Crystal Structures of the CP1 Domain from Thermus thermophilus Isoleucyl-tRNA Synthetase and Its Complex with l-Valine*

Isoleucyl-tRNA synthetase (IleRS) links tRNAIle with not only its cognate isoleucine but also the nearly cognate valine. The CP1 domain of IleRS deacylates, or edits, the mischarged Val-tRNAIle. We determined the crystal structures of the Thermus thermophilus IleRS CP1 domain alone, and in its complex with valine at 1.8- and 2.0-Å resolutions, respectively. In the complex structure, the Asp328 residue, which was shown to be critical for the editing reaction against Val-tRNAIle by a previous mutational analysis, recognizes the valine \batchmode \documentclass[fleqn,10pt,legalpaper]{article} \usepackage{amssymb} \usepackage{amsfonts} \usepackage{amsmath} \pagestyle{empty} \begin{document} \(\mathrm{N}\mathrm{H}_{3}^{+}\) \end{document} group. The valine side chain binding pocket is only large enough to accommodate valine, and the placement of an isoleucine model in this location revealed that the additional methylene group of isoleucine would clash with His319. The H319A mutant of Escherichia coli IleRS reportedly deacylates the cognate Ile-tRNAIle in addition to Val-tRNAIle, indicating that the valine-binding mode found in this study represents that in the Val-tRNAIle editing reaction. Analyses of the Val-tRNAIle editing activities of T. thermophilus IleRS mutants revealed the importance of Thr228, Thr229, Thr230, and Asp328, which are coordinated with water molecules in the present structure. The structural model for the Val-adenosine moiety of Val-tRNAIle bound in the IleRS editing site revealed some interesting differences in the substrate binding and recognizing mechanisms between IleRS and T. thermophilus leucyl-tRNA synthetase. For example, the carbonyl oxygens of the amino acids are located opposite to each other, relative to the adenosine ribose ring, and are differently recognized.

group. The valine side chain binding pocket is only large enough to accommodate valine, and the placement of an isoleucine model in this location revealed that the additional methylene group of isoleucine would clash with His 319 . The H319A mutant of Escherichia coli IleRS reportedly deacylates the cognate Ile-tRNA Ile in addition to Val-tRNA Ile , indicating that the valine-binding mode found in this study represents that in the Val-tRNA Ile editing reaction. Analyses of the Val-tRNA Ile editing activities of T. thermophilus IleRS mutants revealed the importance of Thr 228 , Thr 229 , Thr 230 , and Asp 328 , which are coordinated with water molecules in the present structure. The structural model for the Val-adenosine moiety of Val-tRNA Ile bound in the IleRS editing site revealed some interesting differences in the substrate binding and recognizing mechanisms between IleRS and T. thermophilus leucyl-tRNA synthetase. For example, the carbonyl oxygens of the amino acids are located opposite to each other, relative to the adenosine ribose ring, and are differently recognized.
Aminoacyl-tRNA synthetases (aaRSs) 1 catalyze the esterification of an amino acid to its cognate tRNA. This reaction proceeds in two steps: the synthesis of an aminoacyladenylate, as an activated intermediate, from the amino acid and ATP, and the transfer of the aminoacyl moiety to the 3Ј-terminal of the cognate tRNA to yield the aminoacyl-tRNA (1). To maintain accurate protein biosynthesis, each aaRS must discriminate between its cognate amino acid and other similar amino acids (2,3). Some aaRSs, including the isoleucyl-, leucyl-, and valyl-tRNA synthetases (IleRS, LeuRS, and ValRS, respectively), have a specific editing activity that hydrolyzes the misaminoacylated tRNAs ("post-transfer editing") (4 -6). For example, IleRS also recognizes valine, which is smaller than the cognate isoleucine by only one methylene group, and mischarges it with tRNA Ile . Then, the mischarged Val-tRNA Ile is hydrolyzed to valine and tRNA Ile in the post-transfer editing pathway. As for IleRS, another editing pathway (pre-transfer editing) also exists, in which the misactivated Val-AMP is directly hydrolyzed to valine and AMP in the presence of tRNA Ile (7,8). Biochemical experiments linked the specific location of the editing site to the connective polypeptide 1 (CP1) domain, a large insertion in the aminoacylation catalytic Rossmann-fold domain (9).
