Sequestration of p53 in the Cytoplasm by Adenovirus Type 12 E1B 55-Kilodalton Oncoprotein Is Required for Inhibition of p53-Mediated Apoptosis

Abstract

The adenovirus E1B 55-kDa protein is a potent inhibitor of p53-mediated transactivation and apoptosis. The proposed mechanisms include tethering the E1B repression domain to p53-responsive promoters via direct E1B-p53 interaction. Cytoplasmic sequestration of p53 by the 55-kDa protein would impose additional inhibition on p53-mediated effects. To investigate further the role of cytoplasmic sequestration of p53 in its inhibition by the E1B 55-kDa protein we systematically examined domains in both the Ad12 55-kDa protein and p53 that underpin their colocalization in the cytoplasmic body and show that the N-terminal transactivation domain (TAD) of p53 is essential for retaining p53 in the cytoplasmic body. Deletion of amino acids 11 to 27 or even point mutation L22Q/W23S abolished the localization of p53 to the cytoplasmic body, whereas other parts of TAD and the C-terminal domain of p53 are dispensable. This cytoplasmic body is distinct from aggresome associated with overexpression of some proteins, since it neither altered vimentin intermediate filaments nor associated with centrosome or ubiquitin. Formation of this structure is sensitive to mutation of the Ad12 55-kDa protein. Strikingly, mutation S476/477A near the C terminus of the Ad12 55-kDa protein eliminated the formation of the cytoplasmic body. The equivalent residues in the Ad5 55-kDa protein were shown to be critical for its ability to inhibit p53. Indeed, Ad12 55-kDa mutants that cannot form a cytoplasmic body can no longer inhibit p53-mediated effects. Conversely, the Ad12 55-kDa protein does not suppress p53 mutant L22Q/W23S-mediated apoptosis. Finally, we show that E1B can still sequester p53 that contains the mitochondrial import sequence, thereby potentially preventing the localization of p53 to mitochondria. Thus, cytoplasmic sequestration of p53 by the E1B 55-kDa protein plays an important role in restricting p53 activities.

The adenovirus (Ad) E1B 55-kDa protein is important for Ad reproduction and cooperates with the E1A and E1B 19-kDa proteins in inducing cell transformation. One role for the E1B 55-kDa protein in transformation is thought to be that of inactivating the p53 pathway. Indeed, it is well established that the E1B 55-kDa protein from Ad2 or Ad5 physically interacts with the transactivation domain (TAD) of p53, thereby impairing p53-mediated transcription (30, 31). Several hydrophobic residues within the p53 TAD are critical for p53 binding to 55-kDa. For instance, the L22Q/W23S, W23S, and P27Y p53 point mutants cannot bind to the 55-kDa protein in an immunoprecipitation assay (15). In particular, p53 L22Q/W23S neither binds to MDM2 nor activates transcription as potently as wild-type p53 does. Therefore, it was suggested that these hydrophobic residues are required for interaction with the transcriptional machinery and that binding of these residues by the Ad2/5 55-kDa protein and MDM2 may prevent p53 from recruiting transcription factors, thereby abolishing its transactivation activity (15). In an elegant study, Yew et al. found that the 55-kDa protein is targeted to DNA-bound p53 and directly represses transcription (31). A later biochemical study indicated that as long as the E1B 55-kDa protein is tethered to the promoter, it can repress transcription and that this also requires a corepressor associated with RNA polymerase (20). The Ad2/5 E1B 55-kDa protein also forms large cytoplasmic aggregates with p53 in Ad-transformed cells, and it was suggested that this cytoplasmic sequestration plays a role in inhibiting p53 activity, presumably through removing p53 from its sites of action in the nucleus, where it regulates the transcription of genes involved in cell cycle control and apoptosis. The relative contribution of cytoplasmic sequestration of p53 and direct inhibition of p53-dependent transcription at p53-responsive promoters by the 55-kDa protein is unknown. Nonetheless, cytoplasmic sequestration of p53 by the 55-kDa protein could play a major role in inhibiting apoptosis, since p53 might directly trigger apoptosis by localizing to mitochondria, which results in cytochrome c release and caspase activation (3, 21).

