Trax (translin-associated factor X), a primarily cytoplasmic protein, inhibits the binding of TB-RBP (translin) to RNA.

Trax (Translin-associated factor X) has been shown to interact with TB-RBP/Translin by its coimmunoprecipitation and in yeast two-hybrid assays. Here we demonstrate that Trax is widely expressed, does not bind to DNA or RNA, but forms heterodimers with TB-RBP under reducing conditions. The heterodimer of TB-RBP and Trax inhibits TB-RBP binding to RNA, but enhances TB-RBP binding to specific single stranded DNA sequences. The in vitro interactions between TB-RBP and Trax are confirmed by similar interactions in the yeast two-hybrid system. Cell fractionation and confocal microscope studies reveal that Trax is predominantly cytoplasmic. In contrast, TB-RBP is present in both the nuclei and cytoplasm of transfected cells and uses a highly conserved nuclear export signal to exit nuclei. In addition to a leucine zipper, two basic domains in TB-RBP are essential for RNA binding, but only one of these domains is needed for DNA binding. Trax restores DNA binding to TB-RBP containing an altered form of this domain. These data suggest that Trax-TB.RBP interactions modulate the DNA- and RNA-binding activity of TB-RBP.

The process of mammalian spermatogenesis is highly organized spatially and temporally. Highly controlled transcription and protein expression occur in each developmental stage. During the haploid interval, spermiogenesis, the spermatids become transcriptionally inactive, although there is a need for the synthesis of many proteins essential for the formation of spermatozoa (1). The sex chromosomes encode numerous genes essential for gametogenesis. Because the spermatids are haploid cells, they contain either the X or Y chromosome. Thus, intercellular transport of mRNA in the haploid cells is a critical process to ensure genetic equivalency.
The testis brain RNA-binding protein (TB-RBP) 1 was iden-tified and cloned on the basis of its ability to bind H and Y sequence elements in the 3Ј-untranslated repeats of mouse protamine 1 and 2 mRNAs (2). TB-RBP is the mouse orthologue of human Translin, a single-stranded DNA-binding protein that binds consensus sequence breakpoint junctions of chromosomal translocations in lymphoid malignancies (3). The TB-RBP/Translin consensus binding sequences are also found in TLS-CHOP reciprocal translocations, in therapy-related translocations in acute myeloid leukemias, and in BCR-ABL translocations in chronic myeloid leukemia (4 -6). Sequence analysis in a meiotic recombination hot spot region of human chromosome 16 shows TB-RBP/Translin binding sequences near the breakpoint (7). TB-RBP/Translin has also been proposed to act as a single-stranded DNA-binding transcription factor, which activates early response gene expression in the brain (8). TB-RBP/Translin also functions as a RNA-binding protein mediating intracellular and intercellular mRNA transport (9,10). RNA binding of TB-RBP has been observed in brain and testis, and the binding is dependent upon Y and H sequence elements (2). Many testis-and brain-specific mRNAs have Y and H consensus sequences, and specific RNA⅐TB-RBP interactions have been demonstrated for testis mRNAs encoding protamine 1 and 2 and AKAP 82 and in brain for myelin basic protein mRNA, ␣-calmodulin kinase II mRNA, Tau mRNA, and BC1 RNA (11)(12)(13). Translationally suppressed mRNAs are bound to microtubules by TB-RBP in cellular extracts (14) and recombinant TB-RBP binds specific mRNAs in vitro to reconstituted microtubules (11). The presence of TB-RBP in the cytoplasmic bridges that connect germ cells in a syncytium, together with the association of TB-RBP with transported mRNAs and the cytoskeleton (10), suggest that TB-RBP acts as a transport molecule in the testis for mRNAs in intracellular (from nucleus to cytoplasm) and intercellular (between spermatids) mRNA transport. The shift of subcellular localization of TB-RBP in meiotic and post-meiotic mouse germ cells (10,15), the dendritic translocation of BC1 RNA and TB-RBP in rat hippocampal neurons (13), and the role of TB-RBP for mRNA sorting in dendrites (16) support this hypothesis.
