Evidence for a Transition State Analog, MgADP-Aluminum Fluoride-Acetate, in Acetate Kinase from Methanosarcina thermophila *

Aluminum fluoride has become an important tool for investigating the mechanism of phosphoryl transfer, an essential reaction that controls a host of vital cell func-tions. Planar AlF 3 or AlF 4 (cid:1) molecules are proposed to mimic the phosphoryl group in the catalytic transition state. Acetate kinase catalyzes phosphoryl transfer of the ATP (cid:2) -phosphate to acetate. Here we describe the inhibition of acetate kinase from Methanosarcina thermophila by preincubation with MgCl 2 , ADP, AlCl 3 , NaF, and acetate. Preincubation with butyrate in place of acetate did not significantly inhibit the enzyme. Several NTPs can substitute for ATP in the reaction, and the corresponding NDPs, in conjunction with MgCl 2 , AlCl 3 , NaF, and acetate, inhibit acetate kinase activity. Fluorescence quenching experiments indicated an increase in binding affinity of acetate kinase for MgADP in the presence of AlCl 3 , NaF, and acetate. These and other characteristics of the inhibition indicate that the transition state analog, MgADP-aluminum fluoride-acetate, forms an abortive complex in the active site. The protection from inhibition by a non-hydrolyzable ATP analog or acetylphosphate, in conjunction with the strict dependence of inhibition on the presence of both ADP and acetate, supports a direct in-line mechanism for acetate kinase. Acetate kinase, which catalyzes the phosphoryl transfer of the ATP (cid:1) -phosphate to acetate, was

Acetate kinase, which catalyzes the phosphoryl transfer of the ATP ␥-phosphate to acetate, was one of the earliest phosphoryl transfer enzymes to be recognized. Since the discovery of acetate kinase in 1944 by Lipmann (1) and isolation in 1954 by Ochoa (2), most research on the catalytic mechanism has been performed with the Escherichia coli enzyme. The early research led to two proposed catalytic mechanisms, a tripledisplacement mechanism via a covalent phosphoenzyme intermediate and a direct in-line mechanism. The direct in-line transfer mechanism is consistent with steady state kinetics (3) and also stereochemical evidence (4), which indicates that there are an odd number of in-line phosphoryl transfers (5,6). The triple displacement mechanism was proposed when a phosphoenzyme was reported after incubation with radiolabeled ATP or acetylphosphate (7), and this phophoenzyme was found to be chemically competent to transfer the phosphoryl group to ADP or acetate (7). However, the triple displacement mecha-nism was challenged when it was shown that phosphorylated acetate kinase from E. coli is a phosphoryl donor to Enzyme I of the bacterial phosphotransferase system, which suggested an alternate function for phosphorylated acetate kinase in sugar transport (8). There has been very little mechanistic work on acetate kinase, since this intriguing triple displacement mechanism was proposed by Spector in 1980 (5).
The thermostable Methanosarcina thermophila acetate kinase has been cloned and hyper-produced in E. coli making it the first available for modern biochemical approaches including site-directed mutagenesis (9 -11) and a crystal structure (12). The crystal structure identifies acetate kinase as a member of the superfamily now identified as the acetate and sugar kinases/Hsc70/actin (ASKHA) superfamily (12,13). Glucose binding to hexokinase results in partial closure of the active site cleft and the subsequent binding of ATP is proposed to result in even greater closure of the cleft (14,15). When glucose is absent, the phosphates and metal of the metal-nucleotide complex are proposed to be disordered possibly to reduce the inherent ATPase activity of hexokinase (14 -16). Acetate kinase is expected to undergo an analogous conformational change upon substrate binding. Consistent with this hypothesis, the ␤-phosphate appears to be displaced in the current crystal structure of acetate kinase in complex with ADP (12); hence, identification of residues involved in catalysis from the crystal structure is precluded. Thus, a crystal structure of acetate kinase in the transition state bound to both MgADP and acetate as well as an analog of the ␥-phosphoryl group of ATP is key to the identification of residues involved in transition state stabilization and the orientation of the substrates for catalysis.
