Identification of the nucleocytoplasmic shuttling sequence of heterogeneous nuclear ribonucleoprotein D-like protein JKTBP and its interaction with mRNA.

JKTBP proteins are related to a family of heterogeneous nuclear ribonucleoproteins (hnRNPs) that function in mRNA biogenesis and mRNA metabolism. JKTBP proteins constituted of isoforms 1, 2, and 1Delta6 are localized in the nucleus. We show that the dominant form JKTBP1 shuttles between the nucleus and the cytoplasm and interacts with mRNA. Immunofluorescence microscopy and immunoblotting of the subcellular fractions and overexpression of JKTBP tagged with green fluorescent protein indicated that JKTBP1 and JKTBP1Delta6, but not JKTBP2, accumulate in the cytoplasm upon polymerase II transcription inhibition. After release from inhibition, the return of accumulated cytoplasmic JKTBP to the nucleus was temperature-dependent. In heterokaryons, green fluorescent protein-tagged JKTBP1 and JKTBP1Delta6 migrated from the HeLa nucleus to the mouse nucleus, but JKTBP2 did not. Using various JKTBP deletion mutants, the 25-residue C-terminal tail was identified as a shuttling sequence like M9. It is conserved in the C-terminal tails of hnRNP D/AUF1 and type A/B hnRNP/ABBP-1. Analysis of its sequence-specific interacting protein indicated that JKTBP nuclear import is mediated by the receptor transportin 1/karyopherin beta2. UV cross-linking revealed the increased occurrence of JKTBP1 directly interacting with poly(A)(+) RNA in the cytoplasm following actinomycin D treatment. We discuss a role of JKTBP in mRNA nuclear export.

JKTBP proteins are related to a family of heterogeneous nuclear ribonucleoproteins (hnRNPs) that function in mRNA biogenesis and mRNA metabolism. JKTBP proteins constituted of isoforms 1, 2, and 1⌬6 are localized in the nucleus. We show that the dominant form JKTBP1 shuttles between the nucleus and the cytoplasm and interacts with mRNA. Immunofluorescence microscopy and immunoblotting of the subcellular fractions and overexpression of JKTBP tagged with green fluorescent protein indicated that JKTBP1 and JKTBP1⌬6, but not JKTBP2, accumulate in the cytoplasm upon polymerase II transcription inhibition. After release from inhibition, the return of accumulated cytoplasmic JKTBP to the nucleus was temperature-dependent. In heterokaryons, green fluorescent protein-tagged JKTBP1 and JK-TBP1⌬6 migrated from the HeLa nucleus to the mouse nucleus, but JKTBP2 did not. Using various JKTBP deletion mutants, the 25-residue C-terminal tail was identified as a shuttling sequence like M9. It is conserved in the C-terminal tails of hnRNP D/AUF1 and type A/B hnRNP/ABBP-1. Analysis of its sequence-specific interacting protein indicated that JKTBP nuclear import is mediated by the receptor transportin 1/karyopherin ␤2. UV cross-linking revealed the increased occurrence of JKTBP1 directly interacting with poly(A) ؉ RNA in the cytoplasm following actinomycin D treatment. We discuss a role of JKTBP in mRNA nuclear export.
Pre-mRNAs transcribed in the nucleus are processed by a variety of processes into mRNAs that are exported into the cytoplasm for translation. In these processes, including alternative splicing, mRNA nuclear export, translational regulation, and mRNA turnover, Ͼ20 different heterogeneous nuclear ribonucleoproteins (hnRNPs) 1 are associated with pre-mRNA and mRNA (1)(2)(3). hnRNPs are groups of proteins consisting of multiple RNA-binding domains, each containing conserved RNP-1 and RNP-2 or hnRNP K homology motifs, and of divergent amino-and carboxyl-terminal domains (1)(2)(3). Although they are primarily nuclear, hnRNPs A1, A2, D, E, I, and K shuttle between the nucleus and the cytoplasm, whereas hnRNPs C and U are always retained in the nucleus (4 -6). The nuclear localization of hnRNPs A1 and A2 bearing the M9 shuttling sequence is mediated by the import receptor transportin 1 (Trn-1) and is polymerase II transcription-dependent (7)(8)(9)(10)(11)(12)(13). The nuclear localization of hnRNP C bearing a classical nuclear localization signal (NLS) is transcription-independent (14,15). Differences in their subcellular movements are connected with their different roles in cells (2,3,15,16). Cytoplasmic shuttling hnRNP A1 is strongly correlated with mRNA nuclear export, and nuclear persisting hnRNP C is associated with prevention of pre-mRNA from moving to the cytoplasm (5,14,15,17).
