p53 is localized to a sub-nucleolar compartment after proteasomal inhibition in an energy-dependent manner

The p53 tumor suppressor protein is a central target of inactivation in human cancer and a key regulator of genotoxic stress-induced growth arrest or apoptosis (Appella and Anderson, 2001; Giaccia and Kastan, 1998; Ko and Prives, 1996; Levine, 1997; Oren, 1994). Several reports have suggested a role for p53 in ribosome biogenesis (Budde and Grummt, 1999; Cairns and White, 1998; Chesnokov et al., 1996; Marechal et al., 1994; Sugimoto et al., 2003; Zhai and Comai, 2000). Mutations of the p53 core domain are one of the most frequent alterations detected in cancer cells (Levine, 1997). Interestingly, this region is needed to repress transcription by RNA polymerase I (Pol I) or RNA polymerase III (Pol III) (Stein et al., 2002a; Stein et al., 2002b; Zhai and Comai, 2000). Indeed, several point mutations in the core domain abrogate the capacity of p53 to repress rRNA synthesis (Budde and Grummt, 1999). Furthermore, in fibroblasts from p53 knock-out mice (p53^–/–) the amount of pre-rRNA and tRNA is significantly higher than in wild-type animals (Budde and Grummt, 1999; Cairns and White, 1998). In accordance with these results p53-null epithelial cells have an elevated Pol I transcriptional activity compared to epithelial cells that express p53, indicating that repression of Pol I transcription by p53 occurs by an active repression mechanism (Zhai and Comai, 2000).

Multiple studies have demonstrated that under certain conditions p53 can be found in the nucleolus in several cell lines. The first study documenting p53 in the nucleolus revealed the presence of two different subsets of p53 protein in prostate and bladder carcinoma cells where the wild-type protein accumulates in the nucleolus whereas the mutant subgroup localizes to the nucleoplasm and is excluded from the nucleolus within the same cell (Benninghoff et al., 1999). p53 associates with nucleoli in detergent-permeabilized cells (Rubbi and Milner, 2000) and the proteasome inhibitor MG132 causes p53 to localize in the nucleolus and nucleoplasm (Karni-Schmidt et al., 2007; Klibanov et al., 2001; Latonen et al., 2003; Pokrovskaja et al., 2001). Nucleolar accumulation of p53 was also shown to be important for the reactivation of p53 in HeLa cells after cisplatin treatment (Horky et al., 2001; Wesierska-Gadek et al., 2002). Furthermore, in fibroblasts derived from spinal muscular atrophy (SMA) patients [caused by mutation in the survival motor neuron gene (SMN1)] with markedly reduced levels of SMN protein, p53 accumulated exclusively in the nucleolus (Young et al., 2002). Also, it has been reported that PRIMA-1, a compound that restores DNA binding properties and conformation of mutant p53, induces nucleolar translocation of mutant p53 (Rokaeus et al., 2006).

There are multiple potential mechanisms that might facilitate nucleolar localization of p53. p53 can bind sequences within the ribosomal gene cluster (Kern et al., 1991) and can associate with various nucleolar components such as L5 ribosomal protein (Marechal et al., 1994), topoisomerase I (Gobert et al., 1996), nucleolin (Daniely et al., 2002), nucleostemin (Tsai and McKay, 2002) and nucleophosmin (Colombo et al., 2002). Mdm2, the central negative regulator of p53 and an E3 ubiquitin ligase which ubiquitinates p53 and targets p53 for proteasome degradation, is not only able to form a complex with ribosomal proteins L5, L11 and L23 but itself is capable of nucleolar localization (Dai and Lu, 2004; Dai et al., 2004; Lohrum et al., 2003; Zhang et al., 2003). Nucleolar localization of Mdm2 and its association with the ribosomal proteins inhibit Mdm2 functions and help prevent p53 proteasomal degradation (Dai et al., 2004).