Using Translin as bait in a yeast two-hybrid assay, a protein of unknown function, Translin-associated factor X (Trax), was identified (7). TB-RBP/Translin and Trax are encoded by single-copy genes that are evolutionarily conserved. In addition to having highly conserved sequences in mammals, they are also found in frogs (Xenopus laevis), plants (Arabidopsis thaliana and Oryza sativa), insects (Drosophila), and yeast (Schizosaccharomyces pombe) (17). This extraordinary conservation from yeast to mammals suggests that these molecules play important biological functions. In the adult mouse, TB-RBP mRNA is widely expressed, with its highest levels in testis and brain (15). The subcellular localization of TB-RBP protein is both developmentally and subcellularly regulated during spermatogenesis. During meiosis, the TB-RBP protein primarily localizes in the nuclei of pachytene spermatocytes, whereas in late meiotic prophase and in all subsequent stages of germ cell differentiation, it is in the cytoplasm (10). TB-RBP/Translin also is predominantly a nuclear protein in malignant lymphoid cell lines but not in nonlymphoid cell lines (3). Treatment of nonlymphoid cells with DNA-damaging agents was reported to cause a shift of the TB-RBP/Translin from the cytoplasm into nuclei (18).
TB-RBP and Trax share a 28% identity at the protein level with a conservation of 38% in the C-terminal regions. Both have putative leucine zipper domains at the C terminus of TB-RBP and in the mid-region of Trax (3,7). Interaction between TB-RBP and Trax has been demonstrated by coimmunoprecipitation (19) and in the yeast two-hybrid assay (7,8). Yeast two-hybrid and in vitro binding studies indicate that TB-RBP dimers are the minimum unit needed for DNA or RNA binding (20). Unlike TB-RBP (20), when Trax is used as bait in the yeast two-hybrid system, it does not select itself, suggesting it does not homodimerize. TB-RBP/Translin contains two putative basic domains in the N-terminal region from amino acids 56 to 64 and from amino acids 86 to 97. Changes in the amino acid sequence in the 86 -97 region abolish the DNA-binding activity of Translin (21).
Analysis of the Trax sequence has suggested it contains a putative nuclear localization signal (NLS) (7). The subcellular localization changes in meiotic and post-meiotic cells (10,15) and in nonlymphoid cells after DNA damage (18) indicate a need to shuttle TB-RBP between the nucleus and cytoplasm. Comparison of TB-RBP sequences with known nuclear export signals (NES) suggests the presence of a putative leucine-rich Rev-like NES sequence in its C terminus. Although the Revlike NES is one of the most commonly described NES (22,23), none of the Rev-like NES-containing proteins have been shown to interact with cellular mRNAs.
To date, little is known of the biological function(s) of Trax. Because heterodimeric partners often modulate the activity of proteins that can homodimerize such as c-Fos/c-Jun (25), we set out to determine whether TB-RBP⅐Trax heterodimers could alter the nucleic acid recognition properties of TB-RBP. Here we show that Trax alone does not bind to either DNA or RNA, but Trax does form heterodimers with TB-RBP. The heterodimer of TB-RBP and Trax is unable to bind to RNA, but binds to DNA. Thus, the heterodimerization modulates the substrate specificity in a very unique manner. Cell fractionation and transfection studies reveal that most of Trax is in the cytoplasm. TB-RBP appears to use a Rev-like nuclear export signal to exit the nucleus and requires two basic regions in its N terminus in addition to its leucine zipper to bind RNA.

EXPERIMENTAL PROCEDURES
Expression of Trax Protein in Escherichia coli-A cDNA encoding the complete open reading frame of human Trax was subcloned in-frame with a thioredoxin-6X His-S-peptide N-terminal tag in a pET32a vector (Novagen) and transformed into BL21(DE3) cells. The expressed fusion protein was purified by nickel-nitrilotriacetic acid-agarose column chromatography, and the thioredoxin-6X His tag was removed by thrombin digestion to obtain Trax with an N-terminal S-peptide tag.
Northern and Western Blotting-Total RNA preparations were hybridized with 32 P-labeled Trax cDNA under conditions previously described (15). For Western blotting, tissue extracts were prepared from sexually mature CD-1 male mice using the protocol of Wu et al. (2). Aliquots (30 g) of protein were electrophoresed in 10% SDS-polyacrylamide gels, and the proteins were transferred onto nylon membranes.
The membranes were incubated overnight with TBS containing 5% nonfat dry milk at 4°C and then incubated with a polyclonal antibody to Trax (1:2000) in TBS containing 0.25% nonfat dry milk for 1 h at RT. After washing, the membranes were incubated with protein A conjugated with horseradish peroxidase, and Trax was detected with the enhanced chemiluminescence protocol of Amersham Pharmacia Biotech.
