The N-terminal Nuclear Export Sequence of IκBα Is Required for RanGTP-dependent Binding to CRM1*

Nuclear export of IκBα is mediated by the CRM1 nuclear export receptor. However, the identity of the nuclear export sequences NES(s) in IκBα that are responsible for binding of IκBα to CRM1 is controversial. Both a N-terminal NES-like region (amino acids 45–54) and a C-terminal NES-like region (amino acids 265–280) have, in a number of reports from different laboratories, been implicated in CRM1-dependent nuclear export of IκBα. We now demonstrate that the N-terminal NES-like region, but not the C-terminal NES-like region, is required for RanGTP-dependent binding of IκBα to CRM1. IκBα is a relatively weak substrate for CRM1, with an affinity for CRM1 that is 100-fold less than the minute virus of mice NS2 protein, a high affinity cargo protein for CRM1. We also demonstrate that IκBα functions as a physical adaptor between CRM1 and NFκB/Rel proteins. Both free IκBα and Rel-associated IκBα have comparable affinities for CRM1, suggesting that CRM1 does not discriminate between free IκBα and Rel-associated IκBα. Nuclear export of c-Rel by IκBα requires the N-terminal NES-like sequence of IκBα but is not affected by alanine substitutions within the C-terminal NES-like sequence of IκBα. In contrast, nuclear export of the v-Rel oncoprotein by IκBα is disrupted by alanine substitutions within either the N-terminal or the C-terminal NES-like sequences. However, alanine substitutions within the C-terminal NES-like sequence significantly reduce the affinity of IκBα for v-Rel, suggesting that loss of export function for this mutant is secondary to reduced association between IκBα and v-Rel. Taken together, our results demonstrate that the N-terminal NES-like sequence in IκBα is required for RanGTP-dependent binding of both free IκBα and NFκB/Rel-associated IκBα proteins to CRM1.

The NFB 1 /Rel transcription factor family has provided a powerful experimental paradigm for understanding how eukaryotic gene expression is regulated (reviewed in Refs. [1][2][3]. Mammalian cells have five NFB/Rel proteins, termed p50, p52, RelA(p65), c-Rel, and RelB. The NFB/Rel proteins are defined by the presence of an N-terminal, highly conserved region of 300 amino acids known as the Rel homology domain. The Rel homology domain is responsible for dimerization, nuclear localization, and DNA binding by NFB/Rel proteins. NFB/Rel proteins are regulated, in large part, through association with members of the Inhibitory of kappa B (IB) family of ankyrin repeat proteins.
A major regulatory mechanism for controlling NFB/Rel proteins is sequestration in the cytoplasm. Crystallographic analysis of the ternary p50⅐p65⅐IB␣ complex has revealed that NLSs within the dimeric p50⅐p65 complex are masked by IB␣, suggesting that IB␣ inhibits nuclear import of NFB/Rel proteins by blocking access of the nuclear import machinery to the NLSs of NFB (4,5).
Cytoplasmic sequestration of NFB/Rel proteins arises from a dynamic equilibrium between nuclear import and nuclear export. For example, treatment of unstimulated cells with leptomycin B results in the relocalization of both NFB/Rel and IB␣ proteins to the nucleus (6,7). In addition, leptomycin B treatment results in sustained nuclear localization of NFB/ Rel and IB␣ proteins following transient stimulation with activating agents such as tumor necrosis factor-␣ (6). Because the CRM1 nuclear export receptor is specifically inactivated by leptomycin B (8), CRM1-dependent nuclear export is important for cytoplasmic sequestration of NFB/Rel and IB␣ proteins under conditions of both basal and post-induction repression.
CRM1 binds short leucine-rich nuclear export sequences (NESs) in its cargo proteins (8 -12). Cooperative binding of RanGTP is required for formation of an export-competent complex between CRM1 and a NES-bearing cargo protein (13). IB␣ contains two leucine-rich regions that resemble previously characterized NESs, amino acids 45-54 and 265-280 ( Fig. 1). However, there is contradictory evidence in the literature regarding the requirement of these NES-like sequences for nuclear export of IB␣ (7, 14 -18).
Dargemont and co-workers (14) were the first to show that IB␣ could be exported from the nucleus. In these experiments, alanine substitutions for hydrophobic residues within the Cterminal NES-like sequence reduced the rate of IB␣ export when injected into nuclei of Xenopus oocytes (14). Subsequently, our laboratory demonstrated that IB␣ mediates nuclear export of the v-Rel oncoprotein and that this nuclear export function of IB␣ is reduced by mutations within the C-terminal NES-like region of IB␣ (15).
