Structural Basis for Substrate Recognition and Dissociation by Human Transportin 1

Molecular Cell

Article Structural Basis for Substrate Recognition and Dissociation by Human Transportin 1 Tsuyoshi Imasaki,1,4 Toshiyuki Shimizu,1 Hiroshi Hashimoto,1 Yuji Hidaka,2 Shingo Kose,3 Naoko Imamoto,3 Michiyuki Yamada,1 and Mamoru Sato1,* 1Field of Supramolecular Biology, International Graduate School of Arts and Sciences, Yokohama City University, 1-7-29 Suehiro-cho, Tsurumi-ku, Yokohama 230-0045, Japan 2Department of Life Science, Faculty of Science and Engineering, Kinki University, 3-4-1 Kowakae, Higashi-Osaka, Osaka 577-8502, Japan 3Cellular Dynamics Laboratory, Discovery Research Institute, RIKEN, Wako, Saitama 351-0198, Japan 4Present address: Department of Biochemistry and Molecular Biology, Indiana University School of Medicine, Indianapolis, IN 46202, USA. *Correspondence: [email protected] DOI 10.1016/j.molcel.2007.08.006

SUMMARY

Transportin 1 (Trn1) is a transport receptor that transports substrates from the cytoplasm to the nucleus through nuclear pore complexes by recognizing nuclear localization signals (NLSs). Here we describe four crystal structures of human Trn1 in a substrate-free form as well as in the complex with three NLSs (hnRNP D, JKTBP, and TAP, respectively). Our data have revealed that (1) Trn1 has two sites for binding NLSs, one with high affinity (site A) and one with low affinity (site B), and NLS interaction at site B controls overall binding affinity for Trn1; (2) Trn1 recognizes the NLSs at site A followed by conformational change at site B to interact with the NLSs; and (3) a long flexible loop, characteristic of Trn1, interacts with site B, thereby displacing transport substrate in the nucleus. These studies provide deep understanding of substrate recognition and dissociation by Trn1 in import pathways. INTRODUCTION The transport of macromolecules between the nucleus and the cytoplasm through nuclear pore complexes (NPCs) is mediated via several transport pathways by transport receptors that are most commonly members of the importin-b family (Go¨rlich and Kutay, 1999; Tran and Wente, 2006). Transport receptors form complexes with their transport substrates (‘‘cargoes’’) through cognate nuclear localization signals (NLSs) for import substrates or nuclear export signals (NESs) for export substrates and target substrates to NPC components termed nucleoporins. Transport directionality and interactions between the transport receptor and substrate are regulated by RanGTP through its nucleotide state and, in the nuclear

import system, binding of RanGTP to the receptor in the nucleus is associated with substrate dissociation. Of the several transport pathways, the best characterized is an import pathway mediated by importin-b (karyopherinb1), for which the crystal structure has been determined in complex with various substrates, nucleoporin, and RanGTP (Cingolani et al., 1999, 2002; Lee et al., 2003; Bayliss et al., 2000; Vetter et al., 1999; Lee et al., 2005; Conti et al., 2006). Transportin 1 (Trn1) (karyopherin-b2) is a transport receptor that belongs to the importin-b family and has 24% sequence similarity to importin-b (Siomi and Dreyfuss, 1995). Trn1 was first identified as a nuclear transport receptor for heterogeneous nuclear ribonucleoprotein A1 (hnRNP A1) (Siomi and Dreyfuss, 1995; Pollard et al., 1996; Iijima et al., 2006). Other substrates whose transport is mediated by Trn1 have since emerged, and these include hnRNP D (Suzuki et al., 2005), JKTBP (Kawamura et al., 2002), TAP (Truant et al., 1999), HuR (Gu¨ttinger et al., 2004), a ribosomal protein L23a (Jakel and Go¨rlich, 1998), c-Fos (Arnold et al., 2006b), HIV Rev (Arnold et al., 2006a), and others. NLSs of these transport substrates have little sequence similarity, but recent mutational analyses have shown the importance of two successive proline and tyrosine residues (called a PY motif) conserved in the NLSs of hnRNP D, TAP, JKTBP, hnRNP A1, and hnRNP M for recognition by Trn1 (Figure 1) (Iijima et al., 2006; Suzuki et al., 2005). The structure of Trn1 in complex with RanGppNHp (referred to as Trn1-RanGTP complex) (Chook and Blobel, 1999; Chook et al., 2002) and the recently determined structure of a H8 loop-truncated Trn1 mutant bound to hnRNP A1 NLS and hnRNP M NLS (Figure 1) (Lee et al., 2006; Cansizoglu et al., 2007) showed Trn1 to be similar to importin-b, but with a much longer H8 loop, important for substrate dissociation in the nucleus. This has led to the proposal of a consensus sequence for NLS recognition by Trn1. However, the consensus sequence is still ambiguous, and detailed molecular mechanism of the H8 loop is still unclear. In order to better understand

Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc. 57

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Figure 1. Sequence Alignment of Five Known NLSs with PY Motif For each NLS, three consensus residues (H/R, P, and Y) recognized in site A and one hydrophobic or basic residue (V/A/P/K) in site B are colored red. hnRNP A1 NLS is also called M9 NLS.

