The Relative Levels of Translin-associated Factor X (TRAX) and Testis Brain RNA-binding Protein Determine Their Nucleocytoplasmic Distribution in Male Germ Cells*

Testis brain RNA-binding protein (TB-RBP), the mouse orthologue of human translin, is an RNA and single-stranded DNA-binding protein abundant in testis and brain. Translin-associated factor X (TRAX) was identified as a protein that interacts with TB-RBP and is dependent upon TB-RBP for stabilization. Using immunohistochemistry to investigate the subcellular locations of TB-RBP and TRAX during spermatogenesis, both proteins localize in nuclei in meiotic pachytene spermatocytes and in the cytoplasm of subsequent meiotic and post-meiotic cells. An identical subcellular distribution is seen in female germ cells. Western blot anal-ysis of germ cell protein extracts reveals an increased ratio of TRAX to TB-RBP in meiotic pachytene spermatocytes compared with the post-meiotic round and elongated spermatids. Using COS-1 cells and mouse embryonic fibroblasts derived from TB-RBP null mice as model systems to examine the shuttling of TB-RBP and TRAX, we demonstrate that TRAX contains a functional nuclear localization signal and TB-RBP contains a functional nuclear export signal. Coexpression of both proteins in COS-1 cells and TB-RBP-deficient mouse embryonic fibroblasts reveals that the ratio of

Spermatogenesis is the dynamic developmental process in the testis where the male gamete differentiates into spermatozoa. Following spermatogonial proliferation and differentiation, recombination occurs during meiosis and haploid daughter cells are produced from the two meiotic divisions. This is followed by spermiogenesis where the haploid cells mature into spermatozoa. Throughout spermatogenesis, many testis-spe-cific and testis-enriched mRNAs are temporally and spatially expressed in a tightly controlled manner (1)(2)(3). A number of testicular nucleic acid-binding proteins believed to be involved in post-transcriptional regulation have been identified (1,(3)(4)(5)(6).
One of the nucleic acid-binding proteins, testis brain RNAbinding protein (TB-RBP), 1 the mouse orthologue of human translin (7)(8)(9), has emerged as a highly versatile molecule. Translin, believed associated with chromosomal translocations, is in the nucleus in human lymphoid cell lines where active nuclear transport has been proposed to be involved in processes such as Ig/TCR rearrangements (7). DNA damage has also been proposed to stimulate transport of translin into nuclei (9). TB-RBP is a 28-kDa protein with several domains including a putative nuclear export signal (NES), a leucine zipper domain, and two basic domains in its N terminus (10). TB-RBP functions as both a RNA-binding protein and a singlestranded DNA-binding protein in the testis and brain. As an RNA-binding protein, it links specific mRNAs to microtubules (11,12), mediates intracellular and intercellular mRNA transport, and regulates timing of translation of specific mRNAs in male germ cells (13,14). As a DNA-binding protein, TB-RBP/ translin has been proposed to bind to breakpoint junctions of chromosomal translocations (7)(8)(9)(15)(16)(17) and regulate cell proliferation rates (18,19). The official nomenclature for the mouse TB-RBP gene is Tsn.
Translin-associated factor X (TRAX) has been identified by yeast two-hybrid assays (20) and by immunoprecipitation (21) as a protein that interacts with TB-RBP. TRAX is a 33-kDa protein that shows high amino acid sequence homology to TB-RBP (20). Like TB-RBP, TRAX is also highly expressed in testis and brain, but does not bind nucleic acids directly (10). TRAX contains a putative bipartite nuclear localization signal (NLS) and a leucine zipper domain (10,20). The leucine zipper domain of TRAX is essential for interactions with not only TB-RBP but also with C1D, an activator of the DNA-dependent protein kinase involved in the repair of DNA-double strand breaks and V(D)J recombination (22,23). Although the function of TRAX is unknown, it has been postulated that TRAX complexed with translin is involved in dendritic RNA processing (24) and in DNA double-strand break repair as an interacting partner with C1D (23). All or half of the TRAX protein is absent in null and heterozygous mice lacking both or one allele of TB-RBP, suggesting a close relationship between TB-RBP and TRAX (18). Transfecting TRAX and TB-RBP cDNAs into embryonic fibroblasts derived from TB-RBP null mice has confirmed that TB-RBP is required to stabilize TRAX protein (25). In the absence of TB-RBP, TRAX is ubiquitinated and degraded (25). The official nomenclature for the mouse TRAX gene is Tsnax.
