View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector Cell, Vol. 105, 245–255, April 20, 2001, Copyright 2001 by Cell Press Structural Basis for Guanine Nucleotide Exchange on by the Regulator of Condensation (RCC1)

Louis Renault,† Ju¨ rgen Kuhlmann, phenotype in the hamster cell line tsBN2, which shows Andreas Henkel, and Alfred Wittinghofer* premature chromosome condensation or arrest in the Max-Planck-Institut fu¨ r Molekulare Physiologie G1 phase of the cell cycle (Ohtsubo et al., 1987; for a Postfach 50 02 47 review see Seki et al., 1996). RCC1 homologs have since 44202 Dortmund been identified in many eukaryotes and mutations of Germany the in both S. pombe and S. cerevisiae show many different phenotypes, such as defects in mRNA trans- port and splicing, transport, chromatin deconde- Summary nsation, and cell cycle progression. The sequence of RCC1 had indicated an internal 7-fold repeat pat- RCC1 (regulator of chromosome condensation), a ␤ tern in the amino acid sequence and the three-dimen- ␤ propeller chromatin-bound protein, is the guanine nu- sional structure of RCC1 revealed a seven-bladed cleotide exchange factor (GEF) for the nuclear GTP propeller where the blades correspond to a 51–68 resi- ␤ binding protein Ran. We report here the 1.8 A˚ crystal due repeat that is different from the WD40 propeller structure of a Ran•RCC1 complex in the absence of motif (Renault et al., 1998). nucleotide, an intermediate in the multistep GEF reac- Biochemically, RCC1 is the GEF for Ran (Bischoff and tion. In contrast to previous structures, the phosphate Ponstingl, 1991) and increases guanine nucleotide dis- binding region of the nucleotide binding site is per- sociation by more than five orders of magnitude, just turbed only marginally, possibly due to the presence as the RanGAP stimulates the GTPase reaction of Ran of a polyvalent anion in the P loop. Biochemical experi- by the same magnitude (Klebe et al., 1995a). Since ments show that a sulfate ion stabilizes the Ran•RCC1 RCC1, via its N-terminal end, is bound to chromatin complex and inhibits dissociation by guanine nucleo- (Ohtsubo et al., 1987), and since RanGAP is confined to the cytoplasmic compartment of the cell, Ran•GTP tides. Based on the available structural and biochemi- concentration is high in the nucleus and low in the cyto- cal evidence, we present a unified scenario for the plasm. In interphase cells, this Ran•GTP gradient is be- GEF mechanism where interaction of the P loop lysine lieved to drive much of the two-way traffic of macromol- with an acidic residue is a crucial element for the over- ecules between the nucleus and the cytoplasm (for all reaction. review, see Gorlich and Kutay, 1999). In brief, macromo- lecular cargos to be transported through pores in the Introduction nuclear envelope associate with carriers which in turn bind to pore called nucleoporins. Proteins of The 24 kDa protein Ran (for Ras-related nuclear) belongs the importin ␤ (karyopherin-␤) superfamily constitute to the superfamily of GTP binding proteins that use a import (importin) and export (exportin) carriers and are structurally-conserved G domain to act as molecular Ran effectors, defined as proteins specifically recogniz- switches cycling between the GDP- and GTP-bound ing only the GTP-bound form of Ran (Fornerod et al., state (for reviews, see Bourne et al., 1990, 1991). Due to 1997; Gorlich et al., 1997). Whereas binding of Ran•GTP their low intrinsic hydrolysis and nucleotide dissociation to importins in the nucleus triggers release of imported rates, their cycling is controlled by two types of regulators. cargo, loading of export cargo onto export carriers re- GTPase-activating proteins (GAPs) increase the otherwise quires activated Ran. RanBP1 or RanBP2 belong to low intrinsic GTP hydrolysis, whereas guanine nucleotide a second set of Ran effectors defined by proteins exchange factors (GEFs) induce rapid dissociation of containing a Ran binding domain (RanBD). RanBD- bound nucleotide and thus allow fast activation to the containing proteins in the cytoplasm side destabilize GTP-bound form. For Ras and Ran, the rates of the exported cargo-carrier complexes by competing for nucleotide dissociation and GTP hydrolysis are similar, binding to Ran•GTP and by promoting hydrolysis of and are stimulated in both cases by similar orders of Ran•GTP by RanGAP1. Ran•GDP is recycled from the magnitude (Klebe et al., 1995a; Lenzen et al., 1998; cytoplasm into the nucleus by interacting with a Ran•GDP- Scheffzek et al., 1998). Although a significant amount specific transporter called NTF2 (nuclear transport fac- of biochemical and structural information has shed light tor 2). Questions remain about the role of RanBD pro- on numerous aspects of their function, the multistep teins and NTF2 in RCC1-catalyzed nucleotide exchange interaction between GTP binding proteins and their cog- on Ran and its implication for establishing a Ran•GTP nate GEFs is incompletely understood (Sprang and gradient. Coleman, 1998; Cherfils and Chardin, 1999). RCC1 (reg- Beyond this well characterized role of Ran in in- ulator of chromosome condensation) has been identified terphase cells, recent experiments on metaphase- as a gene that complements the temperature-sensitive arrested Xenopus egg extracts in which no nuclear membrane is present, have clearly implicated the Ran GTPase cycle in microtubule assembly and spindle for- * To whom correspondence should be addressed (e-mail: alfred. mation in M phase (Carazo-Salas et al., 1999; Kalab et [email protected]). † Present address: Laboratoire d’Enzymologie et Biochimie Struc- al., 1999; Ohba et al., 1999; Wilde and Zheng, 1999). turales, Baˆ t.34, CNRS, 1 Av. de la Terasse, 91 198 Gif-sur-Yvette Recently, Ran has been demonstrated to have yet an- Cedex, France. other role in the cell cycle, where the nuclear envelope Cell 246

