The Escherichia coli YadB Gene Product Reveals a Novel Aminoacyl-tRNA Synthetase Like Activity

J. Mol. Biol. (2004) 337, 273–283

doi:10.1016/j.jmb.2004.01.027

The Escherichia coli YadB Gene Product Reveals a Novel Aminoacyl-tRNA Synthetase Like Activity Vale´rie Campanacci1, Daniel Y. Dubois2, Hubert D. Becker3 Daniel Kern3*, Silvia Spinelli1, Christel Valencia1, Fabienne Pagot1 Aurelia Salomoni1, Sacha Grisel1, Renaud Vincentelli1 Christophe Bignon1, Jacques Lapointe2, Richard Giege´3 and Christian Cambillau1* 1

Architecture et Fonction des Macromole´cules Biologiques UMR 6098, CNRS and Universite´s d’Aix-Marseille I and II, 31 chemin J. Aiguier F-13402 Marseille Cedex 20 France 2

De´partement de Biochimie et Microbiologie, Faculte´ de Sciences et de Ge´nie, CREFSIP Universite´ Laval, Quebec Canada G1K 7P4 3

De´partement “Me´canismes et Macromole´cules de la Synthe`se Prote´ique et Cristallogene`se” UPR 9002, Institut de Biologie Mole´culaire et Cellulaire du CNRS, 15 rue Rene´ Descartes F-67084 Strasbourg Cedex France *Corresponding authors

In the course of a structural genomics program aiming at solving the structures of Escherichia coli open reading frame products of unknown ˚ using molfunction, we have determined the structure of YadB at 1.5 A ecular replacement. The YadB protein is 298 amino acid residues long and displays 34% sequence identity with E. coli glutamyl-tRNA synthetase (GluRS). It is much shorter than GluRS, which contains 468 residues, and lacks the complete domain interacting with the tRNA anticodon loop. As E. coli GluRS, YadB possesses a Zn2þ located in the putative tRNA acceptor stem-binding domain. The YadB cluster uses cysteine residues as the first three zinc ligands, but has a weaker tyrosine ligand at the fourth position. It shares with canonical amino acid RNA synthetases a major functional feature, namely activation of the amino acid (here glutamate). It differs, however, from GluRSs by the fact that the activation step is tRNA-independent and that it does not catalyze attachment of the activated glutamate to E. coli tRNAGlu, but to another, as yet unknown tRNA. These results suggest thus a novel function, distinct from that of GluRSs, for the yadB gene family. q 2004 Elsevier Ltd. All rights reserved.

Keywords: YadB; glutamyl-tRNA synthetase; zinc cluster; aaRSs; structural genomics

Introduction Although following diverse approaches, most structural genomics programs share two features; namely, a high-throughput approach and a “discovery” instead of function-based motivation. With these two features in mind, we have set up a medium-scaled structural genomics program aiming at solving the structures of as many bacterial unknown open reading frame (ORF) products as possible among 110 targets (see the web site†). Abbreviations used: aaRS, amino acid RNA synthetase; ORF, open reading frame; GoA, glutamol-AMP. E-mail addresses of the corresponding authors: [email protected]; [email protected] † http://afmb.cnrs-mrs.fr

The rationale for choosing the targets was their wide distribution among the bacterial reign, their unknown function (“y” identifier) and an amino acid identity of less than 30– 35% with any protein sequence of known function.1 The general approach for expression and crystallization of these 110 ORF products is described elsewhere.2,3 Among the selected targets, the yadB gene (target 12, internal numbering2,3) from Escherichia coli was of particular interest. Its putative product shares 65– 88% identity with several gene products from other bacteria with similar length (, 300 amino acid residues), defining thus a putative class of orthologous genes. The YadB protein shares 32% identity with Thermus thermophilus glutamyl-tRNA synthetase (GluRS; Figure 1) and 34% with its homologue from E. coli. GluRSs are members of

0022-2836/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.

274

the family of aminoacyl-tRNA synthetases (aaRSs), a group of enzymes ranked in two classes4 that catalyze specific attachment of amino acids on their cognate tRNAs and thereby are essential factors for correct expression of the genetic code at the translational level.5,6 Interestingly, the YadB gene product shares also 23% identity with E. coli GlnRS, a protein that is structurally and evolutionarily related to GluRSs.7 GluRSs and GlnRSs form the GlxRS superfamily7 – 9 and constitute subclass Ib of aaRSs.6 However, YadB is generally closer to GluRS than to GlnRS. But in contrast to the genes of GlxRSs and of other class I aaRSs that code for protein monomers having , 500 amino acid residues,10 the yadB gene codes for only 298 amino acid residues. Thus, the YadB gene product is 173 amino acid residues shorter than the 471 amino acid residue long E. coli GluRS.8 It lacks the last helix of the stem-contact fold (SC-fold) and the two domains constituting the anticodon-binding region which in many classical aaRSs, including E. coli class Ib GluRS and GlnRS, plays a major role in tRNA recognition11 when interacting with the identity determinants of the tRNA anticodon loop.12 The yadB ORF is conserved in many bacterial genomes and is transcribed in E. coli under aerobic conditions during the exponential growth phase.13 All these considerations are in favor of a functional role of YadB. However, the absence of an anticodon-binding domain indicates that YadB likely does not play the role of a classical aaRS, at least with our present knowledge on this class of enzymes. We decided therefore to clone, express, crystallize and determine the structure of the YadB gene product, in view of possibly getting some functional clues from its structure, and performing

Structure and Activity of the E. coli YadB

initial activity exploration. We could solve the structure of YadB by molecular replacement using the GluRS from T. thermophilus (PDB entry 1GLN)14 as the starting model. Its structure has ˚ resolution, with final R and been refined to 1.5 A Rfree of 14.6% and 16.9%, respectively. We have discovered that YadB possesses a non-classical Zncluster, coordinated by three cysteine residues and one tyrosine moiety. A crevice in the structure, comparable to that seen in other GluRSs, could accommodate nicely an analogue of glutamylAMP, the glutamol-AMP (GoA), without clash, while analogy docking with tRNAGlu was less straightforward. To complement the structural information, the activity of YadB has also been explored for possible amino acid activation and tRNA aminoacylation.

