GTP Hydrolysis Links Initiation and Termination of Nuclear Import on the Nucleoporin Nup358*

Binding of GTP-bound Ran (RanGTP) to karyopherin β1 (Kapβ1) releases import cargo into the nucleus. Using an ultrastructural, biochemical, and functional approach, we have studied the mechanism by which Kapβ1·RanGTP is recycled at the nuclear pore complex for repeated rounds of import. In vitro, Kapβ1 bound to the RanBP1-homologous (RBH) domains of Nup358 in the presence of either RanGTP or RanGDP, forming trimeric complexes. The Kapβ1·RanGTP·RBH complex resisted dissociation by RanBP1 and GTP hydrolysis by Ran GTPase activating protein 1. Ran-dependent binding of gold-conjugated Kapβ1 to the cytoplasmic fibers of the nuclear pore complex in digitonin-permeabilized cells and RanBP1 competition confirmed thein vitro binding data. Interaction of karyopherin α and a classical nuclear localization sequence peptide with the Kapβ1·RanGTP·RBH complex stimulated GTP hydrolysis by Ran GTPase activating protein 1 both in vitro and in permeabilized cells. This GTP hydrolysis was required for reinitiation of import of a nuclear localization sequence-bearing substrate in permeabilized cells. These data suggest that GTP hydrolysis on the RBH domains of Nup358 couples the termination of one cycle of nuclear import with the initiation of the next.

Nuclear import of proteins bearing a classical nuclear localization sequence (NLS) 1 has been recently reviewed in Refs. 1-3. In digitonin-permeabilized cells, nuclear import of NLSbearing proteins is mediated by a karyopherin ␣⅐karyopherin ␤1 (Kap␣⅐Kap␤1) heterodimer. Kap␣ and Kap␤1 are also called importins ␣ and ␤, respectively. Kap␣ binds to the NLS, whereas Kap␤1 interacts with the nuclear pore complex (NPC), resulting in docking of the cargo at the NPC. Other soluble factors including the small GTPase Ran, p10 (also called NTF2), and the small Ran-binding protein RanBP1 help to translocate the cargo-carrier complex through the NPC and release the cargo inside the nucleus in the presence of free GTP. The NPC is a large complex of proteins (nucleoporins) that spans the nuclear membrane and has filamentous projections into both the nucleus and cytoplasm (3)(4)(5)(6). Although several nucleoporins interact with Kap␤1 in vitro, the exact sequence of interactions between Kap␤1, nucleoporins, and other soluble factors that result in translocation from the cytoplasmic to the nuclear side of the NPC and back is not well understood.
Ran binds to guanine nucleotides and cycles between a GDPbound form (RanGDP) and a GTP-bound form (RanGTP). Nuclear transport is thought to require interconversion between these two forms (7)(8)(9). The interconversion of RanGTP and RanGDP is regulated by two factors. Ran GTPase activating protein 1 (RanGAP1), located predominantly in the cytosol and the cytoplasmic surface of the NPC, activates Ran GTPase activity and thus converts RanGTP to RanGDP. RanGAP1 is targeted to the NPC by covalent modification with a small ubiquitin-like molecule (SUMO-1) (10,11). SUMO-RanGAP1 is attached to the C terminus of Nup358 and is the only species of RanGAP1 retained in digitonin-permeabilized cells (10,12,13). RCC1, a predominantly intranuclear guanine nucleotide exchange factor for Ran, allows the release of bound nucleotide and reloading of Ran (7, 8, 14 -16). The cellular concentration of GTP is higher than that of GDP; therefore, RCC1 is thought to favor the formation of RanGTP. The cellular distribution of these two factors suggests that nuclear Ran is predominantly in the GTP-bound form, whereas cytoplasmic Ran is mostly GDP-bound.
RanGTP binds to Kap␤1 leading to the release of Kap␣ along with the bound cargo into the nucleus (17)(18)(19). The Kap␤1⅐RanGTP complex then travels back through the NPC (20) presumably to initiate another cycle of import. Binding of Kap␤1 to RanGTP inhibits RanGAP1-mediated GTP hydrolysis (21). Yet nuclear import is favored when Ran is provided in its GDP bound form and is inhibited by nonhydrolyzable analogs of GTP and mutants of Ran that do not hydrolyze GTP (19,(22)(23)(24)(25)(26)(27)(28). The site and mechanism by which GTP hydrolysis occurs after formation of the Kap␤1⅐RanGTP complex are not well understood. RanBP1 binds strongly to RanGTP (29 -31) and weakly to RanGDP (32,33) and forms trimeric complexes with Kap␤1 and either RanGDP or RanGTP (34,35). In vitro studies with yeast proteins have shown that RanBP1 in concert with Kap␣ and a C-terminal portion of the nucleoporin Nup1p can reverse the inhibition of RanGAP activity imposed by binding of RanGTP to Kap␤1 (36). Similar results have been reported with mammalian factors (37). However, it is not clear whether and where these reactions occur in cells. Recent studies provide evidence that GTP hydrolysis is not required for translocation of cargo through the NPC (38 -40). This leaves open the possibility that GTP hydrolysis is required for recycling of the Kap␤1⅐RanGTP complex for repeated rounds of import.