In contrast, more recent experiments have suggested that the N-terminal NES-like sequence in IB␣ is critically required for CRM1-dependent nuclear export of IB␣. For example, Hope and co-workers (7) have shown that alanine substitutions within the N-terminal NES-like sequence abolished nuclear export of IB␣ proteins following microinjection into the nuclei of 3T3 cells. Miyamoto and co-workers (16) showed that mutations within the N-terminal NES-like sequence markedly in-creased nuclear accumulation of a GFP-IB␣ protein in transiently transfected COS cells. In agreement with both of these results, our laboratory (17) has recently shown that the entire N-terminal domain of IB␣, including the N-terminal NES-like sequence, is required for nuclear export of GST-IB␣ in digitonin-permeabilized HeLa cells. Finally, using both yeast and mammalian cells, Sen and co-workers (18) have showed that the N-terminal NES-like sequence is required for CRM1-dependent nuclear export of IB␣.
There is also contradictory evidence in the literature regarding the requirement of these two NES-like sequences for binding of IB␣ to CRM1. In one set of experiments, the C-terminal NES-like sequence of IB␣ was shown to be required for binding to in vitro translated CRM1 (12). On the other hand, the N-terminal NES-like sequence has been shown to be required for interaction with CRM1 in the yeast two-hybrid assay (16).
In this study, we now demonstrate that purified IB␣ binds to purified CRM1 in a RanGTP-dependent manner. Quantitative analysis of an export-competent ternary complex between IB␣, CRM1, and RanGTP indicates IB␣ has a relatively weak affinity for CRM1. The N-terminal NES-like sequence, but not the C-terminal NES-like sequence, is required for the formation of a ternary IB␣⅐CRM1⅐RanGTP complex. We also examined the role of the respective NESs of IB␣ in IB␣mediated nuclear export of NFB/Rel proteins. The N-terminal NES-like sequence of IB␣ is required for IB␣-mediated nuclear export of c-Rel, whereas both NES-like sequences of IB␣ are required for IB␣-mediated nuclear export of v-Rel. However, alanine substitutions within the C-terminal NES-like sequence markedly reduce the affinity of IB␣ for v-Rel, suggesting that loss of IB␣-mediated nuclear export of v-Rel is secondary to reduced binding of IB␣ to v-Rel. Taken together, our data indicate that the N-terminal NES-like sequence of IB␣ is required for both binding of IB␣ to CRM1 and for CRM1-mediated nuclear export of the IB␣⅐Rel complex.

EXPERIMENTAL PROCEDURES
Construction of Recombinant DNA Molecules-The construction of recombinant DNA molecules was performed by standard techniques (19). An EcoRI fragment containing the avian IB␣ cDNA was used as the progenitor for all of the mutants IB␣ utilized in this study (20). All point mutations were constructed from phagemid single-strand DNA and sequenced to confirm the presence of the desired mutation. The IB␣ genes were expressed in monkey COS cells by using a cytomegalovirus (CMV)-derived vector. The CMV-based expression vectors for c-Rel and v-Rel have been described previously (15). The GST expression vectors for IB␣ and IB␣-CA4 have been described previously (17). The GST expression vector for IB␣-NA4 was constructed by insertion of the mutant cDNA into pGEX-2T. The GST-NS2 vector contains the NS2 open reading frame inserted into pGEX-5X and was a gift from David Pintel (University of Missouri). The expression vector for His-tagged IB␣ contained the IB␣ open reading frame inserted into the NdeI site of pET15b. The expression vectors for His-c-Rel contained codons 1-320, and His-v-Rel contained codons 1-331 inserted into the BamHI site of pET15b. The pQE32-derived expression vectors for wild-type Ran, RanQ69L, and RanGAP proteins were obtained from Dirk Gorlich (University of Heidelberg). A pET-based expression vector for the CRM1 protein was obtained from Peter Askjaer (University of Aarhus) (13). A baculovirus encoding HA-CRM1 was the generous gift of Alan Diehl (University of Nebraska Medical Center).
