Transportin Is a Major Nuclear Import Receptor for c-Fos

c-Fos, a component of the transcription factor AP-1, is rapidly imported into the nucleus after translation. We established an in vitro system using digitonin-permeabilized cells to analyze nuclear import of c-Fos in detail. Two import receptors of the importin β superfamily, importin β itself and transportin, promote import of c-Fos in vitro. Under conditions where importin β-dependent transport was blocked, c-Fos still accumulated in the nucleus in the presence of cytosol. Inhibition of the transportin-dependent pathway, in contrast, abolished import of c-Fos. Furthermore, c-Fos mutants that interact with transportin but not with importin β were efficiently imported in the presence of cytosol. Hence, transportin appears to be the predominant import receptor for c-Fos. A detailed biochemical characterization revealed that the interaction of transportin with c-Fos is distinct from the interaction with its established import cargoes, the M9 sequence of heterogeneous nuclear ribonucleoprotein A1 or the nuclear localization sequence of some basic proteins. Likewise, the binding sites on importin β for its classic import cargo and for c-Fos can be separated. In summary, c-Fos employs a novel mode of receptor-cargo interaction. Hence, transportin may be as versatile as importin β in recognizing different nuclear import cargoes.

Import of proteins into the nucleus is mediated by importins, members of the importin ␤ superfamily. The "classic" signals for nuclear import (nuclear localization signal; NLS) 2 are short basic stretches of amino acids with lysines as characteristic components, which may occur either as a single or a bipartite motif. These classic NLSs are recognized by the adapter molecule importin ␣, which forms a heterodimer with the actual transport receptor, importin ␤ (for reviews see Refs. [1][2][3]. The importin ␣/␤-NLS-cargo complex is then translocated through the nuclear pore complex. Nucleoporins, the protein components of the nuclear pore complex, may interact with the transport receptor, facilitating the translocation of the complex. Importin ␤ interaction with RanGTP on the nuclear side of the nuclear pore complex results in dissociation of the transport complex (4). Hundreds of individual nuclear proteins or proteins that shuttle between the nucleus and the cytoplasm are thought to use this classic importin ␣/␤ nuclear import pathway. Over the last couple of years, however, a number of proteins have been described that do not depend on the adapter protein importin ␣. Some of these bind directly to the classic import receptor, importin ␤, without requiring any adapter protein. These include the proteins Rev and Rex of the human immunodeficiency virus (5,6) and the human T-cell leukemia virus (7), respectively, the T-cell protein tyrosine phosphatase (8), the parathyroid hormone-related protein (PTHrP) (9), cyclin B1 (10), the sterol regulatory element-binding protein 2 (SREBP-2 (11)), the zinc finger protein Snail (12), and the transcription factor CREB (13). Many of these proteins also contain basic NLSs. In contrast to the classic NLS, however, arginine is the predominant basic amino acid in these motives. Other import substrates do not bind to importin ␤ but to other members of the importin ␤ superfamily. The best described member of this family (besides importin ␤ itself) is transportin, which has originally been shown to serve as an import receptor for the M9 sequence, a 38-amino acid domain in the heterogeneous nuclear ribonucleoprotein A1 protein that functions as an NLS (14). Transportin is also an import receptor for other heterogeneous nuclear ribonucleoproteins (15) as well as for ribosomal proteins (16) and the nucleoporin Nup153 (17). Other members of the family are, for example, importin 5 and importin 7, with ribosomal proteins and core histones as import substrates (16), and importin 8, which mediates nuclear import of a protein of the signal recognition particle (18). Some cargo molecules like histones or certain ribosomal proteins can be imported by more than one receptor (16). For some importins, no import cargo has been identified so far (3).
AP-1 (activator protein-1), one of the first mammalian transcription factors to be identified, converts short-term signals to long-lasting responses in a large variety of complex cellular processes like transformation, proliferation, differentiation, and apoptosis (for reviews see Refs. 19 and 20). The best characterized components of AP-1, which functions as a dimer of basic leucine zipper proteins, are c-Jun and c-Fos. c-Jun may form homodimers, but it interacts preferentially with c-Fos proteins, forming heterodimers. c-Fos, unable to form homodimers, is a nuclear proto-oncogene that is expressed at low levels in most exponentially growing cells. Its expression level can be increased by a variety of stimuli. After translation, c-Fos accumulates in the nucleus very rapidly. Its nuclear import appears to be regulated by extracellular signals, as newly synthesized c-Fos accumulates in the cytoplasm in serum-deprived cells (21,22). Different regions in the c-Fos protein are required for its efficient nuclear import (23,24). Nuclear transport of c-Fos and c-Jun appears to be independent of importin ␣, because both proteins interact directly with the transport receptor importin ␤. The apparent affinities of the transcription factors for importin ␤, however, are rather low (13), and importin ␤-dependent transport has not been demonstrated in a permeabilized cell system so far.
