Mutations within the Ran/TC4 GTPase. Effects on regulatory factor interactions and subcellular localization.

Ran, a member of the Ras superfamily of GTPases, is predominantly localized in the nucleus and is a necessary component in the active transport of proteins through nuclear pores. Disruption of Ran function affects the regulation of mitosis, DNA synthesis, and RNA processing and export. To explore the mechanisms of Ran function, mutants of the Ran GTPase were characterized, several of which are capable of dominantly interfering with nuclear protein import. Unlike wild-type Ran, the putative gain-of-function mutant (G19V Ran) was not sensitive to the exchange factor, RCC1. In addition the G19V Ran and effector domain mutants (L43E and E46G Ran) were not sensitive to the GTPase-activating protein, Fug1. Epitope-tagged G19V Ran and L43E Ran isolated from transfected BHK21 cells were each about 50% GTP-bound, whereas the wild-type and a C-terminal deletion mutant (Δ-DE Ran) were primarily bound to GDP. While G19V Ran interacted with known Ran-binding proteins and with an isolated Ran-binding domain, the T24N Ran did not, and binding by L43E Ran was substantially reduced. Wild-type HA1-tagged Ran expressed in BHK21 cells was nuclear, whereas the G19V, T24N, L43E, and E46G forms of Ran were predominantly localized at the nuclear envelope, and Δ-DE Ran was primarily cytosolic. Similar results were observed when permeabilized BHK21 cells were incubated with extracts of COS cells expressing the mutants. Thus mutations that affect the interaction of Ran with regulatory proteins and effectors can disrupt the normal subcellular localization of Ran, lending support for the current model of Ran-mediated nuclear import.

Ran/TC4 is a 25-kDa protein belonging to the Ras superfamily of GTPases and is highly conserved among all eukaryotes (1,2). Ran is unlike other small GTPases in its predominantly nuclear localization and lack of any prenylation sites at the C terminus (3). Instead, the C terminus of Ran consists of a string of negatively charged residues (-DEDDDL).
Ran functions are modulated by the binding and hydrolysis of GTP and are influenced by its interaction with the GTP exchange factor RCC1 1 and with the GTPase-activating protein (RanGAP1 or Fug1) (4 -6). Blocking RCC1 function has profound effects on multiple nuclear events, including progression of the cell cycle, RNA transport, and DNA replication (7). Ran-GAP1 and Fug1 are human and mouse homologs of the budding yeast protein Rna1p, which is necessary for proper RNA processing and export (5) and nuclear protein import (8).
A family of proteins has been described that interacts with Ran in the GTP-bound state and inhibits GTP release (9). Some of these Ran-binding proteins (RanBPs) have been identified. One is a 29-kDa cytosolic protein referred to as HTF9A/ RanBP1 (10), and another is a 350-kDa protein localized at the nuclear membrane termed Nup350 or RanBP2 (11,12). Alignment of these and other similar proteins has revealed a 135amino acid residue Ran-binding domain (RanBD), which is necessary and sufficient for binding to Ran (13). In addition, Ran can associate, through a different type of domain, with the ␤-subunit of the nuclear cargo receptor, ␤-karyopherin/␤-importin/p97 (14). The interaction of Ran with ␤-karyopherin is not inhibited by the RanBD, suggesting that Ran may interact with a complex of nuclear transport proteins including a RanBP and ␤-karyopherin (15)(16)(17).
The Ran GTPase has been shown to be necessary for the in vitro nuclear import and export of synthetic karyophiles possessing nuclear localization signals (NLSs) (for review, see Refs. 14 and 18). Recently, our laboratory demonstrated that several mutants of Ran can dominantly interfere with the dexamethasone-stimulated nuclear import of glucocorticoid receptors expressed in BHK21 cells (19). A putative gain-offunction mutant (G19V Ran), a loss-of-function mutant (T24N Ran), and a C-terminal deletion mutant (⌬-DE Ran) all inhibited protein nuclear import. However, an effector domain mutant, L43E Ran, had no effect on import, yet was markedly toxic to cell growth. Ran, therefore, possesses an essential function, dominantly inhibited by the L43E Ran mutant, that is independent of nuclear protein import. To understand the molecular mechanisms underlying these cell biological effects, it was essential to characterize the Ran mutants biochemically. The present study describes the subcellular distribution of the mutants and their interactions with known regulatory factors. subcloned into the bacterial expression vector, pGEX2T, and the mammalian expression vector, pCH3. The phenotypes of each individual mutant were the same in all experiments as the double mutant (data not shown). The pCH3 plasmid is a pUC-based vector that provides three sequential HA1 epitope tags at the N terminus of the protein (20). The C-terminal deletion mutant, ⌬-DE Ran, was created as described previously (9).