We previously isolated JKTBP cDNAs by DNA affinity screening of human myeloid leukemia cDNA libraries using a cis-element (JKT41) in intron 9 of the human myeloperoxidase gene (18,19). JKTBP proteins are composed of two RNA-binding domains arranged tandemly and a glycine-and tyrosinerich carboxyl-terminal domain and are more closely related to hnRNP D/AUF1 and type A/B hnRNP/ABBP-1 and more distantly related to hnRNPs A1 and A2 (18,19). There are three isoforms of JKTBP, 1 and 2 (major forms) and 1⌬6 (a minor form), which are abundant in HeLa and HL-60 cells and in tissues, especially in brains and testes (18 -20). JKTBP is abundant in nuclei, but its nuclear localization pathway is not yet known (20). Recombinant JKTBP binds preferentially to poly(A) and poly(G) (18) and supposedly to an AU-rich element of the 3Ј-untranslated region of mRNA (21). This study reports that the predominant isoform JKTBP1 shuttles between the nucleus and the cytoplasm in a pathway consisting of a 25residue shuttling sequence and Trn-1 and of the interaction of the cytoplasmic shuttling JKTBP with mRNA.
Indirect Immunofluorescence Microscopy-Approximately 10 4 HeLa cells were seeded on 15-mm glass coverslips and grown for 1 day. The cells were then fixed with 4% paraformaldehyde in Mg 2ϩ -and Ca 2ϩ -free phosphate-buffered saline at 4°C for 30 min. The cells were incubated with 0.5% Triton X-100 for 3 min. After preincubation with Mg 2ϩ -and Ca 2ϩ -free phosphate-buffered saline containing 2% bovine serum albumin, 2% normal goat serum, and 0.1% Triton X-100, the cells were incubated with 4000-fold diluted rabbit anti-GST-JKTBP1 serum in the same buffer overnight at 4°C (19). The cells were incubated with 200-fold diluted goat anti-rabbit IgG (H ϩ L)-biotin conjugate (Wako Pure Chemicals) in the same buffer and with 500-fold diluted streptavidin-Cy3 conjugate (Sigma) for 30 min at room temperature for each step. The nucleus was stained with 1 g/ml 4Ј,6-diamidino-2-phenylindole (DAPI; Sigma) in Mg 2ϩ -and Ca 2ϩ -free phosphate-buffered saline.
Subcellular Fractionation-HeLa subcellular fractions were prepared as described (22). Cells were harvested, washed, and suspended in buffer A (20 mM Tris-HCl (pH 7.6), 5 mM MgCl 2 , 1.5 mM KCl, 1 mM phenylmethanesulfonyl fluoride (Sigma), 2 mM dithiothreitol, and 0.1% Nonidet P-40). The cells were disrupted in a Dounce homogenizer by 20 strokes using a type A pestle. The homogenate was centrifuged at 760 ϫ g at 4°C for 3 min. The supernatant fluid was used as the cytoplasmic fraction. The pellet was resuspended in buffer B (20 mM Tris-HCl (pH 7.6), 5 mM MgCl 2 , 1.5 mM KCl, 75 mM NaCl, 175 mM sucrose, 1 mM phenylmethanesulfonyl fluoride, 2 mM dithiothreitol, and 0.5% Nonidet P-40) and sonicated in a Model UR-20P Handy Sonic (Tomy Seiko) at the maximum scale three times for 10-s periods. The resultant supernatant was used as the nuclear fraction.
