A 2.1-Å-Resolution Crystal Structure of Unliganded CRM1 Reveals the Mechanism of Autoinhibition

A 2.1-Å-Resolution Crystal Structure of Unliganded CRM1 Reveals the Mechanism of Autoinhibition

Natsumi Saito1 and Yoshiyuki Matsuura1,2 1 - Division of Biological Science, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8602, Japan 2 - Structural Biology Research Center, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8602, Japan

Correspondence to Yoshiyuki Matsuura: Division of Biological Science, Graduate School of Science, Nagoya University, Furo-cho, Chikusa-ku, Nagoya 464-8602, Japan. [email protected] http://dx.doi.org/10.1016/j.jmb.2012.11.014 Edited by I. Wilson

Abstract CRM1 mediates nuclear export of numerous proteins and ribonucleoproteins containing a leucine-rich nuclear export signal (NES). Binding of RanGTP to CRM1 in the nucleus stabilizes cargo association with CRM1, and vice versa, but the mechanism underlying the positive cooperativity in RanGTP and NES binding to CRM1 remains incompletely understood. Herein we report a 2.1-Å-resolution crystal structure of unliganded Saccharomyces cerevisiae CRM1 (Xpo1p) that demonstrates that an internal loop of CRM1 (referred to as HEAT9 loop) is primarily responsible for maintaining the NES-binding cleft in a closed conformation, rendering CRM1 incapable of NES binding in the absence of RanGTP. The structure also shows that the C-terminal tail of CRM1 stabilizes the autoinhibitory conformation of the HEAT9 loop and thereby reinforces autoinhibition. Comparison with the structures of CRM1–NES–RanGTP complexes reveals how binding of RanGTP is associated with a series of allosteric conformational changes in CRM1 that lead to opening of the NES-binding cleft, allowing for stable binding of NES cargoes. © 2012 Elsevier Ltd. All rights reserved.

Introduction Macromolecular exchange between the nucleus and the cytoplasm is an essential cellular process in all eukaryotes and occurs through the nuclear pore complexes (NPCs) embedded in the nuclear envelope. Most transport events through the NPCs are mediated by multiple families of soluble transport receptors. 1,2 The cargo macromolecules bind to specific transport receptors in either the cytoplasm or the nucleus and are then translocated through NPCs, after which the transport receptors release their cargoes and are recycled to the original compartment to participate in another transport cycle. The largest class of the nuclear transport receptors is the family of importin-β-like transport factors designated as karyopherins, which can be classified into two types, importins and exportins, depending on the directionality of transport. Importins carry cargoes to the nucleus, whereas exportins carry cargoes to the cytoplasm.

The small GTPase Ran regulates karyopherin– cargo interactions and the directionality of karyopherin-mediated transport. Like other Ras-family GTPases, Ran cycles between GTP- and GDPbound states, and the two surface loops in Ran, referred to as the switch I and switch II loops, undergo significant conformational changes between the GTP- and GDP-bound states. 3 In addition, Ran has a C-terminal extension that is disordered in the GTP-bound state but folds back against the body of Ran as an α-helix in the GDP-bound state. 3 The conformations of these three regions (switch I, switch II, and the C-terminal extension) that are sensitive to the nucleotide state of Ran are important in determining its interactions with other proteins. 4 The low intrinsic rates of nucleotide exchange and hydrolysis on Ran are stimulated by specific factors in vivo. Ran GTPase activity is stimulated by the order of 10 5 by RanGAP, 5 whereas nucleotide exchange is stimulated by the Ran guanine nucleotide exchange factor, RCC1. 6 Based on the

0022-2836/$ - see front matter © 2012 Elsevier Ltd. All rights reserved.

J. Mol. Biol. (2013) 425, 350–364

Autoinhibition of CRM1

localizations of RanGAP in the cytoplasm and RCC1 in the nucleoplasm, cytoplasmic Ran is primarily in the GDP-bound state whereas nucleoplasmic Ran is kept primarily in the GTP-bound state. This gradient of Ran nucleotide state is an important determinant of the directionality of nuclear transport. 7 In nuclear import, RanGTP competes with the cargoes to bind to importins, allowing the cargo binding in the cytoplasm and RanGTP-mediated cargo dissociation in the nucleus. By contrast, in nuclear export, RanGTP and cargoes bind cooperatively to exportins in the nucleus, and the exportin–RanGTP–cargo complexes are disassembled in the cytoplasm, where RanGTPase is activated by RanGAP and the Ran-binding proteins RanBP1/2. Thus, the association and dissociation of karyopherin–cargo complexes are regulated by direct binding of Ran in a compartment-specific manner. CRM1 (also known as exportin 1 or Xpo1) is the most versatile exportin that facilitates nuclear export of a broad range of cargoes. 8–11 The majority of the export substrates of CRM1 contain a short peptide sequence (10–15 residues), the so-called leucinerich nuclear export signal (NES) that was first identified in cAMP-dependent protein kinase inhibitor (PKI) 12 and HIV-1 Rev. 13 Leucine-rich NESs typically harbor four or five characteristically spaced hydrophobic residues that are crucial for the binding to CRM1. Although the NESs show considerable sequence diversity, 14,15 recent structure determination of CRM1–cargo complexes with and without RanGTP 16–18 showed that at least three NESs (leucine-rich NES of snurportin, PKI, and Rev) share the ability to bind specifically to the same site: a hydrophobic cleft of CRM1. CRM1 is a ringshaped molecule that is constructed from 21 tandem HEAT repeats, each of which consists of two antiparallel α-helices, designated A-helix and Bhelix, connected by loops of varying length. The A-helices form outer convex surface whereas the Bhelices form the inner concave surface. The hydrophobic cleft on the convex outer surface of CRM1, formed between the A-helices of HEAT repeats 11 and 12, constitutes the NES-binding site. The hydrophobic side chains of NESs fit into five hydrophobic pockets along this cleft. The binding site of Leptomycin B, a potent inhibitor of CRM1mediated nuclear export, 9 is located in this hydrophobic cleft, 19 and so it is likely that this cleft is the general binding site for NESs. CRM1 is unusual among karyopherins in that it has a cargo-binding site on its outer surface (instead of its inner surface), 4,20–23 but this is important for CRM1 to carry a broad range of cargoes that vary greatly in size and shape, including huge cargoes such as ribosomal subunits. In contrast to the cargoes that bind to the outer surface of CRM1, RanGTP binds to the inner surface of CRM1. 17,18 Four distinct regions of CRM1 contribute to RanGTP binding, and CRM1