assembly in Xenopus egg extracts requires the cycling mutagenesis data, and in terms of a unified mechanism of guanine nucleotide on Ran and is promoted by RCC1 by which GEFs such as RCC1 induce an increase in the (Hetzer et al., 2000; Zhang and Clarke, 2000). It thus dissociation rate. appears that Ran has a general role in the integration of cellular processes downstream of chromatin at different Results and Discussion stages of the nuclear and the cell cycle, and that high concentrations of Ran•GTP created by chromatin- Structure Determination and Overview bound RCC1 may constitute a general marker for the The nucleotide-free Ran•RCC1 complex crystallized in nucleus in interphase cells and for condensed or decon- space group P1 with two complexes per asymmetric densed chromatin during mitosis (Hetzer et al., 2000). unit and was solved to a resolution of 1.76 A˚ by the The reaction of nucleotide exchange catalyzed by Molecular Replacement method with the Ran•GDP•Mg2ϩ GEF is a multistep reaction which involves binary and (Scheffzek et al., 1995) and RCC1 (Renault et al., 1998) ternary complexes between GTP binding protein, gua- structures as search models. Data collection and refine- nine nucleotide, and GEF (Klebe et al., 1995b; Lenzen ment statistics are summarized in Table 1. The two com- et al., 1998; Wittinghofer, 1998). During the reaction, a plexes in the asymmetric unit show minor differences in ternary G protein•GEF•nucleotide complex of low nu- their structures, and only complex one will be discussed cleotide affinity is formed due to the mutual competition here. To distinguish RCC1 residues from Ran residues, of GEF and nucleotide (Klebe et al., 1995a; Lenzen et the latter are identified by a superscript. The final model al., 1998). This ternary complex relaxes into a binary G comprises residues 8R to 31R,37R to 177R (out of 216 protein•GEF complex that is stable in the absence of residues for wt human Ran), and residues 24 to 232 and nucleotide and reverts back to the binary G protein•nucle- 239 to 417 for RCC1 (out of 421). The model includes otide complex in the presence of nucleotide. The partial overall 475 water molecules and one polyanion per nu- kinetic steps of the overall reaction have been thoroughly cleotide-free Ran structure (Figure 1). Most of the Ran investigated for the Ran-RCC1 interaction, since this reac- residues from 125R to 143R are poorly defined in both tion does not occur on membranes or even require them complexes. to proceed either in vitro or in vivo. Here, it was shown The overall structure of the complex is illustrated in that the affinity of the nucleotide in the ternary complex Figure 1A. The interaction with Ran is indeed located on is reduced by five orders of magnitude from 10 pM to 1 the face of the RCC1 ␤ propeller opposite the chromatin ␮ M and that the rate of dissociation is increased by a binding site, as suggested previously (Renault et al., similar magnitude (Klebe et al., 1995b). Since RCC1 (Klebe 1998; Azuma et al., 1999). Superimposition with free et al., 1995b), Cdc25 (Lenzen et al., 1998) and, in all RCC1 indicates very few structural differences on com- likelihood, other GEFs work equally well in stimulating plex formation (rmsd 0.89 A˚ over 364 Ca atoms). The GDP or GTP dissociation, they act as catalysts to in- face of RCC1 interacting with Ran is fairly even except crease the rate at which equilibrium between G protein, for the ␤ wedge, a small prominently protruding ␤ sheet GEF and GDP/GTP, and other factors is achieved. in blade 3 (residues 146–153). Comparison between Ran At the structural level, determinants for the overall in the complex and isolated Ran•GDP show differences exchange reaction have been inferred from the struc- in switch II, ␣3/␣4, and around the NKxD base binding tures of GEF•G protein complexes of EF-Ts•EF-Tu (Ka- motif that are likely relevant to the GEF reaction (Figures washima et al., 1996; Wang et al., 1997), Sos1•Ras (Bori- 2A and 2C). Differences to the GTP-bound form found ack-Sjodin et al., 1998), Sec7•Arf1 (Goldberg, 1998), and in Ran•RanBD (Vetter et al., 1999a) or Ran•Importin ␤ recently, Tiam1•Rac1 (Worthylake et al., 2000). In the (Chook and Blobel, 1999; Vetter et al., 1999b) complexes first three binary complexes, the phosphate/Mg2ϩ bind- are in the same regions (Figures 2B and 2C). Ran in ing site of the nucleotide binding pocket is disturbed and the complex retains the Ran•GDP conformation with its inaccessible, whereas the nucleotide base and ribose extra ␤ strand (␤2 ) in switch I. Switch I is flexible in the binding site are unperturbed and freely accessible to E present structure since residues 32R to 36R are not visible nucleotide. This is achieved in many different ways—by direct intrusion of GEF residues, by displacing the switch and residues that are defined by the electron density I region, and by remodeling of switch II residues. For have high B factors. The C-terminal Ran residues from R Tiam1•Rac1, the nucleotide binding pocket is partially 178 onward are not visible. stabilized by a sulfate ion. The kinetic mechanism of the Ran-RCC1 reaction has Complex Interface been investigated in greater detail than of any other On complex formation, the two proteins bury a large -A˚ 2. Twenty 2700ف small G protein and the presence of a micromolar low solvent-accessible surface area of affinity Ran-RCC1-nucleotide complex has been docu- four residues of Ran and 25 of RCC1 are involved in mented (Klebe et al., 1995a; 1995b). Furthermore, ala- contact formation (using a cut-off limit of 3.5 A˚ ; Figure nine-scanning mutagenesis, in conjunction with bio- 3). The sites at which Ran interacts with its GEF are chemical experiments (Azuma et al., 1999), has identified concentrated around the P loop (19R to 20R), Switch II one face of the RCC1 ␤ propeller as the area interacting (67R to 76R), and helices ␣3 (93R to 110R) and ␣4 (134R, with the GTPase Ran opposite to the chromatin binding 137R, and 140R). Unlike other known GEF/small G protein face. Here, we report the 1.8 A˚ X-ray crystallographic complex structures (Boriack-Sjodin et al., 1998; Gold- structure of a complex formed between Ran and its GEF berg, 1998; Worthylake et al., 2000), switch I (residues RCC1. The structural details of the interaction are rather 29 to 46 in Ran) and neighboring residues are not partici- different from those observed in similar complexes and pating in the GEF interaction. However, in the GTP- will be discussed in the light of available kinetic and bound conformation, switch I would clash with RCC1. Ran•RCC1 Structure and Mechanism 247

Table 1. Crystallographic Data Statistics Data collections and phase determination by molecular replacement method. Crystal space group: P1 (a ϭ 50.334 A˚ b ϭ 71.447 A˚ c ϭ 77.729 A˚ ␣ϭ100.917Њ␤ϭ92.045Њ␥ϭ104.475Њ) Parameter Native constituted by merging data sets of 4 crystals Resolution (A˚ )* 20–1.63 (1.67–1.63) X-ray source BM30 and ID2, ESRF Wavelength (A˚ ) 0.9801 and 0.9903 Completeness (%)* 92.3 (56.1) Unique reflections* 118 194 (5 032) Redundancy* 4.1 (1.5) † Rsym (%)* 14.6 (31.6) † Rmeas (%)* 16.5 (44.7) I/␴* 6.8 (0.8) Refinement Statistics. Resolution (A˚ )* 20–1.76 Completeness* (%) 99.8 (1.79–1.76) (98.9) † Rwork (%)* 19.2 For 101 433 reflections (30.4) (for 4636 refl.) ‡ Rfree (%)* 22.3 For 8466 reflections (33.2) (for 432 refl.) Protein atoms 8500 Sulfate ions 2 No. of water molecules 475 Average B value for all atoms (A˚ 2) 2 Rans: 38.0 2 RCC1s: 32.0 Waters: ‡Rmsd between related-proteins Ran: 0.79/1.41 RCC1: 0.65/1.00 39.2 of the A.U. (A˚ ) §Rmsd 0.005 A˚ 1.3Њ 25.6Њ 0.73Њ || Ramachandran plot 88.1 10.3 0.7 0.9 * Values in parentheses correspond to the highest resolution shell. † † Rsym ϭ⌺h⌺i|I(h) Ϫ Ii(h)|/⌺h⌺iIi(h), where Ii(h) and I(h) are the i-th and mean measurements of the intensity of reflection h. Rwork ϭ⌺h|Fo Ϫ Fc|/

⌺hFo, where Fo and Fc are the observed and calculated structure factor amplitudes of reflection h. Rfree is the same as Rwork, but calculated on the 8.3% of data set aside from refinement. ‡ Root mean squared deviations for main chain atoms and for all protein atoms, respectively, between related-proteins of the two complexes of the asymmetric unit (A.U.). § Rms deviations for bond lengths (A˚ ), bond angles (Њ), torsion angles (Њ) and improper torsion angles (Њ), respectively. || Percentage of residues in most favored, additionally allowed, generously allowed and disallowed regions of the Ramachandran diagram, respectively.

