Báo cáo khóa học: The zinc-binding site of a class I aminoacyl-tRNA synthetase is a SWIM domain that modulates amino acid binding via the tRNA acceptor arm

doi:10.1111/j.1432-1033.2003.03976.x

Eur. J. Biochem. 271, 724–733 (2004) (cid:1) FEBS 2004

The zinc-binding site of a class I aminoacyl-tRNA synthetase is a SWIM domain that modulates amino acid binding via the tRNA acceptor arm

Rajat Banerjee1,*, Daniel Y. Dubois2,*, Joelle Gauthier2, Sheng-Xiang Lin3, Siddhartha Roy1 and Jacques Lapointe2 1Department of Biophysics, Bose Institute, Calcutta, West Bengal, India; 2De´partement de Biochimie et de Microbiologie, Universite´ Laval, Que´bec, Canada; and 3Laboratoire d’Oncologie et Endocrinologie Mole´culaire, Centre de Recherche du CHUL (CHUQ) et Universite´ Laval, Que´bec, Canada

residue

stabilizes

close structural and mechanistic similarities between GluRS and GlnRS, these results indicate that the GluRS SWIM domain modulates glutamate binding to the active site via its interaction with the tRNAGlu acceptor arm. Phylogenetic analyses indicate that ancestral GluRSs had a strong zinc- binding site in their SWIM domain. Considering that all GluRSs require a cognate tRNA to activate glutamate, and that some of them have different or no putative zinc-binding residues in the corresponding positions, the properties of the C100Y variant suggest that the GluRS SWIM domains evolved to position correctly the tRNA acceptor end in the active site, thereby contributing to the formation of the glutamate binding site.

Keywords: glutamyl-tRNA synthetase; zinc-binding site; SWIM domain; glutamate-binding site; tRNA acceptor arm.

In its tRNA acceptor end binding domain, the glutamyl- tRNA synthetase (GluRS) of Escherichia coli contains one atom of zinc that holds the extremities of a segment (Cys98-x-Cys100-x24-Cys125-x-His127) homologous to the Escherichia coli glutaminyl-tRNA synthetase (GlnRS) loop where a leucine the peeled-back conformation of tRNAGln acceptor end. We report here that the GluRS zinc-binding region belongs to the novel SWIM domain family characterized by the signature C-x-C-xn-C-x-H (n ¼ 6–25), and predicted to interact with DNA or proteins. In the presence of tRNAGlu, the GluRS C100Y variant has a lower affinity for L-glutamate than the wild-type enzyme, with Km and Kd values increased 12- and 20-fold, respectively. On the other hand, in the absence of tRNAGlu, glutamate binds with the same affinity to the C100Y variant and to wild-type GluRS. In the context of the

Studies with truncated tRNAs and aminoacyl-tRNA synthetases (aaRS, EC 6.1.1) indicate that early in the evolution of life, short RNA molecules having a hairpin- like structure and homologous to the acceptor-TwC stem- loop domain of contemporary tRNAs were specifically recognized and aminoacylated by small proteins homo- logous to the class-defining catalytic domain of contem- porary aaRSs [1–3].

In several class I aaRSs, this catalytic domain comprises essentially a dinucleotide-binding fold [4]. Another domain is inserted in the connection between the two halves of this doubly wound a/b parallel domain (or Rossmann fold); in the case of Escherichia coli glutaminyl-tRNA synthetase (GlnRS, EC 1.1.1.18) in its complex with tRNAGln and ATP, this inserted domain interacts with the acceptor stem of tRNA, and was thus named (cid:1)acceptor-binding domain(cid:2), or ABD [5,6]. The role of ABD in tRNAGln recognition and in the exclusion of noncognate tRNAs was revealed by the mischarging properties of GlnRS variants altered in either one of two motifs known, from the 3D structure, to interact with the acceptor end of tRNAGln: the extended helix E region (helix/loop E) and loop 1 [7]. In a unified nomen- clature proposed for the GlxRS family which includes the GlnRSs and the evolutionarily related glutamyl-tRNA synthetases (GluRS, EC 6.1.1.17) [8], this helix/loop E is called helix/loop D (Fig. 1). The corresponding domain in E. coli GluRS ([9], for a review see [10]) contains a zinc atom located far from the active site, in a structure different from those of the other known zinc fingers [11] (Fig. 1A), and whose unusual spacings between the zinc-binding residues (Cys98-x-Cys100-x24-Cys125-x-H127) show that it belongs to the recently identified family of SWIM domains, whose signature is C-x-C-xn-C-x-H (where n ¼ 6–25) [12]. These domains were named after the bacterial ATPases of the

Correspondence to J. Lapointe, De´ partement de Biochimie et de Microbiologie, Centre de Recherche sur la Fonction, la Structure et l’Inge´ nierie des Prote´ ines (CREFSIP), Universite´ Laval, Que´ bec, Canada G1K7P4. Fax: + 1 418 6563664, Tel.: + 1 418 6562131 ext 3411, E-mail: Jacques.Lapointe@bcm.ulaval.ca Abbreviations: aaRS, aminoacyl-tRNA synthetases; ABD, acceptor- binding domain; GlnRS, glutaminyl-tRNA synthetase; GluRS, glutamyl-tRNA synthetase; NJ, neighbor-joining; OP, 1,10-phenanthroline; w, pseudo-uridine; TCA, trichloroacetic acid. Enzymes: glutamyl-tRNA synthetase (EC 6.1.1.17); glutaminyl-tRNA synthetase (EC 6.1.1.18). *R.B. and D.Y.D. contributed equally to this work. (Received 22 October 2003, revised 16 December 2003, accepted 22 December 2003)

SWIM domain of glutamyl-tRNA synthetases (Eur. J. Biochem. 271) 725

(cid:1) FEBS 2004

their C-terminal were constructed (Dubois, D.Y., unpub- lished results). These enzymes were purified to homogeneity by affinity chromatography on Ni-nitrilotriacetic acid superflow resin (Quiagen) and used to estimate their zinc content and to confirm the values of the kinetic parameters obtained with the variant purified without His-tag.

Aminoacylation of tRNAGlu and kinetic measurements

The kinetic constants were determined at 37 (cid:2)C and pH 7.2 by measuring the rate of formation of L[14C]glutamyl-tRNA as previously described [14], at saturating concentrations of each fixed substrate and variable concentrations of the substrate tested. E. coli tRNAGlu was purified from the overproducing strain DH5a(pKR15) [15]. Data were ana- lyzed with the Hanes plot [16], using SPSS Science’s SIGMA PLOT 8.0(cid:3) with the ENZYME KINETICS MODULE version 1.10.

