Crystal Structure of Glutamyl-Queuosine tRNAAsp Synthetase Complexed with l-Glutamate: Structural...

doi:10.1016/j.jmb.2008.06.053 J. Mol. Biol. (2008) 381, 1224–1237

Available online at www.sciencedirect.com

Crystal Structure of Glutamyl-Queuosine tRNAAsp

Synthetase Complexed with L-Glutamate: StructuralElements Mediating tRNA-Independent Activation ofGlutamate and Glutamylation of tRNAAsp Anticodon

Mickaël Blaise1, Vincent Olieric1, Claude Sauter1, Bernard Lorber1,Bappaditya Roy2, Subir Karmakar2, Rajat Banerjee2,Hubert Dominique Becker1 and Daniel Kern1⁎
1UPR 9002 ‘Architecture etRéactivité de l'ARN,’ Institut deBiologie Moléculaire etCellulaire du Centre National dela Recherche Scientifique,15 Rue René Descartes,and Université Louis Pasteur,F-67084 Strasbourg Cedéx,France2Dr. B.C. Guha Centre forGenetic Engineering andBiotechnology, University ofCalcutta, Ballygunge CircularRoad, Calcutta 700 019,West Bengal, India
Received 14 April 2008;received in revised form13 June 2008;accepted 19 June 2008Available online26 June 2008

*Corresponding author. E-mail [email protected] used: Glu-Q-RS, gl

tRNAAsp synthetase; GluRS, glutamtRNA, transfer RNA; aaRS, aminoacaa, amino acid; 3D, three-dimensionpeptide; SC, stem contact; PDB, Pro

0022-2836/$ - see front matter © 2008 E

Glutamyl-queuosine tRNAAsp synthetase (Glu-Q-RS) from Escherichia coli isa paralog of the catalytic core of glutamyl-tRNA synthetase (GluRS) thatcatalyzes glutamylation of queuosine in the wobble position of tRNAAsp.Despite important structural similarities, Glu-Q-RS and GluRS divergestrongly by their functional properties. The only feature common to bothenzymes consists in the activation of Glu to formGlu-AMP, the intermediateof transfer RNA (tRNA) aminoacylation. However, both enzymes differ bythe mechanism of selection of the cognate amino acid and by the mechanismof its activation. Whereas GluRS selects L-Glu and activates it only in thepresence of the cognate tRNAGlu, Glu-Q-RS forms Glu-AMP in the absenceof tRNA. Moreover, while GluRS transfers the activated Glu to the 3′accepting end of the cognate tRNAGlu, Glu-Q-RS transfers the activated Gluto Q34 located in the anticodon loop of the noncognate tRNAAsp. In order togain insight into the structural elements leading to distinct mechanisms ofamino acid activation, we solved the three-dimensional structure of Glu-Q-RS complexed to Glu and compared it to the structure of the GluRS·Glucomplex. Comparison of the catalytic site of Glu-Q-RS with that of GluRS,combined with binding experiments of amino acids, shows that a restrictednumber of residues determine distinct catalytic properties of amino acidrecognition and activation by the two enzymes. Furthermore, to explore thestructural basis of the distinct aminoacylation properties of the twoenzymes and to understand why Glu-Q-RS glutamylates only tRNAAsp

among the tRNAs possessing queuosine in position 34, we performed atRNA mutational analysis to search for the elements of tRNAAsp thatdetermine recognition by Glu-Q-RS. The analyses made on tRNAAsp andtRNAAsn show that the presence of a C in position 38 is crucial forglutamylation of Q34. The results are discussed in the context of theevolution and adaptation of the tRNA glutamylation system.

© 2008 Elsevier Ltd. All rights reserved.

Keywords: glutamyl-queuosine tRNAAsp synthetase; glutamyl-tRNAsynthetase; tRNA aminoacylation; queuosine; tRNA-dependent aminoacid activation
Edited by J. Doudna
ess:

utamyl-queuosineyl-tRNA synthetase;yl-tRNA synthetase;al; CP, connectivetein Data Bank.

lsevier Ltd. All rights reserve

Introduction

Aminoacyl-tRNA synthetase (aaRS) constitute aubiquitous and essential family of enzymes thatcatalyze the attachment of each genetically encodedamino acid (aa) onto the 3′ end of homologoustransfer RNAs (tRNAs) to form various aa-tRNAsrequired for protein synthesis. These enzymes have

d.

http://dx.doi.org/10.1016/j.jmb.2008.06.053

1225Glutamyl-Queuosine tRNA Asp Synthetase

been grouped into two distinct families, classes I andII, differing by structural and functional propertiesthat probably evolved from distinct ancestors.1 Theydisplay two main functional modules: (i) the catalyticcore catalyzing the activation of amino acid with ATPto form aminoacyl AMP, followed by the transfer ofthe amino acid onto the 3′ accepting end of tRNA, and(ii) the anticodon-binding domain involved in therecognition of tRNA anticodon.2,3 Additional mod-ules have been recruited by these enzymes through-out evolution to increase their catalytic efficiency andspecificity, such as the CP1 module that promotespretransfer and posttransfer proofreading,4–6 or theEMAPII and YqeY domains that increase theaffinity of the aaRS for the cognate tRNA.7,8 MostaaRS, however, contain appended domains that arenot involved in tRNA aminoacylation whose func-tions remain undefined until now.9,10

It has been accepted that fidelity of translation isensured by the presence of 20 aaRS in a givenorganism—one particular aaRS to charge each of the20 natural amino acids in each of the 20 families ofisoaccepting tRNAs.11,12 However, analysis of thesequenced genomes revealed that there are feworganisms that possess a complete and single set of20 aaRS.13 Three types of widespread anomalies havebeen reported. First, a particular aaRS can be lackingin the organism. The homologous aa-tRNA is thenformed by a two-step pathway resembling that ofthe formation of Sec-tRNASec,14 in which one of theremaining aaRS charges the orphan tRNA beforethe tRNA-dependent conversion of the amino acidinto that homologous to the tRNA. This pathway isoften exclusively used by prokaryotes deprived ofGlnRS and AsnRS to form Gln-tRNAGln and Asn-tRNAAsn.15,16 Second, and more exceptionally, twoaaRS of identical specificity are encoded by twodistinct genes,13 leading to an apparent duplicatedtRNA charging activity. In some cases, this redun-dancy could be related to distinct functions of theduplicates, such as adaptation of the organism todifferent growth conditions or to distinct regulation ofthe activity of each enzyme, or sometimes to distinctpathways able to form the cognate aa-tRNA.17–19

However, most of the aaRS duplications are still notunderstood. Third, various prokaryotes contain openreading frames encoding paralogs of autonomousdomains present in aaRS expressed in free form.Functional analysis has shown that these proteinsusually exert functions not related to canonical aa-tRNA formation, suggesting that they evolved froman aaRS whose function has been deviated to a novelactivity. This is exemplified by the archaeal paralog ofthe catalytic core of AsnRS shown to be, in fact, anarchaeal Asn synthetase that amidates free Asp intoAsn.20 Likewise, the paralog of the catalytic core ofHisRS was also shown to be involved in the synthesisof its cognate amino acid.21 More recently, it has beenreported that prokaryotes display freestanding ortho-logs of the aaRS-editing domains that hydrolyzemischarged aa-tRNAs released by aaRS.22,23We recently discovered an open reading frame,

present inmost proteobacteria, that encodes a protein,

YadB, resembling the catalytic core of glutamyl-tRNAsynthetase (GluRS) but deprived of its anticodon-binding domain.24–26 Structural and functional inves-tigations demonstrated that, like GluRS, this paralogactivates Glu to form Glu-AMP and catalyzes thetransfer of Glu onto tRNA. However, whereas GluRStransfers the activated Glu onto the 3′ terminaladenosine of the cognate tRNAGlu, YadB transfersGlu onto queuosine 34 of tRNAAsp anticodon.26–28

Thus, this enzyme, catalyzing the hypermodificationof tRNAAsp anticodon, has been renamed glutamyl-queuosine tRNAAsp synthetase (Glu-Q-RS).28 Inter-estingly, in contrast to GluRS, which requires thehomologous tRNAGlu as an obligate cofactor toactivate Glu before it catalyzes its glutamylation,YadB activates the amino acid substrate in the absenceof tRNA.24,25