Results and Discussion Overall structure description The YadB structure comprises 273 amino acid residues visible in density from a total of 298 in sequence. Compared to GluRS, YadB includes the connective-peptide domain (CP-domain), the two halves of the Rossmann-fold domain and part of the SC-fold domain (Figure 2A and B). The structure of YadB is of excellent quality as can be judged from crystallographic and geometric criteria (Table 1; Figure 2C). The map is generally well defined, except at the N and C termini (residues 1 –3 and 290 –298) and for the residue stretch 224 – 236, which have no density at all (Figure 2). This latter missing loop domain bears the KMSKS motif (228 KLSKQ 232 in YadB and 243 KISKR 247

Figure 1. Sequence alignment of YadB. A, Alignment with the closest neighbor, the T. thermophilus GluRS (1GLN).14 The residues not visible in the electron density map of YadB are displayed on a green background and the Zn-cluster ligands are identified with green filled circles. The residues in complex with the aminoacyl-adenylate in the modeled complex are marked with red filled circles. B, Alignment of the YadB sequence 98 – 121 with its equivalent in E. coli GluRS. The Zn-cluster ligands of YadB (Cys101, 103, 119 and Tyr115) are identified with green filled circles, and the fourth ligand in the E. coli GluRS cluster (His127) is identified with a black filled circle.

Structure and Activity of the E. coli YadB

275

Figure 2. X-ray structure of E. coli YadB. A, Ribbon representation of the YadB structure made with MOLSCRIPT.52 The N and C termini are identified with blue letters and the residues flanking the disordered loops are indicated by red numbers. The Zn atom is identified by a red sphere (top). B, Stereo view of the Ca trace of YadB (yellow) superimposed to that of the T. thermophilus GluRS (1GLN) (blue).14 The aminoacyl-adenylate has been modeled according to its position in the E. coli GlnRS ternary complex (1QTQ).25 The Zn atom is identified by a pink sphere (top). C, Stereo view of the 2Fo 2 Fc electron density map (1s contour) of the Zn-cluster of YadB. D, Schematic representation ˚ , angles are in degrees (italics). of the Zn cluster geometry; distances (bold) are in A

in T. thermophilus GluRS) (1GLN), a signature motif of class I aaRSs which is known to help correct positioning of the adenine base of ATP in the Rossmann-fold domain.14 This domain is very often hardly visible in synthetase structures and has high B-factors in those structures without ATP or aminoacyl-AMP, which is the case for YadB. Since the mass spectra indicate that YadB is intact, we believe that the KMSKS-bearing subdomain is disordered and not cleaved. In contrast, the HIGH motif (19-HFGS-22 in YadB and 15-HVGT-18 in

T. thermophilus GluRS) (1GLN), the other signature sequence in class I aaRSs including GluRS, that is also involved in binding the adenine base of ATP, is well seen in the YadB structure. The Rossmannfold is very similar to that of the other class I synthetases (Figure 2A), and more particularly to the related subclass Ib T. thermophilus GluRS (1GLN)14 and to a lesser extent to E. coli GlnRS (1QTQ) complexed with tRNA and a glutaminyladenylate analogue.15 The limited crystallographic contact surfaces are in agreement with the DLS

276

Structure and Activity of the E. coli YadB

Table 1. Data collection and refinement statistics of the YadB structure A. Data collection Space group Unit cell parameters ˚) a, b, c (A b (deg.) Beamline Temperature (K) ˚) Wavelength (A Number of images ˚ )a Resolution range (A Number of observations Unique reflections Redundancya Completenessa (%) I/sIa Rsym b (%) B. Refinement ˚) Resolution range (A Unique reflections No. residues/water molecules/ions R=Rfree a Rmsd ˚) Bond angles (A Bond angles (deg.) ˚ 2) Mean B-value (A

C2 115.9, 39.3, 69.15, 116.2 ID14-EH1 100 0.934 203 (18 osc) 25.0– 1.50 164,575 43,648 3.7 (3.5) 99.4 (99.4) 7.6 (2.5) 5.7 (28.3) 25.0– 1.50 43,648 273/356/1 0.146/0.169 0.009 1.2 12.6

a Numbers in parentheses correspond to the last resolution ˚ ). shell (1.58–1.50 A P P P P b Rsym ¼ h i ðIðh;IÞ 2 kIlh Þ= h i kIlh :

results, indicating that YadB, like GluRSs and GlnRSs, is a monomeric molecule. Comparison with other aminoacyltRNA synthetases YadB shares 32% sequence identity with the Rossmann domain of the T. thermophilus GluRS (1GLN; Figure 1) and 23% with that of E. coli GlnRS (1QTQ).15 Compared to both enzymes, YadB lacks the large C-terminal anticodon binding domain (Figure 2B). We do not suspect any annotation problem, however, since the yadB ORF and the next ORF are separated by only a few tens of bases, which would not account for a ,170-residue domain. Furthermore, yadB orthologues have been identified in several bacterial genomes, all having comparable lengths. These elements definitively confirm that the lack of the anticodon domain is not artefactual. Superimposition of YadB and the Rossmann domain of free T. thermophilus GluRS (residues 1 – 303; Figure 2B) and E. coli GlnRS complexed to tRNA yields root-mean-square (r.m.s.) deviations calculated on the Ca atoms of the 273 residues of ˚ and 2.9 A ˚ , respectively. These average r.m.s. 1.9 A deviations take into account significant conformational differences in subdomains between YadB and the GlxRSs, such as the stretches 121 –127 (helix versus strand in GluRS), 136 –158 and 254 – 270 (different tracks). Insertions with respect to YadB, spread along the GluRS sequence between residues 71 –79, 108 – 122 and 128– 142. Indeed, the core of both proteins is closer than considering the ˚ , 161 whole structure: using a cutoff value of 1.5 A ˚ Ca residues are within an r.m.s. deviation of 0.9 A