Interestingly, the E1B 55-kDa protein from the highly oncogenic Ad12 does not bind directly to p53, although it has a high level of sequence identity to its Ad2/5 counterpart and can similarly inhibit p53 function. Consistent with this, whereas the Ad2 55-kDa protein binds to p53 in the yeast two-hybrid assay, we were unable to detect any direct interaction between p53 and the Ad12 E1B 55-kDa protein by using the same method (16). Immunoprecipitation using various antibodies against p53 or the Ad12 E1B 55-kDa protein failed to detect an interaction between them (28, 33). Replacing the p53-binding region of the Ad2 55-kDa protein (amino acids [aa] 224 to 354) with corresponding sequence from the Ad12 55-kDa protein (aa 210 to 341) dramatically reduced its affinity to p53, although the Ad2 and Ad12 55-kDa proteins exhibit very high sequence identity in this region (5). In addition, it was shown previously that the Ad12 E1B 55-kDa protein does not sequester p53 in the cytoplasmic body (28, 33). Nonetheless, the epitopes for antibodies used in previous studies might not be exposed, so that cytoplasmic colocalization between the Ad12 55-kDa protein and p53 was not detectable. Indeed, using different anti-Ad12 55-kDa antibodies, colocalization of p53 and Ad12 E1B was clearly demonstrated (16, 29). Another difference between the Ad2 and Ad12 55-kDa proteins is the preferential nuclear localization of the Ad12 55-kDa protein (5, 14, 28), which may be explained by the presence of a functional nuclear export signal (NES) in the Ad2/5 55-kDa protein but not in the Ad12 protein (12, 14). The Ad2/5 55-kDa protein is modified by SUMO-1, a ubiquitin-like protein modifier, whereas it is not certain whether Ad12 is subjected to the same modification (4). Additionally, the Ad12 55-kDa protein is involved in the induction of chromosomal fragility at specific loci (14, 32).

Thus, precisely how the Ad12 E1B 55-kDa protein inhibits p53-mediated activities is still not clear. In this study, we have examined in detail the requirements for the colocalization of the Ad12 E1B 55-kDa protein and p53 as well as the biological effects of the Ad12 55-kDa protein on the p53 pathway.

MATERIALS AND METHODS

Plasmids and antibodies.

The green fluorescent protein (GFP)-E1B fusion was constructed by inserting the full-length Ad12 E1B 55-kDa protein coding sequence into the EcoRI site of pEGFP-C2 (Clontech). Various E1B and p53 deletion and point mutants were constructed by the QuikChange protocol (Stratagene). The mitochondrial import leader sequence from human ornithine transcarbamylase, as described by Horwich et al. (9), was generated by PCR using a cDNA clone as template (GenBank accession no. AA428033). The PCR product was cloned between Flag and p53-coding sequences to make mito-l-p53, which also contains the Flag epitope at the N terminus. Some p53 mutants with point mutations within the TAD were obtained from J. Lin, and p73 expression plasmids were provided by M. Kaghad.

Antibodies against p53 (DO-1 and anti-p53-393FL) were purchased from Santa Cruz Biotechnology. Anti-Flag mouse monoclonal (M2) and rabbit polyclonal antibodies were purchased from Sigma. Anti-ubiquitin antibody (FK2) was purchased from Affiniti. Anti-Ad12 E1B 55-kDa rabbit polyclonal antibody was described previously (14). Anti-giantin rabbit polyclonal antibody was provided by E. Chan (22). Antiserum to the human centrosome was provided by J. Rattner. Antiserum to protein disulfide isomerase (PDI) was as described previously (2).

Immunofluorescence microscopy.

Cells grown on glass coverslips were transfected using Effectene reagent (Qiagen) and 24 h later were fixed with 4% paraformaldehyde for 20 min at room temperature and permeabilized with 0.2% Triton X-100 in phosphate-buffered saline (PBS). The slides were then incubated with blocking buffer (2% fetal bovine serum, 0.1% sodium azide, and 0.1% Tween 20 in PBS). After incubation with primary antibodies, the cells were washed with PBS containing 0.1% Tween 20 and then incubated with appropriate secondary antibodies conjugated with fluorescent dyes. The cells were washed and mounted in medium containing 4',6-diamidino-2-phenylindole (DAPI). The processed cells were examined using a Zeiss Axiophot microscope.

Luciferase reporter gene assays.

Saos2 human osteosarcoma cells (p53-deficient) were cultured in Dulbecco modified Eagle medium with 10% fetal bovine serum in a 24-well plate. The luciferase reporter construct PG13 contains multiple copies of the p53 DNA-binding site upstream of a TATA box cloned into pGL3-Basic (Promega). The control reporter carries the liciferase gene from the sea pansy (Promega). The reporter plasmids were transiently transfected into cells alone or with other plasmids by using SuperFect transfection reagent (Qiagen). The transfected cells were harvested 48 h posttransfection and processed for dual luciferase assays (Promega). The firefly luciferase activity was normalized against the sea pansy luciferase activity.

Colony formation assay.

Subconfluent Saos2 cells grown in a six-well plate were transfected with a combination of various plasmids. The cells were split in triplicate into 10-cm dishes and grown under puromycin (2 μg/ml) selection 24 h after transfection. The resulting colonies were stained with methylene blue solution and counted in ∼2 weeks under puromycin selection. The experiments were repeated at least three times in triplicate.

Apoptosis assay.

Saos2 cells grown on coverslips were transfected with the plasmids expressing various GFP-E1B and p53 constructs (0.5 μg each); 24 h later, 5-fluorouracil was added to a final concentration of 0.4 mM, and the culture was continued for an additional 24 h. The cells were then fixed with 4% paraformaldehyde and mounted on antifade medium containing DAPI. Using a fluorescence microscope, apoptotic cells were identified by their rounded and shrunken morphology with condensed and fragmented nuclei, in contrast to the spread-out appearance of nonapoptotic cells with well-preserved nuclei. In flow cytometry analysis, cells were similarly transfected and treated with 0.4 mM 5-fluorouracil for 48 h posttransfection; the cell cycle profiles of GFP-positive cells were analyzed, and cells with sub-G₁ DNA contents were considered apoptotic.

