Identification of a Novel Cytoplasmic Protein That Specifically Binds to Nuclear Localization Signal Motifs*

Active transport of proteins into the nucleus is mediated by interaction between the classical nuclear localization signals (NLSs) of the targeted proteins and the NLS receptor (importin) complex. This nuclear transport system is highly regulated and conserved in eukaryotes and is essential for cell survival. Using a fragment of BRCA1 containing the two NLS motifs as a bait for yeast two-hybrid screening, we have isolated four clones, one of which is importin α. Here we characterize one of the other clones identified, BRAP2, which is a novel gene and expressed as a 2-kilobase mRNA in human mammary epithelial cells and some but not all tissues of mice. The isolated full-length cDNA encodes a novel protein containing 600 amino acid residues with pI 6.04. Characteristic motifs of C2H2 zinc fingers and leucine heptad repeats are present in the middle and C-terminal regions of the protein, respectively. BRAP2 also shares significant homology with a hypothetical protein from yeastSaccharomyces cerevisiae, especially in the zinc finger region. Antibodies prepared against the C-terminal region of BRAP2 fused to glutathione S-transferase specifically recognize a cellular protein with a molecular size of 68 kDa, consistent with the size of the in vitro translated protein. Cellular BRAP2 is mainly cytoplasmic and binds to the NLS motifs of BRCA1 with similar specificity to that of importin α in both two-hybrid assays in yeast and glutathione S-transferase pull-down assays in vitro. Other motifs such as the SV40 large T antigen NLS motif and the bipartite NLS motif found in mitosin are also recognized by BRAP2. Similarly, the yeast homolog of BRAP2 also binds to these NLS motifs in vitro. These results imply that BRAP2 may function as a cytoplasmic retention protein and play a role in regulating transport of nuclear proteins.

The passage of macromolecules between the nucleus and the cytoplasm occurs through nuclear pores. Small macromolecules can diffuse through the nuclear pores at a rate inversely proportional to their mass. Proteins with molecular masses greater than 40 -60 kDa are actively transported through the nuclear pores. To be transported into the nucleus, the protein must either contain a nuclear localization signal or, if not, be bound to another protein that does (1,2). This process requires at least four different factors acting in two distinct steps. The first step is mediated by importin ␣ (also termed karyopherin ␣) and importin ␤ (also termed karyopherin ␤). The ␣ subunit is primarily responsible for NLS 1 recognition, whereas the ␤ subunit appears to mediate docking to the nuclear pore complex. The second translocation step requires the small G protein Ran/TC4 and an interacting partner, p15 (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13).
The presence of a nuclear localization signal may not be sufficient to direct nuclear import. The target efficiency of NLS motifs can be modified by the presence of multiple NLS motifs within a protein, by modifications of the flanking sequences, and by the accessibility of the NLSs to the import machinery (14 -17). These regulatory mechanisms provide for delicate control over the transport of many nuclear proteins. Failure of such control can have profound effects on the cells (18,19).
Several tumor suppressor genes encode nuclear proteins, the correct nuclear localization of which is critical to their function. For example, the tumor suppressor p53 is ordinarily a nuclear protein. However, wild-type p53 has been shown to be mislocated in the cytoplasm of several different cancer cells, whereas mutant p53 remains in the nucleus (20). Such aberrant localization of wild-type p53 implicates its functional inactivation and reflects its importance in carcinogenesis. Similarly, BRCA1 is also a nuclear protein in normal breast epithelial cells and is mislocated to the cytoplasmic compartment of many advanced breast cancer cells, such as established cell lines, cancer cells from malignant pleural effusions, and some primary breast cancer specimens (21,22). This suggests that the aberrant subcellular localization of BRCA1 may be responsible for its inactivation in some sporadic breast cancers. Recently, the WT1 tumor suppressor has also been shown to be mislocated in the cytoplasm of breast cancer cells (23).
