Regulation of Cdc2/cyclin B activation by Ran, a Ras-related GTPase

In the past few years, significant advances have been made in understanding the molecular mechanism controlling the pro-gression of the eukaryotic cell cycle (Murray and Hunt, 1993). Central components of this mechanism are the cyclin-dependent protein kinases (CDKs). CDKs consist of a protein-serine/threonine kinase catalytic subunit associated with a cyclin subunit that is required for kinase activity and function. Several different kinase subunits and cyclins have been iden-tified; each kinase complex consists of a certain combination and is thought to catalyse the phosphorylation of key substrates that effect progression through a particular transition in the cell cycle (Pines, 1993). Several mechanisms that control the activity of CDKs have been determined. The synthesis and degradation of their cyclin subunits is regulated during the cell cycle, so the availability of cyclin to form a kinase complex is restricted to certain phases (Hunt, 1991). Once formed, the activity of CDKs may be controlled by the binding of other proteins whose synthesis and/or degradation is regulated; such proteins have been shown to inhibit CDKs required for pro-gression through G₁ into S-phase (Nasmyth and Hunt, 1993). In addition, CDKs are regulated by the phosphorylation state of the kinase subunit (Clarke, 1994). These control mecha-nisms allow the activity of CDKs and hence the progression of the cell cycle to be influenced by external signals, and also intracellular feedback signals, called checkpoints, that monitor the status of cell cycle processes (Hartwell and Weinert, 1989; Murray, 1992).

One of the major transitions in the cell cycle is the entry into mitosis. To prevent genetic defects from being passed on to the daughter cells, DNA replication should be completed and any damage that has occured should be repaired before segregation of the chromosomes is attempted. In most cells, there are checkpoint controls that monitor the progress of S-phase and prevent entry into mitosis if S-phase is not successfully completed (Roberge, 1992). The operation of these check-points can be revealed by delaying the completion of S-phase: inhibition of DNA replication using hydroxyurea or aphidi-colin, or DNA damage induced by UV irradiation or chemicals blocks entry into mitosis. One of the key features of cell trans-formation leading to cancer may be the relaxation of such checkpoints and the accumulation of chromosome damage (Weinert and Lydall, 1993).

Entry into mitosis is induced by the activation of the protein kinase complex consisting of a Cdc2 (CDK1) catalytic subunit and a B-type cyclin (Nurse, 1990). Genetic studies using Schizosaccharomyces pombe have shown that a checkpoint which prevents entry into mitosis during S-phase acts through a network of protein kinases and phosphatases to control the timing of Cdc2 dephosphorylation at an inhibitory site, Tyr15 (Gould and Nurse, 1989; Enoch and Nurse, 1991; Al-Khodairy and Carr, 1992; Enoch et al., 1992). Similarly, in vertebrate cells, inhibition of DNA repli-cation or induction of DNA damage results in the accumula-tion of Cdc2 complexed to cyclin B in an inactive form phos-phorylated on Tyr15 (Dasso and Newport, 1990; Lock, 1992; Smythe and Newport, 1992; Tsao et al., 1992; O’Connor et al., 1993; Kharbanda et al., 1994) and also Thr14 (Kornbluth et al., 1994b). In a hamster cell line (tsBN2), a temperature-sensitive mutation in RCC1 results in loss of the functional protein at the restrictive temperature leading to premature Cdc2/cyclin B activation and chromosome condensation in S-phase cells (Nishitani et al., 1991). Mutation of an RCC1 homologue in S. pombe, pim1, has also been reported to result in premature mitosis, even when DNA replication is inhibited by hydroxyurea (Matsumoto and Beach, 1991; but see also Sazer and Nurse, 1994). In addition, the RCC1 protein seems to be required for progress through G₁ into S-phase, at least in vertebrates (Nishitani et al., 1991; Dasso et al., 1992). RCC1 catalyses guanine-nucleotide exchange on Ran, a Ras-related GTPase that is present at high concentrations in nuclei (Bischoff and Ponstingl, 1991a,b). Ran is a highly conserved protein: the human sequence (denoted TC4; Drivas et al., 1990) is 81% identical to the S. pombe homologue, the product of the spi1 gene (Matsumoto and Beach, 1991). RCC1 is associated with the chromatin (Ohtsubo et al., 1989; Bischoff et al., 1990), making it a good candidate for a detector molecule that senses the progression of S-phase and controls the activation of Cdc2/cyclin B, possibly signalling through Ran.