The location of a protein is often indicative of its function (26,27). RNA-binding proteins that participate in post-transcriptional regulation of mRNA frequently exhibit a dynamic subcellular distribution between the nucleus and the cytoplasm (28,29). Many nucleocytoplasmic shuttling proteins contain both a nuclear localization signal and a nuclear export signal, domains that are recognized by specific receptors and adaptors in nuclear pore complexes (30,31). Translin/TB-RBP contains a leucine-rich nuclear export signal that may be dependent on the CRM1 cellular export receptor for subcellular movement (10,32,33), whereas TRAX contains putative bipartite nuclear sequences suggesting an involvement of TRAX in the nuclear transport of translin/TB-RBP (20).
To gain insight into the cellular movements of TB-RBP and TRAX during spermatogenesis, we have investigated their subcellular localizations in mouse testes and have used COS-1 cells and mouse embryonic fibroblasts (MEF) from TB-RBP null mice as model systems to investigate protein shuttling. Immunohistochemistry demonstrates that both proteins are in the nuclei of pachytene spermatocytes and in the cytoplasm of diplotene/diakinesis spermatocytes and the post-meiotic spermatids. An identical subcellular distribution is seen in female germ cells. Transfecting COS-1 cells and TB-RBP-deficient MEFs, we find that the NLS of TRAX and the NES of TB-RBP are functional and provide a means for the two proteins to shuttle between the nucleus and cytoplasm as a complex. Moreover, the ratio of TB-RBP to TRAX determines the "steady state" subcellular locations of the two proteins.

MATERIALS AND METHODS
Immunohistochemistry-Testes from adult male mice and ovaries from pregnant females at days 16 and 18 post-coitum were fixed and processed by the Histological Core Facility of the Children's Hospital of Pennsylvania and the Jackson Laboratory. Immunohistochemistry was performed as previously described (13,34).
Preparation of Testis and Germ Cell Extracts-Total testis extracts were prepared from adult and 17-day-old male CD-1 mice (Charles River Laboratories, Wilmington, MA) using a modification of the procedure of Wu et al. (35). Testes were decapsulated, washed with phosphate-buffered saline buffer, and resuspended in 300 l of RIPA buffer (10 mM sodium phosphate, pH 7.2, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl) containing 0.5 mM phenylmethylsulfonyl fluoride, 0.5 g/ml leupeptin, 0.7 g/ml pepstatin, and 2 g/ml aprotinin. Testes were homogenized in a Teflon glass homogenizer on ice until most of cells were lysed and homogenates were centrifuged at 14,000 rpm at 4°C for 15 min. The supernatants were stored at Ϫ80°C until use. Testicular germ cells were separated by STA-PUT centrifugation as previously described (36).
Western Blot Analyses-For protein analyses of mouse testes, aliquots from total testis (10 g) or germ cell extracts (70 g) were separated by 10% SDS-PAGE. Whole cell lysates from COS-1 cells and TB-RBP-deficient MEFs were prepared in RIPA buffer as previously described (37). Protein concentrations were determined with a BCA Protein Assay Kit (Pierce). For Western blotting, proteins separated by SDS-PAGE were blotted onto polyvinylidene difluoride membranes (Millipore, Bedford, MA), and anti-TB-RBP antibody (1:10,000 dilution) and anti-TRAX antibody (1:1,000 dilution) were used as primary antibodies. TB-RBP and TRAX were visualized with horseradish peroxidase-conjugated protein A (1:3,000 dilution) and the enhanced chemiluminescent (ECL) detection kit (Amersham Biosciences).
Plasmid Construction-The open reading frame of wild type TB-RBP was amplified by PCR and subcloned to generate the expression plasmid pEGFP-TB-RBP (10). Wild type mouse TRAX cDNA was obtained by RT-PCR from testis RNA and inserted into pEGFP-C2 (Clontech, Palo Alto, CA) to generate the expression plasmid pEGFP-TRAX. To create mutated forms of TB-RBP and TRAX expression plasmids as well as coexpression plasmids, wild type or mutated forms of TB-RBP and TRAX were amplified by PCR and inserted into plasmids, pEGFP-C2 or pIRES (Clontech) using restriction enzyme sites introduced with PCR primers. For the NES deletion form of TB-RBP, the plasmid was constructed by insertion of two PCR products (one for amino acids 1-145 and the other for 158 -228) into the multiple cloning sites of pEGFP-C2, which resulted in replacing the entire NES with two amino acids, glycine and threonine. To make the leucine zipper (LZ) deletion form of TRAX, the PCR-mediated mutagenesis method (38) was applied using two sets of primers, 5Ј-GCCATGAACGGCAAAGAAGGACCA-3Ј and 5Ј-CATGTCTTCCCCCGATATCTCCTCCATA-3Ј for the amplification of the upstream open reading frame from the LZ and 5Ј-GATATGGAG-GAGATATCGGGGGAAGACATGC-3Ј and 5Ј-GCCTTAAGAAATGCTC-TCTTCCTGATC-3Ј for the downstream open reading frame from the LZ. Both amplified PCR products lack the LZ but contain an overlapping sequence (about 30 bp) by which two PCR products can be partially annealed and used as the template for the second round of PCR to generate the LZ deletion form of TRAX. All constructs were sequenced before use. The PCR primers and templates used for the creation of each plasmid are described in Table I.