In Vitro Interactions between TB-RBP and Trax-Recombinant mouse TB-RBP (200 ng) was incubated with recombinant human Speptide-tagged Trax for 30 min at 4°C in 200 l of TBS-T (20 mM Tris, pH 7.6, 137 mM NaCl containing 1% Tween 20) or in 200 l of 20 mM HEPES, pH 7.5, containing 1.5 mM MgCl 2 with or without 5 mM DTT. S-protein-agarose beads (20 l) were added to the mixture, and the incubation was continued for 30 min at 4°C with gentle shaking. The mixture was centrifuged for 1 min at 1000 rpm at 4°C, and the pellet was washed three times with TBS-T (1 ml) interspersed with centrifugation at 1000 rpm. The pellets were boiled in SDS loading buffer for 3 min, and proteins were resolved on a 10% SDS-polyacrylamide gel. The gel was stained with SYPRO Orange protein dye (Bio-Rad), and the proteins were visualized by fluorescence according to the manufacturer's protocol.
RNA and DNA Mobility Shift Assays-Electrophoretic mobility shift assays were performed with a DNA probe, Bcl-CL1, or an RNA probe, transcript c, as described previously by Wu et al. (20). For the RNA gel shifts, the RNA⅐protein complexes were routinely digested with T1 RNase (1 unit/assay) and incubated with heparin (5 mg/ml). Bcl-CL1 was labeled with [␥-32 P]ATP using T4 polynucleotide kinase, whereas transcript c was transcribed from a pGEM 3Z plasmid using SP6 RNA polymerase and [␣-32 P]CTP. Recombinant proteins were incubated with 40,000 cpm of DNA or RNA probe for 10 min at RT in 20 l of binding buffer (20 mM HEPES, pH 7.6, 3 mM MgCl 2 , 40 mM KCl, 2 mM DTT, 5% glycerol), and DNA⅐protein and RNA⅐protein complexes were detected following electrophoresis in 4% polyacrylamide gels in TBE buffer. All electrophoretic mobility shift assays were performed after incubation of TB-RBP and Trax at 10 times the final concentration of the proteins in 20 mM HEPES (pH 7.5) containing 5 mM DTT and 1.5 mM MgCl 2 for 30 min followed by 10ϫ dilution of the mixture to reduce the DTT concentration to 0.5 mM.
Site-directed Mutagenesis of Putative Domains of TB-RBP-Sitedirected mutagenesis of the two basic domains and a putative nuclear export signal of TB-RBP was carried out using the QuikChange sitedirected mutagenesis kit (Stratagene, La Jolla, CA) according to the manufacturer's protocol. The wild-type TB-RBP, TB-Nb mutant, and TB-Cb mutant cDNAs were subcloned into a pET28a vector for protein expression in E. coli.
Wild type TB-RBP and TB-RBP NES mutant cDNAs were subcloned in-frame to the C terminus of GFP in the pEGFP C2 vector (CLON-TECH). We named the fusion proteins from these constructs EGFP-TB-RBP and EGFP-TB-RBP NES , respectively. The Trax cDNA was subcloned into the pEBFP C1 or pDsRed1-N1 vectors, which produce fusion proteins with the blue fluorescence protein at the N terminus of Trax (EBFP-TRAX) or the red fluorescence protein at the C terminus of Trax (TRAX-DsRed), respectively. The Trax cDNA was also cloned into the pEGFP C2 vector.
Yeast Two-hybrid Assays-The complete cDNAs of TB-RBP, its mutant alleles, and Trax were subcloned into the EcoRI/SalI sites of pBD-GAL4cam and pAD-GAL4 (Stratagene, La Jolla, CA). Pairs of binding domain and activation domain plasmid constructs were cotransfected into the yeast strain YRG-2. Transformants were selected on SD medium lacking leucine and tryptophan. Protein⅐protein interactions were detected by growth on SD medium lacking leucine, tryptophan, and histidine, and by the 5-bromo-4-chloro-3-indolyl ␤-D-galactopyranoside (X-gal) filter lift assay on SD medium lacking leucine and tryptophan. Strength of interaction was determined by the addition of 3-amino-1,2,4-Triazole (Sigma Chemical Co., St. Louis, MO) to the medium at concentrations of 5-100 mM. The transformation and filter lift assay procedures were performed following the manufacturer's instructions (Stratagene).
Cell Culture, Transfections, and Confocal Fluorescence Microscopy-NIH 3T3 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum and streptomycin. Cells were transiently transfected with the plasmid constructs described above using FuGENE 6 reagent (Roche Molecular Biochemicals) according to the manufacturer's protocol. Cells were grown on two-well chamber slides (Lab-Tek) and fixed 18 h post-transfection using 4% paraformaldehyde in PBS, washed three times in PBS and mounted using Vectashield mounting medium (Vector Scientific).

