Structure of Importin-α from a Filamentous Fungus in Complex with a Classical Nuclear Localization Signal

Neurospora crassa is a filamentous fungus that has been extensively studied as a model organism for eukaryotic biology, providing fundamental insights into cellular processes such as cell signaling, growth and differentiation. To advance in the study of this multicellular organism, an understanding of the specific mechanisms for protein transport into the cell nucleus is essential. Importin-α (Imp-α) is the receptor for cargo proteins that contain specific nuclear localization signals (NLSs) that play a key role in the classical nuclear import pathway. Structures of Imp-α from different organisms (yeast, rice, mouse, and human) have been determined, revealing that this receptor possesses a conserved structural scaffold. However, recent studies have demonstrated that the Impα mechanism of action may vary significantly for different organisms or for different isoforms from the same organism. Therefore, structural, functional, and biophysical characterization of different Impα proteins is necessary to understand the selectivity of nuclear transport. Here, we determined the first crystal structure of an Impα from a filamentous fungus which is also the highest resolution Impα structure already solved to date (1.75 Å). In addition, we performed calorimetric analysis to determine the affinity and thermodynamic parameters of the interaction between Imp-α and the classical SV40 NLS peptide. The comparison of these data with previous studies on Impα proteins led us to demonstrate that N. crassa Imp-α possess specific features that are distinct from mammalian Imp-α but exhibit important similarities to rice Imp-α, particularly at the minor NLS binding site.


Introduction
The filamentous fungus Neurospora crassa has been studied by classical and molecular genetics, providing several insights into cellular processes, which include cell signaling, growth and differentiation, secondary metabolism, circadian rhythm and genome defense [1]. Together by dialysis. NcImpα was stored under cryogenic temperatures in a buffer composed of 20 mM Tris-HCl, pH 8, and 100 mM NaCl.

Synthesis of NLS peptides
The peptide corresponding to SV40 NLS ( 125 PPKKKRKV 132 ) was synthesized by Proteimax (Brazil) with a purity of higher than 99%. The peptides contained additional residues at the Nand C-termini compared with the minimally identified NLS [18].

Isothermal titration calorimetry
ITC measurements were performed using a MicroCal iTC200 microcalorimeter (GE Healthcare) calibrated according to the manufacturer 0 s instructions. NcImpα and the SV40 NLS peptide were prepared and dialyzed in buffer (20 mM Tris-HCl, pH 8.0, and 100 mM NaCl). The sample cell was loaded with 50 μM NcImpα that was titrated with the SV40 NLS peptide at a concentration of 1 mM (protein:peptide molar ratio of 1:20). Titrations were conducted at 10°C and consisted of 20 injections of 2.0 μL in an interval of 240 s with a 1000 rpm homogenization speed. The heat of dilution was determined by titration of the peptide sample into the protein sample buffer (20mM Tris-HCl, pH 8.0, and 100 mM NaCl) in separate control assays and was subtracted from the corresponding titrations. The assays temperature was chosen to avoid protein aggregation displayed for higher temperatures and to permit direct comparison to previously Impα/SV40NLS ITC studies performed in the same condition [19]. The data were processed using MicroCal Origin Software to obtain values for stoichiometry (N), dissociation constants (K d ), enthalpy (ΔH), and binding-type input parameters were adjusted to obtain the best fitting model. The values of K d and ΔH were used to calculate free energy (ΔG) and entropy (ΔS) values.
Crystallization, X-ray data collection and structure determination NcImpα was concentrated to 12 mg/ml using an Amicon 30 kDa cutoff filter unit (Millipore) and stored at -20°C. Crystals of NcImpα in complex with SV40 NLS were obtained at a 1:8 protein:peptide molar ratio in 20 mM Bicine, pH 8.5, and 20% (w/v) polyethylene glycol 6000 at 4°C [20] using MRC2 Well Crystallization Plates and an Orix4 system (Douglas Instruments).
X-ray diffraction data were collected from a single crystal of NcImpα/SV40NLS at a wavelength of 1.0 Å using a synchrotron radiation source (X25 beamline, National Synchrotron Light Source, NSLS, Upton, NY, USA) and a PILATUS detector. A crystal was mounted in a nylon loop and flash-cooled in a stream of nitrogen at -173.15°C using 20% (v/v) glycerol as the cryoprotectant. The crystal-to-detector distance was 270 mm with an oscillation range of 0.5°, resulting in the collection of a total of 720 images. The data were processed using the HKL2000 suite [21]. The crystal belonged to the space group P2 1 2 1 2 1 (Table 1) and was isomorphous to previously obtained crystals [20].
NcImpα/SV40NLS crystal structure was determined by molecular replacement using the program Phaser [22] and the coordinates of Impα from Mus musculus in complex with the nucleoplasmin NLS (PDB ID: 3UL1, chain B; [15]) as the search model. Rounds of manual modeling were performed using the program Coot [23] and the crystallographic refinement (positional and restrained isotropic individual B factors with an overall anisotropic temperature factor and bulk-solvent correction) was performed using the program phenix.refine [24], considering free R factors. Structure quality was evaluated using the program MolProbity [25], and interactions were analyzed using the program LIGPLOT [26]. Figures were generated using the program PyMOL [27]. Superposition of Impα structures was performed using LSQ algorithm [28] present in the software package CCP4 [29].

