Several Phenylalanine-Glycine Motives in the Nucleoporin Nup214 Are Essential for Binding of the Nuclear Export Receptor CRM1*

Background: Nup214 interacts with the nuclear export receptor CRM1 and promotes export of certain cargos. Results: Several FG motives in the C-terminal region participate in CRM1 binding. Conclusion: CRM1, like other transport receptors, makes multiple contacts to nucleoporins. Significance: Elucidation of the details of nucleoporin-receptor interactions is essential for our understanding of the transition process of transport complexes through the nuclear pore. Nucleoporins containing phenylalanine glycine (FG) repeats play an important role in nucleocytoplasmic transport as they bind to transport receptors and mediate translocation of transport complexes across the nuclear pore complex (NPC). Nup214/CAN, a nucleoporin that is found at the cytoplasmic side of the NPC, interacts with both import and export receptors. In functional assays, dominant-negative fragments of Nup214 inhibited CRM1-dependent nuclear export, as the export receptor became rate-limiting. Several nuclear import pathways, by contrast, were not affected by the Nup214 fragments. We now characterize the CRM1-binding region of Nup214 in detail and identify several FG motives that are required for this interaction. Our results support a model where CRM1, like other transport receptors, contacts FG-Nups via multiple binding sites.

receptors (NTRs) that mediate the translocation of transport substrates across the nuclear envelope (3). These NTRs mostly belong to the family of importin ␤-like proteins, also referred to as karyopherins. The prototype is importin ␤ itself, which together with its adapter importin ␣ interacts in the cytoplasm with proteins carrying a classical nuclear localization signal (cNLS; for review see Ref. 4). Another major importin with a large number of identified substrates is transportin (5), which recognizes a specific localization signal, the PY-NLS (for review see Ref. 6). For other import receptors like importin 5, importin 7, importin 9, or importin 13, only a limited number of substrates have been found so far and their cognate NLSs remain to be described in detail (6). Exportins, on the other hand, interact with their cargo in the nucleus via nuclear export sequences (NESs; for review see Ref. 7). Binding of transport cargoes to their receptors is controlled by the small G-protein Ran, which interacts with all importin ␤-like proteins. In nuclear import, nuclear RanGTP binds to a conserved region present in all importins and exportins and dissociates the binding of incoming NLS substrates, terminating transport (8). The import receptor can then recycle back to the cytoplasm in a complex with RanGTP. In nuclear export, by contrast, RanGTP is an integral component of a trimeric transport complex as it promotes the interaction of NES cargos with exportins in the nucleus (9). After translocation through the NPC, GTP hydrolysis on Ran then initiates the dissociation of the export complex. The localization of GTP hydrolysis and GTP loading of Ran is largely restricted to the cytoplasm and the nucleus, respectively. This results from the strict compartmentalization of the cytoplasmic GTPase-activating protein for Ran, Ran-GAP, and the chromatin-bound nucleotide exchange factor RCC1 (regulator of chromosome condensation 1; for review see Ref. 10). Besides controlling the cargo-receptor interaction, RanGTP also affects binding of NTRs to FG-Nups, either negatively (8) or positively (11). Understanding the molecular details and consequences of the Nup-NTR interaction will be required to mechanistically explain nucleocytoplasmic transport through the NPC. In the selective phase model (12), FG-Nups were suggested to be cohesive, resulting in a sieve-like structure that restricts passage of proteins unable to interact with the FG motives. Transport complexes containing NTRs, by contrast, bind to these motives, leading to disengagement of the FG-FG contacts and to partitioning of the complex into the sieve. RanGTP binding on the nuclear side (import) or GTP hydrolysis on Ran on the cytoplasmic side of the NPC (export) then confers directionality to the transport event. In other models, like the virtual gate model (13), FG-Nups that are predominantly noncohesive are postulated to form FG filaments, repelling proteins that do not interact with FG motives by entropic forces.
