Characterization of Isoleucyl-tRNA Synthetase from Staphylococcus aureus

The kinetic mechanism for the amino acid activation reaction of Staphylococcus aureus isoleucyl-tRNA synthetase (IleRS; E) has been determined from stopped-flow measurements of the tryptophan fluorescence associated with the formation of the enzyme-bound aminoacyl adenylate (E·Ile-AMP; Scheme FS1). Isoleucine (Ile) binds to the E·ATP complex (K 4 = 1.7 ± 0.9 μm) ∼35-fold more tightly than to E(K 1 = 50–100 μm), primarily due to a reduction in the Ile dissociation rate constant (k -1 ≈ 100–150 s−1,cf. k -4 = 3 ± 1.5 s−1). Similarly, ATP binds more tightly toE·Ile (K 3 = ∼70 μm) than to E (K 2 = ∼2.5 mm). The formation of the E·isoleucyl adenylate intermediate, E·Ile-AMP, resulted in a further increase in fluorescence allowing the catalytic step to be monitored (k +5 = ∼60 s−1) and the reverse rate constant (k -5 = ∼150–200 s−1) to be determined from pyrophosphorolysis of a pre-formed E·Ile-AMP complex (K 6= ∼0.25 mm). Scheme FS1 was able to globally predict all of the observed transient kinetic and steady-state PPi/ATP exchange properties of IleRS by simulation. A modification of Scheme FS1could also provide an adequate description of the kinetics of tRNA aminoacylation (k cat,tr = ∼0.35 s−1) thus providing a framework for understanding the kinetic mechanism of aminoacylation in the presence of tRNA and of inhibitor binding to IleRS.

Activation of the carboxyl group of amino acids with ATP to form the corresponding aminoacyl adenylate prior to transfer to tRNA is the first step in protein biosynthesis. The activation of each amino acid is catalyzed by a specific aminoacyl-tRNA synthetase (aaRS) 1 which first catalyzes the activation (and stabilization) of the amino acid as a mixed anhydride adenylate, and the subsequent acyl transfer to the corresponding cognate tRNA (for a review, see Ref. 1). The high level of phylogenic divergence between these enzymes in prokaryotes and eukaryotes (2,(3)(4)(5)(6), together with their essential role in protein synthesis, makes aminoacyl-tRNA synthetases excel-lent targets for the development of selectively acting antibacterial agents.
The best known example of such a compound is pseudomonic acid A (PS-A) which specifically inhibits bacterial isoleucyl-tRNA synthetase (IleRS) around 10,000-fold more potently that the corresponding mammalian enzyme (7,8) and is a highly effective antibiotic. Staphylococcus aureus infections, particularly of the respiratory tract, are a major clinical problem and, because of the emergence of resistance to "classical" antibiotics (e.g. ␤-lactams), this organism is an important target for the development of new antibiotics. However, resistance to PS-A itself is emerging although not yet a clinically relevant problem (e.g. Refs. 9 and 10). Therefore, we have considerable interest in understanding the reaction cycle and inhibition of S. aureus IleRS by PS-A in order to aid the design of novel inhibitors and to understand the mechanisms of PS-A resistance. Following overexpression of S. aureus IleRS (11), reagent quantities of this enzyme are now available. In this paper, we describe the construction of a minimal reaction mechanism for amino acid activation by IleRS which adequately describes all of the steady-state and transient kinetic properties of this enzyme. Accompanying papers describe a detailed characterization of the kinetics and mechanism of inhibitor binding (12), and an analysis of the effects of ligand binding upon proteolysis of IleRS (13).

MATERIALS AND METHODS
The preparation of IleRS (12) and the methods used for steady-state ATP/PP i exchange and tRNA aminoacylation reactions (15) were as described in the accompanying paper (12). Isoleucinyl adenylate (Ile-ol-AMP; Fig. 3) was synthesized at SmithKline Beecham. Stoichiometric E⅐Ile-AMP complex was prepared by incubating IleRS with excess [Mg⅐ATP] and [Ile] followed by gel filtration (Pharmacia Fast Desalt) to remove excess substrates. Adenylated IleRS was stable for several hours when stored on ice (data not shown). Stopped-flow studies, and the resulting data analysis (16 -19), were performed with an Applied Photophysics SM17MV instrument at 22°C in 50 mM Tris-HCl, pH 7.9, 10 mM MgCl 2 (buffer B) as described (12).

Steady-state Measurements of IleRS Tryptophan Fluorescence-Although
IleRS (E) from S. aureus possesses 18 tryptophan residues distributed fairly uniformly throughout the sequence (11), examination of the equilibrium enzyme fluorescence of various complexes showed that, with the exception of E⅐ATP, these differed markedly in their fluorescence yields (Fig. 1). This provided the opportunity to determine the rates and equilibria for their inter-conversion directly by stopped-flow techniques.
