Kap104p-mediated Nuclear Import

Kap104p is a Saccharomyces cerevisiaenuclear import receptor for two essential mRNA-binding proteins, Nab2p and Nab4p/Hrp1p. We demonstrate direct binding of Kap104p to each of these substrates. We have defined the nuclear localization signals in both Nab2p and Nab4p/Hrp1p by Kap104p binding in vitroand KAP104-dependent nuclear import in vivo. The nuclear localization signals map to similar arginine/glycine-rich RNA-binding domains in both proteins and are thus termed rg-nuclear localization signals to distinguish them from classical nuclear localization signals. We also demonstrate that Kap104p, like other known β-karyopherins (or importins), interacts directly with the small GTPase Ran/Gsp1. However, unlike other known import factors, Ran binding is not sufficient to mediate release of substrates from Kap104p; efficient Ran-GTP-mediated substrate release requires RNA. Also, addition of Kap104p to Nab2p and Nab4p/Hrp1p prebound to single-stranded DNA-cellulose stimulated release of both proteins from the resin. We suggest a simple cycle in which Nab2p and Nab4p/Hrp1p, upon import, are released in the nucleus at sites of transcription by the concerted action of Ran-GTP and binding to newly synthesized mRNA. The resulting ribonucleoprotein complexes are exported to the cytoplasm, where Kap104p rebinds to Nab2p and Nab4p/Hrp1p, contributing to their release from mRNA.

In eukaryotic cells, the physical separation of nuclear DNA from cytoplasmic protein synthesis necessitates the bidirectional transport of a diverse set of macromolecules between the nucleus and the cytoplasm. Proteins destined for the nucleus carry a nuclear localization signal (NLS) 1 (1), whereas substrates to be exported from the nucleus harbor nuclear export signals (NESs) (2,3). The signals are recognized by soluble, mobile transporters that mediate transport through the nuclear pore complexes (NPCs) embedded in the nuclear envelope (4,5). These transporters are collectively termed karyopherins (kaps), but specific members of the family have different names (see below). The first kaps to be identified were shown to be responsible for importing basic (1) (now termed classical) cNLS-bearing proteins into the nucleus. In this case, the cNLS is recognized in the cytoplasm by a heterodimer of kap ␣ (6) (also termed importin ␣ (7)) and kap ␤ (6) (also termed importin ␤ (8)). The ␣ subunit binds to the cNLS, whereas the ␤ subunit is responsible for docking the complex to the NPC (6, 9 -12). It is proposed that a series of docking and release steps facilitates movement of the ␤/␣/NLS-cargo from the cytoplasmic filaments through the NPC (13).
Other macromolecules transported through the NPC follow a similar fate but are specifically recognized by different kap family members. The Saccharomyces cerevisiae genome sequence suggests that the yeast kap family may contain as many as 14 members, identified by their predicted primary structural similarity to yeast kap ␤, Kap95p (5,14,15). Although many members have been shown to be bona fide transporters, others remain to be characterized. The first kap ␤-related proteins shown to be transport factors were a mammalian protein termed karyopherin ␤2 or transportin (kap ␤2/Trn) (16,17) and its yeast orthologue, Kap104p (18). Both proteins import mRNA-binding proteins into the nucleus. Kap ␤2/Trn recognizes a 38-amino acid residue sequence termed M9 within hnRNP A1 and mediates its nuclear import (16). Similarly, Kap104p transports the nuclear RNA-binding proteins Nab2p and Nab4p/Hrp1p (18). Nab4p/Hrp1p is the yeast protein most similar in sequence to mammalian hnRNP A1 (19).
In addition to the kaps, the small GTPase Ran and its interacting protein, p10/NTF2, play a critical role in providing directionality to the import and export processes (20,21). In the nucleus, Ran is believed to be maintained in its GTP-bound state by the nuclear-restricted GTP exchange factor RCC1 (22). In contrast, the localization of the Ran GTPase-activating protein to the cytoplasm and the cytoplasmic filaments of the NPC (23,24) is thought to ensure that cytoplasmic Ran is maintained in its GDP-bound form. This separation of the two pools of Ran is thought to play a role in vectorial transport across the NPC, acting as a molecular switch in the binding and release steps that occur during transport through the NPC (4,20,21). For example, the formation of an import complex containing kaps ␤/␣ and the cNLS is stable in the presence of Ran-GDP, but Ran-GTP triggers its disassembly (13,25). In contrast, the formation of an export complex, for example, between an NEScontaining protein and its export factor, Crm1/Exportin 1, is stabilized by Ran-GTP in a trimeric complex (26). Presumably, as this complex reaches the cytoplasm and GTP hydrolysis on Ran is catalyzed by Ran-GTPase-activating protein, the complex disassembles. This mechanism is thought to ensure that cargoes are loaded and released from their appropriate carriers in the correct locations. Exactly how energy is utilized to drive translocation is not clearly understood, but because GTP hydrolysis on Ran is not required for (at least) a single round of either import or export (27)(28)(29)(30), the energy input may, in part, come from maintaining this gradient of Ran, thereby indirectly driving translocation.
