Cooperation of a Single Lysine Mutation and a C-terminal Domain in the Cytoplasmic Sequestration of the p53 Protein*

Cytoplasmic sequestration of the p53 tumor suppresser protein has been proposed as a mechanism involved in abolishing p53 function. However, the mechanisms regulating p53 subcellular localization remain unclear. In this report, we analyzed the possible existence of cis-acting sequences involved in intracellular trafficking of the p53 protein. To study p53 trafficking, the jellyfish green fluorescent protein (GFP) was fused to the wild-type or mutated p53 proteins for fast and sensitive analysis of protein localization in human MCF-7 breast cancer, RKO colon cancer, and SAOS-2 sarcoma cells. The wild-type p53/GFP fusion protein was localized in the cytoplasm, the nucleus, or both compartments in a subset of the cells. Mutagenesis analysis demonstrated that a single amino acid mutation of Lys-305 (mt p53) caused cytoplasmic sequestration of the p53 protein in the MCF-7 and RKO cells, whereas the fusion protein was distributed in both the cytoplasm and the nucleus of SAOS-2 cells. In SAOS-2 cells, the mutant p53 was a less efficient inducer of p21/CIP1/WAF1 expression. Cytoplasmic sequestration of the mt p53 was dependent upon the C-terminal region (residues 326–355) of the protein. These results indicated the involvement of cis-acting sequences in the regulation of p53 subcellular localization. Lys-305 is needed for nuclear import of p53 protein, and amino acid residues 326–355 can sequester mt p53 in the cytoplasm.

Mutation of the p53 gene is among the most common genetic disorders in human cancers, including those of breast, colon, lung, and liver origin (16,17). Although mutations disrupt the function of p53, it has been shown in cultured cells that cytoplasmic sequestration of wild-type p53 protein could impair its function (18). Certain p53 mutations result in cytoplasmic sequestration of the mutant protein (19). An analysis of p53 in human breast cancers indicated that besides mutational inac-tivation, 37% of the breast cancer cells showed cytoplasmic localization of wild-type p53, suggesting an alternative mechanism of inhibiting p53 function via nuclear exclusion (20). Cytoplasmic sequestration of wild-type p53 has also been reported in neuroblastoma (21) and colon carcinoma cells (22). It appears that subcellular distribution of the p53 protein in untransformed Balb/c 3T3 and NIH 3T3 cells is tightly regulated during the cell cycle (23). Although the regulating mechanism remains unclear, these data indicate that the regulation of subcellular localization of p53 is an important mechanism in controlling p53 function.
To execute transcriptional activation, p53 must enter the cell nucleus. In p53, three potential nuclear localization signals (NLSs) 1 reside in the C terminus of the protein (Ref. 24; Fig.  1A). The major one, NLSI (PQPKKKP), when fused to cytoplasmic proteins, is able to direct the fusion protein to the nucleus (25). However, it is by no means a guarantee that p53 will enter the nucleus even with an intact NLS (26,27). This suggests that other factors could influence or regulate the subcellular localization of p53.
Several studies have suggested ways in which p53 might be sequestered in the cytoplasm. It has been reported that conformational changes may play a major role in p53 subcellular localization (28). It has also been suggested that short-lived proteins may be involved in anchorage of p53 in the cytoplasm of rat embryo fibroblasts (18,19). In neuroblastoma cells, cytoplasmically sequestered wild-type p53 was not detected by an antibody specific for the C terminus, suggesting the C-terminal domain of sequestered p53 is masked (29). Our previous data have shown that co-expression of bcl-2 and c-myc can totally inhibit p53 functions by cytoplasmic sequestration in murine erythroleukemia cells (30). It was speculated that Myc induces or activates a factor which cooperates with the membrane protein Bcl-2 to modulate the transport pathway responsible for the entry of p53 into the nucleus (31).
Taken together, these observations suggest that a cis-domain of p53 could be involved in its compartmentalization within the cell. To address this issue, we developed a system to study p53 subcellular localization by fusing p53 to the jellyfish green fluorescent protein (GFP) (32). Mutation analysis revealed that two domains, other than the NLS, interact to regulate p53 subcellular trafficking. Amino acid, Lys-305, is required for nuclear import of full-length p53, and this lysine interacts with a "cytoplasmic sequestration domain" located in the C terminus of p53.

