Nuclear Export Receptor Xpo1/Crm1 Is Physically and Functionally Linked to the Spindle Pole Body in Budding Yeast

Abstract

The spindle pole body (SPB) represents the microtubule organizing center in the budding yeast Saccharomyces cerevisiae. It is a highly structured organelle embedded in the nuclear membrane, which is required to anchor microtubules on both sides of the nuclear envelope. The protein Spc72, a component of the SPB, is located at the cytoplasmic face of this organelle and serves as a receptor for the γ-tubulin complex. In this paper we show that it is also a binding partner of the nuclear export receptor Xpo1/Crm1. Xpo1 binds its cargoes in a Ran-dependent fashion via a short leucine-rich nuclear export signal (NES). We show that binding of Spc72 to Xpo1 depends on Ran-GTP and a functional NES in Spc72. Mutations in this NES have severe consequences for mitotic spindle morphology in vivo. This is also the case for xpo1 mutants, which show a reduction in cytoplasmic microtubules. In addition, we find a subpopulation of Xpo1 localized at the SPB. Based on these data, we propose a functional link between Xpo1 and the SPB and discuss a role for this exportin in spindle biogenesis in budding yeast.

In eukaryotes, the movement of proteins and RNAs between the nucleus and the cytoplasm is an essential cellular process that is mediated by soluble transport receptors (26, 52). In addition, the small GTPase Ran (Gsp1 in yeast) and its cytoplasmic and nuclear effectors are needed (12). Transport receptors involved in nuclear import reactions are usually termed importins, and they recognize their cargoes via a specific amino acid motif, the nuclear localization sequence, or NLS. By analogy, nuclear export reactions are mediated by so-called exportins. Xpo1 (also known as Crm1) was one of the first nuclear export receptors to be identified (23, 27, 50, 64) and recognizes a short leucine-rich amino acid sequence on its cargo, called a nuclear export signal, or NES. Numerous export cargoes have now been identified for the mammalian Crm1 protein, including protein kinase A inhibitor PKI, p53, and the human immunodeficiency virus protein Rev (20, 25, 66). In addition, Crm1 is involved in the nuclear export of several classes of RNAs such as viral pre-mRNAs, ribosomal RNAs, and snRNAs (54). In contrast, few NES-containing proteins have been identified in Saccharomyces cerevisiae. Although the budding yeast Crm1 orthologue Xpo1 is involved in nuclear export of Hog1 (19), Yap1 (76), Ssb1 (62), Ace2 (36), and Prp40 (46), few of the NESs involved have been characterized at a molecular level.

Recently, alongside its well-established role in nuclear export, mammalian Crm1 was shown to be involved in the control of centrosome duplication (71). Centrosomes function as microtubule (MT) organizing centers in mammalian somatic cells and direct the formation of a bipolar spindle during mitosis (reviewed in reference 6). In another study, Crm1 was found to bind to kinetochores, complex protein assemblies that are needed to establish the attachment of MTs to chromatin (4). These studies point to an unanticipated role of the exportin Crm1 in control of spindle assembly and cell cycle progression (8) and add another level of complexity to these processes since, so far, only certain importins and components of the Ran machinery are known to be involved in this process (11, 12, 28).

In budding yeast, the MT organizing center is known as the spindle pole body (SPB). Under the electron microscope, it is a multilayered high-molecular-weight structure embedded in the nuclear envelope (9). Cytoplasmic and nuclear surfaces of the SPB are known as outer and inner plaques and anchor both cytoplasmic and nuclear spindle fibers, which are spatially and functionally distinct (35). Nucleation of MTs at the SPB is mediated by the γ-tubulin (Tub4) complex that binds to the inner and outer plaques of the SPB via two specific Tub4 complex receptors, Spc110 and Spc72, respectively (56). Spc110 is located at the inner plaque of the SPB facing the nucleoplasm, and Spc72 binds at the cytoplasmic side of the SPB. Since budding yeast undergoes closed mitosis with no nuclear envelope breakdown, components of the nuclear spindle apparatus need to be imported into the nucleus from their site of synthesis in the cytoplasm. In addition, nuclear export might be required to clear the interior of the nucleus from spindle components after nuclear spindle disassembly. Curiously, information about this issue is scarce. For example, it has been shown that the γ-tubulin complex, which contains Spc97 and Spc98 as well as Tub4, is assembled in the cytoplasm and is subsequently imported into the nucleus via an NLS signal in Spc98 (53). However, given the fact that Tub4 complex is also needed at the cytoplasmic face of the SPB to nucleate cytoplasmic MTs, regulatory events must ensure the efficient and quantitative sorting of the Tub4 complex between the two compartments. How this is achieved and whether other components of the yeast SPB can act as import or export substrates is currently unknown.

In this study, we provide evidence that Spc72, a cytoplasmically localized protein of SPBs, interacts with the exportin Xpo1 in the yeast two-hybrid system and in in vitro binding experiments. Moreover, binding of Spc72 to Xpo1 is Ran dependent and mediated by an NES sequence in Spc72, which confers export activity to a reporter protein in vivo. Mutations of this NES in SPC72 and XPO1 alleles have severe consequences for spindle morphology, suggesting that in budding yeast the nuclear export machinery is involved in the biogenesis of the spindle apparatus.

MATERIALS AND METHODS

Microbial techniques and plasmids.

Standard methods were used for the transformation and genetic manipulation of yeast (7). Rich medium (yeast extract, peptone, dextrose [YPD]) and synthetic minimal medium (synthetic dextrose [SD]) were prepared as described previously (60). Plasmids used in this study are listed in Table 1. Construction details are available on request.

TABLE 1.

Plasmids

Name	Insert or descriptiona	Source or reference
pACT-II	AD	37
pASF125	pRS306-GFP-Tub1	65
pGAD	AD	Clontech
pGBD	BD	Clontech
pDH5	YFP-HIS3	YRC
pFA series	GFP-KanMX cassette	41
pGEX-6P	GST	Amersham Biosciences
pGS978	YCp GAL-Gsp1	55
pGS979	YCp GAL-Gsp1G21V	55
pGS980	YCp GAL-Gsp1T26N	55
pGS488	GST-PKI-NES-GFP2	42
pGS512	Xpo1-His6	42
pGP63	AD-Spc721-271	37
pIA204-1	BD-Xpo1	P. Preker, UCSFb
pKS63	pRS313-xpo1-101	This study
pKS102	AD-Yap184-640	This study
pKS103	AD-Sgm1218-707	This study
pKS104	AD-Spc723-433	This study
pKS105	AD-Hpa3	This study
pKS106	AD-Ltv155-463	This study
pKS107	AD-Sla2309-720	This study
pKS108	AD-Nup49	This study
pKS109	AD-Nup421-379	This study
pKS186	GST-Spc723-433	This study
pKS187	GST-Spc723-433*	This study
pKS188	GST-Spc723-622	This study
pKS189	GST-Spc723-622*	This study
pKS192	AD-Spc723-433*	This study
pKS193	AD-Spc72	This study
pKS194	AD-Spc72*	This study
pKS195	NLS-Spc72-NES-GFP2	This study
pKS197	GST-Spc723-96	This study
pKS198	GST-Spc723-129	This study
pKS199	GST-Spc723-163	This study
pKS202	GST-Spc723-398	This study
pKS223	MET3-Spc72-GFP	This study
pKS224	MET3-Spc72*-GFP	This study
pKS225	NLS-CtermSpc72-GFP2	This study
pKS226	NLS-CtermSpc72*-GFP2	This study
pKS227	MET3-NtermSpc72-GFP	This study
pKS228	MET3-NtermSpc72*-GFP	This study
pKW430	NLS-PKI-NES-GFP2	62
pKW435	pBS-xpo1::LEU2	62
pKW440	pRS313-XPO1	62
pKW457	pRS313-xpo1-1	62
pKW581	His6-Gsp1	42
pMK15	AD-Spc97	37
pMK151	AD-Spc1101-204	37
pSG26	AD-Spc98	37
pSG46	AD-Tub4	37

