Crystal structure of pea Toc34, a novel GTPase of the chloroplast protein translocon

The ribbon structure of the Toc34 dimer (Fig. 1a) resembles a butterfly whose wings, antennae and body are represented by -helices and -sheets of the dimer, the two longest loops and the dimer interface (Fig. 1a,b), respectively. The overall fold of the Toc34 monomer is divided into a globular GTP-binding domain (G-domain) and a small C-terminal -helical part. Toc34 consists of nine -strands, of which six form a -sheet core flanked by six - and two 3₁₀-helices. In contrast, the canonical GTP-binding protein H-ras-p21 (Ras)¹⁵ contains six -strands and five -helices (Figs 2a, 3). The most obvious structural difference between Toc34 and Ras is the additional 36 N-terminal residues of Toc 34, which form the first short -strand (0), two helices (-1 and 0) and the two loops connecting the strand and the helices (Figs 1a, 3). The next 212 residues constitute a modified G-domain, with a number of insertions that significantly enlarge the structure in comparison to Ras (170 residues) (Fig. 2a,b). The structural superposition of Toc34−GDP and Ras−GppNHp reveals that Toc34 has six inserts, I1−I6 (Fig. 3), which, except for I5, have implications for dimerization and GTPase activity (discussed below).

Figure 1. Structure of the GDP-bound form of the Toc34 dimer and its interface. a, A ribbon drawing of the Toc34 dimer. The two molecules are related to one another by a crystallographic two-fold axis. Monomer1 is colored green, dark blue and yellow; monomer1' is colored orange, violet and light blue. GDP is represented as a ball-and-stick model, and Mg2+ is shown in magenta. Switches I and II, and the longest loop connecting 6 and 6 are labeled. b, Schematic representation of part of the residues involved in dimerization. The residues from interacting monomers and their GDPs are labeled as A and B. MOLSCRIPT37 and RASTER3D38 were used for generating Figs 1a, 2, 4a,b, and LIGPLOT39 for (b). Full Figure and legend (73K)

Figure 2. Stereo view of the superposition of the C and the G motifs of Toc34 and H-ras-p21. a, Cs of H-rasp-21−GDP (PDB accession code 4Q21) and Toc34 are shown in magenta and green, respectively. Switches I and II, and the longest loop of Toc34 are labeled. b, The structural overlay of the G motifs of Toc34−GDP and Ras−GDP, colored as in (a). Two candidates, Ser 68 or Ser 72 and Gln 71 or Asn 104, in Toc34 have been proposed to function like Thr 35 and Gln 61 of Ras, respectively (see text). Full Figure and legend (64K)

Figure 3. Sequence alignment of the Toc34/Toc159-G subfamily. The secondary structural elements, deduced from this work (green), and those of Ras−GppNHp (PDB 5P21) (magenta) are shown above the sequences. Red and black asterisks represent residues from one monomer and the other monomer, respectively, interacting with GDP. Blue wavy lines above the sequences depict the G1−G5 motifs. Residues involved in Toc34 dimerization, shown below the sequences, are depicted as follows: red triangles for hydrogen bonding, blue stars for hydrophobic interaction and solid orange circles for van der Waals forces. The putative switches I and II are enclosed by light blue boxes, and the inserts I1−I6 are shown with a gray background. The longest loop (residues 209−230) between 6 and 6 is depicted as an orange line above the sequences, and the location of NKxD motif of Toc34 is shown by a red line below the sequences. All alignments were obtained using Pileup and Prettybox in GCG. The partial rice Toc34 sequence was obtained from the Monsanto Rice Genome Database. Full Figure and legend (145K)

Although Toc34 was purified in the absence of GDP, extra electron density in the crystal fits perfectly to a GDP molecule. A Mg²⁺ ion was observed at the expected nucleotide binding region (Fig. 4a), suggesting that Toc34 has a high affinity for GDP. The G1 motif (P-loop), Gly 46-X₄-Gly-Lys-Ser 53, of Toc34 has a topology very similar to that of Ras−GDP in their superposition (Fig. 2b). However, the replacement of Ala 18 of Ras by Ser 54 in Toc34 led to the formation of an additional polar interaction between the side chain of Ser 54 and O2 of the GDP (Fig. 4a).