Measurement of GluRS activity after gel electrophoresis

Following electrophoresis of wild-type and C100Y GluRS in a 7.5% polyacrylamide gel under nondenaturing condi- tions [17], a Whatman 3MM paper soaked with GluRS aminoacylation mixture [14] containing L[14C]glutamate at 262 mCiÆmmol)1 was placed on the gel between two glass plates; this sandwich was wrapped in a plastic film (Saran) and incubated at 37 (cid:2)C for 30 min. The paper was washed in 5% trichloroacetic acid (TCA) at 0 (cid:2)C as described previously [14], dried and autoradiographed with a Fuji type BAS-II Imaging plate.

Fluorescence methods

SWI2/SNF2 family and the plant MuDR transposases, where they were first found.

We report here that the GluRS C100Y variant, altered in one of the zinc-binding residues, has a substantially reduced affinity for glutamate in the aminoacylation reaction, in spite of having conserved its zinc atom and its affinity for tRNAGlu. Moreover, phylogenetic analyses show that this site diverged rapidly during GluRS evolution. We propose that the plasticity of the GluRS SWIM domain was used during the evolution of this enzyme to position correctly the acceptor end of its cognate tRNA in the active site, thereby contributing to the formation of the glutamate binding site.

Fig. 1. Model of the zinc-binding region or SWIM domain of E. coli GluRS (A), and 3D-structure of a part of E. coli GlnRS ABD domain complexed to tRNAGln and ATP (B). The GluRS Tyr96-Glu136 seg- ment includes the four zinc-ligands (adapted from [11]); the GlnRS segment corresponding to the GluRS SWIM domain is drawn in yellow (adapted from [6]).

Experimental procedures

All the fluorescence spectra were measured at 25 (cid:2)C in a Hitachi F3010 spectrofluorometer in 10 mM Na Hepes pH 7.2 and 16 mM MgCl2. The excitation wavelength was 295 nm and the emission was recorded at 340 nm. The excitation and emission bandpasses were 5 nm. Acrylamide quenching of tryptophan fluorescence was carried out by addition of a freshly prepared solution of three times recrystallized acrylamide. The protein concentration was 1 lM. The absorbance of acrylamide was measured and inner filter effects due to the addition of acrylamide were corrected. Binding of tRNAGlu to wild-type and C100Y GluRS was determined by fluorescence quenching at 25 (cid:2)C. The protein concentration was 0.2 lM. The inner filter effect was corrected as stated above. The data were fitted using KYPLOT to a single site binding equation.

Overproduction and purification of the C100Y GluRS

Circular dichroism

Circular dichroism was measured at 25 (cid:2)C in a JASCO J-600 spectropolarimeter in a 0.1 cm pathlength cuvette. The protein concentration was 1 lM in 10 mM Na Hepes pH 7.2 and 16 mM MgCl2.

Measurement of Kd for L-glutamate by nuclear magnetic resonance

the a-protons of L-glutamate The NMR spectra of (150 lM) were measured at 27 (cid:2)C in a 500 MHz Bruker

The plasmid pTZ7612 [11] carrying E. coli gltX was used to obtain the C100Y GluRS variant, using the method of Kunkel [13]. The 2.2 kb EcoRI fragments containing the mutated gltX was transferred to pBR322 (Ampr, Tetr) yielding plasmid pJG7612. E. coli DH5a transformed with pJG7612 was grown in LB medium supplemented with 100 lgÆmL)1 ampicillin, at 18 (cid:2)C to increase the yield of overproduced soluble protein. The C100Y variant was purified to near homogeneity (> 95% pure) by a procedure previously described for wild-type GluRS [14]. Alternatively, vectors allowing the overproduction of wild-type GluRS and its C100Y variant carrying removable (His)6-tags at

726 R. Banerjee et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

DRX-500 NMR spectrometer in a 8 mm proton probe. The experiments were conducted in 20 mM potassium phosphate buffer pH 7.2, containing 1 mM EDTA and 16 mM MgCl2. WATERGATE water suppression pulse (using Z-pulsed field gradient) was used to suppress water signal. Line broadening of 0.5 Hz was used for processing. For each spectrum, 512 scans were averaged. The T2s were deter- mined from fit of the NMR data to Lorentzian line-shape using KYPLOT. The T2 of free glutamate and that in the tRNA–aaRS complex were used with presence of the equation developed by Redfield [18], to calculate the dissociation constants. The exact T (cid:2)1 2 was measured by fitting the a-proton resonance (triplet) to a sum of three Lorentzians. T (cid:2)1 2P was obtained by subtracting the contri- bution of the free ligand. Equations 8b and 9 of Redfield [18] have been rearranged to yield the following equation:

T2P ¼ C (cid:3)Kd þ C(cid:3)½ligand(cid:6)

where, C is (sM + T1M)/Et. The symbols are defined by Redfield [18]. The ratio intercept/slope of the plot of T2P vs. ligand concentration gives the dissociation constant.

Sequence alignments and phylogenetic analyses

Sixty-two bacterial ORFs homologous to E. coli GluRS were found in publicly available databases using the programs BLASTP and TBLASTN [19] through the ENTREZ portal at the NIH (http://www3.ncbi.nlm.nih.gov/Entrez/ index.html). Preliminary sequence data for Acidithiobacillus ferrooxidans and Wolbachia sp. (endosymbiont of Droso- phila melanogaster) were obtained from The Institute for Genomic Research website at http://www.tigr.org. The corresponding amino acid sequences were aligned using the program CLUSTAL X [20] and edited manually in BIOEDIT [21]. The neighbor-joining (NJ) phylogeny was based on pairwise distances between amino acid sequences using the programs NEIGHBOR and PROTDIST of the PHYLIP 3.6a2.1 package [22] with a replacement model based on the Dayhoff 120 matrix [23]. The programs SEQBOOT and CONSENSE were used to estimate the confidence limits of branching points from 1000 bootstrap replications, and the program FITCH was used to recalibrate branch lengths. Maximum-likelihood (ML) analysis was carried out using the program TREE-PUZZLE 5.0 [24]. Parameters used were 5000 puzzling quartets, the JTT matrix [25], gamma- distributed rates over eight categories, and the a parameter estimated from the data. Trees from all analyses were visualized and subsequently converted into figures using the program TREEVIEW [26].