Nup358 (also termed RanBP2) is a 358-kDa nucleoporin located at the tips of the cytoplasmic fibrils of the NPC (41,42). Its domains include a leucine-rich region, four RanBP1-homologous (RBH) domains, a zinc finger region, scattered FG and FXFG repeats characteristic of a subset of nucleoporins, and a cyclophilin-homologous domain (41,42). RanGTP binds to Nup358 in overlay and microtiter plate assays (25,41,42). Electron microscopy of rat liver nuclear envelope preparations shows that Ran loaded with GTP or its nonhydrolyzable analog GMPPNP interacts with the cytoplasmic side of the NPC (25,41). These findings are consistent with an interaction between RanGTP and Nup358 at the NPC. In addition, Nup358 appears to form a trimeric complex with RanGTP and Kap␤1 in vitro (43). The binding properties of Nup358 are reminiscent of those of RanBP1, suggesting that Nup358 may be involved in GTP hydrolysis at the NPC. On the other hand, the location of Nup358 at the tips of the cytoplasmic fibrils of the NPC suggests that it may be involved in the initial docking of substrate prior to import.
Many of the recent advances in characterizing nuclear import factors were accomplished by using fluorescent import substrates in a digitonin-permeabilized cell assay (44). One of the major tasks that currently face the field is dissecting the events that occur at the NPC during import (3), a challenge that cannot be met at the resolution level of this assay. Ultrastructural studies of import to date (19,(45)(46)(47)(48)(49)(50)(51)(52) have focused mainly on microinjection experiments in intact cells, in which identifying the functions of individual factors is difficult.
In this study, we bring ultrastructural analysis and Ran GTPase activity measurements to the digitonin-permeabilized cell system. This approach is combined with in vitro binding and GTPase assays, as well as standard import assays, to examine where and how the Kap␤1⅐RanGTP complex is recycled at the NPC. We show that Nup358 is a site at which the first (docking of Kap␣⅐NLS) and the last (recycling of Kap␤1⅐RanGTP) steps of nuclear import are linked and that the resulting GTP hydrolysis by Ran is required for subsequent import. Using ultrastructural and in vitro solution binding assays, we show that the RBH domains of Nup358 serve as RanGDP-or RanGTP-dependent docking sites for Kap␤1 on the cytoplasmic fibers of the NPC. The Kap␤1⅐RanGTP⅐RBH complex, a likely intermediate in the recycling of Kap␤1⅐RanGTP, is not dissociated by RanBP1, suggesting that GTP hydrolysis by Ran is a prerequisite for the dissociation of Kap␤1 from Nup358 during transport. Kap␣ and NLS peptide bound to the Kap␤1⅐RanGTP⅐RBH complex resulting in GTP hydrolysis by free and NPC-associated RanGAP1. This GTP hydrolysis was necessary for the initiation of further cycles of nuclear import. These findings are integrated into a model for the termination of one cycle of import and the initiation of the next.

EXPERIMENTAL PROCEDURES
Cell Culture and Permeabilization-HeLa cells were cultured in Dulbecco's modified Eagle's medium supplemented by 10% fetal calf serum, 2 mM glutamine, and 100 units/ml each of penicillin and streptomycin. For electron microscopy and GTPase assays, 50 -80% confluent cells were digitonin-permeabilized and stored essentially as described (53). Cells were nonenzymatically dissociated in dissociation solution (Sigma C-5914) for 15 min at 37°C and washed three times by low speed centrifugation in cold transport buffer (TB) (20 mM HEPES, pH 7.4, 110 mM potassium acetate, 2 mM magnesium acetate, 1 mM EGTA). They were then suspended in TB with 35 g/ml digitonin and incubated for 5 min on ice followed by three washes in cold TB. They were resuspended in TB with 10 mg/ml bovine serum albumin (Sigma A-7030) and 5% Me 2 SO at 10 7 cells/ml and stored in aliquots at Ϫ80°C.
Recombinant Proteins-Human Ran was expressed in Escherichia coli BL21(DE3) from a pET-9C construct (29) and purified as described (21). Recombinant human Ran was loaded with either GDP, GTP, or GMPPNP as described (21) prior to use for gold conjugation, binding assays, or import assays. The murine RanGAP1 open reading frame was amplified by polymerase chain reaction from a construct kindly provided by Mark Rush (54) and subcloned into pGEX 4T3 (Amersham Pharmacia Biotech), from which it was expressed in E. coli as a GST fusion protein, purified by binding to glutathione-Sepharose, and eluted by thrombin cleavage according to the manufacturer's instructions. The Kap␤1 open reading frame was amplified from a human cDNA library (CLONTECH), subcloned into pGEX4T3, expressed in E. coli as a GST fusion protein, and purified by binding to glutathione-Sepharose. For electron microscopic and binding studies, Kap␤1 was eluted with thrombin; for GTPase assays, it was eluted with glutathione according to the manufacturer's instructions as a GST fusion protein. Murine Kap␣2 (55), expressed with a C-terminal His tag from a pET21A construct, was a kind gift from Michael Matunis. The Nup358-4 protein (12) was a kind gift from Jian Wu. The Nup358-1 construct was produced by amplifying the region of the Nup358 cDNA that encodes amino acids 996 -1963 from a plasmid containing the 7-4 fragment of Nup358 (41) provided by Jian Wu. It was subcloned into pGEX4T3 and expressed as a GST fusion protein in E. coli. Recombinant RanBP1, expressed from a pET11 construct (kindly provided by Elias Coutavas), was a gift from Yuh-Min Chook and Tschong-Uk Park.