Expression of Trax in Mouse
Tissues-To determine the extent of Trax expression in various mouse tissues, a Northern blot hybridization was performed with the complete open reading frame of Trax as probe (Fig. 1A). In the mouse, Trax is present as a single transcript of about 2.5 kb. As previously reported for TB-RBP (15), brain and testis contain the highest levels of Trax mRNA, whereas heart, kidney, liver, lung, and spleen contain lower amounts of Trax mRNA. These differences are not due to RNA degradation or unequal loading, because hybridization of the blot to a ␤-actin cDNA reveals similar amounts of undegraded mRNAs on the filter (Fig. 1B).
To determine the protein levels of Trax in mouse tissues, a Western blot was performed using a polyclonal rabbit anti-Trax antibody to detect a protein band of 34 kDa that comigrates with recombinant Trax (Fig. 1C, lane 8). As seen for the mRNAs, the highest levels of Trax expression are seen in testis and brain. In general, Trax protein levels reflect the Trax mRNA levels, and the level of Trax protein in the mouse tissues examined was similar to the pattern reported for TB-RBP (2,15). Quantitation of Western blots using recombinant Trax and recombinant TB-RBP as standards revealed TB-RBP to be present at about a 2-to 3-fold molar excess compared with Trax in these tissue extracts (data not shown).
To define the subcellular distribution of Trax, Western blot analyses of cytoplasmic and nuclear extracts from mouse testis and from NIH 3T3 fibroblasts were performed (Fig. 1D). In extracts from testis and transfected cells, Trax is predominantly found in the cytoplasm (Fig. 1D, lanes 1 and 3), although lower levels are seen in the nuclear fractions (Fig. 1D, lanes 2  and 4). Similar results are seen with confocal microscopy (see below).

TB-RBP Forms a Heterodimer with Trax in Vitro under Reducing Conditions-Because
Trax coprecipitates with TB-RBP with an affinity-purified antibody to mouse recombinant TB-RBP (19) and the proteins interact in yeast two-hybrid assays (7), we set out to determine the conditions needed for Trax-TB⅐RBP interactions. An S-peptide-tagged recombinant Trax protein was mixed with recombinant TB-RBP and incu-bated in a series of buffers. Proteins precipitated with S-protein-agarose beads were then separated on SDS-polyacrylamide gels and stained using a SYPRO-Orange dye (Fig. 2). Trax and TB-RBP do not stably interact in buffers such as TBS or HEPES at neutral pH (Fig. 2, lanes 1 and 2). The presence of divalent cations such as Ca 2ϩ or Mg 2ϩ in PBS, TBS, or HEPES buffer also does not facilitate interaction. However, the addition of 5 mM DTT in HEPES buffer induces the proteins to interact at roughly a ratio of 1:1 suggesting a heterodimer, although a larger oligomer may also be formed (Fig. 2, lane 3). The heterodimer is maintained when the DTT level is diluted to 0.5 mM, but prolonged dialysis of the protein mixture against HEPES buffer in the absence of DTT dissociates the heterodimer (data not shown).
Trax Inhibits TB-RBP⅐RNA Interactions but Enhances DNA Binding-To determine whether the TB-RBP⅐Trax interaction alters the nucleic acid binding of TB-RBP, gel shift assays were performed. RNA electrophoretic mobility shift assays were performed with transcript c and DNA electrophoretic mobility shifts assays with the single-stranded DNA, Bcl-CL1, two probes previously used to characterize Translin/TB-RBP binding to nucleic acids (20). Although Trax itself does not bind to RNA (Fig. 6, lane 2), the addition of increasing amounts of Trax under conditions where heterodimers are formed leads to about a 75% decrease in RNA binding at approximately a 1:1 ratio of Trax and TB-RBP (Fig. 3A, lane 5). The free RNA in these samples is degraded by the T1 RNase incubation routinely performed for the RNA gel shifts. (Fig. 3, lanes 2-5). In the absence of T1 RNase digestion, the RNA probe remains undegraded (Fig. 6).
The addition of increasing amounts of Trax to TB-RBP enhances binding to one of the DNA target sequences for TB-RBP, Bcl-CL1 (Fig. 3B). Severalfold increases in DNA binding are seen at a Trax:TB-RBP ratio of 2:1 (Fig. 3B, lane 4). A higher level of Trax (4:1) does not elicit any further increase in DNA binding (Fig. 3B, lane 5). As seen for the RNA binding assay, Trax alone does not bind DNA (Fig. 7, lane 2).