351 directly binds to both switch I and switch II loops of RanGTP, accounting for the ability of CRM1 to discriminate between GTP- and GDP-bound Ran. Interestingly, the C-terminal α-helix of CRM1 (the Chelix) adopts dramatically different positions depending on whether or not RanGTP is bound to the CRM1–cargo complexes. 16,17 In the binary CRM1–snurportin complex, the C-helix lies across the central cavity of the CRM1 ring with its Cterminus located close to the NES-binding site, whereas in the ternary CRM1–snurportin–RanGTP complex, the C-helix is located on the outer surface of the CRM1 ring, indicating that the C-terminus of CRM1 could regulate the affinity of NES in a way that is sensitive to RanGTP. In the cytoplasm, the ternary CRM1–cargo– RanGTP complex is disassembled by the action of Ran-binding proteins RanBP1/2 and RanGAP. A recent kinetic study showed that the Ran-binding domains (RanBDs) of the cytoplasmic proteins RanBP1 and RanBP2, but not RanGAP, accelerate dissociation of NES from CRM1 and RanGTP by over 2 orders of magnitude. 24 Crystal structure of yeast CRM1–RanBP1–RanGTP complex showed that the NES- and RanBD-binding sites on CRM1 are distinct and that the binding of RanBD induces the movement of a long internal loop in HEAT repeat 9 (referred to as HEAT9 loop), from switch I of RanGTP to the concave side of the NES-binding cleft (the inner surface of HEAT repeats 11 and 12 of CRM1), driving rotations and translations of the α-helices constituting the NESbinding site. This results in closure of the hydrophobic cleft to dissociate NES. 24 Structure-based mutagenesis of crucial hydrophobic residues of the HEAT9 loop provided strong support for this allosteric mechanism of NES release and also indicated that the HEAT9 loop functions as an allosteric autoinhibitor to stabilize CRM1 in a conformation that is unable to bind NES cargo in the absence of RanGTP. 24 Mutational analyses also indicated that the C-terminus of CRM1 plays an important role in stabilizing CRM1 in an autoinhibited state in the absence of RanGTP, 25,26 and so it has been proposed that the HEAT9 loop and the C-terminus of CRM1 cooperate to stabilize the NES-binding cleft in a closed state in the absence of RanGTP. 24,26 Nevertheless, the absence of a high-resolution structure of CRM1 in isolation meant that the structural basis for autoinhibition remained obscure. Here, we describe a 2.1-Å-resolution crystal structure of unliganded CRM1 that provides definitive structural data to establish the precise mechanism of autoinhibition of CRM1. This structure provides direct evidence that the NES-binding cleft of the unliganded CRM1 adopts a closed conformation, which is stabilized by the binding of the HEAT9 loop to the inner surface of HEAT repeats 11 and 12 in exactly the same way as observed in the CRM1– RanBP1–RanGTP complex. Moreover, our new

352

Autoinhibition of CRM1

structure shows that the C-terminus interacts with the inner surface of CRM1 beneath the NES-binding site in such a way that stabilizes the HEAT9 loop in the autoinhibitory conformation. Thus, the structure reported here highlights the crucial role of the HEAT9 loop as the primary determinant of the conformation of the NES-binding cleft and also reveals how the autoinhibitory function of the HEAT9 loop is reinforced by the C-terminus. Comparison with the structure of CRM1–cargo– RanGTP complexes reveals how binding of RanGTP induces movement of both the HEAT9 loop and the C-terminus of CRM1, enabling the transition of the NES-binding cleft from the closed state to the open state.

Results and Discussion Crystallization and structure determination of unliganded CRM1 To understand the structural basis for autoinhibition of CRM1, we attempted to crystallize full-length

CRM1 from various species including Homo sapiens, Saccharomyces cerevisiae, and Schizosaccharomyces pombe. Although extensive attempts to crystallize full-length, wild-type CRM1 were unsuccessful, optimization of the S. cerevisiae CRM1 (Xpo1p) construct yielded diffraction-quality crystals suitable for high-resolution structure determination. We employed a strategy to delete unnecessary surface loops and protrusions that might hinder formation of well-ordered crystals. Care was taken to engineer the construct for crystallization without compromising the functionality of CRM1. In a previous structural study of the CRM1–RanBP1– RanGTP complex, we used a functional deletion mutant of S. cerevisiae CRM1, in which residues 377–413 (a long, poorly conserved nonessential loop connecting HEAT repeats 8 and 9) are deleted. 24 Hereafter, S. cerevisiae, mouse, and human CRM1 will be abbreviated as yCRM1, mCRM1, and hCRM1, respectively. This deletion mutant of yCRM1, when crystallized in isolation, yielded only clusters of needle crystals not suitable for X-ray data collection. Therefore, we went on to make a double deletion mutant of yCRM1, in which

Table 1. Data collection and refinement statistics Crystal

Native

SeMet

Data collection Space group Unit cell dimensions a, b, c (Å) α, β, γ (°)

P41

P41

116.8, 116.8, 119.0 90, 90, 90

Wavelength (Å) Resolution (Å)a Total observations Unique reflections Completeness (%)a Rmergea Mean I/σa Multiplicitya

1.00000 48.3–2.1 (2.21–2.10) 1,191,624 93,097 99.8 (99.8) 0.11 (0.77) 14.4 (2.4) 12.8 (6.9)

116.9, 116.9, 119.8 90, 90, 90 Inflection 0.97927 53.3–2.4 (2.53–2.40) 467,833 62,790 98.0 (92.6) 0.102 (0.616) 12.7 (2.4) 3.8 (3.1)

Refinement No. of reflections Rcryst/Rfree (%) No. of atoms Protein Water Average B-factors (Å2) Protein Water RMSD from ideality Bond lengths (Å) Bond angles (°) Protein geometryb Rotamer outliers (%) Ramachandran outliers (%) Ramachandran favored (%) Cβ deviations N 0.25 Å Residues with bad bonds (%) Residues with bad angles (%) a b

88,349 18.7/21.7 7802 378 63 51 0.016 1.7 2.41 0 96.49 1 0 0.10

Values in parentheses are for the highest-resolution shell. MolProbity was used to analyze the structure.50

Peak 0.97896 53.3–2.4 (2.53–2.40) 468,538 62,795 98.2 (93.5) 0.088 (0.501) 14.3 (3.0) 3.8 (3.1)

Remote 0.96405 53.3–2.4 (2.53–2.40) 467,359 62,754 98.0 (93.1) 0.109 (0.664) 11.9 (2.3) 3.8 (3.1)

353

Autoinhibition of CRM1

residues 377–413 and residues 971–984 are deleted. Residues 971–984 of yCRM1 constitute a loop connecting HEAT repeats 19 and 20 that forms a protruding bulge on the outer surface of yCRM1 (Supplementary Fig. S1a). A pull-down assay showed that the double deletion mutant [yCRM1 (Δ377–413/Δ971–984)] is functional in vitro and retains the ability to bind RanGTP in a NESdependent manner (Supplementary Fig. S1b). This double deletion mutant [yCRM1 (Δ377–413/Δ971– 984)] readily crystallized. Crystals grew as thick rods to a length of 1.0 mm and a width and thickness of 0.2 to 0.3 mm in 1 week. These crystals belonged to the space group P41 with one molecule in the asymmetric unit and diffracted to 2.1 Å resolution using synchrotron radiation at SPring-8 beamline BL41XU (Table 1). The structure was solved by selenomethionine (SeMet) multi-wavelength anomalous diffraction (MAD) phasing method and was refined to a crystallographic R-factor of 18.7% (Rfree = 21.7%) at 2.1 Å resolution (Table 1). In the final round of the crystallographic refinement, translation/libration/screw (TLS) refinement was performed to analyze the dynamic properties of CRM1. Systematic variation of the number of TLS

(a) yCRM1

(b) mCRM1

H11

groups showed that Rfree decreased as progressively more TLS groups were added, and the minimum Rfree was obtained when CRM1 was partitioned into 14 TLS groups, with the size of each TLS group being smaller than two HEAT repeats (Supplementary Fig. S2). This indicates that CRM1 is a relatively continuously flexible molecule analogous to importin-β, 27 rather than being constructed from a small number of rigid subdomains connected by flexible hinges. The refined model consists of yCRM1 residues 47–56, 64–264, 271–376, 414–970, and 985–1084 and 378 water molecules. The N-terminal third of yCRM1 had a higher B-factor than the remainder of the molecule (Supplementary Fig. S3), and HEAT repeat 1 (residues 1–46) was not modeled because the electron density was not strong enough to trace the polypeptide chain unambiguously. Overall structure of unliganded CRM1 The HEAT repeats of unliganded yCRM1 are organized into a ring-shaped molecule with the Cterminal helix (the C-helix) lying across the ring, reminiscent of the HEAT repeat architecture of hCRM1 in the binary hCRM1–snurportin complex

RanGTP

snurportin ti

(c)

hCRM1

snurportin

H12 H13

H14 H21

H15

H18

H20

H16 H17

H19 80° H8

H9

H10 H11

80° H12

80°

NES

HEAT9 loop p

NES

H7 H6 H5 H4 H3 switch I H1 H2

switch II

Fig. 1. Overall view of the CRM1 structures. (a) Unliganded yCRM1 [this study; Protein Data Bank (PDB) code 3VYC]. (b) mCRM1–snurportin (an NES cargo)–RanGTP complex (PDB code 3GJX). 17 (c) hCRM1–snurportin complex (PDB code 3GB8). 16 The HEAT repeats of CRM1 are labeled H1–H21. CRM1 is colored yellow, except that the HEAT9 loop, HEAT repeats 11 and 12 (that constitute the NES-binding site), and the C-terminal region beyond the A-helix of HEAT repeat 21 are highlighted in magenta, orange, and green, respectively. Ran is colored cyan, with its switch I and switch II loops highlighted in pink and gray, respectively. Snurportin is colored light gray, with its leucine-rich NES at the N-terminus highlighted in purple. Shown in stick representation in purple are the hydrophobic side chains of NES that fit into the hydrophobic pockets along the NES-binding cleft formed between the outer helices of HEAT repeats 11 and 12.