Expression and Purification of Recombinant Proteins-The recombinant GST-IB␣ proteins were expressed in Escherichia coli strain BL21(DE3)pLysS and purified with glutathione-agarose as described previously (17). The His-tagged IB␣, c-Rel, and v-Rel proteins were expressed in BL21(DE3)pLysS and purified as described in the Novagen pET manual. The Ran proteins were expressed in BLR pRep4 cells and purified by metal-chelate affinity chromatography (Invitrogen) as described (13). The His-CRM1 protein was expressed in BL21(DE3)pLysS cells and purified by metal-chelate affinity chromatography. For His-CRM1 production, a single colony from a freshly transformed plate was inoculated into 50 ml of TB containing 200 g/ml ampicillin. The cells were grown at 37°C until A 600 reached to 0.1, collected by centrifugation, and resuspended in 2 ml of fresh TB. The resuspended cells were inoculated in 750 ml of TB containing 200 g/ml ampicillin, grown at 37°C to an A 600 of 0.1, and induced with 1 mM isopropyl-␤-D-thiogalactopyranoside (Sigma Chemical Co.) and 1 mM phenylmethylsulfonyl fluoride for 4 h at 20°C. The cell pellets were lysed with Bugbuster (Novagen) containing 5 mM 2-mercaptoethanol and a protease inhibitor mixture mix (Calbiochem). The His-tagged CRM1 proteins were purified by metal-chelate affinity chromatography (Invitrogen). All purified proteins were dialyzed against 20 mM HEPES, pH 7.3, 100 mM potassium acetate, 20 mM magnesium acetate, and 1 mM dithiothreitol. HA-CRM1 was purified from Tini insect cell lysates infected at high multiplicity with a baculovirus encoding HA-CRM1. Lysates were collected in 50 mM HEPES, pH 7.5, 300 mM KCl, 10% glycerol, 5 mM 2-mercaptoethanol and clarified by centrifugation. A protease inhibitor mix was included during the lysis step. Insect cell lysates were incubated with an anti-HA affinity matrix (Covance) at 4°C for 2 h. After incubation, the affinity matrix was extensively washed with incubation buffer (10 mM sodium phosphate, pH 7.3, 150 mM NaCl, 0.1% Triton X-100, and 1 mM 2-mercaptoethanol).
Cell Culture and Transfection-COS and 293T cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum. Transfections were performed with 1 g of each pCMV4-based expression plasmid and 15 l of LipofectAMINE (Life Technologies, Inc.) per 35-mm dish. 293T cells were transfected with 10 g of a pCMV4-based expression plasmid encoding c-Rel or v-Rel per 100-mm dish using calcium phosphate. 293T lysates were collected in lysis buffer (50 mM HEPES, pH 7.9, 250 mM sodium chloride, 5 mM EDTA, and 1 mM dithiothreitol) containing a protease inhibitor mix. The cellular localization of Rel and IB␣ proteins was analyzed 36 -48 h after transfection by indirect immunofluorescence as previously described (15). Polyclonal anti-IB␣ rabbit antiserum was followed by anti-rabbit FITCconjugated antibody (Jackson Laboratories) for detection of the ectopically expressed IB␣ protein. A monoclonal mouse antiserum directed against Rel (3C1; provided by Henry R. Rose, Jr., University of Texas) was followed by anti-mouse Cy3-conjugated antibody (Jackson Laboratories) for detection of the ectopically expressed Rel protein.
RanGAP Protection Assay-3 g of Ran was incubated for 30 min on ice with [␥-32 P]GTP (10 mCi/ml, 6000 Ci/mmol) in the presence of 20 mM HEPES, pH 7.3, and 100 mM potassium acetate and 5 mM EDTA. Loading was stopped by adding MgCl 2 to a final concentration of 20 mM followed by gel filtration on a Bio-Spin 6 column (Bio-Rad) equilibrated with hydrolysis buffer (20 mM HEPES, pH 7.3, 100 mM potassium acetate, 20 mM magnesium acetate, and 1 mM dithiothreitol) containing 0.5 mg/ml bovine serum albumin. Reaction mixtures containing 50 pM Ran[␥-32 P]GTP, 500 nM CRM1, and 0 -15 M GST-IB␣ or GST-NS2 proteins in 25 l of hydrolysis buffer were incubated at 15°C for 30 min. The samples were placed on ice, 10 nM RanGAP was added, and the samples were immediately placed at 30°C for 2 min. The reactions were diluted into 1 ml of ice-cold hydrolysis buffer and filtered through nitrocellulose (0.45 Micron, Micron Separations Inc.). Filters were washed twice with 3 ml of hydrolysis buffer, and radioactivity was counted in the presence of 3 ml of scintillation fluid (Ready Safe, Beckman) on a Packard Tri-Carb 2100TR liquid scintillation analyzer.
In Vitro Binding Assays-For HA-CRM1 pull-down assays, 100 ng of immobilized HA-CRM1, 1 g of GST-IB␣, and 1-2 g of RanQ69L GTP in incubation buffer was rotated at 4°C for 2 h. The matrix was extensively washed with incubation buffer prior to addition of SDS-PAGE sample buffer.
For GST-IB␣ pull-down of HA-CRM1, the GST-IB␣ proteins were loaded onto glutathione-agarose beads in incubation buffer at 4°C for 30 min. The beads were extensively washed and incubated with 20 l of insect cell lysates containing HA-CRM1 in the presence of 1 g of RanQ69L GTP at 4°C for 2 h. The matrix was extensively washed with incubation buffer prior to addition of SDS-PAGE sample buffer. The cells were lysed 48 h after transfection in incubation buffer containing a protease inhibitor mixture.