The intrinsic protein fluorescence intensity is increased by around 7% in the E⅐Ile binary complex and by around 17% upon the catalytic formation of E⅐Ile-AMP (Fig. 1). However, enzyme fluorescence was not altered by the presence of Ͼ5 mM Mg⅐ATP, Mg⅐PP i , or Mg⅐Ap 4 A (adenosine 5Ј-tetraphospho-5Ј-adenosine; data not shown). Interestingly, binding of Ile-ol-AMP (Fig. 10a) yielded an identical enzyme fluorescence to E⅐Ile-AMP formed during the reaction cycle, suggesting that the conformation of the enzyme-ligand complex is similar in both cases. The fluorescence change induced by Ile-ol-AMP was used to monitor those interactions that are spectroscopically silent (e.g. Mg⅐ATP binding, see Fig. 4) using a kinetic competition approach (see below). Transient Kinetics of Substrate and Inhibitor Binding to S. aureus IleRS-We have characterized the elementary rate and/or equilibrium constants involved in substrate binding, activation, and Ile-ol-AMP binding using stopped-flow according to the minimal mechanism shown in Scheme 1. The rate and equilibrium constants derived (Table I, Scheme 1) provide an adequate description of all of the experimental steady-state and transient kinetic data as described below.
Binding of Ile-ol-AMP (k ϩi and k Ϫi )-See below 2 for explanation of the notation used throughout this and accompanying (12,13) papers. The binding of excess Ile-ol-AMP to IleRS was monitored directly by rapid mixing of the enzyme and inhibitor in a stopped-flow apparatus ( Fig. 2A). The observed rate constant, k obs , of the exponential increase in protein fluorescence (amplitude Ϸ17%) varied linearly with Ile-ol-AMP concentration (Fig. 2B), consistent with a simple bimolecular reaction (Scheme 1; k obs ϭ k ϩi ⅐[Ile-ol-AMP] ϩ k Ϫi ). Although k ϩi ϭ 2.4 Ϯ 0.2 ϫ 10 6 M Ϫ1 s Ϫ1 (n ϭ 3), the intercept defining k Ϫi was too small to be determined accurately (k Ϫi Ͻ 0.5 s Ϫ1 ) from Fig. 2B. However, direct measurement of k Ϫi for Ile-ol-AMP via displacement experiments yielded a value of ϳ0.07 s Ϫ1 (12), de-fining an overall K i ϳ 30 nM.
Binding of Isoleucine (k ϩ1 and k Ϫ1 )-Similar experiments to those described above but with the substrate L-isoleucine (Ile), also resulted in a single exponential increase in protein fluorescence (Fig. 3A). The experiment shown in Fig. 3 provided an estimate of k ϩ1 ϭ 3.1 Ϯ 0.3 ϫ 10 6 M Ϫ1 ⅐s Ϫ1 , k -1 ϭ 142 Ϯ 14 s Ϫ1 ( Fig. 3B; K 1 ϭ 46 Ϯ 6 M). Six independent repeat experiments yielded mean (ϮS.D.) values for each parameter of k ϩ1 ϭ 2.2 Ϯ 0.5 ϫ 10 6 M Ϫ1 ⅐s Ϫ1 , k -1 ϭ 131 Ϯ 50 s Ϫ1 , and hence K 1 ϭ 60 Ϯ 27 M. This large error in K 1 limits the reliability of other equilibrium constants in the mechanism, particularly K 4 , which have been estimated, in part, from a thermodynamic linkage argument involving the estimate of K 1 (see below). Nevertheless, these rate and equilibrium constants are consistent with the [Ile] dependence of the fluorescence amplitudes (Fig. 3C, K 1 ϭ 56 Ϯ 10 M) and provide unambiguous evidence that IleRS can bind Ile (K 1 Ϸ 50 -100 M) in the absence of Mg⅐ATP. The value for k ϩ1 is lower than expected for a diffusion limited binding process (typically Ͼ10 8 M Ϫ1 s Ϫ1 (37)), although several mechanisms can give rise to such behavior (14,20,36) and for the purposes of Scheme 1, Ile binding can be considered an elementary step.