The three-dimensional (3D) structure of Escherichiacoli Glu-Q-RS, solved at 1.5 Å resolution, revealed astrong similarity to the catalytic core of GluRS andconfirmed the lack of theanticodon-bindingdomain.24

Furthermore, alignments showed that most residuesof Thermus thermophilus GluRS involved in thebinding of Glu and ATP are conserved in Glu-Q-RS.25,28 However, these observations contrast withthe distinct functional properties of Glu activationand tRNA aminoacylation by the two enzymes. Thepresent work was designed to investigate thestructural basis of both Glu-Q-RS tRNA-independentGlu activation and tRNAAsp-Q34 glutamylation.To investigate the structural basis of the distinct

mechanisms of amino acid activation by Glu-Q-RSand GluRS, we crystallized and solved the 3Dstructure of Glu-Q-RS complexed to Glu and com-pared it to that of T. thermophilusGluRS bound toGlu.The comparison reveals strong similarities in thebinding of Glu by the two enzymes, but alsopeculiarities that may explain the distinct mechan-isms of Glu-AMP formation. Binding measurementsof cognate and noncognate amino acids by fluores-cence quenching reveal differentmodes of selection ofL-Glu byGlu-Q-RS andGluRS. The structural scaffoldthat triggers the binding of Glu and ATP to eachenzyme and that may promote selection and activa-tion of L-Glu byGlu-Q-RS in the absence of tRNA andby GluRS only when tRNAGlu is present will bediscussed. Furthermore, we investigated the tRNAelements that determine the glutamylation of onlytRNAAsp among the four E. coli tRNAs possessingQ34 (tRNAAsp, tRNAAsn, tRNATyr, and tRNAHis).Analysis of the glutamylation capacity of tRNAvariants expressed in vivo shows that nucleotide C38plays a prevalent role in the glutamylation of Q34 intRNAAsp by Glu-Q-RS.

Results and Discussion

Description of the overall structure of theGlu-Q-RS·Glu complex

Among the 298 residues of Glu-Q-RS, 292 are seenon the electron density map. As already reported, the

1226 Glutamyl-Queuosine tRNA Asp Synthetase

structure is highly homologous to that of the catalyticcore of GluRS from T. thermophilus.24 It comprises thetwo half parts of the Rossmann fold, the connectivepeptide (CP) domain, and part of the stem contact(SC) domain (Fig. 1a). As reported previously, it con-tains a zinc ion coordinated to Cys101, Cys103,His119, and Cys119, which are also found in someGluRS such as those ofE. coli and Bacillus subtilis29 butnot in GluRS from T. thermophilus.30 With theexception of residues 1–3 and 296–298 for which theelectron density is missing or not well defined for theside chain, the map is of good quality. The electrondensities of residues Pro224-Pro236 absent in thestructure of the apoenzyme24 are present in the Glu-Q-RS·Glu complex. The data allowed us to build thispart of the sequence that folds into a loop (Fig. 1). Itincludes sequence 228KLSKQ232 corresponding to theclass I KMSKS sequence signature, which comprisesresidues involved in ATP binding. The high B-factorvalues indicate that this loop is mobile, similar to T.thermophilus GluRS. The second class I sequencesignature, “HIGH” (19HFGS22 in Glu-Q-RS), whichin GluRS also contains residues involved in ATPbinding, is defined as well as in the apoenzyme. Theelectron density of the C-terminal part of the enzymeallowed the building of residues ranging from Val289to Cys298,which are absent in the structure of the freeenzyme. However, the high B-factors mean that thisregion is flexible (Fig. 1b). An additional electrondensity is well seen in the vicinity of Tyr172, a residueexposed in the putative catalytic site of the enzyme inwhich L-Glu could easily be fitted (Fig. 1c and d).

Glutamate-binding site

The Glu-binding pocket encompasses eight resi-dues: Arg9, Ala11, Ser13, Glu45, Tyr172, Arg190,Asp193, and Leu194 (Fig. 1c). These residues interactwith Glu through H-bonding either via directinteractions or via water molecules. The γ-carboxylgroup of Glu contacts three residues: the guanidi-nium groups of Arg9 and Arg190, and the OH groupof Tyr172. Arg9 contacts the Glu side chain via a saltbridge stabilized by Asp180, which establishes aninteraction maintaining the side chain of the Arg9residue. The γ-carboxyl group of Glu45, the OHgroup of Ser13, and the carbonyl group of Ala11contact the α-amino group of Glu. The interaction ofthe α-carboxyl group of Glu is mediated via anetwork involving sevenwater molecules, which arecontacted by the side chain of Asp193 and the mainchains of Ser13, Gly21, and Gly191. Finally, Leu194in the close vicinity of Glu completes the amino-acid-binding pocket.

Highly conserved glutamate-binding sites inGlu-Q-RS and GluRS

Superimposition of the 3D structures of T. thermo-philus GluRS and E. coli Glu-Q-RS confirms the highlevel of similarity of the catalytic sites of bothenzymes. Most residues of Glu-Q-RS contactingGlu are strictly conserved in GluRS; only one residue

differs between the two enzymes: Trp209 of GluRS isreplaced by Leu194 in Glu-Q-RS (Fig. 2a–c). Bindingof the side chain of Glu involves four residues ofT. thermophilus GluRS: Arg5, Tyr187, Asn191, andArg205. The Arg and Tyr residues are conserved inGlu-Q-RS, where they correspond to Arg9, Tyr172,and Arg190, but Asn191 is replaced by Val176.However, since Val can also be found in otherGluRS, the implication of that residue in Glu bindingshould be minimized. Residues Ala11, Ser13, andGlu45 contacting the α-amino group of Glu in Glu-Q-RS correspond to Ala7, Ser9, and Glu41 in T.thermophilus GluRS (Fig. 2b). Surprisingly, in theGluRS·tRNAGlu·Glu complex, Ser9 does not contactGlu but stabilizes the accepting A76 nucleotide fromtRNA on which Glu is bound, thus positioningindirectly the tRNA-boundGlu residue. Interestingly,the conformation of Glu in Glu-Q-RS is closer to thatin GluRS bound to tRNAGlu than in the free GluRS(Fig. 2c). It has been shown that, in T. thermophilusGluRS, the guanidinium group of Arg5 contacts theGlu substrate onlywhen tRNA ispresent.31 It isworthnoting that in Glu-Q-RS, the guanidinium group ofthe equivalent residue Arg9 contacts the side chain ofGlu via a salt bridge even in the absence of tRNA.The interaction of Glu with Glu-Q-RS is reinforced

by the solvent network maintained by Ser13, Gly21,Gly191, and Asp193 that contacts the α-carboxylgroup of Glu. These residues and thewater moleculesmay play a role similar to that of A76 of tRNAGlu bypositioning theGlu substrate in the samemanner as inthe ternary GluRS·tRNAGlu·Glu complex.31 Theseproperties explain why Glu-Q-RS, in contrast toGluRS, is fully competent to activate Glu in theabsence of tRNA. Indeed, Glu binds to Glu-Q-RS inthe absence of the other substrates in a productivemode, in the same manner as in GluRS complexed totRNAGlu31.

Glu-Q-RS binds D-Glu and L-Gln

It has been shown thatT. thermophilusGluRS exhibitsrelaxed specificity for the amino acid substrate31,32 andthat tRNAGlu triggers specific binding of L-Glu.32

Thus, in addition to its essential role as cofactor for Gluactivation, tRNAGlu also promotes selection of thecognate amino acid by GluRS.To analyze whether the specificity of Glu-Q-RS for

amino acid and the mechanism of selection of thecognate amino acid differ from those of GluRS, weanalyzed the binding of L-Glu, D-Glu, and L-Gln toGlu-Q-RS by fluorescence quenching. The fluores-cence intensity of the enzyme decreases by increas-ing the amino acid concentration by up to 17 mM.The plot representing the fluorescence of the proteinin the presence of amino acid over that in its absenceas a function of the amino acid concentration showsthat the cognate L-Glu induces stronger quenchingcompared to the noncognate amino acid (Fig. 3a).Thus, the noncognate amino acid either interactsdifferently with the protein or promotes conforma-tional changes different from those promoted by thecognate one. The Kd of the Glu-Q-RS·L-Glu complex

Fig. 1. Crystal structure of the E. coli Glu-Q-RS complexed to L-Glu. (a) Ribbon representation of the Glu-Q-RS·Glucomplex. The CP domain, the two half parts of the Rossmann fold, and the SC domain appear, respectively, in orange,blue, and green. The “KLSKQ” loop is in red. The zinc ion is in gray, and the L-Glu substrate is represented as a yellowCPK model. (b) Structure of Glu-Q-RS complexed to L-Glu colored by B-factor. Color turns from light blue to red withincreasing B-factor; light blue indicates low B-factors (minimal value, 7.63 Å2) and red indicates high B-factors (maximalvalue, 67.36 Å2). (c) Catalytic site of Glu-Q-RS. Strands and helices are represented as a blue ribbon; catalytic residues andL-Glu are represented, respectively, as yellow and green sticks. (d) 2lFo−Fc electron density of the catalytic site of Glu-Q-RScontoured at 2σ showing the L-Glu and three water molecules in its vicinity.