354 (Fig. 1a and c). 16 The C-helix was not involved in crystal contacts, indicating that this distinctive conformation of the C-terminal segment is not influenced by crystal packing forces. The largest dimension (Dmax) and radius of gyration (Rg) calculated from the atomic coordinates of unliganded yCRM1 are 10.7 nm and 3.7 nm, respectively. These values are in agreement with a previous small-angle X-ray scattering study 28 that reported Dmax and Rg values of 11 ± 1 nm and 3.9 ± 0.1 nm, respectively, for unliganded yCRM1 in solution. Thus, the crystal structure is consistent with the low-resolution structural information of unliganded yCRM1 in solution.

Autoinhibition of CRM1

(a)

HEAT12

HEAT11

F583 M556

Mechanism of autoinhibition of CRM1 The high-resolution structure of unliganded yCRM1 allows us to describe precisely how intramolecular interactions stabilize CRM1 in the lowaffinity form for NES binding in the absence of RanGTP. As previously predicted by Koyama and Matsuura based on mutational analyses, 24 the NESbinding cleft is closed in unliganded yCRM1 (Fig. 2), and the HEAT9 loop binds to the inner surface of HEAT repeats 11 and 12 that constitute the NESbinding site (Fig. 3a), burying a surface area of 673 Å 2, in exactly the same way as observed in the structure of the yCRM1–RanBP1–RanGTP complex. 24 The conformation of the HEAT9 loop and the HEAT repeats 11 and 12 of unliganded yCRM1 is essentially identical with that of the yCRM1–RanBP1–RanGTP complex with an allatom root-mean-square deviation (rmsd) of only 1.1 Å (see Supplementary Fig. S4 for superposition of the two structures). In this conformation, as described previously, 24 particularly important interactions are made by the highly conserved hydrophobic residues in the HEAT9 loop (Val441, Leu442, Val443, and Ile451 of yCRM1 that are equivalent to Val430, Leu431, Val432, and Val440 of mCRM1; see Supplementary Fig. S5 for sequence alignment) that cluster together and pack intimately against the nonpolar patch on the inner surface of CRM1. These four hydrophobic residues bury 48% of the interaction interface between the HEAT9 loop and the inner surface of CRM1, forming a hydrophobic core at the central region of the interface. This indicates that this hydrophobic interaction makes substantial energetic contribution to the binding of the HEAT9 loop to the inner surface of CRM1. The intimate packing of hydrophobic side chains at this interface is enabled by the side chains of Met594 (on 12B helix) and Met556 (on 11B helix) of yCRM1 (Met583 and Met545 of mCRM1) pointing towards the HEAT9 loop and adopting conformations different from the NES-bound form (Fig. 2a). As a consequence, the NES-binding cleft between helices 11A and 12A is closed in unliganded CRM1 (Fig. 2b).

M594

(b) Cleft closed

Cleft open NES

HEAT12 HEAT11

(Unliganded yCRM1)

HEAT11

HEAT12

(mCRM1-snurportin -RanGTP complex)

Fig. 2. The NES-binding cleft of unliganded CRM1 adopts a closed conformation. (a) Superposition of HEAT repeats 11 and 12 of unliganded yCRM1 (yellow) and the mCRM1–snurportin–RanGTP complex (orange). Arrows indicate the conformational changes of the NES-binding site when snurportin (an NES cargo) and RanGTP bind to CRM1. (b) Surface representation of HEAT repeats 11 and 12 (left, unliganded yCRM1; right, mCRM1–snurportin– RanGTP complex). The residues that directly interact with NES are white, whereas the other residues are yellow. The N-terminal NES of snurportin is shown in purple.

The structure of unliganded CRM1 also reveals how the C-terminus of CRM1 is involved in autoinhibition (Fig. 3a). In this structure, the C-helix terminates in a loop (residues 1073–1084; hereafter referred to as the C-terminal tail) that has welldefined electron density until the very C-terminus of yCRM1 (Supplementary Fig. S6). The C-terminal tail interacts with the HEAT9 loop and also with the Bhelices of HEAT repeats 8–12, burying a surface area of 660 Å 2. The interaction interface can be divided into three regions, referred to as regions 1 to 3, as marked by dashed circles in Fig. 4a. In region 1, four consecutive nonpolar amino acid residues (Ile1073, Gly1074, Gly1075, and Leu1076) make a

355

Autoinhibition of CRM1

tight turn and dock into a nonpolar depression on the inner surface of HEAT repeats 8–10, formed by Phe374, Tyr375, Val473, Tyr474, His477, and Gly517 (Figs. 3a and 4a). Essentially the same interaction was observed in the hCRM1–snurportin complex (Figs. 3c and 4b), suggesting that the interactions in region 1 can be formed irrespective of whether the NES-binding cleft is open or closed. By contrast, the interactions in region 2 strongly influence the conformation of the NES-binding cleft and thus have clear implications on the role of the Cterminal tail in the autoinhibition of CRM1. Intriguingly, Lys1078 in region 2 makes salt bridges with two consecutive acidic residues (Glu439 and

(a)

Glu440) that are immediately adjacent to the crucial hydrophobic residues (Val441, Leu442, Val443, and Ile451) in the HEAT9 loop (Fig. 3a; see also Supplementary Fig. S6). Glu440 is also hydrogen bonded to S1080 and K601 via a water molecule (Fig. 3a; see also Supplementary Fig. S6). The conformations of Glu439 and Glu440 (equivalent to Glu428 and Glu429 of mCRM1) change dramatically upon cooperative binding of NES and RanGTP binding 17 (Fig. 3b). In the mCRM1–snurportin– RanGTP complex, Glu428 of mCRM1 forms a salt bridge with Lys590 of mCRM1, and Glu429 of mCRM1 forms a hydrogen bond with Tyr155 of Ran (Fig. 3b). In terms of these conformational NES

(b)

D447

yCRM1

hCRM1

mCRM1

V441

L442

NES

(c)

D436

Y155

E429

E440

E429

K37

K601

E439

E428

K1078 S1080 V473

K605

V462

K594 59

D1084

Y474

F1060

Y463

F374

V1056 I1074 Y375

Y381

NES

C-helix

C helix C-helix HEAT12

90° 90

90° 90

HEAT11

HEAT12

HEAT11

90° 90

NES HEAT12

HEAT11

F583 M545

M592

V441 1 V443

F572

F572

M545

M556

K590

M583

M583 E429

E428

L442

V430

HEAT9 loop E428

HEAT9 lloop

E429 D447

HEAT9 loop

V432

RanGTP

L431 Y1 Y155

C helix C-helix

h li C-helix C D436 K37

Fig. 3. Atomic details of the interactions involved in the control of the transition between the closed state and the open state of the NES-binding site of CRM1. (a) Unliganded yCRM1. (b) mCRM1–snurportin–RanGTP complex. (c) hCRM1– snurportin complex. Only HEAT repeats 8–10 (yellow), HEAT repeats 11–12 (orange), and the C-terminal region (the Chelix and the C-terminal tail, green) of CRM1 are shown for clarity. HEAT repeats 11 and 12 constitute the NES-binding site, and the NES at the N-terminus of snurportin is shown in purple in (b) and (c). HEAT9 loop is colored magenta. Ran is shown in cyan, with bound GTP shown as space-filling spheres. The cyan dotted lines indicate hydrogen bonds or salt bridges. A cyan sphere in (a) represents a water molecule. See also Supplementary Fig. S6 for electron density map of unliganded yCRM1 around the C-terminal tail.

