Impaired p53 binding to importin: A novel mechanism of cytoplasmic sequestration identified in oxaliplatin-resistant cells

Abstract

Previous studies have described one nuclear localization signal (NLSI) in p53 and speculated on two additional sites termed NLSII and NLSIII. Drug resistant KB cells selected with cisplatin or oxaliplatin were found to have increased p53 levels and in oxaliplatin-selected cells, a larger p53 predominantly in the cytoplasm. In oxaliplatin-selected cells a single nucleotide deletion in the sequence encoding amino acid 382, part of NLSIII, resulted in a frame shift and a 420 amino acid protein (p53⁴²⁰). We investigated explanations for the cytoplasmic sequestration of p53⁴²⁰ while assessing the role, if any, of NLSII and NLSIII in p53 nuclear import. We found that neither NLSII nor NLSIII are essential for p53 nuclear localization. Furthermore we confirmed p53⁴²⁰ is able to tetramerize, transactivate a p21 promoter, bind dynein and that the reduced nuclear accumulation is not a consequence of increased p53 nuclear export. However, the association of p53⁴²⁰ with importin-β, essential for nuclear import, was significantly impaired. We conclude that despite sequence similarity to consensus NLSs neither NLSII nor NLSIII have roles in p53 nuclear transport. We also identified impaired association with importin as a novel mechanism of p53 cytoplasmic sequestration that impairs nuclear transport rendering cells functionally deficient in p53.

INTRODUCTION

The level and localization of p53, often referred to as “the guardian of the genome”, is tightly regulated (O'Brate and Giannakakou, 2003). In the absence of stress, p53 levels are low and the protein likely exists in an inactive conformation. Following DNA damage p53 is activated and translocates to the nucleus where it initiates sequence-specific transcription of genes involved in cell cycle arrest or apoptosis (el-Deiry, 1998). The critical role of p53 in maintaining homeostasis in normal cells is evidenced by the fact that over 50% of all human cancers harbor mutant p53 proteins (Hollstein et al., 1991; Levine et al., 1991). It is not surprising that the localization of p53 is tightly controlled and greatly complex.

Normally, p53 shuttles between the cytoplasm and the nucleus. In the cytoplasm, p53 oligomerization is followed by its association with dynein, the minus-end directed motor protein that travels on microtubules (Giannakakou et al., 2000). As dynein cargo, p53 reaches the peri-nuclear region where it associates with importin, through one or possibly more than one nuclear localization signal (NLSI, NLSII, NLSIII) and translocates to the nucleus (Liang and Clarke, 1999). Previous studies have concluded that NLSI is most important in the interaction with importin, assigning supportive roles to NLSII and NLSIII (Shaulsky et al., 1990). In the nucleus, p53 initiates sequence-specific gene transcription of numerous genes including Mdm2. p53 ubiquitination by Mdm2 induces a conformational change that allows exportin 1 to bind to the nuclear export signal (NES) of p53 (Stommel et al., 1999). Exportin 1 translocates p53 to the cytoplasm where it undergoes proteosomal degradation (Woods and Vousden, 2001). Alterations in this elaborate process can lead to uncontrolled cell growth and malignant transformation.

An extensive body of data indicates acquired mutations in p53 confer drug resistance (Bristow et al., 2003; Oggionni et al., 2005; Scata and El-Deiry, 2007). Additionally drug exposure both in vitro and in patients, often leads to the emergence of clones harboring p53 mutations, an acquired change that is viewed as advantageous. In examining a panel of platinum resistant cell lines we had previously demonstrated acquired mutations in the DNA-binding domain of p53 in cisplatin resistant human ovarian carcinoma cells. However, in oxaliplatin selected human cervical carcinoma cells, we discovered a mutation that alters the intracellular distribution of p53, sequestering it in the cytoplasm. Using this as a model, the present study was designed to further our understanding of the process of p53 intracellular trafficking.

RESULTS

Cisplatin and oxaliplatin resistant KB cell lines possess a mutant p53 protein

KB is a subclone of human cervical carcinoma HeLa cells. We began by selecting KB-3-1 cells, a subclone of KB cells with either cisplatin or oxaliplatin. Incremental drug increases resulted in the isolation of one cisplatin and three oxaliplatin resistant cell lines designated KB-CP20, KB-OX20, KB-OX60 and KB-OX80. The three oxaliplatin-selected sublines were of the same lineage. The cross-resistance profile of the selected cell lines is shown in Table 1. Compared to parental KB-3-1 cells, KB-CP20 cells were more than 100-fold resistant to cisplatin while KB-OX80 cells were almost 300-fold resistant to oxaliplatin. As shown in figure 1A, immunoblot analysis demonstrated increased levels of p53 protein in all the resistant sublines, and in the oxaliplatin selected cell lines the p53 protein had a higher molecular weight. As positive controls, we used A549 and KB-3-1 cell lines that harbor a wild type p53. DNA sequencing identified in the KB-CP20 cells an acquired missense mutation (V172F) in the DNA binding domain and in the oxaliplatin selected cells a deletion of a single nucleotide in the sequence encoding for amino acid 382 in the putative nuclear localization signal III (NLSIII) region. The deletion altered the putative NLSIII and caused a frame-shift that resulted in a p53 protein larger by 27 amino acids designated p53⁴²⁰, explaining the larger size observed in the immunoblot analysis. Since this deletion is present in all three oxaliplatin resistant sublines, we assume it occurred early in the selection.

