Dissecting the Interactions between NTF2, RanGDP, and the Nucleoporin X F X FG Repeats*

We have used a range of complementary biochemical and biophysical methods to investigate the interactions between nuclear transport factor 2 (NTF2), the Ras family GTPase Ran, and X F X FG nucleoporin repeats that are crucial for nuclear trafficking. Microcalorimetry, microtiter plate binding, and fluorescence quenching in solution are all consistent with the binding constant for the NTF2-RanGDP interaction being in the 100 n M range, whereas the interaction between NTF2 and X F X FG repeat-containing nucleoporins such as Nsp1p is in the 1 m M range. Although the accumulation of NTF2 at the nuclear envelope is enhanced by RanGDP, we show that Ran binding does not alter the affinity of NTF2 for nucleoporins nor does the binding of nucleoporins alter the affinity of NTF2 for RanGDP. These results indicate that, instead, Ran increases the binding of NTF2 to nucleoporins by another mechanism, most probably by Ran itself binding to nucleoporins and NTF2 binding to this nuclear pore-associated Ran. (1 m M ) by monitoring the emission at 340 nm after excitation at 290 nm. Excitation and emission slidths were 2.5 and 5 nm, respectively. To test the effect of Ran on the affinity of the NTF2/Nsp1 interaction, a 2-fold molar excess of Ran (2 m M ) over NTF2 (1 m M ) was added. The 100 n M K d for the Ran-NTF2 complex indicated that 92% of the NTF2 should be complexed to RanGDP under those conditions. The interaction of NTF2 with Ran was assayed using NBD-labeled NTF2 (1 m M ) and monitoring the changes in NBD fluorescence emission followed at 506 nm after excitation at 435 nm. Excitation and emission slits were 15 and 20 nm, respectively. To test the effect of the presence of 18R-NSP1 on the affinity of the NTF2/Ran interaction, the same titration was repeated with a 50-fold molar excess of 18R-Nsp1 (50 m M ) over NTF2 (1 m M ). With a binding constant for the Nsp1-NTF2 complex of 25 m M (based on the concentration of a single X F X FG repeat), virtually all of the NTF2 should be complexed to 18R-Nsp1under those conditions. 50% ethylene glycol, and PBS containing 1% Triton X-100. Fractions were analyzed after Western blotting using a mono-clonal anti-NTF2 antibody (Transduction Laboratories) diluted 1/1000 in PBS-T and an anti-mouse horseradish peroxidase diluted 1/3000 in Tris-buffered saline with Tween 20 as secondary antibody. The reaction was visualized using the ECL kit (Amersham Pharmacia Biotech). For quantitation, films were digitized on a Molecular Dynamics Personal Densitometer and quantitated using Imagequant (Molecular Dynam-ics). The amount of NTF2 present in the purified material was deduced from a calibration curve produced with known amounts of NTF2 present on the same gel. The determination of the concentration of NTF2 in the cells was based on the dilution factor observed with radioactive NTF2 and applied to the amount of NTF2 as a fraction of the dry mass, assuming a density of about 1 ( i.e. 1 m g 5 1 m l) for both HeLa cells and rat liver homogenates. Assays— Import into nuclei of permeabilized HeLa cells was performed in suspension at 18 °C as described by others (14). Transport buffer was 20 m M Hepes-KOH, pH 7.5, 120 m M acetate, 5 m M acetate, 250 m M sucrose, 0.5 m M EGTA. I-Block from Tropix, Bedford, MA. NBD-Cl from Molecular Probes, Eugene, OR. PD10 or NAP10 columns were obtained from Pierce. All the concentrations given for NTF2 are given according to the amount of monomeric chains present.

ported from the nucleus and must be recycled back to the nucleus for exchange. HeLa cells contain 7 M Ran, being mainly nuclear (4). The nuclear import of RanGDP is mediated by NTF2, 1 a small dimeric protein that binds both RanDGP and to a number of nuclear pore proteins (nucleoporins) that contain tandem sequence repeats based on a series of core XFXFG motifs joined by variable hydrophilic linkers (5)(6)(7)(8). NTF2 is localized primarily to the nuclear rim (9 -11). NTF2 is an essential protein in yeast (11,12), and studies using engineered mutants of both Ran and NTF2 have shown that both the NTF2-RanGDP and NTF2-XFXFG repeat interactions are required for efficient nuclear import of Ran (13)(14)(15).