In this study, we investigate nuclear import of c-Fos in detail, using digitonin-permeabilized cells. Surprisingly, c-Fos still accumulated in the nucleus, even when importin ␤-dependent transport was strongly inhibited. We identified transportin as the major import receptor for c-Fos. Under certain conditions, importin ␤ may also function as a transport receptor for the transcription factor. A detailed biochemical analysis revealed that the binding sites for c-Fos and classic import cargoes are distinct on both transportin and importin ␤. Likewise, distinct regions in c-Fos are involved in the interaction with either transport receptor. Regarding transportin, these results clearly point to a novel mode of receptor-cargo interaction.
Protein Purification and Labeling-Protein expression was induced with 0.25 mM (MBP-tagged proteins) or 0.5 mM (all others) isopropylthio-␤-⌬-galactoside at 25°C for 2-4 h. GST fusion proteins were expressed in BL21(-DE3) cells and natively purified by single-step affinity chromatography using glutathione-Sepharose beads (Amersham Biosciences), according to the instructions of the manufacturer. MBP fusion proteins were expressed in BL21 cells in the presence of 2% glucose. For MBP-transportin, MBP-transportin-N, MBP-transportin-C, and MBP-IBB, bacterial pellets were resuspended in buffer A (20 mM Tris, pH 7.4, 200 mM NaCl, 5% glycerol) containing 1 mM dithiothreitol and 1 g/ml each of leupeptin, aprotinin, and pepstatin. After sonification, lysates were cleared by centrifugation at 100,000 ϫ g and incubated with amylose beads (New England Biolabs) for 4 h at 4°C. The beads were washed with buffer A, and bound proteins were eluted with buffer A containing 10 mM maltose, 1 mM dithiothreitol, and protease inhibitors. For purification of MBP-M9, the bacterial pellet was resuspended in buffer B (20 mM Tris, pH 7.4, 50 mM NaCl, 5% glycerol) containing 1 mM dithiothreitol and protease inhibitors as above. After sonification and centrifugation at 100,000 ϫ g, the lysate was incubated with DEAE-Sepharose (Sigma). The flow-through was collected and subjected to a 40 -75% saturation ammonium sulfate cut, followed by gel-filtration on a S200 column (Amersham Pharmacia Biotech). His-Fos was expressed in BL21(-DE3) cells and natively purified using nickel-nitrilotriacetic acid-agarose (Qiagen), according to the instructions of the manufacturer. Importin ␤ 1-396 and 304 -876 (28) were expressed in BL21(-DE3) and purified via their N-terminal His tags using nickel-nitrilotriacetic acid-agarose. Bound proteins were eluted with 300 mM imidazole and supplemented with 30 mM dithiothreitol, inducing cleavage of the C-terminal intein tag. The tag was removed by addition of chitin beads (New England Biolabs). His-S-Importin ␤ (43), His-transportin (27), His-importin 5 (16), His-importin 7 (16), wild-type Ran, and RanQ69L (29) were purified as described. All GST fusion and MBP fusion proteins, except of MBP-M9, were dialyzed against 20 mM Tris, pH 7.4, 200 mM NaCl, 5% glycerol. MBP-M9, transport receptors, Ran, and RanQ69L were dialyzed against transport buffer (20 mM Hepes, pH 7.3, 110 mM KOAc, 2 mM Mg(OAc) 2 , 1 mM EGTA). For binding studies, RanQ69L was loaded with GTP as described (30). His-Fos was dialyzed against 100 mM sodium carbonate buffer, pH 9.3, and labeled with Cy3 (Amersham Biosciences), according to the instructions of the manufacturer. Preparation of FITC-BSA-NLS was essentially as described (31). All proteins were frozen in liquid nitrogen and stored at Ϫ80°C.
