The Yeast Nucleoporin Nup2p Is Involved in Nuclear Export of Importin α/Srp1p*

The importin α·β heterodimer mediates nuclear import of proteins containing classical nuclear localization signals. After carrying its cargo into the nucleus, the importin dimer dissociates, and Srp1p (the yeast importin α subunit) is recycled to the cytoplasm in a complex with Cse1p and RanGTP. Nup2p is a yeast FXFG nucleoporin that contains a Ran-binding domain. We find that export of Srp1p from the nucleus is impaired inΔnup2 mutants. Also, Srp1p fusion proteins accumulate at the nuclear rim in wild-type cells but accumulate in the nuclear interior in Δnup2 cells. A deletion of NUP2shows genetic interactions with mutants in SRP1 andPRP20, which encodes the Ran nucleotide exchange factor. Srp1p binds directly to an N-terminal domain of Nup2p. This region of Nup2p is sufficient to allow accumulation of an Srp1p fusion protein at the nuclear rim, but the C-terminal Ran-binding domain of Nup2p is required for efficient Srp1p export. Formation of the Srp1p·Cse1p·RanGTP export complex releases Srp1p from its binding site in Nup2p. We propose that Nup2p may act as a scaffold that facilitates formation of the Srp1p export complex.

Transport of macromolecules between the cytoplasm and nucleus in eukaryotic cells is mediated by soluble transport receptors. These receptors bind to specific protein or RNA cargoes and ferry them across the nuclear pore complex (NPC) 1 (reviewed in Refs. 1 and 2). Several different nuclear import and export pathways have been identified (reviewed in Ref. 3). The first nuclear transport pathway that was described mediates import of proteins that contain a classical nuclear localization signal (NLS) (4). The NLS is recognized by importin ␣/Srp1p, which forms a heterodimeric complex with importin ␤/Kap95p. By virtue of its interaction with importin ␤, importin ␣ and its associated NLS-containing protein are carried through the NPC into the nucleus (5,6). Binding of the GTPbound form of the small GTPase Ran to importin ␤ causes dissociation of the import complex in the nucleus (7,8). Importin ␣/Srp1p then must be recycled to the cytoplasm to allow for multiple rounds of NLS protein import. Importin ␣/Srp1p is exported from the nucleus as part of a trimeric complex containing the exportin CAS/Cse1p and RanGTP (9 -12). The export complex is ultimately dissociated in the cytoplasm by RanBP1/Yrb1p (13,14) and RanGAP1/Rna1p (15,16), acting to trigger GTP hydrolysis on Ran (9,11).
All nucleocytoplasmic transport of macromolecules takes place through the NPC (17). The NPC consists of a large number of proteins termed nucleoporins (18). Many of the nucleoporins have repeated sequence motifs containing the dipeptide Phe-Gly (e.g. FXFG and GLFG). Numerous binding interactions have been identified between FG nucleoporins and soluble transport factors (7, 19 -25). However, in most cases, the functional significance of these interactions is unclear. We previously showed that Srp1p binds to the yeast nucleoporins Nup1p and Nup2p (26). Nup2p is an FXFG nucleoporin that contains a Ran-binding domain (RanBD) homologous to that of RanBP1 (14,27,28). Nup2p is the only yeast nucleoporin that contains such a Ran-binding domain. In this report, we further investigate the interaction of Srp1p with Nup2p and its functional significance. Our results indicate that Nup2p is involved in the nuclear export of Srp1p and that Nup2p may act as a scaffold to facilitate formation of the Srp1p export complex at the NPC.

Reagents
Enzymes for molecular biology were purchased from New England Biolabs (Beverly, MA), Roche Molecular Biochemicals, and Amersham Pharmacia Biotech. 5-Fluoroorotic acid was obtained through the Genetics Society of America consortium. pRS vectors (29) were furnished by Dr. Phil Hieter. Rabbit-anti-␤-galactosidase antibodies were obtained from Organon Technica (West Chester, PA), and anti-Xpress mAb was from Invitrogen (Carlsbad, CA).