Although the mechanisms of mRNA export are yet to be elucidated, it is likely that proteins bound to the mRNA act as adaptors for export factors. The human immunodeficiency virus protein, Rev is an excellent example of such an adaptor. It binds specifically to incompletely spliced viral transcripts (31)(32)(33)(34) and, in turn, binds to cellular Crm1/Exportin 1 through its NES, thereby mediating export of the viral mRNA (26,35,36). Because several proteins bind to mRNA and are also exported from the nucleus, these shuttling proteins are candidate adaptors mediating mRNA export (3,37,38). The best characterized such protein is hnRNP A1. hnRNP A1 is one of the most abundant hnRNP proteins in mammalian cells and is essential for efficient mRNA export (38,39). Interestingly, the 38-amino acid residue M9 sequence, which is recognized by Kap ␤2/Trn (16,17), is sufficient to confer shuttling activity upon a passenger protein (40). A direct role, however, for M9 or Kap ␤2/Trn in mRNA export has not been shown.
In yeast, the Kap104p substrates Nab2p and Nab4p/Hrp1p (18) also bind mRNA in the nucleus and are essential for efficient mRNA processing and export (19,(41)(42)(43). To further understand the function of Kap104p in nuclear import, and potentially mRNA export, we have undertaken a molecular approach to characterizing the interactions between Kap104p and Nab2p and Nab4p/Hrp1p. We provide evidence for a simple cycle in which, upon import, the Nabs are released from Kap104p by the concerted action of Ran-GTP and newly transcribed RNA. We suggest the Nabs are then exported with the mRNA, at which point, Kap104p rebinds to RNA-binding domains of Nab2p and Nab4p/Hrp1p, facilitating their release from the mRNA.
KAP104 and NAB2 open reading frames were amplified from yeast genomic DNA by polymerase chain reaction, cloned into the BamHI site of pGEX-2TK (Amersham Pharmacia Biotech), and termed Nab2-pGEX2TK. Likewise, NAB4/HRP1 was cloned into the BamHI site of pGEX-4T1 and termed Nab4-pGEX4T1. Kap95p-GST (13) was a gift from M. Rexach, Stanford University. Full-length Nab2p tagged with green fluorescent protein (GFP) was a gift from C. Guthrie (University of California at San Francisco, CA).

Nab2p and Nab4p Deletion Constructs
Nab2p deletion constructs ⌬8 -188, ⌬200 -264, and ⌬306 -499 were constructed by polymerase chain reaction. Oligonucleotides containing SacI restriction sites were designed to amplify NAB2 and pGEX2TK while omitting the desired site of insertion. The fragments were purified, digested with SacI, and ligated. Recombinant plasmids were sequenced to confirm deletions. Nab4p fragments encoding amino acid residues 492-534, 392-534, and 241-400 were amplified from yeast genomic DNA (Promega, Madison, WI) using polymerase chain reaction and oligonucleotides containing restriction sites for cloning as described below. Fragments were cloned into the BamHI and EcoRI sites in pGEX-2TK or into the HindIII and EcoRI sites of NES-GFP2-NLS (pKW430) or p12-GFP2-NLS (pKW431) (36). PKW431 was a gift from K. Weis (University of California at San Francisco, CA). In each case, the cNLS was removed and replaced with NAB2 or NAB4/HRP1 coding sequence upstream of, and in-frame with, a mutant (p12; nes) or wild type NES followed by two GFP coding sequences (2ϫGFP). The carboxyl-terminal deletion of Nab4p was constructed by digesting full-length Nab4-pGEX4T1 with NdeI and SmaI. T4 DNA polymerase was then used to create blunt ends, and the sample was ligated.