The TB-RBP⅐Trax Heterodimer Is the Cause of the Changes in RNA and DNA Binding-To demonstrate that it is the heterodimer of Trax and TB-RBP that is causing changes in TB-RBP binding to RNA and DNA, heterodimers were prepared using thioredoxin-tagged Trax and recombinant TB-RBP. This heterodimer migrates more slowly than either a TB-RBP homodimer or a Trax⅐TB-RBP heterodimer, because of the additional amino acids at its N terminus. Gel shift assays carried out using this protein nucleic acid complex reveal that it is the heterodimer of Trax and TB-RBP that binds poorly to RNA (Fig. 4, lanes 4 and 5) and the heterodimer that binds well to specific single-stranded DNA probes such as Bcl-CL1 (Fig. 4,  lanes 9 and 10).
Mutations in Either of Two Basic Domains, Nb and Cb, of TB-RBP Prevent RNA Binding-The TB-RBP protein has been shown to contain a leucine zipper important for DNA and RNA binding (20,21) and two putative basic domains, one of which has been reported to be essential for DNA binding (21) (Fig. 5). Mutations in the basic domains of TB-RBP were made to analyze the TB-RBP-nucleic acid binding in greater detail and to determine their effect on RNA binding. Using radiolabeled transcript c, recombinant Trax, and recombinant TB-RBP as controls, electrophoretic mobility shift assays were performed to evaluate nucleic acid binding changes caused by altering these two basic domains of TB-RBP. As previously seen, the addition of Trax to TB-RBP reduces TB-RBP RNA binding (Fig.  6, compare lane 3 to lane 4). A mutation of the N-basic domain of TB-RBP (TB-Nb) drastically reduces the RNA binding in the absence (Fig. 6, lane 5) or presence of Trax (Fig. 6, lane 6). The protein⅐RNA complex formed by the TB-Nb mutant also migrates more rapidly than the wild type TB-RBP⅐RNA complex. A mutation of the C-basic domain of TB-RBP (TB-Cb) leads to the complete abolition of RNA binding (Fig. 6, lane 7). The addition of Trax to the TB-Cb protein does not show any effect on the reduction in RNA binding of TB-Cb (Fig. 6, lane 8). We conclude that both the N-basic and C-basic domains of TB-RBP are essential for RNA binding and the addition of Trax does not substantially alter its RNA interactions with these mutated proteins.
Mutation in the C-basic Domain of TB-RBP Abolishes DNA Binding but Trax Restores DNA Binding-DNA gel shift assays were carried out using an ␣-32 P-labeled Bcl-CL1 probe with control recombinant TB-RBP and Trax and with the two TB-RBP proteins with altered basic domains, TB-Nb and TB-Cb (Fig. 7). As previously demonstrated, DNA binding of TB-RBP is enhanced by heterodimerization of TB-RBP with Trax (Fig.  7, compare lane 3 to lane 4). The mutation in the N-basic domain of TB-RBP does not appear to affect binding to Bcl-CL1, although a more rapidly migrating DNA⅐protein complex is seen (Fig. 7, lane 5). Interestingly, when Trax is added to TB-Nb under conditions in which heterodimers are formed, no enhanced DNA binding is seen (Fig. 7, lane 6). The mutation in the second basic domain of TB-RBP, TB-Cb, abolishes DNA binding completely (Fig. 7, lane 7). Surprisingly, the addition of Trax, forming a Trax⅐TB-Cb heterodimer, restores DNA binding of the TB-Cb mutant (Fig. 7, lane 8). These data suggest that Trax can induce changes in mutant TB-RBP conformation, which influence its binding to DNA.
TB-RBP and Trax Interact in Vivo-To verify that TB-RBP, TB-Nb, TB-Cb, TB-NES , and Trax interact in vivo, a yeast two-hybrid assay was employed (Fig. 8). As previously reported, TB-RBP readily forms homodimers in yeast (20) (Fig.  8A) and heterodimers with Trax ( Fig. 8B) (7). Trax, however, does not homodimerize (Fig. 8C). The TB-Nb, TB-Cb, and TB-NES mutants can also homodimerize, dimerize with wildtype TB-RBP, and heterodimerize with Trax in yeast (Fig. 8,  D-L), suggesting that the changes in nucleic acid binding we detect by the gel mobility shift assays are specific and mediated by changes produced in the heterodimer as a result of mutations in TB-RBP not due to nonspecific interactions.
