The tRNA A76 Hydroxyl Groups Control Partitioning of the tRNA-dependent Pre- and Post-transfer Editing Pathways in Class I tRNA Synthetase*

Background: Isoleucyl-tRNA synthetase uses cognate tRNA to stimulate hydrolysis of non-cognate aminoacyl-adenylates within the synthetic site. Results: The 3′-terminal hydroxyl groups of tRNAIle have no role in pre-transfer editing. Conclusion: The tRNAIle body, rather than the 3′-end of tRNAIle alone, promotes assembly of the improved ribonuclear protein synthetic site. Significance: Isoleucyl-tRNA synthetase acts as a ribonuclear protein to adjust amino acid recognition to the cellular environment. Aminoacyl-tRNA synthetases catalyze ATP-dependent covalent coupling of cognate amino acids and tRNAs for ribosomal protein synthesis. Escherichia coli isoleucyl-tRNA synthetase (IleRS) exploits both the tRNA-dependent pre- and post-transfer editing pathways to minimize errors in translation. However, the molecular mechanisms by which tRNAIle organizes the synthetic site to enhance pre-transfer editing, an idiosyncratic feature of IleRS, remains elusive. Here we show that tRNAIle affects both the synthetic and editing reactions localized within the IleRS synthetic site. In a complex with cognate tRNA, IleRS exhibits a 10-fold faster aminoacyl-AMP hydrolysis and a 10-fold drop in amino acid affinity relative to the free enzyme. Remarkably, the specificity against non-cognate valine was not improved by the presence of tRNA in either of these processes. Instead, amino acid specificity is determined by the protein component per se, whereas the tRNA promotes catalytic performance of the synthetic site, bringing about less error-prone and kinetically optimized isoleucyl-tRNAIle synthesis under cellular conditions. Finally, the extent to which tRNAIle modulates activation and pre-transfer editing is independent of the intactness of its 3′-end. This finding decouples aminoacylation and pre-transfer editing within the IleRS synthetic site and further demonstrates that the A76 hydroxyl groups participate in post-transfer editing only. The data are consistent with a model whereby the 3′-end of the tRNA remains free to sample different positions within the IleRS·tRNA complex, whereas the fine-tuning of the synthetic site is attained via conformational rearrangement of the enzyme through the interactions with the remaining parts of the tRNA body.

The major players, aminoacyl-tRNA synthetases (aaRSs), 2 catalyze covalent coupling of cognate amino acids with the cognate tRNAs in a two-step reaction of aminoacylation (1). The first step comprises amino acid activation, whereby the amino acid is condensed with ATP to form an aminoacyl-adenylate (aa-AMP) intermediate and pyrophosphate (PP i ). In the second step, the aminoacyl moiety is transferred to the 3Ј-end of the cognate tRNA with the release of AMP (Fig. 1). This step proceeds by nucleophilic attack of either the 2Ј-OH or 3Ј-OH group of the terminal ribose of the tRNA substrate on the carbonyl carbon atom of the aminoacyl-AMP intermediate. Both steps are catalyzed within the same synthetic site located at the aaRS catalytic core. aaRSs are divided into two classes, class I and class II, based on different topology of the synthetic sites and the characteristic sequence motifs (2,3).
Arginyl-, glutaminyl-, glutamyl-, and class I lysyl-tRNA synthetases are unable to activate amino acids in the absence of tRNA. In these enzymes, a conformational change induced by tRNA binding is essential for the productive juxtaposition of the ATP-and amino acid-reactive moieties. Other aaRSs from both classes may activate the amino acid substrate independently of the presence of tRNA, yet the activation step in these enzymes may be modulated by tRNA (4 -8). It thus appears that many aaRSs act as ribonuclear proteins regardless of whether the tRNA portion of the ribonuclear protein is required for amino acid activation.
Some aaRSs are unable to establish a high level of specificity against structurally similar amino acids in the synthetic reaction alone. For these enzymes, threshold levels of aminoacylation fidelity, set up by the overall accuracy of protein synthesis (1 in 10 4 ) (9), are ensured by the subsequent proofreading. AaRSs challenged with the less discriminative synthetic sites evolved inherent hydrolytic editing (1,10) to hydrolyze non-cognate aminoacyl-AMP intermediate (pre-transfer editing; Fig. 1, pathways 1 and 3) and/or misaminoacylated tRNA (posttransfer editing; Fig. 1, pathway 4). Post-transfer editing is located in the well defined domain dedicated to proofreading only (editing domain). It is a prominent aaRS editing pathway that coevolved with the acquisition of the additional hydrolytic module appended to the ancient aaRS catalytic core. In this pathway, the transiently misaminoacylated 3Ј-end of the tRNA translocates from the synthetic to the distant editing site, bringing along the non-cognate amino acid. Conversely, pre-transfer editing occurs within the synthetic site (11)(12)(13)(14)(15)(16), where the aminoacyl-AMPs are formed. This activity may or may not be stimulated by tRNA (tRNA-independent and tRNA-dependent pretransfer editing; Fig. 1, pathways 1 and 3, respectively). Recent findings indicate that tRNA-independent pre-transfer editing is a common editing pathway that is not strictly correlated with the presence or absence of the editing domains in the corresponding aaRSs (11,14,16). This activity is generally weak (contributes less than 5% of total editing) and seems to be a remnant of such an activity that may have existed in the ancient aaRSs. In contrast, tRNA-dependent pre-transfer editing is very rarely used by aaRSs, although it may contribute more significantly to total editing (see below).