Disruption of NUP2
NUP2 was disrupted in W303 by transformation with a KAN R cassette (33) created by PCR using oligonucleotides 5Ј-CATACATCATTTT-TCATACAAGTCCTTGTTAAGCAGCTGAAGCTTCGTACGC and 5Ј-G-GGTTCTATTCTATTTAAAATTGTTAACTGGCATAGGCCACTAGTG-* This work was supported by National Institutes of Health Grant GM-54768-04 (to L. I. D.) and by a Postdoctoral Fellowship from the Natural Sciences and Engineering Research Council of Canada (to J. W. B.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18

Yeast Expression Plasmids
SRP1-GFP-Superglow GFP was cut out of pJK19 -1 (40) with NheI and inserted into the XbaI site of pRS305 (LEU2) to generate pLDB349. A promoterless SRP1 gene was amplified by PCR with oligonucleotides introducing a BamHI site at the start codon and an SpeI site before the stop codon and was inserted into pLDB349 at XhoI-SpeI to generate an in-frame fusion of SRP1 with GFP (pLDB520). pLDB520 was linearized with PstI, which cuts in SRP1, and transformed into W303 or LDY680 (⌬nup2::KAN R ) to integrate at SRP1, generating JBY1 and JBY14, respectively.
SRP1-LacZ-A fusion of the SRP1 coding region to the Escherichia coli LacZ gene was created as follows. A fragment containing the SRP1 promoter and initiator methionine was amplified by PCR and cloned into YIp368R (41) to produce an in-frame fusion with LacZ. Next, the region of SRP1 encoding residues 463-542 was amplified and inserted into this plasmid. The resulting plasmid was linearized at a site between the SRP1 promoter and the C-terminal region, and transformed into W303. In vivo gap repair followed by integration of the vector resulted in the formation of full-length SRP1-LacZ fusions (transformants screened by Western blotting). The plasmid was rescued from this strain by cutting genomic DNA with PstI, followed by ligation and transformation into E. coli. The rescued plasmid (pLDB360) was cut with PstI and transformed into W303 or LDY627 (⌬nup2::TRP1) to integrate at SRP1, generating LDY970 and LDY969, respectively.

Immunofluorescence and Fluorescence Microscopy
Rabbit polyclonal anti-Srp1p antibodies were raised by injection with affinity-purified His 6 -Srp1p (28 -542). The antibodies were affinitypurified from serum using His 6 -Srp1p (28 -542) bound to nitrocellulose (46). Cells were treated for immunofluorescence essentially as described by Wente et al. (37) except that they were fixed for 12 min in buffer containing 10% formaldehyde and 10% methanol. Primary antibodies were used at a dilution of 1:200 and secondary antibodies at a dilution of 1:100. For visualizing Srp1p-GFP or Yap1p-GFP, cells were grown overnight in selective medium and then diluted to OD 600 ϭ 0.1 and grown for 6 h in YPD to log phase. Cells were then harvested by centrifugation, washed once with phosphate-buffered saline, resuspended in ice cold 75% ethanol, incubated on ice for 10 min, and then washed once with phosphate-buffered saline and resuspended in phosphate-buffered saline containing DAPI (0.2 mg/ml) for viewing using a Nikon Optiphot microscope.
For analysis of Nup2p protein expression, yeast cell extracts were prepared by vortexing with glass beads in trichloroacetic acid (48). Western blotting was performed essentially as described (26) with either mAb 9E10 or mAb 414 ascites fluids, both at a dilution of 1:1000.

Overlay of GST-Nups
Glutathione-Sepharose beads were used to precipitate GST-Nup fusion proteins from bacterial lysates. The beads were washed, and bound protein was eluted in SDS sample buffer containing 6 M urea and subjected to SDS-PAGE without boiling. Two hundred nanograms of purified GST fusion was loaded in each lane, as determined by Coomassie Blue staining. Blot overlays were performed as described by Lee and Melese (49), using affinity-purified His 6 -Srp1p (74 -542) as a ligand at a concentration of 125 ng/ml. Overlays were probed with anti-Xpress antibody diluted 1:5000, and bound antibody was detected using the ECL detection system (Amersham Pharmacia Biotech).