Binding of Mg⅐ATP (k ϩ2 and k Ϫ2 )-The steady-state protein fluorescence intensity of IleRS was not detectably changed upon addition of 5 mM Mg⅐ATP (Fig. 1), a concentration approx-2 Equilibrium (dissociation) or steady-state constants for step n are termed K n and K n Ј for reactions performed in the absence and presence of tRNA, respectively. The rate and equilibrium constants are defined explicitly in Scheme 1 ( Table I). The internal equilibrium constant, K 5 , and the external equilibrium constant, K 6, are written in the forward direction. K i (ϭ k Ϫi /k ϩi ) is the equilibrium dissociation constant for inhibitor binding to E. Where elementary rate and equilibrium constants cannot be directly extracted from derivative plots of the observed rate constants, k obs,1 and k obs,2 against concentration, the limiting rate constants and apparent equilibrium constants have the suffix app (e.g k ϩ5,app or K 3,app ). The k cat values for the aminoacylation and ATP/PP i exchange reactions are k cat,tr and k cat,ex , respectively. Step Rate/equilibrium constants imately equal to 2ϫ K 2 Ј and ϳ10-fold higher than the K m,ATP Ј measured in the steady state tRNA aminoacylation reaction (see below, Fig. 9), suggesting that ATP binding was spectroscopically silent. We therefore employed a transient kinetic competition approach (21) to characterize the binding of Mg⅐ATP to E. In these experiments, ATP is pre-mixed in the syringe of the stopped-flow apparatus with a ligand (e.g. Ile-ol-AMP) which is both competitive with ATP ( Fig. 10) (12) and which induces a fluorescence change upon binding to E (Figs. 1 and 2). This mixture is then rapidly mixed with E such that the two ligands bind to E simultaneously and in parallel according to Scheme 2 and the transient kinetics of E⅐Ile-ol-AMP formation are monitored spectroscopically. The kinetics and/or thermodynamics of ATP binding are inferred from the effect of ATP on the kinetics of Ile-ol-AMP binding (21). A series of kinetic competition experiments were performed by mixing 0.1 M E with a sample of 0.5 M Ile-ol-AMP plus varying concentrations of Mg⅐ATP. Mixing E into Ile-ol-AMP alone yielded, as expected (Fig. 2), a single exponential increase in protein fluorescence (k obs Ϸ 2.5 s Ϫ1 , amplitude Ϸ16%, Fig. 4A). When Mg⅐ATP and Ile-ol-AMP were mixed with E, k obs for the formation of E⅐Ileol-AMP decreased with increasing [ATP] (Fig. 4), as expected for the competition between the two ligands where the binding of the spectroscopically silent ligand (ATP) is rapidly reversible relative to the rate of Ile-ol-AMP binding (21). Analysis of the data according to Equation 1 yields an estimate for K 2 of ϳ2.5 Ϯ 0.2 mM (mean Ϯ S.D., n ϭ 3) and an estimate of k Ϫi (ϳ0.1 s Ϫ1 ) consistent with that determined directly (Fig. 2), as shown, As predicted, the estimate of K 2 was independent of the concentration of Ile-ol-AMP used (defining k -2 Ͼ 10 s Ϫ1 ; see also Fig. 9), and was similar to that obtained from the activation of the aminoacylation reaction by ATP (see below, 2.7 mM; Table  III). These data therefore provide clear evidence that ATP can bind to E in the absence of the co-substrate, Ile (and vice versa, see Fig. 3), such that a minimal mechanism must include random order of addition of both substrates. Simulations of the experiment shown in Fig. 4A by numerical integration techniques (using Scheme 1, Table I) provide an adequate description of the experimental data (compare Fig. 4, A and C). Binding of Ile and Mg⅐ATP and Subsequent Catalytic Events-Rapid mixing of IleRS with mixtures of Ile and Mg⅐ATP to initiate the amino acid activation reaction resulted in a 16 -17% increase in enzyme fluorescence (cf. Fig. 1). Although the time course of the change in species concentration predicted by Scheme 1 is complex (see below), due to the particular combination of fluorescence yields associated with each of the intermediates, the observed fluorescence transients could, under most conditions, be described by the sum of two exponentials (e.g Fig. 5A). The fast phase of the transient (observed rate constant k obs,1 ) had an [Ile] dependence consistent with that observed previously (Fig. 3) and an equivalent transient was observed in the absence of ATP. Furthermore, this rate constant was invariant when IleRS was mixed into varying ATP concentrations in the presence of a fixed concen- tration of Ile (Fig. 5B, displayed on a logarithmic scale for clarity). The first phase therefore corresponds to the binding of Ile to the fraction of E that is nucleotide-free (at the concentration of ATP used relative to K 2 ) to form E⅐Ile. As described below, the binding of ATP to form E⅐ATP leads to a reduction in the dissociation rate constant for Ile binding but no significant change in the association rate constant. If k obs,1 arose from the binding of Ile to E⅐ATP, k obs,1 would be predicted to be lower than the equivalent rate constant observed in the absence of added ATP, in contrast to that observed experimentally. The slower transient observed in these mixing experiments, k obs,2 (⌬F Ϸ 7%), showed a hyperbolic dependence in either [ATP] (Fig. 5C)  ) to obtain estimates of the maximal observed rate constant, k max ϭ k ϩ5,app ϩ k -5,app , the apparent K d of the binding step, K 3,app and the apparent reverse rate constant for the first order step, k -5,app . We emphasize the apparent nature of these terms as opposed to the elementary nature of the rate and equilibrium constants reported in Table  I 2 and derived as described, The value of k -5,app was too low to be determined reliably in these experiments, but had a value of Ͻ2 s Ϫ1 such that k max (ϭ k ϩ5,app ϩ k -5,app ) approximates to k ϩ5,app . According to Scheme 1, the apparent irreversibility of the chemical cleavage step  Table I. (k -5,app Ͻ 2 s Ϫ1 ) is a consequence of the weak binding of PP i , the rapid and thermodynamically favorable release of which from the E⅐Ile-AMP⅐PP i complex makes the observed reverse rate negligible.
To determine if the maximal apparent rate constant observed at high [ATP] in Fig. 5C (57 s Ϫ1 ) was the elementary rate constant for E⅐Ile-AMP⅐PP i formation, we conducted a series of experiments in which the concentration dependence of k obs,2 in either Ile or ATP was measured in the presence of fixed concentrations of the other substrate (Table II). The maximal observed rate constant, k max , was dependent upon [Ile], reaching a limiting value near 55 s Ϫ1 at high concentrations of both ATP and Ile (Table II). This rate constant likely reflects the true elementary rate constant for E⅐Ile-AMP⅐PP i formation which was confirmed in a number of other experiments conducted at saturating concentrations of Ile and ATP (not shown). Kinetic simulations of these experiments (Scheme 1, Table I) yielded fluorescence transients similar to those observed experimentally (compare Fig. 5, A and D) and furthermore, predicted accurately the apparent maximal rate constants for k obs,2 at different [Ile] (Table II).
