A minimalist glutamyl-tRNA synthetase dedicated to aminoacylation of t
http://www.100md.com
《核酸研究医学期刊》
Département ‘Mécanismes et Macromolécules de la Synthèse Protéique et Cristallogenèse’, UPR 9002, Institut de Biologie Moléculaire et Cellulaire du CNRS, 15 Rue René Descartes, F-67084 Strasbourg Cedex, France, 1 Architecture et Fonction des Macromolécules Biologiques, UMR 6098 CNRS and Universités d’Aix-Marseille I et II, 31 Chemin J. Aiguier, F-13402 Marseille Cedex 20, France and 2 Département de Biochimie et Microbiologie, Faculté de Sciences et de Génie, CREFSIP, Université Laval, Québec, Canada G1K 7P4
*To whom correspondence should be addressed. Tel: +33 3 88 41 70 92; Fax: +33 3 88 60 22 18; Email: d.kern@ibmc.u-strasbg.fr
Correspondence may also be addressed to Richard Giegé. Tel: +33 3 88 70 58; Fax: +33 88 60 22 18; Email: r.giege@ibmc.u-strasbg.fr
ABSTRACT
Escherichia coli encodes YadB, a protein displaying 34% identity with the catalytic core of glutamyl-tRNA synthetase but lacking the anticodon-binding domain. We show that YadB is a tRNA modifying enzyme that evidently glutamylates the queuosine residue, a modified nucleoside at the wobble position of the tRNAAsp QUC anticodon. This conclusion is supported by a variety of biochemical data and by the inability of the enzyme to glutamylate tRNAAsp isolated from an E.coli tRNA-guanosine transglycosylase minus strain deprived of the capacity to exchange guanosine 34 with queuosine. Structural mimicry between the tRNAAsp anticodon stem and the tRNAGlu amino acid acceptor stem in prokaryotes encoding YadB proteins indicates that the function of these tRNA modifying enzymes, which we rename glutamyl-Q tRNAAsp synthetases, is conserved among prokaryotes.
INTRODUCTION
Aminoacyl-tRNA synthetases (aaRSs) constitute a family of essential and ubiquitous enzymes which synthesize 20 families of aminoacyl-tRNAs (aa-tRNAs) that are used by ribosomes for translation of mRNAs into proteins. These enzymes are modular proteins composed of functional units always containing a catalytic domain that activates amino acids before their transfer to tRNA and additional domains which include, in most cases, an anticodon-binding domain (1). As a consequence of this structural organization, the anticodon-binding domain alone binds the tRNA anticodon whereas the catalytic domain exclusively accommodates the accepting arm of tRNA and ensures aminoacylation exclusively at its 3'-end (2).
The accepted scheme of evolution of aaRSs proposes that the catalytic domains are derived from ancient enzymes that activated the amino acid and could transfer them to an acceptor RNA (3). The ability of minihelices derived from modern tRNAs (4,5), including tRNAAsp (6), to be aminoacylated supports this scheme. Both molecules co-evolved by fusion of additional modules. A two-domain L-shaped tRNA emerged from the primitive tRNA by addition of the anticodon stem–loop, while aaRSs evolved by acquiring additional modules, including the anticodon-binding domain, which enhanced specificity for binding of the cognate full-length tRNA and efficiency of its aminoacylation (7). Despite the strong functional interdependence existing between the catalytic core and the anticodon-binding domain, some aaRSs still display functional reminiscences of the ancestral enzyme, since they can aminoacylate acceptor minihelices. However, depriving a tRNA of its anticodon branch or an aaRS of its anticodon-binding domain has effects ranging from a change in specificity of the reaction to a severe drop in aminoacylation efficiency and, often, its abolition (6,8). With regard of these findings, it came as a surprise to see that in sequences emerging from whole genome sequencing projects, a vast number of prokaryotes display open reading frames encoding pieces and even modules of aaRSs (9). Among these truncated aaRSs, six are composed of only the catalytic core of an aaRS. Unexpected functions have been assigned to those which have been studied so far, such as involvement in amino acid metabolism (reviewed in 10) or regulation of translation (11). However, they are all incapable of performing tRNA aminoacylation (10,12).
Escherichia coli encodes a protein, YadB, sharing 34% identity with the catalytic core of glutamyl-tRNA synthetase (GluRS) but deprived of the anticodon-binding domain (13). Unexpectedly, this mini-GluRS is able to aminoacylate tRNA. In contrast to the other GluRSs, YadB activates Glu in the absence of tRNA and, more strikingly, does not transfer the activated Glu to tRNAGlu but to tRNAAsp (14). We here provide a body of evidence indicating that this truncated GluRS attaches the activated Glu to the queuine (Q base) in the wobble position of this tRNA. A comparison of the sequences of tRNAGlu and tRNAAsp and modeling of the YadB–tRNAAsp complex suggests that aminoacylation by YadB is mediated by a structural mimicry between the tRNAAsp anticodon stem and loop and the tRNAGlu acceptor helix.
MATERIALS AND METHODS
General
Thin layer chromatography (TLC) cellulose plates (20 x 20 cm2) and TSK HW-65 S were from Merck, hydroxyapatite CHT5 and UNO-Q1 columns from Bio-Rad and DEAE-cellulose DE52 from Whatman. L-Asp (207 mCi/mmol), L-Glu (238 mCi/mmol), L-Glu (43 Ci/mmol) and ATP (5000 Ci/mmol) were from Amersham. Sodium periodate and dithionitrobenzoate were from Sigma, unfractionated E.coli tRNA and nuclease P1 from Boehringer, venom phosphodiesterase (VPH) from Worthington, apyrase from Sigma, T4 polynucleotide kinase from USB Corp. and RNase T2 was prepared by a classical procedure (15). Escherichia coli YadB was purified as described (14) and E.coli AspRS was a gift from Dr G.Eriani (IBMC, Strasbourg). The E.coli SJ1505 strain tgt– was a gift from Drs K.Reuter and H.Kersten (Erlangen, Germany).
Purification of E.coli tRNAAspQ34 and tRNAAspG34
Escherichia coli strains overexpressing native tRNAAspQ34 and queuosine-lacking tRNAAspG34 were gifts from Drs F.Martin and G.Eriani (IBMC, Strasbourg) (16). The tRNAs were extracted from overexpressing cells after overnight culture, using phenol saturated with 50 mM sodium acetate buffer pH 5.0, 0.1 mM MgCl2 and 0.1 mM EDTA, followed by chloroform extraction and ethanol precipitation. Dried tRNA was dissolved in sodium acetate pH 5.0, adjusted to 2.5 M ammonium sulfate and precipitated on a 100 ml TSK HW-65 S column. Elution was performed with a linear gradient from 2.5 to 1.5 M ammonium sulfate in sodium acetate pH 5.0. The fractions containing tRNAAsp were pooled, adjusted to 2 M ammonium sulfate and loaded on a 2 ml DEAE-cellulose column before washing with 20 ml of water followed by elution with 1.5 M NaCl. tRNA, precipitated with ethanol and redissolved in 50 mM Tris–HCl pH 7.5 was loaded on a UNO-Q1 column equilibrated with the same buffer and eluted with a linear gradient from 0 to 1 M NaCl. Pure tRNAAspQ34 and tRNAAspG34 (accepting capacities 32 nmol/mg) were finally obtained by chromatography on a hydroxyapatite CHT-5 column and elution with a linear gradient from 10 to 200 mM potassium phosphate pH 6.8. Escherichia coli, Thermus thermophilus and yeast tRNAAsp transcripts were obtained as described (17).
Periodate oxidation of E.coli tRNA
One milligram of unfractionated E.coli tRNA was incubated for 20 min at room temperature in 1 ml of potassium phosphate buffer pH 6.8 containing 0.3 mM dithionitrobenzoate, precipitated with ethanol and treated for 30 min in the dark with 0.1 M sodium periodate in 0.25 ml of 50 mM sodium acetate buffer pH 5.0. After addition of 0.5 M KCl and 5% glycerol, the tRNA was ethanol precipitated, dialyzed, incubated for 3 h on ice in 0.3 ml of 0.1 M dithiothreitol, precipitated, redissolved in water and treated for 20 min with 0.1 M NaBH4 in 0.5 ml of water. The tRNA precipitate was dissolved in 0.l ml of water and dialyzed. In some experiments, tRNAAsp was aspartylated or glutamylated, respectively, by AspRS or YadB prior to periodate oxidation and after oxidation, then deacylated by incubation for 30 min at 37°C in 1.8 M Tris–HCl pH 8.0.
Snake venom phosphodiesterase digestion of E.coli tRNA
An aliquot of 0.1 mg of unfractionated tRNA was incubated for 20 min at room temperature with 10 μg snake venom phosphodiesterase in 0.5 ml of 50 mM Tris–HCl pH 8.0. After phenol and chloroform extraction, the tRNA was precipitated with ethanol and dissolved in water.
