Phosphorylation of human p53 at serine 46 determines promoter selection and whether apoptosis is attenuated or amplified.

The capacity of DNA damaging agents to induce apoptosis is regulated by target gene induction by p53. We found that p53 targeted MDM2 in cells in which DNA repair was occurring, but persistent DNA damage induced by chemotherapy led p53 to selectively target PTEN. High dose chemotherapy induced the phosphorylation of p53 on serine 46, whereas low dose chemotherapy did not. A nonphosphorylatable serine 46 to alanine p53 mutant (S46A) targeted the MDM2 promoter in preference to that for PTEN. A serine 46 to aspartate mutant (S46D, a phosphorylation mimic) targeted PTEN in preference to MDM2. These observations show that phosphorylation of serine 46 in p53 is sufficient for it to induce the PTEN (phosphatase and tensin homolog deleted on chromosome ten) tumor suppressor protein in preference to MDM2. S46A induced significantly less cell death than the S46D in cells. The phosphorylation-induced change of p53 promoter targeting suppresses the induction of MDM2 and the formation of the autoregulatory feedback loop. Induction of PTEN by p53 followed by expression of PTEN inhibits AKT-induced translocation of MDM2 into the nucleus and sustains p53 function. The protection of p53 from MDM2 by PTEN and the damage-induced activation of PTEN by phosphorylated p53 leads to the formation of an apoptotic amplification cycle in which p53 and PTEN coordinately increase cellular apoptosis.

The capacity of DNA damaging agents to induce apoptosis is regulated by target gene induction by p53. We found that p53 targeted MDM2 in cells in which DNA repair was occurring, but persistent DNA damage induced by chemotherapy led p53 to selectively target PTEN. High dose chemotherapy induced the phosphorylation of p53 on serine 46, whereas low dose chemotherapy did not. A nonphosphorylatable serine 46 to alanine p53 mutant (S46A) targeted the MDM2 promoter in preference to that for PTEN. A serine 46 to aspartate mutant (S46D, a phosphorylation mimic) targeted PTEN in preference to MDM2. These observations show that phosphorylation of serine 46 in p53 is sufficient for it to induce the PTEN (phosphatase and tensin homolog deleted on chromosome ten) tumor suppressor protein in preference to MDM2. S46A induced significantly less cell death than the S46D in cells. The phosphorylation-induced change of p53 promoter targeting suppresses the induction of MDM2 and the formation of the autoregulatory feedback loop. Induction of PTEN by p53 followed by expression of PTEN inhibits AKT-induced translocation of MDM2 into the nucleus and sustains p53 function. The protection of p53 from MDM2 by PTEN and the damage-induced activation of PTEN by phosphorylated p53 leads to the formation of an apoptotic amplification cycle in which p53 and PTEN coordinately increase cellular apoptosis.
p53 is a short-lived, nonabundant protein that regulates the response of cells to DNA damage, in part through transcriptional activation of genes involved in cell cycle control, DNA repair, and apoptosis (1). p53 is a tumor suppressor protein; consequently, mice in which the p53 gene has been disrupted develop tumors with high frequency (2,3), and deletions or mutations in the p53 gene are prevalent in a majority of human cancers (4,5). The protein level and activity of p53 are regulated, and this is accomplished by MDM2 (murine double minute 2) (6 -13). The MDM2 gene is induced by p53, and MDM2 prevents apoptosis by inhibiting p53 activity and promoting its degradation (6,14,16). Once formed, the p53-MDM2 complex shuttles from the nucleus to the cytoplasm (12,(17)(18)(19)(20)(21), in which MDM2 targets p53 for degradation (9,11,14). Thus, p53 and MDM2 form an autoregulatory feedback loop in which p53 induces MDM2, and MDM2 negatively regulates p53 (22) thereby protecting cells from apoptosis.
