The ATPase Mechanism of ArsA, the Catalytic Subunit of the Arsenite Pump*

The ArsA ATPase is the catalytic subunit of a novel arsenite pump, with two nucleotide-binding consensus sequences in the N- and C-terminal halves of the protein. The single tryptophan-containing Trp159 ArsA was used to elucidate the elementary steps of the ATPase mechanism by fluorescence stopped-flow experiments. The binding and hydrolysis of MgATP is a multistep process with a minimal kinetic mechanism (Mechanism 1). Mechanism 1 A notable feature of the reaction is that MgATP binding induces a slow transient increase in fluorescence of ArsA, which is independent of the ATP concentration, indicative of the build-up of a pre-steady state intermediate. This finding, coupled with a phosphate burst, implies that the steady-state intermediate builds up subsequent to product release. We propose that the rate-limiting step is an isomerization between different conformational forms of ArsA.k cat is faster than the phosphate burst, indicating that both nucleotide binding sites of ArsA are catalytic. Consistent with this interpretation, approximately 2 mol of phosphate are released per mole of ArsA during the phosphate burst.

Resistance to arsenical and antimonial salts in Escherichia coli is conferred by the ars operon of conjugative R-factor R773 (1). This operon encodes an ATP-coupled efflux pump that actively transports the trivalent arsenicals and antimonials out of the cell; reducing the intracellular concentration of those metalloid oxyanions to subtoxic levels produces resistance (2). The pump is composed of two subunits, ArsA, a 63-kDa cata-lytic subunit, and ArsB, a 45-kDa integral membrane protein that is the membrane anchor for ArsA and the oxyanion-translocating sector of the pump (3). ArsA can be purified as a soluble ATPase in the absence of ArsB (4). ArsA is arranged into two homologous halves, the N-terminal (A1) (residues 1-282) and C-terminal (A2) (residues 321-583) domains, which are connected by a flexible linker (residues 283-320) (5,6). Each domain contains the consensus sequence for the phosphate binding loop (P-loop) of an ATP-binding site (7). Sitedirected mutagenesis of these sequences indicates that both of the putative nucleotide binding sites are required for catalysis and resistance (8,9). Intergenic complementation and intragenic suppression studies on ArsA were suggestive of a model in which a single catalytic site was formed at the interface of an A1 and an A2 ATP binding site, possibly within a homodimer (10,11).
To investigate the interaction of nucleotides with the A1 and A2 sites, we have conducted stopped-flow fluorescence experiments on ArsA. We have previously shown that intrinsic tryptophan fluorescence can be used to investigate the interaction of ArsA with its ligands (12). However, the presence of multiple tryptophan residues in ArsA decreased the signal-to-noise response to ligand binding, leading us to construct single tryptophan derivatives of ArsA that are optically responsive to the binding of substrates or products (13). Mutant arsA genes were constructed containing single tryptophan codons (13). Tryptophan residues 253, 522, and 524 were changed to tyrosine residues, and a six-histidine tag was added to the C terminus, producing a histidine-tagged ArsA containing only Trp 159 (W159H6) (13). The single tryptophan-containing W159H6 ArsA gave approximately a 4-fold increase in the signal-tonoise ratio in response to MgATP/mol of tryptophan residue when compared with the wild type enzyme (13). We have used this ArsA derivative for the transient kinetic studies reported herein. It has proved possible to monitor the interaction of MgATP with W159H6 by stopped-flow fluorescence spectroscopy. Based on these analyses, we propose a model for the mechanism of MgATP binding and hydrolysis.

MATERIALS AND METHODS
Purification of His6-tagged ArsA ATPase-ArsA W159H6 ArsA was purified as described previously (13), quickly frozen, and stored in small aliquots at Ϫ80°C. The concentration of purified ArsA was determined by UV absorbance at 280 nm. The extinction coefficient for W159H6 ArsA was calculated to be 20,250 M Ϫ1 cm Ϫ1 (14).
ATPase Assays-A continuous assay was used to monitor phosphate production by ArsA. Essentially, the absorbance change at 360 nm associated with the phosphorolysis of 2-amino-6-mercapto-7-methylpurine by the inorganic phosphate generated by the ATPase activity was monitored (15). The phosphorolysis reaction was catalyzed by purine nucleotide phosphorylase. The components of the assay were provided as part of an EnzCheck phosphate assay kit (Molecular Probes, Eugene, OR) and used according to the manufacturer's recommendations. Assays were performed in 40 mM Tris-HCl (pH 7.5), 2 mM MgCl 2 , contain-ing 0.2 mM sodium azide. The change in absorbance with time was measured in a Unicam (UV2) UV-visible spectrometer. Absorbance changes were converted into phosphate concentrations with ⌬E 360 ϭ 12 mM Ϫ1 cm Ϫ1 .