between YadB and T. thermophilus GluRS and 126 ˚ for E. coli GlnRS. This residues are within 1.0 A explains well why the YadB structure could be solved by molecular replacement using the structure of a GluRS (1GLN). The Zn-cluster A high electron density level was observed near Cys101, Cys103, Cys119 and Tyr115, suggesting the presence of a metal ion in YadB (Figure 2C and D). The GluRSs of E. coli and Bacillus subtilis contain a zinc atom,16 and that of T. thermophilus does not.14,17 In E. coli GluRS, the zinc is essential for catalysis of the aminoacylation reaction,16 and is located in the tRNA acceptor stem-binding domain.18 The ligands of this zinc (Cys98, Cys100, Cys125 and His127) were identified by a mutagenesis analysis. This identification is consistent with the detection by EXAFS of three sulfur atoms and one nitrogen atom in the coordination environment of this zinc, and was used in the modelling of the Zn-binding site structure using the known crystal structures of T. thermophilus GluRS14 and E. coli GlnRS.19 The resulting model of the E. coli GluRS Zn-cluster18 is probably not fully similar to that of YadB reported here. Sequence alignments between YadB and E. coli GluRS (Figure 1b) indicate that Tyr115/121 is conserved while E. coli GluRS His127 is valine in YadB, suggesting a different arrangement of the fourth ligand in the cluster. Other aaRSs belonging to class I possess Zn2þ at a comparable position. This is the case for the T. thermophilus MetRS20 which possesses Zn2þ liganded by cysteine and histidine residues and for E. coli MetRS, IleRS, LeuRS and ValRS in which the Zn2þ is liganded by four cysteine residues.21 The metal ion density of YadB is fully compatible with the presence of Zn2þ. Its ligation, however, differs from that of the above-mentioned aaRSs, since the fourth coordination site, the hydroxyl group of a tyrosine side-chain, might very likely be labile, since tyrosine side-chains are less strong ligands than cysteine or histidine residues. Other aaRSs, such as CysRS22 belonging to class Ia or ThrRS23 belonging to class II possess Zn-clusters with a labile ligand: a glutamic acid in the former case and a water molecule in the latter. These positions have been shown to be of functional importance, since the cysteine substrate can bind the Zn2þ of CysRS, and the threonine substrate that of ThrRS. In both cases, however, the Zn-cluster is very close to the ATP-binding position, making a direct transfer of the amino acid to ATP possible. This is not the case with YadB, since the position ˚ away from the ATP first of the cluster is 30 A phosphate position (see below), making transfer impossible (Figure 2A and B). Possible interactions between YadB and glutamyl-AMP, GoA, ATP and tRNAGlu Several coordinate sets from structures closely

277

Structure and Activity of the E. coli YadB

related to YadB are presently available, namely the free GluRS from T. thermophilus (1GLN)14 and its complex with tRNAGlu (1G59),24 ATP (1N75),25 ATP and Glu (1J09),25 ATP and tRNAGlu (1N77)25 and the glutamyl-AMP analogue GoA and tRNAGlu (1N78).25 They represent an extraordinary source of information illustrating the mechanisms of amino acid activation and tRNA aminoacylation by GluRS. In addition, the coordinates of the complex between the E. coli GlnRS, tRNAGln and glutaminyl-AMP (1QTQ)15 are also available. We have tried to co-crystallize YadB with ATP, but despite numerous efforts we did not succeed. This may be due to the lower affinity of ATP for YadB as compared to GluRS (Table 2). Since the structures of YadB and GlxRSs are very close in ˚ with their core domain (r.m.s. deviation of , 0.9 A the five GluRS structures mentioned above), we have superimposed the models of YadB and T. thermophilus native and complexed GluRS, thus bringing the ligands (ATP, ATP and Glu, aminoacyl-AMP, its analogue GoA, and tRNA) into positions similar to those observed in the experimental complexes. The ATP and glutamate positions (from 1J09) do not clash with YadB. The glutamate moiety is well fitted in a crevice in a position close to that of the GoA analogue (Figure 2b). The interaction of the glutamate side-chain with YadB retains all the bonds observed in GluRS. It is worth noticing that the O11 and O12 atoms of glutamate fall close to two water molecules bound to Arg9 in YadB. As reported,25 the ATP molecule position in the GluRS active site depends on the presence of the tRNA. In the GluRS/Glu/ATP complex (1J09), the ATP position is not as deep in the pocket as in the GluRS/ATP/tRNAGlu complex (1N77). In the latter case, the bound tRNA moves the side-chain of Trp209 deeper in the active site crevice due to short contacts. This displacement allows the ribose moiety of the ATP to establish a hydrogen-bond with Ala206 main-chain NH at the bottom of the crevice, in a favorable position for the glutamylation reaction to occur. The new contact and the neighboring residues induce a rotation of the ribose moiety, bringing the triphosphate motif into contact with the tRNA and with a different part of the GluRS, Thr11, Thr18 and Arg47, and no more with the flap. In YadB, the residue corresponding to Trp209 is Leu194. As a consequence, even in the absence of

tRNA, the modelled ATP molecule is able to establish an hydrogen bond with Gly191 NH (equivalent to Ala206 in GluRS). Many of the ATP contacts observed with GluRS in the GluRS/ATP/tRNAGlu complex (1N77) are kept in YadB, except those that involve the flap (non-visible in YadB). The triphosphate moiety displays also several contacts in common between the model with YadB and the GluRS ternary complex, except those with Arg47 in GluRS, since the equivalent residue in YadB is not an arginine and the main-chain is far from ATP. To conclude with the ATP, its position in YadB can be modelled similar to that in the ternary complex GluRS/ATP/tRNAGlu (1N77) meaning that YadB would not need tRNA activation to perform the glutamylation. The GoA analogue fits exquisitely well in YadB (Figure 2B), since only a few too short or too long contacts are observed, compared to GluRS, which could be easily removed by side-chain rotations. Only three contacts out of 22 are not observed between GoA and YadB: one involving a residue of the flap, and two involving the last residue of the tRNAGlu. Attempts to accommodate E. coli tRNAGlu on the YadB structure were less straightforward and did not allow us to predict a potential interaction of the amino acid accepting arm of the tRNA with the protein. Due to the complexity of the recognition pattern and to the presence of short contacts difficult to remove by simple manipulations, it was not possible to obtain a satisfying docking model of tRNAGlu on YadB. Functional similarities and differences with GluRSs The structural similarity between YadB and the catalytic domain of E. coli GluRS indicates a common evolutionary origin of the two proteins and suggests functional mimicry with GluRS. Therefore, we tested the capability of the enzyme to activate glutamate. Experiments displayed in Figure 3a and b and in Table 2 unambiguously demonstrate that YadB promotes the glutamate-dependent ATP-PPi exchange and thus activates glutamate. However, in strong contrast with what occurs with GluRS that requires the tRNAGlu to activate glutamate, YadB catalyzes this reaction in the absence of tRNA.

