Plasticity of Recognition of the 3′-End of Mischarged tRNA by Class I Aminoacyl-tRNA Synthetases*

Certain aminoacyl-tRNA synthetases prevent potential errors in protein synthesis through deacylation of mischarged tRNAs. For example, the close homologs isoleucyl-tRNA synthetase (IleRS) and valyl-tRNA synthetase (ValRS) deacylate Val-tRNAIle and Thr-tRNAVal, respectively. Here we examined the chemical requirements at the 3′-end of the tRNA for these hydrolysis reactions. Single atom substitutions at the 2′- and 3′-hydroxyls of a variety of mischarged RNAs revealed that, while acylation is at the 2′-OH for both enzymes, IleRS catalyzes deacylation specifically from the 3′-OH and not from the 2′-OH. In contrast, ValRS can deacylate non-cognate amino acids from the 2′-OH. Moreover, for IleRS the specificity for a 3′-O location of the scissile ester bond could be forced to the 2′-position by introduction of a 3′-O-methyl moiety. Cumulatively, these and other results suggest that the editing sites of these class I aminoacyl-tRNA synthetases have a degree of inherent plasticity for substrate recognition. The ability to adapt to subtle differences in mischarged RNAs may be important for the high accuracy of aminoacylation.

The genetic code is based on the accurate aminoacylation of tRNAs by aminoacyl-tRNA synthetases (1,2). These enzymes synthesize aminoacyl-tRNA in two steps. The amino acid is first reacted with ATP to give an activated aminoacyl adenylate, and then transesterified to the 3Ј-end of the tRNA. Aminoacyl-tRNA synthetases must precisely recognize both amino acid and tRNA substrates to yield the correct product. While the structural diversity of tRNA molecules allows for rigorous selection based on RNA-protein interactions, differentiating between closely related amino acids is more challenging. Years ago, Pauling (3) noted the intrinsic difficulty for isoleucyl-tRNA synthetase in the recognition of isoleucine over valine through simple binding interactions.
Valine, which differs from isoleucine by a single methylene unit, is activated by Escherichia coli IleRS 1 only 180-fold less efficiently than isoleucine (4). However, the substitution of valine for isoleucine at isoleucine codons in the cell is less than 1 in 3000 (5). The increased specificity is a result of the RNAdependent editing of misactivated valine by IleRS (6,7). A highly related class I aminoacyl-tRNA synthetase, valyl-tRNA synthetase, faces a similar dilemma in the accurate aminoacylation of tRNA Val . Threonine, an isostere of valine, is activated at a rate 250-fold reduced from that of valine (8). Like IleRS, ValRS prevents the misincorporation of threonine into proteins through the RNA-dependent editing of misactivated threonine (9).
These reactions strictly require the presence of the cognate tRNA (6,10). In the absence of tRNA, the enzymatically generated misactivated adenylates remain in the active site, sequestered from hydrolysis. Upon addition of cognate tRNA the misactivated amino acids are hydrolyzed, regenerating the free tRNA and amino acid, while converting 1 equivalent of ATP to AMP. A prominent mechanism for editing misactivated amino acids is the rapid hydrolysis of transiently mischarged tRNA (7,9). This reaction is catalyzed at a second active site on IleRS and ValRS. This site is located within a large insertion (termed CP1) into the canonical class I aminoacyl-tRNA synthetase active-site fold. The CP1 domain as an isolated polypeptide hydrolyzes its cognate mischarged tRNA (11). Crystallographic analysis of Thermus thermophilus IleRS pinpointed the editing site to a pocket of essentially invariant amino acids within CP1 located ϳ30 Å from the aminoacylation active site (12). This site binds valine but sterically excludes isoleucine. A co-crystal structure of Staphylococcus aureus IleRS with tRNA Ile suggested how the tRNA may place its 3Ј-end in the editing site (13). However, specific interactions in the editing site could not be observed. More recently, the co-crystal structure of T. thermophilus ValRS bound to tRNA Val demonstrated a similar mode of tRNA binding (14). Although a mischarged amino acid could be modeled at the end of the tRNA, neither this model nor mutational analysis established a hydrolytic mechanism (15,16).