Fluorescence Measurements-Fluorescence measurements were performed in a Jasco FP777 spectrofluorometer maintained at 20°C. The excitation wavelength was set at 292.5 nm for selective excitation of tryptophan fluorescence, and the emission was monitored at a wavelength of 340 nm. The bandwidths for both emission and excitation monochromators were 3 nm. The concentration of ArsA was 5 M unless otherwise noted in 50 mM MOPS 1 -KOH (pH 7.5), 0.25 mM EDTA.
In addition, time-resolved fluorescence measurements were also carried out in an Applied Photophysics (London, UK) SX.18MV stoppedflow instrument, operated at 20°C. For measurements of the change in tryptophan fluorescence, the samples were excited with light at 292.5 nm, selected with a monochromator, and the emission monitored at wavelengths above 335 nm, using a cut-off filter. Invariably, equal volumes of the reactants were mixed together in the stopped-flow instrument, using two syringes of equal volume. Standard conditions for multiple turnover experiments were 5 M ArsA and for single turnover experiments, 25 M ArsA. All concentrations are for the mixing chamber, unless stated otherwise, so that the concentrations in the syringe were twice those quoted for the mixing chamber.
Data Analysis-The phosphate burst data was analyzed in terms of an exponential plus linear function using the nonlinear regression routine of the software package, SIGMAPLOT 4.0 (SPSS Software Inc., Chicago, IL). Stopped-flow traces were analyzed by fitting to single or multiple exponential functions, as appropriate, using the nonlinear regression software with the Applied Photophysics stopped-flow. Concentration dependence data were analyzed by nonlinear regression fitting to hyperbolic functions, using SIGMAPLOT 4.0. Kinetic simulations were set up using the program Pro-K (Applied Photophysics), which uses the Marquardt-Levenberg algorithm for global optimization of the reaction parameters.

RESULTS
The Kinetics of ATP Binding and Turnover-Previously we established that the tryptophan fluorescence of ArsA (W159H6) was sensitive to the binding of MgATP (13). Fig. 1A shows the change in tryptophan fluorescence when 5 M ArsA was manually mixed with 1 mM MgATP; the reaction was initiated by the addition of 5 mM MgCl 2 to 5 M ArsA, 1.0 mM ATP. There was a rapid biphasic increase in the tryptophan fluorescence, to reach a plateau, before the fluorescence decayed back to the baseline level (as defined by the fluorescence of 5 M ArsA, 1.0 mM ATP), suggesting that little product MgADP remains bound to the ArsA. However, the addition of EDTA at the end of the reaction, to chelate the Mg 2ϩ , caused a further decrease in tryptophan fluorescence, suggesting that some product MgADP remained bound at the end of the reaction (Fig.  1A). A potentially plausible explanation for the fluorescence profile of the reaction is that the binding of MgATP to the ArsA induces an enhancement in the tryptophan fluorescence, which remains constant until the ATP has been depleted, at which point the fluorescence decays as the products are released. To test this hypothesis, the mixing experiment was repeated with higher concentrations of ArsA and ATP. The period during which the enhanced fluorescence remained constant should have been diminished for higher ArsA and extended for higher ATP if this phase was determined by the time taken for the steady-state turnover of the ATP. As shown in Fig. 1, B and C, no significant effect was observed upon this phase with either higher ATP or ArsA concentrations, respectively. This suggested that the enhancement in tryptophan fluorescence was due to a transient build up and decay of a reaction intermediate. The addition of excess EDTA rapidly reduced the ArsA fluorescence, presumably due to the dissociation of ADP in the absence of Mg 2ϩ (Fig. 1), thus suggesting that the intermediate was an ArsA-Mg 2ϩ nucleotide complex because MgCl 2 alone did not produce a change in the fluorescence of the ArsA (data not shown). The addition of EDTA to ArsA or ArsA/Mg 2ϩ did not produce any change in the fluorescence of the ArsA and only a small decrease with ArsA/MgADP (data not shown). This behavior suggests that the intermediate was ArsA-MgADP⅐P i . 2 We have investigated the kinetics of formation of this intermediate in detail using stopped-flow fluorescence spectroscopy. A typical stopped-flow trace for the mixing of 5 M ArsA, 0.5 mM ATP with 5 mM MgCl 2 (mixing chamber concentrations) is shown in trace A of Fig. 2. The trace was similar to that observed in manual mixing experiments but the rapid-mixing experiment identified more complex behavior at shorter times. Over the first 10 s, the profile was clearly multiphasic with a very fast increase in fluorescence, followed by a moderately fast decrease and then a slow increase in fluorescence (Fig. 3). Although the two fast phases merged, they were well resolved from the slow phase. Accordingly, the initial part of the trace was analyzed as a double exponential (during the first 4 s) and then as a single exponential over the remainder of the trace. The three phases occurred with rate constants of 52 s Ϫ1 (phase 1), 5.2 s Ϫ1 (phase 2), and 0.026 s Ϫ1 (phase 3), respectively (Fig.  3). The entire data set was also fitted to a triple exponential, which indicated similar rate constants of 48 s Ϫ1 , 5.2 s Ϫ1 , and 0.027 s Ϫ1 for the three phases, respectively. The slow decay of the enhanced fluorescence back toward the baseline occurred with a rate constant of approximately 2.4 ϫ 10 Ϫ3 s Ϫ1 (phase 4). Previous studies apparently indicated that the slow decay in fluorescence (phase 4) correlates with the rate of P i production under limited turnover conditions (e.g. 5 M ArsA and 20 M 1 The abbreviation used is: MOPS, 4-morpholinepropanesulfonic acid. 2 These data also indicate that no residual ADP remains bound to the ArsA protein following its purification. MgATP), suggesting that this phase is attributable to the product release step (13).