356

Autoinhibition of CRM1

(a) E439 E440

region 2

K601

K1078 S1080 K605

L1076 G1075

D1084

region 3 Y375

I1073

region 1

yCRM1 CRM1

(b)

N1061 I1059

V1056

region 1

hCRM1

changes, it is noteworthy that the large movement of the HEAT9 loop when snurportin and RanGTP bind to CRM1 can be described as an ~ 100° rotation of residues 443–454 as a pseudo rigid body (C α rmsd of 0.75 Å), around the hinge region formed by residues 438–442 and 455–459 (Supplementary Fig. S7). It therefore seems likely that the extensive interactions in region 2 lock the conformation of the crucial hinge residues and thereby prevent the HEAT9 loop from adopting the NES- and RanGTPbound conformation. The residues of the C-terminal tail in region 3, on the other hand, do not interact directly with the HEAT9 loop. Instead, in region 3, the main-chain carbonyl oxygen atoms of Pro1079 and Leu1082 make hydrogen bonds with N ζ of K605 (Fig. 3a; Supplementary Fig. S6). K605 also makes a

Fig. 4. The electrostatic surface potential of the inner surface of HEAT repeats 8–12 is shown together with the interacting C-terminal tail (stick model in green) and the C-helix (green C α -trace) of CRM1. (a) Unliganded yCRM1. (b) hCRM1–snurportin complex. Solvent-accessible surface of CRM1 is colored by electrostatic potential: positively charged regions are colored blue, negatively charged regions are colored red, and neutral regions are colored white. The color ramp is from − 0.3 to + 0.3 V. The electrostatic potential was calculated using CCP4MG. 45 Marked by dashed circles are the surface areas involved in the interactions with the C-terminal tail. The location of several residues labeled in Fig. 3 are indicated. In region 1, nonpolar residues of the C-terminal tail fit into the hydrophobic pocket on the inner surface of HEAT repeats 8–10 [essentially the same interactions are observed in both (a) and (b)]. In region 2, the C-terminal tail directly interacts with the HEAT9 loop via salt bridges and hydrogen bonds, and this is possible only when the HEAT9 loop adopts the autoinhibitory conformation. In region 3, the C-terminal tail binds to a lysine residue of HEAT repeat 12 via hydrogen bonds and a salt bridge. The interactions in regions 1 and 3 anchor the C-terminal tail on the CRM1 inner surface and this would stabilize the interactions in region 2 and hence reinforce autoinhibition of CRM1.

salt bridge with Asp1084 (Fig. 3a; Supplementary Fig. S6). However, these interactions with K605 are probably not sufficient to regulate the opening or closing of the NES-binding cleft, because the side chain of this lysine (K594 in mCRM1) is exposed to solvent in the mCRM1–snurportin–RanGTP complex and adopts the same position and orientation as observed in the structure of unliganded yCRM1 (Fig. 3b). Therefore, as similarly argued for region 1, it is unlikely that the interactions in region 3 alone are sufficient to stabilize the closed conformation of the NES-binding cleft. Rather, it appears that the roles of the interactions in regions 1 and 3 are to anchor the C-terminal tail on the CRM1 inner surface and hold the key residues in region 2 (Lys1078 and Ser1080) at the appropriate positions for the interactions with

357

Autoinhibition of CRM1

the HEAT9 loop, so that the interactions in region 2 are stabilized and hence autoinhibition is strengthened. Taken together, the structure of unliganded yCRM1 and its comparison with the structures of CRM1–snurportin complexes with and without RanGTP suggest that the HEAT9 loop is primarily responsible for autoinhibition of NES binding and that the C-terminal tail strengthens the autoinhibitory function of the HEAT9 loop by stabilizing the HEAT9 loop in the appropriate conformation for the autoinhibitory interactions with the inner surface beneath the NES-binding cleft. The involvement of the C-terminal tail in stabilizing CRM1 in the autoinhibited state is consistent with previous mutational and biochemical studies, which showed that deletion of C-terminal residues beyond A1026 of hCRM1 (G1043 of yCRM1) or beyond H1063 of mCRM1 (S1080 of yCRM1) results in increase in the affinity of CRM1 for NESs in the absence of RanGTP. 25,26 To examine the contribution of key residues more specifically, we engineered point mutants of yCRM1. The yCRM1 mutation K1078A substantially increased the binding of yCRM1 to PKI (an NES cargo) in the absence of RanGTP (Fig. 6a). The yCRM1 mutation K605A also increased the binding to PKI in the absence of RanGTP, but the effect of K605A mutation was

1040

:::

C-helix 1060

1050

|

|

|

..

*:

milder than that of K1078A mutation (Fig. 6a). Thus, the mutational analyses support the idea that the interactions in region 2 are crucially important for autoinhibition, with the interactions in region 3 contributing secondarily. We also verified the functional role of the interactions in region 1 by engineering a point mutation G1075A in yCRM1. This glycine intimately packs against a nonpolar pocket on the CRM1 inner surface and there is no room to accommodate a larger side chain of this residue at this interface, explaining why this glycine is absolutely conserved (Fig. 5). Therefore, substitution of this glycine with any other amino acid should substantially disrupt the binding of the Cterminal tail to the CRM1 inner surface and this would in turn destabilize the interactions between the HEAT9 loop and the C-terminal tail, leading to weakening of autoinhibition. Indeed, a pull-down assay showed that the G1075A mutant of yCRM1 bound to PKI strongly even in the absence of RanGTP (Fig. 6a). The fact that the G1075A mutation was more effective in relieving autoinhibition than the K605A mutation implies that the free energy librated by the interactions in region 1 is greater than that in region 3. Although two of the key residues for autoinhibition, Lys1078 and Ser1080 of yCRM1, in the C-terminal

: :

1070 |

1080 |

: * :. .*

Fig. 5. Amino acid sequence alignment of the C-terminal residues of CRM1. Residue numbers of S. cerevisiae CRM1 are shown above the sequence alignment. The C-helix of unliganded yCRM1 is represented by a green cylinder. The amino acid residues are colored according to their physicochemical properties as follows: aliphatic/hydrophobic (I, L, V, A, M), pink; aromatic (F, W, Y), orange; basic (K, R, H), blue; acidic (D, E), red; hydrophilic (S, T, N, Q), green; conformationally special (P, G), magenta; cysteine (C), yellow. Hyphens indicate the positions of gaps in the multiple alignment. Asterisks, colons, and periods below the sequence alignment denote absolutely conserved residues, highly conserved residues, and moderately conserved residues, respectively. Sequences were aligned using ClustalW, 48 and its output was visualized with the program Jalview. 49

358

Autoinhibition of CRM1

(a) wild type

K1078A

K605A

G1075A RanGTP RanGDP - yCRM1 y

- GST-PKI - RanGTP 1

2

3

(b) wild type

4

5

7

6

8

9

10 11 12

N1061A/ H1063A RanGTP R GTP RanGDP - hCRM1

Mechanism of cooperative RanGTP and cargo binding to CRM1

- GST GST-PKI PKI - RanGTP 1

2

3

4

5

6

Fig. 6. Mutational analyses of (a) yCRM1 and (b) hCRM1. Bound proteins in the GST pull-down assays were visualized by SDS-PAGE and Coomassie staining. (a) Immobilized 47 μg GST-PKI was incubated with 100 μg yCRM1 with and without 50 μg Ran. (b) Immobilized 50 μg GST-PKI was incubated with 78 μg hCRM1 with and without 50 μg Ran. Ran was charged with GTP or GDP as indicated.