Cloning of JKTBP1⌬6 cDNA-A pair of primers was designed for the 5Ј-end of the coding sequence of JKTBP1 (23 nucleotides, ATG GAG GAT ATG AAC GAG TAC AG) and the 3Ј-untranslated region sequence (TCA ATG TCG TCC TGC AAG ATG) linked with a SmaI linker at their 5Ј-ends. With the Titan One-tube reverse transcription-PCR system (Roche Molecular Biochemicals), the reaction mixture (50 l) containing 500 ng of HL-60 total RNA and 0.4 M primers was incubated at 50°C for 30 min and then processed by PCR in 35 cycles. After agarose gel electrophoresis of the PCR product, the minor band DNA (810 bp) was further amplified with the same pair of primers and 5% dimethyl sulfoxide, cloned into pBluescript SK(Ϫ), and then sequenced. The sequence determined was identical to JKTBP1 except for deletion of nucleotides 773-943. The cDNA clone that compared with the JKTBP genomic sequence was identified as a variant lacking exon 6 and called JKTBP1⌬6 (19).
In Vivo Nuclear Import Assay and Heterokaryon Assay-HeLa cells (4 ϫ 10 6 cells) suspended in 0.5 ml of RPMI 1640 medium were transfected with EGFP-JKTBP plasmids (50 g) by electroporation (200 V, 1180-microfarad capacitance, and low ohm in a Cell-Porator, Invitrogen). About 1 ϫ 10 5 cells were grown on a 15-mm glass coverslip, and the rest were grown for analysis of expression of EGFP-JKTBP fusion proteins. For in vivo import assay, after 24 h, the subcellular distributions of EGFP-JKTBP fusion proteins were studied by fluorescence microscopy. Heterokaryon assay was performed as described by Cá ceres et al. (26). The above-cultured cells were overlaid with 8 ϫ 10 4 NIH3T3 cells and cultured for 3 h. They were then fused by exposure to 100 l of 50% (w/v) polyethylene glycol 3400 (Polyscience) in RPMI 1640 medium for 2 min at 37°C and, after washing, incubated in Dulbecco's modified Eagle's medium for 1 h in the presence of cycloheximide added to the culture 15 min before cell fusion. The cells were fixed and stained with 25 g/ml Hoechst 33342 (Sigma) and studied by fluorescence microscopy. Fluorescent signals and cell images recorded with a cooled CCD camera (SenSys 1400, Photometrics Ltd.) were pseudo-colored. Expression of EGFP-JKTBP fusion proteins was checked by immunoblotting using rabbit anti-GFP serum (CLONTECH).
In Vitro Nuclear Import Assay-HeLa cells on a 15-mm coverslip were permeabilized by treatment with 40 g/ml digitonin transport buffer for 5 min as described by Adam et al. (27). The import reaction mixture (20 l) containing 8 l of rabbit reticulocyte lysate (Promega), 0.1 M GST-JKTBP-(226 -420) as an import substrate with or without 3 M competitor, and 5 l of 4ϫ transport buffer (80 mM HEPES (pH 7.3), 8 mM magnesium acetate, 20 mM sodium acetate, 440 mM potassium acetate, and 4 mM EGTA) containing 2 g/ml each aprotinin, leupeptin, and pepstatin was overlaid on the cells and incubated at 30°C for 30 min. After the reaction, the cells were fixed and probed with 0.5 g/ml diluted rabbit anti-GST IgG overnight at 4°C, and then bound IgG was detected using goat anti-rabbit IgG (H ϩ L)-biotin conjugate and streptavidin-Cy3 conjugate as described above. Anti-GST IgG was prepared by GST-Sepharose affinity chromatography of anti-GST-peptidylarginine deiminase serum (28).
