Molecular Basis for the Recognition of Snurportin 1 by Importin β*

The nuclear import of uridine-rich ribonucleoproteins is mediated by the transport adaptor snurportin 1 (SNP1). Similar to importin α, SNP1 uses an N-terminal importin β binding (sIBB) domain to recruit the receptor importin β and gain access to the nucleus. In this study, we demonstrate that the sIBB domain has a bipartite nature, which contains two distinct binding determinants for importin β. The first determinant spans residues 25-65 and includes the previously identified importin α IBB (αIBB) region of homology. The second binding determinant encompasses residues 1-24 and resembles region 1011-1035 of the nucleoporin 153 (Nup153). The two binding determinants synergize within the sIBB domain to confer a low nanomolar binding affinity for importin β (Kd ∼ 2 nm) in an interaction that, in vitro, is displaced by RanGTP. We propose that in vivo the synergy of Nup153 and nuclear RanGTP promotes translocation of uridine-rich ribonucleoproteins into the nucleus.

The nuclear import of uridine-rich ribonucleoproteins is mediated by the transport adaptor snurportin 1 (SNP1). Similar to importin ␣, SNP1 uses an N-terminal importin ␤ binding (sIBB) domain to recruit the receptor importin ␤ and gain access to the nucleus. In this study, we demonstrate that the sIBB domain has a bipartite nature, which contains two distinct binding determinants for importin ␤. The first determinant spans residues 25-65 and includes the previously identified importin ␣ IBB (␣IBB) region of homology. The second binding determinant encompasses residues 1-24 and resembles region 1011-1035 of the nucleoporin 153 (Nup153). The two binding determinants synergize within the sIBB domain to confer a low nanomolar binding affinity for importin ␤ (K d ϳ 2 nM) in an interaction that, in vitro, is displaced by RanGTP. We propose that in vivo the synergy of Nup153 and nuclear RanGTP promotes translocation of uridine-rich ribonucleoproteins into the nucleus.
Nuclear import of proteins and nucleic acids is an active signal-mediated process that requires, in most cases, soluble transport factors and GTP hydrolysis by the GTPase Ran (1)(2)(3)(4). Nuclear transport factors are sorted into two categories: importins and exportins (also known as karyopherins), which include 14 members in budding yeast and at least 20 in humans (5,6). All karyopherins fold into superhelical solenoids, which present an external convex surface involved in nucleoporin binding and a concave internal face that interacts with transport cargos and RanGTP. The two best characterized importins, human importin ␤1 and yeast karyopherin ␤2, have been visualized using crystallographic methods bound to specific import cargos (7)(8)(9)(10) and RanGTP (11,12). These structures have provided invaluable information on the structural flexibility of karyopherins and their ability to undergo conformational changes during the import process (13)(14)(15)(16).
The best understood nuclear import pathway involves cargos bearing a classical SV40-like nuclear localization sequence (PKKKRKV) (17). These cargos are imported into the nucleus by the adaptor importin ␣ and the receptor importin ␤ with the aid of RanGTP. GTP hydrolysis by Ran energizes the import reaction by promoting both release of the import complex from nucleoporins as well as unloading of import cargos into the nucleus (1)(2)(3). In addition to classical nuclear localization sequence cargos, uridine-rich ribonucleoproteins (U snRNPs) 2 represent an important class of import cargos. Mature U snRNP particles are assembled in the cytoplasm and imported into the nucleus in at least two distinct pathways, both dependent on importin ␤ (18). In the first pathway, the import signal is in the proteinaceous core of the U snRNP formed by the Sm proteins (18 -21). In the second pathway, the trimethylated guanosine cap of the mature U snRNP is recognized by the adaptor snurportin 1 (SNP1) (18 -21), which, in turn, recruits importin ␤. SNP1 consists of an N-terminal IBB domain (sIBB) similar to that found in importin ␣ (␣IBB) and a large C-terminal trimethylated guanosine cap-binding region that resembles the GTP-binding domain of mRNA-guanylyltransferase (22). In permeabilized cells, SNP1 and importin ␤ promote Ran-and energy-independent nuclear import of at least two specific spliceosomal U snRNPs, namely U1 and U5 (23). After nuclear entry, SNP1 is recycled back to the cytoplasm by Crm1 in complex with RanGTP (24). In this study, we have used a combination of crystallographic, biochemical, and biophysical techniques to define the composition and recognition of the sIBB domain by importin ␤.