Previously, we have shown that BRCA1 contains two functional NLS motifs that direct its transport into the nucleus (24). In an attempt to elucidate mechanisms that might regulate this process, we used a fragment of BRCA1 containing both nuclear localization signals as a bait for a yeast two-hybrid screen. Several interacting proteins were identified. One of these is importin ␣, which interacts specifically with the NLS motifs of BRCA1 (24). In this communication, we characterize BRAP2, which is a cytoplasmic protein that binds to the two functional NLS motifs of BRCA1. These results suggest that BRAP2 may serve as a cytoplasmic retention protein for regulating nuclear targeting.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Screen and Isolation of Full-length BRAP2 cDNA-The plasmid pAS-BRCA3.5 (24), which contains the GAL4 * This work was supported by Grants from the National Institutes of Health (P50-CA58183 and P01-CA30195). 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF035620.
RNA Blotting Analysis-Total RNA was extracted by the guanidine isothiocyanate-CsCl method (30). Total RNAs from T47D and HBL100 cells were then used to prepare poly(A) ϩ RNA using the poly(A) Tract mRNA isolation system (Promega, Madison, WI). About 10 g of total RNA or 2 g of poly(A) ϩ RNA was denatured in 50% formamide, 2.2 M formaldehyde, 1 ϫ MOPS (0.2 M MOPS (pH 7.0), 0.5 M sodium acetate, 0.01 M EDTA), and analyzed by 1.2% agarose gel electrophoresis (30). The RNA was then transferred to Hybond paper (Amersham, Buckinghamshire, United Kingdom), and immobilized by UV cross-linking. Prehybridization and hybridization were carried out in 40% formamide, 10% polyethylene glycol 8000, 0.25 M Na 2 HPO 4 /NaH 2 PO 4 (pH 7.2), 0.25 M NaCl, 1 mM EDTA, 7% SDS, 100 g of salmon sperm DNA per ml. The 32 P-labeled 0.8-kb BRAP2 cDNA was used to probe the blot at 42°C for 18 h. The 1.0-kb G␤-like cDNA was also used as a probe to serve as a control for RNA quality. This gene is ubiquitously expressed at a relative high level in most mouse tissues (31). 2 The initial washing was done with 2ϫ SSC (1ϫ SSC is 0.15 M NaCl plus 0.015 M sodium citrate), 0.1% SDS at room temperature, and the final washing was done with 0.1ϫ SSC, 0.1% SDS at 65°C for 30 min.
Immunoprecipitation-To identify endogenous protein synthesized in vivo, about 5 ϫ 10 6 HBL100 cells were metabolically labeled with [ 35 S]methionine for 2 h and subsequently lysed in ice-cold lysis 250 buffer. The clarified lysate was incubated with anti-BRAP2C antibody at 4°C for 1 h, then protein A-Sepharose beads were added, and the mixture was incubated for 1 h with constant rotation. After washing extensively with lysis 250 buffer, the beads were boiled in SDS sample buffer, and the immunoprecipitates were separated by 7.5% SDS-polyacrylamide gel electrophoresis. For double immunoprecipitation, the immunoprecipitates were boiled in 200 l of dissociation buffer (20 mM Tris-Cl (pH 7.4), 50 mM NaCl, 1% SDS, and 5 mM dithiothreitol), and the denatured proteins were diluted with 1 ml of lysis 250 buffer and re-immunoprecipitated with other antibodies.
Cell Fractionation-The protocols to separate membrane, nuclear, and cytoplasmic fractions were adapted from those previously published (22,33). All three fractions and total cell lysate were then assayed for p84 and BRAP2 by immunoprecipitation subsequently as described above. The same fractions were also incubated with glutathione-agarose beads, then separated by SDS-polyacrylamide gel electrophoresis, and stained with Coomassie Blue to detect endogenous GST.