substrate recognition and dissociation by Trn1, structural analyses of full-length wild-type Trn1 in a substrate-free form and bound to multiple NLSs of import substrates are absolutely essential. Here we describe four crystal structures of human full-length wild-type Trn1 corresponding to Trn1 in a substrate-free form and bound to the following three NLS peptides, which all have a PY motif: hnRNP D NLS (residues 332–355), TAP NLS (residues 53–82), and JKTBP NLS (residues 396–420) (Figure 1). RESULTS AND DISCUSSION Overall Structure of Trn1 Full-length wild-type Trn1 and importin-b have yet to be crystallized in a form free of both substrate and RanGTP, due to inherent flexibility of the protein molecules. In the present study, we have succeeded in determining the crystal structure of the Trn1, together with the structures of Trn1 bound to the three NLSs of hnRNP D, TAP, and JKTBP (Table 1). Trn1 is a superhelical S-like molecule formed by two overlapping arches (N- and C-terminal arches) and is constructed by helical stacking of the 20 HEAT repeats (H1–H20) (Figures 2A, 2D, and 2E). Each HEAT repeat is composed of two antiparallel helices (referred to as A and B helices) located at the respective convex and concave surfaces of the superhelical molecule. The H8 loop (residues 312–374) connecting the H8A and H8B helices (Figure 2E) is mostly disordered in the four Trn1 structures (substrate-free form and bound to three NLSs) (Figure 2A and Figure S1 in the Supplemental Data available with this article online). The structure of the N-terminal arch (HEAT repeats 1– 13) is almost the same in the four Trn1 structures, whereas that of the C-terminal arch (HEAT repeats 8–20) changes depending on the NLS bound (Figure 2D). Furthermore, we also determined the crystal structure of Trn1 bound to TAP NLS in a crystal form different from that listed in Table 1 and showed the conformation of the C-terminal arch also changes depending on crystal packing (data not shown). By superimposing the Trn1 structures of free, RanGTP bound, and NLS bound, we observed that the overlapping region of the N- and C-terminal arches (HEAT repeats 8–13) (Figure 2E) show no conformational change (Figure 2D and Figure S2A). The conformational differences in the C-terminal arch are attributable to the fact that the loop connecting HEAT repeats 13 and 14 works as a hinge whose angle is affected by various NLS binding events and crystal packing. This indicates that the region corresponding to HEAT repeats 14–20 is inherently flexible.

Comparison of the four structures with the structure of the Trn1-RanGTP complex (Figure 2C) demonstrates that RanGTP binding to the N-terminal arch produces substantial conformational changes in both the N- and C-terminal arches (Figure 2D and Figure S2A). The conformational change in the C-terminal arch is caused by the intramolecular interaction between the H8 loop and the inherently flexible HEAT repeats 14–20 in the C-terminal arch. This intramolecular interaction is observed in the Trn1-RanGTP complex, but not in the structures of Trn1 in substrate-free or NLS-bound form (Figures 2A and 2C). It is, however, noticeable that no changes occur in the conformation of the overlapping region of the N- and C-terminal arches (HEAT repeats 8–13) irrespective of whether RanGTP or NLS is bound. These results indicate that the inherently large flexibility of the Trn1 molecule is mostly attributable to flexibility in the region corresponding to HEAT repeats 14–20. These conformational features observed in Trn1 are characteristic of importin-b family, because similar features are also observed in other proteins belonging to importin-b family (Figures S2B–S2D). Interaction between NLS and Trn1 Electron densities corresponding to the three NLSs (Figure 2B) are observed in the concave surface of the C-terminal arch of the superhelical S-like Trn1 molecule (Figure 2A). In the structure of hnRNP D NLS bound to Trn1, 15 residues from K341hnRNP_D to Y355hnRNP_D interact with HEAT repeats 8–18 (Figures 2B and 3A and Figure S3A), and the electron density from K341hnRNP_D to G346hnRNP_D in the region of HEAT repeats 14–18 is slightly lower than that from G347hnRNP_D to Y355hnRNP_D, which are in the region of HEAT repeats 8–13. In the case of TAP NLS bound to Trn1, the electron density from P68TAP to P79TAP is observed around HEAT repeats 8– 13, but unlike the case in the hnRNP D NLS bound to Trn1, there is no detectable electron density of TAP NLS around HEAT repeats 14–18 (Figures 2B and 3A and Figure S3B). In the case of JKTBP NLS bound to Trn1, the electron density of JKTBP NLS is so weak that the side chains of the amino acid residues could not be assigned unambiguously, but the electron density is significantly observable around HEAT repeats 8–13 (Figure 2B). However, similar to the situation with TAP NLS bound to Trn1, there is no detectable electron density of JKTBP NLS around HEAT repeats 14–18. In the import pathway mediated by importin-a-importinb complex, the import rate of an NLS-bearing substrate is known to be related to the binding affinity of the substrate

58 Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc.

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Table 1. Crystallographic Data and Refinement Statistics NLS-free Trn1

Trn1 Bound to hnRNP D NLS

Trn1 Bound to TAP NLS

Trn1 Bound to JKTBP NLS

P41212

P212121

P212121

P21212

a (A˚)

107.4

69.1

69.6

132.3

b (A˚)

107.4

119.1

119.7

169.8

c (A˚)

194.7

151.1

147.8

68.4

Resolution (A˚)

3.4

3.2

2.6

3.2

Total observations

116,880

110,635

188,910

128,589

Unique reflections

15,598

19,610

37,522

25,190

Crystallographic Data Space group Cell dimensions

a,b

Rmerge (%)

8.5 (37.4)

8.2 (54.1)

4.4 (30.8)

7.1 (43.4)

Completeness (%)a

94.8 (72.2)

91.9 (56.4)

96.3 (87.4)

94.9 (82.8)



10.4 (1.7)

10.2 (1.7)

13.4 (2.6)

13.2 (3.6)

14,795

18,626

35,626

23,913

23.2/30.3

23.1/29.5

22.0/26.7

23.5/27.8

Trn1

6652

6691

6832

6640

peptide

0

126

108

50

Bond length (A˚)

0.008

0.008

0.010

0.009

Bond angle ( )

1.133

1.120

1.186

1.112

Core region

83.0

85.2

92.3

89.4

Allowed

15.3

13.2

6.8

9.2

Generously allowed

1.7

1.6

0.9

1.3

Disallowed

0

0

0

0

Refinement Statistics Reflections used c,d

Rwork/Rfree (%)

Number of atoms

Rmsd

Ramachandran plot (%)