To generate GST-Ran fusion proteins, DH5␣ Escherichia coli transformed with pGEX2T-Ran were grown to A 600 ϭ 1, and then induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 16 h at 23°C. GST fusion proteins were purified from cell lysates using a glutathione-Sepharose matrix (Pharmacia Biotech Inc.) and concentrated to 1 mg/ml using a Centricon-10 concentrator (Amicon). Protein concentrations were determined by a Bradford protein assay. Protein degradation during preparation was monitored by SDS-polyacrylamide gel electrophoresis and Coomassie staining.
GAP Assays-The plasmid encoding the murine homolog of human RanGAP1, Fug1, was supplied by J. DeGregori (NCI, National Institutes of Health) (6). GST-Fug1 was expressed and purified as described above. GTP hydrolysis stimulated by GST-Fug1 was measured as described previously (21). Briefly, 0.5 g of GST-Ran was loaded with [␣-32 P]GTP as described (9). Radiolabeled GST-Ran was diluted 15-fold with GAP buffer (40 mM Tris-HCl, pH 8.0, 8 mM MgCl 2 , 80 M BSA, and 2 mM GTP) in a final volume of 150 l at 23°C. RanGAP activity was supplied by the addition of GST-Fug1 at a final concentration of 21.2 pg/l; 21.2 pg/l GST was used as a negative control. Fractions (25 l) were removed at specified times and filter-bound as above. Nucleotides were released from the filters and separated by thin layer chromatography (22). Quantitation was performed using a Bio-Rad GS-250 PhosphoImager.
Nucleotide Release Assays-Release of GDP from Ran was measured as described previously (21). Briefly, 2 g of Ran (thrombin cleaved to remove GST) was loaded with 20 Ci of [␣-32 P]GDP (3000 Ci/mmol, DuPont NEN) then diluted 10-fold in buffer containing 50 mM MOPS, pH, 7.1, 2 mM GDP, 1 mM dithiothreitol, 10 mM MgCl 2 , and 0.1 mg/ml BSA in a final volume of 50 l. RCC1 (a gift from T. Nishimoto Kyushu University, Fukuoka, Japan) was expressed in DH5␣ E. coli and added to the assay as a crude cytosolic fraction (1:100 dilution). Bacterial lysate not containing RCC1 was added to control samples. At specified times, 5 l of the sample was subjected to nitrocellulose filter binding and the retained ␣-32 P-labeled nucleotide was quantitated by scintillation counting. 100% bound represents the amount of ␣-32 P-nucleotide bound at t ϭ 0.
RanBP Interactions-The Ran overlay was performed as described in Ref. 9 with the following modifications. BHK21 cell extract was prepared by adding boiling SDS-sample buffer directly to cells and shearing the DNA by passage through a 26-gauge needle. Approximately 100 g of BHK21 extract or 1 g of recombinant GST-Ran-binding domain (13) were separated by 8% SDS-PAGE and transferred to nitrocellulose. Transfers were renatured and overlaid with equal amounts of [␣-32 P]GTP-labeled GST fusions of Ran (approximately 20,000 cpm/ml bound counts). Nucleotide labeling was performed as described previously (9), except that the free nucleotide was removed prior to use by applying the sample to a Centrisep G-25 spin column (Princeton Separations).
Binding of ␤-karyopherin/␤-importin and RanBP1 by Ran mutants was measured by incubating 1 g of GTP-loaded Ran and 50 l of glutathione-Sepharose beads, with 1 g of recombinant ␤-karyopherin or RanBP1 (GST fusion protein cleaved with thrombin; Ref. 21), in a final volume of 1 ml of Ran binding buffer (15). Following 1 h of agitation at 4°C, the beads were collected by centrifugation and washed with Ran binding buffer without BSA. Pellets were resuspended in Laemmli sample buffer, separated by 10% SDS-PAGE, and immunoblotted with anti-T7-Tag (Novagen) antibodies or subjected to the Ran overlay assay as described above.