The nucleolus is the site of rRNA transcription, processing and ribosome assembly (Andersen et al., 2005; Carmo-Fonseca et al., 2000; Lam et al., 2005; Lamond and Sleeman, 2003; Leung et al., 2003; Olson and Dundr, 2005). As diagrammed in Fig. 1A the nucleolus consists of at least three morphologically distinct regions: the fibrillar center (FC), a dense fibrillar center (DFC) and a granular component (GC) (Scheer and Hock, 1999). Although, still not fully described, several studies indicate that the FC or the FC-DFC boundaries are the sites of pre-rRNA transcription (reviewed by Boisvert et al., 2007; Lamond and Sleeman, 2003). Ribosomal proteins and assembly factors are found in the GCs (Huang, 2002; Koberna et al., 2002; Lamond and Sleeman, 2003). FCs are particularly enriched in Pol I, topoisomerase I and the upstream binding factor (UBF, an activator of Pol I), whereas most processing molecules localize to the DFC and/or GC including nucleolin, nucleophosmin and fibrillarin [Casafont et al. (Casafont et al., 2007) and references therein].

In a previous study we showed that p53 accumulation in the nucleolus, after proteasome inhibition, requires two regions within the carboxyl terminus of p53 (Karni-Schmidt et al., 2007). We show now that after proteasome inhibition, p53 can associate with a discrete sub-nucleolar component, the FC. We demonstrate that p53 association with the FC is ATP dependent and it is abrogated by genotoxic stress. Our results also suggest that monoubiquitination of p53 does not preclude its association with nucleoli. Perhaps most relevantly we have found that p53 associates with nucleoli in human bladder and lung carcinomas.

To extend our observations and those of others that the treatment of cells with a proteasome inhibitor causes nucleolar localization of p53 (Karni-Schmidt et al., 2007; Klibanov et al., 2001; Latonen et al., 2003; Pokrovskaja et al., 2001) we examined the precise location of ectopic p53 within the nucleolus. H1299 cells were transfected with a construct expressing wild-type p53 and after 24 hours, cells were treated with different concentrations of MG132 for various times or left untreated. The levels of transfected p53 either did not change or were modestly reduced by treatment with MG132 in our system (Fig. 1B). Nucleolar localization of p53 could be detected by 8 hours after MG132 treatment (Fig. 1C). Surprisingly, when we compared the p53 staining pattern to that of nucleolin, a major nucleolar phosphoprotein that localizes to the DFC and GC (Bugler et al., 1982; Shaw and Jordan, 1995), nucleolin formed a visible ring around p53, indicating that in addition to weak or strong nucleoplasmic staining, p53 is localized in the FC following MG132 treatment (Fig. 1C). In cases where nucleoli contained more than one FC, p53 was detected in several FCs within the same nucleolus. DIC imaging confirmed our results, clearly showing p53 localization at the FC (Fig. 1C bottom row, E bottom row). In our experimental setting, the optimal concentration of MG132 was 10 μM, leading to approximately 10-15% of cells displaying nucleolar p53. However, in all cells with nucleolar p53, we found that p53 localized to the FCs (Fig. 1D). Using UBF as a well validated FC marker (Fig. 1E) (Casafont et al., 2007) we confirmed that the staining pattern with the UBF antibody is virtually identical to that observed with the p53 antibody after treatment with MG132. Furthermore, p53 localization to the FC following MG132 treatment was assessed in several cell lines expressing both wild-type and mutant forms of p53 (Table 1). Our results show that a subset of cell lines have detectable nucleolar p53.