All blades of the RCC1 ␤ propeller participate in the in Ran nucleotide complexes. The P loop, which is a Ran interaction. One key element of the interface, the circular structure that normally wraps around the ␤ protruding ␤ wedge, buries 423 A˚ 2 of accessible surface phosphate (Saraste et al., 1990), contains a polyanion upon complex formation. It interacts with P loop, switch (Figure 1), supporting the notion of it being a giant anion II, and helix ␣3. The geometry of the P loop is intact, hole (Dreusicke and Schulz, 1986). The anion is shifted potentially accessible to the nucleotide, and contains relative to the position normally occupied by the ␤ phos- a sulfate or phosphate anion (Figure 1B), presumably phate (1.2 A˚ ) such that its oxygens still contact main arising from the purification or crystallization buffer. chain nitrogen atoms with similar distances. A putative Switch II region represents the second most extensive anion in the ␤ phosphate binding site has previously part of the Ran interaction with 752 A˚ 2 of buried interface been found to stabilize the structure of Ras•guanosine and is a site of major rearrangements (Figure 2). The complexes (Scheffzek et al., 1994), and is also found in most extensive part of the interaction is provided by a the Rac-Tiam complex (Worthylake et al., 2000). A˚ 2 of solvent- The switch II region, as found in Ran•GDP or 1590ف region around ␣3 which buries accessible surface area in the complex. It contacts all Ran•GppNHp, would severely clash with RCC1. It is blades except blade 5, which interacts specifically with rearranged and involved in a number of interactions (see Switch II. Finally, the region around helix ␣4 makes a below). Large conformational changes are also located few contacts with blade 1 and 2 of RCC1. in the guanine base binding sites of Ran (Figures 2 and 4). Similar large changes in the base binding site have Structural Changes on Complex Formation also been observed for the Ras•Sos complex (Boriack- The ␤ wedge of RCC1 forms an apparently very rigid Sjodin et al., 1998). In the latter, the GEF does not inter- structure that is stabilized by a number of interactions act with any of these sites or neighboring residues, and between main and side chain groups. It shows a small their conformations seem to be a consequence of the overall shift of 1.1 to 1.8 A˚ on complex formation. Asn149 absence of nucleotide. In the RCC1•Ran complex, the and Asn150 and Gly19R and Gly20R from the P loop are GEF interaction with helix ␣4 and neighboring residues in close contact. (Figure 4). The P loop is shifted as a is close to the conserved base binding motif 122NKxD of rigid body toward the base binding sites such that the Ran, with a shift of 4.5 A˚ for residues 123–125R (Figures R C␣ of Gly19 moves by 1.4 A˚ , and by 1.7 A˚ the side chain 2 and 4). The extensive interface of the ␣3/␣4 region O atom of residue Thr24R, which coordinates the Mg2ϩ produces a number of significant changes in the struc- Cell 248

Figure 1. Overall View of the Ran•RCC1 complex (A) Ribbon diagram with RCC1 in gray and the ␤ wedge in green. Ran is yellow, with its switch I and II in light and dark blue, respectively, P loop in violet, and base binding motifs (122NKxD and 149SAK) in black. Figure 2. Structural Changes in Ran on Complex Formation The anion is green-red; nucleotide is not present but its virtual (A) Superimposition of Ran•GDP (red) with Ran from the Ran•RCC1 position in Ran•GDP is indicated as a dotted ball-and-stick model. complex (yellow). RCC1 is as in Figure 1. (B) Electron density simulated-annealing omit map, contoured at (B) Superimposition of Ran from the complex with Ran•GppNHp 1.1 sigma, of the active site region with the model as ball-and-stick. (green) on the left, and with Ran from the Ran•GppNHp•RanBD Green: RCC1, violet: Ran, and polyanion: yellow-red. The coordina- complex on the right, indicating that RCC1 and RanBD (red) can ˚ tion (cutoff distance of 3.25 A) of the polyanion with the P loop is bind simultaneously. shown by black dotted lines. (C) Rms deviation plot of Ran (complex 1) against Ran (complex 2) from the Ran•RCC1 complexes (black), Ran from Ran•GDP (red) and Ran•GppNHp from the complex with RanBD1 (green). Gray ture of Ran in this area. The protruding Lys99R on ␣3 diamonds on the X axis define Ran residues contacting RCC1 (cutoff ˚ would, upon interaction with RCC1, produce an exten- distance 3.5 A). sive steric clash with the 75GGMH motif on blade 2, which is the most highly conserved continuous motif in RCC1. After relocation, Lys99R becomes buried close to the (Chook and Blobel, 1999; Vetter et al., 1999a, 1999b) central tunnel of the RCC1 propeller and interacts with have shown that this C-terminal switch is detached from Asp128 and Asn181. Many other residues of ␣3 are in- the G domain and flexible in the GTP-bound state. In volved in the interface and experience structural the Ran•RCC1 complex, residues from 178R onward are changes. Lys130R preceding ␣4 shows a drastic change, disordered. In the region anchoring the C-terminal end and Val137R and Arg140R from ␣4 are directly involved onto the G domain, large conformational changes are in the interaction (Figures 3 and 4). observed compared to Ran•GDP and Ran•GTP (Figure 2) structures. Although RCC1 does not itself interfere The Ran C-Terminal Switch with the C terminus, its interaction with residues of the A difference of Ran with other members of the Ras ␤5/␣4 region such as the clash with Lys130 (Figure 4B), superfamily is the presence of a long C-terminal exten- repositions residues 129–142R such that they would ste- sion that ends in a conserved C-terminal DEDDDL motif. rically interfere with the C-terminal helix in Ran•GDP, In the Ran•GDP conformation (Scheffzek et al., 1995), leading to the release of the C terminus from the G it consists of a linker and a 16 residue ␣ helix, situated domain core. This suggests that RCC1 might have an opposite to the switch I region. Biochemical (Richards active role in inducing the C-terminal switch and is in et al., 1995; Hieda et al., 1999) and structural studies agreement with in vitro experiments showing that RCC1- Ran•RCC1 Structure and Mechanism 249

Figure 3. Contact Residues in the Complex Contacting residues (distance cutoff of 3.5 A˚ ) of Ran (A) and RCC1 (B) are shown as balls, black in the case of Ran; colored for RCC1 with a color code indicating the region of interaction of Ran as in Figure 1. Interacting residues of RCC1 are listed with their Ran contact partners (underneath). The virtual position of the nucleotide is indicated. catalyzed GDP release is significantly faster with a ⌬211 DEDDDL Ran mutant (Richards et al., 1996). Figure 4. Details from the Ran•RCC1 Interaction (A) The ␤ wedge and the rearrangement of Switch II induce a P loop Correlation with Mutational Data shift supported by ␣3, coloring as in Figure 2A. The Glu70R side The Ran•RCC1 structure is consistent with mutational from Ran•GDP structure (red) is reoriented (black arrow) toward the data (Klebe et al., 1995a; Azuma et al., 1996, 1999; Louns- helix ␣3, where steric clash (red star) induces a 180Њ flip of the R bury et al., 1996; Murphy et al., 1997). For Ran, only the Trp104 side chain such that its ring nitrogen atom interacts with the carboxylate, which is further stabilized by numerous contacts. G19V mutation has been found to inhibit the RCC1- The new position of Trp104R brings it too close to Asp18R of the P catalyzed exchange reaction (Lounsbury et al., 1996). loop. This, together with the action of the ␤ wedge, induces the P In the structure, G19R is in close contact with the ␤ loop to move by 1.3 A˚ . The lower part of (A) has been created by a wedge. Insertion of a valine side chain would provoke 90Њ rotation as indicated. steric clashes with both Asn149 and Asn150. Alanine (B) The base binding site of Ran in Ran•GDP (red) and the Ran•RCC1 scanning mutagenesis of RCC1 has identified four resi- complex (yellow), shown on the van der Waals surface of RCC1. The interaction of residues from ␣3 such as Lys99 contribute to dues that significantly reduce the activity (Azuma et al., tight binding with RCC1, while the clashes with ␣4 residues such 1996, 1999). While the D44A mutation only changes the as Lys130R and the 1.3 A˚ move of the nucleotide induce movement affinity of ternary or binary complex formation, D128A, of Asp125R, an important element for interaction with the guanine

