XMog1, a nuclear Ran-binding protein in Xenopus, is a functional homologue of Schizosaccharomyces pombe Mog1p that co-operates with RanBP1 to control generation of Ran-GTP

Ran is an abundant and evolutionarily highly conserved small GTPase of the Ras superfamily, found mainly in the nucleus of eukaryotic cells (Bischoff and Ponstingl, 1991aEF4; Bischoff and Ponstingl, 1991bEF5; Drivas et al., 1990EF12). Like other GTPases, Ran exists in GTP- and GDP-bound states that interact differently with regulators and effectors (Avis and Clarke,1996EF2). The intrinsic GTPase activity of Ran is very low, but it is greatly stimulated by a GTPase-activating protein (RanGAP) that is localised in the cytoplasm and on the cytoplasmic side of the nuclear pore complex (Mahajan et al.,1997EF26; Matunis et al.,1996EF27; Richards et al.,1996EF46). By contrast, RCC1, the only guanine nucleotide exchange factor (GEF) for Ran that has been identified, is localised in the nucleus (Bischoff and Ponstingl,1991aEF4; Bischoff and Ponstingl,1991bEF5; Klebe et al.,1995bEF24; Ohtsubo et al.,1989EF37). The compartmentalised localisation of these regulators is thought to maintain a high concentration of Ran-GTP in the nucleus, in contrast to a low concentration of Ran-GDP in the cytoplasm of interphase cells (Görlich and Mattaj, 1996EF14;Görlich et al.,1996EF15). This gradient across the nuclear envelope is critical for the directionality of transport of many macromolecules through the nuclear pores (Izaurralde et al.,1997EF20; Nachury and Weis,1999EF32).

Ran also plays important roles during cell division (Rush et al.,1996; Sazer,1996). In model cell-free systems made from Xenopus laevis eggs, elevated levels of Ran-GTP stabilise microtubule asters and promote mitotic spindle formation(Carazo-Salas et al., 1999;Kalab et al., 1999; Ohba et al., 1999; Wilde and Zheng,1999; Zhang et al.,1999). Ran-GTP is also required for nuclear envelope assembly in Xenopus egg extracts(Hetzer et al., 2000; Zhang and Clarke, 2000) and human somatic cell extracts (Zhang and Clarke,2001). Thus, generation of Ran-GTP is critical for the functions of this protein. In solution, RCC1 catalyses the exchange of GDP to GTP and vice versa with similar efficiency,whereas Ran has a higher intrinsic affinity for GDP than GTP (Klebe et al.,1995b). The ability of RCC1 to generate Ran-GTP in vivo would therefore depend critically on a high concentration of free GTP relative to GDP. Alternatively, additional factors may determine the directionality of the nucleotide exchange reaction catalysed by RCC1.

Several other proteins have been identified that are involved in controlling the localisation or nucleotide-bound state of Ran. Ran-binding protein 1 (RanBP1) is a mainly cytoplasmic protein that contains a nuclear export signal (Coutavas et al.,1993; Richards et al.,1996). In the dimeric complex,RanBP1 is highly specific for the GTP-bound form of Ran (Bischoff et al.,1995; Coutavas et al.,1993; Hayashi et al.,1995) and acts as a co-activator of GTPase activity with RanGAP (Bischoff et al.,1995). RanBP1 and related domains on the nucleoporin RanBP2 (Yokoyama et al.,1995) may thereby contribute to GTP hydrolysis on Ran and dissociation of nuclear export complexes in the cytoplasm (Bischoff and Görlich,1997; Kehlenbach et al.,1999). However, RanBP1 is a highly mobile protein that rapidly shuttles between the cytoplasm and the nucleus (Plafker and Macara,2000). Interestingly, deletion of the RanBP1 homologous gene sbp1 in the fission yeast, Schizosaccharomyces pombe, is rescued by mammalian or yeast Ran-binding domains that are restricted to the nucleus, indicating that RanBP1 does not require cytoplasmic localisation for its primary function in S. pombe (Novoa et al.,1999). Over-production of human RanBP1 in yeast or a hamster cell line is antagonistic to a function of RCC1 (Hayashi et al., 1995). Similarly, high concentrations of RanBP1 inhibit nuclear assembly(Nicolás et al.,1997; Pu and Dasso,1997) and oppose the effects of RCC1 on microtubule aster and spindle assembly in Xenopus egg extracts (Carazo-Salas et al.,1999; Kalab et al.,1999; Zhang et al.,1999). In assays using purified proteins, RanBP1 inhibits the exchange of GTP to GDP on Ran catalysed by RCC1, with a lesser effect on the exchange of GDP to GTP, and forms a trimeric complex with Ran and RCC1 in the absence of guanine nucleotides(Bischoff et al., 1995). Nevertheless, the RanBP1 homologue Yrb1p is required for nuclear protein import and mRNA export in S. cerevisiae (Schlenstedt et al.,1995), indicating a role complementary to the RCC1 homologue Prp20p (Amberg et al.,1993) in vivo.

