Yeast tRNA Asp Charging Accuracy Is Threatened by the N-terminal Extension of Aspartyl-tRNA Synthetase*

This study evaluates the role of the N-terminal extension from yeast aspartyl-tRNA synthetase in tRNA aspartylation. The presence of an RNA-binding motif in this extension, conserved in eukaryotic class IIb amino-acyl-tRNA synthetases, provides nonspecific tRNA binding properties to this enzyme. Here, it is assumed that the additional contacts the 70 amino acid-long appendix of aspartyl-tRNA synthetase makes with tRNA could be important in expression of aspartate identity in yeast. Using in vitro transcripts mutated at identity positions, it is demonstrated that the extension grants better aminoacylation efficiency but reduced specificity to the synthetase, increasing considerably the risk of noncognate tRNA mischarging. Yeast tRNA Glu(UUC) and tRNA Asn(GUU) were identified as the most easily mischarged tRNA species. Both hav e a G at thediscriminator position, and their anticodon differs only by one change from the GUC aspartate anticodon. Aminoacylation of tRNA is a key step during translation of genetic information. It is performed by aminoacyl-tRNA synthetases (aaRSs) 1 and involves specific recognition of their cognate tRNAs and amino acids (reviewed in Ref. 1). Fidelity of translation depends on Full-length transcripts correctly ending with the CCA sequence were purified by preparative electrophoresis on 12% polyacrylamide dena-turing gels (8 M urea) followed by electroelution in Tris-borate-EDTA (Schleicher and Schuell apparatus). The concentration of tRNA transcripts was determined by absorbance at 260 nm. Purification of Yeast tRNAs— Countercurrent fractions enriched in tRNA Asn , tRNA Asp or tRNA Glu were gifts from G. Keith. Pure tRNA Asn and tRNA Glu were obtained by purification on a 10% polyacrylamide gel (19/1), in semidenaturing conditions (4 M urea). Migration was per- formed at 300 V for 48 h at 4 °C. Bands containing pure tRNAs were handled as described for tRNA transcripts. Pure tRNA Asp was a gift from A. The´obald-Dietrich. The purity of tRNAs ( (cid:3) 95%) was established by testing their cognate amino acid acceptance.

Aminoacylation of tRNA is a key step during translation of genetic information. It is performed by aminoacyl-tRNA synthetases (aaRSs) 1 and involves specific recognition of their cognate tRNAs and amino acids (reviewed in Ref. 1). Fidelity of translation depends on accurate tRNA aminoacylation, but unexpectedly, this reaction does not occur with absolute specificity (2). This dilemma is resolved since synthetases developed kinetic artifices ensuring destruction of the mischarged tRNAs (editing), thereby reducing error levels in protein synthesis (3). Furthermore, specialized proteins, by interacting with certain synthetases, increase their affinity for the cognate tRNA and thus ensure increased aminoacylation efficiency. This is the case in the association of Arc1p with MetRS and GluRS in yeast (4) and of p43 with ArgRS in mammals (5).
A similar "helper" role seems to be played by the extra domains within eukaryotic synthetases. Indeed, these enzymes differ usually from their prokaryotic counterparts by appendices located in their N-or C-terminal regions (6), outside the structural core defining the two classes of synthetases (7). In mammals, the extensions are involved in the formation of the multi-synthetase complex (6), but such architecture has not been detected in unicellular eukaryotes until now. In fact, little is known about the functional role of the extensions in unicellular eukaryotic synthetases, except that in two different yeast aminoacylation systems, they were shown to provide nonspecific tRNA binding properties to GlnRS (8,9) and AspRS (10).
Yeast AspRS contains an N-terminal appendix of 70 amino acids that adopts a helical structure (11,12). It encompasses an RNA-binding motif (XSKXXLKKXK) conserved in eukaryotic AspRSs but also in eukaryotic LysRSs and AsnRSs (10), the two other class IIb synthetase families. The appendix, located next to the anticodon-binding module, was shown to considerably increase the stability of the complex between AspRS and cognate tRNA Asp as shown by gel-shift binding assays and footprinting experiments. Moreover, the AspRS appendix can be replaced by the appendices of LysRS or AsnRS (10). Likewise, it was recently demonstrated that the extension in human LysRS shares the same properties of those in yeast AspRS (13). However, its presence in AspRS is not mandatory for tRNA aminoacylation (11,14). As seen in the crystal structure of yeast AspRS complexed with tRNA Asp (15), and in that recently solved of the free enzyme truncated of its first 70 amino acids (16), only the core of the enzyme is required for tRNA aspartylation.