carrying an endogenous thermosensitive GluRS [11], we replaced Cys100 with a tyrosine residue, the most conser- vative replacement for Cys [values of 1.0 vs. 0.7 for Ser (over 1.5 for identity in the Dayhoff table [23])]. The C100Y GluRS variant, altered in the second zinc-ligand of E. coli GluRS, was active and was characterized. Surprisingly, it migrates exclusively as the zinc-depleted form of wild-type GluRS (Fig. 2A), but still catalyzes tRNAGlu aminoacyla- tion. The zinc content of the purified C100Y GluRS is fivefold lower than that of wild-type GluRS [28], as determined by atomic absorption spectroscopy (results not shown). Zinc-containing and zinc-free C100Y GluRS have the same electrophoretic mobility (Fig. 2A, lane 1), sug- gesting that zinc does not give to the C100Y GluRS variant the structure that allows the faster migration of the zinc- containing wild-type GluRS [28]. This lower mobility may be due to a decreased interaction of the SWIM domain with the other domains, making the radius of gyration of C100Y GluRS larger than that of wild-type GluRS.

Fig. 2. Electrophoretic mobility (A and B) and aminoacylation activity (C) of wild-type (WT) and C100Y GluRS in nondenaturing polyacryl- amide gel. The migration origin is at the top. (B and C) Detection of protein (with Coomassie blue) and GluRS activity, respectively: lane 1 (control), 20 lg bovine serum albumin; lane 2, 2.0 lg wild-type E. coli GluRS; lane 3, 1.8 lg GluRS C100Y. The two arrows indicate the positions of wild-type GluRS with zinc (fastest) and without zinc (slowest) (see [28]). The GluRS activity shown in (C) is an autoradio- gram of [14C]Glu-tRNAGlu transferred on Whatman paper after the aminoacylation reaction conducted in the gel (B), as described under Experimental procedures.

Results and discussion

The active zinc-containing C100Y GluRS migrates as the zinc-depleted wild-type enzyme

By carrying out the aminoacylation reaction directly on the polyacrylamide gel after electrophoresis under non- denaturing conditions (Fig. 2B,C), we confirmed that the activity of this variant is associated with the bulk of slowly migrating GluRS (Fig. 2C, lane 3) and not with a small proportion of molecules not visible on the gel that would have migrated as zinc-containing wild-type GluRS. More- over, the rapid reduction of the enzymatic activity of the C100Y variant after preincubation with the strong zinc- chelator OP (Fig. 3A) shows that zinc is essential for the aminoacylation activity of the C100Y GluRS and that it is

Wild-type GluRS depleted of its zinc atom by the strong zinc-chelator 1,10-phenanthroline (OP) [27] is inactive in the aminoacylation reaction and migrates more slowly than zinc-containing GluRS during electrophoresis under non- denaturing conditions [28]. As the conservative variants C98S, C100S, C125S and H127Q did not complement at the restrictive temperature (42 (cid:2)C) the E. coli strain JP1449

SWIM domain of glutamyl-tRNA synthetases (Eur. J. Biochem. 271) 727

(cid:1) FEBS 2004

Fig. 4. Structural comparison of wild-type E. coli GluRS (s) and of its C100Y variant (d). (A) Circular dichroism spectra. (B) Stern–Volmer plot of acylamide quenching of tryptophan fluorescence.

for the C100Y variant, which is about 12-fold higher than that of wild-type GluRS (0.105 mM, Fig. 3B). The kcat values for wild-type and C100Y GluRS obtained from this Hanes’ plot are, respectively, 6.8 s)1 and 0.41 s)1; considering that only 20% of the C100Y GluRS mole- cules are active (see above), the lower limit for kcat of this variant is about 2 s)1, which indicates that its active site is minimally perturbed.

Minor changes in the GluRS structure due to the C100Y substitution

more weakly bound to the C100Y variant than to wild- type GluRS. Attempts at inserting a zinc atom in C100Y GluRS that contains no zinc (80% of the molecules, see above) did not increase their activity (results not shown), suggesting that while the C100Y substitution does not prevent zinc binding, it reduces its incorporation into GluRS during the folding of this polypeptide. Reduced affinity for zinc was also reported for variants of E. coli isoleucyl-tRNA synthetase (IleRS) whose Cys4 zinc-binding quadruplet was replaced by CSCC, CCSC or CCCS; mutants carrying these IleRS variants do not grow under normal conditions, but do grow when 25 mM ZnCl2 is added to the growth medium, probably because this high concentration overcomes the weaker binding of zinc to these altered sites [29].

To determine whether the C100Y substitution caused a localized or a widespread structural change, we compared the circular dichroism (CD) spectra and the Stern–Volmer plot of acrylamide quenching of the tryptophan fluorescence of the wild-type and C100Y GluRS at identical protein concentrations. The mean residue molar ellipticity of the C100Y protein is slightly reduced with respect to the wild- type protein (Fig. 4A), compatible with a disorder of the zinc-containing region, and indicating the absence of major global unfolding of the structure. The Stern–Volmer plots are )1 for the very similar with average Ksv values of 5.3 and 6.3 M wild-type and the variant protein (Fig. 4B), suggesting little change in dynamic accessibility. The emission maximum remains around 336 nm in both the wild-type and the variant, suggesting little change in tryptophan exposure.

Kinetic parameters of the GluRS C100Y variant in the aminoacylation reaction

twice the K Glu

m

m

m

As we expected that this substitution in the ABD would influence primarily the interaction of GluRS with its tRNA substrate (see the introduction), we first measured its Km for tRNAGlu in the presence of 0.2 mM L-glutamate (about for wild-type GluRS), and obtained for this variant a K tRNA of 0.23 lM. The wild-type enzyme has a similar K tRNA (0.32 lM), but the (apparent) kcat value of C100Y GluRS was 100-fold lower than that of wild-type GluRS. Considering that tRNAGlu influences strongly the interaction between GluRS and glutamate (reviewed in [30]), the unexpected finding that the C100Y substitution in the ABD decreased substan- tially the kcat of the glutamylation reaction without altering the Km for tRNAGlu suggested that the low apparent kcat value obtained in the presence of 0.2 mM L-glutamate could be due to an increased K Glu m , and to the relatively low glutamate concentration used. This possibility was confirmed, as we found a K Glu of 1.2 mM