Recombinant Protein Preparation
GST fusion proteins were expressed separately in Escherichia coli (BL21 DE3 pLYS-S) (Novagen, Madison, WI). Overnight cultures were diluted and grown for 3-4 h at 37°C. Expression of the chimeras was induced by addition of isopropyl-1-thio-␤-D-galactopyranoside to a final concentration of 2 mM, and the cells were incubated at 37°C for an additional 2 h. Cells were harvested and washed with water, and pellets were rapidly frozen with dry ice ethanol. Pellets were then resuspended in 30 ml of transport buffer (20 mM HEPES-KOH, pH 7.4, 110 mM KOAc, 2 mM MgCl 2 , 1 M CaCl 2 , 1 M ZnCl 2 , 1 mM EDTA, 1 mM dithiothreitol, 0.1% Tween 20) containing 300 l of solution P (0.4 mg/ml pepstatin A, 18 mg/ml phenylmethylsulfonyl fluoride (18)) and immediately lysed by sonication. The resulting lysate was clarified by centrifugation at 2500 ϫ g for 15 min and then 17,000 ϫ g for 20 min at 4°C. The lysates were frozen with dry ice ethanol and stored at Ϫ80°C.

Ran Preparation
Yeast Ran (Gsp1) (45) was a gift from M. Floer and G. Blobel (The Rockefeller University, New York, NY). Ran was loaded with three forms of guanine nucleotide, GTP, GTP␥S (guanylyl imidodiphosphate, nonhydrolyzable form of GTP), and GDP as described (13). A reaction mixture containing 30 mM EDTA, 4 mM dithiothreitol, 1.2 mM guanine nucleotide in binding buffer (20 mM HEPES-KOH, pH 6.8, 2 mM MgOAc 2 , 150 mM KOAc) was incubated with equal volume of Ran (0.8 g/l) at room temperature for 90 min. Magnesium acetate was added to a final concentration of 30 mM, and samples were incubated at 4°C for 15 min and stored at Ϫ80°C.

In Vitro Solution Binding Assay
Protein Purification and Binding-Protein lysates were thawed in ice water and incubated with 20 l of glutathione (GT) resin (Amersham Pharmacia Biotech) preequilibrated in transport buffer for 1 h at 4°C on a platform rocker. Beads were pelleted for 5 s and washed with transport buffer four times. To remove proteins from the resin, the GST chimeras were cleaved with 0.3 National Institutes of Health units of thrombin (Sigma) and incubated at room temperature for 15 min. Thrombin was deactivated after the incubation time by the addition of 1 unit of hirudin (Sigma). Cleaved proteins were collected by removing GT-bound GST-containing proteins by centrifugation as above. The supernatant was transferred to a new tube and cleared of residual GST-containing proteins by addition of 10 l of fresh resin, followed by a 30-min incubation at 4°C and centrifugation.
For assays involving interactions between two or more proteins, the initial binding steps were done as above, and cleaved proteins were incubated with GST fusions, prebound to the GT resin, for 1 h at 4°C. Samples were centrifuged, and the supernatant precipitated with trichloroacetic acid. The bound fractions were washed three times prior to cleavage from the resin as described above and analyzed by SDS-PAGE and Coomassie Blue staining.
Single-stranded DNA (ssDNA) Binding-Recombinant proteins were purified and cleaved as above. ssDNA-cellulose (Amersham Pharmacia Biotech) was prepared according to the manufacturer's instructions and stored at 4°C. Before use, 100 l of slurry was washed twice with transport buffer. ϳ7.5 g of cleaved proteins was incubated in 500 l with ssDNA (35 l of 50% slurry) for 1 h at 4°C. Samples were centrifuged briefly and washed three times with transport buffer. Aliquots of the ssDNA-protein complex were made and incubated with additional protein samples prepared as described above for 1 h at 4°C. Samples were centrifuged, and the unbound fractions were collected and precipitated with trichloroacetic acid. Bound fractions were boiled with SDS sample buffer, and samples were analyzed by SDS-PAGE and Coomassie Blue staining.
Ran Binding-Protein complexes were formed as above and separated into aliquots. Three g of Ran-GTP␥S, Ran-GTP, Ran-GDP, or buffer alone was added, and samples were incubated for 30 min at room temperature. Samples were centrifuged for 3 s, and unbound fractions were collected. The beads were washed, and bound proteins were cleaved from the resin with thrombin. Bound and unbound fractions were analyzed by SDS-PAGE and Coomassie Blue staining.

GFP Analysis
Wild type or mutant cells containing GFP-tagged proteins were grown in selective media, and cells were visualized directly by fluorescent microscopy using a Zeiss Axioskop 2. Images were captured using a Spot camera (Diagnostic Instruments Inc., Sterling Heights, MI). In the case of temperature-sensitive cells, cultures were grown overnight at 23°C, shifted to 37°C for the indicated time, and examined for GFP distribution. Sample temperatures were maintained on the microscope using a heated stage (Minitube, Minitube Canada, Woodstock, ON, Canada).
