Nuclear Export Is Required for Degradation of Endogenous p53 by MDM2 and Human Papillomavirus E6

Abstract

The MDM2 oncoprotein targets the p53 tumor suppressor protein for degradation when the two proteins are expressed in cells. The regulation of p53 levels by MDM2 requires the ability of MDM2 to be exported from the nucleus by utilizing its nuclear export signal (NES). The drug leptomycin B (LMB) blocks the formation of nuclear export complexes consisting of CRM1, RanGTP, and NES-containing proteins. It is predicted that LMB should inhibit nuclear-cytoplasmic shuttling by MDM2 and subsequently stabilize p53. This communication demonstrates that LMB treatment of various cell lines led to an increase in the steady-state levels of the p53 protein as a result of an increase in its stability. The stabilized p53 protein localized to the nucleus and was an active transcription factor. These results indicate that the low steady-state levels of p53 in the absence of DNA damage result from p53’s nuclear export for cytoplasmic degradation. LMB also led to p53 stabilization in cell lines that contain human papillomavirus (HPV) DNA and express HPV E6, a protein that targets p53 for degradation. MDM2 is not necessary for E6-dependent degradation of p53, as evidenced by the observation that E6 promoted p53 degradation in cells lacking endogenous MDM2. In addition, LMB reduced E6’s ability to degrade p53 in the absence of MDM2, demonstrating that complete degradation of p53 by E6 requires nuclear export and therefore likely occurs in cytoplasmic proteasomes. These data suggest that the nuclear export of p53 to the cytoplasm for degradation is a general mechanism for regulating p53 levels.

The p53 tumor suppressor protein is present at low levels in the cell. In response to physiological stress such as DNA damage, p53 protein levels rise as a result of a posttranslational mechanism that leads to its stabilization (25). As a consequence, p53 becomes active as a transcription factor, inducing the transcription of genes such as bax, p21, and GADD45 that help p53 mediate its functions of growth arrest and apoptosis (8, 16, 26, 29). These activities contribute to p53’s role as a tumor suppressor by preventing the establishment of mutations in future generations of cells. The activated p53 protein also induces the transcription of the Mdm2 oncogene (1, 2), whose protein product binds to p53 and inhibits its functions as a transcription factor and tumor suppressor (4–6, 31). This inhibition of p53 by MDM2 can occur by at least two mechanisms. The first is an inhibition of p53’s ability to activate transcription of its downstream target genes; MDM2 binds directly to the transactivation domain of p53 and blocks its interaction with the TBP-associated factors TAF_II70 and TAF_II31 (22, 23, 41). Consistent with this observation, MDM2 can block p53 activity in an in vitro transcription assay (42). In addition, recent experiments demonstrate that the MDM2 oncoprotein can target p53 for degradation (15, 17). MDM2 is thus hypothesized to form an autoregulatory feedback loop with p53, in which p53 can control both its own levels and its own activity by inducing the expression of its negative regulator, MDM2 (20, 45).

The Mdm2 oncogene is amplified or overexpressed in a variety of human tumors (3, 7, 18, 19, 30, 35) and can function as an oncogene in tissue culture systems (9, 10). Its ability to inhibit p53 may well contribute to its activity as an oncogene. MDM2 can bind directly to p53 via its amino-terminal domain of approximately 115 amino acids, and this domain is sufficient to block p53’s transcriptional activation (4). MDM2’s ability to target p53 for degradation, however, requires an additional domain of the MDM2 protein: its nuclear export signal (NES) sequence, which is also required for its ability to shuttle between the nucleus and the cytoplasm (37). Because nuclear export by MDM2 is required for its ability to target p53 for degradation, MDM2’s nuclear-cytoplasmic shuttling presumably mediates the degradation of p53 by cytoplasmic proteasomes.