Fig. 2. Comparison of the Glu-Q-RS and GluRS catalytic sites. (a) Stereoview of the bound L-Glu in the Glu-Q-RS·Glucomplex. The lateral chains of the catalytic residues are in green, and the Glu substrate is in yellow. The hydrogen bondsare represented as black dashed lines, and water molecules are represented as pink spheres. (b) Stereoview of L-Glubinding in the T. thermophilus GluRS·Glu structure (PDB ID 2CUZ). The lateral chains of the catalytic residues and L-Gluare, respectively, in gray and mauve. (c) Stereoview of bound L-Glu in the T. thermophilus GluRS·tRNAGlu·Glu complex(PDB ID 2CVO). The lateral chains of the catalytic residues and L-Glu are, respectively, in gray and green, and thetRNAGlu terminal adenosine A76 is in pink. (d and e) Comparison of the conformation of L-Glu in the Glu-Q-RS·Glu andGluRS·tRNAGlu·Glu complexes (d) and in the Glu-Q-RS·Glu and GluRS·Glu complexes (e). The catalytic residues of theactive sites of the two enzymes have been superimposed.


Fig. 3. Interaction of L-Glu, D-Glu, and L-Gln with Glu-Q-RS. (a–c) Fluorescence quenching of Glu-Q-RS by L-Glu,D-Glu, and L-Gln in the absence (a) or in the presence (b and c) of ATP. E. coli Glu-Q-RS (2 μM) was incubated in 100 mMTris–HCl buffer (pH 7.2) containing 10% glycerol and 1mMMgCl2 with increasing concentrations of L-Glu (○), D-Glu (△),and L-Gln (□) at 20 °C in the absence (a) or in the presence (b and c) of 1 mM ATP. The fluorescence intensity ratios F/Fo(where Fo and F are the fluorescence intensities of Trp, respectively, in the absence and in the presence of ligands) aredetermined. Four experiments were conducted under the same conditions, and the plots were constructed using averagevalues and standard deviations. Dissociation constants were determined as described in Materials and Methods. (d–f)Comparison of L-Glu, L-Gln, and D-Glu binding in the Glu-Q-RS catalytic site. (d) Binding of L-Glu; the arrow indicates thesalt bridge between its side chain and the Arg9 residue. (e) A docking model of L-Gln in the catalytic site; the arrowindicates the loss of the salt bridge. (f) The docking model of D-Glu in the catalytic site; the steric hindrance is indicated bythe arrow. To construct the models, the oxygen of the γ-carboxyl group of L-Glu was replaced by a nitrogen group to formthe L-Gln, and the α-carboxyl group of L-Glu was oriented to form the D-isomer.


(about 2.1 mM) reflects the poor affinity of Glu-Q-RSfor Glu. The noncognate amino acids L-Gln and D-Glu bind, respectively, fourfold and sixfold weaker,as revealed by the increased dissociation constantsof the heterologous complexes (8.1 and 12.3 mM)(Fig. 3a, Table 1).The high Km values determined by ATP–PPi

exchange and tRNA aminoacylation support thepoor affinity of Glu-Q-RS for L-Glu (Table 1), butcontrast with the values in the 10–100 μM rangeusually determined for aaRS.However, ATP enhancesthe affinity for the amino acid by 50-fold (Kd=2.1 and0.047 mM in the absence and in the presence of ATP,respectively; Fig. 3a and c) by promoting Glu-AMPformation. In addition, ATP promotes specific acti-vation of amino acid, since Glu-AMP is formed onlywith L-Glu.The affinity of Glu-Q-RS for L-Glu is not affected by

tRNAAsp, since identicalKm values are determined byATP–PPi exchange performed either in the absence orin the presence of tRNAAsp.24,25 In contrast, tRNAGlu

alters the binding properties of E. coli and T.thermophilusGluRS for the amino acid, but differentlyon each enzyme (Table 1). In the absence of tRNA, L-

Glu, D-Glu, L-Asp, and L-Ala bind T. thermophilusGluRSwith similar affinities (Kd=0.15mM, as deduced fromfluorescence quenching curves32), but L-Glu and D-Glu promote stronger quenching compared to thetwo other amino acids.32 tRNAGlu promotes specificbinding of L-Glu to the protein without a change inaffinity, butwithweaker quenching.32 TheKd value ofthe GluRS·L-Glu complex in the presence of tRNAagreeswith theKmvalue in tRNAaminoacylation33,34(Table 1). In contrast, E. coli GluRS binds L-Glu 10times weaker in the absence of tRNAGlu than in itspresence (Kd=1.0 and 0.1 mM, respectively, andKm=0.1 and 0.03 mM, respectively, in tRNA aminoa-cylation and ATP–PPi exchange; Table 1).33,34 Thedifferent fluorescence quenchings of GluRS and Glu-Q-RS agree with the fact that, in the absence of tRNA,GluRS binds L-Glu (like the noncognate amino acid)nonfunctionally and is unable to activate it, whereasGlu-Q-RS binds L-Glu functionally and activates it inthe presence of ATP. The decreased affinity of Glu-Q-RS for L-Glu, as compared to that of T. thermophilusGluRS,may be related to themore open conformationof its catalytic cleft, which (i) increases the rate ofdissociation more than that of association, and (ii)

Table 1. Binding properties of L-Glu, D-Glu, and L-Gln on E. coli and T. thermophilus GluRS and E. coli Glu-Q-RS

Amino acid

E. coli and T. thermophilus GluRS E. coli Glu-Q-RS

− tRNA +tRNA − tRNA (−ATP/+ATP) +tRNA

Kd of the enzyme·amino acid complex (mM)L-Glu 1.1a,33 0.15b,32 0.1a,33 0.15b,32 2.1/0.047 n.d.L-Gln 0.15b,32 No bindingb,32 8.1/6.1 n.d.D-Glu 0.15b,32 No bindingb,32 12.3/8.3 n.d.

Reaction

Km for L-Glu (mM)tRNA aminoacylation No reaction 0.1a,34,35 No reaction 328

ATP–PPi exchange No reactiona,34 0.03a,24–26,28 5.328 528

tRNA aminoacylation No reaction 0.08b,32

Ki for glutamol-AMP (μM)tRNA aminoacylation No reaction 3a,36 No reaction 0.40a,28

ATP–PPi exchange No reactiona,34 2a,28 n.d. 0.44a,28

n.d.: not determined.The references of the constants not determined in this work are given in the index.

a E. coli GluRS.b T. thermophilus GluRS.


prevents any effect of tRNA. However, formation ofGlu-AMP is accompanied by a strong increase in theaffinity of the amino acid for both enzymes (50-foldfor Glu-Q-RS; Table 1).The increased affinity of Glu-Q-RS for L-Glu after

adenylation is further confirmed by the low Ki valuesof glutamol-AMP, the nonhydrolyzable analogue ofGlu-AMP that competitively inhibits GluRS and Glu-Q-RS with respect to Glu (Ki=2 and 0.44 μM).24–26,36

Interestingly, adenylation increases the affinity of GluforGlu-Q-RS by 2 orders ofmagnitudemore than thatfor GluRS, since theKd values of GluRS andGlu-Q-RScomplexed to L-Glu are in a ratio of 0.03, whereas theratios of the Ki values of GluRS and Glu-Q-RS forglutamol-AMP are in a ratio of 4.5 in ATP–PPiexchange and in a ratio of 7.5 in tRNAaminoacylation(Table 1). As a consequence, the exceptionally lowaffinity of Glu for Glu-Q-RS with respect to that forGluRS is compensated by a stronger synergistic effectpromoted by the AMPmoiety for the binding of Glu-AMP to Glu-Q-RS.Altogether, these results indicate that acquisition by