Early work demonstrated an important role for the 3Ј-end of tRNA Ile and tRNA Val in editing (17)(18)(19). However, the exact nature of this role was not determined. Model studies of uncatalyzed ester hydrolysis demonstrated that a cis-hydroxyl stimulates hydrolysis, likely via intramolecular hydrogen bonding (20). While its effect may be important for the efficiency of deacylation, the importance of the terminal hydroxyls could be related to the rapid transacylation known to occur between the two cis-hydroxyls (with a rate of ϳ10 s Ϫ1 ) (21). Thus, while aminoacylation is specific to a particular hydroxyl (2Ј-OH in the cases of IleRS and ValRS (22)) deacylation could potentially occur from either the 2Ј-or 3Ј-OH group.
To address issues that could not be taken up in earlier work because of then existing technical limitations, we constructed a variety of mischarged RNA substrates having different substitutions at the positions of the terminal hydroxyl groups. Using these substrates we were able to test directly whether deacy-lation occurred from a specific hydroxyl, whether transacylation from the 2Ј-to the 3Ј-OH was required for deacylation, and finally, whether a clear chemical role for vicinal hydroxyls could be identified.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-Wild-type E. coli IleRS was overexpressed in E. coli strain MV1184 from plasmid pKS21, which contains the gene for IleRS under control of the lac promoter (23). The T242P mutant of E. coli IleRS was overexpressed in E. coli strain PS2766 (⌬ilv, ⌬ileS203::kan ϩ ) from plasmid pVDC434, a derivative of pBAD18 containing the gene for T242P IleRS under the control of an arabinose-inducible promoter (24). Purification of the wild-type and mutant IleRS was as described previously (25). IleRS concentrations were determined by active site titration (26).
Wild-type E. coli ValRS was overexpressed as a C-terminal His 6tagged protein in E. coli strain BL21(DE3) from a pET-21b (Novagen, Madison, WI) derivative containing the gene for ValRS (27). Standard protocols were used for purification. The T222P mutant of E. coli ValRS was overexpressed in E. coli strain PS2801 (⌬valS::kan ϩ ) from plasmid pVDC447, a pBAD18 derivative containing the gene for T222P ValRS (28). Purification was essentially identical to the protocol used for IleRS. Wild-type and the T379A,T380A double mutant (these mutations are in the editing site and do not affect aminoacylation activity with valine) yeast (Saccharomyces cerevisiae) ValRS were overexpressed as C-terminal His 6 -tagged proteins in the yeast strain CW1, which has the chromosomal copy of ValRS deleted. 2 Both proteins were purified using standard methods for His 6 -tagged proteins. ValRS concentrations were determined by the Bradford dye-binding assay.
The plasmid pET-22-CCA, which encodes a C-terminal His 6 -tagged E. coli tRNA nucleotidyltransferase (tRNA NTase), was a generous gift from the laboratory of Nancy Maizels and Alan Weiner (29). This plasmid was transformed into E. coli BL21(DE3) cells and the resulting strain was used for overexpression and purification. Protein concentration was determined by the Bradford assay.
RNA Substrates-Mature E. coli tRNA Ile (GAU) was isolated from E. coli strain MV1184 containing the plasmid pES300, which allows for the isopropyl-1-thio-␤-D-galactopyranoside-inducible overexpression of tRNA Ile (30). Purification was as described previously (31). The resulting preparations of tRNA Ile were generally comprised of 50 -70% tRNA Ile with the remainder being other cellular tRNAs. Transcripts of tRNA Ile missing the terminal A76 nucleotide (⌬A76 tRNA Ile ) were found to be superior to full-length transcripts for 3Ј-end modification using tRNA NTase. The plasmid pBNt02 was constructed for use as a template for ⌬A76 tRNA Ile transcription in vitro using T7 RNA polymerase. A small segment of DNA was inserted into plasmid pMF15 (32) to place a FokI recognition site, such that cleavage with FokI generated a template strand terminating with C75. Linearization with FokI was done at 37°C for 2 h at an enzyme concentration of 0.3 units/g of DNA.
Transcription reactions and purification were as described previously (33). Mature tRNA Val was purchased from Sigma. Concentrations of tRNA preparations were determined by measuring the plateau level of charging with cognate amino acid.