GTP:GDP Ratios of Ran Mutants Expressed in BHK21 Cells-BHK21 cells were transfected using CaPO 4 (20) with 10 g of pCH3-Ran/10-cm plate. Transfected cells were incubated in phosphate-free Dulbecco's modified Eagle's medium for 30 min and labeled by incubating for 4 h with 0.5 mCi of [ 32 P]orthophosphate (DuPont NEN) per 10 cm dish. Cells were washed with phosphate-buffered saline then harvested into 100 l of unlabeled lysis buffer containing 50 mM MOPS, 0.5% Triton X-100, 150 mM NaCl, 10 mM NaPO 4 , 20 mM MgCl 2 , 100 M phenylmethylsulfonyl fluoride, 10 M aprotinin, and 10 M leupeptin. Immunoprecipitation with the monoclonal anti-HA1 antibody, 12CA5 (Boehringer Mannheim), was performed as described (20) using 3 l of antibody bound to 50 l of protein A-Sepharose (Sigma) per sample. Nucleotides were released from the precipitates by the addition of buffer containing 1% SDS and 25 mM EDTA and separated by thin layer chromatography as described previously (22). The relative amounts of GTP to GDP were quantitated using a Bio-Rad PhosphoImager GS-250, and the proportion of GTP bound was calculated using the formula %GTP ϭ GTP/(GTP ϩ 1.5 ϫ GDP) to account for the number of phosphates per nucleotide.
As a control for GTP hydrolysis during immunoprecipitation, unlabeled BHK21 extracts were incubated with 3 g of GST-Ran loaded with 10 Ci of [␣-32 P]GTP (as above) and precipitated with glutathione-Sepharose beads using the same conditions and analysis as for the 12CA5 immunoprecipitation.
In Vitro Localization Assay-COS-7 cells transfected with pCH3 or pCH3-Ran constructs (10 g of DNA/10-cm plate) (20) were washed with Buffer A (20 mM HEPES-KOH, pH 7.3, 2 mM magnesium acetate, 2 mM dithiothreitol, 10 M leupeptin, 10 M aprotinin, and 1 mM phenylmethylsulfonyl fluoride) then washed with 500 l of Buffer A containing 0.035% digitonin and aspirated to near dryness. Cells were scraped into a microcentrifuge tube and incubated on ice for 15 min. Lysates were passed through a 26-gauge needle then centrifuged at 12,000 ϫ g for 2 min to remove particulates. Approximately 2 mg of cytosol protein was obtained from five 10-cm dishes.
Nuclear accumulation of Ran was observed using the digitonin-permeabilized import assay of Moore and Blobel (23) with the following modifications. BHK21 cells grown on eight-well glass slides (Nunc) were permeabilized for 5 min on ice with digitonin (35 g/ml) in Buffer A. Wells were rinsed twice with Buffer A, and incubations were performed at 30°C for 15 min. A standard assay mixture contained BSA (1 mg/ml), COS cell digitonin extract (1.5 mg/ml), 1 mM ATP, 5 mM phosphocreatine, and 20 units/ml creatine phosphokinase in a final volume of 100 l with Buffer A. Nuclear import substrate (NLS-HSA) at 20 g/ml, or wheat germ agglutinin (0.5 mg/ml) were added where indicated. The condition "No ATP" included the removal of ATP and the ATP regeneration system and the addition of 5 mM glucose and 16 units/ml hexokinase. To generate NLS-HSA, tetramethyl rhodamine-5isothiocyanate labeled human serum albumin was conjugated to a peptide containing the nuclear localization sequence of the SV40 Tantigen (Multiple Peptide Systems) (24) by the protocol used in (23).
Following incubations, cells were washed three times with ice-cold Buffer A and fixed with 3% paraformaldehyde in Buffer A for 15 min on ice. Slides were then either mounted with Vectashield (Vector) or first processed for indirect immunofluorescence using the anti-HA1 antibody 12CA5.
Immunologic Techniques-For immunofluorescence, cells were fixed with 3% paraformaldehyde for 15 min, permeabilized with Ϫ20°C methanol for 2 min, and blocked with 2% milk in Tris-buffered saline plus 0.1% Tween 20 (TBS-T) for at least 30 min. Cells were then incubated with a 1:250 dilution of 12CA5 antibody in 2% BSA/TBS-T for 1 h at 37°C. After washing with TBS-T, detection of antibody was accomplished by incubating the cells with a 1:500 dilution of Cy3-or fluorescein-conjugated anti-mouse secondary antibody (Jackson Laboratories) in 2% BSA/TBS-T for 1 h. Cells were then washed extensively with TBS-T and mounted with Vectashield. Fluorescence was detected using a Nikon inverted microscope at 40ϫ magnification. Confocal microscopy was performed using an Olympus BX50 microscope at 40ϫ magnification and an MRC-1000 laser scanning imaging system (Bio-Rad). Immunoblotting with the anti-HA1 antibody, 12CA5, was performed as described previously (19).