Aminoacylation reactions
Standard reaction mixtures of 50–250 μl for determination of plateaus or initial rates of aminoacylation contained 100 mM Na HEPES buffer pH 7.2, 30 mM KCl, 10 mM MgCl2, 25 μM L-Asp (270 c.p.m./pmol) or L-Glu (330 c.p.m./pmol), 0.1 mg/ml bovine serum albumin, 0.5 mg/ml unfractionated E.coli tRNA or 1–2 μM pure tRNAAspQ34 or tRNAAspG34 and appropriate concentrations of E.coli AspRS or YadB. The Km for tRNAAspQ34 was determined in the presence of 10 μM L-Glu (2000 c.p.m./pmol) and 0.03–0.3 μM tRNAAspQ34 and the Ki for tRNAAspG34 with the concentration fixed at 0.18 μM. Reactions were conducted at 37°C and the radiolabeled aa-tRNA formed after incubation times ranging from 1 to 40 min was determined in 20 μl aliquots (18).
Analysis of the nucleotide content of tRNAAspQ34 and tRNAAspG34
One microgram of tRNA was digested for 2 h at 37°C with 0.2 U RNase T2 in 3 μl of 50 mM sodium acetate buffer pH 4.5. The nucleotides were 5'-32P-labeled by incubation for 30 min at 37°C in the presence of 5 μCi (5 μl) ATP and 25 U T4 polynucleotide kinase. After 15 min treatment with 0.1 μg apyrase at 37°C, the 3'-phosphate of the nucleotides was dephosphorylated by incubation for 30 min at 37°C with 0.75 μg nuclease P1. One-third of the mix was analyzed by 2-dimensional TLC on cellulose plates developed with isobutyrate/ammonia/water (50/28/9) in the first dimension and HCl/isopropanol/water (15/70/15) in the second dimension and the nucleotides were revealed by autoradiography (19).
Docking of tRNAAsp into YadB
The coordinates of YadB, tRNAAsp and the GluRS-adenylate analog employed for docking are those deposited in the PDB, entries 1NZJ , 1EFW and 1N78 . The loop 224–238, missing in the electron density map of YadB, has been modeled from that of T.thermophilus GluRS (1N78 ). The adenylate analog (GoA) and position of loop 224–238 were kept as provided by the rigid body superposition of YadB with GluRS complexed to a Glu-adenylate analog (1N78 ). The docking of tRNAAsp into YadB was performed with the Turbo-Frodo program (20) by manual rigid body rotation/translation; only two exposed sidechain conformations were modified.
RESULTS AND DISCUSSION
YadB does not aminoacylate the 3' accepting end of tRNAAsp
We have previously shown that YadB aminoacylates pure E.coli tRNAAsp as efficiently as AspRS (14). Similar aminoacylation plateaux are obtained when pure tRNAAsp is charged by each enzyme (80% by AspRS and 75% by YadB; Fig. 1A). Surprisingly, aminoacylation with both enzymes results in a plateau of 155% of charged tRNA, suggesting that both Asp and Glu acylate tRNAAsp (Fig. 1A). The extent of total aminoacylation is independent of whether tRNA is charged first by AspRS or YadB. We confirmed that the two amino acids are attached to tRNA by TLC analysis after extraction of the aa-tRNA followed by hydrolysis of the ester bonds (Fig. 1B). Thus YadB can glutamylate an aspartylated tRNAAsp (Asp-tRNAAsp) and, conversely, AspRS can aspartylate a glutamylated tRNAAsp (Glu-tRNAAsp). An immediate interpretation could be that the two OH groups of the terminal adenosine are acylated with Asp and Glu, respectively, similarly to that reported for T.thermophilus PheRS able to catalyze the attachment of two Phe residues on the 3'-end of tRNAPhe (21). However, for structural considerations, we felt this interpretation unlikely and favored attachment of the Glu residue on another part of the molecule.
Figure 1. Analysis of the amino acid accepting capacity of E.coli tRNAAsp when aminoacylated by AspRS, YadB and by both. (A) Extents of pure tRNAAsp charging with E.coli AspRS (gray bar), YadB (white bar) or AspRS and YadB (black bar). (B) The amino acids acylating tRNAAsp are identified by TLC as described (10).
To verify the latter possibility, we treated unfractionated E.coli tRNA with snake venom phosphodiesterase that hydrolyzes single-stranded RNAs starting from the 3'-OH end and thus removes the CCA end and the discriminator base of the tRNA molecules (Fig. 2A and B). As expected, E.coli tRNAAsp lost its ability to be aspartylated by AspRS, but fully retained its capacity to be glutamylated by YadB. (Fig. 2A and B).
Figure 2. Protection of tRNAAsp amino acid accepting capacity by E.coli AspRS or YadB catalyzed aminoacylation prior to periodate oxidation. (A) Extent of tRNAAsp aspartylation by AspRS (gray bars) and glutamylation by YadB (white bars). (B) The same experiment with tRNAAsp lacking its single stranded 3' terminus. (C) Same experiments with tRNAAsp treated with periodate.
So far, all known aaRSs aminoacylate their cognate tRNA by forming an ester bond between the activated -COOH group of the amino acid and the 2'- or 3'-OH group of the ribose cis-diol of the tRNA terminal adenosine (2,22). Periodate oxidation by opening the ribose ring deprives tRNA of its accepting capacity. However, aminoacylation protects the cis-diol of the terminal adenosine against oxidation and thus protects the tRNA against inactivation (23). Therefore, to test whether glutamylation of tRNAAsp by YadB proceeds by esterification of a OH group, we oxidized unprotected and protected E.coli tRNAAsp with periodate and analyzed their remaining accepting capacities. Figure 2C shows that YadB, like AspRS, is unable to charge periodate oxidized tRNAAsp. However, tRNAAsp protected by glutamylation with YadB prior to periodate oxidation is still aminoacylated by YadB, but not by AspRS. Likewise, tRNAAsp aspartylated by AspRS prior to periodate oxidation retains its aspartylation ability, but is no longer glutamylatable by YadB (not shown).
Altogether, the obtained data demonstrates that aminoacylation by YadB does not occur at the terminal adenosine of tRNAAsp and should take place at another place on the tRNAAsp molecule. This conclusion is supported by a previous result showing that the mischarged tRNAAsp cannot bind the elongation factor EF-Tu (14), despite the capability of a canonical Glu-tRNAAsp, i.e. when charged on the 3'-end aa-tRNA, to bind the factor (24).
YadB aminoacylates the queuosine located in the first position of tRNAAsp anticodon
Analysis of the chemical structure of the modified nucleosides present in E.coli tRNAAsp (25) reveals that the queuosine (Q) in the first position of the anticodon (position 34 of tRNA) possesses a cyclopentenyldiol ring containing a cis-diol that could possibly mimic the 2'- and 3'-OH groups of the ribose of the terminal adenosine (26–28) (Fig. 3A). Introduction of queuosine into tRNA is achieved by a complex cascade of enzymatic reactions involving tRNA-guanine transglycosylase (TGT), which results by its exchange with G34 (29).
Figure 3. Amino acid accepting capacities of E.coli native tRNAAspQ34 and queuosine-lacking tRNAAspG34. (A) Kinetics of aminoacylation of tRNAAspQ34 (top) and tRNAAspG34 (bottom) by E.coli AspRS (black triangles) and YadB (black squares). (Insets) 2-Dimensional TLC analysis of the 5' phosphate nucleosides present in the two tRNAs (the arrows indicate the positions where pQ migrates). The structures of Q and G are shown at the left with the arrows indicating the OH groups of the cyclopentenyldiol ring that are the putative Glu acceptors on Q. (B) Lineweaver–Burk plots allowing determinations of the Km (open triangles) of YadB for native tRNAAsp and Ki (open squares) of tRNAAspQ34 for aminoacylation of native tRNAAsp by YadB.
To check whether glutamylation of tRNAAsp by YadB is dependent on the presence of queuosine we first attempted to glutamylate the E.coli tRNAAsp transcript deprived of modified nucleotides. Table 1 shows that pure E.coli tRNAAsp is fully charged by YadB, whereas the transcript, deprived of post-transcriptional modifications but instead containing G34, is not charged by YadB. Likewise, Saccharomyces cerevisiae and T.thermophilus tRNAAsp deprived of Q (30,31) are fully aspartylated but not charged by YadB.
Table 1. Aspartylation and glutamylation of tRNAAsp of various origins possessing either G34 or Q34
To unambiguously prove that YadB aminoacylates the queuosine of E.coli tRNAAsp, we purified a variant of it containing all modified nucleosides except Q34 (tRNAAspG34) and compared its glutamylation capacity with that of the wild-type tRNAAsp (tRNAAspQ34). tRNAAspG34 was purified from an E.coli tgt– strain, unable to introduce Q into tRNAs. As expected, both tRNAs are completely aspartylated by E.coli AspRS (Fig. 3A). However, while YadB fully aminoacylated tRNAAspQ34, this enzyme was unable to glutamylate tRNAAspG34 (Fig. 3A). As confirmed by 2-dimensional TLC analysis of tRNA hydrolysates, the two tRNA variants differ only in the absence of Q in tRNAAspG34 (Fig. 3A, insets). A further confirmation that Q34 is the acceptor of the activated Glu was obtained by inhibition of glutamylation of wild-type tRNAAsp in the presence of tRNAAspG34. Figure 3B shows that tRNAAspG34 competitively inhibits charging of tRNAAspQ34 by YadB, which displays the same affinity for both tRNAs, as indicated by the kinetic constants (Km for tRNAAspQ34 and Ki for tRNAAspG34 of 0.19 μM). Since both tRNAs bind with the same affinity to YadB, whereas the protein charges only the species containing Q, it is concluded that glutamylation by YadB occurs on Q.