Interaction of the GluRS C100Y variant with tRNAGlu, glutamate or ATP Binding of cognate tRNAGlu to C100Y and wild-type GluRS was monitored by the quenching of tryptophan fluorescence (Fig. 5A). When the profile is fitted to a single site binding equation, extracted dissociation constants are 85 ± 20 nM and 44 ± 11 nM, respectively, indicating that the perturbation of the zinc finger region by the C100Y substitution does not affect substantially tRNA binding; nevertheless, the small reduction in tRNA binding due to the C100Y substitution is consistent with an interaction of the SWIM domain of E. coli GluRS with the acceptor arm of tRNAGlu. Similarly, L-glutamate binding is not significantly altered by this substitution (Fig. 5B), with Kd values of 1.10 ± 0.07 mM for wild-type GluRS and 1.08 ± 0.04 mM for the C100Y variant, nor is ATP binding (data not shown).

m

Fig. 3. Comparison of the aminoacylation activities of wild-type E. coli GluRS (empty symbols) and of its C100Y variant (filled symbols). (A) Sensitivity to the zinc-chelating agent 1,10-phenanthroline (OP): residual activity after various periods of preincubation at 37 (cid:2)C in the presence (squares) and in the absence (circles) of 10 mM OP in 20 mM Na Hepes, pH 7.2, 0.2 mM dithiothreitol and 20% glycerol. (B) Hanes plot of the kinetics of the tRNA aminoacylation with glutamate as the variable substrate, under saturating concentration of ATP and tRNA.

728 R. Banerjee et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Fig. 6. NMR spectra of the a-protons of L-glutamate (150 lM) in the absence (- - - -) and the presence (——) of 15 lM GluRS/tRNAGlu complex; (A) wild-type GluRS, and (B) C100Y GluRS. The dissociation constants of L-glutamate from each of these enzyme/tRNA complexes were estimated with the formalism developed by Redfield [18], as described under Experimental procedures. These complexes were titrated with increasing concentrations of L-glutamate, and the line- width was determined at each concentration. Kd values of 1–10 mM and 0.1 mM (insert) were obtained for the interaction of L-glutamate with the C100Y GluRS/tRNAGlu and the wild-type GluRS/tRNAGlu complexes, respectively.

L-glutamate with free wild-type GluRS or C100Y GluRS (see above). On the other hand, a Kd value of 0.10 ± 0.08 lM, close to the Km value, was obtained for the interaction of L-glutamate with the wild-type GluRS/ tRNAGlu complex.

The fact

that C100Y substitution does not affect substantially the binding of tRNAGlu, ATP or L-glutamate (in the absence of tRNAGlu), suggests that the small changes detected by CD and fluorescence spectroscopy must be confined to the SWIM domain or to its orientation with respect to the other structural domains. Thus, the reduction of L-glutamate binding in the presence of tRNAGlu must result from the disruption of the SWIM domain interaction with the acceptor stem of tRNAGlu.

Presence and evolution of (putative) zinc binding motifs in a-type GluRSs

The GluRS/GlnRS family can be divided into two struc- tural types corresponding, respectively, to the bacterial/ organellar GluRSs (named a-type), and to the eucaryotic/ archaeal GluRSs and all GlnRSs (named b-type) [31]. Each type bears a specific pattern of insertions and deletions in their catalytic N-terminal domain, and a radically different C-terminal domain architecture consisting mainly of a-helices for the a-type and of b-sheets for the b-type

In order to substantiate the high Km value of the variant for glutamic acid, we have estimated its dissociation from the enzyme/tRNAGlu/glutamate ternary constant complex by NMR, under conditions where L-glutamate was in 10-fold excess of the GluRS/tRNAGlu complex. Under such conditions, L-glutamate signals are relatively free from interference from the GluRS/tRNAGlu complex. When a free ligand is in fast exchange with a ligand/ macromolecule complex, there is a lowering of longitudinal (T1) and transverse (T2) relaxation times of the ligand (under fast exchange only one average line is seen for the ligand). Figure 6 shows the CaH signal of L-glutamate in the presence and the absence of either wild-type or C100Y variant GluRS/tRNAGlu complex. In the case of the wild- type enzyme, the line-width is significantly broadened, but very little in the case of the C100Y variant. Similar experiments were also carried out at different enzyme- substrate ratios to confirm that the line broadening was due to binding of substrate to the enzyme (see the insert in Fig. 6A). For the interaction of L-glutamate with the C100Y GluRS/tRNAGlu complex, the high Kd value obtained (of the order of 1–10 mM) is compatible with the precise values obtained by fluorescence quenching for the interaction of

Fig. 5. Quenching of tryptophan fluorescence upon binding of tRNAGlu (A) or L-glutamate (B) to wild-type (s) and C100Y (d) GluRS. The curves (- - - - for wild-type, —— for C100Y) were fitted according to a single binding site equation as described in Experimental procedures.

SWIM domain of glutamyl-tRNA synthetases (Eur. J. Biochem. 271) 729

(cid:1) FEBS 2004

related, with a few exceptions, to the 16S rRNA tree (data not shown) and also therefore to the official taxonomic classification of the corresponding organisms (indicated to the right of Fig. 8). This observation, consistent with previous phylogenetic studies of GluRS [31,32], confirms their high degree of canonicity reported by Woese et al. [10]. The NJ tree was rooted with a b-GluRS (the GluRS of Aeropyrum pernix, one of the deepest branching Archaea) in an attempt to identify its closest counterpart in the bacterial domain. Inclusion into that tree of the quadruplets of residues aligned with the four zinc ligands of E. coli GluRS identified in Fig. 7 reveals that their evolution is not canonical (Fig. 8). Closely linked GluRSs often have similar quadruplets, but drastic differences between neighboring taxons are frequent (e.g. compare the c-proteobacteria

[8,31]. The alignment of the amino acid sequences of a large number of a-type GluRSs reveals that when a (putative) zinc binding motif is present in the ABD, the spacing of one residue between the first two ligands and between the last two ligands found in E. coli GluRS (C98-x-C100-x24-C125- x-H127) [11] is always conserved, as shown in Fig. 7 (segment corresponding to the E. coli GluRS zinc-binding motif in several a-type GluRSs); moreover, when residues at these positions are not putative zinc ligands, we did not find proximal residues that could play that role. The spacing between ligands 2 and 3 is also well preserved (around 24 residues) while the sequence itself is more variable (Fig. 7). A neighbor-joining (NJ) phylogenetic tree (see (cid:1)Experi- mental procedures(cid:2)) obtained from the aligned complete amino acid sequences of a-type GluRSs (Fig. 8) is closely