Table 1.

Tabulation of absolute and relative resistance of the cisplatin- and the oxaliplatin-resistant cells

Tabulation of absolute and relative resistance
	Cisplatin IC50	RR	Oxaliplatin IC50	RR	Adriamycin IC50	RR
KB-3-1	6.9 × 107 M ± 0.1	1	9.0 × 107 M ± 0.01	1	3.2 × 108 M ± 0.03	1
KB-CP20	8.3 × 105 M ± 14.3	120	4.4 × 105 M ± 5.3	49	5.2 × 107 M ± 0.04	16
KB-OX20	1.1 × 105 M ± 1.0	16	5.7 × 105 M ± 4.6	64	4.4 × 108 M ± 0.01	1
KB-OX60	3.3 × 105 M ± 4.9	48	1.4 × 104 M ± 10.0	155	3.1 × 108 M ± 0.03	1
KB-OX80	3.3 × 105 M ± 1.0	48	2.5 × 104 M ± 20.8	274	1.6 × 107 M ± 0.06	5

Figure 1.

Protein expression and sub-cellular fractionation of p53 in A549, KB-3-1 cells and its drug resistant sublines. (A) Western blot analysis shows increased p53 levels of in all four resistant cell lines. In the KB-OX cells, the p53 bands run slower indicative of a longer protein length for this mutant. (B) Sub-cellular fractionation and western blot analysis were performed on untreated (−) or on treated (+) cells. Cells were treated with 400 ng/ml of adriamycin for four hours prior to separation into a cytoplasmic fraction (C) and a nuclear fraction (N), and probed for p53, tubulin and PARP. Tubulin was used as a cytoplasmic marker and PARP as a nuclear marker so as to ensure optimal fractionation. As can be seen, the p53 protein in the KB-OX cell lines is found primarily in the cytoplasm and does not translocate to the nucleus following DNA damage.

Because previous studies have suggested both NLSII and NLSIII have supportive roles in the nuclear transport of p53, we explored the sub-cellular distribution of p53 under unchallenged conditions and after the administration of the DNA-damaging agent, adriamycin. The quality of the separation was assessed using tubulin and PARP as markers of the cytoplasmic and nuclear fractions, respectively. As shown in figure 1B and expected for these mutant proteins, treatment with adriamycin had no effect on the levels of p53. Furthermore, treatment with adriamycin had no effect on the sub-cellular distribution of p53, with 60% of the p53 protein in KB-CP20 cells localized in the nucleus, compared with a predominant cytoplasmic distribution of p53 in the oxaliplatin selected cell lines. This finding was confirmed by immunohistochemistry (data not shown). Thus it appeared that the oxaliplatin-selected sublines harbored a p53 mutation that altered NLSIII and led to its sequestration in the cytoplasm.

p53⁴²⁰ is able to oligomerize and bind to dynein

Because we felt the published data has assigned at best an ancillary role to NLSII and NLSIII in promoting the nuclear localization of p53, we sought other explanations for the relative cytoplasmic sequestration seen in the oxaliplatin-selected cells. We began by examining the steps in the cytoplasm that lead to accumulation of p53 in the nucleus. Earlier studies from our laboratory have demonstrated that p53 traffics on microtubules to the peri-nuclear region and that this transport occurs on dynein (Giannakakou et al., 2000). Subsequent studies also demonstrated that oligomerization of p53 is necessary for dynein association and that this association is independent of microtubules (Trostel et al., 2006). To examine the ability of p53⁴²⁰ to oligomerize, we performed experiments with the cross-linking reagent, diamide. As positive controls, we used KB-CP20 and A549 lung carcinoma cells that harbor a wild type p53 and had been treated with adriamycin so as to raise the endogenous levels of p53. As shown in figure 2A, in the adriamycin treated A549 cells as well as the cisplatin- and all oxaliplatin-selected cell lines, cross-linked p53 oligomers were demonstrated as discrete bands on an immunoblot in the diamide treated samples – consistent with the existence of closely associated p53 molecules in all cell lines. Further evidence of the existence of such oligomers, a pre-requisite for dynein binding, was obtained by demonstrating the co-immunoprecipitation of dynein and p53 as shown in figure 2B. Both p53⁴²⁰, found in the oxaliplatin-selected cell lines, and p53^V172F, found in cisplatin-selected KB-CP20 cells, could be immunoprecipitated with dynein. This finding was further confirmed by transfecting PC3 prostate cancer cells that lack p53, with either wt or mutant p53 protein prior to performing the co-immunoprecipitation. The transfectant designated p53⁴²⁰ encodes a protein that is identical to that found in the KB-OX cell lines, while the p53^FS382S-393 transfectant possesses the same deletion at amino acid 382 and the resultant frame-shift found in p53⁴²⁰, but enocdes a protein that like wild type p53 is 393 amino acids long (see table 2). As shown in the lower panel of figure 2B in all cases the transfected p53 protein co-precipitated with dynein. Thus we concluded that the preferential cytoplasmic sequestration of p53⁴²⁰ is not a consequence of impaired dynein binding or impaired trafficking on microtubules.

