The Saccharomyces cerevisiae Nucleoporin Nup2p Is a Natively Unfolded Protein*

Little is known about the structure of the individual nucleoporins that form eukaryotic nuclear pore complexes (NPCs). We report here in vitro physical and structural characterizations of a full-length nucleoporin, the Saccharomyces cerevisiae protein Nup2p. Analyses of the Nup2p structure by far-UV circular dichroism (CD) spectroscopy, Fourier transform infrared (FTIR) spectroscopy, protease sensitivity, gel filtration, and sedimentation velocity experiments indicate that Nup2p is a “natively unfolded protein,” belonging to a class of proteins that exhibit little secondary structure, high flexibility, and low compactness. Nup2p possesses a very large Stokes radius (79 Å) in gel filtration columns, sediments slowly in sucrose gradients as a 2.9 S particle, and is highly sensitive to proteolytic digestion by proteinase K; these characteristics suggest a structure of low compactness and high flexibility. Spectral analyses (CD and FTIR spectroscopy) provide additional evidence that Nup2p contains extensive regions of structural disorder with comparatively small contributions of ordered secondary structure. We address the possible significance of natively unfolded nucleoporins in the mechanics of nucleocytoplasmic trafficking across NPCs.

In eukaryotic cells, the translocation of biomolecules between the nucleus and cytosol occurs through nuclear pore complexes (NPCs), 1 supramolecular protein structures embedded in the double lipid membrane of the nuclear envelope (1)(2)(3). The Saccharomyces cerevisiae NPC is a 60-MDa structure (4) formed by 30 different Nups present in multiple copies per NPC (5); 13 of these Nups contain regions with multiple phenylalanine-glycine repeats (FG Nups) that are believed to be involved directly in the transport mechanism (6). The yeast NPC contains a core ring structure with 8-fold symmetry measuring 95 nm in diameter and 35 nm in depth (4). Most of the Nups are distributed symmetrically on the nucleoplasmic and cytoplasmic faces of the ring, forming a central transporter that is the conduit for macromolecular transport. However, a subset of Nups form filaments that extend into the cytoplasm, and another subset form a basket structure that extends into the nucleoplasm (5). The Nups of the cytoplasmic fibrils and nuclear basket are proposed to function in the initiation and termination of karyopherin-mediated transport reactions (7)(8)(9)(10)(11).
In S. cerevisiae, proteins, mRNA, tRNA, and other biomolecules (collectively referred to as "cargo") larger than ϳ30 kDa are actively transported across the NPC via association with karyopherins (Kaps: importins, exportins, and transportins) (12)(13)(14). Kaps bind the import or export signals of cargo destined for transport across the NPC and interact with the FG Nups, which may serve as stepping stones for Kaps as they traverse the NPC (6,(15)(16)(17). The GTPase Gsp1p (the yeast homolog of Ran) governs the interaction of Kaps with transport cargo and Nups by binding directly to Kaps (6,15,18). In export reactions, Gsp1p-GTP stabilizes the cargo-exportin interaction and enhances the binding of this export complex to FG Nups (19); in import reactions, Gsp1p-GTP induces the release of cargo from importins and prevents the association of importins and FG Nups (6,7,15). A high concentration of Gsp1p-GTP is maintained in the nucleus by Prp20p, the Ran guanine nucleotide exchange factor (RanGEF) (20, 21); the Ran GTPase activating protein (RanGAP), Rna1p, maintains low cytoplasmic concentrations of Gsp1p-GTP (22)(23)(24).
The S. cerevisiae protein Nup2p is a 78-kDa nucleoporin that contains 720 amino acid residues, including 16 FxFG peptide repeats (25). Although Nup2p is not essential under normal physiological conditions, yeast lacking Nup2p (nup2⌬) exhibit defects in nuclear import of Kap60p⅐Kap95p (importin ␣␤ or karyopherin ␣␤)-dependent cargoes and defects in the Cse1pmediated recycling of Kap60p (Srp1p/importin ␣/Kap ␣) from the nucleus to the cytoplasm (26 -28). Nup2p interacts with Kap95p and Los1p via its FxFG repeats (15,27,29). In addition, the N terminus of Nup2p binds directly to Kap60p (27,28) and functions as a KaRF (karyopherin release factor) to induce the dissociation of cargoes from Kap60p (7). Nup2p also contains a low affinity binding site for Gsp1p-GTP at its C terminus (30,31). Electron microscopic studies have localized Nup2p to the nuclear basket structure of the NPC where it binds directly to Nup60p in a Gsp1p-GTP-sensitive manner (27,28,31,32). At this location, Nup2p is proposed to function in the termination of import reactions involving the Kap95p⅐Kap60p heterodimer and the initiation of Kap60p export (7,26). Interestingly, Nup2p associates with Nup60p in a dynamic manner dependent on Gsp1p-GTP, allowing Nup2p to shuttle between the nucleoplasm and cytoplasm under conditions of low nuclear Gsp1p-GTP concentrations (31,32).
In this study, we employ a variety of biophysical methods to examine the physical and structural characteristics of the nucleoporin Nup2p as a model for nucleoporin structure. From these studies, we have concluded that Nup2p is a natively unfolded protein. We speculate that the intrinsic disorder of Nup2p is important for its functions, as it may facilitate rapid association and dissociation reactions with Kaps while allowing simultaneous interactions with binding partners such as Nup60p, Gsp1p, and Prp20p.