In Vitro Binding Assay-For in vitro transcription and translation of BRCA1, either wild-type BRCA1 containing amino acids 303-701 or the same fragment containing the KLP, KLS, KLN, or KLP ϩ KLS mutations (24) was cloned into the pBSKF vector translationally in frame with the flag epitope. The flag epitope in the vector additionally provides the first methionine for the proteins. For in vitro transcription and translation of importin, the PstI fragment of the importin cDNA (encoding amino acids 64 -529) was cloned into the PstI site of pBSKF. For in vitro transcription and translation of the yeast BRAP2 homolog, the full-length yeast BRAP2 cDNA was cloned from yeast genomic DNA by polymerase chain reaction into pBSK under the control of the T3 promoter. Glutathione-Sepharose beads containing about 20 g of GST or GST fusion proteins were preincubated with Tris-buffered salinebovine serum albumin buffer (25 mM Tris-HCl (pH 8.0), 120 mM NaCl, 10% bovine serum albumin, 1 g/ml of protease inhibitors including leupeptin, antipain, aprotinin, and pepstatin) for 30 min at room temperature with rotation. The beads were then incubated with an equal amount of in vitro translated products in standard lysis buffer (100 mM NaCl, 50 mM Tris (pH 7.4), 5 mM EDTA, 0.5% Nonidet P-40, and protease inhibitors) for 1 h at room temperature with rotation. Complexes were washed extensively with the standard lysis buffer, boiled in protein loading buffer, separated by SDS-polyacrylamide gel electrophoresis, and detected by fluorography. Quantitation of binding efficiency was done using the personal densitometer SI (Molecular Dynamics, Sunnyvale, CA).
Transfection and Immunostaining-Cells were transfected using the standard calcium phosphate precipitation protocol with expression vectors for either green fluorescent protein (GFP) alone, GFP fused to full-length BRAP2, or flag-tagged full-length BRAP2. After transfection, cells were trypsinized and replated on coverslips in a tissue culture dish. 24 h after transfection, the cells were either observed directly for fluoresence under a fluoresence microscope or prepared for immunostaining as follows. The cells were washed in phosphate-buffered saline (PBS) and fixed for 30 min in 4% paraformaldehyde in PBS with 0.5% Triton X-100. After treating with 0.05% saponin in water for 30 min and washing extensively with PBS, cells were blocked in PBS containing 10% normal goat serum. The cells were then incubated with M2 antiflag monoclonal antibody (Eastman Kodak Co.) for 1 h, followed by three washes with PBS. They were then incubated with Texas redconjugated secondary anti-mouse antibody (Amersham) for 1 h. After washing extensively in PBS with 0.5% Nonidet P-40, cells were further stained with the DNA specific dye 4, 6-diamidino-2 phenylindole and mounted in Permafluor (Lipshaw-Immunonon, Inc., Pittsburgh, PA).

Expression of BRAP2 mRNA in Breast Epithelial Cells and
Adult Mouse Tissue-As described previously, BRAP2 is the strongest binding clone isolated by a yeast two-hybrid screen for BRCA1-associated proteins (24). To determine the size of the full-length mRNA for BRAP2 and its expression profile in breast epithelial cells and mouse tissues, the 0.8-kb cDNA insert of the original clone was prepared as a probe for RNA hybridization. Initially, poly(A) selected RNA from the breast epithelial cell line, HBL100, and breast cancer cell line, T47D, were used for RNA blotting, and a single 2.0-kb mRNA was detected (Fig. 1A). This suggests that the isolated cDNA contains only a partial sequence of BRAP2 and that it is expressed in breast epithelial cells. The expression pattern of BRAP2 in adult mouse tissues was also examined by the same method. As shown in Fig. 1B, BRAP2 is abundantly expressed in testis and detectable in other tissues such as kidney, lung, liver, and brain. This transcript is expressed at very low levels in spleen, thymus, and small intestine. As with the cell lines analyzed, only a single 2.0-kb mRNA was detected in mouse tissues. The biological significance of the varied expression level in different tissues has yet to be explored.
Isolation and Sequence Analysis of the Full-length BRAP2 cDNA-To obtain the full-length cDNA, the 0.8-kb cDNA was used as a probe to screen a human fibroblast cDNA library. Two overlapping clones were isolated, which together defined a cDNA of about 2.0-kb, consistent with the mRNA size. The full-length cDNA was sequenced completely, and the longest open reading frame was found to encode a protein of 600 amino acid residues with pI 6.04 (Fig. 2). A termination codon was found prior to the first initiation codon, indicating that the 5Ј-coding sequence is within the cloned cDNA. A computerassisted homology search of GenBank TM at the National Center for Biotechnology Information found the predicted protein to be novel. Characteristic motifs of C2H2 zinc fingers and leucine heptad repeats were identified in the middle and Cterminal regions of the protein, respectively. Interestingly, this protein shares significant homology (23.7% overall similarity) with a hypothetical 585 amino acid protein from Saccharomyces cerevisiae (Fig. 3, GenBank TM accession number P38748). In particular, the zinc finger region (amino acids 317-389) shares 62% homology with the yeast protein (amino acids 301-378). These results suggest that BRAP2 encodes a novel protein that is conserved in yeast.