Aluminum fluoride has recently become a powerful tool in the study of phosphoryl transfer enzymes. In several phosphatases and kinases, AlF 3-4 mimics the planar phosphoryl group in the catalytic transition state (17)(18)(19). Here we present evidence for a transition state analog, MgADP-aluminum fluoride-acetate, for the M. thermophila acetate kinase. To our knowledge, this work represents the first use of aluminum fluoride to mimic a transition state in an acetate kinase or in any kinase from the ASKHA superfamily. In conjunction with recent site-directed replacement studies, our results favor a direct in-line transfer of the ATP-␥-phosphoryl group to acetate and provide a new direction in which to investigate other unanswered questions regarding the mechanism of acetate kinase.

EXPERIMENTAL PROCEDURES
Chemicals-The following chemicals were of the highest purity commercially available. Aluminum chloride hydrate, magnesium chloride hexahydrate, sodium fluoride, and potassium hydroxide (for pH adjustment) were purchased from Aldrich. The monopotassium salt of adenosine 5Ј-diphosphate was purchased from Roche Molecular Biochemi-cals. Potassium acetate, E. coli acetate kinase, and BisTris 1 were purchased from Sigma. Sodium fluoride solutions were made and stored exclusively in plastic containers. All reactions were performed in plastic tubes. The remaining chemicals were of high purity. The sodium salts of inosine 5Ј-diphosphate, cytidine 5Ј-diphosphate, and propionic acid were purchased from Sigma. The tetralithium salt of ATP␥S (78% pure) and the lithium potassium salt of acetylphosphate (90% pure) were also from Sigma. The sodium salt of butyric acid was purchased from Aldrich. Beryllium chloride was purchased from Alfa Aesar. The dipotassium salt of bis-ANS was purchased from Molecular Probes.
Heterologous Production and Purification of the M. thermophila Acetate Kinase-Recombinant acetate kinase with a 6-residue N-terminal histidine tag was heterologously produced in E. coli BL21(DE3) as described previously (10). The recombinant acetate kinase was purified using a Ni-nitrilotriacetic acid silica spin kit (Qiagen) according to manufacturer's instructions. The enzymes were eluted in 50 mM NaH 2 PO 4 buffer (pH 7.0) containing NaCl (300 mM) and imidazole (250 mM). Protein concentrations were determined by the Bradford method (20), using protein dye reagent (Bio-Rad) and bovine serum albumin as the standard.
Enzyme Activity Assays-Acetate kinase activity was routinely determined with the hydroxamate assay (21), which detects the formation of acetylphosphate from acetate and ATP. An enzyme-coupled assay (21), in which ATP formation is linked to the reduction of NADP through hexokinase and glucose-6-phosphate dehydrogenase, was used to determine the K m value of ADP.
Inhibition of Enzyme Activity-M. thermophila acetate kinase (0.4 M dimer) was preincubated in 100 mM BisTris buffer (pH 5.5) with MgCl 2 (10 mM), ADP (10 mM), AlCl 3 (0.1 mM), NaF (0.5 mM), and potassium acetate (25 mM) at ambient temperature for 30 min (200 l total). A portion of the preincubation mix (20 l) was then assayed for activity (350 l total) unless otherwise noted. E. coli acetate kinase (5 units) was preincubated and assayed in the same manner as described for the M. thermophila enzyme. Fluorescence Measurements-A Hitachi F-2000 fluorescence spectrophotometer was used to measure bis-ANS fluorescence. The excitation and emission wavelengths were 390 and 500 nm, respectively. Incubation mixes (1 ml total) contained the indicated concentrations of MgADP, M. thermophila acetate kinase (0.4 M dimer), and bis-ANS (0.01 mM) in 100 mM BisTris buffer (pH 5.5). When noted, AlCl 3 (0.1 mM), NaF (0.5 mM), and potassium acetate (25 mM) were also present in the incubation mixes. Fluorescence intensities were measured after a 2-h incubation at ambient temperature.