Cell Culture, Transfections, and Microscopy-COS-1 cells and TB-RBP-deficient MEFs were maintained in 10-cm tissue culture dishes containing Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and streptomycin. For transient transfections of COS-1 cells and TB-RBP-deficient MEFs, FuGENE 6 reagent (Roche Applied Science) was used according to the manufacturer's protocol. When leptomycin B (LMB) (3 ng/ml) was used, it was added to the transfected COS-1 cell cultures 4 h before fixation. For fluorescence microscopy, cells grown on two-well chamber slides (Lab-Tek, Champbell, CA) were fixed using methanol and mounted using 95% glycerol in phosphate-buffered saline. Green fluorescent (GFP) fusion proteins in the transfected COS-1 cells and TB-RBP-deficient MEFs were detected by fluorescence microscopy and their subcellular locations were confirmed following 4Ј,6-diamidino-2-phenylindole counterstaining. Similar subcellular distributions were obtained with other fixation methods (4% paraformaldehyde).

TB-RBP and TRAX Are in Nuclei of Meiotic Pachytene Spermatocytes and in the Cytoplasm of Subsequent Germ Cells-To
help define the cellular roles of TB-RBP and TRAX during spermatogenesis, we have investigated by immunostaining the subcellular steady state locations of both proteins in mouse testes using affinity purified antibodies to TB-RBP and TRAX (Fig. 1A). As previously shown (13,18), TB-RBP is localized in the nuclei of pachytene spermatocytes and in the cytoplasm of subsequent cell types (Fig. 1A, a and b). A more detailed analysis of the stages of meiotic prophase after pachytene reveals TB-RBP rapidly appears in the cytoplasm in diplotene/diakinesis cells where it remains in the post-meiotic round spermatids (Fig. 1A, c). TRAX shows a similar pattern of nuclear localization in pachytene spermatocytes and cytoplasmic localization in later stage male germ cells (Fig. 1A, d and f).
A similar pattern of subcellular localization of TB-RBP is seen in female germ cells. At the time mouse oocytes are in pachytene (day 16 post-coitum) (Fig. 1B, a), TB-RBP is in nuclei, whereas at day 18 post-coitum ( Fig. 1B, b) when oocytes are arrested in diplotene/diakinesis, TB-RBP is in the cytoplasm. These data suggest that TB-RBP and TRAX have nuclear functions in pachytene-stage meiotic germ cells and cytoplasmic functions in subsequent meiotic and post-meiotic germ cells. The change in subcellular location of both proteins in the testis germ cells as the germ cells differentiate raises the possibility of a coordinated shuttling of proteins between the nucleus and cytoplasm in mouse germ cells.
The TRAX to TB-RBP Ratio Is Increased in Pachytene Spermatocytes-Extracts from testes of 17-day-old mice and from adult mice were prepared for Western blotting and analyzed for TB-RBP and TRAX using anti-TB-RBP and anti-TRAX ( Fig.  2A). The testes of 17-day-old mice contain pachytene spermatocytes, but lack post-meiotic germ cells. The ratio of TRAX to TB-RBP was increased in testes of 17-day-old mice compared with the testes of adult mice ( Fig. 2A).
To more precisely compare the ratio of TRAX to TB-RBP in germ cells where the two proteins are predominately either in nuclei or cytoplasm, we quantitated the level of TB-RBP and TRAX in Western blots of enriched populations of pachytene spermatocytes, round spermatids, elongated spermatids, and mixed germ cells (Fig. 2, B and C). Although similar amounts of TB-RBP are seen in all the cell types, pachytene spermatocytes contain a higher relative amount of TRAX compared with TB-RBP than the other germ cell types examined. The post-meiotic cell types (round and elongated spermatids) maintain similar levels of TB-RBP, but lower amounts of TRAX relative to TB-RBP.