To clarify the mechanism, the dependence of the rate and amplitude of each phase upon the ATP concentration was determined. The amplitude of phase 2 decreased with the ATP concentration to a level around 200 M ATP where it could no longer be distinguished from phases 1 and 3. At these concentrations, the rate constant for phase 1 was determined either by fitting to a single exponential function over the first few seconds of the trace or to a double exponential function over the entire trace (e.g. 200 s). The rate constant for the very fast phase (phase 1) increased in an apparently hyperbolic manner with the ATP concentration up to a maximum around 1.5 mM and then decreased at higher concentrations (Fig. 4). 3 The data obtained at ATP concentrations up to 1.5 mM could be fitted to a hyperbola, indicating maximal and minimal rates of binding of 53.9 s Ϫ1 and 7.3 s Ϫ1 , and a K d of 178 M. A minimal value can be calculated for the second-order association rate constant for the binding of ATP to the ArsA from (k max ϩ k min )/K d , yielding a value of 0.34 ϫ 10 6 M Ϫ1 s Ϫ1 . For a fixed ATP concentration, the rate constant for phase 1 was independent of the Mg 2ϩ concentration (data not shown). Over the ATP concentration range that phase 2 could be reliably measured, the rate constant for this phase had little concentration dependence, varying between 4.5 s Ϫ1 and 6.5 s Ϫ1 (Fig. 4, f). This behavior is consistent with the simple three-step mechanism shown in Scheme 1, where the first step is a rapid equilibrium with binding constant K 2 (K 2 ϭ k Ϫ2 /k 2 ) followed by transitions to states with high fluorescence enhancement (e.g. ArsA 4 -MgATP) and lower fluorescence enhancement (e.g. ArsA 5 -MgATP). The fluorescence transient fits two exponential terms with observed rate constants, The ratio of the amplitudes of the two exponential phases at high substrate concentration will provide a measure of the ratio of enhancements for the two states, calculated as 2.6 (16).
The concentration dependence of the amplitude of phase 1 indicated an overall equilibrium constant of 760 M (Fig. 5). A plausible explanation for this behavior is that the ArsA protein exists in two interconverting conformations (e.g. ArsA 1 and ArsA 2 ) that differ in their affinities for ATP 4 (see Scheme 2).
The overall K d is a function of both K 1 and K 2 : K d ϭ K 2 (1 ϩ K 1 )). Accordingly, K 1 can be calculated, from the overall K d and K 2 , as 3.33, indicating that 77% of the ArsA is in the ArsA 1 conformation before the binding of MgATP. Because the rate of phase 1 increased with the ATP concentration but was independent of the Mg 2ϩ concentration, this implies that we were monitoring the binding of MgATP to the ArsA. A possibility is that the ArsA 1 state is stabilized by the nonproductive binding of ATP in the absence of Mg 2ϩ , which must dissociate before MgATP can bind.
Neither the rate constant for, nor the amplitude of, the slow increase in fluorescence (e.g. phase 3) had any apparent dependence upon the ATP concentration (data not shown). The rate constant varied nonsystematically between 0.025 and 0.05 s Ϫ1 (data not shown). The rate constant for the slow decay in fluorescence back to the baseline (phase 4) occurred with a rate constant of 2.3 ϫ 10 Ϫ3 s Ϫ1 that was also independent of the ATP concentration (data not shown). Consequently suggesting that these phases are attributable to a further isomerization of the ArsA-nucleotide complex (e.g. the formation and decay of ArsA 6 ).