Figure 4. The GTP-binding region of Toc34. All contacts depicted in (a,b) have hydrogen bond distances 3.5 Å. a, Stereo view of the Toc34 residues in contact with GDP and Mg2+. The residues contributing to the nucleotide binding by the other interacting monomer are colored orange. An electron density (2Fo - Fc)c simulated annealing map (blue) was calculated after omitting GDP and Mg2+, and contoured at 1.0 . b, Stereo view of the superposition of GDP of Toc34 and Ras−RasGAP complex (PDB accession code 1WQ1). Only GDP of Ras is shown. AlF3 and the RasGAP residues are colored light blue and purple, respectively; and those of Toc34, colored orange. Although Lys 133 superimposes well on Lys 789, Tyr 132 does not on Phe 788. c, Rate of GTP hydrolysis of the wild type (square) and the R128A mutant (diamond) Toc34 as a function of GTP concentration. d, Rate of GTP hydrolysis of the wild type (solid circle) and the R128A mutant (open circle) Toc34 as a function of protein concentration at a saturating concentration of GTP (200 M). Full Figure and legend (60K)

Toc34 does not contain the classic G2 motif signified by the invariant Thr 35 of Ras, which interacts with the -phosphate of the GTP¹⁶. In the structural superposition (Fig. 2b), Ser 72 of Toc34, which is in a similar location to Thr 35 of Ras, is a good candidate to perform the task similarly to that of Ras Thr 35. The more conserved Ser 68 is another potential candidate, although it is farther from the position of Ras Thr 35 than Ser 72. Another unique feature of the G2 motif of Toc34 is Glu 73 interacting with Mg²⁺ and O2 of the GDP (Fig. 4a).Unlike Ras, where Ser 17 of G1 is the only residue coordinating Mg²⁺, both Ser 53 and Glu 73 are ligands for Mg²⁺ in Toc34. Because the side chain of Glu 73 evidently occupies the presumed location of -phosphate, it has to move away so that the site can accommodate GTP.

For the G3 motif, Toc34 has Asp 93-X-X-Gly-Leu 97 instead of the classic Asp 57-X-X-Gly-Gln 61 sequence of Ras. Asp 93 of Toc34 superimposes well on Asp 57 of Ras (their Cs are 1.59 Å apart). Similarly, a water molecule mediates the interaction of Asp 93 with Mg²⁺ in Toc34 (Fig. 4a). The invariant Gly 96 of Toc34 superimposes well on Gly 60 of Ras-GppNHp (their Cs are 1.88 Å apart) and Gly 100 of hGBP1-GppNHp (their Cs are 1.02 Å apart) but diverges from the corresponding Gly of Ras−GDP (Fig. 2b), suggesting that the GTP-induced conformational change would not be significant either at switch II (residues 97−120) or at Gly 96 of Toc34. More important, Gly 96 of Toc34 is followed by Leu 97 instead of a Gln residue, which is absolutely essential for GTP hydrolysis in Ras. This implies that the mechanism of GTP hydrolysis in Toc34 differs from that in Ras-like GTPases. Nonetheless, either Gln 71 or Asn 104 may be the catalytic residue in Toc34. The strictly conserved Asn 104 is relatively stable (low B-factor) and superimposes moderately well on Gln 61 of Ras-GDP (their Cs are 3.85 Å apart) (Fig. 2b). In contrast to Asn 104, Gln 71 of I1, which is conserved among Toc34/Toc33 orthologs, is mobile (high B-factor) and close enough to interact with Arg 133, which may have a catalytic role in Toc34 (see below). In the structural superposition of Toc34-GDP and hGBP1-GppNHp, Gln 71 of Toc34 lies behind His 74 of hGBP1 with respect to GppNHp. This large GTPase also has a Leu residue at the position corresponding to Gln 61 of Ras, and His 74 was suggested to play a catalytic role in hGBP1, whose GTP hydrolysis mechanism is apparently different than that of Ras-like GTPases.