Specificity of tRNA recognition by yeast AspRS is primarily ensured by the tRNA identity determinants. Only six such determinants, located in three distinct regions of tRNA Asp (Fig.  1A), are important to confer the aspartate identity. They are the discriminator base G73, four bases in the anticodon loop (G34, U35, C36, and C38), and base pair G10-U25 in the core region of the tRNA (17,18). Mutation of these elements leads to losses in aspartylation efficiencies (17), the strongest determinants being those leading to the strongest effects upon mutation (e.g. G73 and the aspartate anticodon G34, U35, C36). Except for identity base pair G10-U25, which is involved in the correct folding of the tRNA (19) and has no direct contact with the enzyme, all other identity bases make specific interactions with amino acid residues on AspRS (20). Nonspecific contacts, however, are much more numerous. As shown by biochemical and x-ray data, they extend along the side of the tRNA comprising the variable region (15,21,22) (Fig. 1B, left). Additional nonspecific contacts are provided by the N-terminal extension of native AspRS and are found along the opposite side of the tRNA (Fig. 1B, right). They increase the stability of the complex and the global aminoacylation efficiency (10).
Considering the better affinity of tRNA Asp for native AspRS and the increased aspartylation efficiency correlated with the presence of the N-terminal extension in the native enzyme, it can be conjectured that the additional contacts between the two macromolecular partners play a role in the mechanisms ensuring the specificity of genetic code expression at the translational level. By mutational analysis of tRNA Asp and AspRS, it is shown that the N-terminal extension is responsible for a decrease in aminoacylation specificity, thereby favoring aspartylation of noncognate tRNAs. The identified mischarged tRNAs are related with tRNA Asp by their anticodon sequence, in agreement with the concept of partial conservation of determinants in mischarged tRNAs. Altogether, a mechanistic role of the yeast AspRS N-terminal extension in the aspartylation reaction of tRNA is established. This has functional implications for the specific expression of aspartate identity in vivo.

Preparation of Pure AspRSs and of Crude Synthetase Extracts
Native AspRS 1-557 from yeast and truncated AspRS 71-557 , deleted of its N-terminal extension, were expressed and purified as described previously (10). In short, the genes were cloned into the Escherichia coli expression vector pQE70 (Qiagen) fused to a C-terminal His 6 tag. As-pRS variants were purified according to the Qiagen protocol, quantified by Bradford assay (Bio-Rad), and stored at Ϫ20°C in 50 mM phosphate buffer, pH 7.2, 150 mM KCl, 50% glycerol, and 10 mM ␤-mercaptoethanol. AspRS preparations yielded only one band on SDS-polyacrylamide gel, even when overloaded, and can be estimated to be 99% pure. Specific activities, expressed as turnover numbers, correspond to 0.20 and 0.19 s Ϫ1 for AspRS 1-557 and AspRS 71-557 , respectively (assays conducted in the presence of 25 M aspartate, a subsaturating amino acid concentration, see Ref. 23) or to 3.4 and 3.1 s Ϫ1 for AspRS 1-557 and AspRS 71-557 , respectively (assays conducted in the presence of 500 M aspartate, a concentration close to the physiological level of aspartate in yeast cells, see Ref. 24). An enriched protein extract deprived of nucleic acids and ribosomes was made by chromatography on a Mono-Q column (Bio-Rad) and used as a crude enzyme preparation containing the activity of aminoacyl-tRNA synthetases (except for prolylation and glutaminylation).