MATERIALS AND METHODS
Preparation of S. cerevisiae Extract-GPY60 yeast were grown in 1 liter of YPD medium at 30°C to an A 600 ϭ 2.0. Yeast were harvested by sedimentation at 5,000 ϫ g for 10 min and resuspended in 20 mM Hepes, pH 6.8, 150 mM KOAc, 250 mM sorbitol, 2 mM Mg(OAc) 2 , and protease inhibitors (0.1 mg/ml phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, 2 g/ml aprotinin, and 2 g/ml leupeptin) to a final volume of 40 ml. Yeast were lysed in a French Press cell and cell debris was removed by sedimentation at 30,000 ϫ g for 30 min at 4°C. The supernatant was desalted in a Sephadex G-25 fine column (Amersham Biosciences) pre-equilibrated in 20 mM Hepes, pH 6.8, 150 mM KOAc, 2 mM Mg(OAc) 2 . Pooled fractions were supplemented with 0.1% Tween, 2 mM dithiothreitol, and protease inhibitors.
Construction and Purification of Recombinant Proteins-Recombinant Nups and Kaps were expressed as glutathione S-transferase (GST) fusions using vector pGEX-2TK (Amersham Biosciences) that incorporates a thrombin cleavage site at the fusion junction. The NUP2, KAP60, KAP95 genes and portions of NUP2 were amplified from yeast genomic DNA (Promega) by PCR. The PCR products were ligated into vector pGEX-2TK and transformed into BL21 codon plus Escherichia coli (Novagen). E. coli were grown in 1 liter of 2ϫ YT medium with ampicillin (0.1 mg/ml) and 2% glucose at 37°C to A 600 ϭ 1.0. Protein production was induced with isopropyl-1-thio-␤-D-galactopyranoside for 15 min to 2 h at 24°C. E. coli were sedimented and resuspended in lysis buffer (50 mM Tris, pH 8.0, 100 mM NaCl, 2 mM EDTA, 5 mM dithiothreitol) with protease inhibitors (0.1 mg/ml phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, 2 g/ml aprotinin, and 2 g/ml leupeptin). Cells were lysed in a French press (SLM Instruments), cell debris was removed by centrifugation at 30,000 ϫ g for 15 min at 4°C, and Tween 20 (0.1%) was added to the supernatant. Recombinant GST-fusion proteins were purified on glutathione-Sepharose beads according to the manufacturer's instructions (Amersham Biosciences). The GST moiety was cleaved off by incubation with thrombin (Calbiochem) at 25°C (1 NIH unit of thrombin, 100 g of GST fusion) for controlled times. A molar excess of hirudin (Calbiochem) was added to neutralize thrombin, and free GST was removed using glutathione-Sepharose beads. Proteins were further purified by fast protein liquid chromatography in gel filtration columns equilibrated with 20 mM Hepes, pH 6.8, 150 mM KOAc, 2 mM Mg(OAc) 2 .
Gel Filtration and Stokes Radii Calculations-Purified proteins were resolved in a Superose-6 gel filtration column (Amersham Biosciences) equilibrated with 20 mM Hepes, pH 6.8, 150 mM KOAc, and 2 mM Mg(OAc) 2 using the AKTA fast protein liquid chromatography system (Amersham Biosciences). The elution volumes (V e ) of proteins were determined by UV absorbance and/or collection of 0.5-ml fractions followed by SDS-PAGE and staining with Coomassie Blue (Bio-Rad). The Stokes radius (R S ) of Nup2p with and without bound Kaps was calculated using the method of Porath (33). Briefly, the V e of four standard proteins (aldolase, 48.1 Å; catalase, 52.2 Å; ferritin, 61.0 Å; and thyroglobulin, 85.0 Å) were plotted in relation to their Stokes radii, allowing for estimation of the Nup2p R S from the resulting linear regression.
Sucrose Gradients and Sedimentation Coefficient Calculations-Linear sucrose gradients (5-20 or 10 -40% sucrose) were poured using a gradient mixer (Hoefer) and recombinant Nup2p (5 g), with or without recombinant Kap95p and/or Kap60p (5 g each), was layered on top. Tubes were subjected to centrifugation at 259,000 ϫ g for 15 h (5-20% sucrose gradients) or 24 h (10 -40% sucrose gradients) at 4°C in a TLS-55 rotor (Beckman). Eighteen fractions were collected from the top of each gradient. Purified BSA (4.7 S), aldolase (7.5 S), and catalase (11.3 S) were used in parallel gradients as size standards. Proteins in each fraction were resolved by SDS-PAGE and visualized with Coomassie Blue. Sedimentation coefficients (s 20, w ) were calculated by the method of McEwen (34) using integration tables and the predicted specific volumes of Nup2p, Kap95p, and Kap60p. Molecular weight estimates for Nup2p and Nup2p-containing complexes were calculated as described (33). Calculations were based on experimentally deter-mined Stokes radii and s 20,w values and were calculated using the following equation, where 0 is the viscosity of the solvent (g/cm⅐s); , the density of the solvent (g/cm 3 ); , the partial specific volume of the protein (cm 3 /g); and n, Avogadro's number. Circular Dichroism (CD) Measurements-Far-UV CD spectra of purified Nup2p and Nup2p fragments were obtained on AVIV 60 DS and AVIV 62A DS spectrophotometers (Lakewood, NJ). Spectra were recorded using 0.1-or 0.01-cm cuvettes from 250 to 190 nm with a step size of 1 nm and an averaging time of 10 s. For all spectra, an average of three scans was obtained, and the background spectra of the buffer was removed.