Figure 2.

p53 oligomerization and dynein binding are intact in the KB-OX cell lines. (A) 50 μg lysates were incubated with (+) or without (−) 10 mM diamide, a crosslinker, for 10 minutes and then resolved on a gradient gel. The arrows show the monomeric and the oligomerized forms of p53. The existence of oligomers is consistent with close physical association of p53 monomers. (B) Lysates of KB-3-1, KB-CP20, and the three KB-OX cell lines as well as PC3 cells transfected with wild-type or mutant p53 for 14 hours were immunoprecipitated with anti-dynein intermediate chain and the immunoprecipitates were probed for p53 and dynein. The lysate only lane contains cell lysate immunoprecipitated only with Protein A antibody and the antibody only lane contains Protein A immunoprecipitated with anti-dynein without lysate. Western blot of 25 μg of cell lysate shows p53 and dynein expression again with p53 from the KB-OX cell lines migrating slower as indicated by the upper arrow. The bottom panel of 2B shows similar results obtained using protein harvested from PC3 cells (p53 null) transfected with either wt p53, p53^FS382–393, or the longer p53⁴²⁰ (p53^FS382–393 has the frame shift mutation, and the associated change in amino acids, but is truncated at the normal length of 393 amino acids).

Table 2.

Sequence of the nine p53 constructs used to explore the role of NLSII and III in the nuclear translocation of p53

	NLSII	…	NLSIII
Wild Type	K S K K G Q	…	H K K L M F K T E G P D S D -
KB-OX	K S K K G Q	…	H K N S C S R Q K G L T Q T D I L H F L F P T D S L P P P S L P P L P F W V L G L -
p53K381T	K S K K G Q	…	H T K L M F K T E G P D S D -
p53K382T	K S K K G Q	…	H K T L M F K T E G P D S D -
ΔNLSIII	K S K K G Q	…	H T T L M F K T E G P D S D -
p53FS382–393	K S K K G Q	…	H K N S C S R Q K G L T Q T -
p53420	K S K K G Q	…	H K N S C S R Q K G L T Q T D I L H F L F P T D S L P P P S L P P L P F W V L G L -
ΔNLSII	K S T T G Q	…	H K K L M F K T E G P D S D -
ΔNLSII/III	K S T T G Q	…	H T T L M F K T E G P D S D -
4G	Impaired tetramerization as a consequence of four glycine substitution: F338G, R342G, L344G, E346G

The nuclear export of p53⁴²⁰ is not augmented

Previous studies have reported that in some cancers, including colorectal and breast carcinomas, as well as in neuroblastoma, wt p53 is rendered non-functional because of its cytoplasmic sequestration (Bosari et al., 1995; Lilling et al., 2002; Moll et al., 1995). The nuclear export protein, exportin 1, is reported to be responsible for the transfer of p53 to the cytoplasm (Stommel et al., 1999). To exclude the possibility that the low nuclear levels of p53⁴²⁰ were secondary to increased nuclear export we sought to investigate the effect of leptomycin B, a drug that blocks the exportin 1 function, on the sub-cellular distribution of p53 in the KB-OX cell lines. As shown by immunofluorescence in figure 3, 10 ng/ml leptomycin B treatment increased the intensity of p53 nuclear staining in the KB-CP20 cell line (p < 0.001 comparing KB-CP20 ± Leptomycin B), but had no effect in the KB-OX60 cell line. These data indicate that the preferential cytoplasmic localization of p53⁴²⁰ is not a result of an augmented nucleo-cytoplasmic transport.

Figure 3.

