were incubated for an additional 5 days. To control for potential artifactual staining of apoptotic nuclei by anti-BrdUrd antibody, aphidicolin was added to control or rapamycin-treated cells prior to addition of BrdUrd. Cells were briefly washed with PBS and fixed with 70% ethanol (in 50 mm glycine buffer, pH 2.0) for 30 min at −20°C. Incorporation of BrdUrd into nuclear DNA was detected using an immunofluorescence assay kit (Boerhinger Mannheim GmbH, Mannheim, Germany). After washing with PBS, slides were incubated with mouse monoclonal anti-BrdUrd antibody containing nucleases (1:100, 66 mm Tris buffer, 0.66 mm MgCl₂, and 1 mm 2-mercaptoethanol) for 30 min at 37°C, followed by washing with PBS and immunostaining with sheep antimouse IgG-FITC conjugates (1:100, diluted in PBS) for 30 min at 37°C. To visualize nuclear morphology, slides were further counterstained with DAPI (4 μg/ml in deionized water; Sigma) for 5 min at room temperature. Following a brief washing with PBS, slides were mounted in glycerol/PBS (1:1, v/v) containing 2.5% 1,4-diazabiclo-(2,2,2)octane (Sigma) and photographed by fluorescence microscopy using Insight software. Data were statistically analyzed with a two-tailed paired Student’s t test. The labeling index (cells stained for BrdUrd/total cells) was calculated from counting approximately 1500 cells.

Cells were infected with adenovirus vectors expressing β-gal, p53, or p21^CIP1 for 24 h as described above, then incubated another 24 h in N2E medium with or without rapamycin (100 ng/ml). After two brief washes with cold PBS, cells were lysed in RIPA buffer. The lysates were cleared of insoluble material by centrifugation (14,000 rpm, 10 min, 4°C). Protein concentration in the supernatants was determined by the bicinchoninic acid assay using BSA as the standard (Pierce). Equivalent amounts of protein (300 μg/sample) were immunoprecipitated at 4°C with rabbit anti-cdk2, cdk4, or cdk6 polyclonal antibodies (all from Santa Cruz Biotechnology, Inc.) and protein A-coupled agarose beads (Santa Cruz Biotechnology, Inc.) for 4 h. The immunoprecipitates were washed twice with RIPA buffer, followed by two washes with fresh kinase buffer (20 mm Tris, pH 7.4, 10 mm MgCl₂, 1 mm DTT, and 10 mm β-glycerophosphate). Phosphorylation reactions were initiated by addition of 50 μl of kinase buffer containing 10 μCi of [γ-³²P]ATP (6000 Ci/mmol; Amersham, Arlington Heights, IL), 20 μm ATP, and 0.5 mg of histone H1 (Sigma, for cdk2 assay) or 0.5 μg of recombinant pRb fusion protein (Santa Cruz Biotechnology, Inc.) for cdk4 or cdk6 assay. After 30 min at 30°C, the reaction was terminated with 1 volume of ice-cold EDTA (20 mm, pH 8.0). Duplicate aliquots of the sample supernatant were spotted onto phosphocellulose paper squares (Upstate Biotechnology, Inc., Lake Placid, NY). The papers were immersed in 0.75% H₃PO₄ solution and washed for 6 × 5 min. Radioactivity incorporated into paper-bound histone or pRb was determined by liquid scintillation counting.

Previously we reported that rapamycin induced growth arrest and p53-independent apoptosis when Rh1 and Rh30 cells were exposed to drug under serum-free conditions (13). Expression of a rapamycin-resistant mTOR (Ser²⁰³⁵ → Ile²⁰³⁵) that has reduced binding of FKBP-rapamycin prevented growth inhibition (13). Consistent with these results, rapamycin (100 ng/ml) did not induce apoptosis in Rh30 and Rh1 cells expressing mutant mTOR (data not shown). These results strongly support rapamycin-induced apoptosis being mediated through its interaction with mTOR and not through inhibition of a secondary target.