We previously found that there are two discrete regions within the p53 C terminus that contribute to its ability to associate with nucleoli (Karni-Schmidt et al., 2007) although neither is homologous to the well characterized HIV-1 Rev NoLS (Hope, 1999; Meyer and Malim, 1994). HIV-1 Rev protein was shown to localize efficiently to nucleoli and to associate specifically with the nucleolar DFC and GC regions (Dundr et al., 1995). It was of interest to determine whether p53 nucleolar localization could be made more efficient by attaching it to the HIV-1 Rev NoLS. Moreover, we wanted to ascertain whether addition of a different well-defined NoLS to the p53 protein would have an impact on its association with the FCs portion of the nucleolus. To this end we compared nucleolar localization of p53 to that of p53 with the HIV-1 Rev NoLS attached to its C terminus (Rev NoLSC'p53; Fig. 2A). Constructs expressing either wild-type p53 or Rev NoLSC'p53 were introduced into H1299 cells with and without MG132 treatment. Levels of p53 protein were assessed by western blotting analysis (Fig. 2B). MG132 treatment again caused either no significant change or a modest reduction in ectopic p53 protein levels both with and without the Rev NoLS. By contrast, treatment with this inhibitor led to a marked increase in the levels of endogenous HSP70, in agreement with previous results showing induction of heat-shock response by proteasome inhibitors (Lee and Goldberg, 1998). At present we do not know the reason for the difference found in p53 and HSP70 levels in response to MG132, although we can conclude that the localization of p53 to the nucleolus after proteasome inhibition is not due to increased levels of p53. Strikingly, in contrast to the FC nucleolar localization of untagged wild-type p53 after MG132 treatment, the Rev NoLSC'p53 protein associated with the DFC-GC within the nucleolus. The expression levels of p53 and of Rev NoLSC'p53 are similar, hence, the difference in the sub-nucleolar localization is due to the differences in the NoLSs directing those proteins to the nucleolus. Furthermore, proteasome inhibitor treatment had no effect on the efficiency of localization of Rev NoLSC'p53, which was somewhat more than that of wild-type p53 (20–25%). Thus our results indicate that the HIV-1 Rev NoLS has a dominant effect over the regions in p53 required for p53 FCs localization. It is also of interest that whereas only 20–25% of the cells expressing p53 Rev NoLSC'p53 showed p53 accumulation in the nucleolus, full-length HIV-1 Rev protein could be detected within the nucleolus in virtually every cell (Fig. 3B and data not shown). Our data thus indicate that p53 possesses both determinants for nucleolar localization that differ from those of the HIV-1 Rev protein and regions that negatively regulate such association. Thus, we conclude that distinct partners or unique pathways are involved in MG132-induced p53 nucleolar localization.

To determine whether genotoxic stress can impact sub-nucleolar accumulation of p53 and Rev NoLSC'p53, cells were treated with the topoisomerase II inhibitor and DNA intercalating agent, daunorubicin (D) and/or with MG132 (MD or M, respectively). Cells were either untreated (Fig. 3B panel 1) or treated with MG132 (Fig. 3B panel 2) or were treated with daunorubicin for 16 hours and then MG132 was added 8 hours before fixing the cells (Fig. 3B panel 3). Daunorubicin treatment diminished sub-nucleolar localization of untagged wild-type p53 and Rev NoLSC'p53 (Fig. 3B last panels, C). After daunorubicin treatment, nucleolin and Rev NoLSC'p53 localized primarily to the nucleoplasm or cytoplasm, respectively, whereas HIV-1 Rev protein localized to both the nucleolus and to the cytoplasm. Our results are consistent with previously described localization of HIV-1 Rev following treatment with α-amanitin, a Pol II and Pol III inhibitor or DRB, a Pol II inhibitor (Dundr et al., 1995). Furthermore, no p53 nucleolar localization was detected after MG132 treatment in cells that were pre-treated with either NCS (500 ng/ml) or UV [(25 J/m²), data not shown].