The apparent equilibrium constants for ATP and Ile (K 3,app ϭ 75 M and K 4,app ϭ 15 M, respectively; Table II) only approximate to the true equilibrium constants, K 3 and K 4 , when the pseudo first-order association and first-order dissociation rate constants are much larger than the rate constant for the subsequent (signal generating) chemical step. For ATP binding to E⅐Ile, kinetic simulations were able to predict the observed fluorescence transients using a value of K 3 (70 M) approximately equal to the K 3,app value from Table II (75 M). However, similar kinetic simulations indicate that the K 4,app value for Ile binding to E⅐ATP (15 M) is an overestimate of the true K 4 (1.7 Ϯ 0.9 M, Table I). Indeed, experiments described below suggest that the dissociation of Ile from E⅐Ile⅐ATP (k -4 ϳ 3 Ϯ 1.5 s Ϫ1 ) occurs more slowly than the rate of the chemical cleavage step (ϳ60 s Ϫ1 ). As such, the non-equivalence of the best estimate of K 4 and the K 4,app for Ile activation of the chemistry step (Table II) is to be expected. Finally, based on reasonably reliable estimates for three of the four equilibrium constants in Scheme 1 defining the formation of the E⅐ATP⅐Ile complex, thermodynamic linkage with associated error propagation (primarily in K 1 , see above), suggests K 4 ϭ 1.7 Ϯ 0.9 M. Despite this slight uncertainty, the major conclusion is that the binding of either substrate to E reduces the K d for the formation of the ternary E⅐ATP⅐Ile complex by Ͼ10-fold compared with the K d obtained for either substrate binding to free E (Figs. 3 and 4).
Binding of Isoleucine to the E⅐ATP Binary Complex (k ϩ4 and k Ϫ4 )-Although Table II predicts that Ile binds more tightly to E⅐ATP than to the free enzyme, there is no information concerning the rate constants k ϩ4 and k -4 . Initial experiments to monitor the association kinetics of Ile with E⅐ATP (prior to the chemical step) proved difficult to interpret since, at low [Ile], k obs,1 (defining the formation of the ternary E⅐ATP⅐Ile complex) had a similar rate constant and amplitude to k obs,2 (defining the catalytic formation of E⅐Ile-AMP⅐PP i ). In an attempt to kinetically resolve the bimolecular Ile binding step from the first order catalytic step, we conducted experiments at high [Ile] (up to 300 M) although this necessarily led to rapid transients and associated uncertainty (Fig. 6A).
The apparent observed rate constant increased approximately linearly with [Ile] (Fig. 6B), yielding an intercept of 22 Ϯ 13 s Ϫ1 and a slope of 1.6 ϫ 10 6 M Ϫ1 s Ϫ1 . These values only represent k -4 and k ϩ4 when the bimolecular formation of E⅐ATP⅐Ile occurs much more rapidly than the decay of the ternary complex (at ϳ60 s Ϫ1 ). However, the predicted species distribution (based on Scheme 1, Table I) suggests that both processes are likely to be kinetically linked below 150 M Ile (Fig. 6D). Only the data between 150 and 250 M Ile (which yielded k obs ϳ 450 s Ϫ1 ; close to the upper limit for reliable measurements on our apparatus) could be used to estimate the elementary rate constants (k ϩ4 ϳ 1.6 ϫ 10 6 M Ϫ1 s Ϫ1 ) and, therefore, we can estimate only k -4 Ͻ10 s Ϫ1 . This value is, as we noted above, significantly less than the estimate of the rate of the chemistry step (about 60 s Ϫ1 ) such that rapid equilibrium assumptions associated with Ile binding to E⅐ATP are invalid. Despite the difficulty in obtaining a reliable estimate for k -4 , and hence K 4 , from either Table I or Fig. 6 or by thermodynamic linkage with other more well defined equilibrium constants we obtain an estimate of K 4 ϭ 1.7 Ϯ 0.9 M and hence k -4 ϳ 3 Ϯ 1.5 s Ϫ1 from an estimate of k ϩ4 . Kinetic simulation of the experiments (Fig. 6A, Scheme 1, Table I, k -4 ϳ 3 Ϯ 1.5 s Ϫ1 ) provided an adequate description of the experimental transients (Fig. 6C) and of the apparent rate constant for the chemistry step at different [Ile] (Table II). We believe the combined data justifies the estimate of k -4 and hence K 4 with appropriate caveats regarding the uncertainty associated with these value.