Table 2. Kinetic constants of ATP-[32P]PPi exchange catalyzed by E. coli YadB and GluRS kcat (s21)

Km (mM) Glutamate Enzyme YadB GluRS

ATP

2tRNA

þtRNA

2tRNA

þ tRNA

2tRNA

þtRNA

6^2 No reaction

6^2 0.030 ^ 0.01

0.50 ^ 0.1 No reaction

0.50 ^ 0.1 0.060 ^ 0.02

14 ^ 0.2 No reaction

10 ^ 0.2 10 ^ 0.2

The kinetic constants were determined as described in Materials and Methods. The values are the mean values of at least three independent determinations.

278

Figure 3. Amino acid activation and tRNAGlu aminoacylation catalyzed by YadB and GluRS from E. coli. a and b, Amino acid activation catalyzed by YadB (a) and GluRS (b) as measured by ATP-[32P]PPi exchange. Reactions were conducted in the absence (B) or the presence (A) of crude E. coli tRNA and 0.4 mM YadB (a) or 0.05 mM GluRS (b). The [32P]ATP formed was determined from 30 ml aliquots. c, Aminoacylation of pure E. coli tRNAGlu by YadB (V) and GluRS (V). The [14C]glutamyltRNA formed was determined from 20 ml aliquots.

YadB activates the amino acid with a rate constant similar to that of the canonical E. coli GluRS in the presence of tRNA ðkcat ¼ 14 and 10 s21 Þ: However, the Km of YadB for glutamate in ATPPPi exchange is significantly higher than that of GluRS (6 mM and 0.03 mM, Table 2). This increased Km does not reflect an instability of glutamyl-adenylate formed by YadB as a consequence of the enzyme-promoted hydrolyis of the intermediate or its release from its binding site, since the Km for ATP, the second ligand involved in glutamyl-adenylate formation, is much lower ðKm ¼ 0:5 mMÞ: As a consequence YadB binds glutamate with a much lower affinity than GluRS. Similarly, the Km of YadB for ATP is one order of magnitude

Structure and Activity of the E. coli YadB

higher than that of GluRS (0.5 mM and 0.06 mM). Thus, affinity of YadB for ATP is lower than that of GluRS. The low affinity for glutamate prompted us to investigate the specificity of YadB for its amino acid. We found that beside glutamate, no one of the other natural amino acids promotes the ATPPPi exchange. Glutamate activation is also stereospecific, since D -glutamate does not promote the reaction and does not inhibit activation of L -glutamate. Surprisingly, the ATP-PPi exchange accounting for glutamate activation into glutamyl-adenylate differs from that catalyzed by canonical GluRSs, in particular E. coli GluRS, that requires tRNAGlu.26,27 Here, it is clearly shown that glutamate activation by YadB is tRNA-independent (Figure 3a) and further that tRNAGlu does not significantly alter the catalytic constants of the ATP-PPi exchange, in particular the Km for glutamate and ATP (Table 2). This exchange reaction catalyzed by YadB resembles that promoted by GluRS, since both are inhibited by glutamol-AMP (Table 3). The non-hydrolyzable analogue of glutamyl-adenylate inhibits competitively the reaction catalyzed by both enzymes with respect to both glutamate and ATP. Interestingly however, the Ki of the analogue for YadB is about one order of magnitude lower than for GluRS (Table 3), reflecting a higher affinity of YadB than GluRS for GoA. The independence of glutamate activation by YadB from tRNA presence can be easily rationalized by comparing YadB structure with those of the GluRS in complex with ATP or ATP/tRNAGlu. In GluRS Trp209 plays a key role in the absence of tRNA by preventing the ATP molecule from moving deeper in the pocket and establishing a hydrogenbond with Ala206 NH. Since the less bulky Leu194 is the corresponding residue in YadB, the activation movement of ATP is possible even in the absence of tRNA. Indeed, a few less interactions of ATP with YadB are observed in our model, reflecting the lower affinity of ATP for YadB as compared to GluRS. In contrast, glutamate or GoA fit with the same extent of positive interactions in YadB as in GluRS and the experimental data indicate similar affinities. Because of the structural resemblance of YadB with GluRS, a first possibility for a function of this protein would be that it charges tRNAGlu. Assays

Table 3. Inhibition of ATP-[32P]PPi exchange catalyzed by E. coli YadB and GluRS by glutamol-AMP Ki (mM) of glutamol-AMP for Substrate

YadB

GluRS

Glutamate ATP

0.44 ^ 0.05 0.20 ^ 0.04

2.0 ^ 0.2 1.0 ^ 0.1

In the presence of the concentrations of substrates and analogue tested, the inhibition by GoA was competitive. The Ki were determined from Lineweaver– Burk plots.

Structure and Activity of the E. coli YadB

displayed in Figure 3c show that tRNAGlu is not aminoacylated by YadB, even at a very low plateau, under the conditions where it is fully charged by E. coli GluRS. This inability of YadB to transfer the activated glutamate on tRNAGlu agrees with the absence of a strong effect of tRNA on the catalytic constants of the ATP-PPi exchange (see Table 2). This is not surprising, taking into account the absence of the anticodon-binding domain in YadB, and that the modified nucleotide U34 from tRNAGlu anticodon constitutes a strong identity element for recognition by GluRS.28,29 From another perspective, since some eubacterial and all archaeal GluRSs are able to recognize tRNAGln in addition to tRNAGlu,30 – 32 a possible charging of tRNAGln by YadB could be hypothesized. As for tRNAGlu, the assays for charging tRNAGln were negative (result not shown). Another possible role of the activated glutamate would be that YadB is an enzyme from the glutamine biosynthetic pathway that transforms glutamate into glutamine by a mechanism involving activation of glutamate by ATP prior to conversion by amidation into glutamine. This possibility is supported by the structural properties of E. coli asparagine synthase A mimicking the catalytic domain of AspRS33 and the discovery in the archaeal kingdom of an asparagine synthase exhibiting a high level of sequence identity with the catalytic domain of AspRSs but deprived of anticodon-binding domain.34 Because of the structural and functional resemblance between AspRSs and asparagine synthases, by analogy, YadB could be a glutamine synthase restricted to the catalytic domain of GluRS that would be able to activate the g-carboxyl group of glutamate prior to amidation with an ammonia group to form glutamine. The first evidence that the glutamyl-adenylate formed by YadB is not an intermediate involved in glutamine formation came from the inability of ammonia group donors (NH4Cl, asparagine and glutamine) to inhibit the ATP-PPi exchange, showing that these compounds are not competitors of PPi for the reaction with glutamyl-adenylate (result not shown). The inability of YadB to amidate the activated glutamate into glutamine was explicitly demonstrated by the absence of formation of glutamine under the experimental conditions using as amido group donors NH4Cl and asparagine that allowed amidation of free glutamate and tRNAdependent amidation of aspartate and glutamate (Figure 4).35 It was a great surprise when glutamylation assays of unfractionated E. coli tRNA with YadB yielded significant charging plateaus corresponding to , 4% of the crude tRNA despite the absence of charging of tRNAGlu and tRNAGln (Figure 5). When assays were conducted in the presence of both pure YadB and GluRS, the plateaus were , twofold higher than those measured in the presence of the sole YadB protein (Figure 5). This confirms the presence of tRNAGlu in crude tRNA and suggests that this tRNA does not inhibit the