The PTEN (23)(24)(25) tumor suppressor is a dual specificity phosphatase that dephosphorylates phosphatidylinositol 3-phosphate, a lipid second messenger produced by phosphatidylinositol 3-kinase (24,26,27). Thus, PTEN-deficient tumor cells and immortalized fibroblasts from PTENϪ/Ϫ mice have elevated levels of phosphatidylinositol 3-phosphate (26 -28). PTEN mutants that retain protein tyrosine phosphatase activity but lose the capacity to dephosphorylate phosphatidylinositol 3-phosphate are found in tumors, indicating that the lipid phosphatase activity of PTEN is required for its tumor suppressor function (24,26,27,29). By inhibiting phosphatidylinositol 3-kinase function, PTEN inhibits the activation of a downstream target of phosphatidylinositol 3-kinase, the AKT/ protein kinase B serine threonine kinase, a survival factor that induces genes that inhibit apoptosis and suppress the functions of proapoptotic proteins (30 -35).
p53, PTEN, MDM2, and AKT form a network that regulates cell viability (36). The induction of MDM2 prevents inappropriate apoptosis by suppressing p53 function. AKT phosphorylates MDM2 and induces its movement into the nucleus, in which it down-regulates p53 (37)(38)(39). By blocking the nuclear entry of MDM2, PTEN sustains p53 activity (40). In damaged or mutated cells p53 activates genes that promote apoptosis; among these is PTEN (41). Thus, p53 can induce MDM2, which inhibits p53 function. Alternatively, p53 can induce PTEN, which stabilizes p53 by segregating it from MDM2. This facilitates formation of a cycle in which p53 induces PTEN and PTEN stabilizes p53. PTEN and p53 facilitate the activities of each other to induce cell death. The mechanism that leads p53 to alter target gene selection is fundamental to understanding its actions and the subject of this study.
Here we demonstrate that as the concentration of etoposide to which cells are exposed increases, an alteration from the induction of antiapoptotic MDM2 to proapoptotic PTEN occurs, and this dictates whether cells live or die. This change of protein induction by p53 results from the phosphorylation of serine 46 and an alteration of gene promoter selection. A consequence of the change of promoter targeting is that cells exposed to low levels of stress respond with the induction of MDM2, a protective response (22), whereas high levels of stress abrogate the protective response and induce an apoptotic amplification cycle that promotes death.
For etoposide treatment, the medium was changed to Dulbecco's modified Eagle's medium to which etoposide was added, and incubation proceeded for 2 h. Cells were washed twice with PBS, 1 cells were harvested at the zero time point, and Dulbecco's modified Eagle's medium/1% fetal bovine serum was added to the remaining plates. Cells were harvested at the indicated times thereafter. The concentrations of etoposide used were based on previous in vivo and in vitro observations. The aim in patients is to achieve concentrations of 1-3 g of the drug/ml (ϳ1.5 M) (42)(43)(44)(45)(46). Studies with cultured cells employ etoposide concentrations of 1 to about 200 M (47-50), and these produce variable, cell type-specific results. The present study assayed changes of signaling associated with etoposide-induced apoptosis, leading us to evaluate etoposide effects over a concentration range of 4.7 to 470 nM. These were the lowest concentrations of etoposide that would induce changes of promoter selection in the cell lines employed.
Transfections and Reporter Assays-Saos-2 and H1299 cells were transfected using Lipofectamine. Some cells were transfected with the pHook vector (Invitrogen) that encodes a single chain surface antibody (c-MYC-tagged sFv) that was used for normalization of transfection efficiency as described in figure legends. For reporter assays, cells were transfected with ␤-galactosidase and vectors using the calcium phosphate method, and reporter activity was normalized to ␤-galactosidase activity. The PTEN reporter was a gift from Dr. Ian de Belle and has been described recently (51).
Apoptosis-Cells were fixed in 1% formalin/PBS, stained with 20 g/ml acridine orange, and visualized using a fluorescence microscope with a blue-green filter. Apoptotic cells showed cytoplasmic and nuclear shrinkage and chromatin condensation or fragmentation. At least 200 cells were analyzed, and the fraction of apoptotic cells is reported as percentage of the total (the apoptotic ratio). For 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assays, cells were treated as described, and viability was determined (40).

RESULTS
To define conditions that lead p53 to induce MDM2 or PTEN, asynchronous U2OS cells were treated with 4.7, 47, or 470 nM etoposide for 2 h, the medium was changed, and cells were harvested. This regimen was chosen to simulate how cells in a tumor would be exposed to chemotherapy. Each concentration of etoposide increased the levels of p53, but the magnitude and duration of response increased with the dose of etoposide ( To test whether the increase of the MDM2 and PTEN proteins was due to increased levels of mRNA, RNA was isolated from cells treated with 4.7 or 470 nM etoposide, and MDM2 and PTEN expression was assessed by RT-PCR. MDM2 was high in cells exposed to 4.7 nM etoposide and lower after treatment with 470 nM etoposide; PTEN was almost undetectable in cells treated with 4.7 nM etoposide but greatly augmented by 470 nM etoposide (Fig. 1B).