GST Pull-down Assay-Growing HeLa cells were harvested and disrupted in binding buffer (50 mM HEPES (pH 7.6), 150 mM NaCl, 75 mM potassium acetate, 5 mM magnesium acetate, 0.1 mM phenylmethanesulfonyl fluoride, and 0.1% Nonidet P-40) by sonication 15 times for 10-s periods; and after centrifugation at 14,000 ϫ g for 10 min, the resultant supernatant was used to interact with immobilized JKTBP. GST-tagged JKTBP produced by E. coli in the extract was immobilized on glutathione-Sepharose beads (20-l packed volume) by mixing at 4°C for 1 h, and the beads were washed three times with buffer C containing 1 M NaCl and 5 mM dithiothreitol and twice with binding buffer without Nonidet P-40. The beads were mixed with HeLa cell extract (6 mg) in a total volume of 1.5 ml of binding buffer without Nonidet P-40 at 4°C for 4 h and washed with binding buffer containing 200 mM NaCl. Bound proteins were eluted by boiling in SDS sample buffer and were analyzed by immunoblotting using 1000-fold diluted anti-transportin monoclonal antibody D45 (29).
UV Cross-linking of Poly(A) ϩ RNA to Proteins-UV cross-linking was performed as described by Piñ ol-Roma et al. (30). HeLa cells grown on a 10-cm dish were washed, immersed in 2 ml of cold Mg 2ϩ -and Ca 2ϩfree phosphate-buffered saline containing 1 mM CaCl 2 and 0.5 mM MgCl 2 , and irradiated with UV light at 254 nm at a dose of 65 mJ/cm 2 . The cells were replaced in buffer A with 10 mM MgCl 2 , 10 mM vanadyl adenosine, 0.5% Nonidet P-40, 1% Tween 40, and 0.5% sodium deoxycholate. The cells were scraped off and homogenized by four passages thorough a 25-gauze needle with a syringe. After centrifugation at 3000 ϫ g for 5 min, the resultant supernatant was brought to 1% SDS, 1% 2-mercaptoethanol, and 10 mM EDTA and called the cytoplasmic fraction. The pellet was suspended in buffer B and 10 mM vanadyl adenosine, sonicated four times for 10-s periods, and centrifuged at 16,000 ϫ g for 10 min. The resultant supernatant was adjusted to 0.5% SDS, 1% 2-mercaptoethanol, and 10 mM EDTA and called the nuclear fraction. Cytoplasmic and nuclear poly(A) ϩ RNAs were prepared by two repeats of oligo(dT)-cellulose column chromatography (0.18 g; Collaborative Research), and the precipitated RNAs were digested with a mixture of RNase A 1 (25 g/ml) and RNase T 1 (0.25 g/ml) at 37°C for 30 min. The digests were analyzed by immunoblotting using anti-JKTBP serum.

RESULTS
Translocation of Nuclear JKTBP to the Cytoplasm-For examination of whether the nuclear localization of JKTBP is perturbed by RNA synthesis inhibition, HeLa cells were treated with or without actinomycin D for 3 h in the presence of cycloheximide and stained with anti-JKTBP serum. As shown in Fig. 1A, in the untreated cells, the JKTBP signal was mostly confined to the nucleus stained with DAPI (panel a), whereas in the treated cells, the JKTBP signal was found uniformly throughout the cells (panel b), indicating that RNA synthesis inhibition results in cytoplasmic accumulation of JKTBP. Next, we treated cells with a reversible inhibitor of RNA synthesis, DRB (Fig. 1B). Like actinomycin D, DRB also resulted in JK-TBP accumulation in the cytoplasm (panels a and b). After removal of the drug by culturing cells at 37°C for 1 h, the cytoplasmic JKTBP signal decreased, and the signal was found only in the nucleus, like that in the untreated cells, whereas at 4°C, the cytoplasmic signal persisted (panels c and d). These results suggest that JKTBP shuttles between the cytoplasm and the nucleus.