Effects of Mutations within Ran on Interactions with RCC1
and RanGAP-Ran mutations were examined that can dominantly interfere with nuclear transport and/or cell growth in an in vivo assay (19). The G19V and T24N mutations of Ran were based on analogous residues that affect either the guanyl nucleotide state or interactions with regulatory proteins of Ras (25). These mutants of Ran exhibit dominant effects on cell cycle progression and nuclear functions (26 -28). The L43E Ran and E46G Ran "effector domain" mutants are analogous to mutations at residue Val-55 and Asp-58 of Rab3A. Mutation at Val-55 of Rab3A disrupts interaction with all regulatory proteins and downstream effectors, whereas the Asp-58 mutant retains sensitivity to Rab3-GRF and a downstream effector, Rabphilin (22,29,30).
To determine the ability of these Ran mutants to interact with RanGAP, Ran proteins loaded with [␣-32 P]GTP were in-cubated with a recombinant GST fusion protein of Fug1. Fug1 is a murine gene with strong sequence similarities to human RanGAP1 and Saccharomyces cerevisiae RNA1p (5, 6) but had not previously been shown to function as a RanGAP. The sensitivity of the Ran mutants to RCC1 was measured as a function of GDP released in the presence of recombinant RCC1 and Mg 2ϩ .
GST-Fug1 exhibited potent RanGAP activity for wild-type GST-Ran (Fig. 1), and RCC1 stimulated GDP release from wild-type GST-Ran with a t 1/2 of 3 min (Fig. 2A). The GST-G19V Ran mutant exhibited minimal GTPase activity even in the presence of Fug1 (Fig. 1) and also was insensitive to GDP release stimulated by RCC1 ( Fig. 2A). The GST-T24N Ran mutant did not bind nucleotide in a stable fashion even in the presence of Mg 2ϩ ; thus the hydrolysis activity and interaction with RCC1 were not measurable by these assays. As predicted from the analogous mutations in Ras and Rab3A, the GST-L43E Ran and E46G Ran (data not shown) mutants were insensitive to RanGAP-stimulated GTP hydrolysis ( Fig. 1). At a 1000-fold higher concentration of GST-Fug1, L43E Ran did exhibit a low level of GAP-stimulated GTP hydrolysis, suggesting that the mutation did not totally block the interaction of Fug1 with Ran (data not shown). Additionally, unlike Rab3A, both Ran effector domain mutants and wild-type proteins showed similar sensitivities toward the guanine nucleotide exchange factor RCC1 ( Fig. 2A).
T24N Ran can dominantly inhibit the wild-type Ran-RCC1 interaction (31), and, in fluorescence measurements, T24N Ran potently inhibits RCC1-catalyzed nucleotide exchange on Ran (28). As shown in Fig. 2B, T24N Ran was able to compete for the binding of RCC1 with an affinity approximately 100-fold higher than that of the wild-type protein. These data confirm the hypothesis that a stable complex can be formed between T24N Ran and RCC1, allowing T24N Ran to compete for RCC1 with greater efficacy than wild-type Ran (28).
Effect of Ran Mutations on Interaction with RanBPs-To determine if the mutations in Ran described above are important for the interaction of Ran with RanBPs, GTP-Ran overlay assays of cell extracts and precipitations of purified RanBPs by the addition of glutathione-Sepharose beads were performed (Fig. 3). For the overlay assay, BHK21 cell extract and an isolated Ran-binding domain of RanBP2 (RanBD3) (11,13) were immobilized on nitrocellulose and then incubated with [␣-32 P]GTP-labeled GST-Ran proteins. Both wild-type Ran and G19V Ran clearly detected RanBP1, RanBP2, and the RanBD by the overlay assay (Fig. 3A). Using glutathione-Sepharose precipitation from solution, G19V precipitated more RanBP1 than wild-type Ran, but less ␤-karyopherin, suggesting that the requirements for the binding to RanBP1 and ␤-karyopherin may differ (Fig. 3, B and C). The binding of GST-T24N Ran to RanBPs was not observed by the overlay assay or by precipitating recombinant RanBP1 or ␤-karyopherin (Fig. 3).