EXPERIMENTAL PROCEDURES
Cell Culture-Three human tumor cell lines were used in this study. The MCF-7 breast cancer and RKO colon cancer cell lines express wild-type p53, whereas the SAOS-2 osteosarcoma cell line lacks p53. All * This work was supported by National Institutes of Health Grant CA67140 from the NCI. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Plasmid Construction and Mutagenesis-Full-length and truncated p53 were amplified by PCR using pC53-SN (kindly provided by A. Levine, Princeton University) as the template (33). The PCR fragments were subcloned into the BamHI and EcoRI restriction sites of the jellyfish GFP expression plasmid pK7-GFP (kindly provided by I. G. Macara, University of Virginia) (34). For the functional assay of induction of p21 expression, wild-type and Lys-305 mutated p53 genes were inserted into the pcDNA3 vector (Invitrogen). Internal deletion and point mutation were conducted by the PCR SOEing technique (35). Three PCR reactions and four primers are needed for each mutation. Two of the primers defined the boundaries of the PCR product for insertion to the expression vector, pK7-GFP. The other two primers are complementary oligonucleotides containing the desired mutation. For the point mutation of Lys-305 of p53 to Ala, for example, the first PCR reaction amplified the upstream fragment using the 5Ј boundary primer (5Ј-AATTAAGATCCATGGAGGAGCCGCAG-3Ј) and one of the mutant primers (5Ј-CCCCCAGGGAGCACTGCACGAGCACTGCCCAAC-3Ј) reading from 3Ј to 5Ј. The second PCR reaction amplified the downstream fragment using the other mutant primer (5Ј-GTTGGGCAGT-GCTCGTGCAGTGCTCCCTGGGGG-3Ј) reading from 5Ј to 3Ј and the 3Ј boundary primer. These upstream and downstream PCR products, which overlap at the mutation region, were mixed together and used as templates in the last PCR reaction with 5Ј and 3Ј boundary primers. The end PCR product with desired mutation was gel-purified and digested with BamHI and EcoRI before ligation to pK7-GFP. All constructs generated in this way were sequenced to ensure no mutation occurred except the designed mutation.
Cell Transfection and Fluorescence Microscopy-MCF-7, RKO, and SAOS-2 cells were grown on glass coverslips in six-well plates (35-mm diameter) to ϳ80% confluence and transfected using LipofectAMINE (Life Technologies, Inc.) according to manufacturer's instructions. For each transfection, 2 g of plasmid DNA and 6 l of LipofectAMINE were mixed in 1 ml of Opti-MEMI (Life Technologies, Inc.) and overlaid onto the cells. Following 4-h incubation at 37°C, 1 ml of Dulbecco's modified Eagle's medium containing 20% fetal bovine serum was added to the transfection mixture. At 48 h following the start of transfection, cells were washed with PBS and fixed by 4% paraformaldehyde for 10 min at room temperature. Cells were then washed with PBS three times and rinsed with water. After a brief air drying, cells were mounted using 50% glycerol in PBS and examined by fluorescence microscopy using a fluorescein isothiocyanate filter to detect expression of GFP fusion proteins. Fluorescent photomicrography was performed using Nikon photomicrographic equipment model H-III, and images were taken by a constant exposure time on a Kodak slide film. Subcellular distribution of the p53/GFP fusion was determined by the relative intensity of GFP in the cytoplasm and the nucleus of each individual cell.