For GST-IB␣ pull-down of v-Rel, 100 l of 293T cell lysates was incubated with GST-IB␣ bound to glutathione-agarose beads in the presence of 0 -1 mM His-IB␣ protein. The reaction mixtures were incubated at 4°C for 1 h. After incubation, the beads were extensively washed with incubation buffer prior to the addition of SDS-PAGE sample buffer.
In all binding assays, immunoblot analysis was used to measure proteins that remained bound to the respective solid supports after extensive washing. The samples were electrophoresed through SDSpolyacrylamide gels and transferred to nitrocellulose. HA-CRM1 was detected by anti-HA mouse monoclonal antibody (Covance), GST-IB␣ proteins were detected by anti-GST mouse monoclonal antibody (Santa Cruz), His-IB␣ was detected by a polyclonal rabbit anti-IB␣ antiserum, His-v-Rel was detected by a polyclonal mouse antiserum directed against the v-Rel NLS, full-length v-Rel was detected by a polyclonal rabbit anti-Rel antiserum, and Ran was detected by a polyclonal rabbit antiserum (Covance). All primaries were used at dilutions of 1:2000 to 1:4000. The secondary antibodies were either horseradish peroxidase-coupled anti-mouse IgG (New England BioLabs) used at a dilution of 1:20,000 or horseradish peroxidase-coupled anti-rabbit IgG (Sigma) used at a dilution of 1:10,000. To detect multiple proteins on the same nitrocellulose filter, some filters were stripped with IgG Elution Buffer (Pierce) before subsequent antibody incubations. The proteins were visualized using Amersham Pharmacia Biotech ECL reagents followed by exposure to XAR-5 film. For quantification, proteins transferred to nitrocellulose membrane were visualized using Super Signal West Dura Extended Duration Substrate (Pierce), and intensity of bands was quantified by phosphorimaging analysis (FXImager, Bio-Rad) using Bio-Rad's Quantity One software.

RESULTS
RanGTP-dependent Binding of IB␣ to CRM1-The formation of an export competent complex between CRM1 and a NES-bearing cargo protein requires cooperative binding of RanGTP (13). To determine if IB␣ binds to CRM1 in a RanGTPdependent manner, the ability of purified IB␣ to bind purified CRM1 was determined in the absence or presence of RanGTP. HA-tagged CRM1 (HA-CRM1) was immobilized on anti-HA agarose beads, and His-tagged IB␣ (His-IB␣) was added in the presence or absence of RanQ69L GTP. The agarose beads were extensively washed, and the presence of bound IB␣ was determined by immunoblot analysis. Binding of His-IB␣ to HA-CRM1 was markedly increased in the presence of RanQ69L GTP ( Fig. 2A, lanes 3 and 4). A 16-fold increase in the amount of His-IB␣ bound to CRM1 was observed when RanQ69L GTP was included in the binding reaction (Fig. 2B).
The N-terminal NES-like Sequence, but Not the C-terminal NES-like Sequence, Is Critical for RanGTP-dependent Binding of IB␣ to CRM1-To determine which NES-like sequence in IB␣ is required for RanGTP-dependent binding to CRM1, the ability of GST-tagged IB␣ proteins to associate with HA-CRM1 was determined. Wild-type and mutant GST-tagged IB␣ proteins were incubated with immobilized HA-CRM1 in the absence or presence of RanQ69L GTP. After extensive washing, the amount of bound GST-IB␣ proteins was determined by immunoblot analysis. As expected, binding of GST-IB␣ to HA-CRM1 was markedly increased in the presence of RanGTP (Fig. 3A, lanes 1 and 2). Mutant GST-IB␣ proteins containing alanine substitutions within either the N-terminal (GST-IB␣-NA4) or the C-terminal (GST-IB␣-CA4) NES-like sequences were still able to associate with CRM1, although a substantial decrease of CRM1 binding was observed with GST-IB␣-NA4 (Fig. 3A, compare lanes 2, 3, and 4). To closely mimic the in vivo situation where CRM1 interacts with multiple cargo proteins, lysates from 293T cells were included during the  1-320). B, the same Ran GAP protection assay using His-v-Rel-(1-331) instead of His-c-Rel was performed as described in A. C, anti-HA-agarose beads containing no protein (lane 1) or 100 ng of HA-CRM1 were incubated with 1 g of His-tagged IB␣ (lanes 1, 3, and 4), 1 g of His-tagged v-Rel-(1-331) (lanes 1, 2, and 4) and 1 g of RanQ69L GTP (lanes 1-4). After extensive washing, bound proteins were electrophoresed through an SDS-7.5% polyacrylamide gel and subjected to immunoblot analysis with a monoclonal anti-HA antibody (top panel), a polyclonal anti-Rel-NLS antiserum (middle panel), and anti-Ran (bottom panel) sera.
binding reaction. Under these conditions, the GST-IB␣ or GST-IB␣-CA4 proteins were still able to bind HA-CRM1, although at reduced levels compared with the level of binding observed with the purified proteins (Fig. 3A, compare lanes 2 and 3 with lanes 5 and 6). In contrast, binding of the GST-IB␣-NA4 protein to HA-CRM1 was abolished by the inclusion of 293T cell lysates in the binding reaction (Fig. 3A, lane 7).