TRAX Is in Nuclei and TB-RBP in the Cytoplasm in Singly Transfected COS-1 Cells-To determine whether the subcellular localization of TB-RBP could be influenced by TRAX, COS-1 cells were used as a model system. Plasmid constructs expressing TB-RBP and TRAX fused to GFP were individually transfected into COS-1 cells. In transfected COS-1 cells, TRAX is predominantly nuclear (about 91%) ( Fig. 3A) with the remaining transfected cells showing both nuclear and cytoplasmic localization of GFP-TRAX (Table II). This is in agreement with previous findings (23).
To determine whether the bipartite NLS at the N terminus of TRAX (amino acids 11-27) is functional, we fused GFP to a construct lacking the N-terminal 27 amino acids of TRAX (TRAX-  NLS). After transfection into COS-1 cells, TRAX-NLS remained in the cytoplasm (about 94%), suggesting that the NLS is needed for TRAX to move to the nucleus (Fig. 3C, Table II).
In contrast to GFP-TRAX, GFP-TB-RBP is mostly in the cytoplasm (about 96%) (Fig. 3E, Table II), consistent with previous observations that TB-RBP contains a functional leucinerich NES (10). To determine whether the predominantly cytoplasmic localization of GFP-TB-RBP in the transfected COS-1 cells resulted from the export of TB-RBP from nuclei, LMB, the inhibitor for the cellular export receptor CRM1, was added to the COS-1 cell cultures 4 h before fixation. Leptomycin B treatment did not change the subcellular distribution of GFP-TB-RBP in the transfected COS-1 cells (Table II), suggesting that TB-RBP had not shuttled out of the nucleus. Thus, the cytoplasmic localization of GFP-TB-RBP could result from insufficient levels of TRAX for the GFP-TB-RBP in the transfected COS-1 cells, preventing the formation of GFP-TB-RBP complexes that could move into the nucleus.
TB-RBP Translocation to the Nucleus Is Dependent upon TRAX-Because TB-RBP and TRAX interact forming oligomeric complexes via their leucine zipper domains (21,23) and TRAX is dependent upon TB-RBP for its cellular stabilization (25), we reasoned that TB-RBP could translocate to the nucleus as a complex dependent upon the NLS of TRAX. To test whether GFP-TB-RBP could be transported into nuclei by increasing the amount of TRAX in the transfected COS-1 cells, we utilized a coexpression plasmid, pTRAX-IRES-EGFP-TB-RBP, which allows the expression of two consecutive open reading frames from one mRNA. In plasmid pTRAX-IRES-EGFP-TB-RBP, the internal ribosome entry site (IRES) is located between two open reading frames of TRAX and GFP-TB-RBP (Fig. 4A, a). Because of a partially disabled IRES sequence, the rate of translation initiation of the second open reading frame is reduced to about 20% of the first open reading frame. Therefore, in the plasmid pTRAX-IRES-EGFP-TB-RBP we obtained attenuated expression of the second gene, GFP-TB-RBP, compared with TRAX (Fig. 4B, lane 5). When relatively high levels of TRAX are produced, about 37% of transfected COS-1 cells show nuclear localization of GFP-TB-RBP, 46.5% cytoplasmic, and 16.5% show both nuclear and cytoplasmic localization (Fig. 4C, a, Table II), suggesting that TRAX enhances the translocation of TB-RBP into the nucleus. Furthermore, when these transfected cells were incubated with leptomycin B, the percentage of cells showing nuclear localization of GFP-TB-RBP increased to about 87% (Fig. 4C, c, Table  II). This indicates that export of GFP-TB-RBP from the nucleus is the reason about half of the cells transfected with pTRAX-IRES-EGFP-TB-RBP show a cytoplasmic localization of TB- RBP, after being shuttled to the nucleus in a complex with TRAX (see below).
The hypothesis of TRAX-dependent TB-RBP translocation to the nucleus is further supported when a TRAX protein lacking the NLS is used in place of the wild type TRAX. COS-1 cells transfected with a coexpression plasmid, pTRAX(ϪNLS)-IRES-EGFP-TB-RBP, showed a predominant cytoplasmic localization (96.6%) of GFP-TB-RBP (Table II). Moreover, when TRAX and TB-RBP interactions are prevented by deleting the leucine zipper domain of TRAX by using a coexpression plasmid,  pTRAX(⌬LZ)-IRES-EGFP-TB-RBP, the GFP-TB-RBP was predominantly localized in the cytoplasm (95.4%) (Table II). Taken together these results strongly indicate that TB-RBP translocation to the nucleus is dependent upon its interaction with TRAX forming a complex via their leucine zipper domains.