Values in parentheses are for the highest resolution shells (3.52–3.40, 3.31–3.20, 2.70–2.60, and 3.31–3.20 A˚ for NLS-free Trn1, Trn1 bound to hnRNP D NLS, Trn1 bound to TAP NLS, and Trn1 bound to JKTBP NLS, respectively). b Rmerge = ShSi jI(h)i  < I(h) > j / ShSi I(h)i. c Rwork/Rfree = SjjFoj  jFcjj / SjFoj, where Rwork and Rfree are calculated by using the working and free reflection sets, respectively. Rfree reflections (5% of the total) were held aside throughout refinement. d Rwork/Rfree (%) calculated without the NLS peptide is 25.1/32.6 for Trn1 bound to hnRNP D NLS, 22.5/27.2 for Trn1 bound to TAP NLS, and 23.9/28.8 for Trn1 bound to JKTBP NLS. a

for importin-a and thus related to the accumulation of the substrate in the nucleus (Hodel et al., 2006). The difference in the electron density appearance of the NLSs shown in Figure 2B is correlated with KD, where KD is the dissociation constant for the NLS interaction with Trn1 (Table 2). Thus, hnRNP D NLS binds to Trn1 with KD = 3.2 nM, and a clear electron density for the NLS is visible in the region of HEAT repeats 8–18 (Figures 2B and 3A and Figure S3A). TAP NLS binds to Trn1 with KD = 17 nM, and a clear electron density for the NLS is visible in the region of HEAT repeats 8–13, but no electron density for the NLS in the

region of HEAT repeats 14–18 can be seen (Figures 2B and 3A and Figure S3B). Finally, JKTBP NLS binds to Trn1 with KD = 1.0 mM, and a weak electron density for the NLS is visible in the region of HEAT repeats 8–13, but no electron density for the NLS can be observed in the region of HEAT repeats 14–18 (Figure 2B). These results indicate that an NLS with a PY motif is recognized by Trn1 at two sites, namely site A (HEAT repeats 8–13) with high affinity and site B (HEAT repeats 14–18) with low affinity, and NLS interaction at site B controls overall binding affinity for Trn1.

Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc. 59

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Figure 2. Structures of Trn1 (A) Overall structures of NLS (substrate)-free Trn1 (red) and Trn1 bound to hnRNP D NLS (green), TAP NLS (blue), and JKTBP NLS (yellow), where a helices are represented as cylinders. Twenty consecutive HEAT repeats (H1–H20), each of which is composed of two antiparallel helices, are labeled on the NLS-free Trn1 structure. Each H8 loop is shown as a black cylinder (a helix) and a black dotted line (disordered region). Fo  Fc map contoured at 2.5 s for each NLS is also shown. (B) The CNS composite simulated annealing omit map of the NLS region bound to Trn1. The map was calculated with coefficient 2Fo Fc and contoured at 1.0 s. NLSs of hnRNP D, TAP, and JKTBP are shown as stick models colored green, blue, and yellow, respectively. (C) Overall structure of Trn1-Ran complex, where a helices are represented as cylinders and the H8 loops are colored black. Structure was drawn with the refined coordinates deposited in the Protein Data Bank (accession code 1QBK). (D) Trn1 structure showing the conformational changes upon either RanGTP or NLS binding. The structures of NLS-free Trn1 (red), Trn1 bound to hnRNP D NLS (green), Trn1 bound to TAP NLS (blue), Trn1 bound to JKTBP NLS (yellow), and Trn1-RanGTP complex (brown) were superimposed

60 Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc.

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Figure 3. NLS Recognition by Trn1 (A) Schematic illustrations of Trn1 interactions with hnRNP D NLS (left) and TAP NLS (right). Trn1-NLS contacts less than 3.8 A˚ are shown. HEAT repeats 8–13 correspond to site A, and HEAT repeats 14–18 to site B. (B) Structures of hnRNP D NLS (green), TAP NLS (blue), hnRNP A1 NLS (orange), and hnRNP M NLS (purple) bound to Trn1. Two close-up views of the structures at the right side and one close-up view at the left side show the interactions with Trn1 at sites A and B, respectively. Structures of hnRNP A1 NLS and hnRNP M NLS bound to an H8 loop-truncated Trn1 mutant were drawn with the refined coordinates deposited in the Protein Data Bank (accession codes 2H4M and 2OT8).

in the overlapping region of the N- and C-terminal arches (HEAT repeats 8–13). The 20 consecutive HEAT repeats in each Trn1 molecule are represented by straight lines. Structure of Trn1-RanGTP complex was drawn with the refined coordinates deposited in the Protein Data Bank (accession code 1QBK). (E) Schematic illustration of Trn1 structure and domain organization. Two antiparallel helices (A and B helices) in each HEAT repeat are colored orange and yellow, respectively. The H8 loop (residues 312–374) defined as the long loop connecting the H8A and H8B helices is colored black.

Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc. 61

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Table 2. Kinetic Parameters of Trn1 Interacting with NLS Soluble Ligand (Trn1) WT

Immobilized Ligand (NLS) hnRNP D WT

ka1(Ms)