Although the interactions between RanGDP, NTF2, and XFXFG nucleoporins are essential for mediating the nuclear import of RanGDP, there is little quantitative information on the strength of these interactions or how they are modulated and orchestrated. We have therefore used a range of biochemical methods to assess these interactions and here show the strength of the interaction between NTF2-Ran to be in the 100 nM range, whereas that between NTF2 and the XFXFG repeats of a representative nucleoporin Nsp1p is in the micromolar range. We also show that the presence of Ran does not affect the strength of the interaction between NTF2 and the Nsp1 XFXFG repeats; conversely, the presence of Nsp1 does not affect the interaction between Ran and NTF2. The results obtained on purified proteins contrast with observations on permeabilized cells, where RanGDP increased the accumulation of NTF2 at the nuclear envelope (14). However, this effect is not seen with R76E-Ran, which does not bind NTF2 (16), indicating that accumulation of NTF2 at the nuclear envelope is mediated through a direct Ran-NTF2 interaction. The different models accounting for these results are evaluated.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-Rat NTF2, canine RanGDP, and the 18 XFXFG repeat-containing fragment of nucleoporin Nsp1p (18R-Nsp1) were expressed and purified as described (7,17,14). Nterminally S-tagged canine Ran was obtained by subcloning wild-type cDNA in pET30a and was expressed after induction with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 2 h in 2TY-kanamycin medium and purified using a slight modification of the protocol used to purify wild-type canine Ran, i.e. a 0 -100 mM linear NaCl gradient to elute it from the DE-52 column.
Microtiter Plate Binding Assay-NTF2 (0.7 M) in PBS was aliquoted onto a 96-well plate (Falcon 3911) and incubated for 2 h with gentle rocking. After blocking with PBS-I-block (0.2%), S-tagged Ran diluted in PBS containing 0.2% I-Block was added to the wells at a range of concentrations (0 -8 M) and incubated for either 2 h at room temper-ature or overnight at 4°C. After two washes in PBS, the plates were incubated with S-protein alkaline phosphatase conjuguate (Novagen) diluted 1/2000 in PBS, 3% BSA, 1 mM NaN 3 . After two washes with PBS buffer and 10 mM diethanolamine, pH 9.5, 0.5 mM MgCl 2 , the reaction was developed by incubating Sigma 104 substrate at 1 mg/ml dissolved in 10 mM diethanolamine, pH 9.5, 0.5 mM MgCl 2 for 10 -30 min at room temperature and the absorbance read at 405 nm. To determine binding constants, the data were fitted by nonlinear least-squares analysis using the program Scientist (MicroMath. Inc., Salt Lake City, UT).
NBD Labeling of NTF2-NTF2 dialyzed against PBS buffer containing 2 mM MgCl 2 was incubated with a 10-fold excess of NBD-Cl overnight at 4°C. The reaction was stopped and unreacted dye removed by passage through a PD-10 desalting column (Pierce) equilibrated in the same buffer. The labeled material absorbed strongly at 430 nm but showed negligible absorbance at 475 nm, indicating that the labeling had occurred primarily at cysteine residues (18). Using ⑀ M430 ϭ 13000 for NBD-Cys (18) indicated 1.1 NBD molecule was bound per NTF2 chain. NBD-labeled NTF2 formed a complex with Ran, as assessed by gel filtration on a HR 10/30 Superdex-75 FPLC column (Amersham Pharmacia Biotech) (data not shown) and also retained the ability to bind to 18R-NSP1 coupled to CNBr Sepharose (data not shown). We therefore concluded that NBD-NTF2 was equivalent to wild type in terms of its binding properties.
Binding Assays Using Fluorescence-Fluorescence measurements were performed on a Perkin-Elmer LS-50B spectrofluorimeter. The interaction of NTF2 with XFXFG nucleoporins was assayed using the intrinsic tryptophan fluorescence of NTF2 (1 M) by monitoring the emission at 340 nm after excitation at 290 nm. Excitation and emission slidths were 2.5 and 5 nm, respectively. To test the effect of Ran on the affinity of the NTF2/Nsp1 interaction, a 2-fold molar excess of Ran (2 M) over NTF2 (1 M) was added. The 100 nM K d for the Ran-NTF2 complex indicated that 92% of the NTF2 should be complexed to RanGDP under those conditions. The interaction of NTF2 with Ran was assayed using NBD-labeled NTF2 (1 M) and monitoring the changes in NBD fluorescence emission followed at 506 nm after excitation at 435 nm. Excitation and emission slits were 15 and 20 nm, respectively. To test the effect of the presence of 18R-NSP1 on the affinity of the NTF2/Ran interaction, the same titration was repeated with a 50-fold molar excess of 18R-Nsp1 (50 M) over NTF2 (1 M). With a binding constant for the Nsp1-NTF2 complex of 25 M (based on the concentration of a single XFXFG repeat), virtually all of the NTF2 should be complexed to 18R-Nsp1under those conditions.