If the hypothesis that a higher relative amount of TRAX directs TB-RBP to the nucleus is correct, reversing the positions of TRAX and TB-RBP in the co-expression plasmid should reverse the subcellular location. When COS-1 cells were transfected with a pTB-RBP-IRES-EGFP-TRAX, a construct (Fig.  4A, b) giving higher relative levels of TB-RBP than GFP-TRAX (Fig. 4B, lane 4), GFP-TRAX was predominantly confined to the cytoplasm (Fig. 4C, e, Table II). This cytoplasmic localization of GFP-TRAX in the presence of high levels of TB-RBP is dependent upon TB-RBP and TRAX interactions because GFP-TRAX is predominantly found in the nucleus when high levels of TB-RBP lacking its LZ are expressed in COS-1 cells following transfection with the coexpression plasmid, pTB-RBP-(157)-IRES-EGFP-TRAX (Table II). The cytoplasmic localization of GFP-TRAX was also seen following incubation with leptomycin B (Fig. 4C, g, Table II), suggesting that a high ratio of wild type TB-RBP to TRAX holds most of the GFP-TRAX in the cytoplasm and prevents its import to the nucleus. Western blotting confirms that transfected COS-1 cells that show a predominantly cytoplasmic localization of GFP-TB-RBP (Fig. 3E) express much more exogenous GFP-TB-RBP than endogenous TRAX (Fig. 4B, lane 3). This establishes that the nuclear localization of both proteins is highly dependent upon sufficient wild type TRAX levels to shuttle TB-RBP into nuclei.
To rule out the effect of endogenous TB-RBP and TRAX on the localization of exogenously expressed proteins, the same coexpression plasmids were transfected into MEFs prepared from TB-RBP null mice. In addition to lacking TB-RBP, the null mice also lack TRAX (18,25). In the null MEFs, singly expressed GFP-TB-RBP and GFP-TRAX were predominantly localized in the cytoplasm and nucleus, respectively (Fig. 5, Table III). When both TB-RBP and TRAX were coexpressed in MEFs, where TRAX was more abundant, the GFP-TB-RBP was predominantly in the nucleus (Fig. 5, Table III). In contrast, GFP-TRAX was predominantly localized in the cytoplasm in the presence of a higher level of TB-RBP (Fig. 5, Table III). This distribution is dependent upon TB-RBP and TRAX interactions because mutants of TRAX and TB-RBP lacking their leucine zipper domains showed singly transfected patterns of GFP-TB-RBP and GFP-TRAX, i.e. cytoplasmic localization of GFP-TB-RBP and nuclear localization of GFP-TRAX in transfected null MEFs (see pTRAX(⌬LZ)-IRES-EGFP-TB-RBP and pTB-RBP-(157)-IRES-EGFP-TRAX in Table III). Moreover TRAX containing its LZ but lacking its NLS could not translocate GFP-TB-RBP into the nucleus (see pTRAX(-NLS)-IRES-EGFP-TB-RBP in Table III), indicating that TB-RBP translocation into the nucleus requires functional TRAX. Western blotting confirmed the same differential expression of TB-RBP and TRAX in the null MEFs as seen with COS-1 cells in Fig. 4B (data not  shown). Taken together, these results establish that in TB-RBP and TRAX null cells the subcellular localization of TB-RBP and TRAX is dependent upon their relative amounts as seen in COS-1 cells.