^aMutations and the amino acid positions included in protein fragments are shown as superscripts. Plasmids containing two moieties of GFP are marked as GFP₂.

^bUCSF, University of California, San Francisco.

Yeast strains, PCR-based gene tagging, and gene modification.

PCR-based C-terminal green fluorescent protein (GFP) tagging of XPO1 was done according to Longtine et al. (43) using either plasmid pFA6a-GFP-TRP1 (GSY835) or pFA6a-GFP-His3MX6 (tetrads from K842 background, resulting in KSY460 and KSY461, respectively) as a template. Yellow fluorescent protein (YFP) tagging of XPO1 was done using plasmid pDH5 (Yeast Resource Center [YRC]) as a template and transforming the PCR product into CRY1 (MATa ade2-1oc can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1) (YRC), resulting in strain KSY592 [Xpo1-YFP (HIS3) MATa ade2-1oc can1-100 his3-11,15 leu2-3,112 trp1-1 ura3-1]. This strain was crossed to DHY45A [Spc29-cyan fluorescent protein (CFP) (KanMX) MATα ade2-loc ade3Δ100 can1-100 his3-11,15 leu2-3,112 ura3-1 TRP] (YRC), resulting in the diploid KSY456. C-terminal tagging of Spc72 with GFP was done in a DF5 diploid using pFA6a-GFP-KanMX6 as a template, resulting in KSY581. Expression of GFP-fused Spc72 was confirmed under the microscope, and tetrads were dissected, resulting in strain KSY457 (Spc72-GFP). To obtain a GFP-tagged NES-mutated version of Spc72, DNA from KSY457 was isolated and used for a PCR containing a mutagenic forward primer resulting in KSY458 (Spc72*-GFP). For the construction of XPO1 alleles with a GFP-tagged Spc72, KSY581 was transformed with a plasmid containing an xpo1::LEU2 deletion construct (pKW435) (64). Correct integration at the XPO1 locus was confirmed by PCR and Southern blotting. Subsequently, plasmids with XPO1 (pKW440), xpo1-1 (pKW457), and xpo1-101 (pKS63) alleles were transformed into these cells, and after sporulation, tetrads were isolated, resulting in strains KSY582, KSY583, and KSY584, respectively. To obtain SPC72 strains in which the NES was mutated, a PCR was set up using mutagenic primer sets and wild-type (WT) genomic DNA as a template. In a parallel reaction, the KanMX6 cassette was amplified from pFA6a-KanMX6. In a second PCR, products from the first PCRs were combined and amplified using a third set of primers. The resulting PCR product was gel purified, ethanol precipitated, and transformed into diploid yeast cells (DF5). Subsequently, cells were plated onto YPD-G418 plates (200 mg/liter geneticin). All PCR products obtained in this way were subcloned and subsequently sequenced to confirm the respective mutations. In all cases, strains were checked by PCR and Southern blotting for correct integration events. Strains testing positive on two successive rounds on YPD-G418 plates were used for targeted integration of pASF125 (67) and subsequent tetrad analysis, resulting in strains KSY462 (SPC72 with GFP-Tub1) and KSY463 (spc72* with GFP-Tub1). WT and XPO1 strains with GFP-tagged tubulin were obtained by targeted integration of pASF125 in JD-4713c (65), XPO1 (KWY120 [64]), xpo1-1 (KWY121 [64]), and xpo1-101 (KSY447), resulting in strains KSY318, KSY453, KSY454, and KSY455, respectively. For K842-derived WT and spc72Δ strains with GFP-Tub1, pASF125 was integrated into KSY361 (K842; MATa/α SPC72/spc72::klTRP1; ade2-1 can1-100 leu2-3,112 his3-11,15 ura3 GAL [psi⁺] ssd1-d2) and tetrad analysis was done, resulting in KSY451 (SPC72 with GFP-Tub1) and KSY452 (spc72Δ with GFP-Tub1). For the expression of Spc72-GFP versions under the MET3 promoter, KSY361 was transformed with either pKS223 or pKS224. After sporulation, tetrads were dissected, and Ura⁺ Trp⁺ strains were isolated, resulting in KSY544 and KSY545, respectively.

Two-hybrid analyzes.

Two-hybrid screening was performed as described previously (34). The bait plasmid pIA204-1 (pGBD-Xpo1, where pGBD is a GAL4 binding domain [BD] vector) was a gift of Pascal Preker (University of San Francisco, San Francisco, CA). pIA204-1 and a yeast genomic DNA library on pGAD (a GAL4 activation domain [AD] vector; Elizabeth Craig, University of Wisconsin, Madison, WI) were transformed separately into PJ69-4a (MATa trp1-901 leu2-3,112 ura3-52 his3-200 gal4Δ gal80Δ LYS2::GAL1-HIS3 GAL2-ADE2 met2::GAL7 lacZ), resulting in 3 × 10⁶ transformants representing five copies of the yeast genome. After 10 days at 30°C on SD reporter plates lacking Trp, Leu, and His (SD-Trp⁻-Leu⁻-His⁻), 264 colonies were isolated. Only 62 of these grew on SD-Trp⁻-Leu⁻-Ade⁻ medium. Plasmids were rescued from these strains and retransformed into strain PJ69-4a containing plasmid pIA204-1. Thirty-two plasmids that tested positive for interaction with pIA204-1 were subsequently sequenced. For the experiments shown in Fig. 1, interactions were tested on SD-Trp⁻-Leu⁻-His⁻reporter plates. For the experiments in Fig. 4C, yeast cells were transformed with the plasmid combinations indicated in the figure. Plasmid-dependent growth was tested on SD-Trp⁻-Leu⁻ plates; interaction was checked on SD-Trp⁻-Leu⁻-His⁻reporter medium.

FIG. 1.