The cytotoxic effect of rapamycin in these malignant cell lines is in contrast to most reports in which rapamycin exerts a cytostatic effect. We considered whether the cellular response to rapamycin (cytostasis or apoptosis) was determined by loss of a G₁ checkpoint function. Both cell lines express mutant p53 alleles (Rh1: Tyr²²⁰ → Cys²²⁰; Rh30: Arg²⁷³ → Cys²⁷³) and fail to induce p21^CIP1 in response to ionizing radiation (13), prompting us to focus on the loss of this tumor suppressor as being a common characteristic of these malignant cells. Rapamycin-mediated G₁ cell cycle arrest appears to be independent of p53 (49), and consistent with this is the finding that both Rh1 and Rh30 cells accumulate in G₁ phase in the presence of rapamycin (13). Conversely, we wondered whether rapamycin-induced apoptosis is a consequence of slowed, but continued progression through G₁ phase in the absence of a p53-mediated checkpoint. Under serum-free conditions of culture, rapamycin (100 ng/ml, 24 h) slowed growth (∼70–80%) and caused a significant increase in the proportion of Rh1 and Rh30 cells in the G₁ phase (Table 1). For Rh1, drug treatment increased the fraction of cells in G₁ phase from 67 to 83%. For Rh30, the proportion of cells in G₁ phase increased from 51.5 to 70.3%. However, there was no obvious change in the expression of G₁ cyclins, D1, E, and A in the presence of rapamycin (Fig. 1). Importantly, expression of cyclin A, a marker of late G₁-S phase interface was expressed at very high levels relative to cyclin D1 or cyclin E in control and rapamycin-treated cells.

Together, these data suggested that cells were progressing through G₁ phase, and potentially entering S phase, and committing to apoptosis. To test this hypothesis, Rh30 cells were grown under serum-free conditions with or without rapamycin (100 ng/ml) for 24 h. BrdUrd was then added with or without aphidicolin. Cells were grown for an additional 5 days, fixed, and processed for BrdUrd immunohistochemistry and stained with DAPI to determine nuclear morphology. Apoptotic cells were characterized as exhibiting either membrane blebbing, shrinkage, chromatin condensation, or micronucleation. As shown in Table 2, the BrdUrd labeling index for control cultures was 100% (all cells had undergone at least one round of replication within the 5-day labeling period). Rapamycin had only a slight effect on the labeling index, indicating continued DNA synthesis during the period of rapamycin exposure. Rapamycin increased the incidence of apoptosis from 8 to 25%, with >93% of apoptotic cells having incorporated BrdUrd in the presence of rapamycin. In contrast, ∼2% of apoptotic cells incorporated BrdUrd into DNA when treated with aphidicolin with or without rapamycin treatment. Thus, nuclear immunostaining for BrdUrd was not an artifact of nuclear condensation, as apoptotic cells did not score positive for nucleotide incorporation in the presence of aphidicolin, a DNA polymerase inhibitor. These data strongly support the contention that apoptosis is a consequence of progression through G₁ to S phase in rapamycin-treated tumor cells.

To determine whether expression of p53 altered the cell cycle response to rapamycin, cells were infected with Ad-p53 or a control adenovirus Ad-β-gal. Infection of Rh1 cells with Ad-p53 resulted in dramatic apoptosis, preventing further analysis. In contrast, Rh30 cells tolerated infection. Ad-p53 infected Rh30 cells, but not those infected with Ad-β-gal accumulated in G₁ phase (Table 3). One day postinfection with Ad-p53, 80% of Rh30 cells were in G₁ phase in contrast to 47% for Ad-β-gal-infected cells. Rapamycin (100 ng/ml, 24 h) increased the G₁ fraction to 86% in Ad-p53-infected cells and to 70% in control cultures (Ad-β-gal infected). The mechanism of p53-induced G₁ arrest involves up-regulation of cdk inhibitor p21^CIP1 (28, 29, 30). As shown in Fig. 2,A, when infected with Ad-p53, Rh30 cells expressed p21^CIP1 in a dose-dependent manner. A high level of p21^CIP1 protein could be detected 24 h postinfection, which could persist for at least 4 days (Fig. 2 B). In contrast, when Rh30 cells were mock infected, or infected with Ad-β-gal, the p21^CIP1 band was very faint or undetectable (data not shown).

Apoptosis was determined by the ApoAlert FACS analysis after 6 days of exposure to rapamycin (100 ng/ml). As shown in Fig. 3, control cells (uninfected) or those infected with Ad-β-gal demonstrated very low levels of annexin V-positive cells (∼4–6%). In contrast, rapamycin induced a significant increase in the proportion of cells positive for annexin V and propidium iodide in uninfected cultures or after infection with Ad-β-gal virus (26–28% viable as judged by annexin V-negative staining and exclusion of propidium iodide). Infection of Rh30 cells with Ad-p53 did not result in any increase in cells positive for annexin V or propidium iodide, but expression of p53 completely abrogated rapamycin-induced apoptosis in these cells (93% viable).