The shuttling assay was performed as described by (36,46) with the following modifications: kap104 -16 cells containing Nab2p-GFP were grown at 23°C overnight and then shifted to 37°C or left at 23°C in the presence or absence of 0.1 mg/ml cycloheximide for 4.5 h. Nab2p-GFP was visualized by fluorescent microscopy.

Ran-GTP/RNA Release
Kap104p-GST was expressed and purified in E. coli as described above. Immobilized Kap104p-GST was incubated for 1 h with recombinant Nab4p/Hrp1p at 4°C. Complexes were then incubated with 3 g of Ran-GTP, 10 g of total yeast RNA, or 3 g of Ran-GTP ϩ 10 g of total yeast RNA for 30 min at room temperature. Unbound fractions were collected, and bound fractions were washed and cleaved with thrombin. Samples were analyzed by SDS-PAGE and Coomassie Blue staining.

RESULTS
Identification of an rg-NLS in Nab2p and Nab4p/Hrp1p-Immunopurification studies showed that Kap104p isolated from the yeast cytoplasm copurified with its two import substrates, Nab2p and Nab4p/Hrp1p (18). However, it is unclear from these results whether this represented a trimeric complex or a pair of two member complexes containing Kap104p and Nab2p or Nab4p/Hrp1p. To examine this, GST chimeras of Kap104p, Nab2p, and Nab4p/Hrp1p were expressed in E. coli. Nab2p-GST and Nab4p/Hrp1p-GST were immobilized on GT-Sepharose and incubated with recombinant Kap104p. In both cases, Kap104p bound directly to the fusion proteins in vitro (Fig. 1A). The relative amounts of protein, as detected by Coomassie Blue staining, suggests that under these conditions, Kap104p binds to Nab2p and Nab4p/Hrp1p with similar capacity. Under the same conditions, we were unable to detect the formation of a trimeric complex (data not shown). Thus, the cytoplasmic interactions previously observed were likely two independent, cytoplasmic, Kap104p-substrate import complexes.
To identify the NLS recognized by Kap104p, we first examined the protein sequence of Nab2p. Nab2p contains two distinct functional RNA-binding motifs: amino acid residues 200 -249 are rich in glycine and arginine residues and contain a sequence that fits the consensus RGG box sequence, and amino acid residues 249 -474 contain a CCCH zinc finger domain. Nab2p also contains a Gln-rich region between amino acid residues 101 and 172 (38,43,47). We expressed GST chimeras harboring deletions of these regions in E. coli and tested their ability to bind Kap104p using in vitro binding assays (Fig. 1B). Nab2p lacking the Gln-rich region (⌬8 -188; Fig. 1B) or the CCCH zinc finger domain (⌬306 -499; Fig. 1B) maintained their ability to bind Kap104p. In contrast, Nab2p lacking the RGG box, within amino acid residues 201-264, failed to bind Kap104p (⌬201-264; Fig. 1B), suggesting that the RGG motif of Nab2p is necessary for Kap104p binding. To test whether this domain is sufficient for binding Kap104p, the RGG motif (amino acid residues 200 -249) fused to GST was assayed for binding to Kap104p. As shown in Fig. 1B (200 -249)  Similarly, we assayed Nab4p/Hrp1p to identify the region responsible for binding to Kap104p. Nab4p/Hrp1p contains two RNA recognition motifs between amino acid residues 161 and 291 and an RG-rich carboxyl terminus (19, 38, 50, 51). Deletion of the carboxyl terminus (⌬321-534; Fig. 1B) of Nab4p/Hrp1p abolished its ability to interact with Kap104p. Amino acid residues 241-400 of Nab4p/Hrp1p fused to GST, containing a highly basic region (pI 10.01) between amino acid residues 281-350, and a short stretch (amino acid residues 329 -350; pI 12.1) rich in Gly, Arg, and Asn residues, failed to bind to Kap104p in the solution binding assay (data not shown). In contrast, the carboxyl-terminal fragments consisting of amino acid residues 392-534 (pI 9.11) or amino acid residues 492-534 (pI 11.44) of Nab4p/Hrp1p fused to GST were sufficient to support Kap104p binding in the solution binding assay (Fig.  1B). The extreme carboxyl terminus (amino acid residues 506 -534) is similar to the RGG box of Nab2p but does not fit the RGG box RNA binding consensus sequence (38,47). Fig. 1C shows the amino acid residue sequence alignment between the Nab2p RGG box and Nab4p/Hrp1p. In Nab4p/Hrp1p, the do-FIG. 1. Nab2p and Nab4p/Hrp1p bind Kap104p through a conserved