Although certain FG-Nups may play a more prominent role in establishing the selective barrier than others (14), multiple nucleoporins are thought to act together as a functional entity in all current models. Besides this collaborative role of FG-Nups as components of the general permeability barrier of the NPC, individual nucleoporins were shown to play roles in specific transport pathways as defined by specific NTRs. The cytoplasmic nucleoporin Nup358, for example, plays a general role in nuclear protein import, but not in export (15)(16)(17). In addition, it is required for efficient import of a subset of proteins (18). Nup214, also known as CAN, is also found at the cytoplasmic side of the NPC, but has been implicated in nuclear protein export rather than import (19,20), although the protein was also shown to interact with nuclear import receptors (21,22). It is an essential protein, as Nup214 Ϫ/Ϫ mice die early in development (23). Interestingly, Nup214 was found in multiple fusions with nuclear proteins like the histone chaperone Set or the chromatin-binding protein Dek, leading to oncogenic transformation (for review see Ref. 24). The predicted domains of Nup214 (Fig. 1A) are not very well characterized at the structural level. An exception is the N-terminal domain, which was shown to form a seven-bladed ␤-propeller (25) that interacts with the RNA-helicase Dbp5/Ddx19, suggesting a role in mRNA export (26,27). The predicted coiled-coil domain(s) are required for interaction with Nup88, another nucleoporin that forms a functional complex with Nup214 (28,29). The C-terminal part of Nup214 was shown to interact with the most prominent nuclear export receptor, the importin ␤-family member CRM1 (29), and depletion of Nup214 inhibited CRM1-dependent export of certain cargoes (19,20). CRM1 mediates nuclear export of hundreds of proteins carrying typical hydrophobic sequences that were originally described as leucine-rich NESs (30,31). CRM1 is also a major RNA-export factor that transports rRNAs and other, less abundant species, out of the nucleus (for review see Ref. 7). The crystal structure of CRM1 in complexes with Ran and export substrates has recently been solved (32)(33)(34)(35). Surprisingly little, however, is known about the interaction of CRM1 with nucleoporins. We now characterize the region in Nup214 that is required for CRM1 binding in detail and identify FG motifs in the nucleoporin that participate in complex formation. Our data suggest that multiple regions in CRM1 are involved in FG-Nup binding, similar to what has been described for importin ␤ (36, 37).
GST-Ran was expressed in BL21(DE3) by induction at 20°C with 0.2 mM isopropyl 1-thio-␤-D-galactopyranoside. Bacteria were harvested and lysed in buffer containing 50 mM Tris, pH 6.8, 200 mM NaCl, 0,25 mM EDTA, and 10% glycerol. The protein was purified with glutathione-Sepharose beads (High performance, GE Healthcare) according to the instructions of the manufacturer and dialyzed against Tris buffer as above. For GDP/GTP loading (39), GST-Ran was incubated for 30 min at room temperature with 4.5 mM EDTA and 10 mM GDP/GTP in Tris buffer. Afterward, 30 mM MgCl 2 was added, followed by incubation for 15 min on ice.
Cell Culture and Immunofluorescence Microscopy-HeLa P4 cells (44) were grown at 37°C and 5% CO 2 in Dulbecco's modified Eagle's medium containing 10% fetal calf serum, 100 units/ml of penicillin, 100 g/ml of streptomycin, and 2 mM L-glutamine. Transfections were performed with calcium phosphate (9.25 mM final concentration (45)) and HEPES-buffered saline (50 mM HEPES, pH 6.98, 250 mM NaCl, 1.5 mM Na 2 HPO 4 ). HeLa cells were transfected with 0.3 g of a plasmid coding for GFP-SPN1 (43) or 0.1 g of a plasmid coding for NC2␤-GFP 2 (46) and 0.5-0.6 g of plasmids coding for RFP-Nup214-cNLS fragments. For CRM1 overexpression experiments, HeLa cells were transfected with 0.1 g of a plasmid coding for GFP-SPN1 or 0.05 g of a plasmid coding for NC2␤-GFP 2 , 0.1 g of a plasmid coding for Myc-Nup214(1859 -2090), and 1 g of the HA-CRM1 plasmid. Images were collected with an Axioskop 2 (Zeiss, Jena) or with a laser-scanning microscope (LSM 510 meta; Carl Zeiss MicroImaging, Inc.) and processed with AxioVision (LE) Rel 4.8 or the LSM Image Browser (Rel 4.0 SP2; Carl Zeiss MicroImaging, Inc.) and Photoshop (Adobe). For quantification of the localization of proteins, at least 100 cells with comparable expression levels of the Nup214 fragments, as judged by the RFP fluorescence, were counted and divided into three categories (NϾC, mostly nuclear; NϭC, equally distributed between the nucleus and the cytoplasm; NϽC, mostly cytoplasmic) and data were presented as % distribution of the individual categories.