Whether this model truly relates to the regulation of endogenous p53 levels in cells is the subject of this paper. Previous experiments have relied upon transfection of MDM2 and p53 expression plasmids and do not therefore determine whether nuclear export regulates the normally low steady-state levels of endogenous p53 in cells in the absence of DNA damage (37). This paper demonstrates that the low steady-state level of p53 in a variety of cell lines is in fact dependent on nuclear export through the use of a nuclear export inhibitor, leptomycin B (LMB) (11, 13, 33). The addition of LMB to various cell lines led to an increase in the levels of the p53 protein, which was localized in the cell nucleus and transcriptionally active. The effect of LMB on p53 levels results from an increase in the half-life of the p53 protein, suggesting that nuclear export is required for the efficient degradation of p53. LMB also elevated the p53 levels in cell lines that contain human papillomavirus (HPV) E6 protein, and it is shown here that HPV E6 can target p53 for degradation by a mechanism that at least partially depends on nuclear export and not on the presence of the MDM2 protein. This result suggests that the HPV E6 protein, its cellular partner E6-AP, or another cellular protein mediates the shuttling of p53 to the cytoplasm for degradation. Thus, the use of nuclear export for the degradation of p53 may be a general mechanism that regulates the levels and therefore the activity of the p53 tumor suppressor protein.

MATERIALS AND METHODS

Cell lines and LMB.

The SiHa and CaSki cell lines were a gift from Peter Howley (Harvard University). The 1KO and 2KO cell lines were derived from primary embryo fibroblasts (27) provided by Guillermina Lozano (M. D. Anderson Cancer Center); the lack of endogenous MDM2 in the 2KO cell line was confirmed by Northern analysis as well as by immunoprecipitation followed by Western blotting. LMB, a gift from Patrick Chene and Michael Becker (Novartis), was stored at −20°C as a 1-mg/ml stock solution in ethanol, and fresh working dilutions in ethanol were made weekly.

Immunoprecipitation and Western analysis.

Cell lysates were prepared by resuspension of cell pellets and vortexing in lysis buffer (50 mM Tris-HCl [pH 7.5], 150 mM NaCl, 0.5% Nonidet P-40, and 1 mM EDTA, with the fresh addition of 1 mM dithiothreitol, 100 μM phenylmethylsulfonyl fluoride, 1 μM pepstatin A, and 1 μM E-64) followed by centrifugation to remove insoluble debris. The concentration of protein in the lysates was determined by Bradford analysis (Bio-Rad), and equal amounts of protein were either immunoprecipitated or separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to nitrocellulose (Hybond ECL [enhanced chemiluminescence] system; Amersham Life Science). Immunoprecipitations were performed by incubating lysates with protein A-Sepharose (Sigma) and a polyclonal antibody against MDM2. The beads were then pelleted and washed twice in SNNTE (50 mM Tris [pH 7.4], 5 mM EDTA, 5% sucrose, 1% Nonidet P-40, 0.5 M NaCl) before products were separated by SDS-PAGE and transferred to nitrocellulose. MDM2 was then detected on the blots by incubation with a combination of the monoclonal antibodies (MAbs) 2A10 and 4B11. For straight Western analysis without immunoprecipitation, p53 was detected with a combination of MAb 421 and MAb 1801, p21 with polyclonal serum specific for the human p21 protein, and β-catenin and Ran with MAbs from Transduction Laboratories. Blots were then incubated with antimouse immunoglobulin G (IgG) conjugated to horseradish peroxidase (Transduction Laboratories) or goat anti-rabbit IgG peroxidase (Cappel Organon Teknika Corp.), followed by detection by chemiluminescence (Amersham Life Science).

Indirect immunofluorescence.

Cells were treated with 5 ng of LMB per ml 20 h prior to fixation in 100% methanol at −20°C for 5 min. The p53 protein was detected by indirect immunofluorescence as described previously (12) with 15 μg of protein A-purified MAb 421 per ml, followed by incubation with goat anti-mouse IgG-Alexa568 (Molecular Probes, Inc.). Cell staining was observed by confocal microscopy.

Pulse-chase analysis.

Half of the cell cultures were treated with 5 ng of LMB per ml for 2.5 h. All of the cells were then prestarved for 30 min in Dulbecco’s modified Eagle’s medium (DMEM) minus methionine with 2% dialyzed fetal bovine serum, maintaining the same LMB concentration. ³⁵S-Express Easy Tag (NEN Life Science Products) at 75 μCi/ml was then added to each plate. After the 3-h labeling pulse, the cells were rinsed three times with phosphate-buffered saline and chased in DMEM with 10% fetal bovine serum and 2 mM cold methionine, either with or without the addition of 5 ng of LMB per ml. Cells were harvested at various times after the chase, lysates were made, and equal numbers of counts were precipitated with MAb 421. Immunoprecipitates were washed three times, each for 5 min in 1 ml of SNNTE and were then separated by SDS-PAGE and visualized by autoradiography. Quantification was performed on a phosphoimager.

