Phosphatase and Tensin Homologue Deleted on Chromosome 10 (PTEN) Has Nuclear Localization Signal–Like Sequences for Nuclear Import Mediated by Major Vault Protein Free

Alterations of the tumor suppressor PTEN/MMAC1/TEP1 are common events in diverse human cancers, including carcinomas of the breast, endometrium, and prostate and glioblastoma. In addition to somatic alterations in common cancers, germ line mutations in phosphatase and tensin homologue deleted on chromosome 10 (PTEN) are associated with the dominantly inherited Cowden and Bannayan-Riley-Ruvalcaba syndromes, which are characterized by the formation of multiple benign tumors and by an increased risk of malignant breast, thyroid, and endometrial tumors. Heterozygous disruption of PTEN in mice leads to neoplasia in multiple tissues reminiscent of human Cowden syndrome (1).

PTEN is composed of a NH₂-terminal phosphatase domain within which lies the consensus phosphatase signature motif composed of a C2 domain that binds lipid vesicles and a COOH-terminal “tail” that contains a PDZ binding domain (2, 3). PTEN can dephosphorylate tyrosine-, serine-, and threonine-phosphorylated peptides in vitro and has been shown to dephosphorylate focal adhesion kinase in vivo. PTEN also dephosphorylates phosphatidylinositol (3, 4, 5)-triphosphate with specificity for the phosphate group at the D3 position of the inositol ring. Phosphatidylinositol (3, 4, 5)-triphosphate is a lipid second messenger and a regulator of the phosphatidylinositol 3-kinase/Akt pathway (1, 4). PTEN mediates G₁ cell cycle arrest and apoptosis and inhibits cell migration, spreading, and focal adhesion formation (4, 5). PTEN also regulates activation of the mitogen-activated protein kinase pathway (5).

Although many studies have observed PTEN localize to nuclear and cytoplasmic compartments, the function of nuclear PTEN and a mechanism to explain nuclear entry and exit has yet to be identified. Alterations in the localization of proteins within subcellular compartments may be a more important mechanism of activation or inactivation of cellular processes than previously thought (6). Mutations that prevent or promote subcellular localization may have serious effects on DNA repair or cell cycle checkpoints controlled by these proteins (6). Recently, for example, Rodriguez et al. (7) reported that nuclear-cytoplasmic shuttling of BARD1 regulates its proapoptotic activity and is regulated by dimerization with BRCA1. Although nuclear PTEN has been shown by several research groups, including ours, to have effects on the cell cycle (8–11), further work will be necessary to examine the relation of PTEN nuclear compartmentalization to the function of PTEN.

In a previous study, we directly showed that PTEN localizes to the nucleus and that this localization occurs with the G₀-G₁ phases of the cell cycle. In addition, we showed increases in cytoplasmic PTEN during S phase (12). Our data are corroborated by incidental observations by other researchers (13–16) who have also described differential localization with nonnaturally occurring truncated protein (Δ354-403) and serine/threonine mutants (5). In addition, PTEN has also been shown in isolated nuclei from pig muscle (17) and in developing brain cells in the mouse (18).

Despite observations of nuclear PTEN, a traditional nuclear localization sequence (NLS) has not yet been reported. Traditional sequences, such as the classic SV40 T-antigen virus motif KKKRK and alternative sequences of basic amino acids, have been identified in other proteins, such as BRCA1 (19). Bipartite sequences with basic regions (KRX10-12KRRK) separated by 10 to 12 residues have been identified in VHL (6) and p53 (20, 21) and bipartite signals in BRCA2 (22). APC protein has NLS-directed import as well as NLS-independent transport (6). Other signaling proteins have also shown nontraditional sequences, including an extended bipartite sequence in SMAD4, which directs nuclear transport coordinated with the cell cycle (23). A recent review of the known NLS and nuclear export sequences of tumor suppressor proteins described the importance of nuclear-cytoplasmic compartmental isolation to a variety of protein activities, including the ability of these proteins to coordinate the cell cycle (24). For example, mutations in TP53 cause alterations in p53 nuclear-cytoplasmic shuttling (25), whereas the phosphatase CDC25B has nuclear import and export sequences that regulate nuclear-cytoplasmic shuttling during the cell cycle in HeLa cells (26). Because we and others have observed PTEN in the nucleus, and because we have shown nuclear-cytoplasmic transport associated with the cell cycle, we hypothesize that a nontraditional NLS controls PTEN nuclear transport.