Identification of Cellular BRAP2 Protein-To identify the cellular protein encoded by BRAP2, we prepared two specific antibodies that recognize either the N-terminal region (amino acids 12-291) or the C-terminal region (amino acids 455-570) of BRAP2. Purified GST fusion proteins encoding these regions served as antigens for immunizing mice. The antibodies generated were used to immunoprecipitate the in vitro translated, [ 35 S]methionine-labeled BRAP2 protein. As shown in Fig. 4, the anti-BRAP2C antibody, but not pre-immune serum, immunoprecipitates the in vitro translated BRAP2 protein (Fig. 4,  lanes 2 and 3). When metabolically [ 35 S]methionine-labeled HBL100 cells were used for immunoprecipitation, anti-BRAP2C antibody recognized a 68-kDa protein that was absent in the immunoprecipitates of the same cell lysates brought down by pre-immune serum (Fig. 4, compare lanes 4 and 5). fusion protein abolished the specific immunoprecipitation, whereas incubation with GST alone had no effect (Fig. 4, lanes  6 and 7). Specificity of the antibody for the 68-kDa protein was further confirmed by reimmunoprecipitation of proteins obtained from the first immunoprecipitates following their recovery by denaturation (Fig. 4, lane 8). The cellular BRAP2 detected by anti-BRAP2C antibody migrates as two major bands, the higher form of BRAP2 may represent a posttranslational modification and is marked by an asterisk (Fig. 4, lane 8). Similar results were also obtained with the other polyclonal antibody, anti-BRAP2N (data not shown). The size of the protein translated in vitro from the full-length cDNA is identical to that of the cellular protein immunoprecipitated by the anti-BRAP2C antibody (Fig. 4, compare lanes 1 and 5), suggesting that the 68-kDa protein is the authentic gene product of BRAP2.
BRAP2 Is a Cytoplasmic Protein-To explore its potential function, the subcellular localization of BRAP2 was examined by both biochemical fractionation and epitope-tagged expression. HBL100 cells were biochemically fractionated into nuclei, cytoplasm, and membrane components as described previously (33), and these fractions were immunoblotted with anti-BRAP2N antibody to detect BRAP2. It was found that BRAP2 distributed mainly in the cytoplasm, including membrane and cytosolic fractions, but was not detected in nuclei (Fig. 5A). p84, a nuclear matrix protein (34) and glutathione S-transferase, a cytoplasmic protein (35) served as controls for the fractionation procedure (Fig. 5A). Since the currently available anti-BRAP2 antibodies are not suitable for immunostaining, an alternative method to confirm the biochemical fractionation results was needed. For this purpose, an expression plasmid containing the full-length BRAP2 cDNA fused in-frame with the flag epitope was constructed under the regulation of the CMV immediate early gene promoter (CEPF-BRAP2), permitting detection of exogenous protein using the anti-flag monoclonal antibody M2 (Kodak). By this method, flag-tagged BRAP2 was mainly localized in the cytoplasm when the plasmid was transfected into Saos2 cells (Fig. 5B). Cells transfected with the control plasmid containing the flag-epitope without BRAP2 did not show any staining (data not shown).
To further confirm BRAP2 localization, the BRAP2 cDNA was fused to a GFP reporter construct. This expression vector also contains a myc-epitope at the 5Ј end of GFP to permit specific detection of the exogenous protein. HBL100 cells transfected with the GFP construct (CHPL-GFP) alone or CHPL-GFP-BRAP2 were observed directly under a fluorescence microscope. Expression of GFP was observed both in nuclei and cytoplasm (Fig. 5C, panel b), whereas expression of GFP-BRAP2 was seen only in the cytoplasm (Fig. 5C, panel d). Similar results were also observed when other cell lines such as T47D and Saos2 were used for the transfection (data not shown). When the transfected cells were lysed and immunoprecipitated with anti-myc antibody, and then immunoblotted with anti-GFP antibody, bands of the expected molecular weights were detected, confirming the authenticity of the expressed proteins (Fig. 5D). Together, these results indicate that BRAP2 is a cytoplasmic protein.