We previously discovered that ATP hydrolysis is required for p53 accumulation in the nucleolus in detergent-permeabilized cells (Karni-Schmidt et al., 2007). To assess the possible effect of reduction in ATP levels on p53 localization to the FCs after MG132 treatment, wild-type p53, HIV-1 Rev NoLSC'p53 and HIV-1 or Rev-GFP protein, as a positive control were transfected into H1299 cells, which were then treated with a combination of sodium azide (to inhibit cytochrome oxidase) and 2-deoxyglucose (to inhibit glycolysis) in order to reduce ATP levels in cells (Fig. 3A,B). Treatment with azide and 2-deoxyglucose (Az/DOG) had no effect on the overall levels of p53 and nucleolin (Fig. 3A, lanes a and Ma for cells treated with Az/DOG or MG132 and Az/DOG, respectively). Interestingly, treatment with Az/DOG eliminated MG132-induced p53 nucleolar association (Fig. 3B panel 5, C) whereas Rev NoLS-C'p53 still accumulated in the DFC-GC of cells treated with Az/DOG (Fig. 3B middle panel 5, C). HIV-1 Rev-GFP protein localization was also not affected by Az/DOG treatment (Fig. 3B panel 5). Thus the p53 NoLSs but not the Rev NoLS require ATP for nucleolar translocation. This result extends our findings that the Rev NoLS signal differs from that of p53 and confirms that the Rev NoLS has a dominant effect on the regions in p53 required for its nucleolar localization.

In summary, genotoxic stress had a significant inhibitory effect on the ability of wild-type p53 and Rev NoLSC'p53 but not intact HIV-1 Rev to localize to the nucleolus. p53 and Rev NoLS-C'p53 were both found to associate with the nucleoplasm whereas Rev protein associated both with nucleoli and the cytoplasm, and was excluded from the nucleoplasm. We cannot absolutely rule out that because of features present in HIV-1, the extensive nucleolar localization of Rev allows some of it to remain in the nucleolus. It is interesting that nucleolin, a well described nucleolar protein, behaves more like p53 in daunorubicin-treated cells.

The proteasome inhibitor MG132 can increase the steady state levels of ubiquitinated p53 in cells (Klibanov et al., 2001; Latonen et al., 2003; Maki et al., 1996). The primary E3 ligase for p53 is Mdm2 and monoubiquitination of p53 by Mdm2 has been reported to have profound consequences on p53 sub-cellular localization (Brooks and Gu, 2006; Li et al., 2003). To investigate a potential role of monoubiqitination on p53 nucleolar localization after treatment with proteasome inhibitor, we ectopically introduced p53 and p53 fused at its C terminus to a single ubiquitin molecule (p53-ubiqAA). We examined the localization of p53-ubiqAA with and without treatment with the proteasome inhibitor MG132. We found that in the absence of MG132 treatment p53-ubiAA was localized to the cytoplasm as well as the nucleoplasm as has been shown before (Li et al., 2003). Interestingly, however, after MG132 treatment p53-ubiqAA localized to the nucleolus and was associated with the FCs to the same extent as wild-type p53 (∼15%; Fig. 4). Thus, although we cannot conclude that monoubiquitination of p53 is required for its nucleolar localization, our data clearly show that monoubiquitinated p53 is not precluded from associating with nucleoli.

Previously, Benninghoff et al. reported that p53 can be detected in nucleoli of bladder and prostate cell lines (Benninghoff et al., 1999). As described above, p53 can be detected in the H1299 lung epithelial cancer cell line. It was therefore of interest to determine whether p53 can be detected in nucleoli from human bladder or lung carcinoma tissue sections. Indeed, in a subset of these tissues (five out of 48 bladder, five out of five lung), we detected p53 association with ∼10% of nucleoli in invasive areas, using immunohistochemistry (Fig. 5A). When we co-stained the tissues with anti-p53 and anti-UBF antibodies (as described in Materials and Methods) we detected co-localization of p53 and UBF by immunofluorescence (Fig. 5B). These experiments demonstrate that p53 can associate with nucleoli in some human invasive cancer subtypes.