In the early embryonic cell cycles of Xenopus laevis, DNA replication and Cdc2/cyclin B activation are not normally coupled, and the rapid oscillation between S-phase and mitosis in these cell cycles depends upon the intrinsic timing of the cell cycle control mechanism (Murray, 1992). Nevertheless, the components of the checkpoint that blocks Cdc2/cyclin B activation during S-phase can be reconstituted in concentrated extracts of Xenopus eggs by the addition of nuclei that replicate their DNA. It seems that the checkpoint does not normally function because the amount of replicating genomic DNA is too low to generate a signal. When the signal is enhanced by adding nuclei, Cdc2/cyclin B is kept tyrosine-phosphorylated and inactive (Dasso and Newport, 1990). This block can be overcome by addition of active Cdc25 proteins to the extract or inhibition of an okadaic acid-sensitive protein phosphatase (Kumagai and Dunphy, 1991; Clarke et al., 1993).

Such cell-free extracts provide a potentially useful system for the molecular dissection of checkpoints. We have used Xenopus egg extracts supplemented with nuclei to examine the putative role of RCC1 and Ran in the coupling between S-phase and control of Cdc2/cyclin B activation. In contrast to some previous predictions, we find that a mutant form of Ran locked in the GTP-bound state (Q69L) does not inhibit Cdc2/cyclin B, but a dominant inactive mutant that inhibits RCC1 activity (T24N) prevents kinase activation in the presence of nuclei. This role for Ran in the regulation of Cdc2/cyclin B activation seems to be distinct from other roles in nuclear import, nuclear formation and DNA replica-tion.

Recombinant human RCC1 and Ran proteins were expressed in Escherichia coli and purified as described by Klebe et al. (1993). The presence of the T24N and Q69L mutations in Ran proteins were confirmed by mass determination using mass spectrometry, and in the case of Ran T24N, by sequencing the amino-terminal residues of the protein. These procedures also confirmed that the purity of the proteins was >90%.

Low speed (10,000 g) supernatants of Xenopus eggs containing cyclo-heximide, prepared as described by Clarke et al. (1993), were sup-plemented with 5% (v/v) glycerol and 10 μg/ml aprotinin before freezing and storage in aliqouts (100 μl) in liquid nitrogen. These extracts do not contain cyclin B and show no activation of histone H1 kinase upon incubation without addition of cyclins. They were estimated to contain approximately 1.2 μM Ran and 0.2 μM RCC1, assuming that the endogenous Xenopus proteins and recombinant human proteins are recognised with similar efficiency by antibodies on western blots (P. R. Clarke, unpublished).

Xenopus demembranated sperm heads (Gurdon, 1976) were added in 1 μl to 7 μl of thawed extract and incubation was carried out at room temperature (21-23°C) for 70 minutes. After this time, the chromatin had fully decondensed, a nuclear envelope had formed (monitored by phase-contrast and fluorescence microscopy) and DNA replication (assayed as described below) was just being initiated, as described in previous studies (e.g. see Blow and Laskey, 1986; Hutchison et al., 1987; Newport, 1987; Sheehan et al., 1988). The activation of Cdc2 was initiated by the addition of recombinant cyclin B (sea urchin cyclin BΔ90, an N-terminal truncated form that is non-degradable; Glotzer et al., 1991) in 1 μl of buffer A (50 mM Tris-HCl, 100 mM KCl, 5 mM MgCl₂, 1 mM dithiothreitol and 10 μg/ml leupeptin) to give a final concentration of 200 nM. At the times shown, samples (1 μl) were removed, diluted in extraction buffer (80 mM β-gly-cerophosphate, pH 7.3, 20 mM EGTA, 15 mM MgCl₂, 1 mM Na₃VO₄, 1 mM dithiothreitol, 25 μg/ml aprotinin, 1 mM benzami-dine, 0.5 mM phenylmethylsulphonyl fluoride) and assayed for Cdc2/cyclin B activity using histone H1 as substrate (Clarke et al., 1992). Unless stated otherwise, additions of Ran proteins, aphidicolin and okadaic acid were made in 1 μl of buffer B (50 mM Tris-HCl, 50 mM NaCl, 1 mM dithiothreitol, pH 7.8) after incubation of the extracts for 70 minutes following addition of sperm heads and imme-diately before cyclin addition. In the absence of an addition, the appro-priate buffer was substituted.