Antibodies-Anti-CRM1 antibodies against a C-terminal peptide (GIFNPHEIPEEMCD) were raised in a goat and purified by affinity chromatography. The mouse anti-myc antibody was obtained from Santa Cruz (USA), the rabbit anti-HA antibody from Sigma, the mouse anti-penta-His antibody from Qiagen, the rat anti-RFP-antibody from ChromoTek (Planegg-Martinsried, Germany), and the rabbit anti-tubulin antibody from Proteintech Group (USA). For indirect immunofluorescence, antibodies from Molecular Probes were used. For Western blots, horseradish peroxidase-coupled secondary antibodies from Dianova (Germany) were used. The enhanced chemiluminescence system (Millipore) was used for visualization of proteins.
Nuclear Transport Assays-To analyze induced nuclear import in vivo, cells expressing RevNES-GR 2 (511-795)GFP 2 -M9, RevNES-GR 2 (511-795)GFP 2 -cNLS, or GR 2 -GFP 2 -NC2␤ fusion proteins were grown on coverslips and treated with 5 M dexamethasone (Sigma) for 15 min at 37°C. The in vitro nuclear transport assay with GFP-NFAT was performed as described previously (47). Briefly, the expression of GFP-NFAT was induced with 1 M trichostatin A overnight. Nuclear import of GFP-NFAT was induced by the addition of 1 M ionomycin for 20 min. Digitonin-permeabilized cells were incubated with cytosol, Ran, and the import substrate BSA-Cy5-cNLS (47) in a final volume of 40 l at 30 or 4°C. After 30 min, the reactions were stopped by the addition of cold transport buffer. Nuclear fluorescence of GFP-NFAT and BSA-Cy5-cNLS was analyzed by flow cytometry using a FACS Canto TM II (BD Bioscience).
Binding Assays-Pulldown assays were performed in binding buffer (50 mM Tris-HCl, pH 7.4, 200 mM NaCl, 1 mM MgCl 2 , 5% glycerol, 2 mM DTT, 1 g of each aprotinin, leupeptin, and pepstatin). Per sample, 5 g of GST-tagged proteins were immobilized on 15 l of glutathione-Sepharose beads that had been preincubated with 20 mg/ml of BSA in binding buffer. Prior to coupling to beads, GST-Ran was loaded with GDP or GTP (39). After immobilization and a washing step to remove free nucleotides, it was incubated with 2.5 M NES peptide (NS2 protein of minute virus of mice, CVDEMTKKFGTLTIH-DTEK), 108 nM CRM1-His, and 160 nM His-Nup214 fragments. GST-Nup214 beads were incubated with 108 nM CRM1-His, 255 nM His-SPN1, and 960 nM RanGDP or RanGTP. Beads were incubated for 1 h at 4°C and washed three times with binding buffer. Bound proteins were eluted with SDS sample buffer and subjected to SDS-PAGE, followed by colloidal Coomassie staining or Western blotting.

Overexpression of a C-terminal Nup214 Fragment Leads to Mislocalization of Various Transport
Receptors-X-ray structures revealed that the interaction of NTRs with nucleoporins is mediated by FG motives (3,50). In vitro studies showed that the nucleoporin Nup214, which contains many FG motives in the C-terminal region, interacts with CRM1 (51), TAP/p15 (52), and importin ␤ (21). Overexpression of the C-terminal CRM1binding region of Nup214 caused nuclear accumulation of CRM1 and inhibited nuclear export in vivo (53). Similarly, importin ␤ colocalized with certain Nup214 fragments in this study. We now used this overexpression approach and investigated whether a C-terminal Nup214 fragment would also affect the localization of other NTRs. For the analysis, we used HA-or FLAG-tagged NTRs and an RFP-tagged C-terminal Nup214 fragment (aa 1861-2090). A cNLS was fused to the C terminus of Nup214 to ensure nuclear localization (Fig. 1A). In control cells expressing RFP-cNLS, importin ␣, importin ␤, importin 5, and importin 9 were detected in the cytoplasm, whereas transportin was located in the nucleus, and importin 7 and importin 13 were detected in both compartments (Fig. 1, B and C). CRM1 was mostly nuclear, but in cells expressing higher levels, a considerable portion could also be found in the cytoplasm. Similar to a previous study (53), CRM1 accumulated in the nucleus in the presence of the C-terminal Nup214 fragment (RFP-Nup214-cNLS), and the cytoplasmic CRM1 staining was completely lost. Likewise, the localization of importin ␣, importin ␤, importin 7, importin 9, and importin 13 was clearly shifted toward the nucleus in cells expressing RFP-Nup214-cNLS, suggesting that these importins can also interact with Nup214. Importin 5 and transportin, by contrast, were not affected by the Nup214 fragment.