In a parallel set of experiments, the binding of MgATP to ArsA was investigated. A typical stopped-flow trace for the mixing of 5 M ArsA with 0.5 mM ATP,5 mM MgCl 2 is shown in Fig. 2B. The binding of MgATP to ArsA was characterized by a slow, apparently monophasic, fluorescence enhancement that occurred with a rate constant of 0.0334 s Ϫ1 . Thus, this phase occurred at a rate comparable with the slow fluorescence enhancement that occurred when the reaction was initiated by mixing ArsA/ATP with Mg 2ϩ (cf. 0.026 s Ϫ1 ). However, the fast increase (phase 1) and decrease (phase 2) in fluorescence, which occurred within 4 s of initiating the reaction with Mg 2ϩ , were not apparent. There was an indication of a hyperbolic increase in the rate constant, between 0.022 s Ϫ1 and 0.058 s Ϫ1 , for the slow enhancement in fluorescence when ArsA was mixed with MgATP. This behavior is consistent with a two-step binding mechanism. The rapid equilibrium binding of the ATP, which is rate limited by a slow isomerization of the ArsA⅐ATP complex (e.g. The data could be fitted to a hyperbolic equation for such a model: k obs ϭ k Ϫ2 ϩ k 2 [MgATP]/([MgATP] ϩ K 1 ), yielding values for K 1 , k 2 , and k Ϫ2 of 624 M, 0.038 s Ϫ1 and 0.022 s Ϫ1 , respectively (Fig. 6A). The overall K d would be given by the following function: The amplitude of the ATP-induced fluorescence enhancement increased in a hyperbolic manner with the ATP concentration, indicating an overall dissociation constant (K d ) of 427 (Ϯ62.3) M (Fig. 6B). Hence, there is a small discrepancy between the predicted and measured overall K d , which could readily be reconciled by postulating a further conformational transition of the ArsA⅐MgATP complex. However, this seems unwarranted at this time considering the difficulty in measuring the small increase in the rate constants k 2 and k Ϫ2 and the consequent error in K 1 . This behavior contrasts with that for the Mg 2ϩ initiated reaction, where the amplitude of the slow increase in fluorescence was apparently independent of the ATP concentration. A plausible explanation is that the MgATP initially binds to an ArsA state that does not produce any change in fluorescence but which is followed by a slow transition to a state with an enhanced fluorescence. When the ArsA was pre-equilibrated with Mg 2ϩ before initiating the reaction by mixing with ATP, the traces were similar to those generated by mixing ArsA with MgATP (data not shown). This indicates that if Mg 2ϩ can bind to ArsA in the absence of nucleotides then this must be a rapid equilibrium process. There was an increase in fluorescence, which occurred with a rate constant of 3.6 ϫ 10 Ϫ2 s Ϫ1 , and a subsequent decay in fluorescence that occurred with a rate constant of 7.8 ϫ 10 Ϫ4 s Ϫ1 . The protein fluorescence decayed back to a level lower than the start of the trace, indicating that there is a very fast increase in fluorescence probably due to the binding of ATP. Indeed, we consistently noted a rapid increase in fluorescence at the start of the single turnover traces that occurred with a rate constant of approximately 20 s Ϫ1 . In comparison with multiple turnovers, the amplitude of the decay phase was greater in relation to that of the enhancement phase, suggesting that ADP remained bound to the ArsA after multiple turnovers.
In contrast, when 25 M ArsA was incubated with 25 M ATP before mixing with 5 mM MgCl 2 in the stopped-flow instrument, there was a rapid increase in the protein fluorescence (phases 1 and 2), followed by a slow decay (phase 4) (Fig. 7B). There was no slow increase in fluorescence (phase 3). It was possible to fit the data to a triple exponential function with the increase and the biphasic decrease in fluorescence occurring with rate constants of 12.3 s Ϫ1 (phase 1), 6.22 s Ϫ1 (phase 2), and 2.0 ϫ 10 Ϫ3 s Ϫ1 (phase 4), respectively. The rate constants for phases 1, 2, and 4 were comparable with the calculated rate of ATP binding (cf. 14 s Ϫ1 for 25 M ATP) and with the rate of fluorescence decay under multiple turnover conditions (cf. 4.5-6.5 s Ϫ1 for under these single turnover conditions, the amplitude of the decay phases (e.g. phase 4) when ArsA/ATP was mixed with Mg 2ϩ was similar to that when ArsA was mixed with MgATP (e.g. 3.0% versus 2.3% fluorescence change, respectively). This slow decay in fluorescence cannot be attributed to multiple turnovers, which depleted the concentration of this intermediate, during the latter stages of the reaction, because this phase was independent of the ATP concentration (cf. from measurements of single and multiple turnovers). The slow increase and decrease in fluorescence presumably represents the transient formation of an intermediate that forms and decays at rates of 6.5 s Ϫ1 and 2.3 ϫ 10 Ϫ3 s Ϫ1 , respectively. If this is an in-line intermediate then the pre-equilibration with ATP (in the absence of Mg 2ϩ ) must induce the ArsA to adopt a similar conformation.