BRAP2 Binds to NLS Motifs of BRCA1-The idea that BRAP2 may bind to the NLS motifs of BRCA1 was based on two considerations. First, the BRCA1 bait used for screening contains two functional NLS motifs (a third potential NLS motif was found to be nonfunctional). Second, one of the other interacting clones is importin ␣ (24). To test this hypothesis, either the wild-type BRCA1 cDNA fragment encoding amino acids 1-1142 (BRCA3.5) or the same region containing one of the three mutated NLS sequences (KLP, KLS, and KLN) as described previously (24), was used in a yeast two-hybrid assay with BRAP2 (Fig. 6A). Importin ␣ (amino acids 220 -529) and one other BRCA1 associated protein, BRAP12, isolated from our original two-hybrid screen (24) were used as positive and negative controls, respectively. Both BRAP2 and importin ␣ bind to BRCA3.5 or the KLN mutant, but not to the KLP mutant (Fig. 6B). Previously, we have shown that the NLS deleted in the KLP mutant is required for nuclear transport of BRCA1. The interaction between importin-␣ and the KLS mutant was decreased about 6-fold compared with the wild-type, whereas BRAP2 interacts well with the KLS mutant (Fig. 6B). In contrast, BRAP12 binds equally well to the wild-type and all three mutants.
To further confirm this interaction, an additional in vitro GST pull-down assay was also performed. The wild-type BRCA1 cDNA encoding amino acids 303-701 or the same region containing the mutated NLS sequences (KLP, KLS, KLN, and KLP ϩ KLS) was translationally fused with the flag epitope in the pBSK vector. These constructs were used for in vitro transcription and translation using the TnT-coupled reticulocyte lysate system (Promega, Madison, WI). The in vitro synthesized BRCA1 proteins were then tested for binding to GST-BRAP2 (amino acids 343-524) or GST-importin ␣ (amino acids 12-529). As shown in Fig. 7, both importin ␣ and BRAP2 bind efficiently to the wild-type and the KLN mutant (Fig. 7A,  lanes 1-3 and 10 -12; and Fig. 7C, lanes 1-3 and 10 -12, respectively). However, a significant reduction was observed in bind- ing to both the KLP and the KLS mutants (Fig. 7A, lanes 4 -6  and 7-9; and Fig. 7C, lanes 4 -6 and 7-9, respectively). The KLS mutant seems to interact well with BRAP2 in the yeast two-hybrid assay, but has decreased affinity for BRAP2 in the GST pull-down assay. Since different fragments of BRCA1 were used in these two assays, it is possible that the overall conformation of the protein may be important for this interaction, although the precise reason is presently unknown. Moreover, the double mutant of KLP ϩ KLS completely failed to bind either importin ␣ or BRAP2 (Fig. 7A, lanes 13-15; and Fig.  7C, lanes 13-15, respectively). These results further indicate the specificity of the interaction of BRAP2 with the NLS motifs of BRCA1.
BRAP2 Binds to Other NLS Motifs-To test whether BRAP2 can bind other kinds of NLS, we tested binding to the NLS motifs from the SV40 large T antigen (36) and a bipartite NLS from mitosin (37). For this purpose, the following GST fusion constructs were generated: GST-T, containing the SV40 large T antigen NLS motif (PKKKRKV); GST-mitosin, encoding the mitosin C terminus from amino acids 2927 to 3113, which contains a bipartite NLS motif (KRQRSSGIWENGGGPT-PATPESFSKKSKK); GST-BRCA1Bgl, encoding the BRCA1 BglII fragment from amino acids 341 to 748, which contains both of the functional NLS motifs of BRCA1. Another GST fusion construct encoding amino acids 762-1315 of BRCA1 (without functional NLS motifs, GST-BRCA1) was also prepared as a negative control. The purified GST and GST fusion proteins were used for binding in vitro translated [ 35 S]methionine-labeled BRAP2, importin ␣ (amino acids 12-529) or the yeast BRAP2 homolog. As shown in Fig. 8, BRAP2, importin ␣, and the yeast BRAP2 homolog bind to GST-T, GST-mitosin, and GST-BRCA1Bgl efficiently (Fig. 8A, lanes 3-5; Fig. 8B, lanes 3-5; and Fig. 8C, lanes 3-5, respectively) but not to GST or GST-BRCA1, both lacking NLS motifs (Fig. 8A, lanes 2 and  6; Fig. 8B, lanes 2 and 6; and Fig. 8C, lanes 2 and 6). These data suggest that BRAP2 has general affinity for different NLS motifs, and this function may be conserved since the yeast homolog shows similar properties.