Statistical analysis.

The fraction of p53 remaining at each time point of each pulse-chase experiment, compared to that in the relevant treated or untreated unchased sample, was calculated. The natural log of the average of four experiments was plotted versus time, and linear regression was performed to derive the half-life of p53 in each case. The statistical significance of the difference between the fractions of p53 remaining with and without LMB at each time point was determined by multivariate analysis of covariance with Quick STATISTICA 4.0 (StatSoft). The Scheffé test was employed for post hoc analysis. A P level of less than 0.01 was considered significant.

HPV E6-mediated degradation of p53.

Cells were transfected with combinations of 2 μg of pRC/CMV-hp53; 4 μg of pCMV-HDM2 or its empty vector, CMV-neo-bam₃; 2 μg of p1436 (E6) and 2 μg of its empty vector, p1318; or 4 μg of p1318 (28) and 2 μg of pGL2-Control (Promega) by lipofection with Lipofectamine reagent (Life Technologies) following the procedure supplied by the manufacturer. When indicated, LMB was added 8 h after the start of the transfection, following washing of the cells with PBS. Cells were lysed 24 h posttransfection, and luciferase activity in each sample was determined by using the Enhanced Luciferase Assay kit (Analytical Luminescence Laboratory) as described in the procedures provided by the manufacturer. Equal relative light units were run on each lane of SDS-PAGE for detection of p53 by Western analysis. Quantification was performed with the NIH Image program.

RESULTS

Inhibition of nuclear export increases the steady-state levels of p53.

The drug LMB blocks the formation of ternary complexes in the nucleus, consisting of CRM1, RanGTP, and proteins that contain NES sequences (11, 13, 33). The formation of these complexes is required for the nuclear export of NES-containing proteins; LMB therefore blocks their export from the nucleus (11, 13, 33). MDM2 contains an NES sequence that mediates its nuclear export and is required for its ability to target p53 for degradation (37). The addition of LMB to cells is then expected to prevent the nuclear export of MDM2, which in turn should result in the stabilization of the p53 protein in these cells. Consistent with this model, the addition of LMB to a variety of cell lines that contain wild-type or mutant p53 led to a marked increase in the steady-state levels of the p53 protein (Fig. 1). Total cellular protein from untreated cells or from cells 6 or 20 h after treatment with 5 ng of LMB per ml was fractionated by SDS-PAGE, and the p53 protein was quantitatively detected by immunoblotting. As expected, no p53 was detected in the p53-null human lung carcinoma cell line H1299 (Fig. 1). LMB addition caused approximate increases in the levels of wild-type p53 20 h post-LMB treatment of 16-fold in the human osteosarcoma cell line SJSA, 5-fold in the human breast carcinoma cell line MCF-7, and 4-fold in the immortalized murine cell line 12(1). The levels of mutant p53 in the human melanoma cell line SK-mel-2 increased approximately fivefold (Fig. 1). The addition of LMB to each of these cell lines, however, did not lead to an increase in the steady-state levels of the β-catenin and Ran proteins in the same lysates (Fig. 1), indicating that the effect is specific for p53. This increase in p53 levels, seen as early as 3 h after the addition of the drug and at a concentration as low as 1 ng/ml (data not shown), suggests that LMB may in fact stabilize the p53 protein by blocking its export to the cytoplasm for degradation.

FIG. 1.

LMB increases p53 protein levels in several cell lines. LMB (5 ng/ml) was added 6 or 20 h prior to the harvest of the cells. The steady-state levels of p53, β-catenin, and Ran in each lysate were observed by separation of total protein by SDS-PAGE and immunoblotting with antibodies specific for each of the proteins. The name of each cell line is indicated at the top; less protein was used for the determination of p53 levels in SK-mel-2, due to the high levels of the mutant p53 protein in these cells. The addition of LMB led to an increase in the levels of the p53 protein in each cell line, except in the p53-null H1299 cell line. Such an effect of LMB is not seen in the levels of the β-catenin and Ran proteins, demonstrating the specificity of the observation for p53.

