Cse1p Is Required for Export of Srp1p/Importin-α from the Nucleus in Saccharomyces cerevisiae

In metazoan cells, the CAS protein has been shown to function as a recycling factor for the importin-α subunit of the classical nuclear localization signal receptor, exporting importin-α from the nucleus to allow its participation in multiple rounds of nuclear import. CAS is a member of a family of proteins that bear homology to the larger subunit of the nuclear localization signal receptor, importin-β, and that are found in all eukaryotes from yeast to humans. Sequence similarity identifies the product of theSaccharomyces cerevisiae CSE1 gene as a potential CAS homologue. Here we present evidence that Cse1p is the functional homologue of CAS: Cse1p is required to prevent accumulation of Srp1p/importin-α in the nucleus, it localizes to the nuclear envelope in a pattern typical of nuclear transport receptors, and it associatesin vivo with Srp1p in a nucleotide-specific manner. We show further that mutations in CSE1 and SRP1 have specific effects on their association and on the intracellular localization of Cse1p.

In metazoan cells, the CAS protein has been shown to function as a recycling factor for the importin-␣ subunit of the classical nuclear localization signal receptor, exporting importin-␣ from the nucleus to allow its participation in multiple rounds of nuclear import. CAS is a member of a family of proteins that bear homology to the larger subunit of the nuclear localization signal receptor, importin-␤, and that are found in all eukaryotes from yeast to humans. Sequence similarity identifies the product of the Saccharomyces cerevisiae CSE1 gene as a potential CAS homologue. Here we present evidence that Cse1p is the functional homologue of CAS: Cse1p is required to prevent accumulation of Srp1p/importin-␣ in the nucleus, it localizes to the nuclear envelope in a pattern typical of nuclear transport receptors, and it associates in vivo with Srp1p in a nucleotide-specific manner. We show further that mutations in CSE1 and SRP1 have specific effects on their association and on the intracellular localization of Cse1p.
Bi-directional transport of macromolecules through nuclear pore complexes (NPCs) 1 is an energy-dependent process that involves substrate recognition by a soluble transport receptor, binding of the receptor to nuclear pore proteins, and subsequent translocation of the receptor-cargo complex through the NPC. All transport receptors identified thus far are members of a family of proteins with similarity to the importin-␤ subunit of the dimeric classical nuclear localization signal (NLS) receptor (Refs. 1-3 and reviewed in Refs. 4 and 5). Importin-␤ family members are most homologous in their N termini (3), which contain a conserved domain that binds to the small GTPase Ran, an essential component of the transport machinery (6,7). Proteins containing basic NLSs of the SV40 or bipartite variety bind to importin-␣, which in turn binds to importin-␤, serving as an adapter between these import substrates and their transport receptor (8 -10).
After crossing the NPC, the importin-␣/␤ dimer dissociates due to the interaction of importin-␤ with Ran-GTP instead of Ran-GDP, thus leaving importin-␣ free in the nucleus (11,12).
Recently, the importin-␤-like protein CAS (cellular apoptosis susceptibility protein) from Xenopus and HeLa cells has been identified as a recycling factor for importin-␣ that mediates export of the protein from the nucleus to allow its participation in further rounds of import (13).
The high degree of conservation of nuclear transport pathways between yeast and higher eukaryotes, combined with the genetic tractability of yeast, has made Saccharomyces cerevisiae a valuable tool for the identification of new transport factors as well as the elucidation of functional interactions between known factors (reviewed in Ref. 14). A search of the completed S. cerevisiae genome identified CSE1 as the potential yeast homologue of the CAS gene. The predicted protein products of the two genes are 40% identical (13). CSE1 is an essential gene that was originally recovered in a screen for mutations that affect chromosome segregation (15). SRP1, the gene that encodes yeast importin-␣ (10,16), was found to be a high copy suppressor of the cold-sensitive cse1-1 mutation (15). Here we show that Cse1p is required for export of Srp1p from the nucleus and that the two proteins interact in a GTP-dependent manner, as expected for a receptor-substrate export complex.