What are the structural elements that promote aminoacylation of the E.coli tRNAAsp anticodon by YadB?
In E.coli, tRNAAsp, tRNAAsn, tRNAHis and tRNATyr contain Q in the wobble position of their anticodons (32), yet only Q34 of tRNAAsp is glutamylated by YadB (14). This indicates that Q is not solely responsible for the recognition of tRNAAsp. In fact, as shown above, it does not contribute to the affinity of tRNAAsp for YadB.
To identify potential determinants responsible for the interaction of tRNAAsp with YadB, we compared the tRNAAsp and tRNAGlu sequences from E.coli. This comparison revealed a striking mimicry between the tRNAAsp anticodon stem and loop and the tRNAGlu acceptor arm (Fig. 4A). The first four base pairs of the tRNAGlu acceptor arm are identical to the first four base pairs of the tRNAAsp anticodon stem, while the discriminator G73 residue and C74 of tRNAGlu align with G39 from the tRNAAsp anticodon stem and C38, the last nucleotide of the anticodon loop, respectively.
Figure 4. Sequence comparison of tRNAGlu acceptor arm and tRNAAsp anticodon stem and loop from bacteria encoding YadB and expansion of the comparison to tRNAAsn, tRNAHis and tRNATyr anticodon stems and loops. tRNAs from 17 bacterial species were compared. (A) Escherichia coli tRNAAsp and tRNAGlu are presented in a 2-dimensional projection of their L-shaped architecture. Nucleotides targeted by the comparison are displayed and the amino acid-accepting nucleotides (Q34 and A76) indicated by an arrow. Numbering of nucleotides is according to Sprinzl et al. (32). For each bacterial species tRNAAsp anticodon stem and loop positions (base pairs 27–43 to 30–40 and nucleotides 39 and 38) were compared to the corresponding positions of the tRNAGlu acceptor stem (base pairs 4–69 to 1–72 and nucleotides 73 and 74) from the same species. When nucleotides are the same as in tRNAAsp and tRNAGlu from E.coli tRNAs, they are displayed in white on a black box. For clarity, only nucleotides 43–38 from tRNAAsp and 69–74 from tRNAGlu are displayed. Consensus sequences of the 38–43 stretches in tRNAAsp and 74–68 stretches in tRNAGlu are given together with the consensus of the complementary residues (30–27 and 1–4 stretches). Residues in italic are partially conserved; R stands for purine and Y for pyrimidine. The sequences displayed are ranked according to the YadB classification (14). (B) 2-Dimensional representation of E.coli tRNAAsp and the consensus sequences in the anticodon branch of the 17 bacterial species. When a nucleotide is the same as its counterpart from the consensus sequence of tRNAGlu it is displayed in white on a black box.
Expanding the comparative analysis to all tRNAGlu and tRNAAsp species of bacteria encoding YadB genes, we observed partial conservation of the sequence mimicry (Fig. 4A). Interestingly, the sequence conservation in the two tRNAs follows the phylogenetic classification of the YadB proteins in three structure-based subgroups (14). Thus complete mimicry between the two tRNAs is found in the two -proteobacteria, Yersinia pestis and Salmonella enteritica, that are most closely related to E.coli with respect to their YadB proteins (Fig. 4A). Mimicry between tRNAAsp and tRNAGlu, as found in E.coli, becomes looser in bacteria, where the divergence between YadB proteins increases, but interestingly high mimicry exists between tRNAs of organisms with YadB proteins from the same subgroup.
Altogether, this comparative analysis indicates that four nucleotides from the 5'-side of the anticodon stem–loop of the tRNAAsp species are conserved (namely C38 from the anticodon loop as well as G39, G42 and G43 from the anticodon stem) and three are present in a 6 nt sequence stretch from the 3'-terminus of the corresponding tRNAGlu species (namely single-stranded C74 from the CCA terminus and discriminator residues G73 and G69). This sequence mimicry with the anticodon branch of tRNAGlu is never found in tRNAAsn, tRNAHis and tRNATyr that possess Q in position 34 but in which G39 and C38 are always replaced by U39 and A38 (Fig. 4B). Likewise, no such mimicry is found between eukaryotic tRNAAsp and tRNAGlu species. In conclusion, this suggests that those residues from the consensus highlighted in Figure 3, especially C38, G39, G42 and G43, are critical for recognition of tRNAAsp by YadB and, conversely, their non-conservation in tRNAAsn, tRNAHis and tRNATyr contribute to the rejection of these Q-containing tRNAs by YadB proteins.
To gain a deeper insight into the structural elements that could promote tRNAAsp anticodon recognition and aminoacylation by YadB, the minimal GluRS, we constructed a docking model of E.coli YadB and E.coli tRNAAsp (Fig. 5). By aligning the two similar helical parts of the tRNAAsp anticodon stem (residues 24–45) and tRNAGlu acceptor stem (without 73GCCA76) we observed that in tRNAAsp Q34 protrudes in a loop, while in tRNAGlu the GCCA stem residues fold back in the opposite direction. As a consequence, the amino acid accepting OH groups of tRNAAsp (Q34) and tRNAGlu (A76) are presented in different conformations. Therefore, simple docking of tRNAAsp to YadB, without structural deformation of the anticodon loop, is not compatible with ‘classical’ tRNA binding in the catalytic domain of a class I synthetase. However, the tRNA could be fitted with only the 30–38 segment in the groove (Fig. 5). This position is correct with regard to shape complementarity and to reactive group closeness (Fig. 5, inset). Note that the interacting surface between YadB and tRNAAsp is quite large in the tentative docking model: 1100 ?2 of YadB (10% of the total surface) are shielded from solvent in the complex. In this model the interacting protein surface is slightly reduced as compared to the complexes involving class I GluRS in which 1600 ?2 are shielded from the solvent (33).
Figure 5. A tentative model of the complex of E.coli tRNAAsp with cognate YadB. (A) The YadB molecular surface is displayed in dark blue, the loop in lighter blue and adenylate in sphere representation. The docking procedure was based on the constraint that the reactive group, the cyclopentenyldiol ring of Q34, like the ribose from A76, should be close to the Glu–adenylate phosphate–Glu scissile bond. The Q34 cyclopentenyldiol ring was therefore used as a pivot around which the tRNA rigid body moved. The tRNAAsp is displayed as an atom color coded molecular surface. (Inset) Close-up of the interaction illustrating the vicinity of the adenylate donor molecule and the acceptor group, Q34 of tRNAAsp. Adenylate and the tRNAAsp domain with Q base are displayed in stick representation. (B) Rotation of 120° around a vertical axis and 30° clockwise in the plane.
Functional and evolutionary considerations
Data reported here break down the current paradigm underlying all the research on tRNA recognition and aminoacylation by aaRSs, namely that the catalytic domain of synthetases recognizes the amino acid branch of tRNAs and attaches the activated amino acid via an ester bond to their 3'-CCAOH termini. All the literature on structural and functional biology in the field has been exclusively based on this paradigm. Therefore, it was a great surprise to discover that a protein closely mimicking the catalytic domain of E.coli GluRS, namely the YadB protein from E.coli, recognizes the anticodon branch of a tRNA and attaches an amino acid on a modified base from the tRNA anticodon. Interestingly, YadB has the ability to activate Glu (14), although without the assistance of tRNAGlu, as is the case of canonical GluRSs (34), but also the unexpected ability to interact with the distal extremity of E.coli tRNAAsp and to acylate the cyclopentenyldiol ring of the Q base post-transcriptionnally inserted at position 34 of the anticodon. This ring, with two cis-diol residues, mimics the structure of the 3'-terminal ribose of tRNA and thus Glu is attached to the Q base via an unstable ester bond (half-life 7.5 min) (14). This explains why the presence of a glutamylated Q base (which can be abbreviated as Glu-Q) in E.coli has escaped identification up to now. Docking of the tRNAAsp anticodon branch on YadB (Fig. 5) suggests that this branch has physical contact with the protein, a conclusion in agreement with the competition of tRNAAsp deprived of Q for glutamylation of native tRNAAsp with Q34 (Fig. 3B). This a priori unpredicted biological function of YadB as a tRNA modifying enzyme recognizing the anticodon stem and loop of tRNA can be generalized to other YadB-containing microorganisms, as supported by sequence comparison of their tRNAAsp and tRNAGlu (Fig. 4). Therefore, YadB proteins can be renamed glutamyl-Q tRNAAsp synthetases (Glu-Q-RSs).