Binding Studies-Binding of cytosolic proteins to c-Fos: 200 g of GST, GST-Fos, or GST-Fos3-160 were immobilized on 100 l of glutathione-Sepharose (Amersham Biosciences) and incubated with 600 l of HeLa cell cytosol (32); 10 mg/ml, in binding buffer (50 mM Tris pH 7.4, 150 mM NaCl, 1 mM MgCl 2 , 5% glycerol), that had been preincubated with or without 10 M RanQ69LGTP. After incubation for 4 h at 4°C, beads were washed three times with binding buffer, and bound material was eluted with 120 l of 2ϫ SDS buffer. 5 l each were analyzed by SDS-PAGE, followed by Western blotting or silver staining. 60 l were analyzed by SDS-PAGE followed by Coomassie staining and mass spectrometry. Binding of recombinant transport receptors to c-Fos: GST or GST fusion proteins were immobilized on glutathione-Sepharose beads that had been preincubated with 10 mg/ml BSA to a final concentration of 1 g/l. 5 g of recombinant His-tagged or MBPtagged transport receptors in 400 l of binding buffer (50 mM Tris, pH 7.4, 200 mM NaCl, 1 mM MgCl 2 , 5% glycerol, 2 mg/ml BSA) were added to 5 l of beads. In some reactions, transport receptors were preincubated with 2 M RanQ69LGTP or a 20-fold molar excess (relative to the immobilized proteins) of MBP-IBB and MBP-M9 as competitors. After incubation for 2 h at 4°C, beads were washed three times with binding buffer without BSA, and bound proteins were eluted with 2ϫ SDS buffer. In experiments where relative affinities between c-Fos and import receptors were measured, importin ␤ without an affinity tag was used (purified by ammonium sulfate precipitation and gel filtration; a kind gift from F. Melchior). In binding studies with His-S-importin ␤, 5 g of this import receptor was immobilized on 5 l of S-protein-agarose (Amersham Biosciences) and incubated with an equimolar amount of GST-Fos, preincubated with or without an equimolar amount or a 10-fold excess of transportin.
Mass Spectrometry-Protein digestion was performed essentially as reported (33), and peptides were desalted and concentrated using C18 extraction tips (34). The peptide mixture was analyzed with an Ultimate nanoHPLC system (LC Packings) coupled to an ion trap mass spectrometer (LCQ, Thermo Finnigan) using a C18 reversed-phase column (PepMap, LC Packings). Peak lists were searched against the Swiss-Prot Protein Sequence Data Base.
In Vitro Import Assays-For nuclear import assays (36), HeLa cells were grown on 10-well slides (Roth), permeabilized with 30 g/ml digitonin (Calbiochem) in transport buffer on ice, and washed. In some experiments, permeabilized cells were preincubated with 200 g/ml wheat germ agglutinin (WGA, Sigma) in transport buffer for 12 min on ice. Each import mixture contained 250 -500 nM import substrates, 4 M Ran or RanQ69L, 2 mM GTP, 3 mg/ml BSA, an ATP-regenerating system (32), and either 2 mg/ml HeLa cytosol (in transport buffer) or 500 nM His-tagged importins. In competition experiments, HeLa cytosol was preincubated with 3 l of a monoclonal anti-importin ␤-antibody (Affinity BioReagents, clone 3E9 (32)) or with 20 M of MBP-M9 as competitor. Import reactions were performed for 30 min at 25°C for c-Fos or at 30°C for GST-NLS, BSA-NLS, and GST-M9. After washing, cells incubated with Cy3-Fos or FITC-BSA-NLS were fixed with 3.7% formaldehyde and analyzed directly by fluorescence microscopy. Cells incubated with GST-tagged import substrates were first subjected to indirect immunostaining using an anti-GST-antibody (Amersham Biosciences) and Alexa 594 donkey-anti-goat (Molecular Probes). Cells were analyzed by fluorescence microscopy using a Zeiss Axioskop2 microscope. Pictures were processed using Adobe Photoshop.
SDS-PAGE and Western Blotting-Proteins were separated by SDS-PAGE and subjected to Western blotting using standard methods. Rabbit anti-importin ␤ and anti-importin 7 antibodies were a kind gift from D. Görlich (35,52). The rabbit-anti-transportin antibody was a kind gift from L. Gerace. Rabbit anti-maltose-binding protein (MBP) was purchased from New England Biolabs. The mouse anti-His antibody was from Covance. HRP-coupled goat anti rabbit or goat anti mouse IgG (Dianova) was used as secondary antibody. The ECL system (Pierce) was used for visualization of proteins.

