An ATP-dependent activity that releases RanGDP from NTF2.

The small GTPase Ran functions in several critical processes in eukaryotic cells including nuclear transport, nuclear envelope formation, and spindle formation. A RanGDP-binding protein, NTF2, facilitates translocation of RanGDP through the nuclear pore complex and also acts to stabilize RanGDP against nucleotide exchange. Here, we identify a novel activity that stimulates release of GDP from Ran in the presence of NTF2. Hydrolyzable ATP enhances the GDP dissociation activity, and this enhancement is inhibited by nonhydrolyzable ATP analogues. In contrast, neither hydrolyzable ATP nor nonhydrolyzable ATP analogues affect GDP dissociation from Ran catalyzed by recombinant RCC1 or inhibition of GDP dissociation from Ran by recombinant NTF2. The ATP-dependent RanGDP dissociation activity therefore has the properties of a RanGDP dissociation inhibitor (GDI) displacement factor (RanGDF) where the GDI is NTF2. A protein phosphatase inhibitor mixture stimulates the RanGDF activity, suggesting the activity is regulated by phosphorylation. We propose that the ATP-dependent NTF2 releasing factor may have a role in the RanGDP/GTP cycle.

The small GTPase Ran functions in several critical processes in eukaryotic cells including nuclear transport, nuclear envelope formation, and spindle formation. A RanGDP-binding protein, NTF2, facilitates translocation of RanGDP through the nuclear pore complex and also acts to stabilize RanGDP against nucleotide exchange. Here, we identify a novel activity that stimulates release of GDP from Ran in the presence of NTF2. Hydrolyzable ATP enhances the GDP dissociation activity, and this enhancement is inhibited by nonhydrolyzable ATP analogues. In contrast, neither hydrolyzable ATP nor nonhydrolyzable ATP analogues affect GDP dissociation from Ran catalyzed by recombinant RCC1 or inhibition of GDP dissociation from Ran by recombinant NTF2. The ATP-dependent RanGDP dissociation activity therefore has the properties of a RanGDP dissociation inhibitor (GDI) displacement factor (RanGDF) where the GDI is NTF2. A protein phosphatase inhibitor mixture stimulates the RanGDF activity, suggesting the activity is regulated by phosphorylation. We propose that the ATP-dependent NTF2 releasing factor may have a role in the RanGDP/GTP cycle.
Ran is an abundant and evolutionarily highly conserved small GTPase of the Ras superfamily, found mainly in the nucleus of eukaryotic cells (1)(2)(3). Like other GTPases, Ran exists in both GTP-and GDP-bound states, abbreviated as RanGTP and RanGDP. These two forms of Ran interact differently with regulators and effectors (4). The intrinsic GTPase activity of Ran is very low but is greatly stimulated by a GTPase-activating protein (RanGAP) that is localized in the cytoplasm and on the cytoplasmic side of the nuclear pore complex (5)(6)(7)(8). In contrast, RCC1, the only identified guanine nucleotide exchange factor (GEF) 1 for Ran, is localized in the nucleus (1,2,9,10). The compartmentalized localization of these regulators maintains a high concentration of RanGTP in the nucleus and a low concentration in the cytoplasm of interphase cells (11)(12)(13)(14). This gradient of RanGTP concentration across the nuclear envelope is critical for the directionality of transport of many macromolecules between the nucleus and cytoplasm (15).
Unlike some small GTPases, where only the GTP-bound form has binding partners, both RanGTP and RanGDP interact with nuclear transport factors. In the context of this manuscript, the RanGDP-binding protein NTF2/p10 (16 -18) is important. NTF2 exists as a homodimer and has two functions; one is to promote the import of RanGDP into the nucleus (19,20) by facilitating diffusion through the NPC. Second, NTF2 interacts directly and specifically with RanGDP and stabilizes RanGDP against RCC1-mediated nucleotide exchange (21). The crystal structure of the NTF2-RanGDP complex (22) reveals direct contacts between the hydrophobic cavity of NTF2 and the switch II region of RanGDP.
Independently of its binding to RanGDP, NTF2 also interacts with FG repeat-containing nucleoporins (17,23) and is found concentrated at NPCs at steady state. The interactions between NTF2 and both RanGDP and FG repeats are essential for efficient nuclear import of RanGDP (24 -27). The recently solved crystal structure of the RCC1-Ran complex (28) indicates that NTF2 and RCC1 cannot bind simultaneously to RanGDP. Thus, RanGDP has to dissociate from NTF2 for RCC1-mediated nucleotide exchange to occur.
