A family of proteins that stabilize the Ran/TC4 GTPase in its GTP-bound conformation.

Ran/TC4, referred to here as Ran1, is a 25-kilodalton nuclear GTP-binding protein with an acidic C terminus that lacks any consensus prenylation sites. Here, we use a nitrocellulose overlay assay to identify potential effector proteins that bind specifically and with high affinity to the GTP-bound form of Ran1. GTP-Ran1 is shown to bind a variety of proteins, present in many eukaryotic tissues and cell extracts. A 28-kDa protein is cytosolic, whereas others, consisting of proteins of 86-300 kDa, are primarily localized in the nucleus. Binding is highly specific and is not detected by other small GTPases, such as c-Ha-Ras or Rab3A. Both deletion of the C-terminal-DEDDDL acidic sequence or alteration of the N terminus of Ran1 inhibits binding. However, these altered forms of Ran1 maintain the capacity to bind guanyl nucleotides and interact with the nucleotide exchange factor. The Ran1-binding proteins potently inhibit release of GTP from Ran1. These proteins can therefore maintain Ran1 in the "on" state and are potential down-stream effectors for Ran1-dependent cellular processes.

0 These two authors contributed equally in the performance of experiments.
In light of the possibility that additional proteins exist in the "Ran family" and to prevent future confusion, we refer to the gene product of TC4 as Ranl based upon its high sequence homology with the polypeptide originally designated Ran (6).
The abbreviations used are: RCC1, regulator of chromosome condensation 1; GDI, guanine nucleotide dissociation inhibitor; CHO, Chinese hamster ovary; MOPS, 3-(N-morpholino)propanesulfonic acid; PAGE, polyacrylamide gel electrophoresis.  temperature-induced loss of RCCl occurs during S phase or G1 phase (11,12). The demonstration that RCCl acts as a nucleotide exchange factor for Ranl suggests that the GTP/GDP ratio of R a n l m a y be an important parameter in the control of cell cycle progression (13). Similar conclusions have been reached from studies on spil andpiml in Schizosaccharomyces pornbe, which are fission yeast homologs of Ranl and RCC1,respectively (14).
RCCl and Ranl have also been linked to functions involving transport across the nuclear membrane. The Saccharomyces cerevesiae homolog of RCC1, PRP2O/SRMl, is required for mRNA processing and nuclear export (15,16), and overexpression of the Ranl homolog, GSP1, or mammalian RCCl can complement the temperature-sensitive phenotype of prp20/ srml mutants (17). The abrogation of mRNA export has also been demonstrated in tsBN2 cells upon temperature-induced loss of RCCl (18). Whether these functions are related to the RCC1-associated cell cycle phenomena is not known; however, the mRNA transport effects occur in minutes versus the cell cycle effects, which occur in hours. In addition, whereas the cell cycle events are dependent upon p34cdc28 kinase activation, the nuclear transport effects are not (11). Additional evidence for a role of Ranl in nuclear transport stems from the purification of R a n l as a cytosolic component in Xenopus oocytes necessary to stimulate import of a nuclear localization signalcontaining protein (19).
To elucidate components of the pathways in which Ranl operates, it is necessary to identify specific target proteins with which it interacts. By analogy with Ras, in which the GTPbound state confers an oncogenic phenotype (20), it is likely that downstream targets of R a n l will interact preferentially with the GTP-bound state. We have used a nitrocellulose overlay assay to detect putative targets and to determine the mechanism by which they interact with Ranl. While this manuscript was in preparation, Coutavas et al. (21) described the cloning of a 28-kDa cytosolic Ran-binding protein, RanBPl, which is almost identical to a previously identified open reading frame, HTF9A (22). With our assay, we have identified numerous RanBPs, including a 28-kDa cytosolic protein potentially representing RanBPl. Information is presented showing that GTP-Ran1 binds to multiple proteins that exhibit a regulatory function by inhibiting the EDTA-induced release of GTP from Ranl.