Cytoplasmic sequestration of p53⁴²⁰ is not caused by augmented nuclear export. Upper panel: Confocal localization of p53 in the KB-CP20 and KB-OX60 cells was used to assess p53 nuclear export. Untreated KB-CP20 and KB-OX60 cells [(−) LMB] or cells treated with 10 ng/ml Leptomycin B [(+) LMB] for 4 hours to block the function of Exportin 1 were examined. In the KB-CP20 cells nuclear p53 levels increase; however, in KB-OX60 cells the p53 cellular distribution remains unchanged following Leptomycin B treatment. Lower Panel: For quantification, the nuclear intensity of 100 cells was measured by confocal microscopy and their averages were calculated. The nuclear p53 level of the KB-CP20 cell line was increased significantly (p < 0.001) by Leptomycin B. In contrast, the nuclear intensity of p53⁴²⁰ in the KB-OX60 cells did not change after blocking the nuclear export of p53 with Leptomycin B.

Neither NLSII nor NLSIII are required for p53 nuclear import

Given the above, we decided to better assess the role of NLSII and NLSIII in the accumulation of p53 so as to evaluate the possibility that the cytoplasmic sequestration occurred as a consequence of reduced nuclear import. To this end nine p53 constructs were generated with either one or both NLSII and NLSIII destroyed and of either normal length or 420 amino acids in length - the length of p53⁴²⁰. The nine constructs included: (1) p53^K381T, (2) p53^K382T, (3) ΔNLSIII (p53^K381T/K382T), (4) p53^FS382S-393 (FS = frame shift), (5) p53⁴²⁰, (6) ΔNLSII (p53^K372T/K373T), (7) ΔNLSII/III (p53^{K372T/K373T, K381T/K382T}), (8) p53 (4G), (9) wt p53, as well as a vector control (Table 2). The control constructs included wt p53 as well as p53 (4G), a construct in which amino acid residues F338G, R342G, L344G and E346G are replaced with glycines and in previous studies has been shown to be unable to oligomerize and in turn associate with dynein, and thus unable to translocate to the nucleus (Trostel et al., 2006). Following transfection into p53-null PC3 prostate cancer cells the distribution of p53 was determined by sub-cellular fractionation as shown in figure 4A. The quality of the separation was assessed using tubulin and lamin B as markers of the cytoplasmic and nuclear fractions, respectively. As quantitated in figure 4B, impaired nuclear accumulation was observed only with the p53 (4G) and the p53⁴²⁰ constructs indicating that neither NLSII nor NLSIII alone or in combination are essential for the nuclear translocation of p53. Thus the cytoplasmic sequestration of p53⁴²⁰ cannot be ascribed to destruction of NLSIII.

Figure 4.

Neither NLSII nor NLSIII are necessary for p53 nuclear translocation; however, p53⁴²⁰ shows decreased nuclear accumulation. (A) PC3 cells (p53 null) were transiently transfected with mutant p53 vectors, an empty vector or a wild-type p53 vector. Six hours after transfection, cellular extracts were fractionated into cytoplasmic (C) and nuclear (N) fractions and probed for p53, tubulin and lamin B. Tubulin was used as a cytoplasmic marker and lamin B as a nuclear marker so as to ensure the fractionation had been well done. The nuclear fraction of p53 in each sample (%N) is shown as percentage and calculated as %N = [N/(C+N)] × 100. The numbers represent the nuclear p53 amount of the individual experiment showed in the immunoblot. (B) Quantitations were performed using the results from three different transfections and 2 immunoblots from each. Six out of the eight mutants translocate into the nucleus efficiently. p53⁴²⁰ [p = 0.0037] and p53(4G) [p < 0.001] show significantly reduced nuclear accumulation when compared to wild-type p53.

Rapid degradation results in reduced stability of p53⁴²⁰

We had observed in the course of the transfection experiments that cells expressing p53⁴²⁰ constructs often had less protein expression on immunoblot analysis than cells transfected with either wt p53 or other mutant proteins. We thus considered the possibility that differential stability of p53⁴²⁰ might account for more rapid degradation and the apparently lower protein levels. To examine the stability of the p53⁴²⁰, PC3 cells transiently transfected with either p53⁴²⁰ or wt p53 vectors were treated with the protein synthesis inhibitor, cycloheximide for variable times prior to harvesting protein. As shown in figure 5A, the levels of p53⁴²⁰ declined after cycloheximide administration but remained stable if cycloheximide was not administered. By comparison, the levels of wild type p53 appeared more stable in this system, remaining stable in this short incubation even though cycloheximide was added and increasing if cycloheximide was not administered (the increase is not unexpected following a transfection since levels usually increase over time). To confirm this finding, we examined the stability of p53 in KB-OX60 and KB-CP20 cells. Cell lysates were extracted followed by variable times of cycloheximide treatment up to 5 hours. In the presence of cycloheximide the level of p53⁴²⁰ in the KB-OX60 cells decreased more rapidly than that of p53^V172F in KB-CP20 cells (80 versus 50 per cent decrease over 5 hours) as shown in figure 5B. One explanation for these results is the possibility that the additional 27 amino acids at the C-terminus of p53⁴²⁰ destabilize the protein to some extent; however, destabilization is not to an extent sufficient to preclude accumulation of the mutant protein in these cells to levels much higher than those found in parental KB-3-1 cells that harbor a wt p53. Alternatively, it is possible that simply because degradation of p53 occurs primarily in the cytoplasm (Freedman and Levine, 1998; Roth et al., 1998) and p53⁴²⁰ is primarily confined to the cytoplasm, the total levels fall more rapidly.