Previously, we determined the crystal structures of the Thermus thermophilus full-length IleRS complexed with isoleucine and with valine and showed that the editing site is in the highly conserved threonine-rich region of the CP1 domain (10). In addition, the crystal structure of Staphylococcus aureus IleRS complexed with tRNA Ile and mupirocin (an analogue of isoleucyl-adenylate) revealed that the 3Ј-terminal of tRNA Ile is located in the CP1 domain (although it was not completely resolved) (11). This suggested that when a nearly cognate amino acid is charged to a tRNA, the acceptor stem flips from the aminoacylation site to the editing site, while the rest of the tRNA remains bound. However, the B factors of many atoms in the CP1 domains of these structures were high and some residues were disordered, since the CP1 domain is quite mobile relative to the rest of the protein (10,11). Furthermore, in our previous study (10), the omit map electron density for valine was not clear enough for us to determine its orientation precisely, and the coordinates were determined by analogy with the related ValRS system (12). The crystal structure of T. thermophilus ValRS complexed with tRNA val accurately revealed the location of the 3Ј terminus of tRNA val in the CP1 domain with complete resolution, although that of threonine (which is edited by ValRS) remains to be elucidated (12). Based * This work was supported by a grant-in-aid for Scientific Research S from the Japan Society for Promotion of Science (JSPS); a grant-in-aid for Scientific Research in Priority Areas from the Ministry of Education, Culture, Sports, Science and Technology (MEXT) of Japan; the RIKEN Structural Genomics/Proteomics Initiative (RSGI); and the National Project on Protein Structural and Functional Analyses, MEXT. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The on the location of valine in the IleRS CP1 domain and that of the tRNA 3Ј-terminal adenosine in the ValRS CP1 domain, we built the structural models for Val-tRNA Ile in IleRS and Thr-tRNA val in ValRS, assuming that their binding modes are analogous (12). However, differences between their binding modes were subsequently reported (13). In addition, no water molecules could be assigned around the valine bound to the CP1 domain in the full-length T. thermophilus IleRS structure. Therefore, the precise valine-binding mode and the mechanisms by which the IleRS CP1 domain selectively recognizes valine and catalyzes the hydrolysis of Val-tRNA Ile remained to be elucidated.
Recently, the crystal structure of T. thermophilus LeuRS complexed with the post-transfer editing substrate analogue was determined (14). In comparison with IleRS, the CP1 domain is inserted at a different point in the enzyme, and its rotational orientation differs by nearly 180°from IleRS (14,15). The structure provided the precise substrate-binding mode, in which conserved Thr residues recognize the substrate (14). However, the LeuRS crystal structures revealed that both the pre-and post-transfer editing analogues are bound in one site in the CP1 domain, although a mutational analysis showed that distinct residues are needed between the pre-and posttransfer editing reactions in IleRS (16). Therefore, the substrate-binding modes may differ between IleRS and LeuRS.
In the present study, to study specifically the post-transfer editing mechanism of IleRS, we isolated the T. thermophilus IleRS CP1 domain (201-384 amino acids), which contains the post-transfer editing site, but lacks a part of the pre-transfer editing site, based on the previous model (12,16). We determined its crystal structure at 1.8-Å resolution. Then, we determined its cocrystal structure with valine at 2.0-Å resolution. In the complex structure, the valine is located in the same pocket as in the case of the modeled Val-tRNA Ile (12,16); however, its orientation within the pocket is quite different. This new valine-binding mode can explain the mutational data well, unlike the previous model. Furthermore, we examined the post-transfer editing activity of some mutant IleRSs. Based on the high resolution structural data and these and other mutational data, we created a new structural model of the Val-adenosine moiety of Val-tRNA Ile in the editing state. In this model, the substrate recognition differs from that in the LeuRS system.