Additional Ran-binding proteins, unrelated in sequence to RanBP1, have been identified that may regulate the localisation and guanine nucleotide bound state of Ran. NTF2/p10 (Moore and Blobel,1994; Paschal and Gerace,1995), binds to Ran-GDP(Paschal et al., 1996) and promotes its import into the nucleus (Ribbeck et al.,1998; Smith et al.,1998). NTF2/p10 also stabilises Ran-GDP against nucleotide exchange (Yamada et al.,1998). Oki and Nishimoto (Oki and Nishimoto, 1998) have shown that, in Saccharomyces cerevisiae, NTF2 and a novel gene MOG1 suppress the growth defect of temperature-sensitive allelles of the Ran homologue, gsp1. Deletion of MOG1 made the yeast temperature sensitive for growth, a defect that was suppressed by over-expression of NTF2, suggesting a functional interaction between their products. Although both Ntf2p and Mog1p are required for nuclear protein import in S. cerevisiae, over-expression of MOG1 does not rescue ntf2 mutants, indicating that their functions are distinct. Mog1p, unlike Ntf2p, binds preferentially to Gsp1p-GTP and is localised to the nucleus when over-expressed (Oki and Nishimoto,1998). Stewart and Baker have recently determined the crystal structure of Mog1p to 1.9Å resolution and have provided evidence that Mog1p interacts with Ran through a site similar to that bound by NTF-2(Stewart and Baker, 2000). S. cerevisiae Mog1p and a related protein from mouse cause the release of nucleotide from Ran-GTP (Oki and Nishimoto,2000; Steggerda and Paschal,2000) although this activity does not appear to be sufficient to explain the biological function of Mog1-related proteins.

To identify putative regulators and effectors of Ran in a vertebrate system, we have screened a Xenopus oocyte cDNA library using the two-hybrid system of interaction in yeast(Nicolás et al.,1997). Here, we report the identification of a Xenopus protein, named XMog1, which has sequence similarity to S. cerevisiae Mog1p, a related gene product in Schizosaccharomyces pombe, as well as plant and mammalian homologues. XMog1 and Mog1p both rescue the growth arrest caused by deletion of the related gene in S. pombe, showing the conservation of their functions. XMog1 is localised to the nucleus and interacts preferentially with Ran-GTP, causing nucleotide release and forming a complex with nucleotide-free Ran. However, in conjuction with RanBP1, XMog1 promotes the exchange of GDP for GTP by causing the release of GDP and the selective binding of GTP to Ran. XMog1 and RanBP1 also promote the selectivity of the guanine nucleotide exchange reaction on Ran catalysed by RCC1. Thus, XMog1, together with RanBP1,may facilitate the generation of Ran-GTP in the nucleus.

Xenopus oocyte mRNA was reverse transcribed into cDNA and cloned into the HybriZAP vector (Stratagene) according to the manufacturer's instructions to produce fusions with the activation domain of Gal4. Bait plasmids encoding fusions with the DNA-binding domain of Gal4 were constructed from cDNAs encoding wild-type, Q69L or T24N versions of Ran(Nicolás et al.,1997). In the library screen,interaction of Ran with a hybrid protein induced the expression of the GAL1-HIS3 and GAL1-lacZ reporter genes. This generated colonies that were able to grow in the absence of histidine and were visualised by the generation of a blue colour using X-gal as substrate forβ-galactosidase.

Plasmid DNA from positive colonies was amplified by PCR. The sequences were compared and aligned using the Lasergene program SEQMAN II (DNASTAR Inc,Madison, WI). DNA from the yeast colony carrying the longest cDNA fragment was cloned and sequenced in both directions. The 5′ end of the XMog1 sequence was extended by PCR amplification from the library to confirm the presence of upstream stop codons, indicating that the assigned ATG is the correct initiation codon. The PCR products obtained were cloned in pGEMTeasy vector (Promega) and all the clones were sequenced in both directions.

All S. pombe media were prepared and used as described previously(Moreno et al., 1991). mog1 was deleted from the S. pombe strain h⁺/h^-leu1-32/leu1-32 ade6-M210/ade6-M216 with the use of PCR generated fragments(Bähler et al.,1998), one of the genomic copies of mog1 being replaced with a Kan^Rcassette. The strain obtained this way, FJN1(h⁺/p^-leu1-32/leu1-32 ade6-M210/ade6-M216 mog1/mog1::kan^R), was assayed for tetrad dissection. mog1-related genes from Xenopus, S. pombe and S. cerevisiae were amplified by PCR and cloned in pREP41X under control of the nmt1 promoter (Maundrell,1993; Maundrell,1990).