The induction of p53 by LMB was also observed in these cells by immunofluorescence. In the SJSA cell line, the p53 protein is present at low levels, localized in a diffuse manner throughout the cell (Fig. 2A). As expected, the addition of 5 ng of LMB per ml resulted in increased levels of p53 protein that was predominantly localized to the nucleus at 20 h after the addition of the drug (Fig. 2B). In the case of MCF-7 cells, which contain cytoplasmic p53 (Fig. 2C), the increased levels of p53 protein found 20 h after the addition of 5 ng of LMB per ml were observed mainly in the nucleus (Fig. 2D). One interpretation of this result is that the p53 protein in MCF-7 cells constantly shuttles between the nucleus and the cytoplasm, likely in an MDM2-dependent fashion, and the pool of p53 is partially degraded in the cytoplasm. Upon the addition of LMB, the p53 protein that enters the nucleus is retained in this compartment, where it cannot be degraded by cytoplasmic proteasomes.

FIG. 2.

The p53 protein induced by LMB is localized to the nucleus. SJSA cells (A and B) and MCF-7 cells (C and D) were left untreated (A and C) or were treated with 5 ng of LMB per ml (B and D) 20 h prior to fixation. The p53 protein was detected with MAb 421, followed by goat anti-mouse IgG-Alexa568. In both cases, the induced p53 protein is localized to the nucleus of the cells following LMB treatment.

To determine if the p53 protein that is induced by the addition of LMB is in fact active as a transcription factor, the levels of p53-responsive MDM2 and p21 proteins were determined before and after the addition of LMB. Using the same lysates as in Fig. 1, immunoprecipitation followed by Western analysis was performed for MDM2 and Western analysis was performed for p21. The addition of 5 ng of LMB per ml to SJSA and MCF-7 cells, which contain wild-type p53, led to an increase in the levels of MDM2 and p21 protein (Fig. 3). The levels of these two proteins did not, however, increase in response to LMB addition in the SK-mel-2 cell line, which contains mutant p53, and in the p53-null H1299 cell line (Fig. 3). These results indicate that the increases in MDM2 and p21 proteins are dependent on the presence of functional p53 protein, and thus the p53 protein induced by LMB treatment of cells is active for transcription.

FIG. 3.

The p53 protein induced by LMB is transcriptionally active. Cells were treated with 5 ng of LMB per ml 6 and 20 h prior to harvesting. The steady-state levels of p53 and p21 in each lysate were observed by separation of total protein by SDS-PAGE followed by immunoblotting. MDM2 protein was immunoprecipitated prior to immunoblotting. The levels of MDM2 and p21 increased in response to LMB in wild-type p53-containing SJSA and MCF-7 cells, but not in SK-mel-2 and H1299 cells that lack functional p53 protein. A longer exposure is shown for the left half of the MDM2 blot because of the lower levels of MDM2 in these cells, and a shorter exposure is shown for the p21 blot of MCF-7 cells due to the higher levels of p21 in this cell line.

The addition of LMB to cells led to an inhibition of the cell cycle which was not solely due to the increased levels of p53, since cells that lack the p53 protein were arrested upon the addition of LMB. These p53-negative cells include the H1299 cell line and an immortalized mouse cell line, 10(3) (data not shown).

Inhibition of nuclear export by LMB stabilizes the p53 protein.

To determine whether the change in the steady-state levels of p53 after the addition of LMB was a result of a block of the p53 protein’s degradation in the cytoplasm via its retention in the nucleus, the half-life of the p53 protein was determined in the presence and absence of LMB by using the 12(1) immortalized mouse 3T3 cell line. Half of the cell cultures were treated with 5 ng of LMB per ml in growth media for 2.5 h, in media lacking methionine for 30 min, and then in media containing [³⁵S]methionine for 3 h. These cells were also chased in media containing an excess of unlabeled methionine in the presence of the same concentration of LMB. The remaining cell cultures were incubated in media lacking methionine, labeled, and then chased, all without the addition of LMB. Lysates were immunoprecipitated with MAb 421 against p53 or with MAb 419 against simian virus 40 large T antigen to serve as a negative control. After washing, the immunoprecipitates were separated by SDS-PAGE and visualized by autoradiography (Fig. 4A). After quantification of four separate experiments, the half-life of p53 was found to increase approximately fourfold after LMB addition from an average of 57 min to 236 min (P < 0.005; Fig. 4B). This difference indicates that the increase in the steady-state levels of p53 seen after blocking nuclear export is in fact due to an inhibition of the degradation of p53. As seen in Fig. 4, the p53 protein, although increased in stability, is still being degraded in the presence of LMB. This degradation could be a result of the p53 protein being degraded by an MDM2-independent mechanism using nuclear proteasomes or could result from an incomplete block of nuclear export by LMB. The cellular proteasome inhibitor MG-132 inhibits both nuclear and cytoplasmic proteasomes and leads to a virtually complete block in p53 degradation (24).