Conjugation of Recombinant Proteins to Colloidal Gold Particles-The optimum amount of recombinant protein needed to stabilize colloidal gold beads (BBI International) was determined by titration according to the manufacturer's instructions. Colloidal gold beads, 10 nm in diameter, were incubated with the recombinant protein for 5 min at room temperature and brought up to pH 9 with 0.2 M potassium carbonate. Bovine serum albumin was added to 10 mg/ml and the coated beads were then concentrated 50-fold by centrifugation according to the manufacturer's instructions and stored at 4°C for up to 1 month. Gold conjugates were checked by electron microscopy to ensure that the preparation is monodisperse with no clumping of gold particles.
Electron Microscopy-Approximately 5 ϫ 10 4 digitonin-permeabilized HeLa cells were incubated with 1 l of 10-nm gold conjugate for 40 min at 22°C in 30 l of TB followed by overnight fixation in 5% glutaraldehyde at 4°C. Samples were processed for transmission electron microscopy and viewed with a Jeol 100CX microscope (Jeol USA, Peabody, MA). Two photographs per cell were taken from ultrathin sections at ϫ 26,000 magnification from 5-6 cells for each experimental point. For quantitation of Kap␤1 binding to the NPC, the gold beads associated with all the morphologically preserved NPCs visible in the photograph (approximately 40 NPCs per sample) were counted and divided by the number of NPCs. All electron micrographs were scanned and saved using Adobe Photoshop 4.0 software. Quantitative analysis of the images was carried out with NIH Image 1.61 software.
In Vitro Binding Assays-Binding reactions contained 10 l of glutathione-Sepharose beads (Amersham Pharmacia Biotech) in a total volume of 46 l in transport buffer with Tween 20 (20 mM HEPES, pH 7.4, 110 mM potassium acetate, 2 mM magnesium chloride, and 0.1% Tween 20) and 0.5 mg/ml Pefabloc SC (Roche Molecular Biochemicals). Nup358-1, Nup358-4, and Kap␤1 were present at 0.25 M each; Ran was at 0.8 M. Reactions were allowed to proceed for 1 h at 4°C with rotation. After pelleting the beads, the supernatant was collected, and the beads were washed twice in cold TB-0.1% Tween 20. For the experiment shown in Fig. 4, a further 1 h incubation was carried out in the presence or absence of RanBP1 at 4°C as indicated. The bound and unbound fractions were subjected to SDS-PAGE on 5-20% gradient gels. Proteins were visualized by Coomassie Blue staining of gels, Amido Black staining of blots, or Western blotting as indicated in the text and figure legends. In Western blots, Kap␣, by virtue of its Cterminal His tag, was detected using a nickel-nitrilotriacetic acid-horseradish peroxidase (Ni-NTA horseradish peroxidase) conjugate (Qiagen) according to the manufacturer's instructions (see Fig. 7). Detection was by chemiluminescence using SuperSignal reagents (Pierce). Quantitation of Kap␣ binding was performed by densitometric analysis of blots using Scan Analysis version 2.56 software (Biosoft).
GTPase Assays-Ran was loaded with [␥-32 P]GTP as described (21)  potassium acetate, 20 mM magnesium acetate, 1 mM dithiothreitol, and 0.5 mg/ml bovine serum albumin) (21). Hydrolysis was allowed to proceed for 30 min at 22°C and stopped by adding 300 l of hydrolysis buffer; half of the reaction was then passed through a nitrocellulose filter. The filter was washed with 5 ml of cold hydrolysis buffer, and the radioactivity retained on it was counted in a scintillation counter.
Import Assays-GST-Kap␤1 fusion protein was immobilized on glutathione-Sepharose beads, and an excess of either RanGTP or RanGMPPNP was added. After a 1-h incubation at 4°C, the beads were washed twice, and the Kap␤1⅐RanGTP and Kap␤1⅐RanGMPPNP complexes were eluted with glutathione according to the manufacturer's instructions. The buffer was changed to TB by centrifugation through preequilibrated MicroSpin columns (Amersham Pharmacia Biotech) prior to use in import reactions. HeLa cells were grown on glass coverslips for 24 h, permeabilized with 30 g/ml digitonin in TB for 5 min at room temperature, and washed three times with transport buffer. Import reactions included 0.1 M Kap␤1⅐RanGTP or Kap␤1⅐RanGMPPNP complexes, 0.6 M Kap␣, 5 g/ml tetramethylrhodamine B isothiocyanate-NLS-bovine serum albumin (kindly provided by Mary S. Moore) (38), 0.3 M p10, 10 nM RanGAP1, 1 M RanBP1, and 2 mM GTP or GMPPNP in TB. The mixtures, except for RanGAP1, were preincubated for 15 min on ice. RanGAP1 was added immediately before the mixtures were added to permeabilized cells. Import was allowed to proceed for 15 min at room temperature, and cells were washed twice in TB, fixed in 2% formaldehyde in TB, and visualized using a Zeiss Axiophot microscope using a ϫ 63 objective. Images were captured with Adobe Photoshop 4.0 software.