CRM1-dependent Nuclear Export, but Not Nuclear Import Is
Inhibited by the C-terminal Nup214 Fragment-Nup214 has previously been implicated in TAP/p15-dependent RNA-and CRM1-dependent protein export (19,20,54). Although an interaction of Nup214 with import receptors has been reported (21,22), a function of the nucleoporin in receptor-mediated import remains unclear. In light of the effects described above, we re-addressed the question of whether inhibition of transport by the C-terminal Nup214 fragment (aa 1861-2090) is specific for CRM1-mediated export or whether the change in distribution of other transport factors affects other transport pathways as well. To answer this question, we investigated several transport pathways in cells expressing the C-terminal fragment (RFP-Nup214-cNLS). First, we used two reporter proteins that are able to shuttle between the nucleus and the cytoplasm. Snurportin 1 (SPN1) is a protein that is involved in import of small nuclear ribonucleoprotein particles into the nucleus and its recycling back to the cytoplasm is mediated by CRM1 (55).
The negative cofactor 2 ␤ (NC2␤), on the other hand, is a transcriptional regulator that can be exported to the cytoplasm via the CRM1 pathway (46). GFP-SPN1 and GFP-NC2␤ were largely found in the cytoplasm when expressed alone (data not shown) or together with RFP-cNLS, which lacked any Nup sequence (Fig. 2, A and B). Co-expression of RFP-Nup214-cNLS, however, led to a clear nuclear accumulation of GFP-SPN1 and GFP-NC2␤, indicating inhibition of nuclear export.
For the analysis of nuclear import, we used a dexamethasoneinducible system (56). Shuttling reporter substrates carrying an NES and the hormone-binding domain (aa 511-795) of the glucocorticoid receptor (GR) were fused to either a cNLS (allowing importin ␣/␤-dependent nuclear import) or an M9 sequence (allowing transportin-dependent nuclear import). Furthermore, we fused the GR fragment to NC2␤, which is imported by importin ␣/␤ as well (46). Without dexamethasone, the reporter proteins were equally distributed between the nucleus and the cytoplasm or even absent from the nucleus in cells B, HeLa cells were transiently cotransfected with plasmids coding for RFP-cNLS (R) or RFP-Nup214 (aa 1861-2090)-cNLS (N) and HA-or FLAG-tagged transport receptors as indicated (imp, importin; tra, transportin). After 24 h, the cells were fixed and subjected to indirect immunofluorescence, detecting the HA or FLAG tag (importin/exportin) and, simultaneously, the fluorescence derived from the RFP fragments. C, for quantification of the subcellular distribution of transport receptors in B, cells were divided into three (NϾC, NϭC, NϽC; importin ␣, ␤, 5, 7, 9, and 13) or two (with or without cytoplasmic staining; CRM1, transportin) categories. At least 100 cells co-expressing RFP-cNLS (R) or RFP-Nup214 (aa 1861-2090)-cNLS (N) were analyzed. expressing RFP-cNLS or RFP-Nup214-cNLS (Fig. 2, C and E). After induction of nuclear import with dexamethasone, the reporter proteins accumulated in the nucleus in cells expressing RFP-cNLS, indicating efficient importin ␣/␤or transportin-dependent nuclear import. Co-expression of RFP-Nup214-cNLS did not affect the dexamethasone-induced import of the reporter proteins, indicating that the Nup214 fragment did not inhibit nuclear import. We also examined the zinc finger protein EZI, which is imported into the nucleus by importin 7 (57). In cells expressing either RFP-cNLS or RFP-Nup214-cNLS (Fig. 2D), Myc-EZI was detected in the nucleus, demonstrating functional importin 7-dependent nuclear import in the presence of the C-terminal Nup214 fragment.