Preparation of tRNAs
In Vitro Transcription of tRNA Asp Variants-tRNA transcripts used in this work have been obtained by in vitro transcription of synthetic genes cloned in pUC 118 (17). Each of these genes corresponds to the T7 RNA polymerase promoter region directly connected to the downstream tRNA sequence. In vitro transcriptions were performed in reaction mixtures containing 40 mM Tris-HCl, pH 8.1 (37°C), 22 mM MgCl 2 , 5 mM dithioerythritol, 0.1 mM spermidine, 4 mM each of nucleoside triphosphate, 5 mM GMP, 50 ng/l linearized plasmid, and 5 g/ml T7 RNA polymerase. Transcription mixtures were incubated for 2 h at 37°C, and reactions were stopped by phenol/chloroform extraction. Full-length transcripts correctly ending with the CCA sequence were purified by preparative electrophoresis on 12% polyacrylamide denaturing gels (8 M urea) followed by electroelution in Tris-borate-EDTA (Schleicher and Schuell apparatus). The concentration of tRNA transcripts was determined by absorbance at 260 nm.
Purification of Yeast tRNAs-Countercurrent fractions enriched in tRNA Asn , tRNA Asp or tRNA Glu were gifts from G. Keith. Pure tRNA Asn and tRNA Glu were obtained by purification on a 10% polyacrylamide gel (19/1), in semidenaturing conditions (4 M urea). Migration was performed at 300 V for 48 h at 4°C. Bands containing pure tRNAs were handled as described for tRNA transcripts. Pure tRNA Asp was a gift from A. Théobald-Dietrich. The purity of tRNAs (Ͼ95%) was established by testing their cognate amino acid acceptance.  (25). Kinetic constants (k cat and K m ) were derived from Lineweaver-Burk plots. In general, the concentration of tRNA transcripts ranged from 0.1 to 5 K m when possible, except for mutants with K m over 10 M, where the highest concentration was at most 2-3 K m . Plateaus of tRNA aspartylation were determined by incubating 1 M pure tRNA (transcript or modified species) or 20 M total tRNA in the presence of different AspRS (0.05-5 M) and aspartate (25-500 M) concentrations and otherwise optimal reaction conditions (as above). Reactions were stopped after 2.5, 5, 10, and 20 min incubation at 37°C. Displayed kinetics and plateaus represent an average of at least two independent experiments. Values of k cat /K m ratios for replicate experiments varied by at most 15%.

Identification of Mischarged tRNAs
Total Yeast tRNA-The brewers' yeast tRNA was from Roche Molecular Biochemicals. Titration by aminoacylation assays, using crude synthetases extract and 25 M aspartate, indicated a content of 4.5 Ϯ 0.5% tRNA Asp in this total tRNA (deduced from plateau values).
Protection of Thiolated Bases-To avoid modification of thiolated bases during periodate oxidation (26), the total yeast tRNA was DTNBtreated as follows. Total tRNA (20,000 pmol) was incubated for 20 min at room temperature in 25 mM phosphate buffer, pH 6.8, in the presence of 0.3 mM DTNB (Sigma). DTNB was removed by dialyzing 12 h against milliQ water, and tRNA was recovered by precipitation.
Aspartylation Reactions-Total tRNA (20 M, containing ϳ1 M tRNA Asp ) was incubated with 10 M native or truncated AspRS in the presence of 500 M aspartic acid (ratio [AspRS]/[tRNA Asp ] ϭ 5). Aminoacylation was conducted at 37°C for 25 min; it was stopped by phenol extraction (pH 5.0), and the tRNA was recovered by ethanol precipitation. Controls without AspRS were run in parallel.
Oxidation of Non-aspartylated tRNAs-The periodate treatment of tRNA was adapted from Hansske and Cramer (27). The pellet containing charged and uncharged tRNAs was dissolved in 50 mM sodium acetate buffer, pH 5.0, to a final concentration of 80 M. Oxidation was started by addition of 40 mM m-periodate (Sigma) and was run at room temperature in the dark for 25 min. The reaction was stopped by neutralizing the excess periodate with 5% glycerol. Glycerol and remaining periodate were removed by dialysis (2 h against 5 mM sodium acetate at pH 5.0). The mixture of aspartylated and of oxidized tRNA species was recovered by ethanol precipitation.