As described above, the mechanism by which p53 protected cells could be mediated by p21^CIP1-induced G₁ arrest. We therefore investigated whether direct expression of p21^CIP1 could protect cells from rapamycin-induced apoptosis and whether such protection results in p21^CIP1-induced G₁ block. Rh30 cells infected with Ad-p21 (MOI = 10) for 24–48 h expressed almost the same level of p21^CIP1 as that in cells infected with Ad-p53 (MOI = 1). Expression of p21^CIP1 was related to MOI, and a high level of p21^CIP1 protein could be detected 24 h postinfection and maintained for at least 5 days (Fig. 4, A and B). Expression of p21^CIP1after infection with Ad-p53 or Ad-p21 was not altered by rapamycin treatment (Fig. 4,C). In contrast to the effect of Ad-p53, Rh1 cells tolerated infection with Ad-p21 virus. Infection of Rh1 and Rh30 cells with Ad-p21 caused G₁ accumulation (data not presented) and significantly suppressed DNA replication (determined by [³H]thymidine incorporation). Inhibition of DNA synthesis by p21 (Rh1 and Rh30) or p53 (Rh30 cells) was significantly more effective than that induced by high concentrations of rapamycin, as shown in Fig. 5. This observation is consistent with the relative effects on cdk2 activity. Cells were treated as above or infected with Ad-βgal (control), then grown for 24 h with or without rapamycin (100 ng/ml). Activity of kinase activity of immunoprecipitated cdk2 is shown in Fig. 6.

Infection of Rh1 and Rh30 cells with Ad-p21 virus protected from rapamycin induced apoptosis (Fig. 7). These results suggest that the mechanism of Ad-p53 protection is at least partially through p21^CIP1-induced G₁ cell cycle arrest. To test this, BrdUrd-labeling experiments were repeated in the absence or presence of rapamycin, as described above, in Rh30 cells infected with either Ad-p53 or Ad-p21 or control virus (Ad-β-gal). Results are presented in Table 4. All cells infected with Ad-β-gal incorporated BrdUrd during the 5-day labeling period. The effect of rapamycin was similar to that in uninfected cells (see Table 2). Ninety-seven percent of cells incorporated BrdUrd and ∼25% of cells demonstrated morphological signs of apoptosis. Ad-p21 suppressed BrdUrd labeling (16% labeling) and protected against rapamycin-induced apoptosis. The effect of Ad-p53 was slightly greater with only 7% of cells labeling with BrdUrd over 5 days. Rapamycin did not increase levels of apoptosis over those in cultures infected with Ad-p21 or Ad-p53 alone. Taken together, these data suggest that rapamycin-induced apoptosis is a consequence of continued cell cycle progression, and that inhibiting progression by inducing a G₁ block protects cells.

Results presented above demonstrate that forced expression of p53 or p21^CIP1 can protect against rapamycin-induced toxicity. However, to define the role of these gene products under normal cellular control, we examined the response to rapamycin in wild-type or mutant MEFs that were nullizygous for p53 or p21^CIP1. Under serum-free conditions, rapamycin did not increase the proportion of apoptotic cells in cultures of wild-type MEFs (control 20% versus 17% in treated cultures). In contrast, rapamycin markedly induced apoptosis in cells lacking p53 or p21^CIP1 (Fig. 8,A). As with Rh30 cells, infection of p53^−/− MEFs with Ad-p53 protected against rapamycin toxicity (Fig. 8 B). Similar protection was obtained when p53^−/− MEFs were infected with Ad-p21 (data not shown).

We next determined whether rapmycin induced a more persistent G₁ arrest in wild-type MEFs than in those lacking either p53or p21^CIP1. Cells were grown under serum-containing conditions and treated with rapamycin for 24 h, at which time BrdUrd was added. After an additional 5 days, labeling of cells with BrdUrd was determined as described above. In control cultures (no rapamycin), the BrdUrd labeling index was 49.6, 100, and 100% for wild-type, p53^−/−, and p21^−/− cells, respectively. Rapamycin significantly decreased BrdUrd labeling in wild-type MEFs with only 19% of cells labeling after 5 days. In contrast, 89% of p53^−/− cells and 70% of p21^−/− cells labeled with BrdUrd in the presence of rapamycin (Fig. 9,A). Analysis of cyclin expression after 24-h treatment with rapamycin showed little effect of this agent on levels of cyclin D1 or cyclin E in any MEF strain. However, cyclin A was reduced to far lower levels in wild-type MEFs than in either p53^−/− or p21^−/− strains, consistent with a reduced proportion of cells in S phase (Fig. 9 B).