To further exclude the possibility that a C-terminal Nup214 fragment might compromise the general functionality of the NPC, we used an established in vitro assay that allows a simultaneous quantification of import and export efficiencies in individual cells (Fig. 2F). Digitonin-permeabilized cells expressing the export substrate GFP-NFAT (nuclear factor of activated T-cells) were incubated with BSA-Cy5-cNLS as an import substrate and increasing amounts of a His-tagged Nup214 fragment (aa 1916 -2033). After the reaction, transport efficiencies were analyzed by flow cytometry. Without the Nup214 fragment, the residual nuclear fluorescence of GFP-NFAT was very low, indicating efficient nuclear export of the reporter protein under these conditions. With increasing concentrations of the Nup214 fragment, nuclear fluorescence of GFP-NFAT increased, demonstrating inhibition of nuclear export (Fig. 2F,  left panel). Nuclear import of BSA-Cy5-cNLS by contrast, was not affected by the Nup fragment (Fig. 2F, right panel).
Taken together the data reveal that a C-terminal Nup214 fragment can interact with multiple nuclear transport receptors. Functionally, however, only CRM1-dependent nuclear export seems to be impaired by such a fragment, whereas several nuclear import pathways are not affected.
FG Motives within the C-terminal Portion of Nup214 Are Required for Its Inhibitory Function in Nuclear Export-Transport receptors like importin ␤ interact with the FG repeats of nucleoporins (3), but it has not been shown directly that Nup214 interacts via its FG repeats with CRM1 as well. To address this issue, we constructed a mutant RFP-Nup214 fragment (aa 1859 -2090), where all FGs were replaced by SGs and investigated the localization of the reporter proteins GFP-SPN1 and GFP-NC2␤ in the presence of Nup214 wild type (WT) or the mutant (SG). Consistent with the results shown in Fig. 2, the GFP-tagged proteins accumulated in cells expressing RFP-Nup214-cNLS-WT (Fig. 3, A-D). By contrast, co-expression of RFP-Nup214-cNLS-SG did not cause nuclear accumulation of export cargos. Expression of RFP fragments with the expected sizes was confirmed by Western blotting (Fig. 3E) and flow cytometry (data not shown). This result supports the notion that the FG repeats are responsible for the inhibitory effect of the C-terminal Nup214 fragments on nuclear export. Of course we cannot rule out the possibility that the mutation of a large number of critical sites compromises the biochemical properties of the protein.
We next used a biochemical approach to analyze the binding of Nup214 fragments to CRM1. Proteins were purified from bacteria and subjected to RanGAP assays to monitor the formation of export complexes. As shown before (20), RanGTP became resistant to RanGAP-induced GTP hydrolysis in the presence of CRM1 and the purified Nup214-WT fragment (Fig.  3F), indicating the formation of a trimeric complex of CRM1, RanGTP, and the nucleoporin fragment. An SG mutant of the Nup214 fragment, by contrast, had no effect in this assay, suggesting that it does not bind to CRM1. Taken together, the FG Overexpression of CRM1 Rescues the Inhibitory Effect of the C-terminal Nup214 Fragment-As described above and as reported previously (53), inhibitory Nup214 fragments can lead to an accumulation of CRM1 in the nucleus. We now analyzed endogenous CRM1 in cells expressing C-terminal Nup214 WT (aa 1861-2090) or the SG mutant (aa 1859 -2090; Fig. 4A). In control cells, CRM1 was found in the nucleus, but also a faint staining was observed in the cytoplasm. However, in cells expressing the C-terminal Nup214 WT fragment, the nuclear staining for CRM1 became brighter, indicating complete nuclear accumulation of CRM1. In cells expressing the Nup214 SG mutant, by contrast, the nuclear staining of CRM1 was comparable with that of the control cells. Thus, the FG repeats of Nup214 fragments cause sequestration of CRM1 in the nucleus, presumably leading to the inhibitory effects on nuclear export.