Initial Incubation with ATP␥S or Acetylphosphate-Acetate kinase (0.4 M dimer) was incubated with ATP␥S (0.5-10 mM) or acetylphosphate (0.5-50 mM) in 100 mM BisTris buffer (pH 5.5) for 5 min at ambient temperature (160 l total). At the end of this initial incubation, MgCl 2 (10 mM), ADP (10 mM), AlCl 3 (0.1 mM), NaF (0.5 mM), and potassium acetate (25 mM) were added to the reaction (200 l total). The preincubation was allowed to proceed for another 30 min at ambient temperature before 20 l of the preincubation mix was assayed for acetate kinase activity (350 l total).

RESULTS
Inhibition of M. thermophila Acetate Kinase by MgCl 2 , ADP, AlCl 3 , NaF, and Acetate-Preincubation with a mixture of MgCl 2 , ADP, AlCl 3 , NaF, and acetate inhibited the activity of acetate kinase from M. thermophila. The inhibition was dependent on the preincubation time with a first-order rate constant of 0.24 Ϯ 0.01 min Ϫ1 (Fig. 1). While this inhibition rate is slower than expected for an enzyme-ligand interaction, the rate can be attributed to the slow formation of aluminum fluoride complexes as investigated previously (22). When MgCl 2 , ADP, and acetate were omitted from the preincubation mixture, there was no detectable loss of activity, indicating that AlCl 3 and NaF alone are not inhibitory (Fig. 1, inset). The results indicate that MgCl 2 , ADP, AlCl 3 , NaF, and acetate are required The enzyme activity was then assayed and plotted relative to 100% activity (600 mol of acetylphosphate/min/mg of protein) obtained from a control mixture in which AlCl 3 and NaF were omitted. Inset, the time course of the activity assay after preincubation for 30 min with: complete inhibitory mixture (q), minus NaF and AlCl 3 (f), minus MgCl 2 , ADP and acetate (ࡗ), minus all components except buffer (OE).
for maximum inhibition (Table I). There was a corresponding decrease in enzyme activity as the concentration of each component of the preincubation mixture was increased in the presence of fixed concentrations of all other components (Fig. 2). Preincubation of acetate kinase with MgCl 2 , ADP, acetate, and AlCl 3 (1-100 M) failed to inhibit enzyme activity (data not shown) verifying fluoride-dependent inhibition and demonstrating AlCl 3 is not inhibitory to the enzyme at the concentrations tested. Replacement of AlCl 3 with BeCl 2 (1-200 M) in the preincubation mix did not inhibit acetate kinase activity (data not shown). The data for Fig. 2 were fit to an analog of the Hill equation (Equation 1) (23), where i is percent inhibition, i max is the maximum percent inhibition, K i app is the apparent inhibition constant, [I] is the concentration of inhibitor, and n is the Hill coefficient (plots not shown). The apparent K i acetate was determined to be 2.9 Ϯ 0.8 mM, 7-fold less than the previously reported K m value of 20 mM (21), and an n value equal to 1.0 Ϯ 0.1 was determined for acetate. The apparent K i ADP value was determined to be 18.4 Ϯ 0.7 M, 130-fold less than the experimental K m value of 2.4 mM (21). The n value for NaF, was found to be 3.0 Ϯ 0.8, suggesting three fluoride atoms are required for inhibition. The apparent K i values for magnesium and AlCl 3 were found to be 0.95 Ϯ 0.02 mM and 3.4 Ϯ 1.1 mM, respectively.
ATP␥S was tested as a substrate of acetate kinase and found to be a non-hydrolyzable inhibitor (data not shown). Acetate kinase was incubated with ATP␥S or acetylphosphate before and during preincubation with MgCl 2 , ADP, AlCl 3 , NaF, and acetate to determine whether ATP␥S or acetylphosphate protected against the inhibition. A corresponding decrease in the inhibition of enzymatic activity was observed as the concentration of ATP␥S (Fig. 3) or acetylphosphate (Fig. 4) increased. There was an apparent lag in the relief of inhibition by ATP␥S, while acetylphosphate protection leveled off at 30% inhibition. One possibility for the lag is that the ATP analog ATP␥S binds the enzyme with a lower affinity than ATP, thus affording less protection from the transition state analog. The residual 30% inhibition can possibly be attributed to turnover of the enzyme during preincubation with ADP and acetylphosphate, which allowed for formation of the transition state complex.