Effect of Order of Mixing of Substrates-Binding of ATP to free E was sufficiently weak (K 2 ϭ 2.5 mM) to be expected to be a rapid equilibrium on the stopped-flow time scale, an assumption we have incorporated into simulations of Scheme 1 (k ϩ2 ϭ 10 7 M Ϫ1 s Ϫ1 , k -2 ϭ 2.5 ϫ 10 4 s Ϫ1 , e.g. Fig. 6D). As such, we would expect no difference in the kinetics of E⅐Ile-AMP formation when the E⅐ATP complex was mixed with Ile and when both substrates were mixed with E simultaneously. To test this hypothesis (Fig. 7), a fixed concentration of the pre-bound substrate was maintained in both stopped-flow syringes, so that any pre-established equilibria were not perturbed upon subsequent mixing with the second substrate. As expected, the stopped-flow time course obtained was identical whether or not ATP was pre-mixed with E (Fig. 7A, note that the traces have been offset for clarity). In contrast, the fluorescence amplitude observed following pre-binding of Ile to E reflects the difference in fluorescence intensity between E⅐Ile and E⅐Ile-AMP (Fig. 1). Kinetic simulation of the experiments in Fig. 7A (Scheme 1, Table I) provided a reasonable description of the experimental data ( Fig. 7B) with exponential rate constants (46 s Ϫ1 ) comparable to those observed experimentally (40 -44 s Ϫ1 ).
Pyrophosphorolysis of E⅐Ile-AMP-Since the E⅐Ile-AMP complex is relatively stable (rate constant for decay of E⅐Ile-AMP Ͻ 3 ϫ 10 Ϫ4 s Ϫ1 ; data not shown), it could be isolated using rapid gel-filtration allowing measurement of the reverse rate constant for the catalytic step (k -5 in Scheme 1) via pyrophosphorolysis of E⅐Ile-AMP. Rapid mixing of freshly isolated E⅐Ile-AMP with Mg⅐PP i resulted in a decrease in enzyme fluorescence (Fig. 8A) corresponding to the reaction E⅐Ile-AMP⅐PP i 3 E⅐Ile⅐ATP (⌬F Ϸ-8%). The observed rate constant for this process showed a hyperbolic dependence on [PP i ] reaching a limiting observed rate constant of 100 Ϯ 10 s Ϫ1 and a yielding a value for K 6 (ϭk ϩ6 /k -6 ) of ϳ250 M. When the binding of PP i is followed by the one-step irreversible formation of E⅐Ile⅐ATP, the limiting rate constant of 100 s Ϫ1 would correspond to k -5 . However, simulations of the experiments in Fig.  8A with k -5 ϭ 100 s Ϫ1 yield simulated transients significantly slower than observed experimentally. We have used a value of k -5 Ϸ 150 -200 s Ϫ1 to obtain transients consistent to those observed experimentally (data not shown). Although the mechanistic basis for this albeit minor difference (1.5-2-fold) is not readily evident, the complex, bisubstrate reversible nature of the pyrophosphorolysis reaction will necessarily lead to a net flux back toward E⅐Ile-AMP⅐PP i to a certain extent and hence reduce the macroscopic apparent maximal rate of pyrophosphorolysis (k max,app ϭ 100 s Ϫ1 ) relative to the true rate constant (k -5 ϭ 150 -200 s Ϫ1 ). Regardless of the absolute estimate of k -5 (100 -200 s Ϫ1 ), the major conclusions are that PP i binding is weak (K 6 ϭ 250 M), the chemistry step is readily reversible and it has an internal equilibrium constant (ϭk ϩ5 /k -5 ) significantly less than unity (0.3-0.4).
Steady-state Kinetics of tRNA Aminoacylation and PP i /ATP Exchange-The results above allowed the construction of a kinetic mechanism for substrate activation in the absence of tRNA (Scheme 1) which could adequately describe all of the transient kinetic data. It was of interest, however, to determine whether the same mechanism could also describe the steadystate kinetics of PP i /ATP exchange and, to a lesser extent, of the full tRNA aminoacylation reaction. The steady-state parameters describing both the exchange and aminoacylation reactions were determined at saturating concentrations of one substrate while varying the concentration of the other substrate (Table III). Unfractionated Escherichia coli MRE600 tRNA (ϳ4 M functionally active tRNA Ile ) was used as the acceptor species in aminoacylation measurements. Analysis of progress curves at a range of tRNA concentrations using an integrated form of the Michaelis-Menten equation confirmed that the K m,tRNA was Յ0.1 M (data not shown) consistent with similar values reported for E. coli IleRS (Ͻ0.1 M) (22). In addition, the steady-state kinetic parameters (Table III) are broadly comparable to those reported previously for E. coli IleRS (22). Therefore, S. aureus IleRS is able to utilize tRNA-Ile from E. coli effectively as a substrate.
The k cat for PP i /ATP exchange performed in the absence of tRNA (k cat,ex ϭ 18 s Ϫ1 at 22°C, Table III) was Ϸ50-fold higher than for tRNA aminoacylation (k cat,ex ϭ 0.35 s Ϫ1 , Table III). Scheme 1 was able to provide an adequate description of the steady-state kinetics of ATP/PP i exchange (e.g simulated k cat,ex ϭ 22 s Ϫ1 , data not shown, observed k cat,ex ϭ 18 Ϯ 2 s Ϫ1 , Table  III). Since both the internal equilibrium constant for the chem-  Table I. istry step (0.3-0.4) and the rate constant k ϩ5 define the k cat,ex for the enzyme (where chemistry is rate-limiting), the data in Table I suggest a k cat,ex of 18 -24 s Ϫ1 , consistent with that observed experimentally assuming 100% active enzyme (16 -20 s Ϫ1 ). Furthermore, since the rate constant for Ile dissociation from E⅐Ile⅐ATP (k -4 ϳ 3 Ϯ 1.5 s Ϫ1 ) is less than the rate constant for the catalytic step (k ϩ5 Ϸ 60 s Ϫ1 ), the k cat,ex /K m ,Ile determined at saturating concentrations of ATP (1.8 ϫ 10 6 M Ϫ1 s Ϫ1 , Table III) should approximate to the true association rate constant, k ϩ4 measured by stopped-flow (1.7 ϫ 10 6 M Ϫ1 s Ϫ1 , Fig.  6B).