From another perspective, one can question the evolutionary origin of the YadB proteins. Whether an ancestral YadB protein was a precursor of modern prokaryotic GluRSs or whether the YadB family originates from an ancestral GluRS remains an open question. However, their contemporary biological function has likely a recent evolutionary origin, given the reaction that YadB enzymes catalyze. Because the presence of YadB proteins is restricted to prokaryotic organisms, glutamylation of the Q base in tRNAAsp, which only occurs in such organisms, is intriguing, since Q base is also found in eukaryotes. However, hypermodified Q residues have been detected in higher eukaryotes, with a mannosyl or galactosyl residue instead of a Glu fixed on the cyclopentenyldiol ring. Note also that the mannosyl and galactosyl moieties, like the ribose and the cyclopentenyldiol ring, possess two adjacent cis-diols that could also be potential amino acid acceptors via catalysis by aaRS-mimicking proteins. These observations raise interesting questions about the structure of the enzymes attaching mannose or galactose to Q bases and about the possible existence of proteins that may esterify these sugars. Like YadB, such proteins could be aaRS-mimicking proteins.
To conclude, let us comment on tRNA structure in the light of the new paradigm and of concepts underlying the origin of tRNA and the genetic code. Here it is shown that a mimic of an aaRS catalytic domain makes no fundamental difference in recognizing a tRNA anticodon branch instead of a tRNA amino acid accepting branch. This property of YadB is in line with the idea that contemporary tRNA originated by fusion of two primordial minihelix or stem–loop structures (3). Likewise, primordial aaRSs were simplified structures restricted to their catalytic core, existing in nature as two versions (the class I and the class II versions), that became more complex in later evolution by the addition of extra domains, the most important being the anticodon-binding domain (1). In the course of such structural tinkering it is likely that primitive aaRSs were able to recognize a variety of RNA architectures with tRNA features. This potential recognition was maintained in some modern proteins as a remnant of their early evolution. YadB (Glu-Q-RS) is such a protein. We anticipate that other aaRS mimics remain to be discovered and believe that such proteins will use their catalytic and RNA recognition potential in participating in metabolic pathways as yet undeciphered.
ACKNOWLEDGEMENTS
This work was supported by Centre National de la Recherche Scientifique (CNRS), Association de la Recherche contre le Cancer (ARC), Université Louis Pasteur (Strasbourg), the French Ministry of Industry (grant ASG) and Marseille Genopole. Additional support came by grant OG0009597 from the Natural Sciences and Engineering Research Council of Canada (NSERC). M.B. is the recipient of a fellowship from Ministère de l’Education Nationale de la Recherche et de la Technologie.
REFERENCES
Ibba,M., Francklyn,C. and Cusack,S. (eds) (2004) The Aminoacyl-tRNA Synthetases, Landes Bioscience, Georgetown, TX.
S?ll,D. and Schimmel,P.R. (1974) Aminoacyl-tRNA synthetases. Enzyme, 10, 489–538.
Schimmel,P., Giegé,R., Moras,D. and Yokoyama,S. (1993) An operational RNA code for amino acids and posible relationship to genetic code. Proc. Natl Acad. Sci. USA, 90, 8763–8768.
Francklyn,C. and Schimmel,P. (1989) Aminoacylation of RNA minihelices with alanine. Nature, 337, 478–481.
Frugier,M., Florentz,C. and Giegé,R. (1992) Anticodon-independent aminoacylation of an RNA minihelix with valine. Proc. Natl Acad. Sci. USA, 89, 3990–3994.
Frugier,M., Florentz,C. and Giegé,R. (1994) Efficient aminoacylation of resected RNA helices by class II aspartyl-tRNA synthetase dependent on a single nucleotide. EMBO J., 13, 2219–2226.
Schimmel,P. and Ribas de Pouplana,L. (1995) Transfer RNA: from minihelix to genetic code. Cell, 81, 983–986.
Schwob,E. and S?ll,D. (1993) Selection of a ‘minimal’ glutaminyl-tRNA synthetase and the evolution of class I synthetases. EMBO J., 12, 5201–5208.
Schimmel,P. and Ribas De Pouplana,L. (2000) Footprints of aminoacyl-tRNA synthetases are everywhere. Trends Biochem. Sci., 25, 207–209.
Roy,H., Becker,H.D., Reinbolt,J. and Kern,D. (2003) When contemporary aminoacyl-tRNA synthetases invent their cognate amino acid metabolism. Proc. Natl Acad. Sci. USA, 100, 9837–9842.
Kaniga,K., Compton,M.S., Curtiss,R. and Sundaram,P. (1998) Molecular and functional characterization of Salmonella enterica serovar typhimurium poxA gene: effect on attenuation of virulence and protection. Infect. Immun., 66, 5599–5606.
Sissler,M., Delorme,C., Bond,J., Ehrlich,S.D., Renault,P. and Francklyn,C. (1999) An aminoacyl-tRNA synthetase paralog with a catalytic role in histidine biosynthesis. Proc. Natl Acad. Sci. USA, 96, 8985–8990.
Campanacci,V., Dubois,D.Y., Becker,H.D., Kern,D., Spinelli,S., Valencia,C., Pagot,F., Salomoni,A., Grisel,S., Vincentelli,R. et al. (2004) The Escherichia coli YadB gene product reveals a novel aminoacyl-tRNA synthetase like activity. J. Mol. Biol., 337, 273–283.
Dubois,D.Y., Blaise,M., Becker,H.D., Campanacci,V., Keith,G., Cambillau,C., Giegé,R., Lapointe,J. and Kern,D. (2004) An aminoacyl-tRNA synthetase-like protein encoded by the Escherichia coli yadB gene glutamylates specifically tRNA(Asp). Proc. Natl Acad. Sci. USA, in press.
Hiramaru,M., Uschida,T. and Egami,F. (1966) Ribonuclease preparation for the base analysis of polyribonucleotides. Anal. Biochem., 17, 135–142.
Martin,F., Eriani,G., Eiler,S., Moras,D., Dirheimer,G. and Gangloff,J. (1993) Overproduction and purification of native and queuine-lacking Escherichia coli tRNA(Asp). Role of the wobble base in tRNA(Asp) acylation. J. Mol. Biol., 234, 965–974.
Becker,H.D., Giegé,R. and Kern,D. (1996) Identity of prokaryotic and eukaryotic tRNA(Asp) for aminoacylation by aspartyl-tRNA synthetase from T. thermophilus. Biochemistry, 35, 7447–7458.
Kern,D. and 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, 1257–1272.
Keith,G. (1994) Nucleic acid chromatographic isolation and sequence methods. In Gehrke,C.W. and Kuo,K.C. (eds), Chromatography and Modifications of Nucleosides. Elsevier, Amsterdam, Vol. 45A, pp. A103–A141.
Roussel,A. and Cambillau,C. (1991) The TURBO-FRODO graphics package. Silicon Graphics, 81, 77–78.
Stepanov,V.G., Moor,N.A., Ankilova,V.N. and Lavrik,O.L. (1992) Phenylalanyl-tRNA synthetase from Thermus thermophilus can attach two molecules of phenylalanine to tRNA(Phe). FEBS Lett., 311, 192–194.
Chladek,S. and Sprinzl,M. (1985) The 3'-end of tRNA and its role in protein biosynthesis. Angew. Chem. Int. Ed. Engl., 24, 371–391.
Kern,D., Giegé,R. and Ebel,J.-P. (1972) Incorrect aminoacylation catalyzed by the phenylalanyl- and valyl-tRNA synthetases from yeast. Eur. J. Biochem., 31, 148–155.
Asahara,H. and Uhlenbeck,O.C. (2002) The tRNA specificity of Thermus thermophilus EF-Tu. Proc. Natl Acad. Sci. USA, 99, 3499–3504.
Sekiya,T., Mori,M., Takahashi,N. and Nishimura,S. (1980) Sequence of the distal tRNA(Asp)1 gene and the transcription termination signal in the Escherichia coli ribosomal RNA operon rrnF (or G). Nucleic Acids Res., 8, 3809–3827.
Kersten,H. and Kersten,W. (1990) Biosynthesis and function of queuine and queuosine. In Gehrke,C. and Wand Kuo,K.C.T. (eds), Chromatography and Modification of Nucleosides, PartB: Biological Roles and Functions of Modification. Elsevier, Amsterdam.
Yokoyama,S., Miyazawa,T., Iitaka,Y., Yamaizumi,Z., Kasai,H. and Nishimura,S. (1979) Three-dimensional structure of hyper-modified nucleoside Q located in the wobbling position of tRNA. Nature, 282, 107–109.
Grosjean,H. and Benne,R. (eds.) (1998) Modification and Editing of tRNA. ASM Press, Washington, DC.
Reader,J.S., Metzgar,D., Schimmel,P. and de Crecy-Lagard,V. (2004) Identification of four genes necessary for biosynthesis of the modified nucleoside queuosine. J. Biol. Chem., 279, 6280–6285.
Gangloff,J., Keith,G., Ebel,J.P. and Dirheimer,G. (1971) Structure of aspartate tRNA from brewer’s yeast. Nature New Biol., 230, 125–127.
Keith,G., Yusupov,M., Briand,C., Moras,D. and Kern,D. (1993) Sequence of tRNA(Asp) from Thermus thermophilus HB8. Nucleic Acids Res., 21, 3216–3230.
Sprinzl,M., Horn,C., Brown,M., Ioudovitch,A. and Steinberg,S. (1998) Compilation of tRNA sequences and sequences of tRNA genes. Nucleic Acids Res., 26, 148–153.