Mutations in Ran that do not directly perturb guanine nucleotide binding also had effects on the binding of Ran to RanBPs. GST-L43E Ran exhibited reduced binding to all Ran-BPs on overlays and by precipitation, despite its ability to bind GTP and its insensitivity to RanGAP (Fig. 3). This result demonstrates that interactions with RanBPs are sensitive to the conformation of the Ran effector domain. As described previously (15), the ⌬-DE mutant interacted in an overlay assay more strongly than wild-type Ran with ␤-karyopherin/␤-importin, and a novel 110-kDa RanBP. Association of ⌬-DE Ran with RanBP1 was not detected by overlay or by precipitation from solution using GSH-Sepharose (Fig. 3, A and C).
GTP:GDP Ratios of Cells Expressing Ran and Mutants of Ran-Given the dramatic effects of several of the Ran mutants on cell growth and nuclear protein import (19), it was important to determine whether the mutants are able to bind and hydrolyze nucleotide when expressed in intact cells. BHK21 cells were therefore transfected with pCH3-Ran plasmids to express HA1-tagged proteins. Transfected cells were incubated with [ 32 P]orthophosphate then harvested and immunoprecipitated with the anti-HA1 antibody, 12CA5. Labeled nucleotides associated with the HA1 Ran proteins were then separated by thin layer chromatography. The extracted wild-type HA1-Ran was 99% GDP-bound, whereas significant proportions of the HA1-G19V Ran and HA1-L43E Ran mutants (57% and 49%, respectively) were in the GTP-bound state (Fig. 4).
One cannot conclude from these experiments that wild-type Ran is predominantly GDP-bound within the intact cells, because significant hydrolysis of Ran-GTP occurs during the immunoprecipitation. For example, when GST-Ran loaded with [␣-32 P]GTP was incubated with BHK21 cell extract under conditions and for a time equal to that required for immunoprecipitation, 90% of the GTP bound to the Ran was converted to GDP. However, the results do indicate that both HA1-G19V Ran and HA1-L43E Ran are predominantly associated with GTP in intact cells. This conclusion is supported by our previous observation that transiently expressed HA1-G19V Ran coimmunoprecipitates as a stable complex with endogenous RanBP1 (13). HA1-⌬-DE Ran exhibited 15% GTP binding, possibly as a result of its increased interaction with RCC1 and reduced affinity for RanBP1 (21).
Localization of Mutant HA1-Ran Proteins by Immunofluorescence-To determine the effect of mutations within Ran on the subcellular localization of the protein, HA1-tagged Ran pro- teins were transiently expressed in BHK21 cells and detected with the anti-HA1 antibody, 12CA5 (Fig. 5). Wild-type HA1-Ran was primarily nuclear, a location identical to that of endogenous Ran (32). This result indicates that the HA1 tag does not interfere with expression or targeting of the protein. In other studies (19), expression of wild-type HA1-Ran was not toxic and did not interfere with nuclear protein import.
HA1-G19V Ran, HA1-E46G Ran (data not shown), and HA1-L43E Ran proteins were all localized predominantly at the nuclear envelope, suggesting that mutants locked in the GTPbound state bind tightly to proteins at the nuclear pore. The HA1-T24N Ran mutant was predominantly nuclear, but was also concentrated at the nuclear envelope. The C-terminal deletion, HA1-⌬-DE Ran, exhibited a diffusely cytoplasmic distribution in most cells.
ATP-dependent in Vitro Import of HA1-Ran into the Nucleus-Using an in vitro nuclear import assay, Moroianu and Blobel (33) have shown that a synthetic NLS-containing substrate and recombinant Ran can accumulate in the nucleus of permeabilized cells in an ATP-dependent and wheat germ agglutinin-sensitive manner. We used a similar assay to examine the import of HA1-Ran isolated from transfected COS cells. Extracts of COS cells expressing HA1-Ran stimulated nuclear import of both the nuclear import substrate, NLS-HSA, and HA1-Ran into the nucleus of digitonin-permeabilized BHK21 cells (Fig. 6). Import occurred within 15 min at 30°C and was dependent on the presence of ATP. The HA1-Ran used in these studies was likely to be primarily in the GDP-bound state (Fig.  4). Import of both HA1-Ran and NLS-HSA was inhibited both by wheat germ agglutinin and by chilling at 0°C (data not shown), indicating a facilitated active transport mechanism. HA1-Ran formed a ring at the nuclear envelope and was diffuse within the nucleus. Import of the NLS-HSA did not require the presence of HA1-Ran in the extract, indicating that the amount of endogenous Ran in the COS cell digitonin extracts was sufficient to support nuclear import. In other experiments, we observed that Ran import was not ATP-dependent or wheat germ agglutinin-sensitive in assays where a GST fusion of Ran was employed, although GST-Ran import was inhibited by incubating at 4°C (data not shown). GST fusion proteins of Ran mutants may not, therefore, be appropriate for studies of Ran function in nuclear import.