Effect of Alternative Substrates on the AlCl 3 and NaF-dependent Inhibition-Several NTPs can replace ATP for the M. thermophila enzyme. The reported specific activities (mol of acetylphosphate/min/mg of protein) with ATP, ITP, UTP, or CTP are 412, 342, 330, and 218, respectively (21). Preincuba-tion of acetate kinase with MgCl 2 , AlCl 3 , NaF, acetate, and either IDP, UDP, or CDP in place of ADP resulted in almost complete inhibition of activity (Table II). To determine the phosphoryl acceptor requirements for formation of the inhibitory complex, acetate was substituted with propionate or butyrate. Activity of the M. thermophila enzyme with propionate is 60% of that with acetate, whereas butyrate is not a substrate (21). Preincubation with MgCl 2 , ADP, AlCl 3 , NaF, and propionate resulted in almost complete inhibition of activity; however, preincubation with butyrate in place of acetate did not significantly inhibit the enzyme (Table II).
Fluorescence Monitoring of MgADP Binding-Buffered solutions of either acetate kinase or bis-ANS had very low intrinsic fluorescence; however, when these two components were incubated together there was an ϳ20-fold enhancement of fluorescence indicative of bound bis-ANS. The fluorescence quenching dependent on the MgADP concentration was used to examine the MgADP binding to the acetate kinase in the absence or presence of AlCl 3 , NaF, and acetate (Fig. 5). When AlCl 3 , NaF, and acetate were absent, the pattern of fluorescence quenching was biphasic in response to increasing MgADP concentrations.  A robust first phase of fluorescence quenching below 10 M MgADP was followed by a weaker second phase at higher concentrations. There was a similar pattern when AlCl 3 , NaF, and acetate were present; however, the first phase was more pronounced than in the absence of AlCl 3 , NaF, and acetate (Fig.  5). Two possible explanations for the first phase of fluorescence quenching are: (i) a MgADP-induced environment change around a bound bis-ANS molecule or (ii) displacement of bound bis-ANS by MgADP. In favor of the latter mechanism, it has been reported that bis-ANS binds to hydrophobic pockets and with particularly high affinity to nucleotide binding sites (24,25). Although the results do not distinguish between these two possibilities, the first phase of fluorescence quenching most likely reflects nucleotide binding at the high affinity catalytic site. The weaker second phase may be attributed to a nonspecific site that binds bis-ANS and has low affinity for MgADP. Consequently, the first phase of fluorescence quenching was used to examine MgADP binding affinity. The results indicate an increase in binding affinity of acetate kinase for MgADP in the presence of AlCl 3 , NaF, and acetate. When either acetate or AlCl 3 plus NaF were omitted from the reaction mixture, the pattern of fluorescence quenching was similar to that observed in the absence of all three of these components (data not shown). This result suggests that the increase in binding affinity for MgADP is dependent on the presence of AlCl 3 , NaF, and acetate but not AlCl 3 plus NaF alone or acetate alone.
Inhibition of E. coli Acetate Kinase by MgCl 2 , ADP, AlCl 3 , NaF, and Acetate-E. coli acetate kinase activity was fully inhibited by the mixture of MgCl 2 , ADP, AlCl 3 , NaF, and acetate (Table I) using the same experimental conditions as for the M. thermophila enzyme. The E. coli enzyme retained 80% or higher activity when each component was individually omitted from the preincubation mixture (Table I), indicating that all of the components are necessary for maximum inhibition. DISCUSSION Studies on several kinases including F 1 -ATPase and UMP kinase indicate that AlF 3-4 mimics the planar phosphoryl group in the catalytic transition state and, with the substrates, forms an abortive active site complex (17,19). Here we report the inhibition of M. thermophila acetate kinase activity by MgCl 2 , ADP, AlCl 3 , NaF, and acetate. The characteristics of the inhibition and other experiments indicate that a transition state analog, MgADP-aluminum fluoride-acetate, forms an abortive complex in the active site of acetate kinase.