RNA minihelices were synthesized on an Expedite 8909 synthesizer (PE Biosystems, Foster City, CA) and deprotected as described (34). Minihelices with modified 3Ј-ends were synthesized directly from the corresponding modified nucleoside CPG resins (ChemGenes, Boston, MA). Following deprotection the minihelices were purified on a DNA-Pac PA-100 ion exchange high performance liquid chromatography column (Dionex, Sunnyvale, CA). Minihelix concentrations were determined from absorbance at 260 nm.
tRNA 3Ј-End Modification Reactions-The tRNA NTase-catalyzed pyrophosphorolysis and nucleotide exchange procedure developed in the Hecht laboratory was used exclusively for the preparation of tRNAs with modified 3Ј-terminal bases (35). The only significant change to their procedure was use of E. coli tRNA NTase, instead of the yeast enzyme. The exchange reaction was performed in 20 mM Tris-HCl (pH 8.5), 20 mM MgCl 2 , 1 mM sodium pyrophosphate, 6 mM ATP (or analog), 15 M tRNA, and 5 M tRNA NTase. In all assays where A76 tRNA is reported as the substrate, the tRNA was put through the tRNA NTase exchange procedure using ATP to regenerate the wild-type 3Ј-end. The 3Ј-deoxy-ATP (3Ј-dATP) was purchased from Sigma. The 2Ј-fluoro-and 3Ј-fluoro-ATP were purchased from Purimex (Göttingen, Germany). The reaction was allowed to proceed for 4 h at 37°C and then ethanol precipitated, resuspended, and the product purified on a 10% polyacrylamide, 8 M urea gel. Final products were generally found to be between 98 and 99% modified as judged by aminoacylation assays. The nucleotide exchange procedure was found to work well with the ⌬A76 tRNA Ile transcript, and found to be necessary (as opposed to a simple addition reaction) due to an approximate 10% contamination of ⌬A76 tRNA Ile with full-length transcript, likely from improper T7 RNA polymerase termination. Concentrations of 3Ј-end modified tRNAs were determined (where possible) from their plateau levels of aminoacylation with cognate amino acid.
Aminoacylation Assays-The aminoacylation of RNA substrates was determined by measuring the trichloroacetic acid precipitable radioactivity generated via the attachment of tritiated amino acid to RNA (25). The assays were performed at room temperature in 20 mM HEPES (pH 7.5), 150 mM NH 4  Mischarging Reactions-Mischarged RNA substrates for IleRS were prepared using two primary techniques. Because IleRS and ValRS both charge the 2Ј-OH, producing mischarged RNA substrates that have altered 2Ј-OH groups (i.e. 2Ј-dA) is challenging. A search for reaction conditions where this mischarging could take place revealed that, under conditions of no monovalent cations and 30% dimethyl sulfoxide (Me 2 SO), yeast ValRS had a low, but easily detectable mischarging activity toward 2Ј-dA76 tRNA Ile . Additionally, it was found that the T379A,T380A double mutant yeast ValRS had an equivalent activity under these conditions. (Larger quantities of the mutant ValRS were available to use in mischarging preparations.) Preparative mischarging reaction conditions were 25 mM Tris-HCl (pH 8.5), 10 mM MgCl 2 , 30% Me 2 SO, 1 mM ATP, 10 nM inorganic pyrophosphatase, 5 M [ 3 H]valine (23 mCi/mol), 500 nM yeast ValRS (or double mutant), and 5 M tRNA or 100 M minihelix. Reactions were incubated at room temperature for 4 -6 h, then extracted with phenol/chloroform (1:1) twice, ethanol precipitated, and resuspended in 10 mM sodium acetate (pH 4.5). This method was used for mischarging tRNA Ile transcripts, minihelices, and their respective 3Ј-end variants. By using transcripts and synthetic RNAs it could be guaranteed there was no trace contamination of tRNA Ile and minihelix Ile substrates with tRNA Val species.
For the preparative mischarging of mature tRNA Ile and the related 3Ј-end variants, an enzyme that has no cross-reactivity with any potential remaining tRNA Val in the tRNA samples was required. This requirement was met by using an editing-deficient mutant (T242P) (16) of E. coli IleRS under standard conditions where tRNA recognition is intact. Mischarging reactions were performed in 20 mM HEPES (pH 7.5), 100 M EDTA, 150 mM NH 4 Cl, 10 mM MgCl 2 , 1 mM ATP, 10 nM inorganic pyrophosphatase, and 5-10 M [ 3 H]valine (23 mCi/mol), with 5 M tRNA substrate and 20 M T242P IleRS. Mischarging reactions were incubated at room temperature for 30 -60 min and worked up as described above.