Rould,M.A., Perona,J.J., S?ll,D. and Steitz,T.A. (1989) Structure of E. coli glutaminyl-tRNA synthetase complexed with tRNA(Gln) and ATP at 2.8 ? resolution. Science, 246, 1135–1142.
Kern,D. and 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.(Micka?l Blaise, Hubert Dominique Becker,)
*To whom correspondence should be addressed. Tel: +33 3 88 41 70 92; Fax: +33 3 88 60 22 18; Email: d.kern@ibmc.u-strasbg.fr
Correspondence may also be addressed to Richard Giegé. Tel: +33 3 88 70 58; Fax: +33 88 60 22 18; Email: r.giege@ibmc.u-strasbg.fr
ABSTRACT
Escherichia coli encodes YadB, a protein displaying 34% identity with the catalytic core of glutamyl-tRNA synthetase but lacking the anticodon-binding domain. We show that YadB is a tRNA modifying enzyme that evidently glutamylates the queuosine residue, a modified nucleoside at the wobble position of the tRNAAsp QUC anticodon. This conclusion is supported by a variety of biochemical data and by the inability of the enzyme to glutamylate tRNAAsp isolated from an E.coli tRNA-guanosine transglycosylase minus strain deprived of the capacity to exchange guanosine 34 with queuosine. Structural mimicry between the tRNAAsp anticodon stem and the tRNAGlu amino acid acceptor stem in prokaryotes encoding YadB proteins indicates that the function of these tRNA modifying enzymes, which we rename glutamyl-Q tRNAAsp synthetases, is conserved among prokaryotes.
INTRODUCTION
Aminoacyl-tRNA synthetases (aaRSs) constitute a family of essential and ubiquitous enzymes which synthesize 20 families of aminoacyl-tRNAs (aa-tRNAs) that are used by ribosomes for translation of mRNAs into proteins. These enzymes are modular proteins composed of functional units always containing a catalytic domain that activates amino acids before their transfer to tRNA and additional domains which include, in most cases, an anticodon-binding domain (1). As a consequence of this structural organization, the anticodon-binding domain alone binds the tRNA anticodon whereas the catalytic domain exclusively accommodates the accepting arm of tRNA and ensures aminoacylation exclusively at its 3'-end (2).
The accepted scheme of evolution of aaRSs proposes that the catalytic domains are derived from ancient enzymes that activated the amino acid and could transfer them to an acceptor RNA (3). The ability of minihelices derived from modern tRNAs (4,5), including tRNAAsp (6), to be aminoacylated supports this scheme. Both molecules co-evolved by fusion of additional modules. A two-domain L-shaped tRNA emerged from the primitive tRNA by addition of the anticodon stem–loop, while aaRSs evolved by acquiring additional modules, including the anticodon-binding domain, which enhanced specificity for binding of the cognate full-length tRNA and efficiency of its aminoacylation (7). Despite the strong functional interdependence existing between the catalytic core and the anticodon-binding domain, some aaRSs still display functional reminiscences of the ancestral enzyme, since they can aminoacylate acceptor minihelices. However, depriving a tRNA of its anticodon branch or an aaRS of its anticodon-binding domain has effects ranging from a change in specificity of the reaction to a severe drop in aminoacylation efficiency and, often, its abolition (6,8). With regard of these findings, it came as a surprise to see that in sequences emerging from whole genome sequencing projects, a vast number of prokaryotes display open reading frames encoding pieces and even modules of aaRSs (9). Among these truncated aaRSs, six are composed of only the catalytic core of an aaRS. Unexpected functions have been assigned to those which have been studied so far, such as involvement in amino acid metabolism (reviewed in 10) or regulation of translation (11). However, they are all incapable of performing tRNA aminoacylation (10,12).
Escherichia coli encodes a protein, YadB, sharing 34% identity with the catalytic core of glutamyl-tRNA synthetase (GluRS) but deprived of the anticodon-binding domain (13). Unexpectedly, this mini-GluRS is able to aminoacylate tRNA. In contrast to the other GluRSs, YadB activates Glu in the absence of tRNA and, more strikingly, does not transfer the activated Glu to tRNAGlu but to tRNAAsp (14). We here provide a body of evidence indicating that this truncated GluRS attaches the activated Glu to the queuine (Q base) in the wobble position of this tRNA. A comparison of the sequences of tRNAGlu and tRNAAsp and modeling of the YadB–tRNAAsp complex suggests that aminoacylation by YadB is mediated by a structural mimicry between the tRNAAsp anticodon stem and loop and the tRNAGlu acceptor helix.
MATERIALS AND METHODS
General
Thin layer chromatography (TLC) cellulose plates (20 x 20 cm2) and TSK HW-65 S were from Merck, hydroxyapatite CHT5 and UNO-Q1 columns from Bio-Rad and DEAE-cellulose DE52 from Whatman. L-Asp (207 mCi/mmol), L-Glu (238 mCi/mmol), L-Glu (43 Ci/mmol) and ATP (5000 Ci/mmol) were from Amersham. Sodium periodate and dithionitrobenzoate were from Sigma, unfractionated E.coli tRNA and nuclease P1 from Boehringer, venom phosphodiesterase (VPH) from Worthington, apyrase from Sigma, T4 polynucleotide kinase from USB Corp. and RNase T2 was prepared by a classical procedure (15). Escherichia coli YadB was purified as described (14) and E.coli AspRS was a gift from Dr G.Eriani (IBMC, Strasbourg). The E.coli SJ1505 strain tgt– was a gift from Drs K.Reuter and H.Kersten (Erlangen, Germany).
Purification of E.coli tRNAAspQ34 and tRNAAspG34
Escherichia coli strains overexpressing native tRNAAspQ34 and queuosine-lacking tRNAAspG34 were gifts from Drs F.Martin and G.Eriani (IBMC, Strasbourg) (16). The tRNAs were extracted from overexpressing cells after overnight culture, using phenol saturated with 50 mM sodium acetate buffer pH 5.0, 0.1 mM MgCl2 and 0.1 mM EDTA, followed by chloroform extraction and ethanol precipitation. Dried tRNA was dissolved in sodium acetate pH 5.0, adjusted to 2.5 M ammonium sulfate and precipitated on a 100 ml TSK HW-65 S column. Elution was performed with a linear gradient from 2.5 to 1.5 M ammonium sulfate in sodium acetate pH 5.0. The fractions containing tRNAAsp were pooled, adjusted to 2 M ammonium sulfate and loaded on a 2 ml DEAE-cellulose column before washing with 20 ml of water followed by elution with 1.5 M NaCl. tRNA, precipitated with ethanol and redissolved in 50 mM Tris–HCl pH 7.5 was loaded on a UNO-Q1 column equilibrated with the same buffer and eluted with a linear gradient from 0 to 1 M NaCl. Pure tRNAAspQ34 and tRNAAspG34 (accepting capacities 32 nmol/mg) were finally obtained by chromatography on a hydroxyapatite CHT-5 column and elution with a linear gradient from 10 to 200 mM potassium phosphate pH 6.8. Escherichia coli, Thermus thermophilus and yeast tRNAAsp transcripts were obtained as described (17).
Periodate oxidation of E.coli tRNA
One milligram of unfractionated E.coli tRNA was incubated for 20 min at room temperature in 1 ml of potassium phosphate buffer pH 6.8 containing 0.3 mM dithionitrobenzoate, precipitated with ethanol and treated for 30 min in the dark with 0.1 M sodium periodate in 0.25 ml of 50 mM sodium acetate buffer pH 5.0. After addition of 0.5 M KCl and 5% glycerol, the tRNA was ethanol precipitated, dialyzed, incubated for 3 h on ice in 0.3 ml of 0.1 M dithiothreitol, precipitated, redissolved in water and treated for 20 min with 0.1 M NaBH4 in 0.5 ml of water. The tRNA precipitate was dissolved in 0.l ml of water and dialyzed. In some experiments, tRNAAsp was aspartylated or glutamylated, respectively, by AspRS or YadB prior to periodate oxidation and after oxidation, then deacylated by incubation for 30 min at 37°C in 1.8 M Tris–HCl pH 8.0.
Snake venom phosphodiesterase digestion of E.coli tRNA
An aliquot of 0.1 mg of unfractionated tRNA was incubated for 20 min at room temperature with 10 μg snake venom phosphodiesterase in 0.5 ml of 50 mM Tris–HCl pH 8.0. After phenol and chloroform extraction, the tRNA was precipitated with ethanol and dissolved in water.
Aminoacylation reactions
Standard reaction mixtures of 50–250 μl for determination of plateaus or initial rates of aminoacylation contained 100 mM Na HEPES buffer pH 7.2, 30 mM KCl, 10 mM MgCl2, 25 μM L-Asp (270 c.p.m./pmol) or L-Glu (330 c.p.m./pmol), 0.1 mg/ml bovine serum albumin, 0.5 mg/ml unfractionated E.coli tRNA or 1–2 μM pure tRNAAspQ34 or tRNAAspG34 and appropriate concentrations of E.coli AspRS or YadB. The Km for tRNAAspQ34 was determined in the presence of 10 μM L-Glu (2000 c.p.m./pmol) and 0.03–0.3 μM tRNAAspQ34 and the Ki for tRNAAspG34 with the concentration fixed at 0.18 μM. Reactions were conducted at 37°C and the radiolabeled aa-tRNA formed after incubation times ranging from 1 to 40 min was determined in 20 μl aliquots (18).
