Mechanism and evolution of multidomain aminoacyl-tRNA synthetases revealed by their inhibition by analogues of a reaction intermediate , and by properties of truncated forms

Many enzymes which catalyze the conversion of large substrates are made of several structural domains belonging to the same polypeptide chain. Transfer RNA (tRNA), one of the substrates of the multidomain aminoacyl-tRNA synthetases (aaRS), is an Lshaped molecule whose size in one dimension is similar to that of its cognate aaRS. Crystallographic structures of aaRS/tRNA complexes show that these enzymes use several of their structural domains to interact with their cognate tRNA. This mini review discusses first some aspects of the evolution and of the flexibility of the pentadomain bacterial glutamyl-tRNA synthetase (GluRS) revealed by kinetic and interaction studies of complementary truncated forms, and then illustrates how stable analogues of aminoacylAMP intermediates have been used to probe conformational changes in the active sites of Escherichia coli GluRS and of the nondiscriminating aspartyl-tRNA synthetase (ND-AspRS) of Pseudomonas aeruginosa.


THE MULTIDOMAIN ENZYMES: ASPECTS OF THEIR EVOLUTION AND OF THEIR FLEXIBILITY
Multidomain enzymes probably evolved from an ancestral domain containing the active site, to which were added during evolution of other domains which increased their catalytic efficiency (k cat ), and/or improved their specificity for their substrate(s).A beautiful and relatively simple illustration of this model, reviewed by Branden and Tooze [1] in their excellent "Introduction to protein structure", is the structure and mechanism of chymotrypsin, made of two antiparallel β-barrel domains; this serine protease which cleaves peptide bonds using the catalytic triad serine/histidine/aspartate (residues not adjacent in this polypeptide), harbors in one domain the serine (Ser) residue, which is the most important one for this catalysis, and harbors in the other domain the histidine (His) and aspartate (Asp) residues whose presence strongly increases the k cat without altering significantly the K M of this enzyme for its substrates.These results suggest that the His and Asp residues of the catalytic triad are not absolutely essential for the catalytic activity, and therefore the ancestral chymotrypsin may have been a single domain containing only the catalytic Ser residue.
The aminoacyl-tRNA synthetases (aaRS) are multidomain enzymes which play a central role in the correct translation of genetic information into proteins (reviewed by Ataide and Ibba [2], and by Giegé et al. [3]).Each member of this family of about 20 enzymes interacts with ATP and a specific amino acid to catalyze its activation in the form of an aminoacyladenylate, a high-energy intermediate where the α-COOH group of the amino acid forms an acid anhydride bond with the phosphate of AMP.This enzyme-bound intermediate then esterifies an OH group at the 3'-end of a tRNA corresponding to that amino acid according to the genetic code, yielding an aminoacyl-tRNA (reviewed by First [4]) which is used either for the polymerisation of amino acid residues on the ribosome, or for other metabolic processes not related to translation biosynthetic (reviewed by Lapointe and Giegé [5]).As noted by Schimmel, Giegé, Moras and Yokoyama in their 1993 review [6], the aaRS structures are organized, to a rough approximation, into two major domains: one containing the active site which interacts with the ancestral part of tRNA including the acceptor end, and one generally less or not conserved which provides for interactions with the second domain of tRNA, including the anticodon.Some structural domains in several multidomain enzymes change their relative orientations upon binding substrates or other ligands, often by rigid-body motions allowed by the flexible hinges linking adjacent domains.Using a bioinformatic method named "computational solvent mapping", Chuang et al. [7] compared ten different crystallographic structures of a multidomain enzyme, and were able to detect significant changes in binding sites and interdomain crevices at a higher resolution than that provided by superposing these X-ray structures, revealing conformational changes even at an overall root mean square deviation (RMSD) that is close to the expected error in the atomic coordinates.In the case of the pentadomain glutamyl-tRNA synthetase (GluRS) of Thermus thermophilus, crystallographic structures determined in the absence and in the presence of the cognate tRNA Glu substrate [8] revealed the reorientation, without changing their folds, by a 6˚ rotation of domain 4 relative to domain 3 (at the transition between the active site domains 1 to 3, and the anticodon arm-binding domains 4 and 5), and by a 8˚ rotation of the C-terminal domain 5 relative to domain 4.These tRNA Glu -triggered domain reorientations result in a switch between ATP binding to a nonproductive site in the absence of tRNA Glu (glutamate is not activated by GluRS alone) to a productive site in its presence.