kd1(1/s) 5

4.6 3 10

5

ka2(1/s) 2

2.7 3 10

2

kd2(1/s) 2

2.0 3 10

3

c2

KD(M) 3

1.1 3 10

3

9

13.6

8

3.2 3 10

WT

hnRNP D V342G

1.1 3 10

2.3 3 10

7.0 3 10

1.8 3 10

4.2 3 10

3.9

WT

hnRNP D V342A

1.8 3 104

1.3 3 102

1.0 3 102

1.5 3 103

9.4 3 108

0.1

hnRNP D H348A

5

2

3

3

8

16.9

9

WT

1.5 3 10

5

1.5 3 10

2

5.2 3 10

2

1.0 3 10

3

1.7 3 10

WT

hnRNP D P354A

1.2 3 10

4.5 3 10

5.7 3 10

1.3 3 10

7.3 3 10

2.0

WT

hnRNP D Y355A

1.2 3 105

2.7 3 102

1.1 3 102

2.4 3 103

4.3 3 108

3.3

5

2

2

3

9

0.1

6

R343A/V345G

hnRNP D WT

4.9 3 10

2

3.7 3 10

2

1.7 3 10

3

1.2 3 10

4

5.0 3 10

W460A

hnRNP D WT

2.6 3 10

1.4 3 10

4.6 3 10

3.1 3 10

3.6 3 10

1.9

E509A

hnRNP D WT

1.7 3 105

6.2 3 102

3.8 3 103

2.2 3 103

1.3 3 108

4.9

hnRNP D WT

4

2

3

3

7

16.7

8

W730A

5.8 3 10

5

3.4 3 10

2

1.3 3 10

3

1.2 3 10

4

2.9 3 10

WT

TAP WT

1.0 3 10

1.3 3 10

6.2 3 10

9.1 3 10

1.7 3 10

1.1

WT

TAP V59G

4.6 3 104

1.7 3 102

5.5 3 103

8.3 3 104

5.0 3 108

5.3

WT

TAP V59A

4

2.8 3 10

2

1.5 3 10

3

9.8 3 10

3

1.1 3 10

8

5.3 3 10

12.4

WT

TAP R71A

3.7 3 104

1.8 3 102

5.9 3 103

9.2 3 104

6.7 3 108

4.9

TAP P74A

4

2

3

4

8

4.1

8

WT

4.5 3 10

4

1.8 3 10

2

5.4 3 10

3

8.2 3 10

4

5.5 3 10

WT

TAP Y75A

3.4 3 10

1.6 3 10

5.7 3 10

8.5 3 10

6.3 3 10

3.3

R343A/V345G

TAP WT

7.8 3 104

1.9 3 102

1.4 3 102

8.9 3 104

1.5 3 108

0.6

W460A

TAP WT

ND

ND

ND

ND

ND

5

2

3

3

ND 7

E509A

TAP WT

3.2 3 10

1.2 3 10

3.9 3 10

1.6 3 10

1.1 3 10

3.7

W730A

TAP WT

1.0 3 104

3.3 3 102

1.3 3 103

5.7 3 104

9.7 3 107

1.3

4

2

3

3

6

13.9

6

WT

JKTBP WT

5.2 3 10

4

8.2 3 10

2

1.7 3 10

3

3.3 3 10

3

1.0 3 10

WT

JKTBP Q399A

4.2 3 10

8.9 3 10

1.7 3 10

3.6 3 10

1.9 3 10

9.9

WT

JKTBP G404A

2.2 3 105

3.6 3 101

1.3 3 103

6.8 3 103

1.4 3 106

0.9

WT

JKTBP H413A

4

1.4 3 10

1

1.2 3 10

3

4.0 3 10

3

9.4 3 10

6

6.3 3 10

9.9

WT

JKTBP P419A

9.5 3 102

3.7 3 102

3.4 3 103

2.0 3 103

1.4 3 105

0.1

WT

JKTBP Y420A

ND

ND

ND

ND

ND

4

2

1

3

ND 8

R343A/V345G

JKTBP WT

2.2 3 10

2.4 3 10

1.4 3 10

1.2 3 10

8.7 3 10

W460A

JKTBP WT

ND

ND

ND

ND

ND

E509A

JKTBP WT

4.5 3 104

1.8 3 102

5.4 3 103

2.9 3 103

3.5 3 106

1.9

JKTBP WT

3

2

3

3

6

2.1

W730A

1.3 3 10

2.5 3 10

7.3 3 10

1.2 3 10

4.2 ND

2.6 3 10

a

Dissociation constants KD = 1/{(ka1/kd1)(1+ka2/kd2)} were derived with both association and dissociation rate constants (ka1, ka2) (kd1, kd2). b WT, wild type; ND, not detected. c R343A/V345G Trn1 mutant may bind JKTBP NLS more tightly by artificial interactions between double-mutated residues (R343A/V345G) on the H8 loop and JKTBP NLS.

NLS Recognition by Trn1 In the structures of the three NLSs, hnRNP A1 NLS, and hnRNP M NLS (Figure 1) bound to Trn1, the NLSs take up different conformations without any secondary structure and with their peptide directions being opposite to the direction of the Trn1 superhelix (Figure 3A and Figure S4). A bioinformatics search for transport sub-

strates with a PY motif as recognized by Trn1 highlighted a C-terminal consensus sequence motif (R/K/H-X(2-5)-P-Y, where X represents any residue) following a hydrophobic motif (f-G/A/S-f-f, where f is a hydrophobic residue) or a basic-enriched(5-8) motif (Lee et al., 2006). Our structural data show that three consensus residues in hnRNP D NLS and TAP NLS and one hydrophobic residue in the hnRNP

62 Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc.

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Figure 4. Effect of Site A and Site B Mutation on In Vitro Nuclear Transport Digitonin-permeabilized cells were incubated with 1 mM GST-GFP-hnRNP D NLS for 20 min at 30 C in the presence of (A) 0.5 mM Trn1 wild-type, (B) 0.5 mM Trn1 W373A (NLS-binding site unrelated mutant), (C) 0.5 mM Trn1 W460A, (D) 0.5 mM Trn1 E509A, and (E) 0.5 mM Trn1 W730A.