Immunofluorescence-Srp1p was localized in CSE1 and cse1-1 strains at 30 and 15°C (12-h shift) using indirect immunofluorescence microscopy. Cells were fixed in 3.7% formaldehyde for 30 min, washed in 0.1 M potassium phosphate buffer, pH 6.5, and resuspended in P solution (0.1 M potassium phosphate buffer, pH 6.5, 1.2 M sorbitol). Cell walls were digested by incubation at 30°C with 0.3 mg/ml zymolyase, 25 mM dithiothreitol, and cells were applied to slides precoated with 0.3% polylysine. Cells were permeabilized by treatment with 0.5% Nonidet P-40 in P solution and blocked with 1 mg/ml bovine serum albumin in PBS. Rabbit polyclonal anti-importin-␣ antibodies (gift from * This work was supported by grants from the National Institutes of Health and the Novartis/Dana-Farber Cancer Institute Drug Discovery Program (to P. A. S.). 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 U.S.C. Section 1734 solely to indicate this fact.
‡ Supported by a National Science Foundation pre-doctoral fellowship and a National Institutes of Health training grant.
Anti-Cse1p Antibodies-To make a GST-Cse1p expression plasmid, the CSE1 open reading frame was PCR amplified, and BamHI and XhoI sites were introduced at the ends using Vent polymerase with primers JKH38 (5Ј-CGGGATCCATGTCCGATTTGGAAACCGTAGC-3Ј) and JKH31 (5Ј-CCGCTCGAGTTAATTACCAACTAATAATTG-3Ј). The PCR product was digested with BamHI and AvaI and was cloned into the BamHI/XhoI sites of pGEX-4T-1 (Amersham Pharmacia Biotech) to generate pPS1611. GST-Cse1p was expressed from pPS1611 in ompT Ϫ Escherichia coli by 3 h isopropyl-1-thio-␤-D-galactopyranoside (0.1 mM) induction at 30°C and was purified by binding to glutathione-Sepharose (Amersham Pharmacia Biotech) and elution with 50 mM Tris⅐Cl, pH 8.0, 5 mM reduced glutathione. Rabbit polyclonal antibodies were raised against full-length GST-Cse1p (Covance Research Products, Inc.). The antibody was used at a 1:5000 dilution for immunoblotting.
Immunoprecipitations-100-ml cultures of Cse1-GFP or cse1-1-GFP strains were grown in rich medium to approximately 2 ϫ 10 7 cells/ml and then were pelleted and washed once with water. Cells were lysed in 250 l of ice-cold PBSMT (2 mM MgCl 2 , 1 mM EDTA, 0.5% Triton X-100 in PBS) plus protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 3 g/ml each pepstatin A, leupeptin, aprotinin, and chymostatin) using glass beads in a FastPrep bead beater (Savant). After lysis, 500 l more of PBSMT were added, and lysates were clarified by two 10-min microcentrifuge spins at 4°C. Anti-GFP beads were prepared by incubating 300 l of IgG-Sepharose (Amersham Pharmacia Biotech, 50% slurry in PBS) with 100 l 1 mg/ml rabbit polyclonal anti-GFP antibodies (gift of J. Kahana) for 15 min on a room temperature rocker. Beads were washed three times with PBS, ending with a 50% slurry. 10 l of anti-GFP beads were used to immunoprecipitate GFP-tagged proteins from 250 l of lysate (approximately 1 mg/ml total protein concentration by Bio-Rad protein assay) by rocking at 4°C for 1 h. The GTP analogue GMPPNP (Sigma) was added to some lysates along with the beads (1 mM GMPPNP/2 mM MgCl 2 ). Beads were washed four times with 500 l of ice-cold PBSMT and once with 500 l of ice-cold PBSM (2 mM MgCl 2 , 1 mM EDTA in PBS). Samples were prepared for gel analysis by adding 50 l of 1ϫ SDS sample buffer and boiling for 2 min.