Effects of Ran Mutations on Their in Vitro Nuclear Accumulation-HA1-tagged Ran mutants, isolated from transfected COS cells, were tested for their ability to enter the nucleus using the in vitro nuclear import assay. As shown in Fig. 7A, the HA1-G19V and HA1-T24N Ran were able to accumulate in the nucleus to a similar extent as wild-type Ran but with a more pronounced ring at the nuclear envelope. HA1-L43E Ran and HA1-E46G Ran (data not shown) bound to the nuclear envelope with little nuclear accumulation. HA1-⌬-DE Ran, bound only weakly to the nuclear envelope and did not enter the nucleus. The diminished HA1-⌬-DE Ran signal did not result from low protein expression as is shown by an immunoblot of the COS extracts used in the import assays (Fig. 7B).
In other studies, when GST fusion proteins were tested in these assays, all of the mutants entered the nucleus even in the absence of ATP. Additionally, the transient expression of Ran as a fusion protein with the Aquoria green fluorescent protein (which at 27 kDa is of similar size to GST) in BHK21 cells also led to nuclear accumulation irrespective of the mutant form of Ran used (data not shown). These observations suggest that the presence of a large tag on Ran may interfere with the normal regulation of Ran import. DISCUSSION These results demonstrate that the interaction of Ran with effectors and regulatory proteins is sensitive to mutations in distinct regions of the GTPase structure. They demonstrate also that the subcellular distribution of Ran is determined by the nucleotide state of the protein and by its ability to bind specific RanBPs. Mutants that cannot cycle between the GDP- and GTP-bound states accumulate at the nuclear envelope and support a model for Ran function in which nuclear protein import and shuttling between the nuclear and cytoplasmic compartments is coupled to the GTP-GDP cycle. This model is reminiscent of that for the shuttling of Rab proteins such as Sec4p between the cytosol and vesicle membranes (34).
The biochemical characterization of Ran mutants reveals differences between the interactions of Ran with regulatory factors versus those of other GTPases such as Ras or Rab3A. Unlike G12V Ras, which remains sensitive to Ras exchange factors (35), G19V Ran was insensitive to RCC1-catalyzed guanine nucleotide exchange. Nonetheless, G19V Ran isolated from transfected cells possessed a much higher GTP:GDP ratio than did wild-type Ran, suggesting that G19V Ran is an activating mutation. This result is important in light of the fact that G19V Ran is a dominant interfering mutant of nuclear protein import (19). The ability of G19V Ran to accumulate in the GTP-bound state despite insensitivity to RCC1 may be a consequence of the high GTP:GDP ratio in the cell, and of the low endogenous GTPase activity of the G19V mutant.
The analysis of the dominant-negative mutation of Ran (T24N) supports and confirms results from other laboratories (28). Like T17N Ras, T24N Ran binds nucleotide poorly and likely for this reason is unable to interact stably with either class of RanBPs.
The effector regions of Ran are not yet defined. Effector domain mutants of other small GTPases, such as Ras and Rab3A, have been valuable in distinguishing the contributions of multiple regulators and downstream effectors to their cellular functions. Given the overall similarity between the structures of Ras and Ran (36,37), the amino acid residues of Ran corresponding to the Switch 1 domain of Ras are likely to be involved in effector interactions. Mutation of Ran at residue 43 (L43E Ran) is analogous to Ile-36 effector domain mutants of Ras (25) and the V55E mutants in Rab3A, while the E46G mutation is analogous to a D58V mutation in the Rab3A effector domain, described previously (29). The Ran L43E Ran mutant was insensitive to GTP hydrolysis stimulated by recombinant Fug1, the murine homolog of RanGAP (Fig. 1) (6). Unlike the Rab3A V55E, however, L43E Ran was sensitive to nucleotide exchange stimulated by RCC1, and a significant fraction of this mutant was GTP-bound when isolated from intact cells.  4. Proportion of GTP bound to HA1-Ran and HA1-Ran mutants expressed in BHK21 cells. BHK21 cells were transfected with 10 g of pCH3 (vector), pCH3-Ran, or the indicated mutant, labeled with 32 P i , and harvested into Triton X-100 lysis buffer. Extracts were immunoprecipitated using the anti-HA1 monoclonal antibody, 12CA5, coupled to protein A-Sepharose. Precipitates were washed extensively with lysis buffer. Nucleotides were released with buffer containing 1% SDS and 25 mM EDTA and separated by thin layer chromatography. A mixture of [ 32 P]GTP and -GDP was used as a nucleotide marker. The proportion of GTP present was quantitated by a phosphoimager.