The continuous regeneration of RanGTP from RanGDP is critical for all the known functions of Ran. Recently, the activities of the known components of the Ran system were analyzed by the construction of a mathematical model (13). The model predicted that the functioning of the Ran system is particularly sensitive to two parameters. The first is the GTP:GDP ratio of the cell, in part a reflection of the fact that Ran has a significantly higher affinity for GDP than GTP (10,13). The second is the delivery of RanGDP to the exchange factor RCC1 (13). In interphase, the latter process involves facilitation of RanGDP diffusion through the NPC by NTF2 followed by its dissociation from NTF2. Both steps are necessary before RCC1-mediated nucleotide exchange can occur. The fact that NTF2 binds RanGDP tightly enough to inhibit RCC1-mediated nucleotide exchange on Ran (21) suggested that a factor might exist that would help dissociate the RanGDP-NTF2 complex. Here, we have identified a novel activity that releases RanGDP from NTF2 in an ATP-dependent manner. We propose that this activity plays a role in the function of the RanGDP/GTP cycle.

Expression and Purification of Recombinant Proteins-Wild-type
Ran, RanT24N, RanQ69L, wild-type NTF2, and RCC1 were expressed in Escherichia coli BL21(DE3) and purified as described previously (21,29). The pET 15b-W7A NTF2 mutant (27) was kindly provided by Dr. Murray Stewart. The protein, like wild-type NTF2, was expressed and purified as described previously (21). Proteins were exchanged into transport buffer (TB) (20 mM HEPES, pH 7.3, 110 mM potassium acetate, 5 mM sodium acetate, 2 mM magnesium acetate, 1 mM EGTA, 2 mM dithiothreitol, and Complete EDTA-free (Roche Applied Science)) using a PD10 column (Amersham Biosciences) and then concentrated * This work was supported by the European Molecular Biology Laboratory, the Human Frontier Science Program Organization and a grant-in-aid for Centers of Excellence Research from the Japanese Ministry of Education, Science, Sports, and Culture. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
using Centricon YM3 (Amicon). Aliquots were frozen in liquid nitrogen and stored at Ϫ80°C.
Assay for Dissociation of [ 3 H]GDP from Ran-[ 3 H]GDP dissociation was assayed by a rapid filtration method as described previously (21).
Depletion of Endogenous RCC1 from Ehrlich Ascites Tumor Cells Extract-For depletion of endogenous RCC1 from the SP200, RanT24N coupled to Affi-Gel-10 (Bio-Rad) was used as described previously (29). RanQ69L or bovine serum albumin coupled to Affi-Gel-10 were used for control experiments. The ⌬RCC1-SP200 was recovered and immediately used for the filter trap assay.
Immunoblot Analysis-Fractions from the Ehrlich ascites tumor cells extract were resolved by SDS-PAGE, and the proteins were transferred to nitrocellulose filters (Schleicher & Schuell). Anti-NTF2 monoclonal antibodies (BD Transduction Laboratories), anti-histone H2A polyclonal antibodies (Santa Cruz Biotechnology), anti-histone H2B polyclonal antibodies (Santa Cruz Biotechnology), and anti-RCC1 rabbit serum were used as primary antibodies. . Comparison with the endogenous situation is not straightforward. However, virtually all of the RanGDP that enters the cell nucleus, and thus might encounter RCC1, is bound to NTF2 dimers (13,19,20). The W7A NTF2 mutant (27), which binds poorly to XFXFG-repeat containing nucleoporins but retains binding affinity for RanGDP, inhibited [ 3 H]GDP dissociation from Ran like wildtype NTF2 (Fig. 1A).

An ATP-dependent Ran GDP Dissociation
We next investigated whether a factor that would stimulate NTF2 release from RanGDP could be detected. For the assay, we used a Q200 fraction prepared from Ehrlich ascites tumor cells extract (see "Experimental Procedures"). The Q200 fraction, but not the other Q fractions (QFT, Q350, and Q550), slightly stimulated [ Fig. 2A). With each dose (3.3 or 6.6 nM) of RCC1, we examined the effects of the presence of an ATP regeneration system, two nonhydrolyzable ATP analogues, or ATP alone. RCC1 at 3.3 or 6.6 nM induced 43 or 65% dissociation, respectively, of [ 3 H]GDP from Ran. Neither the ATP regeneration system, ATP␥S (3 mM), AMP-PNP (3 mM), nor ATP (3 mM) affected the activity of recombinant RCC1 ( Fig. 2A). Similarly, neither the ATP regeneration system, ATP␥S, AMP-PNP, nor ATP had an effect on the RanGDI activity of 1.7 M of recombinant NTF2 (Fig. 2B).