279

Figure 4. Analysis of glutamine synthase activity of YadB. The assays were conducted as described in Materials and Methods with E. coli YadB or Gln synthase and either NH4Cl (lines 1 and 4) or Asn (lines 2 and 5) or without amido group donors (lines 3 and 6). The products were analyzed by TLC on cellulose plates as described in Materials and Methods.

glutamylation reaction catalyzed by YadB, a hypothesis explicitly demonstrated by the absence of inhibition by pure tRNAGlu of this reaction. Altogether, this shows that YadB is an atypical GluRS that does not recognize tRNAGlu, but is able to recognize and charge another tRNA present within crude E. coli tRNA. Glutamylation by YadB of this unknown tRNA (or of several tRNAs) occurs with a catalytic efficiency (kcat/Km) , 1000fold reduced compared to that of tRNAGlu charging by cognate GluRS (kcat ¼ 0:01 s21 versus 5.5 s21 for GluRS, apparent Km for tRNA ¼ 1.3 mM versus 1.0 mM for GluRS and apparent Km for glutamate ¼ 300 mM versus 140 mM for GluRS, Table 4). This reduced efficiency is kcat -dependent, a feature resembling what is often observed when aaRSs charge non-cognate tRNAs.36 The reaction requires a glutamyl-adenylate intermediate, as supported by the observation that the YadB glutamylation activity is inhibited by the adenylate analog GoA even more efficiently than that of canonical GluRS (competitive inhibition with Ki ¼ 3 mM with respect to glutamate and ATP; result not shown).

Figure 5. Glutamylation of unfractionated E. coli tRNA with YadB and GluRS. The reactions were conducted with crude E. coli tRNA (1 mg ml21) and initiated either by a mixture of 0.4 mM GluRS and 3 mM YadB (X) or by 1.5 mM YadB alone (W). In the latter case, assays were supplemented after 30 minutes by 0.4 mM GluRS (arrow). The [14C]glutamyl-tRNA formed was determined from 20 ml aliquots.

280

Structure and Activity of the E. coli YadB

Table 4. Kinetic constants of glutamylation of E. coli unfractionated tRNA by YadB and GluRS Kinetic constant kcat =Km (mM s21)

Km (mM) Enzyme YadB GluRS

21

tRNA

ATP

Glutamate

kcat (s )

tRNA

Glutamate

1.3 ^ 0.2 1.0 ^ 0.2

80 ^ 20 n.d.

300 ^ 100 140 ^ 40

0.01 ^ 0.005 5.5 ^ 1.0

8.4 ^ 5.1 £ 1023 5.9 ^ 2.2

4.4 ^ 3.1 £ 1025 4.5 ^ 2 £ 1022

The conditions are described in Materials and Methods; n.d. not determined.

Conclusion and Perspectives Both the sequence and catalytic data show that the YadB protein constitutes a minimalist GluRS structure restricted to the catalytic domain of canonical GluRS. The functional analysis shows further that this protein has enzymatic activity and like GluRS does catalyze activation of the cognate amino acid, the first step in a tRNA aminoacylation reaction. In contrast to canonical aaRSs, however, YadB does not glutamylate the cognate tRNAGlu. By these features, YadB resembles the catalytic domains of AlaRS and GlnRS generated by genetic engineering methods that keep their amino acid activation function but lose totally37 or almost totally38 their tRNA charging capacity. The high degree of similarity between the glutamate and ATP binding sites of E. coli YadB and T. thermophilus GluRS, and the strong inhibition by the GluRS-specific inhibitor GoA, suggest a priori that both enzymes activate glutamate via similar catalytic mechanisms. The two mechanisms, however, are not identical, since YadB exhibits an affinity for glutamate two orders of magnitude lower than GluRS and in contrast to GluRS it activates glutamate in the absence of tRNA. Therefore, YadB consists of an altered version of the catalytic domain of GluRS that has conserved the capacity to form glutamyl-adenylate, but in the absence of tRNA, and has lost or not acquired the capacity to glutamylate tRNAGlu. The role of the activated glutamate and the cellular process in which it takes part remain unknown. As expected by the absence of the tRNA anticodonbinding domain, YadB does charge neither tRNAGlu nor tRNAGln, indicating that the enzyme is functionally not related to GluRSs of strict or relaxed specificity. However, taking into account the high degree of similarities between the glutamate and ATP-binding sites of the E. coli mini-GluRS and T. thermophilus GluRS, and the strong inhibition by the GluRS-specific inhibitor GoA, suggest that both enzymes share a similar catalytic mechanism, despite the unexpected fact that adenylate formation by YadB is tRNA-independent. From another viewpoint, taking into account the present view on tRNA aminoacylation that is governed by identity rules,12 one can conjecture about the identity of the tRNA glutamylated by YadB. Considering the small size of YadB that is unable to make

contacts with a tRNA anticodon branch, suggests that the glutamylated tRNA acts as a minihelix39 with identity determinants recognized by YadB present in its amino acid acceptor stem. In conventional glutamate systems this is indeed the case, since GluRS in addition to determinants located in the anticodon loop, recognizes the three first basepairs of the tRNAGlu acceptor stem.14,40 In YadB, however, sequence features present in the acceptor stem of tRNAGlu must act as antideterminants,12 thereby preventing binding of this tRNA. Related to this view, notice that GluRS from Bacillus stearothermophilus, another close relative of YadB, is able to glutamylate several non-cognate tRNAs.41 Among the tRNA species that are the most easily glutamylated one finds E. coli tRNAGlu and noncognate initiator tRNAMetf and tRNALeu2 as well as yeast initiator tRNAMet and tRNAAsp. It is likely that the E. coli tRNA species recognized by YadB share(s) sequence resemblance with the amino acid acceptor stem of the above-listed non-cognate tRNAs. Biochemical investigations combined with RNA sequencing are underway to identify the (or these) tRNA(s) actually recognized and charged by E. coli YadB (D.Y.D. et al., unpublished results). Concluding, it is worth recalling that aaRSs are multidomain proteins of ancient origin that likely derive from minimalist structures restricted to catalytic cores that could aminoacylate RNA molecules.39 In the course of evolution additional domains were appended to or removed from such cores. As a result of such tinkering, it is not surprising that structural modules present in contemporary aaRSs are found elsewhere in life, either as part of other proteins or as individual proteins having acquired novel functions.6,34,42 YadB, the mini-GluRS from E. coli, is another example of such a protein. The fact that YadB homologues are widespread in bacterial genomes, suggests a conserved biological function of this family of proteins.