To confirm that the N-terminal and not the C-terminal NESlike sequences are required for association with CRM1, the respective GST-IB␣ proteins were immobilized on glutathione-agarose and incubated with insect cell lysates containing HA-CRM1. Equivalent amounts of RanQ69L GTP were added to each sample. The agarose beads were extensively washed, and the amount of bound HA-CRM1 was determined by immunoblot analysis. Binding of HA-CRM1 to GST-IB␣-NA4 was markedly reduced compared with either the GST-IB␣ or the GST-IB␣-CA4 proteins (Fig. 3B, compare lanes 2, 3, and 4). A 3-fold reduction in binding of HA-CRM1 to the immobilized GST-IB␣-NA4 protein was observed (Fig. 3C).
IB␣ Is a Relatively Weak Substrate for CRM1-To provide a quantitative measure of the affinity of IB␣ for CRM1, we used a RanGAP protection assay that has previously been used to measure the affinity between CRM1 and NES-bearing cargo proteins (11,(21)(22)(23). The slow intrinsic GTPase activity of Ran is markedly stimulated by a GTPase activating protein, Ran-GAP. However, upon formation of an export-competent ternary complex between RanGTP, CRM1, and a NES-bearing cargo protein, RanGAP is no longer able to stimulate the GTPase activity of Ran. In this assay, the ability of a NES-bearing cargo protein to inhibit RanGAP-stimulated GTP hydrolysis in a concentration-dependent manner provides an estimate of the affinity of the cargo protein for CRM1.
In the presence of CRM1 alone, minimal protection of Ran-GAP-stimulated GTP hydrolysis was observed (Fig. 4). To confirm that our CRM1 preparation was competent for binding a NES-bearing cargo protein, the NS2 protein from the minute virus of mice (MVM) was used. The MVM NS2 protein contains a leucine-rich NES that binds with high affinity to CRM1 (11). A GST-NS2 fusion protein was able to inhibit RanGAP-stimulated GTP hydrolysis by Ran in a concentration-dependent manner (Fig. 4). Consistent with previously published results (11), the GST-NS2 protein was a high affinity substrate for CRM1, because 0.07 M GST-NS2 was sufficient to provide 50% protection against RanGAP-stimulated GTP hydrolysis by Ran. In contrast, 8 M GST-IB␣ was required to achieve 50% protection against RanGAP-stimulated GTP hydrolysis. Thus, the affinity of IB␣ for CRM1 is relatively weak, ϳ100-fold less than that of the MVM NS2 protein.
The ability of the mutant GST-IB␣ proteins to inhibit Ran-GAP-stimulated GTP hydrolysis by Ran was determined. The GST-IB␣-CA4 protein was equally effective as the wild-type GST-IB␣ protein for protection against RanGAP, with an apparent affinity of 8 M. In contrast, 15 M GST-IB␣-NA4 protein did not provide significant protection against RanGAPstimulated GTP hydrolysis by Ran.
Taken together, the binding assays and the RanGAP protection assays provide strong evidence that the N-terminal NES-like sequence of IB␣ is required for formation of an export-competent complex with CRM1 and RanGTP. In contrast, mutations of hydrophobic residues within the C-terminal NES-like sequence do not affect RanGTP-dependent association of IB␣ with CRM1.
IB␣ Is an Adaptor That Bridges CRM1 and Rel Proteins-There is strong experimental evidence in support of the notion that nuclear export of NFB/Rel proteins is mediated by CRM1. For example, cytoplasmic sequestration of p65 by IB␣ is CRM1dependent (7,16). Likewise, our laboratory has previously reported that cytoplasmic retention of the v-Rel oncoprotein by IB␣ is abolished by leptomycin B (15). However, it is not clear if IB␣ is an adaptor that mediates binding of NFB/Rel proteins to CRM1 or if NFB/Rel proteins can bind directly to CRM1. Importantly, a functional NES has been identified in p65 (24).
We utilized the RanGAP protection assay to characterize the adaptor function of IB␣. For these experiments, we used Cterminal truncated c-Rel-(1-320) or v-Rel-(1-311) proteins that were readily purified as His-tagged proteins from E. coli. In this assay, purified His-c-Rel or His-v-Rel protein did not have a measurable affinity for CRM1 in the RanGAP protection assay (Fig. 5, A and B). Significant protection of RanGAPinduced hydrolysis of GTP was observed only in the presence of IB␣. The presence of His-c-Rel or His-v-Rel did not alter the weak affinity of IB␣ for CRM1.