To further examine the two JKTBP1 and JKTBP2 isoforms for subcellular localization in cells treated with or without actinomycin D, we prepared their cytoplasmic and nuclear fractions and analyzed them by immunoblotting using anti-JKTBP serum. On the blots of whole cell lysates, two major bands of ϳ53 kDa (JKTBP2) and 38 kDa (JKTBP1) and at least two minor bands of ϳ48 and 36 kDa were detected. No appreciable difference in the JKTBP contents between the untreated and treated cells was observed (Fig. 1C). In the untreated cells, the relative amounts of JKTBP1 recovered in the cytoplasmic and nuclear fractions were 25 and 75%, respectively (Fig. 1C,  upper panel, lanes 1-3). In the treated cells, the percentage of JKTBP1 in the cytoplasmic fraction increased to 50% (lanes 4 -6). Unlike JKTBP1, JKTBP2 was recovered only in the nuclear fractions from the untreated and treated cells. The minor proteins of 48 and 38 kDa became obscure in the subcellular fractionation. hnRNPs A1 and C as controls for polymerase II transcription-dependent and -independent nuclear localization, respectively, were detected on the same blots. Upon actinomycin D treatment, hnRNP A1 in the cytoplasm increased (middle panel), whereas hnRNP C was always recovered only in the nuclear fraction of the untreated and treated cells (lower panel), as reported previously (5). These results indicated that the increased amount of cytoplasmic JKTBP1 was not due to leakage from the nucleus. In the DRB-treated cells, cytoplasmic JKTBP increased compared with that in the untreated cells (Fig. 1D, lanes 1-6). Upon deprivation of the drug, the increased cytoplasmic JKTBP1 decreased to a low level at 37°C, but not at 4°C (lanes 8 -12). This indicated that the accumulated cytoplasmic JKTBP1 was re-imported into the nucleus.
Differential Effects of Actinomycin D on Nuclear Localization of the Three JKTBP Isoforms-The three JKTBP isoforms are depicted in Fig. 2A. JKTBP1 and JKTBP2 have been characterized previously (18,19), and JKTBP1⌬6 was characterized by cDNA cloning in this work. A JKTBP1⌬6 cDNA encodes a 244-residue polypeptide (27,161 Da) and is generated by alternative splicing of exon 6. JKTBP1⌬6 produced by in vitro transcription and translation was ϳ36 kDa as determined by SDS-PAGE (data not shown). JKTBP1⌬6 produced by cells was found as a minor 36-kDa band on the immunoblots (Fig. 1C). To determine the nucleocytoplasmic movements of the three distinct isoforms, we prepared plasmid constructs consisting of three JKTBP cDNAs fused to a 3Ј-end of the EGFP gene and used them to transfect HeLa cells. After 22 h of cell cultivation, expression of the three EGFP-tagged JKTBP isoforms was confirmed as proteins with the expected molecular masses on the immunoblots probed with anti-GFP serum (Fig. 2B). Parallel cultures were incubated with or without actinomycin D in the presence of cycloheximide for 3 h, and then their subcellular localization were examined by fluorescence microscopy. The fluorescent signal of control EGFP was found diffusely in the cytoplasm and the nucleus in the untreated and treated cells (Fig. 2, C and D, panel a). The signals of EGFP-tagged JKTBP1 and JKTBP1⌬6 were confined solely to the nucleus in the untreated cells (Fig. 2C, panels c and d), whereas following actinomycin D treatment, the JKTBP1 and JKTBP1⌬6 signals appeared to be distributed evenly over the cells (Fig. 2D, panels  c and d). The JKTBP2 signal was always retained in the nucleus before and after treatment (Fig. 2, C and D, panel b). These results indicated that JKTBP1 and JKTBP1⌬6 differed in nucleocytoplasmic movement from JKTBP2.