(a)

tail, are not absolutely conserved (Fig. 5), there is always a basic residue in either of these two positions and the other residue is almost always a hydrophilic residue (Fig. 5), and so the way the Cterminal tail contributes to the autoinhibitory mechanism may be conserved universally. For instance, Lys1078 and Ser1080 of yCRM1 are equivalent to Asn1061 and His1063 of hCRM1, and these two residues of hCRM1 could form hydrogen bonds to Glu428 and Glu429 of hCRM1 (Glu439 and Glu440 of yCRM1) either directly or via a water molecule and thereby stabilize the autoinhibitory conformation of the HEAT9 loop in hCRM1. Consistent with this hypothesis, the binding of hCRM1 to PKI in the absence of RanGTP was substantially increased by N1061A/H1063A mutations in hCRM1 (Fig. 6b).

RanGTP

When RanGTP and snurportin bind CRM1 to form a ternary nuclear export complex, RanGTP binds extensively to the inner surface of CRM1 at four distinct binding surfaces: HEAT repeats 1–5, HEAT repeats 7–8, HEAT9 loop, and HEAT repeats 17– 19. 17 Although the overall dimensions of the toroidal structures of unliganded yCRM1 and mCRM1 in complex with snurportin and RanGTP appear similar (Fig. 1a and b), a 6.6-Å C α rmsd suggests that the binding of RanGTP to all of the four binding areas requires substantial conformational changes of CRM1. The most conspicuous conformational changes, vital for the RanGTP-induced release from autoinhibition, occur in the HEAT9 loop and the C-terminal region beyond the A-helix of HEAT21.

(b)

RanGTP

clash l h

yCRM1

yCRM1

Fig. 7. Superposition of unliganded yCRM1 and the mCRM1–snurportin–RanGTP complex suggests that the binding of RanGTP induces movement of the C-terminal region of CRM1 by steric clash. Unliganded yCRM1 is shown in ribbon representation (green, the C-terminal helix and the C-terminal tail; the other regions, yellow). Shown in (a) is the position of RanGTP when unliganded yCRM1 and the mCRM1–snurportin–RanGTP complex are superposed at HEAT repeat 3, whereas in (b), the position of RanGTP when the two structures are superposed at HEAT repeat 19 is shown. Ran is shown in ribbon representation, with bound GTP shown as space-filling spheres. In (a), RanGTP does not clash with anywhere of CRM1, whereas in (b), RanGTP severely clashes with the C-helix of CRM1.

Autoinhibition of CRM1

Fig. 8. The gap between the N- and C-terminal regions of CRM1 closes upon movement of the C-terminal region associated with the binding of RanGTP and cargo. Molecular surface representation of (a) unliganded yCRM1 and (b) mCRM1 in the mCRM1–snurportin– RanGTP complex. (c) A superposition between unliganded yCRM1 (green) and mCRM1 (magenta) in the mCRM1–snurportin–RanGTP complex. The two structures were superposed at HEAT repeat 20. Orientation is the same as in (a) and (b).

359 The HEAT9 loop moves from the inner surface behind the NES-binding site to the switch I loop of RanGTP, and, in the cargo- and RanGTP-bound form, the C-helix is located on the outer surface of CRM1 (instead of lying across the CRM1 ring). When unliganded yCRM1 is superposed with the mCRM1–snurportin–RanGTP complex at HEAT19, RanGTP severely clashes with the C-helix of unliganded yCRM1 (Fig. 7b). Thus, the autoinhibited structure of unliganded yCRM1 is incompatible with the binding of RanGTP to HEAT17–19, and the movement of the C-helix to the outer surface of CRM1 is required to alleviate the steric clash and allow for access of RanGTP to the inner surface of HEAT17–19. This movement of the C-helix is associated with displacement of the C-terminal tail away from the NES-binding site, and this abolishes the interactions between the HEAT9 loop and the Cterminal tail that stabilize the autoinhibitory conformation of the HEAT9 loop, suggesting that the movement of the C-helix would facilitate the conformational transition of the HEAT9 loop. Thus, the binding of RanGTP is coupled with the movement of the two structural elements (the HEAT9 loop and the C-terminal tail), and this would in turn destabilize the closed conformation of the NES-binding cleft, allowing for opening of the cleft and binding of cargo. In addition to the large movement of the C-helix, the intimate association of RanGTP with all of the binding surfaces on the inner surface of CRM1 requires a series of sequential conformational changes along the 21 HEAT repeats, closing the gap between the N- and C-terminal regions of CRM1 in the ternary export complex (Fig. 8). A close examination of the contacts between the N- and Cterminal regions suggests that the movement of the C-helix facilitates the ring closure (Fig. 9). In the mCRM1–snurportin–RanGTP complex, the loop connecting the 21A-helix and the C-helix (referred to as the 21B-helix by Monecke et al. 17) binds to the loop connecting the A- and B-helices of HEAT repeat 2. Furthermore, in the ternary complex, the loop connecting HEAT repeats 4 and 5 binds to HEAT repeat 21. These direct contacts between the N- and C-terminal regions of CRM1 are possible only when the C-helix is on the outer surface of CRM1, and the interactions between the N- and C-terminal regions of CRM1 result in the closure of the CRM1 ring in the ternary complex (Fig. 8b). Thus, the movement of the C-helix facilitates the closure of the CRM1 ring, which would be important to stabilize the RanGTP binding, because it is only when the CRM1 ring is closed that RanGTP can bind all of the four binding surfaces (HEAT repeats 1–5, HEAT repeats 7–8, HEAT9 loop, and HEAT repeats 17–19) simultaneously. We examined the functional significance of the closure of the CRM1 ring by structure-based mutagenesis. In the unliganded yCRM1, D1034 on the 21A-helix is exposed to solvent (Fig. 9a), but in

360

Autoinhibition of CRM1

(a)

(c)

T175

wild type D1034A L1050D RanGTP RanGDP - yCRM1

4B 3A

3B

Q176

yCRM1

D1034

F1051 R1033

V1037

- GST-PKI

L1050 Y1049 L1036

- RanGTP

F952

20B

1 2 3 4 5 6

7 8 9

19B 18B

19A

(b) 4B

3B 3A 5A

mCRM1 C

RanGTP T187

Q188 D1017 21A

L1033

Q1021 19B 20B

18B

19A

Fig. 9. Interactions around the N-terminal base of the C-helix involved in the conformational transition associated with the binding of RanGTP and cargo. (a) Unliganded yCRM1. (b) mCRM1–snurportin–RanGTP complex. Ran is shown in cyan, with bound GTP shown as space-filling spheres. The C-terminal region of CRM1 beyond the A-helix of HEAT repeat 21 is colored green, whereas the other regions of CRM1 are shown in yellow. The cyan dotted lines indicate hydrogen bonds. (c) Mutational analyses of yCRM1. Bound proteins in the GST pull-down assays were visualized by SDS-PAGE and Coomassie staining. Immobilized 47 μg GST-PKI was incubated with 100 μg yCRM1 with and without 50 μg Ran. Ran was charged with GTP or GDP as indicated.