Crystallization and Structure Determination-Importin ␤ bound to a chemically synthesized sIBB-(25-65) peptide was crystallized under 20% polyethylene glycol 8000, 50 mM sodium chloride, at pH 6.0. Crystals of the importin ␤-sIBB-  complex diffracted weakly to 5Å resolution. Prolonged dehydration of crystals by soaking in 38% polyethylene glycol 8000 dramatically improved the diffraction to 2.35 Å resolution. Approximately 50 crystals were soaked for various times and screened at beamline X6A at the Brookhaven National Light Source (BNLS) on a Quantum Q4 CCD detector. Diffraction data were reduced to intensities using the programs DENZO and SCALEPACK (28) (see Table 1). Two distinct crystal forms were obtained, both in space group P2 1 . Crystal form I (see ds2.35 in Table 1) contains one importin ␤-sIBB-(25-65) complex in the asymmetric unit. Crystal form II (see ds3.2 in Table  1), which diffracted weakly to 3.2 Å resolution, contains two importin ␤-sIBB-  complexes in different conformations per asymmetric unit. Both structures were solved by molecular replacement in MolRep (29) using human importin ␤ (Protein Data Bank (PDB) code 1QGK) as a search model. The initial solution was refined in CNS (30) using rigid body refinement, simulated annealing, and grouped B-factor refinement to an R free ϳ 35% (calculated using 10% of the observed reflections). The sIBB-  domain was built in F o Ϫ F c electron density difference maps using the program Coot (31). After several rounds of manual building alternated with positional and B-factor refinement, the final model for crystal form I has a R work and R free of 22.7 and 25.0%, respectively. For crystal form II, the final model was refined to a R work of 30.3% and R free of 32.8%, using all reflections between 40 and 3.2 Å resolution and includes a well defined closed importin ␤-sIBB domain complex (complex A) and an open conformation of the complex (complex B). All structural figures were made using PyMOL (32).
Native Gel Electrophoresis and Pull-down Assay-Native binding assay on agarose gel was carried out as described in Ref. 33. In the assay, 25 g of importin ␤, a gel filtration-purified complex of importin ␤ bound to either sIBB-(1-65) or sIBB-(25-65) peptides, were separated on a 1.5% agarose gel at room temperature for 45 min. For the Ran displacement, gel filtrationpurified importin ␤-sIBB-(1-65) or sIBB-(25-65) complexes were incubated with 0.25-5-fold molar excess of RanGppNHp and incubated for 30 min on ice prior to electrophoretic separation on agarose gel. The pull-down assay was performed on glutathione-agarose beads (Amersham Biosciences) coupled to GST-sIBB- . In the assay, 25 l of GST-sIBB-(1-65) beads were incubated with 16 g of importin ␤ for 15 min at room temperature. For Ran displacement, an excess of either RanGppNHp or RanGDP was added to the beads after the addition of importin ␤. The mixtures were incubated overnight at 4°C followed by extensive washing in pull-down buffer (20 mM HEPES, pH 7.4, 150 mM sodium chloride, 3 mM ␤-mercaptoethanol, 0.005% Tween 20). After washing, all samples were dissolved in SDS-loading buffer and analyzed on SDS-PAGE.
Isothermal Titration Calorimetry (ITC)-ITC experiments were carried out at 30°C in a ITC calorimeter (Microcal). sn-(1-24) dissolved in ␤-buffer (10 mM HEPES, pH 7.4, 150 mM sodium chloride, 3 mM EDTA, 3 mM ␤-mercaptoethanol) at a concentration of 450 M was injected in 9-l increments into the calorimetric cell containing 1.8 ml of importin ␤ in ␤-buffer at a concentration of 60 M. The spacing between injections was 360 s. Titration data were analyzed using the Origin 7.0 data analysis software (Microcal Software, Northampton, MA). Injections were integrated following manual adjustment of the baselines. Heats of dilution were determined from control experiments with the ␤-buffer and subtracted prior to curve fitting using a single set of binding sites model. The curve fitting yields a K d ϭ ϳ30 Ϯ 14 M and ⌬H ϭ 6948 Ϯ 1491 cal/mol at 30°C for titration of the sn-(1-24) peptide into full-length importin ␤ and a K d ϭ ϳ30 Ϯ 15 M and ⌬H ϭ 6548 Ϯ 1191 cal/mol for titration of the sn-(1-24) peptide into importin ␤ (residues 1-445).