To further determine if NFB/Rel proteins bind directly to CRM1 or require IB␣ to act as an adaptor protein, we examined the ability of His-v-Rel to associate with CRM1 in the presence or absence of IB␣. HA-CRM1 was immobilized on anti-HA-agarose in the presence of RanQ69L GTP, IB␣, and v-Rel, and the presence of bound v-Rel was determined by immunoblot analysis after extensive washing of the agarose beads. Binding of v-Rel was only observed in the presence of IB␣ (Fig. 5C, lane 4), consistent with our RanGAP protection assay results. It is likely that IB␣ also functions as a direct physical adaptor between CRM1 and c-Rel, although the specific formation of a CRM1⅐IB␣⅐c-Rel complex could not be demonstrated due to high nonspecific binding of the His-c-Rel  E, I, J, N, and O). The cellular localization of singly transfected IB␣ protein was determined by single-label indirect immunofluorescence with anti-IB␣ sera (A, F, and K). The cellular localization of Rel proteins and cotransfected IB␣ proteins within the same cell was determined by double-label indirect immunofluorescence with anti-Rel and anti-IB␣ sera.
protein to anti-HA-agarose beads (data not shown).
N-terminal NES-like Sequence of IB␣ Is Required for the Nuclear Export of Rel-To determine which NES-like sequence of IB␣ is required for nuclear export of Rel proteins, we examined the ability of wild-type and mutant IB␣ proteins to retain c-Rel or v-Rel in the cytoplasm when coexpressed in COS cells. Because the full-length c-Rel protein is predominantly cytoplasmic when singly expressed in a variety of mammalian cell types (3), we utilized a C-terminal truncated c-Rel protein comprising amino acids 1-546. This truncated c-Rel protein lacks a C-terminal transcriptional activation domain and, like v-Rel, efficiently localizes to the nucleus in COS cells (15).
The localization of the wild-type or mutant IB␣ proteins when singly overexpressed in COS cells was first determined. The wild-type IB␣ protein and both mutant IB␣ proteins were predominantly localized to the nucleus (Fig. 6, panels A,  F, and K).
The localization of both IB␣ and Rel proteins following coexpression in COS cells was next determined. Coexpression of either c-Rel or v-Rel with the wild-type IB␣ protein results in the relocalization of both IB␣ and the respective Rel protein to the cytoplasm (Fig. 6, panels B-E, Table I). To assess the extent to which cytoplasmic relocalization of the Rel proteins by IB␣ requires CRM1-dependent nuclear export, the localization of the coexpressed proteins was determined following a 1-h treatment with leptomycin B. As previously noted (15), cytoplasmic relocalization of c-Rel by IB␣ was partially disrupted by leptomycin B treatment, whereas cytoplasmic relocalization of v-Rel by IB␣ was completely abolished by leptomycin B (Table I).
The ability of the mutant IB␣ proteins to relocalize either c-Rel or v-Rel to the cytoplasm was determined. The IB␣-NA4 protein was not able to completely relocalize c-Rel to the cytoplasm, because both the IB␣-NA4 protein and the c-Rel protein displayed a whole-cell pattern of staining (Fig. 6, panels G and H; Table I). In contrast, coexpression of the IB␣-CA4 protein with c-Rel resulted in the complete relocalization of both the IB␣-CA4 and the c-Rel proteins to the cytoplasm (Fig.  6, panels L and M; Table I). Neither mutant IB␣ protein was able to relocalize v-Rel to the cytoplasm (Fig. 6, panels I, J, N,  and O).
The failure of mutant IB␣ proteins to retain c-Rel or v-Rel in the cytoplasm could arise from a defect in association with CRM1 or from a defect in association with Rel proteins. Our previous analysis of IB␣⅐CRM1 interactions demonstrated that the IB␣-NA4 protein, but not the IB␣-CA4 protein, is defective in association with CRM1 (Figs. 3 and 4). To determine if the mutant IB␣ proteins were able to associate with c-Rel or v-Rel, we determined the ability of GST-tagged IB␣ proteins to pull-down c-Rel or v-Rel from lysates of 293T cells transfected with an expression vector for c-Rel or v-Rel. The binding assays were performed in the presence of increasing amounts of His-IB␣ as a competitor to obtain an estimate of the relative affinity of the wild-type and mutant GST-IB␣ proteins for either c-Rel or v-Rel.