Glu-Q-RS of the ability to activate Glu in the absenceof tRNAwas accompanied by the ability of the proteinto select the cognate amino acid in the absence oftRNA, but with a strongly decreased affinity. Thischange, however, did not significantly affect kcatvalues (18 and 10 s−1, respectively, for Glu-Q-RSand GluRS in tRNA aminoacylation).24–26

The open question is: Why does Glu-Q-RS fail toactivate L-Gln and D-Glu? Docking models of Glu-Q-RS complexed to the noncognate amino acid showthat L-Gln and D-Glu are able to fit the catalytic site ofGlu-Q-RS, but that L-Gln, in contrast to L-Glu, isunable to form a strong salt bridge with Arg9 becauseof the presence of the γ-amide group in Gln sub-stituting the γ-carboxyl group in Glu (Fig. 3d–f). As aconsequence, the α-carboxyl group of L-Gln probablyprotrudes in the ATP-binding pocket. Furthermore, asshown in Fig. 2b, the α-carboxyl group of L-Gluinteracts with several residues via water molecules;

after D-Glu binding, a reorientation of this group mayabolish these interactions. Since ATP does not signifi-cantly affect the Kd of the enzyme bound to non-cognate amino acid (Table 1), the steric hindrancepromotedby thenoncognate aminoacidonlypreventsfunctional binding of ATP and amino acid activation.

ATP-binding pocket

The crystallographic data of the Glu-Q-RS·Glucomplex allowed the building of the loop insertedbetween Pro224 and Pro236, which was not resolvedin the crystalline structure of the free enzyme. Thehigh B-values indicate that this loop is mobile (Fig. 1aand b). In T. thermophilus GluRS, this loop containsseveral residues involved in ATP binding.37 Compari-son of this loop with that of GluRS gives clues aboutthe distinct ATP-bindingmechanisms and differencesin Glu activation by the two enzymes. In T. thermo-philus GluRS, 17 residues mediate recognition of ATP(Fig. 4a) either in the productive mode or in thenonproductive mode.37 Among them, seven areconserved in Glu-Q-RS, eight are semiconserved,and only two (Tyr20 and Trp209) differ. Tyr20, which,in GluRS, participates in the formation of the ATPhydrophobic binding pocket, is replaced, in Glu-Q-RS, by Ile25. The contribution of this residue may,however, be weak, since it is not conserved in allGluRS or in all Glu-Q-RS. In T. thermophilus GluRS,Trp209 seems to play a crucial role in Glu activation.After tRNA binding, this residue is moved deeperinto the active site, allowing the ribose moiety of ATPto reach a position suitable for reacting with Glu. InGlu-Q-RS, this Trp residue is replaced by Leu194.Active site comparison suggests that this replacementwill permit ATP to be accommodated deeper into thebinding pocket of the enzyme and in a productivemode for Glu activation. However, further analysis ofthe 3D structure of the Glu-Q-RS·Glu complexsuggests that the Trp209→Leu194 replacement can-not solely account for tRNA-independent Glu

Fig. 4. Residues involved in the binding of ATP to Glu-Q-RS. (a) Sequence alignment of E. coli Glu-Q-RS andT. thermophilus GluRS based on the 3D structures. The secondary structures of Glu-Q-RS are indicated on the sequences;strands are represented by red arrows, helices are represented by blue ovals, and coils are in green. Residues of GluRSinvolved in ATP binding in the productivemode or in the nonproductivemode are indicated by black boxes. Asterisks anddots indicate conserved and semiconserved residues, respectively. (b) Comparison of the “KLSKQ” loop ofE. coliGlu-Q-RSand the “KISKR” loop in the T. thermophilus GluRS·ATP·Glu complex (PDB ID 1J09). The two structures have beensuperimposed. The structure of GluRS complexed toATP and L-Glu is in light green, and that of Glu-Q-RS is in yellow; ATPand L-Glu are represented. The Glu-Q-RS C-terminal part close to the KLSKQ sequence is indicated. The red arrow showsthat the loop of Glu-Q-RS pushes ATP deeper into the catalytic site than does the loop in the GluRS·ATP·Glu complex.



activation. Indeed, tRNA binding triggers manystructural rearrangements in GluRS. In particular,both the SC domain and the KISKR loop (correspond-ing to the consensus aaRS class I motif KMSKS) havebeen reported tomove. InGlu-Q-RS, the SCdomain isreplaced by the C-terminus of the protein. Thisdomain is very close to the KLSKQ loop, whichseems to be pushed deeper into the direction of thecatalytic site. Consequently, due to these movements,even in the absence of tRNA, ATP probably entersdeeper into the cavity and becomes competent topromote efficient Glu activation (Fig. 4b).

Identification of tRNAAsp elements thatdetermine glutamylation by Glu-Q-RS

The model of the 3D structure of the Glu-Q-RS·tRNAAsp complex suggests binding of the sub-strate by the anticodon stem loop.26,28 Among thetRNA species from E. coli possessing Q34 in the anti-codon (tRNAAsp, tRNAAsn, tRNAHis, and tRNATyr),only tRNAAsp is a substrate ofGlu-Q-RS. SinceGlu-Q-RS is present in other prokaryotes that synthesizequeuosine such as Neisseria meningitidis and since theenzyme from this organism exhibits kinetic propertiesthe same as those from E. coli,25 it may be expectedthat the Glu-Q-RS from various prokaryotes promoteall specific glutamylations of tRNAAsp Q34. For thisreason, the elements of tRNAAsp determining recog-nition byGlu-Q-RS should be conserved in the tRNA-Asp of the organisms expressing this enzyme, whereasthey should be absent in the three other Q34-contain-ing tRNAs.To identify the potential nucleotide candidates of

tRNAAsp that may be involved in recognition by Glu-Q-RS, we aligned the four E. coli tRNA speciespossessing Q34. Figure 5a shows that, among the 17positions of the anticodon stem loop, 6 positions arestrictly conserved, namely, U33, G34, U35, A37, andthe G30-C40 pair. Thus, tRNAAsp, the substrate ofGlu-Q-RS, differs from tRNAAsn, not the substrate, by8 positions. The base pairs C27-G43, U29-A41, C31-G39, C36, and C38 in tRNAAsp are substituted intRNAAsn by the base pairs G27-U43, G29-C41, A31-U39, U36, and A38. The U29-A41 pair that is notconserved in the tRNAAsp of the organisms expres-sing Glu-Q-RS was excluded from set of the potentialidentity elements. Since tRNAAsn is closer to tRNAAsp

than to tRNATyr and tRNAHis containing alsoQ34,weintroduced the potential identity elements for gluta-mylation of tRNAAsp into tRNAAsn by nucleotidesubstitutions and tested the glutamylation capacity ofthe variants.Characterization of the tRNA elements promoting

recognition by Glu-Q-RS through nucleotide substi-tutions encountered three major experimental diffi-culties. First, the tRNAAsp transcripts deprived ofqueuosine are not substrates of Glu-Q-RS and areeven deprived of significant affinity for the enzyme(not shown). Therefore, the recognition of tRNAAsp

variants by Glu-Q-RS could not be investigated bymeasurement of the inhibition capacity of variantstranscribed in vitro, but only by analysis of the

glutamylation capacity of posttranscriptionallymodified variants synthesized in vivo. Second, thetRNA isolated from strains expressing the tRNAvariants contains also the endogenously modifiedtRNAAsp, which should be separated from variants.Third, purification of the tRNAAsp variants synthe-sized in vivo is not easy because their chromato-graphic and electrophoretic properties are similar tothose of the wild-type tRNAAsp. Indeed, all attemptsto separate an overexpressed tRNAAsp in which weintroduced a single mutation from wild-type tRNA-Asp failed (not shown). The inability to obtain puremodified tRNAAsp variants restricted the determina-tion of their glutamylation kinetic constants by Glu-Q-RS. For this reason, we designed a plateau assaythat only gives a qualitative identification of thetRNAAsp identity elements for glutamylation by Glu-Q-RS, but not their relative importance and theirinvolvement in binding or catalysis. However, theextent of glutamylation of the tRNA variants reflectsthe integrity of the determinants for recognition byGlu-Q-RS only (i) if the mutation introduced into thetRNA has no effect on the level of its expression, and(ii) if the mutation does not affect Q34 queuosinyla-tion. The tRNAAsp and tRNAAsn variants tested werenot altered in the essential identity elements govern-ing aspartylation,38 and all were excellent substratesfor AspRS. This means that the aspartylation extentsof tRNA isolated from strains overexpressing nativetRNAAsp or tRNAAsn or their variants reflect theamounts of accumulated aspartylable tRNA.We alsoshowed that the mutations introduced in tRNAAsp

and tRNAAsn do not significantly affect the level ofexpression and aspartylation. Furthermore, neitheroverexpression of tRNAAsp and tRNAAsn nor themutations introduced in the anticodon stem loopaffect their queuosinylation, since it has been reported(i) that a 15-fold overexpression of E. coli tRNAAsp

does not alter the yield of its posttranscriptionalmodifications, in particular of its queuosinylation,39

and (ii) that the structural elements of tRNAdetermining recognition by tRNA guanine transgly-cosylase are solely constituted by the anticodon andby the presence of a pyrimidine in position 32.40