PDB accession code
Coordinates and structure factors from NcImpα/SV40NLS have been deposited in the PDB under accession code 4RXH.

Structure of NcImpα in complex with SV40NLS
NcImpα was expressed as an N-terminal truncation lacking residues 1-74 (which corresponds to the IBB domain; [30]) that are responsible for autoinhibition. Furthermore, crystallization was performed in the presence of an NLS ligand to stabilize the truncated protein, as previously reported [17,20], and to investigate the protein-NLS binding features.
X-ray diffraction data collection and refinement statistics of the NcImpα/SV40 NLS complex are summarized in Table 1. The crystals are not isomorphous to other Impα/SV40  , where jFobsj and jFcalcj are the observed and calculated structure-factor amplitudes, respectively. § R free is equivalent to R cryst but was calculated with reflections (5%) omitted from the refinement process.
Calculated based on the Luzzati plot with the program SFCHECK [31]. † † Calculated with the program PROCHECK [31]. a Values in parentheses are for the highest-resolution shell. complexes [10] and diffracted to high resolution (1.75Å), which is the highest resolution obtained for an Impα to date.
The final model of the NcImpα/SV40NLS complex consists of 428 residues of NcImpα (79-507), the peptide ligands bound to the major (seven residues) and minor (four residues) sites, and 349 water molecules ( Table 1). The structure exhibits an elongated and curved shape composed of ten tandem Arm repeats, each containing three α-helices (H1, H2 and H3;), as observed in other Impα structures [9,10,16,30,32] (Fig 1). The loop containing residues 461-465 could not be modeled due an absence of electron density. The concave surface of the protein maintains the conserved array of Trp and Asn residues and negatively charged residues that interact with positively charged residues from the NLS ligands.

Binding of the SV40 NLS peptide at the NcImpα major binding site
The SV40 NLS peptide binds the major binding site of NcImpα in an extended conformation with an orientation that is antiparallel to the Arm repeats. The electron density for the peptide in the major binding site is well defined, allowing for the unambiguous modeling of the eight peptide residues ( 125 PPKKKRKV 132 , Fig 2a). The SV40 NLS peptide exhibits a conserved binding mode at the major binding site, which is analogous to mouse (MmImpα; [10]), yeast (ScImpα; [9]) and rice (OsImpα; [16]) Impα proteins. The N-terminal residue (P125) also exhibits a conformation that is similar to that observed in the structures of OsImpα/SV40 NLS and MmImpα in complex with an extended SV40 NLS peptide (G110-G132, referred as  Structure of Importin-α from a Filamentous Fungus CN-SV40NLS) [37]. ScImpα/SV40 NLS and MmImpα/SV40 NLS were crystallized using truncated version of the SV40 NLS peptide (P126-V132).
The average B-factor of the peptide at the major binding site (33.2 Å 2 ) is lower than the average B-factor for the protein (37.7 Å 2 ), indicating stability for the interaction of the peptide at this site. The conserved asparagines N150, N192 and N235 of NcImpα stabilize the backbone of the SV40 NLS peptide via hydrogen bonds, whereas the residues W146, W188 and W231 form pockets for the side chains of K129 and K131 at the P 3 and P 5 positions of the SV40NLS peptide, respectively. The K127 of SV40NLS participates in hydrophobic interactions with G195 of NcImpα at the P 1 binding site. K128 of SV40 NLS forms hydrogen bonds with G154, A152 and T159 and a salt bridge with D196 of NcImpα, as observed for lysine residues occupying the P 2 position in previous structures [9,10]. In the P 4 site, the side chain of R130 interacts with L109 and K111 of NcImpα via hydrogen bonds and hydrophobic interactions with P115 and S153. Finally, K131 (position P 5 ) forms hydrogen bonds with Q185 and hydrophobic interactions with F142, and V132 (position P 6 ) forms a hydrogen bond with S110 of NcImpα (Fig 3a, S1 Fig).
Interestingly, the N-terminal residue P125 of the SV40NLS peptide, that is not in the P 1 − P 6 positions, forms hydrogen bonds and hydrophobic interactions with R238 and D270 of NcImpα. P125 aids in the stabilization of interactions between the protein and a peptide NLS at the major binding site, as observed previously for MmImpα/CN-SV40NLS and OsImpα/ SV40NLS structures [16,37].