Immunoblotting-Anti-GFP immunoprecipitates were resolved in 8 or 10% SDS-polyacrylamide gels (19), and proteins were transferred to nitrocellulose membranes using standard techniques (20). Anti-GFP antibodies were used at a 1:4000 dilution, and anti-importin-␣ antibodies were used at a 1:2000 dilution. Horseradish peroxidase-conjugated anti-rabbit secondary antibodies (Jackson ImmunoResearch) were used at a 1:5000 dilution, and immunoreactive bands were visualized using enhanced chemiluminescence (Renaissance kit, NEN Life Science Products).
Visualization of GFP-tagged Proteins-CSE1-GFP and cse1-1-GFP strains were grown in rich medium to 1 ϫ 10 7 cells/ml. For testing cold sensitivity, strains were shifted to 15°C for 6 h; for temperature sensitivity, strains were shifted to 37°C for 1 or 2 h. 1-ml culture aliquots were spun down; cells were washed once with water and examined immediately under a fluorescence microscope (Nikon) using a GFPspecific filter. Images were captured using Metamorph imaging software on the autoscale setting.

RESULTS AND DISCUSSION
By amino acid sequence similarity, Cse1p is the closest CAS protein homologue in S. cerevisiae. We sought to determine whether it is also a functional homologue, that is, whether Cse1p is the recycling factor that returns Srp1p to the cytoplasm after it enters the nucleus along with a classical NLScontaining substrate. To address this question, we first examined the localization of Srp1p in the cold-sensitive cse1-1 strain by indirect immunofluorescence. In wild type cells at 30°C, anti-Srp1p staining shows a diffuse pattern throughout the cell (Fig. 1A), indicative of its continued movement in and out of the nucleus. At 15°C, some wild type cells show a partial accumulation of Srp1p in the nucleus, suggesting that export is slowed in the cold. However, they still retain significant cytoplasmic Srp1p signal (Fig. 1G). In contrast, even at 30°C, cse1-1 cells show significant accumulation of Srp1p in the nucleus (Fig.  1D); at the nonpermissive temperature of 15°C, anti-Srp1p signal is restricted to the nucleus in virtually 100% of the cells (Fig. 1J). This result implies that Cse1p plays a role in the nuclear export of Srp1p. Although cse1-1 cells show nuclear accumulation of Srp1p even at the permissive temperature, they do not grow significantly slower than wild type cells at 30°C (data not shown). This apparent contradiction may be explained by the fact that Srp1p appears to be a very highly expressed protein. Thus, even if a majority of the Srp1p in a cell is nuclear at the permissive temperature, its rate of recycling to the cytoplasm may be sufficient to allow nearly normal rates of nuclear import and population doubling.
To gain further evidence that Cse1p is the export receptor for Srp1p, we localized Cse1p in strains in which the endogenous CSE1 or cse1-1 locus was replaced with sequence encoding Cse1p with GFP fused to its C terminus. The integration strategy was such that the fusion protein is the sole form of Cse1p expressed in these strains (see "Experimental Procedures"). Fig. 2 shows the endogenous Cse1 and cse1-1 proteins detected by immunoblotting with anti-Cse1p antibodies (lanes 1 and 3) and their respective GFP fusion proteins visualized by blotting with anti- Cse1p (lanes 2 and 4) or anti-GFP (lanes 6 and 8) antibodies. The predicted molecular masses of endogenous Cse1p/cse1-1p and the GFP fusion proteins are 109 and 135 kDa, respectively. Cse1p migrates slightly faster than expected on a 10% SDS-polyacrylamide gel, running closer to the 100-kDa marker than the 110-kDa marker (Fig. 2, lanes 1 and 3). There is an approximately 115-kDa protein that also reacts with the anti-Cse1p antibody in all lysates (Fig. 2, lanes 1-4). We do not know the identity of this band, but it is a formal possibility that it could be another importin-␤-like protein. The sequenced yeast genome predicts at least two potential homologues, Lph2p and YGL241W, which would migrate at this approximate molecular size. The integrated Cse1-GFP strain showed no growth defects compared with its parental strain (data not shown), indicating that the fusion protein was completely functional. Furthermore, when Cse1-GFP was integrated into cse1-1, the resulting strain retained its cold sensitivity (data not shown), confirming that the integration strategy did not eliminate the original cse1-1 mutation.