MATERIALS AND METHODS
Cell Culturing and Protein Extraction-Chinese hamster ovary (CHO) cells were maintained in F-12 medium (Life Technologies, Inc.) supplemented with 5% fetal bovine serum. Subcellular fractionation was accomplished by harvesting cells into hypotonic lysis buffer containing 10 m~ HEPES pH 8.0, 5 m~ KCl, and 2 m~ MgCl,. Cells were lysed by expression through a 26.5-gauge needle six times, and nuclei were separated by centrifugation at 1,000 x g for 5 min. The supernatant was then centrifuged at 100,000 x g for 30 min to generate SlOO and PlOO fractions. Soluble nuclear proteins were extracted from the 11285 Proteins That Stabilize GTP-bound Ran I TC4 1,000 x g pellet by incubation with 0.5% Triton X-100,5 m?? MgCI,, and 1 mg/ml DNase (Sigma) for 30 min on ice followed by treatment with 2 M NaCI. Insoluble nuclear proteins were then separated by centrifugation a t 1,200 x g for 5 min. For experiments requiring total cell extract, cells were harvested in the presence of 20 m~ HEPES pH 7.4, 0.1 m~ EDTA, 150 mM NaCI, and 0.5% Triton X-100. After 10 min of solubilization, lysis was completed by passing the extract 10 times through a 26.5-gauge needle. Insoluble particulate material was removed by centrifugation a t 10,000 x g for 5 min. Ranl Overlay Assay-Proteins were separated by 8% SDS-PAGE and transferred to nitrocellulose for 3 h a t 800 mA. Protein transfers were incubated a t 4 "C for 1-2 h in "renaturation buffer" containing 20 m~ MOPS, pH 7.1,lOO m~ sodium acetate, 5 m~ magnesium acetate, 0.25% Tween 20, 0.5% bovine serum albumin, and 5 mM dithiothreitol and then preincubated for 30 min at room temperature in "binding buffer" consisting of 20 mM MOPS, pH 7.1, 100 m~ potassium acetate, 5 mM magnesium acetate, 0.05% Tween 20, 0.5% bovine serum albumin, and 5 m~ dithiothreitol in the presence of 100 p GTP. Blots were rinsed with binding buffer alone and then overlaid with [a-32P1GTP-Ranl in binding buffer for 30 min at room temperature. Nonspecific binding was removed by five successive rinses in binding buffer, the second containing 50 p~ cold GTP. A similar assay was reported by Coutavas et al. (21) while this manuscript was in preparation. Radioactivity associated with the protein bands was quantitated using a GS-250 PhosphorImaging System (Bio-Rad).
The Ranl used in these assays was produced by isopropyl-l-thio-p-D-galactopyranoside-induced expression from a pETlla vector in BL21(DE3)LysS cells and subsequently purified by Mono Q or DEAE ion-exchange chromatography. [a-"PIGTP-Ran1 was generated by loading 2 pg of recombinant Ranl with 10 pCi [a-"PlGTP (DuPont NEN) for 20 min on ice in the presence of 10 mM MOPS, pH 7.1,l m~ EDTA, and 1 mg/ml bovine serum albumin in a 15-pl reaction. The complex was subsequently trapped by addition to 15 ml of binding buffer. Deletion Mutagenesis and Expression of Glutathione S-fiansferase Fusion Proteins-The C-terminal deletion mutant of Ranl (ADEDDDL-Ranl) was generated by PCR using pUCl9-TC4, a gift from Dr. Peter DEustachio, as a template with a 3'-primer, which excludes the last seven codons (PDEDDDL) of TC4. The product was then subcloned into the bacterial expression vector, pETlla, and expressed as described above. The glutathione S-transferase fusion protein of Ranl was produced by subcloning an amplified TC4 sequence into PGEX2T (Pharmacia LKB Biotechnology Inc.). Protein expression was stimulated by 1 mM isopropyl p-o-thiogalactopyranoside, and purification was accomplished by glutathione-Sepharose chromatography.
Nuclear Guanine Nucleotide Release Factor Actiuity-Guanine nucleotide release was measured by the filter binding method of Burstein and Macara (23) with the following modifications. The buffer control consisted of hypotonic lysis buffer containing 10 m~ HEPES pH 8.0, 5 mM KCI, and 2 m~ MgCI,. Activity was measured using 2 mg/ml nuclear extract. Residual GDP bound after 30 min can be attributed to nonspecific binding to the nitrocellulose filters by the nuclear extract.
