|Home | About | Journals | Submit | Contact Us | Français|
Threonyl-tRNA synthetase (ThrRS) plays an essential role in protein synthesis by catalyzing the aminoacylation of tRNAThr and editing misacylation. ThrRS generally contains an N-terminal editing domain, a catalytic domain and an anticodon-binding domain. The sequences of the editing domain in ThrRSs from archaea differ from those in bacteria and eukaryotes. Furthermore, several creanarchaea including Aeropyrum pernix K1 and Sulfolobus tokodaii strain 7 contain two genes encoding either the catalytic or the editing domain of ThrRS. To reveal the structural basis for this evolutionary divergence, the two types of ThrRS from the crenarchaea A. pernix and S. tokodaii have been overexpressed in Eschericha coli, purified and crystallized by the hanging-drop vapour-diffusion method. Diffraction data were collected and the structure of a selenomethionine-labelled A. pernix type-1 ThrRS crystal has been solved using the MAD method.
Aminoacyl-tRNA synthetases (aaRSs) play a crucial role in the biosynthesis of proteins by charging a cognate amino acid onto its cognate tRNA. This reaction takes place in two steps: (i) synthesis of an aminoacyladenylate intermediate by fusing ATP to the cognate amino acid and (ii) transfer of the amino-acid moiety to the 3′-terminus of the cognate tRNA to generate the aminoacyl-tRNA. In principle, aaRSs are not allowed to make any mistakes when catalyzing this two-step reaction. However, some aaRSs misactivate and misaminoacylate noncognate amino acids that are isosteric and chemically related to the cognate amino acids (Fersht & Kaethner, 1976 ; Igloi et al., 1977 ; Schmidt & Schimmel, 1994 ). To maintain high fidelity of protein synthesis, these aaRSs usually translocate and hydrolyze the incorrect products at their editing domains1.
A typical example is threonyl-tRNA synthetase (ThrRS), which generally consists of three domains: an N-terminal editing domain, a catalytic domain and a C-terminal anticodon-binding domain (Sankaranarayanan et al., 1999 ; Torres-Larios et al., 2003 ). This enzyme first synthesizes the threonyl-AMP intermediate and then transfers threonine onto tRNAThr. ThrRS must discriminate threonine from valine and serine. A conserved zinc ion in the catalytic domain precludes the misactivation of noncognate valine, resulting in the discrimination of valine (Sankaranarayanan et al., 1999 ). In Escherichia coli, the N-terminal editing domain of ThrRS removes mischarged serine from Ser-tRNAThr (Dock-Bregeon et al., 2004 ). However, the N-terminal domains of archaeal ThrRSs have no sequence homology to those of eukaryotic and bacterial ThrRSs. Nevertheless, it has been reported that the archaeal N-terminal domain is likewise able to remove serine mischarged onto tRNAThr (Beebe et al., 2004 ). The structure of the N-terminal domain of Pyrococcus abyssi ThrRS (Hussain et al., 2006 ) reveals similarities to that of d-aminoacyl-tRNA deacylase from E. coli, which specifically removes a d-amino acid mischarged on the tRNA.
Interestingly, several crenarchaeal species have been found to possess two different genes encoding ThrRSs (Woese et al., 2000 ). Sequence alignments of these ThrRSs have shown that they diverge into two types, one resembling bacterial-type ThrRSs and the other containing archaeal-type sequences (Korencic et al., 2004 ). In Sulfolobus solfataricus, the bacterial-type ThrRS possesses highly conserved catalytic and anticodon-binding domains, but has a large deletion in the N-terminal editing domain. This type of ThrRS, known as ThrRS-cat, can synthesize both Thr-tRNAThr and Ser-tRNAThr but does not discriminate against Ser-tRNAThr. In contrast, the archaeal-type ThrRS shows high conservation in the N-terminal editing domain but lacks the catalytic domain. This type of ThrRS, called ThrRS-ed, can hydrolyze misaminoacylated Ser-tRNAThr but does not have aminoacylation activity.