RESULTS
Importin ␤-independent Nuclear Import of c-Fos in Vitro-The transcription factor c-Fos has been suggested to be imported into the nucleus by the transport receptor importin ␤, independent of the adapter protein importin ␣ (13). Importin ␤-dependent import of c-Fos in permeabilized cells, however, has not been demonstrated so far. To analyze nuclear transport of c-Fos in detail, we established an import assay in digitonin-permeabilized cells (36). As shown in Fig. 1A, c-Fos was efficiently imported into nuclei of permeabilized cells at 25°C in the presence of cytosol and wild-type Ran. Transport was energy-dependent, as it was blocked at 4°C, and required the addition of soluble factors, as no nuclear accumulation was observed in cells incubated with buffer alone. To test whether nuclear import of c-Fos depends on transport receptors of the importin ␤-superfamily, a reaction was performed in the presence of RanQ69L, a Ran-mutant that is predominantly in the GTP-bound form (37), preventing the interaction of import substrates with import receptors (4). RanQ69L strongly inhibited nuclear accumulation of c-Fos, demonstrating the specificity of the nuclear transport reaction. We next used our in vitro transport assay to test if import of c-Fos depends on the activity of importin ␤. Importin ␤-dependent transport can be inhibited with a monoclonal antibody against this transport receptor (38). Preincubation of cytosol with this antibody resulted in strong inhibition of nuclear import of GST-NLS, a substrate that uses the importin ␣/␤ dimer as a transport receptor (Fig. 1B). In contrast, nuclear import of GST-M9, a substrate for the import receptor transportin, was not affected by the anti-importin ␤ antibody. Likewise, nuclear import of c-Fos was not significantly reduced upon preincubation of cytosol with the anti-importin ␤ antibody. These results strongly suggest that cytosolic transport factors distinct from importin ␤ can mediate nuclear import of c-Fos.
c-Fos Interacts with Different Import Receptors-To identify transport receptors that specifically interact with c-Fos, a GST fusion protein of the transcription factor was immobilized on Sepharose beads and incubated with cytosol in the absence or presence of RanQ69L. This Ran-mutant inhibited nuclear import of c-Fos (Fig. 1A), and we therefore expected it to release transport-relevant interaction partners from the immobilized bait. As shown in Fig. 2A, two proteins ϳ100 and 120 kDa specifically bound to GST-Fos in the absence but not in the presence of RanQ69L. To identify these potential import receptors for c-Fos, the bands were cut from the gel and tryptic digests of proteins were analyzed by mass spectrometry. In each band, two proteins could unequivocally be identified. The strong band at 100 kDa contained importin ␤ and transportin (specifically, transportin 1 (14)), whereas the weaker band at 120 kDa contained importins 7 and 8. We next used specific antibodies against three of these four import receptors to verify their specific interaction with GST-Fos. Importin ␤, importin 7, as well as transportin specifically bound to GST-Fos in the absence but not in the presence of RanQ69L (Fig. 2B). No binding to immobilized GST was detected. Strikingly, a strong accumulation of transportin on the GST-Fos beads was observed, whereas only a rather small proportion of importin ␤ or importin 7 was bound to the beads (compare "input" and "GST-Fos"). c-Fos contains a leucine-zipper and forms stable dimers with proteins like c-Jun. To exclude the possibility that the identified import receptors bound indirectly to c-Fos via proteins that interact with its leucine-zipper, we expressed an N-terminal fragment of c-Fos, lacking this protein interaction domain (GST-Fos 3-160; see also Fig. 5A). Binding of importin 7, importin ␤, and transportin to this fragment was as strong as to the full-length protein (Fig. 2B), suggesting a direct interaction. To demonstrate such a direct interaction of import receptors with c-Fos, importin 7, importin ␤, and transportin, as well as importin 5 as a negative control, were expressed in bacteria and incubated with immobilized GST-Fos. Importin 5 did not interact with GST-Fos under these conditions (Fig. 2C). Importin ␤ and transportin bound efficiently to GST-Fos, whereas importin 7 bound rather weakly (compare "GST-Fos" to "input"). Binding of importin ␤ and transportin was abolished when RanQ69L was included in the reaction, demonstrating the specificity of the interaction. Binding of importin 7 to GST-Fos was less sensitive to RanQ69L, probably reflecting the lower affinity of this transport receptor for RanGTP, compared with importin ␤ (35). Importin 7 has been shown to form a heterodimer with importin ␤ (35). This dimer serves as an import receptor for histone H1 (39). We therefore combined importin ␤ and importin 7 in the binding reaction and now observed increased binding of importin 7 (together with importin ␤) to GST-Fos (Fig. 2C). This result suggests that at least part of the cytosolic importin 7 that interacted with GST-Fos ( Fig. 2A) was indirectly bound via importin ␤. Taken together, different transport receptors of the importin ␤ superfamily specifically interact with c-Fos and may therefore function as physiological import receptors.