RESULTS

Ad12 E1B 55-kDa protein and p53 colocalize in cytoplasmic body.

To examine in detail the cytoplasmic colocalization of p53 and the Ad12 55-kDa protein, we fused GFP to the N terminus of the Ad12 55-kDa protein. The fusion protein formed large cytoplasmic body, as opposed to the even subcellular distribution of GFP (Fig. 1A, compare panel d with panel a). Formation of the cytoplasmic body does not require p53 since it was observed in p53-null Saos2 cells with or without p53 contransfection (data not shown). The cytoplasmic body appeared elongated and filamentous and was located at variable distances from the nucleus: some were juxtanuclear, and others were near the plasma membrane. The number and size of bodies varied widely, with most E1B-expressing cells having around two large bodies.

FIG. 1.

Colocalization of Ad12 E1B 55-kDa protein and p53 in the cytoplasmic body. (A) p53 aa 11 to 27 are required for localization of p53 to the cytoplasmic body. Vectors for fusion of GFP and the Ad12 55-kDa protein and various p53 constructs as indicated were transfected into p53-null Saos2 cells. p53 was detected using goat polyclonal antibody raised against full-length p53 (Santa Cruz Biotechnology) and rabbit anti-goat immunoglobulin G-rhodamine conjugate. The p53 mutants are as follows: K8R, lysine residues at 319 to 321, 372 to 373, 381 to 382, and 386 were converted to arginine; 1-375, aa 1 to 375; 1-355, aa 1 to 355; 1-315, aa 1 to 315; 83-393, aa 83 to 393; 83-355, aa 83 to 355; D11-27, aa 11 to 27 are deleted; D61-75, aa 61 to 75 are deleted; 22/23, L22Q/W23S. The cells were counterstained with DAPI. (B) p73 does not localize to the E1B cytoplasmic body. Flag-p73α or Flag-p73β vector was cotransfected with plasmid for the GFP fusion with Ad12 E1B, and the transfected cells were stained with rabbit anti-Flag polyclonal antibody (Sigma) and rhodamine-conjugated secondary antibody. (C) Summary of results shown in panels A and B. p53 and its mutants, as well as p73α and p73β are schematically drawn, and whether they localize to the E1B cytoplasmic body is indicated with + and −, denoting the presence or absence of colocalization, respectively. The individual domains in both p53 and p73 are illustrated: DBD, DNA-binding domain; OD, oligomerization domain; RD, regulatory domain; PxxP, proline-rich motif; SP, spacer sequence with unknown function; SAM, sterile-α domain. The numbers denote the amino acid position. The black box in the C terminus of p73β indicates distinct amino acids.

We then systematically examined which domain of p53 was responsible for colocalizing with the Ad12 E1B 55-kDa protein in the cytoplasmic body. Point mutations (K8R, lysine residues at 319 to 321, 372 to 373, 381 to 382, and 386 were converted to arginine [Fig. 1A, panels g to i]) in the C-terminal regulatory domain, or deletion from the C terminus (aa 1 to 375 [panels j to l], 1 to 355 [panels m to o], or 1 to 315 [panels p to r]) did not affect the colocalization of p53 with E1B. Therefore, the entire C-terminal regulatory domain of p53 is not necessary for complexing with E1B in the cytoplasm. It is noteworthy that p53 construct 1 to 315 lacks the sequence responsible for oligomerization (OD). Thus, the p53 monomer may be sufficient for localizing to the cytoplasm bodies. N-terminally truncated p53 constructs (aa 83 to 393 and 83 to 355) were not retained in the cytoplasmic body (panels s to x). While deletion of aa 61 to 75 had no effect (panels b1 to d1), deletion of aa 11 to 27 (panels y to a1), or even point mutations (e.g., L22Q/W23S) within this stretch of sequence (panels e1 to g1), abolished the localization of p53 to the cytoplasmic bodies. Thus, the N-terminal sequence between aa 11 and 27 is essential for p53 to form complexes with the Ad12 55-kDa protein in the cytoplasm. This viral protein is highly specific to p53, since p73α and p73β, members of p53-related transcriptional regulators, were not retained in the E1B cytoplasmic body (Fig. 1B, a to f), consistent with previous findings that viral oncoproteins do not interact with and inactivate p73 (18, 25).

The observed E1B cytoplasmic body was not due to some unexpected property of the GFP-E1B fusion, since the wild-type Ad12 55-kDa protein expressed in the human rhabdoid kidney tumor cell line G401 also formed a large, elongated cytoplasmic body and the endogenous p53 localized to it (Fig. 2A, panels a to c). Likewise, cytoplasmic clusters containing Ad12 E1B and p53 were observed in the transfected human non-small cell lung carcinoma cell line H1299 (29). Hence, formation of E1B cytoplasmic bodies is independent of cell types.

FIG. 2.