Recently, the C2 domain in PTEN has been shown to interact with major vault protein (MVP) in HeLa cells (27), and MVP was hypothesized as a general carrier molecule for nuclear-cytoplasmic transport (28). Although the cellular function of vaults is still unknown, their subcellular localization and distinct morphology point to a role for vaults in intracellular, particularly nuclear-cytoplasmic, transport (28) and suggests that it may be an important component in PTEN localization.

In an attempt to identify the mechanism of nuclear-cytoplasmic transport of PTEN, we identified putative NLS that are required for nuclear import and show that nuclear import is mediated by MVP. Our findings provide insights into the molecular mechanism by which PTEN is localized and identify the target of MVP nuclear-cytoplasmic translocation.

Cell lines and culture conditions. MCF-7 breast cancer cells expressing wild-type (WT), protein and lipid phosphatase–dead mutant PTEN:C124S (PTEN:CS), and lipid phosphatase–dead mutant PTEN:G129E (PTEN:GE) were generated as described previously (10). The MCF-7 Tet-Off cell line (BD Clontech, Inc., Palo Alto, CA), a breast cancer cell line containing a tetracycline-controlled cassette, was grown in DMEM-high glucose supplemented with 10% fetal bovine serum, 1% penicillin-streptomycin, and 100 μg/mL G418 (Invitrogen, Inc., Carlsbad, CA) at 37°C under an atmosphere containing 5% CO₂. Stable clones and transient clones incorporating the experimental pTre2Hyg and pTre2Hyg-HA constructs were selected with 400 μg/mL hygromycin B (Invitrogen). Vector expression was controlled with 2 μg/mL tetracycline. All cell lines for preparation of total cell-free extracts were collected by scraping after synchronization with 3 mmol/L hydroxyurea for 16 to 24 hours following 48-hour absence of tetracycline. The MCF-7 cell lines were used for transient transfection for plasmids encoding PTEN dephosphorylation-mimicking mutants.

Plasmid construction and transfection. Four sequences in PTEN were selected as NLS-like sequences by similarity to known nontraditional NLS and are described as NLS1 (amino acids 10-14, RNKRR → RNGNR), NLS2 (amino acids 160-164, RTRDKK → RTNDGK), NLS3 (amino acids 233-237, RREDK → RNEDK), and NLS4 (amino acids 265-269, KKDK → KNDN). To create mutations in the NLS-like sequence, WT PTEN cDNA was cloned from a previously reported vector into the BSSK+ vector (Stratagene, Inc., La Jolla, CA). Site-directed mutagenesis (QuickChange Mutagenesis, Stratagene) was done using the following primers:

• NLS1M-F: GAGATCGTTAGCAGAAACACAGGGACATATCAAGAGGATGGG

• NLS1M-R: RCCATCCTCTTGATATGTCCTGTGTTTCTGCTAACGATCTC

• NLS2M-F: GGGAAGTAAGGAACCATAGACATACAGGGAGTAACTATTCCCA

• NLS2M-R: CTGGGAATAGTTACTCCCTGTATGTCTATGGTCCTACTTCCCC

• NLS3M-F: CAGGACCCACACTACTGGAAGACATGTTCATGTACTTTGAGTTC

• NLS3M-R: AACTCAAAGTACATGAACATGTCTTCCAGTAGTGTGGGTCCTGA

• NLS4M-F: CAGAACAAGGATGCTAACACAGGACACAATGTTTCACTTTTGG

• NLS4M-R: FCCCAAAAGTGAAACATTGTGTCCTGTGTTAGCATCTTGTTCTG

The mutant cDNA was then subcloned and inserted into the pTre2Hyg and pTre2Hyg-HA vectors (BD Clontech). The constructs were confirmed by restriction mapping and sequencing. The pTre2Hyg PTEN-NLSM1, pTre2Hyg-NLSM2, pTre2Hyg-NLSM3, pTre2Hyg-NLSM4, pTre2Hyg PTEN-NLSM1M2, pTre2Hyg-NLSM1M3, pTre2Hyg-NLSM1M4, pTre2Hyg-NLSM2M3, pTre2Hyg PTEN-NLSM2M4, pTre2Hyg-NLSM3M4, pTre2Hyg-HA-PTEN, pTre2Hyg-HA-PTEN-NLSM2M3, and pTre2Hyg HA-PTEN-NLSM2M4 constructs were stably transfected into the MCF-7 Tet-Off cell line as recommended by the manufacturer's instructions. The MCF-7 and Tet-Off system were used previously in our investigation of WT PTEN in the nucleus (12) and have been part of our ongoing investigations of cell cycle components following transfection of WT PTEN and naturally occurring PTEN mutants associated with neoplasia.