To test which of the identified import receptors promote efficient nuclear transport of c-Fos, we performed import reactions in digitoninpermeabilized cells, using recombinant proteins. Cells were incubated either at 4°C or 25°C with equimolar concentrations of the various transport receptors. At 25°C, transportin promoted the strongest accumulation of c-Fos in the nucleus (Fig. 3A). Importin ␤ also mediated nuclear import of c-Fos, yet to a somewhat lower extent. Importins 5 and 7 did not promote nuclear import of c-Fos, although they were transport-competent for their established import cargo, the ribosomal protein L23 (data not shown (16)). Combining importin ␤ with its binding partner importin 7 did not further promote nuclear import, compared with a reaction with importin ␤ alone. Thus, transportin and importin ␤ appear to be the major transport receptors for nuclear import of c-Fos. Transport was clearly specific, because it was temperature-dependent (Fig. 3A) and could be inhibited by RanQ69L and by WGA, a lectin that binds to O-linked nucleoporins (40) and blocks various nucleocytoplasmic transport pathways (41,42) (Fig. 3B).
Characterization of the Interaction between c-Fos and Transportin or Importin ␤-Having established that at least two importins, transportin and importin ␤, may serve as import receptors for c-Fos, we next analyzed the interaction of the transcription factor with these receptors in detail. We first expressed N-and C-terminal fragments of both importin ␤ and transportin and tested them for interaction with either an established binding partner or with c-Fos. A GST fusion of the importin ␤ binding (IBB-) domain of importin ␣ was immobilized on beads and incubated with full-length importin ␤, or with its N-or C-terminal

. c-Fos directly interacts with various import receptors. A, identification of import-relevant binding partners of c-Fos. GST or GST-Fos was immobilized on beads and incubated with cytosol
in the absence or presence of 10 M RanQ69LGTP, as indicated. Proteins in regions 1 and 2 (asterisks) were cut from the gel, digested with trypsin, and analyzed by mass spectrometry. The band marked by the arrowhead corresponds to GST-Fos. B, importin 7, importin ␤, and transportin specifically interact with GST-Fos. GST, GST-Fos, or GST-Fos 3-160 were immobilized on beads and incubated with cytosol and RanQ69LGTP as in A. The cytosolic input corresponds to 1% of the protein used per binding reaction. Note the relative accumulation of transportin on GST-Fos beads, compared with importin 7 and importin ␤. C, recombinant import receptors directly interact with c-Fos. 5 g of recombinant transportin, importin 5, importin 7, importin ␤, or a combination of importin ␤ and importin 7 were pretreated with or without 20 g of RanQ69LGTP and incubated with 5 g of immobilized GST or GST-Fos, as indicated. The input corresponds to 25% of the import receptors used per binding reaction. BSA was included as a blocking reagent. Interacting proteins were analyzed by SDS-PAGE, followed by silver-staining (A and C) or Western blotting (B), using specific antibodies against the transport receptors.
fragment. As expected (43), the IBB-domain preferentially bound to the C-terminal fragment of importin ␤ (Fig. 4A). In contrast, when GST-Fos was immobilized, the N-terminal, but not the C-terminal fragment of importin ␤ was retained on the beads. No binding was observed when GST alone was used as a bait.
In a similar set of experiments, we immobilized GST, GST-M9, or GST-Fos on beads and incubated them with full-length transportin or its N-or C-terminal fragments (Fig. 4B). In these experiments, transportin fragments fused to the maltose binding protein (MBP) were used, because His-tagged fragments were insoluble. Binding of the MBPtransportin fragments to GST beads was not observed. GST-M9 interacted with full-length transportin as well as with the C-terminal fragment, but not with the N-terminal fragment. GST-Fos, in contrast, interacted with all three proteins. Binding of free MBP to the beads was not detected. Interaction of full-length transportin with GST-Fos was specific, as it could be blocked by RanQ69L (see Fig. 2C), suggesting that the interaction of c-Fos with the two individual fragments of transportin is specific, too. We could directly test this assumption for the N-terminal fragment, which contains the Ran-binding site of transportin. As shown in Fig. 4C, the interaction of GST-Fos with this fragment was largely inhibited when RanQ69L was included in the reaction. These results show that c-Fos interacts with different regions of both importin ␤ and transportin compared with their classic binding partners, the IBB-domain of importin ␣ and the M9 sequence, respectively.