730 R. Banerjee et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Fig. 7. Alignment of sequences corresponding to the zinc-binding region of E. coli GluRS in 62 a-type GluRSs. Organism name color legend: green if the (putative) zinc-binding quadruplet is CCCX, where X is H (as in E. coli GluRS), E or D; blue, if the quadruplet is a weaker but still good candidate for binding a zinc atom with at least three residues frequently used as zinc ligands [CCCX, with X ¼ T, R, N, K, D or H, and CCH(D/N)]; orange for quadruplets which are still weaker can- didates for binding a zinc atom; red for organisms whose GluRS are unlikely to have a zinc atom at this site. The BioEdit standard amino acid color scheme, which reflects their physical and chemical proper- ties, was used. M, mitochondrial GluRSs; C, Chloroplastic GluRSs; Magnetospiri. magnetotacti., Magnetospirillum magnetotacticum; endo. of D. mela., endosymbion of Drosophila melanogaster. The Swiss-Prot/ TrEMBL accession numbers (when available) for GluRS amino acid sequences used in the present work are: E. coli, P04805; Haemophilus influenzae, P43818; Yersinia pestis, Q8ZCK0; (Acidi)Thiobacillus ferrooxidans, http://www.tigr.org (gltX-1); Streptomyces coelicolor, O86528; Thermobifida fusca, http://www.jgi.doe.gov; Aquifex aeolicus, O67271; Pseudomonas aeruginosa, Q9XCL6; Agrobacterium tumefac- iens, Q8U7H5; Saccharomyces cerevisiae (M), P48525; Treponema pallidum, O83679; Bacillus subtilis, P22250; Mycoplasma pneumoniae, P75114; Buchnera sp. P57173; Mycoplasma genitalium, P47700; Borrelia burgdorferi, O51345; Ureaplasma urealyticum, Q9PPP0; Clostridium difficile, ftp://ftp.sanger.ac.uk/pub/pathogens/cd/ Entero- coccus faecalis, Q839V7; Campylobacter jejuni #2, Q9PP78; Bacillus stearothermophilus, P22249; Rhizobium meliloti, P15189; Wolbachia sp. (endo. of D. mel.), http://www.tigr.org; Synechocystis sp. Q55778; Magnetospirillum magnetotacticum, http://www.jgi.doe.gov; Nicotiana tabacum (C), Q43794; Hordeum vulgare (C), Q43768; Rhodobacter sphaeroides, Q9ZFA3; Rickettsia prowazekii #1, Q9ZCT8; Rickettsia prowazekii #2, Q9ZDK3; Bacillus halodurans, Q9KGF6; Staphylo- coccus aureus, Q99W75; Bacillus anthracis, Q81VV3; Campylobacter jejuni #1, O52914; Helicobacter pylori J99 #1, Q9ZLZ7; Caulobacter crescentus, Q9A721; Chloroflexus aurantiacus, http://www.jgi.doe.gov; Neisseria meningitidis MC58, Q9K1R6; Bordetella pertussis, Q7W3X9; Azospirillum brasilense, P45631; Prochlorococcus marinus, Q7VDB2; Brucella melitensis, Q8YHG4; Chlamydophila pneumoniae, Q9Z7Z3; Vibrio cholerae, O31153; Chlorobium tepidum, Q9F724; Pseudomonas fluorescens, http://www.jgi.doe.gov; Pseudomonas putida, Q88LF6; Pseudomonas syringae, Q884C8; Helicobacter pylori J99 #2, Q9ZLJ1; Mycobacterium tuberculosis, O53241; Mycobacterium leprae, O33120; Xylella fastidiosa, Q9PF56; Cytophaga hutchinsonii, http://www.jgi. doe.gov; Lactobacillus bulgaricus, O86083; Lactococcus lactis, Q9CDZ7; Streptococcus pneumoniae, Q97NG1; Streptococcus pyo- genes, Q9A1J8; Thermus thermophilus, P27000; Mycoplasma pulmonis, P53662; Deinococcus radiodurans, Q9RX30; Thermotoga maritima #1, Q9X172; Thermotoga maritima #2, Q92I8.

V. cholerae and H. influenzae). A similar situation is found for the closely related Mycobacterium leprae and Mycobacterium tuberculosis whose quadruplets differ only in the distal position (ASDD and ASDH), but differ completely from the CCCE of the related Streptomyces coelicolor. These results indicate a rapid evolution of the (putative) zinc-binding region of a-type GluRSs, possibly as a response to the (cid:1)environment(cid:2) [33]. The rapid rate of the evolution of the SWIM domains of a-GluRSs, documented here, is similar to that of the other known SWIM domains [12].

Out of the 62 quadruplets of the a-type GluRSs shown in Fig. 8, 11 sequences are CCCX, where X is H in six of them, E in three cases and D in two cases; we wrote the corresponding organism names in green, to indicate that we

consider them excellent candidates for the presence of zinc in their GluRS, considering the involvement of D and E as zinc ligands in several proteins of known 3D-structure [34]. Quadruplets likely to bind zinc, either because the four residues are frequent zinc-binders, such as CCHD of

Fig. 8. Rooted NJ tree of the whole sequence of a-type GluRSs. Bootstrap value (> 50%) obtained from 1000 replicates for the NJ or 5000 puzzling steps in ML analysis (see Experimental procedures) are indicated in that order at their corresponding nodes separated by (cid:1)/(cid:2). The tree is rooted with the archaeal b-type GluRS of Aeropyrum pernix (not shown). See the legend of Fig. 7 for the organism name’s color signification. The putative zinc-binding or non zinc-binding residues aligned with the four zinc ligands of E. coli GluRS are shown on the right. The one letter amino-acid code are colored according to their frequency as zinc binder in known 3D structures (35): green for very frequent, blue for less frequent, orange for unknown, and red for absence of binding. Organism abbreviations are Magneto., Magneto- spirillum; Ric., Rickettsia; Azo., Azospirillum; Chlamydo., Chlamydo- phila; Sac., Saccharomyces.; endo. of D. mela., endosymbion of Drosophila melanogaster. Taxon abbreviations are Aqui., Aquificales; Proteo., Proteobacteria; Molli., Mollicutes; Bac./Clos., Bacillus/Clos- tridium group; Bac./Staph., Bacillus/Staphylococcus subgroup; Entero., Enterococcaceae; Lacto., Lactobacillaceae; Strepto., Streptococcaceae; Cyano., Cyanobacteria; Chloro., Chloroplasta; Gr. non-S., Green nonsulfur bacteria; Spiro., Spirochaetales; The./Dei., Thermus/Deino- coccus group; Chlam., Chlamydiales; Clos., Clostridiaceae; Mito., Mitochondria; CFB/Gr. S., Cytophaga-Flexibacter-Bacteroides group/ Green sulfur bacteria; Actino., Actinobacteria; Thermo., Thermoto- gales; a, b, c, e: alpha, beta, gamma and epsilon subdivision. Taxons whose names are red contain a noncanonical GluRS.