Xpo1 interacts with the SPB proteins Spc72 and Spc110 by the yeast two-hybrid method. (A) Summary of the results obtained in the yeast two-hybrid screen. The following proteins or protein fragments were identified as Gal4-activation domain fusions in the screen: Yap (residues 184 to 640), Nup42 (1 to 379), Nup49, Sgm1 (218 to 707), Sla2 (309 to 720), Spc72 (3 to 433), Ltv1 (55 to 463), and Hpa3. No interaction can be detected between Gal4-AD and BD-Xpo1 in the empty vector control. For simplicity, gene names rather than protein fragment length were used. Interactions were tested on SD-Trp⁻-Leu⁻-His⁻ plates. (B) Xpo1 interacts with the N terminus of Spc110 but not with components of the γ-tubulin complex. Gal4-AD fusions of full-length Spc97, Spc98, and Tub4 show no interaction when tested against BD-Xpo1. Gal4-AD-Spc110 comprises amino acids 1 to 204 from the N terminus of this protein (Spc110²⁰⁴). For Gal4-AD fusions of Spc72, a 271-amino-acid (residues 1 to 271; Spc72²⁷¹) and a 430-amino-acid (residues 3 to 433; Spc72⁴³³) fragment of the N terminus were tested. Interactions were tested on SD-Trp⁻-Leu⁻-His⁻ plates.

FIG. 4.

Spc72⁴³³ contains a leucine-rich NES-like amino acid motif at position 418 to 429 that mediates binding to Xpo1 in vitro and in vivo. (A) Schematic drawing of the full-length Spc72 sequence with known binding partners (black bars) and 12 putative NES-like motifs (black squares) indicated. Regions shown in dark gray indicate putative coiled-coil motifs. Spc72 fragments of indicated lengths were expressed as GST fusions in E. coli and used in the binding assays summarized in panel B. (B) Binding assay with Spc72 fragments. Binding assays were done exactly as described in the legend of Fig. 3A, and only binding to the beads is shown. GST fusions of Spc72 fragments of indicated lengths (lanes 3 to 8; 18 μg each) were bound to GSH beads and incubated with Gsp1-GTP (15 μg) and Xpo1 (12 μg). Lane 8 contains an NES-mutated version of GST-Spc72⁴³³ (Spc72⁴³³*). As positive and negative controls for binding, GST-PKI-NES (18 μg; lane 1) and GST alone (18 μg; lane 2) are shown. (C) Yeast two-hybrid analysis of Spc72 fragments and full-length proteins. PJ69-4a cells were transformed with either pGBD or pGBD-Xpo1, respectively, and a WT (pGAD-Spc72⁴³³) or NES-mutated (pGAD-Spc72⁴³³*) version of Spc72⁴³³. pGAD-Spc72 and NES-mutated pGAD-Spc72* were also tested. Interactions were tested on SD-Trp⁻-Leu⁻-His⁻ medium (SD −TLH). Plasmid-dependent growth was controlled on SD-Trp⁻-Leu⁻ medium (SD −TL).

In vitro binding assays.

The purification of His₆-Gsp1 (yeast Ran) and Xpo1-His₆ from Escherichia coli was done as described previously (44). The loading of Gsp1 proteins with either GDP or GTP and subsequent purification has been described previously (5). Spc72 expression vectors were constructed as follows: the 430-amino-acid fragment of Spc72, residues 3 to 433 (Spc72⁴³³) was subcloned from the pGAD vector into pGEX-6P-1 (Amersham Biosciences). To produce shorter versions of this protein, stop codons were introduced at the positions indicated in Fig. 4A by PCR mutagenesis using a QuikChange Kit (Stratagene). The same strategy was used to mutate the NES sequence at position 418 to 429 from LEKQINDLQIDK to LEKQINDAQADK. For expression of full-length Spc72, a PCR was set up using WT genomic DNA as a template. The resulting PCR product was used to replace Spc72⁴³³ in pGEX-6P-1. Mutation of the NES was achieved by site-directed mutagenesis. All glutathione (GSH) S-transferase (GST) fusion proteins were purified using GSH-Sepharose beads (Amersham Biosciences). Solution binding assays were done as described previously (44). WT and mutant PKI-NES peptides (64) used in competition assays were from Biosynthan.

Microscopy.

Yeast cells with GFP-, YFP-, and CFP-tagged proteins were analyzed in vivo with an Axioplan II microscope equipped with an Axiocam digital camera and Axiovison III software (Zeiss) using appropriate filter setups. Processing of images was done using Adobe Photoshop, version 7. For CFP-YFP colocalization in the presence of Gsp1 mutants, strains were grown in synthetic minimal medium lacking uracil (SD-Ura⁻) and containing galactose to select for the presence of plasmids and to induce protein expression. For each Gsp1 mutant, 100 random cells showing an Spc29-CFP signal were checked for the presence or absence of Xpo1-YFP at precisely the same spatial location in each cell. Results from three independent experiments were combined. To visualize nuclei, 1 μl of DAPI (4′,6′-diamidino-2-phenylindole; 1 mg/ml) was added to 1 ml of culture volume. To test the localization of GFP-Tub1 (KSY318) and Xpo1-GFP (KSY68) (64) in the presence or absence of nocodazole, strains were grown at 25°C to mid-log phase in SD-Ura⁻ medium. Cultures were split in half, and either dimethyl sulfoxide alone or nocodazole in dimethyl sulfoxide was added to a final concentration of 30 μg/ml for 13 h according to the method of Amon (3). Expression of either WT or NES-mutated versions of Spc72-GFP under the control of the MET3 promoter was achieved as follows. First, yeast cells (strains KSY544 and KSY545) were grown to mid-log phase in SD-Ura⁻ medium supplemented with 1 mM methionine at 25°C to repress Spc72-GFP expression from plasmids pKS223 or pKS224, respectively. Subsequently, cells were washed into fresh medium lacking methionine for 1 h at 25°C to induce Spc72-GFP expression. Prolonged incubation under these conditions (more than 60 min) led to aggregate formation. Experiments with temperature shifts were done as described previously (64). For fluorescence intensity quantification of SPBs in strains KSY457 and KSY458, yeast were grown to mid-log phase at 25°C. For each strain, small budded cells with a GFP signal at the SPB were chosen, and a z-stack of nine sections at 0.2-μm intervals was captured for each individual cell. Only stacks covering the whole GFP signal (at least 30 stacks per strain) were included in the analysis. ImageJ software (1) was used to sum all nine sections of each stack. The integrated density around the SPB (circular region of interest; 21-pixel diameter) was recorded. For each cell, a corresponding background intensity was recorded and subtracted from the SPB intensity. The absolute values for the WT strains were set to 100 arbitrary units. For fluorescence intensity quantification of the SPB in yeast strains KSY582, KSY583, and KSY584, strains were grown to mid-log phase at 25°C, shifted to 37°C for 15 min, and analyzed within 10 min thereafter.

RESULTS

The SPB protein Spc72 is a novel interaction partner for Xpo1.