Sepharose was detected using the Ran overlay assay as in A. Note: GST-Ran represents the GST fusion of wild-type Ran and mutant Ran proteins are designated by their respective amino acid changes.
The biochemical properties of both L43E Ran and E46G Ran were similar to the D58V mutation in Rab3A, suggesting that residues necessary for interactions at the Ran effector domain differ slightly in function from those at the Rab3A effector domain. The L43E Ran mutation also disrupted binding to RanBPs and ␤-karyopherin. This defect may account for the inability of L43E Ran to dominantly interfere with nuclear protein import despite the fact that it is trapped in the GTP-bound state within the cell (19).
Together, these results suggest the existence of multiple effector domains on Ran, a situation analogous to that of the Rac/Rho GTPase family (38). Association of Ran with RanBDs appears to be mediated, at least in part, through a region adjacent to and probably including ␤ 2 effector region of Ran. The region on Ran that interacts with ␤-karyopherin must, however, be distinct from the RanBD binding site; this region is exposed either by removal of the C terminus of Ran or by association of GTP-Ran with a RanBD (15). The surface of Ran that interacts with the RanBD must also be distinct from that for RCC1 binding, because these two factors can also form a ternary complex with Ran (21,27,39), and the Leu-43 muta- FIG. 6. In vitro import of HA1-Ran into BHK21 cell nuclei. BHK21 cells were permeabilized with 0.035% digitonin, then incubated under standard nuclear import conditions with 20 g/ml tetramethyl rhodamine-5-isothiocyanate-NLS-HSA import substrate and a digitonin extract of COS cells (1.5 mg/ml) expressing HA1-Ran (see "Experimental Procedures"). After 15 min at 30°C, cells were washed with ice-cold Buffer A, fixed with 3% paraformaldehyde, and subjected to immunofluorescence with the anti-HA1 antibody, 12CA5, and fluorescein-anti-mouse secondary antibody. The "No ATP" condition included the removal of ATP and the ATP-regeneration system and the addition of glucose and hexokinase. Cells were visualized by confocal microscopy. Bar, 25 m.

FIG. 7.
In vitro import of HA1-Ran and HA1-Ran mutants into BHK21 cell nuclei. A, BHK21 cells were permeabilized with 0.035% digitonin, then incubated with digitonin extracts of COS cells expressing HA1, HA1-Ran (wild type), or HA1-Ran mutants (G19V, T24N, L43E, or ⌬-DE) under standard import assay conditions (see "Experimental Procedures") for 15 min at 30°C. Cells were fixed and proteins visualized as in Fig. 6. Bar, 25 mm. B, proteins contained in 100 g of the COS extracts described in A were separated by 12% SDS-PAGE and transferred to nitrocellulose. Transfers were immunoblotted using the anti-HA1 antibody 12CA5 and reactive bands detected by chemiluminescence.
tion that interferes with RanBD binding does not reduce the sensitivity of Ran to RCC1-catalyzed exchange. Finally, Ran-BDs must also be capable of forming ternary complexes with Ran and RanGAP because RanBDs act as co-stimulators of GTP hydrolysis on Ran in the presence of RanGAP (13,40,41).
The subcellular locations of the Ran mutants discussed in this report differ in significant aspects from those described by others. For example Ren et al. (26) found that a G19V/Q69L mutant of Ran exhibited a predominantly nuclear localization when overexpressed in 293 Tag cells, and in budding yeast a GST fusion of G21V GSP1p (the yeast homolog of Ran; Ref. 1) was also nuclear (42). Our results indicate that G19V Ran associates with the nuclear envelope when expressed as an HA1-tagged protein in BHK21 cells; however, some nuclear accumulation is evident for G19V in an in vitro assay. The cause for the nuclear pore localization in transfected cells may thus be due to the inability of G19V Ran to exit the nucleus. G19V Ran is predominantly GTP-bound in cells, so it is likely that the binding at the nuclear envelope is due to interaction with the RanBDs of RanBP2, a component of the nuclear pore complex (11,12). Our observations are supported by in vitro analysis of the interaction of Ran bound to a non-hydrolyzable analog of GTP with a nuclear pore complex (43).