SWIM domain of glutamyl-tRNA synthetases (Eur. J. Biochem. 271) 731

(cid:1) FEBS 2004

B. subtilis GluRS which binds zinc [28], or because they include a less frequent zinc-binder (N, S, T, K) or the putative binder R together with three frequent ones, have their corresponding organism names in blue. This includes the CYCH of the C100Y variant of E. coli GluRS presented here. Several cases of zinc-binding sites including these less frequent zinc-binding residues are known from 3D-struc- tures of other proteins [34]. The fact that these eight types of residues which occupy the distal position in these 24 CC(C/ H)X quadruplets are not similar by their structure or their charge, nor by their evolutionary proximity [23] suggests that their selection may be due to their capacity to share electrons with a metal ion. These residues cover a large spectrum of metal-binding properties [35] which might generate metal- binding sites with a wide diversity of strengths and geo- metries, mostly for zinc but possibly for another metal (e.g. copper or magnesium if N or M are present; see [36,37]). These facts suggest that the loop corresponding to the zinc- binding segment of E. coli GluRS in other a-type GluRSs may have different conformations, thus participating to the fine tuning of the interaction between the ABD of GluRSs and their cognate tRNAs during bacterial evolution. Finally, the presence of only one or two putative zinc-binding residues in some quadruplets indicates that they do not bind zinc, as is known for AEAN of Thermus thermophilus GluRS [38].

in tRNA recognition sites, and of tRNAGln variants altered in identity elements for GlnRS has shown that the specific interactions between the identity and recognition elements of the GlnRS/tRNAGln complex determine the efficiency of glutamine recognition [45]; in particular, the GlnRS L136A variant, altered in the leucine residue which stabilizes the peeled-back conformation assumed by the acceptor end of tRNAGln in its complex with GlnRS [5], has a fivefold increase in Km for glutamine [45]. This residue is located in the helix/loop D segment (in the unifying GlxRS nomen- clature, see above) homologous to the zinc-binding region of E. coli GluRS. Furthermore, the adenine ring and the ribose of A76 in tRNAGln make specific interactions with Tyr211 and Phe233 in E. coli GlnRS (which correspond to Tyr187 and His203 in E. coli GluRS) thereby positioning them and creating part of the glutamine-binding site [46,47]. The fact that L-glutamate binds with identical affinities to wild-type GluRS and to the C100Y variant in the absence of tRNAGlu (Fig. 5B) indicates that the reduction of L-glutamate affinity in the aminoacylation reaction cata- lyzed by this variant is not due to direct effects of the mutation on the L-glutamate binding site, but to the removal of a specificity-determining interaction of GluRS with the acceptor arm of tRNAGlu, which is involved in the formation of the glutamate-binding site [48]. As observed here for the C100Y GluRS, the loss of an identity element does not always affect the aaRS/tRNA affinity: such cases include E. coli GlnRS variants with slightly decreased affinities for tRNAGln, but with five- to 10-fold increases in their Km for glutamine [49], and the unaltered binding of E. coli methionyl-tRNA synthetase to variant microhelices of methionine tRNA altered in identity elements [50].

Discussion

Zinc is a metal of borderline (cid:1)hardness(cid:2) which easily accommodates nitrogen, oxygen and sulfur atoms in its biological coordination polyhedra [35]. The diversity of the observed and of the putative zinc-binding residues of several bacterial GluRSs (Figs 7 and 8) suggests that this flexibility of zinc was used in the modulation of the structure of the loops homologous to the Cys100-Cys125 loop of E. coli GluRS, by influencing both the strength and the orientation of the ligand-zinc bonds. Indeed, besides the well established role of C, H, E and D residues in zinc coordination [39] the involvement of N, M, T, S, Y and W in metal binding has been observed, although less frequently (reviewed in [34,35,40,41]). For instance, the side chain hydroxyl group of threonine is coordinated to the zinc ion located in the active site of E. coli ThrRS [42]; zinc coordination by N in H385N ThrRS is suggested by the fact that this variant binds zinc better than does the H385A variant [43]. The fact that the zinc-ligand orientation often differs for carbonyl, nitrogen and sulfur ligands [40] suggests that there may be significant differences in the orientation of the loops closed by zinc (or another metal) in GluRSs using different residues as metal ligands.

The presence of distinct quadruplets in two GluRSs present in the same organism could be due either to the recent acquisition of a noncanonical GluRS by horizontal transfer from a distantly related organism, or to its early acquisition followed by the conservation of its distinct quadruplet for functional reasons.

Roles of tRNAGlu and of tRNAGln in the formation of the amino acid substrate binding sites of E.coli GluRS and GlnRS, respectively

The C100Y substitution in the E. coli GluRS SWIM domain, included in the tRNA acceptor stem-binding domain, reduces only twofold the binding affinity for tRNAGlu, while reducing more than 10-fold the affinity for glutamate in the presence of tRNAGlu. Those results the SWIM/tRNAGlu interaction plays a indicate that minor role in the overall binding of tRNA, but is needed the acceptor the proper orientation of locally for 3¢-terminal adenosine, which contributes in T. thermophi- lus GluRS to the formation of the active site [48]. In this context, we propose that the evolution of the GluRS SWIM domains was selected for the fine tuning of their interaction with the tRNA substrate acceptor arm, which is directly involved in the formation of the glutamate- binding site [48]. We propose that the SWIM domain in GluRS or its functional homologues in other aaRSs may have been inserted to restructure the active site so as to impart a much higher selectivity of the cognate amino acid through the interaction with the acceptor arm of the cognate tRNA. This is an aspect of the parallel evolution of the two components of active GluRS: the protein apoenzyme and its tRNA cofactor-substrate. The presence of SWIM domains in GluRSs at this critical position is consistent with the rapid evolution of all known SWIM domains [12]. Their plasticity may have facilitated the adaptation of the protein apoenzyme to the independently evolving tRNA cofactor–substrate.