The wild type and mutant fusion proteins were localized by fluorescence microscopy. Cse1-GFP shows a strong punctate nuclear envelope localization (Fig. 3A, top panels), similar to that normally observed for NPC proteins (e.g. see Ref. 21). The association of Cse1-GFP with nuclear pores was confirmed by localizing the fusion protein in a rat3-1 strain. This strain has a defect in mRNA export at high temperatures and exhibits clustering of NPCs at one or a few regions of the nuclear envelope at all temperatures (22). As shown in Fig. 3B, Cse1-GFP also shows a clustering pattern in rat3-1 cells, indicating that the localization seen in wild type cells represents authentic NPC association.
Most of the yeast importin-␤-like proteins examined thus far show a similar concentration at nuclear pores but with varying degrees of background cytoplasmic and nuclear localization (23)(24)(25)(26)(27)(28)(29). For Cse1-GFP, the non-pore-associated signal was essentially undetectable. The mutant cse1-1-GFP protein also localized to the nuclear envelope but showed higher intranuclear signal than the wild type fusion protein (Fig. 3A, lower panels) at both 30 and 15°C, suggesting that the mutant protein cannot exit the nucleus as well as wild type Cse1p.
Similar intranuclear accumulation of Cse1-GFP was seen in a prp20-1 strain shifted to the nonpermissive temperature of 37°C (Fig. 3C, middle panels). The PRP20 gene product is the guanine nucleotide exchange factor for Gsp1p (30,31), which is the yeast Ran protein (32)(33)(34). prp20-1 cells exhibit both protein import and mRNA export defects (30,31,35,36). The accumulation of Cse1-GFP inside the nucleus of prp20-1 cells is indicative of a requirement for Gsp1p-GTP for translocation of Cse1p through the NPC. We also localized Cse1-GFP in rna1-1 cells. RNA1 encodes the GTPase-activating protein for Gsp1p (37)(38)(39). The mutant rna1-1 protein fails to stimulate Gsp1p-GTP hydrolysis, and therefore rna1-1 cells have elevated levels of Gsp1p-GTP (39). In the rna1-1 strain, the NPC localization of Cse1-GFP was diminished, and the protein was localized throughout the cell (Fig. 3C, right panels). Because Srp1p shows a similar diffuse localization, this result may indicate that, in the presence of Gsp1p-GTP, Cse1p constitutively associates with its transport substrate.
We next looked for a biochemical interaction between Cse1p and Srp1p, which would be expected if the two form a genuine receptor-substrate export complex. Rabbit polyclonal anti-GFP antibodies bound to IgG-Sepharose beads were used to immunoprecipitate Cse1-GFP from lysates prepared from the integrated Cse1-GFP and cse1-1-GFP strains (Fig. 4A, top panel,  lanes 4 -9). Co-immunoprecipitation of Srp1p was assayed by immunoblotting with anti-importin-␣ antibodies (Fig. 4A, bottom panel, lanes 4 -9). In the absence of any nucleotide analogue, Cse1-GFP immunoprecipitates contained only background levels of Srp1p, similar to a control in which beads were incubated with lysate that did not contain any GFP fusion protein (Fig. 4A, compare lanes 4 and 6). However, when the nonhydrolyzable GTP analogue GMPPNP (1 mM) was added to the Cse1-GFP lysate, co-immunoprecipitation of Srp1p was vastly stimulated (Fig. 4A, compare lanes 5 and 7). Srp1p did not co-immunoprecipitate with two other importin-␤-like proteins expressed as C-terminal GFP fusions (Xpo1-GFP and Sxm1-GFP) or with a fusion of GFP to the N terminus of ␤-galactosidase (Fig. 4B, lanes 4 -9). Therefore, the interaction between Cse1-GFP and Srp1p is specific; it does not reflect an affinity of Srp1p either for GFP or for multiple members of the importin-␤ family. The failure of Srp1p to interact with Crm1/ Xpo1-GFP is especially significant because Crm1p/Xpo1p is also an export receptor, known to mediate export of proteins that contain leucine-rich nuclear export signals (40,41).