The E1B cytoplasmic body is distinct from aggresomes. (A) Intracellular distributions of wild-type Ad12 E1B 55-kDa protein in relation to various cellular proteins. G401 cells that constitutively express the Ad12 E1B 55-kDa protein were stained with rabbit polyclonal antibody raised against the N-terminal domain of Ad12 E1B (14). Antibodies to other proteins were as described in Materials and Methods. The nuclei were visualized by DAPI staining. (B) Staining pattern of GFP-E1B, vimentin, and the Golgi marker giantin. Saos2 cells were transfected with GFP-E1B, and the transfected cells were stained with antibodies to vimentin and giantin and an appropriate secondary antibody, respectively. The cells were counterstained with DAPI.

E1B cytoplasmic body is distinct from aggresomes.

Overexpression of some proteins leads to their aggregation in juxtanuclear locations in the cytoplasm. Such structures are called aggresomes and exhibit a number of common features including ubiquitination of constituent proteins, close association with the centrosome, reorganization of vimentin intermediate filaments, and low copy number (normally 1 per cell) (10). Several viral proteins have been found to form aggresomes (8, 13). To investigate whether the E1B cytoplsmic body has the characteristics of aggresomes, G401-CC3 cells that constitutively express the Ad12 55-kDa protein were stained with antibodies to ubiquitin, vimentin, and centrosomal proteins. Ubiquitin did not colocalize with the E1B body (Fig. 2A, panels d to f). Likewise, the E1B body did not associate with centrosomes (panels j to l). The vimentin intermediate filaments were not distorted around the E1B body (Fig. 2A, panels g to i; Fig. 2B, panels a to c). By contrast, vimentin is radically redistributed to form a cage-like network surrounding the exterior of the large aggregate in aggresome-containing cells (8, 10), and this feature is regarded as the hallmark of aggresomes (11). In addition, as noted above, the copy number of the E1B body is quite variable, ranging from 1 to more than 10 in E1B-expressing cells, and the elongated and filamentous morphology of the E1B body is also in striking contrast to the largely spherical aggresomes. Therefore, the E1B body is distinct from the aggresome.

We also examined the relationship between the E1B body and the Golgi apparatus and endoplasmic reticulum (ER). E1B-expressing cells were stained with antibodies to the ER marker PDI (2) or the Golgi marker giantin (22). The E1B body did not obviously associate with or distort the ER structure (Fig. 2A, panels m to o [note that the E1B body was revealed with p53 staining using mouse monoclonal antibody DO-1, because only rabbit antibodies were available for both Ad12 E1B and PDI]). Similarly, the 55-kDa protein was not coincident with the Golgi marker (Fig. 2B, panels d to f). Therefore, the E1B body is different from and is not closely associated with the Golgi apparatus or the ER.

The conserved C-terminal phosphorylation sites S476/7 of the Ad12 55-kDa protein are required for the formation of the cytoplasmic body.

We examined structural or sequence elements in the Ad12 E1B 55-kDa protein that may be required for the formation of the E1B cytoplasmic body. N-terminal deletion constructs of the Ad12 55-kDa protein abolished the cytoplasmic structure (Fig. 3A, panels a to c). Interestingly, removal of ∼100 aa from the N terminus markedly increased the nuclear presence of the Ad12 55-kDa protein (panels a and b), suggesting that, like the Ad2/5 protein, a potential NES may exist in the N-terminal sequence of the Ad12 55-kDa protein and that active nuclear export might be important for the formation of the E1B body. The N-terminal sequences are poorly conserved among the 55-kDa proteins from various Ad serotypes, and the identified NES in Ad2/5 E1B is absent in the Ad12 protein (14). Nonetheless, leucine-rich elements resembling typical NES are present within the Ad12 E1B N-terminal domain. For example, sequence 101-LsrLtVnLm-109 may be a potential NES. We therefore mutated L101/104 alone or in combination with L106/V108 into alanine and found that these mutations did not affect the formation of the cytoplasmic body (Fig. 3A, panels d to i). While it remains uncertain whether the sequence we mutated represents a functional NES, it is unlikely that active nuclear export is required, since treatment of E1B-expressing cells with the export inhibitor leptomycin B did not affect the appearance of E1B cytoplasmic bodies (data not shown).

FIG. 3.

An S476/7A mutation in the Ad12 55-kDa protein abolishes the formation of the E1B cytoplasmic body. (A) Subcellular distributions of p53 and GFP fusions with the Ad12 E1B 55-kDa protein and its mutants in Saos2 cells. Various mutants of the Ad12 55-kDa protein were fused with GFP, and their expression vectors were transfected alone or together with p53 plasmid into Saos2 cells. p53 was revealed with antibody to p53 (DO-1) and rabbit anti-mouse immunoglobulin G-rhodamine conjugate. The E1B mutants are as follows: 136-482, aa 136 to the C terminus; 341-482, aa 341 to the C terminus; 432-482, aa 432 to the C terminus; 101/4A, substitutions with alanine at L101 and L104; 101/4/6/8A, substitutions with alanine at L101, L104, and L108 as well as V106; 476/7A, S476 and S477 mutated to alanine; 476/7/481A, S476, S477, and D481 mutated to alanine; 481A, D481 mutated to alanine. The nuclei were revealed by DAPIstaining. (B) Summary of the results shown in panel A. The GFP fusion of wt Ad12 E1B 55-kDa protein and its various mutants are schematically depicted. Their ability to form a cytoplasmic body is indicated by + or − signs.