the mCRM1–snurportin–RanGTP complex, the equivalent aspartic acid (D1017 on the 21A-helix) forms hydrogen bonds with T187 and Q188 in the HEAT4–5 linker loop (Fig. 9b). The binding of the D1034A mutant of yCRM1 to PKI was weak and was not enhanced by RanGTP (Fig. 9c), confirming that the interactions involved in the ring closure contributes to stabilize the cargo- and RanGTP-bound conformation of CRM1. Mutagenesis also provided evidence for the functional significance of the drastic conformational changes that occur at the N-terminal base of the C-helix, which are required for the movement of the C-helix and the ring closure of CRM1. In unliganded yCRM1, the central portion of the C-helix is exposed to solvent, but the N-terminal base of the C-helix is involved in interactions with 19B-helix and 21A-helix (Fig. 9a). These interactions

stabilize the orientation of the C-helix so that its distal end reaches the inner surface of HEAT repeats 8– 12. The hydrophobic side chain of Leu1050 (an absolutely conserved residue; see Fig. 5) at the Nterminal base of the C-helix makes nonpolar contacts with Phe952, Leu1036, Val1037, Tyr1049, Phe1051, and the aliphatic part of Arg1033, and thereby forms the hydrophobic core at the N-terminal base of the C-helix (Fig. 9a). This hydrophobic core is disrupted by the cooperative binding of NES cargo and RanGTP. In the mCRM1–snurportin–RanGTP complex, the side chain of Leu1033 of mCRM1 (Leu1050 of yCRM1) points to 20B-helix and 21Ahelix and is within hydrogen-bonding distance from the hydrophilic side chain of Gln1021 of mCRM1 (Fig. 9b). Therefore, substitution of Leu1050 of yCRM1 with a hydrophilic residue such as aspartic

361

Autoinhibition of CRM1

acid would be expected to disrupt formation of the autoinhibited conformation but instead stabilize the cargo- and RanGTP-bound conformation. Indeed, the L1050D mutant of yCRM1 bound to PKI and RanGTP with high affinity, and the binding to PKI in the absence of RanGTP was substantially increased by the L1050D mutation (Fig. 9c), confirming the importance of the hydrophobic interactions at the base of the C-helix in stabilizing the autoinhibited state of CRM1. Although the comparison of the crystal structures of CRM1 in its two limiting states (autoinhibited state and cargo- and RanGTP-bound state) allows only inference on the precise order of events that happen during the cooperative binding process, the somewhat “open” architecture of the CRM1 ring observed in the crystal structure of unliganded yCRM1 raises a possibility that the assembly of the CRM1–cargo– RanGTP complex occurs in a two-step reaction. Interestingly, a structural overlay of unliganded CRM1 with the RanGTP- and snurportin-bound form at HEAT repeat 3 shows that, even when CRM1 adopts the autoinhibited conformation, RanGTP can freely access the binding site on HEAT repeats 1–5 without requiring the movement of the HEAT9 loop and the C-helix and the closure of CRM1 ring (Fig. 7a). This indicates that the binding of RanGTP to CRM1 could occur in a stepwise induced-fit manner as depicted schematically in Fig. 10. In the first step, RanGTP binds weakly to the N-terminus of CRM1, and this weak binding does not require conformational changes of CRM1. In the second step, RanGTP induces a series of conformational changes in CRM1 to bind all of the binding sites. The second step of RanGTP binding would be concomitant with opening of the NES-binding site, because RanGTP displaces both the HEAT9 loop and the C-terminal tail from the autoinhibitory positions. Thus, by relieving the autoinhibition of CRM1, RanGTP enables stable binding of NES cargoes. However, it should be noted that the observation that the binding of RanGTP alone to CRM1 is very weak (see lane 2 in Supplementary Fig. S1b) suggests that the free energy of RanGTP binding is counterbalanced by the energy required for the global conformational changes of CRM1 so that the binary CRM1–RanGTP complex is unstable. The free energy of NES binding would be required to stabilize the conformational state in which CRM1 wraps around RanGTP using all of the four Ranbinding sites. Thus, the conformational changes in CRM1 can account for the positive cooperativity in RanGTP and cargo binding. At present, both of the structures of the cytosolic form (unliganded exportin) and the nuclear form (exportin–cargo–RanGTP complex) have been determined for three exportins: CRM1, Cse1, and Xpot (Supplementary Fig. S8). 17,18,29–31 Importantly, in all of these three nuclear export pathways, exportin

((NES-binding cleft closed) HEAT9 loop

CRM1

CRM1

(NES-binding cleft open)

CRM1 Fig. 10. A schematic model depicting a possible sequence of events that happen during the cooperative binding of RanGTP and NES cargo to CRM1. The binding of RanGTP may occur in two steps. When CRM1 adopts a conformation observed in the crystal structure of unliganded yCRM1, RanGTP can freely bind the N-terminal region of CRM1, but this binding is probably weak. The intimate association of RanGTP with the inner surface of CRM1 at all of the binding sites requires the movement of the HEAT9 loop and the C-helix away from the inner surface beneath the NES-binding cleft, concomitant with the closure of the CRM1 ring. This would relieve the autoinhibition and open the NES-binding cleft, allowing for stable binding of NES cargo. The energetic penalty associated with the substantial conformational changes of CRM1 would be counterbalanced by the free energy of NES and RanGTP binding.

undergoes substantial conformational changes when exportin forms a ternary complex with its specific cargo and RanGTP. Thus, it appears that the hypothesis that exportin is “spring-loaded” so that energy is stored in the ternary nuclear export complex, 29,32 originally proposed by Matsuura and Stewart based on the first structure of a nuclear export complex, 29 generally holds true.

362

Materials and Methods Expression and purification of yCRM1 for crystallization Glutathione S-transferase (GST)-CRM1 (S. cerevisiae, Xpo1p, a deletion mutant in which residues 377–413 and residues 971–984 are deleted) was expressed from pGEX-tobacco etch virus (TEV) 29 in the Escherichia coli host strain BL21-CodonPlus(DE3)RIL (Stratagene) in 2 × TY medium at 20 °C. After harvesting, the pellet was frozen in liquid nitrogen and stored at − 20 °C until needed. For purification, the frozen pellet of the cells expressing GST-CRM1 were thawed at room temperature and resuspended in phosphate-buffered saline, 1 mM PMSF, and 7 mM β-mercaptoethanol and lysed by sonication on ice. All subsequent purification steps were performed at 4 °C. Tween-20 was added to the clarified lysate to a final concentration of 0.05%. After incubating the clarified lysate with glutathione-Sepharose 4B resin (GE Healthcare) for 6 h, we washed the resin with buffer A [10 mM Tris–HCl (pH 7.5), 100 mM NaCl, 0.05% Tween-20, and 2 mM β-mercaptoethanol]. The GST-tag was removed with HisTEV protease (0.18 mg/ml) overnight in buffer A containing 0.2 mM 4-(2-Aminoethyl) benzenesulfonyl fluoride hydrochloride (AEBSF). CRM1 released from the resin was finally purified by gel filtration over Superdex 200 (GE Healthcare) in buffer B [10 mM Tris–HCl (pH 7.5), 150 mM NaCl, and 2 mM β-mercaptoethanol]. Fractions containing N 95% pure CRM1, as assessed by SDS-PAGE, were pooled and concentrated using a Millipore concentrator (molecular weight cutoff, 10,000). SeMet-substituted yCRM1 SeMet-substituted GST-CRM1 (S. cerevisiae, Xpo1p, a deletion mutant in which residues 377–413 and residues 971–984 are deleted) was expressed in the same strain, BL21-CodonPlus(DE3)RIL (Stratagene), which is not auxotrophic for methionine. Methionine biosynthesis was inhibited by growth conditions as previously described. 33 A preculture, grown in 2 × TY medium at 37 °C, was inoculated into minimal medium containing 1 × M9 supplemented with 20 μg/ml thiamine, 20 μg/ml biotin, and 50 μg/ml ampicillin and grown overnight at 28 °C to an OD600 (optical density at 600 nm) of 0.7. A mixture of Lamino acids was added as solids [per liter of culture: 50 mg of SeMet (Wako); 50 mg of leucine, isoleucine, and valine; and 100 mg of lysine, threonine, and phenylalanine]. After 15 min, protein expression was induced by the addition of 1 mM IPTG and the culture was grown overnight at 20 °C. SeMet-substituted CRM1 was purified as described for the non-substituted CRM1, except that 5 mM DTT was used instead of 2 mM β-mercaptoethanol in buffer A, and β-mercaptoethanol concentration in buffer B was increased to 5 mM. Crystallization and X-ray data collection Native crystals of yCRM1 were grown at 20 °C from 22.9 mg/ml protein by hanging drop vapor diffusion against 0.2 M KF and 16% (w/v) polyethylene glycol 3350. Crystals were cryoprotected using mother liquor containing 15% (v/v) glycerol and flash-cooled in liquid nitrogen. Preliminary