The CAS protein has been shown to bind cooperatively to importin-␣ and RanGTP (13). Yeast cell lysates prepared under normal conditions would be expected to contain Gsp1p primarily in the GDP-bound state due to the GTPase activating activity of Rna1p. The addition of the nonhydrolyzable GTP analogue GMPPNP to these lysates should restore the GTPbound population of Gsp1p. The fact that the Cse1-GFP/Srp1p interaction only occurs in the presence of GMPPNP implies that it represents a Gsp1p-GTP-dependent export complex.
Srp1p did not co-immunoprecipitate from lysates from cse1-1-GFP expressing cells even in the presence of GMPPNP (Fig.  4A, lane 9). This result suggests that the cse1-1 mutation affects the formation of the Cse1p-Srp1p export complex and provides an explanation for the previously observed suppression of cse1-1 by high copy SRP1 (15). The cse1-1 strain contains a mutation that changes amino acid residue 220 from aspartate to asparagine (D220N). 2 This residue lies outside the conserved N-terminal Ran binding motif that is shared by importin-␤ family members; However, the experimentally determined Ran binding domain of importin-␤ extends beyond this conserved region (42,43). Therefore, it is possible that the cse1-1 mutation disrupts Gsp1p-GTP binding and thus interferes with the formation of a cooperative Cse1p-Srp1p-Gsp1p-GTP complex. Alternatively, the D220N mutation may identify a region of Cse1p that is directly involved in binding to Srp1p.
Cse1-GFP was also integrated into two strains containing mutations in SRP1, srp1-31, and srp1-49, and similar immunoprecipitation experiments were carried out using lysates from these strains. Like wild type Srp1p, the mutant srp1-49 protein co-immunoprecipitated with Cse1-GFP in the presence of GMPPNP, but the srp1-31 protein did not (Fig. 4B, compare  lanes 7 and 9). Note that the level of Cse1-GFP expression in the srp1-31 strain was decreased relative to the wild type and srp1-49 strains but that longer exposures of the immunoblot still did not visualize any co-precipitating srp1-31 protein.
Srp1p contains a central domain consisting of eight 42-amino acid tandem repeats of the armadillo (arm) motif found in a family of proteins that includes the Drosophila armadillo protein and the vertebrate ␤-catenin and plakoglobin proteins (17). X-ray crystal structures have recently been determined for the arm repeats of ␤-catenin (44) and importin/karyopherin-␣ (45); both structures exhibit an extensive shallow surface groove that seems to function as a specific proteinprotein interaction module. In the case of importin/karyo-pherin-␣, the surface groove constitutes the binding site for positively charged NLS peptides (45). A conserved 41-amino acid domain not present in the crystallized protein has been identified as the importin-␤ binding domain (46).
The srp1-49 mutation lies within the first arm repeat (17), whereas the srp1-31 mutation falls just N-terminal to the start of the arm domain (17) and C-terminal to the importin-␤ binding domain. The different interaction properties of the two mutant Srp1 proteins may indicate that the srp1-31 mutation lies within a region of the protein that is required for binding to Cse1p, but further studies will be required to elucidate the precise domain in Srp1p that interacts with Cse1p. srp1-31 cells exhibit a strong nuclear import defect for classical NLScontaining substrates, whereas srp1-49 cells show a much weaker defect (16). This observation is consistent with the failure of the srp1-31 protein to bind to Cse1p, but paradoxically, the mutant protein does not mislocalize to the nucleus. 3 This contradiction may be explained if the srp1-31 mutation abrogates binding to both Cse1p and importin-␤. In this case, the mutant protein would never enter the nucleus and thus could not be trapped by a lack of interaction with its recycling factor.
The CAS protein has been shown to export importin-␣ in metazoan cells. Here we present evidence that Cse1p, the S. cerevisiae CAS homologue, is the export receptor for Srp1p, which is the yeast importin-␣ protein. We have demonstrated a biochemical interaction between Cse1p and Srp1p that requires the presence of GTP and have identified mutations in both Cse1p and Srp1p that abolish this interaction. The facility of genetic and biochemical analysis in yeast will make it an excellent system for further study of the importin-␣ recycling pathway.