Deacylation and Deprotection of tRNAs-The tRNA pellet was dissolved in 1.8 M Tris-HCl, pH 8.0, to a final concentration of 80 M and incubated for 30 min at 37°C for complete deacylation of aspartylated tRNA species. After precipitation, tRNAs were first incubated for 3 h on ice in 0.1 M dithioerythritol (reduction of thiol groups) and for 20 min at room temperature in 0.1 M NaBH 4 (reduction of periodate-generated aldehyde groups on the terminal ribose of non-aspartylated tRNAs). NaBH 4 was removed by an overnight dialysis in milliQ water. tRNAs were recovered by ethanol precipitation and quantified by absorbance at 260 nm.
Identification of Aspartylated/Protected tRNAs-Three batches of partially aminoacylated tRNAs (a control and two batches that were aspartylated by native AspRS 1-557 or truncated AspRS 71-557 ) were tested for their amino acid acceptance. Aminoacylation activities were measured under the same conditions as those described above, with 25 M amino acid, 20 M treated tRNA, and 0.15 g/l crude enzymatic extract containing aminoacyl-tRNA synthetase activities. Eighteen amino acids were tested: L-[ 14 C]-radiolabeled amino acids were alanine, arginine, glutamic acid, glutamine, histidine, isoleucine, leucine, methionine, phenylalanine, proline, serine, tyrosine, and valine; those [ 3 H]-labeled were asparagine, aspartic acid, glycine, lysine, and threonine (all from Amersham Biosciences). Purity of [ 3 H]-labeled aspartic acid and asparagine as well as of cold aspartic acid, asparagine, and glutamic acid was checked by TLC for possible contamination with other amino acids. No visible trace of such contamination has been detected.

RESULTS AND DISCUSSION
The N-terminal Extension Confers Increased Aminoacylation Efficiency but Decreased Specificity to AspRS-Aspartylation efficiencies of tRNA Asp transcripts mutated at their major identity positions (the discriminator base G73 and the anticodon bases G34, U35, C36) were monitored. For each mutant, the kinetic constants (k cat and K m ) were derived from aminoacylation reactions with the truncated and native AspRS (AspRS 71-557 and AspRS 1-557 ) ( Table I). For each AspRS form, aminoacylation efficiencies (k cat /K m ) were calculated for mutated tRNAs. For an easier comparison, results will be discussed as losses of aminoacylation efficiency calculated as (k cat /K m ) wild-type /(k cat /K m ) mutant . The higher the losses of aminoacylation efficiency, the better the discrimination control of the synthetase toward the mutated position in tRNA Asp .
As shown previously, kinetic constants determined for aspartylation of fully modified tRNA Asp are only slightly different from those measured for the corresponding wild-type transcript (25). For truncated AspRS, both k cat and K m of the aspartylation reaction stay almost the same. With native AspRS, only the K m of the modified tRNA Asp is lowered ϳ17 times as compared with the wild-type transcript (Table I).
With truncated AspRS 71-557 , mutations at single identity positions in tRNA transcripts have strong effects on the global aminoacylation efficiency, which can be reduced up to ϳ3000fold. At position 35, replacement of the wild-type U by any other base leads to sharp decreases in aminoacylation efficiency (between 700-and 1400-fold). At the three other major identity positions, 73, 34, and 36, all mutations except one lead to lower decreases of aminoacylation efficiencies with effects fluctuating between 4-and 340-fold depending on the mutation introduced. Noteworthily, the G34 3 C exception is the mutation leading to the strongest effect (ϳ3000-fold). In general, comparison of the individual kinetic constants indicates that K m is the most affected (exceptions are mutations G73 3 A and G34 3 C where both k cat and K m are changed).
The same experiments, done with native AspRS 1-557 under identical conditions, gave significantly different results. Although as expected, mutations of tRNA Asp identity positions still cause decreases in aspartylation efficiency, their effects are quantitatively and qualitatively changed. Indeed, with the native synthetase, the losses of aminoacylation efficiency reach at most ϳ500-fold (as compared with the ϳ3000-fold decrease for the strongest loss with AspRS 71-557 ), and the involved positions are different (e.g. strongest effect for U35 3 G with native AspRS 1-557 and for G34 3 C with truncated AspRS 71-557 ). For 8 out of 12 mutants, losses in aminoacylation efficiency varies between 1-and 22-fold, and only four tRNA Asp mutants are poorly aminoacylated, with losses in aminoacylation efficiencies above 80. They correspond to transcripts containing mutations G73 3 C, G34 3 C, or U35 3 A or G. Here again, affinity (reflected by K m ) between native AspRS 1-557 and the tRNA substrates is mostly affected, whereas k cat is only moderately affected.