RESULTS
Snurportin IBB Domain Is Bipartite-Nuclear import SNP1 is mediated by the N-terminal sIBB domain (residues 1-65) (19). This region alone is necessary and sufficient to promote nuclear import in the absence of Ran and energy in a permeabilized cell assay (23). Analysis of the primary sequence of the sIBB domain suggests the presence of a bipartite signal (Fig. 1a). Region 25-65 of the sIBB domain closely resembles the classical IBB domain of importin ␣ (␣IBB), which promotes Ran/ energy-dependent nuclear import (23,34). Similar to the ␣IBB domain (residues 11-54), all basic residues critical for importin ␤ binding are conserved in the sIBB-(25-65) domain (supplemental Fig. 1). The latter is, however, shorter due to a 3-aminoacid gap between residues 46 and 47 (Fig. 1a). The region 25-65 of the sIBB domain will be defined as the ␣IBB region of homology. In contrast to importin ␣s, which show poor sequence conservation upstream of the IBB domain, SNP1 residues 1-24, sn-(1-24), are well conserved in snurportins (supplemental Fig.  1). Blast analysis reveals that sn-(1-24) shares 42% sequence identity and 79% similarity to a region of Nup153 spanning residues 1011-1035. This same region of Nup153 binds importin ␤ with high affinity, likely via interactions with phenylalanine-glycine (FXFG) repeats (25).
Structure of Importin ␤ Bound to the sIBB-  Domain-To shed light on the interaction of SNP1 with importin ␤, we crystallized human importin ␤ bound to a chemically synthesized peptide spanning residues 25-65 of SNP1. Although crys-tals of the importin ␤-sIBB-(25-65) complex were obtained under the same conditions used for crystallization with the ␣IBB-(11-54) peptide (8), crystals diffracted weakly to 5 Å resolution. Aiming at high resolution diffraction studies, we subjected the importin ␤-sIBB-(25-65) complex crystals to dehydration in highly concentrated solutions of polyethylene glycol 8000. Diffraction analysis confirmed a dramatic improvement in the diffraction quality and resolution limit as a function of the time of dehydration. Dehydration for 3-12 h yielded high quality diffraction past 2.3 Å resolution. This crystal form, referred to as crystal form I, displayed a small primitive monoclinic unit cell (   long C-terminal helix (residues 41-65), connected to an N-terminal 3 10 helix (residues 27-30) by a 7-residue spoon-shaped linker (Fig. 1b). The overall B-factor for the peptide is ϳ65 Å 2 , slightly higher than that for the importin ␤ (ϳ45 Å 2 ). Crystal form II was obtained upon short dehydration (ϳ30 min) of importin ␤-sIBB-(25-65) crystals and diffracted to ϳ3.2 Å resolution. Interestingly, this second crystal form contains two importin ␤-sIBB-(25-65) complexes in different conformations per asymmetric unit (Fig. 2a). In one complex (referred to as complex A), importin ␤ adopts a closed conformation nearly identical to crystal form I. In the second complex (complex B), importin ␤ is distinctly open, with C-terminal HEAT repeats 12-19 swung up to 20 Å away from the corresponding position in the closed conformation (Fig. 2b). The tertiary structure of the importin ␤ molecule in complex B resembles Kap95p, the yeast homologue of importin ␤, in complex with RanGTP (11). This open conformation of importin ␤ likely represents the structure adopted by the protein in the process to unload the import cargo. Accordingly, the electron density for the sIBB-(25-65) domain is less defined in the complex B, where only the C-terminal helix is visible.
sIBB-  versus ␣IBB Domain-The structure of importin ␤ bound to the ␣IBB-(11-54) domain was previously deter-mined to a comparable resolution of 2.3 Å (8), allowing for direct comparison with the crystal form I described in this study. Despite the profound functional differences between the sIBB and ␣IBB domain, the conformation adopted by importin ␤ in the two complexes is virtually identical (r.m.s. deviation 0.965 Å) (Fig. 3a). It is noteworthy that significant differences exist in the structure of the two IBB domains and in the way importin ␤ positions them inside the cargo-binding surface (Fig. 3a). First, the sIBB-(25-65) domain is slightly shorter than the ␣IBB domain (35 versus 41 Å), which is consistent with the 3-amino-acid gap in the C-terminal helix between residues 46 and 47 (Fig. 1a). Second, the sIBB C-terminal helix is ϳ5°tilted with respect to the ␣IBB helical axis, which positions it closer to the importin ␤ concave surface (Fig. 3a). Third, in the sIBB-  domain, the linker between the C-terminal helix and the N-terminal 3 10 helix is significantly longer than in the ␣IBB domain (7 versus 3 amino acids), which suggests in the sIBB domain that these two helices have a much higher degree of flexibility. This hypothesis is corroborated by the structural plasticity seen in the 3.2 Å crystal form II, where in the more open complex B (Fig. 2a) only the C-terminal helix of the sIBB domain is visible, whereas the N-terminal 3 10 helix is likely flexible and thus disordered in the crystal structure.