Fourier Transform Infrared Spectroscopy (FTIR) Measurements-Attenuated total reflectance data were collected on a Nicolet 800SX FTIR spectrometer equipped with an MCT detector. The IREC (72 ϫ 10 ϫ 6 mm, 45°germanium trapezoid) was held in a modified SPECAC out-of-compartment attenuated total reflectance apparatus. The hydrated thin films were prepared as described previously (35,36). Typically 1024 interferograms were co-added at 4 cm Ϫ1 resolution. Data analysis was performed with GRAMS32 (Galactic Industries). Secondary structure content was determined from curve fitting to spectra deconvoluted using second derivatives and Fourier self-deconvolution to identify component band positions. Hydrated thin films were prepared by drying 50 l of 1 mg/ml Nup2p solution in 50 mM Hepes, pH 6.8, on a germanium crystal with dry N 2 . The IR spectra were collected, followed by Fourier transformation using the spectrum of the clean crystal as a background. Background water (liquid and vapor) components were subtracted from the protein spectrum.
Proteinase K Sensitivity Assays-Recombinant Nup2p and/or recombinant Kap95p and Kap60p were digested in solution with 100 ng/ml proteinase K (Sigma) at 37°C. At various time points, aliquots were removed and mixed with Laemmli sample buffer containing 1 mM phenylmethylsulfonyl fluoride (Sigma) to stop proteolysis. Samples were then resolved by SDS-PAGE and stained with Coomassie Blue to detect the products of proteinase K digestion. For protease digestion of native Nup2p in the context of intact NPCs, yeast nuclei were purified as described (37) and resuspended to 3.6 mg/ml in 20 mM Hepes, pH 6.8, 150 mM KOAc, 250 mM sorbitol, 2 mM MgOAc, 50% glycerol. Proteinase K (Sigma) was added to the nuclei at 300 ng/ml. The reaction was incubated at 37°C, and aliquots were removed at the indicated time points. Digestions were stopped with 2ϫ Laemmli sample buffer plus 1 mM phenylmethylsulfonyl fluoride followed by heating at 95°C for 10 min. The fraction of full-length Nups remaining was quantified by SDS-PAGE followed by Western blotting with anti-Nup antibodies and 125 I-Protein A (Amersham Biosciences). Radioactive blots were exposed to phosphorscreens and then digitized and quantified using Image-Quant software (Amersham Biosciences). Anti-Nup2p antibodies were generated as described previously (31). Anti-Kap95p, -Kap60p, -Nup85p, and -Pom152p rabbit polyclonal antibodies were raised against Kap95p (full-length), Kap60p (full-length), Nup85p (fulllength), and Pom152p (AA 210 -533 and AA 1182-1337).

RESULTS
The S. cerevisiae NPC is formed by 30 different Nups that are present in multiple copies per NPC (5). Cryoelectron microscopy images of purified NPCs and immuno-EM studies have illustrated the overall architecture and assembly of the NPC as well as the regional localization of individual Nups within each NPC (4,5). However, little is known about the structure of individual Nups. The experiments described below provide a detailed physical and structural characterization of the yeast nucleoporin Nup2p.
Two-step Purification of Nup2p from Yeast Extract-As a part of our ongoing biochemical investigations into the mechanism of Kap95p (importin ␤/karyopherin ␤) movement across the yeast NPC, we incubated S. cerevisiae extract with a GST-Kap95p affinity resin to isolate specific Kap95p-interacting proteins. Among the most abundant proteins that eluted with 250 mM MgCl 2 was the nucleoporin Nup2p (Fig. 1A, lane 2). Nup2p is a member of the FxFG family of nucleoporins and is tethered primarily to Nup60p in the nuclear basket structure of the NPC (31,32); Nup2p also shuttles between the nucleoplasm and cytoplasm (32). A second chromatography step was sufficient to obtain purified Nup2p in complex with its binding partner Kap60p (data not shown). Briefly, the MgCl 2 eluate from the GST-Kap95p affinity resin was separated by gel filtration in a Superose-6 column equilibrated in physiological buffer. Nup2p co-purified with Kap60p, and no additional proteins were visible in the same fractions as detected by Coomassie stain (data not shown). To purify Nup2p to homogeneity without Kap60p, the GST-Kap95p eluate was applied to a Superose 6 column equilibrated in buffer with 250 mM MgCl 2 . Under such conditions, Nup2p that eluted from the Superose-6 column was pure as verified by Coomassie (Fig. 1A, lane 3) and silver stains (data not shown).
Gel Filtration of Nup2p Suggests a Structure with Low Compactness-Purified Nup2p eluted from the Superose 6 gel filtration column (24 ml bed volume) at elution volumes (V e ) from 9.25 to 10.75 ml (data not shown). We calculated the R S of Nup2p via the method of Porath (33), using globular proteins with known hydrodynamic dimensions (aldolase, 48.1 Å; catalase, 52.2 Å; ferritin, 61.0 Å; and thyroglobulin, 85.0 Å) as standards. The calculated R S of 101 Å for purified yeast Nup2p was unexpectedly large for a protein with a molecular weight of 78,000 (Table I); in comparison, the R S of BSA, a monomeric protein with a similar molecular weight (66,000), is only 35.5 Å. The presence of Kap60p in complex with Nup2p altered slightly the V e of Nup2p (data not shown), resulting in a smaller R S of 97 Å (Table I).