Figure 5.

The degradation of p53⁴²⁰ is more rapid than that of wild type p53. (A) PC3 cells (p53 null) were transiently transfected with wild-type p53 or the p53⁴²⁰ mutant vector for 6 hours. After the transfection, cells were treated (+) or not treated (−) with 10 μg/ml cycloheximide, a protein biosynthesis inhibitor, in a time course experiment. The degradation of p53⁴²⁰ occurred more rapidly than wild-type p53. The table shows the degradation of the p53⁴²⁰ and wild type p53 with or without cycloheximide treatment as a percentage of the 0 minute time point. (B) KB-CP20 and KB-OX60 cell lines were treated with 10 μg/ml cycloheximide in a time course experiment. As the immunoblot and the graph show the degradation of p53⁴²⁰ in the KB-OX60 cells occurred more rapidly than p53^V172F in the KB-CP20 cell lines.

The association of p53⁴²⁰ with importin-α/β is impaired

Having excluded oligomerization, dynein binding and trafficking on MTs, exportin-mediated transport, protein stability and any role for NLSII or NLSIII in the cytoplasmic sequestration of p53⁴²⁰ we next assessed whether impaired binding to the importin-α/β complex with reduced nuclear import might explain the observations. A search of the literature found only indirect evidence for the role of importin in the transport of most proteins and no direct evidence for a role in the transport of p53 (Kim et al., 2000; Li et al., 2007; Liang and Clarke, 1999). We conducted immunoprecipitation experiments to assess the extent of importin association with p53 and its putative role in the nuclear transport of p53. As shown in the upper panel of figure 6, the binding of p53⁴²⁰ to importin was impaired in the KB-OX60 cell line compared to control cells, KB-CP20 and A549 treated with adriamycin. Higher levels of importin-β in the KB-OX60 pulled down less p53⁴²⁰ from lysates containing p53⁴²⁰ than from the KB-CP20 or the A549 adriamycin treated cell lines (Tabulation of the results of relative and absolute interactions is shown in the lower panel). We thus conclude that p53⁴²⁰ is sequestered in the cytoplasm because it is excluded from entering the nucleus as a consequence of poor importin binding.

Figure 6.

p53⁴²⁰ binding to importin is impaired. Upper Panel: Immunoblots on the top of 50 μg of cell lysate show p53 and importin β expression. Immunoblots on the bottom show the results when lysates were immunoprecipitated with anti-beta importin intermediate chain and probed for p53. Lanes include A549 (treated with 400 ng/ml adriamycin for four hours to enrich for p53), KB-CP20 and KB-OX60. Controls include the lysate only lane in which cell lysate was immunoprecipitated only with Protein A sepharose and the antibody only lane in which Protein A was immunoprecipitated with anti-importin without lysate. Lower Panel: The graph shows the relative amount of p53 in the KB-OX60 cell line normalized to KB-CP20 in the lysate (left bar) and in the immunoprecipitate (right bar). Significantly less p53⁴²⁰ binds to importin than expected. Whereas both KB-CP20 and KB-OX60 cells have much higher levels of p53 than parental KB-3-1 cells, the levels in KB-OX60 cells are lower than in KB-CP20 (16.6%). However, despite comparable importin beta immunoprecipitations, the amount of p53 recovered from the KB-OX60 cells is much less (5.0%) than that recovered from KB-CP20 cells, consistent with a poor association of p53⁴²⁰, found in KB-OX60 cells, to importin beta.

p53⁴²⁰ retains the ability to trans-activate a p53 target gene

Although the p53⁴²⁰ protein is somewhat less stable than the wt p53 protein, as noted above, and as shown in the immunoblots in figures 1 and 2, p53⁴²⁰ accumulates to high levels in the resistant cells. Mutant proteins such as p53^V172F found in cisplatin selected KB cells harbor a mutation in the DNA binding domain that impairs their ability to trans-activate target genes, and hence can safely cross the nuclear membrane since they cannot effectively trans-activate target genes. However, p53⁴²⁰ retains the ability to trans-activate target genes, in this case p21, and hence its sequestration in the cytoplasm is essential to inhibit its activity (supplementary data). Thus in the case of p53⁴²⁰ impaired nuclear import precludes a protein that can trans-activate genes from reaching its targets achieving a similar outcome to that most commonly occurring in cells harboring mutant p53 proteins – impaired ability to trans-activate target genes.