If only the N-terminal sequence is required for the genesis of the cytoplasmic bodies, mutations in other parts of the 55-kDa protein would be inconsequential. Surprisingly, mutations of the conserved S476/7 into alanine also abolished the formation of the structures (Fig. 3A, panels j to o). However, mutating another conserved residue, D481, had no effect (panels p to r). The corresponding residues to Ad12 E1B S476/7 in Ad2/5 were shown to be crucial for this viral protein to inhibit p53-mediated transactivation and apoptosis as well as for cooperation with the E1A and E1B 19-kDa proteins in cell transformation (26). Therefore, our results suggest a correlation between E1B-mediated biological effects and the formation of the cytoplasmic bodies (see below). Of note, the substitutions of the conserved serine residues at the C terminus also strikingly increased nuclear accumulation of the viral protein (panels j to o). Although this is purely speculative, these serine residues or phosphorylation at these sites might also be involved in nuclear export of the E1B 55-kDa protein.

The S476/7A mutation abolishes E1B-mediated inhibition of p53 functions.

Although the C-terminal sequences are highly conserved among the various E1B 55-kDa proteins and the residues corresponding to S476/7 of the Ad12 55-kDa protein are important for Ad5 E1B to inhibit p53 (26), it is still necessary to study whether the S476/7A mutation would have similar effects. We first examined whether such a mutation would affect p53-dependent transactivation. p53-responsive luciferase reporter PG-13 was transfected into Saos2 cells alone or with various combinations of vectors expressing p53, wild-type Ad12 E1B 55-kDa protein, and its mutants. Expressing p53 in Saos2 increased the reporter activity about 15-fold, and wt Ad12 55-kDa protein reduced this enhancement by two-thirds (Fig. 4B). S476/7A or S476/7D481A did not affect p53-mediated transactivation at all, but the D481A mutant repressed the reporter activity to the same extent as the wt 55-kDa protein did (Fig. 4B). p53 was well expressed in transfected cells in all cases, and the protein levels of wt Ad12 E1B and its mutants were similar in transfected cells (Fig. 4B, right panel). One interesting result of this experiment is that the wt Ad12 55-kDa protein or the D481A mutant reduced the reporter activities in the absence of p53 cotransfection. In the absence of p53, both versions of the Ad12 55-kDa protein repressed the reporter activities by two-thirds whereas S476/7A had virtually no effect and S476/7D481A mutant slightly increased the reporter activity (Fig. 4B, the middle panel). These data indicate that the Ad12 55-kDa protein can exert transcriptional repression without being tethered to the promoter by p53, in contrast to Ad2/5 55-kDa protein-mediated repression (20, 31). Nonetheless, like Ad2/5 E1B, the C-terminal phosphorylation sites are critical to repressing p53-dependent transcription.

FIG. 4.

S476/7A mutation of the Ad12 E1B 55-kDa protein relieves its inhibition of p53-mediated transactivation. (A) Sequence alignment of the C-terminal sequences of Ad2 and Ad12 E1B 55-kDa proteins. The numbers above or below the sequence indicate the position of the amino acid residues. Identical residues are indicated by a white box, and similar residues are indicated by a gray box. (B) Point mutations near the C terminus of the Ad12 E1B 55-kDa oncoprotein abolish its inhibition of p53 transactivation. Wild-type E1B (lanes 3 and 4) and point mutants (S476A/S477A [lanes 5 and 6], S476A/S477A/D481A [lanes 7 and 8], and D481A [lanes 9 and 10]) were transfected into Saos2 cells alone with reporters (PG13-Luc and pRL-SV40) or together with p53 and reporters, as indicated, and dual luciferase assays were performed 48 h after transfection. The expression of p53 and E1B in transfected cells is shown on the right. The lane numbers in the left and right panels are the same. The middle panel shows relative reporter activities from cells transfected with wt Ad12 E1B and its mutants in the absence of p53 expression in an enlarged format to facilitate a comparison of the activities in each transfection. Error bars represent one standard deviation of two independent transfections.

We further examined the effects of the C-terminal mutations of the Ad12 E1B on the functions of p53 in colony formation assays. As shown in Fig. 5, p53 expression in Saos2 cells dramatically reduced the colony number and coexpression of wt Ad12 55-kDa protein but not of the S476/7A mutant, while p53-mediated reduction of colony number could not completely be reversed by E1B 55-kDa expression (Fig. 5B). As a control, we also examined whether the Ad12 55-kDa protein would affect p73-mediated effects and found that while p73 reduced the colony number similarly to the reduction mediated by p53, Ad12 E1B cotransfection did not relieve this reduction (Fig. 5). Collectively, our data demonstrated that the S476/7A mutation not only eliminated the formation of cytoplasmic body but also abolished the inhibitory effects of E1B on p53; they therefore argue strongly that cytoplasmic colocalization between the Ad12 55-kDa protein and p53 plays an important role in regulating p53 functions.