Precipitation of NLS-Human Serum Albumin (HSA)
HSA was coupled to oligopeptides consisting of the SV40 large T antigen NLS Pro-Lys-Lys 128 -Lys-Arg-Lys-Val or mutant peptides containing a substitution of Thr for Lys 128 . Bacterial lysates containing GST fusion proteins were bound to glutathione-Sepharose and washed one time with wash buffer (250 mM NaCl, 50 mM Hepes, pH 7.0, 5 mM EDTA, 0.5 mM dithiothreitol, 0.1% Tween 20) before further additions. Incubations with His 6 -Srp1p (28 -542) or NLS-HSA were performed in wash buffer plus 1% nonfat dry milk; washes were with wash buffer.

Srp1p Export Defect in ⌬nup2
Mutants-The observation that Srp1p binds to Nup2p (26, 50) suggested that Nup2p might play a role in Srp1p trafficking. To investigate this possibility, we examined the subcellular localization of Srp1p in yeast mutants carrying a disruption in NUP2 by immunofluorescence using anti-Srp1p antibodies (Fig. 1). In wild-type cells, Srp1p is distributed in a largely uniform manner throughout the cytoplasm and nucleus, with some nuclear concentration in some cells (Fig. 1a). In ⌬nup2 mutants, however, there is a pronounced nuclear accumulation of Srp1p, indicating a defect in its export from the nucleus (Fig. 1, panel c).
As additional tools for studying Srp1p localization, fusions of Srp1p to either the green fluorescent protein (Srp1p-GFP) or ␤-galactosidase (Srp1p-LacZ) were constructed. These fusion proteins were expressed in yeast and localized by direct fluorescence (for Srp1p-GFP) or by immunofluorescence with anti-␤-galactosidase antibodies (for Srp1p-LacZ). In wild-type cells, localization of either Srp1p-GFP (Fig. 1e) or Srp1p-LacZ (Fig.  1i) revealed a punctate fluorescence at the nuclear rim typical of NPC localization. This stands in contrast to the lack of nuclear rim accumulation of endogenous Srp1p (Fig. 1a). What-ever the reason for the accumulation of the fusion proteins, their localization in ⌬nup2 cells is markedly different, with both Srp1p-GFP (Fig. 1g) and Srp1p-LacZ (Fig. 1k) showing a uniform nuclear localization. Thus, as is seen with Srp1p, a lack of Nup2p causes nuclear accumulation of Srp1p fusion proteins. Furthermore, the lack of fusion protein accumulation at the nuclear rim in ⌬nup2 mutants suggests that a site critical for Srp1p binding at the NPC has been lost. Srp1p-GFP was able to complement an srp1::KAN R disruption, showing that this fusion protein is functional; Srp1p-LacZ does not complement. Srp1p-GFP and Srp1p-LacZ were expressed at similar levels in wild-type and ⌬nup2 cells (data not shown).
No nuclear accumulation of Srp1p is seen in ⌬nup1 mutants ( Fig. 2A), ⌬nup100 or ⌬nup116 mutants (data not shown), or numerous other nuclear transport mutants (11,51). Moreover, deletion of NUP2 does not cause a general defect in nuclear export, as Crm1p-mediated nuclear export proceeds normally in the absence of Nup2p (Fig. 2B). Thus, the effects of loss of Nup2p on Srp1p export are specific.
Mapping of Srp1p-binding Regions in Nup2p and Nup1p-We previously showed that Srp1p interacts with Nup1p and Nup2p in yeast extracts (26). To determine whether the binding of Srp1p to Nup1p and Nup2p is direct, interactions between bacterially expressed fusion proteins were measured. To this end, regions of Nup1p and Nup2p were fused to GST, whereas amino acids 73-542 of Srp1p were expressed with an N-terminal His 6 tag. Binding of His 6 -Srp1p (73-542) to were grown at 30°C and then prepared for immunofluorescence. Srp1p was localized by probing with polyclonal anti-Srp1p antibodies followed by DTAF-conjugated secondary antibodies (a and c). For comparison, nuclear DNA was visualized using DAPI (b and d). e-h, strains JBY1 (wild-type, e and f) or JBY14 (⌬nup2, g and h) expressing Srp1p-GFP were fixed and viewed by fluorescence microscopy to detect the GFP signal (e and g) or DAPI (f and h). i-l, strains LDY970 (wild-type, i and j) or LDY969 (⌬nup2, k and l) expressing Srp1p-LacZ were grown at 30°C and then prepared for immunofluorescence and probed with anti-␤-galactosidase antibodies followed by FITC-conjugated secondary antibodies (i and k). j and l, DAPI. the GST-Nup fusions was measured by gel overlay.