In an effort to identify interaction partners for the exportin Xpo1, we performed a yeast two-hybrid screen using Xpo1 as a bait. In total, 32 plasmids were found to support growth on reporter medium in a bait-dependent fashion. These plasmids were sequenced and contained Gal4 activation domain in-frame fusions of the following open reading frames or fragments thereof: Nup42 (YDR192C; 11 plasmids), Yap1 (YML007W; 8 plasmids), Sgm1 (YJR134C; 4 plasmids), Spc72 (YAL047C; 2 plasmids), Hpa3 (YEL066W; 3 plasmids), Ltv1 (YKL143W; 2 plasmids), Sla2 (YNL243W; 1 plasmid) and Nup49 (YGL172W; 1 plasmid). A summary of the interactions found in the screen is shown in Fig. 1A. Three previously known Xpo1 interaction partners (Yap1, Nup42, and Nup49) were identified, demonstrating the usefulness of the two-hybrid method for this purpose. Yap1 contains a nuclear export sequence of the leucine-rich type and acts as an export substrate for Xpo1 (76) whereas Nup42 and Nup49 (48) are structural components of the nuclear pore complex (NPC) through which all nucleocytoplasmic trafficking occurs (18, 69). Sgm1, Sla2, Spc72, Ltv1, and Hpa3 represent novel interaction partners for Xpo1. These proteins are involved in a variety of cellular processes such as membrane cytoskeleton assembly (Sla2 [32]), mitotic spindle assembly (Spc72 [35]), and chromatin modification (Hpa3 [77]). For Ltv1, a role in the Xpo1-mediated export of small ribosomal subunits has recently been shown (59). No function has been assigned to Sgm1 so far, and with the exception of Ltv1, none of the proteins found in our screen has been analyzed with respect to a possible function in nuclear export. Although Sgm1 and Spc72 have been identified in other two-hybrid screens as Xpo1 interactors, they were not analyzed further (33, 36).

In higher eukaryotes, the Ran-GTPase system and certain transport receptors, apart from their well-established function in nuclear transport processes, were recently shown to be involved in mitotic spindle assembly (10, 47, 73, 74). Spc72 was particularly interesting since it is a cytoplasmic protein that functions as a structural component of the yeast SPB. This supramolecular protein complex is embedded in the nuclear membrane and is required to nucleate cytoplasmic and nuclear MTs for mitotic spindle assembly (35). During cytoplasmic MT nucleation, Spc72 interacts with the γ-tubulin complex consisting of Spc97, Spc98, and Tub4 (39). We wanted to test whether these proteins would also interact with Xpo1 in the yeast two-hybrid system. To this end, we performed a directed two-hybrid analysis with plasmids obtained from the Knob and Schiebel laboratory (39). Figure 1B shows the result of this experiment: the three subunits of the γ-tubulin complex, Spc97, Spc98, and Tub4, do not interact with Xpo1. However, we find that the N terminus of Spc110 (amino acids 1 to 204), another protein of the SPB, does indeed interact with Xpo1. Interestingly, a shorter version of the N terminus of Spc72 (amino acids 1 to 271 [Spc72²⁷¹]) fails to interact with Xpo1. Thus, in the two-hybrid assay Xpo1 interacts with at least two components of the yeast SPB, Spc72 and Spc110, which are required to nucleate MTs of the mitotic spindle.

Xpo1 is localized at the SPB in vivo.

In higher eukaryotic cells, the receptor for leucine-rich NESs, Crm1, localizes to the nuclear interior and at NPCs, in line with its proposed function as an exportin mediating nuclear export of substrate proteins (24). More recently, Crm1 was shown to bind to kinetochores as well as centrosomes, two high-molecular-weight protein complexes which in concert with chromatin organize assembly of the mitotic spindle (4, 22). Since Xpo1 is the yeast orthologue of Crm1 and since the SPB is the functional equivalent of centrosomes in yeast, we wanted to know whether the two-hybrid data obtained for Spc72 and Spc110 would hint at a role of Xpo1 in the SPB, apart from its well-established function in nuclear export. In an effort to address this issue, we genomically tagged XPO1 with GFP and localized the fusion protein in living yeast. The upper panels in Fig. 2A show that Xpo1-GFP, in addition to its expected presence in the nucleus and at the nuclear envelope, can be detected at distinct foci at the nuclear periphery, identical to the localization pattern of SPBs. The same localization pattern is observed for a mutant allele of XPO1 (Xpo1-1-GFP). We also tested whether Xpo1 localization is MT dependent. To this end, experiments were performed in the presence and absence of nococdazole, an MT-depolymerizing drug. Even after prolonged incubation with nocodazole, Xpo1-GFP signals are detected in SPB-like dots, indicating that its localization pattern is not MT dependent (see Fig. S2 in the supplemental material). Additionally, we performed a double labeling experiment, depicted in the lower row of Fig. 2A, using chromosomally tagged Xpo1-YFP and Spc29-CFP, a protein of the central plaque of the SPB (16). In this experiment, the two fluorescence signals overlap (Fig. 2A, Merge), suggesting the presence of some Xpo1 at SPBs. To test if Spc72 was the binding partner of Xpo1 at SPBs, a yeast strain with a deletion of SPC72 (spc72Δ) was GFP tagged at the XPO1 locus. As can be seen in Fig. 2B, lower panel, Spc72 is not required for Xpo1 binding at SPBs since the Xpo1-GFP signal is still detected at distinct dots at the nuclear rim in spc72Δ cells. This suggests that Xpo1 might have other binding partners at the SPB.

FIG. 2.

Some fraction of Xpo1 localizes at the SPB in vivo. (A) Xpo1-GFP (strain GSY835) and Xpo1-1-GFP (KSY69 [64]) localize at distinct dots at the nuclear periphery (top). Xpo1-YFP and Spc29-CFP colocalize (Merge) at the SPB (KSY456) (bottom). Bar, 5 μm. (B) Xpo1-GFP is not lost from the SPB in an spc72Δ strain. Shown is a comparison of Xpo1-GFP localization in SPC72 (KSY460) and an spc72Δ mutant (KSY461). In the spc72Δ strain, two nuclei are present in one cell, indicative of a defect in nuclear migration (31). Bar, 5 μm. (C) Perturbations in nuclear Gsp1-GTP concentration influence binding of Xpo1-YFP at the SPB. Plasmid-borne WT (Gsp1) and mutant forms of Gsp1 locked in either the GTP-bound (Gsp1-G21V) or nucleotide-free (Gsp1-T26N) state, respectively, were overexpressed from the GAL promoter in KSY456. A total of 100 random cells showing an Spc29-CFP signal were inspected for the presence or absence of Xpo1-YFP at the SPB, as described in the Materials and Methods section. Error bars (standard deviations) are shown. DIC, differential interference contrast optics.