Isothermal Titration Calorimetry-Binding of RanGDP with NTF2 was investigated by isothermal titration calorimetry (MicroCal Inc.). A 30 M solution of NTF2 in 20 mM potassium phosphate, pH 7.5, 5 mM MgCl 2 , 1 mM dithioerythritol was thermostated in the cell of the apparatus to 25°C. RanGDP was added stepwise up to a 3-fold final excess over NTF2 with the injection syringe of the instrument. Between each injection, a sufficient time delay was inserted to integrate all reactiondependent changes in the heat of the measuring cell (volume approximately 1.4 ml). Data were analyzed by the Origin software of the instrument manufacturer with a 1:1 model for the interaction. Experiments were repeated three times.
Cellular Concentration of NTF2-Radioactive NTF2, 35 S-NTF2, was expressed as the wild-type protein in the pET-based vector pMW172 in Escherichia coli BL21 (DE3) without induction, overnight at 34°C in 100 ml of 2TY medium supplemented with 176 mBq of 35 S-Easy Tag Express protein labeling mix (NEN Life Science Products), a mixture of [ 35 S]Met and [ 35 S]Cys. The bacterial cells were broken using lysozyme. After clarification of the lysate with DNase and centrifugation, the lysate was dialyzed against 0.85 M (NH 4 ) 2 SO 4 , 20 mM Tris, pH 8.0, 2 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride and loaded on a 1-ml phenyl-Sepharose column. After five column washes with 20 mM Tris, pH 8.0, 2 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, radioactive NTF2 was eluted with 40 ml of a 0 -50% ethylene glycol linear gradient in the same buffer. Rat liver homogenates were obtained as described (19). HeLa cells were grown in monolayer flasks in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. Cells were harvested with Dulbecco's modified Eagle's medium containing trypsin/EDTA, washed in cold PBS, and stored at Ϫ20°C as tared dry cell pellets. Rat liver homogenates were sonicated in presence of protease inhibitor mixture Set I (Calbiochem), centrifuged for 5 min at 5000 rpm at 4°C, and dialyzed against PBS buffer containing 1 M (NH 4 ) 2 SO 4 , whereas HeLa cells were resuspended directly in the same buffer. The dialysate or cell mixture was clarified by centrifugation for 5 min at 10,000 rpm and loaded onto a 1-ml phenyl-Sepharose column equilibrated in the same PBS buffer containing 1 M (NH 4 ) 2 SO 4 . After washing, elution was stepwise with PBS, PBS containing 50% ethylene glycol, and finally PBS containing 1% Triton X-100. Fractions were analyzed after Western blotting using a monoclonal anti-NTF2 antibody (Transduction Laboratories) diluted 1/1000 in PBS-T and an anti-mouse horseradish peroxidase diluted 1/3000 in Tris-buffered saline with Tween 20 as secondary antibody. The reaction was visualized using the ECL kit (Amersham Pharmacia Biotech). For quantitation, films were digitized on a Molecular Dynamics Personal Densitometer and quantitated using Imagequant (Molecular Dynamics). The amount of NTF2 present in the purified material was deduced from a calibration curve produced with known amounts of NTF2 present on the same gel. The determination of the concentration of NTF2 in the cells was based on the dilution factor observed with radioactive NTF2 and applied to the amount of NTF2 as a fraction of the dry mass, assuming a density of about 1 (i.e. 1 g ϭ 1 l) for both HeLa cells and rat liver homogenates.
Import Assays-Import into nuclei of permeabilized HeLa cells was performed in suspension at 18°C as described by others (14). Transport buffer was 20 mM Hepes-KOH, pH 7.5, 120 mM potassium acetate, 5 mM magnesium acetate, 250 mM sucrose, 0.5 mM EGTA.