Comparison of the extent of glutamylation ofoverexpressed tRNAAsp and tRNAsn variants pro-vides information on whether the tRNA variant is asuitable substrate for Glu-Q-RS. To evaluate the ex-tent of glutamylation of the overexpressed tRNA, wefirst determined the amount of tRNAAsp or tRNAAsn

accumulating in the overexpressing E. coli strain byaspartylation of the total tRNA and subtraction of theendogenous aspartylable tRNA from the total tRNAof the wild-type strain. Since the glutamylationcapacity by Glu-Q-RS of the total tRNA extractedfrom a strain overexpressing tRNAAsp or tRNAAsn

measures the amount of glutamylable overexpressedtRNA and endogenous tRNAAsp, subtraction of theendogenous tRNAglutamylated byGlu-Q-RS revealsthe glutamylation capacity of the overexpressedtRNAAsp or tRNAAsn. Thus, to search the positionsof tRNAAsp involved in recognition by Glu-Q-RS, wecompared the percent glutamylation by Glu-Q-RS of

Fig. 5. Characterization of thetRNA elements determining gluta-mylation of nucleoside Q34. (a)Sequence alignments of the antic-odon arms and loops of E. colitRNAAsn, tRNAAsp, tRNAHis, andtRNATyr. Conserved residues are inwhite against a black background.(b) Aminoacylation extents of wild-type (wt) and mutated tRNAAsn.(c) Aminoacylation extents of wtand mutated tRNAAsp. White andblack histograms indicate, respec-tively, the aspartylation extents oftRNAAsp and tRNAAsn by E. coliAspRS and T. thermophilus non-discriminating AspRS, and the glu-tamylation extents of these tRNAsby the E. coli Glu-Q-RS. The over-expressed tRNAAsp or tRNAAsn isaspartylated at an extent of 100%,and percent glutamylation by Glu-Q-RS of the overexpressed nativetRNAAsp and tRNAAsn and theirvariants is shown. The aminoacyla-tion extents of overexpressed nativetRNAs and their variants wereevaluated as described in the text.The secondary structures of the wttRNA are displayed on the left, andthe anticodon stem loops of thetRNA variants are underneath thehistograms. The mutated positionsare boxed.


the aspartylable overexpressed tRNAAsp or tRNAAsn

in the total tRNA isolated from E. coli strainsexpressing the tRNAAsp or tRNAAsn variants, aftersubtraction of the glutamylable tRNA in the totaltRNA from the wild-type strain.The nature of the substitutions and their effect on

tRNA glutamylation are summarized in Fig. 5b.Substitution in tRNAAsn of base pairs G27-U43 andA31-U39 by the C27-G43 and C31-G39 base pairs oftRNAAsp does not promote glutamylation of thetRNA, suggesting that the nucleotides of the anti-codon stem do not significantly contribute to recogni-tion by Glu-Q-RS. In contrast, exchange of A38 intRNAAsn by C38 of tRNAAsp promotes 55% glutamy-lation of the variant, suggesting that this nucleotideplays a crucial role in tRNA glutamylation by Glu-Q-RS. This observation is confirmedby a 90%decrease inthe glutamylation capacity of the tRNAAsp variant inwhich C38 is substituted byU38 (Fig. 5c). Substitutionof C32 by U32 in tRNAAsp reduces the extent ofglutamylation to 36%, indicating that this position isalso involved—although to a lesser extent than C38—in recognition by Glu-Q-RS. Substitution of bothnucleotides C32 and C38 in tRNAAsp by U32 and U38almost completely abolishes glutamylation by Glu-Q-RS. Mutations of other positions of the tRNAAsp

anticodon loop were either without effect on gluta-

mylation or could not be obtained because thevariants did not accumulate in the cells.

Conclusion

Glu-Q-RS is a rare example of a creation of a novelenzyme activity by deviation of an aaRS. This paralogof the catalytic core of GluRS evolved by conservationof the ability to activate Glu and to form Glu-AMP. Itlost the anticodon-binding domain and the canonicalfunction of GluRS consisting in the attachment of Gluonto the 3′ accepting endof tRNAGlu and acquired theability to bind tRNAAsp and to transfer activated Gluonto the anticodon Q34 residue. This enzyme, whichconserved tRNA aminoacylation capacity, promoteshypermodification of tRNA and thus belongs to thefamily of tRNA-modifying enzymes. The steps thatled to the emergence of this enzyme resemble thosethat led to the archaeal Asn synthetase. This paralogof the catalytic cores of AspRS and AsnRS evolvedfrom an ancestral AspRS that lost the anticodon-binding domain and the ability to bind tRNA, butconserved the capacity to activate Asp into Asp-AMPand acquired that to convert the activated Asp intoAsn.20 The structural characteristics of class II aaRS,as well as most amino acid residues of the catalyticpocket of AspRS, are conserved in Asn synthetase.


Likewise, Glu-Q-RS conserved the structural char-acteristics of class I aaRS and the capacity to activateGlu intoGlu-AMP, in agreementwith conservation ofmost amino acid residues ofGluRS contacting theGluandATP substrates. Nevertheless, these two enzymesdiffer by their affinity for Glu, and they diverge by themechanism of selection and activation of the cognateamino acid. In GluRS, the cognate tRNAGlu eitherincreases affinity for Glu or triggers its specificbinding and activation35,41 (Table 1). Thus, the roleof tRNA in the selection and binding of Glu was lostwhen Glu-Q-RS acquired new tRNA specificity and anovel function.The structural basis of the distinct functional pro-

perties of the two enzymes is puzzling, since most ofthe amino acid residues that contact Glu are con-served. Interestingly, the active sites of the twoenzymes differ by a single residue, which wasproposed to be involved in the binding of Glu andATP in T. thermophilusGluRS.37 Comparison of the 3Dstructure of T. thermophilus GluRS in free form withthat of the protein complexed with the substratessuggests that Trp209 determines tRNA-dependentamino acid activation. By displacing the side chain ofTrp209, tRNAGlu allows ATP to fall into the catalyticcleft and to promote functional binding via inter-actions between the adenine ring and the amino acidside chain. In Glu-Q-RS, the equivalent residueLeu194 does not close the catalytic cleft and permitsATP to bind functionally by contacting directly theadenine ring in the absence of tRNA. However, thekinetic investigations reveal that the capability ofGlu-Q-RS to activate Glu in the absence of tRNA iscorrelated with the lack of effect of tRNA on affinityand on the specificity of binding of the amino acid,which is probably related to the more openedconformation of the catalytic cleft of Glu-Q-RS thanto that of T. thermophilus GluRS.The results reported here reveal distinct modes of

recognition of the accepting end of tRNAGlu and theQ34 residue of tRNAAsp anticodon by, respectively,GluRS and Glu-Q-RS. Recognition of tRNAGlu byGluRS is ruledby theG1-C72andU2-A71basepairs ofthe acceptor arm, the U11-A24 base pair in the D-arm,the U13-G22-A46 tertiary interaction, the lack ofresidue 47, and the posttranscriptional modificationmnm5s2U34, together with nucleotides U34, U35,and C36 of the anticodon loop.42–44 Recognition oftRNAAsp by Glu-Q-RS is determined by nucleotideC38, which plays a prevalent role reinforced by C32.Interestingly, C38 and C32 are present in the antic-odon of E. coli tRNAGlu that, however, does not bindGlu-Q-RS.26,28 Therefore, the contribution of C38 andC32 to Q34 glutamylation depends on the structuralcontext of tRNA, which involves (i) posttran-scriptional modifications, probably Q34, since thetRNAAsp transcript does not bind Glu-Q-RS,26 and(ii) probably also the presence of antideterminants.Contribution of this structural context to Q34 gluta-mylation is supported by the observations that C38,together with C32, promotes Q34 glutamylation intRNAAspmore efficiently than in tRNAAsn (Fig. 5), andthat transplantation of C38 in tRNAHis does not confer