As in the recognition of the ␣IBB-(11-54) domain, importin ␤ makes two distinct sets of contacts with the sIBB domain. The first includes the N-terminal moiety of the sIBB peptide (residues 25-40), which interacts with HEAT repeats 7-11 (Fig. 3b). This region is critical for binding of SNP1 to importin ␤. Substitution of Arg 27 to Ala decreases the affinity of SNP1 for importin ␤ by 20-fold and blocks nuclear import in digitoninpermeabilized cells (35). The second region of interaction involves the sIBB helix (residues 41-65), which presents an extended network of contacts with HEAT repeats 12-19 of importin ␤ (Fig. 3b). In the sIBB domain, this region engages in fewer contacts, 12 versus 20, as compared with the ␣IBB helix (residues 26 -54). Likewise, the polybasic stretch 27 Arg-Arg-Arg-Arg 31 of the ␣IBB domain is shifted in the sIBB helix where only 2 of the 4 arginines contact importin ␤ directly (Fig. 3b).
SNP1 Region  Contains a Binding Determinant for Importin ␤-We next sought to determine whether the Nup153-like moiety of SNP1, sn-(1-24), binds importin ␤ directly. The interaction of the sn-(1-24) with importin ␤ was probed using three independent binding techniques. First, by native electrophoresis on agarose gel, a chemically synthesized sn-(1-24) peptide was found to shift the migration of importin ␤, which is suggestive of a direct physical interaction between the peptide and the protein (Fig. 4a). Second, using ITC, titration of sn-(1-24) in a calorimetric cell containing importin ␤ at 30°C yielded an endothermic binding reaction that is adequately described by a single exponential decay binding model (Fig. 4b). The thermodynamic parameters obtained from the curve fit indicated that sn-(1-24) binds importin ␤ with a K d ϳ 30 Ϯ 14 M and an enthalpy of complex formation of ⌬H ϭ 6948 Ϯ 1491 cal/mol. Interestingly, identical heat release (and thus K d ) was observed using a fragment of importin ␤ comprising HEAT repeats 1-10 (residues 1-445) (Fig. 4c). These evidences lend support to the existence of a single binding site for the sn-  in the N-terminal, RanGTP- (11,27), and nucleo- porin-binding (36) region of importin ␤. However, based on the ITC data, we cannot rule out the possibility that the sn-(1-24) binds with lower or similar affinity to the C terminus of importin ␤ (HEAT [11][12][13][14][15][16][17][18][19] or that the binding of the sn-  to the N-terminal HEAT repeats 1-10 excludes or drastically decrease its binding to the C terminus of the protein. Third, using SPR, we found that the sIBB-(1-65) domain has a 7-fold higher binding affinity for importin ␤ than the shorter sIBB-(25-65) domain lacking the sn-(1-24) (K d ϳ 2.0 Ϯ 0.5 nM versus 15.2 Ϯ 5.8 nM) (Fig. 5). The affinity measured by SPR for the full-length sIBB-(1-65) domain agrees well with the K d values measured by FRET (K d ϳ 2.0 nM) (37) and solid phase binding assay (K d ϳ 4.7 nM) (35), which makes us confident in the accuracy of this technique in measuring the interactions of importin ␤ with the SNP1. In addition to a significant drop in the K d , the sIBB-(25-65) domain had a 4-fold faster rate of dissociation from importin ␤ as compared with the sIBB-(1-65) domain (k off ϭ 8.24 e 4 Ϯ 0.5 e 4 s Ϫ1 versus k off ϭ 2.07 e 4 Ϯ 0.3 e 4 s Ϫ1 ) (Fig.  5), whose off rate from importin ␤ was comparable with that of importin ␣ (38).