ATP-dependent GDP Dissociation from Ran Requires NTF2
and Is Not Mediated by RCC1-To investigate whether endogenous RCC1 is involved in inducing the ATP-dependent [ 3 H]GDP dissociation from Ran, we further fractionated the activity. The Q200 was applied to a HiTrap SP column, and fractions (SP20, SP50, SP100, SP150, SP200, SP350, and SP500) were eluted stepwise (see "Experimental Procedures"). By immunoblot analysis, most endogenous RCC1 and NTF2 was present in the Q200 fractions ( Fig. 3A and data not shown). After separation, NTF2 predominantly eluted in the SP20 fraction (Fig.  3A), whereas RCC1 was eluted mainly in SP20, SP350, and SP500, with minor amounts present in SP200. The [ 3 H]GDP dissociation activity of each SP fraction and its stimulation by ATP were measured in the presence of recombinant RCC1 and the W7A NTF2 mutant (Fig. 3B). The ATP-dependent [ 3 H]GDP dissociation activity was predominantly found in the SP20 (FT) and SP200 fractions. Although much stronger [ 3 H]GDP dissociation activities were present in both the SP350 and the SP500 fractions, these were little affected by the presence of ATP, and therefore their presence was likely because of endogenous RCC1. Histones H2A and H2B, which have been shown to have a mild stimulatory effect on RCC1 activity (31), were also present mainly in the SP350 and 500 fractions (Fig. 3A).

FIG. 3. Separation of the ATP-dependent GDP dissociation activity from endogenous RCC1.
A, immunoblotting profiles of the Q200 and SP fractions (SP20, SP50, SP100, SP150, SP200, SP350, and SP500) of Ehrlich ascites tumor cells extract using anti-RCC1, anti-NTF2, anti-H2A, or anti-H2B antibodies. The Q200 fraction was applied to a HiTrap SP column (see "Experimental Procedures"). The Q200 contained endogenous RCC1 and NTF2. Endogenous RCC1 was eluted mainly in the SP20, SP350, and SP500 fractions. The SP200 fraction also contained smaller amounts of RCC1. Endogenous NTF2 was predominantly eluted in the SP20 and histones H2A and H2B in SP350 and SP500. B, the effects of the SP fractions on NTF2 inhibition of [ 3 H]GDP dissociation from Ran. [ 3 H]GDP-Ran (0.67 M) and the W7A NTF2 mutant (3.3 M) were preincubated for 5 min at 30°C. A fraction, as indicated, was added and incubated for another 5 min at 30°C. Finally, RCC1 (0.01 M) was added, and incubation continued for 5 min at 30°C. Dissociation was assayed as described in the Fig. 1 legend. C, the effects of the Q200, SP20, or SP200 fraction on [ 3 H]GDP-Ran dissociation in the absence of added recombinant RCC1 or NTF2. The reactions were carried out and evaluated as in Fig. 1. To more definitively rule out the participation of endogenous RanGEF in the ATP-dependent activity, we next depleted endogenous RCC1 from the SP200 using RanT24N coupled to Affi-Gel-10 beads (RanT24N-SP200). Most endogenous RCC1 (Ͼ90%) was depleted by this treatment (Fig. 4A), whereas control beads coupled to either RanQ69L or bovine serum albumin caused a less efficient depletion of RCC1 (Fig. 4A). Using these depleted SP200 fractions, the effects of the ATP regeneration system on the [ 3 H]GDP dissociation in the presence of recombinant RCC1 and NTF2 were examined. As expected, total [ 3 H]GDP dissociation activity was reduced by RCC1 depletion, with almost no ATP-independent stimulation remaining (Fig. 4B). However, the ATP-dependent activity of the SP200 fraction (28.7% dissociation) was recovered similarly in the Q69L-depleted (20.0%), the bovine serum albumin-depleted (22.8%), and the T24N-depleted (20.0%) fractions, indicating that it was not due to endogenous RCC1.
Next we examined the effects of the ATP regeneration system on the intrinsic RanGEF activity in the Q200, SP20 and SP200 fractions in the absence of added recombinant RCC1 or NTF2. Endogenous NTF2 is present in the Q200 and SP20 fractions (see above). The ATP regeneration system enhanced [ 3 H]GDP dissociation in the presence of the Q200 (58% stimulation) and the SP20 (54% stimulation) but not the SP200 fraction (9% stimulation; Fig. 3C). These results, in combination with those presented above, suggest that the stimulation of [ 3 H]GDP dissociation from Ran by ATP depends on NTF2. In other words, ATP may regulate RanGDP release from NTF2 by a factor or factors present in the SP200 fraction. In fact, the SP200 fraction in the presence of ATP released NTF2 from RanGDP to a considerable extent (Fig. 5).