The concentration of endogenous CRM1 in HeLa cells is a limiting factor for nuclear export, at least for certain substrates (43). We therefore tested the possibility that CRM1 becomes rate-limiting for nuclear export of our reporter proteins upon expression of the inhibitory Nup214 fragment. GFP-SPN1 or GFP-NC2␤ were expressed in HeLa cells together with a Myctagged Nup214 fragment and with or without HA-tagged CRM1. We used expression conditions, where nuclear export of the reporter proteins was only partially inhibited by the C-terminal Nup214 fragment. Cells expressing Myc-Nup214 alone showed clear nuclear signals for GFP-SPN1 (Fig. 4, B and C) and GFP-NC2␤ (data not shown). In contrast to the strong inhibitory effect observed in Fig. 2, A and B, both reporter proteins were now distributed equally between the nucleus and the cytoplasm in many cells, indicating a nonsaturating CRM1 inhibition. Under these conditions, expression of CRM1-HA reversed the inhibitory effect of the Nup214 fragment, as GFP-SPN1 and GFP-NC2␤ (data not shown) localized to the cytoplasm in the majority of the cells. In summary, CRM1 becomes rate-limiting for nuclear export upon interaction with the FG motives of the C-terminal Nup214 fragment.
Distinct FG Motives in Nup214 Contact CRM1, Leading to Inhibition of Nuclear Export in Vivo-Our results obtained so far point to a specific role of the FG repeats of Nup214 in CRM1 binding. An alignment of various vertebrate Nup214 species (Fig. 5A) revealed a high level of amino acid conservation in the C-terminal part of the nucleoporin (aa 1859 -2090). We now used the in vivo RFP-cNLS approach for a detailed analysis of the CRM1-nucleoporin interaction and fused shortened Nup214 fragments (Fig. 5B) or FG mutants thereof (Fig. 6A) with a C-terminal cNLS to RFP (for a complete list of all fragments see supplemental Table S1). The fragments/mutants were expressed in HeLa cells, together with GFP-SPN1 (Figs. 5 and 6) or GFP-NC2␤ (supplemental Table S1 and data not shown) and the inhibition of nuclear export of the reporter proteins was compared with that seen in the presence of the whole C-terminal part of Nup214 (aa 1861-2090, fragment 1), containing 21 FG motives. No significant differences in expression levels between the RFP constructs were detected by Western blotting (Figs. 5C and 6B) or flow cytometry (data not shown). In the presence of RFP-cNLS, both reporter proteins were located mainly in the cytoplasm, indicating efficient nuclear export (Figs. 5, D and E, and 6, C and D). In the presence of Nup214 fragments comprising residues 1975-2090 (13 FGs left, fragment 3) or 1916 -2033 (12 FGs left, fragment 4), both cargos accumulated in the nucleus, similar to the observed localization in cells expressing the whole C-terminal part of Nup214. Fragments 1968 -2033 (eight FGs left, fragment 5) and 1991-2090 (10 FGs left, fragment 7) led to reduced inhibition of nuclear export. A Nup214 fragment ranging from aa 1861 to 1974 (i.e. the N-terminal portion of the inhibitory C-terminal fragment containing eight FGs) did not affect the localization of GFP-SPN1 (Fig. 5, D and E, fragment 2) and had only a minor effect on GFP-NC2␤ (data not shown). Short fragments ranging from aa 1968 -1990, 1991-2033, or 2034 -2090, containing four to six FG motives had either no or very minor effects on nuclear export of GFP-SPN1 (Fig. 5, D and E) or NC2␤ (data not shown), respectively.
These results suggest that the FG motives in the N-terminal portion of the "original" Nup214 fragment (aa 1861-2090) and at its very C terminus play no or only a minor role in the inhibitory effect on nuclear export. Fragments from the central region of this fragment, on the other hand, clearly affected the subcellular localization of both reporter proteins.
To further narrow down the region(s) in Nup214 that are important for export inhibition/CRM1 binding, we expressed fragments where sets of neighboring FG motifs were mutated to SGs (Fig. 6A). Two different sets of quadruple FG-SG mutants were introduced into the Nup214 aa 1916 -2033 fragment and both were sufficient to abrogate the inhibitory effect of the wild-type fragments on nuclear export (Fig. 6, C and D, compare fragments 4, 4a, and 4b). Likewise, the mutation of four FG motives in the aa 1968 -2033 fragment led to a loss of export inhibition (compare fragments 5 and 5a). In the context of the whole C-terminal part of Nup214 fragment (aa 1861-2090), by contrast, a reduced inhibitory effect on nuclear export was only observed when at least eight FG motives (from aa 1970 -2033; fragment 1b, 13 FGs left) were mutated to SGs (Fig. 6C and supplemental Table S1). Together, both the number and the position of the FG motives are important for the inhibitory effect of Nup214 fragments on nuclear export.