D NLS are recognized at sites A and B, respectively, in the same manner as is observed with hnRNP A1 NLS and hnRNP M NLS (Figure 3B and Figure S4). The importance of the three consensus residues and one hydrophobic residue was investigated by surface plasmon resonance (SPR) analysis (Table 2). All the binding curves from the SPR analysis are fitted to a two-state model (A + B % AB % AB*), which describes the interaction between structurally changeable analyte and immobilized ligand, better than to a simple 1:1 Langmuirien model (A + B % AB) (Catimel et al., 2001). As summarized in Table 2, the dissociation constants (KDs) of wild-type NLSs bound to wild-type Trn1 are different for the three NLSs, but the KDs are smaller than those of NLSs mutated at the three consensus residues and one hydrophobic residue (H348A, P354A, Y355A, and V342G/V342A for hnRNP D NLS; R71A, P74A, Y75A, and V59G/V59A for TAP NLS; and H413A, P419A, Y420A, and G404A for JKTBP NLS). Of these NLS mutants, however, G404A for JKTBP NLS has little effect on the NLS interaction with wild-type Trn1, because the KD of G404A is almost consistent with KDs of wild-type and Q399A mutated at neither three consensus residues nor one hydrophobic residue for JKTBP NLS. This indicates that, unlike the NLSs of hnRNP D and TAP, JKTBP NLS does not bind to site B. Furthermore, the KDs of the wild-type NLSs bound to mutant Trn1, where W460, E509, and W730 that interact with three consensus residues and one hydrophobic residue are replaced by alanines, are much larger than those of wild-type NLSs bound to wild-type Trn1. Functional significance of these three residues having larger effect on the NLS interaction were assessed by in vitro nuclear transport assay using a GFP fusion protein to hnRNP D NLS (Figure 4). Mutants W460A and W730A show very weak ability to mediate nuclear import, and E509A shows a decreased ability, as compared with those of the wild-type and a NLS-binding site unrelated mutant W373A under the condition examined, thus indicating that ability to mediate nuclear import is strongly correlated with the KD value. These results show that NLS interactions at site A (H348, P354, and Y355 for hnRNP D NLS; R71, P74, and Y75 for TAP NLS; and H413, P419, and Y420 for JKTBP NLS) play a crucial role for NLS recognition by Trn1, and those at site B (V342 for hnRNP D NLS; V59 of TAP NLS) control overall binding affinity to Trn1. Because the con-

formation of site A is independent of NLS binding, whereas that of site B is dependent on NLS binding (Figure 2D), we propose that the residues corresponding to the consensus sequence motif interact with Trn1 at site A and then an induced fit hydrophobic interaction occurs between the hydrophobic motif in the still unstructured NLS and the inherently flexible site B via a change in hinge angle of the loop connecting HEAT repeats 13 and 14. NLS recognition at two sites is also observed in an adaptor protein, importin-a, where classical monopartite and bipartite NLSs are recognized by one or two sites, respectively (Kobe, 1999). Role of Site B for Substrate Dissociation The previous structural study clearly has shown that the H8 loop of Trn1-RanGTP complex makes extensive intramolecular interactions with site B, resulting in spatial overlap (collision) with the NLS at site B, which could lead to displacement of the NLSs at site B (Chook and Blobel, 1999; Chook et al., 2002; Lee et al., 2006). This spatial overlap occurs at a portion of NLS bound at site A as well, which could also lead to displacement of NLSs at site A. These structural data prompted us to investigate if disruption of site B-NLS interaction is the key for dissociation mechanism by the Trn1. We generated the Trn1 double mutant in which the residues (R343 and V345) critical for site B interaction were mutated to alanine and glycine, respectively, and tested their contributions for dissociation of the NLSs from the Trn1 by GST pull-down assays. As for the control, we showed that this double mutant (R343A/V345G) did not affect its ability to bind to RanGppNHp (Figure S5). We tested all three NLSs (hnRNP D NLS, TAP NLS, and JKTBP NLS) with different binding affinity to site B. One could expect that, if disruption of the site B-NLS interaction is the key for dissociation, the H8 loop mutant would most affect dissociation of the NLS, which interacts strongly with site B (hnRNP D NLS), but not the ones with weak (TAP NLS) or no interaction at all (JKTBP NLS). Upon binding to RanGppNHp, the double mutations severely affected the dissociation of hnRNP D NLS from Trn1 bound to hnRNP D NLS (Figure 5), in which site B-NLS interaction is strongest (KD = 3.2 nM) (Figure 3A and Table 2). In contrast, it did not affect the dissociation of TAP NLS (KD = 17 nM) and JKTBP NLS (KD = 1.0 mM) (Figure 5),

Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc. 63

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Figure 5. Effect of H8 Loop Mutation on NLS Dissociation Interaction of Trn1 or H8 loop mutant R343A/V345G with various GST-tagged NLSs shown on the left lower corners of the gels in the presence and absence of RanGppNHp and RanGDP was studied by GST pull-down assay. Signs + and + + for RanGppNHp indicate, respectively, the first and second treatments of a bead-bound form of GST-NLS-Trn1 complex with RanGppNHp.

in which site B-NLS interactions are weak or almost lacking (Figure 3A and Table 2). Dissociations of TAP NLS and JKTBP NLS from Trn1 are most likely caused by the interaction between the portion of the NLS bound at site A and the H8 loop as stated above. These results are consistent with the model that the interaction of the H8 loop with site B is essential for dissociation of those NLSs, which rely on the interaction with site B for binding to the Trn1. Proposed Mechanism for NLS Recognition and Dissociation from the Trn1 Our proposed mechanism for NLS recognition and dissociation from the Trn1 is illustrated in Figure 6. In the cytoplasm, Trn1 recognizes NLS R/K/H-X(2-5)-P-Y motif at site A (preloaded state in Figure 6). In some cases, Trn1 changes its conformation to make hydrophobic interaction between site B and the hydrophobic motif of NLS (loaded state in Figure 6). The Trn1-NLS complex is transported through NPC into the nucleus.

In the nucleus, RanGTP binding to the N-terminal arch in NLS-bound Trn1 brings a competing interaction from the H8 loop against site B (see preunloaded state in Figure 6), resulting in a displacement of an NLS from site B. This displacement from site B is critical for NLSs such as hnRNP D that interact strongly with site B. This displacement, however, is not as critical for the NLSs with weak or no interaction at all with site B (such as TAP NLS and JKTBP). After releasing from site B, the NLS is displaced from site A by the spatial overlap of the H8 loop with the part of the NLS at site A, resulting in a complete dissociation from the Trn1 (see preunloaded and unloaded states in Figure 6). Though the H8 loop is characteristic to the Trn1, the proposed NLS dissociation model by the intramolecular interaction could be seen as a more general mechanism. For instance, the NLS dissociation from importin-a is driven by the intramolecular interaction from the N-terminal autoinhibitory segment after the importin-a-NLS

Figure 6. Schematic Illustration of a Proposed Mechanism for the Nuclear Import Pathway Mediated by Trn1 The Trn1 molecule is represented by the S-like cyan ribbon labeled with N and C termini. NPC is the nuclear pore complex. Red and blue ellipsoids on the NLS in transport substrate show the three consensus residues (red ellipsoid) and one hydrophobic residue (blue ellipsoid).