There was no measurable difference between wild-type GST-IB␣ and GST-IB␣-NA4 mutant in their ability to associate with c-Rel (Fig. 7, A and B) or with v-Rel (Fig. 7, C and D). Thus, the failure of IB␣-NA4 to efficiently relocalize either c-Rel or v-Rel to the cytoplasm is not due to a defect in association with either c-Rel or v-Rel.
However, the IB␣-CA4 protein was reduced in its ability to associate with either c-Rel (Fig. 7, A and B) or v-Rel (Fig. 7, C  and D). At a GST-IB␣ concentration of 50 nM, inclusion of 10 nM His-IB␣ in the pull-down assay reduced the amount of v-Rel bound to the GST-IB␣-CA4 protein by ϳ30%, whereas the amount of v-Rel bound to the GST-IB␣-CA4 protein was reduced to less than 2%. Because the IB␣-CA4 protein can efficiently relocalize c-Rel, but not v-Rel, to the cytoplasm, these results indicate that a slight reduction in affinity between IB␣ and c-Rel does not significantly decrease the ability of IB␣ to either inhibit import or mediate export of c-Rel, whereas a significant reduction in affinity between IB␣ and v-Rel is sufficient to abolish IB␣-dependent export of v-Rel. DISCUSSION The distribution of NFB/Rel proteins between the nucleus and the cytoplasm is a dynamic balance between nuclear import and nuclear export. Although the molecular mechanisms that underlie these competing processes are not fully understood, it is clear that the IB␣ protein can exert a profound influence on both nuclear import and nuclear export of NFB/ Rel proteins. To provide further understanding of how nuclear export of both free and NFB/Rel-associated IB␣ is accomplished, we have used two complementary approaches to examine the interaction of IB␣ with the nuclear export receptor  CRM1. In one approach, binding of purified IB␣ to purified CRM1 was assessed in the absence or presence of RanQ69L GTP. Because the CRM1 protein in these experiments was purified from baculovirus-infected cell lysates, we utilized the RanGAP-resistant RanQ69L protein to eliminate the possibility that any contaminating cellular proteins, i.e. RanGAP and RanBP1, would stimulate GTP hydrolysis by Ran and thereby dissociate the IB␣⅐CRM1 complex. The RanQ69L protein has been extensively used to characterize CRM1 interactions with NES-bearing cargo proteins and is an effective substitute for the wild-type Ran under conditions where GTP hydrolysis by Ran is undesirable (8,25). We found that binding of IB␣ to CRM1 was markedly enhanced by RanQ69L GTP. In a second approach, we found that IB␣ forms a stable ternary complex with CRM1 and RanGTP that is resistant to RanGAP stimu-lation of GTP hydrolysis by Ran. Our results are consistent with numerous functional assays that have demonstrated that nuclear export of IB␣ is CRM1-dependent (7, 14 -18).
We used this binding assay to determine which of two candidate NESs in IB␣ are required for binding to CRM1. Despite extensive functional characterization of CRM1-dependent nuclear export of IB␣, the sequences in IB␣ that mediate binding to CRM1 have not been conclusively identified (7, 14 -18). Using mutant IB␣ proteins that contain alanine substitutions for hydrophobic residues within each candidate NES, we found that alanine substitutions within only the N-terminal candidate NES reduced binding of IB␣ to CRM1. In contrast, alanine substitutions within the C-terminal candidate NES did not affect association of IB␣ with CRM1. It is somewhat surprising that binding of IB␣ to CRM1 was not completely abolished by alanine substitutions within the N-terminal NES, because similar mutations have been reported to completely abolish nuclear export of IB␣ (7). However, because the wildtype IB␣ protein is already a relatively weak substrate for CRM1, a further reduction in the affinity of IB␣ for CRM1 would likely reduce the affinity of IB␣ for CRM1 such that an IB␣ protein containing a mutant NES would no longer effectively compete with other cellular proteins for binding to CRM1. Indeed, the GST-IB␣-NA4 protein was unable to bind to CRM1 when 293T cell lysates were included in the binding reaction.
That the N-terminal NES-like sequence is required for binding of IB␣ to CRM1 was substantiated in the RanGAP protection assay. In this assay, protection of RanGAP-induced GTP hydrolysis by Ran requires the formation of a ternary complex between Ran-GTP, CRM1, and a NES-bearing cargo. Alanine substitutions within the N-terminal NES-like sequence markedly reduced protection against RanGAP-induced GTP hydrolysis. In contrast, alanine substitutions within the C-terminal NES-like sequence had no effect on the ability of IB␣ to form a ternary complex with RanGTP and CRM1. Taken together, our results provide strong evidence that the N-terminal NESlike sequence of IB␣ is an authentic NES that mediates binding of IB␣ to CRM1.