Identification of Nuclear Localization and Nuclear Export Signals of JKTBP Proteins-First, we searched for potential NLSs of JKTBP by sequence homology. Sequence homologous to neither basic type NLSs nor the nucleocytoplasmic shuttling sequence of M9 was present in JKTBP. To delineate an NLS sequence of JKTBP, N-and C-terminal deletion mutants were prepared by PCR with parent JKTBP cDNAs as a template. All of the mutants are represented according to JKTBP2 numbering (Fig. 3A). The mutant cDNAs were fused to the 3Ј-end of an EGFP-GST gene encoding an EGFP-GST fusion protein. In these constructs, a tripartite fusion gene of EGFP, GST, and mutant genes rather than a bipartite fusion gene of EGFP and mutant genes was constructed because the smallest mutant gene examined encoded a 23-residue peptide. These constructs were used to transfect HeLa cells and were expressed for 22 h. Expression of all EGFP-GST mutant fusion proteins with the expected sizes was confirmed by immunoblotting using anti-GFP serum (Fig. 3B). In parallel cultures, the subcellular localization of mutant proteins was studied by fluorescence microscopy (Fig. 3C). The fluorescent signal of control EGFP-GST (ϳ58 kDa) spread out from the cytoplasm to the nucleus (panel a). In the C-terminal deletion mutants 227-341, 227-398, and 226 -408, the signals were all confined to the cytoplasm (panels b-d), indicating that the 12 C-terminal residues are indispensable for the nuclear localization of JKTBP. In the N-terminal deletion mutants 226 -420, 323-420, 396 -420, and 398 -420, the signals were located in the nucleus (panels e-h). These results indicate that the 23-residue C terminus has NLS activity. A single amino acid substitution of Gly 404 with Ala in JKTBP-(226 -420) reduced its nuclear localization weakly, but significantly (data not shown).
We next examined whether the NLS sequence has nuclear export signal activity. For heterokaryon assay, HeLa cells were labeled with EGFP-tagged JKTBP and EGFP-GST-tagged Nterminal deletion mutants and then fused with unlabeled mouse NIH3T3 cells. After cell fusion, cell culture was continued for 1 h in the absence of new protein synthesis. In the heterokaryon, hnRNP C expressed as a control for a nuclear retention protein remained in the HeLa nucleus ( Fig. 4a; ar- rows indicate the mouse nucleus, and dashed lines approximate outlines of the heterokaryon). Like hnRNP C, JKTBP2 persisted in the HeLa nucleus (Fig. 4b). Unlike JKTBP2, JK-TBP1 and JKTBP1⌬6 were found in both the HeLa and mouse nuclei (Fig. 4, c and d). The N-terminal deletion mutants JKTBP-(226 -420), JKTBP-(323-420), JKTBP-(396 -420), and JKTBP-(398 -420) were all found in both HeLa and mouse nuclei (Fig. 3A). Of the deletion mutants examined, representative JKTBP-(396 -420) and JKTBP-(398 -420) are shown in Fig. 4 (e and f). JKTBP-(396 -420) migrated between the two nuclei in all heterokaryons examined, but JKTBP-(398 -420) migrated in only 40% of the heterokaryons examined. These results indicate that a 25-residue C-terminal tail is necessary and sufficient for nuclear import and export of JKTBP1 and JKTBP1⌬6. Although JKTBP2 has the same C-terminal sequence as JKTBP1, it did not shuttle between the nucleus and the cytoplasm. The 25-residue shuttling sequence 396 -420 contains 5 glycine, 4 glutamine, 3 asparagine, 4 tyrosine, and 3 serine residues. This carboxyl-terminal tail sequence was 72 and 60% identical to the 24-residue carboxyl-terminal tails of hnRNP D/AUF1 and type A/B hnRNP/ABBP-1, respectively, and was weakly homologous to M9 (see Fig. 6C).