After incubation with the appropriate additions for 120 minutes, extracts (30 μl) were diluted in 200 μl of extraction buffer and Cdc2/cyclin B complexes were partially purified using p13^suc1-Sepharose beads as described by Clarke et al. (1992). Kinase complexes were solubilised, run on SDS-PAGE and immunoblotted using anti-phosphotyrosine (4G10, Upstate Biotechnology) or anti-PSTAIRE (a gift from M. Yamashita) monoclonal antibodies.

Guanine nucleotide exchange on Ran catalysed by RCC1 was measured taking advantage of the altered tryptophane fluorescence signal of Ran when the fluorescent GDP analogue 2′,3′-bis-O-(N-methylanthraniloyl)guanosine diphosphate (mantGDP; John et al., 1990) is bound (Klebe et al., 1993); 1 μM wild-type Ran complexed to mantGDP was incubated with 10 nM RCC1 in 30 mM KP_i (pH 7.4), 5 mM MgCl₂ and 2 mM β-mercaptoethanol at 25°C in the presence of either 0.6 μM wild-type Ran or 0.6 μM Ran T24N, in both cases complexed to GDP. The exchange reactions were started by addition of 1 mM free GDP and the decrease in fluorescence (excitation wavelength=370 nm/emission wavelelength=450 nm) as mantGDP is exchanged for GDP was followed with time.

DNA synthesis was assayed by the incorporation of radioactivity from dCTP into acid-precipitable DNA (Leno and Laskey, 1991). Incuba-tions (20 μl) containing 500 nuclei/μl or 4 ng M13 DNA/μl were stopped by the addition of 180 μl of 20 mM Tris-HCl, 20 mM EDTA, 5 mg/ml sodium dodecyl sulphate, pH 8.0. Proteins were digested by the addition of proteinase K to 0.5 mg/ml and incubation at 37°C for 60 minutes. The samples were extracted successively with phenol, phenol:chloroform:isoamyl alcohol (25:24:1, by vol.) (both equili-brated at pH 8.0) and chloroform. DNA was precipitated with 10% (w/v) trichloroacetic acid containing 2% (w/v) sodium pyrophosphate on ice for 30 minutes and collected on filter discs (Whatman GF/C). The filters were washed with 5% (w/v) trichloroacetic acid (3 times), 95% (v/v) ethanol (2 times) and dried, and the radioactivity was measured in a liquid scintillation counter. Nuclear DNA synthesis requires the formation of an intact nuclear envelope from membrane components in the extracts (Newport, 1987; Sheehan et al., 1988); it began at about 60 minutes and was mostly completed by 180 minutes, although a low level of synthesis continued after that time. M13 phage DNA synthesis is from single-stranded to double-stranded DNA; this does not require a nuclear envelope (Méchali and Harland, 1982) and was rapidly initiated, reaching a peak at 20-30 minutes and declining thereafter.