D182A, and H304A significantly affect kcat. Asp44 in base. It also leads to clashes with the C terminus of Ran and induces blade 1 is ion pairing with the highly conserved Arg140R it to be released from the surface of the G domain. in helix ␣4, Asp128 ion pairs with the invariant Lys99R and forms an H bond with Asn181, whereas His304 interacts the importance of the protruding extra ␤ sheet of RCC1. with the invariant R320 and Lys71R of Switch II. Although The D182 residue stabilizes the base of the ␤ wedge by the D182A mutation is the most drastic, Asp182 is not making a hydrogen bond with the main chain NH of the directly in contact with Ran in the structure. Together invariant Arg147, which in turn is involved in contact with the effect of the Ran G19V mutation, it highlights with the Ran P loop (Figure 4A). Cell 250

Implications for the Exchange Mechanism Several structures of binary complexes between GEFs and GTP binding proteins in the absence of nucleotides are now available. The GEF-catalyzed guanine nucleo- tide exchange reaction is, however, a multistep kinetic process with additional—usually unstable—ternary G protein•nucleotide•GEF intermediates. The challenge therefore remains to relate the kinetic and mechanistic data with the structural snapshots in order to under- stand how the approach of GEF induces release of the nucleotide, and why GEF or nucleotide binding is antag- onistic. The rate-limiting step in the overall reaction most likely corresponds to one of the conformational changes in the ternary complex, which in turn changes the bind- ing affinity and thus the dissociation rate by several orders of magnitude (Klebe et al., 1995b; Lenzen et al., 1998; Wittinghofer, 1998). Usually, the trimeric complexes are of very low affin- ity—in the case of Ras•GDP•Cdc25, approximately 300 ␮M (Lenzen et al., 1998), in contrast to RCC1, where its affinity is in the micromolar range (Klebe et al., 1995b). Here, we have identified a sulfate or phosphate poly- anion bound in the phosphate binding site. It mimics the binding of the ␤ phosphate by forming strong hydrogen bonds with the P loop main chain nitrogens, and with Lys23R and Thr24R side chains. The P loop is basically intact. It has moved however, together with the bound anion, as a rigid body by about 1.3 A˚ relative to its position in Ran•nucleotide complexes. Previously, the absence of nucleotide was found to induce a collapse of the P loop such that it becomes unavailable for phos- phate binding. We thus believe our structure to corre- spond not to the binary nucleotide-free complex but rather to mimic an intermediate of the exchange reac- tion, carrying some features of a ternary low affinity G protein–GEF–nucleotide complex. We have reasons to believe that the presence of the anion is meaningful and that it is sulfate. When we analyze the effect of several polyanions on the overall rate of RCC1-catalyzed nucleotide exchange Figure 5. Influence of Polyanions on the Ran/RCC1 Reaction of Ran•mGDP against nonlabeled GDP and follow the (A and B) Fluorescence measurements on the effect of polyanions decrease in fluorescence intensity of the mant-group on the (A) overall rates of RCC1 catalyzed nucleotide exchange ␮ ␮ upon release from Ran in a stopped-flow device, we with 200 nM Ran•mGDP, 1 M RCC1, and 40 M GDP or (B) the dissociation of 200 nM Ran•RCC1 with 200 nM mGDP in an APP find that it is strongly inhibited by sulfate, but not by stopped-flow apparatus, at 20ЊC. mono- or polyphosphates (Figure 5A). To find out to (C) Dissociation of the GST-Ran•RCC1 complex in a BIAcore௣ dur- what this effect is due, a nucleotide-free complex be- ing incubation in control buffer (“X”s), in the presence of 20 mM tween Ran•RCC1 was formed, and an equimolar amount sodium sulfate (triangles) or 20 mM potassium phosphate (squares) of mGDP was added to dissociate the complex, by fol- and 100 mM NaCl (diamonds). All lanes were adjusted for the same lowing the increase in fluorescence, in the absence and SPR signal upon start of buffer injection. presence of these polyanions. The observed curves for different buffer compositions were fitted according to a first order rate equation and showed a significant reduc- of RCC1 to be the major determinant of the structural tion with 10 mM sodium sulfate, but not phosphate (Fig- changes that decrease the affinity of nucleotide. It is the ure 5B). Plasmon surface resonance measurements most exposed part of RCC1 and wedges itself between show that the binary Ran•RCC1 complex is stabilized residues of switch II and the P loop. The sequence of also in the absence of nucleotides by sulfate, whereas this element is only partially conserved as an 146FRxxxG phosphate is destabilizing. These experiments indicate motif. However, since only the main chain at the tip of that sulfate ion is stabilizing the nucleotide-free binary this ␤ wedge is in contact with the residues Gly19R and complex in the presence or absence of nucleotides, and Gly20R, only its structural make-up needs to be con- thus slows down the overall reaction. served. The invariant Arg147 is stabilized by Asp182 Whereas Ran shows drastic changes on complex for- and binds to Gly19R of the P loop, whereas Phe146 mation, RCC1 is rather used as a rigid scaffold that anchors the base of the protrusion into the hydrophobic barely changes its structure. For the mechanism of nu- core (Figure 4A). Superimposition with Ran•GDP would cleotide release, we would thus envision the ␤ wedge locate main and side chain oxygen atoms of residues Ran•RCC1 Structure and Mechanism 251