FIG. 5.

Effect of the Ad12 55-kDa protein and its mutant on p53-imediated suppression of colony formation. (A) Representative images of colony formation assays. Saos2 cells were transfected with the indicated plasmids and plated in medium containing puromycin. E1B AA is the S476/7A mutant of the Ad12 E1B 55-kDa protein. (B) Quantification of the data from the colony formation assays. Error bars represent one standard deviation of three independent assays.

Cytoplasmic sequestration of p53 by the Ad12 55-kDa protein is required for inhibiting p53-mediated apoptosis.

To directly assess the relationship between cytoplasmic sequestration of p53 by the Ad12 55-kDa protein and the ability of p53 to induce apoptosis, we transfected Saos2 cells with vectors expressing p53 and various GFP-E1B constructs and then treated the cells with 5-fluorouracil, which has been demonstrated to elicit p53-dependent apoptosis in human tumor cells (1, 34). Consistent with this, apoptotic cells were rarely seen in nontransfected Saos2 cells 24 h after drug treatment (Fig. 6A). In cells transfected with p53 plasmid and the GFP fusion with wt Ad12 E1B, L101/4A, 101/4/6/8A, or D481A mutants, very few apoptotic cells were detected (Fig. 6). By contrast, apoptotic cells were frequently encountered in cells cotransfected with p53 and GFP fusions with S476/7A or the S476/7/D481A mutant (apoptotic cells with condensed and fragmented DNA and shrunken GFP pattern are indicated by arrows in Fig. 6A). Overall, fewer than 1% of transfected cells with the E1B cytoplasmic body exhibited apoptosis, whereas 23 and 31% of apoptotic cells expressed S476/7A and S476/7/D481A mutant, respectively (Fig. 6B). To verify these results independently, expression vectors for GFP-tagged wt Ad12 55-kDa protein and the L101/4A and S476/7A mutants were transfected individually or together with p53-expressing vector into Saos2 cells, the transfected cells were treated with 5-fluorouracil, and the cell cycle profiles of GFP-positive cells were analyzed. As shown in Fig. 6C, significantly higher levels of cells with sub-G₁ DNA content were detected in cells transfected with vectors for p53 and Ad12 E1B S476/7A than in cells expressing p53 together with either wt E1B or mutant L101/4A. To further confirm these observations, we transfected Saos2 cells with vectors for p53, wt Ad12 E1B, and mutants S476/7A and S476/7/D481A individually or with a combination of each E1B construct plus p53 plasmid. The GFP-expressing vector was included in all transfections. The cells were similarly treated with 5-fluorouracil, and GFP-positive cells were analyzed for their cell cycle profiles. The percentage of sub-G₁ cells less that in cells expressing GFP alone is shown in Fig. 6D. The levels of cells with sub-G₁ DNA content in cells transfected with wt Ad12 E1B plus p53 were significantly lower than those in cells expressing p53 alone or together with the E1B mutants with substitutions at S476/7. Note that the E1B mutants appeared to be more cytotoxic than the wt 55-kDa protein in the latter assays (Fig. 6D). This might be associated with toxicity of GFP coexpression (Fig. 7), since mutant S476/7A when fused with GFP was no more toxic than was the GFP fusion with wt E1B or the L101/4A mutant (Fig. 6C). Nevertheless, these flow cytometry experiments consistently showed that mutating the conserved serine residues renders the Ad12 E1B 55-kDa protein far less effective in reducing p53-mediated apoptosis. Therefore, cytoplasmic body formation correlates strongly with the ability of E1B to inhibit p53-mediated cell death.

FIG. 6.

Formation of the E1B cytoplasmic body is required to inhibit p53-dependent apoptosis. (A) Representative micrographs of Saos2 cells transfected with the indicated plasmids. Saos2 cells were transfected with expression vectors for p53 and wt Ad12 E1B 55-kDa protein fusion with GFP or with E1B mutants as specified. Cells were treated with 5-fluorouracil 24 h after transfection and were grown for an additional 24 h before being fixed. The cells were examined under a fluorescence microscope. Apoptotic cells with shrunken morphology and a fragmented nucleus are indicated by white arrows. The mutants of the Ad12 E1B 55-kDa protein are as described in the legend to Fig. 3. The nuclei were visualized by DAPI staining. (B) Quantification of apoptotic cells. Error bars represent one standard deviation of two independent assays. (C and D) Effects of the wt Ad12 E1B 55-kDa protein and its mutants on p53-mediated cell death as assessed by flow cytometry. In panel C, vectors for expressing the GFP fusion of wt E1B and mutants as indicated were transfected alone or together with the p53 plasmid into Saos2 cells, and at 24 h posttransfection the cells were treated with 0.4 mM 5-fluorouracil for 48 h and processed for analysis by flow cytometry. In panel D, expression vectors for the indicated proteins were cotransfected with vector expressing GFP and cells were treated and analyzed in the same way as in panel C. The percentage of cells with sub-G₁ DNA content in various transfections is plotted.