ASPECTS OF GLURS MECHANISM AND EVOLUTION REVEALED BY KINETIC AND INTERACTION STUDIES OF COMPLEMENTARY TRUNCATED FORMS OF THIS PENTADOMAIN ENZYME
The above-mentioned GluRS domain reorientations prompted us to test the importance on GluRS activity of the covalent connectivity between some of these adjacent domains.For answering these questions, we used the GluRS of a mesophilic bacteria, Escherichia coli, because kinetic and interaction studies are performed more easily with it than with the corresponding enzyme of a thermophile.The high level of identity between the amino acid sequences of E. coli and T. thermophilus GluRSs allowed us to use an alignment of their sequences and the known crystalline structure of the latter [8] to identify the position of the hinges linking the structural domains of E. coli GluRS [9].We constructed vectors allowing the overproduction of two sets of truncated GluRSs: domains 1 to 3, and domains 4 + 5; and domains 1 to 4, and free domain 5. Their kinetic characterization showed first that the two C-truncated GluRS (1 − 3) and GluRS (1 − 4) have very low k cat in the tRNA glutamylation reaction (about 2000fold lower than that of full-length GluRS), but that their K M values for their substrates (ATP, glutamate and tRNA Glu ) are nearly identical to those of the full-length GluRS (1 − 5).A similar result for GluRS (1 − 3) was reported by Dasgupta et al. [10].The major importance of k cat in the recognition between aminoacyl-tRNA synthetases and tRNAs was initially reported by Ebel et al. in 1973 [11] (reviewed by Giegé and Springer [12]).
GluRS (1 − 3) glutamylation activity was not complemented by GluRS (4 + 5), indicating the importance of the covalent linkage between domains 3 and 4 for efficient activity.On the other hand, GluRS (1 − 4) activity was stimulated up to 100-fold by free domain 5 [9], located about 70 Å away from the active site.No interacttions between the complementary GluRS (1 − 4) and free domain 5 were detected by white light interferometry with a nanoporous silicon biosensor [13], but K D values of 0.11 and 1.2 µM, respectively, were measured for the interactions with tRNA Glu of these two truncated GluRSs (compared to 0.5 µM for full-length GluRS), using the quenching of the tryptophan fluorescence of these proteins.These results suggest, first, that an ancestral form of GluRS had only the first 4 domains, and that it evolved and became more efficient by the addition of an ancestral C-terminal domain 5, and secondly that at least a part of the information present in the anticodon identity elements of tRNA Glu is transferred to the active site of GluRS through the tRNA Glu backbone.Based on the structure of the T. thermophilus GluRS/tRNA Glu complex, domain 5 interacts with the first nucleotide of the anticodon, which in E. coli tRNA Glu is a modified U (5-methylamino, 2-thioU), which is an important identity element for the recognition of E. coli tRNA Glu by its GluRS ( [14] and references therein).
For E. coli glutaminyl-tRNA synthetase, closely related to GluRS in its evolution and mechanism, but containing anticodon-binding domains with different topologies than those of GluRS (and therefore not related evolutionarily), a model of intraprotein communication between the anticodon-binding domains and the active site has been proposed by Weygand-Durasevic et al. [15].

STABLE ANALOGUES OF AMINOACYL-AMP INTERMEDIATES USED TO REVEAL CONFORMATIONAL CHANGES IN THE ACTIVE SITES OF E. COLI GLURS AND OF THE NONDISCRIMINATING ASPARTYL-TRNA SYNTHETASE (ND-AspRS) OF PSEUDOMONAS AERUGINOSA
As intermediates in the aminoacylation reaction, the aminoacyladenylates (aa-AMP) stand in the active sites of their cognate aaRSs.Stable analogues of some aa-AMP are good inhibitors of the cognate aaRS (reviewed by Chênevert et al. [16], and by Vondenhoff and Van Aerschot [17].
We used some of them as reporters of the influence of tRNA on the structure of the active site.
As mentioned above from crystallographic structures of T. thermophiles GluRS +/− tRNA Glu , the presence of tRNA allows the correct positioning of ATP in the active site.Using isothermal microcalorimetry, we measured the free energy of binding of Glu-AMS (glutamylsulfamoyl adenosine, the strongest known inhibitor of GluRS, with a K i of about 3 nM [18]) to E. coli GluRS, and found that the presence of tRNA Glu increases about 10fold the binding energy of this inhibitor to this enzyme (Blais, S., Bonnaure, G., Kornblatt, J. and Lapointe, J., unpublished results), in qualitative agreement with the structural results.
We noticed a biphasic inhibition by L-aspartol adenylate of the ND-AspRS of Pseudomonas aeruginosa, when unfractionated tRNA containing both tRNA Asp and tRNA Asn from these bacteria was used as substrate in the aminoacylation reaction [19].This nondiscriminating AspRS aspartylates both tRNA Asp and tRNA Asn , and the incorrectly charged Asp-tRNA Asn is then transamidated into Asn-tRNA Asn by a tRNA-dependent amidotransferase [20], these two enzymes catalyzing the indirect pathway of Asn-tRNA biosynthesis (reviewed by Huot et al. [21]).We then separated these two tRNAs and found that the K i of L-aspartoladenylate for this ND-AspRS was 41 µM for tRNA Asp aspartylation, and 215 µM for tRNA Asn aspartylation, indicating that the two different tRNA substrates of this enzyme interact differently with its active site.This result is consistent with the observation of structural changes in the active site of yeast AspRS upon tRNA Asp binding [22].
More generally, we conclude from these results that aminoacyladenylate analogs, which are competitive inhibitors of their cognate aminoacyl-tRNA synthetase, can be used to probe rapidly the role of various structural elements in positioning the tRNA acceptor end in the active site.