The K m and V max values shown in Table III, however, provide no information concerning the kinetically preferred order of substrate addition. We therefore investigated the kinetics of aminoacylation at varying concentrations of ATP and Ile and at a fixed, saturating, concentration of tRNA (Fig. 9). The depen-dences of the initial rate on [Mg⅐ATP] and [Ile] (Fig. 9, A and D) were globally fit to Equation 3 which defines the steady-state equation for an equilibrium-ordered mechanism in which the kinetically preferred path involves the initial rapid equilibrium binding of ATP with a dissociation constant K 2 Ј (ϭ2.7 Ϯ 0.4 mM) followed by the binding of Ile with a Michaelis constant K m,Ile Ј ϭ 3 Ϯ 0.3 M to form the quaternary E⅐tRNA⅐ATP⅐Ile complex (23,24), Other models for bisubstrate addition (including random order of addition such as in Scheme 1) did not yield an adequate fit to the data. Although there appears to be a preferred order of substrate binding in the presence of tRNA, the K 2 Ј measured for ATP binding to E⅐tRNA in steady-state measurements (Scheme 4) is similar to K 2 determined from transient kinetics in the absence of tRNA (2.5 versus 2.7 mM; Figs. 3 and 9). Initial attempts to simulate the tRNA aminoacylation reaction used a modification of Scheme 1 in which a first order step (at 0.35 s Ϫ1 , Table III) corresponding to the rate-limiting chemical transfer reaction was added following PP i release (i.e. E⅐Ile-AMP ϩ tRNA 3 E ϩ Ile-tRNA). Due to the highly favorable release of PP i from the E⅐Ile-AMP⅐PP i intermediate which precedes the rate-limiting step for the reaction, the predicted K m Ј values for both Ile and ATP were Ͼ100-fold lower than observed experimentally (Table III). However, in a mechanism in which the rate-limiting transfer step occurs prior to PP i release (E⅐Ile-AMP⅐PP i ⅐tRNA 3 E ϩ Ile-tRNA ϩ PP i ), the remaining rate constants in Scheme 1 yielded very similar steady-state kinet- . E⅐Ile ϩ ATP, as for E⅐ATP ϩ Ile except that E was preincubated with 30 M Ile (k obs ϭ 34.7 Ϯ 1.6 s Ϫ1 , amp ϭ 7.5%). All fluorescence transients were best described by a single exponential fit (as shown). B, kinetic simulation of the predicted fluorescence transients (arbitrary units) for the experiments shown in A using Scheme 1 and the data in Table I.
ics to those observed experimentally (compare Fig. 9, A versus C, for ATP and D versus F for Ile). Although a solution to the elementary tRNA aminoacylation mechanism was not the main aim of this study and notwithstanding the somewhat arbitrary modification of Scheme 1 to include a tRNA transfer prior to PP i release, the similarity of the experimental data ( Fig. 9) and that predicted from the modified Scheme 1 is rather striking.

Catalysis of Ap 4 A Formation by
IleRS-Scheme 1 provides an adequate description of all the data obtained in the absence of tRNA. However, all aaRSs studied thus far that can catalyze amino acid activation in the absence of tRNA are also able to catalyze the reaction of ATP with the enzyme bound adenylate to form Ap 4 A (e.g. Ref. 34). We therefore investigated whether this reaction was catalyzed by S. aureus IleRS and if so, whether the reaction would impact significantly on the analysis of the stopped-flow data according to Scheme 1. Rapid mixing of E plus saturating [Ile] with Ap 4 A led to a slow exponential increase in enzyme fluorescence, consistent with the formation of the E⅐Ile-AMP⅐ATP complex although the maximal observed rate constant, k ϩ8 , was only 0.04 s Ϫ1 at high concentrations of Ap 4 A (data not shown; K 7 ϭ 200 M; Scheme 3). For those synthetases that have been studied in detail, the affinity of ATP for the E⅐Ile-AMP⅐ATP complex appears to be very low (e.g. K m ϭ 11 and 50 mM for E. coli LysRS and PheRS, respectively) (35). As such, k -8 cannot be determined directly from the experiments we have performed. However, by analogy with other synthetases (e.g. E. coli PheRS) in which synthesis of Ap 4 A is slow (0.25 s Ϫ1 ) and where the internal equilibrium favors ATP rather than Ap 4 A (K 8 Ͼ 1) (35), it is probable that for IleRS, k -8 Ͻ k ϩ8 ϭ 0.04 s Ϫ1 . As such, the catalysis of Ap 4 A formation by S. aureus IleRS is unlikely to contribute significantly to the experimental data and has therefore been disregarded in the construction of the minimal mechanism.