Interaction between JKTBP and Trn-1-First, we examined whether JKTBP and hnRNP A1 can compete for in vitro nuclear import into digitonin-permeabilized HeLa cells. The cells supplemented with reticulocyte lysate were incubated at 30°C for 30 min with the GST-JKTBP-(226 -420) fusion protein as a transport substrate in the presence or absence of a 30-fold molar excess of hnRNP A1 or other competitors. The nuclear localization of GST-JKTBP-(226 -420) was detected by anti-GST IgG staining. In the absence of competitor, its incorpora-tion into nuclei was evident (Fig. 5a) depending on the reticulocyte lysate addition (data not shown). This nuclear localization was strongly inhibited by the addition of hnRNP A1 as well as JKTBP1 (Fig. 5, b and d), but not by a shortened form of hnRNP A1/UP1 lacking the M9 domain (Fig. 5c). This suggested that JKTBP nuclear import might be mediated by the nuclear import receptor Trn-1.
These results indicate that Trn-1 is a nuclear import receptor for JKTBP.
Binding of JKTBP1 to Poly(A) ϩ RNA-To understand the role of nucleocytoplasmic shuttling of JKTBP in cells, we studied the in vivo association of JKTBP with poly(A) ϩ RNA by UV cross-linking. Confluent and sparse HeLa cells were treated with or without actinomycin D in the presence of cycloheximide and irradiated under UV light. Cytoplasmic and nuclear poly(A) ϩ RNAs were isolated and digested with RNases. JK-TBP proteins in the digests were detected using anti-JKTBP serum. As shown in Fig. 7A, in the actinomycin D-treated confluent cells, a measurable amount of JKTBP1, but not JK-TBP2, was UV-cross-linked to the cytoplasmic RNA (lane 3), but not to the nuclear RNA (lane 4), whereas in the untreated cells, none of the JKTBP was UV-cross-linked to either the cytoplasmic or nuclear RNA (lanes 1 and 2). Without UV irradiation, none of the JKTBP was bound to poly(A) ϩ RNA in the cytoplasmic and nuclear fractions, regardless of actinomycin D treatment (lanes [5][6][7][8], indicating that the JKTBP detected by UV irradiation was directly associated with poly(A) ϩ RNA. In the sparse cells, JKTBP1 was UV-cross-linked to the cytoplasmic RNA (Fig. 7B, lane 3), but much less to the nuclear RNA in the actinomycin D-treated cells (lane 4), and its amount in the cytoplasmic RNA was equal to ϳ200-fold less than the cellular JKTBP1 content (lanes 3 and 6). In the actinomycin D-un-treated cells, JKTBP was significantly UV-cross-linked to the nuclear RNA, but not at all to the cytoplasmic RNA (lanes 1, 2, and 5). The amount of poly(A) ϩ RNA-cross-linked JKTBP1 in the cytoplasm of sparse cells was about twice that in the cytoplasm of the confluent cells, suggesting a difference in the poly(A) ϩ RNA contents in the two cultures. These results indicate that actinomycin D-induced cytoplasmic shuttling JK-TBP1 is associated with poly(A) ϩ RNA.

DISCUSSION
This work presents evidence that the predominant isoform JKTBP1 shuttles between the nucleus and the cytoplasm and that the shuttling protein can interact directly with mRNA. Shuttling JKTBP was demonstrated in two ways: polymerase II transcription inhibition and heterokaryon formation. Actinomycin D or DRB induced marked translocation of nuclear JK-TBP into the cytoplasm in the absence of new protein synthesis. This cytoplasmic accumulation could account for the decreased nuclear import caused by transcription inhibition, as found with hnRNP A1 and other pre-mRNA-and mRNA-binding proteins (4,12,26). Nuclear protein import in cells is regulated by post-translational modifications such as methylation and phosphorylation of the nuclear import machinery and import proteins, which may be sensitive to transcription inhibition (31)(32)(33). However, the mechanism of the transcription-dependent nuclear import is still unclear. Nonshuttling JKTBP2, unlike the JKTBP1 and JKTBP1⌬6 sequences, has a 119-residue N-terminal extension that has no homology to nuclear retention sequences of hnRNPs C and D (14,34). Its nuclear retention mechanism remains unknown. The findings obtained by transcription inhibition and with heterokaryons indicate that JKTBP1 and JKTBP1⌬6, but not JKTBP2, shuttle between the nucleus and the cytoplasm. Such an isoform-specific nucleocytoplasmic movement may be associated with different roles in cells, as known for other hnRNPs with multiple functions in the nucleus and cytoplasm (2,3,15,16).