Autoinhibition of CRM1

X-ray diffraction experiments were performed at Photon Factory beamlines BL-5A and BL-17A, and a 2.1-Åresolution data set used for the final structure determination was collected at 100 K at SPring-8 beamline BL41XU. The crystal had P41 symmetry (a = b = 116.8 Å, c = 119.0 Å), with one molecule of CRM1 in the asymmetric unit (VM = 3.4 Å 3/ Da, solvent content ~64%). Crystals of SeMet-substituted yCRM1 were grown at 20 °C by streak seeding hanging drops containing 17.5 mg/ml CRM1, 0.1 M Hepes (pH 7.6), and 13% (w/v) methoxy polyethylene glycol 5000 using the native crystals as seeds. Crystals were cryoprotected using mother liquor containing 15% (v/v) glycerol and flash-cooled in liquid nitrogen. Three-wavelength MAD data sets at the Se absorption edge were collected to 2.4 Å resolution at 100 K at SPring-8 beamline BL41XU. The SeMet-substituted crystal was isomorphous with the native crystal. Structure determination and refinement All data were indexed and integrated using MOSFLM and further processed using CCP4 programs. 34 The structure was determined using SeMet MAD phasing method. The program SOLVE 35 identified 24 Se sites (of 28 SeMet residues in the yCRM1 construct used for crystallization), giving experimental phases with an overall figure of merit of 0.54 at 2.7 Å resolution. The initial phases were improved by density modification and phase extension to 2.1 Å resolution using PARROT. 36 Resulting phases from PARROT were used as input for automatic model building by Buccaneer. 37 Iterative cycles of rebuilding using Coot 38 and refinement using REFMAC5 39 yielded a final model with Rcryst = 18.7% (Rfree = 21.7%). A TLSMD web server 40,41 was used to define TLS groups for the final cycles of refinement. The minimum Rfree was obtained when CRM1 was partitioned into 14 TLS groups (these correspond to residues 47–56, 64–110, 111–166, 167–187, 188–249, 250–356, 357–485, 486–607, 608– 727, 728–777, 778–874, 875–930, 931–1037, and 1038– 1084). The final model contained yCRM1 residues 47–56, 64–264, 271–376, 414–970, and 985–1084 and 378 water molecules. Structural figures were produced using MOLSCRIPT, 42 Raster3D, 43 PyMOL, 44 CCP4MG, 45 and UCSF Chimera. 46 The theoretical value of Rg from the atomic coordinates was calculated using CRYSOL. 47 Expression and purification of proteins for biochemical assays All proteins were expressed in the E. coli host strain BL21-CodonPlus(DE3)RIL (Stratagene). His/S-CRM1 (S. cerevisiae, Xpo1p), His-CFP-YFP-MVM NS2 NES, GSTRan (S. cerevisiae, Gsp1p), GST-PKI (human), and His/SRan (canine) were expressed as previously described. 24,29 His-tagged proteins were purified over Ni-NTA (Novagen) and gel filtration over Superdex 75 or Superdex 200 (GE Healthcare). GST-tagged proteins were purified over glutathione-Sepharose 4B (GE Healthcare). GST-CRM1 (human) was expressed from pGEX-TEV 29 and initially purified over glutathione-Sepharose 4B (GE Healthcare). After removal of GST-tag by His-TEV protease, hCRM1 was finally purified over Superdex 200 (GE Healthcare).

363

Autoinhibition of CRM1

Mutants were created using the QuikChange system (Stratagene). All DNA constructs were verified by DNA sequencing. GST pull-down assay Pull-down assays were performed in binding buffer [10 mM Tris–HCl (pH 7.5), 200 mM NaCl, 5 mM Mg– acetate, 0.05% Tween-20, and 2 mM β-mercaptoethanol] as previously described. 29 GST-fusions were immobilized on 10 μl of packed glutathione-Sepharose 4B beads and each binding reaction was performed by incubating the beads with reaction mixtures in a total volume of 50 μl for 1 h at 4 °C. The amounts of proteins used are indicated in the figure legends. Beads were then spun down and washed twice with 1 ml of binding buffer, and bound proteins were analyzed by SDS-PAGE and Coomassie staining. Accession numbers Atomic coordinates and structure factors for the crystal structure of unliganded S. cerevisiae CRM1 have been deposited with the Protein Data Bank under accession code 3VYC.

Acknowledgements We thank our colleagues in Nagoya, especially Masako Koyama, Junya Kobayashi, Hidemi Hirano, and Tatsuo Hikage, for assistance and discussion. We are also indebted to the staff of Photon Factory and SPring-8 for assistance during data collection. This work was supported in part by the Sumitomo Foundation and also by JSPS/MEXT KAKENHI (18687010, 21770109, and 23770110).

Supplemental Data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jmb.2012.11.014 Received 23 September 2012; Received in revised form 2 November 2012; Accepted 7 November 2012 Available online 16 November 2012 Keywords: nuclear transport; nuclear export; exportin; CRM1; RanGTP Present address: N. Saito, Hamamatsu Photonics, Shizuoka, Japan. Abbreviations used: NPC, nuclear pore complex; NES, nuclear export signal; GST, glutathione S-transferase; SeMet, selenomethionine;

MAD, multi-wavelength anomalous diffraction; PKI, protein kinase inhibitor; RanBD, Ran-binding domain; yCRM1, yeast CRM1; mCRM1, mouse CRM1; hCRM1, human CRM1; TLStranslation/libration/screw; TEV, tobacco etch virus.

References 1. Gorlich, D. & Kutay, U. (1999). Transport between the cell nucleus and the cytoplasm. Annu. Rev. Cell Dev. Biol. 15, 607–660. 2. Weis, K. (2003). Regulating access to the genome. Nucleocytoplasmic transport throughout the cell cycle. Cell, 112, 441–451. 3. Vetter, I. R. & Wittinghofer, A. (2001). The guanine nucleotide-binding switch in three dimensions. Science, 294, 1299–1304. 4. Cook, A., Bono, F., Jinek, M. & Conti, E. (2007). Structural biology of nucleocytoplasmic transport. Annu. Rev. Biochem. 76, 647–671. 5. Klebe, C., Bischoff, F. R., Ponstingl, H. & Wittinghofer, A. (1995). Interaction of the nuclear GTP-binding protein Ran with its regulatory proteins RCC1 and RanGAP1. Biochemistry, 34, 639–647. 6. Bischoff, F. R. & Ponstingl, H. (1991). Catalysis of guanine nucleotide exchange on Ran by the mitotic regulator RCC1. Nature, 354, 80–82. 7. Gorlich, D., Pante, N., Kutay, U., Aebi, U. & Bischoff, F. R. (1996). Identification of different roles for RanGDP and RanGTP in nuclear protein import. EMBO J. 15, 5584–5594. 8. Stade, K., Ford, C. S., Guthrie, C. & Weis, K. (1997). Exportin 1 (Crm1p) is an essential nuclear export factor. Cell, 90, 1041–1050. 9. Fornerod, M., Ohno, M., Yoshida, M. & Mattaj, I. W. (1997). CRM1 is an export receptor for leucine-rich nuclear export signals. Cell, 90, 1051–1060. 10. Fukuda, M., Asano, S., Nakamura, T., Adachi, M., Yoshida, M., Yanagida, M. & Nishida, E. (1997). CRM1 is responsible for intracellular transport mediated by the nuclear export signal. Nature, 390, 308–311. 11. Ossareh-Nazari, B., Bachelerie, F. & Dargemont, C. (1997). Evidence for a role of CRM1 in signalmediated nuclear protein export. Science, 278, 141–144. 12. Wen, W., Meinkoth, J. L., Tsien, R. Y. & Taylor, S. S. (1995). Identification of a signal for rapid export of proteins from the nucleus. Cell, 82, 463–473. 13. Fischer, U., Huber, J., Boelens, W. C., Mattaj, I. W. & Luhrmann, R. (1995). The HIV-1 Rev activation domain is a nuclear export signal that accesses an export pathway used by specific cellular RNAs. Cell, 82, 475–483. 14. la Cour, T., Kiemer, L., Molgaard, A., Gupta, R., Skriver, K. & Brunak, S. (2004). Analysis and prediction of leucine-rich nuclear export signals. Protein Eng. Des. Sel. 17, 527–536. 15. Kutay, U. & Guttinger, S. (2005). Leucine-rich nuclearexport signals: born to be weak. Trends Cell Biol. 15, 121–124. 16. Dong, X., Biswas, A., Suel, K. E., Jackson, L. K., Martinez, R., Gu, H. & Chook, Y. M. (2009). Structural

364

17.