Steady-state Kinetics of Inhibition of by Isoleucinyl Adenylate (Ile-ol-AMP)-Isoleucinyl adenylate (Fig. 10A) is the reduced, and therefore a non-hydrolyzable analogue of the normal activated amino acid intermediate (Ile-AMP) formed during the IleRS reaction cycle. Such amino-alkyl adenylates have been known for some time to be effective inhibitors of aaRSs (25). Ile-ol-AMP, was found to inhibit both the steady-state tRNA aminoacylation (Fig. 10) and PP i /ATP exchange reactions (data not shown) competitively with respect to both ATP and Ile. Analysis of the data in Fig. 10 (in the presence of 5 mM ATP) using Equation 4 yielded an estimate for the apparent K i Since Ile-ol-AMP is competitive with both Ile (Fig. 10) and ATP (Fig. 4), the initial rates subsequently fit to Equation 4 allowed the empirical relationship between K i and K i,app in this system to be defined. The results of these simulations indicated that the K i,app overestimated the true K i by 3.2-fold yielding an estimate of K i ϳ 70 nM. From the direct measurement of k ϩi and k -i for Ile-ol-AMP (Fig. 2 (12)), we obtain K i ϳ 30 nM indicating that our kinetic mechanism is able to quantitatively account, within reasonable margins (45 Ϯ 25 nM), for both the transient and steady-state kinetics of IleRS and for its inhibition by Ile-ol-AMP. DISCUSSION There is considerable interest in the potential therapeutic uses of selective inhibitors of aminoacyl-tRNA synthetases such as PS-A (12, 13) as anti-infective agents (e.g. Ref. 26). We have a long standing interest in the interactions of PS-A and of other inhibitors with IleRS, particularly in relation to the development of compounds with improved anti-bacterial or clinical properties (e.g. Refs. [27][28][29][30]. The recently solved crystal structure of the Thermus thermophilus IleRS⅐PS-A complex (31) has added extra impetus to this endeavor. As a first step to taking a more rational approach to the design of IleRS inhibitors, we have investigated the reaction cycle of this enzyme from S. aureus, one of the key target organisms for PS-A (11).
Minimal Kinetic Mechanism for Amino Acid Activation-Significant changes in enzyme fluorescence (up to 17%) were observed during the reaction cycle or following binding of the reaction intermediate analogue, Ile-ol-AMP, allowing us to use stopped-flow techniques to follow the binding and activation of substrates. Fluorescence resonance energy transfer experiments using chromophoric inhibitors of IleRS that quench the intrinsic protein fluorescence of IleRS suggest that a significant proportion of the observed IleRS tryptophan fluorescence originates from a few residues close to the active site (12).
The transient kinetic experiments described here allowed the solution of a complete minimal reaction scheme for the binding and activation of substrates. With the possible exception of the dissociation rate constant for Ile from the E⅐ATP⅐Ile ternary complex (k -4 , Scheme 1), we have been able to make estimates of most of the elementary rate constants or equilibrium constants that describe the IleRS amino acid activation reaction with reasonable accuracy (Scheme 1, Table I). Based on this minimal mechanism, kinetic simulations were performed which adequately describe the transient kinetics (including both the observed exponential rate constants and flu-orescence amplitudes) and the steady-state kinetics of ATP/PP i exchange.
A key feature of the mechanism is the random order of addition of substrates to form the E⅐Ile⅐ATP ternary complex with thermodynamic linkage between the binding of the first and second substrate (e.g. ATP binds more tightly to E⅐Ile than to free E and vice versa). This confirms previous qualitative evidence for such linkage from steady-state experiments with E. coli IleRS (22).

M
Ϫ2 approximates to the product K 2 ⅐K 4 Ϸ 2-6 ϫ 10 Ϫ9 M Ϫ2 (depending on the value used for K 4 ) and which defines the equilibrium constant for the (path-independent) reaction E ϩ ATP ϩ Ile 7 E⅐ATP⅐Ile. By analogy with tyrosyl-tRNA synthetase (33), the ϳ35-fold difference in equilibrium constants observed in the presence of the co-substrate corresponds to about 2 kcal/mol, and may therefore result from the formation of only two or three additional uncharged hydrogen bonds. A second feature of Scheme 1 is the inherent reversibility of the chemistry step (K 5 ϭ k ϩ5 /k -5 ϭ 0.3-0.4) which accounts for the known reversibility of overall ATP/PP i exchange and which adequately predicts the experimentally observed k cat,ex for this reaction (60 s Ϫ1 ϫ 0.3-0.4 ϭ 18 -24 s Ϫ1 ). However, the very weak affinity of PP i for the E⅐Ile-AMP⅐PP i complex (K 6 ϭ 0.25 mM) drives the reaction toward stoichiometric formation of E⅐Ile-AMP when IleRS is mixed with Ile and ATP at submicromolar concentrations of enzyme.