Purified recombinant Nup2p produced in E. coli also possessed an unexpectedly large Stokes radius of 79 Å in the Superose 6 column (Fig. 1B, top panel; Table I). The addition of equimolar concentrations of Nup2p binding partners (Kap60p and Kap95p) yielded no significant changes in the elution volumes of Nup2p in Nup2p⅐Kap complexes versus Nup2p alone (Fig. 1B, bottom panels; Table I).
It is unclear why the recombinant Nup2p produced in E. coli exhibits a smaller Stokes radius than the Nup2p purified from yeast. Post-translational modifications of Nup2p in S. cerevisiae have not been reported, but sulfation or phosphorylation may account for the difference. However, recombinant and endogenous Nup2p migrate identically in SDS-acrylamide gels (data not shown). It is also possible that structural differences may result from the different purification procedures or the presence of a 9-AA tag at the N terminus of the recombinant Nup2p. Regardless, recombinant and native Nup2p behave identically in their ability to bind Kap60p and Kap95p (Figs. 1 and 2; data not shown), suggesting that the two versions of Nup2p were functionally identical.   The large hydrodynamic dimension of Nup2p is consistent with either (i) a homo-oligomeric assembly of Nup2p or (ii) a nonglobular structure of low compactness (38,39). Sedimentation velocity experiments also provide information regarding the mass, size, and degree of compactness of proteins or protein complexes. To calculate the sedimentation coefficient of Nup2p, recombinant Nup2p with and without its Kap binding partners was layered on linear sucrose gradients (5-20 or 10 -40%) and subjected to centrifugation at 259,000 ϫ g (Figs. 2, A and B). Table I lists the sedimentation coefficients (s 20,w ) for Nup2p and several Nup2p complexes. The s 20,w values were calculated using the method of McEwen (34), and the sedimentation of Nup2p was compared with several standard proteins: BSA (4.7 S/66 kDa), aldolase (7.5 S/160 kDa), and catalase (11.3 S/240 kDa) ( Fig. 2; data not shown). Note that despite the large hydrodynamic dimensions of Nup2p (79 Å) in gel filtration columns, purified Nup2p sediments as a 2.9 S particle, which was significantly slower than BSA (4.7 S/66 kDa) ( Fig. 2A, top  panel). In contrast, the addition of Kap60p or Kap95p caused the resulting Nup2p⅐Kap complex to sediment more rapidly with s 20,w values of 4.8 and 6.2 S, respectively ( Fig. 2A, middle panels; Table I). As expected, the trimeric Nup2p⅐ Kap95p⅐Kap60p complex sediments even faster (7.1 S) (Fig. 2, A, bottom panel, and B, bottom panel; Table I) and was comparable with the sedimentation of Nup2p in yeast extracts (Fig.  2B, top panel). The changes in Nup2p sedimentation caused by Kap95p and/or Kap60p are reasonable considering the known masses and shapes of Kap95p and Kap60p monomers (40,41).
The unexpectedly small sedimentation coefficient of purified Nup2p (2.9 S) was indicative of a low mass particle and/or of high drag forces that slow the migration of Nup2p through the sucrose gradient. Although high molecular weight protein complexes may also sediment slowly as a result of oligomeric arrangements that produce high drag forces, the sedimentation data for Nup2p were most consistent with a nonglobular structure that exhibits a low degree of compactness. To further establish that the large Stokes radius of recombinant Nup2p was not because of a homo-oligomeric assembly, we calculated the molecular weight of Nup2p based on the experimentally determined s 20,w and R S values using the method of Siegel and Monty (33). The calculated molecular weight for Nup2p with and without Kap95p and/or Kap60p was comparable with the predicted molecular weights for monomeric Nup2p and Nup2p complexes containing Nup2p and each Kap in single copy (Table I). We therefore conclude that the large hydrodynamic dimensions of Nup2p and Nup2p⅐Kap complexes in gel filtration columns were likely the result of a disordered, highly flexible Nup2p structure. As detailed below, additional evidence obtained from amino acid analyses of Nup2p (Fig. 3), circular dichroism spectroscopy (Fig. 4), FTIR (Fig. 5), and proteinase K sensitivity experiments (Fig. 6) support the hypothesis that Nup2p is a natively unfolded protein.