RG-rich domain. A, GST fusions of Nab2p and Nab4p/Hrp1p (Nab4p) were expressed in E. coli and immobilized on glutathione-Sepharose 4B beads to yield ϳ0.3 g of protein per l of packed beads. Recombinant Kap104p (10 ng/l) was added and incubated at 4°C for 1 h. After washing, bound fractions were released from GT resin by cleavage with thrombin and analyzed by SDS-PAGE and Coomassie Blue staining. As a control, GST alone was bound to GT to yield ϳ1.5 g of protein per l of packed beads, incubated with 10 ng/l of recombinant Kap104p, and treated as above. No detectable Kap104p bound to GST alone. B, deletion constructs of Nab2p and of Nab4p/Hrp1p were expressed as GST fusions and bound to GT-Sepharose. The line diagram represents the regions of each protein expressed. Structural domains of Nab2p and Nab4p/Hrp1p are indicated and described in the text. Purified recombinant Kap104p was added (10 ng/l), and assayed for its ability to bind each chimera by SDS-PAGE of bound and unbound fractions. Coomassie Blue staining of the Kap104p in each fraction is shown on the right. C, the rg-NLS of Nab2p was compared with fulllength Nab4p/Hrp1p using MegAlign (Lipman-Pearson: ktuple, 2; gap penalty, 4; gap length penalty, 12). main is 37 amino acid residues long, highly basic, and lacking hydrophobic residues. In Nab2p, the RG-rich 41 amino acid domain is similarly basic and contains only 6 hydrophobic amino acid residues.
To test whether these RG-rich motifs in Nab2p and Nab4p/ Hrp1p could serve as functional NLSs, the protein sequences were tagged with a tandem repeat of two GFPs (2ϫGFPs) and visualized in living cells ( Fig. 2A). Although 2ϫGFP alone localized to the cytoplasm (data not shown) each RG-rich Kap104p-interacting domain of Nab2p and Nab4p/Hrp1p targeted the chimeric GFP to the nucleus, as did the cNLS of SV40 large T antigen attached to the same GFP (36). The results demonstrate that these RG-rich Kap104p interacting domains act as functional NLSs in vivo. To distinguish these NLSs from the basic cNLSs we term them rg-NLSs.
In an attempt to determine whether the in vitro binding data reflect cellular Kap104p-mediated import of the rg-NLSs, we expressed each fusion in kap104 ts cells (Fig. 2B). In these cells, at 23°C both rg-NLS-and cNLS-containing chimeras were imported into nuclei. After shifting the cells to 37°C, the rg-NLS-containing chimeras were mislocalized to the cytoplasm, whereas the cNLS-GFP chimera remained nuclear. Similar results were observed when KAP104 was placed under control of a GAL promoter, and its expression repressed by growth of cells on glucose-containing medium (data not shown). Interestingly, like a cNLS (36), the rg-NLS can be overcome by attaching a functional NES to the chimera (Fig. 3). This suggests that the nuclear localization of these chimeras is not simply a result of binding to sites within the nucleus. Taken together, these data demonstrate that Kap104p mediates the import of Nab2p and Nab4p/Hrp1p through its interaction with the rg-NLS domains of each protein.
Kap104p Is Able to Displace Nab2p and Nab4p/Hrp1p from Single-stranded DNA-Nab2p and Nab4p/Hrp1p have been shown to bind mRNA using UV cross-linking studies and are believed to play a role in the export of mature polyadenylated mRNAs (19,(41)(42)(43). The interaction of Nab2p with RNA is likely mediated by both its RGG domain and the CCCH motif, as each is sufficient to bind to ssDNA, but when both are present, the interaction is strengthened (43). Similarly, Nab4p/ Hrp1p RNA binding is likely mediated by both the RNA recognition motifs and the RG-rich carboxyl terminus. Both the amino-terminal fragment (⌬321-534; Fig. 1B) containing RNA recognition motifs 1 and 2 and the carboxyl-terminal fragment (392-534) lacking the RNA recognition motifs but containing the RG-rich domain bind to ssDNA in vitro (data not shown). Because both the RGG-RNA-interacting domain of Nab2p and the RG-rich domain of Nab4p/Hrp1p (at least partially) overlap with the Kap104p-interacting domains, we wished to determine whether Kap104p could bind to Nab2p or Nab4p/Hrp1p while complexed with RNA. The formation of such a complex would support a possible role for Kap104p in mRNA export. Nab2p and Nab4p/Hrp1p were