The requirements for formation of a transition state analog should mimic the requirements for catalysis. If MgADP-aluminum fluoride-acetate is a transition state analog for the acetate kinase, then any substrate that substitutes for acetate should form a transition state analog in conjunction with MgCl 2 , ADP, AlCl 3 , and NaF. Propionate replaces acetate as a substrate (21), and as shown here inhibited the enzyme when preincubated with MgCl 2 , ADP, AlCl 3 , NaF; however, butyrate is not a substrate for acetate kinase (21) and did not form an inhibitory complex with MgCl 2 , ADP, AlCl 3 , and NaF. These results support MgADP-aluminum fluoride-acetate as a transition state analog. A distinguishing characteristic of acetate kinase is the ability to utilize various NTPs (21) in place of ATP. Each of the cognate NDPs (in conjunction with MgCl 2 , ADP, AlCl 3 , NaF, and acetate) resulted in the inhibition of acetate kinase activity, further supporting the formation of a transition state analog. These results also indicate that inhibition of acetate kinase is not the consequence of a nonspecific interaction of aluminum fluoride with the adenosine moiety of ADP to form an inactive nucleotide complex. Although ADP is not a ␤-phosphoryl group donor in the acetate kinase reaction, AMP was able to partially replace ADP in the proposed transition state analog. AMP has previously been shown to be a competitive inhibitor versus ATP in the E. coli acetate kinase (K i of 7.4 mM) (26). MgAMP, aluminum fluoride, and acetate may form a transition state analog despite the absence of the ␤-phosphate.  a The enzyme (0.4 M dimer) was preincubated for 30 min with MgCl 2 (10 mM), the indicated nucleotide (10 mM), AlCl 3 (0.1 mM), NaF (0.5 mM), and acetate, propionate, or butyrate as indicated (25 mM), before the activity was assayed. b 100% (approximately 600 mol of acetylphosphate/min/mg of protein) is the activity of the enzyme preincubated for 30 min with the indicated concentrations of MgCl 2 , nucleotide, and co-substrate.
Either the nucleotide ␤-phosphate group is not essential for formation of the active conformation of acetate kinase or aluminum fluoride substitutes for both the nucleotide ␤and ␥-phosphates in the MgAMP-aluminum fluoride-acetate complex. The ability of MgCl 2 , ADP, AlCl 3 , NaF, and acetate to inhibit acetate kinases from both E. coli (Bacteria domain) and phylogenetically distant M. thermophila (Archaea domain) indicates that the inhibition is related to the common mechanism of catalysis and not a nonspecific event.
Although not always observed, the propensity of enzymes to maximize protein-substrate interactions in the transition state predicts tighter binding of substrates. The fluorescence quenching results indicate that acetate kinase has a higher binding affinity for MgADP when AlCl 3 , NaF, and acetate are present. Consistent with these results are the much lower K i values determined for ADP and acetate (in the presence of MgCl 2 , AlCl 3 , NaF, and the co-substrate) relative to the K m values for ADP and acetate. In conclusion, the fluorescence and K i data are consistent with the formation of a transition state analog, MgADP-aluminum fluoride-acetate.
The primary metal complexes involved in acetate kinase inhibition were MgADP and aluminum fluoride; however, the 23% inhibition of acetate kinase by the preincubation mixture minus AlCl 3 suggests the formation of magnesium fluoride, which is also inhibitory. In support of this scenario, magnesium fluoride, in addition to aluminum fluoride, is proposed to form in the transition state analog of G proteins (27). Consequently, a portion of the metal fluoride in the acetate kinase transition state analog is probably magnesium fluoride. The dependence of acetate kinase inhibition on the AlCl 3 concentration suggests the predominant metal fluoride species in the transition state analog is aluminum fluoride (28). Beryllium failed to replace aluminum in the inhibitory complex. Beryllium fluoride, in association with a bridging ␤-phosphate oxygen, is thought to mimic the tetrahedral nucleotide ␥-phosphate in the ground state (29,30). Consequently, beryllium fluoride is not a true analog of the planar phosphoryl group during catalysis. Despite this fact, preincubation with MgADP, BeCl 2 , and NaF inhibits several phosphoryl transfer enzymes, including adenylate kinase and F 1 -ATPase (17,31). Acetate kinase may differ from these enzymes in the inability to form a locked active conformation without a planar phosphoryl analog in the inhibitory complex.