EDTA-induced Release of GTP-Release of GTP was measured for 1) Ranl bound to proteins immobilized on nitrocellulose, and 2) Ranl itself immobilized on nitrocellulose. 1) Nitrocellulose strips containing CHO cell proteins were incubated with [a-32PlGTP-Ranl in binding buffer as described above. Strips were then subjected, a t room temperature, to "GTP release buffer" (consisting of binding buffer altered to contain 5 mM EDTA, no M e , and 100 p~ cold GDP) for the indicated time periods. After incubations, strips were removed and placed into ice-cold binding buffer, blotted dry, and exposed to x-ray film. 2) Ranl was loaded with [(r-"P]GTP and then filter-bound and washed with binding buffer. Following binding, filters were placed in GTP release buffer for the indicated time after which radioactivity was measured by scintillation counting. To measure the ability of Ranl to rebind GTP after release, Ranl loaded with GTP was first filter-bound to nitrocellulose and incubated with GTP release buffer for 30 min a t room temperature. The filters were then incubated for 10 min a t room temperature with 20 mM MOPS, pH 7.1,l m~ EDTA, 1 mg/ml bovine serum albumin, and 50 pCi/ml [a-32P1GTP. Filters were washed with magnesium-containing buffer and counted for radioactivity. Filters containing no protein were used as an adjustment for background binding of [a-"PlGTP.

RESULTS AND DISCUSSION
Detection of Ranl-binding Proteins-To observe proteins that interact with Ranl, soluble CHO cell lysate proteins were separated by SDS-PAGE and transferred to nitrocellulose. The nitrocellulose was incubated with a complex of recombinant Ranl and [cT-~~PIGTP and then washed and exposed t o x-ray film (Fig. lA). Several bands were detected, corresponding to 28,86,100, and 200-300 kDa. A band corresponding to 50 kDa was sometimes present, depending on the lysis and growth conditions. When a parallel lane was blotted with an equivalent amount of free [CT-~~PJGTP, no bands were detected. Similarly, when Ranl was loaded with an equivalent amount of [cT-~~PIGDP, binding was minimal (Fig.  lA). Therefore, the bands represent CHO proteins that interact with Ranl associated specifically with GTP. To determine whether these CHO proteins interact with other small GTP-binding proteins, similar blots were performed using recombinant c-Ha-Ras and Rab3A loaded with [cT-~~PIGTP. Under these conditions, no bands were detected, even after long exposure (Fig. lA). These results indicate that the proteins exhibit selective binding to Ranl. l mM EDTA, 150 m~ NaC1, and 0.5% Triton X-100. After 10 min of solubilization, lysis was completed by passing the extract 10 times through centrifugation a t 10,000 x g for 5 min. Rat tissues were homogenized a 26.5-gauge needle. Insoluble particulate material was removed by then solubilized and prepared as for the cell extracts. 100 pg of each extract was subjected to SDS-PAGE and nitrocellulose transfer. The Ranl overlay assay was then performed as in Fig. 1.
Avariety of conditions were tested in an effort to characterize the binding of Ranl to proteins that had been transferred from SDS gels to nitrocellulose. Binding of Ranl-GTP to cell proteins increased over time to a maximum at 20 min of incubation (Fig.  LB). Use of acetate buffers gave optimal binding; however, washing the nitrocellulose blots in chloride buffers after Ranl binding had been established did not elute the Ranl from the filter. Inclusion of higher detergent concentrations caused only slight reductions in binding, and reduction of the preincubation time in renaturation buffer from 24 h to 30 min had no significant effect. Additionally, nondenatured (untreated) and denatured (boiled in SDS-sample buffer) cell extracts spotted directly onto nitrocellulose were detectable by Ranl (not shown).
Together, these data suggest that Ranl recognizes a contiguous region within the proteins rather than a three-dimensional conformation and that denaturation is not necessary for Ranl binding to occur.
To establish whether the Ranl-binding proteins are ubiquitous, extracts were tested from a variety of cultured mammalian cells and tissues, as well as Escherichia coli and s. cereuesiae. All extracts, with the exception of bacteria, contain proteins that bind Ranl (Fig. 2). Proteins corresponding to 28 and 86 kDa were observed in all mammalian cell extracts examined. Similar proteins were also detected in extracts of Xenopus oocyte and Sf9 insect cells (not shown). Interestingly, the Ranl-binding proteins above 100 kDa varied in size and intensity among cell type. The intensity of these high molecular weight proteins suggests a strong affinity for Ranl, since these proteins transfer to nitrocellulose with low efficiency. We estimate a minimum dissociation constant of 10"' M based on the amount of Ranl in the assay.