Finally, we analyzed the Nup214-CRM1 interaction at the biochemical level. It was shown before that CRM1 co-precipitated with the C-terminal Nup214 aa 1864 -2090 fragment, but not with shorter fragments from this region (51). Furthermore, this RanGTP-dependent binding was increased in the presence of an NES-cargo (20,39). In pulldown assays, we now investigated the ability of shorter Nup214 fragments (Fig. 7A) from this region to form complexes with CRM1. In a first set of experiments, GST-Ran that had been loaded with GTP or GDP was immobilized on beads and incubated with CRM1, an NESpeptide and Nup214 fragments ranging from aa 1861-1974 (eight FGs) or from aa 1916 -2033 (12 FGs). As expected, CRM1 interacted much stronger with RanGTP compared with RanGDP, indicating the formation of a trimeric export complex (Fig. 7B). Nup214 aa 1861-1974, which had no inhibitory effect on nuclear export, did not bind to the CRM1-RanGTP-NES complex. By contrast, Nup214 aa 1916 -2033, which inhibited nuclear export when expressed as an RFP fusion protein, bound to CRM1 in a RanGTP-dependent manner.
In a second set of experiments, we constructed several short GST-tagged Nup214 fragments, either in a WT or a mutant form, where four FGs were replaced by SGs (Fig. 7A), and investigated RanGTP- (Fig. 7C) and substrate (Fig. 7D)-dependent CRM1 binding. GST-Nup214 aa 1968-1990and 1991-2033 (containing four FGs each) were not able to bind to CRM1 in the presence of RanGTP and His-SPN1 as an export cargo, indicating that these regions alone are not sufficient for CRM1 FIGURE 6. Several FG motives in Nup214 participate in inhibition of nuclear export. A, schematic view of RFP-Nup214-cNLS mutants. Dots within the bars indicate the approximate location of FGs. B, Western blot analysis of the expression levels of RFP-Nup214-cNLS mutants in total cell lysate (ϳ25,000 cells per sample). Note the changes in electrophoretic mobility of the proteins upon mutating FG residues. C, HeLa cells were cotransfected with plasmids coding for GFP-SPN1 and wild-type (WT) or mutant forms of RFP-Nup214-cNLS fragments. After 20 h, the cells were fixed and analyzed by fluorescence microscopy. D, quantification of the subcellular distribution of GFP-SPN1 as in C in the presence of the indicated RFP-cNLS fragments. Bars depict the mean values of at least 100 cells from two independent experiments. Note that for fragment 5, the same quantification as in Fig. 5E is shown. interaction. However, CRM1 bound to the longer GST-Nup214 fragment (aa 1968 -2033) and binding was promoted by either RanGTP or SPN1 as an export cargo. Clearly, the GST-Nup214 mutant (aa 1968 -2033) lacking four of eight FG motives was unable to bind CRM1, demonstrating the importance of the mutated FGs for CRM1 interaction. From these results we conclude that C-terminal Nup214 fragments that bind directly to CRM1 in vitro cause its nuclear accumulation in vivo (data not shown), which finally results in inhibition of CRM1-dependent nuclear export. Several FG motives in the Nup214 fragments are involved in these effects. Furthermore, the N-terminal (aa 1861-1915) and the C-terminal part (aa 2034 -2090) of the originally identified CRM1-binding region of Nup214 (aa 1864 -2090 (51)) are not necessary for export complex formation in vitro and for inhibition of nuclear export in vivo.