In the present study, further sequence analyses have been performed using the KEGG database (http://www.genome.jp/kegg/). Two other crenarchaeal organisms, Aeropyrum pernix (Ap) and S. tokodaii (St), each with two ThrRS genes of different origin (see §3.1 and Fig. 1 ) were identified. The sequences of the encoded ThrRSs are similar to those ThrRS-cat (ApThrRS-1 and StThrRS-1) and ThrRS-ed (ApThrRS-2 and StThrRS-2). In order to elucidate the structural basis of the editing and catalytic mechanisms in the two crenarchaea A. pernix and S. tokodaii, the two ThrRSs from both organisms were overexpressed in E. coli and purified. Crystals of the native and selenomethionine-labelled ThrRSs were obtained and X-ray diffraction data were collected. The structure of the selenomethionine-labelled ApThrRS-1 crystal has subsequently been solved using the Se-MAD method.
The sequences of ThrRS-1 and ThrRS-2 from A. pernix (APE0809.1 and APE0117.1), from S. tokodaii (ST0966 and ST2187) and from other organisms in the Sulfolobaceae family (S. solfataricus, S. acidocaldarius and Metallosphaera sedula) were downloaded from the KEGG database. To analyze the structural modules, these sequences were compared with those of E. coli (Ec) and P. abyssi (Pa) ThrRSs using the program ClustalX (Thompson et al., 1997 ).
The four genes encoding ApThrRS-1 (APE0809.1, 471 residues, 53 122 Da, pI 5.89), ApThrRS-2 (APE0117.1, 421 residues, 45 003 Da, pI 6.74), StThrRS-1 (ST0966, 540 residues, 63 110 Da, pI 6.48) and StThrRS-2 (ST2187, 267 residues, 30 855 Da, pI 6.93) were separately inserted into the pET11a vector (Novagen). The recombinant plasmids were introduced in E. coli Rosetta-gami (DE3) (Novagen) and the cells were grown in LB culture at 310 K. The selenomethionine derivatives were expressed in E. coli B834 (DE3) and BL21 (DE3) (Novagen) using minimal M9 medium containing l-selenomethionine. Following overnight incubation, the cells were harvested by centrifugation at 6000 rev min−1 for 10 min at 277 K and disrupted by sonication. The cell lysates were incubated at 343 K for 30 min to denature the E. coli proteins and centrifuged at 18 000 rev min−1 for 20 min at 277 K. The proteins were purified using the columns and buffer solutions listed in Table 1 . After purification, the proteins were concentrated to final concentrations of 3–15 mg ml−1 using centrifugal filter devices (Microcon and Centricon Ultracel YM-30 or YM-10 membrane from Millipore Co. or Vivaspin 500 from Sartorius Stedim Biotech; Table 1 ). The fractions obtained in all the purification steps as well as the concentrated proteins were analyzed by SDS–PAGE.
All crystallization trials were performed using the hanging-drop vapour-diffusion method by mixing equal volumes (1 µl) of the protein and reservoir solutions and equilibrating the mixed solutions against 700 µl reservoir solution in a 24-well plate at 293 K. Initial screening for potential crystallization conditions was performed using several kits purchased from Hampton Research Corporation (California, USA). The conditions under which crystalline precipitates appeared were optimized by changing the concentrations of the protein, precipitant and salt and by changing the pH of the buffer solution. The pH values and UV-absorption spectra were measured using an F-13 pH meter (Horiba Ltd, Japan) and a BioSpec-mini spectrophotometer (Shimadzu Corporation, Japan), respectively. For crystallizations, 24-well plates (Stem Corporation, Japan), plain glass cover slides (18 mm diameter Matsunami Glass Ind. Ltd, Japan) coated with silicon (L-25, Fuji Systems Corporation, Japan) and high-vacuum grease (Dow Corning Toray Co. Ltd, Japan) were used.
The crystals obtained were soaked in their respective reservoir solutions containing 30% glycerol for 30 s and mounted on a CryoLoop (Hampton Research). X-ray diffraction measurements were performed at 95 K with 1° oscillation per image. Diffraction data were collected on the BL5A, BL6A, BL17A and NW12 beamlines of the Photon Factory (PF; Ibaraki, Japan) using a CCD detector (Area Detector Systems Co., ADSC, San Diego, California, USA). Diffraction images were indexed, integrated and scaled using the HKL-2000 package (Otwinowski & Minor, 1997 ) and intensity data were converted to amplitudes using programs from the CCP4 suite (Collaborative Computational Project, Number 4, 1994 ).