FIG. 7.

The Ad12 E1B 55-kDa protein does not affect apoptosis induced by L22Q/W23S mutant p53. Saos2 cells were transfected with vector for GFP alone or together with indicated plasmids, treated with 5-fluorouracil 24 h later, and harvested for flow cytometry analysis 72 h posttransfection, as was done for Fig. 6C and D. The GFP-positive cells were analyzed, and cells with sub-G₁ DNA content were considered apoptotic. (Top) Representative cell cycle profiles of transfected cells. Cells with G₁ and sub-G₁ DNA content are indicated. (Bottom) Percentage of sub-G₁ cells in various transfections.

Conversely, we reasoned that if keeping p53 in the E1B body is critical for inhibiting p53-mediated apoptosis, E1B might not affect the ability of the p53 mutant that is not retained in the cytoplasmic body to induce cell death. Since the p53 L22Q/W23S mutant does not localize to the cytoplasmic body (Fig. 1A, panels e1 to g1) and was documented to be capable of inducing apoptosis (7), we assessed how the Ad12 55-kDa protein might regulate the function of this mutant p53. As shown in Fig. 7, both wt p53 and the L22Q/W23S mutant elicited significantly higher levels of apoptosis in comparison to GFP-transfected cells, as judged by their sub-G₁ DNA content. Importantly, coexpression of the wt Ad12 55-kDa protein and p53 suppressed sub-G₁ cell population to a level similar to that of control GFP-transfected cells, consistent with results shown in Fig. 6. However, coexpression of the wt Ad12 55-kDa protein with the p53 L22Q/W23S mutant did not result in a reduction of the size of the sub-G1 population. Therefore, it appeared that the Ad12 55-kDa protein could not inhibit apoptosis induced by the p53 L22Q/W23S mutant. Taken together, we conclude that sequestration of p53 in the E1B cytoplasmic body is important for inhibiting p53-dependent apoptosis.

The Ad12 E1B 55-kDa protein sequesters mito-l-p53, which contains mitochondrial import leader peptide.

Recently, it has been shown that p53 can localize directly to mitochondria, where it triggers cytochrome c release and apoptosis (3, 21). Therefore, it is conceivable that sequestration of p53 by the 55-kDa oncoprotein might prevent p53 from entering mitochondria, thereby inhibiting p53-mediated apoptosis. To address this, we fused the N terminus of p53 to the mitochondrial import leader peptide from human ornithine transcarbamylase, which is capable of directing proteins into mitochondria (9), and this p53 fusion protein was shown to be effectively targeted to mitochondria (21). The fusion construct was well expressed in Saos2 cells (Fig. 8). As expected, the cytoplasmic presence of mito-l-p53 was dramatically increased (Fig. 8a and b), suggestive of correct mitochondrial targeting of p53. We then examined whether the Ad12 E1B 55-kDa protein could still retain this fusion in the cytoplasmic body. As shown in Fig. 8c to e, mito-l-p53 was found in the E1B cytoplasmic body, where most of the cytoplasmic p53 was concentrated. Since antibody to the Flag tag was used for detecting p53, the p53 revealed should still have the leader peptide and should not be the processed p53 that had lost the leader peptide. Therefore, it appears likely that as soon as p53 is synthesized by the ribosome, it is recognized and recruited to the E1B cytoplasmic body before being imported into the mitochondria.

FIG. 8.

The Ad12 E1B 55-kDa protein can sequester p53 with the mitochondrial import leader sequence. Saos2 cells were transfected with vector producing mito-l-p53 alone or together with plasmid for the wt Ad12 E1B 55-kDa protein fused with GFP. The mito-l-p53 construct contains the Flag epitope tag followed by the mitochondrial import leader sequence from the human ornithine transcarbamylase. Cells were fixed 16 h after transfection and stained with rabbit polyclonal anti-Flag antibody and goat anti-rabbit immunoglobulin G-rhodamine conjugate. The nuclei were visualized by DAPI staining.