These kinetic differences observed in aspartylation reactions conducted with native or truncated AspRSs have dramatic consequences. To evaluate these consequences, the gain in aminoacylation efficiency consecutive to the presence of the N-terminal extension in AspRS was calculated for each tRNA variant, comparing their aminoacylation efficiencies between both native and truncated AspRSs. Whatever the tRNA substrates, wild-type or mutants, the aminoacylation efficiencies are strongly stimulated when the N-terminal extension is present in AspRS. The strengths of these positive effects, however, are dependent on the mutation introduced in the tRNA. As seen in Table I, these values comprise between 3 and 171. Indeed, the N-terminal appendix triggers only a moderate increase of the aminoacylation efficiency of tRNA Asp transcripts mutated at the discriminator position (at most 11-fold for the G73 3 U mutation). In contrast, its effect is increased at least by one order of magnitude when the mutations are introduced in the anticodon (up to ϳ170-fold for the U35 3 C mutation).
Altogether, this means that native AspRS 1-557 is less discriminative than truncated AspRS 71-557 . This property of native AspRS 1-557 , a priori unexpected, is solely due to the presence of the N-terminal extension. This finding shows that the role of the extension is not marginal and has functional consequences.
Effects of Degradations in AspRS Extension on tRNA Charging-Globally, the data in Table I confirm the mutational analysis on tRNA aspartate identity determinants obtained previously with another AspRS sample (17), but they diverge quantitatively. The differences rely on the chemical nature of the enzymes used in the two studies. The synthetase used formerly was purified directly from Saccharomyces cerevisiae, whereas the two enzymes used in the present study were cloned and overexpressed in E. coli. It was known that during purification from yeast, AspRS is subjected to partial proteolysis within the RNA-binding motif in its N-terminal extension (10,11) with the consequence that the purified synthetase is a heterogeneous protein pool. Thus, the overall effects found with the degraded enzyme (17) are intermediary with those obtained with native AspRS 1-557 and truncated AspRS 71-557 with ami-noacylation efficiencies predominantly k cat -directed, in contrast to the present report where they are mainly K m -directed.
Functional Implications-As a consequence of the presence of the N-terminal extension in AspRS, the anticodon identity nucleotides are recognized with rather low discrimination (Table I). Since identity of a tRNA primarily relies on the presence of the major determinants rather than on the sequence context in which they are embedded (28), this implies that in a native tRNA pool, molecules having anticodon sequences close to the GUC aspartate anticodon should be recognized by native AspRS 1-557 with efficiencies close to that of cognate tRNA Asp . Several yeast tRNA isoacceptors answer this prerequisite, namely tRNA Asn(GUU) , tRNA Glu(UUC) , tRNA His(GUG) , tRNA Tyr -(GUA), tRNA Gly(GCC) , tRNA Ala(GGC) , and tRNA Val(GAC) . Among them, tRNA His , tRNA Tyr , tRNA Gly , tRNA Ala , and tRNA Val could be reasonably eliminated because their discriminator base is A, instead of G as in tRNA Asp . Indeed, it was shown that two mutations in tRNA Asp transcripts distributed between the anticodon and the discriminator position lead to cooperative effects that decrease severely aspartylation (29). Considering these assumptions, only tRNA Asn(GUU) and tRNA Glu(UUC) are left as good candidates for misacylation by AspRS (Fig. 2).