ADP Binding -The binding of MgADP to ArsA was characterized by a small enhancement in the tryptophan fluorescence, precluding a rigorous analysis of the kinetics of binding. However, when ArsA was mixed with a relatively high MgADP concentration (e.g. 10 mM), the stopped-flow traces generated were similar to those for the binding of MgATP to ArsA, which had been pre-equilibrated with ATP. There was a rapid increase and decrease in the fluorescence, followed by a slow increase but not a subsequent decrease (Fig. 8). Recently, we have established that the W141H6 ArsA mutant has greater optical sensitivity to the binding of ADP, and a detailed analysis of the ADP binding mechanism for this mutant will be presented elsewhere.
ADP Dissociation -To test whether the dissociation of MgADP was a rapid or slow process, the ArsA⅐MgADP complex was chased with excess MgATP. Clearly, if MgADP dissociation was a rapid process then we would have expected to observe an increase in the ArsA fluorescence as the MgATP bound. On the other hand, if dissociation of the MgADP was slow, then we would have expected to observe a decrease in the fluorescence as the MgADP dissociated. When 5 M ArsA was equilibrated with 50 M ADP, 5 mM MgCl 2 , and mixed with 500 M ATP in a stopped-flow device, there was a relatively slow decrease in fluorescence, presumably as the ADP was displaced, followed by a slow increase in fluorescence, presumably as the ATP was hydrolyzed (Fig. 9). The amplitude of the decrease in fluorescence increased in a hyperbolic manner with the ArsA/MgADP incubation time (Figs. 9 and 10), indicating that the ADP induced a slow conformational change in the ArsA, which occurred with a rate constant of 1.62 ϫ 10 Ϫ4 s Ϫ1 . This behavior is consistent with the increase in ArsA fluorescence that we have noted upon the binding MgADP to ArsA. This conformational change was to a form that allowed more rapid product release because the rate constant for the dissociation process increased from 0.076 to 0.131 s Ϫ1 over the 8-h incubation time (Fig. 10).
Phosphate Burst-A burst in phosphate production was identified using a continuous assay to monitor phosphate release from ArsA. As shown in Fig. 11, when 1 M ArsA was mixed with 50 M ATP there was an exponential increase in the phosphate concentration during the first 400 s, followed by a  linear steady-state release of phosphate. 5,6 The rate of the pre-steady-state phase only increased slightly with increasing ATP concentration to a value of 3.7 ϫ 10 Ϫ3 s Ϫ1 for 300 M ATP. The pre-steady-state phase could not be readily resolved from the steady-state phase for higher ATP concentrations. In conclusion, during the phosphate burst approximately 2 nmol 7 of phosphate were released at a rate faster than 3.7 ϫ 10 Ϫ3 s Ϫ1 (for near saturating ATP concentrations).
The steady-state rate, as determined from the linear part of the phosphate release time course, increased in a hyperbolic manner with the ATP concentration to a maximal level around 200 M ATP and thereafter was subject to substrate inhibition (Fig. 12). Hence, values for V max , K m , and K i , the substrate inhibition constant, were determined from a fit of the data to the following equation for substrate inhibition: v ϭ V max [ATP]/ K m ϩ [ATP] ϩ [ATP] 2 /K i . This analysis yielded values for V max , K m , and K i of 5.4 ϫ 10 Ϫ3 nmol s Ϫ1 ⅐nmol Ϫ1 ArsA, 72 M and 792 M, respectively, and indicated a k cat of 5.4 ϫ 10 Ϫ3 s Ϫ1 (single site catalysis) or 2.7 ϫ 10 Ϫ3 s Ϫ1 (two-site catalysis).

DISCUSSION
ArsA is the catalytic subunit of the arsenite transporter and is thought to couple the hydrolysis of ATP to the movement of arsenicals and antimonials through the membrane-spanning ArsB protein. Consequently, knowledge of the ArsA ATPase mechanism will provide information of fundamental importance in understanding the energy transduction processes common to many transporters that are driven by ATP hydrolysis, such as those belonging to the ABC superfamily. Utilizing a derivative of ArsA that contains only a single tryptophan residue, Trp 159 , which is optically responsive to the binding of ATP, the ATPase mechanism of ArsA was investigated.