The Amino Acid Composition of Nup2p Predicts a Disordered Structure-The number of proteins shown in vitro to contain a natively unfolded domain(s) under physiological conditions has increased exponentially in the last 10 years. This growing data base has permitted the identification of amino acid compositions peculiar to natively unfolded proteins. Natively unfolded proteins contain a low percentage of hydrophobic amino acids and often have a large net charge at neutral pH as the result of non-neutral pI values (38). Compared with globular proteins in the Protein Data Bank, natively unfolded proteins were significantly depleted of amino acids Ile, Leu, Val, Trp, Phe, Tyr, Cys, and Asn, and were enriched in amino acids Glu, Lys, Arg, Gly, Gln, Ser, Pro, and Ala (39). Hypothetically, the hydrophobic amino acids in the former group contribute "order" to a protein, whereas the charged or polar amino acids of the latter group facilitate unfolding and "disorder." Nup2p contains only six tyrosines (Tyr), one cysteine (Cys), and no tryptophans (Trp); as a result, 25.8% of the Nup2p amino acid sequence was composed of order conferring amino acids, compared with an average of 38.1% for all proteins in the S. cerevisiae proteome (Fig. 3A). Conversely, 57.6% of the Nup2p sequence was derived from the group of AA that confer disorder, compared with 46.0% for all S. cerevisiae proteins (Fig. 3A). Although Nup2p has a high frequency of charged amino acids (Asp, Glu, Lys, and Arg), its 104 positive charges and 104 negative charges compensate at neutral pH, yielding an isoelectric focusing point (pI) of 7 that was uncharacteristic of natively unfolded proteins (Fig. 3A).
A recent study compared the mean hydrophobicity (ϽHϾ) and mean net charge (ϽRϾ) of natively folded and natively unfolded amino acid sequences, establishing that natively unfolded proteins occupy a clearly demarcated region of the charge-hydrophobicity space (42). Fig. 3C shows the location of natively folded proteins (black squares) and unfolded proteins (gray circles) from the original study, as well as Nup2p (large open circle) in the charge hydrophobicity plot; the line represents the boundary between folded and unfolded polypeptides and was defined by the equation, ϽHϾ b ϭ (ϽRϾ ϩ 1.151)/ 2.785. Although Nup2p has a mean net charge of 0, its low mean hydrophobicity (Fig. 3B) places it near the boundary line, but clearly in the natively unfolded region of the charge hydrophobicity plot (Fig. 3C). Thus, the amino acid composition of Nup2p and its low mean hydrophobicity predict that its structure was intrinsically disordered.

CD and FTIR Analyses Show That Nup2p Contains a Large Contribution of Unordered Secondary
Structure-Natively unfolded proteins typically have a low content of ordered secondary structure. To characterize the secondary structure of Nup2p and fragments thereof (see diagram in Fig. 4A), we used CD and FTIR spectroscopic techniques. Far-UV CD and FTIR allow estimations of the ␣-helical, ␤-sheet, and random coil content of a protein. Typically, the far-UV CD spectra of polypeptides with extensive ␣-helical structure have two characteristic minima near 208 and 222 nm; ␤-sheet structure yields a minimum at 215 nm; and random coil is characterized by a negative peak in the vicinity of 200 nm and low ellipticity at 222 nm. The CD spectrum of full-length Nup2p shows an intense minimum at 202 nm that indicates a large unordered contribution (Fig. 4B). However, Nup2p was not composed entirely of random coil as the minimum was shifted from 200 to 202 nm; additionally, the slight negative ellipticity at 222 nm points to a small but detectable contribution of ␣-helical structure.
The frequency and distribution of the apparent unordered and ordered secondary structure (␣-helical and ␤-sheet structures) cannot be determined from the CD spectrum of full-length Nup2p. For example, the ordered secondary structure contribution could be manifested in at least two possible ways: (i) there may be a propensity for the entire length of Nup2p to form short, unstable segments of secondary structure; or (ii) small, specific domains of Nup2p may form relatively stable ordered secondary structure (␣-helices and ␤-strands) with intervening domains that were primarily unordered. To distinguish between the two possible Nup2p conformations, fragments of Nup2p were also analyzed by CD (Fig. 4, A-E). The N terminus of Nup2p (AA 1-185), which binds directly to Kap60p and targets Nup2p to the NPC (26,27), contains a higher fraction of ordered secondary structure than full-length Nup2p as indicated by the emergence of a second minimum at 222 nm (Fig. 4C). The C-terminal fragment of Nup2p (AA 562-720), which includes the Gsp1p-binding domain (30), contains less ordered secondary structure than the N terminus, albeit more than the full-length protein as evidenced by a weak second minimum at 222 nm (Fig. 4E). Despite the content of secondary structure in the N-and C-terminal fragments of Nup2p compared with full-length Nup2p, both termini contained significant contributions of disorder as indicated by their minima at 207 and 204 nm, respectively. In contrast, the large FxFG region of Nup2p (AA 186 -561) that binds Kap95p⅐Kap60p heterodimers (Fig. 4A) displays an intense peak of negative ellipticity at 200 nm characteristic of random coil (Fig. 4E). Thus, full-length Nup2p appears to be composed of N-and C-terminal domains with small contributions of ordered secondary structure, and an extensive FxFG region that was highly unordered. This indicates that the majority of Nup2p was similar to the coil-like class of natively unfolded proteins, which contain mostly random coils (38).
A recent analysis of native coil proteins and native premolten globule proteins established a clear relationship between the molecular weight of proteins from each conforma- tional class and their Stokes radii (38). For comparison, we calculated the predicted Stokes radii (R S ) for monomeric Nup2p (78 kDa) in the five known protein conformations (folded, molten globule, pre-molten globule, natively unfolded "premolten globule-like," and natively unfolded "coil-like") as described in Ref. 38 (Table II). For example, if Nup2p were a natively unfolded pre-molten globule-like protein, its predicted R S would be 54 Å; however, if Nup2p were a natively unfolded coil-like protein, its predicted R S would be 72 Å. The similarity between the experimentally determined R S (79 Å) and the predicted R S (72 Å) suggests that Nup2p is a natively unfolded coil-like protein.