N149 and N150 between 3.3 and 4.3 A˚ from phosphates interaction, where it was proposed that the nucleotide and oxygen atoms of a GDP molecule located in the comes in by first contacting Ras via the base (Boriack- Ran nucleotide binding site (Figure 4). Sjodin et al., 1998; Wittinghofer, 1998). Pushing aside residues of switch II establishes a num- ber of contacts between residues Lys71R, Phe72R, A Generalized Mechanism for GEF Action Gly73R, and Arg76R with several nonconserved residues Since the structures of several complexes have now from blade 5, and the invariant Gln201 and His304 (Fig- been solved, the question of whether different GEFs ure 3). From the number of contacts made, residues work in a similar or totally different manner (Sprang and Lys71R and Phe72R appear to be the most important for Coleman, 1998; Cherfils and Chardin, 1999) can now 2ϩ the interaction. Mg in Ran•GDP is contacted, via water be answered with more confidence. While all the GEF R R R molecules, by residues Asp65 , Thr66 , and Glu70 , the structures are totally different and their complexes with latter two being displaced with switch II upon interaction cognate G proteins induce different conformational 2ϩ with RCC1. This may contribute to the release of Mg changes and use many different residues of the GTP and thus to the affinity of the nucleotide, since release of binding proteins, a common general mechanism is be- the metal induces a 460-fold increase of the nucleotide coming apparent. dissociation rate (Klebe et al., 1993). Mutational analysis The most important element of the binding affinity of indicates, however, that Glu70, at least, does not con- guanine nucleotides is supplied by the interaction of the tribute significantly to GDP affinity and intrinsic dissocia- ␤ phosphate with the P loop, since GMP has a 106 fold tion rate from Ran (A. H., unpublished data), and an lower affinity than GDP or GTP (Rensland et al., 1995). Asp57 (Asp65 in Ran) mutation in Ras increases GDP Another element of high affinity binding is Mg2ϩ coordi- dissociation only 7-fold (John et al., 1993), indicating nation, release of which causes a 500- to 1000-fold re- that switch II is no major determinant of nucleotide af- duction of affinity for Ras (Hall and Self, 1986) and Ran finity. (Klebe et al., 1995a), and the lysine residue of the P loop, The P loop provides a major contribution to the bind- mutation of which drastically reduces nucleotide affinity ing energies of the nucleotides (John et al., 1990; Rens- (Sigal et al., 1986; John et al., 1988; Klebe et al., 1993). land et al., 1995; Muegge et al., 1996). Although the P On the other hand, the relative orientation of the phos- loop itself is intact in the present structure, probably phate and base binding region is similarly important, due to the presence of the sulfate ion, it is pushed to the since polyphosphate alone has no measurable affinity side as a rigid body 1.3 A˚ from its position in Ran•GDP or and ADP/ATP an affinity that is similar to that of GMP Ran•GTP (which superimpose very well) toward the (Rensland et al., 1995). It would thus appear to be a base binding site. Although movement of 1.3 A˚ might good strategy for GEF to displace or modify the P loop seem unimportant, it would nevertheless push the gua- and/or to change the relative orientation of the base or nine base 1.3 A˚ toward the base binding site, clash- phosphate binding sites. ing with the invariant base binding residues such as If we overlay various G protein•GEF complexes (Fig- Asp125R and Ala 151R. This motion would displace resi- ure 6A), we get the impression that this is precisely the dues in this area, and correspondingly lead to a drastic strategy that GEFs use, and that a common feature of change in the affinity of nucleotide and a corresponding these complexes is the collapse of the P loop, after increase in the dissociation rate constant. As can be which it is stabilized by its invariant lysine interacting seen from Figures 2 and 4B, there are large displace- with a negatively charged residue. Inserting Phe82 from ments of base binding residues 122–125R and 149–152R. EF-Ts very close to a helix ␣3 residue (His119) of EF- These displacements are partly due to contacts with Tu leads to a series of rearrangements that ultimately RCC1, but probably also because they become mobile as an effect of the absence of the guanine nucleotide. induces a peptide flip of Ala20 in the P loop that sterically What are the implications for the mechanism of the blocks GDP binding (Kawashima et al., 1996; Wang et overall exchange reaction? We would envision that al., 1997). Incidentally, residue H119 from the G domain ␣ the transition from the present structure—which mimics of EF-Tu (helix 3 in Ran), which is modified by the the weakly bound nucleotide—to the binary nucleotide- insertion of Phe82 of EF-Ts, superimposes with Trp104 ␣ free complex would involve further changes in the P from Ran ( 3), which is modified by the Ran-Glu70 side loop after GDP (or sulfate ion) is released, in line with chain shift induced by RCC1 (Figure 4A). Lys24 from the previous structures of the empty binary complexes EF-Tu P loop is bound to Asp80 (Asp65 in Ran), an where the P loop collapses into a state which is unsuit- invariant residue normally involved in metal binding. In able for entry of the polyphosphate moiety of GDP/GTP. the Arf-Sec7 complex, the CO of Ala27 in the P loop is A conformation of the P loop unsuitable for nucleotide located in the position of the ␤ phosphate (Goldberg, binding is also found in the structure of the GTP binding 1998). An invariant glutamic acid residue, Glu97 in the ϩ proteins GBP-1 in the absence of nucleotides (Prakash Gea2 Sec7 domain, is close to the Mg2 and phosphate et al., 2000). For the reverse reaction, i.e., entry of GDP/ binding site and interacts with the Lys30, which is also GTP into the nucleotide binding site of the Ran•RCC1, stabilized by the Asp67, the residue homologous to we would expect, according to the microscopic revers- Asp80 in EF-Tu. In the Ras•Sos complex (Boriack-Sjodin ibility of the reaction pathway, that the phosphate bind- et al., 1998) Glu942 from Sos is pointing toward the ␤ ing part of the nucleotide comes in first and establishes phosphate position and the Gly13 main chain in the P contact with the P loop, which in turn leads to conforma- loop, analogous to Ala27 in Arf1, is flipped outwards. tional changes that establish tight nucleotide binding The conserved Lys16 is contacted by Glu62 from Ras and releases RCC1. This order of events would be in and by Glu62 from Rac in the Rac-Tiam complex (Wor- contrast to the mechanism proposed for the Ras–Sos thylake et al., 2000). This glutamic acid in switch II is Cell 252

the half-maximal saturation is unaltered (insert). This confirms the structural finding that Glu70, which induces the flop of the Trp104 side chain, is not involved in any contact with RCC1 in the present structure, but is nevertheless crucial for catalysis, just as Glu62 is crucial for the Ras-Cdc25 reaction (Mistou et al., 1992). This might also be the hitherto unknown reason for this resi- due to be conserved in most Ras-like proteins (Valencia et al., 1991), except for Arf, where the glutamic acid is supplied in trans by the GEF.

Implications for Nucleo-Cytoplasmic Transport Transport of Ran into the nucleus and its accumulation requires NTF2 and the exchange of Ran-bound GDP by RCC1 (Ribbeck et al., 1998; Smith et al., 1998). An over- lay of the Ran•GDP•NTF2 structure (Stewart et al., 1998) with the Ran•RCC1 complex shows that Ran can not bind simultaneously to both NTF2 and RCC1 without severe steric conflict in the switch II area. This is in full agreement with the finding that NTF2 inhibits RCC1- catalyzed GDP dissociation from Ran (Yamada et al., 1998). This implies that Ran•GDP needs to dissociate from NTF2, which is feasible considering the low affinity of the Ran•NTF2 complex (Chaillan-Huntington et al., 2000). Trimeric complexes between RCC1/nucleotide-free Ran/RanBP1-related proteins have been demonstrated by the two-hybrid system and isolated in vitro (Bischoff et al., 1995; Yokoyama et al., 1995; Noguchi et al., 1997; Mueller et al., 1998). The role of these interactions in the context of the Ran-induced nucleocytoplasmic transport is presently unclear. In vitro, RanBP1 inhibits Figure 6. Toward a Unified Mechanism for GEF action on GTP Bind- RCC1-induced nucleotide exchange activity of Ran•GTP ing Proteins and keeps the RCC1•Ran complex in a form more resis- (A) Superimposition of switch II and P loop from the Ras•Sos, tant to the dissociation by guanine nucleotides (Bischoff Arf•Sec7, EF-Tu•EF-Ts, and Rac•Tiam complexes with those from Ran•GDP and Ran•RCC1, with the indicated colors. Whereas the et al., 1995). P loop is unavailable for nucleotide binding in the first three com- Model building using the Ran•GppNHp•RanBD1 plexes and modified in the fourth, it is translocated sideways as a complex (Vetter et al., 1999a) and the present structure rigid body by 1.3 A˚ in Ran•RCC1 while leaving its structure intact. shows no overlapping between the binding interfaces The positions of acidic residues which contact the P loop lysine of RanBD1 and RCC1 (Figure 2B). Although RanBP-type (shown only for Lys23 of Ran) in the binary G protein•GEF complexes proteins were originally thought of as proteins with a and in Ran•GDP are indicated. (B) Stopped-flow kinetic analysis, using 0.2 ␮M wt (open circles) or purely cytoplasmic function, nuclear pools of RanBP E70A mutant (closed circles), Ran•mGDP, and increasing concen- proteins have been identified in both higher eukaryotes trations of RCC1 as indicated, and measuring the release of mGDP and yeast (Richards et al., 1996; Pasquinelli et al., 1997; in the presence of excess unlabeled GDP. The rate constants are Schlenstedt et al., 1997; Zolotukhin and Felber, 1997; analyzed as single exponentials and plotted as indicated. Kunzler et al., 2000). Furthermore RanBP3 and yeast Yrb2p are considered nuclear (Noguchi et al., 1997; highly conserved in many Ras-like proteins (not in Arf) Taura et al., 1998; Mueller et al., 1998). It thus evident for no obvious structural reasons. The P loop lysine thus that modulation of RCC1 activity by RanBP might well be always interacts with a negatively charged residue from relevant both for nuclear transport and for Ran function the GEF, or from the DxxGQe motif of Ras-like proteins. during mitosis. In the Ran•RCC1 complex, with the P loop still intact, Lys23 is detached from its normal position and is point- Conclusions ing toward Glu70 and Asp65 (Glu62 and Asp57 in Ras). It could thus easily be envisioned that Glu70 (or Asp65) Due to the presence of a sulfate ion in the P loop of the would be involved in stabilization once the P loop be- Ran•RCC1 complex the present structure is believed to comes empty. Stopped-flow kinetic experiments with be a mimic of the trimeric G protein•GEF•nucleotide the fluorescent mGDP analog show that the RCC1-medi- reaction intermediate of the multi-step GEF reaction with ated guanine nucleotide exchange reaction is severely a loosely bound nucleotide. The structure shows that a compromized in the Ran(E70A) mutation (Figure 6B). surface exposed ␤-wedge on RCC1 plays the decisive Single turnover reaction measurements in a stopped- role in catalysis which pushes aside the P loop as a flow apparatus show that the maximal rate of the reac- rigid body. The pathway of nucleotide release derived tion of 30 sϪ1 is much slower for the mutant but that from the structure is different from that of other GEFs Ran•RCC1 Structure and Mechanism 253