Karyopherin ␤1 Docks at the RBH Domains of Nup358 by Forming Trimeric Complexes with Either RanGDP or
RanGTP-Karyopherin ␤1 (Kap␤1) binds to several nucleoporins in vitro (17,(57)(58)(59)(60)(61)(62). Of these, the nucleoporin Nup358 is located at the tips of the cytoplasmic fibrils of the NPC, where the initial docking of import substrates is believed to occur prior to import (46,49). It contains a leucine-rich region, four domains that are homologous to RanBP1 (RBH domains), eight zinc fingers, a cyclophilin-homologous domain, and scattered FG repeats that are characteristic of a subset of nucleoporins (41,42) (Fig. 1). Trimeric complexes between RanBP1, Kap␤1, and either RanGTP or RanGDP have been demonstrated in solution binding assays (34,35). This would suggest that trimeric complexes could be formed between Kap␤1, RBH domains, and either RanGDP or RanGTP. However, in overlay and microtiter plate assays, binding of Kap␤1 to Nup358 was found to be enhanced by RanGTP but not RanGDP (43).
Ran-dependent binding of Kap␤1 to Nup358 would be of potential significance for both import of NLS-bearing cargo and recycling of Kap␤1 after a round of import. RanGDP is required for import of Kap␣ and NLS-bearing cargo (19,28) in addition to Kap␤1; therefore, complexes involving Kap␤1, RanGDP, and Nup358 may act as intermediates in nuclear import. On the other hand, after a round of import, Kap␤1 is bound to RanGTP and probably exported as a Kap␤1⅐RanGTP complex. This complex may bind to Nup358 as its last station on the NPC before recycling. With these considerations in mind, we carried out solution binding assays to determine whether Kap␤1 can bind to Nup358 in the presence of RanGDP or RanGTP, and which domains of Nup358 are involved in such binding.
For this purpose, two recombinant fragments of Nup358 were expressed in E. coli as fusion proteins with an N-terminal GST tag (Fig. 1). The first fragment, called Nup358-1, contains the RBH1 domain, the eight zinc fingers, and several FG repeats. The second fragment, called Nup358-4, contains the RBH4 domain, the cyclophilin-homologous domain, and several FG repeats (Fig. 1). The constructs were intended to include as much representative sequence from Nup358 as possible within the constraints imposed by the bacterial expression system. We wanted to express the RBH domains in their natural context within Nup358 sequences and to include most of the FG repeats, which have been implicated in the binding of Kap␤1 to other nucleoporins (17,57,58,60,63,64).
We first incubated immobilized Nup358-1 and Nup358-4 with Kap␤1 and either RanGDP or RanGTP. Bound and unbound material was analyzed by SDS-PAGE (Fig. 2). Very little Kap␤1 bound to either of the Nup358 fragments, despite the presence of FG repeats in both fragments; close inspection of the original gels reveals a faint band (see Fig. 7A, lane 2). However, in the presence of either RanGTP or RanGDP, there was strong binding of Kap␤1 to both Nup358-1 and Nup358-4 (Fig. 2). These data indicate that each of the two Nup358 fragments is capable of forming trimeric complexes with Kap␤1 and either RanGTP or RanGDP. When incubations were carried out in the presence of increasing concentrations of RanBP1, there was complete inhibition of Ran-mediated binding of karyopherin ␤1 to Nup358-1 and Nup358-4, indicating that karyopherin ␤1 and Ran form trimeric complexes with the RBH domains of Nup358 (Fig. 3).
The Preformed Kap␤1⅐RanGTP⅐RBH Complex Resists Dissociation by RanBP1-RanBP1 has been implicated in the recycling of Kap␤1⅐RanGTP complexes (36,37). However, these complexes may encounter and bind to Nup358 on their way out of the NPC before they encounter RanBP1, because the latter is a cytosolic protein (65). It was therefore important to determine whether RanBP1 would be able to dissociate a Kap␤1⅐Ran⅐RBH complex after it is formed. Kap␤1 and either RanGDP or RanGTP were incubated with Nup358-4 immobilized on beads in order to form Kap␤1⅐RanGDP⅐RBH and Kap␤1⅐RanGTP⅐ RBH complexes, respectively, as before. The beads were then washed and subjected to a second incubation with increasing concentrations of RanBP1 followed by SDS-PAGE of bound and unbound fractions (Fig. 4). The preformed Kap␤1⅐RanGTP⅐ RBH trimeric complex was resistant to dissociation by RanBP1. In contrast, the preformed trimeric complex of Kap␤1⅐RanGDP⅐ RBH was readily dissociated by RanBP1 (Fig. 4). Identical results were obtained with Nup358-1 (data not shown). Hence, once a trimeric RBH⅐Kap␤1⅐RanGTP complex is formed, RanBP1 is unable to disassemble the complex, whereas a trimeric complex of RBH⅐Kap␤1⅐RanGDP can be disassembled by RanBP1 (or presumably another RBH domain of Nup358). These results suggest that GTP hydrolysis is needed before Kap␤1 can move from one RBH to the next or be released by RanBP1.