SDS-PAGE, Western blot, and antibodies. Following the removal of medium, transfected MCF-7 cells were washed once and then scraped into ice-cold PBS. The cells were used for whole cell–free extracts and nuclear/cytoplasmic protein separation or stored at −80°C for future use. Nuclear and cytoplasmic proteins were isolated with a buffer extraction system and centrifugation according to the manufacturer's recommendations (NE-Per, Pierce Biotechnology, Inc., Rockford, IL). The following inhibitors were added: 0.75 mmol/L phenylmethylsulfonyl fluoride, 0.5 mg/mL benzamidine hydrochloride, 2 μg/mL each of leupeptin, aprotinin, and pepstatin, 10 mmol/L β-glycerophosphate, 0.2 mmol/L sodium orthovanadate, and 25 mmol/L NaF (Sigma Biochemical Co., St. Louis, MO). Whole cell–free protein extracts used for protein analysis were prepared according to the manufacturer's recommendation, except the inclusion of protease inhibitors (M-Per, Pierce Biotechnology). Protein concentrations were determined using BCA reagents (Pierce Biotechnology) with bovine serum albumin as a standard.

SDS-PAGE and Western blot were done according to procedures recommended by the Bio-Rad Protein III System (Bio-Rad, Inc., Hercules, CA). The monoclonal antibody 6H2.1 raised against the last 100 COOH-terminal amino acids of PTEN can recognize the WT PTEN, PTEN:C124S, and PTEN:G129E proteins (10). Antibodies against HA, P-PTEN, 3P-PTEN, Akt, P-Akt, and actin were purchased from Cell Signaling Co. (Beverly, MA) and MVP monoclonal antibody was obtained from Transduction Laboratories (Lexington, KY). Antibodies against importin complex subunits were purchased from Santa Cruz Co. (Santa Cruz, CA). For Western blot, protein (20-40 μg) was fractionated by 8% to 12% SDS-polyacrylamide gels and transferred by Trans-Blot Cell system (Bio-Rad). Nitrocellulose membranes (S&S, Inc., Keene, NH) were washed four times after the first and second antibody reactions with 0.1% TBS containing 0.1% Triton X-100. Secondary antibody was anti-mouse IgG and anti-rabbit IgG conjugated to horseradish peroxidase (Promega, Inc., Madison, WI). Membranes were incubated for 1 minute facedown in enhanced chemiluminescence substrate (Amersham Biosciences, Inc., Chicago, IL) and fluorescent signals were collected by Hyperfilm XR (Amersham Biosciences, Inc.) and quantified by using ImageQuant software version 5.1. Equal loading of wells was evaluated by staining membrane with Ponceau S reagent (Sigma Biochemical) and actin Western blotting.

Immunoprecipitation. Whole cell–free protein extracts used for immunoprecipitation analysis were prepared according to procedures recommended by Cell Signaling. After pretreatment of the sample with protein A-Sepharose beads (Sigma Biochemical), solutions (400 μL) of 5% bovine serum albumin/PBS containing antibody solution (5 μL) were added and the sample was incubated overnight at 4°C with mixing. Protein A-Sepharose [40 μL of a 40% (v/v) suspension] was added to each sample and incubated for 2 hours on ice with mixing. Immunoprecipitates were collected by centrifugation at 5,000 rpm for 5 minutes in a microcentrifuge. Pellets were washed four times in buffer [1% Triton X-100, 1% deoxycholic acid, 50 mmol/L Tris-HCl (pH 7.5), 150 mmol/L NaCl] and precipitated proteins were eluted from the beads by incubation for 10 minutes at 95°C in 100 μL of 4× SDS loading buffer [50 mmol/L Tris (pH 6.8), 8% glycerol, 1.6% SDS, 4% β-mercaptoethanol, 0.04% bromphenol blue]. Half of each precipitate was subjected to SDS-PAGE.

Immunofluorescent microscopy. Cells were prepared for indirect immunofluorescent microscopy by using a published procedure (12). The secondary antibody was goat anti-mouse IgG labeled with fluorescein, and Prolong reagents were used for storage of the signals (Molecular Probes, Inc., Eugene, OR). Fluorescent images were obtained with a Nikon Eclipse E800 fluorescence microscope equipped with a Nikon 100×/1.3 plan fluor oil immersion objective and a Diagnostic Instruments Spot camera controlled by Adobe Photoshop software. Adobe PhotoShop software was used to process images.