Toc34 contains a classic G4 sequence (Asn/Thr 216-Lys-X-Asp 219) but it does not interact with GDP. Instead, Thr 162 makes hydrophobic interactions with GDP, NH of His 163 hydrogen bonds to O6 of GDP and its imidazole ring forms − interactions with the guanine ring (Fig. 4a). His 163 also superimposes well on Lys 117 of G4 in Ras (their Cs are 0.94 Å apart) (Fig. 2b). Thus, Thr 162 and His 163 can be the corresponding G4 motif of Toc34 (Figs 2b, 3).

A significant number of residues in the longest loop, encompassing residues 209−230 (between secondary elements 6 and 6), are highly conserved in all Toc34 and Toc159 homologs identified so far (Fig. 3). This loop of pea Toc34 contains the only Cys residue (Cys 215) that was shown to crosslink to other Toc components with Cys crosslinkers¹⁸, suggesting a role for this strictly conserved residue. Because G5 is located on the longest loop and the guanine exchange factor (GEF) has yet to be identified for the Toc GTPases, the longest loop may have function in the nucleotide exchange process.

Each asymmetric unit of the Toc34 crystal, which belongs to the C2 space group, contains three molecules, and each molecule forms a dimer with one of its adjacent molecules. In other words, monomer1 and monomer1' form a dimer by the crystallographic two-fold symmetry (Fig. 1a), whereas monomer2 with monomer3', and monomer3 with monomer2' form dimers by noncrystallographic two-fold symmetry. Nevertheless, all three dimers are almost identical, and the interface of each buries a surface area of 2,750 Ų using a probe radius of 1.4 Å. This implies the existence of a specific contact between two monomers that may have a biological relevance¹⁹.

Dimerization involves both GDPs, whose sugar rings are buttressed by Pro 169, Asp 170 and Tyr 132 from the other interacting monomer via van der Waals and hydrophobic interactions (Figs 1b, 3). More important, the side chain of Arg 133 of one monomer donates one hydrogen bond to the -phosphate of GDP and two to the amide oxygen of Ser 68, and makes one hydrophobic contact with Gly 49 of the other monomer (Fig. 1b). Consequently, I4, which contains Pro 169 and Asp 170, shields the ribose ring and part of the guanine ring from solvent such that the nucleotide exchange can occur only by at least a partial disruption of the dimer (Fig. 4a).

The dimer interface includes switches I and II (Figs 1a, 3), which is reminiscent of the interface between a Ras-like GTPase and its GAP¹⁶. In addition, there is a catalytic Arg residue, called the 'arginine finger', in the subunit of trimeric G protein (G) and in many GAPs^{16, 20}. In the Toc34 dimer, the side chain of Arg 133 from one monomer is deeply inserted into the GDP-binding subregion of the other monomer, where it hydrogen bonds with the -phosphate (Figs 1b, 4a). Furthermore, the preceding residue to the Arg finger in RasGAPs, RhoGAPs, RapGAPs and YptGAPs is either a Phe (Fig. 4b) or Tyr residue²⁰, and the main chain of the Arg finger is stabilized by another Arg residue. In Toc34, Tyr 132 precedes Arg 133, and the side chain of Arg 181 hydrogen bonds to the amide oxygen of Arg 133 (Fig. 4b). Although Tyr 132 of Toc34 does not superimpose on Phe 788 of RasGAP, its side chain stabilizes the surrounding hydrophobic pocket, similar to Phe 788 in RasGAP. Therefore, the structure of the Toc34 dimer suggests that Arg 133 may function as an Arg finger. The Arg finger in G, and most small GTPases characterized thus far, is provided either by the protein itself or by its corresponding GAP. However, the Arg finger of each monomer in the TOC34 dimer is supplied by its dimeric partner.