The Xenopus XMog1 cDNA obtained in the two-hybrid screen was digested with EcoRI and XhoI and cloned into the GST fusion vector pGEX-4T-1 (Pharmacia). This construct was used to transform E. coli BL21 and the fusion protein was expressed using the manufacturer's standard protocols. The fusion protein was purified on glutatione-Sepharose(Pharmacia), resuspended at approximately 100 μM in 50 mM Tris HCl, pH 8.0,snap frozen in liquid nitrogen and stored in aliquots at -70°C. Protein concentrations were determined by the Bradford assay (Biorad) and Lowry method(Biorad DC protein assay kit).

For removal of the GST moiety, the GST-XMog1 fusion protein bound to glutathione-Sepharose was digested with thrombin as described by the manufacturer (Pharmacia). To produce 6×His-tagged XMog1, the fragment EcoRI/XhoI was cloned in vector pET28b (Novagen), digested with EcoRI and XhoI and used to transform E. coliBL21(DE3). The protein was expressed and purified on NiTa-Sepharose (Qiagen)according to the manufacturer's guidelines, dialysed against PBS and snap frozen in liquid nitrogen and stored in aliquots at -70°C. Xenopus RanBP1 and human Ran were produced as a GST fusion proteins as described previously (Hughes et al.,1998). In some cases, the GST moiety was removed by thrombin cleavage.

Antiserum against Xenopus XMog1 was raised by inoculation of rabbits (Eurogentec, Belgium) with the recombinant protein produced by thrombin digestion of the GST fusion. The serum was purified using an affinity matrix of 6×His-XMog1 covalently attached to CNBr-Sepharose (Pharmacia)as described previously (Harlow and Lane,1999). Antibodies to Xenopus RanBP1 were raised and purified in a similar way. Antibodies to Ran and RCC1 were purchased from Transduction Laboratories and used as described previously (Zhang and Clarke,2000).

Xenopus egg extracts were prepared as 10,000 gsupernatants supplemented with an ATP-regenerating system and cyclohexamide,as described previously (Hughes et al.,1998). Binding assays were carried out by supplementing 50 μl of Xenopus egg extract with GST-XMog1 or GST-Ran fusion proteins to a concentration of 2 μM (except where stated). Where specified, nucleotides were added at 2 mM. Binding was allowed to proceed at 21°C for 30 minutes before diluting to 250 μl with Mg²⁺-buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 2.5 mM MgCl₂, 0.1% Triton X-100, 10% (v/v) glycerol) or EDTA-buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 10% (v/v)glycerol). GST-tagged proteins were precipitated by addition of 20 μl of a 50% slurry of glutathione-Sepharose beads (Pharmacia) in either Mg²⁺-buffer or EDTA-buffer. Samples were then incubated at 4°C for 1 hour with gentle continuous agitation. The beads were recovered by low speed centrifugation, washed three times with the appropriate buffer and eluted by resuspending the beads in SDS-PAGE loading buffer. Proteins were examined on SDS-PAGE by silver staining or immunoblotting. Binding assays using purified proteins were carried out in a similar fashion in Mg²⁺-buffer.

Xenopus XTC cells were cultured on coverslips in L-15 medium Leibovitz (Sigma) plus 10% FCS and 25% H₂O at room temperature. The cells were fixed with 4% paraformaldehyde in Tris-buffered saline (TBS) for 20 minutes on ice and permeabilized with 0.5% Triton X-100 in TBS for 5 minutes. Indirect immunofluorescence labelling was carried out by incubating with the first antibody overnight at 4°C followed by incubation with an FITC-conjugated secondary antibody for 45 minutes at room temperature.

For expression of XMog1 fused to C-terminus of the Aequorea green fluorescent protein (GFP), XMog1 was cloned into pGFP-C2 (Clontech). Cos-1,HeLa and 3T3 cells were cultured in DMEM (Gibco) plus 10% FCS, transfected with pGFP-XMog1 using Lipofectamine (Gibco) and allowed to express GFP-XMog1 for 24 hours. Cells were fixed in acetone for 1 minute, washed twice in PBS and then mounted. Nuclei were stained with DAPI. Images were captured using a cooled CCD camera (Hamamatsu) mounted on a Zeiss Axiovert microscope and processed on an Apple Macintosh computer using Improvision Openlab and Adobe PhotoShop software.

