br Results br Discussion Unraveling the molecular
Discussion Unraveling the molecular details of nucleoporin-karyopherin interactions, which have to be strong enough to promote transport but sufficiently weak to avoid stalling of transport complexes within the pore, is key to our understanding of the mechanisms of nucleocytoplasmic transport. The interactions of CRM1 and Nup214 can serve as a paradigm for karyopherin-nucleoporin interactions in general. In this study, we solved the structure of a CRM1 export complex bound to a 117-amino-acid fragment of Nup214 at multiple FG motifs. We identified three FG regions in the Nup214 fragment containing a total of seven characteristic FG motifs and a similar FS motif (F1–F8). The location of FG-binding patches 2 and 3 on the N-terminal and C-terminal arches of CRM1, respectively, is consistent with a recently reported crystal structure (PDB: 3WYF) of yeast CRM1 (Xpo1p) bound to RanGTP (Gsp1p) and RanBP3 (Yrb2p; Figure 7). RanBP3, however, contains only five FG motifs in two FG regions, and binding to CRM1 occurs mainly via its Ran-binding domain. Almost all of the phenylalanine residues in the Nup214 fragment bind to corresponding mdm2 p53 pockets in CRM1 (P1–P8). In order to insert these large side chains between two HEAT helices and form intricate interactions with hydrophobic residues between them, a kink in the main chain of the nucleoporin is required. To enable such a large (∼180°) bending of the peptide chain, the phenylalanine is followed by an adjacent glycine residue forming a sharp β turn, resulting in hydrogen bonding between the main chain carbonyl of residue n and the main chain amine of residue n + 3. Similar arrangements were previously described for a fragment of importin β binding to an FxFG motif (Bayliss et al., 2000) and for the CRM1-RanBP3 interaction (Koyama et al., 2014). Strikingly, FG-binding pockets from the N- and C-terminal region of CRM1 bind to FG motifs on Nup214 that are separated by a rather short stretch of amino acids. This is possible because, in the context of an export complex, the CRM1 termini are in close proximity, allowing simultaneous binding of the extended Nup214 fragment to the FG-binding pockets in CRM1. CRM1 can also be held together in a closed conformation by the Nup214 fragment in the presence of RanGTP but in the absence of an export substrate (Figure 1A). Thus, Nup214 functions as a molecular clamp, leading to stabilization of the export complex. This function, which requires binding of the nucleoporin to multiple sites of the export receptor, becomes obvious when we compare the effects of mutations within the Nup214 sequence in different assays. The Nup214-X2 mutant, for example, where three phenylalanine residues are changed to serines, still bound the CRM1 export complex in pull-down experiments, similar to the wild-type protein (Figure 4B). However, in assays where we monitored the ability of the nucleoporin fragment to protect CRM1-bound RanGTP from RanGAP-induced GTP hydrolysis, the Nup214-X2 mutant, which according to the structure should only bind to the C-terminal arch of CRM1, showed only a very weak effect compared to the wild-type fragment (Figure 4C). Similar observations were made for the Nup214-X1 mutant, which should only bind to the N-terminal region of CRM1. Thus, simple binding is not sufficient to protect CRM1 from GTP hydrolysis. This effect rather requires the clamp function of Nup214 with simultaneous binding to both ends of the export receptor. Cooperative binding of all four components of the CRM1-RanGTP-cargo-Nup214 complex is further enhanced by subtle changes in FG pockets of CRM1 upon formation of the trimeric export complexes that initially lack the nucleoporin. For export substrate-containing complexes, the stabilizing effect of FG nucleoporins distinct from Nup214 should prevent premature loss of the cargo during transit. The NES-binding site in CRM1 is the most-conserved part of the export receptor (Monecke et al., 2014). Our analysis reveals that regions containing several of the FG-binding pockets in CRM1 are also conserved among species (Figure 6A). Interestingly, mutations in these regions affected binding of RanGTP and/or an NES substrate underlining the allosteric nature of CRM1 and suggesting that the overall CRM1 structure is extremely sensitive with respect to amino acid changes. Thus, for all functional assays in intact cells or in permeabilized systems, possible side effects of even single point mutations in transport receptors must be considered. In light of our observation that binding of CRM1 to Nup214 and other nucleoporins (Nup62) or nucleoporin-like proteins (RanBP3) is mutually exclusive, we conclude that the CRM1 sequence has been optimized during evolution to interact via similar mechanisms with a multitude of FG-containing proteins, as they are encountered during passage of the NPC—without compromising the ability of CRM1 to bind its primary partners, RanGTP and NES cargo. Interestingly, we observed that the FG pocket for Phe1922Nup214 (P1) partially overlaps with the binding site for the C-terminal 12 residues of SPN1, representing its third CRM1-binding epitope in the ternary export complex structure (PDB: 3GJX; Figure S3). It has previously been reported that a truncated version of SPN1, lacking these C-terminal residues, binds CRM1 with a 60% lower affinity (Paraskeva et al., 1999). Due to the cooperative binding of SPN1 and Nup214 to CRM1, it is difficult to distinguish between the respective contributions of SPN1 and Nup214 to complex stability. However, the electron density for Nup214 in this region was significantly weaker when crystals with full-length SPN1 and the same Nup214 fragment were used for structure determination (data not shown). This could indicate a rather dynamic and/or mutual exclusive binding of Nup214 and the SPN1 C terminus. Thus, the overlapping binding sites might function in the release of the export complex, as binding of Nup214 to that site probably lowers the affinity of SPN1 to CRM1.