in detail. Using the comparison we propose however 1.63 A˚ resolution, respectively, and processed with XDS (Kabsch, that the general strategy of GEF consists in a two-sided 1993). The space group of the crystals is P1 and contains two attack to release positive charges, the Mg2ϩ ion and the Ran•RCC1 complexes per unit cell. Data sets from four crystals were merged together to obtain a reasonably complete and redun- invariant P loop lysine, from their interaction with the dant native data set and the statistics of the merged data are pre- phosphates of the nucleotide and that the structure of sented in Table 1. The initial phases were obtained by the molecular the collapsed P loop is stabilized by interactions of the replacement with the program AmoRe (Navaza and Saludjian, 1997) invariant lysine with acidic residues supplied in cis or with human Ran (Scheffzek et al., 1995) truncated from switch re- trans. Furthermore the three-dimensional structures gions and from the flexible C terminus (residues 29–47, 65–78, and suggest that parts of the nucleotide binding site such 208 to the end, respectively) and RCC1 (Renault et al., 1998) struc- tures as search model, resulting in a correlation of amplitudes of as switch I and the base binding area are flexible such 48.4% and a crystallographic R factor of 39.3% using the resolution that it is easily accessible for re-entry of the nucleotide. range from 10 to 3.1 A˚ . Nine percent of the reflections were set The structure of the Ran•RCC1•sulfate complex sug- aside for an R free test before initiating any refinement. The program gests that the base is released first and the phosphate O (Jones et al., 1991) was used for model building and CNS (ver. last, and the nucleotide re-enters with the phosphate 0.5 and 1.0) (Bru¨ nger et al., 1998) with bulk solvent correction used moiety coming in first. Considering the high concen- for automated refinement. tration of Ran in the cell (Bischoff et al., 1995) and Acknowledgments the micromolar dissociation constant of the trimeric Ran•RCC1•GDP complex (Klebe et al., 1995b), it might We thank Doro Ku¨ hlmann, A. Zeipert, C. Nowak and C. Koerner for well be that this is the prevalent species of Ran in the excellent technical assistance; M. Hess for graphical support; M. nucleus. Roth, P. Carpentier, and J.L. Ferrer for access and support at the ESRF beamline BM30 (Grenoble), B. Rasmussen at the ESRF beam- Experimental Procedures line ID02; A. Perrakis at the microfocus EMBL and ESRF beamline ID13; and B. Prakash, J. Becker, I. Vetter, R. Goody, and S. Szedla- Protein Preparation ceck for discussions. L. R. is supported by an EMBO fellowship Human RCC1 and RCC1⌬1–19 (deletion of the RCC1 N-terminal (ALTF 264-1999). A. H. is supported by DFG Schwerpunkt-pro- residues 1 to 19) was expressed from ptac vector in E. coli strain gram 312. CK600K, and purified as described by (Klebe et al., 1993). Expres- sion and isolation of recombinant human wild-type Ran purified in References complex with GDP and Mg2ϩ was as described earlier (Klebe et al., 1993). GST-Ran was cloned in a pGEX-2T vector, expressed in E. Azuma, Y., Seino, H., Seki, T., Uzawa, S., Klebe, C., Ohba, T., Witting- coli BL21, and purified over a glutathione sepharose column. hofer, A., Hayashi, N., and Nishimoto, T. (1996). Conserved histidine A nucleotide-free complex between Ran and RCC1 was obtained residues of RCC1 are essential for nucleotide exchange on Ran. J. Њ by shortly incubating the two purified proteins at 4 Cin25mMKPO4 Biochem. 120, 82–91. (pH 6.5), 25 mM ammonium sulfate, 10 mM EDTA, and 3 mM DTE Azuma, Y., Renault, L.J.A., Valencia, A., Nishimoto, T., and Witting- fold molar excess of Ran, prior to loading onto a-1.4ف using a hofer, A. (1999). Model of the ran-RCC1 interaction using biochemi- Superdex 75 gel exclusion column equilibrated with the same buffer. cal and docking experiments. J. Mol. Biol. 289, 1119–1130. Fractions containing the heterodimeric complex were pooled and concentrated to 90 mg/ml into the previous buffer but with only 2 Bischoff, F.R., and Ponstingl, H. (1991). Catalysis of guanine nucleo- mM EDTA. tide exchange on Ran by the mitotic regulator RCC1. Nature 354, 80–82. Biophysical Experiments Bischoff, F.R., Krebber, H., Smirnova, E., Dong, W., and Ponstingl, The basic experimental setup for biophysical analysis with fluores- H. (1995). Co-activation of RanGTPase and inhibition of GTP dissoci- cence and surface plasmon resonance (SPR) has been described ation by Ran-GTP binding protein RanBP1. EMBO J. 14, 705–715. before (Kuhlmann et al., 1997). Kinetics with mGDP were measured Boriack-Sjodin, P.A., Margarit, S.M., Barsagi, D., and Kuriyan, J. in a SX16MV stopped-flow system (Applied Photophysics, APP). (1998). The structural basis of the activation of ras by sos. Nature Single turnover kinetics were done using 0.2 ␮M Ran•mGDP in 394, 337–343. HEPES buffer (10 mM HEPES [pH 7.4], 5 mM MgCl , and 150 mM 2 Bourne, H.R., Sanders, D.A., and McCormick, F. (1990). The GTPase NaCl), in the presence of varying amounts of RCC1 and polyanions, superfamily: a conserved switch for diverse cell functions. Nature with 40 ␮M unlabeled GDP. Surface plasmon resonance experi- 348, 125–132. ments were measured in a BIAcore௣ 1000 system (BIAcore௣). We used a sandwich-assay with anti-GST-sera (BIAcore௣) as described Bourne, H.R., Sanders, D.A., and McCormick, F. (1991). The GTPase previously (Lenzen et al., 1998) to immobilize GST-fusion proteins. superfamily: conserved structure and molecular mechanism. Nature All experiments were performed in HEPES buffer which was supple- 349, 117–127. mented with 0.005% (w/v) Igepal in the SPR measurement and the Bru¨ nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., additional salts as indicated. Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., et al. (1998). Crystallography & NMR system: A new Crystallographic Analysis software suite for macromolecular structure determination. Acta An initial split Ran•RCC1 crystal was obtained at 20ЊC by the vapor Crystallogr. 54, 905–921. diffusion method with hanging drops after two months from a reser- Carazo-Salas, R.E., Guarguaglini, G., Gruss, O.J., Segref, A., Kar- voir solution containing 20% PEG3350 (polyethylene glycol), 200 senti, E., and Mattaj, I.W. (1999). Generation of GTP-bound Ran by mM KF (Hampton Peg/ion screen). From this crystal, a microseeding RCC1 is required for chromatin-induced mitotic spindle formation. procedure into preequilibrated drops of 12.5%–10% PEG4000, 200 Nature 400, 178–181. mM KF, and 3 mM DTE led to suitable crystals for X-ray analysis, Chaillan-Huntington, C., Braslavsky, C.V., Kuhlmann, J., and Stew- albeit in a single crystallization setup. Crystals grew within 1.5 to 3 art, M. (2000). Dissecting the interactions between NTF2, RanGDP, months to a maximal size of 0.1 ϫ 0.07 ϫ 0.05 mm3 but could not and the nucleoporin XFXFG repeats. J. Biol. Chem. 275, 5874–5879. be reproduced afterward. All data sets were collected at ESRF (Grenoble, France) at 100 K Cherfils, J., and Chardin, P. (1999). GEFs: structural basis for their by cryoprotecting crystals into 30% glycerol, 17.5% PEG4000, 200 activation of small GTP-binding proteins. Trends Biochem. Sci. 24, mM KF, and 3 mM DTE. Several data sets were collected on a MAR- 306–311. imaging plate detector at BM30 and ID2B beamlines to 2.1 and Chook, Y.M., and Blobel, G. (1999). Structure of the nuclear transport Cell 254