Assays were adapted from Dasso et al. (Dasso et al.,1994). To make up a stock of protein loaded with radioactive nucleotide sufficient for several assays,GST-Ran (1 nmol) was incubated with the appropriate nucleotide in 50 μl loading buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1% (v/v) lubrol, 1 mM MgCl₂) containing 6 mM EDTA for 30 minutes at room temperature. The protein was diluted with 450 μl of cold loading buffer and the bound nucleotide stabilised with 20 mM MgCl₂. In exchange assays, free nucleotide was added to the loaded Ran. Similar amounts of nucleotides were used for loading on Ran or added free for exchange: either 0.1 μmol of unlabelled GDP/GTP (Sigma) or 0.185 MBq [³H]GDP/[³H]GTP(Amersham) were used for exchange assays, or 0.037 MBq[γ-³²P]GTP (Amersham) for GTPase assays. For each assay, 25μl of loaded Ran was added to an equal volume of reaction mixture to start the reaction. Reaction mixtures contained RCC1, RanBP1 or XMog1 in exchange buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 0.1% (v/v) lubrol) to give final protein concentrations indicated in the figure legends and 10 mM MgCl₂. Reactions were performed at room temperature (21°C). To stop the reaction, samples were added to 5 ml of stop buffer (20 mM Tris-HCl,pH 7.5, 25 mM MgCl₂, 100 mM NaCl) and immediately filtered through nitrocellulose (Hybond ECL, Amersham). The filters were washed with a further 20 ml of stop buffer, dried and radioactivity bound to the filters was determined by scintillation counting. Where experiments were performed in triplicate, results are shown as mean values with error bars indicating the standard error of the means.

To identify vertebrate proteins that interact with Ran-GTP, we screened a Xenopus oocyte cDNA library using human RanQ69L as the bait in a two-hybrid system of co-expression in yeast. RanQ69L is deficient in GTPase activity, keeping the protein in the GTP-bound state (Bischoff et al.,1994). We screened approximately 2×10⁶ clones and obtained 40 different colonies that grew in the absence of histidine and produced β-galactosidase,indicating an interaction with the bait hybrid. The clones included Xenopus RanBP1 (Nicolás et al.,1997) and a novel sequence of 774 nucleotides that produced a strong positive signal. Analysis of this sequence revealed an open reading frame (ORF) encoding a polypeptide of 187 amino acids with a predicted molecular mass of 20389 Da(Fig. 1). The sequence showed similarity with single ORFs in the Saccharomyces cerevisiae (27%identity with XMog1 at the amino acid level) and Schizosaccharomyces pombe (33% identity) genomes. Related human (53% identity) and Arabidopsis (33% identity) sequences are also present in the GenBank/EMBL/DDBJ databases. The S. cerevisiae gene has been isolated by Oki and Nishimoto as MOG1, a multicopy suppressor of temperature sensitive alleles of Gsp1, which encodes a Ran homologue (Oki and Nishimoto, 1998). Since the product of the Xenopus cDNA is related to S. cerevisiaeMog1p both in primary sequence (Fig. 1) and function (see below), we shall refer to the protein as XMog1. The nucleotide sequence of XMog1 is deposited in the GenBank/EMBL/DDBJ database (accession number AJ278788).

To assess the Ran-binding properties of XMog1, further two-hybrid analysis was carried out using wild-type Ran, RanQ69L and RanT24N as baits. The latter mutant is defective in GTP binding and has a reduced affinity for GDP (Klebe et al., 1995a). In each case,the yeast grew on selective (His^-) media, indicating that XMog1 was able to interact with all of the Ran proteins(Fig. 2). Since both wild-type and RanQ69L are likely to be predominantly GTP-bound when expressed in yeast nuclei, whereas RanT24N may be nucleotide-free, these data suggest an interaction between XMog1 and GTP-bound or nucleotide-free Ran. However, they do not exclude the possibility of an interaction with Ran-GDP.

To test the specificity of the interaction between XMog1 and Ran, we also used several related small GTPases as baits in the two-hybrid system. When Cdc42, Rac and Rho (including mutants locked in the GTP-bound state like RanQ69L or equivalent to RanT24N), were co-transformed with XMog1, all failed to produce strong growth of colonies when plated on selective media(His^-). When colonies grown on non-selective media(His⁺) were tested for β-galactosidase activity, in all cases the activity was negligible, in contrast to the activity given by the XMog1/Ran interaction, showing that the interaction of XMog1 with Ran is specific (data not shown).

To compare the function of XMog1 with the product of S. cerevisiae MOG1 and the related gene in S. pombe, we carried out a complementation analysis in S. pombe. A diploid S. pombestrain was created in which one of the genomic copies of the mog1gene was replaced by insertion of a Kan^R cassette. By tetrad analysis, in which haploid spores are germinated following dissection of asci, we consistently obtained germination and growth of two colonies from four spores (Fig. 3A). This is consistent with the failure of haploid cells lacking mog1(Δmog1) to grow. Microscopic examination of the dissected spores after germination revealed that the Δmog1 spores did germinate to produce a few cells that then failed to divide further (data not shown). Together, these results indicate that mog1 is not required for germination but is essential for continued cell viability in S. pombe. In S. cerevisiae, deletion of MOG1 produces a temperature-sensitive growth defect, whereby cells grow at 26°C but not at 34°C (Oki and Nishimoto,1998). However, growth of S. pombe Δmog1 cells was not rescued at any temperature in the range of 18°C to 36°C (data not shown).