In this study, we have shown that p53 localizes to a specific sub-nucleolar compartment (the FC) after proteasome inhibition. This localization is an active process that occurs in an ATP-dependent manner. Furthermore, we observed that treatment of cells with daunorubicin eliminated the association of p53 with the FCs of the nucleolus. Our results also show that the Rev NoLS differs from the nucleolar determinants on p53 in that the function of the Rev NoLS is unaffected by the proteasome inhibitor MG132 and it localizes within the GC and DFC of nucleoli. Furthermore, the Rev NoLS exerts a dominant effect when fused to p53. In addition, our results suggest that p53 monoubiquitination does not restrict p53 from associating with the nucleolus. Finally, we show that p53 localizes to the nucleolus in some invasive human lung and bladder carcinomas.

Multiple groups have reported that Mdm2-E3 ligase activity is crucial for both nuclear export and degradation of p53 (Boyd et al., 2000; Geyer et al., 2000; Haupt et al., 1997; Honda et al., 1997; Lohrum et al., 2001). Interestingly, Mdm2 primarily catalyzes the attachment of multiple single ubiquitin molecules to p53. Yet, the role of monoubiquitination on p53 is still a subject of investigation. At present, no transcriptional function for the monoubiquitination of p53 has been reported. Furthermore, monoubiquitinated p53 is not a substrate for the proteasome. This raises the question of whether this modification is necessary for p53 cytoplasmic accumulation. A study by the group of W. Gu showed that monoubiquitination of p53 by low levels of Mdm2 might be associated with p53 nuclear export, and high levels of Mdm2 induce polyubiquitination required for p53 degradation (Li et al., 2003). Additionally, these authors showed that a p53-ubiquitin fusion protein similar to the one used in our experiments that mimics monoubiquitinated p53 was found to be localized in the cytoplasm in an Mdm2-independent manner (Li et al., 2003). Other reports confirm that attachment of p53 to ubiquitin causes relocalization of p53 to the cytoplasm in both human and mouse cells (Joseph and Moll, 2003; Joseph et al., 2003). In our assay in which MG132 redirects p53 to nucleoli in a subset of cells, ectopic expression of p53 attached to ubiquitin did not prevent nucleolar localization of p53 after MG132 treatment. Since we could not detect significant amounts of a slower migrating form of p53 after MG132 treatment, this suggest that only a minor amount of the steady state level of p53 in our experiments is monoubiquitinated after this treatment. If that is the case then we suggest that ubiquitin is not required for p53 nucleolar localization, nor does it prevent it. We cannot rule out, however, that ectopic expression of wild-type p53 in H1299 cells results in monoubiquitination of p53 as the levels of endogenous Mdm2 induced by p53 are insufficient for p53 degradation and it is the monoubiquitinated form of p53 that is found associated with nucleoli. Further experiments will be required in order to clarify the role of ubiquitin, if any, in the association of p53 with nucleoli.

Under stress conditions such as proteasomal inhibition where multiple proteins cannot be degraded, limiting ribosomal biosynthesis might be beneficial to the cell and important for its survival. Regulating rRNA biogenesis could be an important step in regulation of ribogenesis. We observed p53 to be associated with the FCs of the nucleolus, a sub-nucleolar domain enriched in Pol I, the polymerase responsible for rRNA synthesis. Indeed, p53 has been shown to repress Pol I transcription in vitro (Budde and Grummt, 1999; Zhai and Comai, 2000). Interestingly, transcription experiments with p53 deletion mutants have indicated that the p53 carboxyl terminus is essential, but not sufficient, to repress transcription of Pol I (Zhai and Comai, 2000). This is in agreement with our previous study showing that the p53 carboxyl terminus is crucial for p53 nucleolar localization (Karni-Schmidt et al., 2007). Our results here suggest another link between p53, Pol I and ribosomal biosynthesis. It is tempting to speculate that p53 localizes to the FCs of the nucleolus in order to repress Pol I. Further investigation of the direct effect of p53 on Pol I followed by MG132 treatment may provide a greater insight into p53-mediated regulation of Pol I. Perhaps regulation of ribosomal biogenesis could be explored as yet another target for therapeutic intervention in cells with deregulated cell proliferation.