The NES in TB-RBP Recognizes a CRM1 Export Receptor-The arrest of TB-RBP in the nuclei in cells incubated with leptomycin B suggests that the NES of TB-RBP utilizes the CRM1 receptor (Fig. 4C, c, Table II). To better define the functionality of the leucine-rich NES of TB-RBP (located between amino acids 146 and 157), we transfected three mutant TB-RBP expression plasmids lacking the NES into COS-1 cells (Fig. 6, A and B). Replacing the NES with glycine and threo-nine, GFP-(ϪNES)TB-RBP, TB-RBP was predominantly in the cytoplasm (Fig. 6B, a, Table IV), consistent with the predominant cytoplasmic localization of GFP-TB-RBP in the presence or absence of leptomycin B (Table II). Although both truncated and full-length TB-RBP can oligomerize with TRAX, we believe that the transfected cells did not contain the sufficient amounts of TRAX needed for a significant amount of nuclear localization of the abundant GFP-TB-RBP. When the C-terminal 82 amino acids containing the NES and the LZ domain were deleted from TB-RBP (GFP-TB-RBP-(145), Fig. 6A), a significant number of transfected cells showed nuclear localization (about 31%) or both nuclear and cytoplasmic localization (about 68%) (Fig. 6B, b, and Table IV). GFP-TB-RBP-(145) cannot multimerize because it lacks the LZ needed for homo-oligomerization (39) or hetero-oligomerization (23) and enters nuclei because it is small enough (about a 46 kDa monomer) to diffuse through the nuclear pore complex (40). This would not occur if TB-RBP multimerizes. Interestingly, the addition of the 11-amino acid NES of TB-RBP to the C terminus of GFP-TB-RBP-(145) creating GFP-TB-RBP-(157) (Fig. 6A) changes its nuclear localization to the cytoplasm, as expected for a functional NES (Fig.  6B, d, Table IV) the NES of TB-RBP is recognized by the CRM1 receptor (Fig.  6B, e, Table IV). Consistent with the hypothesis that forms of TB-RBP that can generate multimeric complexes remain in the cytoplasm, the 37-kDa construct GFP-TB-RBP-(158 -228), which lacks the NES, but contains an intact LZ fused to GFP, is restricted to the cytoplasm (Table IV). DISCUSSION We report here that the relative amounts of TB-RBP and TRAX determine their subcellular distributions. TB-RBP, a DNA-and RNA-binding protein, contains a nuclear export signal (10) and is present in both nuclei and cytoplasm of cells (13,14). TRAX, a protein that interacts with TB-RBP, contains a nuclear localization signal and is also present in both nuclei and cytoplasm (20,21). Because TB-RBP and TRAX readily form oligomers, it has been proposed that the nuclear localization of TRAX facilitates TB-RBP translocation into the nucleus (20). The localization of TB-RBP and TRAX in the nuclei of pachytene stage cells and in the cytoplasm of diplotene/diakinesis stage cells from both male and female mouse germ cells suggests different subcellular functions of both proteins at specific windows of time during meiosis. It further raises the likelihood of required intracellular movement of both proteins during male and female germ cell development. Quantitation of TB-RBP and TRAX by Western blotting in extracts prepared from a population of enriched germ cells has revealed approximately equimolar amounts of each protein (0.06 -0.07%) in  mixed germ cells from adult testes (data not shown). This represents an average from total germ cell protein, because we detect an increase in the TRAX to TB-RBP ratio in pachytene spermatocytes (Fig. 2, B and C).
It has been proposed that TB-RBP binds single-stranded DNA found in breakpoint junctions of chromosomal translocations and functions in DNA repair and V(D)J recombination (7)(8)(9). Although the localization of TB-RBP in meiotic nuclei at the time of recombination (Fig. 1A) is temporally consistent with a DNA repair role for TB-RBP, recent data suggest TB-RBP may have other nuclear functions. TB-RBP appears not to be essential for V(D)J recombination, because normal B and T cell development occurs in TB-RBP-deficient mice (18). Furthermore, cultured primary MEFs derived from TB-RBP-deficient mice show no differences in the number of chromosomal gaps and breaks compared with MEFs from their wild type littermates (data not shown). Although there are likely alternative pathways to protect against chromosomal translocations and breaks, we know that TB-RBP is required to stabilize TRAX (25). TRAX has been proposed to be involved in DNA repair through its interactions with the nuclear matrix protein, C1D. C1D activates the DNA-dependent protein kinase, an essential component for the repair of DNA double strand break repair and V(D)J recombination (22,23). Because TRAX interacts with C1D or TB-RBP through its leucine zipper domain in a mutually exclusive manner (23), TRAX association with proteins such as C1D or other DNA-associated proteins in pachytene spermatocyte nuclei merits investigation.
TB-RBP and TRAX closely interact. Mice deficient in TB-RBP also lack TRAX and exhibit abnormalities in fertility, behavior, skin, and growth (18). Although male fertility is maintained in TB-RBP-deficient mice, the males exhibit a greatly reduced sperm count because of a high level of apoptosis in pachytene spermatocytes. This is consistent with an essential nuclear function for TB-RBP, TRAX, or a combination of the proteins at this stage of meiotic prophase (18). Apoptosis has evolved to remove cells whose repair mechanisms are unable to maintain their cellular integrity and cells with damaged genomes often activate the programmed cell death pathway (41). Thus, the abundant amount of cell death seen in the pachytene spermatocytes in TB-RBP-deficient mice likely results from nuclear deficiencies resulting from the loss of TB-RBP and/or TRAX.