Isoleucyl-tRNA synthetase (IleRS) catalyzes the formation of isoleucyl-tRNA Ile (Ile-tRNA Ile ) in all domains of life. This enzyme discriminates against the structurally similar non-cognate valine with only 200-fold specificity that arises from weak ground state binding alone (12,17). To ensure high fidelity in aminoacylation, IleRS employs tRNA-dependent hydrolytic editing to clear mistakenly recognized valine (18 -20). Structural and kinetic analyses revealed that the large peptide fragment (connective peptide, CP1), inserted into the Rossmann fold catalytic domain, folds into a well defined domain dedicated to post-transfer editing (21)(22)(23)(24). IleRS shares this feature with homologous class Ia leucyl-and valyl-tRNA synthetases (LeuRS and ValRS, respectively) (25)(26)(27). Along with deacylation of valyl-tRNA Ile (Val-tRNA Ile ) at the CP1 domain, IleRS also displays a distinctive capacity to hydrolyze valyl-AMP in a tRNA-dependent manner within the synthetic site (12,28). We have recently performed a comprehensive kinetic analysis of selected mutants within the editing (12) and synthetic sites (28) to demonstrate that tRNA-dependent pre-transfer editing contributes up to 30% of total IleRS editing, with the remaining 70% assignable to the post-transfer pathway.
How tRNA stimulates pre-transfer editing within the IleRS synthetic site and whether this activity is intimately linked with tRNA aminoacylation remain elusive. An early observation that isoleucine and tRNA binding to IleRS are interdependent (29) raised the possibility that tRNA is a more versatile modulator of IleRS amino acid recognition than anticipated. With this in mind, we performed comprehensive steady-state and singleturnover kinetic analyses to establish the role for tRNA in amino acid recognition and editing by IleRS. Our data show that tRNA Ile stimulates aminoacyl-AMP hydrolysis and concomitantly decreases IleRS affinity toward amino acids. Neither of these activities depends on the capacity of the tRNA Ile to accept amino acid. We further demonstrate that the tRNA cofactor does not facilitate discrimination between isoleucine and valine in either of these reactions. Despite these findings, the IleRS⅐tRNA Ile complex is probably a superior catalyst under cellular conditions compared with IleRS alone because (i) its affinity for amino acids is better adjusted to the cellular environment, and (ii) only in the case of valine is the tRNA-dependent rate of hydrolysis comparable with the rate of the aminoacyl transfer step, and thus contributes to proofreading. The separation of the pre-and post-transfer editing pathways by various tRNA Ile analogues clearly demonstrated that tRNA participates in IleRS proofreading by two distinct mechanisms; (i) the intact 3Ј-end of tRNA permits post-transfer editing by supporting the misaminoacylation/deacylation pathway, whereas (ii) the remaining portions of the tRNA body promote a conformational change that stimulates pre-transfer editing within the IleRS synthetic site.

Experimental Procedures
Production and Purification of IleRS and tRNA Ile -IleRS enzymes (wild type (WT) and deacylation-defective variant (D342A)) were overexpressed and purified by standard nickelnitrilotriacetic acid chromatography as described (12). tRNA-GAU Ile (with G 1 :C 72 instead of the native A 1 :U 72 sequence) was overexpressed in Escherichia coli and purified as described (12). Plateau aminoacylation using [ 14 C]isoleucine established that the sample contains ϳ70 -75% functional tRNA Ile . Substitution of the first tRNA Ile base pair was used to enhance transcription by T7 RNA polymerase, because it was shown not to affect isoleucylation parameters (30). The kinetic competence of G 1 :C 72 tRNA Ile in the synthetic and editing pathways was confirmed by testing the native tRNA GAU Ile isoacceptor isolated from the bulk E. coli tRNA by complementary biotinylated oligonucleotides (sequence of the oligonucleotide: 5Ј-ATCAGGGGT-GCGCTCTAACCACCTGAG-3Ј-biotin). No significant difference in the rate of single-turnover aminoacyl transfer step or in the steady-state rate of aminoacylation was observed relative to the G 1 :C 72 tRNA GAU Ile . Plateau aminoacylation using [ 14 C]isoleucine established that the native tRNA GAU Ile sample contains ϳ85% functional tRNA Ile .
tRNA Ile 3Ј-End Modification-Oxidized tRNA Ile (tRNA ox Ile ) was produced by incubating 3 M tRNA Ile in 55 mM NaOAc, pH 4.5, and 6 mM NaIO 4 for 2 h at room temperature in the dark. The excess NaIO 4 was consumed with 300 mM ethylene glycol (incubation for 30 min at room temperature). tRNA ox Ile was ethanol-precipitated, dialyzed against 5-10 mM HEPES, pH 7.5, FIGURE 1. Schematic presentation of editing pathways. Pre-transfer editing can occur through enhanced dissociation of non-cognate aminoacyl-adenylate (2) or its enzymatic hydrolysis (1 and 3), which may be tRNA-independent (1) or tRNA-dependent (3). After transfer, misaminoacylated tRNA can be deacylated through post-transfer editing (4). and stored at Ϫ20°C. Both 2ЈdtRNA Ile (2Ј-OH group of A76-tRNA Ile replaced by a hydrogen atom) and 3ЈdtRNA Ile (3Ј-OH group of A76-tRNA Ile replaced by a hydrogen atom) were produced using tRNA nucleotidyltransferase (31). Preparative A76 exchange reactions were performed by incubating 75 M tRNA Ile with 25 M tRNA nucleotidyltransferase and 6 mM nucleotide (2ЈdATP or 3ЈdATP, titrated to pH 7.0) in 200 mM Tris, pH 8.0, 1 mM NaPP i , 20 mM MgCl 2 , and 0.5 mM DTT for 4 h at 37°C. tRNA was extracted by phenol-chloroform, ethanol-precipitated, and dialyzed against 5-10 mM HEPES, pH 7.5. Residual tRNA Ile present in 2ЈdtRNA Ile or 3ЈdtRNA Ile samples (ϳ10%) was destroyed by oxidation with NaIO 4 , as described above. The control A76-tRNA Ile was prepared by a preparative exchange reaction in the presence of 6 mM ATP, to test whether the exchange procedure affects the editing or aminoacylation activity. We noticed that the exchanged tRNA is prone to instability, aggregation, and precipitation if dissolved in pure water and kept at Ϫ20°C. Therefore, it was stored in 5-10 mM HEPES at pH 7.5, and all kinetic assays contained 150 mM NH 4 Cl to prevent aggregation.