Etoposide induces DNA double strand breaks. To examine whether the persistence of DNA damage corresponded to MDM2 or PTEN induction, we performed a comet assay. Etoposide induced DNA damage in a concentration-dependent manner, and damage that persisted 6 h after treatment was greater in cells treated with 470 nM than with 47 or 4.7 etoposide (Fig. 1C). Confocal microscopy showed that 470 nM but not 4.7 nM etoposide induced irregular morphology and shrinkage characteristic of apoptosis (Fig. 1D, top), and this was quantified by double blind analysis (Fig. 1D, bottom).
We hypothesized that the autoregulatory feedback loop that attenuates p53 function by the induction of MDM2 may be negated during high dose chemotherapy by a change of promoter targeting favoring activation of PTEN. To test this, p53null Saos-2 cells were transfected with transcriptionally active or inactive p53 in the absence or presence of MDM2, and Western blotting showed that MDM2 diminished p53 levels and subsequently a product of one of its target genes, PTEN ( Fig. 2A). Transcriptionally inactive mutant (22/23) p53, which does not bind and is not down-regulated by MDM2, did not induce PTEN (Fig. 2A). To determine whether p53 elevates PTEN in response to DNA damage, HCT116 colon carcinoma cell lines isogenically matched for the presence or absence of p53 were treated with etoposide. Etoposide increased PTEN only in p53-positive cells (Fig. 2B). To reinforce the conclusion that p53 is required for PTEN induction, we took advantage of the fact that p14 arf stabilizes p53 (52) and used U2OS cells engineered to express p14 Arf when treated with isopropyl 1-thio-␤-D-galactopyranoside (53). Induction of p14 arf increased p53 and PTEN protein levels and expression of another p53 target, p21 (Fig. 2C).
Gene reporter assays showed that p53 mediated its effect on PTEN by activation of the PTEN promoter. U2OS cells, wild type for p53, were transfected with vector, catalytically dead (C124S) PTEN or PTEN, and a PTEN promoter luciferase reporter. Ectopic expression of active but not inactive PTEN activated the PTEN promoter, presumably by inhibiting the nuclear movement of MDM2 and thereby sustaining the function of p53 (Fig.  2D). To show that p53 activates the PTEN promoter, isogenically matched HCT116 cells, wild type or null for p53, were transfected with PTEN or catalytically inactive PTEN (C124S) and the PTEN reporter. PTEN activated PTEN only in HCT116 cells that contained p53 (Fig. 2E). The capacity of the p53 antagonist, pifithrin, to block the induction of the PTEN in p53-positive cells also showed that p53 mediated PTEN induction (Fig. 2F). Thus, in addition to inducing MDM2, which protects cells from death, p53 can induce PTEN and promote apoptosis.
p53 is phosphorylated at numerous sites in response to DNA damage. This led us to evaluate changes of p53 phosphorylation in U2OS cells treated with different concentrations of etoposide. Although quantitative evaluation of the relative stoichiometry of protein phosphorylation is difficult, qualitative comparison of the time courses of phosphorylation of various sites in p53 from the same extracts was performed to permit us to ultimately focus on a particular site in p53. Cells were incubated with a high or low concentration of etoposide for various times, and 50 g of lysate from each incubate was fractionated by SDS-PAGE. Western blots were then probed with antibodies that would recognize alterations of any phosphorylation site over time. The time dependence of phosphorylation of serines 6, 9, 15, 20, 37, or 392 (Fig. 3A), did not correlate with the time and concentration dependencies of PTEN induction in these cells (Fig. 1A). Phosphorylation of serine 46 was unaffected by the treatment of cells with 4.7 nM etoposide but increased with time when cells were incubated with 470 nM etoposide (Fig. 3A). The response of normal human mammary epithelial cells to etoposide was also examined (Fig.  3B). Human mammary epithelial cells were incubated with 4.7 or 470 nM etoposide for 5 h. MDM2 levels increased after treatment with 4.7 but not 470 nM etoposide. As with U2OS cells (Fig. 1A), PTEN was induced by 470 but not by 4.7 nM etoposide, and this correlated with phosphorylation of serine 46 in p53.  5. Effect of p53 and p53 mutants on promoter activation. A, U2OS cells were transfected with p53, S46A, S46D, the PTEN reporter and ␤-galactosidase. Luciferase activity from three independent transfections normalized to ␤-galactosidase is shown. B, U2OS cells were transfected with p53, S46A, S46D, the MDM2 reporter, and ␤-galactosidase. Reporter activity from three independent transfections normalized to ␤-galactosidase is shown. Error bars represent standard deviation from the mean.