At the NPC, some nuclear transport receptors interact with nucleoporins via a specific motif found in these nucleoporins, the FG repeat (18). For SPB proteins, FG repeats have not been reported, and it is therefore tempting to speculate that Xpo1 targeting to SPBs is mediated by an NES in one or more of the SPB proteins. Thus, it can be predicted that Xpo1 should bind to SPBs only in the presence of Gsp1-GTP since Xpo1 binds its NES-containing cargoes in a cooperative fashion with Gsp1-GTP (40, 44). To test this hypothesis, we analyzed the colocalization of Xpo1-YFP with Spc29-CFP in the presence of WT Gsp1 and two Gsp1 mutants, Gsp1-G21V and Gsp1-T26N, overexpressed from plasmids in vivo (57). All these versions of Gsp1 accumulate in the nucleus of WT cells; each, however, has a different GTP status. Whereas Gsp1 can bind and hydrolyze GTP normally, Gsp1-G21V is locked in the GTP-bound state. Gsp1-T26N, on the other hand, has a very low affinity for nucleotides and inhibits the Ran cycle via its tight interaction with the RanGEF Rcc1/Prp20 (14). Cells in which Gsp1-G21V is overexpressed from the GAL promoter are viable but show a cytoplasmic mislocalization of nuclear proteins. In addition, these cells have an export defect for poly(A)⁺ RNA, indicative of an inability to hydrolyze GTP by Gsp1-G21V, which is required for normal nuclear transport (57). In contrast, cells with a WT version of Gsp1 show a normal distribution of poly(A)⁺ RNA and nuclear proteins, suggesting that overproduction of Gsp1 is not toxic per se. As is shown in Fig. 2C, 87% (± 5%) of the cells colocalize Xpo1-YFP with Spc29-CFP in the presence of overexpressed WT Gsp1 whereas some Xpo1-YFP signal is lost from the dot at the nuclear rim in the presence of GTP-locked Gsp1-G21V mutant (81% ± 4.8% cells with both proteins colocalizing). However, only 60% (± 6.7%) of the cells show Xpo1-YFP/Spc29-CFP colocalization in the presence of Gsp1-T26N, indicating that binding of Xpo1-GFP at the SPB is indeed dependent on the GTP status of the cell.

Taken together, our in vivo localization data demonstrate that a subpopulation of Xpo1 is localized at SPBs. Since Xpo1-GFP is not lost from SPBs in an spc72Δ mutant, binding must be dependent on one or more of the other SPB components. This binding could be NES dependent since SPB localization of Xpo1 occurs in a Gsp1-GTP-dependent fashion.

Xpo1 binds to a subfragment of Spc72 in a Gsp1-GTP- and NES-dependent manner in vitro.

In combination, the two-hybrid data and the in vivo localization data suggested that a subpopulation of Xpo1 is localized at the SPB and that Spc72 might be one of several interaction partners at this organelle. Since Xpo1 localization is modulated by Gsp1, its SPB association could be dependent on an NES-containing SPB protein. By visual inspection, the Spc72⁴³³ sequence contains several amino acid stretches which resemble leucine-rich NESs. We therefore wanted to test the possibility that Spc72⁴³³ might bind to Xpo1 via one or more of these putative NESs. In the cell, the binding of Ran-GTP and an NES-containing protein to Xpo1 is highly cooperative, and accessory factors support complex formation (17, 42, 49, 68). However, most NESs are rather weak, and the isolation of trimeric complexes from cellular extracts has been technically challenging (40, 41, 44). To circumvent these problems, we expressed the fragment of Spc72 identified in the two-hybrid screen in bacteria and performed in vitro binding assays in the presence of recombinant Gsp1 and Xpo1. In this type of experiment, Xpo1 binds to its substrates in a Gsp1-GTP- and NES-dependent fashion without the addition of accessory proteins. Moreover, this binding can be competed by an excess of NES peptide but not mutated versions thereof (44). The left panel in Fig. 3A shows a control experiment in which a fusion protein between GST and the NES sequence from PKI was bound to GSH beads. Only in the presence of Gsp1-GTP can recombinant Xpo1 bind to the GST-NES beads (lane 5), whereas no binding is observed in the absence of Gsp1 (lane 1) or in the presence of Gsp1-GDP (Fig. 3A, lane 3). In the experiment shown in the right panel of Fig. 3A, an N-terminal fragment of Spc72 (Spc72⁴³³) was expressed as a GST fusion and bound to GSH-Sepharose beads. In the presence of Gsp1-GTP, recombinant Xpo1 can bind to the GST-Spc72⁴³³ beads (lane 9), whereas no binding is observed in the Gsp1-GDP control lane (lane 7). To test whether excess amounts of NES peptide can compete binding of Xpo1 to GST-Spc72⁴³³, we next included a synthetic PKI-NES peptide in the binding reaction. As can be seen in Fig. 3B, binding of Xpo1 to GST-Spc72⁴³³ is reduced when additional NES peptide is present (compare lanes 1 and 3). However, this binding is not reduced when a mutated, nonfunctional version of the PKI-NES peptide (64) is present in the reaction (lanes 5 and 6). From these experiments we conclude that Xpo1 binds to Spc72⁴³³ in a Gsp1-GTP- and NES-dependent fashion, which indicates that these proteins can form a typical export complex.

FIG. 3.

Xpo1 binds a fragment of Spc72 in a Gsp1- and NES-dependent manner in vitro. (A) GST-NES as a control or GST-Spc72⁴³³ (GST fused to amino acids 3 to 433 of Spc72) (18 μg per reaction) was immobilized on GSH-Sepharose and incubated for 30 min at 4°C with 15 μg of recombinant Gsp1-GDP and Gsp1-GTP and 12 μg of Xpo1 as indicated. After binding, the supernatant (S) was precipitated with trichloroacetic acid and resuspended in sodium dodecyl sulfate sample buffer. Bound material was washed three times and subsequently eluted from the beads (B) with sodium dodecyl sulfate sample buffer. All samples were analyzed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and Coomassie blue staining. (B) As in panel A, GST-Spc72⁴³³ was immobilized on GSH beads and incubated with Gsp1-GTP and Xpo1. In addition, synthetic peptides comprising either the WT PKI-NES (89 μg) or a mutated version thereof (NES*; 85 μg) were included in the binding reaction. Samples were further processed as described in panel A. M, molecular mass (in kilodaltons).

Spc72 contains an NES.

The in vitro experiments shown in Fig. 3A and B suggested that Spc72 could act as an export substrate for Xpo1. A prerequisite for such an interaction is a leucine-rich NES. Figure 4A shows a schematic drawing of the complete Spc72 sequence with mapped binding sites for known interaction partners (29, 39, 70) and 12 short leucine-rich amino acid stretches which potentially could serve as NESs. To identify one or more NES(s) in Spc72⁴³³, we expressed the fragments of the protein depicted in Fig. 4A and tested for binding to Xpo1 in the in vitro binding assay. Figure 4B summarizes the results of these experiments: only the longest fragment of Spc72 comprising amino acids 3 to 433 (Spc72⁴³³) can bind to Xpo1 in the presence of Gsp1-GTP (lane 7). All shorter versions fail to bind to Xpo1 in the presence of Gsp1-GTP (Fig. 4B, lanes 3 to 6; compare with the GST-NES control in lane 1 and GST-only control in lane 2), indicating that it is the most-C-terminal NES-like sequence between amino acids 418 to 429 in Spc72⁴³³ that is needed for binding to Xpo1. LEKQINDLQIDK_418-429 conforms to the NES consensus deduced from a variety of NES-containing proteins (30, 41).