Ultrastructural Analysis of Kap␤1 Docking at the NPC-To assess the significance of these in vitro binding data in the context of intact NPCs, Kap␤1 was coupled to 10-nm colloidal gold and incubated with digitonin-permeabilized cells, either alone or in the presence of RanGDP or RanGTP. Electron microscopic inspection of sections of these cells confirmed the in vitro binding data (Fig. 5). There was very little binding to NPCs of Kap␤1-gold alone (Fig. 5, top panel), whereas significant gold decoration of NPCs occurred when either RanGDP (Fig. 5, middle panel) or RanGTP (Fig. 5, bottom panel) was present during the incubation. Localization of the gold particles to the cytoplasmic fibrils of the NPC and competition by RanBP1 (see below) indicate that they are indeed decorating Nup358. Electron microscopic analysis also confirmed the in vitro competition data with concomitantly added RanBP1 (data not shown) and with RanBP1 added after docking (Fig. 6). Kap␤1gold was first allowed to dock in digitonin-permeabilized cells, either alone or with RanGDP or RanGTP. This was followed by a second incubation in the presence or absence of RanBP1, and the average number of Kap␤1-gold particles per NPC under each condition was then determined. There was considerable reduction of NPC-bound gold particles when a preformed Kap␤1-gold⅐RanGDP⅐Nup358 complex was exposed to RanBP1, whereas the preformed Kap␤1-gold⅐RanGTP⅐Nup358 complex remained stable (Fig. 6). Interestingly, Kap␤1 binding that occurred in the absence of exogenously added Ran was also diminished by RanBP1 treatment, suggesting that this binding is mediated in part by the formation of a trimeric complex with residual endogenous RanGDP and Nup358. In fact, residual Ran can be seen at the nuclear envelope in digitonin-permeabilized cells by immunofluorescence (66).
Thus, the binding of Kap␤1⅐RanGTP to Nup358 is stable both in vitro and in permeabilized cells, whereas Kap␤1 docked through RanGDP can be readily displaced. These results suggest that once a Kap␤1⅐RanGTP complex resulting from one round of import binds to Nup358, another round of NLS import may not be possible before GTP hydrolysis occurs. 37). However, with the high concentration of Nup358 RBH domains at the NPC (67) and the cytosolic location of RanBP1 (65), the Kap␤1⅐RanGTP complex is likely to encounter a Nup358 RBH domain and bind to it before it encounters RanBP1. Because the Kap␤1⅐RanGTP⅐Nup358 complex thus formed cannot be dissociated by RanBP1 (see above), GTP hydrolysis would have to occur on Nup358 at the cytoplasmic fibers of the NPC. It was therefore important to determine whether Kap␣ would interact with the Kap␤1⅐RanGTP⅐Nup358 complex and stimulate GTP hydrolysis in the presence of Ran-GAP1. To assess the relevance of these processes to the further import of NLS-bearing substrate, it was also of interest to see whether they are affected by the presence of NLS cargo.

Induction of GTP Hydrolysis by Kap␣ and NLS-The
As shown in Fig. 7A, when Kap␣ and NLS peptide were added to immobilized Nup358-4 alone, they did not bind to a significant extent (Fig. 7A, lane 1). However, binding occurred when both Kap␤1 and Ran were present, indicating the formation of NLS⅐Kap␣⅐Kap␤1⅐RanGTP⅐Nup358 complexes (Fig. 7A,  lanes 2-4). Surprisingly, RanGTP caused even stronger binding of Kap␣ than RanGDP (Fig. 7A, lanes 3-4). GTP remained unhydrolyzed throughout the assay (data not shown). Similar results were seen when Nup358-1 was used instead of Nup358-4 (data not shown). Binding of Kap␣ was diminished in the absence of NLS peptide (Fig. 7B), indicating cooperative binding of Kap␣ and NLS to the rest of the complex. These results were quantitated by scanning densitometry (Fig. 7C).
Ultrastructural studies (46,49) have suggested that the initial docking site in nuclear import is at the tips of the cytoplasmic fibers of the NPC. Because this is where Nup358 is localized as well (41), the NLS⅐Kap␣⅐Kap␤1⅐RanGTP⅐Nup358 complexes may represent initial docking events in classical NLS-mediated nuclear import.
To test whether GTP hydrolysis could be induced within the trimeric and higher order RanGTP-containing complexes described above, we preincubated Kap␤1 and Ran[␥-32 P]GTP at 4°C to allow binary complex formation and then incubated the Kap␤1⅐RanGTP complex at 22°C with RanGAP1 and various combinations of Kap␣, NLS peptide, and Nup358-1. The reactions were stopped and passed through a nitrocellulose filter, and the amount of unhydrolyzed Ran-bound GTP was measured by counting radioactivity that remained bound to the filter after washing (21). The Nup358-1⅐RanGTP⅐Kap␤1 trimeric complexes were resistant to RanGAP1 activity (Fig. 8A). Addition of Kap␣ resulted in partial GTP hydrolysis, and when NLS peptide was included in the reaction, complete hydrolysis occurred (Fig. 8A). This is consistent with the cooperative binding of Kap␣ and NLS peptide in the formation of pentameric complexes (see above). The results were identical when Kap␣ with or without NLS peptide was added to a preformed Kap␤1⅐RanGTP⅐Nup358 complex (data not shown). In addition, identical results were obtained when RanBP1 was substituted for Nup358-1 (data not shown). Thus, the presence of transport cargo favors the hydrolysis of GTP, consistent with a role for GTP hydrolysis by Ran during nuclear import.