It is interesting that in both cell lines and tissues we detected approximately only 10% of cells with nucleolar p53. Although we do not, as yet, have an explanation for this distribution, we can offer some conjectures. First, p53 nucleolar localization may be cell cycle stage dependent. Second, the conformation of p53 protein may play a role as previously shown by others (Benninghoff et al., 1999; Rokaeus et al., 2006). Third, p53 modifications may regulate its association with nucleoli as we previously showed (Karni-Schmidt et al., 2007). These and other possible scenarios may also explain why no nucleolar p53 is detected in some cell lines, as listed in Table 1. It is noteworthy that whereas HIV-1 Rev protein displays highly efficient nucleolar association, p53 fused to HIV-1 Rev NoLS nucleolar localization is a much less frequent event (Fig. 2D, Fig. 3C). Elucidating the mode(s) of p53 nucleolar localization will be a direction for future studies.

IMR90, LNCAP, WI38, U2OS and HCT-116 cells were grown and maintained in DME medium supplemented with 10% fetal bovine serum. H1299 cells (p53-null) expressing tetracycline-regulated wild-type p53, or several tumor-derived mutant forms of p53, including p53^H175, p53^S249, p53^A143 and p53^S248 were generated as previously described (Chen et al., 1996; Resnitzky et al., 1994). Cells were grown and maintained in DME medium supplemented with 10% fetal bovine serum, puromycin (2 μg/ml; Sigma), G418 (300 μg/ml; Invitrogen), and tetracycline (4.5 μg/ml; Sigma). To induce p53, cultures were washed twice with tetracycline-free medium following plating in the above medium lacking tetracycline for 24 hours. HT-29 cells (p53^H273) were grown in McCoy medium supplemented with 10% fetal bovine serum, and T98G (p53^I237) cells were grown in MEM with 2 mM L-glutamine supplemented with 10% fetal bovine serum. To inhibit proteasome function, cells were incubated with MG132 (10 μM) for 8 hours or as described in the text. For ATP depletion experiments, cells were washed twice with PBS and were then incubated for 45 minutes at room temperature with 5 mM sodium azide (NaN₃) and 1 mM 2-deoxyglucose. To induce cellular stress, cells were treated with 0.22 μM daunorubicin (Sigma) for 16 hours.

Cells were washed twice with PBS and then collected by scraping into cold extraction buffer [10 mM Tris (pH 7.5), 1 mM EDTA, 400 mM NaCl, 10% glycerol, 0.5% NP40, 5 mM NaF, 1 mM DTT and 0.1 mM PMSF] followed by centrifugation at 13,000 rpm for 10 minutes at 4°C in a tabletop centrifuge. Supernatants were resuspened in 3× protein sample buffer [30% glycerol, 2 M β-mercaptoethanol, 12% SDS, 500 mM Tris (pH 6.8) and 0.5 mg/ml Bromophenol Blue] and boiled for 10 minutes at 95°C. The samples were loaded onto SDS-polyacrylamide gels and transferred to nitrocellulose for 3 hours. SDS-PAGE and immunoblotting procedures were performed under standard conditions. Monoclonal anti-p53 antibodies PAb 1801 and PAb DO-1 were used to detect p53. The anti-nucleolin antibody was used as described previously (Ghisolfi-Nieto et al., 1996). An anti-actin antibody (Sigma) was used as a loading control and anti-HSP70 (Santa Cruz Biotechnology) was used as a control to proteasome inhibition treatment.