ATP-dependent NTF2-releasing Activity Is Stimulated by Phosphatase Inhibitors-To further clarify the mechanism of the ATP-dependent GDI displacement activity, we tested the effects of two nonhydrolyzable ATP analogues (ATP␥S or AMP-PNP) on [ 3 H]GDP dissociation activity in the SP200 fraction. The ATP regeneration system significantly stimulated dissociation activity in the presence of NTF2 (Fig. 6). The two nonhydrolyzable ATP analogues did not affect the ATP-independent [ 3 H]GDP dissociation activity of the SP200. However, stimulation of the activity by ATP was inhibited (Fig. 6). These results suggest that ATP hydrolysis is essential for ATP-dependent dissociation activity.
To clarify whether the ATP-dependent [ 3 H]GDP dissociation activity might be regulated by phosphorylation, we examined the effects of a protein phosphatase inhibitor mixture on the activity. The phosphatase inhibitor mixture significantly stimulated the [ 3 H]GDP dissociation of the SP200 in the presence of NTF2 (24% stimulation; Fig. 7A). In the absence of recombinant NTF2, the phosphatase inhibitor mixture did not however stimulate [ 3 H]GDP dissociation (Fig. 7B). These data indicate that the ATP-dependent [ 3 H]GDP dissociation activity is probably regulated by phosphorylation and confirm that it requires NTF2. FIG. 4. ATP-dependent GDP dissociation activity is not catalyzed by RCC1. A, immunoblotting profiles of an SP200 fraction that was either untreated or passed over an Affi-Gel column containing RanT24N, RanQ69L, or bovine serum albumin (BSA) (see "Experimental Procedures"). Endogenous RCC1 in the SP200 was almost completely depleted by the RanT24N column. In this case, a 3-fold greater quantity of the SP200 compared with Fig. 3A was applied, and a 10-fold higher concentration of ␣RCC1 rabbit serum was used for the immunoblot analysis. B, the effects of the SP200 fractions from panel A on [ 3 H]GDP dissociation from Ran. The reactions were carried out and evaluated as described in the Fig. 1 legend.

DISCUSSION
NTF2 has two functions. First, it mediates the facilitated translocation of RanGDP through NPCs. Although this is not an intrinsically directional process, the activities of other Raninteracting factors mean that this function of NTF2 acts to import RanGDP into the nucleus, in which RCC1, the RanGEF, is found (19,20,32). Second, NTF2 interacts directly and specifically with the GDP-bound form of Ran and inhibits RCC1-mediated GDP release from Ran (21). For this reason, we reported NTF2 as a RanGDP dissociation inhibitor or RanGDI (21). Here, we investigated the mechanism of release of NTF2 from RanGDP. The fact that in vitro NTF2 binds RanGDP tightly enough to inhibit RCC1-mediated nucleotide exchange on Ran suggested that a factor might exist that would help dissociate the RanGDP-NTF2 complex. We were able to detect such an activity in mouse Ehrlich ascites tumor cells extract. The activity functions in an ATP-dependent manner. We refer to this factor as a RanGDI (i.e. NTF2) displacement factor, or RanGDF for short, adopting the nomenclature from other small GTPase systems (e.g. Refs. 33 and 34). The ATPdependent RanGDF activity is not specific for mouse Ehrlich ascites tumor cells, since a similar activity was detected in fractions prepared from HeLa cells (data not shown). By a combination of methods, we were able to demonstrate that the ATP-dependent activity is distinct from endogenous RCC1, and that it requires NTF2. This suggests a possible mode of action for the RanGDF, to accelerate dissociation of RanGDP from NTF2 and thus to make RanGDP more readily available for RCC1-mediated nucleotide exchange.