Protein Extraction and Immunoblotting-SAOS-2 cells were used for the analysis of induction of p21 expression by wild-type or mutated p53. A plasmid, pRc/CMV-p53R175H (kindly provided by J. Y. Lin, University of Michigan), was used as a negative control for p21 induction. Forty-eight hours after transfection, cells were washed with PBS and lysed by the addition of 180 l of cold lysis buffer (1% Triton X-100, 150 mM NaCl, 50 mM Tris, pH 8.0) containing 0.2 mM phenylmethylsulfonyl fluoride and 1.0 g/ml aprotinin. Following constant agitation for 30 min at 4°C, cells were scraped from the plate and passed several times through a 25-gauge needle to disperse large cell aggregates. Total cell lysate was obtained after centrifugation for 15 min at 4°C in a microcentrifuge, and the protein concentration was determined with the Bio-Rad protein assay reagent on the basis of the Bradford method. Eighty micrograms of protein was separated by SDS-12% polyacrylamide gel electrophoresis, transferred to a polyvinylidene difluoride membrane (Hybond PVDF, Amersham Pharmacia Biotech), and probed with the anti-p21 monoclonal antibody (Santa Cruz Biotechnology) or the anti-p53 PAb1801 antibody (Oncogene). The specific proteins were detected by the ECL protocol (Amersham Pharmacia Biotech).

Utilization of Green Fluorescent Protein as a Reporter to
Analyze p53 Subcellular Localization-MCF-7 human breast cancer cells and RKO colon cancer cells were used here to study p53 subcellular localization, because these two cell lines contain wild-type p53 and can tolerate the exogenous expression of p53 (36 -39). When transfected with plasmids with p53 fused to the C terminus of GFP, both MCF-7 and RKO cells demonstrated three types of staining. In most cells, the p53 was located in only the nucleus or in both the nucleus and the cytoplasm of a cell (Fig. 1B, a and b). In a small population of cells, the p53 was located only in the cytoplasm.
Mutation of Lys-305 Resulted in Cytoplasmic Sequestration of p53-A series of N-terminal deletions of p53 were made by PCR until the deletion reached the NLSI (Fig. 1A). This maintains an intact NLSI so that the truncated protein should enter the nucleus unless the deleted sequences are functionally involved in nuclear import of p53. All the truncated p53 fragments were fused to the C terminus of GFP and transfected into MCF-7 and RKO cells. Because the results were similar between MCF-7 and RKO cells, only data obtained with MCF-7 cells are presented. The subcellular localization of the truncated p53, as determined by GFP fluorescence, are divided into three groups as nuclear (N), cytoplasmic (C), and both nuclear and cytoplasmic (NϩC) localizations. The percentage of each localization was determined from a total of 400 -500 fluorescent cells observed in several fields of a slide. The average data from at least two independent experiments are summarized in Table I. The truncated p53 was seen primarily in the nucleus until amino acid residues 301-305 (PGSTK) were deleted, which resulted in cytoplasmic fluorescence in ϳ85% of the cells (Table I; Fig. 1B, c and d). To check if there are specific amino acids associated with this cytoplasmic sequestration, mutations of single or double amino acids (from residues 300 to 305) were made in a full-length p53/GFP fusion protein. The single mutation of Pro-301 or double mutations of Pro-300 and Pro-301 to Ala did not change the localization of p53 in cell nucleus. The same phenomenon was observed when single substitution of Ser-303 or Thr-304 with Ala or double substitutions of Ser-303 and Thr-304 with Ala were made (data not shown). The mutation of Lys-305 to Ala, however, resulted in cytoplasmic sequestration of p53. Similarly, nuclear exclusion of p53 was observed when Lys-305 was replaced by Arg, Asn, Glu, or Thr (Table I; Fig. 1B, e). These results suggest that Lys-305 has a specific function in mediating nuclear localization of p53.
It has been shown that human p53 is phosphorylated by p34 cdc2 kinase at Ser-315, the residue just adjacent to NLSI (40,41). In an attempt to correlate p34 cdc2 phosphorylation with p53 nuclear transport, we have mutated Ser-315 to Ala in p53. The result indicated that Ser-315 mutation does not affect the p53 localization in MCF-7 cells (Table I), suggesting p34 cdc2 phosphorylation is not necessary for the nuclear import of p53. With respect to the possible effect of p34 cdc2 phosphorylation on cytoplasmic sequestration of Lys-305 mutated p53, a simultaneous mutation of Lys-305 and Ser-315 was made. The result showed that there is no difference between the subcellular localization of only Lys-305 mutated and both Lys-305 and Ser-315 mutated p53 (Table I). This indicated that p34 cdc2 phosphorylation is not involved in cytoplasmic sequestration of Lys-305 mutated p53.