GluRS and GlnRS share the property of requiring their cognate tRNAs for activating their respective amino acid substrates ([44]). A comparison of the kinetics of amino- acylation of wild-type tRNAGln by GlnRS variants altered

732 R. Banerjee et al. (Eur. J. Biochem. 271)

(cid:1) FEBS 2004

Acknowledgements

14. Lin, S.X., Brisson, A., Liu, J., Roy, P.H. & Lapointe, J. (1992) Higher specific activity of the Escherichia coli glutamyl-tRNA synthetase purified to homogeneity by a six-hour procedure. Protein Expr. Purif. 3, 71–74.

15. Madore, E., Florentz, C., Giege, R., Sekine, S., Yokoyama, S. & Lapointe, J. (1999) Effect of modified nucleotides on Escherichia coli tRNAGlu structure and on its aminoacylation by glutamyl- tRNA synthetase. Predominant and distinct roles of the mnm5 and s2 modifications of U34. Eur. J. Biochem. 266, 1128–1135. 16. Cornish-Bowden, A. (1995) Fundamentals of Enzyme Kinetics. Portland Press, London.

17. Davis, B.J. (1964) Disc electrophoresis. II. Method and application to human serum proteins. Ann. NY Acad. Sci. 121, 404–427. 18. Redfield, A.G. (1978) Nuclear magnetic resonance kinetics viewed as enzyme kinetics. Methods Enzymol. 49, 359–369.

This work was supported in part by a grant from CSIR (India) to S. R., and by grants OGP0009597 from the Natural Sciences and Engineering Research Council of Canada (NSERC) and 2003-ER-2481 from the Fonds pour la Formation de Chercheurs et l’Aide a` la Recherche du Que´ bec (FCAR) to J. L. D. Y. D. was a FCAR doctoral fellow. Sequencing of Acidithiobacillus ferrooxidans and Wolbachia sp. (endo- symbiont of Drosophila melanogaster) were accomplished with support from US Department of Energy and NIH, respectively. The sequence data for Thermobifida fusca, Magnetospirillum magnetotacticum, Chlo- roflexus aurantiacus, Pseudomonas fluorescens and Cytophaga hutchin- sonii were produced by the US Department of Energy Joint Genome Institute (http://www.jgi.doe.gov). Clostridium difficile sequence data were produced by the Clostridium difficile Sequencing Group at the Sanger Institute and can be obtained from ftp://ftp.sanger.ac.uk/pub/ pathogens/cd/ 19. Altschul, S.F., Gish, W., Miller, W., Myers, E.W. & Lipman, D.J. (1990) Basic local alignment search tool. J. Mol. Biol. 215, 403–410.

References

20. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. & Higgins, D.G. (1997) The CLUSTAL_X windows interface: flex- ible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25, 4876–4882.

1. Schimmel, P., Giege, R., Moras, D. & Yokoyama, S. (1993) An operational RNA code for amino acids and possible relationship to genetic code. Proc. Natl Acad. Sci. USA 90, 8763– 8768. 21. Hall, T.A. (1999) BIOEDIT: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series 41, 95–98.

2. Schwob, E. & Soll, D. (1993) Selection of a (cid:1)minimal(cid:2) glutaminyl- tRNA synthetase and the evolution of class I synthetases. EMBO J. 12, 5201–5208.

22. Felsenstein, J. (1993) PHYLIP (Phylogeny Inference Package), version 3.6a2. Distributed by the author: http://evolution. genetics.washington.edu/phylip.html Department of Genetics, University of Washington, Seattle. 3. Ribas de Pouplana, L. & Schimmel, P. (2001) Operational RNA code for amino acids in relation to genetic code in evolution. J. Biol. Chem. 276, 6881–6884.

4. Carter, C.W. Jr (1993) Cognition, mechanism, and evolutionary relationships in aminoacyl-tRNA synthetases. Ann. Rev. Biochem. 62, 715–748. 23. Dayhoff, M.O., Eck, R.V. & Park, C.M. (1972) A model of evo- lutionary change in proteins. In Atlas of Protein Sequence and Structure (Dayhoff, M.O., eds), pp. 89–99. National Biomedical Research Foundation, Washington DC.

24. Strimmer, K. & von Haeseler, A. (1996) Quartet puzzling: a quartet maximum likelihood method for reconstructing tree topologies. Mol. Biol. Evol. 13, 964–969.

5. Rould, M.A., Perona, J.J., Soll, D. & Steitz, T.A. (1989) Structure of E. coli glutaminyl-tRNA synthetase complexed with tRNA (Gln) and ATP at 2.8 A˚ resolution. Science 246, 1135–1142. 6. Rould, M.A., Perona, J.J. & Steitz, T.A. (1991) Structural basis of anticodon loop recognition by glutaminyl-tRNA synthetase. Nature 352, 213–218. 25. Jones, D.T., Taylor, W.R. & Thornton, J.M. (1992) The rapid generation of mutation data matrices from protein sequences. Computer Applic. Biosci. 8, 275–282. 26. Page, R.D.W.

7. Weygand-Durasevic, I., Schwob, E. & Soll, D. (1993) Acceptor end binding domain interactions ensure correct aminoacylation of transfer RNA, Proc. Natl Acad. Sci. USA 90, 2010–2014. (1996) TREEVIEW: An application to display phylogenetic trees on personal computers. Computer Applic. Bio- sci. 12, 357–358.

27. Posorske, L.H., Cohn, M., Yanagisawa, N. & Auld, D.S. (1979) Methionyl-tRNA synthetase of Escherichia coli: a zinc metallo- protein. Biochim. Biophys. Acta 576, 128–133. 8. Nureki, O., Vassylyev, D.G., Katayanagi, K., Shimizu, T., Sekine, S., Kigawa, T., Miyazawa, T., Yokoyama, S. & Morikawa, K. (1995) Architectures of class-defining and specific domains of glutamyl-tRNA synthetase. Science 267, 1958–1965.

28. Liu, J., Lin, S.X., Blochet, J.E., Pezolet, M. & Lapointe, J. (1993) The glutamyl-tRNA synthetase of Escherichia coli contains one atom of zinc essential for its native conformation and its catalytic activity. Biochemistry 32, 11390–11396.