Materials and Methods Proteins and tRNA For production of YadB, its ORF was subcloned in the Gateway system (Invitrogen, pDEST17 plasmid) introducing 15 amino acid residues plus six histidine residues

281

Structure and Activity of the E. coli YadB

at the N terminus.3 Expression was performed using the BL21(DE3) E. coli strain as described.2 Purification was ¨ kta FPLC system using performed with a Pharmacia A an IMAC Ni2þ-column, followed by a preparative gel-filtration on Superdex 200. The protein was characterized by four methods: (i) SDS-PAGE, which showed a unique band around 35 kDa; (ii) mass spectroscopy using a MALDI-TOF spectrometer that gave a peak at 36,142 Da (theoretical mass 36,128.8 Da); (iii) circular dichroism (JASCO) that indicated a properly folded protein with a distribution of 29% helices and of 21% b-strands; and (iv) dynamic light-scattering (Dynapro, Protein Solutions), which showed a unique peak centered around 30 kDa, a value compatible with a monomer in solution. GluRS and Gln synthase from E. coli were purified as described;35,43 E. coli tRNAGlu (accepting capacity of glutamate, 36 nmol mg21) was from Subriden and Boehringer and tRNAGln (accepting capacity of glutamine, 35 nmol mg21) from Subriden. X-ray structure determination and refinement Crystals were obtained at 20 8C using vapor-diffusion with sitting-drops by mixing 1 ml of protein at 5 mg ml21 with 1 ml of a precipitant solution containing 13% (w/v) PEG 4000, 2% (w/v) MPD, 12 mM b-mercaptoethanol and 50 mM sodium cacodylate (pH 6.2). Crystals appeared after a few days as thin needles.2 Crystals of YadB were cryocooled in their mother liquor plus 25% (v/v) glycerol and a complete data set was collected ˚ resolution at ID14-EH1 (ESRF, Grenoble) using at 1.5 A an ADSC Quantum 4 detector. Crystals are monoclinic, belong to space group C2 and contain one molecule per ˚ 3/Da). All data were proasymmetric unit (VM ¼ 1:96 A cessed and reduced using DENZO44 and the CCP4 program suite.45 Refinement was performed successively with CNS 1.046 and REFMAC 5.047 using bulk solvent correction followed by manual refitting using Turbo-Frodo.48 The final Rfree and R-values are 16.9% and 14.6%, respectively. Protein geometry was assessed using PROCHECK49 showing 93.6% residues in the most favorable region and 6.4% in the additionally allowed regions. Statistics of data collection and refinement are presented in Table 1. Activity assays The ATP-[32P]PPi exchange reaction mixtures (300 ml) contained 100 mM Na –Hepes (pH 7.2), 10 mM MgCl2, 2 mM ATP, 2 mM [32P]PPi (1 – 2 cpm pmol21) and 25 mM L -glutamate (or 10 mM D -glutamate or another L -amino acid) when YadB was tested or 1 mM when GluRS was tested, 1 mg ml21 unfractionated E. coli tRNA when present, and appropriate concentrations of YadB or GluRS. For Km measurements the concentrations were 0.2 – 4 mM ATP and 3 – 20 mM L -glutamate when YadB was tested and 0.03– 1 mM ATP and 0.01 – 0.5 mM L -glutamate when GluRS was tested. Inhibition assays of the ATP-PPi exchange reaction by glutamol-AMP (GoA), were conducted with variable concentrations of ATP or glutamate in the ranges described above, the second substrate, respectively glutamate or ATP being saturating, and concentrations of GoA fixed between 0.3 mM and 3 mM. The [32P]ATP formed after various incubation times at 37 8C was determined as described.50 The GoA that is a competitive inhibitor of GluRS was synthesized as described.51

For tRNAGlu and tRNAGln aminoacylations, reaction mixtures (150 ml) contained 100 mM Na– Hepes (pH 7.2), 2 mM ATP, 10 mM MgCl2, 30 mM KCl, 25 mM L [14C]glutamate (330 cpm pmol21) and 2 mM pure E. coli tRNAGlu or tRNAGln and appropriate concentrations of YadB or GluRS for initial rate measurements. Both tRNA were fully charged by their homologous aaRS. For crude tRNA aminoacylation, the complete reaction mixtures (200 ml) contained 0.5– 1 mg ml21 of unfractionated E. coli tRNA (glutamate accepting capacity by YadB was 4%) and appropriate concentrations of YadB or GluRS for initial rate or plateau measurements. For Km measurements the concentrations were 0.1 – 2 mM for ATP, 30 –500 mM for glutamate and 0.3– 6 mM for tRNA. The 14C-labeled aminoacyl-tRNA formed at 37 8C was determined as described.50 Glutamine synthase activity was analyzed in a reaction mixture (50 ml) containing 100 mM Na – Hepes (pH 7.2), 10 mM ATP, 12 mM MgCl2, 30 mM KCl, 10 mM 14 21 L -[ C]glutamate (330 cpm pmol ), 2 mM NH4Cl or Asn and 0.04 mg YadB or 0.006 mg E. coli Gln synthase. After three hours reaction at 37 8C the labeled products were separated by TLC of 1 ml aliquots on a cellulose plate with the solvent isopropanol/formic acid/water (80:20:4, by vol.) and revealed after 12 hours migration by image plate with a Fuji Bioimager. Atomic coordinates The coordinates have been deposited in the Protein Data Bank at RCSB† as entry 1NZJ.