FTIR spectroscopy also provides a powerful method for the quantification of protein secondary structure. The primary advantage of FTIR in comparison with CD spectroscopy was that FTIR was much more sensitive to detection of the ␤-structure. Fig. 5 shows the FTIR spectrum (amide I region) of full-length Nup2p at pH 6.8, which was typical for a substantially unfolded polypeptide chain. The dominant feature of this spectrum was the presence of a broad band in the vicinity of 1655 cm Ϫ1 , which corresponds to disordered conformation. Deconvolution (FSD and second derivative) and curve fitting of the FTIR spectrum permitted quantitative analysis of the secondary structure content in Nup2p. The results confirm that the majority of Nup2p (65-80%) is indeed essentially disordered (Table III).
Proteinase K Sensitivity Assays Reveal a Highly Flexible Nup2p Structure-Intrinsically disordered proteins are highly flexible because of their low compactness and a lack of secondary and tertiary structure (38). The flexibility of natively unfolded proteins can be demonstrated indirectly via sensitivity to protease digestion (38,39,43). Here, we treated Nup2p (with and without its Kap binding partners) with proteinase K (100 ng/ml), a nonspecific protease (Fig. 6). Proteinase K rapidly degraded Nup2p, eliminating all of the full-length protein after 30 min of incubation at 37°C (Fig. 6A). Degradation products of Nup2p appear after 10 min; however, they do not accumulate, indicating that they are also hypersensitive to proteinase K digestion. For comparison, Kap60p was resistant to the proteinase K treatment under identical conditions (Fig. 6A); Kap60p is a natively folded globular protein composed almost entirely of ␣-helices (40).
Because disordered proteins frequently acquire folded structure when in complex with other proteins (39,44), it is possible that Nup2p folds upon binding Kap95p and Kap60p, its likely partners in the cell (31,45). However, the binding of Kap95p⅐Kap60p heterodimers failed to improve the resistance of Nup2p to the protease treatment (Fig. 6B), suggesting that bound Kaps do not alter the flexibility of Nup2p or cause it to fold. As expected, the presence of Kap60p or Kap95p monomers also failed to change the sensitivity of Nup2p to the protease treatment (data not shown). Both Kap95p and Kap60p are natively folded proteins with high ␣-helical content (40,41) and were resistant to proteinase K digestion in these experiments ( Fig. 6B and data not shown).
It is also possible that Nup2p folds when bound to Nup60p in the NPC. Previously, it was shown that Nup2p is bound to NPCs in purified yeast nuclei (32). To test the protease sensitivity of Nup2p in its native context within intact NPCs, we exposed purified yeast nuclei to proteinase K and probed Western blots with antibodies against full-length Nup2p and other nuclei proteins. Endogenous Nup2p was highly sensitive to protease digestion (Fig. 6, C and D), indicating that Nup2p retains its flexible and disordered characteristics even when incorporated into the yeast NPC. By comparison, endogenous Kap95p, Kap60p, Nup85p, and Pom152p (all of which co-purify with the nuclei) were highly resistant to proteinase K (Fig. 6, C  and D). Note that Nup85p was partially clipped by proteinase K (Fig. 6D, third panel) but not fully digested (Fig. 6, C and D). This indicates that proteinase K can access Nup85p within the nuclei, but that its structure or interactions with other proteins prevents digestion. Although indirect, these results demonstrate that the structure of Nup2p was highly flexible in purified form and in its native context within the yeast NPC, supporting the conclusion that Nup2p was natively unfolded.
Heat Treatment of Nup2p Does Not Affect Its Binding Properties-Unlike most natively folded proteins, intrinsically disordered proteins retain their activity and are not denatured following high temperature incubations (39). To test its thermotolerance, Nup2p was heated to 90°C for 1 h. Pre-heated Nup2p was then incubated with GST, GST-Kap95p, or GST-Nup60p conjugated to glutathione-Sepharose beads to test its binding activity in solution (Fig. 7). As expected, the heattreated Nup2p bound equally well to GST-Kap95p and GST-Nup60p as compared with untreated Nup2p. Neither Nup2p nor heat-treated Nup2p bound to GST (data not shown). The ability of Nup2p to retain its binding activities following extreme heat treatment was consistent with its classification as a natively unfolded protein.  (AA 1-185). D, the middle FxFG region of Nup2p (AA 186 -561); and E, the C terminus of Nup2p (AA 562-720). All far-UV CD spectra were measured at 25°C at neutral pH in cuvettes with path lengths of 1 mm for Nup2p (0.022 mg/ml) and Nup2p (AA 562-720) (0.11 mg/ml) or 0.1 mm for Nup2p (AA 1-185) (0.69 mg/ml) and Nup2p (AA 186 -561) (0.50 mg/ml). Note the different scales of the spectra. Previous structural and physical analyses of nucleoporin and nucleoporin-containing complexes have been limited to electron microscopic and atomic force microscopic visualization of intact NPCs and purified Nups and Nup complexes (4, 46 -52). Although these images illustrate the overall shape of the NPC and Nups, they are of insufficient resolution to reveal the structural nature of nucleoporins. We therefore subjected Nup2p to biophysical, structural, and conformational analyses to gain insight into the structure of nucleoporins. We specifically chose Nup2p as a model nucleoporin because: (i) it contains a large, characteristic domain of FG repeats (and thus belongs to the family of FG Nups); (ii) much is known about its biochemical and genetic interactions; and (iii) Nup2p exhibits a dynamic interaction with the NPC, existing in both NPC-bound and freely diffusible/mobile forms (25,31,32). Our in vitro studies of purified Nup2p are particularly relevant to the structure of Nup2p when it is in its mobile form as it is possible (although not certain) that Nup2p undergoes structural changes upon incorporation into the NPC. The findings described here constitute the first detailed structural and physical analyses of a full-length nucleoporin.