Concentrated interphasic extracts of Xenopus eggs were prepared without mitotic cyclins by inhibiting protein synthesis following release of the eggs from metaphase arrest. The acti-vation of Cdc2/cyclin B was initiated by the addition of cyclin B protein (mitotic cyclins are the only proteins whose synthesis is required for these extracts to enter mitosis (Murray and Kirschner, 1989). This procedure allows the precise timing of activation of Cdc2/cyclin B controlled by post-translational mechanisms to be studied (Solomon et al., 1990). Activation of Cdc2/cyclin B requires the dephosphorylation of Tyr15 of Cdc2 and, in vertebrates, Thr14. In Xenopus egg extracts, this occurs when there is a switch in the relative kinase and phos-phatase activities that act on these inhibitory phosphorylation sites after a defined lag period (Solomon et al., 1990; Clarke et al., 1993; see Fig. 1A). The coupling between S-phase and the activation of Cdc2/cyclin B could be reconstituted in these extracts: formation of more than 500 nuclei/μl in the extracts increased the lag period before Cdc2/cyclin B activation; at 500 nuclei/μl, Cdc2/cyclin B activation was still coupled to S-phase, but kinase activation was not blocked unless DNA repli-cation was also inhibited by aphidicolin (Fig. 1).

To study the possible roles of RCC1 and Ran in the mechanism coupling Cdc2/cyclin B activation and S-phase, we have produced recombinant RCC1 and Ran proteins in E. coli and purified them to homogeneity (Klebe et al., 1993). The Ran proteins were wild-type or mutant at either glutamine 69 (Ran Q69L) or threonine 24 (Ran T24N). Ran Q69L is defective in its GTPase activity, even when stimulated by a GTPase acti-vating protein (GAP) (Coutavas et al., 1993; Bischoff et al., 1994). This is similar to Ras mutated at glutamine 61, which is a constitutively signalling protein that is locked in the GTP-bound conformation (Polakis and McCormick, 1993). Mutation of threonine 24 in Ran to asparagine was chosen because mutations at the equivalent threonine or serine residues in other related GTPases results in the production of inactive dominant mutant proteins (Feig and Cooper, 1988; Polakis and McCormick, 1993). These residues are involved in the coordination of Mg²⁺ in the nucleotide binding site, and mutation to asparagine or alanine decreases nucleotide affinity (Pai et al., 1989; Farnsworth and Feig, 1991; John et al., 1993). Similarly, the T24N mutation of Ran results in a protein that effectively does not bind GTP and the affinity for GDP is also reduced (C. Klebe and A. Wittinghofer, unpublished).

We tested the effects of these proteins on the activation of Cdc2/cyclin B protein kinase in extracts supplemented with 500 nuclei/μl. When added at 10 μM (Fig. 2a), Ran T24N (pre-loaded with GDP) completely inhibited Cdc2/cyclin B activa-tion, whereas Ran wild-type and the Q69L mutant preloaded with GDP and GTP, respectively, had no effect, even at 60 μM. Similar results were obtained when Ran proteins were added before nuclei (not shown) or at the same time as cyclin B after incubation of nuclei in the extracts for 70 minutes (Fig. 2A). Typically, 5 μM Ran T24N was sufficient to completely block activation, and at 1 μM the lag before kinase activation was lengthened (Fig. 2B). This concentration is similar to the con-centration of Ran endogenous to the extracts (estimated at 1.2 μM; P.R. Clarke, unpublished). The effect of Ran T24N was specific to the native protein, since no effect was produced by Ras or Rab5 proteins with analogous mutations (Ras S17N and Rab5 S34N) or by Ran T24N denatured by heating prior to addition to the extract (not shown).