EXPERIMENTAL PROCEDURES
Protein Preparation-The gene fragment encoding the T. thermophilus IleRS CP1 domain (201-384 amino acids with the initiating methionine) was subcloned into pET26b (Novagen). The construct was designed to have only the post-transfer editing site, but not the pretransfer editing site, based on the previous model (12,16), to study specifically the post-transfer editing mechanism. The plasmid was transformed into E. coli strain JM109(DE3). For protein overexpression, the cells were grown to an OD 600 of 0.8, and the expression was induced with 1 mM isopropyl-␤-D-thiogalactopyranoside for 4 h. The cells were harvested and sonicated in 50 mM K 2 HPO 4 buffer (pH 6.0) containing 5 mM MgCl 2 , 10 mM dithiothreitol, and 1 mM phenylmethylsulfonyl fluoride. The insoluble cell debris was removed by centrifugation at 15,000 ϫ g for 30 min at 4°C. The supernatant was heattreated at 70°C for 30 min to denature the E. coli proteins. The heat-treated mixture was centrifuged at 15,000 ϫ g for 1 h at 4°C. Then, 2.4 M (NH 4 ) 2 SO 4 was added to the supernatant to a final concentration of 0.8 M. The mixture was applied to a 50 ml column of butyl-Toyopearl (Tosoh) equilibrated with 50 mM K 2 HPO 4 buffer (pH 6.0) containing 0.8 M (NH 4 ) 2 SO 4 , 5 mM MgCl 2 , and 1 mM dithiothreitol. The CP1 domain was eluted with a linear gradient of 0.8 -0 M (NH 4 ) 2 SO 4 . The fractions containing the CP1 domain were pooled, dialyzed against 50 mM Tris-HCl buffer (pH 7.5) containing 5 mM MgCl 2 and 1 mM dithiothreitol, and loaded onto a Mono Q column (Amersham Biosciences) using an Amersham fast protein liquid chromatography system. The enzyme was eluted with a linear gradient of 0 -1 M NaCl. The CP1 domain protein was further purified on a UnoQ column (Bio-Rad) using an fast protein liquid chromatography system eluted with a linear gradient of 0 -1 M NaCl and was dialyzed against 10 mM Tris-HCl buffer (pH 8.0) containing 5 mM MgCl 2 and 5 mM ␤-mercaptoethanol. The final purity of the protein was monitored by SDS-PAGE.
Crystallization and Data Collection-For crystallization, the hanging drop vapor diffusion method was used, by mixing 1 l of the protein solution (10 mg⅐ml Ϫ1 ) with 1 l of the reservoir solution (50 mM Hepes-NaOH buffer (pH 7.0) containing 2.0 M (NH 4 ) 2 SO 4 and 5% 2-propanol) and by equilibrating this mixture against 500 l of the reservoir solutions at 20°C. Crystals (space group P4 1 2 1 2; unit-cell parameters a ϭ b ϭ 102.7 Å, c ϭ 83.8 Å) were grown for 3 days to dimensions of ϳ0.3 ϫ 0.3 ϫ 0.3 mm 3 . To obtain cocrystals of the CP1 domain complexed with valine, reservoir solutions containing 100 mM valine were used, and the crystals thus obtained were transferred to a solution containing 55 mM Hepes-NaOH buffer (pH 7.0), 2.2 M (NH 4 ) 2 SO 4 , 5.5% 2-propanol, and 100 mM valine, 24 h before the data collection. The diffraction datasets of the ligand-free crystal were collected at beamline BL41XU at SPring-8 (Harima) to 1.8-Å resolution, and those of the complex crystal were collected at beamline BL44XU at SPring-8 to 2.0 Å resolution. A single crystal, flash-frozen at a temperature of Ϫ173°C, was used for each experiment. Before flash-cooling, the crystals were transferred into a cryoprotective solution containing 20% (v/v) ethylene glycol. The ligand-free data sets were processed using the HKL2000 program (17), and the complex data sets were processed using MOSFLM/SCALA (18). Data collection statistics are summarized in Table I.