To study GTP hydrolysis by Ran within intact NPCs in a context closer to the in vivo situation, we used digitonin-permeabilized HeLa cells (Fig. 8B). RanGAP1 exists in cells in two forms, a free cytosolic form and an NPC-bound form that is covalently modified with a small ubiquitin-like peptide (SUMO-1) (10,11). The latter form (SUMO-RanGAP1) is attached to the C terminus of Nup358 and is the only species retained in digitonin-permeabilized cells (10,13). Assays were carried out in a manner similar to the in vitro assays with preincubation of [␥-32 P]RanGTP with or without Kap␤1. GTP hydrolysis assays were then started by adding digitonin-permeabilized cells with or without Kap␣ and NLS peptide (Fig.  8B). The digitonin-permeabilized cells served as a source of both Nup358 and NPC-associated SUMO-RanGAP1. When RanGTP was added to digitonin-permeabilized cells in the absence of Kap␤1, GTP was readily hydrolyzed (Fig. 8B). When RanGTP was preincubated with Kap␤1, there was maximal inhibition of GAP activity. Thus SUMO modification and NPC association of RanGAP1 do not override the inhibitory effect of Kap␤1 on GTP hydrolysis. As in the in vitro experiment, addition of Kap␣ stimulated partial GTP hydrolysis. The addition of NLS did not further stimulate GTP hydrolysis by NPC-associated SUMO-RanGAP1 (Fig. 8B). The presence of residual unhydrolyzed GTP under these conditions suggests that some Kap␤1⅐RanGTP associated with the four RBH domains of Nup358 might not be accessible to SUMO-RanGAP1, which is attached to the C terminus of Nup358. Indeed, this residual GTP was completely hydrolyzed in a Kap␣ and NLS-dependent fashion when recombinant free RanGAP1 was included in addition to the permeabilized cells (data not shown). Thus, the two forms of RanGAP-soluble and NPC-associated-may play complementary roles in GTP hydrolysis at the NPC.
GTP Hydrolysis Is Required for Recycling of the Kap␤1⅐ RanGTP Complex for Further Rounds of Import-Several previous studies have shown that import of classical NLS bearing substrates requires GTP hydrolysis by Ran (22-28). On the other hand, a recent study indicated that GTP hydrolysis might not be required for the translocation step of the nuclear import of classical NLS-bearing proteins (38). In that study, Ran was supplied in its GDP-bound form in the import reactions. However, as discussed previously, Ran is likely to be in the GTP bound form and in complex with Kap␤1 at the end of a round of import; and GTP hydrolysis may then be required for a recycling step before this Kap␤1⅐RanGTP complex can be utilized for another round of import. To address this question, RanGTP or Ran loaded with the nonhydrolyzable GTP analog GMPPNP was added in excess to GST-Kap␤1 immobilized on glutathione-Sepharose beads. After washing to remove unbound Ran, the Kap␤1⅐RanGTP and Kap␤1⅐RanGMPPNP complexes thus formed were eluted from the beads with glutathione (Fig. 9A) and used in standard import reactions in permeabilized cells in the presence of free GTP and GMPPNP, respectively (Fig. 9B). As shown in Fig. 9B, the Kap␤1⅐RanGTP complex supported import, whereas the Kap␤1⅐RanGMPPNP complex did not. These data show that GTP hydrolysis is required for recycling of a Kap␤1⅐RanGTP complex and reinitiation of nuclear import of a classical NLS-bearing substrate. DISCUSSION Nuclear import of classical NLS-bearing proteins is mediated by a Kap␣⅐␤1 heterodimer. Other soluble factors involved in the process include Ran, p10, RanBP1, RanGAP1, and GTP. Binding of RanGTP to Kap␤1 dissociates Kap␣ from Kap␤1, resulting in the release of Kap␣ and its NLS-bearing cargo into the nucleus. In this study, we analyze how and where the Kap␤1⅐RanGTP complex is recycled at the NPC to initiate another round of import. Data from in vitro binding and GTPase activity assays are correlated with findings from ultrastructural and functional assays in permeabilized cells to provide a clearer picture of the events that actually occur on intact NPCs. With this approach, we show that the RBH domains are a site where docking of NLS cargo is linked to GTP hydrolysis by Ran and that this hydrolysis is a prerequisite for repeated cycles of import.
After translocation through the NPC, RanGTP binds to Kap␤1, leading to the release of the Kap␣⅐NLS complex in the nucleus. The mechanism of reexport of Kap␤1 after a round of nuclear import is not well understood. A leucine-rich nuclear export sequence in yeast Kap␤1 has been reported to mediate its export to the cytoplasm and its interaction with GLFGcontaining yeast nucleoporins (63). It is not known whether other soluble export factors are involved in Kap␤1 export, but it is probably reexported in a complex with RanGTP (20). The Kap␤1⅐RanGTP complex is then presumably recycled to initiate further rounds of import. In vitro, binding of RanGTP to Kap␤1 releases it from some nucleoporins (17,43,61), but it enhances its binding to Nup358 (Fig. 2) (43). Nup358 is the outermost known nucleoporin on the cytoplasmic side of the NPC and is likely to be the first station for nuclear import of NLS-bearing proteins as well as the last station for the reexported Kap␤1⅐RanGTP complex on the NPC.