Binding of the SV40 NLS peptide at the NcImpα minor binding site
The electron density for the SV40 NLS peptide at the minor binding site is also well defined and allows the unambiguous modeling of four residues of the peptide ( 128 KRKV 132 , Fig 2b). As observed for the major binding site, asparagines residues (N319 and N361 in this case) of NcImpα define and guide the backbone of the peptide. The average B-factor for the SV40 NLS peptide bound at the minor binding site of NcImpα is higher than the average B-factor for the protein (39.3 and 37.7 Å 2 , respectively), indicating a lower stability for the peptide at this site compared with the major site.
The interactions between the SV40NLS peptide and NcImpα in the minor site exhibit greater similarities with the interactions between this peptide and OsImpα [16] (Table 2) than those with MmImpα or ScImpα [9,10]. In the NcImpα/SV40NLS structure, K129 at the P 1 ' position forms hydrogen bonds with G323, V321 and T328 of NcImpα. R130, which occupies the P 2 ' position, is accommodated between the hydrophobic side chains of W357 and W399 and interacts with E396 via salt bridges and with S360 via hydrogen bonds, which results in the lowest B-factor value (36.2 Å 2 ) compared to the other residues of the peptide. K131 at the P 3 ' position is stabilized by helix dipoles, negatively charged residues and hydrogen bonds with N283, G281, and T322. Finally, V132 at P 4 ' interacts with D280, R315 and N319 of NcImpα via hydrophobic interactions (Fig 3b, S2 Fig).

Affinity of the SV40 NLS peptide for NcImpα
The affinity and other thermodynamic parameters for the association of NcImpα and the SV40 NLS peptide were evaluated using ITC. A protein:peptide molar ratio of 1:20 was sufficient to yield a sigmoidal titration curve representing an exothermic process during complex formation (Fig 4). A two non-symmetrical binding site interaction model was selected based on the determined structure, which indicates the presence of two NLS binding sites in the Impα protein with different binding modes. The equilibrium dissociation constants of K d = 1.23±0.22 μM and K d = 1.69±0.46 μM (Table 3) indicated that the SV40NLS peptide may bind to both sites but with higher affinity for one site. These results are consistent with previous structural and functional results that indicate the presence of the major and minor binding sites in Impα in complex with the SV40NLS peptide [9,10] and are comparable with ITC experiments performed with MmImpα/SV40NLS complex [19]. Furthermore, the negative contribution for the enthalpy (ΔH) and positive value for the entropy (ΔS) suggest that both hydrogen bonds and hydrophobic interactions play a role in this interaction, whereas conformational changes are unfavorable. These data corroborate with the structural information obtained in the present study.