64 Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc.

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

complex was dissociated from the importin-b upon binding to RanGTP (Kobe, 1999; Chook and Blobel, 2001; Lee et al., 2005). The N-terminal autoinhibitory segment of the importin-a could be functionally equivalent to the H8 loop of the Trn1. EXPERIMENTAL PROCEDURES Crystallization Expression and purification of human full-length Trn1 (residues 1–890) and crystallization of Trn1 bound to hnRNP D NLS (residues 332–355) and JKTBP NLS (residues 396–420) were performed as described previously (Imasaki et al., 2006). Crystallization of NLS-free Trn1 was performed at 293K by hanging-drop vapor diffusion against a solution of 0.1 M CAPSO (pH 9.8) containing 16%–18% (w/v) PEG-8K and 0.1 M NaH2PO4. Hanging drops were prepared by mixing 2 ml of 5 mg/ml Trn1, 2 ml reservoir solution, and 0.4 ml of 0.1 M spermidine. Trn1 bound to TAP NLS was prepared by mixing Trn1 at a final concentration of 4 mg/ml with 5-fold molar excess of TAP NLS (residues 53–82) and incubating overnight at 293 K. Crystals of Trn1 bound to TAP NLS were obtained at 293 K by the hanging-drop vapor diffusion method with equilibrium in a reservoir solution (200 ml) of 0.1 M CAPSO (pH 9.8) containing 16%–20% (w/v) PEG-8K, 0.1 M KCl, and 0.1 M NaH2PO4. Hanging drops (4 ml) were prepared by mixing 2 ml of 4 mg/ml Trn1 bound to TAP NLS and 2 ml reservoir solution. The NLS peptides bound to Trn1 were assessed by GST pull-down assays in the crystallization conditions in the absence of the precipitants (Figure S6). Data Collection Crystals were soaked in reservoir solution containing 20% glycerol for data collection at 100 K. Diffraction data for NLS-free Trn1 and Trn1 bound to hnRNP D NLS and JKTBP NLS were collected with an ADSC Quantum 315 CCD detector on BL41-XU at SPring-8, and those for Trn1 bound to TAP NLS were collected with an ADSC Quantum 210 CCD detector on NW-12 at Photon Factory Advanced Ring (PF-AR). All the diffraction data were processed with HKL2000 (Otwinowski and Minor, 1997). Crystallographic data and data collection statistics are given in Table 1. Structure Determination and Refinement The structure of Trn1 bound to JKTBP NLS was solved by molecular replacement using PHASER (McCoy et al., 2005), where two fragments (residues 171–603 and residues 604–890) of Trn1 in the Trn1RanGTP complex (PDB code 1QBK) were used as search models and refined with O (Jones et al., 1991), CNS (Bru¨nger et al., 1998), and Refmac5 (Murshudov et al., 1997). Structures of NLS-free Trn1 and that bound to TAP NLS and hnRNP D NLS were solved by molecular replacement using PHASER (McCoy et al., 2005), where two fragments (residues 5–603 and residues 604–890) of Trn1 in the refined structure of Trn1 bound to JKTBP NLS were used as search models and refined with O, CNS, and Refmac5. The data collection and refinement statistics are summarized in Table 1. All figures were generated with MOLSCRIPT (Kraulis, 1991), RASTER3D (Merritt and Murphy, 1994), and PyMOL (Delano, 2002). SPR Analysis The protein concentration was determined by absorbance at 280 nm using an extinction coefficient calculated from the amino acid composition (ProtParam). All measurements were performed with a BIAcore3000 biosensor (GE Healthcare). Anti-GST IgG was immobilized onto a carboxymethyldextran sensor (CM5) chip by NHS/EDC coupling as described previously (Suzuki et al., 2005). GST-NLSs were captured by immobilized anti-GST IgG at a concentration of about 1000 resonance units (RU). The samples (100 ml) for analyses were prepared at various concentrations in HBS buffer (10 mM HEPES