Identification of a Cytoplasmic Sequestration Domain at the C Terminus of p53-Postulating a possible interaction of Lys-305 with the C terminus of p53, we performed a series of deletions of the C terminus of Lys-305 mutated p53 (p53K305N) distal to the major NLS (Fig. 1A). It was found that p53K305N⌬356 -393 was still located in the cytoplasm, whereas p53K305N⌬351-393 lost the ability to be sequestered in the cytoplasm and was distributed in both the nucleus and the cytoplasm of a cell (Table I; Fig. 1B, f and g), suggesting a region could be involved in cytoplasmic sequestration of Lys-305 mutated p53. To further define this region, several constructs were made by deleting various amino acids from residues 326 to 350 of p53K305N (Fig. 1A). The data indicated that any deletion between amino acids 326 -350 abrogated the complete cytoplasmic sequestration of p53K305N (Table I; Fig. 1B,  h). These results revealed that a domain extending from amino acids 326 to 355 is necessary for cytoplasmic sequestration of Lys-305 mutated p53.
Lys-305 Mutated p53 Significantly Lost the Ability to Induce the Expression of p21/CIP1/WAF1-Because Lys-305 mutated p53 predominantly remained in the cytoplasm, its transactivation ability should be abolished. To test this hypothesis, immunoblotting was performed to determine the p21 induction by wild-type or Lys-305 mutated p53 in SAOS-2 cells. The SAOS-2 cell line was chosen, because unlike MCF-7 and RKO cells, it possesses no endogenous p53, which may induce p21 expression during the transfection procedure. Using GFP fusion, it was found that the cellular localization of p53 in SAOS-2 cells was similar to that in MCF-7 and RKO cells (Fig. 2A). However, the subcellular distribution of the Lys-305 mutated p53 was somewhat different in SAOS-2 cells from that in MCF-7 or RKO cells. In contrast to the latter cells in which the mutant was located only in the cytoplasm in the vast majority of the cells, the mutated p53 was located in both the nucleus and the cytoplasm of the SAOS-2 cells with 50% of them having stronger fluorescence in the cytoplasm (Fig. 2B), suggesting Lys-305 mutation of p53 also had cytoplasmic sequestration effect in SAOS-2 cells. Although this effect in SAOS-2 cells is not as much as it is in MCF-7 and RKO cells, immunoblotting indicated that Lys-305 mutated p53 resulted in a significant loss of its transactivation activity manifested by the reduced induction of p21 expression (Fig. 3). After normalized to the amount of p53 from the immunoblots of three individual experiments, the p21 induction by Lys-305 mutated p53 was 3-fold less than that of wild-type p53. A negative control with Arg-175 mutation of p53, which loses the ability for DNA binding, showed that no p21 induction was detected by immunoblotting (Fig. 3).
Relationship between Lys-305 Mutation and the Cytoplasmic Sequestration Domain-It is of particular importance to point out that p53⌬326 -393 showed a high percentage (85%) of nuclear localization, whereas p53K305N⌬326 -393 was mainly located in both the nucleus and the cytoplasm (Table I). This indicated that the Lys-305 mutation alone can negatively affect the nuclear import of p53. By cooperating with the C-terminal sequestration domain identified above, the Lys-305 mutation caused the cytoplasmic sequestration of p53. DISCUSSION In the present study, a single amino acid, Lys-305, which can dramatically affect the cellular localization of p53, was identified. Even more interesting, it was found that this lysine interacts with a cytoplasmic sequestration domain (CSD) expanding from amino acids 326 to 355 at the C terminus of p53. This CSD has no effect unless Lys-305 is mutated. Lys-305 is not located within any of the five conserved domains of the p53 protein (42). Nevertheless, Lys-305 is conserved throughout a variety of species from human to Xenopus (42,43) (Fig. 4). In view of this striking evolutionary conservation, Lys-305 is probably a functionally significant amino acid of p53 and is most likely involved in the regulation of p53 cellular localization as evidenced from this report.