The S. pombe strain diploid heterozygous for mog1 was transformed with the empty vector or plasmids that expressed XMog1, S. pombe Mog1p or S. cerevisiae Mog1p. One colony from each transformation was allowed to sporulate and the resulting spores were germinated. Approximately 50 colonies from each transformation were transferred to media with or without thiamine (50 μg/ml). Thiamine represses the nmt1 promoter, blocking the production of plasmid-encoded Mog1 protein. None of the colonies derived from the transformation with the empty vector showed retardation when plated on media with thiamine nor the Kan^R phenotype. However,transformation with the other plasmids resulted in approximately 50% of the colonies showing growth retardation when plated on media with thiamine. After a few rounds of plate selection, growth retardation became evident(Fig. 3B). Colonies that still showed growth retardation in media with thiamine had a Kan^R phenotype, indicating they were Δmog1(this was confirmed by PCR analysis; data not shown). Thus, expression of S. pombe Mog1p, S. cerevisiae Mog1p or XenopusXMog1 rescues the lethal phenotype of Δmog1 cells(Fig. 3B). Together, these data demonstrate a remarkable conservation of the function of Mog1-related proteins between yeasts and vertebrates.

A polyclonal antibody raised against XMog1 and affinity purified against the recombinant protein recognised a single major polypeptide in Xenopus egg and somatic Xenopus XTC cell extracts that migrated on SDS-PAGE with an apparent molecular mass of 34 kDa, identical to the migration of recombinant XMog1 protein(Fig. 4A). XMog1, which has a calculated molecular mass of 20.4 kDa, therefore migrates aberrantly on SDS-PAGE in a similar fashion to S. cerevisiae Mog1p (Stewart and Baker, 2000). Indirect immunofluorescence of Xenopus somatic XTC cells using this antibody revealed a predominantly nuclear localisation of XMog1, with exclusion from nucleoli (Fig. 4B). Consistent with this finding, XMog1 expressed as a fusion with green fluorescent protein(GFP) showed a nuclear localisation in three different mammalian cell lines(Fig. 4C).

In solution binding assays, XMog1 interacted directly with Ran, binding strongly to wild-type Ran loaded with GTP and more weakly with Ran-GDP(Fig. 5A). In the presence of EDTA, which chelates Mg²⁺ ions required for nucleotide binding to Ran, a strong association was also seen, indicating that XMog1 also forms a complex with nucleotide-free Ran. XMog1 interacted strongly with both RanQ69L-GTP, which carries a mutation in the switch II region, and RanT24N,which is deficient in nucleotide binding(Fig. 5B). Similar interactions were seen using either GST-tagged Ran and untagged XMog1(Fig. 5A,B) or untagged Ran and GST-tagged XMog1 (Fig. 5C; data not shown), confirming that the GST moiety does not affect the binding properties of XMog1 in this assay. Consistent with these results and the two-hybrid analysis in yeast, XMog1 present in Xenopus egg extracts interacted more strongly with Ran loaded with GTP than with the GDP-bound form, but also interacted strongly with Ran under conditions that promote the formation of nucleotide-free Ran (Fig. 5B,C).

To investigate the biochemical function of XMog1 further, we examined the effect of XMog1 on the binding of guanine nucleotides to Ran. Initially, we examined the effect on nucleotide exchange by observing the loss of Ran-bound[³H]GTP when incubated with free GDP(Fig. 6A). Under these conditions, XMog1 promoted loss of radioactivity in a concentration-dependent manner when Ran was preloaded with [³H]GTP, but not[³H]GDP (Fig. 6A). Significant release of radioactivity was observed when the amount of XMog1 in the assay was similar or greater than that of Ran (50 pmol). However, XMog1 also caused release of [³H]GTP from Ran in a similar experiment performed without free nucleotide, which is essential for the nucleotide exchange reactions catalysed by RCC1 or EDTA(Fig. 6B). Thus, in the absence of other proteins, XMog1, like S. cerevisiae Mog1p (Oki and Nishimoto, 2000) and mouse Mog1 (Steggerda and Paschal,2000), is a GTP release factor for Ran rather than a guanine nucleotide exchange factor.