The same strategy was used to prepare mischarged tRNA Val and the respective 3Ј-end variants. The T222P mutant of E. coli ValRS was found to effectively mischarge wild-type and other tRNA Val derivatives with threonine (28). The conditions for mischarging were identical to those for the T242P IleRS mischarging of tRNA Ile , except that 8 M [ 3 H]threonine (50 mCi/mol) replaced valine. T222P ValRS was used at a concentration of 5 M, as was the tRNA substrate. The yield of mischarged product depended on the incubation time more critically than seen with tRNA Ile mischarging reactions, likely due to a low level of editing activity with T222P ValRS. Optimal yields were achieved using a 5-min incubation for mischarging wild-type tRNA Val , a 20-30min incubation with 3Ј-dA76 tRNA Val , and a 1-h incubation with 3Јfluoro-A76 tRNA Val . All reactions were performed at room temperature and worked up as described above.
Deacylation Assays-Deacylation was also measured using the trichloroacetic acid precipitation technique. Deacylation of mischarged mature tRNA Ile was done at room temperature in 150 mM Tris-HCl (pH 7.5), 20 mM MgCl 2 , using 1 M mischarged tRNA and 50 nM IleRS. For comparisons of the deacylation activity of Val-2Ј-dA76 tRNA Ile with other tRNAs, tRNA Ile transcripts were used. Not only were transcripts free of tRNA Val contamination, but the 3Ј-end-modified derivatives also had less contamination of their 3Ј-ends with residual A76 because the exchange reactions utilized ⌬A76 tRNA Ile as the source tRNA (generally 99.5% exchange was achieved as opposed to 98 -99%). The deacylation assays using these mischarged transcript-derived substrates 2 C.-C. Wang, unpublished data.
were performed under the same conditions as those conditions for mature tRNA Ile (isolated from cells) and its 3Ј-end variants, except 600 nM mischarged tRNA and 200 nM IleRS were used. ValRS has a slightly higher deacylation activity, so assays measuring the deacylation of mischarged tRNA Val and its derivatives were done with 10 nM ValRS and 1 M mischarged tRNA. Deacylation assays using mischarged minihelices were also done with 600 nM mischarged minihelix and 200 nM IleRS. Deacylation of Val-3Ј-dA76 tRNA Ile was achieved using very high enzyme concentrations under conditions where substrate recognition is altered. Both mischarged and correctly charged 3Ј-dA76 tRNA Ile (1 M) were deacylated in 150 mM Tris-HCl (pH 8.5), 20 mM MgCl 2 , 20% methanol, using 2 M IleRS.

IleRS Mischarges tRNA Substrates with Altered 3Ј-Ends-
Early work established that IleRS and ValRS utilize the 2Ј-OH as the initial site of aminoacylation, a feature now known to be characteristic of all class I aminoacyl-tRNA synthetases (22,36,37). To further investigate the role of the 3Ј-end of tRNA in determining the specificity and mechanism of the deacylation reaction, a variety of substrates with altered 3Ј-ends were constructed. Replacement of the 3Ј-OH group of tRNA Ile by a hydrogen atom (giving 3Ј-dA76 tRNA Ile ) yields a tRNA that is devoid of editing activity (Fig. 1A) (17). As expected, this mod-ification does not significantly affect its aminoacylation activity relative to A76 tRNA Ile (Fig. 1B). The combination of high aminoacylation activity and lack of editing manifests itself in 3Ј-dA76 tRNA Ile being completely mischarged with valine by IleRS. This mischarging is in striking contrast to the cis-diol containing A76 tRNA Ile , which is not detectably mischarged with valine.
To more closely examine the properties of the 3Ј-OH group of tRNA Ile that are critical for editing, a 3Ј-fluoro-A76 tRNA Ile was constructed. Fluorine better emulates the electronegative properties of the OH group and may even serve as a hydrogen bond acceptor (38 -40). If these were the properties of the OH group that promoted editing, then 3Ј-fluoro-A76 tRNA Ile should be difficult to mischarge. However, IleRS rapidly mischarged 3Ј-fluoro-A76 tRNA Ile with valine (Fig. 1A). As expected, we observed no charging or mischarging with 2Ј-dA76 tRNA Ile or 2Ј-fluoro-A76 tRNA Ile .
It is worth noting that the data in Fig. 1 also demonstrate the purity of the tRNA products isolated from the tRNA NTase exchange reactions. As the relative charging plateaus (isoleucine versus valine) for 3Ј-dA76 tRNA Ile and 3Ј-fluoro-A76 tRNA Ile are equal for both tRNA substrates, it is likely that virtually no wild-type tRNA Ile (which only charges with isoleucine) remains in the preparations. The same conclusion can be drawn from the observation that neither 2Ј-dA76 tRNA Ile nor 2Ј-fluoro-A76 tRNA Ile display any charging activity.