Nuclear localization of hnRNP A1 bearing the 38-residue shuttling sequence M9 is mediated by the import receptor Trn-1, and the activities of the NLS and nuclear export signal of M9 are not separated (7,9,10,15,35). Analysis of JKTBP deletion mutants for subcellular localization and their heterokaryon assays indicated that the 25-residue carboxyl-terminal sequence has both NLS and nuclear export signal activities, indicating that it is a shuttling sequence. In this sequence, the amino-terminal 15-residue sequence (positions 396 -410) is partly homologous to the amino-terminal portion in M9 and also to a 12-residue consensus transportin interaction motif in M9 (Fig. 6C) (35). Single amino acid substitution of JKTBP (G404A) revealed a weak but significant reduction of its nuclear import. This seems similar to a characteristic of M9, although a single mutation of M9 (G274A) almost completely abol-ished both the NLS and nuclear export signal activities (9). These 25-and 23-residue NLS sequences of JKTBP interacted with Trn-1. Moreover, JKTBP1 could compete for Trn-1 with hnRNP A1, but not with UP1 lacking M9. These findings give convincing evidence that the nuclear localization of JKTBP is mediated by an M9-transportin pathway. The JKTBP shuttling sequence is highly homologous to the carboxyl-terminal tails of nucleocytoplasmic shuttling hnRNP D/AUF1 and ABBP-1, a nuclear component of apoB mRNA-editing complexes (36). This suggests that their carboxyl-terminal tails are shuttling sequences. JKTBP1 and its N-terminal deletion mutants, when tagged at the carboxyl-terminal end with EGFP, were partially or severely hindered for in vivo nuclear localization. 2 This suggests that the position of a shuttling sequence in molecules is important for interaction with the Trn-1 molecule whose carboxyl-terminal domain is involved in their interaction (10,13). Examination of the structural relationship between Trn-1 and M9 as revealed by structural studies of Trn-1/karyopherin ␤2 complexed with Ran-GTP (13) will be required for precise understanding.
The occurrence of shuttling JKTBP suggested that a possible role of nuclear JKTBP is in mRNA nuclear export. UV crosslinking of JKTBP to poly(A) ϩ RNA showed that, under the transcription inhibition conditions resulting in JKTBP1 cytoplasmic accumulation, the JKTBP1⅐poly(A) ϩ RNA complex was found at higher levels in the cytoplasm than in the nucleus, whereas under transcriptional conditions, the JKTBP1⅐poly(A) ϩ RNA complex was found in the nucleus, but not in the cytoplasm. These results suggest that shuttling JKTBP may carry mRNA from the nucleus to the cytoplasm. The estimated amount of the cytoplasmic JKTBP1⅐poly(A) ϩ RNA complex was far lower than the amount of cytoplasmic JKTBP1. This may primarily account for the low efficiency of UV cross-linking. The cytoplasmic accumulation of the JKTBP⅐poly(A) ϩ RNA complex could be due to either increased stability of mRNAs or decreased nuclear re-import of JKTBP1, or both. In addition to hnRNP A1, the shuttling mRNA-binding proteins TAP, Aly, Y14, SRp20, and 9G8 have been shown to be involved in mRNA nuclear export (37)(38)(39)(40)(41). To understand the meaning of JKTBP1-poly(A) ϩ RNA association, further studies on JKTBP-RNA sequence specificity, compositions of JKTBP⅐mRNA complexes, and relationships to messenger ribonucleoprotein and hnRNP complexes are needed. In addition to nuclear export, various roles of cytoplasmic hnRNPs in translational regulation, mRNA stability, and local localization of mRNA have been uncovered (16,(42)(43)(44). A possible role of JKTBP in the cytoplasm also needs to be studied. JKTBP1 shown as a shuttling mRNA-binding protein is a candidate for mRNA nuclear export in cells.