DISCUSSION
We used an approach based on transport inhibition to further delineate the region in Nup214 that binds to CRM1 (53). Clearly, CRM1-dependent export of several reporter proteins was inhibited in the presence of an isolated Nup214 fragment (aa 1861-2090) containing a classical NLS to ensure nuclear localization. The fragment affected the subcellular localization of CRM1 and also that of several other NTRs. Exceptions were transportin, which was all nuclear also in the absence of the Nup fragment, and importin 5 and importin 13, which were not affected at all by the fragment, demonstrating that strong Nup214 interaction is not a general feature of all NTRs. In contrast to the clear inhibition of CRM1-dependent nuclear export by the Nup214 fragment, however, sequestering Nup214-binding importins in the nucleus did not inhibit nuclear import of several substrates (Fig. 2). This probably relates to the observation that import receptors dissociate from FG-Nups (or fragments thereof) in the presence of RanGTP, i.e. in the nucleus. Hence, the equilibrium of an importin might be shifted toward the nucleus in the presence of the Nup214 fragment. Apparently, however, the receptors are set free in the nucleus to diffuse back to the cytoplasm and to initiate pathway-specific import of the tested cargo proteins. A possible function for the interaction of importins with endogenous Nup214 at the NPC remains to be investigated. Perhaps the nucleoporin supports import of a selected set of substrates in a receptor-dependent manner, similar to Nup358 (18). For the export receptor CRM1, by contrast, the situation is different. Here the transport receptor gets trapped in the nucleus in a complex with RanGTP, any available export cargo, and the Nup fragment. The latter seems to occupy binding sites on CRM1 that are required for FG-Nup interaction and, therefore, for efficient translocation of export complexes through the NPC. As a result, CRM1 gets trapped in the nucleus and becomes rate-limiting for nuclear export, an interpretation that is supported by our CRM1 overexpression experiments (Fig. 4).
What are the characteristics of the CRM1-FG-Nup interaction? Previous studies suggested that binding of the export receptor to Nup214 and other nucleoporins occurs preferentially in the context of a trimeric export complex containing an export cargo and RanGTP. A crystal structure of un-bound CRM1 is not available yet, but structural changes that are induced on CRM1 upon complex formation are likely to expose binding sites for FG-Nups. This is corroborated by the observation that RanGTP, together with an export substrate promotes cooperative binding of CRM1 to Nup214 fragments (Fig.  7). With respect to potential binding sites in Nup214, our data clearly show that several FG motives in the nucleoporin contribute to high affinity interaction. Of particular interest is a region between amino acids 1968 and 2033 of the human Nup214 sequence that is conserved between species. The corresponding fragment was the shortest one that showed clear inhibition of export of GFP-SPN1 and that also interacted with CRM1 in a RanGTP-and cargo-specific manner. In a previous in vivo study, similar fragments (aa 1864 -2052 and aa 1957-2090) did not interact with CRM1 in co-precipitation experiments (51). These discrepancies may result from the very different experimental approach in the two studies. The mutation of four FG motives to SGs in the 1968 -2033 fragment abrogated the CRM1 interaction and, as a result, inhibition of nuclear export. The corresponding mutants of the longer (original) fragment (1861-2090), by contrast, retained their ability to inhibit export. In this context, mutation of eight or more FG motives were required to abrogate the inhibitory effect. These results suggest that CRM1 makes several contacts to Nup214 and all of them are likely to involve the FG motives of the nucleoporin. This is reminiscent of the importin ␤-nucleoporin interaction, where several regions in the import receptor were identified to contact FG repeat peptides. Such NTR-Nup contact sites were initially established using classical biochemical methods (36) and x-ray crystallography (3,50) and were later corroborated by molecular dynamics simulations (37). Using this method, Schulten and co-workers (58) also predicted a number of binding spots in the related transport receptor Cse1p. Together, multiple binding sites, which on their own are likely to result in low-affinity binding, could lead to a strong interaction between the nucleoporin and the NTR. In the central channel of the NPC, where FG-Nups are thought to create the functional permeability barrier, such a high-avidity interaction is not desired, because it could slow down translocation of the transport complex. Nup214, however, is a nucleoporin that localizes to the cytoplasmic periphery of the NPC (59) and is probably not a crucial component of the permeability barrier (14,20,60). The NTR-Nup interaction investigated in this study is, therefore, particularly relevant for CRM1-mediated export but not for other nucleocytoplasmic transport pathways. Previously, we measured affinities/avidities between CRM1 and Nup214 fragments in the low nanomolar range (20), allowing co-precipitation of the two proteins under appropriate conditions (39,51). In the complex milieu of a cell, such interactions may be reduced by competing factors (61). Nevertheless, our results show that even short regions of Nup214 can engage in an inhibitory interaction with CRM1 in vivo, which could be relevant, for example, in the context of oncogenic fusions of various proteins with the C-terminal part of Nup214 (62). Indeed, SET-CAN, a fusion protein where the template activation factor SET is fused to the C terminus of Nup214, occurs in certain forms of acute myeloid leukemia and was shown to inhibit CRM1-dependent nuclear export (63). FEBRUARY 8, 2013 • VOLUME 288 • NUMBER 6 JOURNAL OF BIOLOGICAL CHEMISTRY 3961