I-Block was from Tropix, Bedford, MA. NBD-Cl was from Molecular Probes, Eugene, OR. PD10 or NAP10 columns were obtained from Pierce. All the concentrations given for NTF2 are given according to the amount of monomeric chains present.

RESULTS
K d for the NTF2-RanGDP Interaction -We used three complementary methods to determine the K d for the NTF2-RanGDP interaction. Using microtiter plates coated with 0.7 M NTF2, S-tagged RanGDP showed a K d of 240 Ϯ 90 nM ( Fig.  1A), whereas plates coated with the NTF2 E42K mutant (which does not bind RanGDP (Ref. 8)) did not bind RanGDP (data not shown). We also used microcalorimetry and fluorescence to assess the strength of the NTF2-Ran interaction in solution. Microcalorimetry gave a K d of 150 Ϯ 30 nM (Fig. 1C). Using NTF2 fluorescently labeled on cysteine residue with NBD-Cl (an environment-and conformation-sensitive probe (Ref. 20)), the decrease in fluorescence at 506 nm on addition of Gsp1p (the yeast Ran homologue) was concentration-dependent and gave a K d of 75 Ϯ 40 nM (Fig. 1B). However, canine RanGDP did not produce any change in fluorescence, even though NBD-NTF2 was qualitatively equivalent to wild-type NTF2 in terms of its binding to Ran and Nsp1 using column chromatography or bead-binding assays (data not shown). The crystal structure of the Ran-NTF2 complex (21) shows that the 3 cysteines of NTF2 are located on a face of the molecule opposite and distant from the Ran binding site, and so it is unlikely that the NBD and Ran/Gsp1p interact directly. It is possible that binding NTF2 to Gsp1p produces a slight conformation change reflected by the change in fluorescence emission of the attached NBD group. However, in the case of canine RanGDP, the binding on NTF2 must be slightly different, so that this change in conformation does not occur. Because of the sensitivity of NBD probes to small changes in environment, a faint difference in the way canine Ran and its yeast homologue Gsp1p bind to NTF2 could certainly account for the effect observed. Unfortunately, neither the crystal structure of the Gsp1p nor its complex with NTF2 have yet been solved. None of the binding isotherms obtained gave any indication of cooperativity between the two Ran binding sites on an NTF2 dimer molecule; the curves were all fitted to a simple isotherm with no allosteric curvature shape. This would imply that each binding site for Ran on an NTF2 chain in the dimer behaved independently. Additionally, the stoichiometry of binding deduced from the binding between NBD-NTF2 and Gsp1 (Fig. 1B) was approximately 1 Gsp1 molecule bound per NTF2 chain.
RanGDP Binding Does Not Alter the Affinity of NSP1 XFXFG Repeats for NTF2-We used intrinsic tryptophan fluorescence to determine the K d for the interaction between NTF2 and a bacterially expressed construct containing 18 XFXFG repeats from yeast nucleoporin Nsp1p (18R-NSP1).
Binding isotherms (Fig. 3A) showed no cooperativity, indicating that the XFXFG repeat-binding site on each chain of the NTF2 dimer behaved independently. We therefore, as a first approximation, analyzed the data assuming that all 18 XFXFG repeats in the Nsp1p construct and the two chains in the NTF2 molecule were able to interact independently. A K d value of 1.4 Ϯ 0.21 M for the whole Nsp1 protein or 25 M (based on the concentration of individual XFXFG motifs) was obtained (Fig.  3A). As illustrated in Fig. 3B, when the titration was repeated in the presence of a 2-fold molar excess of RanGDP over NTF2 (when 92% of the NTF2 would be complexed with Ran) we obtained a K d of 1. Addition of the Nsp1p XFXFG repeats to NTF2 produced an increase in fluorescence when added to free NTF2 and a decrease in fluorescence when added to NTF2 in the presence of RanGDP (Fig. 3, A and B). This suggests that Ran binding may have altered the local environment of the XFXFG repeat binding site on NTF2. X-ray crystallography (21) has indicated that, although RanGDP binding causes a small rigid-body rotation of the two chains in the NTF2 dimer, the overall conformation of each chain in the complex is similar to that observed in uncomplexed NTF2 (22). Therefore, the change in tryptophan environment resulting from XFXFG repeat binding is due to the proximity of the bound Ran rather than Ran binding producing a substantial conformational change in NTF2. Indeed, such an explanation would also be consistent with the negligible change in NTF2's affinity for XFXFG repeats observed in the presence of RanGDP. Addition of the 18RNsp1p repeats to RanGDP did not produce any change in fluorescence when tested under the same conditions as for free NTF2 (data not shown) indicating that this effect was not due to the XFXFG repeats altering the Ran intrinsic fluorescence.