DISCUSSION
In this paper, we have identified a novel human cytoplasmic protein, BRAP2, that binds to the NLS motifs of BRCA1. In addition to recognizing the NLS motifs in BRCA1, BRAP2 binds to other NLS motifs, such as those found in SV40 large T FIG. 6. BRAP2 interacts with the NLS motif of BRCA1 in a yeast two-hybrid assay. A, a panel of BRCA1 wild-type and NLS mutants fused to the DNA binding domain of Gal4 in the pAS vector. B, interaction of the BRCA3.5 bait or the mutants with BRAP2, importin ␣, or BRAP12. The ␤-galactosidase activities were quantified using chlorophenol red B-D-galactopyranoside as substrates. The interaction between pAS-KLP and BRAP2 was decreased dramatically.

FIG. 5. Cellular localization of BRAP2.
A, biochemical fractionation of HBL100 cells. About 2 ϫ 10 7 HBL100 cells were harvested, 5 ϫ 10 6 cells were left unfractionated (T), and the remainder were separated into nuclei (N), membrane (M), and cytosolic (C) fractions. BRAP2 is mainly in the membrane fraction and is also detectable in the cytosolic fraction. For control of the fractionation procedure, p84 served as a marker for nuclear distribution and GST for cytoplasmic distribution. B, immunostaining of Saos2 cells transiently transfected with CEPF-BRAP2. Panels a, phase contrast; b, 6-diamidino-2 phenylindole stain; and c, immunostaining with anti-flag antibody (M2) and secondary Texas red-conjugated anti-mouse antibody. C, HBL100 cells were transiently transfected with either the CHPL-GFP or the CHPL-GFP-BRAP2 contruct, and the transfected cells were observed using a fluorescence microscope. GFP protein is found to be both nuclear and cytoplasmic (panel b). However, GFP-BRAP2 protein is only observed in the cytoplasm (panel d). Phase contrast images of GFP and GFP-BRAP2 transfected HBL100 cells are also shown (panels a and c). D, the expression of GFP or GFP-BRAP2 chimeric protein was detected by anti-GFP antibody following immunoprecipitation with anti-myc antibody. Lane 1, HBL100 cells without transfection; lane 2, HBL100 cells transfected with CHPL-GFP vector alone. GFP protein is about 42 kDa. Lane 3, HBL100 cells transfected with CHPL-GFP-BRAP2, GFP-BRAP2 protein is about 110 kDa. antigen and mitosin. This suggests that BRAP2 recognizes NLS motifs in a manner similar to that of the known NLS receptor, importin ␣. However, unlike importin ␣, which cycles between the nucleus and the cytoplasm, BRAP2 is a cytoplasmic protein. These results raise the possibility that BRAP2 may serve as a cytoplasmic retention protein involved in the regulation of nuclear protein transport.
The nuclear transport machinery has been conserved throughout eukaryotic evolution. Interestingly, human BRAP2 shows significant homology to a hypothetical yeast protein. The highest similarity between these two proteins is in the zinc finger domain, and our preliminary results indicate that deletion of the conserved zinc finger domain in BRAP2 diminishes its binding activity to the NLS of BRCA1 in a yeast two-hybrid assay. Consistent with this, we have shown that yeast BRAP2 has similar binding affinities for the various NLS motifs as does human BRAP2, suggesting that this function is conserved. Unlike SRP1, the yeast homolog of importin ␣ (38, 39), inactivation of BRAP2 in yeast does not result in a lethal phenotype (Refs. 18 and 19 and data not shown). This again suggests that the role of BRAP2 may be in regulating nuclear transport rather than being required to chaperone proteins into the nucleus as does importin ␣.