significant glutamylation capacity ofQ34byGlu-Q-RS(result not shown). Thus, the appropriate environmentfor Q34 glutamylation is only found in tRNAAsn

besides tRNAAsp, but not in tRNAGlu and tRNAHis.The mechanism by which E. coli Glu-Q-RS selects

tRNAAsp for glutamylation among the other tRNAspecies possessingQ34 seems to be used byGlu-Q-RSexpressed in the other organisms. Indeed, compara-tive analysis of the tRNA sequences (not shown)reveals that C32 and C38, which we have character-ized as important for tRNA Q34 glutamylation in E.coli tRNAAsp, are conserved in the tRNAAsp of theseorganisms, whereas in tRNAAsn and tRNATyr, C38 issubstituted by A38, and tRNAHis presents an A atposition 32, and an A, C, or U at position 38.GluRS forms the subclass Ia of aaRS with GlnRS

and ArgRS. Besides the peculiar structural charac-teristics of the amino-acid-binding site, these aaRScatalyze ATP–PPi exchange only in the presence ofthe cognate tRNA. This kinetic behavior led to theproposal that aa-tRNAs are formed, at least by theseaaRS, via a concerted mechanism in which the aa-AMP is not an obligatory intermediate.45 Extensiveinvestigations of themechanism of tRNA glutamyla-tion by E. coli GluRS demonstrated that GluRSaminoacylates tRNAGlu via a two-step pathway inwhich Glu is activated before it acylates the tRNA.Thus, the complex formed by GluRS bound totRNAGlu acts like a ribonucleoprotein particle inwhich tRNAGlu triggers the first step of the reaction(i.e., the selection of the homologous amino acid andits activation by ATP before to become the substrateable to react with the activated amino acid).31,35,41,46

However, whereas most conventional ribonucleopro-tein particles in which the RNA partner is not asubstrate are stable, the stability of the tRNA gluta-mylation particle is limited because, after aminoacy-lation, GluRS has to exchange the glutamylatedtRNAGlu with free tRNAGlu. In contrast, the Glu-Q-RS·tRNAAsp complex cannot be qualified as aribonucleoprotein, since tRNAAsp does not induce theactive enzyme conformation that promotes aminoacid binding and activation. These competencieswereacquired by the protein when the role of tRNA wasrestricted to that of the substrate of the activatedamino acid. Thus, the glutamylation system consti-tutes an example illustrating how a ribonucleoproteinparticle evolved into a fully competent protein bytransfer of the competence of RNA to the protein.

Materials and Methods

Expression of Glu-Q-RS in E. coli and purification ofthe protein

E. coli Glu-Q-RS was expressed in BL21(DE3) Rosetta 2strain (Novagen) transformed by the recombined pDest14-yadB vector.24 The cells were grown at 37 °C, and ex-pression of the protein was induced by addition of 0.5 mMIPTG at 0.8 A700 nm. After overnight culture, the cells werepelleted and suspended in 100 mM Tris–HCl buffer(pH 8.0) containing 1 mM β-mercaptoethanol, 0.1 mMNa2 ethylenediaminetetraacetic acid, 1 mM benzamidine,

Table 2. Crystal structure data and refinement statisticsof the E. coli Glu-Q-RS·Glu complex

Data collectionSpace group C2221Cell parameters (Å) a=39.1; b=115.0; c=154.4X-ray source ID14eh2Wavelength (Å) 0.933Resolution (Å) 1.75Number of observations

Unique 34,396Total 128,015

Completeness (%) 96.1 (75.5)a

Rsym 0.046 (0.095)a

Mean I/σ 25.4 (7.0)a

RefinementResolution range (Å) 38–1.75R-factor 18.7Rfree 21.4Number of atoms

Protein 2404Zn+ 1Water 440

r.m.s.d.Bond angle (°) 1,3Bond length (Å) 0.007Average atomic B-value (Å2)

ProteinGlu-Q-RS 21.95Ligand 22.2

Water 35.7Ramachandran plotb (%)

Core region 90.7Allowed regions 8.6Generously allowed regions 0.4a The number inside the parentheses is for the last shell

(1.80–1.75).b Statistics from PROCHECK.32


and 0.1 mM of the protease inhibitor AEBSF (Interchim).Except for the Tris–HCl buffer, all reactants were presentduring the purification. After sonication of the cellsuspension, the crude extract was centrifuged at 105,000gfor 90 min, and the supernatant was dialyzed against20 mM potassium phosphate buffer (pH 7.2; buffer A). Thepresence of 10% glycerol during the purification preventedthe precipitation of Glu-Q-RS. The S105 extract was loadedonto a DEAE cellulose DE52 column (6 cm2×30 cm)equilibrated with the dialysis buffer, and the proteins wereeluted with a linear gradient of potassium phosphate from20 mM (pH 7.2) to 250 mM (pH 6.8). The fractionscontaining Glu-Q-RS were dialyzed against buffer A andloaded onto a phosphocellulose column P11 (6 cm2×30 cm)before elution of the proteins with a linear gradient from 0to 500 mM KCl in buffer A. The active fractions weredialyzed against buffer A and loaded onto a hydroxyapatiteCHT-20 (20 ml; Bio-Rad) column. Interestingly, Glu-Q-RSeluted in the flow-through, whereas the contaminatingproteins were retained on the matrix. The pure protein wasdialyzed against 20 mM Na-Hepes buffer (pH 7.2) and50mMNaCl, concentrated on Centricon 30 K, and stored at4 °C in the presence of 10% glycerol.

Crystallization and X-ray data collection

Crystals were obtained in sitting drops by mixing 2 μl ofthe protein solution concentrated to 9.7 mg/ml and 2 μl ofthe crystallization solution containing 0.1 M Mg-acetateand Na-cacodylate buffer (pH 5.5), 0.2 M KCl, 10%polyethylene glycol 8000, and 2 mM L-Glu. Long andthin platelets (1000×140×20 μm3) were obtained in a fewdays. The crystals were mounted in a nylon loop.Cryobuffer was not necessary before data collection. X-ray data were collected at 100 K at beamline ID14eh2 ofthe European Synchrotron Radiation Facility (Grenoble,France) using an ADSC quantum 4 CCD detector withincident radiation, a wavelength of 0.933 Å, and a crystal-to-detector distance of 180 mm. The diffraction intensitieswere recorded with 0.5° oscillation and a 30-s exposureper image over a range of 180°. X-ray data were indexed,processed, and scaled using HKL2000.47

Structure determination and refinement

The structure was solved by molecular replacementusing MOLREP (CCP4 computational project) and theGlu-Q-RS structure [Protein Data Bank (PDB) ID 1NZJ24]as a search model, and refined to 1.75 Å using CNS,48 with34,396 significant unique reflections (completeness,96.1%). Manual adjustments, rebuilding, and addition ofwater molecules were performed using the programCoot.49 The final model (Rcryst=18.7%; Rfree=21.4%; testset, 5% of the reflections) containing 440 ordered watermolecules was validated using PROCHECK.50 The aver-age atomic B-values are 21.95, 22.2, and 35.7 Å2 for Glu-Q-RS, Glu, and water molecules, respectively. The structureexhibits good geometry, with 90.7% of the residues in thecore region, 8.6% in allowed regions, and 0.4% ingenerously allowed regions. Data statistics and refinementstatistics are summarized in Table 2.