immobilized on ssDNA-cellulose and challenged with Kap104p. As shown in Fig. 4, Kap104p did not bind to preformed complexes and remained in the unbound fractions. Interestingly, however, Kap104p also removed Nab2p and Nab4p/Hrp1p from the ssDNA, in a concentration-dependent manner, shifting them to the unbound fractions. As a control, Kap95p was added in similar amounts but had no effect on the Nab-ssDNA complexes. These results demonstrate that binding of Kap104p to the rg-NLS of both Nab2p and Nab4p/Hrp1p destabilizes the interaction of these proteins with ssDNA. The relatively large amounts of Kap104p needed to produce complete displacement from ssDNA are most likely due to the presence of other RNA-binding motifs contributing to the interaction. Nevertheless these data suggest that FIG. 2. Kap104p-interacting domains are functional Kap104pdependent NLSs. Amino acid residues 200 -249 of Nab2p (Nab2p rg-NLS), residues 492-534 of Nab4p/Hrp1p (Nab4p rg-NLS), and the cNLS of SV40 large T antigen (cNLS) were expressed as 2ϫGFP chimeras in wild type (WT) W303 haploid cells (A). Cells were observed using differential interference contrast (right column), and GFP was detected by fluorescent microscopy (left column). In each case, GFP was localized to the nucleus. The NLS-GFP chimeras were also expressed in kap104 -16 ts cells (B). kap104 -16 cells were maintained at 23°C or grown at 23°C and shifted to 37°C for 4.5 h prior to examination of the GFP chimeras by fluorescent microscopy. Note the cytoplasmic localization of the rg-NLSs after the temperature shift. FIG. 3. NES-rg-NLS chimeras are exported from nuclei in vivo. rg-NLS regions of Nab2p and Nab4p were cloned upstream of a mutant (mutant nes) or a functional nuclear export signal (wild type NES) followed by two GFP tags. GFP chimeras were visualized in wild type cells grown at 30°C. the binding Kap104p to the rg-NLS of newly synthesized Nab2p and Nab4p/Hrp1p in the cytoplasm may have a role in preventing nonproductive cytoplasmic Nab-RNA interactions.
In addition, if Nab2p and Nab4p/Hrp1p exit the nucleus with newly synthesized mRNA, the observations reported above may suggest a role for Kap104p in promoting cytoplasmic disassembly of the mRNP. In support of this model, Nab4p/Hrp1p, like its closest human homologue, hnRNP A1, has recently been shown to exit the nucleus, presumably with newly synthesized mRNA (42,52). We therefore investigated whether Nab2p also exits the nucleus. The distribution of a functional Nab2p-GFP chimera 2 was monitored in kap104 ts cells under conditions where new protein synthesis was inhibited. As expected, and as observed previously for endogenous Nab2p (18), the fusion was nuclear at the permissive temperature and, upon shift to the nonpermissive temperature, accumulated in the cytoplasm (Fig. 5). Protein synthesis was inhibited by the addition of cycloheximide to the culture, and upon temperature shift, Nab2p redistribution was still observed, indicating that the Nab2p-GFP was exported from the nucleus to the cytoplasm. Nab2p-GFP did not change its distribution in wild type cells under the conditions tested, and cycloheximide treatment alone did not affect the localization of Nab2p-GFP. Importantly, Kap104p is rapidly turned over at the restrictive temperature in kap104 -16 cells, and mRNA export is ongoing (18). This suggests that Kap104p is not required for export of mRNA and Nab2p from the nucleus but is specific for the import leg of the Nab2p (and Nab4p/Hrp1p) cycle.
The Role of Ran-GTP on Kap104p-mediated Transport-Ran-GTP is a believed to be a key regulator of nuclear transport (20,21). Its presence in the nucleus is thought to disrupt import-bound substrate/kap complexes while stabilizing export-bound transporter/substrate complexes. As described in the Introduction, this differential effect of Ran-GTP and GDP provides directionality to import and export. To investigate the role Ran-GTP and Ran-GDP in Kap104p-mediated transport, we first tested whether Ran-GTP or -GDP bound to Kap104p. Kap104p-GST was immobilized on GT-Sepharose, and Ran-GTP, Ran-GDP, or buffer alone was added. As shown in Fig. 6, Ran-GTP bound preferentially to immobilized Kap104p, consistent with its role in nuclear import.