The concentration dependence of the effect of XMog1 on the loss of[³H]GTP was similar using wild-type Ran or the Q69L mutant, which is deficient in GTPase activity (Fig. 6B). The rate of loss of radioactivity was also identical for Ran preloaded with [γ-³²P]GTP, rather than [³H]GTP(data not shown). These results indicated that XMog1 causes loss of GTP from Ran without stimulation of GTP hydrolysis. They also showed that the interaction of XMog1 with Ran is not disrupted by the Q69L mutation,consistent with the analysis of their interactions by two-hybrid analysis(Fig. 6A) and co-precipitation experiments (Fig. 6B). The release of [³H]-GTP from Ran by XMog1 was stimulated by chelation of Mg²⁺ by EDTA (Fig. 6C), although EDTA alone did not cause [³H]GTP release under these conditions (data not shown). Taken together with the data showing an affinity of XMog1 for Ran under nucleotide-free conditions(Fig. 5), these results suggest that XMog1 interacts with Ran-GTP, causes loss of nucleotide and forms a complex with nucleotide-free Ran.

Ran interacts strongly with a family of binding proteins related to RanBP1(Avis and Clarke, 1996). Although RanBP1 and XMog1 are unrelated in primary sequence, RanBP1 also interacts strongly with Ran-GTP, including the Q69L mutant(Nicolás et al.,1997). To examine the relationship between the interactions of RanBP1 and XMog1 with Ran, we determined the effect of XMog1 on the binding of RanBP1 to Ran. GST-RanBP1 interacted more strongly with Ran preloaded with GTPγS rather than GDP,as expected. However, addition of XMog1 reduced the interaction between GST-RanBP1 and Ran-GTPγS. By contrast, XMog1 stabilised the otherwise weak interaction between RanBP1 and Ran-GDP(Fig. 7A). Although no direct interaction between XMog1 and RanBP1 occurred (data not shown), XMog1 precipitated with GST-RanBP1 in the presence of Ran preloaded with GTPγS or GDP (Fig. 7A). GST-XMog1 also precipitated both RanBP1 and Ran from Xenopus egg extracts (data not shown). Thus, GST-XMog1, Ran and RanBP1 can form a complex, suggesting that XMog1 and RanBP1 can interact with different regions of Ran.

The ability of XMog1 and RanBP1 to form a complex with Ran simultaneously(Fig. 7A) prompted us to examine the combined effects of these proteins on guanine nucleotide binding to Ran. RanBP1 inhibited the release of [³H]GTP from Ran induced by XMog1 unless high concentrations of XMog1 were added(Fig. 7B). XMog1 or RanBP1 alone did not cause the release of [³H]GDP from Ran. However, when added together, XMog1 and RanBP1 caused a strong stimulation of GDP release,with >50% of the nucleotide released after 30 minutes(Fig. 7C), a rate comparable with the release of GTP by XMog1 alone(Fig. 6B). These experiments were performed without the addition of free nucleotides, so the effect of XMog1 and RanBP1 on Ran-GDP was not due to nucleotide exchange, but rather the release of nucleotides to form nucleotide-free Ran. RanBP1 has a strong affinity for Ran-GTP, but also interacts more weakly with nucleotide-free Ran(Bischoff et al., 1995;Nicolás et al.,1997). Thus, generation of nucleotide-free Ran by a high concentration of XMog1 may explain why the binding between RanBP1 and Ran loaded with GTPγS is decreased, whereas the interaction with Ran loaded with GDP is increased(Fig. 7A).

When Ran-GDP was incubated in the presence of [³H]GTP (1.2μM) and a large excess of GDP (700 μM), the combination of XMog1 and RanBP1 promoted the uptake of [³H]GTP by Ran(Fig. 7D). Under these conditions, free exchange catalysed by RCC1 or EDTA did not cause any detectable loading of Ran with [³H]GTP (data not shown). Thus, the combination of XMog1 and RanBP1 loads Ran with GTP preferentially, unlike the nucleotide exchange reaction catalysed by RCC1 or EDTA.

In the nucleus, guanine nucleotide exchange on Ran is catalysed by RCC1. We therefore examined the effects of XMog1 and RanBP1 on the activity of RCC1 assayed by release of [³H]GDP from Ran(Fig. 8), or by the uptake of[³H]GDP (Fig. 9A,B)or [³H]-GTP (Fig. 9C).

XMog1 alone had no inhibitory effect on nucleotide exchange on Ran-GDP catalysed by RCC1 in the presence of free GDP(Fig. 8A). By contrast, RanBP1 strongly inhibited [³H]GDP release catalysed by RCC1 under these conditions (i.e. GDP to GDP exchange), as reported previously (Bischoff et al., 1995). XMog1 partially overcame the inhibitory effect of RanBP1 on [³H]GDP release(Fig. 8A); this effect may represent the release of [³H]GDP by XMog1 and RanBP1 to form nucleotide free-Ran (Fig. 7C). In the presence of free GTP, the inhibitory effect of RanBP1 on release of[³H]GDP by RCC1 (i.e. exchange of GDP to GTP) was much less pronounced (Fig. 8B; Bischoff et al., 1995). XMog1 overcame the small inhibition by RanBP1, and when added alone, XMog1 stimulated the rate of [³H]GDP release by RCC1(Fig. 8B).