DISCUSSION

It is generally agreed that p53 functions primarily in the nucleus (Ryan et al., 2001; Shaulsky et al., 1991). Active transport of p53 to the nucleus is known to occur following DNA damage and is followed by sequence-specific DNA binding (el-Deiry, 1998). As a transcription factor, p53 is able to initiate cell cycle arrest to allow for DNA repair and when repair fails can assist in eliminating damaged cells through apoptosis (Lane, 1992). Because of its pivotal role protecting the genome, p53 nuclear transport is tightly regulated. Previous studies have reported that p53 has three putative nuclear localization signals (NLSs), and it has been suggested that nuclear translocation of p53 is mediated principally by NLSI with ancillary roles served by NLSII and NLSIII (Dang and Lee, 1989; Shaulsky et al., 1990). For nuclear translocation, the nuclear localization signal(s) of p53 interact with a carrier heterocomplex, importin-α/β. Importin-β establishes the connection between the p53-bound heterocomplex and the nuclear pore complex, while importin-α facilitates the translocation. Following nuclear translocation, p53 binds DNA at p53 responsive elements and modulates gene transcription. In the present study we have identified impaired association with importin as a novel mechanism abrogating p53 function. The resultant sequestration of p53 in the cytoplasm renders nonfunctional a p53 protein capable of trans-activating target genes. Furthermore, we provide evidence that despite their designation as nuclear localization signals, neither NLSII nor NLSIII has a role in nuclear translocation of p53.

The present study began with the characterization of several cell lines developed as models of platinum resistance. Previous studies have described acquired p53 mutations in cisplatin resistant cell lines, and have suggested this as either the sole or a contributing mechanism of tolerance. To assess p53 we screened our isolates by immunoblotting and found increased levels of p53 in all isolates, consistent with a mutant protein. Loss of p53 function is commonly found in cancer cells most frequently as a consequence of mutations in the DNA binding domain (DBD). Although these p53 proteins can translocate to the nucleus, they cannot bind p53 responsive elements and cannot initiate transcription. As a result of this the level of Hdm2, the principal negative regulator of p53, is low or absent and p53 proteins with mutations in the DBD accumulate both in the nucleus and the cytoplasm. We identified such a mutation in our cisplatin-selected subline, and as expected this led to higher nuclear and cytoplasmic p53 levels. However, in the oxaliplatin-selected cells the mutation was in the C-terminus, not in the DNA binding domain, and generated a mutant protein that was confined primarily to the cytoplasm. Sequence analysis identified a deletion in the sequence encoding lysine 382 that both disrupted NLSIII and shifted the reading frame, such that a 420 amino acid protein containing an additional 27 amino acids at its C-terminus was made. Given the disruption of NLSIII we examined the nuclear/cytoplasmic distribution of the mutant protein and were surprised to find that in three oxaliplatin resistant cell lines derived from the same lineage the protein was largely confined to the cytoplasm. The latter was reminiscent of observations in some inflammatory breast cancers, melanoma malignum, colon cancers as well as undifferentiated neuroblastoma that have reported preferential localization of a wild-type p53 protein in the cytoplasm (Bosari et al., 1995; Lilling et al., 2002; Moll et al., 1995; Weiss et al., 1995). In these cases, the potentially functional wild-type p53 is rendered inactive by confinement to the cytoplasm providing advantages for the tumor - in colon cancer for example this is an indicator of poor survival (Bosari et al., 1994; Sun et al., 1992). While most of these studies are descriptive and do not propose explanations for the cytoplasmic sequestration of p53 (Bosari et al., 1995; Imamura et al., 1993; Weiss et al., 1995), insufficient nuclear translocation or augmented nuclear export have been proposed as reasonable explanations (Kim et al., 2000; Stommel et al., 1999). The list of putative contributing factors is long and includes a truncated form of importin alpha (Kim et al., 2000), increased amount of intermediate filaments (Sembritzki et al., 2002), cytomegalovirus infection (Utama et al., 2006), endoplasmic reticulum stress (Qu et al., 2004), amplification of unknown genes inducing protein complex formation with p53 (Ottaggio et al., 2000) and aberrant hyperubiquitination (Becker et al., 2007), among others. Although we were unable to find a study reporting cytoplasmic sequestration of a mutant p53, our data clearly indicated that in our oxaliplatin selected cell lines the p53 that was found primarily in the cytoplasm was mutant and we inferred that the mutation was causative in the aberrant localization.