The kinetics of the binding of MgATP to ArsA were indicative of a multistep mechanism. When ArsA was pre-equilibrated with ATP before mixing with Mg 2ϩ to initiate the reaction, the binding of MgATP to the ArsA could be monitored as an increase in ArsA fluorescence. The rate constant for the binding step increased hyperbolically, indicative of a two-step process for the sequential formation of ArsA 3 ⅐MgATP and ArsA 4 ⅐MgATP. The forward and reverse rate constants for the formation of these intermediates were calculated as k 2 Ͼ 0.34 ϫ 10 6 M Ϫ1 s Ϫ1 , k Ϫ2 ϭ 7.3 s Ϫ1 , k 3 ϭ 53.9 s Ϫ1 and k Ϫ3 ϭ 7.3 s Ϫ1 (Schemes 1, 2, and 4 nomenclature). Formation of the ArsA 4 ⅐MgATP complex was followed by a further isomerization to the less fluorescent ArsA 5 ⅐MgATP state, at a rate of 4.5-6.5 s Ϫ1 . There then followed a further conformational change in which an intermediate with a higher fluorescence was formed at a rate of 0.025-0.05 s Ϫ1 . In comparison, when ArsA was mixed directly with MgATP only a slow monophasic increase in the fluorescence was observed, which occurred at a similar rate 5 Less than 10% of the ATP was hydrolyzed over the studied time course. 6 The data was best-fitted to an exponential plus a linear term with a nonzero intercept (e.g. a nonzero absorbance at the start of the experiment) as judged by a 400% decrease in residual variance of the fit. We have found a small upward drift in the absorbance of the reactants before the addition of the ArsA to initiate the reaction. 7 Eight measurements gave an average burst of 1.7 (Ϯ0.28) nmol. to the formation of this intermediate. However, both the rate constant for, and the concentration of, this intermediate increased in a hyperbolic manner with the ATP concentration. This behavior is consistent with a two-step binding process: the rapid equilibrium binding of ATP, to produce an ArsA⅐MgATP state with unaltered fluorescence, followed by a slow isomerization to a more fluorescent state. Accordingly, there are several lines of evidence to indicate that ArsA adopts a different conformation after equilibration with ATP. The binding of MgATP directly to ArsA (ϪATP state) is optically silent, whereas the binding of MgATP to ArsA pre-equilibrated with ATP (ϩATP state) induces an enhancement in fluorescence. It is conceivable that in the absence of ATP the ArsA adopts a similar conformation to ArsA 5 ⅐MgATP, a "low-fluorescence" state. The ϪATP state has a lower affinity for MgATP (K d ϭ 624 M) than the ϩATP state (K d ϭ 178 M). Pre-equilibration of the ArsA with ATP induces a conformation that has low or no affinity for MgATP, the ArsA 1 state. The formation of this state is not apparent from the kinetics of MgATP binding to ArsA or of ATP binding to ArsA/Mg 2ϩ , suggesting that this is an ArsA 1 -ATP state. Indeed, formation of this state could be the mechanistic basis for the observed substrate inhibition at high ATP concentrations.
Irrespective of the mixing order, the slow formation of an intermediate with enhanced fluorescence is observed under multiple turnover conditions. The kinetic data is indicative of an intermediate that builds up and decays with rate constants of 3.8 ϫ 10 Ϫ2 s Ϫ1 and 1-3 ϫ 10 Ϫ3 s Ϫ1 , respectively (e.g. phases 3 and 4). This cannot be the steady-state intermediate for the reaction; otherwise it would only decay after depletion of the ATP. Accordingly, its formation must precede the rate-limiting step of the reaction. This intermediate has greater fluorescence than the ArsA⅐MgADP complex suggesting that it is the ArsA-MgADP⅐P i complex. An alternative possibility is that it is a  11. Phosphate burst phase of ArsA ATPase. 