The Yeast AspRS Extension Increases the Risk of tRNA Mischarging-To determine the total amount of tRNA species that can be aspartylated, total yeast tRNA was incubated in the presence of different concentrations of native and truncated AspRSs and under saturating aspartate concentrations (500 M). Five aminoacylation conditions were tested with a fixed concentration of total tRNA (20 M) and increasing concentrations of synthetases (from 0.125 to 5 M). Thus, molar ratios between the tRNA Asp fraction present in total yeast tRNA (4.5%, see "Experimental Procedures") and AspRS were 1/0.125, 1/0.5, 1/1.25, 1/2.5, and 1/5. When ratios of tRNA Asp / AspRS 1-557 or tRNA Asp /AspRS 71-557 were 1/0.5, both enzymes show the same charging capacity with 6% of total tRNA aspartylated. This level reflects aspartylation of part of tRNA Asp present in total yeast tRNA (ϳ4.5%) plus a significant amount of mischarged tRNA species (ϳ1.5%). The amount of aspartylated tRNA increases with the AspRS concentration in the aminoacylation assays. However, the raise is much more pronounced in assays conducted with native AspRS 1-557 (Fig. 3). Thus, when present at a 5-fold molar excess as compared with tRNA Asp , the amount of aspartylated tRNA catalyzed by native AspRS 1-557 is more than 2-fold of that of tRNA Asp , whereas with truncated AspRS 71-557 , it reaches only 1.5-fold. This indicates that AspRS aspartylates noncognate tRNA(s) and that the level of mischarging is significantly higher with native AspRS 1-557 . In control experiments conducted in the presence of limiting aspartate concentration (25 M), the mischarging potential of AspRS, even in its native version, is suppressed (Fig. 3).
Identification of Aspartylated Noncognate tRNAs-At first, to prevent false interpretation of results, care was taken that AspRS preparations as well as amino acids were of enough purity to avoid possible misaminoacylation catalyzed by trace contaminations of E. coli synthetases (see "Experimental Procedures"). The mischarged tRNA species were identified in a three-step procedure. First, total tRNA was aspartylated by either native AspRS 1-557 or truncated AspRS 71-557 in assays conducted with 500 M aspartate, i.e. under conditions where the specific activity of both AspRSs is the highest. Further, the AspRS concentration was in 5-fold molar excess as compared with the amount of tRNA Asp present in total yeast tRNA. Second, the pool of aspartylated tRNA (comprising tRNA Asp and putative mischarged noncognate tRNAs) was periodatetreated to inactivate all uncharged tRNAs by oxidation of their 3Ј-terminal ribose. This treatment does not affect the aspartylated species protected by the aspartate residue esterified on their 3Ј-end. Before treatment, care was taken to protect tRNAs containing thiolated bases against inactivation by periodate (see "Experimental Procedures"), i.e. yeast tRNA Glu , a potential candidate for mischarging, which contains a 5-methoxycarbonylmethyl-2-thiouridine (mcm 5 s 2 U) residue in its anticodon (30). Third, the periodate-treated tRNA, still containing the pool of aspartylated tRNA species, was deacylated, and formerly mischarged species were identified by their original amino acid acceptance (see "Experimental Procedures").
Aminoacylation of total yeast tRNA with both truncated AspRS 71-557 or native AspRS 1-557 leads to quantitatively different, although qualitatively comparable, results. As seen in Table II, besides tRNA Asp , three families of noncognate aspartylatable tRNA species can be identified. The first one comprises, as anticipated, tRNA Asn and tRNA Glu and corresponds to molecules abundantly aminoacylated. The second family, with tRNA His , tRNA Lys , and tRNA Arg , comprises molecules moderately aminoacylated. The third family encompasses molecules that were not mischarged, or if so, mischarged to trace levels. With truncated AspRS 71-557 , the amount of aspartylated tRNA was only slightly above the predicted level of tRNA Asp charging (6.25% versus 4.5%). As expected, tRNA Asp was the major species identified, representing 40% of the charged tRNAs. The 60% tRNAs left were mischarged species. Surprisingly, tRNA Glu and tRNA Asn were mischarged to comparable levels representing each as much as half of the charged tRNA Asp . The minor species correspond to tRNAs mischarged to plateau levels comprising between 1 and 8%.
With native AspRS 1-557 , as a consequence of the presence of the N-terminal extension, ϳ50% more tRNA was aspartylated (Table II). The same tRNA species were identified in the aspartylated pool, but they were charged to a higher level. Altogether, aspartylation by the synthetase comprising the extension reaches 74% of the amount of tRNA Asp present in total yeast tRNA, 70% of tRNA Glu , 35% of tRNA Asn , and 3-9% of miscellaneous tRNAs. At this point, it is important to remember that E. coli GluRS cannot charge yeast tRNA Glu (31) and that prokaryotic AsnRSs do not recognize eukaryotic tRNA Asn species (reviewed in Ref. 32). Considering these facts, we can conclude that charging of both tRNA Glu and tRNA Asn did not occur as a consequence of contaminating amino acids and E. coli GluRS and AsnRS present in yeast AspRS preparations.