Preparation of Radiolabeled tRNA-[ 32 P]tRNA Ile or 3Јd[ 32 P]-tRNA Ile was prepared using tRNA nucleotidyltransferase (12,31) in the presence of [␣-32 P]ATP and tRNA Ile or 3Јd[␣-32 P]ATP and 3ЈdtRNA Ile , respectively. Because radiolabeling decreases tRNA Ile plateau in aminoacylation, the measured aminoacylation steady-state rate constants are corrected by a factor that takes into account the proportion of functional [ 32 P]tRNA Ile . Aminoacylated or misaminoacylated [ 32 P]tRNA Ile was prepared by mixing approximately 300 nM [ 32 P]tRNA Ile with 1 M D342A IleRS, 4 mM ATP, 0.004 units/l IPPase, and 2 mM valine or isoleucine in the standard IleRS reaction buffer (see above). tRNA was extracted by acidic phenol and subjected to two subsequent gel filtration steps (Micro Bio-Spin P30 columns) and dialyzed against 10 mM NaOAc, pH 5.0. Aminoacylated and misaminoacylated [ 32 P]tRNAs were inactive in deacylation and therefore subjected to renaturation prior to use by heating to 80°C, adding MgCl 2 to 10 mM final concentration, and slow cooling to ambient temperature. Acidic conditions (5 mM NaoAc, pH 5.0) were preserved during the renaturation period to ensure stability of the ester bond. The percentage of aminoacylated tRNA did not change after renaturation.
ATP-PP i Exchange Assay-The ATP-PP i exchange assay was performed at 37°C in the standard IleRS reaction buffer (see above) supplemented with 100 g/ml BSA for enzyme stabilization, 4 mM ATP, and 1 mM [ 32 P]PP i . Amino acid concentration was varied over the range of 0.1-10 times the K m value, and 25 nM WT or D342A IleRS was used. To assess the role of tRNA in amino acid recognition, ATP/PP i exchange reactions were performed in the presence of 15-20 M non-aminoacylable tRNA variants (tRNA ox Ile , 2ЈdtRNA ox Ile , and Val-tRNA Ile ). When the capacity of Val-tRNA Ile to modulate amino acid activation was tested, the assay was performed in the presence of deacylation-defective D342A IleRS.
Aminoacyl-Adenylate Synthesis Assay-The reactions were conducted as described previously (8,12). Briefly, 25 nM WT IleRS was used at 37°C in the standard IleRS reaction buffer supplemented with 100 g/ml BSA in the presence of 0.5 mM [␣-32 P]ATP (0.01-0.1 mCi/ml), 0.004 units/l IPPase with (tRNA-dependent editing) or without (tRNA-independent editing) 15-20 M tRNA Ile or tRNA Ile analogues. The reactions were started by the addition of amino acids and quenched by mixing 1.5 l of the reaction with 3 l of 1.5 M formic acid. aa-[ 32 P]AMP, [ 32 P]AMP, and [ 32 P]ATP were separated by thin layer chromatography and independently quantitated as described (8,12,26). In the presence of aminoacylated or misacylated tRNAs, AMP formation was followed using the D342A IleRS.
Determination of AMP/Aminoacyl-tRNA Ile Ratios-[ 32 P]AMP and aminoacyl-[ 32 P]tRNA Ile formations were followed in parallel steady-state assays (26,28). The reactions were performed at 37°C in the standard IleRS buffer supplemented with 100 g/ml BSA and 0.004 units/l IPPase, using either 40 nM WT or D342A IleRS, 100 M ATP, 15 M tRNA Ile variant, and 10 mM Ile or 30 mM Val. AMP formation was followed with trace quantities of [␣-32 P]ATP added to the reaction mixtures, whereas aminoacyl-tRNA Ile formation was followed via trace quantities of [ 32 P]tRNA Ile or 3Јd[ 32 P]tRNA Ile . The ratios in the presence of tRNA Ile and the WT or D342A IleRS were lower than previously published (28) because of a modification in the experimental conditions. We have shown that the presence of 150 mM NH 4 Cl (used in this work), slightly impedes the k obs for AMP production, thereby lowering the AMP/aminoacyl-tRNA Ile ratio.
Single-turnover Transfer Step-The IleRS⅐aa-AMP complexes were preformed in situ by incubation of 20 -30 M WT or D342A IleRS with 5 mM isoleucine or 50 mM valine in the standard IleRS reaction buffer supplemented with 8 mM ATP and 0.008 units/l IPPase for 5 min at 37°C, similarly as described (12,26). The single-turnover transfer reactions were then performed with the KinTek RQF-3 instrument by mixing equal volumes of 20 -30 M IleRS⅐aa-AMP with trace amounts of [ 32 P]tRNA Ile or 3Јd[ 32 P]tRNA Ile . Reactions were quenched with NaOAc (pH 4.5, final concentration 0.8 M), and the collection tubes contained SDS (final 0.1% (w/v)). MAY 29, 2015 • VOLUME 290 • NUMBER 22

JOURNAL OF BIOLOGICAL CHEMISTRY 13983
Single-turnover Aminoacylation-Single-turnover aminoacylation was performed with the KinTek RQF-3 instrument by mixing equal volumes of 20 M D342A IleRS, trace amounts of [ 32 P]tRNA Ile , 8 mM ATP, and 0.008 units/l IPPase in the standard IleRS buffer, with various concentrations of isoleucine or valine also present in the standard IleRS buffer. Time courses were fitted to a first-order exponential equation to extract the observed rate constants (k chem ), and the K d values were determined by fitting the observed rate constants to the hyperbolic binding equation, Y ϭ S 0 ϫ k chem /(S 0 ϩ K d ) ϩ c, where S 0 represents the initial amino acid concentration, k chem represents the maximal aminoacylation rate, and c is the intercept (32). Various mixing modes showed no influence on the extracted rate constants.