Our data (Figs. 4 and 5) show that either RanGDP or RanGTP can target Kap␤1 to Nup358 by forming trimeric complexes and that the bulk of Kap␤1 docking on Nup358 is Ran-dependent. Furthermore, although RanGDP can bind either to the zinc fingers or the RBH domains of Nup358 (67), Ran-dependent docking of Kap␤1 occurs through the RBH sites only, as evidenced by the fact that this binding can be prevented by concomitantly added excess RanBP1 (Fig. 3). Ultrastructural studies with Kap␤1-gold conjugates in digitonin permeabilized cells confirm RanGDP-and RanGTP-dependent docking of Kap␤1 at the cytoplasmic fibrils (Fig. 5) and competition by RanBP1 (Fig. 6). It should be noted that, although RanBP1 inhibits Kap␤1 binding to Nup358, it enhances docking and import of NLS-bearing proteins (34). The mechanism of this enhancement is not clear; it is possible that RanBP1 stabilizes NLS docking at NPC sites other than the RBH domains of Nup358. FIG. 8. GTP hydrolysis catalyzed by free RanGAP1 and NPC-associated SUMO-RanGAP1 is triggered by Kap␣ and NLS. A, maximal stimulation of GTP hydrolysis in vitro by free Ran-GAP1 within trimeric complexes requires both Kap␣ and NLS peptide. ␥-32 P-Labeled RanGTP was preincubated with or without Kap␤1 prior to adding the other indicated components. The amount of radioactivity retained on a nitrocellulose filter was expressed as a percentage of control (radioactivity retained in the Nup358-RanGTP-Kap␤1 complex was set at 100%). B, NPC-associated SUMO-Ran-GAP1 catalyzes partial GTP hydrolysis. ␥-32 P-Labeled RanGTP was preincubated with or without Kap␤1 and GTPase assays were carried out in the presence of 5 ϫ 10 4 digitonin-permeabilized HeLa cells per reaction; other components were added as indicated. The amount of radioactivity retained on the filter in the reaction containing RanGTP, cells, and Kap␤1 was set at 100%.
It is of interest in this context to compare our data to those of Delphin et al. (43). Their overlay and microtiter plate assays (in agreement with our solution binding assays) show that RanGTP enhances the binding of Kap␤1 to Nup358. However, when they added recombinant Kap␤1 to isolated nuclear envelopes and detected it by immunoelectron microscopy, the substantial amount of Kap␤1 docking on the cytoplasmic side of the NPC was reduced by RanGTP instead of the increase one might have expected. This is probably due to the presence of other nucleoporins on the cytoplasmic side of the NPC and the fact that RanGTP dissociates Kap␤1 from some nucleoporins (see above). Our data, on the other hand, were obtained by directly coupling recombinant Kap␤1 to colloidal gold and applying it to digitonin-permeabilized cells with or without Ran (Fig. 5). We saw little docking in the absence of Ran, but docking was markedly enhanced in the presence of either RanGDP or RanGTP (Figs. 5 and 6). It thus appears that gold conjugation of Kap␤1 has led to a reduction in docking on other nucleoporins while leaving Ran-dependent docking on RBH domains relatively intact. This may be due to selective masking of certain domains on Kap␤1 or to steric interference by the bulk of the gold particle. In any case, this has allowed us to examine docking on Nup358 in relative isolation from the background of other nucleoporins. We have thus been able to confirm the in vitro binding and competition data and to show that Ran-dependent docking of Kap␤1 occurs on intact NPCs.
As shown diagrammatically in Fig. 10, formation of a Kap␤1⅐RanGTP⅐Nup358 complex is likely to be the last step in the reexport of the Kap␤1⅐RanGTP complex through the NPC. An important feature of the Kap␤1⅐RanGTP⅐Nup358 complex is that once it is formed, it resists dissociation by subsequently added RanBP1 both in vitro and on intact NPCs, indicating that it is relatively stable (Figs. 4 and 6). The RanGDP-containing complex, on the other hand, is readily dissociated by subsequently added RanBP1. These data imply that GTP hydrolysis by Ran is needed in order to allow movement of Kap␤1.
Docking of Kap␣⅐NLS, GTP Hydrolysis, and Import-The stability of the Kap␤1⅐RanGTP⅐Nup358 complex in comparison to the Kap␤1⅐RanGDP⅐Nup358 complex (see above) suggested that GTP hydrolysis is required before translocation and import can occur. Indeed, nonhydrolyzable analogs of GTP and mutants of Ran that do not hydrolyze GTP have been reported to inhibit nuclear import (22)(23)(24)(25)(26)(27)(28). On the other hand, a recent study showed that the nonhydrolyzable GTP analog GMPPNP can support classical NLS-mediated import in the presence of RanGDP (38). However, a round of nuclear import results in a Kap␤1⅐RanGTP complex, and it is this complex that needs to be recycled and used in a subsequent round of import. A hydrolysis requirement for recycling would be bypassed if RanGDP were used as a starting point. Indeed, as shown in Fig. 9, when cells were presented with a Kap␤1⅐RanGMPPNP complex, import failed to occur, whereas the Kap␤1⅐RanGTP complex was able to support import under similar conditions.
Our studies of GTP hydrolysis in vitro and in permeabilized cells (Fig. 8) revealed a requirement for Kap␣ and NLS peptide for maximal GTP hydrolysis at the NPC. The trimeric Kap␤1⅐RanGTP⅐Nup358 complex was resistant to the action of RanGAP1. The addition of Kap␣ and NLS peptide cooperatively induced complete GTP hydrolysis in the presence of RanGAP1. This did not result from a dissociation of Kap␤1 from Nup358 and RanGTP, but rather from cooperative binding of Kap␣ and NLS to the Kap␤1⅐RanGTP⅐Nup358 complex (Fig. 7). Ultrastructural studies indicate that the initial docking site of NLS bearing cargo is at the tips of the cytoplasmic fibrils (46,49), where Nup358 resides. Thus, the pentameric NLS⅐Kap␣⅐Kap␤1⅐Ran⅐Nup358 complexes may represent early intermediates in the nuclear import pathway for classical NLSbearing proteins (Fig. 10). The small amount of Kap␣⅐NLS binding in the absence of Ran could be mediated by interaction of Kap␤1 with the repeat regions present in our Nup358 constructs ( Fig. 1), but the significant increase seen in the presence of Ran, particularly RanGTP, occurs in all probability through the RBH domains. However, even in the presence of RanGTP, the amount of Kap␣⅐NLS bound was small compared with the input, suggesting that the pentameric complexes represent unstable transient intermediates in nuclear import, the function of which is to induce GTP hydrolysis. In summary, the RBH domains of Nup358 appear to be a site where docking of Kap␣⅐NLS results in GTP hydrolysis and hence recycling of the Kap␤1⅐RanGTP complex, thus linking the termination of one cycle of import and the initiation of the next (Fig. 10).