If a stretch of amino acids is predicted to function as an NES, it should lose this function when mutated at key residues of the consensus (30). We therefore introduced leucine/isoleucine-to-alanine mutations at positions 425 and 427, resulting in LEKQINDAQADK_418-429 and tested binding of this NES-mutated Spc72⁴³³ (Spc⁴³³*) to Xpo1 in the presence of Gsp1-GTP. As is shown in Fig. 4B, lane 8, binding to Xpo1 is lost when Spc72⁴³³ is mutated at positions 425 and 427, indicating that this sequence between residues 418 and 429 can indeed function as an NES recognized by Xpo1. We also expressed full-length WT and NES-mutated Spc72 in bacteria although this was technically difficult due to the tendency of Spc72 to form aggregates. In binding experiments, these proteins gave inconsistent binding patterns (data not shown). To circumvent the problems with recombinant full-length Spc72, we performed a directed yeast two-hybrid analysis to test for an interaction with Xpo1. As can be seen in Fig. 4C, Xpo1 interacts only with full-length Spc72 with a WT NES (Fig. 4C, upper row, pGBD-Xpo1/pGAD-Spc72) but not when this sequence is mutated (Fig. 4C, upper row, pGBD-Xpo1/pGAD-Spc72*), indicating that the NES at position 418 to 429 (NES_418-429) is the only functional NES in Spc72. Shorter constructs with Spc72⁴³³ and Spc72⁴³³* were also included in the analysis. Positive Xpo1/Spc72⁴³³ interaction (pGBD-Xpo1/pGAD-Spc72⁴³³) is lost when the NES is mutated (pGBD-Xpo1/pGAD-Spc72⁴³³*), confirming the results obtained in the in vitro binding assays (Fig. 4B). No interaction is seen with the empty vector control (pGBD/pGADSpc72⁴³³). Taken together, our results indicate that amino acids LEKQINDLQIDK_418-429 in Spc72 can function as an NES in vitro and in vivo.

NES_418-429 has nuclear export activity in vivo.

To test whether LEKQINDLQIDK_418-429 had any export function in vivo, we used a GFP reporter assay (64). In this assay, a tandem GFP is fused to an NLS and an NES derived from the protein of interest, for example, PKI (72). In WT cells, this fusion protein shuttles between the nucleus and the cytoplasm and hence can be localized to both compartments. However, when this protein is expressed in xpo1 mutants in which nuclear export is substantially slowed down compared to WT cells, the protein accumulates in the nucleus (64). Figure 5A shows a comparison of XPO1 and xpo1-1 cells expressing two versions of the GFP reporter protein at 37°C. As can be seen in the upper row of Fig. 5A, the NLS-PKI-NES-GFP reporter is detected in the cytoplasm and nucleus of WT cells even at 37°C. The reporter construct in which the PKI-NES was replaced by Spc72 LEKQINDLQIDK_418-429 also localizes to the nucleus and cytoplasm when expressed in XPO1 cells (Fig. 5A, upper row, right panels). However, both GFP reporter proteins accumulate in the nucleus of the xpo1-1 strain when cells are incubated at the nonpermissive temperature (Fig. 5A, lower row). This indicates that LEKQINDLQIDK_418-429 has an export activity in vivo.

FIG. 5.

The leucine-rich NES-like sequence at position 418 to 429 of Spc72 mediates nuclear export in vivo. (A) Summary of the results obtained from a nuclear export reporter assay in vivo. The NES sequence from PKI (NLS-PKI-NES-GFP; pKW430) and the putative Spc72 NES LEKQINDLQIDK_418-429 (NLS-SPC72-NES-GFP; pKS195) were expressed as GFP fusions in living yeast. In XPO1 cells (KWY120), due to the NES and NLS, the proteins shuttle between the nucleus and cytoplasm at 37°C and can be localized to both compartments (GFP, upper row). However, in the xpo1-1 temperature-sensitive mutant (KWY121), export activity is slowed down substantially at 37°C (GFP, lower panels), and therefore even in the presence of an intact NES, the reporter proteins accumulate in the nucleus. Bar, 5 μm. (B) Localization of GFP-tagged N and C termini of Spc72. The N-terminal 433 amino acids of Spc72 (NtermSPC72-GFP) were fused to GFP in either their WT (NES; pKS227) or NES-mutated (NES*; pKS228) version. The Spc72 C terminus comprising amino acids 396 to 622 (NLS-CtermSPC72-GFP) was fused to two moieties of GFP either in its WT (NES; pKS225) or NES-mutated (NES*; pKS226) version. In addition, the C-terminal fusion proteins contain an artificial NLS sequence to allow for nuclear accumulation. All fusion proteins were expressed in cells WT for XPO1 and SPC72 (KSY57) at 25°C. Bar, 5 μm. DIC, differential interference contrast optics.

We also wanted to test NES_418-429 in its native protein context. However, the amount of Spc72-GFP when expressed from its endogenous promoter was rather low (Fig. 6), making it impossible to detect any non-SPB-bound Spc72-GFP which might have accumulated in the nucleus, as would be expected for xpo1 mutants. On the other hand, use of regulatable promoters such as GAL1-10 and MET3 leads to formation of aggregates when Spc72-GFP is overexpressed (63; see also Fig. S5 in the supplemental material). We therefore expressed the C- and N-terminal halves of Spc72 individually as GFP fusions. All fusion proteins are larger than 70 kDa in size to prevent passive diffusion through the NPC. First, we expressed the C terminus of Spc72 (amino acids 396 to 622 [NLS-CtermSPC72-GFP]) in its WT (NES) and NES-mutated version (NES*) as NLS-GFP fusion proteins (Fig. 5B). An artificial NLS was included to allow for shuttling of the proteins between the nuclear and cytoplasmic compartments. Whereas the WT C terminus (NES) is localized to the nucleus, cytoplasm, and nuclear periphery in WT cells, its NES-mutated version (NES*) is completely targeted to the nucleus due to the artificial NLS sequence. This indicates that NES_418-429 can function as an NES within the C-terminal half of Spc72. In a similar experiment, we fused the N terminus of Spc72 (amino acids 1 to 433 [NtermSPC72-GFP]) including either a WT (NES) or a mutated NES_418-429 (NES*) to GFP. In this construct, no artificial NLS is present. As can be seen in Fig. 5B (right panel, upper row), Nterm-SPC72-GFP localizes to the cytoplasm of WT cells, with some GFP dots at the nuclear periphery which could be SPBs. The nucleus is completely devoid of any GFP signal (nuclear exclusion) (Fig. 5B, right, compare DAPI and GFP panels). However, when NES_418-429 is mutated, a significant proportion of NtermSPC72-GFP (NES*) becomes nuclear (Fig. 5B; DAPI and GFP signals overlap). Since in this case no artificial NLS was included in the fusion protein, this indicates that the N-terminal part of Spc72 can be imported into the nucleus. Whether this nuclear localization is mediated by an endogenous NLS in Spc72 or whether the protein is imported via an adaptor (piggyback mechanism) is unknown at present. But in any case, our data indicate that Spc72 has a functional NES_418-429 in vivo.