18.

19.

20. 21. 22. 23. 24.

25.

26.

27.

28.

29. 30.

31. 32. 33.

Autoinhibition of CRM1

basis for leucine-rich nuclear export signal recognition by CRM1. Nature, 458, 1136–1141. Monecke, T., Guttler, T., Neumann, P., Dickmanns, A., Gorlich, D. & Ficner, R. (2009). Crystal structure of the nuclear export receptor CRM1 in complex with Snurportin1 and RanGTP. Science, 324, 1087–1091. Guttler, T., Madl, T., Neumann, P., Deichsel, D., Corsini, L., Monecke, T. et al. (2010). NES consensus redefined by structures of PKI-type and Rev-type nuclear export signals bound to CRM1. Nat. Struct. Mol. Biol. 17, 1367–1376. Kudo, N., Matsumori, N., Taoka, H., Fujiwara, D., Schreiner, E. P., Wolff, B. et al. (1999). Leptomycin B inactivates CRM1/exportin 1 by covalent modification at a cysteine residue in the central conserved region. Proc. Natl Acad. Sci. USA, 96, 9112–9117. Cook, A. G. & Conti, E. (2010). Nuclear export complexes in the frame. Curr. Opin. Struct. Biol. 20, 247–252. Xu, D., Farmer, A. & Chook, Y. M. (2010). Recognition of nuclear targeting signals by Karyopherin-beta proteins. Curr. Opin. Struct. Biol. 20, 782–790. Lee, S. J., Jiko, C., Yamashita, E. & Tsukihara, T. (2011). Selective nuclear export mechanism of small RNAs. Curr. Opin. Struct. Biol. 21, 101–108. Guttler, T. & Gorlich, D. (2011). Ran-dependent nuclear export mediators: a structural perspective. EMBO J. 30, 3457–3474. Koyama, M. & Matsuura, Y. (2010). An allosteric mechanism to displace nuclear export cargo from CRM1 and RanGTP by RanBP1. EMBO J. 29, 2002–2013. Dong, X., Biswas, A. & Chook, Y. M. (2009). Structural basis for assembly and disassembly of the CRM1 nuclear export complex. Nat. Struct. Mol. Biol. 16, 558–560. Fox, A. M., Ciziene, D., McLaughlin, S. H. & Stewart, M. (2011). Electrostatic interactions involving the extreme C terminus of nuclear export factor CRM1 modulate its affinity for cargo. J. Biol. Chem. 286, 29325–29335. Forwood, J. K., Lange, A., Zachariae, U., Marfori, M., Preast, C., Grubmuller, H. et al. (2010). Quantitative structural analysis of importin-beta flexibility: paradigm for solenoid protein structures. Structure, 18, 1171–1183. Fukuhara, N., Fernandez, E., Ebert, J., Conti, E. & Svergun, D. (2004). Conformational variability of nucleo-cytoplasmic transport factors. J. Biol. Chem. 279, 2176–2181. Matsuura, Y. & Stewart, M. (2004). Structural basis for the assembly of a nuclear export complex. Nature, 432, 872–877. Cook, A., Fernandez, E., Lindner, D., Ebert, J., Schlenstedt, G. & Conti, E. (2005). The structure of the nuclear export receptor Cse1 in its cytosolic state reveals a closed conformation incompatible with cargo binding. Mol. Cell, 18, 355–367. Cook, A. G., Fukuhara, N., Jinek, M. & Conti, E. (2009). Structures of the tRNA export factor in the nuclear and cytosolic states. Nature, 461, 60–65. Conti, E., Muller, C. W. & Stewart, M. (2006). Karyopherin flexibility in nucleocytoplasmic transport. Curr. Opin. Struct. Biol. 16, 237–244. Van Duyne, G. D., Standaert, R. F., Karplus, P. A., Schreiber, S. L. & Clardy, J. (1993). Atomic

34.

35. 36. 37. 38. 39.

40. 41. 42. 43. 44. 45.

46.

47.

48.

49.

50.

structures of the human immunophilin FKBP-12 complexes with FK506 and rapamycin. J. Mol. Biol. 229, 105–124. Collaborative Computational Project, Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta Crystallogr., Sect. D, 50, 760–763. Terwilliger, T. C. & Berendzen, J. (1999). Automated MAD and MIR structure solution. Acta Crystallogr., Sect. D, 55, 849–861. Cowtan, K. (2010). Recent developments in classical density modification. Acta Crystallogr., Sect. D, 66, 470–478. Cowtan, K. (2006). The Buccaneer software for automated model building. 1. Tracing protein chains. Acta Crystallogr., Sect. D, 62, 1002–1011. Emsley, P. & Cowtan, K. (2004). Coot: model-building tools for molecular graphics. Acta Crystallogr., Sect. D, 60, 2126–2132. Murshudov, G. N., Skubak, P., Lebedev, A. A., Pannu, N. S., Steiner, R. A., Nicholls, R. A. et al. (2011). REFMAC5 for the refinement of macromolecular crystal structures. Acta Crystallogr., Sect. D, 67, 355–367. Painter, J. & Merritt, E. A. (2006). Optimal description of a protein structure in terms of multiple groups undergoing TLS motion. Acta Crystallogr., Sect. D, 62, 439–450. Painter, J. & Merritt, E. A. (2006). TLSMD web server for the generation of multi-group TLS models. J. Appl. Crystallogr. 39, 109–111. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946–950. Merritt, E. A. & Bacon, D. J. (1997). Raster3D: photorealistic molecular graphics. Methods Enzymol. 277, 505–524. DeLano, W. L. (2002). The PyMOL Molecular Graphics System. http://www.pymol.org. Potterton, L., McNicholas, S., Krissinel, E., Gruber, J., Cowtan, K., Emsley, P. et al. (2004). Developments in the CCP4 molecular-graphics project. Acta Crystallogr., Sect. D, 60, 2288–2294. Pettersen, E. F., Goddard, T. D., Huang, C. C., Couch, G. S., Greenblatt, D. M., Meng, E. C. & Ferrin, T. E. (2004). UCSF Chimera—a visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612. Svergun, D., Barberato, C. & Koch, M. H. J. (1995). CRYSOL—a program to evaluate X-ray solution scattering of biological macromolecules from atomic coordinates. J. Appl. Crystallogr. 28, 768–773. Thompson, J. D., Higgins, D. G. & Gibson, T. J. (1994). CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res. 22, 4673–4680. Waterhouse, A. M., Procter, J. B., Martin, D. M., Clamp, M. & Barton, G. J. (2009). Jalview Version 2— a multiple sequence alignment editor and analysis workbench. Bioinformatics, 25, 1189–1191. Chen, V. B., Arendall, W. B., 3rd, Headd, J. J., Keedy, D. A., Immormino, R. M., Kapral, G. J. et al. (2010). MolProbity: all-atom structure validation for macromolecular crystallography. Acta Crystallogr., Sect. D, 66, 12–21.