Aware of previous reports assigning a principal role in the nuclear translocation of p53 to NLSI and only ancillary functions to both NLSII and NLSIII we wondered whether NLSII and NLSIII alone or in combination might have a more crucial role in the translocation of p53 to the nucleus – allowing us to explain the cytoplasmic preference of the mtp53 in the oxaliplatin resistant cells. However, the experiments described in figure 4 convincingly demonstrated that contrary to prior speculation and despite some sequence homology neither NLSII nor NLSIII has any role alone or together in p53 nuclear translocation. Two classes of NLSs are distinguished: bipartite and monopartite. The bipartite type of nuclear localization signal, first identified in Xenopus nucleoplasmin, has two basic residues followed by ten spacer residues and a second region with at least three basic residues (KRPAATKKAGQAKKK) (Dingwall et al., 1987). NLSI of p53 (³⁰⁵KRALPNNTSSSPQPKKK³²¹) has a sequence similar to the bipartite consensus and as our data indicates is the only NLS involved in nuclear translocation. In contrast, the monopartite NLS, represented by the SV40 large T-antigen NLS, lacks a spacer (K-K/R-X-K/R) (Smith et al., 1985). Because neither NLSII (³⁶⁹LKSKKGQ³⁷⁵) nor NLS III (³⁷⁹RHKKLM³⁸⁴) conforms well to the consensus sequences it is not surprising that our results found no role in nuclear translocation for these regions – indicating they should thus not be designated as NLSs.

As we sought other explanations, we showed that the mutant p53⁴²⁰ protein associates with dynein, allowing it to traffic on microtubules, and to be unloaded in the perinuclear region. Furthermore, increased export as the explanation for the low nuclear levels of p53⁴²⁰ was excluded by the lack of an effect of leptomycin B on nuclear levels. However, the results in figure 6 convincingly demonstrated in the control cells that indeed nuclear import of p53 occurs on the importin complex, and that the association of p53⁴²⁰ with importin-β was impaired. The latter is important because it is believed the rate of formation of the complex strongly determines the import rate (Riddick and Macara, 2005). As noted in Results, the acquired mutation, while reducing protein stability to a small extent, did not inhibit the protein's trans-activating activity, a quandary solved by inhibiting nuclear import. Thus, cytoplasmic sequestration due to impaired association with importin was identified a novel mechanism of inactivating p53. We must conclude that although the stability of p53⁴²⁰ was not markedly affected it was sufficient to alter affinity for importin. The latter was likely caused by the additional 27 amino acids at the C-terminus and not the frame shift since a protein with residues similar to the “shifted” 382 – 393 present in p53⁴²⁰ but 393 amino acids in length (p53^FS382–393) could associate with dynein and was found in the nucleus to the same extent as wt p53.

In summary, we report a novel mutation in oxaliplatin-selected cells that interferes with p53 trans-activation activity by impairing the association of the mutant p53 with importin, sequestering p53 in the cytoplasm. We further demonstrate that NLSII and III have no role in nuclear translocation and suggest this nomenclature be dropped. A mutation that interferes with importin-binding as a mechanism of cytoplasmic sequestration of p53 is a novel mechanism for attenuating p53 function. While this mechanism may be confined to this model or only a few others it seems prudent to investigate its occurrence further. It could also be considered as a mechanism to inactivate other tumor suppressors whose nuclear localization is essential for function and whose confinement to the cytoplasm could be important in cancer etiology or progression.

MATERIALS AND METHODS

Cytotoxicity Assay

Four-day cytotoxicity assays were performed based on the method of Skehan et al. Briefly, cells plated (1,500 – 2,000 cells/well) in 96-well, flat-bottom plates attached overnight at 37 °C in 5% CO₂. Chemotherapeutic agents in triplicate were subsequently added at various concentrations for 96 h. Cells were fixed with 40% TCA and stained with 0.4% sulforhodamine B in 1% acetic acid. After washing plates in 1% acetic acid and drying, the dye was solubilized in 50% Trizma base and read at 540 nm.

Western Blot and Antibodies

Cell extracts were prepared by cell lysis in buffer containing 50 mM Tris (pH 7.5), NP-40 (1% V/V), EDTA 2 mM, 5M NaCl and Complete Mini, EDTA-free protease inhibitor cocktail tablets (Roche). Equal amount of protein pellets were re-suspended in 1× sample loading buffer and heated 99°C for 5 minutes before being resolved by 10% SDS-polyacrylamide gels and immunoblotted. The following antibodies were used: anti-p53 (Ab6, monoclonal Do1, Oncogene Science), tubulin (polyconal rabbit, Abcam), lamin B (polyclonal rabbit, Calbiochem) anti-dynein intermediate chain (IC74 monoclonal, Covance), anti-importin ß (31H4 monoclonal, Sigma), and anti-importin ß (ab2811, 3E9 monoclonal, Abcam), anti-PARP (polyclonal rabbit, Upstate, 06–557), HRP conjugated anti-mouse (NA931, Amersham), HRP conjugated anti-mouse kappa (1050–05, Southern Biotech), anti-mouse kappa (1050–01, Southern Biotech) labeled with SAIVI Alexa Fluor 680 with 0.1 mg labeling kit (Molecular Probes), anti-mouse IRDye 680 (926–32220, Li-cor), anti-rabbit IRDye 800CW (926–32211, Li-cor).