Protein synthesis was inhibited in the extracts, so the effect of Ran T24N must have been mediated through a post-trans-lational mechanism that regulates the timing of Cdc2/cyclin B activation. Indeed, Ran T24N inhibited Cdc2/cyclin B activa-tion by maintaining Cdc2 in the tyrosine phosphorylated form (Fig. 3A). The effect of Ran T24N was completely overcome by addition of okadaic acid, whether added at the same time as Ran T24N or following incubation for 60 minutes with Ran T24N (Fig. 3B), and there was a concomitant loss of tyrosine phosphate from Cdc2 (Fig. 3A). Okadaic acid inhibits a type-2A protein-serine/threonine phosphatase (PP-2A) that, during interphase, suppresses the phosphorylation and activation of Cdc25-C phosphatase (Clarke et al., 1993; Kinoshita et al., 1993). Specific inactivation of this PP-2A seems to be required for a switch in the relative activities of Cdc25-C and the opposing kinases, resulting in the dephosphorylation of Cdc2 at Tyr15 and Thr14, and activation of the Cdc2/cyclin B complex (Kumagai and Dunphy, 1992; Clarke et al., 1993; Tang et al., 1993). Since inhibition of PP-2A with okadaic acid overcomes the effect of Ran T24N, this mutant of Ran may function upstream of PP-2A, maintaining its ability to prevent Cdc2/cyclin B activation in the presence of replicating DNA. In the case of Ras, it has been proposed that a mutant equiv-alent to T24N in Ran (Ras S17N) does not activate downstream effectors and binds tightly to an upstream regulator, presum-ably a Ras guanine nucleotide exchange factor (Farnsworth and Feig, 1991; Schweighofer et al., 1993). We therefore assessed the ability of the guanine nucleotide exchange factor for Ran, RCC1, to interact with Ran proteins by measuring the RCC1-catalysed guanine nucleotide exchange reaction on Ran (Fig. 4). RCC1 catalysed GDP exchange on both Ran wild-type and T24N proteins (data not shown). However, Ran T24N inhibited RCC1-catalysed exchange on the wild-type protein when added to the same reaction, decreasing the observed rate of the reaction by 11-fold (Fig. 4). To investigate whether the ability of Ran T24N to prevent Cdc2/cyclin B activation in egg extracts was due to inhibition of RCC1 activity, the effect of supplementing extracts with active RCC1 protein was investi-gated. RCC1 added alone produced no inhibitory effect on Cdc2/cyclin B activation. When RCC1 was added prior to Ran T24N, the inhibition of Cdc2/cyclin B activation by Ran T24N was antagonized. These results indicate that Ran T24N prevents RCC1 from catalysing nucleotide exchange on the endogenous Ran in the extracts, and that this activity is required for Cdc2/cyclin B activation. However, if Ran T24N was first incubated in the extracts, so that Cdc2/cyclin B acti-vation was inhibited, then addition of RCC1 failed to overcome the block (not shown). Furthermore, the block on Cdc2/cyclin B activation caused by Ran T24N was not relieved by adding a 6-fold excess of Ran Q69L (or Ran wild-type) pre-loaded with GTP (Fig. 5b), suggesting that Ran-GTP is unable to overcome the effect of Ran T24N. These results suggest that Ran T24N may also sequester an effector required for Cdc2/cyclin B activation, possibly in an inactive complex together with RCC1.

We examined whether the role of Ran in regulating Cdc2/cyclin B activation might be an indirect one, perhaps through an effect on DNA replication. When Ran proteins were added to the extracts prior to sperm heads, an inhibition of subsequent DNA replication by Ran T24N (but not Ran Q69L or wild-type protein) was observed (Fig. 6A, t=0). Ran T24N did not affect chromatin decondensation or the accumulation of membrane vesicles, but the nuclei failed to swell as normal (not shown; see also Kornbluth et al., 1994a). However, when Ran proteins were added after the formation of nuclei and the initiation of DNA replication, conditions where Ran T24N inhibits Cdc2/cyclin B activation, none of the Ran proteins had any inhibitory effect on DNA replication (Fig. 6A, t=70). Therefore Ran T24N must exert its effect on the activation of Cdc2/cyclin B by acting downstream of DNA replication. Aphidicolin inhibited DNA replication both when added before sperm heads and, in contrast to Ran T24N, after they had assembled functional nuclei. Furthermore, Ran T24N (as well as Ran Q69L and wild-type protein) had no effect on the replication of M13 single-stranded phage DNA in the extracts (Fig. 6B), a process that does not require the formation of intact nuclei. These results suggest that Ran plays a role in the formation of functional nuclei that are competent for DNA replication, although not during DNA replication itself, and this role is distinct from the effect on Cdc2/cyclin B activation.