Interest in serine 46 was reinforced by the demonstration that its phosphorylation correlated with increased PTEN in cells treated with 470 nM etoposide. Expression of apoptosisinduced protein 1 (p53AIP1) is dependent on serine 46 phosphorylation of p53 and plays a role in the induction of the apoptotic program (54). We observed that 470 nM but not 4.7 nM etoposide augmented p53AIP1 expression, and this correlated with the phosphorylation of serine 46 (Fig. 4A). In addition, activation of caspase 3, another marker for cellular apoptosis, was induced by 470 but not by 4.7 nM etoposide.
On the basis of the correlation of phosphorylation of serine 46 with the induction of PTEN, p53AIP1, caspase 3, and cell death (Figs. 1, 4, and 5) the significance of this site in p53 for the promotion of the apoptotic program was studied. To accomplish this, we mutated serine 46 in p53 to alanine (nonphosphorylatable S46A) or aspartic acid (a phosphorylation mimic, S46D). S46A, S46D, and p53 were tested for activation of PTEN and MDM2 reporters. p53 activated both promoters (Fig. 5). S46A activated the MDM2 reporter but was unable to activate the PTEN reporter. S46D activated PTEN but was unable to target MDM2. These results show that the modification of serine 46 directs promoter targeting by p53.
Examination of endogenous expression of MDM2 and PTEN in H1299 cells showed that p53 induced both genes (Fig. 6A). S46A altered gene induction in favor of MDM2, and S46D induced PTEN. By comparison with p53 or S46A, S46D was ineffective in targeting MDM2 for induction. An analysis of protein expression yielded results consistent with those from gene reporter assays and RT-PCR. Western blot analysis showed that the amount of MDM2 was greater in H1299 cells transfected with S46A or p53 than in cells expressing S46D. Additionally, MDM2 was greater in cells expressing S46A rather than p53. PTEN levels were greater in cells transfected with S46D than with p53 or S46A (Fig. 6B). The level of PTEN induction induced by S46D was comparable with PTEN induction observed by Stambolic et al. (41) after the treatment of cells with radiation. Because MDM2 and PTEN were determined against a background of the endogenous proteins, the differences observed by Western blotting were less pronounced than those observed with RT-PCR.
The functional significance of p53 promoter selection was determined by assaying apoptosis in H1299 cells treated with 4.7 nM etoposide. This low dose of chemotherapy was used as transfection with p53 sufficiently stressed H1299 cells so as to induce phosphorylation of serine 46 (Fig. 7A). Cells expressing p53 and S46D were less viable than cells expressing S46A (Fig.  7B). That cells expressing S46A underwent apoptosis shows that serine 46 phosphorylation is not the only mechanism through which p53 promotes death.