Maximal GTP hydrolysis in vitro required both Kap␣ and NLS (i.e. the formation of the pentameric complex). In permeabilized cells, under the same conditions, NPC-associated SUMO-RanGAP1 catalyzed only about 60% hydrolysis of GTP (Fig. 8B). This is unlikely to be due to inadequate quantity of NPC-associated SUMO-RanGAP1, because near-complete GTP hydrolysis occurred in the absence of Kap␤1 (Fig. 8B). It thus appears that the RanGTP-Kap␤1 complex bound to a subset of the RBH domains on Nup358 may not be accessible to SUMO-RanGAP1 that is tethered to the C-terminal portion of Nup358. The addition of free recombinant RanGAP1 to the cells resulted in complete hydrolysis (not shown). These data suggest that the two forms of RanGAP1 cooperate in GTP hydrolysis at the NPC. Indeed, free RanGAP has been shown to enhance nuclear import in permeabilized cells and becomes essential if Ran is FIG. 9. GTP hydrolysis is required for reinitiation of nuclear import. A, preparation of Kap␤1⅐RanGTP and Kap␤1⅐RanGMPPNP complexes. RanGTP or RanGMPPNP was added in excess to GST-Kap␤1 immobilized on glutathione-Sepharose beads. After washing, the Kap␤1⅐RanGTP and Kap␤1⅐RanGMPPNP complexes were eluted with glutathione, exchanged into transport buffer, and checked by SDS-PAGE and Coomassie Blue staining. B, recycling of the Kap␤1⅐RanGTP complex for nuclear import requires GTP hydrolysis. Import reactions were carried out in permeabilized cells as described under "Experimental Procedures." In the left panel, Kap␤1⅐RanGTP complex and GTP were added, whereas the right panel contained Kap␤1⅐RanGMPPNP complex and GMPPNP. In addition, both reactions contained NLSbovine serum albumin-rhodamine, Kap␣, p10, RanBP1, and RanGAP1. provided in its GTP-bound form (19). Furthermore, yeast Ran-GAP (Rna1p), which is neither SUMO-modified nor targeted to the NPC, has been shown to be essential for import (68).
Another difference between NPC-associated SUMO-Ran-GAP1 and free unmodified RanGAP1 is that NLS peptide did not enhance GTP hydrolysis by the former (Fig. 8B). A comparison with yeast may be relevant to this observation. Hydrolysis of GTP after the formation of a yeast Kap␤1⅐RanGTP complex can be accomplished in a reaction involving RanBP1, Kap␣, and the C terminus of the yeast nucleoporin Nup1 (36). It is interesting to note that in the same study, direct binding of the C terminus of Nup1 to Kap␣ was observed. It may well be that the C terminus of Nup1, by binding to Kap␣, played a role similar to that of the NLS peptide in the present study. By analogy, it is possible that NLS-like sequences in other components of the NPC in permeabilized cells played a similar role with regard to NPC-associated SUMO-RanGAP1, thus obviating the need for NLS peptide (Fig. 8B). In any case, the functional significance of this difference between free and NPCassociated RanGAP1 remains to be determined. One possibility is that free, unmodified RanGAP1 may link GTP hydrolysis to NLS availability, whereas NPC-associated RanGAP1 may be involved in the recycling of Kap␤1 back to the cytosol in the presence of "empty" Kap␣.
Two types of models have been proposed for the initiation and termination of nuclear import. One assumes that the NLS⅐Kap␣⅐Kap␤1 complex forms in the cytosol and then docks at the NPC (69). In this case, it is proposed that GTP hydrolysis (recycling) is mediated by RanBP1 and occurs independently from the docking step of import. The other model postulates that the Kap␤1⅐RanGTP complex binds to Nup358, that GTP hydrolysis occurs there as Kap␣ and NLS release Kap␤1 from RanGTP, and that this is followed by the formation of a more stable RanGDP-containing complex (70). Our data agree with the second model in that GTP hydrolysis occurs on Nup358. However, we show that Kap␣ and NLS bind to the Kap␤1⅐ RanGTP⅐Nup358 complex rather than destabilize it, that RanGTP is more favorable for this binding than RanGDP, and that hydrolysis occurs within a pentameric NLS⅐Kap␣⅐Kap␤1⅐ RanGTP⅐Nup358 complex.
To summarize, we have shown that Nup358 at the cytoplasmic fibers of the NPC provides a site where GTP hydrolysis by Ran terminates one cycle of import as Kap␣ and NLS dock to initiate the next cycle. This may be the first and last of several steps at the NPC that are required to complete a cycle of nuclear import.