Nup2p Is a Natively Unfolded Protein-Intrinsically unordered or natively unfolded proteins are identified by low hydrophobicity, high net charge, low compactness, absence of globularity, lack of secondary structure, and high flexibility (38,39,(42)(43)(44). Consistent with these characteristics, Nup2p contains a very low frequency of hydrophobic amino acids (such as Trp, Tyr, and Cys), which contributes to its low overall hydrophobicity (Fig. 3). The under-representation of hydrophobic amino acids in a protein may inhibit formation of a compact hydrophobic core (38,39,42,43), resulting in unexpectedly large hydrodynamic dimensions. Indeed, gel filtration and sucrose gradient experiments show that Nup2p possesses a large Stokes radius (78 -100 Å) and a small coefficient of sedimentation (2.9 S), characteristics that are indicative of low compactness ( Figs. 1 and 2). CD and FTIR measurements demonstrate that Nup2p and particularly its middle FxFG region (AA 186 -561) are highly disordered, lacking extensive contributions of ordered secondary structure (Figs. 4 and 5). Finally, Nup2p is hypersensitive to proteinase K digestion, consistent with a highly flexible tertiary structure (Fig. 6A). It remains possible that Nup2p acquires structure in complex with binding partners in its cellular context(s) (44). However, the high protease sensitivity of native Nup2p within purified nuclei suggests that Nup2p retains considerable disorder and flexibility even when bound to NPCs (Fig. 6, C and D).
The conclusion that Nup2p behaves as a natively unfolded protein was unexpected and provocative. As discussed below, the structural disorder of Nup2p has interesting implications regarding its function and the mechanism of trafficking across NPCs. It also raises the possibility that other Nups may be natively unfolded.
Why Is Nup2p Natively Unfolded?-Computer analyses of the yeast proteome predict that 30% of S. cerevisiae proteins contain disordered domains of at least 50 amino acids in length (39). Among the hundreds of proteins and protein fragments that have been characterized as intrinsically disordered, there is in an eclectic range of cellular functions including transcription factors, cytoskeletal components, heat shock proteins, chromosome-binding proteins, hormone receptors, and more (39,43,44). The evolutionary impetus for the emergence of natively unfolded proteins is not clear; yet, natively unfolded proteins share functional commonalities that suggest at least two significant advantages of structural disorder.
First, natively unfolded proteins often contain multiple binding domains and are thus capable of simultaneous interactions with multiple protein partners (39,43,44,53). Nup2p binds directly to Kap60p, Kap95p, Gsp1p-GTP, Nup60p, and Prp20p (26 -28, 30, 31). Furthermore, Nup2p binds simultaneously to Nup60p, Gsp1p-GTP, and Kap60p to form a tetrameric complex that facilitates Cse1p-mediated export of Kap60p (31). The flexible, unfolded structure of Nup2p may allow it to serve as a scaffold for the assembly of multisubunit complexes. Additionally, the structural disorder of Nup2p may be related to its ability to assemble onto the NPC during the biogenesis of NPCs.
Second, disorder can dramatically affect association and/or dissociation rates of the protein with binding partners. For example, natively unfolded proteins often associate rapidly with binding partners. Although their large hydrodynamic dimensions slow down diffusion, their size provides a large target for initial molecular collisions that facilitate association with binding partners (39,54). More importantly, the lack of rigid binding pockets permits multiple approach orientations for a binding partner that may increase the probability of productive interactions (39). Rapid association and dissociation rates may be an important feature for interactions of Nup2p with its binding partners. For example, Nup2p displays KaRF activity, as it accelerates the dissociation rates of import cargoes containing the classic nuclear localization signal from Kap60p (7). As disassembly of import complexes is rate-limiting in the Kap95p⅐Kap60p-dependent import pathways, fast binding of Nup2p to Kap60p cargo FIG. 5. Secondary structure analysis of Nup2p by FTIR spectroscopy. Spectrum of the amide I region of Nup2p was measured at pH 6.8 (dashed line). The curve fit spectrum is presented by the interior solid lines. The major ␤-structure peaks are in the 1620 -1640 cm Ϫ1 region, whereas the major disordered peaks are in the 1650 -1660 cm Ϫ1 region. Note the large amount of disorder structure in comparison to ␤-sheet structure. complexes is likely necessary to ensure rapid release of cargo to the nucleus and rapid recycling of Kap60p to the cytoplasm. Thus, large hydrodynamic size and high flexibility of Nup2p may allow it to perform its KaRF activity efficiently at the nuclear basket structure.