Trax Is Predominantly a Cytoplasmic Protein-To determine the subcellular locations of TB-RBP and Trax, transfections were performed in NIH 3T3 mouse fibroblasts using a fusion protein of TB-RBP and green fluorescence protein and Trax with a blue or red fluorescence protein. Confocal fluorescence microscopy reveals that Trax is predominantly cytoplasmic with a high concentration localized around the nucleus (Fig.  9C). TB-RBP is also mostly in the cytoplasm, although a reproducible low level is seen in nuclei, but not in nucleoli (Fig. 9, B  and D). Cotransfections of TB-RBP with Trax also show a predominantly cytoplasmic localization for the two proteins. Identical results are obtained whether the fluorescence protein tag is on the N or C terminus of Trax (Fig. 9, D-F). The staining of the GFP-Trax-transfected cells with the fluorescent Golgi/ER marker BODIPY 558/568 suggests that much of the Trax is Golgi/ER-associated (Fig. 9G). This is supported by  studies where the disruption of the Golgi with brefeldin A leads to a more diffused cytoplasmic localization of Trax (Fig. 9H).
TB-RBP Has a Functional Leucine-rich Nuclear Export Signal-By sequence comparison of TB-RBP with other nuclearcytoplasmic shuttling proteins such as HIV Rev and c-Abl (22, 23), we noted a putative nuclear export signal N-terminal to the leucine zipper of TB-RBP (Fig. 5). This sequence shows high sequence homology to other leucine-rich NES sequences present in a number of shuttling proteins (Fig. 10A) and is highly conserved from Drosophila to humans (Fig. 10B). To test the functionality of this sequence, site-directed mutagenesis was used to disrupt the putative nuclear export signal (from LASELSRLSVN to LASEQSRLSVN) (Fig. 10A). The mutated TB-RBP, TB-RBP NES , was then fused to GFP and transfected into NIH 3T3 cells. In contrast to wild type TB-RBP, which is mostly seen in the cytoplasm (Fig. 9, B and D), the GFP-TB-RBP NES mutant protein localizes mostly in the nucleus with little staining in the nucleoli (Fig. 9I). DISCUSSION Trax is Widely Expressed-As previously shown for TB-RBP (2, 15), Trax is widely expressed in mouse tissues with high levels of mRNA and protein in brain and testis (Fig. 1). By Northern blotting, we detect one transcript of about 2.5 kb, in agreement with the 2.7-kb mRNA reported by Aoki et al. in humans (7). In general, this pattern of expression of Trax in mouse tissues is similar to the expression pattern observed for TB-RBP (2,15), suggesting a functional relationship between these two similar proteins.  Fig. 3B with aliquots of the same protein samples analyzed in Fig. 6. This gel was overexposed to detect any weakly binding complexes. ducing conditions, whereas in the presence of DTT they form a heterodimer or oligomer at roughly a 1:1 ratio (Fig. 2). We believe this interaction is physiologically significant, because Trax was initially isolated as a Translin/TB-RBP interacting protein using TB-RBP as a bait in yeast two-hybrid assays (7) and Trax protein coimmunoprecipitates with TB-RBP with an affinity-purified monospecific anti-TB-RBP antibody (19). A homodimer is the minimal structural unit of TB-RBP that is sufficient for TB-RBP binding to either single-stranded DNA or RNA (20). Heterodimerization of Trax and TB-RBP appears to require the reduction of the cysteine disulfide linkage that stabilizes the TB-RBP homodimer. Incubation of the heterodimer at reduced levels of DTT results in the formation of TB-RBP homodimers, indicating that heterodimer formation is reversible (data not shown). Under the reducing environment of cells, the interconversion of heterodimers and homodimers should occur spontaneously.