The molecular-replacement and multiple anomalous dispersion (MAD) methods were applied to solve the phase problem. For molecular replacement, the programs AMoRe (Navaza, 1994 ), MOLREP (Vagin & Teplyakov, 2000 ) or Phaser (McCoy et al., 2007 ) were employed and the ThrRS structures from E. coli (EcThrRS; PDB code 1qf6; Sankaranarayanan et al., 1999 , 2000 ) and Staphylococcus aureus (SuThrRS; PDB code 1nyq; Torres-Larios et al., 2003 ), truncated according to the sequence alignment, were used as search models. For MAD phasing, the programs SHARP/autoSHARP (Vonrhein et al., 2007 ), SHELXD (Sheldrick, 2008 ), SOLOMON (Abrahams & Leslie, 1996 ) and ARP/wARP (Perrakis et al., 1999 ) were used to calculate the phases, to locate heavy atoms, to modify the electron densities and to build the structures, respectively.
Fig. 1 shows a schematic domain arrangement summarized from sequence comparisons between EcThrRS, PaThrRS, ApThrRS-1, ApThrRS-2, StThrRS-1, StThrRS-2, SfThrRS-1 (containing SsThrRS-1, SaThrRS-1 and MsThrRS-1) and SfThrRS-2 (containing SsThrRS-2, SaThrRS-2 and MsThrRS-2). More detailed diagrams are provided as supplementary material2. As pointed out by Korencic et al. (2004 ), ApThrRS-1 has a shorter sequence that only contains the residues of the catalytic and anticodon-binding domains. When compared with EcThrRS and PaThrRS, it is noteworthy that StThrRS-1 and SfThrRS-1 have a short extension at the N-terminus of the catalytic domain. In contrast, crenarchaeal ThrRS-2s completely lack the catalytic domain. ApThrRS-2 and SfThrRS-2 have an editing domain similar to that of PaThrRS. In the case of StThrRS-2, the corresponding editing domain displays high sequence conservation with PaThrRS, ApThrRS-2 and SfThrRS-2, but is much shorter at the N-terminus. In addition, crenarchaeal ThrRSs contain insertions between the editing and anticodon-binding domains. X-ray analysis revealed that the PaThrRS editing domain forms a dimer (Hussain et al., 2006 ). The interface of the dimer occurs between the shorter sequences commonly conserved in the editing domain of ThrRS-2s. Therefore, the editing domains of ThrRS-2s might form a dimer similar to that of PaThrRS.
As shown in Fig. 2 , every protein was highly purified prior to crystallization. Crystals of the native proteins and their SeMet derivatives (Fig. 3 ) were obtained under the conditions given in Table 2 . In all cases, they do not contain ATP, amino acids or tRNA. The conditions for crystallization of the SeMet derivatives are similar to the corresponding conditions for native proteins, with the exception of SeMet StThrRS-1, for which the reservoir solution added to the protein solution contained 80 mM Na2HPO4. Therefore, the difference in crystal habits between the StThrRS-1 native and SeMet derivative crystals (described in §3.3) may be a consequence of the difference in their crystallization conditions. The conditions under which the ApThrRS-1 and StThrRS-1 crystals were grown are similar, but differ from the reported conditions for EcThrRS (Sankaranarayanan et al., 1999 ) and SuThrRS (Torres-Larios et al., 2002 , 2003 ). Naturally, ThrRSs from different sources may crystallize under different conditions. Moreover, the conditions may vary between the present and previously reported ThrRS crystals because the latter were grown in complex with tRNA, ATP or a threonine analogue.