In light of this differential binding, we next analyzed whether these classic binding partners and c-Fos could interact simultaneously with either importin ␤ or transportin. Immobilized GST-IBB or GST-Fos was incubated with importin ␤, in the absence or presence of an excess of the IBB-domain, coupled to the maltose binding protein (MBP-IBB). As expected, no binding of importin ␤ to GST-IBB was observed when MBP-IBB was present (Fig. 4D, top panel). MBP-IBB also abolished the interaction of importin ␤ with c-Fos, suggesting that binding of IBB and c-Fos to importin ␤ are mutually exclusive. Similarly, an M9 sequence coupled to MBP prevented the interaction of transportin with both GST-M9 and GST-Fos (Fig. 4D, bottom panel). These results show that the two cargo binding sites on importin ␤ and transportin cannot be occupied simultaneously by the classic cargoes of the import receptors and by c-Fos. This may result from conformational changes in the structure of the receptors upon binding one type of ligand, preventing simultaneous binding of a second ligand to a different binding site.
To characterize the regions in c-Fos required for interaction with importin ␤ and transportin, we generated a number of fragments of the transcription factor as well as two point mutations as GST fusion proteins (Fig. 5A). These proteins were immobilized on beads and incubated with either full-length importin ␤ or full-length transportin. A fragment, where the C-terminal part of c-Fos, including the leucinezipper was deleted, interacted with importin ␤ (Fig. 5B) and transportin (Fig. 5C), similar to the full-length protein (compare lanes 1 and 2). Thus, the C-terminal part of c-Fos beyond the basic DNA-binding domain ("ϩϩϩ " in Fig. 5A) is not required for interaction with the transport receptors. The DNA-binding domain has been reported to function as an NLS (23). Indeed, when one, two or three of these basic regions were deleted, the corresponding c-Fos fragment failed to interact with importin ␤ (Fig. 5B, lanes 3-5). Similarly, when two arginines in either the second or the third basic region were mutated to alanines, full-length c-Fos did not bind to importin ␤ (Fig. 5B, lanes 9 and 10). Strikingly, however, all these fragments/mutants still interacted with transportin (Fig. 5C, lanes 3-5, 9, and 10), demonstrating that different regions in c-Fos must be responsible for binding to the different transport receptors. Accordingly, a short fragment comprised of the second and third basic region of c-Fos bound to importin ␤ (Fig. 5B, lane 7) but hardly to transportin (Fig. 5C, lane 7). In an attempt to further narrow down the binding site on c-Fos for transportin, we tested other fragments of the transcription factor. Fragments comprising amino acids 3-124 or 81-160 efficiently bound to transportin (Fig. 5, lanes 5 and 8), whereas a fragment comprising amino acids 3-110 did not (Fig. 5, lane  6). These results suggested that amino acids 81-124 (overlap between fragments 5 and 8) or 111-124 (deleted residues in fragment 6 compared with fragment 5) are required for binding. A fragment comprising amino acids 81-124, however, did not bind to transportin (data not shown), arguing for a more complex mechanism of interaction.
As importin ␤ and transportin bind to different regions of c-Fos, we next asked whether the transcription factor could interact simulta- neously with the two transport receptors. Importin ␤ was immobilized on beads and incubated with GST-Fos in the absence or presence of transportin. As shown in Fig. 5D, an equimolar concentration of transportin was sufficient to abolish binding of GST-Fos to importin ␤, suggesting that the proteins cannot form a trimeric complex.
Transportin Is the Major Import Receptor for c-Fos-Our results have shown that c-Fos can be imported into the nucleus by two transport receptors, importin ␤, and transportin. Is there any preference for one of these factors? To answer this question, we first compared relative affinities of importin ␤ and transportin for c-Fos. GST-Fos was immobilized on beads and incubated either with transportin or importin ␤ alone, or with transportin in combination with increasing concentrations of importin ␤. The amounts of GST-Fos and import receptors were chosen such that GST-Fos would be limiting for the binding reactions. Transportin and importin ␤ efficiently interacted with GST-Fos when they were added separately to the reaction (Fig. 6A). When the two transport receptors were present at equimolar concentrations, transportin still bound efficiently to GST-Fos, whereas importin ␤ was almost undetectable, in agreement with the result presented in Fig. 5D. Even at a 20-fold molar excess of importin ␤ over transportin, there was still substantial binding of transportin to GST-Fos, suggesting that transportin has a much higher affinity for the transcription factor compared with importin ␤. We next tested whether fragments or mutants of c-Fos that interact with transportin but not with importin ␤ could be imported efficiently into nuclei of digitonin-permeabilized cells in the presence of cytosol. As shown in Fig. 6 (B and C), the fragment 3-154 as well as the mutant R157A/R158A (compare Fig. 5A) were imported at 25°C but not at 4°C. Import was specific, as it could be blocked by RanQ69L or by WGA. Under conditions where importin ␤-dependent transport was inhibited, import of c-Fos still occurred (Fig. 1B). We now finally tested, if inhibition of the transportin-dependent nuclear import pathway would abolish transport of c-Fos into the nucleus. Permeabilized cells were incubated with the transport substrates GST-M9, BSA-NLS, or GST-Fos with cytosol as the source of transport factors. All substrates were imported efficiently when no inhibitor was present (Fig.  6D). When we included MBP-M9 in the reaction, a substrate that competes with transportin-dependent cargoes for efficient import, we observed strong inhibition of transport of GST-M9, as well as of c-Fos. The importin ␣/␤-dependent substrate BSA-NLS was not affected by MBP-M9, demonstrating the specificity of inhibition. Thus, under con- ditions where importin ␤-dependent transport still occurs, inhibition of transportin abolishes the nuclear import of c-Fos, suggesting that transportin is the major cytosolic transport receptor for the transcription factor under our experimental conditions.