Analysis of the nucleotide content of tRNAAspQ34 and tRNAAspG34
One microgram of tRNA was digested for 2 h at 37°C with 0.2 U RNase T2 in 3 μl of 50 mM sodium acetate buffer pH 4.5. The nucleotides were 5'-32P-labeled by incubation for 30 min at 37°C in the presence of 5 μCi (5 μl) ATP and 25 U T4 polynucleotide kinase. After 15 min treatment with 0.1 μg apyrase at 37°C, the 3'-phosphate of the nucleotides was dephosphorylated by incubation for 30 min at 37°C with 0.75 μg nuclease P1. One-third of the mix was analyzed by 2-dimensional TLC on cellulose plates developed with isobutyrate/ammonia/water (50/28/9) in the first dimension and HCl/isopropanol/water (15/70/15) in the second dimension and the nucleotides were revealed by autoradiography (19).
Docking of tRNAAsp into YadB
The coordinates of YadB, tRNAAsp and the GluRS-adenylate analog employed for docking are those deposited in the PDB, entries 1NZJ , 1EFW and 1N78 . The loop 224–238, missing in the electron density map of YadB, has been modeled from that of T.thermophilus GluRS (1N78 ). The adenylate analog (GoA) and position of loop 224–238 were kept as provided by the rigid body superposition of YadB with GluRS complexed to a Glu-adenylate analog (1N78 ). The docking of tRNAAsp into YadB was performed with the Turbo-Frodo program (20) by manual rigid body rotation/translation; only two exposed sidechain conformations were modified.
RESULTS AND DISCUSSION
YadB does not aminoacylate the 3' accepting end of tRNAAsp
We have previously shown that YadB aminoacylates pure E.coli tRNAAsp as efficiently as AspRS (14). Similar aminoacylation plateaux are obtained when pure tRNAAsp is charged by each enzyme (80% by AspRS and 75% by YadB; Fig. 1A). Surprisingly, aminoacylation with both enzymes results in a plateau of 155% of charged tRNA, suggesting that both Asp and Glu acylate tRNAAsp (Fig. 1A). The extent of total aminoacylation is independent of whether tRNA is charged first by AspRS or YadB. We confirmed that the two amino acids are attached to tRNA by TLC analysis after extraction of the aa-tRNA followed by hydrolysis of the ester bonds (Fig. 1B). Thus YadB can glutamylate an aspartylated tRNAAsp (Asp-tRNAAsp) and, conversely, AspRS can aspartylate a glutamylated tRNAAsp (Glu-tRNAAsp). An immediate interpretation could be that the two OH groups of the terminal adenosine are acylated with Asp and Glu, respectively, similarly to that reported for T.thermophilus PheRS able to catalyze the attachment of two Phe residues on the 3'-end of tRNAPhe (21). However, for structural considerations, we felt this interpretation unlikely and favored attachment of the Glu residue on another part of the molecule.
Figure 1. Analysis of the amino acid accepting capacity of E.coli tRNAAsp when aminoacylated by AspRS, YadB and by both. (A) Extents of pure tRNAAsp charging with E.coli AspRS (gray bar), YadB (white bar) or AspRS and YadB (black bar). (B) The amino acids acylating tRNAAsp are identified by TLC as described (10).
To verify the latter possibility, we treated unfractionated E.coli tRNA with snake venom phosphodiesterase that hydrolyzes single-stranded RNAs starting from the 3'-OH end and thus removes the CCA end and the discriminator base of the tRNA molecules (Fig. 2A and B). As expected, E.coli tRNAAsp lost its ability to be aspartylated by AspRS, but fully retained its capacity to be glutamylated by YadB. (Fig. 2A and B).
Figure 2. Protection of tRNAAsp amino acid accepting capacity by E.coli AspRS or YadB catalyzed aminoacylation prior to periodate oxidation. (A) Extent of tRNAAsp aspartylation by AspRS (gray bars) and glutamylation by YadB (white bars). (B) The same experiment with tRNAAsp lacking its single stranded 3' terminus. (C) Same experiments with tRNAAsp treated with periodate.
So far, all known aaRSs aminoacylate their cognate tRNA by forming an ester bond between the activated -COOH group of the amino acid and the 2'- or 3'-OH group of the ribose cis-diol of the tRNA terminal adenosine (2,22). Periodate oxidation by opening the ribose ring deprives tRNA of its accepting capacity. However, aminoacylation protects the cis-diol of the terminal adenosine against oxidation and thus protects the tRNA against inactivation (23). Therefore, to test whether glutamylation of tRNAAsp by YadB proceeds by esterification of a OH group, we oxidized unprotected and protected E.coli tRNAAsp with periodate and analyzed their remaining accepting capacities. Figure 2C shows that YadB, like AspRS, is unable to charge periodate oxidized tRNAAsp. However, tRNAAsp protected by glutamylation with YadB prior to periodate oxidation is still aminoacylated by YadB, but not by AspRS. Likewise, tRNAAsp aspartylated by AspRS prior to periodate oxidation retains its aspartylation ability, but is no longer glutamylatable by YadB (not shown).
Altogether, the obtained data demonstrates that aminoacylation by YadB does not occur at the terminal adenosine of tRNAAsp and should take place at another place on the tRNAAsp molecule. This conclusion is supported by a previous result showing that the mischarged tRNAAsp cannot bind the elongation factor EF-Tu (14), despite the capability of a canonical Glu-tRNAAsp, i.e. when charged on the 3'-end aa-tRNA, to bind the factor (24).
YadB aminoacylates the queuosine located in the first position of tRNAAsp anticodon
Analysis of the chemical structure of the modified nucleosides present in E.coli tRNAAsp (25) reveals that the queuosine (Q) in the first position of the anticodon (position 34 of tRNA) possesses a cyclopentenyldiol ring containing a cis-diol that could possibly mimic the 2'- and 3'-OH groups of the ribose of the terminal adenosine (26–28) (Fig. 3A). Introduction of queuosine into tRNA is achieved by a complex cascade of enzymatic reactions involving tRNA-guanine transglycosylase (TGT), which results by its exchange with G34 (29).
Figure 3. Amino acid accepting capacities of E.coli native tRNAAspQ34 and queuosine-lacking tRNAAspG34. (A) Kinetics of aminoacylation of tRNAAspQ34 (top) and tRNAAspG34 (bottom) by E.coli AspRS (black triangles) and YadB (black squares). (Insets) 2-Dimensional TLC analysis of the 5' phosphate nucleosides present in the two tRNAs (the arrows indicate the positions where pQ migrates). The structures of Q and G are shown at the left with the arrows indicating the OH groups of the cyclopentenyldiol ring that are the putative Glu acceptors on Q. (B) Lineweaver–Burk plots allowing determinations of the Km (open triangles) of YadB for native tRNAAsp and Ki (open squares) of tRNAAspQ34 for aminoacylation of native tRNAAsp by YadB.
To check whether glutamylation of tRNAAsp by YadB is dependent on the presence of queuosine we first attempted to glutamylate the E.coli tRNAAsp transcript deprived of modified nucleotides. Table 1 shows that pure E.coli tRNAAsp is fully charged by YadB, whereas the transcript, deprived of post-transcriptional modifications but instead containing G34, is not charged by YadB. Likewise, Saccharomyces cerevisiae and T.thermophilus tRNAAsp deprived of Q (30,31) are fully aspartylated but not charged by YadB.
Table 1. Aspartylation and glutamylation of tRNAAsp of various origins possessing either G34 or Q34
To unambiguously prove that YadB aminoacylates the queuosine of E.coli tRNAAsp, we purified a variant of it containing all modified nucleosides except Q34 (tRNAAspG34) and compared its glutamylation capacity with that of the wild-type tRNAAsp (tRNAAspQ34). tRNAAspG34 was purified from an E.coli tgt– strain, unable to introduce Q into tRNAs. As expected, both tRNAs are completely aspartylated by E.coli AspRS (Fig. 3A). However, while YadB fully aminoacylated tRNAAspQ34, this enzyme was unable to glutamylate tRNAAspG34 (Fig. 3A). As confirmed by 2-dimensional TLC analysis of tRNA hydrolysates, the two tRNA variants differ only in the absence of Q in tRNAAspG34 (Fig. 3A, insets). A further confirmation that Q34 is the acceptor of the activated Glu was obtained by inhibition of glutamylation of wild-type tRNAAsp in the presence of tRNAAspG34. Figure 3B shows that tRNAAspG34 competitively inhibits charging of tRNAAspQ34 by YadB, which displays the same affinity for both tRNAs, as indicated by the kinetic constants (Km for tRNAAspQ34 and Ki for tRNAAspG34 of 0.19 μM). Since both tRNAs bind with the same affinity to YadB, whereas the protein charges only the species containing Q, it is concluded that glutamylation by YadB occurs on Q.