The binding of Ran-GTP to other ␤-kaps, including Kap95p (13), Kap121p (53, 54), Kap123p (54,55), and kap ␤2/Trn (56,57), has been shown to disrupt their interaction with import substrates. To investigate whether Ran-GTP had a similar disruptive effect on the Kap104p-Nab complexes, Nab2p-and Nab4p/Hrp1p-Kap104p complexes were formed on GT-Sepharose and challenged with Ran-GTP, -GDP, or -GTP␥S (Fig. 7A). Addition of Ran-GTP to this complex stimulated a moderate release of Kap104p from both Nab2p and Nab4p/Hrp1p, as shown by a shift of some of the bound Kap104p to the unbound fraction. Ran-GDP did not, however, stimulate Kap104p release over what was observed with buffer alone. Quantitation of these results revealed that Kap104p release was stimulated 3-fold by the presence of Ran-GTP, demonstrating that Ran must be in its GTP-bound form to disrupt the Kap104p-substrate complexes. Furthermore, GTP hydrolysis is not required for Kap104p release from its substrate, as Ran-GTP␥S was also able to dissociate Kap104p from Nab2p and Nab4p/Hrp1p with similar efficiency as Ran-GTP. These results are similar to those found by Rexach and Blobel (13) for the disruption of the Kap95p/Kap60p/cNLS, by Ran-GTP. However, in our hands, the Ran-GTP stimulated release of the Kap104p-Nab complex 2 C. Guthrie, personal communication. is far less complete than that observed for Kap95p/Kap60p (data not shown).
Because Kap104p stimulates release of Nabs from singlestranded DNA (Fig. 4), we investigated whether the Ran-GTPdependent disruption of the Kap104p-Nab complex was stimulated by the addition of RNA (Fig. 7B). In this case, we bound Kap104p to GT-Sepharose and formed a complex with Nab4p/ Hrp1p. Under these conditions, addition of Ran-GTP or RNA alone stimulated a weak release of Nab4p/Hrp1p from the column. Strikingly, Ran-GTP and RNA had a cooperative effect in releasing the substrate. Quantitation of the release under these conditions revealed that RNA and Ran-GTP alone released approximately similar amounts of Nab4p/Hrp1p, but together, their effect on substrate release was cooperative, stimulating substrate release greater than 7-fold. DISCUSSION Here, we show that Nab2p and Nab4p/Hrp1p both interact directly and independently, through a similar RG-rich domain, with the karyopherin Kap104p. Because this domain is both necessary and sufficient for Kap104p binding and acts as a KAP104-dependent nuclear import signal in yeast, we term this domain an rg-NLS to distinguish it from cNLSs recognized by Kap60p/Kap95p (kap ␣ and ␤). The Kap104p-dependent pathway is considered to be homologous to the mammalian Kap ␤2/Trn pathway (14,15,48). Kap104p shares the most sequence similarity to Kap ␤2/Trn; likewise, Nab4p/Hrp1p is the most similar to mammalian hnRNP A1, the prototype import substrate of Kap ␤2/Trn. Both the rg-NLS defined here (see also Refs. 48 and 49) and the Kap ␤2/Trn-interacting (M9) domain of hnRNP A1 are rich in glycines. Recently, however, a consensus Kap ␤2/Trn-interacting motif has been defined, and this motif shows no similarity to the rg-NLSs. This suggests that the NLSs recognized by Kap ␤2/Trn and Kap104p are different and offers an explanation for why Nab2p cannot be imported by Kap ␤2/Trn in vitro and for our observations that overexpression of mammalian Kap ␤2/Trn does not complement kap104 cells. 3 The rg-NLS is similar to the NLS in Npl3p (58), which is recognized by the yeast kap ␤ family member Mtr10p (58,59). Because the deletion of either MTR10 (58,59) or KAP104 (18) is not lethal, it seems reasonable that in the absence of one, the other may substitute for the loss. In support of this, these two transport pathways genetically interact, as cells carrying a deletion of both MTR10 and KAP104 are not viable (data not shown). It is also likely that both Mtr10p and Kap104p import other proteins in addition to the three substrates already defined. Of course, several other nuclear proteins in the yeast proteome contain RG-rich domains, thereby making them candidate substrates; however, whether these are functional NLSs in the context of these proteins remains to be shown.