RNA-binding proteins in the testis have been proposed to play roles in the processing, transport, stabilization, localization, and translational regulation of mRNAs (4 -6). TB-RBP  binds specific mRNAs to microtubules (11,12) and participates in the intracellular and intercellular transport of mRNAs (14). TRAX interacts with TB-RBP (10,23) and colocalizes with TB-RBP during spermatogenesis, implying intracellular movement of both proteins as a complex. Electron microscopy has demonstrated that TB-RBP and the Ter ATPase complex transport specific mRNAs from nuclei to cytoplasm and through intercellular bridges in mouse male germ cells (14). How TRAX could be involved in mRNA transport and localization as a component of TB-RBP and Ter ATPase complex has yet to be elucidated. The proper subcellular location of proteins is crucial for their normal function. Protein mislocalization of SRY, the sex-determining transcription factor, alters the sex of mammals (42). SRY contains two nuclear localization signals in its high mobility group domain. Mutations of the NLSs cause insufficient nuclear localization of SRY and XY sex-reversal, indicating that a critical level of protein in the nucleus is needed for normal sex determination (42). A protein induced by SRY, Sox 9, contains two NLSs in its high mobility group box (43) and a nuclear export signal, domains allowing the shuttling of Sox 9 between the nucleus and the cytoplasm, critical events for sexual differentiation in mammals (26). Cyclin-dependent kinase cyclin complexes that play a role in triggering DNA replication and entry into mitosis in vertebrate cells also exert their specificity when they are properly localized subcellulary (27).
Translocation of macromolecules between the nucleus and the cytoplasm occurs through the nuclear pore complex in the nuclear envelope that provides a 9-nm diffusion channel for molecules smaller than ϳ60 kDa (40). The transport of most larger proteins and ribonucleoproteins through the nuclear pore complex depends on receptor-mediated processes that involve importins or exportins in association with the Ran GTPase (31,44). Importins and exportins recognize and interact with the NLS and NES on proteins undergoing transport, thereby facilitating their translocation. TRAX contains a bipartite NLS domain at the N terminus that consists of two basic amino acid clusters separated by a region of 10 amino acids (20). The nuclear localization of full-length TRAX and cytoplasmic localization of TRAX deletions lacking the NLS indicate that the NLS of TRAX is functional.
TB-RBP contains a HIV Rev-like NES that appears to be recognized by CRM1 exportin (chromosome region maintenance) (10,32,44). Although constructs containing the NES are detected in the cytoplasm, full-length TB-RBP or TB-RBP lacking its NES are also predominantly cytoplasmic in transfected COS-1 cells. Moreover, incubation with leptomycin B, a nuclear export inhibitor specific to the CRM1 receptor (45-47) (Table II), does not substantially alter the cytoplasmic localization of full-length TB-RBP. Because the import of most nucleoproteins to the nucleus depends on their NLS, it is likely that TB-RBP, which lacks a NLS, requires an interacting partner with a functional NLS. Thus, TRAX can provide the NLS for TB-RBP by forming a complex. In fact, when TRAX levels are high relative to TB-RBP, a large number of transfected cells show TB-RBP in the nucleus. This redistribution of TB-RBP appears dependent upon interaction between TRAX and TB-RBP through their leucine zipper domains because TB-RBP localizes in the cytoplasm in the presence of high levels of TRAX mutants that lack the leucine zipper domain (Tables II  and III).
We propose that the increase of TRAX in pachytene spermatocytes facilitates movement of TB-RBP and TRAX to the nucleus in a complex. In cells, TRAX is highly unstable and becomes ubiquitinated and degraded in the absence of TB-RBP (25). Although MEFs from TB-RBP null mice contain normal amounts of TRAX mRNA, TRAX protein is absent until TB-RBP is reintroduced. This suggests that the dissociation of the TB-RBP and TRAX complex would cause TRAX to be degraded and allow TB-RBP to move to the cytoplasm.
TB-RBP also appears to influence the subcellular distribution of TRAX, because high levels of TB-RBP relative to TRAX appear to prevent the translocation of TRAX. This likely occurs because multimers of TB-RBP bind to cytoplasmic components such as microfilaments (21) or to microtubules (11)(12)(13). We believe the predominantly cytoplasmic localization of TB-RBP lacking the NES is because of the insufficient levels of TRAX to form complexes with TB-RBP that can facilitate the nuclear localization of the high levels of TB-RBP (Fig. 6B, Table IV). The cytoplasmic localization of TB-RBP and TRAX appears dependent upon multimerization, because high amounts of mutant forms of TB-RBP that cannot multimerize and thus, do not bind TRAX, do not impede the translocation of TRAX into the nucleus (see pTB-RBP-(157)-IRES-EGFP-TRAX in Tables II  and III).