Fluorescence measurements

The fluorescencemeasurementswere performed at 20 °Con a Hitachi F3010 spectrofluorimeter using an excitationwavelength of 295 nm and by recording emission at

340 nm. The excitation and emission bandpasses were 5and 10 nm, respectively. The absorbance of the solutionwas b0.1 absorption unit/cm at the excitation wavelength,and the fluorescence intensity was corrected for dilutionwhen ligand solution was added. Quenching was evalu-ated from the ratio of the corrected fluorescence intensity inthe presence of the substrate to that in its absence. The datawere fitted to a single binding site equation using KYPLOT(Koichi Yoshioka, 1997–2000, version 2.0, beta 13).

tRNA aminoacylation

The reaction mixture of 50 or 100 μl contained 0.1 MNa-Hepes buffer (pH 7.2), 2 mM ATP, 5 mM MgCl2, 30 mMKCl, 20 μM L-[14C]Glu (330 cpm/pmol) or L-[14C]Asp(270 cpm/pmol), unfractionated tRNA (4 mg/ml), andappropriate concentrations of Glu-Q-RS for plateaumeasurements. After various incubation times, the [14C]Glu-tRNA or [14C]Asp-tRNA formed in 10- or 20-μlaliquots was determined as described.51 The glutamyla-tion capacities of overexpressed tRNAAsp, tRNAAsn, andtheir variants were determined from aminoacylationplateaus obtained by Glu-Q-RS. The values were correctedby subtraction of the glutamylation capacity of the endo-genous tRNAAsp in the crude tRNA extracted from thewild-type strain evaluated at 2.5% (see the text).

Preparation of tRNA mutants

The genes ofwild-typeE. coli tRNAAsp and tRNAAsn andtheir variants were cloned into the pKK223-3 vector under


the control of the ptac promoter using the cassette cloningprocedure.52 The tRNAs were expressed in E. coli JM 103strain. Total tRNAwas isolated from an overnight cultureof 500ml by extractions with phenol saturatedwith 50mMNa-acetate buffer (pH 5.0), 0.1 mM MgCl2, 0.1 mM Na2ethylenediaminetetraacetic acid (buffer B), and chloro-form, followed by ethanol precipitation. The dried tRNAwas dissolved in buffer B, loaded onto a 2-ml DEAE cellu-lose column, and eluted with buffer A containing 1.5 MNaCl. After ethanol precipitation, the tRNAwas dissolvedin water and stored at −20 °C. The contents of tRNAAsp

and tRNAAsn were estimated by charging with E. coliAspRS or nondiscriminating T. thermophilus AspRS2.

Accession codes

Structure factors and coordinates were deposited intothe PDB under accession no. 2ZLZ.

Acknowledgements

This work was supported by the University LouisPasteur (Strasbourg), the Centre National de laRecherche Scientifique, the Association pour laRecherche sur le Cancer, the ACI Biologie CellulaireMolécaire et Structurale, and the Department ofSciences and Technology (New Delhi). EuropeanSynchrotronRadiation Facility teamsoperating beam-lines ID14-2/4, ID23, and BM30 are acknowledged forthe timeallocated to theproject and for their assistanceduring data collection. The authors thank Dr. GautamBasu (Bose Institute, Calcutta) for fluorimetric mea-surement facilities, and Dr. Søren Thirup (AarhusUniversity, Denmark) and Dr. Richard Giegé (IBMC,Strasbourg) for facilitating some steps of thiswork.M.B. was a recipient of a fellowship from the Associationpour la Recherche sur le Cancer.

References

1. Eriani, G., Delarue, M., Poch, O., Dirheimer, G. &Moras, D. (1990). Partition of tRNA synthetases intotwo classes based on mutually exclusive sets of seq-uence motifs. Nature, 347, 203–206.

2. Moras, D. (1992). Structural and functional relation-ships between aminoacyl-tRNA synthetases. TrendsBiochem. Sci. 17, 159–164.

3. Cusack, S. (1997). Aminoacyl-tRNA synthetases. Curr.Opin. Struct. Biol. 7, 881–889.

4. Lin, L., Hale, P. H. & Schimmel, P. (1996). Aminoacy-lation error correction. Nature, 384, 33–34.

5. Nureki, O., Vassylyev, D. G., Tateno, M., Shimada, A.,Nakama, T., Fukai, S. et al. (1998). Enzyme structurewith two catalytic sites for double-sieve selection ofsubstrate. Science, 283, 578–582.

6. Dock-Bregeon, A., Sankaranarayanan, R., Romby, P.,Caillet, J., Springer, M., Rees, B. et al. (2000). TransferRNA-mediated editing in threonyl-tRNA synthetase.The class II solution to the double discriminationproblem. Cell, 103, 877–884.

7. Renault, L., Kerjan, B., Pasqualato, S., Menetrey, J.,Robinson, J. C., Kawaguchi, S. et al. (2001). Structure ofthe EMAPII domain of human aminoacyl-tRNA

synthetase complex reveals evolutionary dimer mimi-cry. EMBO J. 20, 570–578.

8. Deniziak, M., Sauter, C., Becker, H. D., Paulus, C. A.,Giegé, R. & Kern, D. (2007). Deinococcus glutaminyl-tRNA synthetase is a chimer between proteins from anancient and a modern pathways of aminoacyl-tRNAformation. Nucleic Acids Res. 35, 1421–1431.

9. Wolf, Y. I., Aravind, L., Grishin, N. V. & Koonin, E. V.(1999). Evolution of aminoacyl-tRNA synthetases:analysis of unique domain architectures and phylo-genetic trees reveals a complex history of horizontalgene transfer events. Genome Res. 9, 689–710.

10. Becker, H. D., Roy, H., Moulinier, L., Mazauric,M. H., Keith, G. & Kern, D. (2000). Thermus ther-mophilus contains an eubacterial and an archae-bacterial aspartyl-tRNA synthetase. Biochemistry, 39,3216–3230.

11. Schimmel, P. & Söll, D. (1979). Aminoacyl-tRNAsynthetases: general features and recognition oftransfer RNAs. Annu. Rev. Biochem. 48, 601–648.

12. Lapointe, J. & Giegè, R. (1991). Transfer RNAs andaminoacyl-tRNA synthetases. In Translation in Eukar-yotes, (Trachsel, H. ed), pp. 35–69. CRC Press Inc. BocaRaton F.L.

13. Ibba, M. & Söll, D. (2004). Aminoacyl-tRNAs: settingthe limits of the genetic code. Genes Dev. 18, 731–738.

14. Böck, A., Thanbichler, M., Rother, M. & Resch, A.(2005). Selenocysteine. In The aminoacyl-tRNA synthe-tases (Ibba, M., Francklyn, C. S. & Cusack, S., eds),pp. 320–327. Georgetown, TX: Landes Biosciences.

15. Feng, L., Tumbula-Hansen, D., Min, B., Namgoong, S.,Salazar, J., Orellana, O. et al. (2005). Transfer RNA-dependent amidotransferases: key enzymes for Asn-tRNAAsn and Gln-tRNAGln synthesis in nature. In Theaminoacyl-tRNA synthetases (Ibba, M., Francklyn, C. S.& Cusack, S., eds), pp. 314–318. Georgetown, TX:Landes Biosciences.

16. Kern, D., Roy, H. & Becker, H. D. (2005). Asparaginyl-tRNA synthetase: pathways and evolutionary historyof tRNA asparaginylation. In The aminoacyl-tRNAsynthetases (Ibba, M., Francklyn, C. S. & Cusack, S.,eds), pp. 193–209. Georgetown, TX: Landes Bioscience.

17. Brevet, A., Chen, J., Lévêque, F., Blanquet, S. &Plateau, P. (1995). Comparison of the enzymatic pro-perties of the two Escherichia coli lysyl-tRNA synthe-tases species. J. Biol. Chem. 270, 14439–14444.

18. Putzer, H., Grunberg-Manago, M. & Springer, M.(1995). Independent genes for two threonyl-tRNAsynthetases in Bacillus subtilis. J. Bacteriol. 172,4593–4602.

19. Becker, H. D., Reinbolt, J., Kreutzer, R., Giegé, R. &Kern, D. (1997). Existence of two distinct aspartyl-tRNA synthetases in Thermus thermophilus. Structuraland biochemical properties of the two enzymes.Biochemistry, 36, 8785–8797.

20. Roy, H., Becker, H. D., Reinbolt, J. & Kern, D. (2003).When contemporary aminoacyl-tRNA synthetasesinvent their cognate amino acid metabolism. Proc.Natl Acad. Sci. U. S. A. 100, 9837–9842.

21. Sissler, M., Delorme, C., Ehrlich, S. D., Renault, P. &Francklyn, C. (1999). An aminoacyl-tRNA synthetaseparalog with a catalytic role in histidine biosynthesis.Proc. Natl Acad. Sci. U. S. A. 96, 8985–8990.

22. Ahel, I., Korencic, D. & Söll, D. (2003). Trans-editing ofmischarged tRNAs. Proc. Natl Acad. Sci. U. S. A. 100,15422–15427.