Single-turnover Deacylation-Single-turnover deacylation reactions were performed with the KinTek RQF-3 instrument or by hand sampling (slow reactions with D342A IleRS or 3ЈdtRNA Ile ). 100 -500 nM renatured Val-or Ile-[ 32 P]tRNA Ile in 5 mM NaOAc, pH 5.0, and 10 mM MgCl 2 was mixed with 30 M WT IleRS or D342A IleRS in the standard IleRS reaction buffer supplemented with an additional 75 mM HEPES, pH 7.5 (in total 150 mM HEPES, pH 7.5). The reactions were quenched with 0.8 M NaOAc, pH 4.5 (final concentration), and the collection tubes contained SDS (final 0.1% (w/v)). Data were fitted to the single exponential equation, y ϭ Y 0 ϩ A ϫ e Ϫkdeacyl ϫ t , where Y 0 represents the y intercept, A is the amplitude, k deacyl represents the apparent deacylation rate constant, and t is time. We observed a biphasic Val-tRNA Ile deacylation curve with the WT IleRS. The reported rate constant corresponds to the fast phase with 50% amplitude. The origin of the slow phase is unknown and has been previously reported (17,19,26).

Results
The Intact 3Ј-End Is Not Required for the Participation of tRNA in Pre-transfer Editing-IleRS exhibits inherent hydrolytic activity toward aminoacyl-AMP, which can be substantially modulated by the tRNA (Table 1) (12,28). It has been proposed (18, 33) that only tRNAs active in aminoacylation are capable of stimulating IleRS editing. Our finding that post-transfer editing is the major editing pathway in IleRS (28) rationalizes earlier work, yet the question of whether the minor tRNA-dependent pre-transfer editing pathway also requires a tRNA cofactor active in aminoacylation still remains. To experimentally address this question, we performed detailed steadystate analysis of IleRS editing in the presence of various tRNA Ile analogues ( Table 2) using a highly sensitive editing assay (8) that employs [␣-32 P]ATP and allows detection of aminoacyl-[ 32 P]AMP and its hydrolytic product [ 32 P]AMP after separation by thin layer chromatography. Participation of tRNA in IleRS pre-transfer editing was first examined with tRNA analogues that lack amino acid acceptor activity. Replacement of the 2Ј-OH group of tRNA Ile by a hydrogen atom was accomplished by exchange of the terminal A76 with 2ЈdATP using tRNA nucleotidyltransferase (34). As expected, the IleRS complex with 2ЈdtRNA Ile loses aminoacylation capacity (Table 2), because IleRS utilizes the 2Ј-OH moiety as the initial site of aminoacylation (34 -36). The tRNA Ile fraction that had remained unexchanged and thus retained the wild-type activity (ϳ10%) was inactivated by periodate oxidation (4). This yields 2ЈdtRNA ox Ile , which completely lacks isoleucine acceptor activity (Table 2). Next, we produced tRNA ox Ile by periodate treatment of the intact tRNA Ile . In this case, we also observed a concomitant loss of aminoacylation activity as expected due to oxidation of the vicinal diol of the terminal ribose of the tRNA sample (Table 2).
Remarkably, both 2ЈdtRNA ox Ile and tRNA ox Ile analogues stimulated AMP formation by IleRS in the presence of valine. Indeed, the k cat values increased about 3-5-fold relative to the k cat in the absence of tRNA (0.128 and 0.07 s Ϫ1 versus 0.027 s Ϫ1 ; Table 1). An important implication of this finding is that participation of tRNA Ile in pre-transfer editing is not coupled to its capacity to accept the amino acid as previously presumed (37) (see "Discussion"). Neither E. coli tRNA Val nor tRNA Leu is capable of stimulating IleRS pre-transfer editing (Table 3), whereas only weak stimulation of valyl-AMP hydrolysis was observed in the presence of a tRNA Ile in vitro transcript (Table 3), known to be a very weak substrate in isoleucylation (30). Our results show that the operation of the tRNA-dependent pre-transfer editing pathway in IleRS requires interaction with the specific cognate and post-transcriptionally modified tRNA Ile body but does not require intactness of the tRNA 3Ј-end.

tRNA Modulates Amino Acid Recognition by IleRS
The aforementioned tRNA Ile analogues also enhance AMP accumulation in the presence of cognate isoleucine ( Table 1). The observed rate of AMP formation in the presence of the tRNA analogues is about 100-fold higher than the non-enzymatic hydrolysis of isoleucyl-AMP in solution ((2.2 Ϯ 0.2) ϫ 10 Ϫ4 s Ϫ1 ) (13). Consistent with our recent findings (28), these data show that the IleRS synthetic site may exhibit weak tRNAdependent hydrolysis of isoleucyl-AMP. In these experiments, the isoleucyl transfer step was blocked by utilization of the aminoacylation-inactive tRNA Ile analogues, thus allowing isolation and monitoring of the tRNA-dependent isoleucyl-AMP hydrolysis step. Kinetic analysis then reveals that tRNA stimulates cognate and non-cognate aminoacyl-AMP hydrolysis similarly by 5-10-fold (Table 1). It thus appears that tRNA does not contribute to IleRS specificity by preferentially promoting noncognate aminoacyl-AMP hydrolysis. Nevertheless, only the non-cognate amino acid is essentially proofread in the presence of intact tRNA. This is so because even in the presence of tRNA, the rate of isoleucyl-AMP hydrolysis is 2 orders of magnitude slower than the rate of the isoleucyl transfer step. Thus, in the cognate system, aminoacyl transfer step is kinetically favored and outcompetes tRNA-dependent proofreading, as confirmed by the AMP/ Ile -tRNA Ile close to 1 (28). This may explain why this unintended activity remains dormant within the synthetic site. It may also provide some clues as to why synthetic site editing did not evolve toward the major editing pathway (28).