The lipid and protein phosphatase activities of PTEN are not required for its nuclear-cytoplasmic transport. As a first step toward identifying the mechanism involved in nuclear-cytoplasmic transport of PTEN, we examined the role of the protein and lipid phosphatase activities of PTEN in nuclear-cytoplasmic transport. PTEN has been implicated not only in suppressing cancer growth but also in regulating development, cell cycle, cell adhesion, apoptosis, and stem cell growth and differentiation (1). In all cases, it would seem that the phosphatase activity of PTEN is essential. Mutations at C124, a key residue in the phosphatase core motif, results from a single nucleotide missense mutation seen in a subset of Cowden syndrome/Bannayan-Riley-Ruvalcaba syndrome patients and primary sporadic tumors (29). Further, the PTEN:C124S (sometimes called PTEN:CS) point mutation results in a phosphatase-dead protein, with neither lipid nor protein phosphatase activity (10). It has been well documented previously that PTEN containing another missense mutation, Gly¹²⁹ → Glu¹²⁹ (PTEN:G129E or PTEN:GE), which retains protein phosphatase activity but lacks lipid phosphatase activity, can inhibit the migration of some tumor cells (30). In addition, it has been shown that the phosphatase domain is essential for the electrostatic membrane binding of PTEN and plays a role in membrane targeting as supported by experiments using the dominant-negative Cys¹²⁴ → Ser¹²⁴ (PTEN:C124S; ref. 30). These observations may therefore suggest that nuclear entry of PTEN may be related to phosphatase activity via its membrane electrostatic effects. To examine this possibility, we used MCF-7 Tet-Off breast cancer cell lines stably transfected with each of the phosphatase-mutant PTEN constructs (pcDNA3 PTEN:C124S and pcDNA3 PTEN:G129E) and compared these to cells transfected with WT PTEN and to empty vector control cells (12).

If there is a phosphatase-mediated nuclear localization defect, then nuclear PTEN levels in the PTEN:C124S and PTEN:G129E phosphatase-dead cell lines would be expected to be similar to vector control cells (MCF-7 Tet-Off cells also have endogenous WT PTEN in the nucleus) but less than that in the WT PTEN-expressing cell lines in the context that our monoclonal antibody 6H2.1 recognizes WT PTEN and both phosphatase-dead PTEN mutants. Indeed, when we did these experiments, all PTEN-expressing cells lines, irrespective of whether they had WT PTEN or each of the phosphatase mutants (PTEN:C124S or PTEN:G129E), expressed similar levels of PTEN in the nucleus and the cytoplasm (with a relative nuclear/cytoplasmic ratio of ∼1:4; Fig. 1A and B). These observations, therefore, suggest that the lipid and protein phosphatase activities of PTEN in the context of electrostatic membrane binding and membrane targeting of PTEN are not related to PTEN nuclear-cytoplasmic transport.

Phosphorylation of PTEN may not be required for its nuclear import. Phosphorylation of PTEN causes a conformation change that can result in the masking of the PDZ domain-binding site, thereby suppressing the recruitment of PTEN into the PTEN-associated complex in the plasma membrane (31). Recently, it has been shown that the COOH-terminal tail regulates PTEN activity and the mutation of three residues (S380, T382, and T383) in the COOH-terminal tail dramatically lowers the membrane association rate of PTEN (32). Thus, we were interested in examining the relation of phosphorylation to nuclear-cytoplasmic transport. To determine if phosphorylation has a direct effect on nuclear-cytoplasmic transport of PTEN and phospho-PTEN, we used MCF-7 Tet-Off breast cancer cell lines stably transfected with WT PTEN constructs (pTre2Hyg PTEN:WT) and examined nuclear and cytoplasmic distributions of PTEN and phospho-PTEN in WT PTEN-overexpressing and vector control cells.