23. Korencic, D., Ahel, I., Schelert, J., Sacher, M., Ruan, B.,Stathopoulos, C. et al. (2004). A freestanding proof-readingdomain is required for protein synthesis quality


control in Archaea. Proc. Natl Acad. Sci. U. S. A. 101,10260–10265.

24. Campanacci, V., Dubois, D. Y., Becker, H. D., Kern, D.,Spinelli, S., Valencia, C. et al. (2004). The Escherichia coliyadB gene glutamylates specifically tRNAAsp. J. Mol.Biol. 337, 273–283.

25. Dubois, Y. D., Blaise, M., Becker, H. D., Campanacci, V.,Keith, G., Cambillau, C. et al. (2004). An aminoacyl-tRNA synthetase-like protein encoded by the Escher-ichia coli yadB gene glutamylates specifically tRNAAsp.Proc. Natl Acad. Sci. U. S. A. 101, 7530–7535.

26. Blaise, M., Becker, H. D., Keith, G., Cambillau, C.,Lapointe, J., Giegé, R. et al. (2004). A minimalistglutamyl-tRNA synthetase dedicated to aminoacyla-tion of the tRNAAspQUC anticodon. Nucleic Acids Res.32, 2768–2775.

27. Salazar, J. C., Ambrogelly, A., Crain, P. C., McCloskey,J. A. & Söll, D. (2004). A truncated aminoacyl-tRNAsynthetase modifies RNA. Proc. Natl Acad. Sci. U. S. A.101, 7536–7541.

28. Blaise, M., Becker, H. D., Lapointe, J., Cambillau, C.,Giegé, R. & Kern, D. (2005). Glu-Q-tRNAAsp synthe-tase coded by the yadB gene, a new paralog ofaminoacyl-tRNA synthetase that glutamylates tRNA-Asp anticodon. Biochimie, 87, 847–861.

29. Liu, J., Lin, S.-X., Blochet, J.-E., Pézolet, M. & Lapointe,J. (1993). The glutamyl-tRNA synthetase from Escher-ichia coli contains one atom of zinc essential for itsnative conformation and its catalytic activity. Biochem-istry, 32, 11390–11396.

30. Nureki, O., Kohno, T., Sakamoto, K., Miyazawa, T.& Yokoyama, S. (1993). Chemical modification andmutagenesis studies on zinc binding of aminoacyl-tRNA synthetases. J. Biol. Chem. 268, 15368–15373.

31. Sekine, S.-I., Shichiri, M., Bernier, S., Chênevert, R.,Lapointe, J. & Yokoyama, S. (2006). Structural bases oftransfer RNA-dependent amino acid recognition andactivation by glutamyl-tRNA synthetase. Structure,14, 1791–1799.

32. Hara-Yokoyama, M., Yokoyama, S. & Miyazawa, T.(1986). Conformational change of tRNAGlu in thecomplex with glutamyl-tRNA synthetase is requiredfor the specific binding of L-glutamate. Biochemistry,25, 7031–7036.

33. Banerjee, R., Dubois, D. Y., Gauthier, J., Lin, S.-X.,Roy, S. & Lapointe, J. (2004). The zinc-binding siteof a class I aminoacyl-tRNA synthetase is a SWIMdomain that modulates amino acid binding via thetRNA acceptor arm. Eur. J. Biochem. 271, 724–733.

34. Kern, D., Potier, S., Boulanger, Y. & Lapointe, J. (1979).The monomeric glutamyl-tRNA synthetase of Escher-ichia coli. J. Biol. Chem. 254, 518–524.

35. Kern, D. & Lapointe, J. (1980). The catalytic mechan-ism of the glutamyl-tRNA synthetase from Escherichiacoli. Detection of an intermediate complex in whichglutamate is activated. J. Biol. Chem. 255, 1956–1961.

36. Desjardins,M., Garnerau, S., Desgagnès, J., Lacoste, L.,Yang, F., Lapointe, J. et al. (1998). Glutamyl-adenylateanalogues are inhibitors of glutamyl-tRNA synthetase.Bioorg. Chem. 26, 1–13.

37. Sekine, S., Nureki, O., Dubois, D. Y., Bernier, S.,Chênevert, R., Lapointe, J. et al. (2003). ATP binding byglutamyl-tRNAsynthetase is switched to the productivemode by tRNA binding. EMBO J. 22, 676–688.

38. Giegé, R., Florentz, C., Kern, D., Gangloff, J., Eriani, G.& Moras, D. (1996). Aspartate identity of transferRNAs. Biochimie, 78, 605–623.

39. Martin, F., Eriani, G., Eiler, S., Moras, D., Dirheimer, G.& Gangloff, J. (1993). Overproduction and purifica-tion of native and queuosine-lacking Escherichia colitRNAAsp. Role of the wobble base in tRNAAsp

acylation. J. Mol. Biol. 234, 965–974.40. Nakanishi, S., Ueda, T., Hori, H., Yamazaki, N.,

Okada, N. & Watanabe, K. (1994). A UGU sequencein the anticodon-loop is a minimum requirement forrecognition by Escherichia coli tRNA-guanine trans-glycosylase. J. Biol. Chem. 269, 32221–32225.

41. Kern, D. & Lapointe, J. (1980). Catalytic mechanism ofglutamyl-tRNA synthetase from Escherichia coli. Reac-tion pathway in the aminoacylation of tRNAGlu.Biochemistry, 19, 3060–3068.

42. Sekine, S., Nureki, O., Sakamoto, K., Niimi, T., Tateno,M., Go, M. et al. (1996). Major identity determinants inthe ‘augmented D helix’ of tRNAGlu from Escherichiacoli. J. Mol. Biol. 256, 685–700.

43. Kern, D. & Lapointe, J. (1979). Glutamyl-transferribonucleic acid synthetase of Escherichia coli. Effect ofalteration of the 5-methylaminomethyl-2-thiouridinein the anticodon of glutamic acid transfer ribonucleicacid on the catalytic mechanism. Biochemistry, 18,5819–5826.

44. Madore, E., Florentz, C., Giegé, R., Sekine, S.,Yokoyama, S. & Lapointe, J. (1999). Effect of modifiednucleotides on Escherichia coli tRNAGlu structure andon its aminoacylation by glutamyl-tRNA synthetase.Eur. J. Biochem. 266, 1128–1135.

45. Loftfield, R. B. (1972). The mechanism of aminoacyla-tion of transfer RNA. Progress in Nucleic Acid Researchand Molecular Biology, vol. 12, pp. 87–128. AcademicPress, New York.

46. Schimmel, P. & Yang, X.-L. (2006). Perfecting thegenetic code with RNP complex. Structure, 14,1729–1730.

47. Otwinowskiu, Z. & Minor, W. (1996). Processing ofX-ray diffraction data collected in oscillation mode.Methods Enzymol. 276, 307–326.

48. Brünger, A. T., Adams, P. D., Clore, G. M., DeLano,W. L., Gros, P., Grosse-Kunstleve, R. W. et al. (1998).Crystallography and NMR system: a new softwaresuite for macromolecular structure determination.Acta Crystallogr. Sect. D, 54, 905–921.

49. Emsley, P. & Cowtan, K. (2004). Coot: model-buildingtools for molecular graphics. Acta Crystallogr. Sect. D,60, 2126–2132.

50. Laskowsky, R. A., Rullmann, J. A., MacArthur, M. W.,Kaptein, R. & Thornton, J. M. (1996). AQUA andPROCHEK-NMR: programs for checking the qualityof protein structures solved by NMR. J. Biomol. NMR,8, 477–486.

51. Kern, D. & Lapointe, J. (1979). The twenty aminoacyl-tRNA synthetases from Escherichia coli: generalseparation procedure and comparison of the influenceof pH and divalent cations on their catalytic activity.Biochimie, 61, 1257–1272.

52. Becker, H. D., Giegé, R. & Kern, D. (1996). Identity ofprokaryotic and eukaryotic tRNAAsp for aminoacyla-tion by aspartyl-tRNA synthetase from Thermusthermophilus. Biochemistry, 35, 7447–7458.

Crystal Structure of Glutamyl-Queuosine tRNAAsp Synthetase Complexed with l-Glutamate: Structural...

Documents

Transcript of Crystal Structure of Glutamyl-Queuosine tRNAAsp Synthetase Complexed with l-Glutamate: Structural...