His 6 -Srp1p (73-542) binds directly to regions in both Nup1p and Nup2p (Fig. 3). A nup1 mutant has previously been shown to have a defect in Srp1p-mediated protein import (44). For Nup1p, the strongest Srp1p-binding region is found at the C terminus of the protein (Fig. 3, lanes 7, 9, and 10). Binding of Srp1p to the C terminus of Nup1p has also been reported by Floer et al. (52). We note that two nonoverlapping stretches at the C terminus both bind to Srp1p (Fig. 3, lanes 9 and 10), indicating the presence of multiple Srp1p binding sites. With Nup2p, Srp1p binds directly to the region N-terminal of the FXFG repeats (lane 2). A fusion of GST to the full-length Nup2 protein is highly degraded (lane 1), and Srp1p binds to the degradation products, which presumably represent a series of truncations of GST-Nup2p all containing the N terminus of the protein.
Srp1p does not bind to the FXFG repeat-containing regions of Nup1p or Nup2p (Fig. 3, lanes 3 and 6). Such repeats have been implicated in binding to importin ␤ family members (7,19,21). We also note that the experiment shown in Fig. 3 was performed with Srp1p protein lacking the N-terminal 72 amino acids of Srp1p, which contain the IBB domain required for binding to Kap95p (5, 6). This demonstrates that binding of Srp1p to Nup1p and Nup2p does not require the IBB domain.
An identical pattern of binding affinities for the GST-Nup fusions was seen using a His 6 -tagged full-length Srp1p, although binding to all the Nups was weaker, possibly due to the presence of Srp1p breakdown products in the protein preparation (data not shown).
One possible reason for Srp1p binding to Nup1p and Nup2p is that it binds to NLS-like sequences in these nucleoporins. No NLS consensus sequences are found in the Srp1p-binding regions of Nup1p or Nup2p; however, to further address this possibility, we tested whether Srp1p could simultaneously bind to an NLS and to Nup1p or Nup2p by assaying for the ability of Srp1p to precipitate an NLS-HSA conjugate when bound to GST-Nup fusions (Fig. 4). His 6 -Srp1p (28 -542) was incubated with GST-Nup fusions adsorbed to glutathione-Sepharose and then washed to remove unbound Srp1p. The Srp1p⅐Nup complex was then incubated with HSA conjugated to either a wild-type or mutant NLS peptide. Wild-type NLS-HSA was precipitated by each of the GST-Nup fusions when His 6 -Srp1p was present (Fig. 4, lanes 1 and 4) but was not detected in those samples to which His 6 -Srp1p was not added (lanes 2 and 5). A NLS-HSA conjugate containing a mutant NLS peptide was not precipitated (lanes 3 and 6). These data indicate that Srp1p can simultaneously bind to an NLS and to Nup2p or Nup1p. Because Srp1p has been shown to contain two adjacent NLSbinding sites (53), it is conceivable that these sites could mediate simultaneous binding to NLS-HSA and to an NLS in Nup1p or Nup2p; however, we consider this unlikely due to steric considerations. Thus, these results strongly suggest that binding of Srp1p to the nucleoporins is not via NLS-like sequences.
Domains of Nup2p Involved in Srp1p Export-In addition to the N-terminal Srp1p-binding domain and the central FXFG repeat domain, Nup2p contains a C-terminal RanBD (14,28). To determine which domains are involved in Srp1p export, we constructed Nup2p truncations lacking the RanBD or the RanBD and the FXFG repeats and tested their ability to restore the wild-type pattern of Srp1p-GFP localization to a ⌬nup2 strain (Fig. 5A).