FIG. 4.

LMB leads to an increase in the half-life of p53 in 12(1) cells. (A) Five plates of 12(1) cells were pretreated with 5 ng of LMB per ml for 2.5 h prior to being prestarved for 30 min in media lacking methionine and then were labeled with 75 μCi of ³⁵S-Express per ml for 3 h, both in the presence of the same concentration of LMB. Six plates were prestarved and labeled similarly, but without the addition of LMB. Cell were washed, chased with media containing excess cold methionine with or without 5 ng of LMB per ml, and harvested at different times. Samples were immunoprecipitated with MAb 419 (419 IP) against simian virus 40 large T antigen as a negative control or with MAb 421 against p53, washed, and run on SDS-PAGE. Molecular weights (MW) are shown to the left (thousands). (B) The levels of p53 remaining at each time were quantified from four separate experiments on a phosphoimager, background was subtracted, and values were normalized to the samples that were not chased in the same experiment. The average fractions of p53 remaining and their associated standard deviations are shown at the various chase times on the left-hand plot, in the absence of LMB (solid line) and in the presence of 5 ng of LMB per ml (dashed line). The natural log (ln) transform of this plot is also shown on the right; linear regression of this plot was used to calculate the average half-life of p53 without (57 min) and with (236 min) the addition of 5 ng of LMB per ml. The difference between the fraction of p53 remaining at each time point in the presence or absence of LMB was tested by multivariate analysis of covariance with the Scheffé test for post hoc analysis. Asterisks (∗) denote time points that are significantly different between the two conditions (P < 0.01).

HPV E6-mediated degradation of p53 also involves nuclear export.

In order to explore whether mechanisms for p53 degradation other than that which are mediated by MDM2 also require nuclear export, cell lines containing integrated DNA sequences from HPV types 16 and 18 (HPV-16 and -18, respectively) were used. Such cells express both HPV E6 and E7 proteins; the E6 protein from these viral isotypes binds to p53 and targets it for degradation (38, 43). The p53 protein levels in HeLa (HPV-18), SiHa (HPV-16), and CaSki (HPV-16) cells increased 20 h after the addition of 5 ng of LMB per ml (Fig. 5), suggesting that E6-mediated degradation of p53 utilizes a nuclear export step. One possible explanation for this observation is that the export of p53 is mediated by the low levels of the MDM2 protein in these cells; thus MDM2 and E6 could cooperate to target p53 for cytoplasmic degradation. To test this possibility, a transfection experiment was performed with the MDM2-null 2KO cell line, which was obtained by passage on a 3T3 schedule of primary mouse embryo fibroblasts derived from a mouse null for both the p53 and Mdm2 genes. Cotransfection of HPV-16 E6 and p53 expression plasmids into these cells led to a decrease in the amount of p53 detected in these cells by immunoblotting as compared to the p53 levels obtained by transfection of the p53 plasmid alone (Fig. 6, lanes 3 and 4). Similarly, cotransfection of an MDM2 expression plasmid with p53 led to a decreased amount of p53 protein (Fig. 6, lanes 3 and 5). The reduction in the p53 protein levels after transfection of E6 or MDM2 plasmids (approximately 10- and 3-fold, respectively) was similar to that observed in the 1KO cell line, which is null only for p53 and contains functional MDM2 (data not shown). This result indicates that the E6 protein does not require MDM2 for the degradation of p53.

FIG. 5.

LMB induces the p53 protein in HPV E6-expressing cells. Cells were treated, lysates were prepared, and p53 was detected by immunoblotting as described in the legend to Fig. 1. Baculovirus-purified human p53 was loaded in the first lane as a positive control. HeLa cells contain HPV-18 sequences, while SiHa and CaSki cells contain HPV-16 sequences. The p53 protein levels in each of these cell lines increased after 20 h in the presence of 5 ng of LMB per ml. Molecular weights (MW) are shown to the left (thousands).