When dimeric partners serve as each other's GAP, the dimer should bind to aluminum fluoride (AlF_x), mimicking the transition state of GTP hydrolysis¹⁶. In contrast, the disruption of the dimer should deprive Toc34 of its GAP, leading to a significant reduction in its GTPase activity. To test this, we mutated Arg 128, one of the key residues for dimerization (Fig. 1b), to an Ala residue. Gel filtration experiments revealed that the mutant molecules existed only as monomers in solution (data not shown). Furthermore, the wild type Toc34 showed a broad ¹⁹F NMR signal at 66 p.p.m., arising from GDP-AlF_x complex²², whereas the mutant did not (data not shown). GTPase activity assays also revealed that the mutant had a reduced activity compared to that of the wild type, even though their substrate binding affinities (K_m) were similar (Fig. 4c). To investigate whether the mutation did not impair the GTPase activity of the mutant Toc34, an additional GTPase assay was performed by increasing concentration of the wild type and the mutant Toc34 at a saturating concentration of GTP (200 M). The results suggested that the mutant protein could still hydrolyze GTP, although with a reduced efficiency (Fig. 4d). Taken together, our results are consistent with the proposal that dimerization is required for mutual activation of the two Toc34 GTPases. However, we cannot exclude the possibility that there are additional conformational changes necessary for GTP hydrolysis and that the Arg mutation may have more affects than disrupting dimerization.

Reciprocal activation of two GTPases has been shown to occur between the signal recognition particle (SRP) and the -subunit of the SRP receptor (SR), which share a high sequence homology in their G-domains²³. Two strictly conserved Arg residues from each were proposed as candidates for the Arg fingers in SRP and SR²⁴. Thus, the Toc34 dimer may provide a structural ground, to support the proposed mechanism of the reciprocal activation between SRP and SR.

The G-domain of Toc159 (Toc159-G) shares a high degree of sequence homology with the Toc34 G-domain (Fig. 3). The most conserved regions, besides the GTP-binding motifs, lie in the regions responsible for dimer formation²⁵ (Fig. 3). Arg 133 of Toc34 has its counterparts in both atToc159 (Arg 953) and pea Toc159 (Arg 926), even though they are one residue out of register relative to Arg 133 (Fig. 3). Furthermore, there is a conserved Arg 978 in pea Toc159 that may have a role similar to Arg 181 of Toc34. Crosslinking experiments performed at the initial stage of the precursor binding also showed that Toc159 was crosslinked in 2:1 with Toc75 (ref. 26). Therefore, Toc159 may also form a dimer through which it attains GTPase activation, similar to the Toc34 dimer. However, Hiltbrunner et al.²⁷ recently reported that atToc159 interacted with atToc33. Therefore, atToc159 and atToc33 also possibly form heterodimers, either in cytosol or on the outer membrane. Interchangeable homodimeric and heterodimeric states may even serve as a regulatory mechanism.

GTPase assays were carried out as described²⁸. A time course experiment was carried out in a 20 l reaction mixture containing 50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 10% (v/v) glycerol, 3.4 M of the wild type or the R128A mutant Toc34, 1.7 nM [-³²P]GTP (1 Ci) (NEN), a specified concentration of GTP and the same concentration of MgCl₂ (Fig. 4c). The reaction was incubated at 37 °C. At various time points, an aliquot was removed and the reaction was stopped by adding 70 M EDTA. A fraction of the stopped reaction (0.35 l) from each time point was spotted on a polyethyleneimine cellulose thin layer chromatography plate and developed in 1 M formic acid and 0.5 M LiCl. The spots were quantified using the Phosphoimager SP (Molecular Dynamic). The rate of GTP hydrolysis at each GTP concentration was calculated and plotted. A similar assay was performed but with 200 M of GTP and MgCl₂ and various concentrations of the wild type or the mutant Toc34 (Fig. 4d). The reactions were stopped after incubation at 37 °C for 30 min.