Structure Determination and Refinement-We carried out molecular replacement with MOLREP (19), starting with the coordinates of the CP1 domain from the full-length T. thermophilus IleRS, which we previously determined at 2.5-Å resolution (Protein Data Bank ID, 1ILE) (10). The rotated and translated model was first subjected to rigid-body refinement using 20 -2.5 Å data sets collected from the ligand-free crystal. The model refinement was carried out using the program CNS (20). After several rounds of Cartesian coordinate energy minimization, simulated annealing, B factor refinement, automatic water picking, and manual revision of the model, using the program O (21), R and R free decreased to 20.3% and 25.0%, respectively. A random sample containing 5% of the total reflections in the data sets was excluded from the refinement, to calculate R free . Using the data sets of the complex crystals, molecular replacement was carried out with the coordinates of the ligand-free CP1 domain structure. The structure refinement was performed in the same way. After building the valine ligand, further rounds of refinement were performed. The model refinement of the complex structure finally converged to an R value of 20.3% and an R free of 24.8%, with good stereochemistry. Ramachandran plot analysis using the program PROCHECK (22) showed that all of the residues in both where I j (hkl) and ͗I j (hkl)͘ are the intensity of measurement j and the mean intensity for the reflection with indices hkl, respectively. R factor ϭ ⌺(F obs Ϫ kF calc )/ ⌺ hkl F obs , where k is a scale factor and R free is the R factor for the test set of reflections not used during refinement (5% of the data set). the ligand-free and complex structures, except for one residue in each structure, were in the most favored or additionally allowed regions. Omit maps for the valine and surrounding water molecules were calculated at the end of the refinement.

Post-transfer Editing Assay of Mutant IleRSs-
The full-length T. thermophilus IleRS gene was cloned into pET28c (Novagen). This plasmid encodes IleRS with a His tag on its C-terminal end. The  mutations, T228A, T229A, T230A, T233A, T228A/T230A, T229/T230A,  T230A/T233A, and D328A, were introduced using a QuikChange mutagenesis kit (Stratagene). Plasmids were transformed into E. coli strain BL21(DE3). Cells were cultured and harvested by the same procedure used for the CP1 domain. Cells were sonicated in 30 mM Tris-HCl buffer (pH 7.5) containing 500 mM NaCl, 5 mM MgCl 2 , 10 mM imidazole, 5 mM ␤-mercaptoethanol, and 1 mM phenylmethylsulfonyl fluoride (buffer A). The insoluble cell debris and the E. coli proteins were removed by the same procedure used for the CP1 domain. The supernatant was applied to a 1-ml HiTrap chelating column (Amersham Biosciences) chelated with Ni 2ϩ ions and equilibrated with buffer A. After the column was washed with buffer A, the mutant IleRSs were eluted with 30 mM Tris-HCl buffer (pH 7.5) containing 500 mM NaCl, 5 mM MgCl 2 , 500 mM imidazole, and 5 mM ␤-mercaptoethanol and were dialyzed against 150 mM Tris-HCl buffer (pH 7.5) containing 150 mM KCl and 10 mM MgCl 2 . The final purity of the proteins was monitored by SDS-PAGE.

RESULTS AND DISCUSSION
Overall Structure-We determined the high resolution crystal structure of the isolated T. thermophilus IleRS CP1 domain, containing the post-transfer editing site, at 1.8 Å resolution (Protein Data Bank ID, 1UDZ). The root-mean-square deviation between the isolated CP1 domain and the CP1 domain in the full-length IleRS over all C ␣ atoms is 0.91 Å, which is small enough to show that these two structures are quite similar to each other. Indeed, the secondary structures and the locations of the individual residues are almost the same, and therefore, the structure of the CP1 domain in this study reflects that of the full-length IleRS. Next, we determined the cocrystal structure of the CP1 domain complexed with valine at 2.0-Å resolution (Protein Data Bank ID, 1UE0) (Fig. 1A). Its overall structure is also quite similar to that of the CP1 domain in the full-length IleRS (Fig. 1B), and thus, no large conformational change occurs upon the valine binding.