Expression of a full-length Myc-tagged version of Nup2p (Nup2p myc ) in ⌬nup2 cells restored the wild-type pattern of Srp1p-GFP fluorescence at the nuclear rim, with no accumulation in the nuclear interior (Fig. 5A, c). Expression of Nup2p myc (1-546), lacking the C-terminal RanBD, resulted in increased fluorescence at the nuclear rim. However, significant nuclear accumulation of Srp1p-GFP remained relative to what was observed with full-length Nup2p myc (Fig. 5A, e). Thus, Nup2p lacking the RanBD but containing the N-terminal Srp1p-binding domain appears to be sufficient to allow docking of Srp1p at the pore, but insufficient to restore wild-type levels  (Nup2-N), or GST alone were adsorbed to glutathione-Sepharose beads and incubated with 0.5 g/ml His 6 -Srp1p(28 -542) (lanes 1, 3, 4, 6, 7, and 9) or buffer alone (lanes 2, 5, and 8) for 1 h at 4°C. Samples were washed one time and then incubated with 2 mg/ml HSA conjugated to NLS peptide (lanes 1, 2, 4 , 5, 7, and 8) or to mutant NLS peptide (lanes 3, 6, and 9) for 1 h at 4°C. The beads were washed six times, and bound material was subjected to SDS-PAGE, transferred to nitrocellulose, and probed with anti-HSA antiserum.
of export. We confirmed that Nup2p myc (1-546) is expressed at the same level as full-length Nup2p myc by Western blotting (Fig. 5B).
A further truncation of Nup2p, Nup2p(1-174), contains only the region N-terminal of the FXFG repeats. When this truncation was expressed in ⌬nup2 cells, Srp1p-GFP remained largely nuclear, but some fluorescence at the nuclear rim was still visible in many cells (Fig. 5A, g, arrows). The difference between Nup2p(1-174) and Nup2p myc (1-546) may indicate a role for the FXFG repeats in Srp1p export, or it may be that Nup2p(1-174) is expressed or targeted to the NPC to a lesser extent than Nup2p myc . In any case, these observations suggest that the N-terminal Srp1p-binding domain alone is sufficient to provide partial docking function, although export is very inefficient. Another indication that Nup2p(1-174) is partially functional comes from our observation that replacement of Srp1p with Srp1p-GFP results in slow growth in ⌬nup2 cells, but not wild-type cells, and that this growth defect is largely suppressed by expression of either Nup2p(1-174) or Nup2p myc (1-546). Thus, our observations indicate that the Nterminal Srp1p-binding domain of Nup2p is sufficient to provide partial function, but that efficient export requires the Ran-binding domain in addition.
Genetic Interactions-Independent evidence for a role for Nup2p in Srp1p export comes from genetic considerations. Deletion of NUP2 has no effect on growth (34), despite the observed defect in Srp1p export, implying that the residual level of Srp1p export is sufficient to sustain wild-type growth rates. If so, one might expect that deletion of NUP2 would have synergistic effects in combination with other mutants that affect the export or function of Srp1p.
To test this we crossed ⌬nup2 to several nuclear transport mutants. First, ⌬nup2 cells were crossed to cells bearing the srp1-31 temperature-sensitive mutation (35) and carrying wild-type SRP1 on a URA3-containing plasmid. ⌬nup2 srp1-31 haploid progeny were unable to grow on plates containing 5-fluoroorotic acid (Fig. 5A), indicating that they were unable to lose the SRP1 URA3 plasmid and thus that ⌬nup2 and srp1-31 are synthetically lethal. Moreover, as noted above, ⌬nup2 cells but not wild-type cells show a growth defect when endogenous Srp1p is replaced with Srp1p-GFP. These synthetic phenotypes would be expected if Nup2p plays a role in Srp1p export, because any defect in Srp1p function could be exacerbated by inhibition of its cycling between the nucleus and cytoplasm.