[pH 7.4] containing 3 mM EDTA, 0.15 M NaCl, and 0.005% [v/v] Tween 20) and were injected over the sensor surface at a flow rate of 20 ml min1. After completion of the injection phase, dissociation was monitored in HBS buffer for 300 s at the same flow rate. All values were determined from three to six independent assays. Bound proteins were eluted, and the surface was regenerated between injections, using 40 ml of 10 mM NaOH. Regeneration conditions did not denature the immobilized antigen as shown by equivalent signals upon reinjection of the ligand. The apparent association and dissociation rate constants were calculated with BIAevaluation version 4.1 (GE Healthcare) (Catimel et al., 2001). The binding curves were fitted to a two-state conformational change binding model, which describes the binding of a conformational changeable analyte to immobilized ligand, where one analyte molecule can bind in two or more binding modes (A + B % AB % AB*). The apparent association (ka1, ka2) and dissociation constant (kd1, kd2) rates are shown in Table 2. Dissociation constants were derived with both association and dissociation rates 1/{(ka1/kd1)(1+ka2/ kd2)}. The goodness of the fit between experimental data and fitted curves was estimated from the coefficient of correlation by c2 values, which is a statistical measure of how closely the model fits the experimental data. In general, c2 values lower than about 10 signify a good fit. GST Pull-Down Assay All the reaction mixture (50 ml) containing HBS buffer and 5 mM MgCl2 was incubated for 1 hr at 277 K. Various NLSs fused with GST (15 mg) were each immobilized with either Trn1 (5 mg) or Trn1 R343A/V345G (5 mg) on a 10 ml slurry of glutathione Sepharose. Unbound Trn1 was washed away with 2 3 1 ml of binding buffer. To assay NLS dissociation from Trn1 on RanGppNHp binding, the beads were incubated with or without either RanGppNHp or RanGDP (15 mg) for 1 hr and then centrifuged. The precipitated beads were washed with 2 3 1 ml of binding buffer. In the second dissociation, the beads were again treated with or without RanGppNHp in the same way. The bead-bound proteins were eluted with SDS-sample buffer and analyzed by SDSPAGE. Proteins on gels were stained with Coomassie brilliant blue. Sample Preparation for SPR Analysis and GST Pull-Down Assay GST-fused NLSs were used for binding assays with Trn1. The amplified cDNAs of hnRNP D NLS and TAP NLS were subcloned at a BamHI/SalI site in expression vectors pGEX-6P-1 and pGEX-2T to prepare GST-fused NLSs for hnRNP D and TAP, respectively. GSTfused JKTBP NLS was prepared as described previously (Kawamura et al., 2002). Point mutants of GST-fused NLSs and Trn1 were prepared with the QuikChange Site-Directed Mutagenesis Kit (Stratagene) according to the manufacturer’s recommendations and confirmed by DNA sequencing. The GST-fused NLSs and point mutants were expressed and purified as previously described (Suzuki et al., 2005; Imasaki et al., 2006). Ran was cloned into pQE30 vector and expressed in E. coli strain BL21(DE3). The cells were suspended in buffer A (20 mM HEPES [pH 7.3], 110 mM CH3COOK, 10 mM b-mercaptoethanol, and 5 mM MgCl2). After sonication on ice, Ran was purified by a combination of Ni-affinity and ion-exchange chromatographies. RanGppNHp complex was prepared by incubating Ran on ice in a solution of 10 mM EDTA, 1 mM GppNHp for 1 hr, and then adding MgCl2 to the solution to make the concentration up to 25 mM MgCl2. In Vitro Nuclear Transport Assay HeLa-S3 cells on an 8-well multitest slide (ICN Biochemicals, Inc.: cat. No. 6040805E) were permeabilized by treatment with transport buffer (TB: 20 mM HEPES [pH 7.3], 110 mM potassium acetate, 2 mM magnesium acetate, 5 mM sodium acetate, 1 mM EGTA, 1 mM DTT, and 1 mg/ml each of aprotinin, leupeptin, and pepstatin) containing 40 mg/ml digitonin for 5 min on ice (Suzuki et al., 2005). After rinsing with TB twice, the slide was immersed in TB for 5 min on ice.

Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc. 65

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Digitonin-permeabilized cells were incubated with 1 mM GST-GFPhnRNP D (337-355) NLS (Suzuki et al., 2005) in the presence of 0.5 mM Trn1 wild-type or its mutants for 20 min at 30 C. All reactions were performed in the presence of 2.5 mM RanGDP and ATP regenerating system containing 1 mM ATP, 5 mM creatine phosphate, and 20 U/ml creatine phosphokinase. After the cells were fixed in 3.7% formaldehyde in TB, fluorescent proteins were detected by epifluorescence microscopy (Olympus BX51). Supplemental Data Supplemental Data include Supplemental Experimental Procedures and six figures and can be found with this article online at http:// www.molecule.org/cgi/content/full/28/1/57/DC1/.

Chook, Y.M., Jung, A., Rosen, M.K., and Blobel, G. (2002). Uncoupling Kapß2 substrate dissociation and Ran binding. Biochemistry 41, 6955–6966. Cingolani, G., Petosa, C., Weis, K., and Muller, C.W. (1999). Structure of importin-b bound to the IBB domain of importin-a. Nature 399, 221– 229. Cingolani, G., Bednenko, J., Gillespie, M.T., and Gerace, L. (2002). Molecular basis for the recognition of a nonclassical nuclear localization signal by Importin b. Mol. Cell 10, 1345–1353. Conti, E., Muller, C.W., and Stewart, M. (2006). Karyopherin flexibility in nucleocytoplasmic transport. Curr. Opin. Struct. Biol. 16, 237–244. Delano, W.L. (2002). The PyMOL Molecular Graphics System (San Carlos, CA: Delano Scientific).

ACKNOWLEDGMENTS

Go¨rlich, D., and Kutay, U. (1999). Transport between the cell nucleus and the cytoplasm. Annu. Rev. Cell Dev. Biol. 15, 607–660.

We thank Drs. M. Yamamoto, M. Kawamoto, H. Sakai, K. Hasegawa, and N. Shimizu for the data collection at SPring-8. Thanks are also due to Drs. T.D. Hurley, M.M. Georgiadis, and Y. Takagi for valuable discussions and comments and to Miss H. Ishida for her technical assistance. This work was supported by Grants-in-Aid for Young Scientists (B) to H.H. (14780515, 16770080) from the Japan Society of the Promotion of Science (JSPS), Grant-in-Aid for Scientific Research on Priority Areas to H.H. (16048226), T.S. (17054035), and M.S. (18054026), national project on protein structural and functional analyses (Protein 3000 project) to M.S., T.S., and H.H. from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), a Kaneko-Narita grant to H.H. from the Protein Research Foundation., and Grants-inAid for Research on Health Sciences focusing on Drug Innovation to M.Y. (SH24402) from the Japanese Health Science Foundation.

Gu¨ttinger, S., Muhlhausser, P., Koller-Eichhorn, R., Brennecke, J., and Kutay, U. (2004). Transportin2 functions as importin and mediates nuclear import of HuR. Proc. Natl. Acad. Sci. USA 101, 2918–2923.

Received: March 29, 2007 Revised: June 22, 2007 Accepted: August 6, 2007 Published: October 11, 2007 REFERENCES Arnold, M., Nath, A., Hauber, J., and Kehlenbach, R.H. (2006a). Multiple importins function as nuclear transport receptors for the Rev protein of the human immunodeficiency virus type I. J. Biol. Chem. 281, 20883–20890.