DISCUSSION

In this report, we showed that the E1B 55-kDa oncoprotein from the highly oncogenic serotype Ad12 forms a large cytoplasmic body that also contains p53 when expressed. The N-terminal sequence between aa 11 and 27 in p53 is critical for retaining p53 in the cytoplasmic body. These residues are also implicated in a direct interaction of p53 with the Ad2/5 55-kDa protein (15). Therefore, the 55-kDa oncoproteins from different serotypes might interact with p53 in the same way. However, immunoprecipitation assays and yeast two-hybrid experiments failed to detect, or, at best, weakly detected, the interaction between the Ad12 55-kDa protein and p53 (5, 16, 33), whereas the interaction between p53 and Ad2/5 55-kDa can easily be detected in vitro and in vivo (19, 31) or in yeast (16). Thus, in spite of striking colocalization of the Ad12 55-kDa protein and p53 in the cytoplasmic body, it is still possible that a cellular protein(s) might be required for a stable interaction between the Ad12 55-kDa protein and p53. Interestingly, whereas S476/7A mutations in the Ad12 55-kDa protein completely abolished the formation of the cytoplasmic body (Fig. 3), mutations of corresponding residues in the Ad5 55-kDa protein did not significantly affect its affinity of binding to p53 (27). Thus, localization of p53 to the cytoplasmic body may not necessarily require its direct interaction with the 55-kDa protein in support of a view that one or more unidentified cellular proteins might be required for targeting p53 to the cytoplasmic body. Cellular proteins that both localize to the E1B body and bind to p53 can presumably bring p53 to the cytoplasmic structure. The WT1 tumor suppressor might be such a potential protein, because it binds to p53 and also localizes to the E1B cytoplasmic body (17). MDM2 might be another potential cellular protein, since it binds to the same p53 N-terminal sequence that is required for its localization to the E1B body. However, MDM2 did not localize in the body (data not shown); therefore, this cellular oncoprotein is not involved in bringing p53 to the cytoplasmic structure.

Our data indicated that aa 11 to 27 in the p53 N terminus are required for localizing p53 to the cytoplasmic body; other parts of the N-terminal TAD, the C-terminal regulatory domain, or the tetramerization domain are dispensable (Fig. 1). Therefore, monomeric p53 is sufficient for its targeting to the cytoplasmic body. Interestingly, aa 11 to 27 of p53 are a functional NES (35). This raises the possibility that nuclear export of p53 may be necessary for localizing p53 to the cytoplasmic body. Indeed, the E1B 55-kDa oncoprotein can actively shuttle between the nucleus and cytoplasm and is involved in the transport of late viral mRNA in cooperation with the 34-kDa E4orf6 protein (12). This property could enable E1B to relocate nuclear p53 to the cytoplasmic body, which would require active nuclear export. However, we found that p53 can still localize to the cytoplasmic body in the presence of the nuclear export inhibitor leptomycin B (data not shown), which is known to block shuttling of the E1B 55-kDa protein (12). As such, a more likely scenario is that p53 is recruited to the E1B body immediately after translation and before its import into the nucleus. Nonetheless, it remains possible that both nuclear export and cytoplasmic recruitment may contribute to localization of p53 to the cytoplasmic body.

Interestingly, mutation of S490/1 in the Ad2/5 55-kDa protein results in a mutant that retains the ability to bind to p53 (27). One might envision that this mutation should not affect the E1B-mediated repression of p53-dependent transactivation. However, the mutation essentially abolished the inhibitory effects of E1B on the transcriptional activity of p53 (26, 27). This is in agreement with previous conclusions that interaction between p53 and the 55-kDa protein is necessary but not sufficient for E1B to repress p53-dependent transcription (31). Therefore, it appears likely that the mutation may disrupt the interaction between the 55-kDa protein and the corepressor(s) that is required for E1B-mediated repression. Although interacting with the 55-kDa protein (23), histone deacetylases may not be the corepressors, since E1B could still exert repression in an in vitro system that contains low levels of histones(19). Other corepressors, e.g., mSin3a and an unidentified protein that copurified with RNA polymerase (20), could potentially be involved in the repression mechanism. Surprisingly, we showed here that corresponding mutation of the Ad12 55-kDa protein not only rendered it defective in inhibiting p53-mediated transactivation (Fig. 4) but also abolished the appearance of the E1B cytoplasmic body (Fig. 3). These observations suggest that cytoplasmic sequestration of p53 might contribute at least in part to E1B-mediated repression of p53, for example by reducing the p53 concentration in the nucleus. Alternatively, cytoplasmic sequestration of p53 by E1B inhibits the p53-mediated apoptotic mechanism only in the cytoplasm (3, 21), and the S476/7A mutation may independently affect cytoplasmic body formation and the interaction of E1B with corepressors.

Recently, it was shown that p53 can be found in the mitochondria, apparently through direct interaction with BclXL and Bcl2, which eventually results in cytochrome c release and cell death (3, 21). We showed above that p53 carrying a mitochondrial import leader peptide could still be retained in the E1B cytoplasmic body (Fig. 8). We also observed that the cytoplasmic body did not colocalize with mitochondria (data not shown). These observations are consistent with a model that the E1B 55-kDa protein recruits p53 to the large cytoplasmic body before it is imported into mitochondria. Therefore, the E1B 55-kDa oncoprotein is a particularly powerful inhibitor of p53; it uses multiple ways to inhibit p53 activities. In the nucleus, it tethers repressors to chromatin-bound p53, thereby inhibiting p53-dependent transactivation (31). It also interferes with acetylation of p53, which plays an important role in activating p53 by binding to acetylase and coactivator PCAF (16). In cooperation with the E4orf6 34-kDa protein, the 55-kDa protein can mediate ubiquitination and degradation of p53 by tethering it to SCF E3 ubiquitin ligase (6, 24). Furthermore, this viral oncoprotein can suppress p53-mediated apoptosis in the cytoplasm by sequestering p53 in the E1B cytoplasmic body and potentially blocking the import of p53 into mitochondria.