PTEN and phospho-PTEN levels were measured in nuclear and cytoplasmic fractions, in the presence and absence of tetracycline, by Western blot analysis using a monoclonal antibody against PTEN (6H2.1) and polyclonal antibodies against phospho-PTEN (S370 and S380/S382/S384). If a nuclear localization defect was present, we would expect nuclear PTEN and phospho-PTEN expression to be at similar levels (i.e., not increased) to that of empty vector controls cells given that MCF-7 Tet-Off cells have endogenous PTEN and phospho-PTEN in the nucleus. We observed very similar results in PTEN-overexpressing and control cells (Fig. 2A). PTEN and phospho-PTEN expression in the nuclear fractions was ∼25% of that in the cytoplasmic fractions and these ratios were maintained in the absence of tetracycline in WT PTEN and vector control cells (Fig. 2A and B).

PTEN has five putative phosphorylation sites, but by using antibodies specific for PTEN phosphorylation, we were limited to examining three phosphorylation sites. To fully examine the effects of phosphorylation on nuclear translocation, HA-tagged dephosphorylation-mimicking mutants (S370A, 3A: S380A/T382A/T385A, 4A: S380A/T382A/T383A/S385A, and 5A: S370A/S380A/T382A/T383A/S385A) were transiently transfected into the MCF-7 breast cancer cell line. Subcellular localization was examined by using indirect immunofluorescence microscopy with anti-HA antibody and anti-mouse IgG fluorescein conjugates (Fig. 2C). 4′,6-Diamidino-2-phenylindole (DAPI) staining was also done to unequivocally visualize the nucleus. In PTEN:WT cells, HA-PTEN localized to the cytoplasm and nucleus and some portions of the plasma membrane. Yet, no nuclear localization defects were detected in the cells containing the phosphorylation mutants, an observation that was further confirmed by Western blot analysis (data not shown). These results verify those shown in Fig. 2A. Thus, based on the combined data, we conclude that phosphorylation of PTEN does not affect PTEN nuclear localization.

Putative single site nuclear localization sequence mutations in PTEN do not affect its nuclear-cytoplasmic transport. To determine if the sequences identified as NLS-like can act as part of the targeting signal for the movement of PTEN across the nuclear membrane, four putative NLS sequences within PTEN were selected based on sequence similarity to other nontraditional NLS. Mutant constructs were generated using PCR-directed mutagenesis to incorporate mutations in the regions encompassing the putative NLS sequence as follows: NLSM1, amino acids 11 to 15 RNKRR → RNGNR; NLSM2, amino acids 159 to 164 RTRDKK → RTNDGK; NLSM3, amino acids 233 to 237 RREDK → RNEDK; and NLSM4, amino acids 266 to 269 KKDK → KNDN (mutated amino acids are indicated in bold). WT PTEN constructs and empty vector controls described above were always used for comparison as positive and negative controls, respectively (12). Each of the clones was tested for the presence of PTEN in cell-free extracts, nuclear and cytoplasmic fractions, and in the presence (+Tet) and absence (−Tet) of tetracycline, by Western blot analysis. Nuclear and cytoplasmic proteins were examined in clones of each of the four different NLS mutant cell lines.

In initial experiments, the expression of PTEN and phospho-PTEN was different in each of the clones. We found that each of the mutant NLS PTEN constructs were active because phosphorylation of Akt decreased, whereas Akt expression remained steady following the increase of PTEN expression (ref. 10; Fig. 3C). If a nuclear localization defect resulted from each NLS mutation, it would be expected that nuclear PTEN and phospho-PTEN levels would be decreased compared with WT PTEN-expressing cells but would be similar to that in empty vector control cells given that MCF-7 Tet-Off cells have endogenous PTEN and phospho-PTEN in the nucleus as well. Instead, similar levels of nuclear and cytoplasmic PTEN were observed in cell lines expressing WT PTEN and each of the NLS mutant proteins (Fig. 3A and B). These observations suggest that single site NLS-like sequences do not mediate PTEN import into the nucleus.

NLS4 (amino acids 265-269 KKDK), together with NLS2 (amino acids 160-164 RTRDKK) or NLS3 (amino acids 233-237 RREDK), is required for nuclear import of PTEN. Because many tumor suppressor proteins have bipartite NLS, consisting of two basic amino acid groups separated by one spacer of 12 amino residues, we wanted to examine the effects of combining the NLS mutations. Each of the single PTEN NLS mutant constructs was used for double NLS mutant constructions.