FIG. 6.

HPV E6 degradation of p53 is MDM2-independent and LMB sensitive. The 2KO cell line, which lacks endogenous MDM2, was transfected with pGL2-Control which leads to constitutive expression of luciferase (lanes 2 to 8), empty vectors (lane 2), human p53 (lanes 3 to 8), HPV-16 E6 (lanes 4 and 7), and/or human MDM2 (lanes 5 and 8). Eight hours after transfection, 5 ng of LMB per ml was added to half of the samples (lanes 6 to 8). At 24 h posttransfection, lysates were made and luciferase activity in each sample was determined in order to normalize for transfection efficiency. Equal relative light units were run on SDS-PAGE, and p53 was visualized by immunoblotting as described in the legend to Fig. 1. E6-mediation degradation of p53 does not require MDM2, but is partially blocked by LMB. Molecular weights (MW) are shown to the left (thousands).

To determine if the degradation of p53 by E6 was dependent on nuclear export even in the absence of MDM2, LMB was added to the cells 8 h posttransfection. As expected, the addition of 5 ng of LMB per ml to cells transfected with MDM2 and p53 almost completely blocked the ability of the MDM2 protein to lower the steady-state levels of p53 in the transfected cells (Fig. 6, lanes 6 and 8; a less than 20% reduction in p53 levels). In contrast, the addition of LMB only partially blocked the ability of E6 to lower the steady-state levels of p53 in transfected cells, such that a three- to fourfold reduction was still observed (Fig. 5, lanes 6 and 7). One reason for the incomplete block in E6-mediated p53 degradation may be that E6 can target p53 for degradation in both the nucleus and cytoplasm. Alternatively, it may be that there is too much E6 protein present in this overexpression assay for a full block of p53 degradation by LMB. One final explanation for the incomplete effect of LMB on E6-mediated degradation in this assay could be the speed with which E6 can target p53 for degradation in the cytoplasm, such that some degradation occurs before p53 can enter the nucleus. The addition of LMB does, however, consistently decrease the level of p53 degradation by E6 by at least twofold in this assay, indicating that E6-mediated degradation of p53 utilizes a nuclear export function that can be observed in the absence of MDM2. It remains to be determined if the HPV E6 protein itself has nuclear export ability or whether some other cellular protein performs this function.

DISCUSSION

The ability of LMB to stabilize the p53 protein as shown in this communication clearly demonstrates that the low steady-state levels of p53 in the absence of DNA damage are maintained by CRM1-dependent nuclear export. Although it is possible that LMB could, in addition to blocking nuclear export, be causing DNA damage or providing a DNA damage-like signal to p53, several observations indicate that this is unlikely. For instance, the induction of p53 by LMB occurs in cell lines that do not respond to certain types of DNA damage. The HeLa cell line does not respond after UV irradiation by increasing levels of p53; however, this cell line responds to LMB addition with an increase in levels of p53 (Fig. 5 and data not shown). In addition, one effect of DNA damage, the p53-independent decrease in MDM2 protein levels (44), does not occur after the addition of LMB to the SJSA cell line, whereas it does occur after UV irradiation (data not shown). This result demonstrates that the cellular response to DNA damage can be differentiated from the response to LMB.

As seen in Fig. 4, the addition of LMB does not completely block the degradation of p53 in the 12(1) immortalized murine cell line. One possible explanation for this observation is that the block of nuclear export by LMB is incomplete, allowing some p53 protein to be exported to the cytoplasm, where it is rapidly degraded. Alternatively, it is possible that p53 is degraded, albeit with slower kinetics, in the nucleus. Such degradation is likely to be MDM2-independent, for reasons discussed below, and may represent a second pathway for the regulation of p53 levels.

Three lines of evidence suggest that MDM2’s degradation of p53 occurs exclusively in the cytoplasm. The first is that a mutant form of MDM2 that is unable to be exported from the nucleus is also unable to target p53 for degradation (37). Second, the NLS-rex protein, which efficiently competes with MDM2 for shuttling from the nucleus to the cytoplasm, blocks MDM2’s ability to leave the nucleus and its ability to mediate p53 degradation (37). The third line of evidence is that the addition of LMB, which blocks nuclear export, blocks MDM2-mediated degradation of p53 (Fig. 6). Proteasomal complexes are located in both the cytoplasm and the nucleus, and there is some indication that the two populations have a unique subunit composition (34, 36, 40). It is likely that the proteasomal components recognized by MDM2 after shuttling p53 are exclusively cytoplasmic. Alternatively, a factor may be present in the nucleus that blocks MDM2’s interaction with nuclear proteasomes.