The ⌬nup2 mutation was also crossed to a temperaturesensitive mutant in PRP20/MTR1, the nucleotide exchange factor for Ran. mtr1-1 ⌬nup2 double mutants failed to grow at 30.5°C, whereas mtr1-1 single mutants grew at this temperature (Fig. 5B). prp20-1 mutants, like ⌬nup2 mutants, have been shown to have a defect in Srp1p export (51), presumably because of a decrease in nuclear RanGTP, which is required for export complex formation. In contrast, we saw no obvious synthetic growth phenotype in ⌬nup2 cse1-1 double mutants, where cse1-1 is a cold-sensitive allele of the exportin Cse1p that carries Srp1p out of the nucleus (39). However, ⌬nup2 cse1-1 spores appeared to be slow to germinate (data not shown). The idea that Nup2p functions in export of Srp1p also suggests a possible explanation for the previously observed lethality of ⌬nup2 ⌬nup1 double mutants (34). As ⌬nup1 mutants are defective for Srp1p-mediated protein import (44), a decrease in the cytoplasmic pool of Srp1p in the absence of Nup2p may be lethal in the context of the ⌬nup1 mutation. ⌬nup2 showed no synergistic growth defects with rna1-1 or nup82 ts mutations (Fig. 5B) or with ⌬nup100 or ⌬nup116 (data not shown), suggesting that the interactions we observe with srp1 and prp20 are specific.
Competition for Srp1p Binding-Srp1p export from the nucleus depends on its incorporation into a trimeric complex with Cse1p and RanGTP (10 -12). To better understand the role of Nup2p during export of Srp1p, we performed solution binding assays with recombinant proteins to examine the modulation of Srp1p binding to Nup2p by Cse1p and RanGTP. To this end, Srp1p and Cse1p were expressed as His 6 -tagged proteins. Gsp1p (yeast Ran) and Nup2p(9 -172) (containing the Srp1pbinding region), were expressed as GST fusion proteins. The latter were either used as fusion proteins or were cleaved from GST with thrombin as necessary.
First, we confirmed that the recombinant Srp1p, Cse1p, and Gsp1p proteins were competent for forming trimeric export complexes in vitro (see Fig. 7A). GST-Gsp1p was bound to glutathione-Sepharose beads and loaded with GTP. Neither His 6 -Srp1p nor His 6 -Cse1p bound to GST-Gsp1p-GTP when added alone (lanes 1 and 2), but when added together, they bound efficiently (lane 3), consistent with the cooperative for- mation of trimeric export complexes (9,11). Binding was not observed if GST-Gsp1p was loaded with GDP instead of GTP (data not shown).
To confirm the competitive nature of the binding of Cse1p⅐Gsp1p-GTP or Nup2p to Srp1p, a converse experiment was also performed in which GST-Gsp1p-GTP on beads was incubated with His 6 -Srp1p and His 6 -Cse1p in the presence of free Nup2p (9 -172). Increasing amounts of Nup2p(9 -172) progressively inhibited formation of the Srp1p⅐Cse1p⅐Gsp1p complex (Fig. 7C), presumably by titrating Srp1p. DISCUSSION Our observations indicate that Nup2p is involved in Srp1p export, yet it binds to free Srp1p rather than to the Srp1p⅐Cse1p⅐RanGTP export complex. This suggests two possibilities for Nup2p function: either Nup2p acts at an early step in export, during export complex assembly, or at a late step, during complex disassembly. In the first case, Nup2p may facilitate formation of the Srp1p⅐Cse1p⅐RanGTP export complex at the NPC. Nup2p would provide a site to dock Srp1p, targeting complex formation directly to the NPC. In addition, docking of Srp1p and RanGTP near each other at two sites on Nup2p may improve the kinetics of assembly of the Srp1p⅐Cse1p⅐RanGTP complex. Export complex formation is a highly cooperative event in which all three components must FIG. 6. Genetic interactions of ⌬nup2. A, NOY612 (srp1-31) (35) carrying a SRP1 URA3 CEN plasmid was crossed to LDY627 (⌬nup2::TRP1). After sporulation and dissection of the resulting tetrads, haploid progeny with the indicated genotypes were spotted onto a uracil drop-out plate (-ura) or a plate with 1 mg/ml 5-fluoroorotic acid (5FOA) and grown at 25°C. B, yeast strains of the indicated genotypes were obtained by crosses described under "Experimental Procedures." Cells were spotted onto YPD plates and grown at 25, 30.5, or 37°C. assemble with each other, despite the fact that no two of the components interact pairwise in a stable manner. This process could be enhanced by bringing two of the binding partners into close proximity. Molecular scaffolds that bind to multiple proteins and thus facilitate their interaction are important in a number of systems (54,55). Formation of the Srp1p⅐Cse1p⅐RanGTP export complex would release Srp1p from Nup2p (Fig. 7B), allowing subsequent translocation of the complex through the NPC. The coupling of complex formation to dissociation from Nup2p would prevent Nup2p from being an unproductive sink for these complexes. Instead, it could act catalytically to facilitate complex formation.