3ЈdtRNA Ile Is a Key Analogue for Studying Interplay of Aminoacylation and Pre-transfer Editing-tRNA Ile lacking the 3Ј-OH at the terminal adenosine of tRNA was produced by exchanging A76 with 3ЈdATP using tRNA nucleotidyltransferase. The 3ЈdtRNA sample was further oxidized by periodate to inactivate traces of unexchanged tRNA Ile . Single-turnover kinetics ( Fig. 2 and Table 4) and steady-state analysis ( Table 2) were then used to provide quantitative assessment of kinetic competence of the 3ЈdtRNA analogue in the synthetic and editing IleRS reactions.
The capacity of 3ЈdtRNA to participate in the aminoacyl transfer step was tested with both cognate and non-cognate amino acid. Use of the IleRS variant with inactivated posttransfer editing is required in this assay to allow for accumulation of Val-tRNA Ile . Substitution of the conserved Asp-342, located at the CP1 editing site, with alanine, results in an enzyme incapable of Val-tRNA Ile deacylation (12,37). The enzyme⅐aa-AMP complex was formed in situ by preincubation of the wild-type or deacylation-defective IleRS with ATP and isoleucine or valine, respectively, in one syringe of the rapid chemical quench instrument and mixed with the limiting amount of either [ 32 P]3ЈdtRNA Ile or [ 32 P]tRNA Ile . The aminoacyl-[ 32 P]tRNA formation in time fits best to the single exponential equation, allowing extraction of the k trans values (Fig. 2). The rates of isoleucyl transfer to tRNA with the intact 3Ј-end and to 3ЈdtRNA were the same (3.8 s Ϫ1 versus 3.3 s Ϫ1 ; Fig. 2A), indicating full kinetic competence of this analogue in Ile-tRNA Ile synthesis. In accordance, transfer of the non-cognate valyl moiety to tRNA was also independent of the presence of the A76 3Ј-OH group (Fig. 2B). This supports the conclusion that the specificity of the aminoacyl transfer step in IleRS is not modulated by the 3Ј-OH group at the 3Ј-terminal ribose of its tRNA substrate.
Next, hydrolysis of misaminoacylated 3ЈdtRNA was followed under single-turnover conditions, and the deacylation rate constants (k deacyl ) were extracted from the exponential decay of aminoacyl-tRNA in time. IleRS displayed a 10 6 -fold drop in the k deacyl for Val-3ЈdtRNA Ile relative to Val-tRNA Ile (2.3 ϫ 10 Ϫ5 s Ϫ1 versus 64 s Ϫ1 ; Table 4). Interestingly, the D342A substitution affects the rate of Val-tRNA Ile deacylation less than does the lack of the A76 3Ј-OH group (0.014 s Ϫ1 versus    Table 4). This cements the notion that the A76 3Ј-OH group plays an essential role in IleRS post-transfer editing. As IleRS rapidly aminoacylates 3ЈdtRNA Ile and cannot deacylate Val-3ЈdtRNA Ile , the 3ЈdtRNA analogue allows isolation of pretransfer editing. This is consistent with the previous findings (34) and provides a rationale for use of this analogue in analyzing the interdependency of aminoacylation and pre-transfer editing (see below). IleRS displays weak hydrolytic activity toward the cognate Ile-tRNA Ile product within the editing domain (0.058 s Ϫ1 ; Table 4). This confirms that the CP1 domain exhibits high specificity against the cognate amino acid, as expected. This feature is shared with LeuRS (26) and ValRS (38).
The A76 2Ј-OH Group Does Not Organize the Pre-transfer Editing Subsite-The participation of the A76 2Ј-OH group in pre-transfer editing was tested by monitoring the initial rates of AMP and aminoacyl-tRNA formation in parallel in the presence of 3ЈdtRNA ox Ile . The strength of this methodology, recently developed in our work on LeuRS (26) and IleRS (28), resides in its capacity to define the AMP fraction that originates particularly from editing. The initial rates of [ 32 P]AMP and Val-3Јd[ 32 P]tRNA ox Ile formations were extracted from two parallel assays relying on either [ 32 P]ATP or 3Јd[ 32 P]tRNA, respectively. In the presence of wild-type IleRS and 3ЈdtRNA ox Ile , the AMP formation (k AMP ϭ 0.14 s Ϫ1 ; Table 5) results from both valyl-AMP hydrolysis (tRNA-dependent pre-transfer editing) and Val-3ЈdtRNA Ile synthesis. Because Val-3ЈdtRNA Ile is not deacylated by IleRS (Table 4), post-transfer editing does not contribute to AMP accumulation in this case. The independently determined rate of Val-3Јd[ 32 P]tRNA synthesis (k Val-tRNA ϭ 0.083 s Ϫ1 ; Table 5) quantitates the fraction of total AMP that originates from aminoacylation. The rate constant associated with tRNA-dependent pre-transfer editing alone is then obtained by subtracting k Val-tRNA from k AMP (k ed ϭ 0.057 s Ϫ1 ; Table 5). The calculated k ed obtained in the presence of 3ЈdtRNA thus represents pre-transfer editing stimulated with aminoacylation-active tRNA. Conversely, when tRNA inactive in aminoacylation is used, such as 2ЈdtRNA ox Ile , k ed equals k AMP (k AMP ϭ k ed ϭ 0.068 s Ϫ1 ; Table 5) in value. Comparison of the k ed values obtained with 3ЈdtRNA ox Ile and 2ЈdtRNA ox Ile reveals no significant difference (0.057 s Ϫ1 versus 0.068 s Ϫ1 , respectively). This clearly demonstrates that pre-transfer editing is not mod-ulated by the terminal 2Ј-OH group of the tRNA cofactor. It further appears that the 2Ј-OH or 3Ј-OH group at the terminal ribose of the tRNA has little or no role in tRNA-dependent pre-transfer editing.