Valine-binding Mode-The editing active site is formed mainly by one ␤-strand (␤8) and two almost parallel ␣-helices (␣1 and ␣5). In the F o Ϫ F c omit map of the complex structure, there is a strong and clear electron density that could be attributed to the valine molecule in the editing active site ( Fig.  2A), and there are some electron densities that could be attributed to water molecules around the valine. The high resolution complex structure revealed the precise valine recognition mechanism (Fig. 2, B and C, and Fig. 4A). The valine NH 3 ϩ hydrogen bonds to the O␦ of Asp 328 and the ␣-CO of His 319 , and the valine COO Ϫ group hydrogen bonds to the O␥ atoms of Thr 229 and Thr 230 , and the ␣-CO groups of Trp 227 and Phe 324 , through water molecules. The valine side chain is surrounded by the side chains of Thr 233 , His 319 , Ala 321 , and Phe 324 , and the ␣-CO of Phe 324 . All of these residues are highly conserved in the IleRSs from other species (Fig. 6A). We previously proposed a valine-binding mode in the post-transfer editing state, which was modeled on the basis of the electron density corresponding to valine (although it was not clear enough to locate it unambiguously) in the full-length IleRS structure, and the 3Јterminal adenosine of tRNA Val in the T. thermophilus ValRS⅐ tRNA Val cocrystal structure (12). In the present structure, the valine is located in the same pocket as in the case of the modeled Val-tRNA Ile , but its orientation within the pocket is different from that in the model (Fig. 3).
This valine-binding mode can nicely explain the mutational data of the E. coli H319A mutant IleRS, which the previous model could not. The H319A mutant hydrolyzes not only the nearly cognate Val-tRNA Ile but also the cognate Ile-tRNA Ile (16). In the present structure, valine fits well in the editing active site (the distance between the N⑀2 of His 319 and the C␥ of valine is 3.5 Å) (Fig. 4A), but when the isoleucine model is placed in the same site, the additional methylene group clashes with the side chain of His 319 (the distance between the N⑀2 of His 319 and the C␦ of isoleucine would be 2.1 Å) (Fig. 4B). If this His 319 residue was mutated to Ala, then the distances between the Ala C␤ and the valine C␥ or the isoleucine C␦ would be 5.4 FIG. 1. Overview of the T. thermophilus IleRS CP1 domain crystal structure. A, the structure of the T. thermophilus IleRS CP1 domain complexed with valine, determined in this study. The active site area is formed mainly by one ␤-strand (␤8) and two almost parallel ␣-helices(␣1 and ␣5). B, the CP1 domain structure determined in this study, shown by a red line. The observed valine is shown as a yellow ball and stick model. The previously reported overall structure of the fulllength T. thermophilus IleRS is shown by the black line. The HVGH motif (colored in green) and the KMSKS motif (colored in blue) are shown. These two motifs compose the aminoacylation site. The overall structure of the CP1 domain in this study is almost the same as that in the full-length IleRS. or 4.7 Å, respectively. These data support the idea that the location of the valine in the determined structure represents the post-transfer editing state, while the previous model for the valine-binding mode could not explain this mutational analysis. Therefore, we concluded that the valine-binding mode in the present structure precisely reflects that of the valyl moiety of Val-tRNA Ile in the editing state.
Post-transfer Editing Assay of Mutant IleRSs-Previous biochemical experiments using E. coli IleRS showed that the T242A (corresponding to Thr 229 in T. thermophilus IleRS) mu-tant is defective in the total editing activity (whether it affects post-or pre-transfer editing or both is unknown), while the T241A, T243A, and T246A E. coli IleRS mutants (Thr 228 , Thr 230 , and Thr 233 in T. thermophilus IleRS, respectively) can still perform the total editing activity (10,25). In addition, the D342A E. coli IleRS mutant (Asp 328 in T. thermophilus IleRS) is defective in the post-transfer editing activity (26). Thus far, all of the mutant analyses have been carried out using E. coli IleRS, with most examining only the aspect of total editing and not distinguishing between pre-and post-transfer editing. For a more precise understanding of the post-transfer editing reaction, we analyzed the post-transfer editing activity of Alareplaced mutants of T. thermophilus IleRS. As a result, the T228A, T229A, T228A/T230A, T229A/T230A, and D328A mutants had some defects, while T230A, T233A, and T230A/ T233A retained the full activity (Fig. 5). These data show that Thr 228 , Thr 229 , Thr 230 (detectable only when Thr 228 is also mutated to Ala), and Asp 328 (Thr 241 , Thr 242 , Thr 243 , and Asp 342 , in E. coli IleRS, respectively) play some role in the post-transfer editing reaction.