Comparison of NcImpα with other Impα structures
The crystal structure of NcImpα resembles other Impα structures [9,10,16] but exhibits a conformation that is more concave compared with MmImpα (Fig 5a). Its superposition (residues 83-504) with other Impα proteins reveals a higher r.m.s.  (Fig 5b). The primary sequence identities between Table 2. Binding to specific pockets of Impα/NLS structures from different organisms.
Several studies have shown that Impα proteins from different families exhibit preferences for specific NLSs [45][46][47][48]. Thus, examining the binding mode of the SV40NLS peptide, which has been crystallized with various Impα proteins, may provide insights into the binding specificities of these proteins. Crystal structures of Impα/SV40NLS complexes have been  Structure of Importin-α from a Filamentous Fungus determined for ScImpα [9], MmImpα (in complex with SV40NLS and with an extended SV40NLS peptide, referred to as CN-SV40- [10,37]), OsImpα [16] and NcImpα (the present study). In all of these structures, the SV40NLS peptide binds strongly to the major site via several interactions, resulting in well-defined electron density and B-factor values that are similar to or lower than the average B-factor value for the protein. These results are consistent with the determined affinities for the MmImpα/SV40NLS [19] and NcImpα/SV40NLS complexes using ITC, in which both studies demonstrated the presence of two binding sites for the peptide Structure of Importin-α from a Filamentous Fungus exhibiting with different affinities. In all of these structures, SV40NLS binding to the major site is essentially identical, in which all six positions (P 1 − P 6 ) of the peptide ( 127 KKKRKV 132 , Table 2) are occupying similar regions at the protein. The main differences are related to the number of peptide residues bound before the P 1 position (OsImpα exhibits 4 residues before P 1 , NcImpα exhibits 2 residues, MmImpα exhibits 1 residue and MmImpα/CN-SV40NLS exhibits 4 residues). However, these differences result from the length of the SV40NLS peptide used in these structural studies rather than differences in Impα proteins from different organisms.
The role of the minor binding site for NcImpα specificity The major site was usually associated with the binding of monopartite NLSs, whereas the minor site was associated with a secondary binding site for bipartite NLSs [7,10,40]. Recently, studies have demonstrated that some monopartite NLSs may bind to only the minor site [16,38], indicating the importance of this binding site for Impα specificity for cargo proteins.
In contrast to the binding of SV40NLS to the major site (Fig 6a), the binding mode of the peptide at the minor binding site is different among different Impα proteins. As shown in Table 2 and Fig 6b, SV40NLS binds to ScImpα and MmImpα via Lys and Arg residues at the P 1 ' and P 2 ' positions, whereas the peptide binds OsImpα and NcImpα via Lys residues at both P 1 ' and P 2 ' positions, i.e., the binding of the peptide is shifted one position in the OsImpα and NcImpα structures.
Interestingly, Lys and Arg residues at the P 1 ' and P 2 ' positions, respectively, are observed for the majority of monopartite (hPLSCR1, hPLSCR4, c-Myc, and TPX2 [11,38,39,49]) and bipartite NLS complexes (nucleoplasmin, N1N2, RB and FEN1 [10,37,40]), and it is also the predominant binding mode for MmImpα/CN-SV40NLS [37]. Furthermore, the authors of MmImpα/SV40NLS structure [10] also observed staggering of peptide one position N-terminally which may lead to an alternative binding mode with Lys and Arg residues at the P 1 ' and P 2 ' positions. The binding of KK residues at the P 1 ' and P 2 ' positions for SV40NLS in MmImpα and ScImpα seems to be exceptions in these particular cases.
An additional interesting characteristic of SV40NLS binding to Impα is the presence of additional interactions between both the N-and the C-termini of the peptide and MmImpα in the minor site: two positions before P 1 ' (P126, N-terminus) and in the P 5 ' position (V132). These interactions are observed in only the mammalian receptor despite all SV40NLS peptides used for crystallization with Impα proteins have these same residues. Structural studies on OsImpα have indicated that the plant-specific NLS peptide preferentially binds at the minor site in OsImpα and at the major site in MmImpα [16]. The similarity of SV40NLS binding at the minor site between OsImpα and NcImpα raises questions concerning the importance of this NLS binding site for the import mechanism of N. crassa.
Chang and colleagues [16] observed the presence of non-conserved residues between OsImpα and MmImpα (S394, R427, E434, E480, and K484 for OsImpα) in the region near the minor site (Arm repeats 8 and 9) and toward the C-terminus may be responsible for the specificity of NLS binding for OsImpα. Multiple sequence alignment of Impα proteins (Fig 7a) indicates that some of these residues that are present in OsImpα (S394, E480 and K484) are also present in NcImpα (S402, E493, and K497) and in other Impα proteins from the α1 family (S408, R443 and K500 in ScImpα and E491 and K495 in HsImpα). Structural comparison between NcImpα and MmImpα (Fig 7b) shows that these three residues (S402, E493, and K497 in NcImpα) are substituted by T402, S483 and A487 for MmImpα indicating that favorable interaction of particular peptides may occur. Particularly, T402 in MmImpα (S402 in NcImpα) has been reported to lie in an identical position as S394 in the OsImpα structure, sterically preventing the binding of the plant-specific NLS peptide to MmImpα [16]. However, these substitutions may also prevent the binding of K127 of SV40NLS to NcImpα and OsImpα, explaining the presence of this interaction in MmImpα, which belongs to the α2 family, and the absence of this interaction in NcImpα, which belongs to the α1 family.