In the presence of IleRS and tRNA with the intact 3Ј-end, the calculated k ed represents total pre-and post-transfer editing (0.504 s Ϫ1 ; Table 5) and, as expected, significantly overcomes in value the k ed that accounts for tRNA-dependent pre-transfer editing alone (0.057-0.068 s Ϫ1 ; Table 5). For the wild-type reaction, k ed is very similar to k AMP in value (0.57 s Ϫ1 ; Table 5), because only a small fraction of Val-tRNA Ile evades proofreading and contributes to AMP formation unrelated to editing. To more closely examine the full competence of the 3ЈdtRNA ox Ile analogue in stimulation of pre-transfer editing, the same set of experiments was performed with the post-transfer editing-deficient D342A IleRS. This variant exhibits a similar level of pretransfer editing activity (compared at the level of calculated k ed ) with tRNA Ile and 3ЈdtRNA ox Ile (0.102 and 0.150 s Ϫ1 ; Table 5), thus verifying that 3ЈdtRNA ox Ile stimulates pre-transfer editing analogously as tRNA with an intact terminal adenosine. Both wild-type and D342A IleRS in the presence of 3ЈdtRNA ox Ile exhibit reasonably comparable pre-transfer editing (0.057 and 0.150 s Ϫ1 , respectively; Table 5). The slight difference between the 2ЈdtRNA ox Ile and 3ЈdtRNA ox Ile analogues observed in these experiments may be a consequence of the different sugar pucker.
tRNA Ile Adjusts IleRS Affinity for Amino Acids without Modulating the Enzyme's Specificity-Prompted by the early finding that tRNA ox Ile may modulate isoleucine-IleRS interactions (29), we sought to examine whether tRNA Ile -dependent organization of the synthetic site for pre-transfer editing concomitantly modulates amino acid discrimination. Therefore, the tRNA analogues devoid of aminoacylation activity were tested in an ATP-PP i exchange assay to assess a possible role of tRNA in activation of isoleucine and valine. In the presence of 2ЈdtRNA ox Ile , IleRS exhibits a substantial decrease in affinity for isoleucine as judged by the 15-fold increase in the K m for isoleucine (0.0069 mM versus 0.104 mM; Table 6). Conversely, no significant effect on the k cat was observed (40 s Ϫ1 versus 33 s Ϫ1 ; Table 6). The data reporting valine activation (12-fold increase in the K m for valine; a lack of effect on k cat ; Table 6) closely parallel those for the cognate amino acid. As a consequence, the

tRNA Modulates Amino Acid Recognition by IleRS
discrimination factor (k cat /K m (Ile)/k cat /K m (Val)), describing the IleRS capacity to distinguish among isoleucine and valine in the activation step, remained unchanged in the presence of tRNA (188 versus 176). However, tRNA contributes to amino acid recognition because the ribonuclear protein, displaying decreased amino acid affinity relative to the IleRS alone, is less amenable to misactivation under cellular conditions than the free-standing protein (see "Discussion"). Interestingly, tRNA ox Ile , lacking both 2Ј-and 3Ј-A76 hydroxyl groups, did not affect isoleucine and valine activation differently from 2ЈdtRNA ox Ile (Table 6).
To test whether the presence of both 2Ј-and 3Ј-OH groups at the aminoacylation-active tRNA influences the amino acid recognition step, we followed the two-step aminoacylation under single-turnover conditions to extract the K d for the amino acid and maximal rate of aminoacyl-tRNA synthesis on the enzyme. Use of the IleRS variant with inactivated post-transfer editing is required in this assay to allow for accumulation of Val-tRNA Ile . Thus, to monitor formation of aminoacyl-tRNA on the enzyme, a large surplus of D342A IleRS was mixed with the limiting amount of 32 P-labeled tRNA. Inspection of the time course curves reveals no sign of deviation from the single exponential behavior at any concentration of the used amino acid (Fig. 3, A  and C). This strongly indicates that binding of the amino acid is a rapid equilibrium, thus allowing extraction of the K d constants. The K d values for isoleucine and valine (Fig. 3) highly parallel the corresponding K m values (Table 6) from the ATP-PP i exchange assay performed in the presence of 2ЈdtRNA ox Ile (0.11 mM versus 0.104 mM and 10 mM versus 12 mM, respec-tively). This shows that the 2Ј-OH group at the terminal A76 does not contribute to the organization of the active site prior to aminoacylation. The presented data thus provide evidence that tRNA does not operate locally via its terminal 2Ј-and 3Ј-OH groups in modifying IleRS affinity toward amino acids. A similar result is observed for tRNA-dependent pre-transfer editing. Taken together, this argues that a global conformational change, accompanying IleRS⅐tRNA Ile recognition, concomitantly amends synthetic and editing features of the synthetic site.
Aminoacylated tRNA Ile Modulates Synthetic and Editing Capacities of the Catalytic Site-Next, we tested the ability of aminoacylated tRNA to modulate amino acid activation (Table  6) and pre-transfer editing (Fig. 4). Use of the IleRS variant with inactivated post-transfer editing is required in this assay to ensure stability of misaminoacylated tRNA Ile . Aminoacylated tRNAs were purified from the remaining non-aminoacylated tRNA Ile by reverse phase chromatography. In the presence of Val-tRNA Ile , a 7-fold increase in the K m for valine and a 1.5-fold decrease in the k cat were observed relative to the conditions where tRNA was lacking (Table 6). This is highly similar to the effect imposed by non-aminoacylated tRNA. The participation of aminoacylated tRNAs in pre-transfer editing was also tested. Both aminoacylated and misaminoacylated tRNAs Ile retained the ability to stimulate pre-transfer editing of valine similar to other tRNA Ile analogues (Fig. 4). It thus appears that the participation of tRNA in organizing the synthetic site for amino acid activation and pre-transfer editing is not disturbed by the presence of the aminoacyl moiety at its 3Ј-end. This implies that  MAY 29, 2015 • VOLUME 290 • NUMBER 22 assembly of the ribonuclear protein synthetic site is not controlled by positioning of the 3Ј-end of the tRNA within the synthetic site.