The stoichiometry of the aluminum fluoride complex that forms in acetate kinase is uncertain. It has been proposed that pH plays a significant role in the aluminum to fluoride ratio of the transition state analog (32). The acetate kinase was inhibited in a preincubation mixture containing MgCl 2 , ADP, AlCl 3 , NaF, and acetate at a pH value of 5.5. At this pH, AlF 4 Ϫ is proposed to be the dominant species in the transition state analogs of kinases (32) and thus may be the species formed in acetate kinase. A planar AlF 4 Ϫ molecule would not strictly mimic the planar phosphoryl group during catalysis. Nevertheless, transition state analogs containing AlF 4 Ϫ have been shown to be reasonable mimics of the transition state (29,33,34). The nature of the nucleophilic attack may also help to determine the aluminum fluoride species in the transition state analog (30). Crystal structures of G␣ i (33) and G␣ t GTPases (34), as well as myosin (29), indicate that an octahedrally coordinated aluminum is present in the transition state analog with four equatorial fluoride atoms, a nucleotide ␤-phosphate oxygen as one apical ligand, and a water molecule as the second apical ligand. Hydrolytic nucleotidases such as these may have more structural flexibility to accept AlF 4 Ϫ than would enzymes that utilize a protein functional group as the nucleophile (30). In nucleoside diphosphate kinase for example, in which an active site histidine attacks the ␥-phosphate, steric considerations were invoked to explain the presence of AlF 3 instead of AlF 4 Ϫ in the crystal structure (30). If the above scenario is correct, the requirement for acetate in the proposed acetate kinase transition state analog suggests that this molecule acts as the nucleophile and further suggests that the aluminum fluoride species is AlF 3 , not AlF 4 Ϫ . This supposition is supported by a binding coefficient of ϳ3 that was determined for NaF in formation of the inhibitory complex.
Since the 1970s, there has been dispute over a single versus triple displacement mechanism for acetate kinase. The single displacement mechanism assumes direct in-line phosphoryl transfer from ATP to acetate, whereas the triple displacement mechanism involves phosphoryl transfer to two enzyme sites before transfer to acetate. Here we report the requirement for both ADP and acetate in formation of the transition state analog. The most straightforward interpretation of these results is that the phosphorylation of acetate by ATP occurs by a direct in-line mechanism. The protection from inhibition by non-hydrolyzable ATP␥S or acetylphosphate suggests that the terminal phosphate of ATP, the phosphate group of acetylphosphate, and the aluminum fluoride of the transition state analog all share the same binding site. These results are consistent with a direct in-line mechanism. Other recent site-directed mutagenesis studies reported for the M. thermophila enzyme support the direct in-line mechanism. These studies have eliminated the possibility of active site histidine residues acting as phosphorylation sites in the catalytic mechanism (11). In addition, active site residues Arg 91 and Arg 241 were found to be essential for catalysis (10). These arginines are well positioned in the crystal structure to stabilize the transition state in a direct in-line mechanism (12). If the acetate kinase reaction involves a direct in-line phosphoryl transfer, an alternative role must be proposed for active site residue Glu 384 , previously found to be essential for catalysis and postulated as the phosphorylation site of acetate kinase (9). Instead of acting as a catalytic phosphorylation site, Glu 384 may be involved in magnesium binding. In support of this proposal, variant Glu 384 3 Ala requires a 30-fold increase in the concentration of magnesium necessary for half-maximal velocity relative to that of the wild-type enzyme (35). Thus, the results presented here combined with recent reports support a direct in-line mechanism. Finally, the work presented here lays the foundation for further research to elucidate the role of active site residues and the catalytic mechanism for this enzyme.