Acknowledgements The ESRF is greatly acknowledged for beam time allocation. We thank the structural genomics team of the Marseille laboratory for technical assistance. Jean-Michel Claverie and the IGS laboratory (Marseille) are gratefully acknowledged for their bioinformatics analysis. This study was supported by the French Ministry of Industry (grant ASG) and is a collaboration with the IGS laboratory and the Aventis company. Additional support came by grant OG0009597 from the Natural Sciences and Engineering Research Council of Canada (NSERC) and by the French Centre National de la Recherche Scientifique (CNRS), the Association de la Recherche contre le Cancer (ARC) and Universite´ Louis Pasteur (Strasbourg).

References 1. Claverie, J.-M., Monchois, V., Audic, S., Poirot, O. & Abergel, C. (2002). In search of new anti-bacterial target genes: a comparative/structural genomics approach. Comb. Chem. High Throughput Screen, 5, 511 – 522. 2. Vincentelli, R., Bignon, C., Gruez, A., Sulzenbacher, G., Canaan, S., Tegoni, M. et al. (2003). Mediumscale structural genomics: strategies for protein † http://www.rcsb.org/pdb/

282

3.

4.

5. 6.

7.

8.

9.

10.

11. 12. 13.

14.

15. 16.

17.

18.

expression and crystallization. Accts. Chem. Res. 36, 165– 172. Sulzenbacher, G., Gruez, A., Roig-Zamboni, V., Spinelli, S., Valencia, C., Pagot, F. et al. (2002). A medium-throughput crystallization approach. Acta. Crystallog. sect. D, 58, 2109– 2115. Eriani, G., Delarue, M., Poch, O., Gangloff, J. & Moras, D. (1990). Partition of tRNA synthetases into two classes based on mutually exclusive sets of sequence motifs. Nature, 347, 203– 206. Ibba, M. & So¨ll, D. (2000). Aminoacyl-tRNA synthesis. Annu. Rev. Biochem. 69, 617– 650. Francklyn, C., Perona, J. J., Puetz, J. & Hou, Y.-M. (2002). Aminoacyl-tRNA synthetases: versatile players in the changing theater of translation. RNA, 8, 1363– 1372. Lamour, V., Quevillon, S., Diriong, S., N’Guyen, V. C., Lipinski, M. & Mirande, M. (1994). Evolution of the Glx-tRNA synthetase family: the glutaminyl enzyme as a case of horizontal gene transfer. Proc. Natl Acad. Sci. USA, 91, 8670– 8674. Breton, R., Sanfac¸on, H., Papayannopoulos, I., Biemann, K. & Lapointe, J. (1986). Glutamyl-tRNA synthetase of Escherichia coli. Isolation and primary structure of the gltX gene and homology with other aminoacyl-tRNA synthetases. J. Biol. Chem. 261, 10610– 10617. 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. Schimmel, P. (1987). Aminoacyl tRNA synthetases: general scheme of structure– function relationships in the polypeptides and recognition of transfer RNAs. Annu. Rev. Biochem. 56, 125– 158. Martinis, S. A., Plateau, P., Cavarelli, J. & Florentz, C. (1999). Aminoacyl-tRNA synthetases: a new image for a classical family. Biochimie, 81, 683– 700. Giege´, R., Sissler, M. & Florentz, C. (1998). Universal rules and idiosyncratic features in tRNA identity. Nucl. Acids Res. 26, 5017–5035. Tjaden, B., Saxena, R. M., Stolyar, S., Haynor, D. R., Kolker, E. & Rosenow, C. (2002). Transcriptome analysis of Escherichia coli using high-density oligonucleotide probe arrays. Nucl. Acids Res. 30, 3732– 3738. Nureki, O., Vassylyev, D. G., Katayanagi, K., Shimizu, T., Sekine, S.-i., Kigawa, T. et al. (1995). Architectures of class-defining and specific domains of glutamyl-tRNA synthetase. Science, 267, 1958– 1965. 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. Liu, J., Lin, S.-X., Blochet, J.-E., Pe´zolet, 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. Nureki, O., Kohno, T., Sakamoto, K., Miyazawa, T. & Yokoyama, S. (1993). Chemical modification and mutagenesis studies on zinc binding of aminoacyltRNA synthetases. J. Biol. Chem. 268, 15368– 15373. Liu, J., Gagnon, Y., Gauthier, J., Furenlid, L., L’Heureux, P.-H., Auger, M. et al. (1995). The zincbinding site of Escherichia coli glutamyl-tRNA synthetase is located in the acceptor-binding domain. Studies by extended X-ray absorption fine structure,

Structure and Activity of the E. coli YadB

19.

20.

21. 22.

23.

24.

25.

26.

27.

28.

29.

30.

31. 32.

33.

34.