Discussion
Decoupling of Aminoacylation and Pre-transfer Editing within the IleRS Synthetic Site-Isoleucyl-tRNA synthetase is a unique aaRS whose mechanisms of proofreading remained a puzzle for more than 3 decades. Its distinctive feature is utilization of tRNA-dependent pre-transfer editing, a pathway that up to now was clearly demonstrated only for this enzyme. Here we show that tRNA promotes IleRS pre-and post-transfer editing by different mechanisms relying on different parts of the tRNA molecule. Whereas the aminoacylation active 3Ј-end of the tRNA is essential for post-transfer editing, the remaining tRNA body, rather than the terminal adenosine alone, appears to modulate the pre-transfer editing activity. This important contribution resolves dilemmas regarding the requirement for an intact tRNA 3Ј-end in promoting pre-transfer editing (18,33,37).
Pioneering work by Baldwin and Berg (18) clearly demonstrated the capacity of IleRS to proofread valine in a tRNA-dependent manner, postulating that error correction occurs at the level of valyl-AMP. Early work by Fersht (17) then enforced a view whereby tRNA-dependent pre-transfer editing presents a dominant pathway, whereas the post-transfer step serves as a mopping-up activity. Interestingly, this was concluded solely on kinetic modeling of the wild-type reaction in the presence of valine. In that work, a lack of transient accumulation of Val-tRNA Ile to a concentration consistent with the independently measured rates of valyl transfer to tRNA and Val-tRNA Ile deacylation was interpreted to mean that valyl-AMP must be hydrolyzed before transfer to tRNA. It is, however, worth noting that Fersht (17) acknowledged a possibility that post-transfer editing measured in trans does not reflect directly the process occurring in cis, and this might complicate the interpretation of the modeling data. Early mutational analysis (39,40) of the IleRS hydrolytic editing site (22) at the distant CP1 domain (21,23) resulted in a "postpre-prepre" model (37) according to which the post-transfer editing step primes the enzyme for the successive rounds of valyl-AMP hydrolysis. Coupling of pre-transfer editing with the post-transfer step in that model presupposes that tRNA needs to participate in ami-noacylation in order to stimulate pre-transfer editing. An independent analysis of the tRNA analogues with substituted 3Ј-cis diol of the terminal adenosine (33,34) was further provided to support the model. Importantly, these studies failed to isolate the tRNA-dependent pre-transfer editing step, because inactivation of the post-transfer activity by the D342A substitution in the editing domain resulted in an IleRS variant that apparently lost the ability to edit at any level (37). Thus, the conclusions about the pre-transfer editing pathway were drawn indirectly and not by measuring the activity directly. The model was subjected to numerous criticisms (8,41,42), as discussed in detail by Dulic et al. (12).
A major step toward understanding the IleRS editing mechanism came from our recent in-depth analysis of the isolated tRNA-dependent pre-transfer editing pathway (12,28). In these studies, we applied a novel highly sensitive editing assay (8) to perform rigorous transient and steady-state kinetic analyses of the wild-type and deacylation-defective D342A IleRS enzymes. A model was proposed where the balance between the pre-and post-transfer editing pathways is dictated by the kinetic partitioning of aminoacyl-AMP between aminoacyl transfer to water and to tRNA, respectively. Finally, by monitoring in parallel AMP and aminoacyl-tRNA formation using deacylation-defective IleRS, the rate of AMP formation representing the pre-transfer editing reaction alone was calculated. This provided the first direct evidence that tRNA-dependent pre-transfer editing is the less prominent pathway (28).
Our new model implies no need for the priming aminoacylation step in the mechanism of pre-transfer editing. However, it is possible that the tRNA terminal 2Ј-OH group, while acting as the prime nucleophile, also participates in organization of the synthetic site for the hydrolytic step. With this in mind, we sought to examine whether indeed tRNA relies on the functional terminal adenosine to promote pre-transfer editing, thus intimately linking the aminoacyl transfer step with aminoacyl-AMP hydrolysis. To address this question, we produced the tRNA Ile analogues missing either the A76 2Ј-OH (2ЈdtRNA) or A76 3Ј-OH (3ЈdtRNA) group or both (tRNA ox ). Remarkably, extensive kinetic analysis of these analogues demonstrated that the 2Ј-OH or 3Ј-OH group at the terminal ribose of the tRNA cofactor has little or no role in pre-transfer editing. Apparently, tRNA does not act in pre-transfer editing by using free 3Ј-terminal diol to either position or activate substrates or to organize the hydrolytic subsite. Instead, the remaining parts of the tRNA body participate in attaining the synthetic site conformation with improved capacity to hydrolyze aminoacyl-AMP (see below). The established uncoupling of aminoacylation and pretransfer editing further supports a view whereby the pre-and post-transfer editing pathways are distinct IleRS activities linked by a common requirement that tRNA Ile be present. The considerable confidence in our conclusions arises from the fact that they are based on the comparison of rigorously determined kinetic rate constants (12,28). In sharp contrast, earlier conclusions (33,34,37) were predominately based on time course analysis rather than determined kinetic rate constants.