DISCUSSION
Transportin Is the Major Import Receptor for c-Fos-Different regions in c-Fos required for its efficient import into the nucleus have been identified (23,24). Transport factors that promote nuclear import of c-Fos in the classic in vitro system, however, have not been described. In the course of this study, we analyzed nuclear import of c-Fos in digitonin-permeabilized cells. In this system, the import receptors transportin and importin ␤ are able to promote specific import of c-Fos into the nucleus. Several lines of evidence, however, point to a dominant role of transportin for nuclear import of c-Fos. (i) Transportin was highly enriched on GST-Fos beads incubated with a cytosolic extract. In contrast, only a small fraction of other import receptors like importin ␤ and importin 7 bound to immobilized GST-Fos. (ii) Recombinant transpor-tin promoted nuclear import of GST-Fos more efficiently than importin ␤. Importin 7 did not promote nuclear import of c-Fos. Like importin 8, importin 7 interacts with importin ␤ (35). Thus, these two proteins may have interacted with GST-Fos via importin ␤ ( Fig. 2A). (iii) Competition experiments revealed that GST-Fos interacts more efficiently with transportin than importin ␤, suggesting a much higher affinity. The published dissociation constant of the c-Fos-importin ␤ complex is rather high (65 nM (13)) compared with the dissociation constant of the PTHrP-importin ␤ complex (3.5 nM (9)) or to the binding affinity of importin ␣ to importin ␤ (5-18 nM (44)). (iv) Fragments or mutants of c-Fos that do not interact with importin ␤ (but with transportin) were efficiently imported into the nucleus in the presence of cytosol. (v) Most importantly, under conditions where transportin-dependent import was inhibited by a substrate-competitor, c-Fos did not enter the nucleus, whereas an importin ␣/␤ substrate was efficiently transported. In contrast, inhibition of the importin ␤-dependent pathway with an antibody against the receptor did abolish import of GST-NLS but not that of c-Fos. Taken together, we believe that transportin is the major import receptor for c-Fos. Under certain conditions of activation or in certain cell types, however, importin ␤ may serve as an importin for c-Fos as well. This interpretation is in agreement with earlier results that showed that a 22-amino acid basic region (i.e. the importin ␤-interaction domain, see below) can function as an NLS when fused to pyruvate kinase as a reporter protein (23). Deletion of this basic region, however, did not affect the nuclear localization of c-Fos, suggesting alternative import pathways.
In living cells, nuclear import of c-Fos does not occur by default, because the newly synthesized protein is retained in the cytoplasm in serum-starved cells (21). One example for a stimulus-dependent transport is the UV-induced phosphorylation of c-Fos by p38 kinases and the subsequent import of the transcription factor into the nucleus (45). Our bacterially produced c-Fos (i.e. unphosphorylated protein) readily binds to transportin and importin ␤, suggesting that phosphorylation of c-Fos may rather regulate its cytoplasmic retention than its interaction with transport receptors. Nevertheless, different signaling pathways could regulate nuclear import of c-Fos by different importins. Furthermore, nuclear import pathways of c-Fos could differ from cell type to cell type, depending on relative expression levels of potential import receptors.