The two major misacylated tRNAs, namely tRNA Glu and tRNA Asn , both have a G at the discriminator position (as in tRNA Asp ), and their anticodon differs only by one base change from the GUC aspartate anticodon (U34 in tRNA Glu(UUC) and U36 in tRNA Asn(GUU) ). The less efficient charging of tRNA His and tRNA Lys can be explained by less conservation of aspartate identity elements in these tRNAs (only U35 and G73 in tRNA Lys(UUU) and G34 and U35 in tRNA His(GUG) ). The moderate mischarging of arginine-specific tRNA is more difficult to understand but could be accounted by the sole presence of G73.
Aspartylation of Pure tRNA Asn and tRNA Glu -Pure fully modified tRNA Asn(GUU) and tRNA Glu(UUC) from S. cerevisiae were tested in mischarging assays with native AspRS 1-557 and their aspartylation capacity as compared with that of tRNA Asp . Mischarging plateaus were determined in the presence of low (0.05 M) and high (5 M) AspRS 1-557 concentrations (data not shown). The results confirm the experiments conducted with total yeast tRNA. At a low AspRS 1-557 concentration, only tRNA Asp is charged to a high level (ϳ80%); mischarging of tRNA Glu did not exceed a plateau of 20%, and no aspartylation of tRNA Asn was detected. On the contrary, as soon as the AspRS concentration was increased, 100% of tRNA Asp was charged, and excessive mischarging was detected with plateaus reaching 75% of tRNA Glu and 30% of tRNA Asn aspartylated.
Besides conservation of at least three of the four major aspartate identity elements in tRNA Glu (G73, U35, and C36) and in tRNA Asn (G73, G34, and U35), one finds also in these tRNAs large sequence homologies with tRNA Asp (Fig. 2). In tRNA Glu , they comprise, among others, residue C38, which was shown to influence aspartylation efficiency (18). However, differences appear at the level of the post-transcriptional modifications (Fig. 2). In tRNA Glu , only residue U34 in the anticodon loop is hypermodified (mcm 5 s 2 U34), whereas in tRNA Asn , four specific modifications are spread all over its L-shaped structure (m 2 G10 and m 2 2 G26 in the core, t 6 A37 in the anticodon loop, and m 1 A58 in the T-loop). In both tRNAs, these modifications could influence negatively their recognition by AspRS. Nevertheless, both tRNA Glu and tRNA Asn become significantly aspartylated when the AspRS concentration is increased, which seems to be sufficient to overcome the slight sequence differences that exist between them and tRNA Asp . Thus, a high AspRS concentration is a threat for accurate expression of yeast aspartate identity.
General Conclusion-During protein synthesis, errors in amino acid incorporation have been detected (33) that are tolerated by the cell if their level is low enough. Mischarging of tRNA was considered as an obvious cause leading to such errors. It can originate as the result either of misactivated amino acids transferred to the cognate tRNA or of noncognate a The percentages of charging of each tRNA are relative to its total amount in the bulk. The 16 tested aminoacylation activities have titrated 84% of the tRNA molecules present in the bulk; the amount of tRNA Asp was 4.5% (see "Experimental Procedures"), tRNA Ala was 0.8%, tRNA Glu was 4.2%, tRNA Phe was 17.8%, tRNA Gly was 1.6%, tRNA His was 2.6%, tRNA Ile was 4.5%, tRNA Lys was 5.3%, tRNA Leu was 6.8%, tRNA Met was 2.6%, tRNA Asn was 8.6%; tRNA Arg was 5.1%, tRNA Ser was 5%, tRNA Thr was 4.4%, tRNA Val was 6.9%, and tRNA Tyr was 3.9%. ND, nondetermined (these activities could not be tested because of technical difficulties). tRNA misacylations. Different mechanisms have been considered to overcome these errors. Editing is an option for synthetases having low specificity toward their cognate amino acid, as E. coli IleRS (34) and ValRS (35) and its structural foundation explicitly established for prokaryotic IleRS (36) and ThrRS (37). Correction or prevention of tRNA mischarging due to misrecognition of tRNA is less documented. The presence of antideterminants in tRNAs, responsible for blocking noncognate aminoacylation, has been studied. For example, methylation of G37 in yeast tRNA Asp blocks arginylation (38).