Under serum-free conditions, malignant rhabdomyosarcoma cells maintain viability and proliferate through autocrine mechanisms (2, 50, 51). However, under these conditions the cytostatic agent, rapamycin, induces p53-independent apoptosis (13). Rapamycin is generally regarded as a cytostatic agent as in most reports the cellular response to rapamycin is G₁ arrest without apoptosis. Thus, we were interested in determining whether rapamycin-induced apoptosis in these malignant cells was a consequence of continued progression through G₁ phase and entry into S phase in the presence of drug.

Extending our previous findings (13), expression of the rapamycin-resistant Ser²⁰³⁵ → Ile²⁰³⁵ mutant mTOR prevented rapamycin-induced apoptosis, strongly suggesting that apoptosis is a consequence of attenuated mTOR signaling in rapamycin-treated cells. As shown previously, within 24 h rapamycin caused an increased fraction of Rh1 and Rh30 cells to accumulate in G₁ phase of the cell cycle. However, this was not associated with any detectable change in expression of cyclin D1, E, or A, which is in contrast to other reports (52, 53, 54). For example, rapamycin inhibits the expression of interleukin 2-dependent cyclin A and E and the activity of cyclin-A or E-associated kinases p33^cdk2 and p34^cdc2, causing a mid to late G₁ block (52). In NIH3T3 cells, although rapamycin does not affect cyclin D- or E-dependent kinases, it delays the expression of cyclin A and its associated kinase activities (53). Furthermore, rapamycin decreases the cyclin D1 mRNA level and protein stability, resulting in inhibition of the G₁ to S transition in NIH3T3 cells (54). It has also been shown that rapamycin induces G₁ arrest, either by inhibiting cyclin E-associated kinase by preventing interleukin 2-mediated elimination of the cdk inhibitor p27^Kip1 in T lymphocytes (55) or by inhibiting pRb phosphorylation in vascular smooth muscle cells (56). However, we did not find any effect of rapamycin either on the levels of p21^CIP1, p27^Kip1, or p57^Kip2 or the phosphorylation levels of pRb, p130, and p107 in the tumor cells (data not shown). Since all above-mentioned experiments were done using p53 wild-type cells, we speculated that loss of p53 function may allow rapamycin-treated cells to progress through G₁ and enter S phase before initiating apoptosis. To test this, Rh30 cells were used as an experimental model, since Rh1 did not tolerate Ad-p53 infection for unknown reasons. Rh30 cells were exposed to rapamycin for 24 h to accumulate cells in G₁, then exposed further to rapamycin in the presence of BrdUrd. Cells were scored for morphological signs of apoptosis and immunostained to detect BrdUrd incorporation into DNA. Virtually all (>93%) apoptotic cells were positive for BrdUrd, indicating that these cells had initiated DNA synthesis in the presence of rapamycin. In addition, when rapamycin-treated Rh30 cells were exposed continuously to BrdUrd for 5 days, almost 100% of the cells were BrdUrd labeled, suggesting that rapamycin only slowed, but could not stop cell cycle progression of the p53 mutant tumor cells.