It has been proposed that the transfer of RanGDP from NTF2 to RCC1 might happen either at the NPC or in the nucleoplasm (35). NTF2 has been shown to bind to both RanGDP and XFXFG repeat-containing nucleoporins, such as yeast Nsp1p and vertebrate p62 (17,23). Using the crystal structure of rat NTF2 in complex with RanGDP, Bayliss et al. (27) designed an NTF2 mutant, W7A, in which the affinity for XFXFG repeat-containing nucleoporins is reduced but which retains wild-type binding to RanGDP. It was possible that an XFXFG repeat-containing nucleoporin, like p62, might be the RanGDF. However, GDF activity levels measured in the presence of the W7A mutant and wild-type NTF2, either alone or in the presence of SP200 (which does not contain endogenous NTF2 but which does contain the RanGDF), were identical (Figs. 1A and 3B). In addition, we prepared recombinant p62 and assayed its effect in the filter-binding assay. p62 did not affect [ 3 H]GDP dissociation from Ran pretreated with wildtype NTF2 either in the presence or absence of ATP (data not shown). It has also been reported (36) that the nucleoporin Nup153, which is located on the nucleoplasmic face of the NPC, is a RanGDP-binding protein. The zinc-finger region of Nup153 mediates interaction between RanGDP and Nup153 (36). We therefore prepared the zinc-finger domain of Nup153 in recombinant form. It, however, had no effect on [ 3 H]GDP dissociation from Ran prebound to wild-type NTF2 (data not shown). Our preliminary data therefore suggest that neither p62 nor Nup153 functions as a RanGDF.
Another candidate that interacts genetically with both Ran , preincubated with wild-type NTF2 (0.93 M) for 5 min at 30°C, and washed extensively. The SP200 fraction was added in the presence or absence of an ATP regeneration system, incubated for another 5 min at 30°C, and washed. Proteins bound to the beads were resolved by SDS-PAGE and analyzed by immunoblotting using anti-Ran or anti-NTF2 monoclonal antibodies. Wild-type NTF2 directly and specifically bound to RanGDP but not to RanGTP. The SP200 fraction, in the presence of an ATP regeneration system, displaced most of the NTF2 from RanGDP. were preincubated for 5 min at 30°C. The SP200 fraction was added in the presence or absence of the various ATP forms and incubated for another 5 min at 30°C. Finally, RCC1 (0.01 M) was added, and incubation continued for 5 min at 30°C. The reactions were carried out and evaluated as described in the Fig. 1 legend. and NTF2 is MOG1. The temperature-sensitive phenotype caused by deletion of MOG1 from yeast was suppressed by over-expression of NTF2 (37). Although both Ntf2p and Mog1p are required for optimal nuclear protein import in Saccharomyces cerevisiae, over-expression of MOG1 does not rescue ntf2 mutants, indicating that the functions of the two proteins are distinct. Stewart and Baker (38) have determined the crystal structure of Mog1p to 1.9 Å resolution and have suggested that Mog1p interacts with Ran through a site similar to that bound by NTF2. We therefore investigated whether Mog1 could function as a RanGDF using recombinant mouse Mog1 protein.
However, there was no detectable effect of Mog1 on the RanGDI activity of NTF2 (data not shown).
Our attempts to further purify the RanGDF have not met with success. Although further purification was sometimes achieved, this was not reproducibly the case. In addition, we failed to identify any protein bands within fractions separated by SDS-PAGE for which the presence was correlated with GDF activity. If, as our results suggest, a component of the RanGDF activity acts catalytically, as a kinase, the GDF may be present in active fractions in extremely small quantities. In any event, as summarized above, we have been unable to identify the GDF either by directed approaches or by further biochemical fractionation.
Finally, we propose some possible models for RanGDF function (Fig. 8). A RanGDF might specifically bind to NTF2 (1) or to RanGDP (2) and thereby dissociate the NTF2-RanGDP complex. In these models, RCC1 would interact with free RanGDP. There are two remaining formal possible modes of action for the activity. RanGDF might bind the NTF2-RanGDP complex and destabilize GDP binding to Ran directly (3). Alternatively, in an NTF2-dependent manner, RanGDF could interact directly with RCC1 and enhance its catalytic activity (4). Although we have not identified the ATP-dependent NTF2-releasing factor, the data presented provide the first evidence for the existence of a factor that releases RanGDP from NTF2. Our data further suggest that the function of the RanGDF is regulated by phosphorylation. Our future effort will be to continue the attempts to identify the factor and to characterize in detail its function in RanGTP-dependent processes. where indicated for 5 min at 30°C. The SP200 and the protein phosphatase inhibitor mixture were added as indicated and incubated for another 5 min at 30°C. Finally, RCC1 (0.01 M) was added, and incubation continued for 5 min at 30°C. The reactions were evaluated as described in the Fig. 1 legend. B, the effects of the protein phosphatase inhibitor mixture on the SP200. [ 3 H]GDP-Ran (0.67 M) was incubated with either 0.01 M RCC1 or the SP200 fraction for 5 min at 30°C as indicated in the presence (black bars) or absence (white bars) of protein phosphatase inhibitors. GDP dissociation was evaluated as described in the Fig. 1