The Presence of Nsp1p XFXFG Repeats Does Not Change the Affinity of NTF2 for Ran-To establish whether the nucleoporin XFXFG repeats would alter the interaction between NTF2 and RanGDP, we determined the value for the K d between Gsp1p and NTF2 in the presence of an excess of 18R-Nsp1 (10 times the K d ). As before, we then followed the decrease in fluorescence at 506 nm upon titration of NBD-NTF2 with Gsp1p. In this assay, we are detecting only the binding of RanGDP to NTF2, since we are monitoring the change in fluorescence of the NBD-labeled NTF2 molecule, regardless of whether or not Gsp1p is capable of binding to Nsp1p as well. The K d obtained for the interaction between NTF2 and Gsp1p in the presence of 18R-Nsp1p was indistinguishable from that obtained for NTF2 and Gsp1p alone (Fig. 2). Thus, NTF2 free in solution or bound to Nsp1p has the same affinity for RanGDP.
Total NTF2 Concentration in Rat Liver Homogenates and HeLa Cells -To place the K d values obtained for the interaction between NTF2 and Ran in cellular context, we determined the total cellular concentration of NTF2 using isotope dilution. Ran is present at 7 M in HeLa cells, of which 80% is thought to be nuclear (4). For NTF2, we found a total concentration of 2 M both in rat liver homogenates and HeLa cells. NTF2 is located primarily at the nuclear rim (10,11) and concentrated there by about 120-fold compared with total NTF2 (9), which correlates with the determined cytosolic concentration of NTF2 of 0.3 M (6). Calculations based on these concentrations and the 100 nM K d for the NTF2-RanGDP interaction indicate that the NTF2-Ran complex will be found mainly localized at the nuclear envelope (Table I).
Effect of Ran or R76E-Ran on the Accumulation of NTF2 at the Nuclear Envelope -RanGDP enhances accumulation of NTF2 at the nuclear envelope (14), raising the possibility of a cooperative interaction between NTF2, RanGDP, and a NPC constituent. Our results on purified proteins show that Ran does not have any cooperative effect on the binding of NTF2 to 18R-Nsp1, since the K d obtained for the interaction between NTF2 and 18R-Nsp1 was the same, whether NTF2 was complexed with RanGDP or not (see Fig. 3). To assess whether the Ran-induced increase in NTF2 accumulation at the nuclear envelope required a direct interaction between Ran and NTF2, we used the R76E Ran mutant, which does not interact with NTF2 although it retains the ability to bind nucleoporin XFXFG repeats and other components of the nuclear trafficking machinery (16). Unlike wild-type Ran, addition of the R76E mutant did not increase the accumulation of fluorescently labeled NTF2 at the nuclear envelope (Fig. 4), demonstrating that direct interaction between Ran and NTF2 is necessary to obtain the enhanced accumulation of NTF2 at the nuclear envelope.

DISCUSSION
The results using three separate methods were broadly comparable giving a K d value in the order of 100 nM (Fig. 1) for the RanGDP-NTF2 interaction. Although the microtiter plate assay gave a slightly weaker interaction (by 2-or 3-fold), this difference was probably not significant as the standard deviations obtained with this assay were quite large. It was therefore reassuring that two different assays (microplate and solution fluorescence quenching) gave comparable results. Similar K d values were obtained for canine Ran and the yeast homologue Gsp1, whereas addition of sequences at the Ran N terminus such as the S and His tags decreased the affinity of RanGDP for NTF2 by a factor not more than 2. The on-rate association between NTF2 and Ran GDP would be expected to be in the range 10 7 to 10 8 s Ϫ1 M Ϫ1 (23), and so a 100 nM K d would imply an off-rate of the order of 1-10/s, which is somewhat slower than the rate of nuclear trafficking, which is thought to be of the order of 10 -100/s. Thus, the binding constant for the NTF2-RanGDP complex is sufficiently strong to ensure that the complex remains intact during nuclear import but sufficiently weak for it to dissociate on a time scale rapid relative to nucleotide exchange once in the nucleus. Moreover, combined with the likely concentrations of Ran and NTF2, a K d in the 100 nM range indicated that, whereas RanGDP would be expected to be only partially complexed in the cytoplasm and nucleus, at the nuclear envelope (primarily in nuclear pore complexes) virtually all the RanGDP present  a The calculated concentration of NTF2 at the NE is a minimal approximation considering total cell concentration: 2 M (this study), a 10 times accumulation factor of NTF2 at the NE (9), and the negligeable volume occupied by the NE.