complex karyopherin-beta 2-Ran center dot GppNHp. Nature 399, Lounsbury, K.M., Richards, S.A., Carey, K.L., and Macara, I.G. 230–237. (1996). Mutations within the Ran/TCV4 GTPase—effects on regula- Dreusicke, D., and Schulz, G.E. (1986). The glycine-rich loop of ade- tory factor interactions and subcellular localization. J. Biol. Chem. nylate kinase forms a giant anion hole. FEBS Lett. 208, 301–304. 271, 32834–32841. Fornerod, M., Vandeursen, J., Vanbaal, S., Reynolds, A., Davis, D., Mistou, M.Y., Jacquet, E., Poullet, P., Rensland, H., Gideon, P., Murti, K.G., Fransen, J., and Grosveld, G. (1997). The human homo- Schlichting, I., Wittinghofer, A., and Parmeggiani, A. (1992). Muta- logue of yeast crm1 is in a dynamic subcomplex with can/nup214 tions of Ha-ras p21 that define important regions for the molecular and a novel nuclear pore component nup88. EMBO J. 16, 807–816. mechanism of the SDC25 C-domain, a guanine nucleotide dissocia- Goldberg, J. (1998). Structural basis for activation of arf — tion stimulator. EMBO J. 11, 2391–2397. mechanisms of guanine nucleotide exchange and gtp-myristoyl Muegge, I., Schweins, T., Langen, R., and Warshel, A. (1996). Electro- switching. Cell 95, 237–248. static control of GTP and GDP binding in the oncoprotein p21(ras). Gorlich, D., and Kutay, U. (1999). Transport between the cell nucleus Structure 4, 475–489. and the cytoplasm. Ann. Rev. Cell Dev. Biol. 15, 607–660. Mueller, L., Cordes, V.C., Bischoff, F.R., and Ponstingl, H. (1998). Gorlich, D., Dabrowski, M., Bischoff, F.R., Kutay, U., Bork, P., Hart- Human RanBP3, a group of nuclear RanGTP binding poteins. FEBS mann, E., Prehn, S., and Izaurralde, E. (1997). A novel class of rangtp Lett. 427, 330–336. binding proteins. J. Cell Biol. 138, 65–80. Murphy, G.A., Moore, M.S., Drivas, G. Perez, d.l.O., Villamarin, A., Hall, A., and Self, A.J. (1986). The effect of Mg2ϩ on the guanine D’Eustachio, P., and Rush, M.G. (1997). A T42A Ran mutation: differ- nucleotide exchange rate of p21N-ras. J. Biol. Chem. 261, 10963– ential interactions with effectors and regulators, and defect in nu- 10965. clear protein import. Mol. Biol. Cell 8, 2591–2604. Hetzer, M., Bilbao-Cortes, D., Walther, T.C., Gruss, O.J., and Mattaj, Navaza, J., and Saludjian, P. (1997). AMoRe: An automated molecu- I.W. (2000). GTP hydrolysis by Ran is required for nuclear envelope lar replacement package. Methods Enzymol. 276, 581–594. assembly. Mol. Cell 5, 1013–1024. Noguchi, E., Hayashi, N., Nakashima, N., and Nishimoto, T. (1997). Hieda, M., Tachibana, T., Yokoya, F., Kose, S., Imamoto, N., and YRB2P, a NUP2P-related yeast protein, has a functional overlap Yoneda, Y. (1999). A monoclonal antibody to the COOH-terminal with RNA1P, a yeast Ran-GTPase-activating protein. Mol. Cell. Biol- acidic portion of Ran inhibits both the recycling of Ran and nuclear ogy 17, 2235–2246. protein import in living cells. J. Cell Biol. 144, 645–655. Ohba, T., Nakamura, M., Nishitani, H., and Nishimoto, T. (1999). Self- John, J., Frech, M., and Wittinghofer, A. (1988). Biochemical proper- organization of microtubule asters induced in Xenopus egg extracts ties of Ha-ras encoded p21 mutants and mechanism of the autophos- by GTP-bound Ran. Science 284, 1356–1358. phorylation reaction. J. Biol. Chem. 263, 11792–11799. Ohtsubo, M., Kai, R., Furuno, N., Sekiguchi, T., Sekiguchi, M., Haya- John, J., Sohmen, R., Feuerstein, J., Linke, R., Wittinghofer, A., shida, H., Kuma, K., Miyata, T., Fukushige, S., Murotsu, T., et al. and Goody, R.S. (1990). Kinetics of interaction of nucleotides with (1987). Isolation and characterization of the active cDNA of the hu- nucleotide-free H-ras p21. Biochemistry 29, 6058–6065. man cell cycle gene (RCC1) involved in the regulation of onset of John, J., Rensland, H., Schlichting, I., Vetter, I., Borasio, G.D., Goody, chromosome condensation. Genes Dev. 1, 585–593. R.S., and Wittinghofer, A. (1993). Kinetic and structural analysis of Pasquinelli, A.E., Powers, M.A., Lund, E., Forbes, D., and Dahlberg, ϩ the Mg(2 )-binding site of the guanine nucleotide-binding protein J.E. (1997). Inhibition of mrna export in vertebrate cells by nuclear p21H-ras. J. Biol. Chem. 268, 923–929. export signal conjugates. Proc. Natl. Acad. Sci. USA 94, 14394– Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, S. (1991). Im- 14399. proved methods for binding protein models in electron density maps Prakash, B., Praefcke, G.J.K., Renault, L., Wittinghofer, A., and Herr- and the location of errors in these models. Acta Crystallogr. 47, mann, C. (2000). Structure of human guanylate-binding protein 1 110–119. representing a unique class of GTP-binding proteins. Nature 403, Kabsch, W. (1993). Automatic processing of rotation diffraction data 567–571. from crystals of initially unknown symmetry and cell constants. J. Renault, L., Nassar, N., Vetter, I., Becker, J., Klebe, C., Roth, M., and Appl. Crystallogr. 26, 795–800. Wittinghofer, A. (1998). The 1.7 A crystal structure of the regulator of Kalab, P., Pu, R.T., and Dasso, M. (1999). The Ran GTPase regulates chromosome condensation (RCC1) reveals a seven-bladed propel- mitotic spindle assembly. Curr. Biol. 9, 481–484. ler. Nature 392, 97–101. Kawashima, T., Berthet-Colominas, C., Wulff, M., Cusack, S., and Rensland, H., John, J., Linke, R., Simon, I., Schlichting, I., Witting- Leberman, R. (1996). The structure of the Escherichia coli EF-Tu•EF- hofer, A., and Goody, R.S. (1995). Substrate and product structural Ts complex at 2.5 A˚ resolution. Nature 379, 511–518. requirements for binding of nucleotides to h-ras p21—the mecha- Klebe, C., Nishimoto, T., and Wittinghofer, F. (1993). Functional ex- nism of discrimination between guanosine and adenosine nucleo- pression in Escherichia coli of the mitotic regulator proteins p24ran tides. Biochemistry 34, 593–599. and p45rcc1 and fluorescence measurements of their interaction. Ribbeck, K., Lipowsky, G., Kent, H.M., Stewart, M., and Gorlich, D. Biochemistry 32, 11923–11928. (1998). NTF2 mediates nuclear import of Ran. EMBO J. 17, 6587– Klebe, C., Bischoff, F.R., Ponstingl, H., and Wittinghofer, A. (1995a). 6598. Interaction of the nuclear GTP-binding protein Ran with its regula- Richards, S.A., Lounsbury, K.M., and Macara, I.G. (1995). The C tory proteins RCC1 and RanGAP1. Biochemistry 34, 639–647. terminus of the nuclear RAN/TC4 GTPase stabilizes the GDP-bound Klebe, C., Prinz, H., Wittinghofer, A., and Goody, R.S. (1995b). The state and mediates interactions with RCC1, RAN-GAP, and HTF9A/ kinetic mechanism of Ran—nucleotide exchange catalyzed by RANBP1. J. Biol. Chem. 270, 14405–14411. RCC1. Biochemistry 34, 12543–12552. Richards, S.A., Lounsbury, K.M., Carey, K.L., and Macara, I.G. Kuhlmann, J., Macara, I., and Wittinghofer, A. (1997). Dynamic and (1996). A nuclear export signal is essential for the cytosolic localiza- equilibrium studies on the interaction of Ran with its effector, tion of the Ran binding protein, RanBP1. J. Cell Biol. 134, 1157–1168. RanBP1. Biochemistry 36, 12027–12035. Saraste, M., Sibbald, P.R., and Wittinghofer, A. (1990). The P loop—a Kunzler, M., Gerstberger, T., Stutz, F., Bischoff, F.R., and Hurt, E. common motif in ATP- and GTP-binding proteins. Trends Biochem. (2000). Yeast Ran-binding protein 1 (Yrb1) shuttles between the Sci. 15, 430–434. nucleus and cytoplasm and is exported from the nucleus via a CRM1 (XPO1)-dependent pathway. Mol. Cell. Biology 20, 4295–4308. Scheffzek, K., Kabsch, W., Schlichting, I., Pai, E.F., Lautwein, A., Lenzen, C., Cool, R.H., Prinz, H., Kuhlmann, J., and Wittinghofer, A. Frech, M., Wittinghofer, A., and Goody, R.S. (1994). Crystallization (1998). Kinetic analysis by fluorescence of the interaction between and preliminary X-ray structure analysis of thermally unstable p21H- Ras and the catalytic domain of the guanine nucleotide exchange ras guanosine complexes. Acta Crystallogr. 50, 521–526. factor Cdc25Mm. Biochemistry 19, 7420–7430. Scheffzek, K., Klebe, C., Fritz-Wolf, K., Kabsch, W., and Wittinghofer, Ran•RCC1 Structure and Mechanism 255