The Toc34 crystals were grown at 4 °C by hanging drop vapor diffusion. A 2 l protein solution (10 mg ml^-1 in 50 mM Tris-HCl, pH 8.0, and 0.1 M NaCl) was mixed with a 2 l reservoir solution containing 22% (w/v) PEGMME 5000 and 10% (v/v) glycerol in 0.1 M HEPES, pH 6.5. The crystals belong to space group C2, with cell dimensions a = 143.16 Å, b = 78.67 Å, c = 67.28 Å and = 91.1°, and diffract to 2.0 Å. The structure of the GDP-bound Toc34 was determined using multiwavelength anomalous diffraction (MAD) phasing²⁹ applied to the SeMet analog³⁰. The MAD experiments for SeMet-Toc34 were conducted at the BL18B synchrotron beam line of Photon Factory (PF) in High Energy Accelerator Research Organization (KEK), Tsukuba, Japan. A single crystal with approximate dimensions of 0.1 0.2 0.3 mm³ was frozen at 110 K using PEGMME 5000 and glycerol as cryoprotectants. The MAD data were collected at four wavelengths. Two energies were chosen near the absorption peak and edge of selenium, 0.9794 Å and 0.9796 Å, that correspond to the maximum f'' and the minimum f', respectively. Two remote energies were selected as reference wavelengths in 0.9680 Å and 0.9802 Å. The diffraction data were collected using Quantum 4R CCD (Area Detector System Corporation), and indexed and integrated using DPS/MOSFLM³¹.

SOLVE³² was used to locate selenium sites and to generate the initial MAD phases at 3.0 Å resolution. The initial phases were extended and further improved to 2.5 Å by DM³³. XtalView³⁴ and O³⁵ were used to examine the electron density maps and molecular models. CNS³⁶ iterative cycles of refinement were performed using the average of the four data sets. During the refinement, noncrystallographic symmetry was imposed initially but released after three monomers in the asymmetric unit were built (Table 1). The current model contains residues 8−262 of monomer1, residues 7−262 of monomer2 and residues 7−263 of monomer3. The C root mean square (r.m.s.) deviations between monomers 1 and 2, 1 and 3, and 2 and 3 are 0.46 Ų, 0.61 Ų and 0.62 Ų, respectively. 90.1% of the residues are in the most favored regions, with the remaining ones in the additional allowed regions.