We confirmed that these double NLS mutants maintained phosphatidylinositol 3-kinase–dependent PTEN activity by examining the P-Akt levels. We found that all the double mutants resulted in a decrease in P-Akt levels when expressed (Fig. 4). The PTEN proteins in each of these double NLS mutants, like those of the NLS single mutants, were active. Yet, the expression level of exogenous PTEN differed with each clone. However, double NLS mutants NLSM2-NLSM4 (NLSM2M4) and NLSM3-NLSM4 (NLSM3M4) exhibited defective PTEN nuclear import (Fig. 4B) as evidenced by the level of nuclear PTEN being the same in the presence or absence of tetracycline and PTEN levels in cytoplasmic fractions were increased ∼4-fold in the absence of tetracycline. This indicates that the NLSM4 (amino acids 265-269 KKDK) is necessary but not sufficient for nuclear localization of PTEN. NLSM4 requires either NLSM2 (amino acids 160-164 RTRDKK) or NLSM3 (amino acids 233-237 RREDK) to mediate PTEN nuclear import.

To confirm this notion further, each of the double NLS mutants was transferred to pTre2Hyg-HA and stably transfected into MCF-7 Tet-Off breast cancer cell lines. Nuclear-cytoplasmic localization was examined using Western blot analysis and indirect immunofluorescence microscopy with anti-HA antibody and anti-mouse IgG fluorescein conjugates. The mutated and HA-tagged PTEN proteins in each of the double NLS mutants were active because phosphorylation of Akt decreased, whereas Akt expression remained steady following the increase of PTEN expression (Fig. 4A). As shown in Fig. 4C, HA-tagged PTEN was in the nucleus and cytoplasm in WT PTEN cells but did not localize to the nucleus in NLSM2M4 and NLSM3M4 double mutants. According to the indirect immunofluorescence microscopic study, anti-HA antibodies localized to the cytoplasm and nucleus with additional focus to portions of the plasma membrane in WT PTEN cells. HA-PTEN:NLSM2M4 and HA-PTEN:NLSM3M4 exhibited nuclear localization defects, which were confirmed by Western blot analysis (Fig. 4D). These data confirm that NLS4 (amino acids 265-269 KKDK) together with either NLS2 (amino acids 160-164 RTRDKK) or NLS3 (amino acids 233-237 RREDK) provide NLS for PTEN.

Importin complex may not be required in the nuclear-cytoplasmic transport of PTEN. Because proteins of the importin family organize the transport of NLS-containing proteins into the nucleus, we were interested in investigating the possible influence of the importin-α/β complex on the nuclear import of PTEN and phospho-PTEN. To assess the influence of PTEN NLS mutations on importin-α/β binding, the interaction between importin-α and importin-β was analyzed using immunoprecipitation. To determine the mechanism underlying the nuclear import of PTEN, PTEN NLS double mutants were examined. In initial experiments, we examined each of the importin complex proteins in cell-free extracts and found that each of the cell lines had similar levels of each subunit (Fig. 5A). Surprisingly, there were small variations in the intensity of bands from experiment to experiment, but overall our results indicated that PTEN and phospho-PTEN equally interact with importin-α and importin-β in each of the four cell lines regardless of mutation status, indicating that the NLS does not directly interact with importin. It may stand to reason that importin proteins interact with basal levels of PTEN and phospho-PTEN or that importin proteins mediate another function of PTEN (Fig. 5B and C).

Major vault protein mediates the nuclear import of PTEN. Several groups have reported the association of vaults with the nucleus and particularly the nucleoli, the nuclear membrane, and/or the nuclear pore complex (33). Recently, it has been reported that PTEN associates with MVP (27) and MVPs have been hypothesized as carrier molecules for nuclear-cytoplasmic transport (28). We examined if MVPs interact with PTEN and mediate PTEN nuclear-cytoplasmic transport. In initial experiments, MVP expression was examined by Western blotting using anti-MVP monoclonal antibody. Interaction with PTEN using immunoprecipitation was examined in total cell-free extracts in the absence of tetracycline.