Fig. 4 shows examples of diffraction patterns obtained from the native crystals. The data-collection statistics for the ApThrRS-1, StThrRS-1 and StThrRS-2 crystals are summarized in Tables 3 , 4 and 5 , respectively. Although diffraction data were collected to 3.7 Å resolution from the ApThrRS-2 crystal, data-processing attempts were unsuccessful. The crystal data suggest that the native and selenomethionine-labelled pairs are isomorphous except for the StThrRS-1 pair. The native and selenomethionine forms of StThrRS-1 belong to space groups P21 and P1, respectively, and their unit-cell parameters differ. The calculated Matthews coefficients (Matthews, 1968 ) and the estimated solvent contents, which are also listed in Tables 3 , 4 and 5 , suggest that there is one subunit of ApThrRS-1 and two subunits of StThrRS-1 or StThrRS-2 in their respective asymmetric units. ThrRSs belong to the class II aaRSs (Eriani et al., 1990 ; Cusack et al., 1990 ), which generally form a homodimer through contacts between the catalytic domain of one subunit and the anticodon-binding domain of another subunit. Therefore, in the case of the ApThrRS-1 crystal the two subunits must be related by crystallographic twofold symmetry in the C2221 space group, even though the editing domain is missing.
Although several structures derived from ThrRS data published in the PDB were used as search models in molecular-replacement trials, it was difficult to obtain significant solutions. However, we have successfully completed phase determination for the selenomethionine derivative of ApThrRS-1 using the MAD method. 11 Se atoms with occupancies greater than 50% were found. After refinement of the atomic parameters, the electron-density map modified by the solvent-flattening technique shows a definite protein backbone, as shown in Fig. 5 (a). The final correlation coefficient in |E|2 was 0.771. After structure model building, more than 96% of protein residues were assigned. The statistics of structure modeling at the current stage are listed in Table 6 . In parallel to refining the structure of ApThrRS-1, we are currently attempting to obtain crystals of higher quality for the selenomethionine-labelled ApThrRS-2, StThrRS-1 and StThrRS-2 proteins.
As seen in Fig. 5 (b), ApThrRS-1 forms a homodimer of subunits which are related to each other by crystallographic twofold rotation symmetry along the a axis. The amino-acid residues at the interface interact with each other through hydrogen bonds and van der Waals contacts between the two subunits. This structural feature is consistent with the dimer formation of the class II ThrRSs (Eriani et al., 1990 ; Cusack et al., 1990 ). The dimers are reasonably packed according to the space-group symmetry in the unit cell, with no molecular collisions, as seen in Fig. 5 (c). More detailed structural features will be described elsewhere after full refinement.
The sequence comparison in Fig. 1 shows that the editing domain is completely missing in ApThrRS-1, while the other ThrRS-1s have additional small domains consisting of up to 145 amino-acid residues. In the ThrRS-2s, the editing domain of ApThrRS-2 is similar to that of PaThrRS. The domains of PaThrRS form a dimer similar to that of d-aminoacyl-tRNA deacylase from E. coli (Hussain et al., 2006 ). Therefore, ApThrRS-2 might form a dimer between the editing domains, as the catalytic domain for dimerization is missing. This is consistent with the results of our gel-filtration experiment, which showed that the molecular size of ApThrRS-2 is equivalent to twice that of the subunit (data not shown). Here, it is interesting to note that the corresponding editing domain is extremely small in StThrRS-2 but this part is just a dimerization interface to form a β-sheet between the two subunits. Furthermore, ThrRS-2 contains an additional domain just after the editing domain. This domain might be inserted to stabilize dimer formation, although the sequence of ApThrRS-2 is slightly different from the others. Both ThrRS-2 crystals contain two subunits in their asymmetric units, suggesting the possibility of dimer formation. These structural questions will be resolved by X-ray analyses of the four proteins ApThrRS-1, ApThrRS-2, StThrRS-1 and StThrRS-2.
This work was supported in part by Grants-in-Aid for the Protein3000 Research Program from the Ministry of Education, Culture, Sports, Science and Technology of Japan. We thank S. Kuramitsu for organizing the research group in the program and N. Igarashi and S. Wakatsuki for facilities and help during data collection.
1Two exceptions, known as nondiscriminating aaRSs, are aspartyl-tRNA synthetase (AspRS) and glutamyl-tRNA sunthetase (GluRS). The nondiscriminating AspRS (Sato et al., 2007 ) produces Asp-tRNAAsp and Asp-tRNAAsn to compensate for the missing AsnRS. Similarly, GluRS produces Glu-tRNAGlu and Glu-tRNAGln (Curnow et al., 1997 ). The mischarged amino acids are amidated by other enzymes: GatCAB for Asp-tRNAAsn and GatED for Glu-tRNAGln.
2Supplementary material has been deposited in the IUCr electronic archive (Reference: BO5044).