Trax Inhibits TB-RBP RNA-binding Interactions and Enhances DNA Binding-Using RNA and DNA gel shift assays, we have found that Trax does not bind RNA or DNA by itself and interactions between Trax and TB-RBP reduce the RNA binding of TB-RBP in a concentration-dependent manner (Fig.  3A). In contrast, heterodimerization increases DNA binding of the complex. Heterodimerization of various transcription factors and coactivators often results in changes in DNA sequence specificities and their transcriptional activation. The Drosophila proteins, spineless and tango, interact in a manner similar to their mammalian orthologues, aryl hydrocarbon receptor and aryl hydrocarbon receptor nuclear translocator, causing changes in DNA binding specificity (24). Similar effects are known for c-Jun/c-Fos heterodimers (25). Many of the heterogeneous nuclear ribonucleoproteins (hnRNPs) play a variety of roles in DNA and RNA metabolism. In in vitro assays, the hnRNPs A1, A2/B1, D, and E bind G-rich single-stranded DNA overhangs similar to the G-biased strands or G-strands found in telomeres. The hnRNP A1 appears essential for maintenance of telomere length in mouse cell lines (26) while having important roles in pre-mRNA splicing and mRNA export (27). To date, little is known about how the specificity for RNA and single-stranded DNA binding is regulated in hnRNPs. The G-strand binding protein of Chlamydomonas reinhardtii (Gbp1p) was recently reported to contain atypical RNP motifs and to bind RNA sequences and single-stranded DNA as monomers (28). The homodimeric form of the protein was reported to lose RNA binding completely, whereas the dimeric Gbp1p shows a strong preference for single-stranded DNA. The decrease in RNA binding when the TB-RBP homodimer is replaced by a heterodimer of TB-RBP and Trax offers many regulatory possibilities in the release of transported or stored mRNAs in post-meiotic male germ cells (10). In light of the cessation of transcription in these cells, post-transcriptional regulation of mRNAs plays a prominent role in cellular differentiation (1).
DNA and RNA Binding Domains of TB-RBP-Many proteins, including the hnRNPs A1, A2/B1, D, and K and Gbp1p have shared nucleic acid binding domains for single-stranded DNA and RNA (26 -28). TB-RBP contains two putative basic domains from amino acids 56 to 64 and from 86 to 97. We confirm the observation of Aoki and colleagues (21) that mutation of the Cb domain (amino acids 86 -97), but not the Nb domain (amino acids 56 -64), leads to loss of DNA binding (Fig.  7). In addition, mutation of either of these two domains abolishes RNA binding (Fig. 6). This suggests that more sequence elements in TB-RBP may be needed for RNA binding than for DNA binding.
A modest change in the electrophoretic mobility of the DNA⅐protein complex is seen when Trax interacts with the TB-Nb protein and Trax dramatically alters the binding of TB-Cb to DNA (Fig. 7). Enhanced DNA binding observed upon heterodimer formation of wild type TB-RBP or the TB-Cb mutant with Trax may be due to improved interaction of the N-basic domain with DNA. Thus, modulation of TB-RBP nucleic acid binding by Trax may be through conformational changes in TB-RBP that are induced by Trax. These changes in protein⅐protein interactions are likely due to heterodimer formation, because similar interactions are seen in vivo using the yeast two-hybrid assay. All of the mutant alleles interact as homodimers or with Trax as a partner for a heterodimer. These interactions indicate that, in vivo, the introduced mutations do not significantly alter the interacting capabilities of the fusion proteins (29). Although interaction strength was not quantified, we detect differences among the TB-RBP alleles. Wild type TB-RBP and TB-RBP NES interact strongly as homodimers and heterodimers, whereas TB-Nb interacts noticeably weaker. These results suggest that the effects on nucleic acid binding in the gel mobility shift assays are predominantly due to the mutations in TB-RBP, not gross alteration of protein⅐protein interactions.
Subcellular Locations of TB-RBP and Trax-Using confocal microscopy of N-terminal GFP-tagged TB-RBP in transfected NIH 3T3 cells, we detect the majority of TB-RBP in the cytoplasm with low levels in nuclei. No TB-RBP is seen in nucleoli, suggesting that TB-RBP is not directly involved with ribosomal RNA transcription or transport. It has been reported that treatment of HeLa cells with mitomycin C or etoposide induces endogenous TB-RBP to move into nuclei (18). However, treatment of the transfected NIH 3T3 cells with doxorubicin, a compound that causes DNA double-strand breaks (at concentrations to 30 M) does not alter the location of the TB-RBP fusion protein in NIH 3T3 cells (data not shown).
Trax has been proposed to have a bipartite nuclear localization signal in its N terminus that could facilitate the movement of TB-RBP/Translin into the nucleus (7). Western blot analyses of cytoplasmic and nuclear protein fractions from mouse testis and NIH 3T3 cells indicate that Trax is predominantly a cytoplasmic protein (Fig. 1D), although a low level of Trax is found in the nuclear fractions (Fig. 1D, lanes 2 and 4). Although we cannot exclude the possibility that Trax transiently enters and rapidly exits nuclei as part of a transport function, most of the "steady-state" levels of Trax protein in our Western blot assays appear to be cytoplasmic. Comparison of the Trax protein sequences of S. pombe, Drosophila melanogaster, A. thaliana, mouse, rat, and human indicate that, although Trax is a very conserved protein, the putative NLS sequence is situated in a region of the protein with lower sequence conservation.