In the reverse reaction, XMog1 partially inhibited the exchange of GTP to GDP catalysed by RCC1, assayed by the loading of RanGTP with[³H]GDP (Fig. 9A). RanBP1 also inhibited exchange of GDP for GTP, blocking [³H]GDP uptake; this inhibition by RanBP1 was not competed by XMog1(Fig. 9B). By contrast, XMog1 and RanBP1 promoted loading of Ran-GDP with [³H]GTP by RCC1, the predominant effect being due to RanBP1(Fig. 9C). Under these conditions, where [³H]GTP concentration (1.2 μM) was low relative to GDP (700 μM), RCC1 alone failed to load Ran with[³H]GTP, whereas XMog1 and RanBP1 caused a modest loading with[³H]GTP in the absence of RCC1, as before (Figs 7D, 9C). Thus, XMog1 and RanBP1 promote the exchange of GDP to GTP by RCC1 against an unfavourable gradient in nucleotide concentrations.

Ran GTPase plays several critical roles in the control of nuclear structure and function. The localisation, nucleotide bound state and activity of Ran depend upon the interaction of regulatory proteins. Here, we have used a two-hybrid system of co-expression in yeast to identify a novel Xenopus Ran-interacting protein, XMog1, which shows sequence similarity with Mog1p in S. cerevisiae and a related gene product in S. pombe, as well as mammalian and plant proteins. Expression of Xenopus XMog1 and S. cerevisiae Mog1p can rescue deletion of the S. pombe mog1 gene, which is essential for cell viability,indicating that a critical function of Mog1-related proteins is conserved in eukaryotes. In biochemical assays, XMog1 interacts with Ran-GTP, causing release of GTP and forming a stable complex with nucleotide-free Ran. Similar activites have been reported recently for S. cerevisiae Mog1p (Oki and Nishimoto, 2000) and a Mog1-like protein from mouse (Steggerda and Paschal,2000). Endogenous XMog1 is predominantly nuclear, with exclusion from nucleoli, consistent with the localisation of GFP-XMog1 in mammalian cells (this study), overexpressed Mog1p(Oki and Nishimoto, 2000) and tagged mouse Mog1 (Steggerda and Paschal,2000).

Reduction of Ran-GTP levels by Mog1-related proteins would seem to be inconsistent with the requirement of Mog1p for nuclear protein import in S. cerevisiae (Oki and Nishimoto,1998), because this process also requires the Ran homologue Gsp1p (Oki et al.,1998) and is likely to involve the maintenance of high levels of Gsp1p-GTP in the nucleus, where Mog1p is predominantly localised. Furthermore, high levels of Mog1p do not inhibit the growth of S. cerevisiae (Oki and Nishimoto,2000), which might be expected if nuclear Ran-GTP were depleted. It seems unlikely, therefore, that induction of GTP release from Ran in the nucleus and formation of a stable nucleotide-free complex could account for the principal biological activity of Mog1-related proteins.

In the cellular environment, the interaction of XMog1 with Ran is likely to be influenced by other proteins that interact with Ran. A major Ran-GTP binding protein is RanBP1, which forms a trimeric complex with Ran and XMog1 in Xenopus egg extracts, and inhibits GTP release from Ran. Similarly, RanBP1 homologues inhibit GTP release induced by the mouse(Steggerda and Paschal, 2000)and yeast (Oki and Nishimoto,2000) Mog1 proteins. However,in contrast to the effects of XMog1 alone, we find that when RanBP1 is present, XMog1 causes the release of GDP from Ran and the binding of GTP. In the presence of RCC1, RanBP1 and XMog1 strongly promote loading of Ran with GTP while inhibiting the exchange of GTP to GDP. XMog1 and RCC1 are both localised to the nucleus during interphase, but RanBP1 has been described as a predominantly cytoplasmic protein. However, it is now clear that RanBP1 in fact shuttles rapidly between nucleus and cytoplasm in organisms as diverse as S. cerevisiae (Kunzler et al.,2000), mammals (Plafker and Macara, 2000) and Xenopus (F.J.N., C.Z. and P.R.C., unpublished). Therefore, XMog1,RanBP1 and RCC1 could all interact with Ran in the nucleus.