To determine the mechanism underlying the nuclear import of PTEN, we first analyzed the ability of PTEN NLS single-site mutants to interact with MVP. MVP expression in each cell line was similar to vector control cells and the interaction of overexpressed WT and each single site NLS mutant PTEN with MVP was similar (Fig. 6A and B). These observations indicate that each single site mutation retains the ability to interact with MVP. Next, similar experiments were done but with each double NLS mutant. In these experiments, when PTEN was overexpressed, MVP expression in WT PTEN-expressing cells was similar to that of the vector control but at a higher level than in the nuclear localization defect cell lines expressing NLSM2M3 or NLSM2M4 (Fig. 6A and C). As shown in Fig. 6D, PTEN and phospho-PTEN in the WT PTEN-overexpressing cell lines, and to a lesser extent in vector control cells (which have endogenous PTEN), interact with MVP. In contrast, however, PTEN containing either NLSM2M4 or NLSM3M4 mutations did not interact with MVP. MVP-PTEN interactions were similar in each NLS double mutant cell lines as in the vector control cells, reflecting only the endogenous levels of WT PTEN present (Fig. 6D). To confirm these results, we did similar experiments in HA-tagged NLS double mutant cells. Again, WT PTEN was found to interact with MVP but NLS double mutants did not (Fig. 6C and D). On further analysis, MVP was found to localize densely in the nucleus as is the case with PTEN (Fig. 6E). Taken all together, our observations confirm that MVP mediates PTEN and phospho-PTEN nuclear import.

According to previous observations, PTEN is known to exist in the nucleus and peaks there with G₀-G₁ and nadirs during S phase in MCF-7 cells (12). Based on these observations, we investigated signals and partners for PTEN nuclear transport. We examined if the phosphatase activities of PTEN and/or phosphorylation of PTEN are key regulators of PTEN nuclear localization. PTEN membrane localization is highly dependent on two phosphatase functions of PTEN, and other tumor suppressor proteins, such as p53, can be regulated by phosphorylation (20). If nuclear localization defects exist, nuclear PTEN and phospho-PTEN protein are expected to be similar to (i.e., no higher than in) non-PTEN-overexpressing cells because MCF-7 Tet-Off cells have endogenous PTEN and phospho-PTEN in the nucleus. Our data suggest that the phosphatase activity of PTEN was not related to PTEN nuclear-cytoplasmic transport (Fig. 1) and that the degree of PTEN phosphorylation was not related to the nuclear localization of PTEN (Figs. 2, 3B, and 4B and C). This may be partially explained by PTEN localizing densely in membrane compartments (Fig. 4D, WT PTEN) and that membrane targeting of PTEN may use a different mechanism than nuclear-cytoplasmic shuttling.

In identifying the mechanism for PTEN nuclear-cytoplasmic transport, we looked for four amino acid sequences that could be involved in a nontraditional NLS because PTEN does not have a traditional NLS. The nuclear localization mechanisms of other tumor suppressor proteins often involve the recognition of NLS, binding of importer proteins (karyopherins) and the use of karyopherins to traverse the nuclear pore complex. The traditional NLS is a repeat of positive amino acids with flaking proline, but proteins may have a less traditionally structured region (24). Nontraditional nuclear import sequences have been identified in SMAD2, SMAD3, SMAD4, and Erk2 proteins. The import of these proteins is mediated by lysine-rich sequences, which interact with importin-α in traditional and nontraditional nuclear import. Mutations in the lysine-rich regions have been reported to decrease nuclear import of these proteins (24).

Four NLS-like sequences within PTEN were selected for mutation according to sequence similarity to other nontraditional NLS. NLSM1 (amino acids 10-14 RNKRR) incorporates a change in an arginine that has been reported previously to be important for function (34). NLSM2 (amino acids 265-269 RTRDKK) has a sequence homology surrounding the putative NLS, which is similar to the residues surrounding the CDC25B phosphatase NLS (26) and is a common structural feature of other phosphatases that coordinate water hydrolysis of phosphates at the active site (34). NLSM3 (amino acids 233-237 RREDK) lies within the C2 domain and is in a lysine-rich region, which are key interaction domains in nontraditional transport systems (24). NLSM4 (amino acids 265-269 KKDK) lies within the C2 domain, which may position the active site near the membrane (28) and has been shown to interact with the MVP in HeLa cells (27). Interestingly, however, unlike many proteins with nontraditional NLS, mutations in each of the four NLS-like regions of PTEN do not alter entry into the nucleus (Fig. 3). When individual NLS mutations were combined, we found that PTEN nuclear localization was affected, thereby indicating that nuclear import requires two NLS-like sequences acting in concert (Figs. 4 and 6). In particular, we have shown that the combination of NLS2 and NLS4 or NLS3 and NLS4 is required for nuclear localization.