The Terminal 3Ј-OH of tRNA Ile Is Important for Deacylation of Mischarged tRNA Ile -IleRS has a potent deacylation activity toward mischarged A76 tRNA Ile (Fig. 2A). This activity is completely abolished by the replacement of the terminal 3Ј-OH with either a hydrogen or fluorine atom. (While the plots in Fig.  2A show a slow deacylation rate for Val-3Ј-dA76 tRNA Ile and Val-3Ј-fluoro-A76 tRNA Ile , under these conditions the rates were identical to the rates in the absence of enzyme.) Under conditions designed to promote rapid deacylation rates (4 M mischarged tRNA substrate, 1 M IleRS), the complete lack of deacylation activity toward Val-3Ј-dA76 tRNA Ile is striking (Fig. 2B). The A76 tRNA was completely deacylated in 2-3 min, whereas the 3Ј-dA76 tRNA Ile was only about 3% deacylated after 1 h. This amounts to a more than 750-fold decrease in deacylation rate.
IleRS Deacylates Mischarged 2Ј-dA76 Substrates-The results with 3Ј-dA76 tRNA Ile and 3Ј-fluoro-A76 tRNA Ile confirm and expand upon previous work detailing the importance of the 3Ј-OH group in hydrolytic editing. The question still remains as to what is the mechanistic reason behind the importance of the 3Ј-OH. One possibility is that the 3Ј-OH serves a catalytic function to facilitate deacylation from the 2Ј-OH. Alternatively, the mischarged amino acid may be esterified to the 3Ј-OH just prior to hydrolysis. The well characterized 2Ј-to 3Ј-transacylation reaction on the cis-diol-containing wild-type tRNA could reposition the amino acid. A similar transacylation is not possible for either 3Ј-dA76 or 3Ј-fluoro-A76 tRNA Ile . If transacylation were a prerequisite to deacylation, then a mischarged 2Ј-dA76 tRNA Ile (containing a 3Ј-aminoacyl linkage) might be expected to be a substrate for deacylation.
To investigate this possibility we had to develop a method for producing mischarged 2Ј-dA76 substrates. We attempted to mischarge 2Ј-dA76 substrates using a number of different aminoacyl-tRNA synthetases and reaction conditions. For this purpose, we used both transcripts (as opposed to the mature, cell-isolated tRNA used in the previous experiments) of tRNA Ile and RNA minihelices based on the acceptor-T⌿C stem of tR-NA Ile (Fig. 3). (IleRS, like many aminoacyl-tRNA synthetases, specifically charges minihelices based on its cognate tRNA (41,42).) Most mischarging trial reactions were unsuccessful. Nei- ther editing-deficient mutants of IleRS nor E. coli or Bacillus stearothermophilus ValRS had any charging activity with 2Ј-dA76 substrates. The highest degree of mischarging was achieved using ValRS from yeast. While the reactions were not efficient, about 6 -8% of the input 2Ј-dA76 tRNA Ile and 1% of 2Ј-dA76 minihelix Ile was mischarged in incubations using yeast ValRS in the presence of 30% dimethyl sulfoxide (Me 2 SO) and no monovalent cations. The T379A,T380A double mutant yeast ValRS gave identical yields and was used interchangeably with wild-type yeast ValRS (see "Experimental Procedures").
Because the fraction of RNA that was successfully mischarged in these preparations was low, contamination of starting materials with wild-type (non-exchanged) RNAs was investigated. The 2Ј-dA76 tRNA Ile used in these reactions was derived from ⌬A76 tRNA Ile transcripts. This starting material has the advantage of being free of any tRNA Val species (that might be recognized by ValRS) and of having the lowest possible contamination with the cis-diol containing A76 tRNA Ile that can arise from incomplete exchange of the A76 terminus. Un-der standard aminoacylation conditions with isoleucine, 2Ј-dA76 tRNA Ile transcripts showed about 0.5% of the charging level that would be expected based on its absorbance at 260 nM. However, the large excess of 2Ј-dA76 tRNA Ile in these preparations could be inhibiting the aminoacylation of A76 tRNA Ile and thus masking the amount of A76 tRNA Ile present. To rule out this possibility, a control compared the level of aminoacylation of 2Ј-dA76 tRNA Ile alone to that of 2Ј-dA76 tRNA Ile intentionally spiked with 3% A76 tRNA Ile . The reaction containing the 3% A76 tRNA Ile showed complete charging in 10 min (data not shown), while the 2Ј-dA76 tRNA Ile alone showed almost no charging. Thus, the 6 -8% yield of mischarged 2Ј-dA76 tRNA Ile cannot be due to contamination with A76 tRNA Ile . A similar experiment showed that 2Ј-dA76 minihelix Ile was free of wild-type minihelix Ile . This result was expected, as the starting material for chemical RNA synthesis is homogenous.