Ranl-binding Proteins Are Predominantly Localized in the
Nucleus-Because Ranl localizes to the nucleus (161, it was of interest to determine the subcellular distribution of the proteins to which it binds. CHO cells were first fractionated by lysis in a hypotonic buffer and centrifuged to obtain nuclear (Pl), cytosolic (SlOO), and membrane (P100) fractions. The nuclei were washed and then extracted with 0.5% Triton X-100 plus DNase, and finally with 2 M NaCI. The remaining insoluble fraction is primarily comprised of the nuclear matrix  Fig. 1. The content of radioactivity in specific protein bands was quantitated using a GS-250 PhosphorImaging system (Bio-Rad). B, markers for subcellular fractionation. Lactate dehydrogenase ( L D H ) activity was measured in each fraction according to the procedure of Kornberg (35). The presence of the soluble nuclear retinoblastoma gene product, Rb(plO71, in subcellular fractions was detected by Western blot using anti-retinoblastoma monoclonal antiserum with an ECL detection system (Amersham Corp.) (28). mU, milliunits. C , Coomassie staining of subcellular fractions separated by 8% SDS-PAGE. (24). Equal proportions of each fraction were then analyzed for Ranl-binding proteins as above (Fig. 3A). Interestingly, the majority of Ranl-binding proteins are contained in the soluble nuclear fraction. Greater than 80% of the binding to proteins of 40, 50, and 200-300 kDa is found in this fraction. Proteins corresponding to 28,85, and 100 kDa were distributed throughout all the fractions with at least 50% in the cytosol, and this distribution was not a consequence of cross-contamination of the various fractions, as demonstrated by the presence of lactate dehydrogenase activity in the SlOO fraction and retinoblastoma protein exclusively in the soluble nuclear fraction (Fig. 3B ). A measure of the specificity of Ranl binding is demonstrated by the lack of preferential binding by Ranl to any of the more abundant proteins visualized in a parallel gel stained with Coomassie Blue (Fig. 3 0 . Some Ranl-binding proteins were also present in the insoluble nuclear matrix fraction, but these proteins are not enriched by the fractionation procedure, and they do not correspond to any major proteins in a preparation of purified rat liver nuclear matrix proteins. The association of Ranl in the GTP-bound state to components of the nucleus is consistent with the finding that RCC1, the Ranl exchange factor, is necessary for the localization of Ranl to the nucleus (25). Although localized to the nucleus, the Ranl-binding proteins were unable to bind to dsDNAcellulose, suggesting that, unlike RCC1, they do not represent DNA-binding proteins (8).
To address the possibility that Ranl recognizes a post-translational modification rather than an amino acid sequence in the binding proteins, a variety of treatments designed to remove covalent modifications (incubations with alkaline phosphatase, potato acid phosphatase, hydroxylamine, or 10% acetic acid) were tested. None of these treatments had any effect upon the capacity of Ranl to bind to cell proteins. In addition, treatment of 3T3 cells with tunicamycin (to block glycosylation) or lovastatin (to inhibit prenylation) also had no effect on the amount or pattern of Ranl binding. Furthermore, none of the Ranbinding proteins were immobilized by wheat germ agglutinin, suggesting that they are not glycosylated, as would be the case for nuclear pore proteins (26).
Altering the C or N Terminus of R a n l Inhibits Binding-The presence of an acidic C-terminal sequence in Ranl, absent from other small GTP-binding proteins that do not exhibit specific binding in this overlay assay, suggested that this sequence may be necessary for binding to target proteins. We therefore generated a mutant Ranl protein with a 7-residue deletion of its acidic C-terminal tail. The mutant protein was less soluble than wild type, and its capacity for binding to GTP was 10-fold less, indicating that the C terminus is important for the proper folding of Ranl, and its deletion may lead to denaturation. To account for the lower GTP binding activity, an equal amount of [(U-~~PIGTP counts bound was added to the overlay assay. When tested against wild-type Ranl in the overlay assay, the C-terminal deletion mutant, while able to recognize bands at 85 and 120 kDa, is unable to detect the 28-or 200-300-kDa proteins in either a CHO cell extract or in a preparation of nuclear proteins (Fig. 4). These results demonstrate the importance of the Cterminal acidic tail in the GTP-specific binding of Ranl to a specific class of target proteins.
Interaction of Ranl through the C terminus has implications regarding the recent finding that Ranl is necessary for nuclear import (19). The C-terminal sequence of Ranl (PDEDDDL) is almost a mirror image of the SV40 T antigen nuclear localization sequence (PKKKRKV) (26). In fact, antiserum against the peptide, DDDED, has been demonstrated to block subsequent nuclear import of nucleoplasmin (27). Yoneda et al. (27) described a 69-kDa protein recognized by this antiserum. However, they also noted the recognition of other proteins by this antiserum; Ranl is potentially one of these proteins.