FIG. 6.

Mutations in the NES sequence of Spc72 influence its binding to the SPB and affect spindle morphology in vivo. (A) Localization of WT (Spc72-GFP) and NES-mutated Spc72-GFP (Spc72*-GFP) in vivo. The upper row shows z-axis plane fluorescence images of a WT yeast cell in which SPC72 is genomically tagged with GFP (KSY457). The two lower rows show individual cells of a strain (KSY458) in which the NES_418-429 of Spc72-GFP has been mutated as described in Materials and Methods. Exposure time was the same for all samples. For comparison, the position of the cell nucleus is indicated by DAPI staining. Bar, 5 μm. (B) Fluorescence intensity quantification of Spc72-GFP at the SPB. Haploid strains of SPC72 and XPO1 alleles were grown to mid-log phase and processed for fluorescence intensity quantification as described in Materials and Methods. The left panel compares the GFP fluorescence intensity at the SPB of a WT (Spc72-GFP; strain KSY457) and an NES-mutated (Spc72*-GFP; strain KSY458) version of Spc72-GFP. The right panel shows a similar experiment in which Spc72-GFP fluorescence was measured at SPBs of three different XPO1 alleles: WT (KSY582), xpo1-1 (KSY583), and xpo1-101 (KSY584). The mean intensity ± standard deviation of at least 30 SPBs is shown for each strain. (C) The formation of cytoplasmic MT (cyt. MT) is impaired in an spc72* mutant in vivo. For SPC72 WT (KSY462), spc72* mutant (spc72*; KSY463), another WT (KSY451), and spc72Δ strains (KSY452), a GFP-tagged copy of TUB1 was integrated at the URA locus. In each experiment, 200 cells in the G₁ phase of the cell cycle (see example cells) were visually inspected for the presence or absence of cytoplasmic MTs as described previously (58). Error bars (standard deviations) are shown. Size bar, 5 μm. (D) Formation of cytoplasmic MTs is reduced in XPO1 alleles in vivo. GFP tagging of TUB1 in XPO1 (KSY453), xpo1-1 (KSY454), and xpo1-101 (KSY455); temperature shifts; and cell classification were done as described in Materials and Methods. DIC, differential interference contrast optics.

Mutations in SPC72 and XPO1 result in a defect in cytoplasmic MT formation in vivo.

To test whether NES_418-429 had any impact on the functions of Spc72 in vivo, we generated yeast strains carrying either a WT (Spc72-GFP) or an NES-mutated GFP-tagged version of Spc72 (Spc72*-GFP) under the control of its own promoter. Under normal laboratory conditions, no difference in growth could be observed. Figure 6A shows that WT Spc72-GFP localizes at a distinct dot at the nuclear periphery, in accordance with its proposed function as an SPB protein (upper row). In a strain carrying a mutated NES_418-429 (Fig. 6A, Spc72*-GFP), a strongly reduced GFP signal was visible at the nuclear periphery with a faint increase of GFP fluorescence in the cytoplasm, suggesting that despite its mutated NES_418-429, Spc72*-GFP might still be localized to the cytoplasm but might no longer be able to associate efficiently with SPBs. We used fluorescence intensity quantification to test this possibility. The left panel in Fig. 6B shows the result of this analysis: for Spc72*-GFP, the fluorescence intensity (in arbitrary units) is only half (52.8 ± 12.5) of what can be observed for WT Spc72-GFP (100 ± 15.3), indicating a loss of fluorescence from the SPB. A similar experiment was performed with a WT and two conditional xpo1 alleles. In these strains, Spc72-GFP also localizes to the SPB (data not shown). To exclude the possibility of an allele-specific defect of xpo1-1, we included the xpo1-101 allele in the analysis. This novel xpo1 allele has a defect in nuclear export but carries a different mutation than xpo1-1 (A. Neuber and K. Stade, unpublished data). The right panel in Fig. 6B shows that, compared to a WT XPO1 strain (100 ± 15.9), Spc72-GFP fluorescence is significantly reduced in xpo1-1 (68.5 ± 13.0) and xpo1-101 cells (67.7 ± 15.0), respectively. By immunoblot analysis, the amounts of Spc72-GFP and Spc72*-GFP were comparable, indicating that the loss of GFP fluorescence from the nuclear periphery did not result in degradation of the protein (data not shown). We therefore reasoned that if the amount of NES-mutated Spc72 is reduced at SPBs, the nucleation of cytoplasmic MTs and, hence, mitotic spindle formation should be impaired, as is the case in an spc72 deletion strain (31, 63). To address this question, we integrated a GFP-tagged copy of TUB1 at the URA locus in SPC72 and spc72* strains and analyzed the extent to which cytoplasmic MTs could be formed on SPBs. In the G₁ phase of the cell cycle, the SPB usually appears as a single dot with MTs emanating from it into the cytoplasm (58, 75). Between 150 and 200 cells in the G₁ phase were selected and inspected for the presence or absence of MT asters (Fig. 6C, GFP-Tub1, top panel). As a control, we included an spc72 deletion strain in our analysis which is known to lack a substantial amount of cytoplasmic MTs (31, 63). As can be seen in Fig. 6C, formation of cytoplasmic MTs can be observed in only 60.7% (± 7.1%) of the cells carrying an NES mutation (spc72*) compared to WT cells with 76.3% (± 6.5%). The WT strain derived from the K842 background (SPC72 (K842); 78.6% ± 7.0%) and the corresponding spc72 deletion strain (spc72Δ; 60.4% ± 6.5%) show similar results, indicating that the mutation of NES_418-429 might indeed perturb function of Spc72 at the SPB.