molecular modeling, and site-directed mutagenesis. J. Biol. Chem. 270, 15162 –15169. Rould, M. A., Perona, J. J., So¨ll, D. & Steitz, T. A. (1989). Structure of E. coli glutaminyl-tRNA synthe˚ tase complexed with tRNA(Gln) and ATP at 2.8 A resolution. Science, 246, 1135– 1142. Mechulam, Y., Schmitt, E., Maveyraud, L., Zelwer, C., Nureki, O., Yokoyama, S. et al. (1999). Crystal structure of Escherichia coli methionyl-tRNA synthetase highlights species-specific features. J. Mol. Biol. 294, 1287– 1297. Ibba, M., Franklyn, C. & Cusack, S. (2004). Editors of The Aminoacyl-tRNA Synthetases, Landes Bioscience, Georgetown, TX. In the press. Newberry, K. J., Hou, Y. M. & Perona, J. J. (2002). Structural origins of amino acid selection without editing by cysteinyl-tRNA synthetase. EMBO J. 21, 2778– 2787. Sankaranarayanan, R., Dock-Bregeon, A.-C., Rees, B., Bovee, M., Caillet, J., Romby, P. et al. (2000). Zinc ion mediated amino acid discrimination by threonyltRNA synthetase. Nature Struct. Biol. 7, 461– 465. Sekine, S., Nureki, O., Shimada, A., Vassylyev, D. G. & Yokoyama, S. (2001). Structural basis for anticodon recognition by discriminating glutamyl-tRNA synthetase. Nature Struct. Biol. 8, 203 –206. Sekine, S., Nureki, O., Dubois, D. Y., Bernier, S., Cheˆnevert, R., Lapointe, J. et al. (2003). ATP binding by glutamyl-tRNA synthetase is switched to the productive mode by tRNA binding. EMBO J. 22, 676– 688. Kern, D. & Lapointe, J. (1980). The catalytic mechanism of glutamyl-tRNA synthetase of Escherichia coli. Evidence for a two-step aminoacylation pathway and study of the reactivity of the intermediate complex. Eur. J. Biochem. 106, 137– 150. Kern, D. & Lapointe, J. (1980). The catalytic mechanism of the glutamyl-tRNA synthetase from Escherichia coli. Detection of an intermediate complex in which glutamate is activated. J. Biol. Chem. 255, 1956– 1961. Normanly, J., Kleina, L. G., Masson, J. M., Abelson, J. & Miller, J. H. (1990). Construction of Escherichia coli amber suppressor tRNA genes. III. Determination of tRNA specificity. J. Mol. Biol. 213, 719– 726. Sylvers, L. A., Rogers, K. C., Shimizu, M., Ohtsuka, E. & So¨ll, D. (1993). A 2-thiouridine derivative in tRNAGlu is a positive determinant for aminoacylation by Escherichia coli glutamyl-tRNA synthetase. Biochemistry, 32, 3836– 3841. Gagnon, Y., Lacoste, L., Champagne, N. & Lapointe, J. (1996). Widespread use of the Glu-tRNAGln transamidation pathway among bacteria. A member of the alpha purple bacteria lacks glutaminyl-tRNA synthetase. J. Biol. Chem. 271, 14856– 14863. Ibba, M., Curnow, A. W. & So¨ll, D. (1997). Aminoacyl-tRNA synthesis: divergent routes to a common goal. Trends Biochem. Sci. 22, 39 – 42. Kern, D., Roy, H. & Becker, H. D. (2004). The Aminoacyl-tRNA Synthetases (Ibba, M., Francklyn, C. & Cusack, S., eds), Landes Bioscience, Georgetown, TX. In the press. Nakatsu, T., Kato, H. & Oda, J. (1998). Crystal structure of asparagine synthetase reveals a close evolutionary relationship to class II aminoacyl-tRNA synthetase. Nature Struct. Biol. 5, 15 – 19. Roy, H., Becker, H. D., Reinbolt, J. & Kern, D. (2003). When contemporary aminoacyl-tRNA synthetases

283

Structure and Activity of the E. coli YadB

35.

36.

37.

38. 39.

40.

41.

42. 43.

invent their cognate amino acid metabolism. Proc. Natl Acad. Sci. USA, 100, 9837– 9842. Becker, H. D. & Kern, D. (1998). Thermus thermophilus: a link in evolution of the tRNA-dependent amino acid amidation pathway. Proc. Natl Acad. Sci. USA, 95, 12832– 12837. Ebel, J.-P., Giege´, R., Bonnet, J., Kern, D., Befort, N., Bollack, C. et al. (1973). Factors determining the specificity of the tRNA aminoacylation reaction. Non-absolute specificity of tRNA-aminoacyl-tRNA synthetase recognition and particular importance of the maximal velocity. Biochimie, 55, 547– 557. Buechter, D. D. & Schimmel, P. (1993). Dissection of a class II tRNA synthetase: determinants for minihelix recognition are tightly associated with domain for amino acid activation. Biochemistry, 32, 5267– 5272. Schwob, E. & So¨ll, D. (1993). Selection of a “minimal” glutaminyl-tRNA synthetase and the evolution of class I synthetases. EMBO J. 12, 5201– 5208. 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. Sekine, S., Nureki, O., Sakamoto, K., Niimi, T., Tateno, M., Go, M. et al. (1996). Major identity determinants in the “augmented D helix” of tRNA(Glu) from Escherichia coli. J. Mol. Biol. 256, 685– 700. Grosjean, H., Charlier, J., Darte, C., Dirheimer, G., Giege´, R., de Henau, S. et al. (1976). Purification, characterization and mechanism of action of several aminoacyl-tRNA synthetases from Bacillus stearothermophilus. Experientia Suppl. 26, 347– 362. Schimmel, P. & Ribas de Pouplana, L. (2000). Footprints of aminoacyl-tRNA synthetases are everywhere. Trends Biochem. Sci. 25, 207– 209. Kern, D., Potier, S., Boulanger, Y. & Lapointe, J. (1979). The monomeric glutamyl-tRNA synthetase of Escherichia coli. Purification and relation between

44.

45. 46.

47.

48.

49.

50.

51.

52.

its structural and catalytic properties. J. Biol. Chem. 254, 518– 524. Otwinowski, Z. & Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode: macromolecular crystallography, Part A. Methods Enzymol. 276, 307– 326. Collaborative Computing Project Number 4 (1994). The CCP4 suite: programs for protein crystallography. Acta. Crystallog. sect. D, 50, 760–766. Bru¨nger, A. T., Adams, P. D., Clore, G. M., DeLano, W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998). Crystallography & NMR system: a new software suite for macromolecular structure determination. Acta Crystallog. sect. D, 54, 905–921. Murshudov, G. N., Vagin, A. A. & Dodson, E. J. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallog. sect. D, 53, 240– 255. Roussel, A. & Cambillau, C. (1991). The TURBOFRODO Graphics Package pp. 77 – 78. Silicon Graphics Geometry Partners Directory, vol. 81, Silicon Graphics, Mountain View, USA. Laskowski, R. A., MacArthur, M. W., Moss, D. S. & Thornton, J. M. (1993). PROCHECK: a program to check the stereochemical quality of protein structures. J. Appl. Crystallog. 26, 283– 291. Kern, D. & Lapointe, J. (1979). The twenty aminoacyl-tRNA synthetases from Escherichia coli. General separation procedure, and comparison of the influence of pH and divalent cations on their catalytic activities. Biochimie, 61, 1257– 1271. Desjardins, M., Garneau, S., Desgagne´s, J., Lacoste, L., Yang, F., Lapointe, J. & Cheˆnevert, R. (1998). Glutamyl adenylate analogues are inhibitors of glutamyl-tRNA synthetase. Bioorg. Chem. 26, 1 – 13. Kraulis, P. J. (1991). MOLSCRIPT: a program to produce both detailed and schematic plots of protein structure. J. Appl. Crystallog. 24, 946– 950.

Edited by R. Huber (Received 26 September 2003; received in revised form 8 January 2004; accepted 8 January 2004)