Previous results using E. coli IleRS showed that Thr 228 and Thr 230 were not critical for the total editing (10,25). The Ala-replacing mutation of Thr 230 leads to the defect only when Thr 228 is also mutated to Ala, whereas the previous analysis examined only the single mutation of Thr 230 . The T228A defect is not as severe in the post-transfer editing, and in IleRS, the post-transfer editing was estimated to account for only 25% of the overall editing, while the remaining 75% was attributed to the pre-transfer editing (7). By assuming that the T228A mutant is only defective in the post-transfer editing, but not in the pre-transfer editing, we can account for the previous mutational analysis by not being able to detect the slight defect of the T228A mutant in the total editing. The more severe defect of the T228A/T230A double mutant than that of the T228A single mutant also shows the importance of the Thr 228 residue in the post-transfer editing (Fig. 5).
A Model for the Editing Mechanism-We tried to build a structural model for the Val-adenosine moiety of Val-tRNA Ile . First, we superposed T. thermophilus ValRS complexed with tRNA val onto the T. thermophilus IleRS CP1 domain. However, the 3Ј-terminal adenosine of tRNA val and valine could not be ester bonded, even with a slight adjustment. The distance between the OH group of the tRNA val adenosine ribose and the valine COO Ϫ group was too far (data not shown). Then, we superposed T. thermophilus LeuRS complexed with a posttransfer editing substrate analogue onto the T. thermophilus IleRS CP1 domain (Fig. 6B). This time, although there were some differences in the editing site architecture, especially in the adenine base recognition region and the conserved threonine-rich region, the distance between the OH group of the adenosine ribose and the valine COO Ϫ group was close enough for linkage with a slight adjustment. We translated the aden-osine to allow the ribose 2Ј-OH group to form an ester bond with the valine COO Ϫ group, and rotated it around the 2Ј-C-O bond. We also rotated the angle of the adenine base to adjust for the differences between the adenine base recognition regions of IleRS and LeuRS (Fig. 6A). Finally, we could build a fine model for the Val-adenosine moiety of Val-tRNA Ile , without moving any atom of the valine molecule in the determined structure (Fig. 6C). This modeled substrate is favorably hydrogen bonded by the editing active site residues and does not clash against any residue.
We compared the substrate-binding mode of IleRS with that of LeuRS (Fig. 6B). The locations of the amino acid NH 3 ϩ groups and the Asp residues (Asp 328 and Asp 347 in T. thermophilus IleRS and LeuRS, respectively) are quite similar, and the amino acid NH 3 ϩ groups are recognized by the Asp residues in  (25). Therefore, the interaction between the amino acid NH 3 ϩ group and the Asp residue must be vital for the editing reaction in both IleRSs and LeuRSs and maybe also in ValRSs. However, the orientations of the ester bond in the editing substrates are quite different between IleRS and LeuRS, and the threonine-rich region is closer to the substrate in LeuRS than in IleRS (Fig. 6B). In LeuRS, the amino acid carbonyl group is oriented toward the threonine-rich region and is recognized by Thr 247 (Thr 228 in IleRS) (Fig. 6B). In contrast, in IleRS, it is located opposite to the threonine-rich region and is recognized by the ␣-CO of His 319 through the water molecule, not by the conserved threonine residue (Fig. 6C). In IleRS, only Thr 229 hydrogen bonds to the substrate 5Ј-O, unlike the other conserved thereonine residues (Fig. 6C). These findings can account for the mutational data that Thr 229 plays some role in the post-transfer editing, but Thr 233 does not (Fig. 5). Furthermore, the water molecule, which hydrogen bonds to Asp 328 and the NH 3 ϩ group of the substrate, is favorably located to act as the catalytic nucleophile for the ester bond hydrolysis (Fig. 6C). Conversely, in the determined valine-bound structure, Thr 230 hydrogen bonds to the valine COO Ϫ group through the water molecule (Fig. 2, A  and B), but the water molecule clashes with the ribose 3Ј-OH of the modeled substrate, and cannot exist in the same place in the model (data not shown). We will discuss the importance of Thr 228 and Thr 230 below.
In the aminoacylation reaction by IleRS, valine is first mischarged to the 2Ј-OH of the tRNA 3Ј-terminal adenosine ribose (13). Under normal conditions, valine is transacylated to the 3Ј-OH from the 2Ј-OH, and vice versa, and stays in equilibrium. Until now, we have discussed the 2Ј-OH valylated tRNA Ile model. Biochemical analyses using aminoacyl-tRNAs with a deoxygenized 2Ј-OH or 3Ј-OH of the 3Ј-terminal adenosine ribose, which the amino acid cannot transacylate, suggested that IleRS specifically deacylates valine from the 3Ј-OH but not from the 2Ј-OH, whereas ValRS can deacylate the mischarged threonine from the 2Ј-OH (13).