Wild-type HA1-Ran is nuclear, as is endogenous Ran (3), demonstrating that the abnormal localization of the G19V mutant is not mediated by the epitope tag. The level of heterologously expressed HA1-Ran proteins is only 1.7-fold higher than the level of endogenous Ran in transfected BHK21 cells (19). Therefore the cells are not burdened by massive overexpression (Ͼ10-fold over endogenous levels) as occurs in 293 Tag cells when expression is driven from autonomously replicating plasmids (26). For these reasons, it is likely that the observed subcellular distributions of the HA1-tagged Ran mutants identify physiologically meaningful states of the Ran nuclear transport cycle. On the other hand, the attachment of large epitopes such as GST or green fluorescent protein to Ran results in ATP-independent nuclear accumulation of Ran in vitro and in the nuclear accumulation of Ran in transfected cells, irrespective of the type of Ran mutant employed. These observations suggest that such large fusion proteins are inappropriate for studies of the subcellular localization of Ran.
The localization of the HA1-tagged L43E and E46G Ran mutants were similar to that of G19V Ran. Although a large fraction of L43E Ran in the intact cell is likely to be GTPbound, we could not detect a stable interaction of this mutant with RanBPs in the overlay assay. It remains unclear at present, therefore, whether L43E or E46G Ran associates with RanBP2 in the nuclear pore, or becomes trapped in the pore by interaction with other components of the transport machinery. Because little L43E or E46G Ran enters the nucleus during in vitro import, it is possible that, unlike G19V, L43E Ran cannot enter the nucleus. The location of L43E Ran in the nuclear pore was unexpected as it shows no dominant interference in nuclear protein import (19). Possibly weak interaction with RanBP2 allows association of L43E Ran with the nuclear pore yet reduces its inhibitory potency.
Contrary to expectations, the T24N Ran mutant was also distributed partially at the nuclear membrane, although with significant accumulation in the nucleoplasm. Because T24N Ran does not interact in vitro with RanBP2, which is localized at the nuclear pore, other components of the nuclear pore may exist that can interact with Ran in the GDP-bound state. One candidate protein for the T24N Ran-binding protein is p10/ NTF2 (43,44). The recent demonstration that Ran in the GDPbound state can form a complex with p10, NLS receptor, and nucleoporin is consistent with this idea (17). T24N Ran can potently inhibit nuclear protein import in vivo (19), an effect that may be a consequence of the interaction of T24N Ran with this complex.
Deletion of the C terminus of Ran resulted in a predominantly cytosolic distribution of the protein. Ren et al. (26) have reported previously that ⌬-DE Ran, when expressed in 293 Tag cells, is nuclear. We believe that this difference, as with the G19V Ran results, may be a consequence of the very high levels of heterologous protein expression in the 293 Tag cells. The localization of ⌬-DE Ran is consistent with our recent discovery that this mutant interacts efficiently with the cytoplasmic protein ␤-karyopherin/␤-importin, a component of the NLS receptor (15). Such an interaction may account for the dominant inhibition of nuclear protein import by ⌬-DE Ran in intact cells (19).
Taken together, these results argue that Ran can interact with nuclear pore complexes that are in different states, and which may represent different steps in the nuclear transport mechanism. This hypothesis is consistent with the model described recently by Nehrbass and Blobel (17), in which transient complexes assemble and disassemble within the nuclear pore in a process coupled to the Ran GTP/GDP cycle. Ran•GDP forms a complex with p10, ␣and ␤-karyopherin/importin, nuclear cargo, and nucleoporins. We predict that the T24N Ran inhibits nuclear protein import by stabilizing this complex, preventing nucleotide exchange-coupled disassembly. The G19V Ran mutant may stabilize a Ran•GTP-␤-karyopherin-RanBP2 complex and block further cycling. The possibility that the different Ran mutants described above can trap the nuclear pore complex machinery at different steps in the transport process suggests that they may provide useful tools to test the validity of models for Ran-mediated functions in intact cells.