Fourty-eight hours after plating in 60 mm culture dishes, cells were washed twice with PBS followed by incubation with 4% paraformaldehyde (Sigma) for 15 minutes. Cells were then washed three times with PBS, treated with PBS plus 0.5% Triton X-100 for 1.5 minutes and blocked with 0.5% BSA (Sigma) in PBS for 30 minutes at room temperature prior to incubation with 50 μl of the diluted primary antibodies, PAb DO-1, PAb 1801 and anti-nucleolin or anti-UBF (Santa Cruz) as indicated for 1 hour at room temperature. The coverslips were then washed three times with PBS and double staining was performed by incubating the coverslips for 1 hour with 50 μl of diluted (1:100) secondary Cy5 goat anti-mouse IgG antibody (Jackson Immunoresearch) and Alexa Fluor® 488 goat anti-rabbit (Molecular Probes). Coverslips were washed three times with PBS and mounted with 10 μl cold 50% glycerol. The images were collected and analyzed using confocal laser scanning microscopy (Olympus, model 1X70) and Fluoview software. In order to directly visualize nuclei and nucleoli, differential interference contrast (DIC) images were taken in parallel. To quantify the data, p53 proteins were analyzed by visually screening fields of 200 cells or more and choosing representative images. Cells were scored as positive for p53 nucleolar localization if p53 was detected in one or more nucleoli.

Immunohistochemical and immunofluorescence analyses were conducted on formalin-fixed and paraffin-embedded tissue sections from human lung and bladder carcinomas. Sections (5 μm) were de-paraffinized and submitted to antigen retrieval by steam treatment for 15 minutes in 10 mM citrate buffer at pH 6.0. Subsequently, slides were incubated in 10% normal horse serum (for immunohistochemistry) or normal donkey serum (for immunofluorescence) for 30 minutes, followed by incubation with primary antibody overnight at 4°C. The primary antibodies used were mouse monoclonal anti-p53 (PAb1801; Calciochem) for immunohistochemistry, and rabbit polyclonal anti-p53 (Santa Cruz) and mouse monoclonal anti-UBF (Santa Cruz) for immunofluorescence. Slides were then incubated with Alexa Fluor® 488 anti-rabbit and Alexa Fluor® 594 anti-mouse for 30 minutes (Invitrogen) and then mounted with Mounting Medium for Fluorescence (Vector Laboratories, Inc.). For immunohistochemistry, slides were incubated with biotinylated anti-mouse immunoglobulins for 30 minutes (Vector Laboratories, Inc.) followed by avidin-biotin peroxidase complexes (Vector Laboratories, Inc.) for 30 minutes. Diaminobenzidine was used as chromogen and Hematoxylin as the nuclear counter-stain. The immunohistochemistry images were collected using a standard light microscope and the immunofluorescence images were collected and analyzed as stated above using confocal laser scanning microscopy (Olympus model 1X70) and Fluoview software.

Transfections were performed using Lipofectamine 2000 reagent (Invitrogen) in accordance with the manufacturer's protocols. Expression vectors derived from pcDNA3 were pc53-expressing full-length p53 cDNA and Rev NoLS-C'p53, expressing full-length p53 with the nucleolar localization signal (NoLS) of HIV-1 Rev protein attached to its carboxyl terminus. The Rev NoLS-C'p53 construct was made using standard cloning procedures and was confirmed by sequencing. Rev-GFP was a generous gift from Dirk Daelemans (Rega Institute for Medical Research, Belgium) (Stauber et al., 1998).

A construct expressing p53 fused to ubiquitin at the p53 carboxyl terminus (p53-ubiAA) was generously provided by Wei Gu (Columbia University, New York, NY) and made as described by Li et al. except that the last two amino acids of ubiquitin were mutated to alanine (Li et al., 2003). We are grateful to all the members of the Prives lab and particularly to Ella Freulich, Masha Poyurovsky, Nicole Baptiste, Lynn Biderman, Yingchun Li, Yan Zhu and Igor Matushansky for technical assistance, valuable discussions and great suggestions. This work was supported by grants CA58316, CA87497 and CA009503 from the NIH.