1 M ArsA was manually mixed with 50 M ATP, and the release of phosphate was monitored spectrophotometrically as the change in absorbance at 360 nm associated with the phosphorolysis of 2-amino-6-mercapto-7-methylpurine by phosphate. The curve through the data is the best-fit to a single exponential plus a linear term with a nonzero intercept: y ϭ a(1 Ϫ exp(b ϫ t)) ϩ (c ϫ t) ϩ d, where y is the absorbance at time (t); a is the amplitude of the burst in phosphate release; c is the linear rate of change in the absorbance; and d is the absorbance at the start of the experiment (zero time). A fit to this equation yielded values for a, b, c, and d of 2.94 ϫ 10 Ϫ2 (Ϯ1.50 ϫ 10 Ϫ4 ) absorbance/units, 1.14 ϫ 10 Ϫ3 (Ϯ8.5 ϫ 10 Ϫ6 ) s Ϫ1 , 2.1 ϫ 10 Ϫ5 (Ϯ4.6 ϫ 10 Ϫ8 ) absorbance/units/s Ϫ1 , and 1.6 ϫ 10 Ϫ2 absorbance units, respectively. This correlates with a 2.45 nmol pre-steady-state burst of phosphate release at a rate of 1.14 ϫ 10 Ϫ3 s Ϫ1 and a steady-state release of phosphate at a rate of 1.75 ϫ 10 Ϫ3 nmol s Ϫ1 nmol Ϫ1 of ArsA. state with a higher affinity for ADP relative to that of the final state. Consistent with the former interpretation, phosphate is released with a relatively slow rate constant (e.g. k off ϭ 3.7 ϫ 10 Ϫ3 s Ϫ1 for the phosphate burst with 300 M ATP) compared with the rate of formation of this intermediate (e.g. k ϭ 3.8 ϫ 10 Ϫ2 s Ϫ1 ). The slow release of phosphate suggests that there would not be an appreciable build up of ArsA⅐MgADP during the time course of the build up of the intermediate with enhanced fluorescence. We tentatively conclude that this intermediate is the ArsA-MgADP⅐P i complex. We have previously measured phosphate production during a 4-fold limited turnover of ATP by ArsA (13). A discontinuous assay was used to monitor the hydrolysis of 20 M ATP by 5 M ArsA, with the reaction terminated at set times with trichloroacetic acid to displace bound P i . We found that during the period of enhanced fluorescence (phase 3, 100 -200 s), there was a stoichiometric production of bound phosphate, consistent with the proposal that phase 3 is attributable to the production of ArsA-ADP⅐P i . The slow decay in fluorescence is presumably attributable to the product release steps. ADP is released at a much faster rate than the decay in fluorescence (e.g. for ADP k off Ն 0.08 s Ϫ1 , whereas k obs for phase 4 ϭ 1-3 ϫ 10 Ϫ3 s Ϫ1 ). On the other hand, the decay in fluorescence occurs at a rate similar to that for phosphate release (e.g. for P i k off Ն 3.7 ϫ 10 Ϫ3 s Ϫ1 , whereas k obs for phase 4 ϭ 1-3 ϫ 10 Ϫ3 s Ϫ1 ). If this is the case, then hydrolysis of the ATP must be fast because the phosphate burst includes both the hydrolysis and phosphate release steps. An alternative possibility is that the fluorescence decay is due to a conformational change subsequent to the release of the phosphate. However, the fact that there is a phosphate burst implies that phosphate release is more rapid than the step that is rate-limiting in the steady-state. Because neither the hydrolysis or product release steps are rate-limiting for the steadystate, we conclude that there is a rate-limiting conformational change in ArsA following product release.
The stoichiometry of the phosphate burst is 1.7 (Ϯ0.28), suggesting that both nucleotide sites of ArsA are catalytic. Moreover, the steady-state rate for a single site is apparently faster than the rate constant for the phosphate burst and phase 4 of the fluorescence profile, whereas these steps in the reaction mechanism must be faster than the rate-limiting step. However, if both sites were catalytic, this would in effect halve k cat , which would then have a value reasonably consistent with the other kinetic data. Indeed, computer simulations revealed that the formation of the ArsA-MgADP⅐P i intermediate (with enhanced fluorescence) need only be followed by the relatively fast formation of a further intermediate (e.g. 10ϫ the rate constant for phase 4) that decays with a rate constant marginally slower than does ArsA-MgADP⅐P i . Under these conditions an intermediate would build up and decay at the rates of phases 3 and 4. Thus, there is kinetic evidence to support the supposition that both the A1 and A2 sites of ArsA bind and hydrolyze ATP. Previous studies have shown that the A1 and A2 sites can be covalently labeled with ATP and FSBA, respectively, indicating that both sites are available for nucleotide binding (17,18). However, it was proposed that antimonite might act as a switch in regulating ATP binding to the A2 site (18). Our studies do not support this supposition; we find no evidence for two sites of differing affinity, and the data suggest that both sites are catalytically competent.