These mischarged 2Ј-dA76 substrates were tested in deacylation assays relative to their A76 and 3Ј-dA76 variants. Both Val-2Ј-dA76 tRNA Ile and Val-2Ј-dA76 minihelix Ile were efficiently deacylated by IleRS. Relative to Val-A76 tRNA Ile , the 2Ј-dA76 derivative showed an approximate 5-fold decrease in the initial rate of deacylation (Fig. 4A). This decrease is in striking contrast to the 750-fold or more drop in deacylation rate observed with Val-3Ј-dA76 tRNA Ile . The hydroxyl specificity of deacylation was maintained with regards to the minihelix, as a Val-3Ј-dA76 minihelix Ile was completely resistant to hydrolysis (Fig. 4B). No difference in rate was detected between the deacylation rates for Val-minihelix Ile and Val-2Ј-dA76 minihelix Ile .
The 3Ј-OH of tRNA Val Is Important for Preventing Misacylation by ValRS-To compare how editing specificity may have adapted through evolution, we performed analogous experiments with the closely related ValRS. While, no misacylation of A76 tRNA Val was detected in any of the conditions tested, substitution of the 3Ј-OH group of tRNA Val with either a hydrogen or fluorine atom gave a tRNA that is mischarged with threonine ( Fig 5A). Additionally, these modifications had little effect on the aminoacylation activity with valine (Fig. 5B). Under the assay conditions used (Fig. 5, A and B), the plateau level of charging of 3Ј-dA76 tRNA Val with threonine was significantly lower than with valine. When aminoacylation reactions were performed with higher concentrations of ValRS, the charging plateaus with valine and threonine were identical for 3Ј-dA76 tRNA Val (Fig. 5C). (Thus, the nucleotide exchange reactions with 3Ј-dATP gave essentially pure 3Ј-dA76 tRNA Val , and the observed intermediate threonine mischarging plateau (Fig. 5A) cannot be due to contamination with A76 tRNA Val .) Mischarging of 3Ј-fluoro-A76 tRNA Val with threonine was complete even under the conditions of a lower ValRS/tRNA ratio than needed for complete mischarging of 3Ј-dA76 tRNA Val (Fig. 5A).
ValRS Does Not Require an Intact 3Ј-OH to Edit Mischarged tRNA-In contrast to IleRS, a mischarged 3Ј-dA76 substrate is a substrate for deacylation (Fig. 6A). (A similar finding was reported for yeast ValRS (19).) This modification decreases the editing efficiency of both IleRS and ValRS, yet the observed ValRS catalyzed deacylation of Thr-3Ј-dA76 tRNA Val demonstrates a marked difference in substrate specificity for editing. The mischarging with threonine of 3Ј-dA76 tRNA Val is the result of competing aminoacylation and deacylation reactions (Fig. 5A). Because the deacylation activity is reduced it cannot keep pace with the mischarging.
To further pursue investigation of 3Ј-modified tRNA Val substrates, Thr-3Ј-fluoro-A76 tRNA Val was constructed. Little deacylation of Thr-3Ј-fluoro-A76 tRNA Val was detected over the relatively high rate of non-enzyme catalyzed deacylation (Fig.  6A). This result is consistent with the full mischarging of 3Јfluoro-A76 tRNA Val under conditions where 3Ј-dA76 tRNA Val is only partially mischarged. With higher concentrations of ValRS, Thr-3Ј-fluoro-A76 tRNA Val is indeed deacylated by ValRS (Fig 6B). Thus, ValRS has distinct substrate specificity relative to its close homolog IleRS. Whether ValRS simply has broader hydroxyl specificity or has reversed hydroxyl specificity relative to IleRS is unanswered. Attempts to mischarge 2Ј-dA76 FIG. 4. IleRS deacylates mischarged 2-dA76 RNA substrates. A, deacylation of mischarged tRNAs. The 2Ј-dA76 tRNA Ile is deacylated by IleRS, although at a reduced rate relative to A76 tRNA Ile . B, deacylation of mischarged minihelices. Minihelices based on the acceptor/ T⌿C stem of tRNA Ile are specifically deacylated by IleRS at a somewhat slower rate than are the full tRNAs. The hydroxyl specificity is the same as seen with the full tRNAs.
tRNA Val by a variety of approaches and subsequently measure deacylation of Thr-2Ј-dA76 tRNA Val were unsuccessful.