b Approximation from Ref. 10, and our data (not shown). c Approximation was made from the total known cell content being 7 M (4), of which 80% being nuclear, from that considered that 90% of the Ran in the nucleus would be in the GTP-bound state and 90% of the Ran in the cytosol would be in the GDP-bound state. d Considered the same concentration as the one found in the cytosol. Calculations for the fractions bound were derived from the equation: should be complexed with NTF2 (Table I). Two implications for nuclear import can be drawn from those results. First, the stable nature of the complex between RanGDP and NTF2 (offrate of the order of 1-10/s) would mean that the complex probably stays intact during trafficking. Second, the higher local concentration of NTF2 at the NE implies that 99% of the Ran present at the NE would be found in such a complex; hence, even if dissociation of RanGDP from the NTF2-Ran complex would occur, the reassociation would occur quickly. Those conclusions are consistent with a model for the nuclear import of Ran that would cross the NE in complex with NTF2 (Fig. 6).
The affinity of NTF2 for the Nsp1 repeats is weaker than that between NTF2 and Ran and was of the order of 25 M based on the concentration of individual XFXFG repeats (Fig.  3). However, this lower affinity does not preclude the possibility of this interaction to occur in vivo at the NPC, since the XFXFG repeats are present in high copies number on several nucleoporins at the NE and therefore their local concentration is expected to be high, probably of the order of 50 mM (15). The off-rate for the NTF2/Nsp1 interaction has been estimated to be in the order of 500 -5000/s for an individual repeat (15), implying that this interaction would be more transient than that between NTF2 and RanGDP. This would allow rapid attachment and detachment of NTF2 to the nucleoporins. In the context of the more stable interaction between NTF2 and Ran, a 25 M affinity of NTF2 for nucleoporin XFXFG repeats would also enable rapid attachment and detachment of the RanGDP-NTF2 complex to the nucleoporins during trafficking. Such a mechanism would be consistent with the results showing the crucial importance of a direct interaction with NTF2 for the import of RanGDP (13,14).
Our results showed no synergy between the interactions involving RanGDP, NTF2, and Nsp1. RanGDP had no effect on the affinity of NTF2-Nsp1 interaction, nor did the presence of Nsp1 have an effect on the Ran-NTF2 interaction (Figs. 2 and 3). The crystal structure of NTF2 either free (22) or in complex with RanGDP (21) are certainly consistent with those results, since there is no major conformational changes in the protein backbone resulting from the RanGDP binding on the NTF2 molecule.
Previous work using permeabilized cells had shown that Ran increased the accumulation of NTF2 at the nuclear envelope implying a co-operativity between Ran, NTF2, and a constituent of the NPCs (14). However, our in vitro results with purified proteins do not support such a co-operative role for RanGDP in the NTF2-Nsp1 interaction (Fig. 3). How then can one account for the increased nuclear envelope binding of NTF2 induced by Ran? Certainly, our results ( Fig. 4) with R76E-Ran that does not bind to NTF2 (16) indicated that a direct interaction between NTF2 and Ran was required for this effect. Therefore, as illustrated in Fig. 5, either NTF2 is interacting with another type of nucleoporin in a Ran-sensitive manner or NTF2 is binding to RanGDP that is itself bound to the NPC. Thus, in model A, NTF2 binds a non-XFXFG repeat containing nucleoporin with higher affinity when NTF2 is bound to RanGDP (Fig. 5A). Although results obtained with an NTF2 mutant (W7A-NTF2) with reduced affinity for XFXFG repeat-containing nucleoporins are consistent with this hypothesis (15), it is FIG. 4. Influence of the presence of wild-type Ran or R76E Ran on the accumulation of fluorescently labeled NTF2 at the nuclear envelope. Fluorescein-labeled NTF2 (0.9 M; dimer) was incubated with permeabilized cells. Where indicated, the following additions had been made: energy-regenerating mix (0.5 mM GTP, 0.5 mM ATP, 10 mM creatine phosphate, and 50 g/ml creatine kinase), 2.3 M wild-type RanGDP, 2.3 M R76E RanGDP. Import was initiated at 18°C by the addition of nuclei (10 6 ). Nuclei were fixed after 10 min