The E6 protein, on the other hand, may be able to target p53 to both cytoplasmic and nuclear proteasomes, because the addition of LMB to cells transfected with p53 and E6 plasmids only partially blocks the degradation of p53 by E6 (Fig. 6). This result suggests that E6, unlike MDM2, might recognize both nuclear and cytoplasmic proteasomes. Alternatively, it is possible that the E6 protein is able to degrade a fraction of the p53 protein present in the cytoplasm prior to its initial import into the nucleus. Although these possibilities cannot be distinguished at this point, the partial effect of LMB on E6 degradation of p53 demonstrates that CRM1-dependent nuclear export is required for the complete degradation of p53 by E6.

That E6-mediated degradation of p53 utilizes a nuclear export step and that this can occur in the absence of MDM2 suggests that some NES-containing protein other than MDM2 is shuttling p53 to the cytoplasm in E6-containing cells. Candidates for this function include both HPV E6, which contains a potential NES sequence, and E6-AP, which contains several such potential NES sequences. The shuttling abilities of both proteins remain to be determined, but evidence suggests that E6-AP is localized to both the nucleus and the cytoplasm (14), and the E6 protein has been shown to colocalize with p53 in the cytoplasm of HeLa, SiHa, and CaSki cells (21). In addition, p53 itself contains a putative NES and may be able to mediate its own nuclear export under certain circumstances (39a).

The results presented here support the model presented in Fig. 7 but do not prove that MDM2 is required for the export of p53 from the nucleus. An alternative model in which MDM2 and p53 are exported separately from the nucleus, meeting in the cytoplasm, where MDM2 then targets p53 to cytoplasmic proteasomes, is also consistent with the data presented here. Recent work shows that p53 itself contains a leucine-rich putative NES present in the tetramerization domain (such that only monomers and dimers should be exported) and that exogenous wild-type p53 can be exported from the nucleus in cells lacking MDM2 (39a). These observations indicate that some protein other than MDM2, whether p53 itself or some other cellular protein, might mediate the nuclear export of p53, at least, in an overexpression assay. This conclusion is consistent with the observation that complete p53 degradation by HPV E6 requires a nuclear export function that occurs in the absence of MDM2 (Fig. 6).

FIG. 7.

A model for the degradation of p53 by MDM2. MDM2, p53, CRM1, and RanGTP are predicted to form a ternary complex in the nucleus, which is transported to the cytoplasm through the nuclear pore. p53 is then targeted to cytoplasmic proteasomes, where it is degraded. CRM1, Ran (now in the GDP-bound form), and possibly MDM2 return to the nucleus to repeat the process. The addition of LMB prevents the formation of this complex in the nucleus so that p53 is not shuttled to the cytoplasm for degradation. As predicted by the model, p53 levels in the nucleus rise after LMB treatment.

Several lines of evidence support the model presented in Fig. 7, in which MDM2 shuttles p53 from the nucleus to the cytoplasm, rather than the model in which they exit the nucleus separately. First, it is clear that p53 and MDM2 do interact in the nucleus in many cell lines (4, 32), consistent with the model presented in Fig. 7. Second, p53 is localized to the cytoplasm in some cell lines that overexpress MDM2, such as MCF-7 (Fig. 2C). DNA damage, which modifies p53 such that it can no longer interact with MDM2, leads to p53 accumulation in the nucleus in these cells (39). This observation is also consistent with the model in which MDM2 is responsible for the continual shuttling of p53 molecules from the nucleus to the cytoplasm, localizing most of the p53 protein there and/or targeting it for degradation by cytoplasmic proteasomes. Third, a cytoplasm-localized mutant form (Δ150–230) of MDM2 can efficiently bind to p53 in vitro, but fails to bind to and inhibit the transcriptional activity of p53 in vivo (4), suggesting that p53 is retained in the nucleus in the absence of the shuttling activity of MDM2. Experiments are currently under way to directly demonstrate CRM1-dependent shuttling of p53 by MDM2.