Considering these data, we built the binding mode model of another editing substrate, the 3Ј-OH valylated adenosine moiety of Val-tRNA Ile (Fig. 7). Unlike the case of the 2Ј-OH valylated model, to make the 3Ј-OH valylated model, we moved the valyl moiety from the observed location in the determined valine-bound structure. In the 3Ј-OH valylated model, the recognition modes of the NH 3 ϩ group and the adenine base are similar to those in the 2Ј-OH valylated model (Figs. 6C and 7). The two water molecules are located in the same places as in the determined structure (light blue). Ionic bonds and hydrogen bonds are shown by dashed yellow lines. The water molecule that hydrogen bonds to Asp 328 is favorably located to act as a nucleophile in the ester bond hydrolysis reaction. In addition, the adenosine ribose 3Ј-OH group is favorably located to act as a nucleophile in the transacylation reaction.
In the 3Ј-OH valylated model, Thr 229 hydrogen bonds directly to the adenosine 5Ј-O, as in the 2Ј-OH valylated model, and Thr 228 and Thr 230 hydrogen bond to the adenosine 3Ј-O through the water molecule (Fig. 7), as observed in the determined valine-bound structure (Fig. 2, A and B). This water molecule, which hydrogen bonds to Thr 228 and Thr 230 , is favorably located to act as a catalytic nucleophile for the ester bond hydrolysis (Fig. 7). In this model, Thr 228 and Thr 230 participate in the substrate recognition, in contrast to the 2Ј-OH valylated model, thus confirming the mutational analysis that revealed the importance of these residues in the editing reaction. Although Thr 230 is highly conserved in IleRSs, it is replaced with the highly conserved Arg residue in LeuRSs and ValRSs (Fig.  6A). This is consistent with the data that the mischarged amino acid is deacylated from the 3Ј-OH only in IleRS (13) but is deacylated from the 2Ј-OH in ValRS (13) and maybe also in LeuRS (14). Moreover, these two binding-mode models are not mutually exclusive. Instead, both binding mechanisms might be used. That is, first the 2Ј-OH valylated tRNA Ile is bound to the editing site, and after the valyl moiety is transferred to the 3Ј-OH from the 2Ј-OH, it is deacylated from the 3Ј-OH. In the 2Ј-OH valylated model, the 3Ј-OH is favorably located to act as a catalytic nucleophile for the transacylation reaction (Fig. 6C).
Concluding Remarks-The high resolution crystal structure of the T. thermophilus IleRS CP1 domain complexed with the nearly cognate valine provides a precise understanding of the selective valine recognition mechanism. In addition, we were able to build the binding mode models of the two substrates, the 2Ј-OH valylated and the 3Ј-OH valylated adenosine moieties of Val-tRNA Ile . These structure models illustrate the mutational data well and suggest a specific editing mechanism for IleRS, which is distinct from those of LeuRS and ValRS. Actually, only in IleRS, the pre-transfer editing is predominant (75% of the total editing) (7). However, we still need to determine whether the active site accommodates 2Ј-OH or 3Ј-OH valylated tRNA Ile or both and to elucidate the precise binding mode. Moreover, the binding mode of the pre-transfer editing substrate is still unknown in IleRS, although in LeuRS, both substrates are recognized in the same binding site in similar manners (14). This structural study has provided significant clues toward elucidating the editing mechanism of IleRS. For a more precise understanding of both pre-and post-transfer editing, the de-termination of the IleRS structures complexed with nonhydrolyzable analogs of pre-and post-transfer editing substrates is required. FIG. 7. Structural model of the 3-OH valylated adenosine moiety of Val-tRNA Ile (stereo view). The two water molecules are located in the same places as in the determined structure (light blue). Ionic bonds and hydrogen bonds are shown by dashed blue lines. The water molecule hydrogen bonding to Thr 228 and Thr 230 is favorably located to act as a nucleophile in the ester bond hydrolysis reaction.