It was possible to simulate the time for the fluorescence changes associated with the binding of ATP to ArsA (e.g. 1 mM ATP to 5 M ArsA) using the measured rate constants and the minimal kinetic mechanism shown in Scheme 4 with the kinetic constants,  For curve A the rate constants for the kinetic scheme were fixed, and the fluorescence values for the different ArsA states were optimized during the fitting procedure. Curve B was generated using a model for which an extra isomerization of ArsA 7 was included in kinetic Scheme 4. For curve B the rate constants for the final two (isomerization) steps and the fluorescence values for the different ArsA states were optimized during the fitting procedure. Visual inspection of the curves indicates a better fit of curve (B) to the data than curve (A) (which is accompanied by a 1.3-fold improvement in the residual variance). One vertical division represents a 1.0% change in the fluorescence of ArsA. The reaction time is presented in logarithmic progression on the horizontal axis. ϭ 0.08 s Ϫ1 , k 8 ϭ 0.001 s Ϫ1 , k 1 ϭ fast. 8 The model allows for the formation of an intermediate (e.g. ArsA 6 -MgADP⅐P i ) during the first 100 s and its decay over the following 900 s and for a pre-steady state burst and subsequent steady-state release of phosphate (Fig. 13). The fluorescence of the intermediate decays to about 0.33-fold that of the maximum due to the formation of the subsequent steady-state intermediate (e.g. ArsA 7 ), which only decays after all of the ATP has been depleted. Under single turnover conditions, there is a decay in the fluorescence of the ArsA 6 -MgADP⅐P i intermediate to near the baseline because there is no substantial build up in the steady-state intermediate (Fig. 13). The model incorporates a number of simplifications and assumptions; first, that hydrolysis occurs at step 5, whereas equally this could occur at step 4, which would then be followed by a slow isomerization between different conformations of the ArsA-ADP⅐P i complex. In any event, the hydrolysis step is faster than the subsequent productrelease steps and the overall mechanism would be similar. The build up of the ArsA 6 -MgADP⅐P i intermediate requires the subsequent formation (at moderate rate compared with k 5 ) and decay (at a slower rate than k 5 ) of a steady-state intermediate. To minimize the number of mechanistic steps, we propose the ordered dissociation of phosphate followed by ADP, leading to the build up of the ArsA 7 steady-state intermediate. Both phosphate and ADP release precede the steady-state step, which we propose to be a conformational change in the ArsA protein. The model predicts a return to baseline fluorescence as the ATP is depleted at the end of the reaction, because we have made ADP dissociation irreversible. However, ArsA can bind MgADP, and this is a multistep process. We do not observe a decay in the fluorescence to the baseline under multiple turnover conditions probably because such behavior would be masked by the increase in fluorescence due to the binding of MgADP. As a further test of the validity of the model, we attempted to fit a stopped-flow fluorescence trace directly to kinetic Scheme 4. Clearly, for a single trace the problem would be too ill-conditioned to identify a unique solution in which all the parameters (e.g. the rates of interconversion of the different ArsA states and their relative fluorescence values) were allowed to simultaneously vary during the nonlinear fitting procedure. Instead, the rates were held constant, and the relative fluorescence values of the different ArsA states were optimized. The fitting procedure indicated relative fluorescence values for ArsA 2 , ArsA 3 -MgATP, ArsA 4 -MgATP, ArsA 5 -MgATP, ArsA 6 -MgADP⅐P i , ArsA 7 -MgADP, and ArsA 7 of 1.00, 0.995, 1.052, 1.041, 1.037, 1.797, and 1.019, respectively. This analysis suggested that the maximal fluorescence enhancement observed 100 s after mixing ArsA with MgATP was attributable to the build up of ArsA 6 -MgADP⅐P i (e.g. 3 M) and ArsA 7 -MgADP (e.g. 0.08 M). 9 Although there is only a slight build up of ArsA 7 -MgADP, its high fluorescence enhancement (e.g. 79.7%) is a significant contribution to the overall enhancement. Fig.  14 shows the best-fit curve (e.g. curve A) superimposed upon a semi-logarithmic plot of a stopped-flow trace. There is a deviation of the best-fit curve from the measured data toward the end of the trace that is probably attributable to the fact that the decay in fluorescence is biphasic (rather than monophasic). To test this hypothesis, we introduced a further step (e.g. an isomerization of ArsA 2 ) and allowed a free fit of the rate constants for these last two steps (e.g. steps 8 and 9), indicating respective values of 1.3 ϫ 10 Ϫ2 (Ϯ2.3 ϫ 10 Ϫ3 ) s Ϫ1 and 4.2 ϫ 10 Ϫ5 (Ϯ1.7 ϫ 10 Ϫ5 ) s Ϫ1 . Clearly the latter step is too slow relative to the steady-state rate to be an in-line intermediate but could be an isomerization to a different catalytic form of ArsA, such as ArsA 1 . We conclude that kinetic Scheme 4 provides a minimal model to account for both the kinetics of the ArsA catalyzed ATPase reaction and of the fluorescence profile that results from the ATP-induced conformational changes in ArsA.