Table 1. MAD data and refinement statistics Full Table

	Top

1. Kessler, F., Blobel, G., Patel, H.A. & Schnell, D.J. Science 266, 1035−1039 (1994). | PubMed | ISI | ChemPort |
2. Seedorf, M., Waegemann, K. & Soll, J. Plant J. 7, 401−411 (1995). | Article | PubMed | ISI | ChemPort |
3. Chen, D. & Schnell, D.J. J. Biol. Chem. 272, 6614−6620 (1997). | Article | PubMed | ISI | ChemPort |
4. Keegstra, K. & Froehlich, J E. Curr. Opin. Plant Biol. 2, 471−476 (1999). | Article | PubMed | ISI | ChemPort |
5. Schleiff, E. & Soll, J. Planta 211, 449−456 (2000). | Article | PubMed | ISI | ChemPort |
6. Olsen, L.J., Theg, S.M., Selman, B. R. & Keegstra, K. J. Biol. Chem. 264, 6724−6729 (1989). | PubMed | ISI | ChemPort |
7. Ma, Y., Kouranov, A., LaSala, S.E. & Schnell, D.J. J. Cell Biol. 134, 315−327 (1996). | Article | PubMed | ISI | ChemPort |
8. Akita, M., Nielsen, E. & Keegstra, K. J. Cell Biol. 136, 983−994 (1997). | Article | PubMed | ISI | ChemPort |
9. Perry, S.E. & Keegstra, K. Plant Cell 6, 93−105 (1994). | Article | PubMed | ISI | ChemPort |
10. Schnell, D.J., Kessler, F. & Blobel, G. Science 266, 1007−1012 (1994). | PubMed | ISI | ChemPort |
11. Hinnah, S. C., Hill, K., Wagner, R., Schlicher, T. & Soll, J. EMBO J. 16, 7351−7360 (1997). | Article | PubMed | ChemPort |
12. Reumann, S., Davila Aponte, J. & Keegstra, K. Proc. Natl. Acad. Sci. USA 96, 784−789 (1999). | Article | PubMed | ChemPort |
13. Kouranov, A. & Schnell, D.J. J. Cell Biol. 139, 1677−1685 (1997). | Article | PubMed | ISI | ChemPort |
14. Sveshnikova, N., Soll, J. & Schleiff, E. Proc. Natl. Acad. Sci. USA 97, 4973−4978 (2000). | Article | PubMed | ChemPort |
15. Pai, E. F. et al. Nature 341, 209−214 (1989). | Article | PubMed | ISI | ChemPort |
16. Wittinghofer, A. in Frontiers in molecular biology − GTPases (ed. Hall, A.) 222−311 (Oxford Univ. Press, New York; 2000).
17. Prakash, B., Renault, L., Praefcke, G.J.K., Herrmann, C. & Wittinghofer, A. EMBO J . 19, 4555−4564 (2000). | Article | PubMed | ChemPort |
18. Seedorf, M. & Soll, J. FEBS Lett. 367, 19−22 (1995). | Article | PubMed | ISI | ChemPort |
19. Janin, J. Nature Struct. Biol. 12, 973−974 (1997). | Article |
20. Scheffzek, K., Ahmadian, M.R. & Wittinghofer, A. TIBS 23, 257−262 (1998). | Article | PubMed | ChemPort |
21. Greasely, S.E. et al. Nature Struct. Biol. 9, 797−806 (1995). | Article |
22. Nelson, D.J. & Martin, R.B. J. Inorg. Biochem. 43, 37−43 (1991). | Article | ISI | ChemPort |
23. Powers, T. & Walter, P. Science 269, 1422−1424 (1995). | PubMed | ISI | ChemPort |
24. Montoya, G., Kaat, K., Moll, R., Schafer, G. & Sinning, I. Structure Fold. Des. 8, 515−525 (2000). | Article | PubMed | ISI | ChemPort |
25. Chen, L.-J. & Li, H.-M. Plant Physiol. 110, 714 (1996).
26. Chen, K., Chen, X. & Schnell, D.J. Plant Physiol. 122, 813−822 (2000). | Article | PubMed | ISI | ChemPort |
27. Hilbrunner, A. et al. J. Cell Biol. 154, 309−316 (2001). | Article | PubMed |
28. DeLuca-Flaherty, C., McKay, D.B., Parham, P. & Hill, B.L. Cell 62, 875−887 (1990). | Article | PubMed | ChemPort |
29. Hendrickson, W.A. Science 254, 51−58 (1991). | PubMed | ISI | ChemPort |
30. Hendrickson, W.A., Horton, J.R. & LeMaster, D.M. EMBO J. 9, 1665−1672 (1990). | PubMed | ISI | ChemPort |
31. Rossmann, M.G. & van Beek, C.G. Acta Crystallogr. D 55, 1631−1653 (1999). | Article | PubMed | ISI |
32. Terwilliger, T.C. & Berendzen, J. Acta Crystallogr. D 55, 849−861 (1999). | Article | PubMed | ISI |
33. Cowtan, K.D. Acta Crystallogr. D 52, 43−48 (1996). | Article | PubMed | ISI | ChemPort |
34. McRee, D.E. J. Struct. Biol. 125, 156−165 (1999) | Article | PubMed | ISI | ChemPort |
35. Jones, T.A., Zou, J.Y., Cowan, S.W. & Kjeldgaard, M. Acta Crystallogr. D 47, 110−119 (1991). | Article |
36. Brünger, A.T. et al. Acta Crystallogr. D 54, 905−921 (1998). | Article | PubMed | ISI |
37. Kraulis, P.J. J. Appl. Crystallogr. 24, 946−950 (1991). | Article | ISI |
38. Merrit, E.A. & Murphy, M.E.P. Acta Crystallogr. D 50, 869−873 (1994). | Article | PubMed |
39. Wallace, A.C., Laskowski, R.A. & Thornton, J.M. Protein Eng. 8, 127−134 (1995). | PubMed | ISI | ChemPort |

	Top