DISCUSSION
To determine the molecular basis for the sIBB-(1-65) karyopheric properties, we have dissected the structure of the sIBB-(1-65) domain and its binding interactions with importin ␤. The main conclusion of our work indicates that the sIBB-(1-65) domain has a bipartite organization, which consists of two moieties, both interacting with importin ␤. The first binding determinant of SNP1 spans residues 25-65 and includes an ␣IBB region of homology, which binds importin ␤ with nanomolar affinity (K d ϳ 15 nM). As shown in the crystal structure of the importin ␤ bound to the sIBB-  domain, this region of SNP1 interacts with HEAT repeats 7-19 of importin ␤ similar to an ␣IBB domain and not only with the N terminus of the protein (residues 1-618), as reported previously (20). Within the sIBB-  domain, the C-terminal helix (residues 41-65) makes fewer and weaker contacts with importin ␤ than the N-terminal residues 25-40 of SNP1. This observation partially explains why an N-terminal fragment of importin ␤ (residues 1-618) shows relatively high binding affinity for the sIBB domain (20,35), but it fails to bind the ␣IBB domain (39). Likewise, the paucity of contacts between the sIBB helix (residues 41-65) and importin ␤ may explain the fast rate of dissociation measured by SPR between importin ␤ and the sIBB-  domain as compared with its counterpart, the ␣IBB domain (38).
In addition to the ␣IBB region of homology, SNP1 presents a novel binding determinant for importin ␤ between residues 1 and 24. This region interacts weakly with importin ␤ (K d ϳ 30 M) when isolated from SNP1 but synergizes with the sIBB-(25-65) moiety in the context of the sIBB-(1-65) domain to confer low nanomolar binding affinity and a slow rate of dissociation from importin ␤. Therefore, we propose that the sn-(1-24) acts as a molecular hook to retain SNP1 bound to importin ␤. Interestingly, sn-(1-24) shares high sequence identity to the region 1011-1035 of Nup153, which is localized at the nuclear basket (40) and is known to bind importin ␤ with nanomolar affinity (25). This prompts the intriguing idea of a direct displacement of the sn-(1-24) from importin ␤ within the snRNP-SNP1-importin ␤ import complex at the nuclear basket. Notably, this scenario is conceptually analogous to what was reported for the yeast nucleoporin Nup2p, which is also associated with the nuclear basket (41,42). Residues 1-51 of Nup2p bind tightly to Kap60p, the yeast homologue of importin ␣, and thus accelerate the release of nuclear localization sequence cargos from Kap60p. In addition, the region 1-24 of SNP1 was also shown to be critical for binding to Crm1, which, in complex with RanGTP, recycles SNP1 back to the cytoplasm (24).
How do our data fit with the observation that in digitonin-permeabilized cells, the sIBB domain promotes Ran-and energy-independent nuclear import (23)? The structural and biochemical work presented in this study challenges the idea of a simple Ran-independent nuclear import pathway in at least three ways. First, at the three-dimensional level, the conformation of importin ␤ bound to the sIBB domain is very similar to that adopted in complex with the ␣IBB domain, which requires RanGTP to undergo a dramatic conformational change necessary to release the import cargo (11). Second, the sIBB domain binds importin ␤ with low nanomolar binding affinity, which is incompatible with a Ran-independent cargo release into the nucleus. Third, in vitro, RanGTP, and not RanGDP, specifically disrupts the importin ␤-sIBB complex. To complement these data, we provide evidences that region 1-24 of SNP1 contains a Nup153-like motif that modulates both the strength of interaction and the off rate of the sIBB domain from importin ␤. We propose that the sn-(1-24) plays an important functional role during the import reaction by reducing the avidity of the NPC for the sIBB-import complex. This could be explained, for instance, by a simple intermolecular contact between the SNP1 moiety sn-(1-24), which contain a 12 FSVS 15 repeat, and the major nucleoporin-binding site in importin ␤ (25). Such an interaction would reduce the avidity of the sIBB- import complex for nucleoporins and possibly allow translocation of the import complex through the NPC in permeabilized cells, in the absence of exogenous RanGTP. The idea of a reduced avidity of the NPC for the sIBB-import complex agrees well with the observation that a chimera of the sIBB-(1-65) fused to the ␤-galactosidase, sIBB-␤-galactosidase, remains associated for less time to the nuclear basket than an ␣IBB-␤galactosidase, suggesting a reduced stalling of the sIBB-import complex inside the NPC (20).
However, if the reduced avidity of the sIBB-import complex for the NPC may be sufficient to explain why in permeabilized cells the sIBB domain translocates efficiently without the addition of exogenous Ran, it does not explain in an in vivo setting what energizes the active release (and thus nuclear import) of sIBB-import cargos into the cell nucleus. In agreement with the observation that in vitro the sIBB-importin ␤ interaction is specifically disrupted by RanGTP, we propose that GTP hydrolysis by the small GTPase Ran remains the driving force of the dissociation of SNP1 from importin ␤. Further studies are needed to better understand this important import pathway. In particular, it will be necessary to develop a functional assay that allows discrimination between nuclear passage of the sIBB-import complex and dissociation from importin ␤, and that defines the energetic requirement of these two processes.