Subcellular Fractionation

Cells incubated overnight in 6-well plates were harvested and fractionated using NE-PER nuclear and cytoplasmic extraction reagent (Pierce) according to the manufacturer's instruction. Samples were loaded in equal proportion on 10% SDS-PAGE and Western Blots were performed for p53 (monoclonal mouse anti-p53, 1:1000, Calbiochem), tubulin (polyconal rabbit anti-tubulin, 1:1000, Abcam), lamin B (polyclonal rabbit anti-lamin B, 1:500, Calbiochem), PARP (polyclonal rabbit anti-PARP,1:500, Upstate, 06–557). Densitomery values were obtained using the Odyssey Li-Core program.

Immunohistochemistry

Exponentially growing cells were plated on 12-mm glass coverslips and incubated overnight. The following day, cells were rinsed in PBS and fixed with 3.7% formaldehyde/PBS for 10 min at room temperature. Cells were then fixed in 90% methanol for 5 min and coverslips were rinsed 2 times with PBS and incubated in blocking solution (5% donkey serum/PBS) for 30 minutes at room temperature. Coverslips were rinsed 3 times in PBS between each of the steps. For double-labeling experiments, primary and secondary antibodies were added sequentially for 1 h each at room temperature. The antibodies used were mouse monoclonal anti-tubulin (DM1A) antibody (Sigma) for tubulin/microtubule staining, and sheep polyclonal (Ab7) antibody for p53 (Calbiochem) staining. A FITC-conjugated anti-mouse antibody, and a Texas RedX anti-sheep antibody were used as secondary antibodies. DNA was counterstained with 1 g ml⁻¹ 4,6-diamidino-2-phenylindole (DAPI, Sigma) in PBS.

Cross-linking Assay

For cross-linking experiments, 50–200 ug lysates were incubated with 10 mM diamide for 10 minutes at RT. Samples were heated at 99° for 5 minutes without a reducing agent and then resolved on a precast 4–15% gradient gel (Biorad).

Immunoprecipitation

Cell lines or transiently transfected cells were lysed in IP buffer (50 mM Tris-HCl pH 8.0, 150 mM NaCl, 1% Triton X-100, and Complete Mini, EDTA-free protease inhibitor cocktail tablets (Roche)), homogenized with a 25-gauge needle, and centrifuged at 14,000 rpm for 2 minutes at 4°. Lysates (1 mg) were pre-cleared for 1 hr at 4° then incubated with IC74 anti-Dynein or anti-importin ß (Abcam) overnight at 4°. Protein and antibody were then incubated with ImmunoPure Plus Immobilized Protein A (Pierce) for 2 hours at 4°. Samples were washed 5× with 500 ul of IP buffer. The Protein A pellet was re-suspended in 35 μl of 1× sample loading buffer and heated 99° for 5 minutes before being resolved by 10% SDS-polyacrylamide gels and immunoblotted with antibodies for dynein, p53, or importin ß (Sigma). Results were determined by chemiluminescent detection (SuperSignal West Pico Chemiluminescent Substrate, Pierce) or by detection of IRDye infrared conjugated secondary antibodies using the LICOR Odyssey.

Mutagenesis

Wt and mutant p53 were amplified by polymerase chain reaction using a pcDNA3.1+ vector (Invitrogen) containing wt p53 as a template. The methodology utilized is described in detail in Supplementary materials.

Transfection

2 × 10⁵ PC3 cells/well in RPMI were transiently transfected with p53 constructs using Lipofectamine 2000 (Invitrogen) according to manufacturer's instruction. Following the 6h transfection period sub-cellular fractionation was performed immediately or following drug treatment.

Luciferase reporter assays

70,000 PC3 cells/well in 12-well plates were transiently transfected using TransFast Transfection Reagent (Promega) for 48h. The transfection media (OPTI-MED, Gibco) was replaced with RPMI 5h following transfection. Vectors used included the backbone or the p21-Luc vector (kindly provided by Dr. Bert Vogelstein, Johns Hopkins University), the pRL-TK internal control vector, and either the wt p53 vector or the p53⁴²⁰ vector. Luciferase assays were performed using the Dual-Luciferase Reporter Assay System (Promega).