When Ran is imported into the nucleus, it is likely to be initially in the GDP-bound state. XMog1 may displace the import factor NTF2 from Ran-GDP(Stewart and Baker, 2000EF52),although XMog1 does not form a stable complex with Ran-GDP, but rather increases the affinity for RanBP1 and promotes the loss of nucleotide. Thus,XMog1 does not behave like a bona fide guanine nucleotide exchange factor(GEF) such as RCC1, which acts catalytically by increasing the rate at which equilibrium between GDP- and GTP-bound states is achieved (Klebe et al.,1995bEF24), but rather acts stoichiometrically by destabilising guanine nucleotide binding. A similar mechanism has been suggested for another GTPase-interacting protein, Mss4,which forms a complex with nucleotide-free Rab proteins (Geyer and Wittinghofer, 1997EF13; Nuoffer et al., 1997EF35). RanBP1 favours the binding of GTP to the nucleotide-free complex between Ran and XMog1 by forming a strong association with Ran-GTP, thereby effectively removing Ran-GTP from the equilibrium reaction. Thus, the combination of a nucleotide-destabilising factor (XMog1) and a Ran-binding protein (RanBP1) that interacts with nucleotide-free Ran but has a higher affinity for Ran-GTP can produce a nucleotide-exchange reaction, albeit much less efficiently than RCC1.

RCC1 interacts with Ran-GDP and Ran-GTP with similar affinities, catalysing the exchange of GDP to GTP and vice versa at similar rates (Klebe et al.,1995bEF24). Although XMog1 and RanBP1 inhibit GDP binding catalysed by RCC1, they promote the accumulation of RanGTP. At higher activities of RCC1, the predominant effect is due to RanBP1,probably by stabilisation of Ran-GTP and inhibition of the reverse reaction in which GTP is exchanged to GDP (Bischoff et al.,1995EF8). Thus, XMog1 and RanBP1 act as co-factors that ensure that the direction of the exchange reaction catalysed by RCC1 promotes the generation of Ran-GTP(Fig. 10). These activities of XMog1 and RanBP1 may be important in the nucleus, where RCC1 has to work against an unfavourably high concentration of Ran-GTP relative to that of Ran-GDP. XMog1 and RanBP1 would also ensure that the generation of Ran-GTP by RCC1 is not critically dependent on a high concentration of free GTP relative to GDP.

In this way, the generation of Ran-GTP, critical for maintenance of nuclear structure and function, may be regulated by the interaction of Ran with XMog1 and RanBP1. Nucleocytoplasmic shuttling could provide a mechanism to precisely control the level of RanBP1 in the nucleus, perhaps coupling the activity of RCC1 and the generation of Ran-GTP to the rate of nucleocytoplasmic transport. If the nuclear Ran-GTP concentration was too high or RanBP1 levels declined,perhaps owing to over-active nuclear export, XMog1 could also act as a sink for Ran, sequestering the protein in an inactive, nucleotide-free form. These possibilities may be tested by examining the effects on nuclear function of altering the relative levels of these proteins and disrupting their interactions by mutation.

Could the ability of XMog1 and RanBP1 to promote generation of Ran-GTP in the nucleus explain the phenotype of the deletion of MOG1 in S. cerevisiae? MOG1 is required for nuclear protein import and MOG1 deletion makes the yeast temperature-sensitive. Growth at the restrictive temperature can be rescued by overexpression of the Ran homologue Gsp1p or Ntf2p (Oki and Nishimoto,1998). Increased expression of Gsp1p or promotion of Gsp1p import into the nucleus by expression of Ntf2p may both compensate for the loss of Mog1p by increasing the total level of Ran,and therefore Ran-GTP, in the nucleus. To test this possibility, it will be of interest to examine the genetic interactions between Mog1, RanBP1 and RCC1 homologues in yeast. It also remains possible that Mog1-related proteins have a more direct function in nuclear import by dissociating import complexes between Gsp1p/Ran-GDP, importins/karyopherins and their cargoes in nucleus.

As well as its role during interphase, when maintenance of a high concentration of Ran-GTP is required to distinguish the environment of the nucleus from the cytoplasm and thereby determine the direction of nucleocytoplasmic transport, generation of Ran-GTP plays crucial roles during cell division. Evidence from Xenopus egg extracts (Carazo-Salas et al., 1999; Kalab et al.,1999; Ohba et al.,1999; Wilde and Zheng,1999; Zhang et al.,1999) indicates that localised generation of Ran-GTP is likely to play a critical role in centrosomal nucleation of microtubules and organisation of the mitotic spindle. In addition, generation of Ran-GTP on the surface of chromatin is critical for nuclear envelope assembly at the end of mitosis (Hetzer et al.,2000; Zhang and Clarke,2000; Zhang and Clarke,2001). It is possible that Mog1-related proteins also play roles in these processes.

We are very grateful to J. Ayté for help with S. pombe work, M. A. García for help and reading the manuscript, and A. Wittinghofer for reagents. This work was supported by grants to P.R.C. from the Biotechnology and Biological Sciences Research Council and the Medical Research Council. W.M. was supported by a Biomedical Research Centre postgraduate studentship and F.J.N. was supported by a Human Frontiers Science Programme Organisation long-term fellowship.