We find that Trax fusion constructs either tagged with blue fluorescence protein at their C terminus or with red fluorescence protein at their N terminus localize predominantly to the cytoplasm of 3T3 cells (Fig. 9, C, E, and G). Although we find colocalization of GFP-TB-RBP and BFP Trax in the cytoplasm of doubly transfected cells (Fig. 9, D-F), we only detect substantial amounts of TB-RBP in the nucleoplasm of the cells. The Trax fusion protein localizes predominantly around the nuclei in the transfected cells, reminiscent of a Golgi/ER distribution. Staining the fixed cells with BODIPY 558/568 reveals a colocalization of Trax with the Golgi/ER (Fig. 9G). Moreover, Trax delocalizes from the Golgi into the cytoplasm following disruption of the Golgi with brefeldin A (Fig. 9H). Although the retention of the Trax fusion protein in Golgi may be preventing Trax from utilizing its putative NLS for nuclear entry, following the disruption of the Golgi, we do not see any substantial increase of Trax in the nucleus. This may be due to rapid transport into and out of the nuclei, levels of nuclear Trax too low to be detected by these methodologies, or the presence of cytoplasmic retention sequences in Trax that override the NLS sequences (30). Subcellular fractionation of Translin and Trax in cerebellar extracts also suggests the two proteins are enriched in the cytoplasm (31).
Other cytoplasmic proteins such as the Fanconi anemia complementation group A gene product, FAA, also contain a putative bipartite NLS sequence (30). A hybrid protein containing the NLS of SV40 large T antigen and FAA also localizes in the cytoplasm of transfected human 293 cells, showing a specific cytoplasmic retention. The subcellular distribution of the Drosophila Cubitus interruptus protein, which mediates Hedgehog signaling, appears to be regulated by opposing bipartite NLS and cytoplasmic targeting/retention signals (32). Trax may also contain similar cytoplasmic retention sequences that override its putative NLS. From our confocal studies, Trax appears to predominantly localize to the Golgi/ER, although we cannot exclude low levels of Trax in the nuclei (Fig. 9G). Studies using Trax as bait in yeast two-hybrid assays have primarily detected Trax interactions with cytoplasmic and Golgi resident proteins. 2 The selective localization and movement of proteins between the nucleus and cytoplasm is often influenced by protein phosphorylation. A GFP fusion of MAPKAP kinase 2, which contains a bipartite NLS, exits the nuclei of transfected cells under stress following phosphorylation (33). Phosphorylation of the adenomatous polyposis coli protein is required for its nuclear import (34). We have found that recombinant Trax can be phosphorylated in testis germ cell nuclear extracts (unpublished data). Whether a phosphorylated (or nonphosphorylated) form of Trax localizes transiently in the nucleus, thereby helping TB-RBP to enter the nucleus as proposed by Aoki et al. (7), remains to be determined.
TB-RBP Contains a Nuclear Export Signal-Previous studies have indicated that TB-RBP exits nuclei of male germ cells toward the end of meiosis (10), is associated with various mRNAs in germ cell cytoplasmic extracts (11), and moves through the intercellular bridges in haploid spermatids (10). Movement of proteins from the nucleus to the cytoplasm utilizes specific nuclear export signal sequences. The HIV I Revlike NES sequence binds directly to a nuclear export receptor CRM1/Exportin, which is involved in the export of various proteins such as MAPKK, PKI-alpha, FMRP, and p53 in a RanGTP-dependent manner (22). Various cellular RNA-binding proteins utilize other shuttling signals such as the M9 signal in hnRNP A1, KNS signal in hnRNP K, and HNS signal in HuR (23). Comparing TB-RBP with known NES sequences, we have found a highly conserved putative leucine-rich NES sequence at its C terminus (Fig. 10). We believe this is a functional sequence, because when we alter the NES of TB-RBP by site-directed mutagenesis, the GFP-TB-RBP fusion protein accumulates in the nuclei (Fig. 9I). This suggests that TB-RBP utilizes a Rev-like NES to move from the nucleus to the cytoplasm. We propose that TB-RBP has a role in mRNA binding and export in male germ cells and neuronal cells where it associates with a specific subset of mRNAs. In the cytoplasm, Trax interacts with TB-RBP producing a heterodimer with reduced affinity for RNA leading to the release of mRNAs in the cytoplasm.