9. Breton, R., Watson, D., Yaguchi, M. & Lapointe, J. (1990) Glu- tamyl-tRNA synthetases of Bacillus subtilis 168T and of Bacillus stearothermophilus: cloning and sequencing of the gltX genes and comparison with other aminoacyl-tRNA synthetases. J. Biol. Chem. 265, 18248–18255. 29. Landro, J.A. & Schimmel, P. (1994) Zinc-dependent cell growth conferred by mutant tRNA synthetase, J. Biol. Chem. 269, 20217–20220. 30. Freist, W., Gauss, D.H., Soll, D. & Lapointe, J. (1997) Glutamyl- 10. Woese, C.R., Olsen, G.J., Ibba, M. & Soll, D. (2000) Aminoacyl- tRNA synthetases, the genetic code, and the evolutionary process. Microbiol. Mol. Biol. Rev. 64, 202–236. tRNA sythetase. Biol. Chem. 378, 1313–1329.

31. Siatecka, M., Rozek, M., Barciszewski, J. & Mirande, M. (1998) Modular evolution of the Glx-tRNA synthetase family-rooting of the evolutionary tree between the bacteria and archaea/eukarya branches. Eur. J. Biochem. 256, 80–87.

11. Liu, J., Gagnon, Y., Gauthier, J., Furenlid, L., L’Heureux, P.J., Auger, M., Nureki, O., Yokoyama, S. & Lapointe, J. (1995) The zinc-binding site of Escherichia coli glutamyl-tRNA synthetase is located in the acceptor-binding domain: studies by extended X-ray absorption fine structure, molecular modeling, and site-directed mutagenesis. J. Biol. Chem. 270, 15162–15169. 32. Brown, J.R. & Doolittle, W.F. (1999) Gene descent, duplication, and horizontal transfer in the evolution of glutamyl- and gluta- minyl-tRNA synthetases. J. Mol. Evol. 49, 485–495. 33. Roberts, M.S. & Cohan, F.M. 12. Makarova, K.S., Aravind, L. & Koonin, E.V. (2002) SWIM, a novel Zn-chelating domain present in bacteria, archaea and eukaryotes. Trends Biochem. Sci. 27, 384–386. (1995) Recombination and migration rates in natural populations of Bacillus subtilis and Bacillus mojavensis. Evolution 49, 1081–1094. 34. Auld, D.S. (2001) Zinc coordination sphere in biochemical zinc 13. Kunkel, T.A., Roberts, J.D. & Zakour, R.A. (1987) Rapid and efficient site-specific mutagenesis without phenotypic selection. Methods Enzymol. 154, 367–382. sites. Biometals 14, 271–313.

SWIM domain of glutamyl-tRNA synthetases (Eur. J. Biochem. 271) 733

(cid:1) FEBS 2004

35. Glusker, J.P. (1991) Structural aspects of metal complex enlightens its repressor activity and reveals an essential zinc ion in the active site. Cell 97, 371–381. liganding to functional groups in proteins. Adv. Protein Chem. 42, 1–76.

44. Schimmel, P.R. & Soll, D. (1979) Aminoacyl-tRNA synthetases: general features and recognition of transfer RNAs. Ann. Rev. Biochem. 48, 601–648. 36. Chakrabarti, P. (1989) Geometry of interaction of metal ions with sulfur-containing ligands in protein structures. Biochemistry 28, 6081–6085.

45. Ibba, M., Hong, K.W., Sherman, J.M., Sever, S. & Soll, D. (1996) Interactions between tRNA identity nucleotides and their recognition sites in glutaminyl-tRNA synthetase determine the cognate amino acid affinity of the enzyme. Proc. Natl Acad. Sci. USA 93, 6953–6958. 37. Tibbitts, T.T., Murphy, J.E. & Kantrowitz, E.R. (1996) Kinetic and structural consequences of replacing the aspartate bridge by asparagine in the catalytic metal triad of Escherichia coli alkaline phosphatase. J. Mol. Biol. 257, 700–715.

46. Rath, V.L., Silvian, L.F., Beijer, B., Sproat, B.S. & Steitz, T.A. (1998) How glutaminyl-tRNA synthetase selects glutamine. Structure 6, 439–449. 38. Nureki, O., Kohno, T., Sakamoto, K., Miyazawa, T. & Yoko- yama, S. (1993) Chemical modification and mutagenesis studies on zinc binding of aminoacyl-tRNA synthetases, J. Biol. Chem. 268, 15368–15373.

39. Vallee, B.L. & Auld, D.S. (1990) Zinc coordination, function, and structure of zinc enzymes and other proteins. Biochemistry 29, 5647–5659. 47. Liu, J., Ibba, M., Hong, K.W. & Soll, D. (1998) The terminal adenosine of tRNAGln mediates tRNA-dependent amino acid recognition by glutaminyl-tRNA synthetase. Biochemistry 37, 9836–9842. 40. Christianson, D.W. (1991) Structural biology of zinc. Adv. Protein Chem. 42, 281–355. 41. Vallee, B.L. & Auld, D.S. (1993) Zinc: biological functions and coordination motifs. Acc. Chem. Res. 26, 543–551. 48. Sekine, S., Nureki, O., Dubois, D.Y., Bernier, S., Chenevert, R., Lapointe, J., Vassylyev, D.G. & Yokoyama, S. (2003) ATP binding by glutamyl-tRNA synthetase is switched to the produc- tive mode by tRNA binding. EMBO J. 22, 676–688.

42. Sankaranarayanan, R., Dock-Bregeon, A.C., Rees, B., Bovee, M., Caillet, J., Romby, P., Francklyn, C.S. & Moras, D. (2000) Zinc ion mediated amino acid discrimination by threonyl-tRNA syn- thetase. Nat. Struct. Biol. 7, 461–465. 49. Hong, K.W., Ibba, M., Weygand-Durasevic, I., Rogers, M.J., Thomann, H.U. & Soll, D. (1996) Transfer RNA-dependent cognate amino acid recognition by an aminoacyl-tRNA synthe- tase. EMBO J. 15, 1983–1991.

43. Sankaranarayanan, R., Dock-Bregeon, A.C., Romby, P., Caillet, J., Springer, M., Rees, B., Ehresmann, C., Ehresmann, B. & Moras, D. (1999) The structure of threonyl-tRNA synthetase-tRNA (Thr) 50. Gale, A.J., Shi, J.P. & Schimmel, P. (1996) Evidence that specifi- city of microhelix charging by a class I tRNA synthetase occurs in the transition state of catalysis. Biochemistry 35, 608–615.