We next determined whether enforced expression of p53 could protect Rh30 cells from rapamycin-induced apoptosis. Unmanipulated cells or those infected with Ad-p53 or Ad-β-gal were grown with or without rapamycin. After 5 days, apoptosis was determined using the ApoAlert FACS assay. Viability of control, Ad-p53-, and Ad-β-gal- treated cultures was similar (>93%). Rapamycin induced massive apoptosis in unmanipulated cells and in those infected with Ad-β-gal. Ad-p53 dramatically protected Rh30 cells from rapamycin-induced apoptosis. Since up-regulation of p53 may drive cells to apoptosis, or protect cells from death by inducing cell cycle arrest (18, 19, 20, 21, 22, 23), we further investigated the mechanism by which p53 protected Rh30 cells from rapamycin-induced apoptosis. Western blot analysis showed that infection of Rh30 cells with Ad-p53 induced p21^CIP1 within 24 h and persisted for at least 4 days. In addition, although expression of p53 did not significantly affect levels of cyclin D1 and E and the activity of cyclin D-associated kinases cdk4 or cdk6 (data not shown), expression of p53 dramatically reduced the levels of cyclin A and cdk2 activity. This was accompanied by a sharp decrease of phosphorylation levels of pRb (data not shown). Consistently, labeling experiments revealed that only 7% of Ad-p53-infected Rh30 cells were BrdUrd positive following incubation with BrdUrd for 5 days, indicating that expression of p53 prevented the cell cycle progression. Moreover, expression of p53 resulted in G₁ accumulation, and rapamycin treatment further increased the proportion of cells in G₁ to 86%. Infection of Rh30 cells with control adenovirus (Ad-β-gal) did not affect either levels of p21^CIP1 and cyclin A, or cdk2 activity, or pRb phosphorylation, and did not induce G₁ accumulation. Taken together, these data suggest that rapamycin-induced apoptosis is a consequence of entering S phase while mTOR activity is inhibited by rapamycin. In contrast, forced expression of p53 blocked the cell cycle progression and arrested cells in G₁ phase rescuing from rapamycin-induced apoptosis.

To test whether p21^CIP1 could mimic p53, Rh1 and Rh30 cells were infected with Ad-p21 adenovirus. In contrast to Ad-p53, both cell lines tolerated infection with Ad-p21 and expressed p21^CIP1 at high levels. Ad-p21 infection resulted in accumulation of cells in G₁ phase, significant suppression of both DNA synthesis, as determined by [³H]thymidine and BrdUrd incorporation, and cdk2 activity, causing G₁ arrest. Ad-p21 completely protected both Rh1 and Rh30 cells from rapamycin-induced apoptosis. These results further strengthen the conjecture that arresting cell cycle progression prevented killing by rapamycin. Conversely, continued cell cycle progression in tumor cells results in rapamycin inducing a cytotoxic response.

These results indicate that forced expression of p53or p21^CIP1 can protect tumor cells from rapamycin-induced apoptosis, leading to the speculation that in cells with proficient p53 the response to rapamycin may be cytostasis in G₁ phase. Since tumor cells with wild-type p53 appear to have other defects in the ARF-p53-Rb pathway, we chose to examine the role of p53and p21^CIP1 in MEFs having defined gene disruptions. The response of wild-type MEFs to rapamycin was dramatically different from those with disruptions of p53or p21^CIP1. Rapamycin induced significant apoptosis in p53^−/− and p21^−/− cells, but not in wild-type cells. These results suggest that at least part of the p53 protection is mediated by p21^CIP1 in MEFs treated with rapamycin. Consistent with data derived from Rh30 cells, forced expression of p53or p21^CIP1 in p53^−/− MEFs also protected against rapamycin cytotoxicity. The response to rapamycin under normal culture conditions (serum containing) was also different between wild-type and mutant MEFs. Rapamycin induced growth arrest in wild-type cells and reduced the BrdUrd labeling index more effectively than in p53^−/− or p21^−/− cells. Ninety percent of p53^−/− cells incorporated BrdUrd over 5 days in the presence of rapamycin, compared to 19% in wild-type cells. Results for p21^−/− were intermediate, with 70% of rapamycin-treated cells incorporating BrdUrd. However, this latter result may represent an underestimate, as many p21^−/− cells in rapamycin-treated experiments became apoptotic and detached from the microscope slide. Thus, loss of p53 or p21^CIP1 function decreases the ability of cells to arrest in G₁ phase of the cell cycle in response to rapamycin treatment.

The results presented here support the notion that a p53-dependent event cooperates by enforcing G₁ arrest in response to inhibition of cap-dependent translation caused by rapamycin. The exact mechanism by which this is accomplished is under investigation. However, potentially, loss of p53 function could provide a basis for tumor-selective cytotoxicity of agents that target mTOR. We have shown previously (13) that rapamycin-induced apoptosis can be prevented by insulin-like growth factor I, hence it remains to be determined whether the tumor millieu in situ also protects cells from rapamycin-induced apoptosis in patients treated with the rapamycin ester CCI-779 that is currently in early clinical development. As CCI-779 induced objective tumor regressions in patients enrolled in the Phase I trial (57), it suggests that a cytotoxic response to mTOR inhibition may be achieved in some tumors. However, whether these “responsive” tumors had mutant p53 was not determined.

We thank Richard Ashmun for FACS analysis of apoptotic cells.