with 4% paraformaldehyde, spun onto coverslips, and analyzed by confocal microscopy using a 63ϫ objective and oil immersion. Note that, for all images, scanner laser power and treatment of image were done under the same conditions. NPC binding of NTF2 was greatly enhanced by the addition of RanGDP, but not by the addition of the mutant R76E RanGDP, stating that a direct interaction between NTF2 and RanGDP is crucial for this process. FIG. 5. Different possible models accounting for the cooperative binding of NTF2 at the nuclear envelope upon Ran addition. In all the models described, N stands for NTF2 and is represented as a dimer, Ran stands for RanGDP, and it is understood that the interaction between RanGDP and NTF2 is essential for the cooperative binding. Nucleoporin X stands for a putative nucleoporin potentially interacting with NTF2 via a different set of interaction from the XFXFG repeat type. The increased accumulation of NTF2 at the NE upon Ran addition could result from an increased affinity of NTF2 for nucleoporin X when NTF2 is bound to Ran (model A). Alternatively, Ran is directly anchoring NTF2 on a nucleoporin with or without simultaneous binding of NTF2 to this nucleoporin (model B), or Ran is anchoring NTF2 to the nucleoporin via an intermediate protein (model C).
controversial whether NTF2 binds to nucleoporins other than those with XFXFG repeats. Although a number of putative bands were seen with gel overlays using gold-NTF2 conjugates (24), bead binding and two-hybrid screens have only shown interactions with XFXFG nucleoporins (8). Alternatively, the NTF2-RanGDP complex could interact with another protein anchored to the NPC (Fig. 5C), but no stable complex in solution with importin-␤ (data not shown) nor with transportin-RanGDP-NTF2 complex could be formed (14). The most probable model explaining the accumulation of NTF2 upon addition of RanGDP at the NE in the permeabilized cell assay would then be that illustrated in Fig. 5B, where Ran associated with a nucleoporin also binds NTF2 (with the possibility that RanGDP and NTF2 might simultaneously bind to the nucleoporin). The nucleoporins RanBP2 and Nup153 have been shown to interact with RanGDP through Zn finger domains (25,26); moreover, both nucleoporins also contain XFXFG repeats, which are potential binding sites for NTF2.
In summary, the affinity constant for the Ran/NTF2 complex and the NTF2/XFXFG repeat-containing nucleoporins such as Nsp1p complex are 100 nM and 1 M, respectively. Upon RanGDP binding, there is no alteration for the affinity of NTF2 for nucleoporins. Conversely, upon Nsp1p binding there is no alteration for the affinity of NTF2 for RanGDP. In permeabilized cells, an enhanced accumulation of NTF2 at the NE proceeds through a direct Ran-NTF2 interaction indicating most probably that Ran itself ensures the binding of NTF2 to nucleoporins via a nuclear pore-associated Ran.
FIG. 6. Schematic representation for a model for Ran import. Off-rates for the Ran/NTF2 and NTF2/XFXFG repeat binary complexes have been estimated at 1-10/s (see "Results") and 540 -5400/s, respectively (15). Our results show that Ran does not change the affinity of NTF2 for the nucleoporins, nor does the binding of nucleoporins alter the affinity of NTF2 for RanGDP. Thus, the off-rate for the Ran-NTF2 complex at the nucleoporin is expected to be equivalent to the one for the NTF2/XFXFG interaction. Such differences in stability are consistent with a model where the Ran/NTF2 complex would stay intact during the translocation across the NE (rate of nuclear trafficking of 10 -100/s), whereas there would be a rapid attachment/detachment to the nucleoporins. Local concentrations of Ran, NTF2, and nucleoporins favor the formation of the Ran-NTF2 complex at the cytoplasmic face of the NE (Table I). The off-rates between the different complexes are in favor of a "hopping" of the complex from one nucleoporin to the next along the central channel of the pore bordered with several XFXFG repeats containing nucleoporins (27). Once delivered in the nucleus, the exchange of GTP for GDP on Ran by the exchange factor RCC1 will favor the disruption of the Ran/NTF2 complex.

Stewart Catherine Chaillan-Huntington, Carolina Villa Braslavsky, Jürgen Kuhlmann and Murray
Repeats FG X F X Dissecting the Interactions between NTF2, RanGDP, and the Nucleoporin