The majority of monopartite (hPLSCR4, c-Myc, and TPX2 [11,38,39]) and bipartite NLS complexes (nucleoplasmin, N1N2, RB and FEN1 [7,10,40]) presented KR residues at the P 0 1 and P 0 2 positions which seems to be the most favorable residues to occupy these positions. The binding of KK residues atthe P 0 1 and P 0 2 positions for SV40NLS in MmImpα seems to be an exception which probably occurs because it is also favorable [10] to bind additional residues at Comparison of SV40NLS peptides binding to the major (a) and minor (b) binding sites for Impα structures from different organisms. SV40NLS peptides in complex with NcImpα (magenta), ScImpα (green), OsImpα (blue) and MmImpα (yellow) were superimposed using the Cα atoms of the peptides. Positions binding corresponding to the major (P 1 -P 5 ) and minor (P 1 0 -P 4 0 ) sites are identified along the chains (binding pocket labels were based on peptide binding with NcImpα). doi:10.1371/journal.pone.0128687.g006 Structure of Importin-α from a Filamentous Fungus  Fig 7. Non-conserved residues between Impα proteins near to the minor NLS binding region. (a)Partial alignment of amino acid sequences of Impα proteins from different organisms. Conserved residues are shown in black. The residues S402, K497 and E493 for NcImpα and their equivalent residues for Impα proteins are shown in red. (b) Superposition between Cα of NcImpα and MmImpα structures (performed as described in Fig 5) highlights the residues S402, K497 and E493 of NcImpα (magenta) and the residues T402, S483 and A487 of MmImpα (green). These particular residues may be associated to NLS binding specificity to the minor NLS binding site. E493 side-chain is shown in a hypothetical conformation in this figure because this residue presents lack of electron density in this structure due to its high flexibility. doi:10.1371/journal.pone.0128687.g007 Structure of Importin-α from a Filamentous Fungus the N-and C-terminal regions of the peptide for some Impα proteins. For OsImpα/SV40NLS and NcImpα/SV40NLS complexes, KR binding at the P 0 1 and P 0 2 positions is more favorable because fewer interactions occur with the N-and C-terminal regions of the peptide. This observation accounts for the shift in the binding of SV40NLS to OsImpα and NcImpα compared with MmImpα.
Affinity assays between NLS peptides and Impα proteins have been performed based in indirect affinity measurements [16,40,[49][50][51][52][53][54], using Surface Plasmon Resonance (SPR) [32], and ITC [19,38,55]. Indirect affinity measurements between NLSs and Impα proteins reported dissociation constants in the nM range and permitted the comparison among these molecules. SPR experiments was only able to estimate the peptide-protein affinity in the μM range which was lately confirmed by ITC experiments. The comparison between ITC analysis of MmImpα/SV40NLS [19] and NcImpα/SV40NLS is consistent with the features of the minor binding site observed in the structure. The higher affinity of the peptide at the minor binding site for MmImpα (K d = 0.98±0.08 μM) compared with NcImpα (K d = 1.69±0.46 μM) may be associated with the presence of additional interactions at N-and C-termini in the MmImpα/ SV40NLS complex. The crystal structure of this complex [10] has revealed that the residues K131 and V132 of the SV40 NLS peptide form salt bridges with the conserved E354 and R315 of MmImpα, respectively. Additionally, a hydrogen bond between V132 and R135 aids in the stabilization of the backbone of the SV40 NLS C-terminus. The importance of additional interactions N-and C-termini has been also observed with ITC assays for a phosphorylated NLS peptide which enhanced by 10-fold its affinity to MmImpα compared to unphosphorylated version [54]. Furthermore, it has been observed that the full length nucleoplasmin protein binds to Impα/Impβ complex with a 2-fold increase in affinity compared to just nucleoplamin NLS peptide [53]. Interesting, ITC studies with the hPLSCR4 NLS peptide ( 273 GSIIRKWN 280 ) which binds only the minor site of MmImpα [38] showed lower affinity (K d = 48.7±6.5 μM) compared with that of the SV40NLS peptide. This fact may be mainly attributed to the absence of a positively charged residue in the hPLSCR4 peptide at the P 1 ' position.
In conclusion, despite the structural similarities among Impα proteins, this study and other recent studies with this receptor from different organisms or different isoforms from the same organism clearly demonstrated differences in the binding specificities for cargo proteins. The differences between NcImpα and MmImpα may result from the phylogenetic distance among the proteins and the functions of each protein family in organism development, which results in differences in affinities for NLSs. The elucidation of NcImpα in complex with specific NLSs peptides from fungi may provide an explanation for the differences between these proteins.