Newly transcribed mRNA is bound by proteins in the nucleus, and the resulting heteroribonuclear protein particles are processed and transported out of the nucleus. Studies in metazoan cells demonstrate that whereas some hnRNP proteins are retained in the nucleus, others, such as hnRNP A1, accompany the mRNA into the cytoplasm (3, 37, 38, 60 -62). Because of their ability to exit and reenter the nucleus, these proteins are collectively called shuttling proteins. Previously, we showed that Nab2p and Nab4p/Hrp1p are imported into the nucleus by Kap104p and proposed that both Nab2p and Nab4p/Hrp1p exit the nucleus with RNA (18). This has been shown to be true for Nab4p/Hrp1p (42,52), and here we demonstrate that Nab2p also exits the nucleus, presumably bound to mRNA. We propose that in the cytoplasm, Kap104p binds the rg-NLS, and this destabilizes the Nab interaction with mRNA. This idea is supported by our in vitro binding studies. Addition of Kap104p to Nab2p and Nab4p/Hrp1p bound to ssDNA causes their release from the column. The rg-NLS of Nab2p overlaps with the RGG box. Furthermore, our data suggest the RG-rich carboxyl terminus of Nab4p/Hrp1p contributes to RNA binding. Thus, because both rg-NLSs in Nab2p and Nab4p/Hrp1p reside within these RNA-binding domains, binding Kap104p to this region could lead to displacement of the RNA thereby contributing to mRNP disassembly. However, other factors may also mediate the release of the Nabs from the mRNA in the cytoplasm in vivo. In the cytoplasmic environment of a low Ran-GTP concentration, the overlap of the rg-NLS with the RNA binding 3 J. Aitchison, N. Bonifaci, and G. Blobel, unpublished observations. FIG. 7. Ran-GTP and RNA act cooperatively to remove Nabs from Kap104p. A, recombinant Kap104p (6 ng/l) was incubated with immobilized Nab2p-GST (ϳ1.5 g/10 l of packed resin) (left panel) and Nab4p/Hrp1p-GST (ϳ2 g/10 l of packed resin) (right panel) for 1 h at 4°C. Complexes were incubated with Ran-GTP␥S, Ran-GTP, Ran-GDP, or buffer, as indicated, for 30 min at room temperature. Polypeptides that remained bound to the column and those that were released were analyzed by SDS-PAGE and Coomassie Blue staining. Kap104p release was quantitated using Image-Quant software, revealing that Ran-GTP␥S and Ran-GTP caused a 3-fold increase in the displacement of Kap104p from its substrate over that observed with Ran-GDP or buffer alone. B, immobilized Kap104p-GST (ϳ1 g/10 l of packed resin) was incubated with cleaved Nab4p (6 ng/l). Complexes were then incubated with 1) Ran-GTP, 2) Ran-GTP ϩ total yeast RNA, or 3) total yeast RNA. Released fractions and bound fractions (after cleavage from the resin with thrombin) were analyzed by SDS-PAGE and Coomassie Blue staining. The results of quantitation of Nab2p/Hrp1p in the released fractions in three separate experiments are represented graphically below the gel. Ran-GTPmediated release was normalized to 1.0. activity may drive the reaction to a new round of import, by either direct displacement of the RNA or simply by preventing the rebinding of the Nab to mRNA.
Our results also suggest that upon import of the Nab-Kap104p complexes, a combination of Ran-GTP and newly transcribed RNA act cooperatively to dissociate Nab2p and Nab4p/Hrp1p from Kap104p. This is similar to the release of Npl3p from Mtr10p (58). Imposing a role for RNA, in combination with nuclear Ran-GTP, in the release of Nab2p and Nab4p/ Hrp1p from Kap104p may (as proposed for Npl3p (58)) allow the local unloading of the cargo at the site of transcription. Such a mechanism would prevent the premature release of mRNA-binding proteins until they are required to bind RNA. Similarly, several cNLSs appear to overlap with DNA-and RNA-binding domains (63). Thus, this may be a general mechanism invoked to facilitate intranuclear protein sorting.
Although Nab2p and Nab4p/Hrp1p are required for efficient mRNA export, and they interact directly with both RNA and Kap104p, our results strongly support a model in which Kap104p is specific for the import of the Nabs, and not in mRNA export. Similarly, although hnRNP A1 is required for efficient mRNA export in metazoan cells, kap ␤2/Trn is not likely involved in this process (39,56,64) but is specific for the import of (among other proteins (56)) hnRNP A1 (16,17). In both cases, cells have evolved a system of maintaining separate directionality to import and export. Ran-GTP in the nucleus destabilizes import-substrate/kap complexes; however, as demonstrated here, this may not be sufficient for completely unloading the cargo. Masking of the NLS by competition with overlapping RNA or DNA binidng sites, or by assembly into a multicomponent complex, may also contribute to local release. Interestingly, the M9 domain of hnRNP A1 is not accessible to kap ␤2/Trn in the context of the nuclear hnRNP (56). Because the M9 domain does not bind RNA (40), this masking is likely due to other proteins in the nuclear hnRNP. It would be interesting to determine whether proteins in yeast nuclear hnRNPs also mask NLSs and whether they, as in mammalian cells, differ from cytoplasmic hnRNPs. Thus, upon export, some hnRNP proteins may be shed, allowing access of the NLSs to karyopherins in the cytoplasm and lead to another round of import.