The Editing Site of IleRS Can Adapt to Hydrolyze 2Ј-O-Acyl Linkages-The results with ValRS raised the possibility that editing is inherently "plastic" and that, given the appropriate stimulus, IleRS might also deacylate substrates mischarged at the 2Ј-OH. Here, the minihelix system was of particular advantage relative to the use of tRNA substrates. A 3Ј-O-methyl-A76 minihelix Ile (3Ј-OMe-A76) could be synthesized from commercially available starting materials, whereas 3Ј-OMe-ATP is not a substrate for tRNA NTase. A mischarged 3Ј-OMe-A76 minihelix Ile has a methyl group that prevents a 2Ј-to 3Ј-transacylation from occurring. Although this molecule has an intact 2Ј-OH, it was not aminoacylated by IleRS. Successful aminoacylation was only achieved using yeast ValRS under the same conditions (30% Me 2 SO and no monovalent cations) that were used for mischarging 2Ј-dA76 substrates.
Interestingly, Val-3Ј-OMe-A76 minihelix Ile was efficiently deacylated by IleRS (Fig. 7A). To verify that this deacylation activity was catalyzed by the editing site in CP1, we demonstrated that the editing-deficient mutant T242P IleRS was unable to deacylate Val-3Ј-OMe-A76 minihelix Ile (data not shown). (The T242P mutation is located in the editing pocket of CP1 (12).) The rate of Val-3Ј-OMe-A76 minihelix Ile deacylation is only 2-fold reduced from that of wild-type minihelix Ile and, thus, significantly faster than the deacylation-resistant Val-3Ј-dA76 minihelix Ile . Because the steric bulk of the O-methyl group could alter the positioning of the mischarged RNA terminus in the editing site so as to place the hydrolytic machinery in close proximity to the aminoacyl linkage, a search for conditions that might alter the position of the 2Ј-O-acyl linkage of Val-3Ј-dA76 tRNA Ile within CP1 was undertaken. Importantly, Val-3Ј-dA76 tRNA Ile is efficiently hydrolyzed in the presence of high enzyme concentrations and 20% methanol (Fig. 7B). However, it is worth noting that these conditions also promoted the hydrolysis of Ile-3Ј-dA76 tRNA Ile . Thus, under these conditions the aminoacyl terminus is bound in a modified orientation. DISCUSSION A mischarged 3Ј-dA76 tRNA Ile is completely resistant to deacylation by IleRS. This resistance is most likely because the FIG. 6. Deacylation of mischarged tRNA Val variants. A, using 10 nM ValRS, mischarged tRNA Val is rapidly deacylated. Unlike IleRS, ValRS shows deacylation activity toward a mischarged 3Ј-dA76 tRNA. The deacylation rate for this substrate is reduced relative to A76 tRNA-Val . Under these conditions, the deacylation activity of mischarged 3Ј-fluoro-A76 tRNA Val (Thr-3ЈF-A76) is so low that it is indistinguishable from the background rate. B, the modified tRNAs require higher concentrations of ValRS to attain similar deacylation activity as that of A76 tRNA Val .

FIG. 7. IleRS catalyzed deacylation of 2-O-esters.