Interaction of c-Fos with Importin ␤ and Transportin-The basic regions in the DNA-binding domain of c-Fos have been implicated in nuclear import of the transcription factor (23). Indeed, deletions of one or more of these regions, as well as point mutations exchanging arginines for alanines abolished the interaction of c-Fos with importin ␤. Thus, c-Fos contains a basic, arginine-rich NLS that can be recognized by importin ␤, similar to other proteins (5-7). 19 HEAT repeats in importin ␤ generate an extensive interaction surface that allows binding of a large variety of molecules, including RanGTP, nucleoporins, and different import cargoes. Most prominent for cargo binding is the importin ␣-interaction domain in the C-terminal part of importin ␤ (43), comprising HEAT repeats 7-19 (46). In contrast, PTHrP binds to the concave surface of the N-terminal part of importin ␤, contacting HEAT repeats 2-11 (47). A third interaction mode is used by SREBP-2, which binds as a dimer to importin ␤ (48). SREBP-2 has no typical consecutive basic residues and the binding to importin ␤ is primarily mediated by hydrophobic interactions, with HEAT repeats 7 and 17 being of particular importance (48). c-Fos binds to the N-terminal portion of importin ␤, possibly in a similar way as PTHrP. Unlike c-Fos, however, PTHrP can form a trimeric complex with importin ␣ and importin ␤, at least in vitro (47), suggesting a somewhat different binding mode. A common feature of all importin ␤-cargo complexes, includ-FIGURE 6. Transportin is the major import receptor for c-Fos. A, 3.5 g of GST-Fos was immobilized on beads and incubated with either 5 g of importin ␤ or transportin alone or with transportin and increasing concentrations of importin ␤. Its molar-fold excess (5ϫ, 10ϫ, and 20ϫ) is indicated. Bound proteins were analyzed by SDS-PAGE, followed by Western blotting. The input corresponds to 25% of the import receptors (at 1ϫ) that were used per binding reaction. B, permeabilized cells were incubated with the fragment GST-Fos 3-154 (left panel) or the mutant GST-Fos RR157/158AA (right panel) at 4°C or 25°C with cytosol and wild-type Ran (WT) or RanQ69L (QL), as indicated. For WGA inhibition, cells were preincubated with 200 g/ml of the lectin at 4°C. C, cells were incubated with 250 nM of the import substrates GST-M9, BSA-NLS, or GST-Fos in the absence (ϪMBP-M9) or presence (ϩMBP-M9) of MBP-M9 as a competitor. B and C, GST fusion proteins were detected by indirect immunofluorescence. BSA-NLS was labeled with FITC.
ing the importin ␤-c-Fos complex, is their sensitivity to RanGTP, which binds to HEAT repeats 1-8 at the N terminus of the receptor (49). Thus, regardless of the nature of the receptor-cargo complex, common mechanism are at play for the dissociation of the complex by RanGTP in the nucleus.
The regions in c-Fos that are required for binding to transportin are clearly distinct from the basic regions needed for binding to importin ␤, because their deletion does not affect transportin binding. So far we have been unable to generate point mutations that do not bind to transportin. Amino acids 111-124 of the c-Fos protein appear to be important for binding, because the interaction was lost when these residues were deleted. This region, which is much shorter than the classic M9 sequence and does not contain consecutive basic residues, is not sufficient for binding, as even longer fragments (e.g. Fos 81-124) did not interact with transportin. Of course we cannot exclude problems resulting from misfolded proteins. Together, these results argue for a more complex mode of interaction, possibly involving residues at the N terminus of the transcription factor. By several criteria, however, the interaction of transportin with c-Fos is distinct from its binding to its established import cargoes. First, the N terminus of transportin clearly interacts specifically with c-Fos, in contrast to the classic import cargo, the M9 sequence. We also detected binding of c-Fos to the C-terminal part of transportin, which also interacts with the M9 sequence (50). Hence, there is probably more than one binding site on transportin for c-Fos, explaining the strong interaction of the importin with the transcription factor. Second, the transportin-binding fragment c-Fos 3-124 has no characteristic basic regions, unlike other import cargoes of transportin that do not contain an M9-like NLS (16). Therefore, the interaction of transportin with c-Fos is the first example of a third binding mode of the import receptor to a cargo molecule. The observation that binding of the M9 sequence and of c-Fos to transportin are mutually exclusive argues for extensive conformational changes in the receptor molecule upon cargo interaction, similar to what has been described for importin ␤ (48,51). Alternatively, the first cargo could mask the binding site for the second cargo.
In summary, transportin may be as versatile as importin ␤ when it comes to the accommodation of a variety of import cargoes. A detailed crystal structure analysis will be required to characterize the different transportin-cargo complexes at the molecular level.