In a similar experiment, we used xpo1 alleles for our analysis. If Spc72 is an export substrate for Xpo1, nuclear accumulation of Spc72 in xpo1 mutants should lead to a shortage of the protein in the cytosol and, hence, to a defect in formation of cytoplasmic MTs. As can be seen in Fig. 6D, the different xpo1 alleles indeed have similar phenotypes: at the permissive temperature, 77.1% (± 6.6%) of WT cells (XPO1) form cytoplasmic MTs at the SPB whereas in the xpo1-1 strain (56.0% ± 0.7%) and the xpo1-101 allele (73.8% ± 1.4%), this number is reduced. At 37°C, this phenotype is more pronounced: only 52.7% (± 7.7%) of xpo1-1 cells have cytoplasmic MTs left, and in the xpo1-101 allele, 47.2% (± 1.8%) of the cells are still equipped with cytoplasmic MTs. Taken together, our in vivo data suggest that the mutation of NES_418-429 of Spc72 leads to loss of the protein from SPBs and subsequently to a reduction of MTs on the cytoplasmic side of the nuclear envelope. In addition, mutations in the nuclear export receptor Xpo1 result in a similar phenotype with respect to cytoplasmic MT morphology.

DISCUSSION

In higher eukaryotic cells, evidence is accumulating that components of the nuclear transport machinery are involved in nuclear functions apart from nucleocytoplasmic trafficking. In metazoans, nuclear envelope formation, NPC assembly, and formation of the mitotic spindle have been shown to be dependent on certain nuclear transport receptors and the Ran-GTPase system (12, 45). In fission and budding yeast, the Ran system was identified as a regulator for the MT cytoskeleton and NPC assembly (21, 38, 51, 55). Since the Ran machinery, transport receptors, and NPC proteins are highly conserved, it is reasonable to assume that in budding yeast the nuclear transport apparatus is engaged in other nuclear functions as well. However, experimental evidence is scarce, and since budding yeast cells do not dissolve their nuclear envelopes during mitosis, it has been argued that spindle assembly must be functionally distinct from the analogous process in higher eukaryotes (13).

In this study, we provide evidence that nuclear protein export and spindle morphogenesis may be linked in budding yeast. In an effort to identify novel export substrates for the exportin Xpo1, we performed a yeast two-hybrid screen. Among known interactors of Xpo1 such as Yap1, Nup42, and Nup49 (48, 76), we identified a group of proteins that previously had not been studied in the context of nuclear transport. Among these novel putative export substrates, Spc72, a component of the yeast SPB, was the most interesting. Using a combination of in vitro binding assays and in vivo experiments, we showed that Spc72 fulfills the criteria for being an export substrate for Xpo1. First, binding of Xpo1 to Spc72 fragments is Gsp1-GTP dependent, and the addition of excess NES peptide competes with it. Secondly, we identified a leucine-rich amino acid sequence at position 418 to 429 in Spc72 that conforms to the NES consensus and that when mutated no longer confers binding to Xpo1. When fused to a GFP reporter or analyzed in the context of the N-terminal and C-terminal halves of Spc72, this sequence mediates nuclear export of the respective GFP fusion proteins in living yeast cells. Since Spc72 in concert with the γ-tubulin complex is required to nucleate cytoplasmic MTs at the SPB, we expected that the lack of nuclear export of NES-mutated Spc72 would result in a shortage of the protein in the cytoplasm and hence a deficiency in astral MT nucleation. Using fluorescence intensity quantification of SPBs, we found that Spc72*-GFP signals were reduced to 50% of levels observed for WT Spc72-GFP. We also observed a reduction of cytoplasmic MTs in a strain containing NES-mutated Spc72; this was also true for strains carrying mutations in XPO1. Taken together, these data are consistent with the notion that a functional interaction between Spc72 and Xpo1 might be required for proper spindle assembly in yeast, and they provide a first hint that nuclear export processes might be involved.

We were quite surprised to find that Xpo1 itself localizes to SPBs although the majority of the protein is found at other nuclear locations such as the NPC and the nuclear interior. In principle, this binding could be mediated by an NES-containing protein such as Spc72. However, Spc72 localizes to the cytoplasmic side of the SPB and is not required for this binding event since Xpo1 is still found at the SPB in an Spc72 deletion mutant. This suggests that Xpo1 has one or more other binding partners at the SPB, one of which could be the inner plaque protein Spc110. We found by two-hybrid analysis that it interacts with Xpo1, a fact that had not been previously analyzed. The SPB binding of Xpo1 can be reduced by overexpressing the Ran-mutant Gsp1-T26N, which lowers the Gsp1-GTP concentration in the nucleus. Although preliminary, this in vivo data might indicate that Xpo1 binding to the SPB is Ran dependent and probably mediated by an NES-containing protein other than Spc72.

What could be the function of the binding of Xpo1 to SPBs? An answer could come from what has been observed for Crm1, the Xpo1 orthologue in metazoans. Crm1 has been localized to centrosomes (22), along with Ran-GTP (37) and the NES-containing RanBP1 (15), an effector of the Ran-GTPase cycle. More recently, nucleophosmin, a multifunctional nuclear protein, has been shown to be recruited to centrosomes and Crm1 via its NES sequence (71). Treatment of cells with leptomycin B, an inhibitor of NES-Crm1 interactions, results in loss of nucleophosmin from centrosomes (61, 71). In addition, inactivating the NES of nucleophosmin by mutation had the same effect (71). Integrating all their findings in one model, these authors proposed that formation of a trimeric complex between Crm1, Ran-GTP, and an NES-containing protein such as, for example, nucleophosmin, could prevent unscheduled duplication of the centrosome during the cell cycle (8). In another study, the nuclear export of certain centrosomal proteins was shown to be inhibited by leptomycin B, but no additional data with respect to Crm1 binding or NES identification were provided (37).

Although the yeast SPB is the functional equivalent of the metazoan centrosome, it is not clear whether duplication is an analogous process in both systems (2). In addition, since budding yeast lacks nuclear envelope breakdown, mitotic spindle assembly must involve certain features unique to yeast. Although intriguing with respect to the data obtained in metazoans, our findings concerning Xpo1 localization at the SPB can provide only a first hint with respect to a possible function of Xpo1 at this location. Therefore, alternative models of why and how components of the spindle apparatus and Xpo1 interact in yeast must be considered at this point.

Based on our data for Spc72, we favor the idea that Xpo1/Spc72 interaction could reflect the necessity to keep Spc72 out of the nucleus. It has been shown that fusions between the N terminus of Spc110 and the C terminus of Spc72 are functional in nucleating MTs at the cytoplasmic side of the SPB, whereas expression of the N terminus of Spc72 fused to a C-terminal fragment of Spc110 results in cell death (39). Using Xpo1 as an export receptor, the cell might restrict Spc72 to the cytoplasm, where it binds via its N terminus to the γ-tubulin complex. A sufficient amount of this complex, hence, would be anchored in the cytoplasm, initiating MT nucleation at this side of the SPB. This could explain how sorting of the Tub4 complex is achieved between the nucleus and the cytoplasm, despite the fact that cytoplasmically assembled Tub4 complexes carry an NLS sequence on Spc98, which would otherwise be used to direct all Tub4 complex into the nucleus. However, since our two-hybrid analysis did not reveal an interaction of Xpo1 and components of the Tub4 complex, the situation might be more complicated than anticipated in this model. Definitely more work is needed to characterize the regulatory mechanisms, including nuclear transport reactions that underlie spindle biogenesis in budding yeast.