What are the structural elements that promote aminoacylation of the E.coli tRNAAsp anticodon by YadB?
In E.coli, tRNAAsp, tRNAAsn, tRNAHis and tRNATyr contain Q in the wobble position of their anticodons (32), yet only Q34 of tRNAAsp is glutamylated by YadB (14). This indicates that Q is not solely responsible for the recognition of tRNAAsp. In fact, as shown above, it does not contribute to the affinity of tRNAAsp for YadB.
To identify potential determinants responsible for the interaction of tRNAAsp with YadB, we compared the tRNAAsp and tRNAGlu sequences from E.coli. This comparison revealed a striking mimicry between the tRNAAsp anticodon stem and loop and the tRNAGlu acceptor arm (Fig. 4A). The first four base pairs of the tRNAGlu acceptor arm are identical to the first four base pairs of the tRNAAsp anticodon stem, while the discriminator G73 residue and C74 of tRNAGlu align with G39 from the tRNAAsp anticodon stem and C38, the last nucleotide of the anticodon loop, respectively.
Figure 4. Sequence comparison of tRNAGlu acceptor arm and tRNAAsp anticodon stem and loop from bacteria encoding YadB and expansion of the comparison to tRNAAsn, tRNAHis and tRNATyr anticodon stems and loops. tRNAs from 17 bacterial species were compared. (A) Escherichia coli tRNAAsp and tRNAGlu are presented in a 2-dimensional projection of their L-shaped architecture. Nucleotides targeted by the comparison are displayed and the amino acid-accepting nucleotides (Q34 and A76) indicated by an arrow. Numbering of nucleotides is according to Sprinzl et al. (32). For each bacterial species tRNAAsp anticodon stem and loop positions (base pairs 27–43 to 30–40 and nucleotides 39 and 38) were compared to the corresponding positions of the tRNAGlu acceptor stem (base pairs 4–69 to 1–72 and nucleotides 73 and 74) from the same species. When nucleotides are the same as in tRNAAsp and tRNAGlu from E.coli tRNAs, they are displayed in white on a black box. For clarity, only nucleotides 43–38 from tRNAAsp and 69–74 from tRNAGlu are displayed. Consensus sequences of the 38–43 stretches in tRNAAsp and 74–68 stretches in tRNAGlu are given together with the consensus of the complementary residues (30–27 and 1–4 stretches). Residues in italic are partially conserved; R stands for purine and Y for pyrimidine. The sequences displayed are ranked according to the YadB classification (14). (B) 2-Dimensional representation of E.coli tRNAAsp and the consensus sequences in the anticodon branch of the 17 bacterial species. When a nucleotide is the same as its counterpart from the consensus sequence of tRNAGlu it is displayed in white on a black box.
Expanding the comparative analysis to all tRNAGlu and tRNAAsp species of bacteria encoding YadB genes, we observed partial conservation of the sequence mimicry (Fig. 4A). Interestingly, the sequence conservation in the two tRNAs follows the phylogenetic classification of the YadB proteins in three structure-based subgroups (14). Thus complete mimicry between the two tRNAs is found in the two -proteobacteria, Yersinia pestis and Salmonella enteritica, that are most closely related to E.coli with respect to their YadB proteins (Fig. 4A). Mimicry between tRNAAsp and tRNAGlu, as found in E.coli, becomes looser in bacteria, where the divergence between YadB proteins increases, but interestingly high mimicry exists between tRNAs of organisms with YadB proteins from the same subgroup.
Altogether, this comparative analysis indicates that four nucleotides from the 5'-side of the anticodon stem–loop of the tRNAAsp species are conserved (namely C38 from the anticodon loop as well as G39, G42 and G43 from the anticodon stem) and three are present in a 6 nt sequence stretch from the 3'-terminus of the corresponding tRNAGlu species (namely single-stranded C74 from the CCA terminus and discriminator residues G73 and G69). This sequence mimicry with the anticodon branch of tRNAGlu is never found in tRNAAsn, tRNAHis and tRNATyr that possess Q in position 34 but in which G39 and C38 are always replaced by U39 and A38 (Fig. 4B). Likewise, no such mimicry is found between eukaryotic tRNAAsp and tRNAGlu species. In conclusion, this suggests that those residues from the consensus highlighted in Figure 3, especially C38, G39, G42 and G43, are critical for recognition of tRNAAsp by YadB and, conversely, their non-conservation in tRNAAsn, tRNAHis and tRNATyr contribute to the rejection of these Q-containing tRNAs by YadB proteins.
To gain a deeper insight into the structural elements that could promote tRNAAsp anticodon recognition and aminoacylation by YadB, the minimal GluRS, we constructed a docking model of E.coli YadB and E.coli tRNAAsp (Fig. 5). By aligning the two similar helical parts of the tRNAAsp anticodon stem (residues 24–45) and tRNAGlu acceptor stem (without 73GCCA76) we observed that in tRNAAsp Q34 protrudes in a loop, while in tRNAGlu the GCCA stem residues fold back in the opposite direction. As a consequence, the amino acid accepting OH groups of tRNAAsp (Q34) and tRNAGlu (A76) are presented in different conformations. Therefore, simple docking of tRNAAsp to YadB, without structural deformation of the anticodon loop, is not compatible with ‘classical’ tRNA binding in the catalytic domain of a class I synthetase. However, the tRNA could be fitted with only the 30–38 segment in the groove (Fig. 5). This position is correct with regard to shape complementarity and to reactive group closeness (Fig. 5, inset). Note that the interacting surface between YadB and tRNAAsp is quite large in the tentative docking model: 1100 ?2 of YadB (10% of the total surface) are shielded from solvent in the complex. In this model the interacting protein surface is slightly reduced as compared to the complexes involving class I GluRS in which 1600 ?2 are shielded from the solvent (33).
Figure 5. A tentative model of the complex of E.coli tRNAAsp with cognate YadB. (A) The YadB molecular surface is displayed in dark blue, the loop in lighter blue and adenylate in sphere representation. The docking procedure was based on the constraint that the reactive group, the cyclopentenyldiol ring of Q34, like the ribose from A76, should be close to the Glu–adenylate phosphate–Glu scissile bond. The Q34 cyclopentenyldiol ring was therefore used as a pivot around which the tRNA rigid body moved. The tRNAAsp is displayed as an atom color coded molecular surface. (Inset) Close-up of the interaction illustrating the vicinity of the adenylate donor molecule and the acceptor group, Q34 of tRNAAsp. Adenylate and the tRNAAsp domain with Q base are displayed in stick representation. (B) Rotation of 120° around a vertical axis and 30° clockwise in the plane.
Functional and evolutionary considerations
Data reported here break down the current paradigm underlying all the research on tRNA recognition and aminoacylation by aaRSs, namely that the catalytic domain of synthetases recognizes the amino acid branch of tRNAs and attaches the activated amino acid via an ester bond to their 3'-CCAOH termini. All the literature on structural and functional biology in the field has been exclusively based on this paradigm. Therefore, it was a great surprise to discover that a protein closely mimicking the catalytic domain of E.coli GluRS, namely the YadB protein from E.coli, recognizes the anticodon branch of a tRNA and attaches an amino acid on a modified base from the tRNA anticodon. Interestingly, YadB has the ability to activate Glu (14), although without the assistance of tRNAGlu, as is the case of canonical GluRSs (34), but also the unexpected ability to interact with the distal extremity of E.coli tRNAAsp and to acylate the cyclopentenyldiol ring of the Q base post-transcriptionnally inserted at position 34 of the anticodon. This ring, with two cis-diol residues, mimics the structure of the 3'-terminal ribose of tRNA and thus Glu is attached to the Q base via an unstable ester bond (half-life 7.5 min) (14). This explains why the presence of a glutamylated Q base (which can be abbreviated as Glu-Q) in E.coli has escaped identification up to now. Docking of the tRNAAsp anticodon branch on YadB (Fig. 5) suggests that this branch has physical contact with the protein, a conclusion in agreement with the competition of tRNAAsp deprived of Q for glutamylation of native tRNAAsp with Q34 (Fig. 3B). This a priori unpredicted biological function of YadB as a tRNA modifying enzyme recognizing the anticodon stem and loop of tRNA can be generalized to other YadB-containing microorganisms, as supported by sequence comparison of their tRNAAsp and tRNAGlu (Fig. 4). Therefore, YadB proteins can be renamed glutamyl-Q tRNAAsp synthetases (Glu-Q-RSs).
From another perspective, one can question the evolutionary origin of the YadB proteins. Whether an ancestral YadB protein was a precursor of modern prokaryotic GluRSs or whether the YadB family originates from an ancestral GluRS remains an open question. However, their contemporary biological function has likely a recent evolutionary origin, given the reaction that YadB enzymes catalyze. Because the presence of YadB proteins is restricted to prokaryotic organisms, glutamylation of the Q base in tRNAAsp, which only occurs in such organisms, is intriguing, since Q base is also found in eukaryotes. However, hypermodified Q residues have been detected in higher eukaryotes, with a mannosyl or galactosyl residue instead of a Glu fixed on the cyclopentenyldiol ring. Note also that the mannosyl and galactosyl moieties, like the ribose and the cyclopentenyldiol ring, possess two adjacent cis-diols that could also be potential amino acid acceptors via catalysis by aaRS-mimicking proteins. These observations raise interesting questions about the structure of the enzymes attaching mannose or galactose to Q bases and about the possible existence of proteins that may esterify these sugars. Like YadB, such proteins could be aaRS-mimicking proteins.