To test the possibility that the acidic C-terminal tail of Ranl interacts with basic nuclear localization signals, 1 pg of an estrogen receptor hormone-binding domain, expressed as a glutathione S-transferase fusion protein, was run on an SDS-PAGE gel, transferred to nitrocellulose, and blotted with Ranl. This domain of the estrogen receptor contains two nuclear localization sequences. No binding of Ranl to the fusion protein was observed. These data indicate that more than a polybasic localization signal is necessary for a protein to accommodate specific interaction with Ranl. Additionally, a number of known nuclear proteins, including the retinoblastoma gene product (28), Nu& (29), and a recently identified nuclear matrix protein, p250, were examined for their ability to bind to Ranl, but none gave a detectable signal in the overlay assay.
Manipulations of Ranl at the N terminus were also inhibitory to its ability to bind in the overlay assay. A glutathione S-transferase fusion protein of Ranl was unable to detect any proteins in cell extracts, although its GTP binding activity was unaffected (not shown). Remarkably, the ability to interact with target proteins was not recovered by thrombin treatment of glutathione S-transferase-Ranl. Thrombin cleaves all but 2 residues (Gly-Ser) from the N terminus of the fusion protein. This result indicates that the interactions with target proteins are exquisitely sensitive to small changes a t both the N and C termini of Ranl. Interestingly, however, both the C-terminal deletion mutant of Ranl and the glutathione S-transferase-Ranl fusion protein were able to interact with guanine-nucleotide exchange factor (Fig. 5). Controls performed using the recombinant Ranl expressed from pETlla demonstrated that the loss of [(u-~'P]GDP counts bound to Ranl was not a consequence of protein degradation. Additionally, accelerated binding of nucleotide to Ranl was also observed following addition of nuclear extract, confirming that the extract contains bona fide exchange activity. These data suggest that the epitopes used to interact with guanine nucleotide release factors differ from those involved in interactions with putative target proteins. The data further indicate that both the Nand C-terminal regions of the protein (which are predicted to be in close proximity, based on the Ras crystal structure (30)) are essential for high affinity interaction with most of the observed binding proteins.
Rad-binding Proteins Inhibit GTP Release-To determine whether the Ranl-binding proteins detected by the overlay assay play a functional role in the regulation of nucleotide binding andor hydrolysis, we measured the rate of release of [a-32P]GTP from Ranl bound to these proteins in the presence of excess EDTA. As with most other Ras-like proteins, GTP release is very slow in the presence of magnesium and is rapid in its absence (31). Remarkably, GTP release was almost completely inhibited by association of Ranl with the binding proteins (Fig. 6). Additionally, Ranl-binding proteins that were incubated with a complex of Ranl and cold GTP followed by elution with GTP release buffer were unable to bind [a3'P1GTP-nitrocellulose was rapid in the absence of magnesium. This release might, in principle, however, be a result of denaturation of Ranl, rather than of guanine nucleotide dissociation, and the apparent inhibition of release might, in this case, be ascribed to stabilization of Ranl by the Ran-binding proteins, rather than to a reduction in k,, for nucleotide. To distinguish these possibilities, the ability of Ran on nitrocellulose to rebind [a-32PlGTP was tested. A complex of Ranl bound to [a3'P1GTP was bound to nitrocellulose and treated with EDTA for 30 min to release nucleotide. A binding mixture containing [a-32PlGTP was then added and trapped with magnesium. As can be seen in Fig. 6B, a rapid and complete rebinding of GTP to Ranl was observed.
This result demonstrates that loss of bound GTP was not caused by rapid denaturation of the Ranl on the nitrocellulose. Therefore, the inhibition of GTP release by Ranl-binding proteins must be a functional inhibition of the dissociation rate.
This effect is unique among the guanine nucleotide dissociation inhibitors (GDIs) that have been reported to date. The Rab3-GDI, for example, interacts only with the GDP-bound state of Rab3A (32), while the Rho-GDI interacts equally with Rho-GDP and Rho-GTP (33). The physiological role of a family of GDIs that interact only with the GTP-bound state of Ranl remains to be explored. One possibility is that they are components of a complex involved in nuclear transport and act to maintain Ranl in the GTP-bound state until interaction with a Ranl-GTPase-activating protein triggers nucleotide hydrolysis and complex disassembly. A cyclical association-dissociation of a complex involving a karyophile, nuclear localization signal receptor, hsc70, and other components has been proposed previously to be necessary for nuclear protein import (34). As these binding proteins are identified, a clearer picture of the role Ranl plays in nuclear transport andor mitotic regulation can be established.