The dissociation of Ile from the E⅐Ile⅐ATP complex (k -4 ϭ 3 Ϯ 1.5 s Ϫ1 ) is slower than the k cat,ex in the reverse direction (PP i 3 ATP, 18 s Ϫ1 ). Therefore, if this estimate is accepted with the associated caveats, the kinetically preferred pathway must then involve release of ATP from E⅐Ile⅐ATP at a rate governed by k -3 rather than via the sequential release of Ile (at 3 Ϯ 1.5 s Ϫ1 ) and then of ATP. A second consequence of a low value for k -4 is that the binding of Ile to E⅐ATP has a large commitment to catalysis (since k ϩ5 Ͼ k -4 ) and hence the k cat,ex /K m,Ile (1.8 ϫ 10 6 M Ϫ1 s Ϫ1 , Table III) is predicted to approximate to the true bi-molecular association rate constant for Ile binding to E⅐ATP (1.6 ϫ 10 6 M Ϫ1 s Ϫ1 , Fig. 6B). Of course, Scheme 1 suggests that assumptions invoking rapid equilibrium binding of Ile prior to catalysis are invalid.
Comparison of Scheme 1 with the kinetics of tRNA Aminoacylation-Our main focus has been to define the elementary rate constants that describe the amino acid activation reaction. However, we have also examined the steady-state kinetics of tRNA aminoacylation, primarily to determine how closely Scheme 1 could predict the full tRNA aminoacylation reaction. The parameters determined in these experiments were broadly similar to those previously reported for the E. coli IleRS (22), although our data (Fig. 10) suggest a rapid-equilibrium ordered mechanism in which ATP binds first (Fig. 9) (24). However, analysis of the crystal structures of enzyme-substrate complexes of other aaRSs (e.g. Bacillus stearothermophilus TyrRS) (33), has led to questioning of the validity of an ordered substrate addition mechanism that arose from a steady-state kinetic analysis of TyrRS (33).
When considering the relationship between Scheme 1 and the complete tRNA aminoacylation reaction, it is clear that tRNA could bind to and dissociate from any of the enzyme intermediates. We have only two pieces of indirect evidence concerning their tRNA ligation states. First, proteolysis protection experiments confirm that tRNA binds to unligated E with an affinity of ϳ0.1 M (13), comparable to the K m,tRNA Ј determined here (Table III). The second indirect and more speculative evidence comes from our simulations of the tRNA aminoacylation reaction using Scheme 1. It is intuitive that any mechanism in which the (rate-limiting) transfer of the activated amino acid to the uncharged tRNA occurs subsequent to thermodynamically favorable PP i release will lead to K m values for ATP and Ile far below their equilibrium dissociation constants. In contrast, the comparability of the steady-state parameters (Table III) to their equivalent dissociation constants in Scheme 1 (Table I) led us to consider whether we could simulate, at least broadly, the tRNA aminoacylation reaction using these rate and equilibrium constants. If the tRNA binding and transfer reactions are simulated to occur prior to PP i release (E⅐Ile-AMP⅐PP i ⅐tRNA 3 E ϩ Ile-tRNA ϩ PP i ϩ AMP, Scheme 4), the remaining rate constants in Scheme 1 yield very similar steady-state kinetics to those observed experimentally (Fig. 9). Of course, Scheme 4 does not define the order of tRNA binding and dissociation which was not the purpose of the current study. However, since tRNA can bind to free IleRS with an affinity comparable to K m,tRNA Ј in the tRNA aminoacylation reaction (13), the most simple mechanism appears to involve tRNA binding throughout the reaction cycle as shown in Scheme 4. Although it is possible that the rather striking similarity of the tRNA aminoacylation kinetics predicted from Scheme 4 and of those observed experimentally (Fig. 9) is coincidental, we feel this is somewhat unlikely. Rather, our current working hypothesis is that the rate and equilibrium constants in Scheme 4 which yield the random-order (ATP preferred) mechanism are slightly modified in the presence of tRNA to lead to the ordered (ATP first) mechanism observed experimentally. In addition, the K i for Ile-ol-AMP binding to tRNA-free IleRS determined in direct binding experiments (30 nM, Fig. 2) (12) is similar to that predicted from the inhibition of tRNA aminoacylation according to Scheme 4 (70 nM). Clearly, testing this hypothesis experimentally by using the types of transient stopped-flow approaches described here with an E⅐tRNA complex is the next stage in the construction of a kinetic mechanism for the complete tRNA aminoacylation reaction of S. aureus IleRS.
The data described in this paper represent the most detailed study so far of IleRS and of an aaRS that is the major target for an antibiotic in widespread clinical use. The aims were to provide the basis for further mechanistic studies of the full tRNA aminoacylation reaction and for understanding the mechanism by which IleRS from S. aureus develops resistance to inhibition by clinically used PS-A and its derivatives (e.g. Refs. [27][28][29][30]. In an accompanying paper (13), we describe a companion study of the mechanism of binding of such compounds. However, Schemes 1 and 4 have already proven useful in understanding the structure-activity relationships for the effects of inhibitors on individual rate constants and processes in the reaction cycle (yielding "microscopic structure-activity relationships"). Furthermore, using these techniques and Schemes we have been able to define structure-activity relationships around rate and equilibrium constants (and E⅐inhibitor conformations) describing the interaction of the inhibitors with IleRS which are otherwise inaccessible from traditional steady-state measurements of the inhibition of enzyme catalysis. Finally, the present work provides a dynamic framework to interrogate kinetically conclusions and hypotheses derived from on-going structural and site-directed mutagenesis experiments, including those arising from the recent solution of the crystal structure of the T. thermophilus IleRS⅐PS-A complex (31).