The p53 protein binds to DNA and transcriptionally activates target genes. Structurally, p53 can be divided into an N-terminal transactivation domain (residues 1-43), a central specific DNA-binding domain (residues 102-292), and C-terminal multifunctional domains including nuclear localization, tetramerization, and nonspecific DNA-binding domains. Obviously, regulation of p53 subcellular localization and DNA binding are two major ways in which the cell can control p53 function. Regulation of sequence-specific DNA binding of p53 is involved in the C-terminal region. It has been shown that the last 30 amino acid residues of p53 are the negative regulatory domains for DNA-binding. However, casein kinase II phosphorylation of amino acid 392 and protein kinase C phosphorylation of sites within the PAb421 epitope (residues 371-380) could activate specific DNA-binding ability of p53 (44,45). The fundamental aspect of p53 is its structural flexibility, which governs the biological activity. This structural or protein conformational determination of p53 function could be modified by post-translational modification (46). The involvement of C-terminal domains in regulating the DNA-binding ability of the central core domain indicates a domain-domain interaction or conformation-directed regulation of p53 function. Based on these observations, it is logical to hypothesize that p53 localization is also regulated, although the mechanism is still unclear.
To date, the only functional motifs found to be directly involved in p53 subcellular localization are nuclear localization signals at the C terminus of p53. NLSI, especially, is the key player for nuclear import of p53 (24). The nuclear protein import receptor, importin ␣, would bind to p53 at the NLSI locus. Thereafter, importin ␤ interacts with importin ␣ and mediates docking of the complex at the nuclear pore and then translocates to the nucleus involving GTP hydrolysis by Ran (47,48). Thus, the p53 NLSI needs to be accessible to importin The amino acid sequences from human, monkey, mouse, rat, chicken, and Xenopus gene A are aligned and positioned relative to the human sequence (42,43). Lys-305 of human p53 protein is indicated by a star. a Subcellular localization are catalogued into nuclear (N), a combination of nuclear and cytoplasmic (N ϩ C), and cytoplasmic (C) accumulation. The percentage of each accumulated pattern is determined as described in the text. ␣ in order for nuclear transport to occur. It has been shown that the NLS can function at a variety of positions within a protein. However, its activity could be masked in some locations (49), suggesting the function of the NLS is dependent on the protein context within which it is located. Accordingly, Lys-305 could be an amino acid in a position to help binding of importin ␣ to NLSI of p53, because its mutation negatively affects the p53 nuclear import. Structurally, Lys-305 may also influence the position of the CSD (residues 326 -355). Without the presence of Lys-305, the CSD can crucially mask the NLSI function of p53.
Another potential role for the CSD is its involvement in interactions with cytoplasmic anchoring proteins. In support of this hypothesis, a study has shown that protein synthesis is required to anchor a mutant p53 protein (a temperature-sensitive protein with a point mutation) for nuclear transport (19). It is thus possible that the Lys-305 mutation could somehow open the CSD for binding of anchoring proteins. This binding could block the interaction of importin ␣ with NLSI and subsequently hold p53 in the cytoplasm. It is also possible that Lys-305 and the CSD form a "pocket" in the p53 protein into which the anchoring protein is unable to specifically fit unless Lys-305 is functionally modified or mutated.
Coincidentally, the CSD identified in this study also contains the p53 oligomerization domain (50,51). Because of the opposite effects resulting from the CSD (nonfunctional p53 with cytoplasmic sequestration) and tetramerization domain (functional p53 for specific DNA binding), it is likely that this domain possesses two different functions. The crystal structure reveals that the oligomerization domain consists of a ␤ strand (residues 326 -333) and an ␣ helix (residues 335-354). The ␤ strand and the ␣ helix form a V-shaped structure, which associates with a second structure across an antiparallel ␤ sheet and a helix-helix interface to form a dimer. The tetramer is formed through the interaction between distinct helix-helix interface of two dimers (52,53). It is possible that some other cellular proteins could interact with p53 in a similar manner. In our case, for example, the oligomerization domain could turn into a CSD by binding to another cellular protein when Lys-305 is mutated. Studies are in progress to address these hypotheses.