Recently, Ran has been purified as a factor that promotes the nuclear import of proteins containing nuclear localisation sequences (Melchior et al., 1993; Moore and Blobel, 1993). Ran T24N does not prevent the import of such proteins into nuclei in Xenopus egg extracts containing endogenous wild-type Ran, although this process is inhibited in a dominant manner by Ran Q69L (C. Dingwall, J. Robbins, P.R. Clarke, C. Klebe and A. Wittinghofer, unpublished). This suggests that the effect of Ran on nuclear import does not explain its role in the regulation of Cdc2/cyclin B activation. Furthermore, Ran T24N partially inhibited the activation of Cdc2/cyclin B in the complete absence of nuclei, although higher concentrations were required to produce the same degree of inhibition observed when nuclei were present (Fig. 7).

GTPases such as Ran cycle between GTP-bound and GDP-bound forms that interact differently with effector proteins. The irreversible step in the cycle is the hydrolysis of GTP to GDP, stimulated by GTPase-activating proteins (GAPs). Recharging with GTP occurs when GDP is exchanged for GTP, a reaction catalysed by guanine nucleotide exchange factors (GEFs). Mutations in RCC1 (which encodes a nuclear GEF for Ran) have been reported to cause premature chromo-some condensation in a hamster cell line and in S. pombe (Matsumoto and Beach, 1991; Nishitani et al., 1991). One explanation for these results might have been that loss of func-tional RCC1 prevented the generation of Ran GTP required to inhibit Cdc2/cyclin B during S-phase. However, in this paper, we have shown that the Q69L mutant of Ran, which is locked in the GTP-bound conformation, does not affect the kinetics of Cdc2/cyclin B activation. Although it is possible that the Q69L mutation also prevents the recognition of Ran by an effector, we have found that Ran Q69L is functional in its ability to interact with putative effector proteins and inhibit nuclear import of proteins mediated by nuclear localisation sequences (Dingwall et al., unpublished). Our results indicate, therefore, that Ran GTP has no inhibitory effect upon Cdc2/cyclin B acti-vation. Furthermore, they suggest that protein import into the nucleus is not required during activation of the kinase in Xenopus egg extracts containing nuclei, although it may be required for cell cycle progression in somatic cells (Ren et al., 1993, 1994).

In contrast to Ran Q69L, we have found that the T24N mutant of Ran inhibits Cdc2/cyclin B activation in the presence of replicating DNA. While this paper was in preparation, a similar result for Ran T24N was reported by Kornbluth et al. (1994a). They showed that addition of the T24N mutant to ‘cycling’ extracts (in which the activation of Cdc2 is powered by the synthesis of endogenous cyclins) caused the accumula-tion of inactive, phosphorylated forms of Cdc2. Our results confirm and extend these findings. Since there is no protein synthesis in our experiments, we have demonstrated that the effect of Ran T24N is entirely post-translational, and we have shown that Cdc2 is indeed tyrosine phosphorylated when inhibited. By quantitation of the amount of Ran required to inhibit Cdc2/cyclin B activation, we have demonstrated that Ran T24N increases the lag period before kinase activation, a characterisic feature of the temporal regulation of the process (Solomon et al., 1990). Kornbluth et al. (1994a) found that Ran T24N also prevented the formation of functional nuclei: initial chromatin decondensation and nuclear envelope formation occured, but the nuclei remained small and failed to initiate DNA replication. The possibility remained, therefore, that the inhibition of Cdc2/cyclin B activation by Ran T24N could have been an indirect effect. We have observed similar effects of Ran T24N upon nuclear formation and the initiation of DNA replication, but Cdc2/cyclin B activation is blocked by Ran T24N even when it is added after nuclear formation and the initiation of DNA replication. Under these conditions, Ran T24N does not inhibit DNA synthesis. These results show that inhibition of Cdc2/cyclin B activation by Ran T24N is not due to an indirect effect on the establishment or progression of S-phase. Kornbluth et al. (1994a) also reported that Ran T24N blocked Cdc2/cyclin B activation in the absence of any nuclei. We show that, although this is possible, much higher concen-trations of Ran T24N are required than in the presence of nuclei.