Alternatively, Nup2p might be involved in the terminal step of nuclear export, i.e. complex disassembly. The metazoan RanBD-containing nucleoporin RanBP2/Nup358 (56,57), which is located on the cytoplasmic side of the NPC (58), and the cytosolic protein RanBP1/Yrb1p have been proposed to play such a role, acting to dissociate export complexes after their transit through the NPC by binding to RanGTP (59,60). Subsequent hydrolysis of GTP on the RanBD-bound Ran makes the dissociation irreversible (59). Nup2p may perform a similar function by competing RanGTP and Srp1p off Cse1p; however, we consider it more likely that Nup2p is involved in complex assembly than in disassembly. RanBP2 and RanBP1/Yrb1p bind RanGTP with high affinity, which is necessary for the proteins to function efficiently in complex disassembly. In contrast, Nup2p has a much lower affinity for RanGTP (28,61). In this respect, it is similar to Yrb2p (62, 63). Yrb2p, like Nup2p, is involved in a nuclear export process, namely export of proteins that contain a nuclear export sequence (64). This export pathway is mediated by Crm1p and is distinct from the one used by Srp1p. Yrb2p is located in the nuclear interior (63), implying that it acts at an early step in Crm1p-mediated export. Like Nup2p, Yrb2p binds to two components of a trimeric export complex: RanGTP and Crm1p. It was therefore proposed that Yrb2p might function by stabilizing the Crm1p⅐nuclear export sequence⅐RanGTP export complex (64). Alternatively, it may play a more catalytic role in export complex formation, similar to that proposed above for Nup2p. Binding to RanGTP with low rather than high affinity would allow Nup2p and Yrb2p to act as scaffolds for complex formation without being irreversible sinks for RanGTP. Consistent with the idea that low affinity binding is required, a chimeric protein in which the Yrb2p RanBD is replaced with that of Yrb1p does not rescue the nuclear export sequence export defect of ⌬yrb2 mutants (64). Thus, an attractive model is that Nup2p and Yrb2p facilitate formation of the Srp1p⅐Cse1p⅐RanGTP and Crm1p⅐nuclear export sequence⅐RanGTP export complexes, respectively. Future localization of Nup2p within the NPC may help to suggest whether Nup2p acts at an early or late step in Srp1p export.
The role, if any, of the FXFG repeat domain of Nup2p in Srp1p export is unclear. Because several members of the importin ␤ superfamily have been shown to bind to FG repeats, an attractive possibility would be that this region binds to Cse1p, thus providing a third binding activity to interact with the third member of the Srp1p export complex. However, we saw no binding of Cse1p to the Nup2p FXFG repeats either in vitro or in the two-hybrid assay (data not shown).
One possibility for Nup2p function is that it acts as a scaffold for Srp1p export complex formation. More generally, although assembly and disassembly of import and export complexes can be recapitulated in vitro with soluble components, in vivo these processes may take part largely at the NPC, rather than in the bulk cytoplasm or nucleoplasm. Dissociation of the importin ␣⅐importin ␤ complex occurs at a site on the nuclear side of the NPC (65), and RanGAP1 targeted to the NPC by SUMO-1 modification (66, 67) may dissociate export complexes at the cytosolic face of the NPC (59). Assembly of import/export complexes may also take place at scaffolding sites at the NPC.