A, IleRS can deacylate a mischarged 3Ј-OMe-A76 minihelix Ile . The rate is ϳ2-fold reduced from that of the wild-type or 2Ј-dA76 minihelix. B, with 20% methanol added to the solvent, even a mischarged 3Ј-dA76 tRNA Ile can be deacylated. Under these conditions the molecular recognition in the editing active site is disrupted to the point where properly charged 3Ј-dA76 tRNA Ile is also deacylated. aminoacyl linkage prevents the scissile ester bond from coming into close proximity of the hydrolytic machinery while bound in the editing site. The magnitude of the rate difference (Ն750fold) between deacylation of Val-A76 tRNA Ile and Val-3Ј-dA76 tRNA Ile is larger than one might expect if the 3Ј-OH had a noncovalent role in deacylation. The calculated difference in transition state stabilization of ϳ4 kcal/mol is beyond the range of energies observed for average hydrogen bonds or the electronegative inductive effect of a neighboring hydroxyl. Correspondingly, a fluoro group in place of the 3Ј-OH failed to allow for even partial deacylation activity. Earlier work did not directly investigate the deacylation of Val-2Ј-dA76 tRNA Ile and, given the available data, concluded that the role of the 3Ј-OH was to assist catalysis of deacylation from the 2Ј-OH (43). Here, reasonable quantities of mischarged 2Ј-dA76 tRNA Ile were produced, isolated, and directly shown to be deacylated by IleRS. This mischarged substrate, with a fixed 3Ј-O-aminoacyl linkage, showed only a modest 5-fold decrease in initial rate of deacylation. This rate decrease is in the range of what might be expected for the loss of neighboring group effects due to the missing 2Ј-OH (20). Thus, under normal circumstances, transacylation from the 2Ј-to the 3Ј-OH is required for deacylation of Val-tRNA Ile by IleRS.
Although IleRS cannot deacylate Val-3Ј-dA76 tRNA Ile , ValRS deacylates Thr-3Ј-dA76 tRNA Val with an approximate 10-fold reduction in rate compared with deacylation of Thr-A76 tRNA Val . Still, it is possible that ValRS preferentially deacylates 3Ј-O-aminoacyl esters. Even though the initial site of aminoacylation is the 2Ј-OH, because transacylation is more rapid than deacylation (21), a 3Ј-O-aminoacyl ester could be the main substrate for editing. In that event, a 10-fold reduced deacylation rate of Thr-3Ј-dA76 tRNA Val relative to Thr-A76 tRNA Val would not be surprising.
Alternatively, the role of the 3Ј-OH of tRNA Val may be as a hydrogen bond donor (Fig. 8). In this model, derived from studies of non-catalyzed ester hydrolysis by Bruice and coworkers (20), the neighboring hydroxyl donates a hydrogen to the oxyanion of the tetrahedral intermediate, thereby helping to stabilize the build up of negative charge (Fig. 8, left). The 3Ј-dA analog is obviously unable to fulfill this role (Fig. 8,  middle). Not only is the fluoro group unable to donate a hydrogen bond, but its partial negative charge likely also induces some electrostatic repulsion that distorts the tetrahedral intermediate (Fig. 8, right). This model fits best with the observed trends in deacylation rates for these substrates.
The observed deacylation of Val-3Ј-OMe-A76 minihelix Ile by IleRS demonstrates a shift in substrate specificity for deacylation reminiscent of the substrate specificity ValRS shows. The mechanism by which IleRS is able to hydrolyze this 2Ј-Oaminoacyl ester likely arises from some local structural rearrangement induced by the bulkier 3Ј-end as opposed to a hydrogen bonding role for the 3Ј-OMe group (the 3Ј-OH cannot play a similar hydrogen-bonding role during deacylation of Val-2Ј-dA76 substrates). In the context of a 2Ј-O-aminoacyl ester, the editing active site may be too crowded to accommodate the 3Ј-OMe group in the normal binding orientation. Alternatively, there could be a small hydrophobic interaction that is made with the 3Ј-O-Me group, also altering the position of the scissile ester bond with respect to the catalytic center. Such local structural perturbations are certainly feasible within the IleRS-editing site, as evidenced by the deacylation activity seen in the presence of 20% methanol.
Regardless of the detailed mechanism, the results show that the editing site is inherently plastic with respect to recognition of mischarged tRNA. For IleRS and ValRS to discriminate against valine and threonine, respectively, the editing sites of both enzymes must recognize subtle differences in aminoacyl side chains. Yet this fine structure discrimination must also be flexible, as both enzymes have been shown to edit multiple non-cognate amino acids (44). A third, related synthetase, leucyl-tRNA synthetase, has also been reported to edit multiple amino acids (45,46). This remarkable combination of specificity and plasticity in recognition of the aminoacyl side chain is shown here to include the position of the aminoacyl group on the 3Ј-end of tRNA. This plasticity may have developed over a long period of evolution and may have been particularly important in the early stages of the development of aminoacylation systems and the genetic code, when aminoacyl-RNA substrates were not perfected. Indeed, the CP1 insertion is ancient, as it is conserved throughout evolution and found in deeply rooted organisms of all three kingdoms such as the Thermotogales, Crenarchaeota, and Diplomonads (47).