One explanation for the effect of Ran T24N could be that Ran GDP actively inhibits Cdc2/cyclin B activation, since the T24N mutation abolishes the ability of Ran to bind GTP and the predominant nucleotide bound to Ran T24N would be therefore expected to be GDP. Addition of high concentrations of wild-type Ran complexed with GDP did not affect the kinetics of Cdc2/cyclin B activation, although it is possible that GDP could be exchanged for GTP once the protein is added to the extracts. However, we found that Ran T24N inhibited the ability of RCC1 to catalyse guanine nucleotide exchange on wild-type Ran. Furthermore, the inhibition of Cdc2/cyclin B activation by Ran T24N could be overcome by increasing the amount of RCC1 in the extracts prior to addition of the T24N mutant. This indicates that one of the effects of the T24N mutation may be to sequester RCC1 in an inactive complex, preventing it from catalysing the formation of Ran GTP; this would explain the dominance of Ran T24N over the wild-type Ran present in the extracts. Our results suggest that the GTP-bound form of Ran generated by the activity of RCC1 may be required for Cdc2/cyclin B activation in the presence of nuclei, but do not exclude the possibility that Ran GDP also plays an inhibitory role.

In addition to sequestering RCC1, Ran T24N may also bind a putative effector required for Cdc2/cyclin B activation, possibly in a trimeric complex with RCC1, preventing the interaction of the effector with Ran GTP. This would explain the inability of Ran GTP to overcome the effect of Ran T24N. We have also obtained evidence to suggest that interfering with the interaction between Ran GTP and an effector in the extracts prevents Cdc2/cyclin B activation in the presence of nuclei (P. R. Clarke, unpublished). So far, the putative effector proteins that have been shown to associate with Ran interact specifi-cally with the GTP-bound state and not the GDP-bound state (Coutavas et al., 1993; Lounsbury et al., 1994). However, the affinity for Ran T24N (alone or complexed to RCC1) of any effector that modulates Cdc2/cyclin B activation is unknown. The affinity for Ran T24N would be predicted to be much greater than for Ran wild-type or Q69L, since addition of a 6-fold excess of those proteins failed to overcome the inhibition of Cdc2/cyclin B by Ran T24N. Testing this hypothesis will require the identification and purification of the putative effector.

Ran modulates the activation of Cdc2/cyclin B through control of Cdc2 dephosphorylation at the inhibitory Tyr15 site, and probably also at Thr14. This dephosphorylation involves a switch in the relative activities of Cdc25-C phosphatase and the opposing kinases that act on these sites. These enzymes acting on Cdc2 have been implicated in the coupling of S-phase to mitosis by both genetic and biochemical studies. Trig-gering the switch that initiates Cdc2/cyclin B activation seems to involve the down-regulation of a type-2A protein phos-phatase (PP-2A) that negatively regulates Cdc25-C and possibly also the Tyr15 and Thr14 kinases that oppose Cdc25-C. Our results suggest that Ran GTP may allow the inactiva-tion of PP-2A to occur (Fig. 8). The generation of Ran GTP catalysed by RCC1 might be inhibited during DNA replication, coupling entry into mitosis to the progression of S-phase. In the light of our results, it will be of interest to determine whether the predominant guanine nucleotide bound to Ran changes from GDP to GTP prior to the onset of mitosis.

We thank P. D’Eustachio for the Ran/TC4 cDNA, T. Nishimoto for the RCC1 cDNA, M. Glotzer for cyclin BΔ90-expressing bacteria, M. Yamashita for anti-PSTAIR antibodies, H. Stenmark and M. Zerial for Rab5 proteins, and C. Dingwall, I. Mattaj, F.R. Bischoff and H. Ponstingl for discussion.