Is Nup2p Structure a Model for Other Nups?-A similar biophysical analysis of other yeast nucleoporins to determine whether intrinsic disorder is a common characteristic of Nups is currently underway. 2 Because the middle FxFG region of Nup2p (AA 186 -561) shows extensive disorder (Fig. 4), we hypothesize that other Nups containing FG repeat regions will also prove to be unfolded under physiological conditions. Interestingly, the composite cryo-EM images of Xenopus NPCs suggest that disorder is a general feature of Nups that comprise the central transporter region of the NPC. In these images, an unresolved amorphous protein "plug" occupies the central conduit, reflecting either a heterogeneity of cargoes within the transporter or a structural disorder of the Nups in that region (47). As discussed below, Nup disorder could explain some issues regarding the mechanism of facilitated trafficking across the NPC.
Our "natively unfolded Nups" hypothesis supports the current "entropic exclusion" and "selective phase partitioning" models of nuclear transport (3,5,55). In these models, nucleoporins are hypothesized to form a barrier meshwork that excludes most macromolecules larger than a threshold size from entering the NPC (3,5). Only molecules (such as karyopherins) exhibiting physical characteristics compatible with the meshwork hydrophobicity can penetrate the barrier and diffuse to the opposite side of the NPC (55). Natively unfolded Nups would be ideal building blocks for this barrier meshwork because their flexible structures could re-arrange, contort, and re-pack to accommodate a wide range of sizes for karyopherin⅐cargo complexes. A meshwork barrier composed of unfolded Nups could be maintained behind and in front of the Kap⅐cargo complexes as they displace Nups during facilitated transport, thereby preserving the diffusional barrier.
Finally, the intrinsic disorder of Nups may reconcile two seemingly contradictory observations: (i) the fast rate of karyopherin movement across the NPC (1-10 3 molecules/NPC/s); and (ii) the high binding affinity between certain Kaps and Nups (7,(55)(56)(57)(58). The former observation implies that Nup-Kap interactions must be transient (and therefore of low affinity) to support rapid rates of transport. However, in vitro studies with purified proteins have shown that Kap-Nup interactions can be of very high affinity with dissociation constants (K D values) in the nanomolar and even picomolar range (7,56). Assuming typical association rates for protein-protein interactions, the high affinity measured between Kaps and Nups predict slow 2 D. P. Denning et al., manuscript in preparation.
FIG. 6. Nup2p exhibits high sensitivity to proteinase K digestion. A, recombinant Kap60p (5 g) or Nup2p (5 g) were incubated at 37°C in the presence of 100 ng/ml proteinase K. At various time points, aliquots of the digestion were mixed with Laemmli sample buffer and phenylmethylsulfonyl fluoride to stop the digestion. Samples were then resolved by SDS-PAGE and stained with Coomassie Blue. B, recombinant Nup2p, Kap95p, and Kap60p (2 g each) were mixed and incubated at 37°C in the presence of 100 ng/ml proteinase K. At various time points, aliquots of the digestion were removed and processed as in A. C, fraction of full-length protein remaining in purified nuclei after digestion with 300 ng/ml proteinase K. Following digestion, aliquots of nuclei were resolved by SDS-PAGE, transferred to Immobulon-P, and probed with anti-Nup2p, Kap95p, Kap60p, Nup85p, or Pom152p antibodies and 125 I-Protein A. Each graph represents the average of three to six independent digestions; error bars represent the S.E. Note that Nup85p was partially clipped during the digestion; the top line in the graph represents total Nup85p (full-length and clipped), whereas the bottom line represents only full-length Nup85p. D, representative 125 I-Protein A Western blots used to generate graphs shown in C. Note that in the case of Nup85p, the full-length protein was clipped to yield a large, stable fragment (*) .   FIG. 7. Effect of heat treatment on the binding properties of Nup2p. Recombinant Nup2p (1 g) was heated to 90°C for 1 h or not and incubated with GST-Kap95p or GST-Nup60p (1 g each), which were immobilized on glutathione-Sepharose beads for 30 min at 4°C. Bead-associated (bound) and unbound Nup2p were collected, resolved by SDS-PAGE, and stained with Coomassie Blue. Note that heattreated Nup2p binds well to its Nup and Kap binding partners. dissociation rates that are incompatible with the rapid rates of nuclear transport. To explain the rapid flux of cargo through NPCs, it has been hypothesized that Gsp1p-GTP destabilizes Nup⅐Kap complexes (6). We suggest that structural disorder within Nups also facilitates fast transport rates. Natively unfolded proteins and their binding partners often have unexpectedly fast association and dissociation rates (39,53). Consistent with these observations, Kap95p⅐Kap60p heterodimers dissociate from the yeast FG nucleoporin Nup1p, a functional homolog of Nup2p (25), much faster than predicted for their high affinity of interaction (K D Յ 51 pM) (7). Kap95p⅐Kap60p heterodimers would be expected to dissociate from Nup1p with a half-life of minutes to hours assuming typical association rates (10 7 to 10 8 M Ϫ1 s Ϫ1 ) (54); yet, the measured half-life of Kap95p⅐ Kap60p⅐Nup1p complexes was Յ21 s and their association rate approached the limits set by diffusion (7). Thus, despite the high affinity between Nups and Kaps, their interactions may be transient as a result of fast association and dissociation reactions.
The mechanism of how Kaps move cargoes rapidly across the NPC remains a central question in the study of nucleocytoplasmic trafficking. As the physical and structural analyses of yeast Nups continue, it will be interesting to determine whether the disordered structure of nucleoporins is indeed intimately related to the mechanism of nuclear trafficking.