Optimized Amino Acid Recognition within the IleRS⅐tRNA Ribonuclear Protein-The novel perspective enforced by this study highlights the importance of tRNA Ile as a versatile IleRS tRNA Modulates Amino Acid Recognition by IleRS molecular partner that contributes to the enzyme's amino acid recognition more than anticipated thus far. Besides its role in editing, the work presented here, and in part elsewhere (29), highlights tRNA as a mediator of the amino acid activation step. Here we demonstrate that unexpectedly, tRNA modulates affinity for both cognate isoleucine and non-cognate valine in a highly similar manner, promoting the 10-fold increase in the corresponding K m values. The benefits of decreased affinity toward valine by the IleRS⅐tRNA complex are evident. Valine is among the most abundant amino acids in E. coli, reaching concentrations up to 4 mM (43). Because of IleRS low K m for valine (12,17) (Table 6), the enzyme would operate near saturation with non-cognate valine, keeping the proofreading pathways highly active. However, when IleRS acts as a ribonuclear protein, misactivation at the cellular concentration of valine becomes less frequent, making energetically expensive proofreading less vigorous. This is consistent with our recent suggestion that IleRS may have adopted various mechanisms to keep proofreading under control; the slow transfer step, observed in IleRS but not in many other aaRSs, may provide molecular adaptation that maintains minimal editing without significantly compromising misaminoacylation frequency (28). Interestingly, inspection of the K m values (12, 44 -67) for cognate amino acids for many E. coli aaRSs and putting these values in correlation with the cellular concentrations of amino acids (43,68) show that the majority of aaRSs have K m values, in activation, up to an order of magnitude below the concentration of the corresponding free amino acid. Among some exceptions that have higher c(aa)/K m ratios is IleRS (Fig. 5). Because operating at almost saturating conditions is not kinetically optimal for enzymes (69), it is plausible to assume that the interaction with tRNA Ile , which decreases IleRS affinity for isoleucine, is relevant for efficiency and regulation of isoleucine activation and hence Ile-tRNA Ile synthesis by IleRS. Analogous situations were observed with E. coli TyrRS (70) and GlyRS (44), where tRNA also decreases the affinity toward cognate tyrosine and glycine, respectively. It remains to be seen whether this represents a more general molecular mechanism for adjusting catalysis to cellular conditions. It is intriguing to speculate why interaction with tRNA is not used to improve the enzyme's specificity against non-cognate amino acids. Based on the aforementioned arguments, one may assume that modulation of IleRS affinity by tRNA is pertinent for both cognate and non-cognate amino acids, making evolution of the ribonuclear protein toward higher specificity less likely. This promoted an alternative mechanism, where tRNA participation is expressed through readjustment of the synthetic site to operate in favor of cognate amino acid within the inherent IleRS network of synthetic/pre-transfer editing pathways under cellular conditions.
Assembly of the Synthetic Site within the IleRS⅐tRNA Ile Complex-The finding that aminoacylated and non-aminoacylated tRNAs promote similar effects on the catalytic capacity of the synthetic site suggests that the 3Ј-end of tRNA may sample between the catalytic and editing domains, or it may predominantly reside at the CP1 domain, whereas the synthetic site displays tRNA-dependent features. This conclusion is further supported by the crystal structure of the Staphylococcus aureus IleRS bound to tRNA Ile and synthetic site inhibitor mupirocin (23), which revealed that the 3Ј-end of non-aminoacylated tRNA resides in the editing site. Structural analyses on homologous ValRS and LeuRS enzymes in complexes with cognate tRNAs indicated that the favorable tRNA-binding mode positions the 3Ј-end of tRNA in the CP1 domain (27,71,72). In line with that, our data strongly suggest that the 3Ј-end of the tRNA Ile is not (substantially) engaged in the assembly of the optimized ribonuclear protein synthetic site.
The important implication of that finding is that interactions with the more distant parts of the tRNA Ile body promote global conformational readjustment concomitant with assembly of the synthetic site within the IleRS⅐tRNA complex. A previous study pinpointed three nucleotides in the D-loop of the tRNA Ile as determinants for editing (73). Interestingly, these nucleotides do not interact directly with IleRS (23) and were not required for the tRNA Ile aminoacylation activity and deacylation of Val-tRNA Ile (74). Consequently, their role in editing remained elusive, and their role in the translocation step was proposed. In light of these new findings, it is tempting to speculate whether these nucleotides stabilize the structure of tRNA requiredfororganizingthesyntheticsitetobecapableoftRNA-dependent pre-transfer editing.
The participation of tRNA-dependent pre-transfer editing in overall IleRS editing has now been tested by two complementary approaches, mutational silencing of the IleRS post-transfer editing pathway and ablation of the tRNA Ile acceptor activity. tRNA Ile analogues deprived of aminoacylation capacity stimulate editing up to 25-30% of total editing at the level of k cat /K m (0.238 s Ϫ1 versus 0.058 and 0.075 s Ϫ1 ; Table 2). This is in good correlation with the data obtained by deacylation-defective IleRS (28), thus strengthening the notion that tRNA-dependent pre-transfer editing represents an important proofreading pathway in IleRS. However, importantly, in the absence of posttransfer editing, IleRS exhibits significant misaminoacylation (12). The requirement for the tRNA-dependent pre-transfer FIGURE 5. The ratios of cellular amino acid concentrations and corresponding K m values for amino acids in activation by cognate aaRSs. The K m values were extracted from the literature (12,. The ratios with K m measured in the absence of tRNA (depicted as white rectangles) were compared, when possible, with the ratios with K m measured by an ATP-PP i exchange assay in the presence of tRNA (depicted as black rectangles). Glutaminyl-and arginyl-tRNA synthetase require tRNA for activation; therefore, only the ratios with K m measured in aminoacylation are presented. The ratios that are higher than 10 (Val, Asn, Ile, and Lys) are highlighted with striped rectangles. Concentrations of amino acids are as published in Bennett et al. (43) and, for Gly and Cys, obtained from ECMDB: The E. coli Metabolome Database (68). editing pathway in vivo is completely unknown. Understanding of the physiological background may provide a clue as to whether this activity represents some relic from primordial times or a unique solution for IleRS proofreading (75). The experiments to address this important biological question are under way in our laboratory.