Hodel, A.E., Harreman, M.T., Pulliam, K.F., Harben, M.E., Holmes, J.S., Hodel, M.R., Berland, K.M., and Corbett, A.H. (2006). Nuclear localization signal receptor affinity correlates with in vivo localization in Saccharomyces cerevisiae. J. Biol. Chem. 281, 23545–23556. Iijima, M., Suzuki, M., Tanabe, A., Nishimura, A., and Yamada, M. (2006). Two motifs essential for nuclear import of the hnRNP A1 nucleocytoplasmic shuttling sequence M9 core. FEBS Lett. 580, 1365– 1370. Imasaki, T., Shimizu, T., Hashimoto, H., Hidaka, Y., Yamada, M., and Sato, M. (2006). Crystallization and preliminary X-ray crystallographic studies of transportin 1 in complex with nucleocytoplasmic shuttling and nuclear localization fragments. Acta Crystallogr. Sect. F Struct. Biol. Cryst. Commun. 62, 785–787. Jakel, S., and Go¨rlich, D. (1998). Importin b, transportin, RanBP5 and RanBP7 mediate nuclear import of ribosomal proteins in mammalian cells. EMBO J. 17, 4491–4502. Jones, T.A., Zou, J.Y., Cowan, S.W., and Kjeldgaard, M. (1991). Improved methods or building protein models in electron density maps and the location of errors in these models. Acta Crystallogr. A 47, 110–119.

Arnold, M., Nath, A., Wohlwend, D., and Kehlenbach, R.H. (2006b). Transportin is a major nuclear import receptor for c-Fos: a novel mode of cargo interaction. J. Biol. Chem. 281, 5492–5499.

Kawamura, H., Tomozoe, Y., Akagi, T., Kamei, D., Ochiai, M., and Yamada, M. (2002). Identification of the nucleocytoplasmic shuttling sequence of heterogeneous nuclear ribonucleoprotein D-like protein JKTBP and its interaction with mRNA. J. Biol. Chem. 277, 2732–2739.

Bayliss, R., Littlewood, T., and Stewart, M. (2000). Structural basis for the interaction between FxFG nucleoporin repeats and Importin-b in nuclear traffiking. Cell 102, 99–108.

Kobe, B. (1999). Autoinhibition by an internal nuclear localization signal revealed by the crystal structure of mammalian importin a. Nat. Struct. Biol. 6, 388–397.

Bru¨nger, A.T., Adams, P.D., Clore, G.M., DeLano, W.L., Gros, P., Grosse-Kunstleve, R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., et al. (1998). Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallogr. D Biol. Crystallogr. 54, 905–921.

Kraulis, P.J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. J. Appl. Crystallogr. 24, 946–950.

Cansizoglu, A.E., Lee, B.J., Zhang, Z.C., Fontoura, B.M., and Chook, Y.M. (2007). Structure-based design of a pathway-specific nuclear import inhibitor. Nat. Struct. Mol. Biol. 14, 452–454. Catimel, B., Teh, T., Fontes, M.R., Jennings, I.G., Jans, D.A., Howlett, G.J., Nice, E.C., and Kobe, B. (2001). Biophysical characterization of interactions involving importin-a during nuclear import. J. Biol. Chem. 36, 34189–34198. Chook, Y.M., and Blobel, G. (1999). Structure of the nuclear transport complex karyopherin-b2-Ran GppNHp. Nature 399, 230–237. Chook, Y.M., and Blobel, G. (2001). Karyopherins and nuclear import. Curr. Opin. Struct. Biol. 11, 703–715.

Lee, S.J., Sekimoto, T., Yamashita, E., Nagoshi, E., Nakagawa, A., Imamoto, N., Yoshimura, M., Sakai, H., Chong, K.T., Tsukihara, T., et al. (2003). The structure of importin-b bound to SREBP-2: nuclear import of a transcription factor. Science 302, 1571–1575. Lee, S.J., Matsuura, Y., Liu, S.M., and Stewart, M. (2005). Structural basis for nuclear import complex dissociation by RanGTP. Nature 435, 693–696. Lee, B.J., Cansizoglu, A.E., Suel, K.E., Louis, T.H., Zhang, Z., and Chook, Y.M. (2006). Rules for nuclear localization sequence recognition by karyopherinß2. Cell 126, 543–558. McCoy, A.J., Grosse-Kunstleve, R.W., Storoni, L.C., and Read, R.J. (2005). Likelihood-enhanced fast translation functions. Acta Crystallogr. D Biol. Crystallogr. 61, 458–464.

66 Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc.

Molecular Cell Structures of Human Transportin 1 Bound to NLSs

Merritt, E.A., and Murphy, M.E. (1994). Raster3D Version 2.0. A program for photorealistic molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 50, 869–873. Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallogr. D Biol. Crystallogr. 53, 240–255. Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326. Pollard, V.W., Michael, W.M., Nakielny, S., Siomi, M.C., Wang, F., and Dreyfuss, G. (1996). A novel receptor-mediated nuclear protein import pathway. Cell 86, 985–994. Siomi, H., and Dreyfuss, G. (1995). A nuclear localization domain in the hnRNP A1 protein. J. Cell Biol. 129, 551–560. Suzuki, M., Iijima, M., Nishimura, A., Tomozoe, Y., Kamei, D., and Yamada, M. (2005). Two separate regions essential for nuclear import of the hnRNP D nucleocytoplasmic shuttling sequence. FEBS J. 272, 3975–3987.

Tran, E.J., and Wente, S.R. (2006). Dynamic nuclear pore complexes: life on the edge. Cell 125, 1041–1053. Truant, R., Kang, Y., and Cullen, B.R. (1999). The human Tap nuclear RNA export factor contains a novel transportin-dependent nuclear localization signal that lacks nuclear export signal function. J. Biol. Chem. 274, 32167–32171. Vetter, I.R., Arndt, A., Kutay, U., Go¨rlich, D., and Wittinghofer, A. (1999). Structural view of the Ran-Importin b interaction at 2.3 A˚ resolution. Cell 97, 635–646.

Accession Numbers Refined coordinates and structure factors have been deposited in the Protein Data Bank (accession codes: 2Z5J, NLS-free Trn1; 2Z5N, Trn1 bound to hnRNP D NLS; 2Z5K, Trn1 bound to TAP NLS; and 2Z5O, Trn1 bound to JKTBP NLS).

Molecular Cell 28, 57–67, October 12, 2007 ª2007 Elsevier Inc. 67