Here, in both tRNA Glu and tRNA Asn , base modifications could influence negatively their recognition by AspRS, but they are not sufficient to repress mischarging with aspartate in vitro. In fact, the biological consequences of this kind of mischarging was considered as marginal because it is further repressed in the presence of the cognate tRNA (reviewed in Ref. 39), and this view is strengthened by the observation that in vivo competitions between different synthetases lower the risk of mischarging (40). However, this study demonstrates that yeast AspRS is able to aspartylate in vitro noncognate tRNAs to unexpectedly high levels, even in total tRNA samples containing cognate tRNA Asp . This property is considerably enhanced when the N-terminal extension of the enzyme is present. High mischarging relies on the inherent low specificity of yeast AspRS for its tRNA Asp substrate, as shown by the weak catalytic discrimination toward tRNA Asp molecules mutated at identity positions. This low discrimination is mainly due to the rather high affinities the mutated or noncognate tRNAs retain for the native AspRS. From these considerations, it can be expected that tRNAs favorably mischarged are those having sequence elements closely related to the tRNA Asp identity set. This is actually the case since the aspartate identity set is partially conserved in yeast tRNA Asn and tRNA Glu with a sole difference found in one anticodon position.
In this work, aspartylation of total yeast tRNA and of pure noncognate tRNA species were conducted under an aspartate concentration (500 M) that approximates the physiological level of aspartic acid present in yeast cells (24). Noteworthily, yeast AspRS is rather unique among synthetases with a low affinity (2.5 mM) for its amino acid substrate (23,41). Therefore, this enzyme needs a high aspartate concentration for optimal catalytic activity. Thus, at a low aspartic acid concentration (25 M, the usual concentration in standard aspartylation assays), no significant aspartylation enhancement was observed, but mischarging was easily detected when working at a higher amino acid concentration (Fig. 3). Indeed, enhancing the amino acid concentration from 25 to 500 M enhances by about 17-fold the specific activity of AspRS (see "Experimental Procedures"), which not only favors aminoacylation of cognate tRNA Asp but also facilitates the less efficient mischarging of noncognate tRNAs. The reason for that is the existence of an equilibrium between aminoacylation (amino acid concentrationdependent) and deacylation (amino acid concentration-independent) of tRNAs (42, 43).
The consequence of yeast AspRS mischarging ability would be dramatic misincorparations of aspartic acid residues at asparagine and glutamate positions in polypeptides. This possibility receives support considering that the high level of tRNA Asn and tRNA Glu mischarging was found under conditions converging to the physiological environment. Indeed, the high intracellular concentrations of tRNAs (ϳ1 M for each tRNA species, see Ref. 44) and aspartate (ϳ500 M aspartate, see Ref. 24) are those used in this work. However, as seen in Fig. 3, the level of tRNA mischarging can be modulated since it is dependent upon the AspRS and aspartic acid concentrations. It can even become undetectable if these concentrations are low enough. This latent mischarging of AspRS, enhanced by the nonspecific tRNA binding properties of its N-terminal extension, represents a vital threat for the yeast cell. Thus, one can question whether elongator factor would discriminate efficiently enough the mischarged tRNAs (45,46) and protect the cell against their toxicity. Another way to lower the risk of errors during translation would be to reduce the intracellular concentrations of AspRS, thereby decreasing mischarging possibilities. It is remarkable that traces of this potential mischarging ability remain present in organisms other than yeast. Indeed, an evolutionary proximity between tRNA Asp and tRNA Asn has been explicitly demonstrated in some microorganisms, with AspRS mischarging tRNA Asn and Asp-tRNA Asn being the precursor for the synthesis of Asn-tRNA Asn (47,48). It is also interesting to note that maybe evolution answered to the menace of tRNA Glu mischarging by replacing the nature of its discriminator base in higher eukaryotes (G73 in prokaryotes and unicellular eukaryotes changed to A in higher eukaryotes, see Ref. 49).