Nuclear transport is a highly regulated process. The pres-ence of a nuclear localization signal may not be sufficient to direct nuclear import. The accessibility of NLS motifs to the transport machinery is crucial for the efficient targeting of nuclear proteins. A cytoplasmic retention protein can regulate the transport of specific nuclear proteins by masking their NLS motifs. A paradigm of this control is the regulation of NF-B subcellular localization by IB. NF-B dimers can bind to target DNA elements and activate transcription of genes encoding proteins involved in immune or inflammation responses and cell growth control (see Ref. 40 for review). All members of the NF-B family have highly conserved NLS motifs, and the integrity of their NLS motifs is necessary for their nuclear transport (41,42). IB/MAD-3 protein makes direct contact with the NLS residues of both the p50 and p65 subunits of NF-B, and sequesters both p65 and p50 subunits of NF-B in the cytoplasm (17 1, 4, 7, 10, and 13) were incubated with GST (lanes 2, 5, 8, 11, and 14) or GST-importin ␣ (lanes 3, 6, 9, 12, and 15). KLP and KLS mutants both show decreased binding affinity to GST-importin ␣. Panel B, quantitation of the binding efficiency of Importin ␣ to BRCA1 shown in panel A. Panel C, interaction of GST-BRAP2 with BRAC1 (amino acids 303-701) wild-type or NLS mutants. In vitro translated BRCA1 (amino acids 303-701) wild-type or NLS mutant proteins (lanes 1, 4, 7, 10, and 13) were incubated with GST (lanes 2, 5, 8, 11, and 14) or GST-BRAP2 (lanes 3, 6, 9, 12, and 15). KLP and KLS mutants both show decreased binding affinity to GST-BRAP2. Panel D, quantitation of the binding efficiency of BRAP2 to BRCA1 shown in panel C. quired for its targeting to the nucleus (24). These NLS motifs are recognized by the NLS receptor subunit, importin ␣. Significantly, mutations in these sequences that disrupt importin binding also block nuclear transport of BRCA1 (24). In this report, we have provided data to suggest that BRAP2, a novel cytoplasmic protein, also interacts with these two functional NLS motifs of BRCA1. Although BRCA1 is ordinarily targeted to nuclei, we have consistently found it to be mislocated in the cytoplasm of many breast cancer cell lines and tumor specimens (22,24). This mislocalization appears to be regulatory, rather than due to mutations in BRCA1 itself, since exogenous flag-tagged wild-type BRCA1 fails to translocate to the nuclei of breast cancer cells while localizing to the nuclei of other cell lines (24). Moreover, other nuclear tumor suppressor proteins such as p53 and, most recently, WT1 have also been observed to be mislocated in the cytoplasm of breast cancer cells (20,23). These data suggest that a general mechanism may be responsible for the mislocalization of these proteins in breast cancer. Such mislocalization of nuclear proteins is likely to be functionally equivalent to their inactivation, consistent with the recessive genetic behavior of tumor suppressor genes. Whether this mislocalization is a cause or consequence of tumor progression is a vital question to be addressed. However, the mechanisms regulating the nuclear import of these proteins is unknown. Identification of a novel BRCA1-associated protein, BRAP2, as described here may provide some clues to this process.
Interestingly, our results also point to the possibility that BRAP2 binds stronger to BRCA1 than does importin ␣. In yeast two-hybrid assays, the reporter gene activity resulting from the interaction between BRAP2 and BRCA1 is about 5-fold higher than that of importin ␣ and BRCA1 (Fig. 6B). Likewise, in the GST pull-down assay, 70% of total input BRCA1 was bound to BRAP2 and only 50% of it was bound to importin ␣ (Fig. 7). These results provide a scenario in which BRAP2 may retain newly synthesized BRCA1 protein in cytoplasm. Upon further signaling, such as phosphorylation, BRCA1 will dissociate from BRAP2 and bind to importin ␣ for transport into the nucleus. It was noted that BRAP2 recognizes different kinds of NLS motifs (Fig. 8), and it is quite likely that BRAP2 may not serve specifically to regulate the transport of BRCA1 alone.
Whether BRAP2 indeed serves as cytoplasmic retention protein in cells remains to be established. This experiment is complicated by the scarce amount of BRCA1 protein and its modification by phosphorylation in cell. Moreover, BRAP2 may bind many cellular proteins with different NLS motifs. Nonetheless, it will be possible to address this question using a reconstituted nuclear transport system as described (43) in the future.