A. (1995). Crystal structure of the nuclear Ras-related protein Ran in its GDP-bound form. Nature 374, 378–381. Scheffzek, K., Ahmadian, M.R., and Wittinghofer, A. (1998). GTPase- activating proteins: helping hands to complement an active site. Trends Biochem. Sci. 23, 257–262. Schlenstedt, G., Smirnova, E., Deane, R., Solsbacher, J., Kutay, U., Gorlich, D., Ponstingl, H., and Bischoff, F.R. (1997). YRB4P, a yeast Ran-GTP-binding protein involved in import of ribosomal protein L25 into the nucleus. EMBO J. 16, 6237–6249. Seki, T., Hayashi, N., and Nishimoto, T. (1996). RCC1 in the Ran pathway. J. Biochem. 120, 207–214. Sigal, I.S., Gibbs, J.B., D’Alonzo, J.S., Temeles, G.L., Wolanski, B.S., Socher, S.H., and Scolnick, E.M. (1986). Mutant ras-encoded pro- teins with altered nucleotide binding exert dominant biological ef- fects. Proc. Natl. Acad. Sci. USA 83, 952–956. Smith, A., Brownawell, A., and Macara, I.G. (1998). Nuclear import of Ran is mediated by the transport factor NTF2. Curr. Biol. 8, 1403– 1406. Sprang, S.R., and Coleman, D.E. (1998). Invasion of the nucleotide snatchers—structural insights into the mechanism of G protein GEFs. Cell 95, 155–158. Stewart, M., Kent, H.M., and Mccoy, A.J. (1998). Structural basis for molecular recognition between nuclear transport factor 2 (NFT2) and the GDP-bound form of the Ras-family GTPase Ran. J. Mol. Biol. 277, 635–646. Taura, T., Krebber, H., and Silver, P.A. (1998). A member of the ran- binding protein family, yrb2p, is involved in nuclear protein export. Proc. Natl. Acad. Sci. USA 95, 7427–7432. Valencia, A., Chardin, P., Wittinghofer, A., and Sander, C. (1991). The ras protein family: evolutionary tree and role of conserved amino acids. Biochemistry 30, 4637–4648. Vetter, I.R., Nowak, C., Nishimoto, T., Kuhlmann, J., and Witting- hofer, A. (1999a). Structure of a Ran-binding domain complexed with Ran bound to a GTP analogue: implications for nuclear trans- port. Nature 398, 39–46. Vetter, I.R., Arndt, A., Kutay, U., Gorlich, D., and Wittinghofer, A. (1999b). Structural view of the Ran-Importin beta interaction at 2.3 A˚ resolution. Cell 97, 635–646. Wang, Y., Jiang, Y.X., Meyeringvoss, M., Sprinzl, M., and Sigler, P.B. (1997). Crystal structure of the EF-Tu.Ef-Ts complex from Thermus thermophilus. Nat. Struct. Biol. 4, 650–656. Wilde, A., and Zheng, Y.X. (1999). Stimulation of microtubule aster formation and spindle assembly by the small GTPase Ran. Science 284, 1359–1362. Wittinghofer, A. (1998). Ras signalling: Caught in the act of the switch-on. Nature 394, 317–320. Worthylake, D.K., Rossman, K.L., and Sondek, J. (2000). Crystal structure of Rac1 in complex with the guanine nucleotide exchange region of Tiam1. Nature 408, 682–688. Yamada, M., Tachibana, T., Imamoto, N., and Yoneda, Y. (1998). Nuclear transport factor p10/NTF2 functions as a Ran-GDP dissoci- ation inhibitor (Ran-GDI). Curr. Biol. 8, 1339–1342. Yokoyama, N., Hayashi, N., Seki, T., Pante´ , N., Ohba, T., Nishii, K., Kuma, K., Hayashida, T., Miyata, T., Aebi, U., et al. (1995). A giant nucleopore protein that binds Ran/TC4. Nature 376, 184–188. Zhang, C.M., and Clarke, P.R. (2000). Chromatin-independent nu- clear envelope assembly induced by Ran GTPase in Xenopus egg extracts. Science 288, 1429–1432. Zolotukhin, A.S., and Felber, B.K. (1997). Mutations in the nuclear export signal of human ran-binding protein block the rev- mediated posttranscriptional regulation of human immunodefi- ciency virus type 1. J. Biol. Chem. 272, 11356–11360.

Protein Data Bank ID Code

The atomic coordinates of the Ran•Rcc1 complex described in this paper have been deposited with the under code 1I2M.