Our results indicate that the N-terminal NES-like sequence of IB␣ is also required for IB␣-mediated nuclear export of Rel proteins. In the case of c-Rel, mutations with the N-terminal NES-like sequence disrupted the ability of IB␣ to relocalize c-Rel from the nucleus to the cytoplasm without significantly altering the ability of IB␣ to associate with c-Rel. In contrast, mutations within the C-terminal NES-like sequence had no effect on relocalization of c-Rel by IB␣, despite a slight reduction in the affinity of IB␣ for c-Rel. In the case of v-Rel, a mutant form of c-Rel that has lost critical contacts with IB␣ that are required for efficient masking of the Rel-derived NLS (15), we find that both NES-like sequences of IB␣ are required for relocalization of v-Rel from the nucleus to the cytoplasm. Mutation of the N-terminal NES-like sequence of IB␣ did not alter the affinity of IB␣ for v-Rel, whereas mutation of the C-terminal NES-like sequence of IB␣ resulted in a significant reduction in affinity of IB␣ for v-Rel. This result suggests that the inability of the IB␣-CA4 protein to relocalize v-Rel from the nucleus to the cytoplasm is a consequence of impaired association between v-Rel and IB␣ rather than a consequence of impaired association with CRM1. Because IB␣ is an adaptor protein that bridges Rel proteins and CRM1, loss of association with either the Rel protein or with CRM1 would result in loss of IB␣-mediated export of Rel proteins from the nucleus. We suggest that a reduced affinity of IB␣ for Rel proteins, rather than a reduced affinity for CRM1, provides an explanation for why previous reports in the literature, including one  1-5), GST-IB␣-NA4 (lanes 6 -10), or GST-IB␣-CA4 (lanes [11][12][13][14][15] were incubated with 293T lysates containing the c-Rel-(full-length) protein. Increasing amounts of His-IB␣ was added (lanes 2-5, 7-10, and 12-15). After incubation, beads were extensively washed. Bound proteins were eluted with SDS-PAGE sample buffer, electrophoresed through an SDS-7.5% polyacrylamide gel, and subjected to immunoblot analysis using anti-Rel sera. The location of the c-Rel protein is indicated by the arrow. B, the amount of c-Rel bound was determined by phosphorimaging analysis. The bound c-Rel is plotted against the concentration of His-IB␣ protein. C, the same competition experiment was performed, as described in A, using 293T cell lysates containing v-Rel-(full-length) instead of c-Rel. The location of the v-Rel protein is indicated by the arrow. N.S., nonspecific. D, the amount of v-Rel bound was determined by phosphorimaging analysis. The bound v-Rel is plotted against the concentration of His-IB␣ protein.
report from our laboratory (14,15), had concluded that the C-terminal NES-like sequence of IB␣ was functioning as a bona fide NES.
In the RanGAP assay, the concentration range over which a CRM1-interacting cargo protein provides protection against RanGAP-induced GTP hydrolysis provides an estimate of the affinity of the cargo protein for CRM1 (11). In comparison of IB␣ with the NS2 protein of MVM, we found that the affinity of IB␣ for CRM1 is ϳ100-fold less than that of NS2. It is likely that the affinity of IB␣ for CRM1 is very near the lower limit of exportable substrates, because the GST-IB␣-NA4 protein has only a reduced affinity for CRM1 but is completely defective for CRM1-dependent export.
It is puzzling why IB␣ has a relatively weak affinity for CRM1, in light of the notion that IB␣-mediated nuclear export of NFB/Rel proteins is important for cytoplasmic sequestration of NFB/Rel proteins. One possibility is that CRM1 might discriminate between free IB␣ and NFB/Rel-associated IB␣ proteins by virtue of an increased affinity for NFB/Rel-associated IB␣ proteins as compared with free IB␣. Our results do not support this scenario, because the inclusion of either c-Rel or v-Rel in the RanGAP protection assay did not alter the weak affinity of IB␣ for CRM1. An alternative possibility is that the low affinity of IB␣ for CRM1 is necessary to allow accumulation of sufficient IB␣ to displace NFB/Rel proteins from DNA prior to binding to CRM1 as a NFB⅐Rel-IB␣ complex. Under this scenario, a high affinity interaction between IB␣ and CRM1 would result in futile shuttling of free IB␣ between the nucleus and the cytoplasm, whereas a low affinity interaction between IB␣ and CRM1 would ensure that NFB/ Rel-associated IB␣ proteins are also available for CRM1-dependent nuclear export. In this regard, it is of interest to note that the Rev protein of HIV is also both an adaptor protein for CRM1-dependent export of human immunodeficiency virus RNA and a low affinity substrate for CRM1 (11). Indeed, it may generally be the case that export adaptor proteins such as IB␣ and Rev must be low affinity substrates for CRM1 to enable binding of these adaptor proteins to their targets prior to being removed from the nucleus by CRM1.