To conclude, let us comment on tRNA structure in the light of the new paradigm and of concepts underlying the origin of tRNA and the genetic code. Here it is shown that a mimic of an aaRS catalytic domain makes no fundamental difference in recognizing a tRNA anticodon branch instead of a tRNA amino acid accepting branch. This property of YadB is in line with the idea that contemporary tRNA originated by fusion of two primordial minihelix or stem–loop structures (3). Likewise, primordial aaRSs were simplified structures restricted to their catalytic core, existing in nature as two versions (the class I and the class II versions), that became more complex in later evolution by the addition of extra domains, the most important being the anticodon-binding domain (1). In the course of such structural tinkering it is likely that primitive aaRSs were able to recognize a variety of RNA architectures with tRNA features. This potential recognition was maintained in some modern proteins as a remnant of their early evolution. YadB (Glu-Q-RS) is such a protein. We anticipate that other aaRS mimics remain to be discovered and believe that such proteins will use their catalytic and RNA recognition potential in participating in metabolic pathways as yet undeciphered.
ACKNOWLEDGEMENTS
This work was supported by Centre National de la Recherche Scientifique (CNRS), Association de la Recherche contre le Cancer (ARC), Université Louis Pasteur (Strasbourg), the French Ministry of Industry (grant ASG) and Marseille Genopole. Additional support came by grant OG0009597 from the Natural Sciences and Engineering Research Council of Canada (NSERC). M.B. is the recipient of a fellowship from Ministère de l’Education Nationale de la Recherche et de la Technologie.
REFERENCES
Ibba,M., Francklyn,C. and Cusack,S. (eds) (2004) The Aminoacyl-tRNA Synthetases, Landes Bioscience, Georgetown, TX.
S?ll,D. and Schimmel,P.R. (1974) Aminoacyl-tRNA synthetases. Enzyme, 10, 489–538.
Schimmel,P., Giegé,R., Moras,D. and Yokoyama,S. (1993) An operational RNA code for amino acids and posible relationship to genetic code. Proc. Natl Acad. Sci. USA, 90, 8763–8768.
Francklyn,C. and Schimmel,P. (1989) Aminoacylation of RNA minihelices with alanine. Nature, 337, 478–481.
Frugier,M., Florentz,C. and Giegé,R. (1992) Anticodon-independent aminoacylation of an RNA minihelix with valine. Proc. Natl Acad. Sci. USA, 89, 3990–3994.
Frugier,M., Florentz,C. and Giegé,R. (1994) Efficient aminoacylation of resected RNA helices by class II aspartyl-tRNA synthetase dependent on a single nucleotide. EMBO J., 13, 2219–2226.
Schimmel,P. and Ribas de Pouplana,L. (1995) Transfer RNA: from minihelix to genetic code. Cell, 81, 983–986.
Schwob,E. and S?ll,D. (1993) Selection of a ‘minimal’ glutaminyl-tRNA synthetase and the evolution of class I synthetases. EMBO J., 12, 5201–5208.
Schimmel,P. and Ribas De Pouplana,L. (2000) Footprints of aminoacyl-tRNA synthetases are everywhere. Trends Biochem. Sci., 25, 207–209.
Roy,H., Becker,H.D., Reinbolt,J. and Kern,D. (2003) When contemporary aminoacyl-tRNA synthetases invent their cognate amino acid metabolism. Proc. Natl Acad. Sci. USA, 100, 9837–9842.
Kaniga,K., Compton,M.S., Curtiss,R. and Sundaram,P. (1998) Molecular and functional characterization of Salmonella enterica serovar typhimurium poxA gene: effect on attenuation of virulence and protection. Infect. Immun., 66, 5599–5606.
Sissler,M., Delorme,C., Bond,J., Ehrlich,S.D., Renault,P. and Francklyn,C. (1999) An aminoacyl-tRNA synthetase paralog with a catalytic role in histidine biosynthesis. Proc. Natl Acad. Sci. USA, 96, 8985–8990.
Campanacci,V., Dubois,D.Y., Becker,H.D., Kern,D., Spinelli,S., Valencia,C., Pagot,F., Salomoni,A., Grisel,S., Vincentelli,R. et al. (2004) The Escherichia coli YadB gene product reveals a novel aminoacyl-tRNA synthetase like activity. J. Mol. Biol., 337, 273–283.
Dubois,D.Y., Blaise,M., Becker,H.D., Campanacci,V., Keith,G., Cambillau,C., Giegé,R., Lapointe,J. and Kern,D. (2004) An aminoacyl-tRNA synthetase-like protein encoded by the Escherichia coli yadB gene glutamylates specifically tRNA(Asp). Proc. Natl Acad. Sci. USA, in press.
Hiramaru,M., Uschida,T. and Egami,F. (1966) Ribonuclease preparation for the base analysis of polyribonucleotides. Anal. Biochem., 17, 135–142.
Martin,F., Eriani,G., Eiler,S., Moras,D., Dirheimer,G. and Gangloff,J. (1993) Overproduction and purification of native and queuine-lacking Escherichia coli tRNA(Asp). Role of the wobble base in tRNA(Asp) acylation. J. Mol. Biol., 234, 965–974.
Becker,H.D., Giegé,R. and Kern,D. (1996) Identity of prokaryotic and eukaryotic tRNA(Asp) for aminoacylation by aspartyl-tRNA synthetase from T. thermophilus. Biochemistry, 35, 7447–7458.
Kern,D. and 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, 1257–1272.
Keith,G. (1994) Nucleic acid chromatographic isolation and sequence methods. In Gehrke,C.W. and Kuo,K.C. (eds), Chromatography and Modifications of Nucleosides. Elsevier, Amsterdam, Vol. 45A, pp. A103–A141.
Roussel,A. and Cambillau,C. (1991) The TURBO-FRODO graphics package. Silicon Graphics, 81, 77–78.
Stepanov,V.G., Moor,N.A., Ankilova,V.N. and Lavrik,O.L. (1992) Phenylalanyl-tRNA synthetase from Thermus thermophilus can attach two molecules of phenylalanine to tRNA(Phe). FEBS Lett., 311, 192–194.
Chladek,S. and Sprinzl,M. (1985) The 3'-end of tRNA and its role in protein biosynthesis. Angew. Chem. Int. Ed. Engl., 24, 371–391.
Kern,D., Giegé,R. and Ebel,J.-P. (1972) Incorrect aminoacylation catalyzed by the phenylalanyl- and valyl-tRNA synthetases from yeast. Eur. J. Biochem., 31, 148–155.
Asahara,H. and Uhlenbeck,O.C. (2002) The tRNA specificity of Thermus thermophilus EF-Tu. Proc. Natl Acad. Sci. USA, 99, 3499–3504.
Sekiya,T., Mori,M., Takahashi,N. and Nishimura,S. (1980) Sequence of the distal tRNA(Asp)1 gene and the transcription termination signal in the Escherichia coli ribosomal RNA operon rrnF (or G). Nucleic Acids Res., 8, 3809–3827.
Kersten,H. and Kersten,W. (1990) Biosynthesis and function of queuine and queuosine. In Gehrke,C. and Wand Kuo,K.C.T. (eds), Chromatography and Modification of Nucleosides, PartB: Biological Roles and Functions of Modification. Elsevier, Amsterdam.
Yokoyama,S., Miyazawa,T., Iitaka,Y., Yamaizumi,Z., Kasai,H. and Nishimura,S. (1979) Three-dimensional structure of hyper-modified nucleoside Q located in the wobbling position of tRNA. Nature, 282, 107–109.
Grosjean,H. and Benne,R. (eds.) (1998) Modification and Editing of tRNA. ASM Press, Washington, DC.
Reader,J.S., Metzgar,D., Schimmel,P. and de Crecy-Lagard,V. (2004) Identification of four genes necessary for biosynthesis of the modified nucleoside queuosine. J. Biol. Chem., 279, 6280–6285.
Gangloff,J., Keith,G., Ebel,J.P. and Dirheimer,G. (1971) Structure of aspartate tRNA from brewer’s yeast. Nature New Biol., 230, 125–127.
Keith,G., Yusupov,M., Briand,C., Moras,D. and Kern,D. (1993) Sequence of tRNA(Asp) from Thermus thermophilus HB8. Nucleic Acids Res., 21, 3216–3230.
Sprinzl,M., Horn,C., Brown,M., Ioudovitch,A. and Steinberg,S. (1998) Compilation of tRNA sequences and sequences of tRNA genes. Nucleic Acids Res., 26, 148–153.
Rould,M.A., Perona,J.J., S?ll,D. and Steitz,T.A. (1989) Structure of E. coli glutaminyl-tRNA synthetase complexed with tRNA(Gln) and ATP at 2.8 ? resolution. Science, 246, 1135–1142.
Kern,D. and 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.(Micka?l Blaise, Hubert Dominique Becker,)