The ability of leptomycin B to block movement of TB-RBP from the nucleus to the cytoplasm in COS-1 cells indicates that the NES of TB-RBP is recognized by the CRM1 cellular export receptor. Other types of NES have been identified in RNAbinding proteins including the M9 signal in heterogeneous nuclear ribonucleoprotein A1 (48), the KNS signal in heterogeneous nuclear ribonucleoprotein K (49), and the HNS signal in HuR (50). Unlike the Rev-like NES for the HIV Rev protein that exports the specific HIV RNA sequence, RRE (51), these signals mediate both nuclear import and nuclear export (32).
Lacking the leucine zipper, GFP-TB-RBP-(145) is a 46-kDa monomer small enough to diffuse through the nuclear pore complex and be present in both nuclei and cytoplasm. The two basic domains in the N terminus of TB-RBP (10) may also contribute to the nuclear localization of forms of TB-RBP that are unable to multimerize. Other forms of TB-RBP, GFP-TB-RBP-(158 -228), contain intact leucine zippers and are predominantly localized in the cytoplasm, likely through oligomerization. Biochemical studies from several laboratories have demonstrated that TB-RBP forms octamers (8,16,52). The resultant oligomeric complex of this 37-kDa fusion protein of GFP and other truncated forms of TB-RBP may not be able to diffuse through nuclear pores and thus would be restricted to the cytoplasm.
Based upon the subcellular localization of TB-RBP and TRAX in mouse testes and our transfection data with COS-1 cells and MEFs derived from TB-RBP null mice, we propose the following working model of nucleocytoplasmic shuttling of the TB-RBP/TRAX complex (Fig. 7). In the pachytene spermatocytes where the TRAX to TB-RBP ratio is elevated, TB-RBP and TRAX form a complex in the cytoplasm that likely contains other proteins such as the Ter ATPase (14). Utilizing the NLS of TRAX, the complex is transported into the nucleus. The nuclear translocation of this complex may utilize importin ␣ and ␤ in conjunction with Ran GTPase (30,44). In the nucleus TB-RBP alone or TB-RBP complexed with TRAX or other proteins such as the TRAX interactor, C1D, interacts with DNA or other macromolecules. Although the nuclear function(s) of TB-RBP and TRAX in pachytene spermatocytes or the factors that hold the proteins in the nuclei are unknown, we speculate that TB-RBP and TRAX bind to DNA or to other macromolecules in pachytene nuclei. TB-RBP binding to nuclear meiotic RNAs is equally possible. The nuclear localization of TB-RBP and TRAX is transient, because both move to the cytoplasm in diplotene/ diakinesis in both male and female germ cells. In subsequent round and elongated spermatids, TB-RBP/TRAX complexes are predominantly, but not solely, found in the cytoplasm and are known to contain other proteins such as Ter ATPase and KIF 17b (14,53). These complexes shuttle mRNAs out of the nuclei to the cytoplasm and through intercellular bridges (14,53). TB-RBP and TRAX transiently enter the nuclei of post-meiotic male germ cells utilizing the NLS of TRAX. In nuclei TB-RBP binds to intron-free cAMP-responsive element modulator-regulated mRNAs. Steady state levels of TB-RBP and TRAX in post-meiotic nuclei are barely detectable, suggesting rapid transit out of nuclei in association with specific mRNAs (14). The kinesin motor protein, KIF 17b, is involved in both transcription and transport of these mRNAs because it controls cAMP-responsive element modulator-dependent transcription by regulating the intracellular location of the transcriptional coactivator, ACT (53,54). This mRNA-protein complex, which contains KIF 17b, is exported to the cytoplasm through the nuclear pore complex utilizing the NES of TB-RBP and the CRM1 receptor. In the cytoplasm, the complex transports and localizes mRNAs while suppressing their translation. KIF 17b is released from the complex in the cytoplasm before TB-RBP is released after which protein synthesis of the transported mRNAs occurs. Upon dissociation of the protein components of the mouse ribonucleoprotein complex, TB-RBP and TRAX recycle back to the nucleus. Understanding the complete protein composition of the shuttling complex and its stoichiometry will help define the roles of TB-RBP and TRAX in the posttranscriptional regulation of gene expression during spermatogenesis.