The significance of PTEN is illustrated by greater etoposideinduced apoptosis in glioblastoma cells that express PTEN at a level comparable with that in normal brain, compared with PTEN-null or glioblastoma expressing inactive PTEN. PTEN inactivated AKT, restricted MDM2 to the cytoplasm, and stabilized p53 (Ref. 38, and data not shown). Fig. 7C shows that a significantly greater fraction of PTEN-positive cells underwent apoptosis than PTEN-null cells or cells expressing mutant R130M PTEN in which all lipid phosphatase activity is abrogated (57,58). Thus, the shift of p53 gene targeting from MDM2 to PTEN results in diminished survival as p53 and PTEN coordinate death by facilitating one another's expression and functions. cells were transfected with vector, p53, S46A, or S46D. Twenty-four h thereafter, the cells were treated with 4.7 nM etoposide for 4 h, cell lysates were prepared, and Western blots were probed for phosphoserine 46 p53 (P-Ser46) and p53. C, control; WT, wild type. B, H1299 cells in 6-well plates were transfected with vector, p53 (Wt p53), S46A, or S46D (3 wells/construct) and then selected with G418 for 10 days. The cells were treated with 4.7 nM etoposide for 8 h, the medium was changed, and the cells were grown for 1 week after survival was assayed by crystal violet staining. Data are the average from two independent experiments with data points from each taken in triplicate. C, PTEN-null U87MG, glioblastoma cells, or U87 cells expressing inactive (R130M) or wild type PTEN were grown in the presence of various concentrations of etoposide. After 24 h viability was assayed by a 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide assay and is expressed as percentage survival relative to time zero (prior to the addition of etoposide). Each point is the mean of triplicate determinations. DISCUSSION p53 coordinates the expression of genes that attenuate or promote cell death (59). One p53 target, MDM2, is overexpressed in 5-10% of all human tumors (8). MDM2 immortalizes and transforms primary and cultured cells (7,60), and ectopic expression of an MDM2 transgene leads to tumor formation in mice (61). Overexpression of the MDM2 protein is predictive of high grade, aggressive, metastatic malignancies refractory to chemotherapy. The best defined mechanism through which MDM2 leads to uncontrolled cell growth is by the inhibition of p53. Another p53 target is PTEN, which is frequently mutated in glioblastoma, prostate, endometrial, breast, lung, and other tumor types (62)(63)(64) as well as in several autosomal cancer predisposition syndromes (23,25,(65)(66)(67)(68). PTEN negatively regulates cancer cell survival by inhibiting the activation of AKT, which through numerous pathways allows cancers to evade apoptosis.
Posttranslation modifications of p53 or MDM2 can prevent their interaction, and phosphorylation of p53 also affects its retention in the nucleus, transcriptional activity, DNA binding, degradation, and interaction with co-activators (15, 69 -73). Phosphorylation of serine 46 has no affect on the interaction of p53 with MDM2. This phosphorylation is mediated by HIPK2 and results in increased expression of proapoptotic p53AIP1 (54 -56). The present study shows that chemotherapy induces serine 46 phosphorylation, which abrogates the autoregulatory feedback loop through which MDM2 inhibits the apoptotic function of p53 and induces PTEN, which favors apoptosis. We also show that the phosphorylation status of serine 46 determines gene promoter selection by p53 and whether cells live or die.
MDM2 is induced by low dose etoposide and PTEN by high dose etoposide, and this is due to a change of promoter selection by p53. This conclusion is supported by increased expression of RNA for MDM2 but not PTEN after the treatment of cells with 4.7 nM etoposide and by the induction of PTEN, but not MDM2, RNA by 470 nM etoposide. That high dose but not low dose chemotherapy increases the level of mRNA for PTEN and activates the PTEN promoter suggests that transcription underlies the alteration of protein expression induced by the higher dose of chemotherapy.
Phosphorylation of serine 46 in p53 correlated with the dose of etoposide to which cells were exposed and was associated with the induction of p53AIP1, activation of caspase 3, induction of PTEN gene expression, and cell death. Expression of HIPK2 or S46D activated the PTEN promoter, whereas S46A was unable to do so but rather activated the promoter for MDM2, which S46D was unable to do. This change of promoter targeting is functionally significant because cell death was promoted by expression of S46D and low dose etoposide. S46A was less capable of inducing cell death, although it did so in response to stress. Our observations show that stress sufficient for p53 to target the PTEN promoter results in augmented apoptosis; however, the observation of some, albeit attenuated, apoptosis in PTEN-null cells demonstrates that events other than serine 46 phosphorylation and unrelated to PTEN play a role in p53-mediated apoptosis.
The switch of gene promoter targeting induced by phosphorylation illustrates how a specific posttranslational modification can regulate p53 promoter selection and cell viability. By selecting the promoter for MDM2 an autoregulatory feedback loop is formed that restrains p53 function. By selecting the promoter for PTEN, an apoptotic amplification cycle is formed; PTEN blocks nuclear entry of MDM2 and stabilizes and sustains p53; and under stress, p53 targets the PTEN promoter and increases PTEN expression. Formation of the autoregula-tory feedback loop is blocked by the coordinate activities of p53 and PTEN, which together form an apoptotic amplification cycle that induces cell death.