In attempts to find the partners that mediate PTEN nuclear localization, WT PTEN and NLS single mutants was observed to interact with MVP (Fig. 6A and B) and transported to the nucleus, whereas nuclear localization defect mutants did not (Figs. 4B-D and 6D and E). These results indicate that MVP mediates both PTEN and phospho-PTEN nuclear import. The association of other proteins with vault proteins has shown that this binding can convey the protein near the nuclear membrane for transport. The idea of vaults taking part in a nuclear-cytoplasmic transport route is based on observations in rat fibroblasts. In these cells, vaults were detected in close proximity to the nuclear pore complex (28). Corresponding studies in developing sea urchin embryos, where MVP was copurified with ribosomes, suggested that vaults are involved in the transport of ribosomes (28). In more recent studies, the coiled coil of MVP was found necessary for vault tube formation in the 7 ± 2 μm penetrating nuclear membrane (33), which suggested that vaults could operate as cytoplasmic and/or nuclear membrane-associated transport (28). However, until now, no other studies corroborated this hypothesis and no targets for nuclear-cytoplasmic transport were known. Our studies do support this hypothesis and furthermore identify PTEN as a target for MVP-mediated nuclear-cytoplasmic transportation (Fig. 6).

Interaction via the C2 domain of PTEN and EF hands in MVP is not a usual phenomenon. However, some proteins do contain the C2 domain and EF hands, such as the phospholipase C family proteins. Crystal structures have revealed that the second lobe of the EF hands of phospholipase C-δ1 interact with its C2 domain, but the strength of this interaction is very weak (27). The C2 domain of PTEN has structural similarity to that of phospholipase C-δ1, with a root mean square deviation of 1.9 Å for 75 αC atoms (35). Therefore, a more stable interaction with the M2 and M3 regions might be necessary for MVP-PTEN interactions (Fig. 6D).

PTEN in WT and control cells, as well as in both PTEN nuclear localization defect cells (NLSM2M4 and NLSM3M4), was found to interact with importin-α and importin-β (Fig. 5B) and all have similar steady-state levels of the importin complex subunit (Fig. 5A). The interaction strength and quantity in all of these cells was found to be very similar. This observation may lead us to postulate that PTEN could induce a damage signal and nuclear transport of this signal may be mediated by the importin interaction with PTEN. Importin, once thought to be exclusively a nuclear transport receptor, is emerging as a global regulator of diverse cellular functions. Importin-β acts positively in multiple interphase roles: in nuclear import, as a chaperone for highly charged nuclear proteins, and as a potential motor adaptor for movement along microtubules (36). Importin-β plays a negative regulatory role in mitotic spindle assembly, centrosome dynamics, nuclear membrane formation, and nuclear pore assembly. More recent work has shown that importin-β plays a role in transducing damage signals from the axons of injured neurons back to the cell body (36). However, further work will be necessary to eliminate the possibility that importin is responsible for PTEN nuclear import (Fig. 5).

Despite the observations provided by the nuclear PTEN localization defective mutants, no nuclear export defects were revealed in our mutation studies. Our nuclear localization study carried out on nontraditional NLS does however provide evidence for MVP molecules as carrier proteins. PTEN does not have a traditional nuclear export sequence, so further work will be necessary to examine the mechanism responsible for PTEN nuclear export. According to our previous observations, PTEN is known to exist in the nucleus and peaks there with G₀-G₁ and nadirs during S phase in MCF-7 cells and that it is related with cell cycling (12). More recently, it was shown that PTEN induces cell cycle arrest by decreasing the level and nuclear localization of cyclin D1 (37). However, further work will be necessary to identify the mechanism of regulation of cyclin D1 by nuclear PTEN.

The results presented here are the first to indicate a mechanism for PTEN nuclear import and the identification of a substrate for MVP. Our observations suggest that NLS4 (amino acids 65-269 KKDK) necessarily pairs with NLS2 (amino acids 60-164 RTRDKK) or NLS3 (amino acids 233-237 RREDK) to direct the import of PTEN to the nucleus mediated by MVP. Further, we provide definitive evidence that MVP is the nuclear-cytoplasmic transport protein that mediates PTEN nuclear import.

Grant support: Susan G. Komen Breast Cancer Research Foundation grant BCTR 2000 462, American Cancer Society grant RSG-02-151-01-CCE (C. Eng), National Cancer Institute grant P30CA16058 (Comprehensive Cancer Center), and Doris Duke Distinguished Clinical Scientist Award (C. Eng).

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.

We thank Dr. William R. Sellers for kindly providing the pGL5L PTEN dephosphorylation-mimicking mutant plasmids, Michelle Sinden and Dr. Kristin Waite for helpful discussions, and Dr. Waite for critical review of multiple drafts of the article.